`WATER FOR A HEALTHY COUNTRY FLAGSHIP
Flow dependent ecological
responses
Kingsford, R.T., Watts, R.J., Koehn, J.D., Thompson, R.M. & Sims, N.C.
(eds)
A technical report from the Ecological Responses to Altered Flow Regimes
Flagship Research Cluster (SubProject 3)
Flow dependent ecological responses | i
Water for a Healthy Country Flagship Report series. ISSN: 1835-095X
Australia is founding its future on science and innovation. Its national science agency, CSIRO, is a
powerhouse of ideas, technologies and skills. CSIRO initiated the National Research Flagships to
address Australia’s major research challenges and opportunities. They apply large scale, long term,
multidisciplinary science and aim for widespread adoption of solutions. The Flagship Collaboration
Fund supports the best and brightest researchers to address these complex challenges through
partnerships between CSIRO, universities, research agencies and industry.
Consistent with Australia’s national interest, the Water for a Healthy Country Flagship aims to
develop science and technologies that improve the social, economic and environmental outcomes
from water, and deliver $3 billion per year in net benefits for Australia by 2030.
http://www.csiro.au/org/HealthyCountry.
For more information about the Flagship Program visit www.csiro.au/flagships.
Citation
Kingsford, R.T., Watts, R.J., Keohn, J.D. Thompson, R. & Sims, N.C. (eds) (2014) Flow dependent
ecological responses. A technical report from the Ecological Responses to Altered Flow Regimes
Flagship Research Cluster (SubProject 3). CSIRO Water for a Healthy Country Flagship, Australia.
Copyright
© 2014 CSIRO. To the extent permitted by law, all rights are reserved and no part of this publication
covered by copyright may be reproduced or copied in any form or by any means except with the
written permission of CSIRO.
Important disclaimer
CSIRO advises that the information contained in this publication comprises general statements
based on scientific research. The reader is advised and needs to be aware that such information may
be incomplete or unable to be used in any specific situation. No reliance or actions must therefore
be made on that information without seeking prior expert professional, scientific and technical
advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all
liability to any person for any consequences, including but not limited to all losses, damages, costs,
expenses and any other compensation, arising directly or indirectly from using this publication (in
part or in whole) and any information or material contained in it.
The Ecological Responses to Altered Flow Regimes Collaboration Cluster acknowledgements:
The Ecological Responses to Altered Flow Regimes Cluster is a collaboration between: CSIRO Water
for a Healthy Country Flagship; Griffith University; The University of New South Wales; Monash
University; Charles Sturt University; La. Trobe University; Victorian Department of Environment and
Primary Industry (Arthur Rylah Institute).
ii | Flow dependent ecological responses
Contributors
Cluster Leader
CSIRO Cluster Project Leader
Ecological models
Flow regimes and ecological assets
Flow dependent ecological responses
Assessing ecosystem condition and trend
Project Support
Prof Stuart Bunn (Griffith University)
Dr Neil Sims (CSIRO)
Team leaders: Dr Simon Linke (Griffith University) and
Dr Carmel Pollino (CSIRO)
Other key staff:
Dr Lyndsey Bradford (Griffith University)
Dr Alice Brown (CSIRO)
Prof Jon Olley (Griffith University)
Team leader: Professor Stuart Bunn (Griffith University)
Other key staff:
Dr Nick Bond (Griffith University)
Dr Mark Kennard (Griffith University)
Dr Simon Linke (Griffith University)
Dr Neil Sims (CSIRO)
Dr Doug Ward (Griffith University)
Team leader: Professor Richard Kingsford
(University of New South Wales)
Other key staff:
Dr Leah Beesley (Arthur Rylah Institute/University of
WA)
Rebecca Durant (La Trobe University)
Dr Ben Gawne (La Trobe University)
Dr Mike Grace (Monash University)
A/Prof Alison King (Arthur Rylah Institute/Charles
Darwin Univ)
Dr John Koehn (Arthur Rylah Institute)
Dr Keller Kopf (Charles Sturt University)
Dr Heather McGinness (CSIRO)
Dr Amina Price (La Trobe University)
Dr Shiquan Ren (University of NSW)
A/Prof Robyn Watts (Charles Sturt University)
Dr Ross Thompson (Monash University/University of
Canberra)
Dr Skye Wassens (Charles Sturt University)
Dr Jess Wilson (La Trobe University)
Team leaders: Prof Jenny Davis (Monash
University/University of Canberra) and A/Prof Fran
Sheldon (Griffith University)
Other key staff:
Dr Stephen Balcombe (Griffith University)
Prof Max Finlayson (Charles Sturt University)
Dr Ben Gawne (La Trobe University)
Dr Sally Hladyz (Monash University)
Dr Keller Kopf (Charles Sturt University)
Dr Erin Peterson (CSIRO)
Dr Amina Price (La Trobe University)
Dr Jess Wilson (La Trobe University)
Linda Merrin (CSIRO)
Malcolm Hodgen (CSIRO)
Flow dependent ecological responses | iii
Contents
1
Introduction
8
Background ........................................................................................................................................................ 8
Conceptual framework .................................................................................................................................... 10
In-channel Ecosystems ..................................................................................................................................... 12
Floodplain and wetland ecosystems ................................................................................................................ 17
Precis 20
References ....................................................................................................................................................... 21
2
Research project reports
28
2.1
In-channel environmental watering in the Edward-Wakool system .................................................. 28
Introduction ........................................................................................................................................ 28
The Edward-Wakool system ............................................................................................................... 28
Assessment of ecosystem responses to environmental watering in the Edward-Wakool
system in 2011-12............................................................................................................................... 31
Conclusion .......................................................................................................................................... 32
Acknowledgements ............................................................................................................................ 32
References .......................................................................................................................................... 33
2.2
Lateral connectivity and availability of slackwater habitat in the Edward-Wakool system
during in-channel flows.................................................................................................................................... 34
Introduction ........................................................................................................................................ 34
Methods.............................................................................................................................................. 35
Results 36
Discussion and conclusion .................................................................................................................. 41
Acknowledgements ............................................................................................................................ 42
References .......................................................................................................................................... 42
2.3
Whole stream metabolism ................................................................................................................. 44
Introduction........................................................................................................................................ 44
Methods.............................................................................................................................................. 45
Results and Discussion........................................................................................................................ 47
References .......................................................................................................................................... 54
2.4
Leaf-litter breakdown rates in response to changes in flow regime .................................................. 56
Introduction ........................................................................................................................................ 56
Methods.............................................................................................................................................. 57
Results and Discussion........................................................................................................................ 58
Acknowledgements ............................................................................................................................ 62
References .......................................................................................................................................... 62
2.5
Invertebrate response to in-channel unregulated flow pulses and small in-channel
environmental flows ........................................................................................................................................ 65
Introduction ........................................................................................................................................ 65
Methods.............................................................................................................................................. 66
Results and Discussion........................................................................................................................ 68
Acknowledgements ............................................................................................................................ 74
References .......................................................................................................................................... 74
2.6
Timing of fish spawning and recruitment in lowland river channel ecosystems ............................... 77
Introduction ........................................................................................................................................ 77
Methods.............................................................................................................................................. 77
Results and Discussion........................................................................................................................ 79
Conclusion .......................................................................................................................................... 82
Acknowledgements ............................................................................................................................ 82
References .......................................................................................................................................... 82
iv | Flow dependent ecological responses
2.7
The importance of flow on spawning of riverine native fish in the Barmah-Millewa region,
Murray River .................................................................................................................................................... 84
Introduction ........................................................................................................................................ 84
Methods.............................................................................................................................................. 85
Results and Discussion........................................................................................................................ 87
References .......................................................................................................................................... 90
2.8
Antecedent flow: predicting the condition of young-of-year fish in a temperate lowland
river system? .................................................................................................................................................... 93
Introduction ........................................................................................................................................ 93
Methods: ............................................................................................................................................ 94
Results and Discussion:....................................................................................................................... 97
References .......................................................................................................................................... 99
2.9
Hypoxic blackwater survival thresholds of four Australian lowland river fishes.............................. 101
Introduction ...................................................................................................................................... 101
Methods............................................................................................................................................ 101
Results/Discussion ............................................................................................................................ 103
Conclusion ........................................................................................................................................ 108
Acknowledgements .......................................................................................................................... 109
References ........................................................................................................................................ 109
2.10
Flow changes to the Lowbidgee floodplain from the Murrumbidgee river ..................................... 111
Introduction ...................................................................................................................................... 111
Methods............................................................................................................................................ 111
Results and Discussion...................................................................................................................... 114
References ........................................................................................................................................ 116
2.11
Inundation modelling in the Macquarie Marshes ............................................................................ 119
Introduction ...................................................................................................................................... 119
Methods............................................................................................................................................ 119
Results and discussion ...................................................................................................................... 123
References ........................................................................................................................................ 126
2.12
Wetland and floodplain understory vegetation ............................................................................... 128
Introduction ...................................................................................................................................... 128
Methods............................................................................................................................................ 129
Results and Discussion...................................................................................................................... 130
Conclusions ....................................................................................................................................... 136
References ........................................................................................................................................ 137
2.13
Blackbox (Eucalyptus largiflorens) vegetation and bird communities in the Lowbidgee
Floodplain – Flood regimes driving community transitions in semi-arid floodplain woodlands................... 140
Introduction ...................................................................................................................................... 140
Methods............................................................................................................................................ 141
Results 143
Discussion ......................................................................................................................................... 146
Acknowledgements .......................................................................................................................... 147
References ........................................................................................................................................ 148
2.14
Antecedent flow and juvenile fish abundance in floodplain wetlands of the mid-Murray
River 152
Introduction ...................................................................................................................................... 152
Methods............................................................................................................................................ 153
Results and discussion ...................................................................................................................... 156
Conclusions ....................................................................................................................................... 160
References ........................................................................................................................................ 161
Flow dependent ecological responses | v
2.15
Hydrological and biophysical predictors of frog occupancy patterns across the MurrayDarling Basin .................................................................................................................................................. 168
Introduction ...................................................................................................................................... 168
Methods............................................................................................................................................ 169
Results 171
Discussion ......................................................................................................................................... 173
References ........................................................................................................................................ 174
2.16
The role of antecedent conditions in determining responses to environmental flows ................... 177
Introduction ...................................................................................................................................... 177
The initial condition and natural variability...................................................................................... 178
The role of short term disturbance history ...................................................................................... 181
The role of medium term disturbance history ................................................................................. 182
The role of long term disturbances .................................................................................................. 182
Conclusion ........................................................................................................................................ 184
References ........................................................................................................................................ 184
3
Summary of research implications
187
Changes in benthic wetted area and availability of slackwater in response to in-channel
flows 187
Whole stream metabolism ............................................................................................................... 187
Organic matter breakdown .............................................................................................................. 188
Invertebrate response to in-channel flow pulses ............................................................................. 188
Fish spawning ................................................................................................................................... 188
Predicting the condition of young-of-year fish................................................................................. 189
Hypoxic blackwater fish kills ............................................................................................................. 189
Flow and inundation models ............................................................................................................ 190
Wetland and floodplain vegetation .................................................................................................. 190
Antecedent flows and the fish community ...................................................................................... 191
Frogs in the Murray-Darling Basin .................................................................................................... 192
vi | Flow dependent ecological responses
1 Introduction
Background
Freshwater ecosystems around the world are threatened by many different factors
including changes to flow regimes affecting habitats, pollution, climate change,
overharvesting, pest species and disease (Millennium Ecosystem Assessment 2005;
Dudgeon et al. 2006). Many of the threats interact to produce confounded impacts on
ecosystems, dependent biota and processes. Anthropogenic changes to flow regimes are
widespread and affect all hydrological, geomorphological ecological processes. They affect
flow quantity and temporal and spatial variability, directly impacting on the requirements of
many flow dependent organisms and processes.
Flow regimes define the character of a river and its dependent ecosystems and their
characteristic biota and driving processes. The size, frequency, timing and duration of flows
and their associated floods provide the template for understanding the responses of
different organisms and dependent processes at all the different spatial and temporal scales
associated with a river. Periods of low flow are also important restructuring phases in the
organization of freshwater ecosystems (Lake, 2000b, Lake, 2003). Freshwater ecosystems
are highly dependent on the full range of flows from large structuring floods (booms) to the
periods of low to no flow (bust) as well as flows in between which trigger a subset of
process and responses from dependent organisms (Kingsford et al. 1999; Bunn et al. 2006).
High flows and floods can trigger the germination of aquatic plants and emergence of
invertebrate biota (Brock et al. 2006; Boulton et al. 2006). These form the food web for
higher order organisms such as frogs, fish species and waterbirds which can occur in large
abundances, particularly in wetlands, following large flooding. Floods which move across the
floodplain connect different parts of the river, allowing for the movement of invertebrates
and fish which colonise highly productive aquatic habitats on the floodplain. These
integrated processes are critical to the entire food web, linking the many different
dependent organisms and processes operating at longitudinal and lateral dimension on the
river. Humans have disrupted all of these processes, particularly through the building of
dams, extraction of water and development of structures on floodplains that alter flow
connectivity across floodplains (Kingsford 2000; Bunn and Arthington 2002; Arthington and
Pusey 2003; Steinfeld and Kingsford 2012).
Flow regulation changes the flow regime across a range of temporal and spatial scales
(Figure 1). Specifically, this relates to effects on the size, timing, frequency and duration of
specific flow events. The effects of flow regulation vary among systems and may include
reductions in average or event-specific discharges. River regulation may also change in the
frequency, duration and timing of specific flow events such as floods, cease-to-flow periods
8 | Flow dependent ecological responses
or in-channel flow pulses. Such changes are widespread across rivers of the Murray-Darling
Basin, which has a long history of river regulation. In recent years, governments have
invested in the buyback of water to provide environmental flows across the Murray-Darling
Basin with the aim of restoration of ecosystems as well improving planning and
management frameworks. Such restorative actions will be effected through mechanisms
such as the Murray-Darling Basin Plan, its watering strategies and water sharing plans. There
is increasing need to focus on effective ways of measuring the impacts of different
management approaches on ecosystem responses in freshwater ecosystems.
Figure 1. Conceptual spatial framework used to focus research on the ecological responses to flow in the
Murray-Darling Basin (inset) focusing on the Edward-Wakool in-channel system (EW) and three major
floodplain wetlands: Macquarie Marshes (MQ), Lowbidgee wetland (LB) and the Barmah-Millewa Forest
(BM). These conformed with the spatial conceptual framework for freshwater ecosystems in the MurrayDarling Basin, typical of regulated river systems with large dams in the upper catchment (D) and including
river channels (C) with their in-channel benches (B) and floodplains downstream (F, stippled area) with
distributary creeks and more perennial wetlands (W).
To maximize the value of this research we chose to focus our efforts on systems where
there was already substantial scientific knowledge amongst the project team, which
capitalised on available long-term data on existing knowledge of the organisms and
processes operating at each site. This provided more explanatory power over the high
variability experienced by rivers and wetlands, thereby increasing our ability to quantify
Flow dependent ecological responses | 9
relationships among the considerable spatial and temporal variability always encountered in
freshwater ecosystems. Consequently, the research focus was on four major research areas
in the Murray-Darling Basin including one in-channel system, the Edward-Wakool river
systems and three floodplain systems; the Macquarie Marshes, the Lowbidgee wetland and
the Barmah-Millewa Forest, each respectively supplied by the Macquarie River, the
Murrumbidgee River and the River Murray (Figure 1).
Conceptual framework
In order to best integrate the research being conducted in this large multi-disciplinary
project, we developed a generic conceptual model that could be applied to different biota
or processes within in-channel ecosystems. This model could be applied across different
systems (in-channel, floodplains), organisms (e.g. flood-dependent vegetation, fish) or
processes (e.g. food web productivity). There was clear understanding and agreement that
flow was a major driver of ecosystem responses, but that flow’s influence may be direct or
indirect through its influence on food, competition or predators for instance (Figure 2, Table
1). The team undertaking this work were highly experienced in their respective fields, and
drawn from a range of disciplines. Consequently, while there was an appreciation that
other factors may also influence ecological responses to flows, this model provided a
sufficient framework to guide our collective research efforts.
Figure 2. . Generic conceptual model used to define the responses of different organisms or processes
(target) for measuring ecological responses to flows and inundation of floodplains and wetlands. This
conceptual model was applied across different freshwater ecosystems (in-channel, floodplain and wetland)
and the range of organisms and processes (Table 1).
Flow and inundation are the primary drivers which can be measured by various elements
and provide the input into physical changes; inputs, production and food resource;
competition; outputs, consumption and predators which influence the target organisms or
processes (Table 1; Figure 2). Flow is the major driver of inundation of floodplains (Thomas
et al. 2011) although local rainfall is sometimes important. Both flow and inundation have
key measures which are important for ecological responses including timing, frequency,
10 | Flow dependent ecological responses
duration and magnitude as well as other indices. These can all influence ecosystem
responses in different and subtle ways. Similarly flow is a fundamental factor determining
the type of physical habitat available whether these are floodplains or in-channel benches
(Figure 1) but mediated by connectivity (Table 1).
Together flow and the physical habitat available provide the source for the inputs,
production or food resources which drive much of the biological response of freshwater
ecosystems (Table 1, Figure 2). These are not always available for target organisms but may
be intercepted by competitors which can include exotic species. Ultimately remaining
resources in the form of outputs are available for target organisms or processes, once
predation is also considered. Finally, there are a range of aspects of an organisms biology
that can be measures of ecosystem response to flow as well as various ecosystem processes
(Table 1, Figure 2).
Table 1. Drivers of ecological responses and metrics that can be used to identify ecological responses within
a generic conceptual framework (see Figure 2).
Driver
Measures
Flow and inundation
Timing; frequency; duration; magnitude (area, height); depth; rate of
rise and fall; sequence (pattern of events); time since last event
Physical
Scale; geomorphology; habitat (type, structure, size, connectivity)
Inputs/ production/
food resources
Sediment; carbon; nutrients; organic matter; new growth; nectar;
seeds; invertebrates (micro and macro)
Competition
Invasive species; nomadic species; migratory species; resident
species; chemical competitors
Outputs/
consumption/
predators
Carbon; nutrients; organic matter; new growth; nectar; seeds;
invertebrates (micro and macro); fire; human impacts
Target ecological
response
Processes including productivity, rate of reproduction, dispersal,
survival, persistence and extinction and states including biodiversity
assemblage (diversity, species richness, abundances), breeding effort,
breeding success
We derived specific conceptual models for each of the organisms and processes investigated
in this project. While these organisms and processes are inextricably linked to one another
through river flows and flooding conditions, they can be conveniently defined spatially in
terms of (i) in-channel, and (2) floodplain and wetlands.
Flow dependent ecological responses | 11
In-channel Ecosystems
Flow is a major structuring influence on river in-channel ecosystems, affecting physical
habitat availability (including hydraulic environment, sediment composition and water
quality), patterns of connectivity (longitudinal, lateral and vertical) or inputs to ecosystem
functions (e.g. nutrient cycling, primary production and decomposition). Flow also drives inchannel interactions with other systems, including groundwater, tributaries, wetlands,
floodplains, downstream reaches, the estuary and ocean. As a consequence, there is no
known component of a river channel ecosystem that is not influenced by flow, although the
influence may be either direct or indirect and may act over a large range of spatial and
temporal scales.
Despite the importance of flow to river channels, there is still considerable uncertainty
around the role of flow in determining the condition of in-channel river ecosystems. As the
temporal and spatial scales decrease, the level of uncertainty grows. At the largest
temporal scales of the long-term flow regime (i.e. up to 100 years), there is strong evidence
of the relationship between flow regulation and declines in river condition. At a decadal
time-scale, there is also enough knowledge to identify ecosystem water requirements to
support identification of environmental flow requirements. For environmental flow
management and restoration, there remains considerable uncertainty in understanding
responses of in-channel ecosystems to individual flows or sequences of flows which can
inform decisions at fine temporal scales (i.e. days, months, years).
For instream systems, the most direct influence of flow is on the availability of physical
habitat in rivers, which is one of four principles identified as mechanisms by which flow
affects biodiversity (Bunn and Arthington, 2002). Using fish as an example, the key physical
habitat attributes include hydraulic habitat (i.e. velocity and depth), structure (e.g. instream
vegetation, substrate) and water quality, including water temperature, dissolved oxygen
and salinity (Mathews, 1998). Different species and life-stages display habitat preferences,
reflected in high levels of spatial heterogeneity driving spatial segregation and co-existence
of multiple species (Amoros and Bornette, 2002, Bunn and Arthington, 2002, Schiemer and
Zalewski, 1992, Schlosser, 1991). In riverine systems, the key driver of physical habitat
heterogeneity is the interaction between geomorphology and flow (Copp, 1989, Lake,
2000a, Schlosser, 1991, Ward et al., 1998).
Given the profound influence of flow on river channel ecosystems, Bunn and Arthington
(2002) proposed that aquatic species have evolved life history strategies adapted to natural
flow regimes and their associated patterns of habitat availability (Principle 2). While there is
a wealth of evidence supporting this principle from wetland and floodplains, there is less
certainty concerning the role of individual flow events on Australian river channels.
12 | Flow dependent ecological responses
Flow and physical habitat also influence the availability of resources required to sustain
processes or populations. For example, flow influences the amount of dissolved organic
carbon available to microbial decomposition which can, under appropriate conditions, lead
to major blackwater events (Kerr et al., 2013). Flow also influences the amount of food
available to higher trophic levels. For example, rate of primary production, whether by
algae (Leland, 2003, Sherman et al., 1994) macrophytes (Franklin et al., 2008, Greet et al.,
2013) or trees (Cunningham et al., 2009, Sims et al., 2009) are influenced by flow and then
provide food or resources for microbial or metazoan communities.
Flow also has the capacity to influence competitive interactions. Bunn and Arthington
(2002) proposed that flow mediated changes may facilitate the invasion and success of
exotic and introduced species (Principle 4). For example, loss of flooding creates
opportunities for riparian weeds (Catford et al., 2011) while stabilization of flows can
promote willows (Stokes, 2008, Stokes and Cunningham, 2006). Within the fish community,
both carp and Gambusia are believed to do better when flow regimes are modified (Gehrke
et al., 1999, Walker et al., 1995).
Projects within this study focussed on two broad areas: ecosystem functions (primary
production and decomposition) and responses of the fish community. Primary production
(Figure 3) is the formation of organic matter by plants (macrophytes and algae). Algae and
macrophytes provide a source of food (Bunn et al., 2003, Hladyz et al., 2012, Zeug and
Winemiller, 2008), and habitat for many high level consumers (Humphries, 1996, Warfe and
Barmuta, 2006). The rate of primary production in aquatic ecosystems is spatially and
temporally variable and influenced by many factors besides flow, including light, nutrients,
carbon dioxide (e.g. turbulence, depth, reach residence time) (Reynolds, 1996).
Flow dependent ecological responses | 13
Magnitude / area
Duration
Depth
Aquatic Primary
Production
Flow
Physical
Benthic Algae
Macrophyte
Phytoplankton
Inputs
Habitat Mosaic
Season
Riparian Shading
Topographic shading (e.g.
incised channels)
(Soil/Sediment) Substrate Type
Substrate Stability
Particle size distribution
Organic matter
Bioavailable
Phosphorus &
Nitrogen
+
Light (PAR)
Loss of NOx as
N2
Figure 3. Conceptual model developed for primary production.
Decomposition (Figure 4) is influenced by flow which transports allochthonous organic
matter (as particulate or dissolved material) into the water body (for example, during
floodplain inundation (Junk et al., 1989)) and the location of autochthonous and
allochthonous organic matter decomposition (Webster et al., 1999). Flow also degrades
organic material by abrasion (Wipfli and Merritt, 1994); Ryder et al. 2004) releases organic
matter from the soil matrix (Findlay and Sinsabaugh, 1999) and controls habitat quality for
macro-invertebrates (Figure 5) (Gore et al., 2001) which in turn consume or shred organic
material (Graça, 2001). Photodegradation is also promoted by low, shallow flows but
inhibited by highly coloured or turbid water (Amon and Benner, 1996).
14 | Flow dependent ecological responses
Figure 4. Conceptual model developed for leaf litter breakdown.
The second focus area was the fish community (Figure 6). The native fish community in the
MDB has undergone significant modification with dramatic declines in the abundance and
distribution of a significant proportion of the Murray-Darling Basin’s species. Of the 46
naturally occurring native species, 26 are of conservation concern (Lintermans 2007). These
include species that have undergone localized extinction such as, Trout cod (Maccullochella
macquariensis), and Macquarie Perch (Macquaria australasica) or suffered widespread or
sudden rapid declines such as Silver Perch (Bidyanus bidyanus), Murray Cod (Maccullochella
peelii) and Freshwater Catfish (Tandanus tandanus) (Lintermans 2007). Native fish have
been subjected to a number of stressors, including over-exploitation (Humphries and
Winemiller 2009), habitat loss (e.g. de-snagging, Koehn et al. 2004), population
fragmentation due to barriers (Barrett and Mallen-Cooper 2006) and invasive species
(Koehn 2004).
Flow dependent ecological responses | 15
Figure 5. Conceptual model developed for aquatic macroinvertebrates
In addition to these stressors, flow modification has played a significant role in decline in the
condition of the native fish community (Gehrke et al. 1995, 1999; Kingsford et al. 2011;
Arthington and Pusey 2003). One of the key pathways by which flow modification is
believed to affect native fish is through its influence of reproduction and recruitment. For
example, recruitment of some Australian native fish species is cued by flooding while others
appear dependent on low flows (Humphries et al., 1999). Understanding the ways that flow
has affected native fish will be important in restoring native fish diversity in the MDB.
16 | Flow dependent ecological responses
Magnitude / area
Timing
Duration
Depth
Rate of rise and fall
Frequency
Sequence (pattern of events)
Time since last event
Flow
+
+
+
Physical
habitat
Food
production
Competition
Predators
Condition
+/-
Abundance
+
Water quality:
DO, temp, nutrients,
salinity,
Phytoplankton
Zooplankton
Macroinvertebrates
Fish
+
Body condition
Individual growth
rate
Relative abundance
Spawning (eggs, larval abundance)
Recruitment (young-of-year)
Dispersal
Figure 6. Conceptual model for fish community responses.
Floodplain and wetland ecosystems
Floodplains are ecotone ecosystems that straddle aquatic and terrestrial environments.
The unpredictable and changeable hydrological regime across the floodplain creates a
mosaic of varying habitat conditions in terms of water perenniality, depth and chemistry,
the composition of terrestrial and riparian vegetation communities and submerged
structural complexity (Amoros and Bornette 2002; Boulton and Brock 1999; Ralph and
Rogers 2011). Floodplains may also contain more highly diverse aquatic communities than
adjacent riverine habitats (Boulton and Lloyd 1991; Hillman and Shiel 1991).
Connectivity between the in-channel ecosystem and other flow-dependent ecosystems (e.g.
floodplains) underpins two major conceptualisations of river systems: the River Continuum
Concept (RCC) (Vannote et al., 1980) which emphasized the transport of organic matter
from upstream to downstream, and the Flood Pulse Concept (FPC) (Junk and Wantzen,
2004) which highlights the importance of lateral connectivity in facilitating the exchange of
organic matter between rivers and floodplains. These processes are critical for maintaining
stocks of bioavailable carbon (Baldwin et al. 2013) and for the maintenance of populations
of riverine species (Principle 3; Bunn and Arthington 2002).
In this project we focussed on four main elements of floodplain and wetland ecosystems:
aquatic macrophytes (Figure 7), terrestrial floodplain vegetation (Figure 8), the relationship
between Black Box (Eucalyptus largiflorens) and woodland bird communities (Figure 9) and
frogs (Figure 10).
Flow dependent ecological responses | 17
Magnitude / area
Timing
Duration
Depth
Rate of rise and fall & flow velocity
Frequency
Sequence (pattern of events)
Time since last event
Habitat Types
Woodlands:
•Black Box, Red Gum, Coolibah
Wetlands:
•Open water, Woodland
Understorey/macrophytes:
•Lignum
Nitregoosefoot
•Canegrass
•Phragmites
•Typha
Groundcover:
•Grass /herb/ forblands
Physical
habitat
Flow
+
+
+
+
?
Food
production
Competition
+
Macrophyte primary production
Predators
-/+
-/+
open water wetland, red
Nutrients?
Growth of macrophytes
Invasive spp.
gum and black box
Fixing carbon in to floodplain
Light availability
woodlands have different
ecosystem
historic water regimes, as
a result different
macrophyte community
structures and periods
when habitat is suitable for
macrophyte germination
and establishment
Floodplain inundation is predicted to have a positive on effect on net floodplain primary production by;
a)
increasing soil moisture and producing conditions which enable macrophyte germination and growth.
b)
it is uncertain how inundation will effect soil nutrient status and what effect this has on macrophyte production.
c)
flow will ultimately increase competition and predation but it is uncertain if this will result in increased or
decreased primary production.
The strength of the relationships will change spatio-temporally and depending on habitat type;
a)
Flood duration needs to be sufficient for germination and recruitment.
b)
Flood duration will change interspecies competition (particularly light) as communities move through different
successionally stages possibly having a negative effect of production.
c)
Changes in depth will favour particular species that have different growth rates and change light availability.
d)
Rate of production will be influence by season/temperature (timing)
Figure 7. Conceptual model for Floodplain macrophytes
Habitat Types
Woodlands:
•Black Box, Red Gum, Coolibah
Wetlands:
•Open water, Woodland
Magnitude / area
Timing
Duration
Depth
Rate of rise and fall
Frequency
Sequence (pattern of events)
Time since last event
Understorey/macrophytes:
•Lignum
•Nitregoosefoot
•Canegrass
•Phragmites
•Typha
Groundcover:
•Grass /herb/ forblands
Physical habitat
Geomorphology
Sediment / soil type
Soil moisture
Soil air
Soil temperature
Shear stress
Light availability
Flow
Food production
Organic matter
Nutrients
Competition
Invasive spp
e.g.
•Lippia
•Scleroleana
•Giant rush
•‘Terrestrialisation
Other factors affecting Target
-Grazing
-Fire
-Clearing
-Veg patch size and connectivity
-Extinction debts
Predators
•Invertebrates
•Native mammals
•Bushbirds
•Waterbirds
•Reptiles
•Domestic and
feral mammals
(e.g. rabbits)
Floodplain vegetation
(Vegetation community, metapopulation, population)
Productivity (function)
-Growth, condition
-Reproduction – flowering,
seedset, seedfall, viability
-Dispersal
-Germination
-Seedling establishment
-Survival
Biodiversity (assemblage /
structure)
-Diversity, species richness
-Structural complexity
-Age structure
-Distribution and extent
-Density
Figure 8. Conceptual model for floodplain terrestrial vegetation
18 | Flow dependent ecological responses
Event (on-site or adjacent)
-Magnitude / area
-Timing
-Duration
-Depth
-Rate of rise and fall
-Antecedent conditions (flow history)
Physical
Other factors
-Grazing
-Fire
-Clearing
-Veg patch
size and
connectivity
-Extinction
debts
Allocation
Flow
Food
Woodland type:
•Black Box
•Coolibah
•Red Gum
•Upland
Wetland type:
•Temporary
•Semipermanent
•Permanent
Structure and
composition of:
•Overstorey
•Understorey
•Groundcover
Upstream rainfall
Local rainfall
Breeding hab itat
vs. feeding hab itat
Resident woodland
birds
Predators
Competitors
New growth
Nectar
Seed
Invertebrates
•(Leaf, sap, and
water quality)
•(Aquatic,
terrestrial)
Small
vertebrates
Regime (flow history on-site or
adjacent)
-Time since last event
- Prior wetting frequency
- No. of events in past 10 years
- Degree of alteration from natural
(Community, meta-population,
population, guild)
Invasive spp
Nomadic spp
Migratory spp
Raptors
Reptiles
Foxes
Cats
Humans
Raptors
Reptiles
Foxes
Cats
Bees
Ants
Rabbits
Macropods
Possums
Bats
Productivity (function)
-Abundance
-Breeding effort
-Breeding success (e.g. fledgling
survival)
-Dispersal to floodplain and/or
upland communities
-Survival
-Persistence / Extinction
Biodiversity (assemblage)
-Diversity
-Species richness
Figure 9. Blackbox and woodland bird conceptual model
Magnitude / area
Timing
Duration (hydroperiod)
Frequency
Sequence (pattern of events)
Time since last event
Flow
Physical habitat
Spatial habitat types:
• Floodplain (spatial
arrangement of rain
fed, flood fed and
persistent refuge
habitats
• Flooding pathways
and movement
corridors
• Wetland hydrological
heterogeneity
•
Aquatic vegetation
complexity and
structure
• Tree hollows (tree
frogs)
• Burrowing sites
(burrowing
species)(soil type,
soil moisture,
ground water
quality and depth)
Food production
Terrestrial and
semi-aquatic
Macroinvertebrates
Size ranges and
nutrient value of
macroinvertebrates
Competition
Interspecific
competition
Predators
Predators:
Frogs
Water birds
Snakes
Foxes and pigs
Target
Population persistence
Age structure
Movement patterns
(dispersal/colonisation/movements)
Recruitment into adult population
Other factors affecting Target
climatic patterns
existing community structure,
abundance of adults, reproductive
fitness of adults
habitat patchiness (availability of refuge
habitats, connectivity and dispersal
pathways)
land use and terrestrial habitat
Figure 10. Conceptual model for frogs
Flow dependent ecological responses | 19
Australian floodplain wetlands are areas of extraordinary biodiversity (Kingsford 1995;
Boulton and Lloyd 1991; Roberts and Ludwig 1991). Yet, despite their ecological significance
they are increasingly under threat in Australia (Kingsford 2000) and throughout the world
(e.g. Sparks 1995). Water resource development is the principle threat to the decline in the
number and health of wetlands (Dynesius and Nilsson 1994; Kingsford 2000; Bunn and
Arthington 2002). The construction of dams and weirs, and the cumulative impact of
diversions have either reduced connectivity or entirely isolated many floodplains and
wetlands from their main river channel.
The broad implications of river regulation are widely known (see Bunn and Arthington
2002), and in wetland systems the impacts can vary from localised loss of aquatic species
adapted to unpredictable flooding regimes, to complete transition of wetlands to terrestrial
ecosystems. For taxa such as waterbirds, fish and frogs that utilise wetlands for breeding
and rearing of their young, reduced flooding may have significant impacts on wider,
regional-scale populations (Kingsford and Johnson 1999). However, we have very limited
understanding about the possible lag effects of flooding, or how the less frequent, larger
floods influence longer-term populations or processes (Kingsford et al. 1999).
Precis
Environmental flows, which involves providing water to an ecosystem to maintain its
structure and functioning, aim to alleviate some of the effects of river regulation (Arthington
2012). However, water resource managers must aim to optimise the use and allocation of
environmental water so they can efficiently meet multiple ecological targets. A key part of
this process is to improve our understanding of the responses of organisms and processes to
environmental flows. The aim of this project was to contribute to that improved
understanding through a series of targeted field campaigns, which are described in this
report.
Section 2 contains the research project reports. In general, each study is presented as a
stand-alone report. However, this section commences with a description of the EdwardWakool river system (Watts et al., Section 2.1), which is the study area for several
subsequent research reports contained in this volume (Sections 2.2 to 2.6). Section 2 closes
with a review of the importance of antecedent conditions for determining contemporary
ecological responses to changes in flow.
Section 3 summarises the management and research implications of this work in terms of
broader fields of research.
20 | Flow dependent ecological responses
References
Amon, R. and Benner, R. (1996) Photochemical and microbial consumption of dissolved
organic carbon and dissolved oxygen in the Amazon River system. Geochimica et
Cosmochimica Acta 60, 1783-1792.
Amoros, C. and Bornette, G. (2002) Connectivity and biocomplexity in waterbodies of
riverine floodplains. Freshwater Biology, 47, 761-776.
Arthington, A.H. & Pusey, B.J. (2003) Flow restoration and protection in Australian rivers.
River Research and Applications 19, 377-395.
Atkinson, B.L., Grace, M.R., Hart, B.T., & Vanderkruk, K.E.N. (2008) Sediment instability
affects the rate and location of primary production and respiration in a sand-bed
stream. Journal of the North American Benthological Society 27, 581-592.
Bond, N., Thomson, J. & Reich, P. (2012) Macroinvertebrate responses to antecedent flow,
longterm flow regime characteristics and landscape context in Victorian rivers.
National Water Commission, Canberra.
Boulton, A.J. & Lake, P.S. (1992) The ecology of two intermittent streams in Victoria,
Australia, III, temporal changes in faunal composition’. Freshwater Biology 27, 123–
138.
Boulton, A. J., Sheldon, F. & Jenkins, K. M. (2006) Natural disturbance and aquatic
invertebrates in desert rivers. In: Ecology of Desert Rivers (ed. Kingsford, R.T.) pp 133153. Cambridge University Press, Cambridge.
Brock, M. A., Capon, S. J. & Porter, J. L. (2006) Disturbance of plant communities dependent
on desert rivers. In: Ecology of Desert Rivers (ed. Kingsford, R.T.) pp 100-132.
Cambridge University Press, Cambridge.
Bunn, S. E. & Arthington, A. H. (2002) Basic principles and ecological consequences of
altered flow regimes for aquatic biodiversity. Environmental Management 30, 492507.
Bunn, S. E., Davies, P. M. & Winning, M. (2003) Sources of organic carbon supporting the
food web of an arid zone floodplain river. Freshwater Biology 48, 619-635.
Bunn, S. E., Thoms, M. C., Hamilton, S. K. & Capon, S. J.( 2006) Flow variability in dryland
rivers: Boom, bust and the bits in between. River Research and Applications 22, 179186.
Catford, J. A., Downes, B. J., Gippel, C. J. & Vesk, P. A. (2011) Flow regulation reduces native
plant cover and facilitates exotic invasion in riparian wetlands. Journal of Applied
Ecology 48, 432-442.
Cellot, B., Mouillot, F. & Henry, C. P. (1998) Flood drift and propagule bank of aquatic
macrophytes in a riverine wetland. Journal of Vegetation Science 9, 631-640.
Chessman, B.C. (2009) Climatic changes and 13-year trends in stream macroinvertebrate
assemblages in New South Wales, Australia. Global Change Biology 15, 2791–2802
Copp, G. H. (1989) The habitat diversity and fish reproductive function of floodplain
ecosystems. Environmental Biology of Fishes 26, 1-26.
Flow dependent ecological responses | 21
Crook, D. A. (2004a) Is the home range concept compatible with the movements of two
species of lowland river fish? Journal of Animal Ecology 73, 353-366.
Crook, D. A. (2004b) Movements associated with home-range establishment by two species
of lowland river fish. Canadian Journal of Fisheries and Aquatic Sciences 61, 21832193.
Cunningham, S. A., Mac Nally, R., Read, J., Baker, P., White, M., Thomson, J. R. & Griffioen, P.
(2009) A Robust Technique for Mapping Vegetation Condition Across a Major River
System. Ecosystems 12, 207-219.
Downes, B. J., Bellgrove, A. & Street, J. L. (2005) Drifting or walking? Colonisation routes
used by different instars and species of lotic, macroinvertebrate filter feeders. Marine
and Freshwater Research 56, 815-824.
Dudgeon. D., Arthington, A. H., Gessner, M.O., Kawabata, Z., Knowler, D.J., Lévêque, C.,
Naiman, R.J., Prieur-Richard, A., Soto, D., Stiassny, M.L.J. & Sullivan, C.A. (2006)
Freshwater biodiversity: importance, threats, status and conservation challenges.
Biological Reviews 81, 163-182.
Ebner, B., Johnston, L. & Lintermans, M. (2009) Radio-tagging and tracking of translocated
trout cod (Maccullochella macquariensis: Percichthyidae) in an upland river. Marine &
Freshwater Research 60, 346-355.
Findlay, S. & Sinsabaugh, R. (1999) Unravelling the sources and bioavailability of dissolved
organic matter in lotic aquatic ecosystems. Marine and Freshwater Reseasrch 50, 781790.
Franklin, P., Dunbar, M. & Whitehead, P. (2008) Flow controls on lowland river
macrophytes: A review. Science of the Total Environment 400, 369-378.
Gehrke, P., Astles, K. & Harris, J. (1999) Within-catchment effects of flow alteration on fish
assemblages in the Hawkesbury-Nepean River system, Australia. Regulated Rivers:
Research and Management 15, 181-198.
Gehrke, P. C., Brown, P., Schiller, C.B., Moffatt, D.B. and Bruce, A.M. (1995) River regulation
and fish communities in the Murray-Darling river system, Australia. Regulated RiversResearch & Management 11, 363-375.
George, R.J., McFarlane, D.J. & Speed, R.S. (1995) The consequences of a changing
hydrological environment for native vegetation in South-western Australia. In: The
Role of Networks- Nature Conservation 4 (ed. D.A. Saunders, J.L. Craig and E.M.
Mattiske) pp 9-21. CSIRO Sydney: Surrey-Beatty, Sydney.
Gore, J., Layzer, J. & Mead, J. (2001) Macroinvertebrate instream flow studies after 20 years:
A role in stream management and restoration. Regulated Rivers: Research and
Management 17, 527-542.
Graça, M. (2001) The role of invertebrates on leaf litter decomposition in streams – a
review. International Review of Hydrobiology 86, 383-393.
Greet, J., Cousens, R. D. & Webb, J. A. (2013) Seasonal timing of inundation affects riparian
plant growth and flowering: implications for riparian vegetation composition. Plant
Ecology 214, 87-101.
22 | Flow dependent ecological responses
Gregory, K. J. (2006) The human role in changing river channels. Geomorphology 79, 172191.
Groves, J. H., Williams, D. G., Caley, P., Norris, R. H. & Caitcheon, G. (2009) Modelling of
floating seed dispersal in a fluvial environment. River Research and Applications 25,
582-592.
Gurnell, A., Thompson, K., Goodson, J. & Moggridge, H. (2008) Propagule deposition along
river margins: linking hydrology and ecology. Journal of Ecology 96, 553-565.
Hey, D.L. & Philippi, N.S. (1999) A Case for Wetland Restoration. Wiley-Interscience.
Hladyz, S., Nielsen, D. L., Suter, P. J. & Krull, E. S. (2012) Temporal variations in organic
carbon utilization by consumers in a lowland river. River Research and Applications 28,
513-528.
Hobbs, R.J. & Norton, D.A. (1996) Towards a Conceptual Framework for Restoration Ecology.
Restoration Ecology 4, 93–110
Humphries, P. (1996) Aquatic macrophytes, macroinvertebrate associations and water levels
in a lowland Tasmanian river. Hydrobiologia 321, 219-233.
Humphries, P., King, A. J. & Koehn, J. D. (1999) Fish, flows and flood plains: links between
freshwater fishes and their environment in the Murray-Darling system, Australia.
Environmental Biology of Fishes 56, 129-151.
James, A. B. W., Dewson, Z. S. & Death, R. G. (2008) The effect of experimental flow
reductions on macroinvertebrate drift in natural and streamside channels. River
Research and Applications 24, 22-35.
Jones, M. J. & Stuart, I. G. (2007) Movements and habitat use of common carp (Cyprinus
carpio) and Murray cod (Maccullochella peelii peelii) juveniles in a large lowland
Australian river. Ecology of Freshwater Fish 16, 210-220.
Jordan III, W.R. (2003) The Sunflower Forest: Ecological Restoration and the New
Communion with Nature. University of California Press, California.
Junk, W. J., Bayley, P. B. & Sparks, R. E. (1989) The flood pulse concept in river-floodplain
systems. In: Proceedings of the international large river symposium (ed. Dodge, D. P)
pp 110-127 Canadian Special Publication on Fisheries and Aquatic Sciences 106.
Junk, W. J. & Wantzen, K. (2004) The flood pulse concept: New aspects, approaches and
applications - an update. In: Proceedings of the Second International Symposium on
the Management of Large Rivers for Fisheries. Bangkok. RAP: Food and Agriculture
Organization & Mekong River Commission FAO Regional Office for Asia and the Pacific.
Kerr, J. L., Baldwin, D. S. & Whitworth, K. L. (2013) Options for managing hypoxic blackwater
events in river systems: A review. Journal of Environmental Management, 114, 139147.
King, J., Brown, C. & Sabet, H. (2003) A scenario-based holistic approach to environmental
flow assessments for rivers. River Research and Applications 19, 619-639.
Kingsford, R. T. (2000) Ecological impacts of dams, water diversions and river management
on floodplain wetlands in Australia. Austral Ecology 25, 109-127.
Flow dependent ecological responses | 23
Kingsford, R.T. & Thomas, R.F. (1995) The Macquarie Marshes and its waterbirds in arid
Australia: A 50-year history of decline. Environmental Management 19, 867-878.
Kingsford, R.T. & Thomas, R.F. (2004) Destruction of wetlands and waterbird populations by
dams and irrigation on the Murrumbidgee River in arid Australia. Environmental
Management 34, 383-396.
Kingsford, R.T., Jenkins, K.M. & Porter, J.L. (2004) Imposed hydrological stability on lakes in
arid Australia and effect on waterbirds. Ecology 85, 2478-2492.
Kingsford, R. T., Curtin, A. L. & Porter, J. (1999) Water flows on Cooper Creek in arid
Australia determine 'boom' and 'bust' periods for waterbirds. Biological Conservation
88, 231-248.
Koehn, J.D. (2004) Carp (Cyprinus carpio ) as a powerful invader in Australian waterways.
Freshwater Biology 49, 882-894.
Koehn, J.D, Nicol, S.J. and Fairbrother, P.S. (2004) Spatial arrangements and physical
characteristics of structural woody habitat in a lowland river in south-eastern
Australia. Aquatic Conservation: Marine and Freshwater Ecosystems 14, 457-464.
Koster, W. M. & Crook, D. A. (2008) Diurnal and nocturnal movements of river blackfish
(Gadopsis marmoratus) in a south-eastern Australian upland stream. Ecology of
Freshwater Fish 17, 146-154.
Lake, P. (2000) Disturbance, patchiness, and diversity in streams. Journal of the North
American Benthological Society 19, 573-592.
Larned, S. T., Datry, T., Arscott, D. B. & Tockner, K. (2010) Emerging concepts in temporaryriver ecology. Freshwater Biology 55, 717-738.
Leigh, C., Sheldon, F., Kingsford, R.T. and Arthington, A.H. (2010) Sequential floods drive
'booms' and wetland persistence in dryland rivers: a synthesis. Marine and Freshwater
Research 61, 896-908.
Leland, H. V. (2003) The influence of water depth and flow regime on phytoplankton
biomass and community structure in a shallow, lowland river. Hydrobiologia 506, 247255.
Lintermans M (2007) Fishes of the Murray-Darling Basin: An introductory guide. MurrayDarling Basin Commission, Canberra
Mac Nally, R., Cunningham, S.C., Baker, P.J., Horner, G.J. and Thomson, J.R. (2011) Dynamics
of Murray-Darling floodplain forests under multiple water stressors - the past, present
and future of an Australian icon. Water Resources Research 47, 11pp.
Mathews, W. (1998) Patterns in Freshwater Fish Ecology. Chapman & Hall, New York.
Merritt, D. & Wohl, E. (2002) Processes governing hydrochory along rivers: Hydraulics,
hydrology, and dispersal phenology. Ecological Applications 12, 1071-1087.
Merritt, D. M., Nilsson, C. & Jansson, R. (2010) Consequences of propagule dispersal and
river fragmentation for riparian plant community diversity and turnover. Ecological
Monographs 80, 609-626.
Millennium Ecosystem Assessment (2005) Ecosystems and human well-being: synthesis.
Island Press, Washington.
24 | Flow dependent ecological responses
Moggridge, H. L. & Gurnell, A. M. (2010) Hydrological controls on the transport and
deposition of plant propagules within riparian zones. River Research and Applications
26, 512-527.
Nilsson, C., Andersson, E., Merritt, D. M. & Johansson, M. E. (2002) Differences in riparian
flora between riverbanks and river lakeshores explained by dispersal traits. Ecology
83, 2878-2887.
Palmer, M.A., Liermann, C.A.R., Nilsson, C., Florke, M., Alcamo, J., Lake, P.S. and Bond, N.
(2008) Climate change and the world’s river basins, anticipating management options.
Frontiers in Ecology and the Environment 6, 81–89.
Petts, G. E. & Gurnell, A. M. (2002) Dams and geomorphology: Research progress and future
directions. Geomorphology 71, 27-47.
Pickett, S.T.A and Parker, V.T. (1994) Avoiding the Old Pitfalls: Opportunities in a New
Discipline, Restoration Ecology 2, 75–79.
Poff, N.L., Allan, J.D., Bain, M.B., Karr, J.R., Prestegaard, K.L., Richter, B.D., Sparks, R.E. and
Stromberg, J.C. (1997) The natural flow regime – a paradigm for river conservation
and restoration. Bioscience 47, 769-784.
Postel, S. and Richter, B. (2003) Rivers of life: Managing water for people and nature. Island
Press, Washington.
Prosser, I., Rutherford, I., Olley, J., Young, W., Wallbrink, P. & Moran, C. (2001) Large-scale
patterns of erosion and sediment transport in river networks, with examples from
Australia. Marine and Freshwater Research 52, 81-99.
Puckridge, J. T., Walker, K. F. & Costelloe, J. F. (2000) Hydrological persistence and the
ecology of dryland rivers. Regulated Rivers-Research & Management 16, 385-402.
Ren S.Q. & Kingsford R.T. (2011) Statistically Integrated Flow and Flood Modelling Compared
to Hydrologically Integrated Quantity and Quality Model for Annual Flows in the
Regulated Macquarie River in Arid Australia. Environmental Management 48, 177-188.
Reynolds, C. (1996) The production, biomass and structure of phytoplankton in large rivers.
Archive fur Hydrobiologia. Supplement 113, 161-187.
Robinson, A. T., Clarkson, R. W. & Forrest, R. E. (1998) Dispersal of larval fishes in a
regulated river tributary. Transactions of the American Fisheries Society 127, 772-786.
Rodriguez, M. A. (2002) Restricted movement in stream fish: The paradigm is incomplete,
not lost. Ecology 83, 1-13.
Roy, A.H., Rosemond, A.D.,Paul, M.J., Leigh, D.S. and Wallace, J.B. (2003) Stream
macroinvertebrate response to catchment urbanisation. Freshwater Biology 48, 329 –
346.
Ryder D.S., Watts R.J., Nye E., & Burns, A. 2006. Can flow velocity regulate epixylic biofilm
structure in a regulated floodplain river? Marine and Freshwater Research 57(1), 29-36.
Schiemer, F. & Zalewski, M. (1992) The importance of riparian ecotones for diversity and
productivity of riverine fish communities. Netherlands Journal of Zoology 42, 323-335.
Schlosser, I. (1991) Stream fish ecology: a landscape perspective. Bioscience 41, 704-712.
Flow dependent ecological responses | 25
Schludermann, E., Tritthart, M., Humphries, P. & Keckeis, H. (2012) Dispersal and retention
of larval fish in a potential nursery habitat of a large temperate river: an experimental
study. Canadian Journal of Fisheries and Aquatic Sciences 69, 1302-1315.
Sherman, B. S., Jones, G. J. & Webster, I. T. (1994) Flow,stratification and the growth of
algae and cyanobacteria in Maude weir pool on the Murrumbidgee River.
Environmental Flows Seminar Proceedings. Australian Water and Wastewater
Association, Canberra.
Sims, N., Colloff, M. & Guerschman, J.-P. (2009) Basin-scale vegetation. In: Ecological
outcomes of flow regimes in the Murray-Darling Basin (eds. Overton, I., Colloff, M.,
Doody, T., Henderson, B. & Cuddy, S.). CSIRO, Canberra.
Steinfeld, C.M.M. and Kingsford, R.T. (2012) Disconnecting the floodplain: Earthworks and
their ecological effect on a dryland floodplain in the Murray-Darling Basin, Australia.
Rivers Research and Applications 29, 206-218.
Stokes, K. E. 2008. Exotic invasive black willow (Salix nigra) in Australia: influence of
hydrological regimes on population dynamics. Plant Ecology 197, 91-105.
Stokes, K. E., Cunningham, S. A. (2006) Predictors of recruitment for willows invading
riparian environments in south-east Australia: implications for weed management.
Journal of Applied Ecology 43, 909-921.
Thomas R.F., Kingsford R.T., Lu Y. and Hunter S.J. (2011) Landsat mapping ofannual
inundation (1979−2006) of the Macquarie Marshes in semi-arid Australia.
International Journal of Remote Sensing 32, 4545−4569.
Thomson, J.R., Thompson, R.M., Metzeling, L., Reich, P., Bond, N.R., Cunningham, S.C. and
Mac Nally, R. (2012) The influence of vegetation, flows and climate variation on
stream biota: Lessons from the Big Dry in southern Australia. Global Change Biology
18, 1582-1596.
Thorp, J. H. & Delong, M. D. (1994) The riverine productivity model: an heuristic view of
carbon sources and organic processing in large river ecosystems. Oikos 70, 305-308.
Vannote, R. L., Minshall, G. W., Cummins, K. W., Sedell, J. R. & Cushing, C. E. (1980) The river
continuum concept. Canadian Journal of Fisheries and Aquatic Sciences 37, 130-137.
Walker, K. F., Sheldon, F. & Puckridge, J. T. (1995) An ecological perspective on large dryland
rivers. Regulated Rivers: Research and Management 11, 85-104.
Ward, J. V., Bretschko, G., Brunke, M., Danielopol, D., Gibert, J., Gonser, T. & Hildrew, A. G.
(1998) The boundaries of river systems: the metazoan perspective. Freshwater Biology
40, 531-569.
Warfe, D. M. & Barmuta, L. A. (2006) Habitat structural complexity mediates food web
dynamics in a freshwater macrophyte community. Oecologia 150, 141-154.
Watts R.J., Allan C., Bowmer K.H., Page K.J., Ryder D.S. & Wilson A.L. 2009. Pulsed flows: a
review of environmental costs and benefits and Best practice. Waterlines report,
National Water Commission, Canberra.
Webster, J. R., Benfield, E. F., Ehrman, T. P., Schaeffer, M. A., Tank, J. L., Hutchens, J. J. &
D'angelo, D. J. (1999) What happens to allochthonous material that falls into streams:
26 | Flow dependent ecological responses
A synthesis of new and published information from Coweeta. Freshwater Biology 41,
687-706.
White, J. L. & Harvey, B. C. (2003) Basin-scale patterns in the drift of embryonic and larval
fishes and lamprey ammocoetes in two coastal rivers. Environmental Biology of Fishes
67, 369-378.
Wipfli, M. S. & Merritt, R. W. (1994) Disturbance to a stream food web by a bacterial
larvicide specific to black flies: feeding responses of predatory macroinvertebrates.
Freshwater Biology 32, 91-103.
Zeug, S. & Winemiller, K. O. (2008) Evidence supporting the importance of terrestrial carbon
in a large-river food web. Ecology 89, 1733-1743.
Flow dependent ecological responses | 27
2 Research project reports
2.1
In-channel environmental watering in the Edward-Wakool
system
Watts, R.J.1, Bowen, P.2, Conallin, J.2, Grace, M.3, Hladyz, S.3, Howitt, J.A.1, Kopf, R.K.1,
McCasker, N.1, Thompson, R.4, Wassens, S.1
1
Institute for Land, Water and Society, Charles Sturt University
Murray Catchment Management Authority
3
Water Studies Centre, School of Chemistry and School of Biological Sciences, Monash University
4
Institute for Applied Ecology, University of Canberra
2
INTRODUCTION
This section provides background information on the study area, hydrology and study design
for five CSIRO Flagship Cluster studies on the Edward-Wakool system within this SP3 report:
in-channel inundation modelling (Section 5.2); whole stream metabolism (Section 5.3); leaflitter breakdown rates in response to changes in flow regime (Section 5.4); invertebrate
response to in-channel unregulated flow pulses and small in-channel environmental flows
(Section 5.5) and; timing of fish spawning and recruitment in lowland river channel
ecosystems (Section 5.6). We outline the approach used to assess ecosystem responses to
in-stream environmental flows in the Edward-Wakool system. We describe the use of
‘control’ rivers to assess the benefits of the environmental flows and inform decisions on
the timing, duration and magnitude of flows. The Flagship Cluster projects were conducted
concurrently with other projects in the Edward-Wakool system (see Watts et al. 2013).
THE EDWARD-WAKOOL SYSTEM
The Edward-Wakool system in south-eastern Australia is a major anabranch and floodplain
of the Murray River in the Murray-Darling Basin in southern-eastern Australia. It is a
complex network of interconnected streams, ephemeral creeks, flood runners and wetlands
intersected by irrigation channels. The system begins upstream of Deniliquin in the BarmahMillewa Forest, and travels northwest before discharging back into the Murray River (Figure
1).
This system is considered to be important for its high native species richness and diversity
including threatened and endangered fish, frogs, mammals, and riparian plants. It is listed as
an endangered ecosystem, as part of the ‘aquatic ecological community in the natural
drainage system of the lower Murray River catchment’ in New South Wales (NSW Fisheries
28 | Flow dependent ecological responses
Management Act 1994). Fish kills reported in 2007-08 and 2010 were related to hypoxic
blackwater events and resulted in the loss of many hundreds of native fish.
Figure 1 Location of the Edward-Wakool system in south-eastern Australia.
This system has a long history of regulated flows for irrigation, stock and domestic water
supply and the water regime has been significantly altered with changes to the timing and
volume of flows (Green 2001). Natural flows in the river system would have been high in
spring and very low in summer and autumn. Between February 2006 and September 2010
there were periods of minimal or no flow in the Edward-Wakool system (Figure 2) due to
severe drought conditions. At the break of the drought several large unregulated flow
events occurred in the system between September 2010 and March 2011 (Figure 2).
Figure 2 Daily discharge between 01/01/08 and 28/02/13 in three rivers in the Edward-Wakool system.
Discharge data was obtained from NSW Office of Water website for three stations: Colligen Creek regulator
(409024), Wakool River offtake regulator (409019), Yallakool Creek offtake regulator (409020).
Flow dependent ecological responses | 29
Environmental watering in the Edward-Wakool system in 2011-12
The ecological objectives for environmental watering in the Edward-Wakool system focus on
supporting the on-going recovery of this system following the drought (CEWO 2013), with a
particular focus on supporting the condition and reproduction of native fish. This involves
the delivery of instream flows to maximise available breeding habitat, create flow conditions
favourable for reproduction or contribute to the survival of native fish. When very low flow
periods are experienced, the focus of environmental watering is on the delivery of base
flows to provide refuge habitat for fish and contribute to managing water quality issues.
In 2011-12 environmental water was delivered to this system on three occasions (Figure 3):
Figure 3. Daily discharge in four rivers in the Edward-Wakool system in 2011-12. Discharge data was
obtained from NSW Office of Water for four gauging stations: Colligen Creek regulator (409024), Wakool
River offtake regulator (409019), Yallakool Creek offtake regulator (409020), and Little Merran Creek at
Franklings Bridge (409044). Black triangles indicate timing of monthly sampling (From Watts et al 2013a).
30 | Flow dependent ecological responses
1. November 2011 environmental watering in Colligen Creek. The objective of this
watering was to “encourage movement of large bodied native fish such as Murray
cod, silver perch and golden perch to initiate spawning and recruitment of these
species” (CEWO 2012a) and also to maintain and enhance in-stream habitat.
2. February 2012 environmental watering in Colligen Creek and the Wakool River. The
objectives of this watering were to “provide opportunities for small-bodied fish to
access important breeding and feeding habitat and improve the condition of the
river and riparian ecosystems” (CEWO 2012b). The watering in the Wakool River was
suspended on 6/3/2102 due to rainfall.
3. April to May 2012 environmental watering via irrigation escapes to the Edward and
Wakool Rivers. The objective of this environmental watering action was to “provide
and maintain refuge habitats for remnant fish populations, particularly Murray cod,
from hypoxic blackwater that can severely impact fish” (CEWO 2012c).
ASSESSMENT OF ECOSYSTEM RESPONSES TO ENVIRONMENTAL WATERING IN THE
EDWARD-WAKOOL SYSTEM IN 2011-12
The Edward-Wakool system provided a unique opportunity to assess responses to
environmental watering. There are several distributary rivers in this system with regulators
that control inflows. In addition, water can be delivered to several rivers via escapes from
the Mulwala canal. This facilitates a comparison between rivers in close geographic
proximity that are receiving environmental water (treatment rivers) with those not receiving
environmental water (serving as ‘controls’). Such opportunities are relatively rare in testing
the effectiveness of environmental flows as control systems are often difficult to find
(Konrad 2011).
In 2011-12 four rivers (Figure 4) with differing hydrology were selected to assess ecosystem
responses to environmental watering. Yallakool Creek, Wakool River and Little Merran
Creek were controls for the November 2011 and February 2012 watering events, and Little
Merran Creek served as a control for the environmental watering from March to May 2011
(Figure 4).
Parameters monitored in 2011-12 include water chemistry, whole stream metabolism,
phytoplankton, biofilms, leaf-litter breakdown rates, zooplankton, macroinvertebrates,
frogs, and fish movement, spawning, recruitment and population structure.
Sampling was undertaken at different temporal frequencies depending on the response
time of the indicator. Sampling of fish larvae, zooplankton and water chemistry was
undertaken fortnightly. Other parameters were sampled monthly.
Flow dependent ecological responses | 31
No environmental
watering in 2011-2012
Environmental watering
in November 2011,
February 2012 and
March to May 2012
Environmental
watering in
March to May
2012
Environmental
watering in
February 2012 and
March to May 2012
Figure 4. Location of monitoring reaches for the assessment of ecosystem responses to environmental
watering in the Edward-Wakool system in 2011-12. Blue dots show location of gauging stations.
CONCLUSION
The assessment of ecosystem responses to instream environmental watering in the EdwardWakool River system was undertaken in 2011-12 using multiple indicators in multiple rivers,
with some rivers serving as controls. Results of the Flagship Cluster studies on the EdwardWakool system are contained in this SP3 report in sections 5.2 to 5.6.
ACKNOWLEDGEMENTS
We also thank the Wakool River Association and landholders in the Edward-Wakool river
system for allowing access to their properties and their interest in this project. State Water
Corporation provided information on river and escapes discharge and NSW OEH provided
information on watering actions. Maps were prepared by Simon McDonald and Deanna
Duffy (Charles Sturt University Spatial Analysis Unit). Monitoring and evaluation results
referred to in this paper were funded by the CSIRO Flagship cluster project and the
Commonwealth Environmental Water Office with in-kind contributions from Charles Sturt
University, Murray Catchment Management Authority, Industry and Investment NSW,
Monash University, NSW Office of Environment and Heritage.
32 | Flow dependent ecological responses
REFERENCES
Commonwealth Environmental Water Office (CEWO). 2012a Environmental watering in
Colligen Creek.
http://www.environment.gov.au/ewater/southern/murray/colligen.html {accessed
24th May 2012]
Commonwealth Environmental Water Office (CEWO). 2012b Environmental watering in the
Wakool River and Colligen Creek.
http://www.environment.gov.au/ewater/southern/murray/wakool.html {accessed
24th May 2012]
Commonwealth Environmental Water Office (CEWO). 2012c Environmental watering in the
Edward-Wakool river system (fish refuge habitat flows).
http://www.environment.gov.au/ewater/southern/murray/edward-wakool.html
{accessed 24th May 2012]
Commonwealth Environmental Water Office (2013) Environmental watering in the EdwardWakool River System: Environmental Watering 2012-13 Accessed Feb 25, 2013 from:
http://www.environment.gov.au/ewater/southern/murray/edward-wakool.html
Green D (2001) The Edward-Wakool System: River Regulation and Environmental Flows.
Department of Land and Water Conservation. Unpublished Report.
Konrad, C.P., Olden, J.D., Lytle, D.A., Melis, T.S., Schmidt, J.C., Bray, E.N., Freeman, M.C.,
Gido, K.B., Hemphill, N.P., Kennard, M.J., McMullen, L.E., Mims, M.C., Pyron, M.,
Robinson, C.T., Williams, J.G. (2011). Large-scale flow experiments for managing river
systems. BioScience. 61: 948–959.
Watts RJ, Kopf RK, Hladyz S, Grace, M, Thompson R, McCasker N, Wassens S, Howitt JA,
Burns A and Conallin J (2013). Monitoring of ecosystem responses to the delivery of
environmental water in the Edward-Wakool system, 2011-2012. Institute for Land,
Water and Society, Charles Sturt University, Report 2. Prepared for Commonwealth
Environmental Water.
Flow dependent ecological responses | 33
2.2
Lateral connectivity and availability of slackwater habitat in
the Edward-Wakool system during in-channel flows
Watts, R.J.1, Grace, M.2, McCasker, N.1, Watkins, S. 1, Bowen, P. 3, Conallin, J.3
1
Institute for Land, Water and Society, Charles Sturt University, Albury
Water Studies Centre, School of Chemistry and School of Biological Sciences, Monash University
3
Murray Catchment Management Authority
2
INTRODUCTION
Growing awareness of the impacts of river regulation has led to increased interest in the
delivery of environmental water to restore the ecological function of regulated river systems
(e.g. Poff 2009; Arthington et al. 2010). The two major types of environmental watering are
overbank flows that inundate wetlands and floodplains, and instream flows that are
contained within the channel. Internationally, there have been a few high profile examples
of the monitoring and evaluation of in-stream environmental flows, such as the
experimental flood downstream of the USA‘s Glen Canyon Dam in the mid-1990s (e.g.
Shannon et al. 2001; Valdez et al. 2001) and more recently in 2009 (Cross et al. 2011), and
an experimental flood in the Spöl River in Switzerland (Robinson et al. 2004). In Australia instream flows have historically been used to disperse algal blooms and other contaminants
(e.g. Maier et al. 2004; Mitrovic et al. 2003). Prior to 2010 there were few examples in
Australia where environmental water was used to create in-stream pulsed flows (Watts et
al. 2009). The ecosystem benefits of this type of environmental watering action are not well
understood and they are being assessed through monitoring and evaluation programs.
Inundation of riverbanks is important for river productivity and the creation of low flow
zones (or slackwaters, having low velocities and shallow depths) is important for riverbank
plants and the survival of organisms such as larval fish and larval shrimp through provision
of habitat and food. Modelling is a useful method for comparing estimates of the extent of
inundation under different discharge scenarios because it provides results more cheaply and
efficiently than ground based survey methods.
Previous studies modelling river flow and floodplain inundation have been undertaken for
wetlands on the Darling River (Shaikh et al. 2001), and floodplains on the Murrumbidgee
River (Frazier et al. 2003) and the River Murray (Overton 2005; Overton et al. 2006). These
studies have generally focussed on estimating floodplain inundation during overbank flows.
The use of digital elevation models is considered to not give the best representation of
floodplain inundation, because even small impediments on a predominantly flat floodplain
can affect the models. However, in systems where environmental watering is contained
within the channel, the use of digital elevation models to create flow path assessments
below bankfull is an appropriate approach to compare the extent of in-channel riverbank
34 | Flow dependent ecological responses
inundation under different discharge scenarios. Vietz et al (2013) used two dimensional
hydraulic modelling to examine the impacts of flow regulation on slackwaters in the Broken
River, south-eastern Australia. They demonstrated that the area of slackwaters decreased
with increasing discharge until inundation of higher-elevation bars and benches.
The aim of this project was to undertake hydrodynamic modelling for four river reaches in
the Edward-Wakool system to estimate the extent of wetted benthic surface area and the
area of slackwater (0-0.02 m/sec) for six discharge scenarios, ranging from low flows up to
bankfull flows. The modelling will contribute to the interpretation of ecological responses to
flows, to decision making regarding the magnitude of environmental watering and will assist
the communication of likely outcomes of planned watering events with landholders.
METHODS
Discharge scenarios were modelled for a 2 to 4 km reach of Colligen Creek, Yallakool Creek,
Wakool River and Little Merran Creek. The study area was previously described in section
6.1. Each reach was represented within the hydraulic model using a digital elevation model
(DEM) supplied by the Murray Catchment Management Authority-NSW. Several significant
artefacts were removed from the Little Merran Creek and Colligen Creek DEMs to ensure
normal stream flow was not impeded. Artefacts were removed by identifying erroneous
elevation values and integrating corrected values directly into the elevation surface using a
process of data fusion. To account for vegetation in each reach the surface friction
coefficient (Manning’s n) within the model was set to a value of 0.05 with the exception of
the Yallakool Creek site where a value of 0.04 was deemed more appropriate.
Estimates of daily discharge ML/day) at low flow, base flow, half bank and bankfull scenarios
were provided by D. Green (MDBA). Six discharge scenarios were modelled for Colligen
Creek and the Wakool River ranging from low flow to estimated bankfull flows (Table 1). In
Colligen Creek the estimated half bankfull discharge (800 ML/d) was approximately the
same as an environmental watering action undertaken in 2012. For Yallakool Creek the six
discharge scenarios plus an additional discharge of 560 ML/d was modelled (referred to as
environmental watering scenario) to estimate inundation resulting from an environmental
watering action in Yallakool Creek between October and December 2012. In Little Merran
Creek only three scenarios could be successfully modelled because the LiDAR survey was
undertaken when discharge was approximately 200 ML/day, so the low flow and base flow
scenarios could not be modelled in this system.
Discharge values were converted from ML/day to m3.sec-1 and supplied to the model as
static flow values. Discharge scenarios were modelled using the 2D grid implementation of
Eonfusion Flood (Myriax Software) with model outputs post-processed using the GIS
functionality of Eonfusion (Myriax Software). Post-processing, including surface area
calculations, was achieved using Eonfusion (Myriax Software), Quantum GIS and made
Flow dependent ecological responses | 35
distributable using Google Earth. Post processing was also undertaken to quantify the
spatial configuration of four velocity categories :
•
•
•
•
Zone 1: 0 – 0.02 m.sec-1 (still water/slackwater)
Zone 2: 0.02 – 0.3 m.sec-1 (slow water)
Zone 3: >0.3 m.sec-1 (fast water)
Zone 4: >0 m.sec-1 (entire flow field)
Stable state data frames captured during the modelling were used as the starting point for
determining the benthic surface area exposed to the four water velocity zones. Stable state
data frames were converted into multiband raster data frames containing water depth,
velocity and absolute elevation attributes. Multiband rasters were then converted into
vector format, triangulated and trimmed to produce discrete spatial features for each
velocity zone. Calculation of benthic surface area, and the depth range within each velocity
zone, was undertaken on the 3D surface to take into account the vertical relief of the river
bed. Summary statistics of the depth range (minimum, maximum, mean, standard
deviation) were calculated for each velocity zone under each discharge scenario. Postprocessing, including surface area and depth calculations, was achieved using Eonfusion
Quantum GIS and Excel and made distributable using Google Earth.
Table 1. Summary of discharge scenarios modelled for the four focus reaches; Colligen Creek, Yallakool
Creek, Wakool River and Little Merran Creek.
Scenarios
1 - low flow (estimated)
2 - base flow (estimated)
3 - median flow (2011-2013)
Environmental watering
4 - half bank (estimated)
5 - maximum daily discharge (2011-2013)
6 - bankfull (estimated)
Wakool
25
50
110
500
1442
3000
Discharge (ML/day)
Yallakool Colligen
30
30
170
200
271
314
800
560
800
800
1913
2808
4000
4000
Merran
230
500
1062
RESULTS
The estimates of wetted benthic surface area (Table 2) and representation of the spatial
coverage of the water surface (Figure 1) illustrate that estimated low flow scenario,
estimated base flow scenario and the median flow scenario for 2011-2013 were mostly
constrained within the river channel.
36 | Flow dependent ecological responses
In Colligen Creek, an environmental flow of 800 ML/day resulted in a 14% increase in wetted
benthic surface area relative to the base flow of 200 ML/day. Similarly, in Yallakool Creek an
environmental flow of 560 ML/day resulted in a 22% increase in wetted benthic surface area
relative to the base flow of 170ML/day (Table 2). The estimate of wetted benthic surface
area during the maximum daily discharge scenario experienced in 2011-2012 during high
unregulated flows (maximum daily discharge scenario) was considerably higher than during
the modelled base flow or environmental flow scenarios in these rivers. On the peak of this
unregulated flow event the wetted benthic surface area relative to the base flow 200
ML/day scenario is estimated to have increased by 47.8% in Colligen Creek and 59.7% in
Yallakool Creek. The models estimate there would be a considerable further increase in
wetted benthic surface area during a bankfull flow, however this type of flow event has not
occurred since 2010.
In contrast, in Little Merran Creek the estimated half bankful and maximum discharge 201113 scenarios resulted in considerable lateral connection and inundation of low lying
geomorphic features (Figure 1).
Table 2 Estimates of wetted benthic surface area under a range of discharge scenario in the four focus
reaches; Colligen Creek, Yallakool Creek, Wakool River and Little Merran Creek
Estimate of wetted benthic surface area (m3)
Wakool
Yallakool Colligen Merran
1 - low flow (estimated)
42,196
33,159
43,257
2 - base flow (estimated)
43,587
38,169
48,292
3 - median flow (2011-2013)
46,222
41,510
49,982
160,908
Environmental watering
46,432
55,110
4 - half bank (estimated)
58,547
49,924
55,110
242,902
5 - maximum daily discharge (2011-2013) 200,455
60,989
71,337
542,584
6 - bank-full (estimated)
264,109
86,404
84,820
Scenarios
The relationship between discharge and wetted surface area in these rivers was not linear
(Figure 2). The modelling has demonstrated that if there were a similar increase in discharge
in two rivers, it would not produce in the same increase in wetted benthic surface area in
the two rivers. For example, an increase from 200 ML/day to 800 ML/day in Colligen Creek
resulted in a modelled 14% increase in wetted benthic area, whereas a similar increase from
170 ML/day to 800 ML/day in Yallakool Creek resulted in a modelled 30% increase in wetted
benthic area (Table 2). There was a very large increase in estimated wetted surface area
with increasing discharge in Little Merran Creek.
Flow dependent ecological responses | 37
Figure 1 Maps showing representation the spatial coverage of the water surface in Colligen Creek, Yallakool
Creek, Wakool River and Little Merran Creek under different discharge scenarios. Environmental watering
occurred only in Colligen Creek and Yallakool Creek.
38 | Flow dependent ecological responses
Figure 2 Modelled relationship between discharge (ML/day) and wetted surface area (m3) for study reaches
in Colligen Creek, Yallakool Creek, Wakool River and Little Merran Creek.
The representation of the spatial coverage of the benthic surface extent for each velocity
zone for Colligen Creek is illustrated in Figure 3. There was a large area of slackwater
available in the low flow scenario (63 % of the total available benthic surface area, 27,391
m2), but slackwater was very low during the base flow (9.8 %, 4,712 m2), median flow
(6.4%, 3184 m2) and half bankfull scenarios (3.1 %, 1708 m2). Slackwater was created along
the margins and particularly in low lying geomorphic in-channel features during the
maximum discharge scenario (7.4 %, 5,308 m2) and bankfull scenario (13.6%, 11,575
m2)(Figure 3). The results for Yallakool Creek and the Wakool River are not presented here,
but were similar to those for Colligen Creek.
Flow dependent ecological responses | 39
Figure 3 Maps showing representation the spatial coverage of the water surface extent for each velocity
zone in Colligen Creek under different discharge scenarios.
40 | Flow dependent ecological responses
The spatial coverage of the benthic surface area for each velocity zone in Little Merran
Creek is presented in Figure 4. In contrast to the results for Colligen Creek, in Little Merran
Creek there was a large percent of slackwater available for the median flow scenario (70 %
of the total available water surface area). Slackwater was also abundant during the half
bankfull scenario (25.5 %) and maximum discharge scenario (41.5 %). Only a very small
percent of the area was classed as fast water in each of these scenarios.
Figure 4 Maps showing representation the spatial coverage of the benthic surface extent for each velocity
zone in Little Merran Creek under different discharge scenarios.
DISCUSSION AND CONCLUSION
The in-channel hydrodynamic modelling results demonstrate that the relationship between
discharge and wetted surface area and the area of slackwater in these four rivers in the
Edward-Wakool system is not linear and is most likely strongly influenced by
geomorphology. Similar to the findings of Vietz et al. (2013), we found that in some river
reaches (e.g. Colligen Creek) the area of slackwaters decreased with increasing discharge
until inundation of higher-elevation bars and in-channel benches. However, in another river
Flow dependent ecological responses | 41
reach (e.g. Little Merran Creek) a different pattern was observed with large area of
slackwater available at a range of different discharges. As the data presented here are for
only a small 2-4 km section of each river, it is important to consider that the relationships
may be reach specific and that these relationships need to be examined over longer river
distances in the study rivers.
These results have important implications for studies of ecology-flow relationships for inchannel flows. They suggest it may be more appropriate to examine the relationship
between inundation area and ecosystem responses to in-channel flows rather than
focussing on relationships with daily discharge data, as has commonly been the practise.
In-channel hydrodynamic modelling under different flow scenarios can be used to:
i)
ii)
iii)
better understand the relationship between in-channel flows and ecosystem
responses,
predict the consequences of in-channel flows on slackwater dependent biota and
ecosystem functions, and
facilitate better planning and management of the future in-channel
environmental flows. In particular it may assist managers examine trade-offs
between ecosystem benefits and third party impacts, such as inundation of
private land, prior to the delivery of environmental water. The modelling can
help managers determine the optimum discharge at which there can be an
increase in inundation and creation of slackwater to produce ecosystem
responses to environmental flows, but with minimal third party impacts.
ACKNOWLEDGEMENTS
Thanks to the Wakool River Association and landholders in the Edward-Wakool river system
for allowing access to their properties and their interest in this project. Maps were prepared
by Simon McDonald and Deanna Duffy (Charles Sturt University Spatial Analysis Unit). The
modelling was undertaken by Dr Hugh Pederson from Marine Solutions. The modelling
presented in this paper was partly funded by CSIRO Flagship cluster project with cash
contributions from Charles Sturt University, Murray Catchment Management Authority and
Monash University.
REFERENCES
Arthington AH, Naiman RJ, McClain ME and Nilsson C (2010) Preserving the biodiversity and
ecological services of rivers: new challenges and research opportunities. Freshwater
Biology 55(1), 1-16.
Cross WF, Baxter CV, Donner KC, Rosi-Marshall EJ, Kennedy TA, Hall RO, Wellard Kelly HA
and Rogers RS (2011) Ecosystem ecology meets adaptive management: food web
42 | Flow dependent ecological responses
response to a controlled flood on the Colorado River, Glen Canyon. Ecological
Applications 21, 2016-2033.
Frazier PS, Page KJ, Louis J, Briggs S and Robertson A (2003) Relating Wetland Inundation to
River Flow Using Landsat TM Data. International Journal of Remote Sensing 24 (19),
3755-3770.
Maier HG, Kingston GB, Clark T, Frazer T and Sanderson A (2004) Risk-based approach for
assessing the effectiveness of flow management in controlling cyanobacterial blooms
in rivers. River Research and Applications 20,459–471.
Mitrovic SM, Oliver RL, Rees, G, Bowling LC and Buckney RT (2003) Critical flow velocities for
the growth and dominance of Anabaena circinalis in some turbid freshwater rivers.
Freshwater Biology 48,164–174.
Overton IC (2005) Modelling Floodplain Inundation on a Regulated River: Integrating GIS,
Remote Sensing and Hydrological Models. River Research and Applications. 21 (9),
991-1001.
Overton IC, McEwan K, and Sherrah JR (2006) The River Murray Floodplain Inundation
Model – Hume Dam to Lower Lakes. CSIRO Water for a Healthy Country Technical
Report 2006. CSIRO: Canberra.
Poff NL (2009) Managing for Variability to Sustain Freshwater Ecosystems. Journal of Water
Resources Planning and Management 135(1), 1-4.
Robinson CT, Uehlinger, U and Monaghan MT (2004) Stream ecosystem response to
multiple experimental floods from a reservoir. River Research and Applications
20,359–377.
Shaikh M, Green D and Cross H (2001) A Remote Sensing Approach to Determine the
Environmental Flows for Wetlands of the Lower Darling River, New South Wales,
Australia. International Journal of Remote Sensing, 22, 1737-1751.
Shannon JP, Blinn DW, McKinney, T, Benenati EP, Wilson KP and O'Brien C (2001) Aquatic
food base response to the 1996 test flood below Glen Canyon Dam, Colorado River,
Arizona. Ecological Applications 11(3),672–685.
Valdez RA, Hoffnagle TL, McIvor CC, McKinney T and Leibfried WC (2001) Effects of a test
flood on fishes of the Colorado River in Grand Canyon, Arizona. Ecological Applications
11(3),686–700.
Vietz, GJ, MJ Sammonds, and MJ Stewardson (2013) Impacts of flow regulation on
slackwaters in river channels. Water Resources Research, 49, 1797–1811.
Watts RJ, Allan C, Bowmer KH, Page KJ, Ryder DS and Wilson AL (2009) Pulsed flows: a
review of environmental costs and benefits and Best practice. Waterlines report,
National Water Commission, Canberra.
Flow dependent ecological responses | 43
2.3
Whole stream metabolism
Grace, M.1, Hladyz, S. 2 and Caron, V.2
1
2
Water Studies Centre, School of Chemistry, Monash University - Clayton, Australia
School of Biological Sciences, Monash University - Clayton, Australia
This section examines the effects of stream flow on metabolism. The collected diel dissolved
oxygen profiles are interrogated to explore the relationships between the key variables
controlling metabolism (light, nutrients, organic carbon supply) and the metabolic
parameters (primary production and ecosystem respiration). Flow effects are both direct –
through scouring of biofilms and inundating previously dry bank areas – and indirect,
through mediating supply of nutrients and carbon as well as altering water column light
attenuation through increased turbidity.
INTRODUCTION
‘Whole Stream Metabolism’ provides estimates of rates of the key ecological processes of
photosynthesis and respiration (Odum 1956) by measuring the production and consumption
of dissolved oxygen gas (‘DO’). Aquatic ecosystems need photosynthesis to generate new
biomass (food for higher trophic levels) and respiration to break down plant and animal
detritus to recycle nutrients to enable such growth (Figure 1). Hence metabolism provides
assessment of the energy base underpinning aquatic foodwebs.
Rates vary seasonally as warm temperatures and increasing direct sunlight enhance primary
production; warmer temperatures and fresh organic carbon increase ecosystem respiration.
Streams with naturally higher concentrations of nutrients and with open canopies have
much higher natural rates of primary production than forested streams (Roberts and
Mulholland 2007).
Low process rates limit the amount of food (e.g. bacteria, algae and water plants) available
to consumers, constraining the populations of larger organisms including fish and
amphibians. Unnaturally low rates can indicate introduced agrochemicals (e.g. Ruzycki et al.
1998). Sometimes rates of these processes are particularly high. Such elevated primary
production usually indicates excessive plant growth (e.g. pondweed (Potamogeton sp. and
Azolla sp.) and more problematically algal blooms (often cyanobacteria), with the welldefined range of deleterious effects including blocking of sunlight (mortality of submerged
plants), possible toxin production (affecting drinking water for animals and humans) and
large swings in diel DO (including predawn anoxia with potential fish kills) (Paerl et al. 2001).
When an algal bloom eventually collapses, there is usually widespread anoxia, often
accompanied by toxin release.
44 | Flow dependent ecological responses
Organic Matter
ENERGY (Sunlight)
O2
Photosynthesis
ALGAE / PLANTS
Production of
Organic Material
O2
Respiration
Nutrients (N & P)
CO2
Food Resource
BACTERIA / FUNGI
Breakdown of
Organic Material
N, P, CO2
Figure 1 Relationships between photosynthesis (algae and plants) and respiration (bacteria and fungi) with
inputs of energy and organic matter, highlighting the role of sunlight, nutrients, dissolved gases and food
resources.
Many of the factors controlling stream metabolism are mediated by stream discharge
patterns. For example, Uehlinger (2000) demonstrated that scouring caused by higher water
velocities after rainfall can ‘reset’ primary production to very low rates, which are then
maintained until primary producer biomass is re-established, typically over a period of
several weeks. Resource limitation of metabolic rates may be overcome when materials
(organic carbon, nutrients) are transported into the stream channel by higher flows. These
materials may originate within the stream channel or on the floodplain. Higher discharges
normally result in greater stream power and the ability to carry and resuspend larger
particles (Gordon et al. 2004) which may then inhibit primary production (Oliver et al. 1997).
Vink et al. (2005) examined the effects of irrigation flow releases on the middle reaches of
the Murrumbidgee River and found that these releases stimulated primary production
through addition of bioavailable P and also stimulated ecosystem respiration through
additional organic carbon. Consequently, in this current study, metabolic parameters have
been measured under a range of discharge conditions to assess how flow changes impact on
rates of primary production and ecosystem respiration and hence the energy base of these
aquatic foodwebs.
METHODS
Metabolism Measurements
Rates of primary production and respiration can readily be estimated by continuously
recording DO, temperature and photosynthetically active radiation (PAR) over many
months, under a range of meteorological conditions and flows (Grace and Imberger 2006).
Flow dependent ecological responses | 45
Rates were calculated for each day so that the relative influences of flow, cloud cover and
season could be deconvoluted. DO changes in a sinusoidal pattern over 24 hours (Figure 2),
with DO increasing once the sun comes up, reaching a peak in mid-late afternoon (due to
photosynthesis throughout daylight hours) and then declining overnight, as ecosystem
respiration (which consumes DO) continues 24 hours a day. Gross primary production and
ecosystem respiration rates, plus reaeration (physical re-equilibration of DO to 100%
saturation across the air-water interface) were calculated daily from each diel data set,
using the daytime regression method (Atkinson et al. 2008). Only parameter estimates from
diel curves which reasonably matched the sinusoidal shape (Figure 2), and where the
inverse modelling approach produced excellent fits (r2 > 0.80) to these diel curves, were
included in the subsequent analysis.
9.3
9.2
[Dissolved Oxygen] (mg O2/L)
9.1
9.0
8.9
8.8
8.7
8.6
8.5
8.4
8.3
0:00
3:00
6:00
9:00
12:00
15:00
18:00
21:00
0:00
Time of Day
Figure 2 Typical Dissolved Oxygen (DO) Concentration profile over a 24 hour period. Data were from the
th
Wakool River at Moorabee, on 18 September, 2011.
Battery-powered ZebraTech data sondes, attached to star pickets in the thalweg at least 30
cm below the water surface under base flow conditions, recorded DO and water
temperature every 10 minutes. Every month, data were downloaded and the sondes
recalibrated. PAR data were also recorded at the same frequency in open fields near each of
the stream sites to avoid shading of the sensor by individual trees (hence the PAR sensor
accounts for effects of cloud cover).
Samples for bioavailable nutrient analysis (FRP, ammonia and NOx) were collected monthly
from the thalweg of each stream, then filtered through a 0.2 μm membrane at the time of
sampling. Samples were stored on ice then frozen before transportation to the laboratory.
46 | Flow dependent ecological responses
Nutrient analysis was undertaken by the NATA-accredited laboratory at the Monash
University Water Studies Centre.
Study sites
Two DO sondes were used in each of the four study streams, Wakool River and Little
Merran, Colligen and Yallakool Creeks, with each sonde pair separated by 1-4 river km. The
overall study design and site information are described in section 6.1.
RESULTS AND DISCUSSION
Rates of gross primary production rates (‘GPP’), ecosystem respiration and daily discharge
varied as a function of time for the Wakool River and Yallakool Creek (Figures 3 and 4) and
for Colligen and Little Merran Creeks (Figures 5 and 6).
7.00
1400
Wakool GPP
Yallakool GPP
Wakool Flow
Yallakool Flow
1200
5.00
1000
4.00
800
3.00
600
2.00
400
1.00
200
0.00
0
Flow (ML/Day)
GPP (mg O2/L/Day)
6.00
27/04/2012
24/03/2012
19/02/2012
16/01/2012
13/12/2011
9/11/2011
6/10/2011
2/09/2011
30/07/2011
Figure 3. Gross Primary Production (GPP) rates and stream discharge (ML/Day) for the Wakool River and
Yallakool Creek from July 2011 to April 2012. The green ovals identify GPP during and immediately after the
October 2011 flow event on the Wakool River. Data from both upstream and downstream sondes at each
site are included.
Flow dependent ecological responses | 47
16.00
1600
WakoolERGPP
Wakool
YallakoolER
GPP
Yallakool
Wakool Flow
Yallakool Flow
1400
12.00
1200
10.00
1000
8.00
800
6.00
600
4.00
400
2.00
200
0.00
0
Flow (ML/Day)
ER (mg O2/L/Day)
14.00
27/04/2012
24/03/2012
19/02/2012
16/01/2012
13/12/2011
9/11/2011
6/10/2011
2/09/2011
30/07/2011
Figure 4 Ecosystem Respiration (ER) rates and stream discharge (ML/Day) for the Wakool River and Yallakool
Creek from July 2011 to April 2012. The brown arrows indicate the trend in ER after the flow events in the
Wakool River in September and October 2011; the black arrows show contemporary trends in Yallakool
Creek. Data from both upstream and downstream sondes at each site are included.
4.5
1800
Colligen GPP
Merran GPP
Colligen Flow
Merran Flow
4
1400
3
1200
2.5
1000
2
800
1.5
600
1
400
0.5
200
0
Flow (ML/Day)
GPP (mg O2/L/Day)
3.5
1600
0
27/04/2012
24/03/2012
19/02/2012
16/01/2012
13/12/2011
9/11/2011
6/10/2011
2/09/2011
30/07/2011
Figure 5 Gross Primary Production (GPP) rates and stream discharge (ML/Day) for the Little Merran and
Colligen Creeks from July 2011 to April 2012. The green arrow indicates the trend in GPP after the flow event
in Colligen Ck in Aug-Sep 2011; the black arrow shows the contemporary trend in Little Merran Creek. Data
from both upstream and downstream sondes at each site are included.
48 | Flow dependent ecological responses
2100
7
Colligen ER
Little Merran ER
Colligen Flow
Little Merran Flow
1800
5
1500
4
1200
3
900
2
600
1
300
0
0
Flow (ML/Day)
ER (mg O2/L/Day)
6
27/04/2012
24/03/2012
19/02/2012
16/01/2012
13/12/2011
9/11/2011
6/10/2011
2/09/2011
30/07/2011
Figure 6 Ecosystem Respiration (ER) rates and stream discharge (ML/Day) for the Little Merran and Colligen
Creeks from July 2011 to April 2012. Data from both upstream and downstream sondes at each site are
included.
Rates of primary production and respiration remained relatively constant over the entire
study period. There were longer term (seasonal) trends where both primary production and
respiration rates increased during summer due to increased water temperatures, and for
primary production, more hours of sunshine. For days when flow was rapidly changing,
during and immediately after heavy rainfall or major water releases, the model was unable
to adequately estimate metabolic rates. When successful model fits were obtained during
high flow events, these mostly showed a highly significant depression in rates of primary
production. For example, data within the small green oval compared to the larger oval,
during and after environmental watering in October-November 2011 (Figure 3). The large
green oval (Figure 3) indicated a stimulation of GPP in the two weeks after the October 2011
flow in the Wakool River. The increase was above the seasonally-related upward trend
detected in the Yallakool Creek (Figure 3). This could be attributed to an influx of nutrients
from the flow event. Similarly, ER increased more after the flows in the Wakool River
(brown arrows, Figure 4) than could be attributed to warming temperatures alone (black
arrows for Yallakool Creek). The influx of fresh organic material during the flow probably
caused this enhanced respiration in the Wakool River.
The stimulatory role of a flow event is also seen in the slightly greater enhancement of GPP
in Colligen Creek in the months after the large unregulated flow in late August-early
September 2011 (and subsequent smaller flows) when compared to the GPP increases in
Little Merran Creek: the slope indicated by the green arrow for Colligen Creek is significantly
Flow dependent ecological responses | 49
greater (p = 0.003) than the black arrow for Little Merran Creek (Figure 5). Increases in
Ecosystem Respiration in these two creeks tracked each other over this October-December
period (Figure 6). In summary, the effects of these discharge events on ecosystem
metabolism rates were relatively small and idiosyncratic. This lack of a large response to
flow events is almost certainly due to the constraint of the additional water to the existing
stream channel and the low nutrient and organic carbon concentrations in the water
comprising the flow events. This finding is likely to be replicated in other catchments with
low nutrients (with the proviso that the water introduced by flow events is also low in these
analytes). Reasons for this behaviour are discussed below.
At no stage did any of the loggers record very low dissolved oxygen concentrations (< 20%
DO saturation), even during or straight after high flow events. Elevated flows decreased
%DO maxima each day by 0-20% and also decreased minimum %DO by the same
magnitude. Consequently, the environmental flows in November and February did not
constitute a black water event. Although these flows filled low-lying backwater areas, they
did not spread onto the floodplain. When this happens, water remains for days-weeks and
then returns commonly producing ‘black water’ (high dissolved organic carbon and very low
or no dissolved oxygen; Howitt et al. 2007).
There was considerable variation in GPP and ER rates within each river (Table 1). The P/R
ratio measures the balance between primary production and ecosystem respiration rates; a
value of < 1 indicates that more organic carbon was being consumed in the study reach than
was produced by primary production. The source of this organic carbon may have been
from overbank flow, upstream in the catchment or from riparian vegetation. Primary
production within the stream channel was also an extremely important contributor to
organic carbon supply, sustaining high level consumers in the food web.
The rates (Table 1) were at the lower end of the typical range found for non-polluted, slow
flowing rivers elsewhere in the world (e.g. Bernot et al. 2010; Marcarelli et al. 2011). They
were also similar among these four streams with median primary production rates differing
by less than a factor of two and respiration by less than a factor of three. This is unsurprising
given the similarity in land forms, biogeography and stream size. Median P/R ratios indicate
that for almost all the time these rivers were net heterotrophic (P/R < 1) indicating that
there must be an external source of organic carbon fuelling respiration.
50 | Flow dependent ecological responses
Table 1 Descriptive statistics for primary production (GPP) and ecosystem respiration (ER) rates for the
Wakool River and Colligen, Little Merran and Yallakool Creeks in the Edward-Wakool system, July 2011 April 2012.
COLLIGEN CK
LITTLE MERRAN CK
WAKOOL R
YALLAKOOL CK
227
2.47
2.3 ± 0.7
0.46
4.02
193
1.57
1.6 ± 0.5
0.25
1.70
152
2.31
2.2 ± 1.2
0.32
6.55
211
2.90
2.9 ± 0.9
0.37
5.15
3.48
3.5 ± 1.2
1.20
6.49
2.63
2.8 ± 1.1
0.76
6.42
5.69
6.0 ± 2.8
0.96
14.9
4.56
4.6 ± 1.4
2.14
8.73
0.71
0.71 ± 0.19
0.11
1.26
0.54
0.59 ± 0.25
0.10
1.86
0.36
0.36 ± 0.13
0.09
0.64
0.64
0.63 ± 0.17
0.07
1.09
GPP (mg
O 2 /L/Day)
n
Median
Mean (± 1 std dev)
Minimum
Maximum
ER (mg O 2 /L/Day)
Median
Mean (± 1 std dev)
Minimum
Maximum
P/R
Median
Mean (± 1 std dev)
Minimum
Maximum
As light (PAR) is required for photosynthesis, it might be expected that an increase in light
would increase rates of GPP. However, there were only weak positive relationships (e.g.
Yallakool: GPP = 0.158 x PAR + 0.881, r2 = 0.37, p < 0.001 ; Wakool: GPP = 0.092 x GPP +
1.33, r2 = 0.027, p = 0.044), with an extremely large variation in GPP at any particular PAR.
This variability at a fixed PAR indicates that another factor (or factors) was contributing
heavily to controlling GPP. One such factor was the presence and quantity of organisms
(autotrophs) capable of photosynthesis. In the water column, photosynthesis can only occur
when there is sufficient light. Turbidity or ‘cloudiness’, caused by suspended fine soil and
sediment particles, limits how far light can penetrate into the water column. The point
below which photosynthesis is no longer biologically viable is known as the ‘euphotic
depth’, Z eu . We used the empirical turbidity-euphotic depth relationship from the Darling
River (a large, turbid, lowland river; Z eu = 4.6 / (0.04 x Turbidity + 0.73) (Oliver et al., 2000))
to estimate Z eu in our study rivers. There was considerable variation in the euphotic depth
for each river based on the regular measurements of turbidity (Table 2).
Flow dependent ecological responses | 51
Table 2 Summary of stream turbidity (NTU) plus the median euphotic depth, Z eu , for the Wakool River and
Colligen, Little Merran and Yallakool Creeks in the Edward-Wakool system, July 2011 - April 2012.
Median
Mean
(± 1 std dev)
Minimum
Maximum Z eu (m)
COLLIGEN
LITTLE MERRAN
WAKOOL
YALLAKOOL
45
46 ± 12
24
74
1.82
83
88 ± 26
52
164
1.14
48
55 ± 21
27
124
1.74
61
62 ± 20
26
128
1.45
The euphotic depths indicate that for all river reaches deeper than Z eu , insufficient light will
reach the sediment surface to allow plants (macrophytes, benthic algae) to grow. Biofilms
will be restricted to shallow regions on the stream edges. This will decrease the overall
primary productivity of the rivers. Site observations throughout the study period confirmed
that macrophytes were confined to shallow marginal areas within the stream channel.
Benthic algae were noted in the littoral zones and also seen in drying sediment after the
recession of higher flows.
Increasing water temperatures associated with seasonal change result in faster physiological
rates in individual organisms and hence it was expected that higher temperatures would
result in greater rates of GPP and ER. This hypothesis was confirmed. Table 3 shows that
there were highly significant linear relationships between daily water temperature and GPP
and respiration (Table 3). It is pertinent to note that temperature is one of several
controlling variables (along with light, nutrients, organic carbon), hence a “high” r2 is not
necessarily expected here.
Table 3 Linear temperature relationships with GPP and ER rates for the Wakool River and Colligen, Little
Merran and Yallakool Creeks in the Edward-Wakool system. All linear regressions were highly significant (p <
-6
1 x 10 ).
COLLIGEN
LITTLE MERRAN
WAKOOL
YALLAKOOL
227
0.106
0.37
192
0.071
0.33
143
0.091
0.30
179
0.097
0.28
0.157
0.29
0.158
0.40
0.178
0.21
0.200
0.37
GPP vs Temperature
n
Slope
r2
ER vs Temperature
Slope
r2
52 | Flow dependent ecological responses
The temperature relationship with GPP and ER rates (Table 3), indicated a potential problem
during hot, dry summers. High water temperatures automatically mean lower dissolved
oxygen solubilities. This is rarely problematic as, even at 40oC, solubility is 6.4 mg O 2 /L, well
above the ANZECC trigger value of 4.0 mg O 2 /L. However, as temperature increases, the net
drawdown of O 2 also increases as respiration rates increase more rapidly with temperature
than GPP. This net drawdown increases the likelihood of suboxic, even anoxic conditions,
with fish kills being one likely consequence (Paerl et al. 2001). It is highly pertinent to note
that re-aeration across the air-water interface is significantly higher from a moving water
column than from a still one, especially if this current increases water turbulence (Demars
and Manson 2013). Thus maintenance of even slow flow rates by periodic environmental
releases is far preferable to standing water in the streams as a means of addressing
potential anoxia during hot, dry periods. This is a different scenario to ‘blackwater’ events,
where rapid oxygen drawdown is induced by direct contact with high levels of organic
carbon on the floodplain.
Nutrients, particularly nitrogen and phosphorus, are key controllers of photosynthesis. The
bioavailable nutrient concentrations in the water column of the rivers in this study were
extremely low, as shown in Figure 7. In most cases, the nitrate (NO x ) concentration was
below the detection limit of 1 μg/L. The ANZECC Water Quality Guidelines (ANZECC 2000)
typically recommend that the risk of algal blooms (and other adverse outcomes) in lowland
rivers in south-eastern Australia is minimised when bioavailable concentrations are below
20, 20 and 40 μg/L for ammonia, FRP and NOx respectively. The median concentrations in
the four rivers were nearly an order of magnitude lower than these levels. Occasional high
values coincided with the large flow events in mid-March 2012 which would also have
decreased the overall primary productivity of the rivers.
Flow dependent ecological responses | 53
Ammonia
FRP
NOx
[Nutrient] (µg /L of P or N)
100
10
1
Yallakool
Wakool
Merran
Colligen
Yallakool
Wakool
Merran
Colligen
Yallakool
Wakool
Merran
Colligen
Figure 7 Summary of monthly sampling of bioavailable nutrient concentrations in the four study streams
(September 2011 to April 2012). Note that the concentration axis is on a logarithmic scale for ease of
presentation. All concentrations less than the detection limit of 1 μg/L are plotted as 0.5 μg/L. The boxes
represent the data range 25th to 75th percentile, with the ‘middle’ line in the box being the median. The
“whiskers” indicate 10th and 90th percentiles in the data. Outliers are shown as circles.
This nutrient limitation of primary productivity means that large algal blooms are unlikely in
these streams. It also indicates that at least good land management practices and nutrient
uptake by benthic biofilms and macrophytes are effective in keeping nutrients from human
activities (including fertilizer usage in agriculture) from entering these waterways in
problematic concentrations.
REFERENCES
ANZECC (2000) Australia and New Zealand guidelines for fresh and marine water quality . In
(Australian and New Zealand environment and conservation council and Agriculture
and resource management council of Australia and New Zealand) Accessed at:
http://www.mincos.gov.au/publications/australian_and_new_zealand_guidelines_fo
r_fresh_and_marine_water_quality
54 | Flow dependent ecological responses
Atkinson BL, Grace MR, Hart BT and Vanderkruk KEN (2008) Sediment instability affects the
rate and location of primary production and respiration in a sand-bed stream.
Journal of the North American Benthological Society 27, 581-592.
Bernot MJ, Sobota DJ, Hall Jr RO, Mulholland PJ, Dodds WK, Webster JR, Tank JL, Ashkenas
LR, et al. (2010) Inter-regional comparison of land-use effects on stream metabolism.
Freshwater Biology 55, 1874–1890.
Demars BO and Manson JR (2013) Temperature dependence of stream aeration coefficients
and the effect of water turbulence: a critical review. Water Research 47, 1-15.
Gordon ND, McMahon TA, Finlayson BL, Gippel CJ and Nathan RJ (2004) Stream Hydrology.
An Introduction for Ecologists (2nd Edn). Wiley.
Grace MR and Imberger SJ (2006) Stream Metabolism: Performing and Interpreting
Measurements. Water Studies Centre Monash University, Murray Darling Basin
Commission and New South Wales Department of Environment and Climate
Change. 204 pp. Accessed at: http://monash.edu/science/wsc/
Howitt JA, Baldwin DS, Rees GN and Williams JN (2007) Modelling blackwater: predicting
water quality during flooding of lowland river forests. Ecological Modelling 203, 229242.
Marcarelli AM, Baxter CV, Mineau MM and Hall Jr RO (2011) Quantity and quality: unifying
food web and ecosystem perspectives on the role of resource subsidies in
freshwaters. Ecology 92, 1215-1225.
Odum HT (1956) Primary Production in Flowing Water. Limnology and Oceanography 1(2),
102–117.
Oliver RL, Hart BT, Olley J, Grace MR, Rees CM and Caitcheon CG (1999) The Darling River:
Algal Growth and the Cycling and Sources of Nutrient., Murray Darling Basin
Commission Report M386, CRC for Freshwater Ecology and CSIRO Land and Water,
pp 200.
Paerl HW, Fulton III RS, Moisander PH and Dyble J (2001) Harmful Freshwater Algal Blooms,
with an Emphasis on Cyanobacteria. The Scientific World 1, 76-113.
Roberts BJ. Mulholland PJ and Hill WR (2007) Multiple Scales of Temporal Variability in
Ecosystem Metabolism Rates: Results from 2 Years of Continuous Monitoring in a
Forested Headwater Stream. Ecosystems 10, 588-606.
Ruzycki EM, Axler RP, Owen CJ and Martin TB (1998) Response of Phytoplankton
Photosynthesis and Growth to the Aquatic Herbicide Hydrothol 191. Environmental
Toxicology and Chemistry 17, 1530–1537.
Uehlinger U (2000) Resistance and resilience of ecosystem metabolism in a flood-prone river
system. Freshwater Biology 45, 319–332.
Vink S, Bormans M, Ford PW and Grigg NJ (2005) Quantifying ecosystem metabolism in the
middle reaches of the Murrumbidgee River during irrigation flow releases. Marine
and Freshwater Research 56, 227-241.
Flow dependent ecological responses | 55
2.4
Leaf-litter breakdown rates in response to changes in flow
regime
Hladyz, S.1,2, Thompson, R. 3, Kopf, R.K.4, Watts, R.J.4 and Grace, M.2
1
School of Biological Sciences, Monash University - Clayton, Australia
Water Studies Centre, School of Chemistry, Monash University - Clayton, Australia
3
Institute for Applied Ecology, University of Canberra - Canberra, Australia
4
Institute for Land, Water and Society and School of Environmental Sciences, Charles Sturt
University - Albury, Australia
2
INTRODUCTION
Primary production and leaf-litter breakdown rates are key ecosystem functions which are
well recognised river health metrics in the literature but are not routinely undertaken in
bioassessment monitoring programs (Gessner and Chauvet 2002; Woodward et al. 2012).
Functional assessments usefully synthesise responses across entire ecosystems (Kingsford et
al. 2010) and provide the information needed to understand the processes which underpin
negative impacts, such as hypoxic blackwater events and fish kills (Hladyz et al. 2011b).
Leaf-litter breakdown has been proposed as a useful functional indicator of river health
because it responds to natural (e.g. climate) and anthropogenic stressors (e.g. organic
pollution, land use) and is relatively easy to measure with inexpensive equipment (Gessner
and Chauvet 2002; Young et al. 2008; Hladyz et al. 2011a). The rate of litter break-down
represents an integrated response to both biological (microbial degradation via fungi and
bacteria, invertebrate feeding by shredders) and physical (i.e. abrasion, fragmentation by
current) processes (Webster and Benfield 1986). This makes it useful as a bioassessment
tool (Gessner and Chauvet 2002; Young et al. 2008; Hladyz et al. 2011a). Litter breakdown
rates respond to various aspects of the flow regime (e.g. flow magnitude, flow velocity,
timing of flows, Ferreira et al. 2006; Watkins et al., 2010; Niu and Dudgeon 2011). However,
the understanding of functional responses to altered flow regimes is limited (Poff and
Zimmerman 2010).
We conducted a study that examined the effects of two environmental flow pulses on leaflitter breakdown rates in Colligen Creek, a regulated river in the Edward-Wakool river
system, NSW, south eastern Australia. We hypothesised four different mechanisms whereby
flow pulses could influence leaf-litter breakdown rates (see leaf-litter breakdown conceptual
model in Appendix).
1. Increased flows increase abrasion, increasing breakdown rates.
2. Increased flows disturb shredder communities, reducing breakdown rates.
56 | Flow dependent ecological responses
3. Increased flows increase organic matter supply from river margins, providing an
excess of resources to shredders, decreasing breakdown rates.
4. Increased flows result in wash-out of organic matter, resulting in a shortage of
resources to shredders, increasing breakdown rates.
To test these hypotheses, we compared leaf-litter breakdown rates in Colligen Creek (which
received environmental flows and unregulated flows) with rivers that received only
unregulated flow pulses (Wakool River, Little Merran Creek) or no flow pulses (Yallakool
Creek).
METHODS
Study sites
Study design and study site information is provided in Section 2.1.
Leaf-litter breakdown rates
Leaf-litter breakdown rates assessed overall changes in flow regime among the four rivers.
Two sites were chosen per river to examine the variability within each river. The field study
was done during early spring 2011 (late September) and early autumn 2012 (March) and ran
for six months. Placement of litter bags occurred after the large unregulated flow pulses in
September therefore breakdown rates were not influenced by these flows.
Methods used were based on a standard approach (see Hladyz et al. 2010). Leaves of two
common native trees, river red gum (Eucalyptus camaldulensis) and blackwood (Acacia
melanoxylon), were collected. Browning pre-abscised (leaves were collected from branches
that had fallen recently onto the ground from gusts of wind) river red gum leaves were
collected from floodplain forest. Blackwood leaves were collected immediately after
abscission. These species were chosen because they are widespread in south east Australia
and because they differ markedly in terms of resource quality, with the former having
higher carbon: nutrient ratios (Glazebrook and Robertson 1999; Wedderburn and Carter
1999). All leaves were air-dried to constant mass and five grams (±0.25 g) of air-dried leaf
material was added to experimental plastic-mesh bags of two mesh aperture sizes. The
mesh size of 0.5mm (which excludes macroinvertebrates) was used to measure leaf-litter
breakdown due to microbial breakdown and abrasion by fine sediments. A mesh size of
10mm was used to measure total breakdown caused jointly by microorganisms,
invertebrates and physical processes. Metal stakes were anchored on the river banks and
litter bags were attached and weighed down with lead weights within each reach. Of the 4
rivers each was sampled on 6 occasions (monthly; 2 replicates x 2 mesh sizes x 2 leaf species
x 2 reaches x 4 rivers), thus giving a total of 384 litter bags (64 litter bags per monthly
sampling). On each sampling occasion bags were collected with a 250 µm mesh net (to trap
accompanying macroinverterbates), placed in ziplock bags and then frozen. In the
Flow dependent ecological responses | 57
laboratory after thawing, leaf material was separated from invertebrates (using a 250 µm
sieve), and sorted leaf material was oven-dried at 105°C to constant mass and then
weighed. Breakdown rates were expressed as proportion of mass remaining. Correction
factors derived for leaching losses (determined after 24 h under running tap water in the
laboratory) and moisture content were applied to the initial air-dry masses, to calculate
post-leaching dry mass loss over time (after Hladyz et al. 2009). Macroinvertebrate data
from bags are not presented here.
Data analyses
Exponential breakdown decay coefficients (-k) were estimated using linear regression
analyses on ln(x) transformed dry mass data. This calculated decay coefficient allows
comparisons with other studies and can be used to estimate the time required for 50% of
the leaf litter to breakdown. Decay coefficients were classified according to the scheme of
Petersen and Cummins (1974). “Fast” breakdown groups decay coefficient > 0.010,
“medium” groups are between 0.005 – 0.010 and “slow” groups <0.005. An analysis of
covariance (ANCOVA) was used to test for differences in breakdown rates between
treatments with time as the covariate (Boulton and Boon 1991). Among rivers, leaf species
and mesh-sizes were analysed separately. Within rivers, mesh-sizes were analysed together
to examine varying contributions from differing agents of breakdown (i.e. microbial
degradation vs. macroinvertebrate feeding and mechanical fragmentation).
RESULTS AND DISCUSSION
Breakdown rates were similar among rivers for coarse and fine-mesh bags for both litter
types (Table 2, Figure 1). For river red gum, in comparison to other studies decay
coefficients from coarse-mesh bags in the Edward-Wakool system were categorised with
the “fast” processing category and were similar to values from a study carried out in the
Murray River (Table 3). Values calculated from fine-mesh bags were categorised with the
“medium” processing category and were more similar to processing rates in wetlands and a
temperate rainforest stream (Table 3). Within rivers, red gum breakdown rates in Colligen
Creek were approximately 1.7 times faster in coarse-mesh than fine-mesh bags (Table 4). In
contrast, in the other three rivers there was more variability in breakdown rates and rates
did not differ between coarse and fine-mesh bags (Table 4). These findings suggest that
increases in unregulated flows and environmental flows in Colligen Creek may affect
breakdown by physical fragmentation of litter and potentially increased abrasion from
suspended particulate material (Bird and Kaushik 1992; Lepori et al. 2005; Paul et al. 2006)
or via increased invertebrate shredding (Ferreira et al. 2006). Shredder abundances in the
Edward-Wakool system were relatively low which may suggest that the differences between
mesh sizes may be more attributed to physical fragmentation via increases in flow.
58 | Flow dependent ecological responses
0.025
A)
River Red Gum
Coarse-mesh
Fine-mesh
0.020
-k
0.015
0.010
0.005
0.000
0.0035
B)
Blackwood
0.0030
Coarse-mesh
Fine-mesh
0.0025
-k
0.0020
0.0015
0.0010
0.0005
0.0000
Colligen
Yallakool
Wakool
Merran
River
Figure 11 Leaf-litter breakdown rates (-k) of river red gum (A) and blackwood leaves (B) in four rivers in the EdwardWakool system. Solid bars denote coarse-mesh and hatched bars denote fine-mesh. Data displayed are mean values ± 1
SE.
Flow dependent ecological responses | 59
Table 2 Analysis of covariance (ANCOVA) of leaf-litter breakdown rates comparing among rivers. Results displayed are
the time by leaf-litter interaction term (comparison of slopes)
LITTER TREATMENT
DF
RESIDUAL DF
F-RATIO
P
River Red gum Coarse-mesh
15
24
0.3
0.995
River Red gum Fine-mesh
15
24
1.8
0.106
Blackwood Coarse-mesh
15
24
0.2
0.999
Blackwood Fine-mesh
15
24
1.0
0.501
Blackwood leaves in coarse and fine-mesh were in the “slow” processing rate category and
were slower than values calculated from other streams in south-eastern Australia (Table 3).
Within rivers, blackwood breakdown rates did not differ between coarse and fine-mesh
bags (Table 4). Blackwood leaf-litter breakdown rates in Colligen Creek were similar
between mesh sizes, though these leaves were less refractory than river red gum (slower to
breakdown) and therefore may be more resistant to physical fragmentation via flow.
Blackwood leaf-litter breakdown rates between mesh-sizes were also similar in the other
three rivers suggesting microbial degradation of leaf-litter was the primary agent in
controlling breakdown rates in the Edward-Wakool system (Table 4).
Table 3 Decay coefficients (-k) and half-lives (T 50 in days) of River Red Gum (RRG) and Blackwood (B) leaves compared
with published data. Category refers to the processing categories of Petersen and Cummins (1974) “Fast” breakdown
groups decay coefficient > 0.010, “medium” groups, 0.005 – 0.010 and “slow” groups < 0.005. *Double exponential
models used to calculate decay coefficients
RIVER
MESH (MM)
LEAF
-K(D-1)
T 50 (D)
CATEGORY
REFERENCE
Colligen
10 / 0.5
RRG
0.012 / 0.007
58 / 100
F/M
This study
B
0.003 / 0.002
257 / 447
S
RRG
0.012 / 0.007
60 / 94
F/M
B
0.002 / 0.001
347 / 478
S
RRG
0.015 / 0.005
46 / 146
F/M
B
0.002/ 0.002
330 / 447
S
RRG
0.011 / 0.009
63 /79
F/M
B
0.002 / 0.001
365/ 603
S
Yallakool
Wakool
Merran
10 / 0.5
10 / 0.5
10 / 0.5
This study
This study
This study
Murray
10 /0.3
RRG
0.015 / 0.015
47/46
F
Schulze and Walker (1997)
Stream
Leaf pack
RRG
0.006
110
M
Campbell and Fuchshuber (1995)
Wetland
Leaf pack
RRG
0.496, 0.003*
138 autumn
M
Glazebrook and Robertson (1999)
0.687, 0.004*
132 winter
M
Wetland
1.5
RRG
0.007
80
M
Briggs and Maher (1983)
Wetland
(mesocosms)
Leaf pack
RRG
0.449, 0.005*
57 sum./aut.
M
Watkins et al. (2010)
98.51, 0.004*
134 spring
M
Wetland
10
RRG
10.281, 0.007*
70
M
Janssen and Walker (1999)
Stream
Leaf pack
B
0.005
141
M
Campbell et al. (1992)
B
0.007
M
O’Keefe and Lake (1987)
Stream
60 | Flow dependent ecological responses
Table 4 Analysis of covariance (ANCOVA) of leaf-litter breakdown rates comparing within rivers. Results displayed are
the time by flow treatment interaction term (comparison of slopes)
FLOW
TREATMENT
LITTER TYPE
DF
RESIDUAL DF
F-RATIO
P
Colligen
River red gum
5
12
21.1
<0.001
Yallakool
River red gum
5
12
0.8
0.548
Wakool
River red gum
5
12
1.0
0.443
Merran
River red gum
5
12
1.3
0.328
Colligen
Blackwood
5
12
1.9
0.166
Yallakool
Blackwood
5
12
0.6
0.678
Wakool
Blackwood
5
12
0.4
0.873
Merran
Blackwood
5
12
0.4
0.40
Leaf-litter breakdown rates did not show any marked differences among rivers and were
highly variable and therefore may not be a good indicator for changes in flows. There was
some evidence that differences in rates between the coarse and fine mesh bags were higher
in Colligen Creek which received environmental flows, suggesting that physical abrasion may
be an important underlying mechanism driving processing rates. In general however, there
was no clear evidence that the delivery of environmental water alters rates of leaf-litter
processing in these rivers. This may suggest that leaf-litter breakdown rates are not
sensitive to changes in hydrological regimes or to the delivery of environmental flows in the
Edward-Wakool system, or that the positive and negative impacts on breakdown are acting
to cancel one another out (see hypotheses).
The lack of a response in the coarse mesh bags (measuring macroinvertebrate-mediated
breakdown) may be a due to the low densities of shredders (leaf-eating invertebrates) in the
Edward-Wakool system (Hieber and Gessner 2002). Other studies have also found no effect
on breakdown rates from restoration of natural flow regimes or reduced flow variation
(Muehlbauer et al. 2009; Casas et al. 2000) and have suggested that other factors such as
heavy sedimentation of leaf bags and high magnitude flows can override any differences
between controls and restored or impacted sites. Studies which have found differences in
breakdown have found contrasting responses to reduced flow variation, including faster
litter breakdown rates (Short and Ward 1980) and slower breakdown rates (Nelson and
Roline 2000; Mendoza-Lera et al. 2012). No consistent patterns with provision of flows have
emerged to date from published studies, or the current study, and it is likely that
understandings of underlying mechanisms that drive differences in processes are needed.
Flow dependent ecological responses | 61
ACKNOWLEDGEMENTS
We extend our thanks to the Wakool River Association and landholders in the EdwardWakool river system for allowing access to their properties and for their keen interest in this
project. Field surveys and community liaison were supported by staff at the Murray
Catchment Management Authority, with particular thanks to Josh Campbell and Patricia
Bowen. Vincent Kelly at the State Water Corporation provided information on river and
escapes discharge. We thank Josh Campbell, Anthony Conallin, Philip De Zylva, Chris
McCormack, Kelsey Tucker and Katrina Wilson for assisting with field and/ or laboratory
work. This project was funded by the Cluster Collaboration Fund and the Ecological
Responses to Altered Flow Regimes Research Cluster which represents a collaboration
between the CSIRO Water for a Healthy Country Flagship, Griffith University, the University
of New South Wales, Monash University, Charles Sturt University, La Trobe University and
the Arthur Rylah Institute of the Victorian Department of Sustainability and Environment.
The project also was supported by in-kind contributions from Commonwealth
Environmental Water, Charles Sturt University, Murray Catchment Management Authority,
and Monash University.
REFERENCES
Bird GA and Kaushik NK (1992) Invertebrate colonisation and processing of maple leaf litter
in a forested and an agricultural reach of a stream. Hydrobiologia 234, 65-77.
Boulton AJ and Boon PI (1991) A Review of Methodology Used to Measure Leaf Litter
Decomposition in Lotic Environments - Time to Turn over an Old Leaf. Australian
Journal of Marine and Freshwater Research 42, 1-43.
Briggs SV and Maher MT (1983) Litter fall and leaf decomposition in a river red gum
(Eucalyptus camaldulensis) swamp. Australian Journal of Botany 31, 308-316.
Campbell IC and Fuchshuber L (1995) Polyphenols, Condensed Tannins, and Processing
Rates of Tropical and Temperate Leaves in an Australian Stream. Journal of the North
American Benthological Society 14, 174-182.
Campbell IC, James KR, Hart BT and Devereaux A (1992) Allochthonous coarse particulate
organic material in forest and pasture reaches of two south-eastern Australian
streams. Freshwater Biology, 27: 353–365.
Casas JJ, Zamora-Munoz C, Archila F and Alba-Tercedor J (2000) The effect of a headwater
dam on the use of leaf bags by invertebrate communities. Regulated Rivers: Research
and Management 16, 577-591.
Ferreira V, Graca MAS, De Lima JLMP and Gomes R (2006) Role of physical fragmentation
and invertebrate activity in the breakdown rate of leaves. Archiv Fur Hydrobiologie
165, 493-513.
Gessner MO and Chauvet E (2002) A case for using litter breakdown to assess functional
stream integrity. Ecological Applications 12, 498-510.
62 | Flow dependent ecological responses
Glazebrook HS and Robertson AI (1999) The effect of flooding and flood timing on leaf litter
breakdown rates and nutrient dynamics in a river red gum (Eucalyptus camaldulensis)
forest. Australian Journal of Ecology 24, 625-635.
Hieber M and Gessner MO (2002) Contribution of stream detritivores, fungi, and bacteria to
leaf breakdown based on biomass estimates. Ecology 83: 1026-1038.
Hladyz S, Abjornsson K, Giller PS and Woodward G (2011a) Impacts of an aggressive riparian
invader on community structure and ecosystem functioning in stream food webs.
Journal of Applied Ecology 48, 443-452.
Hladyz S, Gessner MO, Giller PS, Pozo J and Woodward G (2009) Resource quality and
stoichiometric constraints on stream ecosystem functioning. Freshwater Biology 54,
957-970.
Hladyz S, Tiegs SD, Gessner MO, Giller PS, Risnoveanu G, Preda E, Nistorescu M, Schindler M
and Woodward G (2010) Leaf-litter breakdown in pasture and deciduous woodland
streams: a comparison among three European regions. Freshwater Biology 55, 19161929.
Hladyz S, Watkins SC, Whitworth KL and Baldwin DS (2011b) Flows and hypoxic blackwater
events in managed ephemeral river channels. Journal of Hydrology 401, 117-125.
Janssen MA and Walker KF (1999) Processing of riparian and wetland plant litter in the River
Murray, South Australia. Hydrobiologia 411, 53-64.
Kingsford RT, Brandis KJ, Jenkins KM, Nairn LC and Rayner TS (2010) Measuring ecosystem
responses to flow across temporal and spatial scales. In: Saintilan N and Overton I
(eds) Ecosystem response modelling in the Murray-Darling Basin. CSIRO Pub.,
Collingwood, Vic.
Lepori F, Palm D and Malmqvist B (2005) Effects of stream restoration on ecosystem
functioning: detritus retentiveness and decomposition. Journal of Applied Ecology 42,
228-238.
Mendoza-Lera C, Larranaga A, Perez J, Descals E, Martinez A, Moya O, Arostegui I and Pozo J
(2012) Headwater reservoirs weaken terrestrial-aquatic linkage by slowing leaf-litter
processing in downstream regulated reaches. River Research and Applications 28, 1322.
Muehlbauer JD , Leroy CJ, Lovett JM, Flaccus KK, Vlieg JK and Marks JC (2009) Short-term
responses of decomposers to flow restoration in Fossil Creek, Arizona, USA.
Hydrobiologia 618, 35-45.
Nelson SM and Roline RA (2000) Leaf litter breakdown in a mountain stream impacted by a
hypolimnetic release reservoir. Journal of Freshwater Ecology 15, 479-490.
Niu SQ and Dudgeon D (2011) The influence of flow and season upon leaf-litter breakdown
in monsoonal Hong Kong streams. Hydrobiologia 663, 205-215.
O'Keefe M and Lake PS (1987) The decomposition of pine, eucalypt and Acacia litter in a
small upland Victorian stream. Bull. Aust. Soc. Limnol. 11:15-32.
Paul MJ, Meyer JL and Couch CA (2006) Leaf breakdown in streams differing in catchment
land use. Freshwater Biology 51, 1684-1695.
Flow dependent ecological responses | 63
Petersen RC and Cummins KW (1974) Leaf processing in a woodland stream. Freshwater
Biology 4, 343-368.
Poff NL and Zimmerman JKH (2010) Ecological responses to altered flow regimes: a
literature review to inform the science and management of environmental flows.
Freshwater Biology 55, 194-205.
Schulze DJ and Walker KF (1997) Riparian willow and eucalypt trees and their interactions
with littoral invertebrates in the River Murray, South Australia. Regulated Rivers
Research and Management 13, 557-577.
Short RA and Ward JV (1980) Leaf litter processing in a regulated Rocky Mountain stream.
Canadian Journal of Fisheries and Aquatic Sciences 37, 123-127.
Watkins S, Quinn G and Gawne B (2010) Changes in organic matter dynamics and
physicochemistry associated with riparian vegetation loss and river regulation in
floodplain wetlands of the Murray River, Australia. Marine and Freshwater Research
61, 1207-1217.
Webster JR and Benfield EF (1986) Vascular plant breakdown in freshwater ecosystems.
Annual Review of Ecology and Systematics 17, 567-594.
Wedderburn ME and Carter J (1999) Litter decomposition by four functional tree types for
use in silvopastoral systems. Soil Biology and Biochemistry 31, 455-461.
Woodward G, Gessner MO, Giller PS, Gulis V, Hladyz S, Lecerf A, Malmqvist B, Mckie BG,
Tiegs SD, Cariss H, Dobson M, Elosegi A, Ferreira V, Graca MAS, Fleituch T, Lacoursiere
JO, Nistorescu M, Pozo J, Risnoveanu G, Schindler M, Vadineanu A, Vought LBM and
Chauvet E (2012) Continental-Scale Effects of Nutrient Pollution on Stream Ecosystem
Functioning. Science 336, 1438-1440.
Young RG, Matthaei CD and Townsend CR (2008) Organic matter breakdown and ecosystem
metabolism: functional indicators for assessing river ecosystem health. Journal of the
North American Benthological Society 27, 605-625.
64 | Flow dependent ecological responses
2.5
Invertebrate response to in-channel unregulated flow pulses
and small in-channel environmental flows
Hladyz, S.1,2, Grace, M.2, Kopf, R.K.3, Watts, R.J.3, Thomson, J.1 & Thompson, R.4
1
School of Biological Sciences, Monash University - Clayton, Australia
Water Studies Centre, School of Chemistry, Monash University - Clayton, Australia
3
Institute for Land, Water and Society and School of Environmental Sciences, Charles Sturt
University - Albury, Australia
4
Institute for Applied Ecology, University of Canberra - Canberra, Australia
2
INTRODUCTION
Balancing human demands for water and the ecological integrity of regulated rivers is a
complex issue. There is general acceptance in society that the river is a legitimate user of
water to sustain aquatic biodiversity and ecological functions (Naiman et al. 2002). In
support of this, there is increasing investment in the restoration of natural flow regimes
through managed water releases from reservoirs, or through the return of flow to rivers and
wetlands via a reduction in abstraction (Feld et al. 2011; Pittock and Finlayson 2011).
Returning water to regulated rivers as ‘environmental flows’ can improve ecosystem
condition by mimicking a more natural flow regime and potentially restoring natural
ecological processes and patterns (Arthington et al. 2010). The financial costs of delivery of
environmental flows are large (e.g. in terms of opportunity cost for irrigation water) and a
means of demonstrating the ecological benefits from these flows is greatly needed by water
managers.
Few studies have explicitly tested if delivery of environmental flows to regulated rivers can
improve biodiversity. Invertebrates are a major component of the biodiversity of flowing
water systems and are a critical resource to high trophic levels such as fish (Balcombe et al.
2005; Strayer 2006). Invertebrate communities have been widely used as biological
indicators, as they are abundant, diverse, occupy a range of ecological niches and can
respond relatively quickly to changes in environmental management, due to their relatively
rapid life-histories (Norris and Thoms 1999; Chessman 2003). Altered flow regimes in
regulated rivers can influence invertebrate communities through changes in food and
physical habitat availability (Bunn and Arthington 2002; see invertebrate conceptual model
in Appendix). In the few studies that have looked at environmental flows in relation to
invertebrate biodiversity, effects on invertebrate communities are variable. Environmental
flows can increase (Mackie et al. 2003; Chester and Norris 2006) or decrease invertebrate
richness (White et al. 2012), and either alter (Lind et al. 2007) or not alter community
composition (Brook et al. 2011).
We conducted a study that examined the effects of two environmental flows on
invertebrate biodiversity (abundance, biomass, taxon richness, composition) in Colligen
Flow dependent ecological responses | 65
Creek, a regulated river in the Edward-Wakool river system, NSW, south eastern Australia.
We predicted that environmental flows would increase invertebrate biodiversity and alter
community composition through the provision of additional habitat and food resources. To
determine this, we contrasted Colligen Creek with rivers that received only unregulated flow
pulses (Wakool River, Little Merran Creek) or no flow pulses (Yallakool Creek).
METHODS
Study sites
Study design and study site information is provided in Section 2.1.
Invertebrate abundance, biomass, taxon richness and composition
Sampling methods were consistent across all rivers. Macroinvertebrate communities were
sampled each month for seven months, from September 2011 until March 2012. A small
snag bag sampler (70 cm length, 250 µm mesh) was used to estimate ambient invertebrate
abundance and characterise community structure on small natural snags (small woody
debris, approx 50 cm length, surface area: 450 cm2) (after Growns et al., 1999). These snags
were collected by approaching from downstream and placed quickly into the snag bag with
as little disturbance to the invertebrates on the snag as possible. All surfaces were scrubbed
vigorously to dislodge invertebrates into the collecting jar at the bottom of the snag bag
sampler. The surface area of each snag was calculated by measurement of snag length and
diameter after all animals were removed. The invertebrate sample was then frozen until
further processing. Three replicate snag samples were taken per river.
In the laboratory, snag samples were defrosted and sorted under a dissecting microscope at
4-10x magnification. Invertebrates were enumerated and body length measured for all taxa
to the nearest 0.1 mm (except Chironomidae where the first 30 individuals encountered per
sub-family where measured). Animals were identified to genus level in most cases and
assigned to Functional Feeding Groups (FFG) using The Murray-Darling Freshwater Research
Centre online bugguide (Hawking et al. 2009) except for nematodes which were not
identified further, Chironomidae which were identified to sub-family, and mites which were
identified to either the order Oribatidia or other. Some samples included zooplankton taxa
(e.g. copepods, cladocerans) which were identified to family level. These are not typically
considered as macroinvertebrates, but for the purposes of consistency and due to their
periodic high abundance in these samples, we included these taxa. Abundances of
invertebrates were expressed as individuals per 100 cm2 of snag. Biomasses of invertebrates
were calculated using published body length-to-mass relationships to calculate the total
biomass in each sample (after Feller and Warwick 1988; Towers et al. 1994; Benke et al.
1999; Tod and Schmid-Araya 2009; Baumgärtner and Rothhaupt 2003) per 100 cm2 of snag.
66 | Flow dependent ecological responses
Where length–mass regressions were not available for a particular family or genus, the
closest relative was used.
Data analyses
We used a BACI (Before-After-Control-Impact) design to assess differences in flow
treatments. We analysed the differences in invertebrate biodiversity (abundance, biomass,
taxon richness) in the months before and after the delivery of environmental flows in
Colligen Creek and compared these changes in the same time period with rivers that
received unregulated flow pulses (Wakool River and Little Merran Creek = UR) and no flow
pulses (Yallakool Creek). We compared between time periods, ‘before’ and ‘after’ for each
river and then among rivers to examine treatment effects. The statistical model used was:
yirm ~N(µrm , σ2 ); µrm = βr + δr Iafter + εm + ϵrm (model 1).
Where; βr was the mean value of the response y (abundance, biomass etc.) in river r during
the before period, δr is the mean difference in y between before and after periods (Iafter = 1
if month m is in the after period, 0 otherwise), and εm and ϵrm are month and month-byriver random effects, respectively.
The model was fitted as a Bayesian hierarchical model. We calculated the posterior
probabilities that response variables increased after environmental flow delivery, Pr(>0),
and that changes in Colligen Creek exceeded any concurrent changes in Yallakool Creek,
Pr(>), and in the rivers which received unregulated flow pluses, the Wakool River and Little
Merran Creek (UR), Pr(>), (δ�UR = mean of δWakool , δMerran ). We also compared changes in
Yallakool Creek with changes in the Wakool River and Little Merran Creek (UR),
Pr(δ�UR >δYallakool ). Posterior probabilities exceeding 0.9 indicate strong evidence of a
positive difference, while values <0.1 are evidence of a negative difference (these values
correspond to ca. 10-fold change from prior to posterior odds of an effect in each direction).
We assigned uninformative, independent, normal prior distributions to main effects:
βr ~N(0,1000), δr ~N(0,1000). The monthly errors were modelled with a first order
autocorrelation structure, εm |ε−m ~N(0.5(εm−1 + εm+1 ),0.5σ2m ), and the month-by-river
errors were assigned exchangeable normal prior distributions, ϵrm ~N(0, σ2rm ). Error
variances, σ2 , σ2rm , σ2m , were assigned uniform prior distributions on the corresponding
standard deviation, e.g. σ2rm ~U(0,100). All models were fitted via Markov chain Monte
Carlo (MCMC), using WinBUGS software (version 1.4, Lunn et al., 2000), with 3 independent
chains of 50000 iterations each, including 5000 iteration burn-in periods. Inspection of
chain histories and Gelman-Ruben-Brooks diagnostics (Brooks and Gelman, 1998) verified
adequate MCMC mixing and convergence.
Analyses were run separately on zooplankton and macroinvertebrate abundance and
biomass data due to differences in size class which may mask potential differences in
responses. Variables were transformed (e.g. 4th root transformed, log 10 transformed) prior
Flow dependent ecological responses | 67
to statistical analyses when necessarily to normalise data and stabilise variances. Model
residuals were also checked for normality. If model residuals were not normally distributed
the model was re-run using a Poisson distribution for the response variable:
yirm ~Poisson(λrm ); log(λrm ) = βr + δr Iafter + εm + ϵrm (model 2).
Rarefaction was used to compare the taxon richness of sampling sites with varying numbers
of individuals as sample size can affect the number of taxa recorded in a sample (Gotelli and
Colwell, 2001). Taxon richness was estimated using 10 individuals which was the lowest
amount of individuals in one sample. Taxon richness was estimated using Primer 6 (v 6.1.13)
(PRIMER-E Ltd., Plymouth, UK).
Multivariate analyses of invertebrate community composition in the rivers were carried out
using PRIMER 6 (v 6.1.13) and PERMANOVA (v.1.0.3) (PRIMER-E Ltd., Plymouth, UK).
Multivariate analyses were performed on square-root transformed relative abundance and
raw biomass using Bray-Curtis dissimilarity. To examine community differences in relation to
flow treatments a PERMANOVA was performed using a BACI design. The model
incorporated three flow treatments (Colligen Creek, environmental flow treatment; rivers
that received unregulated flow pulses only, Wakool River, Little Merran Creek; no flow
pulses, Yallakool Creek), two time periods (‘before’ and ‘after’) and sampling months nested
within time periods. An impact is indicated by a significant interaction between flow
treatment x time period. The taxon contributing to Bray-Curtis dissimilarities among flow
treatments were investigated using SIMPER in PRIMER 6 (v 6.1.13). To visualise patterns
evident from PERMANOVA, metric multi-dimensional scaling, principal coordinate analysis
(PCO) was performed.
RESULTS AND DISCUSSION
Invertebrate abundance, biomass and taxon richness
There was some evidence that macroinvertebrate abundance increased in Colligen Creek
(Pr(��������� >0) = 0.88) and the Wakool River (Pr(������� >0)= 0.87) between time periods.
However, there were no differences among flow treatments. These findings suggest that
overall abundances of invertebrates were not affected by changes in flow and therefore
may be a poor indicator for changes in flow. This may be a result of differing responses to
changes in flows between taxa and is consistent with a recent quantitative meta-analysis
that found no consistent effect of flow magnitude on macroinvertebrate abundances (Poff
and Zimmerman, 2010).
There was strong evidence that the total biomass of macroinvertebrates increased after
delivery of environmental flows in Colligen Creek (Pr(αColligen >0) = 0.94) (Figure 1a). This
increase in biomass was greater than the changes observed in the rivers that received
unregulated flow pulses, the Wakool River and Little Merran Creek and no flow pulses,
68 | Flow dependent ecological responses
Yallakool Creek (Pr(αColligen >α
�UR ) = 0.89; Pr(αColligen >αYallakool) = 0.93) (Fig. 1a). This
biomass increase in Colligen Creek was largely due to sampling large sized taxa that were
not present before the delivery of environmental flows (e.g. the caseless caddis larvae
Ecnomus spp.) and due to an increase in size of taxa that were present before the
environmental flow delivery (e.g. mayfly larvae Atalophlebia spp., dipteran larvae
Chironominae). Chironominae 50% and Ecnomus spp. 26% made up over 75% of the
biomass increase.
2
Figure 1 Biomass of a) macroinvertebrates and b) zooplankton on woody debris (100cm ) in four rivers in the EdwardWakool system. Colligen Creek denotes environmental flow river (green) where unregulated flow pulse in ‘before’
period was followed by a delivery of environmental flows in the ‘after’ period, Wakool River and Little Merran Creek
denotes rivers that received unregulated flow pulses in ‘before’ period only (red and grey, respectively) and Yallakool
Creek denotes river that received no flow pulses and was run at base flow for the duration of the study (blue). Data
displayed are monthly mean values pooled ‘before’ (solid bar) and ‘after’ (hatched bar) periods ± 1 SE. Letters indicate
important treatment effects
This increase in biomass may be due to environmental flows providing additional wetted
habitat and increasing food availability (see section 5.2). Changes in flow regime can alter
the quantity, quality and size fractions of organic matter (Poff et al. 1997; Small et al. 2008).
The Functional Feeding Groups (FFG) of macroinvertebrates will therefore reflect these
changes in supply of organic matter (Merritt and Cummins 1984). Increases in river flow
should increase pelagic food sources favouring filterers, and reduction in flows should cause
organic matter to settle on the benthos, favouring collector-gatherers, scrapers or
shredders. In this study the increase in biomass of Chironominae (collector-gatherers) may
suggest that environmental flows are transporting organic matter into depositional zones on
the benthos, or inundating areas of benthic organic matter along margins. The caseless
caddis larvae Ecnomus spp. (predators) may be benefiting from the increased biomass of the
chironominae larvae. The increase in Ecnomus spp. may also benefit higher order predators
such as fish.
Flow dependent ecological responses | 69
Between time periods there was some evidence to suggest macroinvertebrate taxon
richness declined in Colligen Creek (Pr(αColligen >0) = 0.18) and Little Merran Creek
(Pr(αMerran >0) = 0.16) and strong evidence that richness declined in the Wakool River
(Pr(αWakool>0) = 0.09) and Yallakool Creek (Pr(αYallakool>0) = 0.09). However, the
magnitude of these declines did not differ among flow treatments. These declines in taxon
richness over time are most likely reflecting seasonal shifts in communities.
There was strong evidence of a reduction in zooplankton abundance and biomass between
time periods in Colligen Creek (abundance; Pr(αColligen >0) = 0.03, i.e. probability negative
change = 0.97; biomass; Pr(αColligen >0) = 0.02) and the Wakool River and Little Merran
Creek (abundance; Pr(α
�UR >0) = 0.02; biomass; Pr(α
�UR >0) = 0.07). Among flow treatments,
the abundance changes in the Wakool River and Little Merran Creek were substantially
different to changes in Yallakool Creek (Pr(α
�UR >αYallakool)= 0.05), but the changes in
Colligen Creek were not (Pr(αColligen >αYallakool )= 0.62). This may be due to the large
variation in zooplankton numbers in Colligen Creek before the delivery of environmental
flows, and the lower sample size for the environmental flow treatment (n = 1 river).
Zooplankton biomass changes in Colligen Creek (Pr(αColligen >αYallakool ) = 0.02) and the
Wakool River and Little Merran Creek (Pr(α
�UR >αYallakool) = 0.06) were greater than changes
in Yallakool Creek (Figure 1b). The zooplankton cladocerans Chydoridae and Ilyocryptidae
were the main taxa that decreased in abundance and biomass in the Wakool River and Little
Merran Creek and Colligen Creek after the unregulated flow pulses.
Zooplankton abundance and biomass were higher in the Wakool River and Little Merran
Creek and Colligen Creek immediately after the period of unregulated flow pulses. When
rivers were exposed to no flow pulses (Wakool River and Little Merran Creek) or small
magnitude environmental flows (in Colligen Creek), in the ‘after’ time period, zooplankton
abundance and biomass were low. It is possible that unregulated flow pulses increased
zooplankton abundance and biomass through transport of animals in from tributaries or
slackwaters, as zooplankton transport increases with flooding (Ning et al. 2013). It is also
possible that unregulated flow pulses increased wetted habitat (see section 6.2) and
subsequently zooplankton emergence from sediments (Boulton and Lloyd 1992), but that
small in-channel environmental flow pulses (in Colligen Creek river) were insufficient to
generate this response.
There was only weak evidence that zooplankton diversity changed between time periods
within rivers. There were marginal declines in zooplankton diversity in Colligen Creek
(Pr(αColligen >αYallakool ) = 0.10) and Wakool River and Little Merran Creek
(Pr(α
�UR >αYallakool ) = 0.06) and a marginal increase in diversity in Yallakool Creek. The
declines in taxon richness in Colligen Creek and the Wakool River and Little Merran Creek
appear to track declines in abundance and biomass of zooplankton taxa.
70 | Flow dependent ecological responses
Invertebrate assemblage composition
Invertebrate assemblages (expressed as patterns of abundance across all taxa, and
separately, patterns of biomass across all taxa) varied between time periods (‘before’ and
‘after’), flow treatments and months within time periods (Tables 2 and 3). A flow treatment
by month interaction was evident in the analyses and in order to examine whether
assemblages differed between the ‘before’ and ‘after’ period for each flow treatment,
separate PERMANOVA analyses were run.
Table 2 Permutation MANOVA (PERMANOVA) of invertebrate species’ relative abundance on small woody debris, based
on Bray–Curtis dissimilarity measure (B-A denotes before vs. after environmental flow, Mo denotes month of sampling)
SOURCE
DF
MS
PSEUDO-F
P (PERM)
B-A
1
19868
7.3
0.001
Flow treatment
2
5212.8
2.7
0.022
Mo (B-A)
5
2725.9
3.3
0.001
B-A x Flow
2
2110.1
1.1
0.349
Mo (B-A) x Flow
10
1938.3
2.4
0.001
Residual
63
818.3
Total
83
Table 3 Permutation MANOVA (PERMANOVA) of invertebrate species’ biomass variables on small woody debris, based
on Bray–Curtis dissimilarity measure (B-A denotes before vs. after environmental flow, Mo denotes month of sampling)
SOURCE
DF
MS
PSEUDO-F
P (PERM)
B-A
1
12090
2.9
0.02
Flow treatment
2
7410.1
3.1
0.004
Mo (B-A)
5
4206.2
2.7
0.001
B-A x Flow
2
3120.1
1.3
0.211
Mo (B-A) x Flow 10
2388.5
1.5
0.005
Residual
63
1563.8
Total
83
For invertebrate relative abundances, Colligen Creek (Pseudo-F = 4.13, P = 0.013), Wakool
River and Little Merran Creek (Pseudo-F = 5.43, P = 0.001) assemblages differed between
before and after time flow periods, whereas Yallakool Creek (Pseudo-F = 1.95, P = 0.126)
assemblages did not differ significantly (Figure 2a). For biomass, Colligen Creek invertebrate
assemblages differed before and after delivery of environmental flows (Pseudo-F = 2.82, P =
0.034) whereas the Wakool River, Little Merran Creek and Yallakool Creek assemblages did
Flow dependent ecological responses | 71
not (Figure 2b). SIMPER analyses revealed 17-18 taxa contributed approximately 90% to the
dissimilarity between the before and after periods in Colligen Creek, the Wakool River, and
Little Merran Creek for abundance, and Colligen Creek for biomass. Chironominae and the
mayflies Caenidae and Atalophlebia spp. increased in abundance in Colligen Creek after
delivery of environmental flows and the Wakool River, and Little Merran Creek after
unregulated flow pulses. The unregulated flow pulses may be driving these changes in taxa
abundances, since similar changes occurred in the Wakool River and Little Merran Creek as
Colligen Creek. In Colligen Creek taxa that increased in biomass after delivery of
environmental flow pulses were Chironominae, Ecnomus spp., Caenidae, and Atalophebia
spp. Many zooplankton taxa decreased in abundance in Colligen Creek, Wakool River and
Little Merran Creek between the before and after flow periods, including cladocerans
(Chydoridae, Ilyocryptidae) and the ostrocod Candonidae. In Colligen Creek there was a
decrease in biomass of ostrocods (Candonidae, Ilyocyprididae), cladocerans (Chydoridae,
Ilyocryptidae) and the copepod Cyclopoida. Most taxa that responded positively to
unregulated flow pulses and delivery of environmental flow pulses were collector-gatherers
and may be benefiting from the delivery of organic matter to the river channel. The
predatory caseless caddis Ecnomus spp. also increased in biomass, potentially due to the
increases in prey abundance and biomass (e.g. chironominae).
This suggests that environmental flow pulses and unregulated flow pulses generate similar
changes in invertebrate assemblages, and that environmental flows effectively mimic
natural flow regimes. However, environmental flows are generally of a smaller magnitude
than natural flows, and there is some evidence that some effects of flow pulses are
dependent on the size of the flow, likely because of the increased amount of habitat
inundated and potential connections with flood runners and anabranches that are not
achieved by small flow pulses. Environmental flows when delivered after unregulated flow
pulses can increase invertebrate biomass and therefore, may provide additional food and
habitat availability for taxa than unregulated flow pulses in isolation or low flows.
72 | Flow dependent ecological responses
Figure 2 Two factor PCO plot of the invertebrate assemblages on small woody debris based on a) abundances and b)
biomass on the basis of the Bray–Curtis dissimilarity measure, showing the factors of flow treatment ( denote Colligen
Creek, denotes Wakool River, denotes Little Merran Creek and denotes Yallakool Creek, respectively) and time
period (open symbols denote before period and closed symbols denote after period)
Flow dependent ecological responses | 73
ACKNOWLEDGEMENTS
We extend our thanks to the Wakool River Association and landholders in the EdwardWakool river system for allowing access to their properties and for their keen interest in this
project. Field surveys and community liaison were supported by staff at the Murray
Catchment Management Authority, with particular thanks to Josh Campbell and Patricia
Bowen. Vincent Kelly at the State Water Corporation provided information on river and
escapes discharge. We thank Josh Campbell, Anthony Conallin, Philip De Zylva, Chris
McCormack, Kelsey Tucker and Katrina Wilson for assisting with field and/ or laboratory
work. This project was funded by the Cluster Collaboration Fund and the Ecological
Responses to Altered Flow Regimes Research Cluster which represents a collaboration
between the CSIRO Water for a Healthy Country Flagship, Griffith University, the University
of New South Wales, Monash University, Charles Sturt University, La Trobe University and
the Arthur Rylah Institute of the Victorian Department of Sustainability and Environment.
The project also was supported by in-kind contributions from Commonwealth
Environmental Water, Charles Sturt University, Murray Catchment Management Authority,
and Monash University.
REFERENCES
Arthington AH, Naiman RJ, Mcclain ME and Nilsson C (2010) Preserving the biodiversity and
ecological services of rivers: new challenges and research opportunities. Freshwater
Biology 55, 1-16.
Balcombe SR, Bunn SE, Mckenzie-Smith FJ and Davies PM (2005) Variability of fish diets
between dry and flood periods in an and zone floodplain river. Journal of Fish Biology
67, 1552-1567.
Benke AC, Huryn AD, Smock LA and Wallace JB (1999) Length-mass relationships for
freshwater macroinvertebrates in North America with particular reference to the
southeastern United States. Journal of the North American Benthological Society 18,
308-343.
Boulton AJ and Lloyd LN (1992), Flooding frequency and invertebrate emergence from dry
floodplain sediments of the River Murray, Australia. Regulated Rivers: Research and
Management 7, 137–151.
Brooks AJ, Chessman BC and Haeusler T (2011) Macroinvertebrate traits distinguish
unregulated rivers subject to water abstraction. Journal of the North American
Benthological Society 30, 419-435.
Brooks SP and Gelman A (1998) General methods for monitoring convergence of iterative
simulations. Journal of Computational and Graphical Statistics 7, 434-455.
Bunn SE and Arthington AH (2002) Basic principles and ecological consequences of altered
flow regimes for aquatic biodiversity. Environmental Management 30, 492–507.
74 | Flow dependent ecological responses
Chessman BC (2003) New sensitivity grades for Australian river macroinvertebrates. Marine
and Freshwater Research 54, 95-103.
Chester H and Norris R (2006) Dams and flow in the Cotter River, Australia: Effects on
instream trophic structure and benthic metabolism. Hydrobiologia 572, 275-286.
Feld CK, Birk S, Bradley DC, Hering D, Kail J, Marzin A, Melcher A, Nemitz D, Petersen ML,
Pletterbauer F, Pont D, Verdonschot PFM and Friberg N (2011) From natural to
degraded rivers and back again: a test of restoration ecology theory and practice. In:
Woodward G (Ed) Advances in Ecological Research, pp. 119-209, Vol. 44.
Feller RJ and Warwick RM (1988) Energetics. In: Higgins RP and Thiel H (eds) Introduction to
the study of meiofauna. Smithsonian Institution Press, Washington.
Gotelli NJ and Colwell RK (2001) Quantifying biodiversity: procedures and pitfalls in the
measurement and comparison of species richness. Ecology Letters 4, 379-391.
Growns JE, King AJ and Betts FM (1999) The Snag Bag: a new method for sampling
macroinvertebrate communities on large woody debris. Hydrobiologia 405, 67-77.
Hawking JH, Smith LM, Le Busque K (editors) (2009) Identification and Ecology of Australian
Freshwater Invertebrates. http://www.mdfrc.org.au/bugguide, Version January 2009,
Accessed September 2011.
Lind PR, Robson BJ and Mitchell BD (2007) Multiple lines of evidence for the beneficial
effects of environmental flows in two lowland rivers in Victoria, Australia. River
Research and Applications 23, 933-946.
Lunn DJ, Thomas A, Best N and Spiegelhalter D (2000) WinBUGS - A Bayesian modelling
framework: Concepts, structure, and extensibility. Statistics and Computing 10, 325337.
Mackie JK, Chester ET, Matthews TG and Robson BJ (2013) Macroinvertebrate response to
environmental flows in headwater streams in western Victoria, Australia. Ecological
Engineering 53, 100-105.
Merritt RW and Cummins KW (1984) An introduction to the aquatic insects of North
America. Second ed. Kendall/Hunt Publishing Company, Dubuque, Iowa, 722pp.
Naiman RJ, Bunn SE, Nilsson C, Petts GE, Pinay G and Thompson LC (2002) Legitimizing
fluvial ecosystems as uses of water: an overview. Environmental Management 30,
455-467.
Ning NSP, Gawne B, Cook RA and Nielsen DL (2013) Zooplankton dynamics in response to
the transition from drought to flooding in four Murray-Darling Basin rivers affected by
differing levels of flow regulation. Hydrobiologia 702, 45-62.
Norris RH and Thoms MC (1999) What is river health? Freshwater Biology 41, 197-209.
Pittock J and Finlayson CM (2011) Australia's Murray-Darling Basin: freshwater ecosystem
conservation options in an era of climate change. Marine and Freshwater Research 62,
232-243.
Poff NL, Allan JD, Bain MB, Karr JR, Prestegaard KL, Richter BD, Sparks RE and Stromberg JC
(1997) The natural flow regime. Bioscience 47, 769-784.
Flow dependent ecological responses | 75
Poff NL and Zimmerman JKH (2010) Ecological responses to altered flow regimes: a
literature review to inform the science and management of environmental flows.
Freshwater Biology 55, 194-205.
Small, M.J., Doyle, M.W., Fuller, R.L. and Manners, R.B. (2008) Hydrologic versus
geomorphic limitation on CPOM storage in stream ecosystems. Freshwater Biology,
53, 1618-1631.
Strayer DL (2006) Challenges for freshwater invertebrate conservation. Journal of the North
American Benthological Society 25, 271-287.
Tod SP and Schmid-Araya JM (2009) Meiofauna versus macrofauna: Secondary production
of invertebrates in a lowland chalk stream. Limnology and Oceanography 54, 450-456.
Towers DJ, Henderson IM and Veltman CJ (1994) Predicting Dry-Weight of New-Zealand
Aquatic Macroinvertebrates from Linear Dimensions. New Zealand Journal of Marine
and Freshwater Research 28, 159-166.
White HL, Nichols SJ, Robinson WA and Norris RH (2012) More for less: a study of
environmental flows during drought in two Australian rivers. Freshwater Biology 57,
858-873.
76 | Flow dependent ecological responses
2.6
Timing of fish spawning and recruitment in lowland river
channel ecosystems
Kopf, R.K.1, McCasker, N.1, Watts, R.J.1, Hladyz, S.2, Grace, M.3, Thompson, R.4
1
Institute for Land, Water and Society, Charles Sturt University, Albury, Australia
School of Biological Sciences, Monash University - Clayton, Australia
3
Water Studies Centre, School of Chemistry, Monash University - Clayton, Australia
4
Institute for Applied Ecology, University of Canberra - Canberra, Australia
2
INTRODUCTION
The timing of reproduction in organisms that live in flowing water environments is tightly
coupled with historical patterns of the natural flow regime associated with flooding, drought
and more regular seasonal cycles. Given the relatively high fecundity of fishes, small
changes to the survival rates of larvae (King et al. 2013) can result in cascading effects on
populations, communities the ecosystems they inhabit. Therefore, improvements in
managing regulated river flow regimes have great potential to benefit native fish
communities. There is, however, little information concerning the environmental flow
requirements for early life history stages of Murray- Darling Basin fishes.
Information from around the world suggests that productivity and habitat derived from
overbank floods is critical to adult spawning and subsequent survival of young fish across a
wide variety of species (Junk et al. 1989; King et al. 2003; Baumgartner et al. 2013), although
many fishes, particularly those in variable climate regions such as the of the Murray-Darling
Basin, may thrive under low-flow conditions (Humphries et al. 1999). The flow requirements
needed to initiate spawning and support survival of early life history stages in most species
of fish in Australia remains unknown or is contested. Given the emphasis on the importance
of floodplains for fish reproduction (King et al. 2003), comparatively less is known about the
spawning and recruitment dynamics of fishes within river channel environments.
The aim of this study was to investigate the impact of in-channel environmental flow pulses
on spawning and recruitment dynamic of fish in lowland rivers.
METHODS
Experimental design and fish sampling
This study utilised a repeated before-after-control-impact (BACI; Underwood 1991) field
experimental design to examine the effects of in-channel environmental flows on the
seasonal timing and magnitude of fish spawning (larval fish) and recruitment (juveniles) in
four lowland rivers of the Edward-Wakool system. Methodological information concerning
Flow dependent ecological responses | 77
the study area can be found in the previous section outlining the Edwards-Wakool project
(section 2.1.1).
Larval and juvenile fish up to 31 mm standard length were sampled fortnightly using
quatrefoil perspex light traps in four rivers in the Edward-Wakool system between 19/9/11
and 5/4/12. Light traps contained green bioluminescent light sticks and were covered with 5
mm diameter exclusion mesh to prevent predation of larvae by larger fish. Each river was
sampled fortnightly using eight light traps distributed across two sites in 2011/12. Light
traps were set at dusk, which varied among months, and retrieved the following morning.
Larvae and juveniles were preserved in 70% ethanol or frozen for species and
developmental stage identification and measurement of standard length (SL) in the
laboratory. A sub-sample of sagittal otoliths was extracted, mounted on microscope slides
and polished for daily age estimation to approximate hatch dates. Developmental stage was
assigned according to Serafini and Humphries (2004) including: protolarvae (with and
without yolksac), flexion, postflexion, metalarvae and juveniles. Relative abundance of
larvae and juveniles was reported as the total number sampled (n) divided by effort
(number of light traps) and were reported as median values at the river-scale.
Statistics
A before-after, control and impact, ‘repeated BACI’ (Underwood, 1991), statistical design
with multiple controls was used to test the effect of the November and February
environmental watering events on fish spawning (larvae) and recruitment (juveniles). The
null hypothesis tested was that mean abundance of larvae within control rivers was not
significantly different to impacted rivers before, during or after environmental watering.
Statistical analyses were carried out in SigmaStat (3.5). Impacted reaches received
environmental water, while controls were represented by all reaches that did not receive
environmental water. Data were square root transformed to achieve equal variance and
normal distribution structure prior to analysis. Differences in square root transformed
values between control/impact reaches and before, during or after environmental watering
events were evaluated statistically using a Two-way ANOVA with data nested fortnightly
sampling periods. Post-hoc comparisons were conducted using the Holm-Sidak method.
For both environmental watering events, we compared abundance of larvae two fortnightly
periods before with two fortnightly periods after the event among control and impacted
reaches. For the November 2011 environmental flow, Little Merran Creek, Wakool River,
and Yallakool Creek represented the controls while Colligen Creek represented the impacted
river. For the February 2012 event, Colligen Creek and the Wakool Downstream represented
the two impacted reaches and controls were represented by Yallakool Creek, Little Merran
Creek. For both environmental flow events, a single fortnightly sampling period occurred
during the flow peak and this was evaluated separately to before and after data.
78 | Flow dependent ecological responses
RESULTS AND DISCUSSION
Eight species and 6717 larval and juvenile fish were sampled on fifteen fortnightly trips
between 19/09/11 and 6/04/12. Carp gudgeon and Australian smelt were the dominant
species sampled in all four rivers and they comprised of over 85% of all larvae and juveniles.
Australian smelt larvae were sampled between September and early December while
breeding of carp gudgeon occurred later in the season between November and April (Figure
1).
There was no evidence of elevated spawning activity or recruitment of Murray cod,
Australian smelt or common carp in response to in-channel environmental flows. These
species, therefore, may not be useful indicators of in-channel environmental flows. Peak
abundance of Murray cod larvae occurred in light traps between 1/11/11 and 15/11/11,
which coincided with daytime water temperatures ranging from 19.9°C to 25.36° prior to
both environmental flows. This observation is consistent with previous studies (King et al.
2009) suggesting that Murray cod do not require high spring flow pulses to spawn but that
survival of larvae or juveniles may depend on productivity derived from these events.
Floodplain and flow dependent specialist fishes including golden perch, silver perch and
southern pygmy perch (Baumgartner et al. 2013) occur in the Edward-Wakool system but
larvae or early stage juveniles of these species were not sampled. Introduced species
sampled as larvae or juveniles included common carp (n=14), redfin (n=18), mosquitofish
(n=7), and oriental weatherloach (n=1), although these species were not sampled in
adequate numbers to test for significant changes in spawning or recruitment.
When compared to control rivers (Figure 2), abundance of carp gudgeon larvae and
juveniles increased significantly (Figure 2) in Colligen Creek during (Holm-Sidak; t = 3.96; P <
0.05) and after (Holm-Sidak; t = 2.64; P < 0.05) the November 2011 environmental watering
(Figure 2) but did not change following the second event later in February. Spawning
activity of carp gudgeon peaked during spring or early summer in all rivers, regardless of
environmental flows. Therefore, environmental flows released in February, after the peak
spring-summer spawning period of carp gudgeon and other species that spawn at this time,
may be of little benefit. Based on a modal size of 8.2 mm, unvalidated otolith age estimates
suggest that the majority of larvae hatched either several days before or after the
November 2011 environmental flow in Colligen Creek. Given the length of time required for
females to develop eggs and several days for eggs to hatch, these data suggest that survival
of larvae may have been enhanced rather than these events increasing breeding activity of
carp gudgeon. However, the significance of this particular event relative to the overall
spawning and recruitment success of this species sampled in this study throughout the year
has yet to be quantified.
Flow dependent ecological responses | 79
Figure 1 Median abundance of the three most dominant larvae and juveniles sampled in the Edward-Wakool system
during 2011/12. (From Watts et al. 2013)
80 | Flow dependent ecological responses
Carp gudgeon are one of the most wide-spread and abundant native fish species in the
Murray-Darling Basin (Lintermans 2007). We suggest that the success of this group of fishes
may be in part attributed to its ability to take advantage of sporadic spawning and
recruitment opportunities provided within regulated river channel environments. This
opportunistic strategy of spawning appears to allow this species complex to thrive both in
low-flow conditions (Humphries et al. 1999) but also take advantage of conditions provided
during in-channel environmental flows or other changes that may influence riverine
productivity.
Figure 2 Results from a two-way ANOVA comparing larval carp gudgeon abundance (mean number per light trap ± SD)
before, during and after an environmental flow in Colligen Creek against control rivers that did not receive an
environmental flow. (From Watts et al. 2013)
The protracted spawning season of carp gudgeon in addition to other common small-bodied
fishes in the Murray-Darling Basin (see Humphries et al. 2013) may be the key to their
success. Depending on river channel geomorphology, in-channel rises in river discharge can
increase the availability of slow-flowing slackwater habitat generally used to describe the
Flow dependent ecological responses | 81
low-flow conditions (Humphries et al. 1999) which support carp gudgeon breeding and
reproduction. Modelling presented in Section 5.2 suggested that environmental flows in the
Edward-Wakool system inundated a relatively small amount of habitat-area and therefore,
it appears that carp gudgeon may be one of few native species able to take advantage of
small in-channel rises.
CONCLUSION
We suggest that the ability of carp gudgeon to take advantage of spawning and recruitment
opportunities within a wide range of flow conditions contributes to the success of this
species as one of the most wide-spread and abundant small-bodied fishes in the MDB. This
study demonstrates that spawning and recruitment of carp gudgeon is not exclusive to lowflow conditions. Other fishes, such as Murray cod and Australian smelt, are also able to
spawn and recruit successfully within the river channel regardless of flow conditions but
further work is required to determine how, or if, overbank environmental or natural flows
augment the magnitude of recruitment.
ACKNOWLEDGEMENTS
We extend our thanks to the Wakool River Association and landholders in the EdwardWakool river system for allowing access to their properties and for their keen interest in this
project. Field surveys and community liaison were supported by staff at the Murray
Catchment Management Authority. State Water Corporation provided information on river
and escapes discharge. We thank CSU and Murray CMA staff for assisting with field and/ or
laboratory work. This project was funded by the Cluster Collaboration Fund and the
Ecological Responses to Altered Flow Regimes Research Cluster which represents a
collaboration between the CSIRO Water for a Healthy Country Flagship, Griffith University,
the University of New South Wales, Monash University, Charles Sturt University, La Trobe
University and the Arthur Rylah Institute of the Victorian Department of Sustainability and
Environment. The project also was supported by contributions from Commonwealth
Environmental Water, Charles Sturt University, Murray Catchment Management Authority,
and Monash University..
REFERENCES
Baumgartner LJ, Conallin J, Wooden I, Campbell B, Gee R, Robinson WA and Mallen-Cooper
M (2013) Using flow guilds of freshwater fish in an adaptive management framework
to simplify environmental flow delivery of semi-arid riverine systems. Fish and
Fisheries DOI: 10.1111/faf.12023
Junk WJ, Bayley PB and Sparks RE (1989) The flood pulse concept in river-floodplain
systems. Canadian Special Publication of Fisheries and Aquatic Sciences, 106, 110-27.
82 | Flow dependent ecological responses
Humphries P, King AJ and Koehn JD (1999) Fish, Flows and Flood Plains: Links between
Freshwater Fishes and their Environment in the Murray-Darling River System,
Australia. Environmental Biology of Fishes, 56, 129-151.
Humphries P, Richardson A, Wilson G and Ellison T (2013) River regulation and recruitment
in a protracted-spawning riverine fish. Ecological Applications, 23, 208-225.
King AJ, Humphries P and Lake PS (2003) Fish recruitment on floodplains: The roles of
patterns of flooding and life history characteristics. Canadian Journal of Fisheries and
Aquatic Sciences, 60, 773-786.
King AJ, Tonkin Z and Mahoney J (2009) Environmental flow enhances native fish spawning
and recruitment in the Murray River, Australia. River Research and Applications, 25,
1205-1218.
King AJ, Humphries P and McCasker N (2013) Reproduction and early life history. In: Ecology
of Australian Freshwater Fish (eds. Humphries P and Walker K). CSIRO.
Lintermans M (2007) Fishes of the Murray-Darling Basin: An Introductory Guide. MurrayDarling Basin Authority. MDBC Publication number 10/07.
Underwood, AJ (1991) Beyond BACI: Experimental designs for detecting human
environmental impacts on temporal variations in natural populations. Australian
Journal of Marine and Freshwater Research, 42, 569-587.
Watts, RJ, Kopf, R.K., Hladyz, S., Grace, M., Thompson, R., McCasker, N., Wassens, S., Howitt,
J.A., A. Burns and Conallin, J. (2013) Monitoring of ecosystem responses to the delivery
of environmental water in the Edward-Wakool system, 2011-2012. Institute for Land,
Water and Society, Charles Sturt University, Report 2. A report to the Commonwealth
Environmental Water Office, Canberra, Australia.
Flow dependent ecological responses | 83
2.7
The importance of flow on spawning of riverine native fish
in the Barmah-Millewa region, Murray River
King, A.J.1,2, Tonkin, Z.1, Mahoney, J.1, Gwinn, D.3 & Beesley, L.1,4
1
Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment,
Heidelberg, Vic., Australia
2
Research Institute for Environment and Livelihoods, Charles Darwin University, Darwin, NT,
Australia
3
Program for Fisheries and Aquatic Sciences, School of Forest Resources and Conservation, The
University of Florida, Gainesville, Florida 32653, USA.
4
Current address: Centre of Excellence in Natural Resource Management, University of Western
Australia, Albany, Australia
Further information about this project can be found in King et al. (in prep).
INTRODUCTION
Enhancing native fish populations is a commonly stated objective of many environmental
flow programs. The flow regime can influence key factors in the life history of freshwater
fish, including pre-spawning condition and maturation, cues for both movement and
spawning, and improving larval and juvenile survival by promoting suitable food and habitat
(e.g. Junk et al. 1989; Humphries et al. 1999; Marchetti and Moyle, 2001; King et al., 2003).
The importance of flow characteristics for native fish of the Murray-Darling Basin has been
debated over many years (e.g. Harris and Gehrke 1994; Humphries et al. 1999; King et al.
2003). For example, spawning and/or recruitment of golden perch has been variously
reported as being linked to flooding, flow rises within the main channel or no flow increase
(Mallen-Cooper and Stuart 2003; Balcombe et al. 2006; Roberts et al. 2008; King et al. 2009;
Kerezsy et al. 2011; Cockayne et al. 2013; Zampatti and Leigh 2013). For other species such
as Murray cod, flow and particularly flooding, may play little role in spawning (Humphries
2005), but may be important in determining strength of recruitment (Harris and Gehrke
1994). However, previous studies have limitations in that they have: (i) used a statistically
weak correlative approach to explore the importance of flow for spawning and recruitment;
(ii) failed to consider the importance of other environmental covariates; (iii) generally
considered only the importance of flow discharge (particularly floods versus within channel
flows) and not other flow attributes; and/ or (iv) have been conducted over limited
hydrological conditions, making extrapolation to other conditions difficult.
This study explored the importance of flow as a determinant of spawning occurrence and
strength for four iconic and/or recreational valuable species of freshwater fish in the Murray
River, Australia. It used standardised catches of eggs or newly hatched larvae collected at
fortnightly intervals over the spawning season from five hydrologically distinct years; and
tested the importance of major flow attributes (such as discharge, rate of change and
number of flood days), relative to other key environmental and sampling correlates. We
84 | Flow dependent ecological responses
hypothesised that characteristics of the flow regime, both immediate and recent antecedent
flow conditions, would be a major influence on the spawning occurrence and strength of all
species, but that other environmental factors such as water temperature would also be
important. We also hypothesised that the importance of flow would vary among species,
depending upon the life history strategy employed.
METHODS
Study area
The Barmah-Millewa Forest (BMF) is a large (70,000 ha), river red gum floodplain forest on
the Murray River, in south-eastern Australia. The Forest is a highly complex wetland
containing a diverse range of aquatic habitats. The Murray River has a highly altered flow
regime, and this is a major threat to the environmental values of the BMF (Ward, 2005) and
the Murray River (Walker and Thoms, 1993). In an attempt to mitigate some of these
effects, the BMF’s water regime is highly managed using a series of offstream regulators and
structures, and the application of an environmental water allocation (EWA) (see King et al.
2010 for discussion).
Fish sampling occurred over five years (2003-2008) of variable hydrology. Water level during
the sampling period (September to February) for the first two years (2003/04 and 2004/05)
was maintained at bank-full height, with a few short duration flood events. Extensive
flooding of BMF occurred in 2005/06 sampling season, when 513 GL of environmental water
was used to extend the duration of a spring flood event (See King et al. 2009, 2010 for more
details). The last two sampling seasons were dominated by extreme drought conditions in
the Murray River catchment, resulting in unprecedented low flows in the main channel
particularly during 2007/08, and large areas of the BMF drying.
Target species and sampling methods
This study was part of a broad investigation into the effects of water management on
spawning and recruitment of the entire fish community at BMF (see King et al. 2009, 2010).
This component focused on the spawning dynamics of four native fish species: golden
perch, silver perch, Murray cod and trout cod. All four species predominantly reside in the
main channel of the Murray River in the BMF region, are large-bodied and have important
recreational, conservation and cultural significance. Golden perch and silver perch produce
100,000’s of small, semi-buoyant pelagic eggs (described as Mode 2 by Humphries et al.
1999); and have been postulated to require flooding for spawning, although evidence is
conflicting (Lake 1967; Mackay 1973; Harris and Gehrke 1994; Mallen-Cooper and Stuart
2003; King et al. 2009; Zampatti and Leigh 2013). Murray cod and trout cod produce up to
10,000 demersal eggs and exhibit parental care of eggs and early larvae (described as Mode
1 by Humphries et al. 1999). Spawning tends to occur predictably each year and is thought
Flow dependent ecological responses | 85
to be linked to both temperature and photoperiod, but not flow (Humphries 2005; Koehn
and Harrington 2006).
The presence and density of eggs and/or larvae (back calculated to likely spawning date)
was used as an indication of spawning occurrence and intensity on a particular date. Eggs
and/or larvae of the four species were collected from three sites in the Murray River:
upstream, mid region and downstream of the BMF (see King et al. 2009 for more details).
Sampling commenced each year in mid-September and continued fortnightly thereafter
until the end of February (~11 trips per breeding season), and was undertaken for five years
(total of 57 sampling trips). Sampling was conducted using two passive drift nets set
overnight at each site, as the eggs and/or larvae of these species are known to exhibit a
drifting dispersal phase (Humphries and King 2004).
Statistical analysis
We hypothesised that temporal predictors of spawning occurrence (indicated by the
presence of eggs/larvae in samples) or intensity (indicated by the relative density of
eggs/larvae), would be influenced primarily by variation in water temperature (temp), moon
phase (moon), daily discharge (flow), flooding history (Flood 90d) and the volume of water
filtered through the sampling net (VolFilt). Daylength (or photoperiod) was also
hypothesised to be an important predictor; however, it was highly correlated with water
temperature (r2 = 0.745) and was removed from the analysis.
We modelled the reproductive output of fish within a Bayesian hierarchical multispecies
framework with both an occupancy and detection model portions (Royle and Dorazio 2008).
Data were pooled among drift nets on each sampling occasion; and data from each site on
each sampling occasion were treated as replicates. Occurrence was defined as a binary
variable, indicating the presence or absence of spawning. Because the true occupancy state
cannot be observed without error, we also modelled the probability of detection of
spawning in the river. We assumed that the probability of detection of spawning would
increase if the density of eggs/larvae was higher and if we sampled with a greater intensity
(Royle and Nichols 2003). Therefore, the volume of water filtered by each drift net (VolFilt)
and environmental factors that have been hypothesised to increase spawning intensity, i.e.
discharge (flow) and number of flood days in the last three months (Flood 90day); were
included in the detection portion of the model. We assumed that the occurrence and
detection of spawning in the river for each species would therefore be affected by
environmental variables and survey characteristics. We incorporated these effects as
covariates in the model using a logit link function (McCullagh and Nelder 1989).
Model uncertainty was accounted for by averaging estimated parameters and derived
values across all possible models weighted by model probabilities using a mixture modelling
approach (Royle and Dorazio 2008). The posterior probabilities of the inclusion parameters
86 | Flow dependent ecological responses
correspond to the probability that the given variable was included in the “best” model.
Posterior probability distributions of the model parameters were estimated using a Monte
Carlo-Markov chain algorithm. All prior distributions were specified as flat normal
distributions with mean equal to zero and standard deviation equal to one hundred and
were specified for the parameters on the logit scale such that they would have no influence
on the posterior probability distributions. Inference was drawn from 30,000 posterior
samples taken from 3 chains of 100,000 samples thinned to every 10. We allowed a burn in
of 30,000 samples to remove the effects of initial values. Convergence was diagnosed for
the full model for each fish species by visual inspection of the MCMC chains for adequate
mixing and stationarity and by using the Gelman-Rubin statistic (with values < 1.1 indicating
convergence; Kery 2010; Gelmin et al. 2004). Further details of the modelling and statistical
methods can be found in King et al. (In prep).
RESULTS AND DISCUSSION
Posterior model probabilities for the top three models for each species were all very low,
and indicated that none of the top three models were probable (see King et al. in prep).
However, posterior probabilities of individual parameter inclusion in both the occupancy
and detection portion in any of the models, showed that a number of parameters were
important in predicting both the occurrence and intensity of spawning of the four species
(Table 1).
Temperature (temp2) was the dominant parameter for all four species in the occupancy
portion of the model, indicating that temperature predicted with a high certainty if
spawning occurred (Table 1). Temperature has often been regarded as the dominant
spawning cue for fishes living in temperate regions in freshwaters, where water
temperature shows marked seasonal change (King et al. 2013). However, there is limited
understanding of the relative role of temperature, compared with other factors in
determining spawning occurrence. Golden perch, trout cod and Murray cod were all
predicted to utilise an optimum temperature window (or range) predicting when spawning
would occur (Figure 1). A threshold response was evident for silver perch. Modelling
suggests that spawning was predicted to occur with a 60% probability for Murray cod
between 18–23⁰C, trout cod 17-24⁰C, golden perch 15-22⁰C and silver perch at
temperatures >19⁰C. The optimum spawning temperatures predicted for Murray cod and
trout cod agree with previous observational and correlative studies on spawning
occurrence, where spawning occurred at temperatures >15⁰C; however no studies have
suggested an upper spawning temperature threshold for these species (Humphries 2005;
Koehn and Harrington 2006). Our estimate of golden perch spawning temperature was well
below previous studies that have more commonly suggested temperatures of > 20⁰C (e.g.
Lake 1967; Roberts et al. 2008; Zampatti and Leigh 2013).
Flow dependent ecological responses | 87
Two other parameters, chwtemp and flood90d were also important for predicting the
spawning occurrence for trout cod (chwtemp: higher spawning probability with increasing
weekly temperature); and silver perch and Murray cod (flood90d: decrease in spawning
probability with more flood days) (Table 1; Figure 1).
We assumed that the probability of detection of spawning would increase if the density of
eggs/larvae was higher and if we sampled with a greater intensity. The probability of
detecting spawning was found not to be related to the volume of water filtered through the
drift nets (Volfilt), however, discharge (flow) affected the probability of detection for three
species (Table 1). Hence for golden perch, silver perch and trout cod, this suggests that an
increase in discharge increased the density of eggs/larvae or spawning intensity.
Table 1 Posterior probabilities of parameter inclusion for the occupancy and detection portion of the
models. Values in bold indicate values were model parameter estimates and their 95% credible intervals do
not overlap zero. Coding for Moon is for 3 categories of moon phase.
Variable
Golden perch Silver perch Trout cod Murray cod
Occurrence model
Temp
0.96
1.00
0.76
0.46
Temp2
0.97
0.75
1.00
0.97
Moon1
0.29
0.29
0.24
0.52
Moon2
0.31
0.29
0.26
0.50
Chwtemp
0.33
0.21
0.63
0.26
Chwflow
0.28
0.36
0.28
0.46
Flood 90d
0.20
0.58
0.15
0.77
Flow
0.36
0.28
0.20
0.32
Detection model
Flow
0.73
0.62
0.86
0.15
VolFilt
0.40
0.40
0.40
0.40
Flood 90d
0.26
0.14
0.17
0.35
The probability of detecting spawning or increased spawning intensity was predicted to
occur with a 60% probability at flows > 15,000ML/d for golden perch, and >10,000ML/d for
silver perch and trout cod. Discharge greater than 10,000ML/d started to connect major
anabranches, creeks and side channels throughout BMF, and discharge >15,000ML/d
inundated significant areas of floodplain within BMF. This was the first study that suggested
high flows may influence spawning intensity or occurrence of trout cod. However, flooding
or high flow events within the main channel has been proposed as influencing the spawning
88 | Flow dependent ecological responses
of golden perch and silver perch (e.g. Mallen-Cooper and Stuart 2003; King et al. 2009,
Cockayne et al. 2013; Zampatti and Leigh 2013). These studies have all been correlative in
nature, only weakly ascribing the influence of flow or flooding on spawning of these species.
Our results suggested that discharge did not influence the probability of when spawning
occurred, but did increase spawning intensity for trout cod, silver perch and golden perch.
Managing flows to maximise spawning intensity should result in a greater probability of a
stronger year class in recruitment. While this study covered a broad range of hydrological
conditions, it did not include a major flood; therefore, further investigation is required to
determine whether spawning intensity does increase linearly with increasing discharge or
whether a flow threshold exists.
Figure 1 Predicted probability of (a) occurrence and (b) detection of spawning for four species in the Murray
River for four key variables, temperature, number of flood days in previous 90 days (Flood days), change in
weekly temperature (Chwtemp) and discharge (flow). The light grey is the 90% probability region and the
dark grey is the 60% probability region
Flow dependent ecological responses | 89
This study clearly demonstrated that temperature was the dominant and overriding
environmental parameter influencing spawning occurrence in all four species, although
attributes of flow were also important and in particular affects the spawning intensity of
three species. This study demonstrates the importance of considering the significance of
considering the timing of water application in the design and management of environmental
flows and could be used to create a stronger conceptual or analytical framework for flow
management for improved conditions for native fish spawning.
REFERENCES
Balcombe SR, Arthington AH, Foster ND, Thoms MC, Wilson GG and Bunn SE (2006). Fish
assemblages of an Australian dryland river: Abundance, assemblage structure and
recruitment patterns in the Warrego River, Murray-Darling Basin. Marine and
Freshwater Research, 57, 619-633.
Cockayne BJ, McDougall AJ, Espinoza T, Burndred KR, Thrupp CL, Broadfoot CD and Finn MA
(2013). Riverine flow and spawning requirements of Macquaria ambigua oriens:
implications for conservation and management. Marine and Freshwater Research, 64,
42-53.
Gelman A, Carlin JB, Stern HS, Rubin DB (2004) Bayesian Data Analysis. Boca Raton:
Chapman and Hall.
Harris JH and Gehrke PC (1994) Development of predictive models linking fish population
recruitment with streamflow. Proceedings of the Australian Society for Fish Biology
Workshop. In: Population Dynamics for Fisheries Management (ed. Hancock DA).
Bureau of Rural Resources Proceedings Canberra, pp. 195-199.
Humphries P (2005) Spawning time and early life history of Murray cod, Maccullochella
peelii peelii (Mitchell) in an Australian river. Environmental Biology of Fishes, 72, 393407.
Humphries P and King AJ (2004) Drifting fish larvae in Murray-Darling Basin rivers:
composition, spatial and temporal patterns and distance drifted. In: Downstream
movement of fish in the Murray-Darling Basin. Statement, recommendations and
supporting papers from a workshop held in Canberra 3-4 June 2003 (eds. Lintermans
M and Phillips B). Murray-Darling Basin Commission Canberra, Australia., pp. 51-58.
Humphries P, King AJ and Koehn JD (1999) Fish, Flows and Flood Plains: Links between
Freshwater Fishes and their Environment in the Murray-Darling River System,
Australia. Environmental Biology of Fishes, 56, 129-151.
Junk WJ, Bayley PB and Sparks RE (1989). The flood pulse concept in river-floodplain
systems. Canadian Special Publication of Fisheries and Aquatic Sciences, 106, 110-27.
Kerezsy A, Balcombe SR, Arthington AH and Bunn SE (2011) Continuous recruitment
underpins fish persistence in the arid rivers of far-western Queensland, Australia.
Marine and Freshwater Research, 62, 1178-1190.
Kery M (2010) Introduction to WinBUGS for Ecologists: Bayesian approach to regression,
ANOVA, mixed models and related analyses: Academic Press. 320 p.
90 | Flow dependent ecological responses
King AJ, Humphries P and Lake PS (2003) Fish recruitment on floodplains: The roles of
patterns of flooding and life history characteristics. Canadian Journal of Fisheries and
Aquatic Sciences, 60, 773-786.
King AJ, Tonkin Z and Mahoney J (2009) Environmental flow enhances native fish spawning
and recruitment in the Murray River, Australia. River Research and Applications, 25,
1205-1218.
King AJ, Ward KA, O'Connor P, Green D, Tonkin Z and Mahoney J (2010) Adaptive
management of an environmental watering event to enhance native fish spawning
and recruitment. Freshwater Biology, 55, 17-31.
King AJ, Humphries P and McCasker N (2013) Reproduction and early life history. In: Ecology
of Australian Freshwater Fish (eds. Humphries P and Walker K). CSIRO.
King AJ, Tonkin Z, Mahoney J, Gwinn D and Beesley L (In prep) Does flow determine
spawning for freshwater fish? Using hierarchical multi-species modeling to inform
optimisation of environmental flows
Koehn JD and Harrington DJ (2006) Environmental conditions and timing for the spawning of
Murray cod (Maccullochella peelii peelii) and the endangered trout cod (M.
macquariensis) on southeastern Australian rivers. River Research and Applications, 22,
327-342.
Lake JS (1967) Rearing experiments with five species of Australian freshwater fishes. I.
Inducement to spawning. Marine and Freshwater Research, 18, 137-154.
Mackay NJ (1973) Histological changes in the ovaries of the golden perch, Plectroplites
ambiguus, associated with the reproductive cycle. Marine and Freshwater Research,
24, 95-102.
Mallen-Cooper M and Stuart IG (2003) Age, growth and non-flood recruitment of two
potamodromous fishes in a large semi-arid/temperate river system. River Research
and Applications, 19, 697-719.
Marchetti MP and Moyle PB (2001) Effects of flow refime on fish assemblages in a regulated
California stream. Ecological Applications, 11, 530-539.
McCullagh P and Nelder JA (1989) Generalized linear models. Chapman and Hall, London,
UK.
Roberts DT, Duivenvoorden LJ and Stuart IG (2008). Factors influencing recruitment patterns
of Golden Perch (Macquaria ambigua oriens) within a hydrologically variable and
regulated Australian tropical river system. Ecology of Freshwater Fish, 17, 577-589.
Royle JA and Nichols JD (2003) Estimating abundance from repeated presence-absence data
or point counts. Ecology 84: 777-790.
Royle JA and Dorazio RM (2008) Hierarchical modeling and inference in ecology: the analysis
of data from populations, metapopulations and communities. Oxford, UK: Elsevier
Academic Press. 444 p.
Walker KF and Thoms MC (1993) Environmental effects of flow regulation on the lower
River Murray, Australia. Regulated Rivers: Research and Management, 8, 103-19.
Flow dependent ecological responses | 91
Ward KA (2005). Water management in the changing Barmah-Millewa wetlands.
Proceedings of the Royal Society of Victoria, 117, 77-84.
Zampatti BP and Leigh SJ (2013) Within-channel flows promote spawning and recruitment of
golden perch, Macquaria ambigua ambigua – implications for environmental flow
management in the River Murray, Australia. Marine and Freshwater
Research,doi.org/10.1071/MF12321
92 | Flow dependent ecological responses
2.8
Antecedent flow: predicting the condition of young-of-year
fish in a temperate lowland river system?
Beesley, L.1, 2, Hartwell, D.1, King, A. J. 1,3 and Gwinn, D.C.4
1
Arthur Rylah Institute for Environmental Research, Department of Environment and Primary
Industries, Heidelberg, Vic., Australia
2
Current address: Centre of Excellence in Natural Resource Management, University of Western
Australia, Albany, Australia
3
Current address: Research Institute for Environment and Livelihoods, Charles Darwin University,
Darwin, NT, Australia
4
Program for Fisheries and Aquatic Sciences, School of Forest Resources and Conservation,
University of Florida, Gainesville, Florida, USA
INTRODUCTION
The nutrients and energy produced by inundating floodplains sustains fish production in
many lowland river systems, particularly in tropical, neo-tropical and arid climates (Junk et
al. 1989; Balcombe and Arthington 2009; Jardine et al. 2011, 2012). However, the
applicability of models such as the Flood Pulse Concept in temperate systems has been
questioned (see King et al. 2003) and less is known about the extent to which floods in
temperate river systems benefit fish. This is because high flows are uncoupled from
maximum water temperatures in many temperate systems, and may not increase food
production or availability (Humphries et al. 1999). High flows may also reduce the
availability of food for fish by spreading prey items, making foraging harder (Humphries et
al. 1999; King et al. 2003; Zeug and Winemiller 2003).
The effect of floods on riverine energy production may also be progressive. Firstly, primary
productivity may decline initially as the system is disturbed (Gawne et al. 2007), but increase
later as populations recover and capitalise on additional available nutrients. Secondly, there
may be a time lag between the pulse of energy associated with flooding and its
incorporation into the high level food chains (e.g., fish) (Kitchell and Carpenter 1996). And
finally, productivity in floodplain habitats may not benefit the main river channel until
secondary, or follow-up, floods wash nutrient-laden water back into the channel.
Currently, we know relatively little about the extent to which flow improves the body
condition (as one measure of productivity) of fish in temperate systems. However, there is
some evidence to suggest that flows and floods in particular, influence the body condition of
small fish (Tonkin et al. 2011; Beesley et al. 2012). This study retrospectively analysed six
years of fish data with differing flows to assess if antecedent flow was associated with the
juvenile body condition of two small-bodied native species in the mid-Murray River. The
examination of juveniles, with limited capacity to store energy and which do not use it for
reproduction, allows us to infer that their condition reflects current system energetics. This
provides insight into the longevity of flow-related energy pulses, rather than the capacity of
Flow dependent ecological responses | 93
organisms to store energy from previous energy pulses (see Balcombe et al.’s 2012 study of
adult golden perch). We also investigated the interaction between antecedent flow and
habitat for one species, Australian smelt.
METHODS:
Study Area, Sampling Design and Antecedent flow
Barmah-Millewa Forest (BMF) is a large, river red gum floodplain forest on the Murray River,
in south-eastern Australia. The forest is a highly complex wetland containing a diverse range
of aquatic habitats. Fish were collected over a five year period (2003/04 to 2007/08) as part
of a larger study examining spawning and recruitment in BMF. Hydrological conditions
varied significantly from bank full in-channel flows, long duration floods, to severe drought
conditions where most floodplain habitats dried (Figure 1). Additional fish were collected
during November 2010 during a large flood event, but before blackwater conditions started
(see King et al. 2012). A detailed description of the data and sampling was provided in King
et al. (2007, 2010). Briefly, early life stage fish were sampled at 14 sites in three habitats;
river, creek, and wetland. Sites were sampled monthly from September to February each
year using light traps and sweep-net electrofishing. All fish collected were euthanased and
stored in 100% ethanol. A sub-set of the fish collected was assessed for body condition.
60,000
50,000
Discharge (ML/d)
40,000
30,000
20,000
10,000
1/06/2011
1/06/2010
1/06/2009
1/06/2008
1/06/2007
1/06/2006
1/06/2005
1/06/2004
1/06/2003
1/06/2002
0
Figure 1 Average daily discharge (ML/day) between 2002 and 2011 in the Murray River at Barmah-Millewa
Forest (Yarrawonga gauge, 30km upstream). Triangles represented sampling occasions for Australian smelt
and unspecked hardyhead. The dashed line showed the discharge where water started flowing down creek
lines into the Forest. Overbank flooding occurred at ~ 28,000 ML/day.
94 | Flow dependent ecological responses
Flow metrics were developed to describe short, medium and long-term antecedent
conditions and are described below and visually represented in Figure 2.
•
•
•
Short-term flow: the summation of average daily discharge from the estimated
spawning date of juvenile fish, until the date of fish collection. This described the
flow conditions that juvenile fish were directly exposed to.
Medium-term flow: the summation of average daily discharge from July 1
(designated as the start of each spawning year) until the approximate date of
spawning. This described flow conditions leading up to spawning. Medium-term
flows were highly positively correlated (r=0.80) with short-term flows and were
excluded from the analysis.
Long-term flow: the summation of daily average discharge from July 1 for the
preceding year. This described flow conditions over the previous year.
The importance of flow metrics was used to assess the level of support for four different
hypotheses about flow-related energy production in the river, see Figure 2.
Flow (discharge)
D
C
Energy
B
A
July 1
July 1
flow the year before spawning
(long-term)
Time
*
flow before flow post spawning
spawning
(short-term)
(mediumterm)
Figure 2 A conceptual relationship between river flow (discharge) and system energetics (energy) through
time, and its effect on juvenile native fish condition caught at time x (triangle) and spawned at time *, for
four scenarios. Scenario (A) system energetics and fish condition are unrelated to flow; (B) system
energetics is affected by flow, but with a short-lag effect, so fish condition is related to short-term flow; (C)
system energetics is affected by flow, but with a medium lag effect, so that fish condition is related to shortterm and long-term flow; (D) system energetics is affected by flow, but with a long lag effect, so that fish
condition is related more to long-term flow than short-term flow.
Flow dependent ecological responses | 95
Fish Condition
Fish condition was assessed for two abundant small-bodied species; Australian smelt and
unspecked hardyhead. These species mature within a year, (Milton and Arthington 1985)
and their larvae and juveniles feed predominantly off zooplankton (King 2005), providing a
relatively direct link to ecosystem production. Juvenile Australian smelt (n=122) were
collected in mid-December, with an average standard length (SLLs) of 22.70 mm. Juvenile
unspecked hardyhead (n=149) were collected in mid-January, with an average length of
18.25 mm SL. A minimum of 20 individuals of each species were assessed for condition each
year (sites within habitat types were pooled). Due to limited numbers of hardyhead in
wetlands, analyses were restricted to the river for this species. In the laboratory, fish were
measured to ± 0.01 mm standard length (SL) using vernier calipers and weighed to four
significant figures. Condition was described using relative weight (Wr), and index with the
formula Wr = (W/Ws)*100, where Ws was a length-specific standard weight predicted from
length-weight relationships of all fish caught (Wege and Anderson 1978; Pope and Kruse
2007). Values between 95 and 100 indicate fish in average condition, above 100 indicated
good condition and below 95 poor condition (Pope and Kruse 2007).
Statistical Analysis
We evaluated the influence of short-term (dst) and long-term discharge (dlt) on relative
weight (Wr) for both species (independently) using linear regression. The influence of
habitat (hab) was only evaluated for smelt as hardyhead only appeared in samples collected
in riverine habitats. Models were specified with main effects and all two-way interactions.
Higher order interactions were not evaluated because of data limitations. The models were
specified as:
wr = β1 + β 2 dst + β 3 dlt + β 4 hab + β 5 hab : dst + β 6 hab : dlt + β 7 dst : dlt + ε
We considered all possible combinations of covariates to be candidate models representing
competing hypotheses. Support for competing hypotheses was determined using the
Bayesian mixture modelling approach (Royle and Dorazio 2008) where each covariate
parameter in the model is multiplied by an ‘inclusion parameter’ (wc, for c covariates).
Model probabilities of competing models were derived from the posterior samples of the
inclusion parameters by calculating the proportion of instances of different combinations of
parameters of wc in the posterior sample. Posterior probability distributions of the model
parameters were estimate using a Monte Carlo-Markov chain algorithm implemented in
JAGS (Plummer 2003) from within program R (R Development Core Team 2010). All prior
distributions were standard uninformative distributions specified to have no influence on
the posterior probability distributions. Priors for parameters were set as flat normal
distributions with mean equal to zero and standard deviation equal to one hundred.
Convergence was diagnosed for the full model by visual inspection of the MCMC chains for
96 | Flow dependent ecological responses
adequate mixing and stationarity and by using the Gelman-Rubin statistic (with values < 1.1
indicating convergence; Gelmin et al. 2004; Kery 2010). All continuous data were normalised
prior to analysis.
RESULTS AND DISCUSSION:
The relationship between antecedent flow and juvenile fish body condition differed
markedly between the two species studied. For Australian smelt, body condition was
positively related to short-term flow and negatively related to long-term flow (Table 1;
Figure 3a, b). For unspecked hardyhead, body condition was unrelated to short-term flow,
and negatively (but mildly) related to long-term flow (Table 1; Figure 3c, d). The different
responses among species with similar life histories and habitat preferences, suggests it is
unwise to make generalisations about how the body condition of different species will
respond to flow.
Table 1 Parameter inclusion probabilities for the general linear model of juvenile fish condition for two
species (Australian smelt and unspecked hardyhead) relative to antecedent flow
Parameter inclusion probabilities
Parameter
Australian smelt unspecked hardyhead
Habitat
0.612
NA
Short-term discharge
0.930
0.034
Long-term discharge
1.000
0.918
Habitat*Short-term discharge
0.827
NA
Habitat*Long-term discharge
0.224
NA
Short-term discharge*Long-term discharge
0.998
0.147
The importance of flows a year before fish were spawned, suggests that antecedent
conditions may affect the energetics of the system in excess of a year later. While
conceptually, an energy lag could follow a flood pulse, as nutrients move off the floodplain,
or up the food chain (see Figure 2), the negative relationships found here did not support
such a conceptual understanding. An alternative explanation is that antecedent low flows
allowed nutrient mineralisation and leaf litter to build up on the floodplain, which fuelled
energetics during proximal flood conditions.
The importance of short-term flow, for Australian smelt, conflicts with previous assertions
that high flows are disadvantageous for larval and small-bodied fish because they dilute the
density of suitable prey items (i.e., microcrustaceans) (Humphries et al. 1999). Recent
research by Ning et al. (2013) has revealed that although increasing flows decrease the
density of microcrustaceans, that the increased volume of water travelling down the river
leads to an increase in the total number of zooplankton being transported (i.e. food
Flow dependent ecological responses | 97
available). One would expect that the extent to which juvenile fish are able to benefit from
high flows will depend on how increased food availability trades off against increased
energy use in fast flowing waters. The relationship between smelt body condition and flow
is likely to be complicated, see Tonkin et al. (2011).
b) Unspecked hardyhead
Condition (relative weight)
Condition (relative weight)
a) Unspecked hardyhead
Long-term discharge (ML/d)
Short-term discharge (ML/d)
d) Australian smelt
Condition (relative weight)
Condition (relative weight)
c) Australian smelt
Short-term discharge (ML/d)
Long-term discharge (ML/d)
Figure 3 General linear model predictions of fish condition (relative weight) versus short (several months
since spawning) and long-term (year preceding spawning) antecedent flow for Australian Smelt and
unspecked hardyhead in the Barmah-Millewa Forest. Solid lines indicate the mean and shaded areas
indicate the 95% confidence of the mean
Short-term flow was also positively related to the condition of smelt in wetland habitats;
however, the relationship was considerably weaker than for smelt occupying the river
(Table 1, Figure 3 c, d). The weaker relationship was probably because main-channel flow
gauges were not an ideal descriptor of wetland changes in food production and habitat. This
98 | Flow dependent ecological responses
highlights the reduced ecological sensitivity when extrapolating among habitat types, using
broad hydrodynamic modelling.
In summary, this retrospective analysis provided preliminary evidence that antecedent flow
conditions may alter system energetics and fish condition. We recommend additional
investigation in this area, and caution against extrapolation of these findings because this
study is based on only six years of flow data. The relatively low number of years limits our
capacity to detect relationships, and particularly to understand how different components
of flow interact.
REFERENCES
Balcombe SR and Arthington AH (2009) Temporal changes in fish abundance in response to
hydrological variability in a dryland floodplain river. Marine and Freshwater Research
60, 146-159.
Balcombe SR, Lobegeiger JS, Marshall SM, Marshall JC, Ly D and Jones DN (2012) Fish body
condition and recruitments success reflect antecedent flows in an Australian dryland
river. Fisheries Science 78, 841-847.
Beesley L, King AJ, Amtstaetter F, Koehn JD, Gawne B, Price A, Nielsen DL, Vilizzi L and
Meredith SN (2012) Does flooding affect spatiotemporal variation of fish assemblages
in temperate floodplain wetlands? Freshwater Biology 57, 2230-2246.
Gawne B, Merrick C, Williams DG, Rees G, Oliver R, Bowen PM, Treadwell S, Beattie G, Ellis I,
Frankenberg J and Lorenz Z (2007) Patterns of primary and heterotrophic productivity
in an arid lowland river. River Research and Applications 23, 1070-1087.
Gelman A, Carlin JB, Stern HS, Rubin DB (2004) Bayesian Data Analysis. Boca Raton:
Chapman and Hall.
Humphries P, King AJ and Koehn JD (1999) Fish, Flows and Flood Plains: Links between
Freshwater Fishes and their Environment in the Murray-Darling River System,
Australia. Environmental Biology of Fishes 56, 129-151.
Jardine TD, Pettit NE, Warfe DM, Pusey BJ, Ward DP, Douglas MM, Davies PM and Bunn SE
(2011) Consumer–resource coupling in wet–dry tropical rivers. Journal of Animal
Ecology 81, 310-322.
Jardine T, Pusey B, Hamilton S, Pettit N, Davies P, Douglas M, Sinnamon V, Halliday I and
Bunn S (2012) Fish mediate high food web connectivity in the lower reaches of a
tropical floodplain river. Oecologia 168, 829-838.
Junk WJ, Bayley PB and Sparks RE (1989) The flood pulse concept in river-floodplain
systems. Canadian Special Publication of Fisheries and Aquatic Sciences 106, 110-27.
Kery M (2010) Introduction to WinBUGS for Ecologists: Bayesian approach to regression,
ANOVA, mixed models and related analyses: Academic Press.
King AJ (2005) Ontogenetic dietary shifts of fishes in an Australian floodplain river. Marine
and Freshwater Research 56, 215-225.
Flow dependent ecological responses | 99
King AJ, Humphries P and Lake PS (2003) Fish recruitment on floodplains: The roles of
patterns of flooding and life history characteristics. Canadian Journal of Fisheries and
Aquatic Sciences 60, 773-786.
King AJ, Tonkin Z and Mahoney J (2007) Assessing the effectiveness of environmental flows
on fish recruitment in Barmah-Millewa Forest. Arthur Rylah Institute for
Environmental Research, Department of Sustainability and Environment. MDBC
Project No. BMF 2004.09. Arthur Rylah Institute for Environmental Research, DSE.
King AJ, Ward KA, O'Connor P, Green D, Tonkin Z and Mahoney J (2010) Adaptive
management of an environmental watering event to enhance native fish spawning
and recruitment. Freshwater Biology 55, 17-31.
King AJ, Tonkin Z and Lieschke J (2012) Short-term effects of a prolonged blackwater event
on aquatic fauna in the Murray River, Australia: considerations for future events.
Marine and Freshwater Research 63, 576-586.
Kitchell JF and Carpenter SR (1996) The trophic cascade in lakes. Cambridge University Press,
Cambridge.
Milton DA and Arthington AH (1985) Reproductive strategy and growth of the Australian
smelt, Retropinna semoni (Weber) (Pisces: Retropinnidae), and the olive perchlet,
Ambassis nigripinnis (De Vis) (Pisces: Ambassidae) in Brisbane, south-eastern
Queensland. Australian Journal of Marine and Freshwater Research 36, 329-341.
Ning NSP, Gawne B, Cook RA and Nielsen DL (2013) Zooplankton dynamics in response to
the transition from drought to flooding in four Murray-Darling Basin rivers affected by
differing levels of flow regulation. Hydrobiologia 702, 45-62.
Plummer M (2003) JAGS: A Program for Analysis of Bayesian Graphical Models Using Gibbs
Sampling. In: Leisch F and Zeileis Z (eds), Proceedings of the 3rd International
Workshop on Distributed Statistical Computing (DSC 2003), Vienna, Austria.
Pope KL and Kruse CG (2007) Condition. In: Guy CS and Brown ML (eds) Analysis and
interpretation of freshwater fisheries data pp. 423-471. American Fisheries Society,
Bethesda, Maryland.
Royle JA, Dorazio RM (2008) Hierarchical modeling and inference in ecology: the analysis of
data from populations, metapopulations and communities. Elsevier Academic Press,
Oxford.
Tonkin ZD, King AJ, Robertson AI and Ramsey DS (2011) Early fish growth varies in response
to components of the flow regime in a temperate floodplain river. Freshwater Biology,
56, 1769-1782.
Wege GJ and Anderson RO (1978) Relative weight (Wr): a new index of condition for
largemouth bass. In: Novinger GD and Dillard JG (eds) New approaches to the
management of small impoundments, pp. 79-91. American Fisheries Society,
Bethesda, Maryland.
Zeug SC and Winemiller KO (2008) Relationships between hydrology, spatial heterogeneity,
and fish recruitment dynamics in a temperate floodplain river. River Research and
Applications 24, 90-102.
100 | Flow dependent ecological responses
2.9
Hypoxic blackwater survival thresholds of four Australian
lowland river fishes
Small, K.1, Kopf, R.K.1, Watts, R.J.1 & Howitt, J.1
1
Institute for Land, Water and Society, Charles Sturt University, Albury, Australia
This report is based on results published in Small et al. (2014), Hypoxic blackwater
survival thresholds of four Australian lowland river fishes. PLoS ONE 9(4) e94524.
INTRODUCTION
Hypoxia has been influential in shaping the evolution of freshwater fish and the
environments in which they live. Hypoxic blackwater events in can be a natural and
relatively common phenomenon associated with floodplain inundation and the subsequent
decomposition of organic material in lowland rivers (Hladyz et al. 2011). Fish employ a
variety of mechanisms ranging from air-breathing to behavioural avoidance, although
severe, widespread, or prolonged hypoxia can lead to sub-lethal stress and mortalities
(Gehrke et al. 1993; McMaster and Bond 2008; La and Cooke 2012).
Given the regular occurrence of drought, flooding and subsequent blackwater events in
lowland rivers of the Murray-Darling Basin (MDB) in south-eastern Australia (Howitt et al.
2007; Whitworth et al. 2012), it is reasonable to assume that native fish in this ecosystem
are well adapted to acute hypoxia and high levels of DOC. However, after 10 years of
drought, the MDB river system experienced multiple large-scale flooding events followed by
blackwater events and fish kills (Whitworth et al. 2012). Hypoxic blackwater mortalities of
keystone species, such as Murray cod (Mitchell), have been the focus national and
international attention (Koehn 2004; La and Cooke, 2010; Whitworth et al. 2012) and bring
into question the hypoxia tolerances of these species.
This study aimed to quantify hypoxia tolerances of juvenile stages of four lowland river
fishes native to the MDB. The species examined included Murray cod, golden perch, silver
perch and eel tailed catfish.
METHODS
Two-hundred to two-hundred and fifty juveniles of each species were sourced from local
hatcheries in south-eastern Australia and were transported the same day to a controlled
temperature room. The juvenile stage of development was selected as a starting point for
minimum oxygen thresholds because critical oxygen concentrations were assumed to be
independent of body size among juvenile and adult fish species (Nilsson and Ostlund-Nilsson
2008).
Flow dependent ecological responses | 101
Hypoxia tolerance of each species was determined using median lethal dissolved oxygen
concentration (LC 50 ) experiments in freshwater (no dissolved organic carbon/leaf litter) and
in simulated blackwater treatments containing four different leaf litter concentrations and a
control (no leaf litter). All leaf litter treatments were repeated with atmospheric air bubbled
in tanks (aeration) and without aeration on each of the four species. Prior to
experimentation each species was acclimated to a constant temperature of 25° C for 9 to 15
days. All experiments were undertaken in glass experimental tanks, 23 cm (L) x 23 cm (W) x
45 cm (H), custom built to achieve minimal surface area and limit oxygen diffusion with the
atmosphere. A static flow during all experiments was utilised to reduce variation in DO
concentration. Prior to each experiment DO sensors were calibrated against temperaturespecific oxygen saturated distilled water.
Experiments started by netting 10 fish and transferring them from 750 l holding tanks to a
pre-determined experimental DO treatment tank. Water from holding tanks was used in all
experimental and control tanks to avoid stress during transfer. Dissolved oxygen,
temperature and pH were measured in each experimental and control tank at intervals of 0
h, 1 h, 4 h, 8 h, 16 h, 24 h, 32 h, 40 h, 48 h and DO concentrations were regulated with
bubbled N 2 gas or atmospheric air as necessary.
Lethal dissolved oxygen concentration
Median lethal dissolved oxygen concentration (LC 50 ) experiments were conducted in
accordance with international guidelines for acute toxicity testing (Anon. 1992). Dissolved
oxygen (DO) was regulated using nitrogen, N 2 , or atmospheric air. Atmospheric air and N 2
were separately regulated via air stones in experimental tanks filled with 17·5 l of holding
tank water. Seven oxygen treatments, each containing 10 fish per experimental tank, were
maintained during LC 50 experiments on each species. Treatments consisted of DO
concentrations of 0·1 mg l-1, 0·7 mg l-1, 1·3 mg l-1, 1·9 mg l-1, 2·5 mg l-1 and 3·1 mg l-1, and a
control (8 mg l-1) that were maintained at 25° C for 48 h.
Blackwater simulation
Four simulated blackwater treatments were developed using naturally abscised dried river
red gum Eucalyptus camaldulensis leaf litter collected from a fallen branch near the Murray
River in Albury, NSW. Leaf litter and twigs, < 5 mm diameter, were dried in oven at 60° C for
24 to 48 hrs and four geometrically increasing masses of dried leaf litter containing 3 g, 9 g,
27 g, and 81 g were placed in separate nylon stockings weighted with 250 g of cobble. These
masses resulted in effective leaf litter concentrations ranging from 0.17 g l-1 to 4.59 g l-1
within experimental tanks and were designed to simulate dissolved organic carbon and
oxygen conditions ranging from oxygen saturated to hypoxic blackwater.
102 | Flow dependent ecological responses
Natural blackwater event
Water quality conditions of experimental treatments were compared to a naturally
occurring hypoxic blackwater event sampled in the MDB from the Edward River (Natural 1)
and Wakool River (Natural 2) in December, 2010. John and Anthony Conallin from the
Murray CMA and Luke Pearce from the NSW Department of Primary Industries confirmed
the presence or absence of fish species mortalities in each river following the event. Water
quality measurements were recorded during the hypoxic black water conditions, while fish
mortalities in the field were verified during and after events. Water quality measurements
of DO, pH and temperature were collected in flowing water within the main river channel
during both events.
Data analysis and statistics
Lethal DO concentrations (LC 50’s ) and aquatic surface respiration (ASR 50’s ) values were
estimated by fitting log transformed 48 h cumulative percent responses to a logistic
regression (Anon., 2006):
y =a+
b−a
x
1 +
ASR
orLC
50
50
− Hillslope
where the fitted parameter a is the minimum y-axis value, b is the asymptote, LC50 or
ASR50 represent the median response and hillslope is the slope at its midpoint and
mortality responses were over the 48 h experimental period of lethal oxygen.
Species-specific LC50’s, ASR50’s were compared independently using One Way Analysis of
Variance’s (ANOVA’s). A Three Way ANOVA was used to examine variation in cumulative
percent mortality among species, leaf litter treatments and the presence/absence of
aeration. All post hoc comparisons were undertaken using Tukey Tests (Zar, 1999).
RESULTS/DISCUSSION
The juveniles of four large-bodied fishes examined here were unexceptional in their ability
to tolerate low dissolved oxygen in freshwater and simulated blackwater. Logistic models
(Figure 1) predicted that hypoxia related mortalities may start at oxygen concentrations as
high as 3.54 mg l-1 and sub-lethal stress, as indicated by the onset of aquatic surface
respiration (ASR), may occur above 5.00 mg l-1 at 25°- 26o C. These results were somewhat
unexpected given the variable water quality conditions and purported regular occurrence of
hypoxic blackwater events in lowland rivers of the MDB (Howitt et al. 2007; Hladyz et al.
2011). There is increasing evidence suggesting that anthropogenic factors including river
Flow dependent ecological responses | 103
flow regulation, hypolimnetic weir discharge and climate change (Whitworth et al. 2012)
may be escalating the severity of hypoxic events in lowland river systems.
Figure 1 Logistic regressions fitted to (a) cumulative percent mortality showing lethal dissolved oxygen
concentrations (LC50’s) and (b) cumulative percent performing aquatic surface respiration (ASR50’s) at
2
different oxygen concentrations for juvenile (black circles; solid thin line ) Murray cod (r = 0·99, F3;4 =
2
2
1331·7, P < 0·001; r = 0·92, F3;4 = 25·0, P < 0·05); (white circles; solid thick line ) golden perch (r = 0·99, F3;4
2
2
= 603·2, P < 0·001; r = 0·99, F3;4 = 439·4, P < 0·001); (black triangles; dashed line) eel-tailed catfish (r = 0·98,
2
2
F3;4 = 11960·5, P < 0·001; r = 0·97, F3;4 = 83·8, P < 0·05); and (white triangles; dotted line ) silver perch (r =
2
0·99, F3;4 = 3393·8, P < 0·001; r = 0·99, F3;4 = 1331·7, P < 0·001).
104 | Flow dependent ecological responses
Our results suggest that low oxygen levels of less than 2.0 mg l-1, in addition to the high DOC
(> 10 mg l-1) and low pH (4.2-6.3), which occurred in the Edward-Wakool system (Table 1)
and across much of the southern basin for several weeks or months during these events
(Whitworth et al. 2012), approached or exceeded lethal threshold limits for all species
except for eel-tailed catfish. Confirmed field mortalities of the remaining three fish species
in the Edward-Wakool system indeed corroborate this assumption (Table 1). It is important
to consider, however, that fish in the wild may be able to avoid hypoxic blackwater by
seeking refuge habitat, while fish in our study were confined within experimental treatment
tanks that may have exaggerated estimates of mortality.
Table 1. Mortality (%) of four Murray-Darling Basin fishes exposed to simulated blackwater experiments
and confirmed mortalities associated with natural hypoxic blackwater events in the Edward-Wakool river
system in 2010/11. Water quality and dissolved organic carbon (DOC) measurements of simulated
-1
blackwater treatments (mass of leaf litter g l ) were aggregated across species for comparison with two
natural blackwater events. Mean ± SE
Blackwater
sample
Natural 1
Natural 2
Control
-1
(0 g l )
Simulated
blackwater
-1
(0.17 g l )
Simulated
blackwater
-1
(0.51 g l )
Simulated
blackwater
-1
(1.53 g l )
Simulated
blackwater
-1
(4.59 g l )
a.
-
DO (mg l
1
)
1.07
(0.08)
1.98
(0.11)
8.4 (0.4)
pH
Temp.
o
( C)
DOC
-1
(mg l )
Mortality (%)
Murray cod Golden perch Eel-tailed
a
catfish
Confirmed Confirmed
-
Silver perch
Confirmed
Confirmed
-
Confirmed
0
0
0
0
0
0
0
0
4.18
(0.04)
6.25
(0.09)
7.41
(0.05)
7.22
(0.04)
19.4
(0.02)
18.7
(0.05)
25.62
(0.03)
26.26
(0.08)
17.7
(0.08)
16.1
(0.07)
2.7
(0.05)
3.58
(0.38)
7.05
(0.01)
26.09
(0.06)
4.93
(0.13)
0
0
0
0
1.8 (0.44) 6.76
(0.03)
25.93
(0.06)
9.45
(0.55)
90
7
8
45
0.46
(0.28)
26.41
(0.05)
30.0
(0.7)
> 99
> 99
> 99
> 99
4.83
(0.38)
4.0 (0.3)
6.05
(0.01)
Confirmed
Historically present in the Edward-Wakool system but not sampled in recent unpublished fisheries surveys .
The range of hypoxia tolerances (Table 2) reported here may be considered intermediate on
a global-scale comparison with other freshwater fishes. For example, the crucian carp
Carassius carassius (L.) and several non-air breathing fishes of the Amazon can withstand
complete anoxia for several months (Kramer et al. 1978; Nilsson and Renshaw 2004), while
some salmonids are vulnerable to mortality when oxygen concentrations fall below 4 mg l-1
(Davis 1975). The hypoxia tolerances observed here ranged from 0.70 mg l-1 to 1.29 mg l-1
at 25°C-26°C and are comparable to other temperate stream fishes. The mean LC50 of 35
Flow dependent ecological responses | 105
species in the mid-western United States ranged from 0.49 mg l-1 to 1.59 mg l-1 (Smale and
Rabeni 1995), while other prairie stream fishes have LC50’s ranging from 0.95 mg l-1 to 2.66
mg l-1 at 25°C (Ostrand and Wilde 2001). The LC50’s of four native New Zealand fishes and
rainbow trout, Oncorhynchus mykiss (Walbaum) ranged from 0.54 mg l-1 to 2.65 mg l-1 at
15°C (Landman et al. 2005).
All four species performed ASR in response to low DO and this behaviour has not previously
been documented in peer-reviewed literature for any of these species. Aquatic surface
respiration was initiated in all fishes before lethal oxygen levels were reached and this is a
common trend in other species in which this behaviour has been documented (Chapman
and McKenzie, 2009). The ASR50’s (Table 2) reported here fell within the range 0.48 to 1.98
mg l-1 reported by McNeil and Closs (2007) at 25°C for other Australian native fishes
including Australian smelt Retropinna semoni (Weber), flat-headed galaxias Galaxias
rostratus (Klunzinger), carp gudgeon compex Hypseleotris spp. and southern pygmy perch
Nannoperca australis (Gunther).
One outstanding question remaining from the present study is whether or not the hypoxia
tolerances of the juveniles examined here are representative of adults that are more
commonly reported in fish kills in the MDB. A review on this subject across multiple species
of marine fish concluded that body size itself plays little or no direct role in determining
hypoxia tolerance and that the tolerance of a given species is usually size-independent after
the larval stage (Nilsson and Ostlund-Nilsson 2008).
Table1 Species and water quality data for 48 h lethal dissolved oxygen (DO) exposures. Logistic functions
were used to predict lethal DO concentrations (LC50’s) and aquatic surface respiration (ASR50’s) values.
Mean ± S.E.
Species
n
Mass
(g)
Temperature
(°C)
pH
DO (mg l )
range
-1
DO (mg l )
LC 50
-1
DO (mg l )
ASR 50
-1
Murray cod
80
1·02 ± 0·02
25·63 ± 0·15
7·92 ± 0·11
0·16 - 8·24
1·29 ± 0·03
1·39 ± 0·24
Golden
perch
Eel tailed
catfish
Silver perch
80
0·48 ± 0·02
25·41 ± 0·09
7·75 ± 0·08
0·07 - 8·14
0·96 ± 0·03
1·40 ± 0·04
80
3·61 ± 0·10
26·35 ± 0·22
7·52 ± 0·18
0·09 - 8·68
0·70 ± 0·02
0·79 ± 0·09
70
0·43 ± 0.02
26·19 ± 0·13
7·52 ± 0·11
0·09 - 8·67
1·18 ± 0·01
1·51 ± 0·07
In contrast to popular anecdote, it appears that hypoxia sensitivity of Murray cod may not
be entirely related to its large maximum attainable body mass (> 100,000 g). The Murray
cod examined here were juveniles (< 1.2 g) and not the largest species, yet they were still
the most sensitive examined. The comparatively low hypoxia tolerance of Murray cod
appears to be species-specific or, at least, not entirely size-specific. We suggest that
Australia’s largest freshwater fish species is sensitive to hypoxia, not because it a largebodied apex predator per se, but because it has evolved a life-history strategy that primarily
utilizes lotic river channel environments and habitats, which are less prone to severe
106 | Flow dependent ecological responses
hypoxic blackwater conditions than floodplains, wetlands or backwater areas commonly
occupied by other species .
This study simulated a relatively narrow range of blackwater types (Figure 2) compared to
the potential array of conditions that fish may experience (see Howitt et al. 2007; Hladyz et
al. 2011). However, the un-aerated leaf litter concentration of 1.53 mg l-1 resembled the
early stages of a severely hypoxic natural blackwater event which occurred in the southern
MDB in 2010/11 (see Whitworth et al. 2012). The mortality rate of large-bodied fish in this
simulation was 37.5% and this result was corroborated by observed fish kills of Murray cod
during the natural event. However, blackwater simulations in the present study were
different to the natural blackwater event in that fish were confined within hypoxic
conditions and experiments were undertaken at higher water temperatures and higher
pH’s.
Figure 2 Experimental leaf litter treatments showing (a) cumulative percent mortality of juvenile Murray
cod, golden perch, eel-tailed catfish, and silver perch; (b) dissolved oxygen concentration; (c) dissolved
organic carbon (DOC) concentration; and (d) pH. Black circles represent non-aerated treatments and white
circles represent aerated treatments. Non-overlapping 95% confidence intervals represent significant
differences between aerated and non-aerated treatments and asterisks indicate significant differences from
respective controls (no leaf litter).
Flow dependent ecological responses | 107
The spatial extent of the natural blackwater event extended across much of the southern
basin, including the main channel of the Murray River (Whitworth et al. 2012), which
probably eliminated the possibility for behavioural avoidance of hypoxia in most areas.
Water quality measurements from the natural blackwater event also showed that the pH
was more acidic than blackwater simulations which would have further reduced the oxygen
affinity of the gills.
Mortality increased significantly with leaf litter concentration in non-aerated treatments
(Three way ANOVA, F 4,12 = 18·8, P < 0·001; Figure 2a) but no significant differences were
detected among species. Mean mortality rates in non-aerated were significantly different
(Tukey; q < 8·98; P < 0·05) to controls starting at leaf litter concentrations of 1.53 g l-1 and a
mean dissolved oxygen concentration of 1.83 ± 0.3 S.E. mg l-1. There was a relatively high
degree of variability in oxygen levels within non-aerated treatments but DO sharply declined
with increasing leaf litter concentration (Figure 2b). Aeration elevated DO concentrations to
saturation and simultaneously neutralised pH levels across all treatments (Figure 2d).
CONCLUSION
Hypoxia is a natural phenomenon in aquatic ecosystems but conservation and management
intervention may be required if the frequency, severity or spatial extent of these events has
changed due to anthropogenic causes such as river regulation. Despite the natural
occurrence of hypoxia and blackwater in lowland rivers of the MDB, large-bodied predatory
fish in this ecosystem are vulnerable to acute mortality induced by low oxygen
concentration and associated water chemistry changes.
Given the extent of natural flow regime alteration and climate change predictions of rising
temperatures and more severe droughts and flooding, acute episodes of hypoxia and
blackwater may represent a formidable risk to aquatic biodiversity and ecosystem
functioning in lowland rivers. Hypoxic blackwater fish kills in combination with the ‘modest’
survival thresholds presented here suggest that large-bodied native fishes of the MDB are
not particularly well adapted to acute hypoxia when behavioural avoidance is restricted.
The thresholds reported here provide a starting point to begin developing hypoxia
mitigation strategies but must be carefully interpreted in light of potential sub-lethal effects,
ambient and antecedent environmental conditions. It is essential to consider that hypoxia
thresholds will change according to temperature, pH and concentration of toxic
polyphenols. Consequently, a better understanding how these variables interact will be
essential in further developing hypoxia management plans to prevent fish kills and the loss
of aquatic biodiversity. One of the proposed management approaches to avoid
anthropogenic hypoxic events is to restore natural ecosystem functions and natural river
flow regimes. In cases where this is not possible, other management measures such as
aeration, environmental water dilution flows (Hladyz et al. 2011), or other regulatory flow
diversion measures (Howitt et al. 2007) may mitigate the effects of hypoxia.
108 | Flow dependent ecological responses
ACKNOWLEDGEMENTS
We thank Charles Sturt University for a Faculty of Science Honours scholarship and the
Faculty of Science laboratory staff for assistance in maintaining fish and the controlled
temperature room. This research was funded in part by the Ecological Responses to Altered
Flow Regimes Research Cluster which represent a collaboration between the CSIRO Water
for a Healthy Country Flagship, Griffith University, the University of New South Wales,
Monash University, Charles Sturt University, La Trobe University and the Arthur Rylah
Institute of the Victorian Department of Sustainability and Environment. This work was
carried out with permission from the CSU Animal Care and Ethics committee protocol
number 11/018.
REFERENCES
Anon (1992) Guideline for Testing of Chemicals 203 Fish, Acute Toxicity Test. Paris, France:
Organisation for Economic Cooperation and Development.
Http://www.oecd.org/chemicalsafety/assessmentofchemicals/1948241.pdf
Davis JC (1975) Minimal dissolved oxygen requirements of aquatic life with emphasis on
Canadian species: a review. Journal of the Fisheries Research Board of Canada 32,
2295-2332.
Gehrke PC, Revell, MB and Philbey AW (1993) Effects of river red gum, Eucalyptus
camaldulensis, litter on golden perch, Macquaria ambigua. Journal of Fish Biology 43,
265-279.
Hladyz S, Watkins SC, Whitworth KL and Baldwin DS (2011) Flows and hypoxic blackwater
events in managed ephemeral river channels. Journal of Hydrology 401, 117-125.
Howitt JA, Baldwin DS, Rees GN and Williams JL (2007) Modelling blackwater: predicting
water quality during flooding of lowland river forests. Ecological Modelling 203, 229242.
Junk JW, Bayley PB and Sparks RE (1989) The flood pulse concept in river flood plain
systems. Special Publication Canadian Journal of Fisheries and Aquatic Sciences 106,
110-127.
Koehn JD (2004) The loss of valuable Murray cod in fish kills: a science and management
perspective. Canberra, Australia: Paper presented at the Management of Murray cod
in the Murray-Darling Basin: Statement, recommendations and supporting papers.
Kramer DL, Lindsey CC, Moodie GE, and Stevens ED (1978) The fishes and the aquatic
environment of the central Amazon basin, with particular reference to respiratory
patterns. Canadian Journal of Zoology 56, 717-729.
La VT and Cooke SJ (2011) Advancing the science and practice of fish kill investigations.
Reviews in Fisheries Science 19, 21-33.
Landman MJ, Van den Heuvel MR and Ling N (2005) Relative sensitivities of common
freshwater fish and invertebrates to acute hypoxia. New Zealand Journal of Marine
and Freshwater Research 39, 1061-1067.
Flow dependent ecological responses | 109
McMaster D and Bond N (2008) A field and experimental study on the tolerances of fish to
Eucalyptus camaldulensis leachate and low dissolved oxygen concentrations. Marine
and Freshwater Research 59, 177-185.
McNeil DG and Closs GP (2007) Behavioural responses of a south east Australian floodplain
fish community to gradual hypoxia. Freshwater Biology, 52, 412-420.
Nilsson GE and Renshaw GMC (2004) Hypoxic survival strategies in two fishes: extreme
anoxia tolerance in the North European crucian carp and natural hypoxic
preconditioning in a coral-reef shark. Journal of Experimental Biology 207, 3131-3139.
Nilsson, G. E. and Ostlund-Nilsson S. (2008). Does size matter for hypoxia tolerance in fish?
Biological Reviews 83, 173-189.
Ostrand KG and Wilde GR (2004). Temperature, dissolved oxygen, and salinity tolerances of
five prairie stream fishes and their role in explaining fish assemblage patterns.
Transactions of the American Fisheries Society 130, 742-749.
Smale MA and Rabeni CF (1995) Hypoxia and hyperthermia tolerances of headwater stream
fishes. Transactions of the American Fisheries Society 124, 698-710.
Whitworth KL, Baldwin DS, and Kerr JL (2012) Droughts, floods and water quality: Drivers of
a severe hypoxic blackwater event in a major river system (the southern MurrayDarling Basin, Australia). Journal of Hydrology 450-451, 190-198.
Zar JH (1999) Biostatistical Analysis, 4th edn. Prentice-Hall, Upper Saddle River, NJ.
110 | Flow dependent ecological responses
2.10
Flow changes to the Lowbidgee floodplain from the
Murrumbidgee river
Ren, S.1 and Kingsford, R.T.1
1
Centre for Ecosystem Science, School of Biological, Earth and Environmental Sciences, University of New
South Wales
INTRODUCTION
Many of the world’s rivers are fragmented by regulatory structures (Nilsson et al. 2005),
affecting flows and impacting on freshwater biodiversity (Vӧrӧsmarty et al. 2010). Analysis
of changes to river flows, supplying major wetlands, is critical to understanding responses of
organisms, particularly long-lived biota, to changes in flows and subsequently management
of environmental flows. These challenges are particularly relevant to Australian rivers which
have relatively low and variable flows, reflecting Australia’s aridity (Puckridge et al. 1998). It
is in the Murray–Darling Basin in the southeastern Australia where there is a strong focus on
these two challenges. The Murray-Darling Basin is the most highly regulated river basin in
Australia with more than 30,000 ML of storage capacity (Kingsford 2000; Kingsford et al.
2011; Leblanca et al. 2012), supplying large irrigated areas. The Australian Government has
invested more than A$ 12 billion in purchasing water from the irrigation industry and
improving efficiencies of irrigation systems which return river flows to the environment,
including major wetland ecosystems (DEWHA 2010). Such wetlands usually lie in the lower
regions of the major catchments.
One of the large river systems targeted for rehabilitation is the Murrumbidgee and its
ecologically important Lowbidgee wetland, subject to considerable alteration from water
resource development (Kingsford 2003; Kingsford and Thomas 2004). The Lower
Murrumbidgee floodplain has exhibited symptoms of ecological decline, as a result of water
resource development on the 1,690 km long Murrumbidgee River that supplies the wetland
(Kingsford and Thomas, 2004; Keith et al. 2013). Understanding the level of hydrological
impact to many rivers, including the Murrumbidgee River, is primarily measured by
governments using the Integrated Quantity and Quality Model (IQQM) in southeastern
Australia (Simons et al.. 1996; Hameed and O'Neill 2005; Brown et al.. 2007). Independent
hydrological analysis of monthly river flow data is possible using time series, but the
residuals are large, limiting predictive power, even though results may be generally
consistent with the integrated quantity and quality model (IQQM) (Wen 2009).
METHODS
To explore long-term changes to flow in Murrumbidgee River, daily observed flow data
(NSW Natural Resources Pinneena Version 9 and NSW Water Information Website
Flow dependent ecological responses | 111
http://waterinfo.nsw.gov.au) were used at seven flow gauges, at increasing distances
upstream of the Lowbidgee wetland, the major wetland in the Murrumbidgee River
catchment (Kingsford and Thomas, 2002; Kingsford and Thomas, 2004; Figure 1): Gundagai
(1887-2010); Wagga Wagga (1892-2010); Narrandera (1893-2010); Darlington Point (19152010); Hay (1914-2010); Maude (1937-2010) and Redbank (1937-2010). Maude and
Redbank were gauges within the Lowbidgee wetland. Also, we assembled annual rainfall
data for five locations, mainly in the upper catchment (Queanbeyan, Cooma, Yass, Tumut
and Cootamundra), where most of the flows (Kingsford and Thomas 2004) to the Lowbidgee
were generated (Figure 1). Daily rainfall data (Bureau of Meteorology Australian Daily
Rainfall CD 2010) were transformed to monthly and then annual data from November to
October to coincide with usual annual spring flooding in the Lowbidgee wetland, retaining
an annual time frame. There were some missing data, so monthly rainfall data were
interpolated, based on the relationship among the rainfall data at the corresponding station
and nearby stations, to provide an annual rainfall record from 1880 to 2010. Annual rainfall
for nine stations in the upper, middle (Wagga Wagga, Narrandera) and lower catchment
(Hay, Maude) were calculated for the 131 year available record.
Figure 1: The Murrumbidgee River (thick central black line) and its tributaries and distributaries (thin black
lines) flow west to the Lowbidgee floodplain wetland (shaded green) in southeastern Australia, where
Burrinjuck and Blowering Dams (part of the Snowy Mountains Scheme) are the major regulatory structures;
long-term models were developed using rainfall stations (filled circles) and flow gauges (open circles).
The Lowbidgee wetland, at the lowest part of the Murrumbidgee catchment, has had its
flow regime affected by upstream river flows after Burrinjuck, Snowy Mountains Scheme
dams and other upper catchment dams were built, followed by water resource
development (Kingsford 2003; Kingsford and Thomas, 2004; CSIRO, 2008a and 2008b; Wen
et al., 2011a and 2011b). We developed annual low regulated flow models, before the river
112 | Flow dependent ecological responses
became highly regulated, at seven flow gauges (Gundagai, Wagga Wagga, Narrandera,
Darlington Point, Hay, Maude and Redbank), using annual rainfall at the catchment rainfall
locations as the independent variables (Figure 1). We used structural change tests to
identify the two different periods of flow change in the river (Zeileis et al. 2002; 2003 and
2010). This calculated the empirical fluctuation of monthly flow at the Redbank gauge, with
cumulative sums of recursive residuals providing the point beyond which there was a
declining trend. This clear break point of annual structural change was in 1957, after which
the most significant dam building occurred on the river (Kingsford 2003). This separated the
flow record into a period of low river regulation, when total storage capacity was 1044.5GL
(1927-1957) and when the river was highly regulated and storage capacity increased
(3028.4GL, 1958-1971) (Kingsford 1995; Kingsford 2003). There were no significant
differences (p>0.05) in annual rainfall between the two periods: 1880-1957 and 1958-2010
for the eight stations in the catchment, except Yass rainfall station (p=0.025) using K-S test.
River regulation was the catalyst for diversion of water to the Murrumbidgee Irrigation Area
and the Colleambally Irrigation Area (Kingsford 2003). The latter was developed when water
was diverted from the easterly flowing Snow River, with the development of the Snowy
Mountains Hydroelectricity Scheme, increasing flows in the river above natural levels for
this part of the river (Kingsford 2003).
We used LOESS to build the relationship between annual observed flow and upstream flow /
rainfall data (Ren et al, 2010), modeled in R language (R Development Core Team, 2013).
LOESS fits local polynomial regressions progressively to points (Cleveland 1979; Cleveland
and Devlin 1988) and has the advantage over other modelling approaches because it does
not rely on the specification of a function to fit the model to all of the sample data. LOESS
(Cleveland et al. 1992) does not produce a regression function, so we calculated quantile
confidence intervals, based on the leave-one-out samples which successively drops the ith
observation from the original sample (Ren et al.. 2010). We used upstream annual flow data
and local rainfall data for the models downstream of the Gundagai gauge, corresponding to
the locations of Wagga Wagga, Narrandera, Darlington Point, Hay, Maude and Redbank
gauges (Figure 1). We progressively developed a series of flow models down the river, using
sequential flow gauges. We developed annual flow models at the Gundagai gauge using
catchment rainfall. For subsequent gauges downstream, we used flow models from the
upstream gauge. At Maude, flow models also included local rainfall at Hay station as well as
flows from the Hay models. Following this process, we independently developed models for
annual low regulated flow at the seven gauges - Gundagai (1887-1957), Wagga Wagga
(1892-1957), Narrandera (1893-1957), Darlington Point (1915-1957), Hay (1914-1957),
Maude (1937-1957) and Redbank (1937-1957); and then correspondingly highly regulated
models at the seven flow gauges for the period 1958-2010. We derived annual low and
highly regulated flow models for Maude and Redbank gauges, within the Lowbidgee (see
Figure 1). Following development of these flow models, which coincided with the time that
Flow dependent ecological responses | 113
flow data were available, we then extended our modelling to the full record (1880-2010),
using the available rainfall data and the predicted flow data from upstream flow models to
estimate long-term changes in low and highly regulated flow. We developed annual low and
highly regulated flow models at the flow gauges of Maude and Redbank with overall results.
RESULTS AND DISCUSSION
The low regulated flow modelling at the Maude gauge was built with annual observed flow
data at Maude, low regulated flow and rainfall data at Hay from 1937 to 1957 using LOESS
and leave-one-out samples. It showed high agreement between observed data and fitted
data because Nash-Sutcliffe coefficient of efficiency as the goodness-of-fit indicator is 99%
(Nash and Sutcliffe, 1970), and there was no significant difference between the observed
and fitted distributions of flow at Maude using Kolmogorov-Smirnov test (p>0.9999). The
highly regulated flow modelling at Maude was developed using the relationship between
annual flow data at Maude and highly regulated flow data at Hay after 1958. Observed and
fitted flows at Maude from 1958 to 2010 were similarly close (goodness-of-fit is 98.7%) with
no significant difference (p>0.9999). With our low and highly regulated flow models at the
Maude gauge, we extended the predictions of these models using annual low and highly
regulated flow and rainfall data at Hay (1880-2010) (Figure 2a).
The low regulated flow modelling at Redbank was built based on the relationship between
annual flow data at the Redbank gauge and low regulated flow data at the Maude gauge
from 1937 to 1957. The highly regulated flow modelling were developed using the
relationship between annual flow data at the Redbank gauge and highly regulated flow data
at the Maude gauge after 1958. Similar results were found in a comparison between the
respective fitted data from the low and high regulated modelling developed for the Redbank
gauge and observed data (goodness-of-fit: 91.5% and 95.1%; Kolmogorov-Smirnov test:
p>0.9999; p>0.9999). And then we extended the low and highly regulated flow modelling to
predict annual flow at the Redbank gauge to the Lowbidgee wetland (1880-2010) (Figure
2b).
We developed two models (low and highly regulated) for annual flows in the Murrumbidgee
River, at different locations, based on annual rainfall data, extending for 131 years from
1880 to 2010. Even though climate change may reduce rainfall and related flow, rainfall
remains the most independent variable available for tracking long term changes to river
systems (Herron et al. 2002), without the confounding effects of river regulation (Kingsford
2000). Therefore such modelling can extend current understanding of the effects of river
regulation to large wetlands and consequently for understanding biotic responses.
We found there was a significant difference (p<0.0001) between low and highly regulated
flows from 1880 to 2010 of the Murrumbidgee at Maude with median highly regulated flow
was 52.8% of the median low regulated flow. Similarly, these effects also occurred at the
114 | Flow dependent ecological responses
downstream Redbank gauge where median annual highly regulated flows were 47.4% of the
low regulated flows (p<0.0001).
Figure 2: Annual low and highly regulated flow (1880-2010) at Maude and Redbank gauges
Such statistical approaches to flow modelling may be useful for rivers around the world and
may be linked to inundation modelling and ecosystem responses. Since April 2007, we have
created and been developing new integrated flow and flood modelling (IFFM) including daily
flow modelling in the Macquarie and Murrumbidgee Rivers, monthly sub-regional inundated
area size modelling in the Macquarie Marshes and Lowbidgee, monthly spatial sub-regional
flood modelling in the Macquarie Marshes and Lowbidgee, long-term annual unregulated
and regulated flow modelling in the Macquarie River (Ren et al, 2010; Ren and Kingsford,
2011), long-term annual inundated area size modelling and long-term annual spatial flood
Flow dependent ecological responses | 115
modelling in the Macquarie Marshes. The series of statistical modelling down the river, from
the upstream gauge, extending progressively downstream, can provide another
independent modelling approach to changes to flow regimes to large wetland systems. Such
modelling can extend to the full record available for a river, allowing assessment of the
frequency and magnitude of floods of large wetlands. Combined with understanding of
requirements of biota in wetlands, this can provide critical information on whether
sufficient flows are provided to sustain extent and character of floodplain wetlands.
REFERENCES
Brown AE, Podger GM, Davidson AJ, Dowling TI and Zhang L (2007) Predicting the impact of
plantation forestry on water users at local and regional scales: An example for the
Murrumbidgee River Basin, Australia. Forest Ecology and Management 251, 82–93.
CSIRO (Australian Commonwealth Scientific and Industrial Research Organization) MurrayDarling Basin Sustainable Yields Project (2008a) Water Availability in the MurrayDarling Basin. A report to the Australian Government from the CSIRO Murray-Darling
Basin Sustainable Yields Project. CSIRO, Canberra.
CSIRO (Australian Commonwealth Scientific and Industrial Research Organization) MurrayDarling Basin Sustainable Yields Project (2008b) Uncertainty in river modelling across
the Murray-Darling Basin.
http://www.clw.csiro.au/publications/waterforahealthycountry/mdbsy/technical/OSurfaceWaterModelUncertainty.pdf
Cleveland WS (1979) Robust locally weighted regression and smoothing scatterplots. Journal
of the American Statistical Association 74, 829-836.
Cleveland WS and Devlin SJ (1988) Locally weighted regression: an approach to regression
analysis by local fitting. Journal of the American Statistical Association 83, 596-610.
Cleveland WS, Grosse E and Shyu WM (1992) Local regression models. In Statistical Models
in S, Chambers JM and Hastie TJ (eds.); Wadsworth and Brooks/Cole: Pacific Grove,
CA; Chapter 8.
DEWHA (2010) Secure our water future. Department of the Environment, Water, Heritage
and the Arts, Website visited June 2011:
http://www.environment.gov.au/water/publications/action/pubs/securingwaterfuture.pdf
Gille ST (2004) Using Kolmogorov-Smirnov statistics to compare geostrophic velocities
measured by the Jason, TOPEX, and Poseidon altimeters. Marine Geodesy 27, 47-57.
Hameed T and O'Neill R (2005) River management decision modelling in IQQM. In MODSIM
2005 International Congress on Modelling and Simulation 1957-1962, Zerger A and
Argent RM (eds). Modelling and Simulation Society of Australia and New Zealand,
http://www.mssanz.org.au/modsim05/papers/hameed.pdf
Herron N, Davis R and Jones R (2002) The effects of large-scale afforestation and climate
change on water allocation in the Macquarie River catchment, NSW, Australia.
Journal of Environmental Management 65, 369-381.
116 | Flow dependent ecological responses
Keith DA, Rodriguez JP, Rodriguez-Clark KM, Nicholson E, Aapala K, Alonso A, Asmussen M,
Bachman S, Basset A, Barrow EG, Benson JS, Bishop MJ, Bonifacio R, Brooks TM,
Mark A, Burgman MA, Comer P, Francisco A, Comin FA, Essl F, Faber-Langendoen D,
Fairweather PG, Holdaway RJ, Jennings M, Kingsford RT, Lester RE, Nally RM,
McCarthy MA, Moat J, Oliveira-Miranda MA, Pisanu P, Poulin B, Regan TJ, Riecken U,
Spalding MD and Zambrano-Martinez S (2013) Scientific Foundations for an IUCN
Red List of Ecosystems. PLoS ONE 8(5): e62111.
Kingsford RT (1995) Ecological effects of river management in New South Wales. In
Conserving Biodiversity: Threats and Solutions (eds.) RA Bradstock, TD Auld, DA,
Keith, RT, Kingsford D Lunney and DP Sivertsen. Surrey Beatty and Sons, Sydney: 144161.
Kingsford RT (2000) Protecting or pumping rivers in arid regions of the world? Hydrobiologia
427, 1-11.
Kingsford RT (2003) Ecological impacts and institutional and economic drivers for water
resource development – a case study of the Murrumbidgee River, Australia. Aquatic
Ecosystem Health and Management 6, 69-79.
Kingsford RT and Thomas RF (2002) Use of satellite image analysis to track wetland loss on
the Murrumbidgee River floodplain in arid Australia, 1975-1998. Water Science and
Technology 45, 45-53.
Kingsford RT and Thomas RF (2004) Destruction of wetlands and waterbird populations by
dams and irrigation on the Murrumbidgee River in arid Australia. Environmental
Management 34, 383-396.
Kingsford, RT, Walker, KF, Lester, RE, Young, WJ, Fairweather, PG, Sammut, J and Geddes,
MC, (2011) A Ramsar wetland in crisis - the Coorong, Lower Lakes and Murray
Mouth, Australia. Marine and Freshwater Research 62, 255-265.
Leblanca, M, Sarah Tweed, S, Dijk, AV and Timbal, B, (2012) A review of historic and future
hydrological changes in the Murray-Darling Basin. Global and Planetary Change 8081: 226–246.
Nash JE and Sutcliffe JV (1970) River flow forecasting through conceptual models, Part I: a
discussion of principles. Journal of Hydrology 10, 282-290.
Nilsson C, Reidy CA, Dynesius M and Revenga C (2005) Fragmentation and flow regulation of
the World’s large river systems. Science 308, 405-408.
Puckridge JT, Sheldon F, Walker KF and Boulton AJ (1998) Flow variability and the ecology of
large rivers. Marine and Freshwater Research 49, 55–72.
R Development Core Team (2013) R: A language and environment for statistical computing.
Vienna, Austria. (http://www.R-project.org)
Ren SQ and Kingsford RT (2011) Statistically Integrated Flow and Flood Modelling Compared
to Hydrologically Integrated Quantity and Quality Model for Annual Flows in the
Regulated Macquarie River in Arid Australia. Environmental Management, 48, 177188.
Ren SQ, Kingsford RT and Thomas FT (2010) Modelling flow to and inundation of the
Macquarie Marshes in arid Australia. Environmetrics, 21, 549–561.
Flow dependent ecological responses | 117
Simons M, Podger G, and Cooke R, 1996. IQQM - A hydrologic modelling tool for water
resource and salinity management. Environmental Software 11, 185-192.
Vorosmarty CJ, McIntyre PB, Gessner MO, Dudgeon D, Prusevich A, Green P, Glidden S,
Bunn SE, Sullivan CA, Reidy LC and Davies PM (2010) Global threats to human water
security and river biodiversity. Nature 467, 555–561.
Wen L (2009) Reconstruction natural flow in a regulated system, the Murrumbidgee River,
Australia, using time series analysis. Journal of Hydrology 364, 216–226.
Wen L, Marsh N, Mackay C, Salbe I, Saintilan N and Ling J (2011) Linking wetland hydrology
to ecological outcomes in the Lowbidgee wetlands in Southern New South Wales. In
MODSIM 2011 International Congress on Modelling and Simulation 2479-2485.
Modelling and Simulation Society of Australia and New Zealand,
http://www.mssanz.org.au/modsim2011/e15/wen.pdf
Wen L, Rogers K, Ling J and Saintilan N (2011) The impacts of river regulation and water
diversion on the hydrological drought characteristics in the Lower Murrumbidgee
River, Australia. Journal of Hydrology, 405, 382-391.
Zeileis A, Leisch F, Hornik K and Kleiber C (2002) Strucchange: An R package for testing for
structural change in linear regression models. Journal of Statistical Software 7(2), 138.
Zeileis A, Kleiber C, Kramer W and Hornik K (2003) Testing and dating of structural changes
in practice. Computational Statistics and Data Analysis 44, 109-123.
Zeileis A, Shah A and Patnaik I (2010) Testing, monitoring, and dating structural changes in
exachange rate regimes. Computational Statistics and Data Analysis 54, 1696-1706.
118 | Flow dependent ecological responses
2.11
Inundation modelling in the Macquarie Marshes
S. Ren1, R. T. Kingsford1 and R. F. Thomas1,2
1
Centre for Ecosystem Science, School of Biological, Earth and Environmental Sciences, University of
New South Wales
2
Water Wetlands and Coast Science, NSW Office of Environment and Heritage
INTRODUCTION
There is an increasing need to understand relationships between river flows and inundation
responses (Thomas et al. 2011). Such understanding of the frequency and extent of
inundation provides requisite information for interpreting how much wetlands may have
declined with river regulation but also serves as a background for providing information for
the management of environmental flows. The growing availability of multi-temporal
satellite data has increased opportunities for monitoring large rivers from space (Smith
1997). A variety of passive and active sensors operating in the visible and microwave range
can estimate inundation area and delineate flood boundaries. High-resolution
visible/infrared sensors provide good delineation of inundated areas, where trees and
floating vegetation do not obscure the water surface (Smith 1997). These data can be used
to develop relationships between flow and flooding, critical to understanding dependencies
and sufficiency of environmental flow for maintaining organisms on floodplain wetlands
(e.g. floodplain eucalypts). We aimed to develop such modelling for the Macquarie Marshes,
a complex floodplain wetland dependent on flows from the Macquarie River. Since April
2007, we have created and been developing new integrated flow and flood modelling
(IFFM) including daily flow modelling in the Macquarie and Murrumbidgee Rivers, monthly
subregional inundated area size modelling in the Macquarie Marshes and Lowbidgee,
monthly spatial subregional flood modelling in the Macquarie Marshes and Lowbidgee,
long-term annual unregulated and regulated flow modelling in the Macquarie River (Ren et
al. 2010; Ren and Kingsford 2011), long-term annual low and highly regulated flow
modelling in the Murrumbidgee River, long-term annual inundated area size modelling and
long-term annual spatial flood modelling in the Macquarie Marshes. These statistical
modelling and approaches are useful for the inundation modelling in the Macquarie
Marshes.
METHODS
The Macquarie Marshes, which flows north, is a large wetland (Kingsford and Thomas 1995;
Ren et al., 2010; Ren and Kingsford, 2011) on the floodplain of the Macquarie River. They
cover 379,330 ha, comprising four regions in the Macquarie Marshes (Figure 1). These
regions accord with different areas interpreted for the management of environmental flows
Flow dependent ecological responses | 119
used by the New South Wales environmental flow management agency (Thomas et al.
2011). We used 95 observed inundated maps (2550 x 6343 pixels, where each pixel is 25m x
25m) from Landsat 5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper
(ETM+) images in the Macquarie Marshes for different dates between 9 February 1989 and
21 January 2011 (Thomas et al. 2012). Each image was divided into the four regions,
comprising different regions: east, north, south and west (Table 1). For each image,
inundated area was classified using remote sensing analyses to provide an observed
estimate of area flooded (Thomas et al. 2011).
Table 1: Descriptive statistics of areas inundated in four regions within the Macquarie Marshes on 95 dates
from 09/02/1989 to 21/01/2011
Region
East
North
South
West
Area (ha)
140,186
68,382
63,757
107,005
Min
46
190
316
127
Max
100,616
53,016
44,712
47,811
Figure 1: The Macquarie Marshes and its stream network and the main flow gauges (left), and the
corresponding four water management regions (East, North, South and West) used to manage
environmental flows in the Macquarie Marshes
To model inundated areas in the four regions, we used daily observed flow data (NSW
Natural Resources Pinneena Version 9 and NSW Water Information Website
http://waterinfo.nsw.gov.au) at five flow gauges: Dubbo (January 1, 1913 - December 31,
2011); Warren (January 1, 1913 - December 31, 2011); Marebone Break (August 1, 1975 -
120 | Flow dependent ecological responses
December 31, 2011); Marebone Weir Downstream (April 1, 1986 - December 31, 2011) and
Oxley (January 1, 1943 - December 31, 2011). Marebone Break, Marebone Weir
Downstream and Oxley were gauges within the Macquarie Marshes (Figure 1). The
described statistics of total flow (ML) at Dubbo, Warren, Marebone Break, Marebone Weir
Downstream and Oxley flow gauges were calculated for the previous 30 days before each
image date (May 1, 1986 to December 31, 2011; Table 2). Some missing data of daily flow
data were based on our own daily flow modelling (Ren and Kingsford, in prep.) using the
daily data from the upstream flow and rainfall. Our IFFM daily flow modelling (The
corresponding Nash–Sutcliffe efficiency are 99.99% at Oxley between 01/01/1943 and 31
November 2011 and 99.99% at Carinda between 1 January 1939 and 31 October 2011,
Figure 2) in the Macquarie River should be better than the integrated quantity and quality
model (IQQM: The corresponding Nash–Sutcliffe efficiency are 95% at Oxley and 76% at
Carinda between 7 January 1991 and 30 June 2009; Wen et al., 2013).
Figure 2: Comparison of the observed and fitted data from IFFM daily flow modelling at Oxley – 1 Jan 1943
to 31 October 2011
We also calculated the total rainfall at Quambone and mean air temperature within the
Macquarie Marshes (Australian Government Bureau of Meteorology
http://www.bom.gov.au) for the previous 30 days of each date (May 1, 1986 - December 31,
2011) as this can also influence inundation extent. Monthly sub-regional inundated area
sizes in the east, north, south and west regions of the Macquarie Marshes were then
Flow dependent ecological responses | 121
modelled using the corresponding upstream total flow and rainfall, and mean air
temperature data.
We developed a series of models linking inundation in the four regions to total flow at the
Warren, Marebone Break and Marebone Weir Downstream flow gauges, total rainfall at
Quambone (Figure 1) and mean air temperature in the Macquarie Marshes for the previous
30 days for each date of the 95 images and then compared these to the observed inundated
area sizes. We used Voronoi-based lifting wavelets (Jansen et al., 2009), LOESS (Cleveland
1979; Cleveland and Devlin 1988) and leave-one-out samples without overfitting (Ren et al.
2010) to build the monthly subregional inundated area size modelling.
A series of subregional inundated area size modelling in the Eastern Macquarie Marshes
linked the total flow at the Warren, Marebone Break and Marebone Wier Downstream flow
gauges, total rainfall at Quambone and mean air temperature in the Macquarie Marshes for
the previous 30 days of these 95 dates to the 95 observed eastern inundated area sizes. We
also built a series of sub-regional inundated area size modelling in the Northern Macquarie
Marshes to describe the relationship of the total flow at the Marebone Break, Marebone
Weir Downstream and Oxley flow gauges, total rainfall and mean air temperature in the
Macquarie Marshes for the previous 30 days of these 95 dates, and the 95 observed
northern inundated area sizes. A series of sub-regional inundated area size modelling in the
Southern Macquarie Marshes was developed to the relationship of the total flow at the
Warren, Marebone Break and Marebone Weir Downstream flow gauges, total rainfall and
mean air temperature in the Macquarie Marshes for the previous 30 days of these 95 dates
and the 95 observed southern inundated area sizes. Similarly, a series of sub-regional
inundated area size modelling in the Western Macquarie Marshes was based on the
relationship of the total flow at the Dubbo and Warren flow gauges, total rainfall and mean
air temperature in the Macquarie Marshes for the previous 30 days of these 95 dates, and
the 95 observed western inundated area sizes.
Table 2: Descriptive statistics of total flows at five flow gauges within the Macquarie Marshes for the
previous 30 days of the 95 image dates from 09/02/1989 and 21/01/2011
Gauge
Dubbo
Warren
Marebone Weir Downstream
Oxley
Marebone Break
122 | Flow dependent ecological responses
median (GL)
63.636
33.678
19.173
14.298
7.56
2.5% quantile
6.792
3.342
1.116
0.591
0.843
97.5% quantile
1116.519
371.601
115.047
80.235
83.652
RESULTS AND DISCUSSION
We compared between modelled and observed inundated area sizes for the 95 dates using
Kolmogorov-Smirnov test (Weiss 1978; Gille 2004). There was no significant difference
between the observed and predicted distributions of inundated area sizes for the eastern
(p=0.5498), northern (p=0.5498), southern (p=0.3374) and western (p=0.5498) regions,
respectively. Reflecting this, there was good agreement between observed and modelled
data using the Nash-Sutcliffe coefficient of efficiency as the goodness-of-fit indicator, based
on the squares of the differences between modelled and observed data (Nash and Sutcliffe
1970) for each of the four regions (Table 3): east (98.3%, Figure 3a), north (98% Figure 3b),
south (97.8%, Figure 3c) and west (96.5%, Figure 3d).
Table 3: Comparison of observed and fitted area sizes (ha) for 95 images of the Macquarie Marshes
Region
East
North
South
West
Median
observed
fitted
1,579
5,282
4,341
2,653
1,601
5,430
4,377
2,351
95% CI (Q 2.5 , Q 97.5 )
observed
fitted
(73, 77,697)
(300, 46,088)
(628, 35,773)
(311, 36,391)
(135, 77,698)
(1,194, 45,809)
(1,207, 35,530)
(833, 36,389)
K-S test*
p value
GOF
(%)
0.5498
0.5498
0.3374
0.5498
98.3
98.0
97.8
96.5
* Kolmogorov-Smirnov goodness-of-fit test
No significant differences between observed and modelled inundated area sizes allowed us
to develop a series of monthly estimates of inundated area sizes (May 1986- December
2011) for each of the regions of the Macquarie Marshes, based on flow, rainfall and
temperature (30 days previously) for each of the regions (Figure 3). We found that there
were uncertainties in this modelling in relation to confidence limits, based on comparisons
with observed data. This probably reflected variability in soil moisture, drying patterns, local
water management as well as uncertainties in the mapping of the wetlands (Thomas et al.
2011). The monthly inundated area sizes are dependent on the predicted inundated area
sizes on the dates of the corresponding month. If the predicted inundated area sizes on all
the dates of a month in each of the regions tend to be significantly increasing or decreasing
using Mann-Kendall trend test (Mann 1945), we then derived an estimate of area inundated
in each of the regions in this month based on the 75% quantile of all the predicted area
sizes; otherwise, the inundated area size in each of the regions in this month is the median
of these predicted area sizes (Figure 4). There were three large flood peaks flood events in
August 1990 (226,775 ha), December 2010 (195,772 ha), and December 2000 (166,119 ha)
in Figure 4.
Flow dependent ecological responses | 123
Figure 3: Comparisons of observed and modelled estimates of wetland area for four different regions of the
Macquarie Marshes for 95 images between 09/02/1989 and 21/01/2011
The monthly sub-regional inundated area size modelling performed well, but if there are
more flood maps on regular dates such as 16 day periods, then these modelling need to be
rebuilt and developed based on their autocorrelation information. These modelling are
critical for us to build and develop the monthly spatial sub-regional flood modelling in the
Macquarie Marshes that related to hydrological and ecological impacts on the large wetland
systems. IFFM monthly spatial subregional flood modelling in the Macquarie Marshes (over
accuracy is 99.6% on 16/12/2000, Figure 5; Ren et al. in prep.) should be better than MIKE
FLOOD (about 75% of the satellite data on 16/12/2000; Wen et al. 2013).
124 | Flow dependent ecological responses
Figure 4: Estimated monthly inundated areas for the four regions in the Macquarie Marshes (May 1986 –
December 2011)
Fine temporal scale inundation data (monthly) provides opportunity to test responses of
different organisms to the flooding regime (frequency, timing, magnitude). These data can
be used to interpret condition of floodplain eucalypts for example to determine whether
there is sufficient inundation to maintain their populations. Also, these data may be used to
investigate duration of inundation over time which is critical for the breeding of organisms
such as colonial waterbird. Further, these modelling can be used to predict inundation
patterns for environmental flows and build a series of scenarios for potential inundation of
different floodplain communities. Finally, such inundation models may save considerable
time in the classification of remote sensing imagery and also provide information to fill
missing data where satellite images cannot be used (e.g. cloud obscured).
Flow dependent ecological responses | 125
Figure 5: Comparison of the observed and IFFM modelled maps on 16/12/2000
REFERENCES
Cleveland WS (1979) Robust locally weighted regression and smoothing scatterplots. Journal
of the American Statistical Association 74, 829-836.
Cleveland WS and Devlin SJ (1988) Locally weighted regression: an approach to regression
analysis by local fitting. Journal of the American Statistical Association 83, 596-610.
Gille ST (2004) Using Kolmogorov-Smirnov statistics to compare geostrophic velocities
measured by the Jason, TOPEX, and Poseidon altimeters. Marine Geodesy 27, 47-57.
Jansen M, Nason GP and Silverman BW (2009) Multiscale methods for data on graphs and
irregular multidimensional situations. Journal of the Royal Statistical Society: Series B
(Statistical methodology) 71, 97-125.
Kingsford RT and Thomas RF (1995) The Macquarie Marshes and its waterbirds in arid
Australia: A 50-year history of decline. Environmental Management 19, 867-878.
Mann HB (1945) Non-parametric test against trend. Econometrica 13, 245-259.
Nash JE and Sutcliffe JV (1970) River flow forecasting through conceptual models, Part I: a
discussion of principles. Journal of Hydrology 10, 282-290.
R Development Core Team (2013) R: A language and environment for statistical computing.
Vienna, Austria. (http://www.R-project.org)
126 | Flow dependent ecological responses
Ren SQ and Kingsford RT (2011) Statistically Integrated Flow and Flood Modelling Compared
to Hydrologically Integrated Quantity and Quality Model for Annual Flows in the
Regulated Macquarie River in Arid Australia. Environmental Management, 48, 177188.
Ren SQ, Kingsford RT and Thomas FT (2010) Modelling flow to and inundation of the
Macquarie Marshes in arid Australia. Environmetrics, 21, 549–561.
Ren SQ and Kingsford RT (in prep.) Daily flow modelling in the Macquarie River.
Ren SQ, Kingsford RT and Thomas FT (in prep.) Monthly spatial subregional flood modelling
in the Macquarie Marshes.
Smith LC (1997) Satellite remote sensing of river inundation area, stage, and discharge: a
review. Hydrological Processes, 11, 1427-1439.
Thomas RF, Lu Y, Cox SJ and Hunter SJ (2012) Inundation response project, executive
summary, NSW Rivers Environmental Restoration Program Subprogram II, Office of
Environment and Heritage NSW, Sydney.
Thomas RF, Kingsford RT, Lu Y and Hunter SJ (2011) Landsat mapping of annual inundation
(1979−2006) of the Macquarie Marshes in semi-arid Australia. International Journal
of Remote Sensing, 32, 4545−69.
Weiss MS (1978) Modification of the Kolmogorov-Smirnov statistic for use with correlated
data, Journal of the American Statistical Association 73, 872-875.
Wen L, Macdonald R, Morrison T, Hameed T, Saintilan N and Ling J (2013) From
hydrodynamic to hydrological modelling: Investigating long-term hydrological
regimes of key wetlands in the Macquarie Marshes, a semi-arid lowland floodplain in
Australia. Journal of Hydrology, 500, 45–61.
Flow dependent ecological responses | 127
2.12
Wetland and floodplain understory vegetation
J.S. Wilson, B. Gawne, A. Price & R. Durant
Murray-Darling Freshwater Research Centre, Wodonga. La Trobe University, Wodonga
INTRODUCTION
Riparian vegetation (encompassing wetlands and floodplains) affect floodplain and
ecosystem processes in many ways, including the provision of habitat, altering
biogeochemical cycles, primary production and fixing carbon (Carpenter and Lodge 1986;
Bragg et al. 2005). The community can be grouped into two components: the canopy
comprised of trees and the understory comprised of shrubs, macrophytes, herbs and
grasses. The understory makes a significant contribution to ecosystem primary production
(Gawne et al. 2007; Baldwin et al. 2013).
Flow events and their characteristics (e.g. magnitude, frequency, duration, depth,
fluctuation, timing/seasonality) affect the germination, distribution, diversity and growth of
riparian vegetation (Casanova and Brock 2000; Deegan et al. 2007; Franklin et al. 2008;
Casanova and Brock 2000; Greet et al. 2011; Webb et al. 2012). Flow regime becomes
increasingly important as the climate becomes more arid, as there is insufficient rainfall to
maintain vegetation communities (Baldwin et al. 2013). Understanding floodplain
production in response to environmental flows and inundation is critical to understanding
energy flow and storage across the floodplain ecosystem.
A conceptualisation of the influence of flow on floodplain production identified a range of
flow characteristics expected to influence production directly through impacts on physical
habitat and indirectly through effects on competitors and herbivores (the model Appendix
1). Of the influences identified, current knowledge suggests that the influence on physical
habitat is the major driver and so, in terms of developing predictive capacity, this causal
relationship was our focus.
The aim of this project was to investigate how primary production of riparian or floodplain
vegetation responds to flow and determine if understory riparian vegetation can predict
ecological response models of floodplain/wetland condition. Remote sensing techniques
can successfully predict changes in wetland and woodland condition in response to
inundation and could be used to calculate primary production (Cunningham et al. 2007;
Sims and Colloff 2012). However it is more difficult to determine the contribution of the
understory because overstory canopy tends to obscure understory vegetation growth. As a
result we aimed to model riparian production response using variables known to drive their
production (season/timing, depth and duration of inundation) as well as determining if this
was consistent across broad geographical floodplain habitat types.
128 | Flow dependent ecological responses
Riparian understory production at the landscape scale was predicted to be dependent on
the floodplain habitat type, reflecting historically different flow regimes (depth and
duration) affecting community composition. In this study we have considered three habitat
types, temporary open water wetlands, red gum (Eucalyptus camaldulensis) woodlands and
black box (Eucalyptus largiflorens) woodlands. Water residence time within each habitat for
a given flow will vary, given water remains longer in open water wetlands than red gum
woodlands and black box woodlands. As a result, understory production was predicted to be
greater in wetlands than in red gum woodlands, followed by black box woodlands.
Both above and below ground biomass, in the form of roots, are important (Baldwin et al.
2013). Below ground carbon is often overlooked in production estimates due to
methodological difficulties. The proportional allocation of above to below ground carbon of
individual species of macrophyte is affected by flood depth and water level fluctuations (Rea
and Ganf 1994; Smith and Brock 2007) however few studies have investigated the whole of
community responses.
METHODS
Two mesocosm experiments were conducted to examine the effects of the timing, depth
and duration of inundation on floodplain plants from three different communities: red gum
(Eucalyptus camaldulensis), black box (Eucalyptus largiflorens) and temporary open water
wetlands. Soil and sediment containing seeds of plants were collected (see detailed
methods in Casanova and Brock 2000 and Nielsen et al. 2007), from the three habitat types
in Yanga National Park and adjoining farmlands on the Lower Murrumbidgee River
floodplain, NSW in November 2011 and June 2012. Surface soil/sediment (2 – 5cm depth)
was collected from within 0.01m quadrats, between 32 and 41 quadrats were collected
from each habitat type and the soil pooled. Soil from each habitat type was lightly crushed,
air dried, mixed thoroughly and sieved to remove vegetation, prior to subsampling.
The experiment was run using a mesocosm at Wonga Wetlands Environmental Centre,
Albury, NSW (Nielsen et al. 2007), over two seasons (summer and winter). The experimental
design (Table 1) had five replicate pots of each habitat by depth combination, sacrificially
sampled one each sampling occasion. Emerging plant seedlings were identified, taxon
richness, number of individuals of each taxon and dry weight of above and below ground
biomass of each taxon recorded for each pot. Identification of vegetation species was based
on Cunningham et al. (2006), Sainty and Jacobs (1981, 2003) and The Royal Botanic Gardens
and Domain Trust online plant guide (PlantNET).
Flow dependent ecological responses | 129
Table 1 Experimental design showing dates experiment 1 and experiment 2 were conducted and the levels
of each of the four factors (Season, Duration, Habitat Types and Depths).
FACTORS/TRATMENTS
EXPERIMENT 1
th
EXPERIMENT 2
rd
Experiment Dates
14 November 2011 to
th
6 April 2012
23 July 2012 to
th
8 October
Season
Summer/autumn
Winter/spring
Duration/sampling occasion (weeks)
4, 8, 12, 16, 20
4, 8, 12
Habitat Types
wetland, red gum, black box
wetland, red gum, black box
Flood Depth (cm)
0.1 (saturated), 5, 15, 30
0.1 (saturated), 5, 30
Data Analysis
Species richness, abundance (count), above ground biomass and belowground biomass data
were analysed within each season with a three-factor analysis of variance (ANOVA) and
Holm-Sidak post hoc test in SigmaPlot v11 after assumption testing. Differences between
seasons were determined by t-test across all treatments (where treatments did not exist in
winter – 15 cm, weeks 16 and 20 – they were removed from the t-test analysis). Below
ground biomass as a percentage of total biomass data was analysed with a three factor
PERMANOVA (Primer + PERMANOVA v.6) (Anderson et al. 2008) due to not meeting the
assumptions of equal variance to enable analysis by ANOVA. Multivariate community
structure data analysis was undertaken within each season, using a three-factor
PERMANOVA of square root transformed above ground biomass. Where significant main
effects were detected, pairwise comparisons were undertaken to determine which
treatment groups differed. Differences between seasons were determined by one way
analysis of similarity (ANOSIM) on square root transformed data, using a Bray Curtis
similarity matrix (Clarke and Warwick 2001). Data were presented in a non-metric
multidimensional scaling (n-MDS) ordination using Bray Curtis similarity.
RESULTS AND DISCUSSION
Seasonal variation
Biomass, both above and below ground, species richness and abundance were all
significantly lower in winter than in summer, averaged across all depth duration and habitat
treatments (p<0.001). The greatest effect was on biomass accumulation where there was
less above and below ground biomass in winter, after 12 weeks growth than recorded after
four weeks growth in summer (Figure 1). There was also lower species richness and seedling
abundance, however this was less pronounced. The community structure was also
significantly different between seasons, with clear separation in the n-MDS ordination of
averaged replicated data (Figure 3a). One way ANOSIM between season of all samples (not
replicate means) showed a significant seasonal effect R=0.812, p<0.001.
130 | Flow dependent ecological responses
Flow seasonality affects processes that generate and sustain riparian vegetation (Greet et al.
2011) but not on the composition of germinating community (Greet et al. 2013).
Significantly less biomass was produced after autumn compared to summer flooding
(Warwick and Brock 2003). Few studies have compared summer to winter, with most
focusing on spring to summer with an emphasis on germination and community
composition rather than biomass and growth (Greet et al. 2011). Seasonal differences in
germination from seed banks have generally been attributed to temperature as a cue for
germination, day length and light quality (Britton and Brock 1994; Warwick and Brock 2003).
Greet et al. (2013) attributed the lack of seasonal effect to the relatively constant
temperatures recorded during the growing period of the glass house between seasons. Our
study was conducted outside and temperature difference, day length and light quality
between seasons may account for differences in germination rate, species composition and
growth. Further experimentation into temperature, day length and light quality is required
to incorporate seasonal variability of growth into predictive models.
(b) Red Gum
Above ground biomass
(g/pot)
(a) Wetland
(c) Black Box
12
12
12
10
10
10
8
8
8
6
6
6
4
4
4
2
2
2
0
0
0
4
8
Duration (weeks)
0.1cm – saturated,
0
0
12 16 20
4
8
12 16 20
0
Duration (weeks)
5cm,
15 cm,
4
8
12 16 20
Duration (weeks)
30cm
Figure 1 Mean (±95% confidence intervals, n=5) above ground biomass in a) wetland, b) red gum and c) black
box habitat types in summer (filled) and winter (open) at five different depths over time. Error bars allowed
comparison of significance between depths, habitat type and duration within each season
Depth, Duration and Habitat within seasons
The main effects of depth, duration and habitat (wetland, red gum and black box) were
significant (p<0.01) for all variables in each season. There were many significant interactions
among the three factors (depth, duration and habitat) and so they could not be interpreted
in isolation (Tables 2 and 3).
In summer, above ground biomass followed a clear pattern (Figure 1), similarly found with
below ground biomass data. Across all soil types and depths above ground biomass
production peaked between 12 and 16 weeks, after which there was a significant decline
Flow dependent ecological responses | 131
(Figure 1). Flooded sites in black box habitat had the greatest peak of above ground
biomass, at 12 weeks (Figure 1c). Above ground biomass of saturated soils across all habitat
types followed a different pattern to flooded soils, peaking at 8 weeks above ground (Figure
1) and 12 weeks below ground. Above ground biomass levels were then maintained for the
duration of the experiment whereas below ground biomass significantly declined at 16
weeks and increased again by 20 weeks. Flooding to a depth of 30cm delayed production in
wetland and red gum habitat types with significantly less production at 30cm after 8 weeks
of inundation, compared to the other flooded depth treatments (Figure 1). The black box
habitat produced the greatest biomass followed by wetlands and then red gum, however
this was not consistent across all depths and durations.
In winter there was little biomass either above or below ground, but there were still
significant interactions between depth, duration and habitat types (Table 3). Saturated soils
had significantly greater above and below ground production than flooded soils in black box
and red gum habitat types but not in the wetland habitat. Black box had significantly less
above and below ground biomass than wetland and red gum for the duration of the
experiment and red gum and wetland habitats were only significantly different after 8
weeks (Figure 1).
Table 2 Summer sampling period F values, Pseudo-F values and associated significance levels for three-factor
ANOVA and PERMANOVA with duration of inundation (‘Duration’), depth of inundation (‘Depth’) and
floodplain habitat type (‘Habitat’) as factors. *=P<0.05, **=P<0.01, ***=P<0.001
DF
ABOVE
GROUND
BIOMASS
BELOW
GROUND
BIOMASS
ABUNDANCE
(COUNT)
SPECIES
RICHNESS
COMMUNITY
STRUCTURE
BELOW
GROUND AS %
OF TOTAL
Duration
4
436.5***
260.4***
10.2***
13.7***
41.5***
18.9***
Depth
3
5.6***
43.8***
69.9***
58.1***
51.7***
25.7***
Habitat
2
8.8***
11.9***
27.8***
14.9***
52.2***
6.2**
Duration x Depth
12
4.8***
4.0***
0.8
1.8*
5.8***
4.14***
Duration x Habitat
8
3.3***
2.5**
2.8**
2.2*
4.1***
6.0***
Depth x Habitat
6
2.6*
2.2*
1.9
3.7**
5.5***
1.9
Duration x Depth x
Habitat
24
1.5
2.3***
1.1
1.7*
2.0***
2.26**
Residual
240
Total
299
132 | Flow dependent ecological responses
Table 3 Winter sampling period F values, Pseudo-F values and associated significance levels for three-factor
ANOVA and PERMANOVA with duration of inundation (‘Duration’), depth of inundation (‘Depth’) and
floodplain habitat type (‘Habitat’) as factors. *=P<0.05, **=P<0.01, ***=P<0.001
DF
ABOVE
GROUND
BIOMASS
BELOW
GROUND
BIOMASS
COUNT
(ABUNDANCE)
SPECIES
RICHNESS
COMMUNITY
STRUCTURE
BELOW
GROUND AS %
OF TOTAL
Duration
2
22.1***
16.4***
115.9***
60.5***
25.2***
5.5***
Depth
2
60.8***
21.4***
65.4***
52.8***
16.0***
7.6***
Habitat
2
52.4***
61.8***
32.4***
72.0***
30.8***
12.3***
Duration x Depth
4
3.1*
3.3*
2.9*
13.5***
10.0***
5.8***
Duration x Habitat
4
2.8*
3.0*
6.5***
17.2***
8.9***
5.0***
Depth x Habitat
4
29.1***
6.2***
17.1***
14.4***
2.9***
2.7*
Duration x Depth x
Habitat
8
1.3
1.3
2.3*
1.8
3.0***
2.2*
Residual
108
Total
134
The combination of depth, duration, and habitat resulted in a complex partitioning of
production between above and below ground biomass (Tables 2 and 3, Figure 2). Overall
there was a tendency for the understory vegetation community to have a greater
proportion of biomass allocated below ground. In summer within the saturated treatment,
the black box habitat had significantly lower proportion of biomass allocated to below
ground than the red gum habitat at 12 weeks. Contrastingly, black box had significantly
greater resources allocated to below ground biomass compared to wetlands at 20 weeks.
The proportion of below ground to total biomass was relatively stable over time in saturated
and 5cm flooded treatments in summer with no significant effect of duration, within
habitats at 5cm but with some pairwise differences in saturated soils (e.g. a greater
proportion of below ground biomass in saturated, black box at week 20 compared to weeks
8, 12 and 16). In soils flooded in summer to 15cm and 30cm, there was a significant
decrease in the proportion of resources allocated to below ground biomass across habitat
types over time (week 20 significantly lower than week 4, across all habitat types when
flooded to 15 and 30cm). However the black box habitat followed a slightly different pattern
compared to red gum and wetland habitats when flooded to 15 and 30 cm. When flooded in
winter (-5cm and -30cm), the little biomass produced was in the form of roots, rather than
above ground. Water depth changes the above ground to below ground biomass allocation
in individual species (Rea and Ganf 1994) but little is known about how this might operate at
the community level or between seasons. Below ground production is important to
floodplain ecosystems, with shifts in production potentially having significant effects on
energy stores/flow which is often overlooked in production estimates (Baldwin et al. 2013).
Flow dependent ecological responses | 133
Below ground biomass as % of total
(a)
(b)
(c)
(d)
100
100
100
100
75
75
75
75
50
50
50
50
25
25
25
25
0
0
0
0
0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
Duration (weeks)
Duration (weeks)
Duration (weeks)
Duration (weeks)
Wetland,
Red Gum,
Black Box,
Figure 2 Mean (±SE, n=5) of below ground biomass as a percentage of total biomass of vegetation
communities from wetland, red gum and black box habitat types a) saturated, or flooded to b)5cm, c) 15cm
and d)30cm over time in different seasons. Note: no samples were flooded to 15cm (part C) in winter
The understory vegetation community structure was affected by depth, habitat and
duration, but like biomass these factors could not be interpreted independently due to the
significant interaction between factors (Tables 2 and 3). The n-MDS ordination clearly
showed a separation between summer and winter communities (Figure 3a). Within each
season, there was a shift in community structure in response to the duration of inundation
(Figure 3b). In winter this was significant for all habitat by depth combinations with the
exceptions of black box habitat flooded to 30 cm where no effect of duration was observed
and in the wetland habitat type where week 4 and 12 were not significantly different. In
summer, the community structure at 4 weeks was different to any other duration across all
habitat types and flood depths. After 4 weeks of flooding, the factors of habitat and depth
were significant (Table 4). In the red gum habitat after 8 weeks, duration was only
significant when flooding occurred to 30cm depth. In the black box habitat duration was
significant for the entire experiment when saturated and only weeks 16 and 20 were not
significantly different in flooded treatments (5cm, 15cm and 30cm). In the wetland habitat,
duration was significant with most pairwise comparisons with a few exceptions (Table 4). In
wetland at 30cm although there are a number of pairwise comparisons that are not
significant, week 12 is different to week 20 indicating there was still a shift in community
structure at 30cm with the increased duration.
The n-MDS showed a clear trajectory of change in community structure with depth in
summer, which was not as clear in winter (Figure 3c). The pairwise comparison of the three
way interaction showed that, across all habitat types, all depths were significantly different
after 12 weeks of inundation in summer and winter. However, this was not consistent
134 | Flow dependent ecological responses
across other habitat type and duration combinations. In summer, saturated soils produced
a different community structure to flooded soil on every sampling occasion, across all
habitat types (Figure 3c). On every sampling occasion, saturated soils produced a different
community structure to flooded soil across all habitat types. Also across all habitats there
were sampling occasions where individual depths were not significantly different. This was
generally on a trajectory: where for example the community structure found in 5cm
treatments was not different to that in 15cm but the community in 5cm and 15cm were
different to that in 30cm.
The effect of habitat on community structure was less clearly defined in the n-MDS
ordination (Figure 3d), however in both seasons, across all sampling weeks and inundation
depths, the black box habitat type had a significantly different community structure,
compared to the wetland and red gum habitats. There was some overlap in the community
structure of red gum and wetland habitat types but this was dependant on season, depth
and duration. Wetland and red gum habitat types were significantly different to each other
on all occasions with some exceptions (Winter, after 4 weeks at 5cm and 15cm, Summer – 8
weeks at 30cm, 12 weeks at 5cm and 20 weeks at 30cm).
Table 4 Effect of duration on community structure. Pairwise significance of the tree way interaction of
community structure PERMANOVA within each season, depth and habitat combination. All pairwise
comparisons are significantly different to the p<0.05 except where indicated being equal (=) Depth
treatments – winter (4, 8 and 12 weeks) – summer (4, 8, 12, 16 and 20 weeks)
Season
Depth
Black Box
Red Gum
Wetland
Winter
0.1
All different
All different
All different
5
4=12
All different
All different
30
4=8=12
All different
4=12
0.1
All different
12=16=20
12=20
5
16=20
8=12, 12=16=20
16 =20
15
16=20
8=12, 12=20, 16=20
8=12, 12=16=20
30
16=20
All different
12=16, 16=20
Summer
(a)
(b)
Flow dependent ecological responses | 135
Stress 0.12
Summer
Stress 0.12
Winter
Summer
Winter
4
4
20
16
12
820
20
1620 121616
8812
12
20
812
12
16
16
20
8 8
16201620
16
12
12
8 8
12 20
20 20
16
12 8
16 16
20 12 8 8
12
8
4 4
4
4 4 4
4
4 444 4
4
8
12
48
44 4
4
4 8
8
12
12 8
12
88 8
8
12
12
Duration increase
(d)
Stress 0.12
Winter
Winter
Depth increase
Summer
Summer
Black Box
Stress 0.12
Red Gum & Wetland
(c)
12
Duration increase
Figure 3 n-MDS ordination of averaged treatment replicates (n=5) square root transformed above ground
biomass, showing separation of the four factors analysed (a) season – symbols filled summer open winter,
(b) flood/growth period numbers = weeks, (c) depth colour – 0.1cm saturated , 5cm , 15cm , 30cm ,
(d) habitat type symbol shapes –black box circle, red gum triangle, wetland square
CONCLUSIONS
The results of the experiments suggest that diversity and productivity responses to
inundation were influenced by interactions between habitat, depth, duration and timing of
inundation. These complex interactions made quantification of the influence of individual
factors difficult, however, in summer, biomass increased over the first three to four months
and then declined. The species richness patterns were less pronounced and more variable,
although the multivariate analysis suggested overlap between wetland and red gum
communities with black box representing a different community.
One of the strongest influences was flow timing that influenced above ground and below
ground biomass, abundance, species richness and community structure. Flow timing also
influenced the influence of other factors reducing or eliminating the effects of depth
changing the influence of duration. Further work is required to identify the specific causal
factors (temperature, day length) that drive these differences. This finding is of particular
136 | Flow dependent ecological responses
significance for the management of environmental flows as the timing of environmental
flows can be controlled by managers.
Perhaps the most striking result was the complexity of vegetation community responses to
environmental flows with significant interactions between depth, habitat type and duration
for all vegetation communities. This complexity was manifest even though mesocosm
experiments represent a simplified floodplain ecosystem. Our study eliminated many factors
known to be important, including antecedent conditions, drying periods between floods
(Smith and Brock 2007), rate of rise and fall, flood frequency (Casanova and Brock 2000),
level fluctuations (Deegan et al.2007; Smith and Brock 2007), nutrient availability (Deegan et
al. 2012), transition zones between habitats, grazing (Blanch and Brock 1994; Crossle and
Brock 2002), competition and changes in climatic zone. We also did not include biomass
production from understory species that seldom germinate from seed banks (e.g. tangled
lignum (Muehlenbeckia florulenta) (Chong and Walker 2005), species like Eleocharis which
produce ribosomes or growth of plants which are already established at the time of an
inundation event. It therefore appears likely that the situation is more complex in natural
environments
The complex interactions between depth, habitat type and duration meant that we were
not able to develop statistical relationships between these factors and either rate of
production or community responses. The objective of the project had been to develop these
statistical relationships in order develop models capable of predicting productivity
responses to environmental flows. As a consequence, the observed complexity does have
implications for the development of predictive models of the response of systems to
individual flows.
Floodplain ecosystems have long been recognised as a dynamic mosaic of habitat patches
over multi-year time-scales (Junk et al. 1989; Ward and Stanford 1995). The results of this
experiment suggest that from a productivity perspective, the most productive places on a
floodplain are likely to vary between inundations events and over a single flow event. This
implies that floodplains and wetlands are also dynamic mosaics of habitat and production
over the course of individual inundation events. This suggests that developing the capacity
to predict vegetation responses to environmental flows is unlikely to emerge from simple
experiments at the scale of individual wetlands, but that flow is clearly a strong driver in
these systems.
REFERENCES
Anderson MJ, Gorley RN and Clarke KR (2008) 'PERMANOVA+ for PRIMER: Guide to
Software and Statistical Methods.' (PRIMER-E Ltd: Plymouth, UK)
Flow dependent ecological responses | 137
Baldwin DS, Rees GN, Wilson JS, Colloff MJ, Whitworth KL, Pitman TL and Wallace TA (2013)
Provisioning of bioavailable carbon between the wet and dry phases in a semi-arid
floodplain. Oecologia 172, 539-550.
Blanch SJ and Brock MA (1994) Effects of grazing and depth on 2 wetland plant-species.
Australian Journal of Marine and Freshwater Research 45, 1387-1394.
Bragg OM, Black AR, Duck RW and Rowan JS (2005) Approaching the physical-biological
interface in rivers: a review of methods for ecological evaluation of flow regimes.
Progress in Physical Geography 29, 506-531.
Britton DL and Brock MA (1994) Seasonal germination from wetland seed banks. Australian
Journal of Marine and Freshwater Research 45, 1445-1457.
Capon SJ, James CS, Mackay SJ and Bunn SE (2009) Environmental watering for understorey
and aquatic vegetation in The Living Murray Icon Sites. A literature review and
identification of research priorities relevant to the environmental watering actions of
flow enhancement and retaining floodwater on floodplains. Murray Darling Basin
Authority, MDBA Publication No. 10/12, Canberra.
Carpenter SR and Lodge DM (1986) Effects of submerged macrophytes on ecosystem
processes. . Aquatic Botany 26, 341-370.
Casanova MT and Brock MA (2000) How do depth, duration and frequency of flooding
influence the establishment of wetland plant communities? Plant Ecology 147, 237250.
Chong C and Walker KF (2005) Does lignum rely on a soil seed bank? Germination and
reproductive phenology of Muehlenbeckia florulenta (Polygonaceae). Australian
Journal of Botany 53, 407-415.
Clarke KR and Warwick RM (2001) Changes in marine communities: an approach to
statistical analysis interpretation 2nd edn, Primer-E Ltd: Plymouth, UK.
Crossle K and Brock MA (2002) How do water regime and clipping influence wetland plant
establishment from seed banks and subsequent reproduction? Aquatic Botany 74, 4356.
Cunningham GM, Mulham WE, Milthorpe PL and Leigh JH (2006) 'Plants of Western New
South Wales.' (Elsevier Australia: Marrickville, NSW)
Cunningham SC, Read J, Baker PJ and Mac Nally R (2007) Quantitative assessment of stand
condition and its relationship to physiological stress in stands of Eucalyptus
camaldulensis (Myrtaceae). Australian Journal of Botany 55, 692-699.
Deegan BM, White SD and Ganf GG (2007) The influence of water level fluctuations on the
growth of four emergent macrophyte species. Aquatic Botany 86, 309-315.
Deegan BM, White SD and Ganf GG (2012) Nutrients and water level fluctuations: A study of
three aquatic plants. River Research and Applications 28, 359-368.
Franklin P, Dunbar M and Whitehead P (2008) Flow controls on lowland river macrophytes:
A review. Science of the Total Environment 400, 369-378.
138 | Flow dependent ecological responses
Gawne B, Merrick C, Williams DG, Rees G, Oliver R, Bowen PM, Treadwell S, Beattie G, Ellis I,
Frankenberg J and Lorenz Z (2007) Patterns of primary and heterotrophic productivity
in an arid lowland river. River Research and Applications 23, 1070-1087.
Greet J, Cousens RD and Webb JA (2013) Seasonal timing of inundation affects riparian plant
growth and flowering: implications for riparian vegetation composition. Plant Ecology
214, 87-101.
Greet J, Webb JA and Cousens RD (2011) The importance of seasonal flow timing for riparian
vegetation dynamics: a systematic review using causal criteria analysis. Freshwater
Biology 56, 1231-1247.
Junk WJ, Bayley PB and Sparks RE (1989) The flood pulse concept in river-floodplain
systems. Canadian Special Publication of Fisheries and Aquatic Sciences 106, 110-127.
Nielsen DL, Brock MA, Petrie R and Crossle K (2007) The impact of salinity pulses on the
emergence of plant and zooplankton from wetland seed and egg banks. Freshwater
Biology 52, 784-795.
Rea N and Ganf G (1994) Water depth changes and biomass allocation in two contrasting
macrophytes. Marine and Freshwater Research 45, 1459-1468.
Sainty GR and Jacobs SWL (1981) Waterplants of New South Wales. Water Resources
Commission, NSW.
Sainty GR, Jacobs SWL (2003) Waterplants in Australia. Sainty and Associates Pty Ltd, Potts
Point NSW.
Sims NC and Colloff MJ (2012) Remote sensing of vegetation responses to flooding of a
semi-arid floodplain: Implications for monitoring ecological effects of environmental
flows. Ecological Indicators 18, 387-391.
Sims NC and Thoms MC (2002) What happens when floodplains wet themselves: vegetation
response to inundation on the Lower Balonne Floodplain. In Dyer F, Thoms MC and
Olley JM (Eds) Proceedings of the IAHS Symposium on the Structure, Function and
Management Implications of Fluvial Sedimentary Systems.' Alice Springs. International
Association of Hydrological Sciences, Wallingford.
Smith RGB and Brock MA (2007) The ups and downs of life on the edge: The influence of
water level fluctuations on biomass allocation in two contrasting aquatic plants. Plant
Ecology 188, 103-116.
The Royal Botanic Gardens and Domain Trust - 'PlantNET’ The Plant Information Network
System of The Royal Botanic Gardens and Domain Trust,.' Viewed 27 May 2012
<http://plantnet.rbgsyd.nsw.gov.au>
Ward JV and Stanford JA (1995) The serial discontinuity concept: extending the model to
floodplain rivers. Regulated Rivers 10, 159-168.
Warwick NWM and Brock MA (2003) Plant reproduction in temporary wetlands: the effects
of seasonal timing, depth, and duration of flooding. Aquatic Botany 77, 153-167.
Webb JA, Wallis EM and Stewardson MJ (2012) A systematic review of published evidence
linking wetland plants to water regime components. Aquatic Botany 103, 1-14.
Flow dependent ecological responses | 139
2.13
Blackbox (Eucalyptus largiflorens) vegetation and bird
communities in the Lowbidgee Floodplain – Flood regimes
driving community transitions in semi-arid floodplain
woodlands
H. M. McGinness1, V.A.J. Doerr1, A. D. Arthur1,2, M. Davies1
1
CSIRO Ecosystem Sciences and Water for a Healthy Country Flagship, GPO Box 1700 Canberra City
ACT Australia 2601;
2
Australian Bureau of Agricultural and Resource Economics and Sciences, Department of Agriculture,
Fisheries and Forestry, GPO Box 1563 Canberra ACT 2601. Australia (current address)
INTRODUCTION
Water resource development in south-eastern Australia has seriously reduced the
frequency, extent and duration of inundation of floodplain woodlands, resulting in
significant habitat change and loss of productivity (Kingsford, 2000, Horner et al., 2009,
Alexander et al., 2008, Bren and Gibbs, 1986, Casanova and Brock, 2000). The effects of
such changes on floodplain fauna are poorly understood. Woodland birds may be useful
indicators of the effects of changing flood regimes on floodplain fauna because they are
relatively numerous, diverse, easily surveyed, and responsive to change translated through
the entire food web (McGinness et al., 2010). Flooding may affect birds through local and
landscape-scale effects on primary productivity and food resources, habitat quality and
configuration, or open water resources for drinking and nest construction. Differences in
site factors can also influence woodland bird abundance through other mechanisms, such as
competitive or invasive species interactions (Clarke and Oldland, 2007, Mac Nally et al.,
2012, Maron, 2007). Different guilds or species groups may respond differently to spatial
differences or temporal changes in flooding, and listed vulnerable or endangered species
may also be sensitive to changes in water availability and its direct and indirect effects. Over
time, the influence of differences in site flood history upon habitat, food and water
availability is likely to be reflected most strongly by resident, sedentary species at a site,
because these are the species most limited by the availability of local resources.
Terrestrial woodland birds have been in decline across Australia (Barrett et al., 2003, Ford et
al., 2001, Olsen, 2008). In recent decades, local extinctions have become increasingly
widespread, including previously common species (Ford et al., 2009, Mac Nally et al., 2009,
Olsen, 2008). Generally, species that are resident, insectivorous, small-bodied and/or
ground-foraging are most at risk (Reid, 1999). McGinness et al. (2010) suggested that
floodplain woodlands and forests may be important refuges for woodland bird populations
because a) floodplain woodlands and forests comprise some of the largest and most
continuous vegetation remnants in south-east Australia; and b) floods intermittently supply
water, sediment and nutrients that drive greater primary and secondary productivity than
140 | Flow dependent ecological responses
found in woodlands not subject to flooding. They suggested that declining woodland bird
species may be disproportionately affected by reduced productivity following changes in
flood regime. Similarly, reductions in ground layer vegetation structure and primary
production that flowed through the food web are considered key drivers of observed bird
population declines (Reid and Fleming (1992). Hence, the resources provided by flooding
may be essential for persistence of declining woodland birds.
This research uses new data to explore the hypothesis that changes in flood regime drive
transitions in vegetation structure and condition that in turn create dramatic shifts in
woodland bird communities. Beginning with a set of conceptual models based on the links
between flooding (frequency and single-event), resource availability (habitat and food) and
woodland bird species presence, density and breeding success (Appendix 1), it examines
the abundance of selected bird species and groups in remnant floodplain woodlands of a
major floodplain and tests the following three predictions:
•
•
•
vegetation condition will improve and structural complexity will increase with
increasing flood frequency and with decreasing time since a site was last wet;
terrestrial bird abundance will increase with improving vegetation condition and
increasing structural complexity and;
terrestrial bird abundance will increase with increasing flood frequency and with
decreasing time since a site was last wet.
METHODS
Remnant black box (Eucalyptus largiflorens) woodlands in the lower Murrumbidgee river
floodplain (Lowbidgee) in south-east Australia were surveyed in 2010 and 2011. Black box
woodland communities in good condition are generally flooded every 2-10 years (George et
al., 2005, Johns et al., 2009, Roberts and Marston, 2000, Taylor et al., 1996). Using a spacefor-time substitution design, sites were selected across a gradient of water history
characteristics, from rarely inundated or experiencing severe drought, to sites regularly
inundated. Sites were also selected for contrasts between recently inundated and not
recently inundated (time since last wet). Multiple inundation history, surface water
proximity and landscape vegetation configuration variables were derived for each site from
information provided by land managers, LANDSAT and SPOT imagery processed within
CSIRO, and NSW OEH GIS mapping (Table 1), all of which were tested during modelling. All
sites were in vegetation remnants greater than 10 ha in size, contained hollow-bearing
trees, and were of similar topography and elevation.
Two intensive survey sessions were conducted in each of the years 2010 and 2011: Early
spring (August-September) and spring-summer (November-December). These are in
addition to pilot surveys at a subset of sites conducted in September and November 2009,
Flow dependent ecological responses | 141
the data from which are being used to examine temporal changes over the 2009-2010-2011
period (from record drought to record wet years) separately.
Four transects 500m in length and 200m apart were surveyed at each site, covering
approximately 56 ha at each site. Bird surveys comprised about 20 minutes walking
observation per 500 m, recording all birds seen and heard and the transect location and
distance and angle from the transect of each bird seen. The same two observers conducted
each survey at each site, with each observer covering two transects. Each site was visited
twice per survey trip: once in the morning, and once in the evening. Morning and evening
data were added together for analysis. Thirty bird species were targeted in order to reduce
observer bias. Selected species were predominantly sedentary or resident, considered to
have strong reliance on the condition of the sites and their surrounds and to be reliable
indicators of factors driving local population dynamics. The species selected ranged from
common species to some that are considered vulnerable, threatened or declining, spanning
insectivorous, nectivorous, granivorous and omnivorous groups. Analyses of bird responses
to flood history and vegetation parameters were conducted first for all surveyed bird
species combined, followed by analyses of various combinations of species in groups, e.g.
social/non-social, granivorous/non-granivorous, patch-specialist/generalist, and
excluding/including competitive miner species (noisy miner Manorina melanocephala and
yellow-throated miner Manorina flavigula). Finally, analyses were conducted for selected
individual species representing competitive and vulnerable groups, together with potential
interactions between species. In this report, we focus on trends in abundance identified for
the main bird groups tested in response to water and vegetation-related factors.
Photographic vegetation assessment was used to improve the efficiency and
representativeness of data collection and reduce field-time. At each site, three standardised
digital photographs were taken at 24 evenly distributed points, aligned with the bird survey
transects: 1) a full-frame photo of the nearest unobscured entire single adult or mature
tree, from base to top and full width, portrait or landscape view; 2) a landscape view
community photograph with the horizon line at the middle of the image; and 3) a
groundcover photograph taken at approximately 1.5 m perpendicular from the ground,
using forced flash, no zoom, and moving away from shaded areas. Images were assessed in
detail on-screen in the laboratory. The main variables recorded for each of the three photo
types together with derivation of scores and indices are described in Table 2 and in
McGinness et al. (2013).
Single-factor and multi-factor generalised linear models were fitted according to the data
distribution (Poisson or negative binomial), and models compared using Akaike information
criterion analysis for small sample sizes (AICc and Akaike weights). The AICc and Akaike
weights measure the relative goodness of fit of individual models, and are used to compare
relative support for the different models, with better models having lower AICc numbers.
142 | Flow dependent ecological responses
Generally, models with an AICc within three points of the best supported model are also
well supported, except where the null model is the better model (Burnham and Anderson,
2002). Zero inflated models were used when required (Zeileis et al., 2008). Models were
fitted using water-related explanatory variables for vegetation responses, and both waterrelated and vegetation-related explanatory variables for bird responses. A null model was
also fitted for each combination where response variables were not related to any factors
(i.e. the mean of the sites). Counts of live, dead, and total numbers of trees and shrubs >1m
tall were log transformed, and groundcover proportions were arcsine transformed. Rather
than presenting AICc results for all models explored, statistical results were presented for
strongly supported models in terms of likelihood ratio (Chi squared) tests between models
with and without the term of interest. Model-predicted values were plotted for the bestsupported models (as ranked by the AICc) and inferences made based on the size of the
effects. All analyses were carried out in R, using the packages bbmle, nlme, pscl, and lmtest
(Jackman, 2011, Pinheiro et al., 2013, Zeileis and Hothorn, 2002).
RESULTS
Models driven by site inundation history factors were consistently stronger than all other
models, for both vegetation and bird response variables. Site inundation history was
ultimately best represented in models by two primary categorical factors: 1)
LowestARI5pccat: The lowest average recurrence interval (ARI) zone recorded that occupied
more than 5% of the site, representing the frequency with which at least part of the site has
received water (even if that water does not inundate the entire site); and 2) TSWW20pc:
The time since the site was last wetted over more than 20% of its area. The
LowestARI5pccat (ARI) categories used were one in <3, 4 to 9, 10 to 19, and >20 years; and
the time since wet categories used were 0-2 years, 2-10 years, and >20 years. An ARI of <3
years means that a site has been flooded at least every 3 years on average, if not more often
(frequent inundation), whereas an ARI of >20 years means that a site has been flooded less
than once every 20 years on average (infrequent inundation).
Vegetation responses to flooding: shifts in condition and structure
Both vegetation condition and vegetation structure in the black box floodplain woodlands
surveyed were clearly influenced by flood history and sometimes also influenced by the
time since a site was last flooded. In general, the best supported models of tree condition
(including crown density, crown death and flowering) were those driven by
LowestARI5pccat or an additive interaction between LowestARI5pccat and TSWW20pc.
Variables affecting habitat structure such as large shrub abundance and subshrub
abundance were also strongly predicted by additive interactions between LowestARI5pccat
and TSWW20pc. In contrast, the best supported models of groundcover variables were
those driven by the time since a site was last flooded (TSWW20pc).
Flow dependent ecological responses | 143
Lower flood frequencies at a site were associated with significantly lower average tree
health (including decreased crown density and increased crown death) and reduced tree
flowering. There were significantly fewer large shrubs where flooding occurred less than
once every 20 years, together with a lower proportion of young trees present
(demonstrating reduced regeneration and recruitment and lower vegetation structural
complexity), increased abundance of dryland subshrubs, and increased numbers of dead
trees.
Vegetation patterns in response to the time since a site was last wet were more complex
and variable than those observed in response to flood frequency. Average tree health and
tree flowering were significantly lower at sites that had not been wet for >20 years than
other sites (including significantly reduced crown density and significantly increased crown
death). Site vegetation complexity/structure was significantly greater at sites that had last
been flooded between 2-10 years before, compared to other sites, partly driven by greater
proportions of young and old trees at these sites. The number of live trees was also greater
at these sites than at others. Groundcover was highly variable, but plant cover was reduced
at sites that had been recently flooded (within two years), with relatively greater
proportions of bare ground and organic litter and a significantly lower proportion of live
plants present than other sites. Live plant cover was significantly greater at sites that had
been flooded 2-10 years before compared to other sites.
Bird responses to vegetation and flooding: shifts in relative abundance and community
composition
Woodland bird abundance in floodplain woodlands was significantly affected by flood
history; however response patterns varied among different bird groups. Models of bird
abundance directly related to flood history significantly outperformed models related to
vegetation or abundance of competitive species such as miners. The strongest predictor of
bird abundance was the lowest ARI recorded that occupied more than 5% of the site
(LowestARI5pccat). In general, variables describing broad flood regime were better
predictors of bird abundance than short-term variables such as the time since a site was last
wet (TSWW20pc), however the latter was still significant and interacted with flood
frequency (LowestARI5pccat) for some bird groups/guilds. The strongest vegetation
predictors of bird abundance, over all analyses, were average site tree health and site
vegetation complexity.
Low flood frequencies at a site were associated with significantly lower abundances of nonsocial bird species and patch specialist species (with granivores excluded), non-granivorous
species, and small-bodied bird species than in other sites (e.g. Figure 1). There appeared to
be a threshold where an ARI of less than one flood event in 10 years resulted in significantly
reduced abundances in these bird groups at a site, whereas bird abundances in these groups
at sites with ARIs of more than one event in 10 years were generally similar. The reverse
144 | Flow dependent ecological responses
120
50
Abundance of granivorous species
Abundance of non granivorous non
social species
was true for granivorous species, with significantly greater abundances of birds in this group
at sites with an ARI of less than one flood event in 10 years. No significant effect of flood
frequency was apparent for social species (excluding miners), but the data trended toward
increasing abundances with increasing ARI.
45
40
35
30
25
20
15
10
5
0
100
80
60
40
20
0
<3
4 to 9
10 to 19
>20
Lowest average flood recurrence interval (years)
(LowestARI5pccat)
<3
4 to 9
10 to 19
>20
Lowest average flood recurrence interval (years)
(LowestARI5pccat)
Figure 1. Lower flood frequency at a site was associated with lower abundances of non-social bird species
(with granivores excluded) and higher abundances of granivorous bird species. LowestARI5pccat is the
lowest average recurrence interval (ARI) category recorded that occupied more than 5% of the site,
representing the frequency with which at least part of the site has received water.
Bird abundance patterns in response to the time since a site was last wet were more
variable than those observed in response to flood frequency. Increasing time since a site
was last wet (particularly an interval of >20 years) was associated with significantly lower
abundances of non-social, patch specialist and small-bodied bird species, when granivores
were excluded (Figure 2). Most other bird groups were more abundant at sites that had
been flooded within 2-10 years, however variability was much greater in this category.
Overall, bird abundances were lower in sites that had not been flooded for more than 20
years, compared with sites that had been flooded within 10 years, except for social and
granivorous species groups.
Increasing average tree health and increasing site vegetation complexity at a site were both
associated with significantly increasing abundance of non-social bird species, patch
specialist species, non-granivorous species, and small-bodied bird species. In contrast,
greater tree health and vegetation complexity were associated with significantly lower
abundances of granivorous species and social species.
Flow dependent ecological responses | 145
160
40
140
Abundance of native species
(excluding miners)
Abundance of non granivorous non
social species
45
35
30
25
20
15
10
5
0
120
100
80
60
40
20
0
0-2 years
2-10 years
Time since wet
>20 years
0-2 years
2-10 years
>20 years
Time since wet
Figure 2. Sites that had not been flooded for more than 20 years were associated with lower abundances of
non-social bird species (with granivores excluded). Sites that were wet within 2-10 years prior to survey
generally had higher abundances of other bird groups.
DISCUSSION
These results suggest that black box floodplain woodlands that do not receive sufficient
flooding undergo transitions in vegetation type or structure, rather than degradation in
condition alone, and that this in turn affects woodland bird relative abundance and
community composition. Black box sites receiving insufficient flooding appear to transition
toward a more terrestrial, dryland state, with associated increases in relative abundance of
dryland understorey species (e.g. subshrubs and grasses). With continued declines in tree
condition the black box tree layer will probably eventually be lost entirely in the long term
at these sites if a suitable flood regime is not reinstated. In the interim, vegetation condition
(and potentially productivity) is reduced at both tree and understorey levels. Woodland bird
community composition has also shifted at these sites – with greater abundance of more
mobile granivorous and social or co-operative species at dry sites, and fewer small-bodied,
non-social residents. Together, these data indicate that reduced flooding has changed
resource availability, in terms of both habitat as structure changes and food type and
abundance as productivity declines. These changes are culminating in significant shifts in
vegetation and fauna community composition and will probably eventually result in an
entirely different community type. Further analysis of species richness and community
composition data across sites and over time will explore the nature of these shifts.
Such changes affect fauna populations and diversity at a site through multiple related
mechanisms. Populations of species that require sites with greater flood frequency and
associated productivity and habitat structure are more likely to decline and eventually
become locally extinct. Increased competition for resources and changes in habitat
146 | Flow dependent ecological responses
structure such as increased openness and edge availability may advantage highly
competitive or co-operative / social species. This may maintain or increase their population
sizes while less effective competitors are reduced in abundance or eliminated. Co-operative
social breeding may increase the chances of a population surviving at a site during dry years,
and allow that population to take rapid advantage of boom conditions during high rainfall or
flood years, quickly increasing population size and spreading to new areas, even where
habitat structure and quality are poor. In contrast, small bodied non-social/non-cooperative species may be more susceptible to competitive and predatory attacks than large
bodied social/co-operative species, and therefore their presence or population sizes may be
disproportionally affected by such changes.
Alterations in flood regimes and associated changes in structure, productivity and condition
of floodplain vegetation have probably contributed to observed declines in woodland bird
populations in Australian drylands (McGinness et al., 2010). Declining bird species in these
regions tend to be small-bodied, resident (or less mobile), insectivorous, and/or groundforagers (Reid, 1999) – similar groups to those identified in this study as occurring in
reduced abundance at sites with low flood frequency and reduced tree health and structural
complexity. In south-east Australia, many of the most dramatic changes in flood regimes
have occurred over the same period and across the same geographical areas in which
significant declines in bird populations have been observed (Barrett et al., 2003, Mac Nally
et al., 2009, Major et al., 2001, Reid, 1999). Woodland bird population declines have
continued in these areas despite a recent general cessation of clearing and increasingly
sensitive management of vegetation, prompting suggestions that continuing degradation of
habitat (in particular food resources) and extinction debts are important drivers (Ford et al.,
2009, Mac Nally et al., 2009).
This study demonstrates that changes in flooding frequency result in shifts in site character
and ultimately transitions to different vegetation and fauna communities, even within the
same broad vegetation type. Importantly, these shifts have a compounding effect resulting
from interactions between vegetation changes, fauna habitat preferences and fauna
behaviour (e.g. competition). Such transitions encompass changes in structure, composition,
productivity and diversity of a community or site; with potentially far-reaching implications
for ecosystem function at multiple scales and for the way a site is valued and managed.
ACKNOWLEDGEMENTS
This research was funded by the CSIRO Water for a Healthy Country Flagship. It would not
have been possible without the generous support of the Yanga National Park NSW staff, the
NSW Office of Environment and Heritage and the kind landholders on whose properties we
worked. We gratefully acknowledge Murray Darling Freshwater Research Centre staff
Flow dependent ecological responses | 147
including Darren Baldwin and Jessica Wilson for introducing us to the study area and sharing
information, and Nick Nicholls (CSIRO) for statistical advice.
REFERENCES
Alexander P, Nielsen DL and Nias D (2008). Response of wetland plant communities to
inundation within floodplain landscapes. Ecological Management and Restoration
9(3): 187-195.
Barrett GW, Silcocks A, Barry S, Cunningham R and Poulter R (2003). The new atlas of
Australian birds. Melbourne, RAOU.
Bren LJ and Gibbs NL (1986). Relationships between flood frequency, vegetation and
topography in a river red gum forest. Australian Forest Research 16: 357-370.
Brock MA, Nielsen DL and Crossle K (2005). Changes in biotic communities developing from
freshwater wetland sediments under experimental salinity and water regimes.
Freshwater Biology 50(8): 1376-1390.
Brock MA, Nielsen DL, Shiel RJ, Green JD and Langley JD (2003). Drought and aquatic
community resilience: the role of eggs and seeds in sediments of temporary wetlands.
Freshwater Biology 48(7): 1207-1218.
Burnham KP and Anderson DR (2002). Model selection and multimodel inference: a practical
information theoretic approach. New York, Springer Science + Business Media, LLC.
Casanova MT and Brock MA (2000). How do depth, duration and frequency of flooding
influence the establishment of wetland plant communities? Plant Ecology 147: 237250.
Clarke MF and Oldland JM (2007). Penetration of remnant edges by noisy miners (Manorina
melanocephala) and implications for habitat restoration. Wildlife Research 34(4): 253261.
Ford HA, Barrett GW, Saunders DA and Recher HF (2001). Why have birds in the woodlands
of Southern Australia declined? Biological Conservation 97(1): 71-88.
Ford HA, Walters JR, Cooper CB, Debus SJS and Doerr VAJ (2009). Extinction debt or habitat
change? - Ongoing losses of woodland birds in north-eastern New South Wales,
Australia. Biological Conservation 142: 3182-3190.
George AJ, Walker KF and Lewis MM (2005). Population status of eucalypt trees on the River
Murray floodplain, South Australia. River Research and Applications 21(2-3): 271-282.
Horner GJ, Baker PJ, Mac Nally R, Cunningham SC, Thomson JR and Hamilton F (2009).
Mortality of developing floodplain forests subjected to a drying climate and water
extraction. Global Change Biology 15: 2176 - 2186.
Jackman S (2011). pscl: Classes and Methods for R Developed in the Political Science
Computational Laboratory, Stanford University. Department of Political Science,
Stanford University. Stanford, California.
Johns C, Reid CJ, Roberts J, Sims N, Doody TM, Overton IC, McGinness HM, Rogers K,
Campbell C and Gawne B (2009). Native trees of the River Murray floodplain:
148 | Flow dependent ecological responses
Literature review and experimental designs to examine effects of flow enhancement
and floodwater retention. Final report to the Murray-Darling Basin Authority.
Wodonga, Murray-Darling Freshwater Research Centre.
Kingsford RT (2000). Ecological impacts of dams, water diversions and river management on
floodplain wetlands in Australia. Austral Ecology 25(2): 109-127.
Mac Nally R, Bennett AF, Thomson JR, Radford JQ, Unmack G, Horrocks G and Vesk PA
(2009). Collapse of an avifauna: climate change appears to exacerbate habitat loss and
degradation. Diversity and Distributions 15(4): 720-730.
Mac Nally R, Bowen M, Howes A, McAlpine CA and Maron M (2012). Despotic, high-impact
species and the subcontinental scale control of avian assemblage structure. Ecology
93(3): 668-678.
Major RE, Christie FJ and Gowing G (2001). Influence of remnant and landscape attributes
on Australian woodland bird communities. Biological Conservation 102(1): 47-66.
Maron M (2007). Threshold effect of eucalypt density on an aggressive avian competitor.
Biological Conservation 136(1): 100-107.
McGinness HM, Arthur AD, Davies M and McIntyre S (2013). Floodplain woodland structure
and condition: the relative influence of flood history and surrounding irrigation land
use intensity in contrasting regions of a dryland river. Ecohydrology 6: 201-213.
McGinness HM, Arthur AD and Reid JRW (2010). Woodland bird declines in the MurrayDarling Basin: are there links with floodplain change? The Rangeland Journal 32: 315327.
Neilsen DL and Chick AJ (1997). Flood-mediated changes in aquatic macrophyte community
structure. Marine Freshwater Resources 48: 153-157.
Olsen P (2008). The State of Australia's Birds 2008. Supplement to Wingspan, Vol. 18, no. 4.,
Birds Australia: 21.
Pinheiro J, Bates D, DebRoy S, Sarkar D and R Development Core Team (2013). nlme: Linear
and Nonlinear Mixed Effects Models.
Reid J and Fleming M (1992). The conservation status of birds in arid Australia. The
Rangeland Journal 14(2): 65-91.
Reid JRW (1999). Threatened and declining birds in the New South Wales sheep-wheat belt:
1. Diagnosis, characteristics and management. Consultancy report to NSW National
Parks and Wildlife Service. Canberra, ACT, Australia, CSIRO Wildlife and Ecology: 69.
Roberts J and Marston F (2000). Water regime of wetland and floodplain plants in the
Murray-Darling Basin: A source book of ecological knowledge. Canberra, CSIRO Land
and Water.
Taylor PJ, Walker GR, Hodgson G, Hatton TJ and Correll RL (1996). Testing of a GIS model of
Eucalyptus largiflorens health on a semiarid, saline floodplain. Environmental
Management 20(4): 553-564.
Zeileis A and Hothorn T (2002). Diagnostic checking in regression relationships. R News 2(3):
7-10.
Flow dependent ecological responses | 149
Zeileis A, Kleiber C and Jackman S (2008). Regression models for count data in R. Journal of
Statistical Software 27: 1-25.
Table 1 Water-related variables used in analyses
Flood frequency
"ARI_0p5"
"ARI_1p5"
"ARI>20"
"ARI1"
"ARI10"
"ARI15"
"ARI2"
"ARI20"
"ARI3"
"ARI4"
"ARI5"
"GARI_0p5to5"
"GARI10to20"
Description
Percentage of the site inundated at an average recurrence interval of 2 floods per year
Percentage of the site inundated at an average recurrence interval of 1.5 floods per year
Percentage of the site inundated at an average recurrence interval of <1 flood every 20 years
Percentage of the site inundated at an average recurrence interval of 1 flood each year
Percentage of the site inundated at an average recurrence interval of 1 flood every 10 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 15 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 2 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 20 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 3 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 4 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 5 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 0.5 to 5 years
Percentage of the site inundated at an average recurrence interval of 1 flood every 10 to 20 years
The lowest average recurrence interval (ARI) recorded that occupied more than 5% of the site, representing the
"LowestARI5pc"
frequency with which at least part of the site has received water (even if that water does not inundate the entire site)
The lowest average recurrence interval (ARI) category recorded that occupied more than 5% of the site, representing
"LowestARI5pccat
the frequency with which at least part of the site has received water (even if that water does not inundate the entire
"
site)
"ARICAT2"
A two-category version of LowestARI5pccat
"MinARI"
The lowest average recurrence interval at the site
"MaxARI"
The highest average recurrence interval zone recorded at the site
"ARIrange"
The range of average recurrence intervals at the site (categorical groups)
"Ctwaterpres_200
The number of times water was recorded as present at a site between the years 2006 and 2011
6_2011"
"Ctwaterpres_3yr The number of times water was recorded as present within the site during the months September and November over
NS"
the preceding three years
"Ctwaterpres_3yr The number of times water was recorded as present at a site during the same month as the survey over the preceding
sm"
three years
"Ctwaterpres20pc
The number of times water was recorded as occupying >20% of the site between the years 2006 and 2011
_2006_2011"
"Ctwaterpres20pc The number of times water was recorded as occuping >20% of the area of the site during the months September and
_3yrNS"
November over the preceding three years
"DomARI"
The average recurrence interval category that covers most of the site
Recent flood extents and time since wet
"TSWW20pc"
The time elapsed since >20% of the site was inundated
"TSW2"
A two-category version of TSWW20pc
"pcFlooded"
Percentage of the site inundated during survey
"Water_ha"
Area of water present within site during survey
"Waterores20pc_
Presence or absence of water inundating >20% of the site during survey
1_0"
"Waterpres_1_0" Presence or absence of water within the site during survey
"Waterpres_YN" Presence or absence of water within the site during survey
"Waterpres20pc_
Presence or absence of water inundating >20% of the site during survey
YN"
"Cumpcflood_3yr The sum of the percentage of the site inundated during the months of September and November over the preceding
NS"
three years
"Cumpcflood_3yr
The sum of the percentage of the site inundated within the same month as the survey over the preceding three years
sm"
"Maxextent_3yrN The maximum percentage of the site inundated during the months of September and November during the preceding 3
S"
years
Water in the surrounding landscape
"DistAnyChan" Distance from the centre of the survey site to the nearest channel of any kind (m)
"DistBidgee"
Distance from the centre of the survey site to the Murrumbidgee River
"DistFarmDam" Distance from the centre of the survey site to the nearest farm dam
"DistMajCk"
Distance from the centre of the survey site to the nearest major creek (m)
"DistPerWater" Distance from the centre of the survey site to the nearest perennial water
"FarmDamCt5km"Number of farm dams within a 5km radius of the centre of the survey site
150 | Flow dependent ecological responses
Table 2 Vegetation variables used in analyses. Each variable was either averaged to site level, or analysed as
proportions of photos on which they were present per site.
Single tree photo
Crown death score
Crown density score
Live or dead
Frequency of tree
flowering
Frequency of new buds
or new leaves
Tree health score
Landscape photo
Description
Percentage of the total crown completely defoliated (0 =0%; 1 = 1-10%; 2 = 11-20%; 3 = 21-40%); 4 = 41-60%; 5 = 61-80%; 6 = 8190%; 7 = 91-100%)
Percentage of skylight blocked by those portions of the tree crown that were foliated (0 =0%; 1 = 1-10%; 2 = 11-20%; 3 = 21-40%); 4
= 41-60%; 5 = 61-80%; 6 = 81-90%; 7 = 91-100%).
Tree alive (1) or dead (0)
Flowers visible on tree (Yes=1/No=0).
New buds or new leaves visible on tree (Yes=1/No=0).
The sum of the above scores (maximum score 17; scores for crown death reversed)
Number of individual live trees visible in each image, excluding those that formed part of the blurred horizon line, were indistinct
because they were too far away or too close to other trees, or that were <50% within the image.
Number of individual dead trees visible in each image, excluding those that formed part of the blurred horizon line, were indistinct
Dead tree count
because they were too far away or too close to other trees, or that were <50% within the image.
Total number of individual trees visible in each image, excluding those that formed part of the blurred horizon line, were indistinct
Total tree count
because they were too far away or too close to other trees, or that were <50% within the image.
Frequency of old trees Number of images in which old trees are visible (mature or aged-mature).
Frequency of young trees Number of images in which young trees are visible (seedling or sapling).
Presence of old trees
Old trees (mature or aged-mature) present at site (Yes=1/No=0)
Presence of young trees Young trees (seedling or sapling) present at site (Yes=1/No=0)
Number of tree size
Of 6: Seedling, sapling, pole, adult, mature, aged-mature.
classes
Total number of individual subshrubs <1m tall visible in each image, excluding those that formed part of the blurred horizon line, or
Shrubs <1m count
that were <50% within the image.
Total number of individual large shrubs >1m tall visible in each image, excluding those that formed part of the blurred horizon line,
Shrubs >1m count
or that were <50% within the image.
Large woody debris
Count of all pieces of woody debris >50% within the image estimated at >10cm diameter. If a complex fallen branch was present, the
(LWD) score
number of points touching the ground was counted and included in the overall score.
Groundcover photo
Bare %
Bare soil, rocks and stones.
Live plants %
Any living plant, including trunks of living trees and upright dormant plants.
Dead plants %
Standing dead plants.
Organic litter %
Dead, detached plant material <2cm diameter.
Woody ground cover % Dead, detached woody plant material >2cm diameter.
Cryptogams %
Any visible cryptogams, e.g. lichen.
Frequency of Sclerolaena
Number of images in which Sclerolaena is visible
spp.
Frequency of Hordeum
Number of images in which Hordeum is visible
spp.
Frequency of
Number of images in which Lignum (Muehlenbeckia) is visible
Muehlenbeckia
GIS-based
"Wood1a"
The area of River Red Gum tall gallery forest (adjacent to rivers) within a 5km radius of the centre of the survey site
"Wood1c"
The area of River Red Gum forest with grass / chenopod shrub groundcover within a 5km radius of the centre of the survey site
"Wood2a"
The area of Black Box with Lignum / Nitre Goosefoot understorey within a 5km radius of the centre of the survey site
"Wood2c"
The area of Black Box with grass / chenopod shrub groundcover within a 5km radius of the centre of the survey site
"Wood1b"
The area of River Red Gum forest with Spike-rush ground cover within a 5km radius of the centre of the survey site
The area of River Red Gum forest with shrubby (Lignum / River Cooba) understorey within a 5km radius of the centre of the survey
"Wood1d"
site
"Wood2b"
The area of Black Box with grass / chenopod shrub groundcover within a 5km radius of the centre of the survey site
"Allwood5km"
The area of woodland of any type within a 5km radius of the centre of the survey site
Composite site-based
All variables rescaled to values between 0 and 1 before summing
variables
The sum of: Shrubs <1m count, Shrubs >1m count, Large woody debris (LWD) score, Live plants %, Organic litter %, Presence of old
"SITECOMPLEXITYA"
trees, Presence of young trees
The sum of: Shrubs <1m count, Shrubs >1m count, Large woody debris (LWD) score, Live plants %, Organic litter %, Presence of old
"SITECOMPLEXITYB"
trees, Presence of young trees, Total tree count, Number of tree size classes, and Woody ground cover
The sum of:Crown death score, Crown density score, Shrubs <1m count, Shrubs >1m count, Large woody debris (LWD) score, Live
plants %, Organic litter %, Presence of old trees, Presence of young trees, Total tree count, Number of tree size classes, and Woody
"SITESCORE"
ground cover
"SUMSCOREALL"
After rescaling scores to values between 0 and 1, the sum of all photo-derived vegetation variables
Live tree count
Flow dependent ecological responses | 151
2.14
Antecedent flow and juvenile fish abundance in floodplain
wetlands of the mid-Murray River
L. Beesley1, 2, A. Price3, A.J. King1,4, J.D. Koehn1, B. Gawne3, D. Nielsen3 & D.C. Gwinn5
1
Arthur Rylah Institute for Environmental Research, Department of Environment and Primary
Industries, Heidelberg, Vic., Australia
2
Current address: Centre of Excellence in Natural Resource Management, University of Western
Australia, Albany, Australia
3
Murray-Darling Freshwater Research Centre, Wodonga. La Trobe University and CSIRO, Wodonga
4
Current address: Research Institute for Environment and Livelihoods, Charles Darwin University,
Darwin, NT, Australia
5
Program for Fisheries and Aquatic Sciences, School of Forest Resources and Conservation,
University of Florida, Gainesville, Florida, USA
INTRODUCTION
One of the greatest threats to lotic ecosystems is flow regulation. The impacts of flow
regulation on native fish are often mitigated using periodic water allocations, termed
environmental flows. The effectiveness of environmental flows has been studied in some
systems, but the role of flow conditions prior to water allocation (i.e. antecedent conditions)
has rarely been quantified. This is somewhat surprising because research on an array of
biota, including fish, suggest that antecedent flows, particularly episodes of flood and
drought, can affect: (i) the dispersal of biota (Labbe and Fausch 2000; Wen et al. 2011), (ii)
the recruitment and abundance of fish (Freeman et al. 2001; Puckridge, Costelloe and Reid
2010; Balcombe et al. 2012), (iii) the productivity of ecosystems (Puckridge, Costelloe and
Reid 2010; Leigh et al. 2010; Wen et al. 2011), (iv) the number of species (Balcombe and
Arthington 2009; Puckridge, Costelloe and Reid 2010; Leigh 2013), (v) the condition of
individuals (Abujanra, Agostinho and Hahn 2009; Balcombe et al. 2012), and (vi) the
condition of refuge habitats (Labbe and Fausch 2000; Leigh et al. 2010).
A significant challenge to informing how antecedent conditions might guide the use of
environmental flows for the benefit of native fishes is the difficulty collecting reliable longterm community data. Field data on fish assemblages are always imperfect and can be
highly variable. Variation in the sampling methods, the behaviour of the study organisms
and the environment, can lead to variation in detection probability, potentially resulting in
spurious conclusions (Hangsleban, Allen and Gwinn 2013; Archaux, Henry and Gimenez
2011). In spite of these sampling issues, few investigations of environmental flows and fish
have attempted to account for incomplete or variable detection (but see Bradford et al.
2011). Fortunately, there have been many recent advances in the applications of Bayesian
hierarchical models that account for incomplete and variable detection (Royle and Dorazio
2008).
152 | Flow dependent ecological responses
We developed a Bayesian hierarchical multiple-species abundance model for six wetlands in
the Murray River (Australia), a flow-regulated system that received variable levels of
inundation, and whose fish populations were sampled seasonally, over five years. We
investigated the relative importance of short-term, medium-term and long-term antecedent
flow conditions on juvenile fish abundance. We also evaluated the factors that affect the
detection probability of wetland fish species. Lastly, we discuss the implications of our
results for informing environmental flow management aimed at enhancing native fish
communities in floodplain wetlands.
METHODS
This study was conducted in six floodplain wetlands of the mid-Murray River, a heavily flowregulated system in south-eastern Australia (Close 1990), sampling three times a year, for
five years (October 2007 to April 2012). Each wetland varied in river-wetland commence-tofill values (Figure 1). Intra-annual sampling was related to season (late October: Spring; late
January: Summer; April: Autumn). The drying of one wetland prevented sampling twice, and
poor access prevented sampling of three wetlands once. Consequently, a total of 85
samples were collected. Further detailed descriptions of study wetlands exist (Beesley et al.
2012).
60,000
Corowa
Yarrawonga
Average Daily Discharge (ML/d)
50,000
40,000
30,000
20,000
5
10,000
2
3
4,6
1
0
Jun-02
Jun-03
Jun-04
Jun-05
Jun-06
Jun-07
Jun-08
Jun-09
Jun-10
Jun-11
Jun-12
Figure 1 Average daily discharge from June 2002 to June 2012 for two gauging stations on the Murray River
(Yarrawonga (grey) and downstream Corowa (dark). Horizontal dashed lines showed the flow threshold
Flow dependent ecological responses | 153
when the six study wetlands start receiving water from the river (wetlands 1-5 relative to Corowa, 6 relative
to Yarrawonga). Open triangles indicated sampling periods.
Fish were sampled using a passive method, single-wing fyke nets, and an active method,
beach seine. Sampling effort was scaled to wetland size to keep sample coverage similar
among wetlands. Generally, four fyke nets (range 2 to 5) were set overnight on each
sampling occasion (1.5-mm mesh, six hoops, two funnels, 5-m wing 0.6m drop); and on
average 4.5 seine hauls (range 3 to 5) were conducted during the day (0.5mm-mesh, 7m
long, 1.5-m drop with a small purse). Fish collected were identified to species, counted and
released. The majority of all specimens collected (80-100%) were ≤ age 1 (Beesley et al.
2012) with the exception of oriental weatherloach (Misgurnus anguillicaudatus) which were
mainly age 1+. A detailed description of the study wetlands and fish sampling methods is
provided in Beesley et al. (2012).
We summarized the indicators of wetland inundation into three flow metrics representing
antecedent flows on three different time scales. This was achieved by summing the number
of days that a wetland was inundated for each time scale. The time scales were defined as
short-term (three months prior = 1 to 91 days), medium-term (four to twelve months prior =
92 to 365 days), and long-term (one year to five years prior = 366 to 1822 days). These time
periods were deliberately not nested as to ease interpretation and avoid issues of
correlation among antecedent flow metrics. Together the short and medium-term metrics
describe wetland connection over the year prior to sampling, and together with the longterm metric, describe wetland connection over five years prior to sampling. All flow metrics
were determined by using average daily discharge data from the nearest gauging station
(Figure 1).
Statistical Analyses
We performed our analysis of antecedent flow and juvenile fish abundance with a multispecies occupancy model (Royle and Dorazio 2008), an extension of the single-species
occupancy models. These models were essentially zero-inflated logistic regressions that
accounted for imperfect detection of individuals, allowing for variation in sampling
efficiency. We hypothesized that flow and seasonal covariates may influence the mean
abundance of juvenile fish and incorporated them into the model with a log link function
(McCullagh and Nelder 1989). We also hypothesized that the detection probability may be
influenced by the type of sampling gear used and the wetland sampled. We incorporated
these covariates into the detection model using a logit link (McCullagh and Nelder 1989).
Finally, the effect of site (wetland) was modelled as a random effect across wetlands,
representing the cumulative influences of the unique characteristics of each wetland
impacting on sampling efficiency (e.g. area, depth, or plant cover).
154 | Flow dependent ecological responses
Our model assumes that the abundance of fish is a latent random variable ��� described by
a Poisson distribution as ��� ~Poisson���� �, where ��� is the Poisson mean for species i at
site j. Our data ��� is the total number of replicate samples where species i was observed at
site j. We assume that ��� is the result of multiple binomial outcomes as,
��� ~Binomial���� , �� �, where ��� is the probability of detecting at least one individual of
species i at site j and �� is the number of replicate samples collected at site j. We specified
the relationship between ��� and ��� per Royle and Nichols (2003) as, ��� = 1 − �1 − ��� �
where ��� represents the detection probability on the individual level as opposed to ���
���
,
which is the probability of detecting one or more individuals of species i at site j. This
formulation essentially models the detection probability ��� as a random effect defined by
the relationship between ��� and ��� and the mixing distribution of ��� (Poisson) to account
for variation in ��� due to variation in abundance of fish among sites.
We incorporated antecedent-flow and seasonal covariates that may influence the mean
abundance of fish (��� ) into the model with a log link (McCullagh and Nelder 1989) as:
log(λij ) = β 1i + β 2 i long j + β 3i medium j + β 4 i short j + β 5 i long j short j +
β 6 i medium j short j + β 7 i season 2 j + β 8i season3 j + β 9 i ,k weland
where, �1� represents the species-specific intercept of the abundance sub-model,
�2� , �3� , & �4� represent the effects of each antecedent-flow variable on the abundance of i
species, �5� and �6� represent the interaction effects of short, medium and long-term
antecedent flows, and �7� and �8� represent species-specific seasonal differences in
abundance. The parameters represented by �9�,� were random effects across i species and
k wetlands to account for differences in the abundance of fish related to each wetland’s
unique characteristics and the non-independence of samples collected through time in the
same wetland.
We hypothesized that the detection probability ��� was influenced by sampling gear,
wetland area, and wetland sampled, and incorporated them into the detection model using
a logit link as:
logit (rij ) = α1i + α 2 i gear j + α 3i area j + α 4 i gear j area j + α 5i ,k wetland
where �1� represents the species-specific intercept and �2� represents the species-specific
gear effect. The parameter �3� represents the species-specific effect of the wetland area on
detection as an index of overall wetland size. The parameter �4� represents the interaction
between the gear and the wetland area to evaluate the possibility that different sampling
gears are affected by the size of the wetland differently. The parameters represented by
�5�,� were modelled as random effects across i species and k wetlands and represent the
cumulative influences of the unique characteristics of each wetland that could impact
sampling efficiency.
Flow dependent ecological responses | 155
All species-specific parameters were specified as random effects drawn from hyperdistributions governed by community level hyper-parameters. All hyper-distributions were
specified as normal distributions. The random effects of wetland on abundance (�9�,� ) and
detection (�5�,� ) were also assumed to be drawn from normal distributions with means
equal to zero and species-specific standard deviations. We evaluated the importance of
covariates on species-specific abundance and detection by sampling the posterior
distributions of the full model. We considered covariates important when 90% credible
intervals did not include zero.
Flow scenarios
We evaluated native and non-native fish responses to wetland inundation to aid in the
interpretation of species-specific results and gain insight into flow characteristics that result
in fish communities dominated by native species or non-native species. We achieved this by,
first, defining five generic flow scenarios based on intervals of wetland inundation as: (i)
zero days of inundation during a five-year period, (ii) 60 days of inundation once at the end
of five years, (iii) 60 days of inundations once every year, (iv) 60 days of inundation once
every six months, and (v) 60 days of inundation every three months (see Figure 2). We
generated simulated flow data based on each flow scenario and summarized it into the
wetland inundation metrics for the long-term, medium-term, and short-term time scales.
We then utilized our fitted model to predict the ratio of total abundance of native fish to
non-native fish for each hypothetical flow scenario.
Figure 2 Five hypothetical flow scenarios. Dashed grey bars indicate wetland inundation events.
RESULTS AND DISCUSSION
Our results offer insight into the relative influence of multiple temporal scales of antecedent
flows on the structure of a wetland fish assemblage. The hierarchical multi-species
modelling approach identified patterns across the entire assemblage and highlighted
156 | Flow dependent ecological responses
species-specific variation. For some species, juvenile fish abundance in these floodplain
wetlands was significantly related to antecedent flow conditions (Table 1). The short-term
responses found by this study, were in general agreement with the flow-related life history
groupings put forward by Humphries et al. (1999). For example, our finding that flathead
gudgeon, carp gudgeon and Australian smelt, showed little to no response to recent high
flows corroborates the findings of King et al. (2003) and is consistent with Humphries et al.
(1999) categorisation of these species as flow generalists.
Of particular interest though, was our finding that medium and longer-term antecedent
flows affected the response to short-term flows for other species (i.e. common carp,
gambusia and oriental weatherloach) (see interaction terms in Table 1). These interactions
are important for management, because they suggest that past flow conditions can
moderate the impact of current environmental watering events on these non-native
species. That common carp juvenile abundance was greatest when recent flows were high
and previous flows were low is probably driven by the fact that recently inundated, shallow
habitats with abundant plant material are ideal carp spawning habitat (Koehn et al. 2000),
and that drying of the floodplain promotes a dense growth of macrophytes and understorey
terrestrial grasses (Briggs and Maher 1985) – a perfect spawning substrate when inundated.
Our evaluation of hypothetical flow scenarios suggests that policies that ensure the
frequent inundation of wetlands will promote native fish and suppress non-native fish
(Figure 2). This prediction agrees with the findings of a recent study by Ho, Bond and
Thompson (2013) who also studied wetland fish in south-eastern Australia, and found that
regularly flooded wetlands supported higher abundances of carp gudgeon (native), and
irregularly flooded wetlands supported higher abundances of eastern gambusia (nonnative). In the southern Murray-Darling Basin, river regulation has reduced the frequency of
floodplain inundation (Maheshwari, Walker and McMahon 2006); so, an improvement for
native species as the frequency of flow increases is in accordance with general expectations
(Poff et al. 1997). The restoration of flow regimes in regulated rivers has benefited native
fish elsewhere in the world (see Travnichek, Bain and Maceina 1995; Brown and Ford 2002;
Kiernan, Moyle and Crain 2012). In the southern Murray-Darling Basin, our results suggest
that policies aimed at avoiding long drying periods by frequent watering will benefit the
native fish community.
Flow dependent ecological responses | 157
Table 1 The significance of antecedent flow covariates at different time scales (short, medium and long) on
species-specific juvenile abundance. The symbol, + represents a significant positive affect on abundance and
the symbol – represents a significant negative affect on abundance. Significance was determined by 90%
credible intervals that do not overlap zero. LT:ST = the interaction between long-term and short-term
wetland inundation; MT:ST = the interaction between medium-term and short-term wetland inundation.
SPECIES
SHORT-TERM MEDIUM-TERM LONG-TERM LT:ST MT:ST
Native species
Australian smelt
+
Flathead gudgeon
+
+
Carp gudgeon
-
Unspecked hardyhead
Non-native species
European perch
Oriental weatherloach
+
-
Eastern gambusia
+
-
Common carp
+
-
Goldfish
+
-
-
Our analysis revealed multiple sources of heterogeneity in the detection process that could
bias analyses that do not account for variable detection. We found that detection
probability varied randomly among species and wetlands, and systematically with gears and
wetland areas (Figure 3). Firstly, our model revealed an aspect of our sampling design that
imparted significant heterogeneity in detection probability, i.e. the use of different gear
types. The use of multiple gears is common when sampling fish communities because it
expands detection opportunities across species and individuals with variable vulnerability to
different detection methods (e.g. Lauretta et al. 2013). However, the use of statistically valid
methods to combine data from multiple gears is rare in the literature (but see Haynes et al.
2013).
158 | Flow dependent ecological responses
Figure 2. Hypothetical flow scenario’s i – v where, (i) zero days of inundation during a five-year period (ii) is
60 days of inundation once at the end of five years, (iii) is 60 days of inundations once every year, (iv) is 60
days of inundation once every six months, and (v) is 60 days of inundation every three months. The y-axis is
the proportion of the total possible number of days of inundation for each time-scale (long-term, mediumterm, and short-term). Nat/Non represents the resulting ratio of total native fish abundance to total nonnative fish abundance of each flow scenario, where higher numbers indicate more natives relative to nonnatives.
One common method of sampling with multiple gears is to standardize the relative sampling
effort of both gears (e.g. Rayner, Jenkins and Kingsford 2009); however, at the end of the
survey it remains unknown how detection varied for each method. Our hierarchical model
combined the data from both gears in a statistically valid way by estimating the detection
probability of each gear. This analysis indicated that the detection probability tended to be
higher for fykes than seines, but that fyke detection probability declined as the area of a
given wetland increased. Variation in detection of this nature can easily lead to spurious
trends and incorrect conclusions. Indeed, in a separate analysis not reported here, we
analysed these data with a Zero-inflated Poisson GLM that ignored variable and incomplete
detections. This analysis resulted in unsubstantiated parameter estimates for most species
that were clearly unreliable (Beesley unpub. data). Thus, sampling designs for evaluating the
effects of future environmental flows in the Murray-Darling River basin (and elsewhere) will
benefit from careful consideration of the detection biases indicated by this analysis.
Flow dependent ecological responses | 159
1.0
Seine
(b)
European perch
Unspecked hardyhead
Carp gudgeon
Flathead gudgeon
Australian smelt
0.6
0.8
(a)
Goldfish
Common carp
Gambusia
Oriental weatherloach
0.0
0.2
0.4
Detection probability (r
Fyke
-1
0
1
2
-1
0
1
2
Wetland area
Figure 3. Species-specific predictions of detection probability for the interaction between each sampling
gear (fyke, seine) and across a wetland area gradient. Dotted lines represent native species while solid lines
represent non-native species.
CONCLUSIONS
Our hierarchical model of juvenile fish abundance revealed that antecedent flows are
important in determining the structure of the wetland fish community. We observed that
native and non-native fishes tended to differ in their response to antecedent flows, and that
the response of some non-native species to recent flow was affected by prior medium and
longer-term conditions. Furthermore, species-specific detection probabilities demonstrated
significant spatial and temporal variation indicating that future research investigating
environmental flow policies for enhancing native fish communities will benefit from
accounting for variable detection. This knowledge will allow managers to develop effective
environmental flow policies that promote the conservation of native fish communities.
160 | Flow dependent ecological responses
REFERENCES
ALEXANDER, P., NIELSEN, D. L. & NIAS, D. 2008. Response of wetland plant communities to
inundation within floodplain landscapes. Ecological Management & Restoration, 9,
187-195.
AMON, R. & BENNER, R. 1996. Photochemical and microbial consumption of dissolved
organic carbon and dissolved oxygen in the Amazon River system. . Geochimica et
Cosmochimica Acta 60:, 1783-1792.
AMOROS, C. & BORNETTE, G. 2002. Connectivity and biocomplexity in waterbodies of
riverine floodplains. . Freshwater Biology, 47, 761-776.
BAILEY, L. L., SIMONS, T. R. & POLLOCK, K. H. 2004. Estimating site occupancy and species
detection probability parameters for terrestrial salamanders. Ecological Applications,
14, 692-702.
BALDWIN, R. F., CALHOUN, A. J. K. & DEMAYNADIER, P. G. 2006. The significance of
hydroperiod and stand maturity for pool-breeding amphibians in forested
landscapes. Canadian Journal of Zoology, 84, 1604-1615.
BALL, I. J., GILMER, D.S., COWARDIN, L.M. AND RIECHMANN, J.H. 1975. Survival of wood
duck and mallard broods in north- central Minnesota. J. Wildl. Manage., 39, 776-780.
BARRETT, G. W., SILCOCKS, A., BARRY, S., CUNNINGHAM, R. & POULTER, R. 2003. The new
atlas of Australian birds. Melbourne: RAOU.
BREN, L. J. & GIBBS, N. L. 1986. Relationships between flood frequency, vegetation and
topography in a river red gum forest. Australian Forest Research, 16, 357-370.
BUNN, S. E. & ARTHINGTON, A. H. 2002. Basic principles and ecological consequences of
altered flow regimes for aquatic biodiversity. Environmental Management, 30, 492507.
BUNN, S. E., DAVIES, P. M. & WINNING, M. 2003. Sources of organic carbon supporting the
food web of an arid zone floodplain river. Freshwater Biology, 48, 619-635.
BURNHAM, K. P. & ANDERSON, D. R. 2002. Model selection and multimodel inference: a
practical information theoretic approach, New York, Springer Science + Business
Media, LLC.
CASANOVA, M. T. & BROCK, M. A. 2000. How do depth, duration and frequency of flooding
influence the establishment of wetland plant communities? Plant Ecology, 147, 237250.
CATFORD, J. A., DOWNES, B. J., GIPPEL, C. J. & VESK, P. A. 2011. Flow regulation reduces
native plant cover and facilitates exotic invasion in riparian wetlands. Journal of
Applied Ecology, 48, 432-442.
CLARKE, M. F. & OLDLAND, J. M. 2007. Penetration of remnant edges by noisy miners
(Manorina melanocephala) and implications for habitat restoration. Wildlife
Research, 34, 253-261.
COPP, G. H. 1989. The habitat diversity and fish reproductive function of floodplain
ecosystems. Environmental Biology of Fishes, 26, 1-26.
Flow dependent ecological responses | 161
CUNNINGHAM, S. A., MAC NALLY, R., READ, J., BAKER, P., WHITE, M., THOMSON, J. R. &
GRIFFIOEN, P. 2009. A Robust Technique for Mapping Vegetation Condition Across a
Major River System. Ecosystems, 12, 207-219.
FINDLAY, S. & SINSABAUGH, R. 1999. Unravelling the sources and bioavailability of dissolved
organic matter in lotic aquatic ecosystems. Marine and Freshwater Reseasrch, 50,
781-790.
FORD, H. A., BARRETT, G. W., SAUNDERS, D. A. & RECHER, H. F. 2001. Why have birds in the
woodlands of Southern Australia declined? Biological Conservation, 97, 71-88.
FORD, H. A., WALTERS, J. R., COOPER, C. B., DEBUS, S. J. S. & DOERR, V. A. J. 2009. Extinction
debt or habitat change? - Ongoing losses of woodland birds in north-eastern New
South Wales, Australia. Biological Conservation, 142, 3182-3190.
FRANKLIN, P., DUNBAR, M. & WHITEHEAD, P. 2008. Flow controls on lowland river
macrophytes: A review. Science of the Total Environment, 400, 369-378.
GEHRKE, P., ASTLES, K. & HARRIS, J. 1999. Within-catchment effects of flow alteration on
fish assemblages in the Hawkesbury-Nepean River system, Australia. Regulated
Rivers: Research and Management, 15, 181-198.
GEORGE, A. J., WALKER, K. F. & LEWIS, M. M. 2005. Population status of eucalypt trees on
the River Murray floodplain, South Australia. River Research and Applications, 21,
271-282.
GORE, J., LAYZER, J. & MEAD, J. 2001. Macroinvertebrate instream flow studies after 20
years: A role in stream management and restoration. Regulated Rivers: Research and
Management, 17, 527-542.
GRAÇA, M. 2001. The role of invertebrates on leaf litter decomposition in streams – a
review. International Review of Hydrobiology, 86, 383-393.
GREET, J., COUSENS, R. D. & WEBB, J. A. 2013. Seasonal timing of inundation affects riparian
plant growth and flowering: implications for riparian vegetation composition. Plant
Ecology, 214, 87-101.
GRIESE, V. 2011. Responses of frogs to changing patterns of habitat availability during
flooding in two regulated streams., Charles Sturt University.
GROWNS, J. E., KING, A. J. & BETTS, F. M. 1999. The Snag Bag: a new method for sampling
macroinvertebrate communities on large woody debris. Hydrobiologia, 405, 67-77.
GULVE, P. S. 1994. Distribution and extinction patterns within a northern metapopulation of
the pool frog, Rana Lessonae. Ecology, 75, 1357.
HAZELL, D., CUNNINGHAM, R., LINDENMAYER, D., MACKEY, B. & OSBORNE, W. 2001. Use of
farm dams as frog habitat in an Australian agricultural landscape: factors affecting
species richness and distribution. Biological Conservation, 102, 155-169.
HAZELL, D., HERO, J.-M., LINDENMAYER, D. & CUNNINGHAM, R. 2004. A comparison of
constructed and natural habitat for frog conservation in an Australian agricultural
landscape. Biological Conservation, 119, 61-71.
162 | Flow dependent ecological responses
HEALEY, M., THOMPSON, D. & ROBERTSON, A. 1997. Amphibian communities associated
with billabong habitats on the Murrumbidgee floodplain, Australia. Austral Ecology,
22, 270-278.
HLADYZ, S., NIELSEN, D. L., SUTER, P. J. & KRULL, E. S. 2012. Temporal variations in organic
carbon utilization by consumers in a lowland river. River Research and Applications,
28, 513-528.
HORNER, G. J., BAKER, P. J., MAC NALLY, R., CUNNINGHAM, S. C., THOMSON, J. R. &
HAMILTON, F. 2009. Mortality of developing floodplain forests subjected to a drying
climate and water extraction. Global Change Biology, 15, 2176 - 2186.
HUMPHRIES, P. 1996. Aquatic macrophytes, macroinvertebrate associations and water
levels in a lowland Tasmanian river. Hydrobiologia, 321, 219-233.
HUMPHRIES, P., KING, A. J. & KOEHN, J. D. 1999. Fish, flows and flood plains: links between
freshwater fishes and their environment in the Murray-Darling system, Australia.
Environmental Biology of Fishes, 56, 129-151.
JACKMAN, S. 2011. pscl: Classes and Methods for R Developed in the Political Science
Computational Laboratory, Stanford University. Department of Political Science,
Stanford University. Stanford, California. R package version 1.04.1 ed.
JAKOB, C., POIZAT, G., VEITH, M., SEITZ, A. & CRIVELLI, A. J. 2003. Breeding phenology and
larval distribution of amphibians in a Mediterranean pond network with
unpredictable hydrology. Hydrobiologia, 499, 51-61.
JANSEN, A. & HEALEY, M. 2003. Frog communities and wetland condition: relationships with
grazing by domestic livestock along an Australian floodplain river. Biological
Conservation, 109, 207-219.
JOHNS, C., REID, C. J., ROBERTS, J., SIMS, N., DOODY, T. M., OVERTON, I. C., MCGINNESS, H.
M., ROGERS, K., CAMPBELL, C. & GAWNE, B. 2009. Native trees of the River Murray
floodplain: Literature review and experimental designs to examine effects of flow
enhancement and floodwater retention. Final report to the Murray-Darling Basin
Authority. Wodonga: Murray-Darling Freshwater Research Centre.
JUNK, W. J., BAYLEY, P. B. & SPARKS, R. E. 1989. The flood pulse concept in river-floodplain
systems. In: DODGE, D. P. (ed.) Proceedings of the international large river
symposium. Canadian Special Publication on Fisheries and Aquatic Sciences.
JUNK, W. J. & WANTZEN, K. 2004. The flood pulse concept: New aspects, approaches and
applications - an update. In: WELCOMME, R. & PETR, T. (eds.) Proceedings of the
Second International Symposium on the Management of Large Rivers for Fisheries.
Bangkok. RAP: Food and Agriculture Organization & Mekong River Commission FAO
Regional Office for Asia and the Pacific.
KERR, J. L., BALDWIN, D. S. & WHITWORTH, K. L. 2013. Options for managing hypoxic
blackwater events in river systems: A review. Journal of Environmental Management,
114, 139-147.
KINGSFORD, R. T. 2000. Ecological impacts of dams, water diversions and river management
on floodplain wetlands in Australia. Austral Ecology, 25, 109-127.
Flow dependent ecological responses | 163
KOLOZSVARY, M. B. & SWIHART, R. K. 1999. Habitat fragmentation and the distribution of
amphibians: Patch and landscape correlates in farmland. Canadian Journal of
Zoology, 77, 1288-1299.
LAKE, P. 2000a. Disturbance, patchiness, and diversity in streams. Journal of the North
American Benthological Society, 19, 573-592.
LAKE, P. S. 2000b. Disturbance, patchiness, and diversity in streams. Journal of the North
American Benthological Society, 19, 573-592.
LAKE, P. S. 2003. Ecological effects of perturbation by drought in flowing waters. Freshwater
Biology, 48, 1161-1172.
LANE, S., HAMER, A. & MAHONY, M. 2007. Habitat correlates of five amphibian species and
of species-richness in a wetland system in New South Wales, Australia. Applied
Herpetology, 4, 65-82.
LELAND, H. V. 2003. The influence of water depth and flow regime on phytoplankton
biomass and community structure in a shallow, lowland river. Hydrobiologia, 506,
247-255.
MAC NALLY, R., BENNETT, A. F., THOMSON, J. R., RADFORD, J. Q., UNMACK, G., HORROCKS,
G. & VESK, P. A. 2009. Collapse of an avifauna: climate change appears to exacerbate
habitat loss and degradation. Diversity and Distributions, 15, 720-730.
MAC NALLY, R., BOWEN, M., HOWES, A., MCALPINE, C. A. & MARON, M. 2012. Despotic,
high-impact species and the subcontinental scale control of avian assemblage
structure. Ecology, 93, 668-678.
MACKENZIE, D. I. & BAILEY, L. L. 2004. Assessing the fit of site-occupancy models. Journal of
Agricultural, Biological, and Environmental Statistics, 9, 300-318.
MACKENZIE, D. I., NICHOLS, J. D., ROYLE, J. A., POLLOCK, K. H., BAILEY, L. L. & HINES, J. E.
2006. Occupancy estimation and modeling, Burlington, MA, Academic Press.
MACKENZIE, D. I., NICHOLS, J. D., SUTTON, N., KAWANISHI, K. & BAILEY, L. L. 2005.
Improving inferences in population studies of rare species that are detected
imperfectly. Ecology, 86, 1101-1113.
MAJOR, R. E., CHRISTIE, F. J. & GOWING, G. 2001. Influence of remnant and landscape
attributes on Australian woodland bird communities. Biological Conservation, 102,
47-66.
MARON, M. 2007. Threshold effect of eucalypt density on an aggressive avian competitor.
Biological Conservation, 136, 100-107.
MATHEWS, W. 1998. Patterns in Freshwater Fish Ecology. , New York, Chapman & Hall.
MCGINNESS, H. M., ARTHUR, A. D., DAVIES, M. & MCINTYRE, S. 2013. Floodplain woodland
structure and condition: the relative influence of flood history and surrounding
irrigation land use intensity in contrasting regions of a dryland river. Ecohydrology, 6,
201-213.
MCGINNESS, H. M., ARTHUR, A. D. & REID, J. R. W. 2010. Woodland bird declines in the
Murray-Darling Basin: are there links with floodplain change? The Rangeland Journal,
32, 315-327.
164 | Flow dependent ecological responses
MCKIBBIN, R., DUSHENKO, W. T., VANAGGELEN, G. & BISHOP, C. A. 2008. The influence of
water quality on the embryonic survivorship of the Oregon spotted frog (Rana
pretiosa) in British Columbia, Canada. Science of The Total Environment, 395, 28-40.
OLSEN, P. 2008. The State of Australia's Birds 2008. Supplement to Wingspan, Vol. 18, no. 4.:
Birds Australia.
PINHEIRO, J., BATES, D., DEBROY, S., SARKAR, D. & R DEVELOPMENT CORE TEAM 2013.
nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-107 ed.
POFF, N. L. & WARD, J. V. 1989. Implications of streamflow variability and predictability for
lotic community structure: A regional analysis of streamflow patterns. Canadian
Journal of Fisheries and Aquatic Sciences, 46, 1805-1818.
POFF, N. L. & ZIMMERMAN, J. K. H. 2010. Ecological responses to altered flow regimes: a
literature review to inform the science and management of environmental flows.
Freshwater Biology, 55, 194-205.
REID, J. & FLEMING, M. 1992. The conservation status of birds in arid Australia. The
Rangeland Journal, 14, 65-91.
REID, J. R. W. 1999. Threatened and declining birds in the New South Wales sheep-wheat
belt: 1. Diagnosis, characteristics and management. Consultancy report to NSW
National Parks and Wildlife Service. Canberra, ACT, Australia: CSIRO Wildlife and
Ecology.
REYNOLDS, C. 1996. The production, biomass and structure of phytoplankton in large rivers.
Archive fur Hydrobiologia. Supplement, 113, 161-187.
ROBERTS, J. & MARSTON, F. 2000. Water regime of wetland and floodplain plants in the
Murray-Darling Basin: A source book of ecological knowledge. Canberra: CSIRO Land
and Water.
RUBBO, M. J. & KIESECKER, J. M. 2005. Amphibian Breeding Distribution in an Urbanized
Landscape. Conservation Biology, 19, 504-511.
SCHIEMER, F. & ZALEWSKI, M. 1992. The importance of riparian ecotones for diversity and
productivity of riverine fish communities. Netherlands Journal of Zoology, 42, 323335.
SCHLOSSER, I. 1991. Stream fish ecology: a landscape perspective. Bioscience, 41, 704-712.
SHERMAN, B. S., JONES, G. J. & WEBSTER, I. T. 1994. Flow,stratification and the growth of
algae and cyanobacteria in Maude weir pool on the Murrumbidgee River.
Environmental Flows Seminar Proceedings. Canberra: Australian Water and
Wastewater Association.
SIMS, N., COLLOFF, M. & GUERSCHMAN, J.-P. 2009. Basin-scale vegetation. In: OVERTON, I.,
COLLOFF, M., DOODY, T., HENDERSON, B. & CUDDY, S. (eds.) Ecological outcomes of
flow regimes in the Murray-Darling Basin. Canberra, ACT: CSIRO.
SPENCER, C., ROBERTSON, A. & CURTIS, A. 1998. Development and testing of a rapid
appraisal wetland condition index in south-eastern Australia. Journal of
Environmental Management, 54, 143-159.
Flow dependent ecological responses | 165
SPENCER, J. A., THOMAS, R., WASSENS, S., LU, Y., WEN, L., ILES, J., KOBAYASHI, Y., HUNTER,
S., LING, J. & SAINTILAN, N. 2011. Environmental flow monitoring in the Lowbidgee
wetlands and Macquarie Marshes in 2009-10. Testing wetland resilience: monitoring
the response of iconic wetlands to re-flooding following historic drought project.
Sydney, NSW, Australia: NSW Department of Environment, Climate Change and
Water.
SPENCER, J. A. & WASSENS, S. 2009. Responses of waterbirds, fish and frogs to
environmental flows in the Lowbidgee wetlands in 2008–09. Final report for the
NSW Rivers Environmental Restoration Program. Sydney, NSW, Australia: Rivers and
Wetland Unit, NSW Department of Environment and Climate Change.
STOKES, K. E. 2008. Exotic invasive black willow (Salix nigra) in Australia: influence of
hydrological regimes on population dynamics. Plant Ecology, 197, 91-105.
STOKES, K. E. & CUNNINGHAM, S. A. 2006. Predictors of recruitment for willows invading
riparian environments in south-east Australia: implications for weed management.
Journal of Applied Ecology, 43, 909-921.
TAYLOR, P. J., WALKER, G. R., HODGSON, G., HATTON, T. J. & CORRELL, R. L. 1996. Testing of
a GIS model of Eucalyptus largiflorens health on a semiarid, saline floodplain.
Environmental Management, 20, 553-564.
THORP, J. H. & DELONG, M. D. 1994. The riverine productivity model: an heuristic view of
carbon sources and organic processing in large river ecosystems. Oikos, 70, 305-308.
VANNOTE, R. L., MINSHALL, G. W., CUMMINS, K. W., SEDELL, J. R. & CUSHING, C. E. 1980.
The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences, 37,
130-137.
VOS, C. C. & STUMPEL, A. H. P. 1996. Comparison of habitat-isolation parameters in relation
to fragmented distribution patterns in the tree frog (Hyla arborea). Landscape
Ecology, 11, 203-214.
WALKER, K. F., SHELDON, F. & PUCKRIDGE, J. T. 1995. An ecological perspective on large
dryland rivers. Regulated Rivers: Research and Management, 11, 85-104.
WARD, D., TEWS, K., KENNARD, M., PETERSON, E. & BOND, N. 2012. Murray–Darling Basin
aquatic system mapping and classification—Aquatic assets and environmental
attribute geodatabases. Report for CSIRO–Cluster project “Ecological Responses To
Altered Flow Regimes”. CSIRO.
WARD, J. V., BRETSCHKO, G., BRUNKE, M., DANIELOPOL, D., GIBERT, J., GONSER, T. &
HILDREW, A. G. 1998. The boundaries of river systems: the metazoan perspective.
Freshwater Biology, 40, 531-569.
WARFE, D. M. & BARMUTA, L. A. 2006. Habitat structural complexity mediates food web
dynamics in a freshwater macrophyte community. Oecologia, 150, 141-154.
WASSENS, S. & AMOS, C. 2011. Assessing the resilience of frog communities within River
Red Gum forest: Report to New South Wales Office of Environment and Heritage.
Albury, NSW, Australia: Institute for Land, Water and Society.
166 | Flow dependent ecological responses
WASSENS, S., ARNAIZ, O. L. & WATTS, R. J. 2007. Assessing the diversity, abundance and
hydrological requirements of frog populations at ‘Burrawang West’ and ‘Yarnel’
Lagoons, two small wetlands on anabranch creeks of the mid-Lachlan River.
WASSENS, S., HALL, A., OSBORNE, W. & WATTS, R. J. 2010a. Habitat characteristics predict
occupancy patterns of the endangered amphibian Litoria raniformis in flowregulated flood plain wetlands. Austral Ecology, 35, 944-955.
WASSENS, S., HEALEY, S. & WATTS, R. 2010b. Evaluation of landscape and patch metrics to
predict the response of frog populations following flooding. Report to NSW Office of
Environment and Heritage. Albury, NSW, Australia: Institute for Land, Water and
Society.
WASSENS, S. & MAHER, M. 2011. River regulation influences the composition and
distribution of inland frog communities. River Research and Applications, 27, 238246.
WASSENS, S., WATTS, R. J., SPENCER, J. A., HOWITT, J., MCCASKER, N. A., GRIESE, V., BURNS,
A., CROFT, R., ZANDER, A., AMOS, C. & HALL, A. 2012. Monitoring of ecosystem
responses to the delivery of environmental water in the Murrumbidgee system.
Albury, NSW, Australia: Institute for Land, Water and Society.
WEBSTER, J. R., BENFIELD, E. F., EHRMAN, T. P., SCHAEFFER, M. A., TANK, J. L., HUTCHENS, J.
J. & D'ANGELO, D. J. 1999. What happens to allochthonous material that falls into
streams: A synthesis of new and published information from Coweeta. Freshwater
Biology, 41, 687-706.
WEYRAUCH, S. L. & GRUBB, T. C. 2004. Patch and landscape characteristics associated with
the distribution of woodland amphibians in an agricultural fragmented landscape: an
information-theoretic approach. Biological Conservation, 115, 443-450.
WILBUR, H. M. 1980. Complex life cycles. Annual Review of Ecology and Systematics, 11, 6793.
WIPFLI, M. S. & MERRITT, R. W. 1994. Disturbance to a stream food web by a bacterial
larvicide specific to black flies: feeding responses of predatory macroinvertebrates.
Freshwater Biology, 32, 91-103.
ZEILEIS, A. & HOTHORN, T. 2002. Diagnostic checking in regression relationships. R News, 2,
7-10.
ZEILEIS, A., KLEIBER, C. & JACKMAN, S. 2008. Regression models for count data in R. Journal
of Statistical Software, 27, 1-25.
ZEUG, S. & WINEMILLER, K. O. 2008. Evidence supporting the importance of terrestrial
carbon in a large-river food web. Ecology, 89, 1733-1743.
Flow dependent ecological responses | 167
2.15
Hydrological and biophysical predictors of frog occupancy
patterns across the Murray-Darling Basin
Trotter, A.1, Hall, A.1, and Wassens, S.1
1
Institute of Land, Water and Society, School of Environmental Sciences, Charles Sturt University,
Albury
INTRODUCTION
Frogs are important components of wetland ecosystems in the Murray–Darling Basin (MDB).
However, the relative importance of hydrology, water quality and habitat features in driving
species occupancy patterns are poorly described. In addition, habitat occupancy modelling
for frogs is typically conducted at relatively small spatial scales, within a single catchment or
wetland complex, with transferability poorly known.
Hydrological and biophysical habitat characteristics are important predictors of frog
distributions (see conceptual model in section 6). Hydrology (e.g. flow characteristics, water
permanence) plays a vital role in driving aquatic ecological processes and hence species
distributions (Poff and Ward, 1989). Both aquatic and terrestrial ecosystem characteristics
influence the quality of habitat for frogs, due to their complex life cycle (Wilbur, 1980).
Hydrology is commonly identified as a predictor of frog occupancy and species richness
;(Gulve, 1994, Kolozsvary and Swihart, 1999, Jakob et al., 2003, Weyrauch and Grubb, 2004,
Rubbo and Kiesecker, 2005, Ball, 1975, Baldwin et al., 2006). Hydrological variability may
affecting timing of successful reproduction for different groups of frog species (Jakob et al.,
2003). Altered flow regimes causing ecological decline may influence habitat suitability for
frogs (Poff and Zimmerman, 2010). Distributions o amphibian species are correlated to
vegetation characteristics (Vos and Stumpel, 1996, Healey et al., 1997, Hazell et al., 2001,
Jansen and Healey, 2003, Bailey et al., 2004, Hazell et al., 2004, Rubbo and Kiesecker, 2005,
Lane et al., 2007, Wassens et al., 2010a), as have water quality measures (Gulve, 1994,
Weyrauch and Grubb, 2004, Lane et al., 2007, Wassens et al., 2007, McKibbin et al., 2008,
Wassens et al., 2010a).
Occupancy modelling is a relatively powerful statistical technique that uses presence–
absence data in relating these to habitat characteristics. The comprehensive incorporation
of detection probabilities into occupancy predictions, as opposed to using simple presence–
absence values, renders occupancy modelling uniquely suited when combining datasets
with unequal survey effort.
The aims of this study were to describe key drivers of frog occupancy at local, regional and
basin-wide levels, and to test the suitability of occupancy modelling for basin-wide frog
distribution models.
168 | Flow dependent ecological responses
METHODS
Species’ detection and habitat data were drawn from 173 sites from the mid-Lachlan (200607) (Wassens and Maher, 2011), mid-Murrumbidgee (2010-12) (Wassens and Amos, 2011,
Wassens et al., 2012), Lowbidgee (2008-2010) (Spencer et al., 2011, Spencer and Wassens,
2009), Murray (2008-11) (Wassens et al., 2010b), and Wakool (2010-11) (Griese, 2011). To
increase coverage of wetland complexes in the northern basin, the Gwydir and Gingham
wetlands were also surveyed in 2011-2012. Sites surveyed in more than one season were
counted as separate records, giving a total of 216 unique records, or site–season
combinations.
The frog data used for analysis were in the form of Boolean presence–absence values for
each site–season–species combination. The original data collected were discrete counts of
individuals on each sampling occasion. Tadpole and fish surveys were carried out during the
day, consisting of a sweep net survey in each 5m–wide transect, with at least three transect
sweeps completed in total. Adult frog surveys were conducted as soon after dark as possible
by two teams of two people. Each team surveyed frogs along a meandering transect
through a site for 15 to 30 minutes, using a 25-watt spotlight. All individuals observed were
identified to species and recorded. The number of calling individuals of each species was
estimated at 5- to 10-minute intervals.
Adequate and reliable data across regions were only available for 27 variables in (Table 1)
included for modelling. Flood return data were extracted from the Murray–Darling Basin
Aquatic Assets and Environmental Attributes GeoDatabases v2.0 (Ward et al., 2012).
Standard freshwater ecosystem measurements were carried out according to methods
developed by Spencer et al. (1998).
Detection histories were created for each record for occupancy modelling (MacKenzie et al.,
2005). Data were collected for seven species, but modelling was only completed for four
species Limnodynastes fletcheri, Limnodynastes tasmaniensis, Litoria peronii, and Crinia
parinsignifera with known ranges across all study regions.
There were two key modeling steps. (1) Single site covariate models, or simple models, (with
survey region as a covariate) were developed and graphed. (2) Individual covariates of high
predictive value were combined in complex models (Table 2). Akaike information criterion
AIC, model weightings (see Mackenzie and Bailey (2004)) were used to rank models.
Goodness of Fit tests were carried out using 100 parametric bootstraps and a model
considered to be a poor fit to the data if the p-value (probability of obtaining a test statistic
≥ observed) ≤ 0.05. A p-value approaching 1 indicated over-fitting (MacKenzie et al., 2006).
Flow dependent ecological responses | 169
Table 1. Descriptions of 27 variables included in univariate models for The four frog species. Variables
included in the top three ranked multivariate models for each species are highlighted in bold. Variables
classified as broad scale describe characteristics of the landscape and hydrological classifications of the
wetland complex drawn from (Ward et al., 2012). Fine scale variables describe vegetation and water quality
variables within 100m of the frog survey point.
SCALE
VARIABLE
MODELLING
ABBREVIATION
TYPE
DESCRIPTION
Region
reg
Categorical
Region that site was located in (e.g. Lachlan or Gwydir)
System type
type
Categorical
(e.g. open wetland or minor stream)
Connection type
conn
Categorical
Manner in which the site was filled (e.g. ground water–
fed, overbank, rain-fed)
Surrounding land use
surr
Categorical
Land use(s) occurring within 50 m of site
Hydrology/water
permanence
hyd
Categorical
Permanent waterbodies flood return 5 in 5 year,
seasonally flooded wetlands ( 1-2 in 5 years), and
ephemeral wetlands ( >3 in 5 years)
Flood return 1 in 1 year
fr1
Continuous
Proportion of 1km-radius site zone expected to flood each
year.
Flood return 1 in 2 years
fr2
Continuous
Proportion of 1km-radius site zone expected to flood
every 2 years
Flood return 1 in 5 years
fr5
Continuous
Proportion of 1km-radius site zone expected to flood
every 5 years
Flood return 1 in 10 years
fr10
Continuous
Proportion of 1km-radius site zone expected to flood
every 10 years
Fringing tall grass
friTallGrass
Continuous
Average percent cover of tall grasses (>0.1 m) in the
fringing zone
Fringing short grass
friShortGrass
Continuous
Average percent cover of short grasses in the fringing zone
Fringing tall herbs
friTallHerb
Continuous
Average percent cover of tall herbs in the fringing zone
Fringing short herbs
friShortHerb
Continuous
Average percent cover of short herbs in the fringing zone
Fringing cover
friCover
Continuous
Average percent total ground cover provided by all
fringing vegetation combined
Fringing layers/complexity
friLayers
Continuous;
discrete
Count of fringing vegetation types present at a site in a
season.
Aquatic tall emergent
aquTallEm
Continuous
Average percent cover of tall emergent vegetation (>0.5
m) in the aquatic zone
Aquatic short emergent
aquShortEm
Continuous
Average percent cover of short emergent vegetation
Aquatic broadleaf
emergent
aquBL
Continuous
Average percent cover of broadleaf emergent vegetation
Aquatic attached floating
aquFloatAttach
Continuous
Average percent cover of short attached floating
vegetation
Aquatic free floating
aquFloatFree
Continuous
Average percent cover of short free floating vegetation
Aquatic submerged
aquSub
Continuous
Average percent cover of submerged aquatic vegetation
Aquatic cover
aquCover
Continuous
Average total percent of water covered by all aquatic
vegetation combined
Aquatic layers/complexity
aquLayers
Continuous;
discrete
Count of aquatic vegetation types present at a site in a
season.
pH
wPH
Continuous
Average water pH at 1 m from water’s edge
Temperature
wTemp
Continuous
Average water temperature at 1 m from water’s edge
Conductivity
wCond
Continuous
Average water conductivity at 1 m from water’s edge
Turbidity
wTurb
Continuous
Average water turbidity at 1 m from water’s edge
Broad
Fine
170 | Flow dependent ecological responses
RESULTS
For all four species, C. parinsignifera, L. tasmaniensis, L. peronii, and L. fletcheri, occupancy
modelling was likely to be reliable and produce unbiased results using the available data.
There were clear differences in detection probability estimates between regions (Figure 1)
which supports the inclusion of region as a survey covariate in all models. AIC values for
complex models were lower than those for simple models, implying an advantage to using
more than one variable to model occupancy for these species
Figure 1 Estimated detection probabilities (± 95% credibility intervals) (n=153) for six regions and four frog
species. *indicated species were always detected, resulting in a detection probability of 1 for which
credibility intervals cannot be calculated.
Of the simple models developed for Limnodynastes tasmaniensis, the categorical variables
of connection type and system type ranked highly, with some aquatic and fringing
vegetation measures. The best complex model (lowest AIC value) included the variables
aquatic vegetation layers and connection type (Table 2). Nearly all of the best candidate
models contained the aquatic layers variable. For all complex models, p-values were ≤ 0.05,
evidence of lack of fit.
Of the simple models developed for Litoria peronii, broad-scale categorical variables ranked
highly, specifically including, connection type, system type, hydrology, surrounding land use,
and 1 in 10 year flood return proportion. All the complex models listed had p-values > 0.05,
indicating adequate fit. Many of the categorical variables were highly correlated, limiting
the options for complex model development because the best individual predictors were
categorical. The p-value for two of the best models was 1.0, indicating overfitting and so
these models were excluded from further consideration. The best viable complex model
included four variables: system type, surrounding land use, water turbidity, and aquatic
layers (Table 2). All of the top ranking complex models included surrounding land use and
either system type or connection type.
Flow dependent ecological responses | 171
Fine-scale variables associated with aquatic vegetation and water quality ranked highly for
Limnodynastes fletcheri in simple models. A positive relationship between predicted
occupancy and short emergent, tall emergent and floating attached vegetation was evident,
although variation was only present at a limited portion of the total range and credibility
intervals were large. Some broad variables ranked (Table 2) including surrounding land use,
connection type, and flood return period, but the rankings indicated they were not as
important as the fine-scale variables. The fine-scale variables of aquatic short emergent
vegetation, aquatic vegetation layers count, and water pH all featured consistently in the
best complex models for L. fletcheri. The categorical variables of connection type and
surrounding land use also improved models. There was evidence of lack of fit for several of
the best complex models (p-value < 0.05).
For Crinia parinsignifera, the broad system variables of connection type and system type
were the most informative of the simple models. There was evidence of lack of for all
complex models (p-values were ≤ 0.05). Nonetheless, the variables present in the highranked complex models included connection type, aquatic layers, flood return, and aquatic
free-floating vegetation cover (Table 2).
Table 2 Top 3 Complex (> 1 site covariate) models ranked from best to worst according to AIC. Model fit
assessed using Goodness of fit test (Mackenzie and Bailey 2004), modified Pearson’s Chi-square statistic; p
< 0.05 indicates lack of fit. Refer to Table 1 for description of variables.
SPECIES
MODEL ( SEE TABLE 1 )
AIC
∆AIC
AIC
WEIGHT
MODEL
LIKELIHOOD
NO.
VARIABLES
-2*
LOGLIKE
MODEL
FIT
C.
parinsignifera
659.44
0.00
0.5336
1.0000
11
637.44
0.0099
psi(aqulayers, conn,
aqufloatfree), p(reg)
660.81
1.37
0.2690
0.5041
12
636.81
0.0099
psi(type, fr1, aqulayers), p(reg)
662.44
3.00
0.1191
0.2231
16
630.44
0.0297
psi(conn, aqufloatattach), p(reg)
654.76
0.00
0.2717
1.0000
11
632.76
0.0198
psi(conn, aqufloatattach,
frishortherb), p(reg)
654.91
0.15
0.2521
0.9277
12
630.91
0.0198
psi(conn, aqufloatattach,
fritallherb), p(reg)
655.20
0.44
0.2180
0.8025
12
631.20
0.0099
psi(type, surr, wturb,
aqulayers), p(reg)
491.98
0.00
0.8476
1.0000
21
449.98
0.1584
psi(conn, surr, wturb), p(reg)
495.51
3.53
0.1451
0.1712
16
463.51
0.3069
psi(conn, surr, fr10), p(reg)
502.68
10.70
0.0040
0.0047
16
470.68
0.1386
psi(aqushortem, wph, surr,
aqubl), p(reg)
471.15
0.00
0.5164
1.0000
15
441.15
0.0099
psi(aqushortem, wph, surr),
p(reg)
471.36
0.21
0.4649
0.9003
14
443.36
0.0099
psi(aqushortem, wph, conn),
p(reg)
479.83
8.68
0.0067
0.0130
12
455.83
0.0693
L. fletcheri
L. peronii
psi(aqulayers, conn), p(reg)
L. tasmaniensis
P-VALUE
172 | Flow dependent ecological responses
DISCUSSION
Robust models were developed for Litoria peronii and Limnodynastes fletcheri, but we were
less successful in developing models for two other common and widespread basin species,
Limnodynastes tasmaniensis and Crinia parinsignifera. Model development for these latter
species was difficult due to their broad habitat requirements and apparent insensitivity to
the hydrological and biophysical variables included, which meant that key drivers of their
distributions could not be described with confidence.
Broad scale hydrological variables influenced occupancy for the species modelled. At fine
scales, vegetation variables were also important, consistent with previous studies (Wassens
et al. 2010a). The predicted relationships with habitat variables were weaker at the basinwide scale than at fine scales, suggesting that region-specific occupancy relationships
cannot be applied to other regions reliably.
For all four modelled species, the key finding was that regional differences in detection
probability far outweighed the influence of site-specific habitat variables. These regional
differences suggest that Basin level assessment of frog communities and their expected
response to flow will be more robust if they incorporate regional variability. For frogs at
least, the Murray–Darling Basin is not homogeneous, so broad-scale management of the
basin should also incorporate region-specific frog habitat management, if species are to be
effectively conserved.
Clear relationships were revealed between predicted occupancy and measured variables,
while broad scale hydrological variables were generally not as important as fine scale
habitat characteristics, flow plays aan import role in shaping habitat at finer scales.The value
of the broad scale flood return classifications could have been improved if complimented by
fine-scale hydrological data for each region. For example, models to predict species’
distribution and persistence with respect to hydrological composition of floodplains provide
opportunities for modelling across the Lowbidgee floodplain.
Flow dependent ecological responses | 173
REFERENCES
Bailey LL, Simons TR and Pollock KH (2004) Estimating site occupancy and species detection
probability parameters for terrestrial salamanders. Ecological Applications, 14, 692702, doi:10.1890/03-5012
Baldwin RF, Calhoun AJ K. and deMaynadier PG (2006) The significance of hydroperiod and
stand maturity for pool-breeding amphibians in forested landscapes. Canadian Journal
of Zoology, 84, 1604-1615, doi:10.1139/z06-146
Griese V (2011) Responses of frogs to changing patterns of habitat availability during
flooding in two regulated streams: Honours Dissertation. Charles Sturt University.
Gulve PS (1994) Distribution and extinction patterns within a northern metapopulation of
the pool frog, Rana Lessonae. Ecology, 75, 1357, doi:10.2307/1937460
Hazell D, Cunningham R, Lindenmayer D, Mackey B and Osborne W (2001) Use of farm dams
as frog habitat in an Australian agricultural landscape: factors affecting species
richness and distribution. Biological Conservation, 102, 155-169, doi:10.1016/S00063207(01)00096-9
Hazell D, Hero J.-M, Lindenmayer D and Cunningham R (2004) A comparison of constructed
and natural habitat for frog conservation in an Australian agricultural landscape.
Biological Conservation, 119, 61-71, doi:10.1016/j.biocon.2003.10.022
Healey M, Thompson D and Robertson A (1997) Amphibian communities associated with
billabong habitats on the Murrumbidgee floodplain, Australia. Austral Ecology, 22,
270-278, doi:10.1111/j.1442-9993.1997.tb00672.x
Jakob C, Poizat G, Veith M, Seitz A and Crivelli AJ (2003) Breeding phenology and larval
distribution of amphibians in a Mediterranean pond network with unpredictable
hydrology. Hydrobiologia, 499, 51-61, doi:10.1023/A:1026343618150
Jansen A and Healey M (2003) Frog communities and wetland condition: relationships with
grazing by domestic livestock along an Australian floodplain river. Biological
Conservation, 109, 207-219, doi:10.1016/S0006-3207(02)00148-9
Kolozsvary MB and Swihart RK (1999) Habitat fragmentation and the distribution of
amphibians: Patch and landscape correlates in farmland. Canadian Journal of Zoology,
77, 1288-1299, doi:10.1139/z99-102
Lane S, Hamer A and Mahony M (2007) Habitat correlates of five amphibian species and of
species-richness in a wetland system in New South Wales, Australia. Applied
Herpetology, 4, 65-82, doi:10.1163/157075407779766688
MacKenzie D. I. and Bailey L. L. (2004) Assessing the fit of site-occupancy models. Journal of
Agricultural, Biological, and Environmental Statistics, 9, 300-318,
doi:10.1198/108571104X3361
MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey LL & Hines JE (2006) Occupancy
estimation and modeling. Inferring patterns and dynamics of species occurrence.
Burlington, MA: Academic Press.
174 | Flow dependent ecological responses
MacKenzie DI, Nichols JD, Sutton N, Kawanishi K and Bailey LL (2005) Improving inferences
in population studies of rare species that are detected imperfectly. Ecology, 86, 11011113, doi:10.1890/04-1060
McKibbin R, Dushenko WT, vanAggelen G and Bishop CA (2008) The influence of water
quality on the embryonic survivorship of the Oregon spotted frog (Rana pretiosa) in
British Columbia, Canada. Science of The Total Environment, 395, 28-40,
doi:10.1016/j.scitotenv.2008.01.050
Poff NL and Ward JV (1989) Implications of streamflow variability and predictability for lotic
community structure: A regional analysis of streamflow patterns. Canadian Journal of
Fisheries and Aquatic Sciences, 46, 1805-1818, doi:10.1139/f89-228
Poff NL and Zimmerman JKH (2010) Ecological responses to altered flow regimes: a
literature review to inform the science and management of environmental flows.
Freshwater Biology, 55, 194-205, doi:10.1111/j.1365-2427.2009.02272.x
Rubbo MJ and Kiesecker JM (2005) Amphibian breeding distribution in an urbanized
Landscape. Conservation Biology, 19, 504-511, doi:10.1111/j.15231739.2005.000101.x
Spencer C, Robertson A and Curtis A (1998) Development and testing of a rapid appraisal
wetland condition index in south-eastern Australia. Journal of Environmental
Management, 54, 143-159,
Spencer JA, Thomas R, Wassens S, Lu Y, Wen L, Iles J, Kobayashi Y, Hunter S, Ling J and
Saintilan N (2011) Environmental flow monitoring in the Lowbidgee wetlands and
Macquarie Marshes in 2009-10. Testing wetland resilience: monitoring the response
of iconic wetlands to re-flooding following historic drought project. Sydney, NSW,
Australia: NSW Department of Environment, Climate Change and Water.
Spencer JA and Wassens S (2009) Responses of waterbirds, fish and frogs to environmental
flows in the Lowbidgee wetlands in 2008–09. Final report for the NSW Rivers
Environmental Restoration Program. Sydney, NSW, Australia: Rivers and Wetland Unit,
NSW Department of Environment and Climate Change.
Vos CC and Stumpel AHP (1996) Comparison of habitat-isolation parameters in relation to
fragmented distribution patterns in the tree frog (Hyla arborea). Landscape Ecology,
11, 203-214, doi:10.1007/BF02071811
Ward D, Tews K, Kennard M, Peterson E and Bond N (2012) Murray–Darling Basin aquatic
system mapping and classification—Aquatic assets and environmental attribute
geodatabases. Report for CSIRO–Cluster project “Ecological Responses to Altered Flow
Regimes”: CSIRO.
Wassens S and Amos C (2011) Assessing the resilience of frog communities within River Red
Gum forest: Report to New South Wales Office of Environment and Heritage. Albury,
NSW, Australia: Institute for Land, Water and Society.
Wassens S, Hall A, Osborne W and Watts RJ (2010a) Habitat characteristics predict
occupancy patterns of the endangered amphibian Litoria raniformis in flow-regulated
flood plain wetlands. Austral Ecology, 35, 944-955, doi:10.1111/j.14429993.2010.02106.x
Flow dependent ecological responses | 175
Wassens S, Healey S and Watts R (2010b) Evaluation of landscape and patch metrics to
predict the response of frog populations following flooding. Report to NSW Office of
Environment and Heritage. Albury, NSW, Australia: Institute for Land, Water and
Society.
Wassens S and Maher M (2011) River regulation influences the composition and distribution
of inland frog communities. River Research and Applications, 27, 238-246,
doi:10.1002/rra.1347
Wassens S, Walcott A, Wilson A and Freire R (2011) Frog breeding in rain-fed wetlands after
a period of severe drought: implications for predicting the impacts of climate change.
Hydrobiologia, 1-12, doi:10.1007/s10750-011-0955-2
Wassens S, Watts RJ, Spencer JA, Howitt J, McCasker NA, Griese V, Burns A, Croft R, Zander
A, Amos C and Hall A (2012) Monitoring of ecosystem responses to the delivery of
environmental water in the Murrumbidgee system, pp. 1-123. Albury, NSW, Australia:
Institute for Land, Water and Society.
Weyrauch SL and Grubb TC (2004) Patch and landscape characteristics associated with the
distribution of woodland amphibians in an agricultural fragmented landscape: an
information-theoretic approach. Biological Conservation, 115, 443-450,
Wilbur HM (1980) Complex life cycles. Annual Review of Ecology and Systematics, 11, 67-93.
176 | Flow dependent ecological responses
2.16
The role of antecedent conditions in determining responses
to environmental flows
Thompson, R.1, Kingsford, R2. and King, A.J.3,4
1
Institute for Applied Ecology, University of Canberra - Canberra, Australia
Centre for Ecosystem Science, School of Biological, Earth and Environmental Sciences, University of
New South Wales
3
Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment,
Heidelberg, Vic., Australia
4
Research Institute for Environment and Livelihoods, Charles Darwin University, Darwin, NT,
Australia
2
INTRODUCTION
There are considerable challenges in understanding and managing flows to ecosystems,
given the background of impacts to natural flow regimes. Natural flow regimes in rivers are
highly variable, particularly in Australia, where sequences of floods are critical in
determining ecosystem responses (Leigh et al. 2010). For example, flows of rivers in the
Lake Eyre Basin variably reach different parts of the river system, often depending on
whether a series of floods has provided the requisite amount of water to reach downstream
rivers channels and floodplains (Kingsford et al. 1999; Puckridge et al. 2000).
Correspondingly, antecedent dry periods are critically important in determining the effects
of a particular flow event on in-stream and floodplain ecosystems. In regulated rivers, there
are managed environmental flows which represent part of the flow regime and are often
targeted at particular ecological outcomes. Ecological outcomes can be diverse (e.g. bird
numbers, fish condition, ecosystem respiration) and we will describe responses generically
in this section. We will also use the term ‘ecological condition’ as an indicator of the
underlying state of an ecosystem, with systems in poor condition less able to show positive
ecological responses to watering events.
Early research on environmental flow research tended to ignore the role of temporal and
spatial antecedent conditions as well as interactions between flows and other stressors in
influencing ecological condition (Postel and Richter 2003; Poff et al. 1997). Ignoring these
factors is unwise for management and research on the ecological responses to flows, given
their importance in dictating ecological condition. Understanding how flow responses vary
with antecedent conditions can help water managers understand their relative importance
of their environmental flow management in delivering on ecological outcomes, including
when to capitalise on previous conditions. There are also the often intractable challenges of
dealing with the confounding factors affecting ecological responses. For example, native fish
populations in the Murray-Darling Basin have declined significantly over time with
regulation of flows identified as a major stressor (Gehrke et al. 1995) but many other
Flow dependent ecological responses | 177
stressors confound this impact in different ways including exotic species, barriers and
harvesting. Ultimately, an environmental flow manager may have a target to improve
native fish population but fail because of these other factors. Understanding long-term
trends in condition and the relationship to non-flow related stressors can provide water
managers with critical information on where and when to prioritise environmental flow
delivery, and what level of response magnitude may be expected.
Provision of environmental flows is best considered in the context of restoration ecology,
which seeks to reverse habitat degradation through active intervention (Jordan 1994).
Restoration seeks to take biodiversity values and/or ecosystem functions (such as
productivity) back towards the levels found in the non-degraded ecosystem (Hobbs and
Norton 1996). Increasingly restoration ecology as a discipline is recognising the difficulties of
defining restoration end-points and in determining efficacy against natural background
variability in condition, and the impacts of unrelated changes such as natural or
anthropogenic climate change (Pickett and Parker 1994; Hey and Philippi 1995). Similarly, it
is increasingly clear that restoration in a river management context requires a landscapescale, rather than site-scale, approach (George et al. 1995). Hobbs and Norton (1996)
summarised the broad conceptual basis of restoration ecology and recognised the
importance of determining all degrading processes and assessing which of these
management interventions are addressing. This is particularly important in systems which
are subject to multiple degrading processes.
The provision of environmental flows occurs against a background of short and medium
term trends in flow regime and ecological condition which naturally differ between different
ecosystems, and which usually interact with flow regimes. The attributes of a particular flow
event (e.g. timing, duration, frequency, magnitude) will also influence the timing and scale
of ecological responses. The intention of this section is to discuss the ways in which
antecedent conditions may influence the potential of particular ecological systems to
respond to environmental flows at two time scales: short (within year) and medium (years).
We aim to provide information on the likely importance of antecedent conditions on likely
response to environmental flows. We also discuss the role of trends in condition which
result from non-flow related stresses (legacies) which may interact with flow impacts.
THE INITIAL CONDITION AND NATURAL VARIABILITY
Ecological responses to environmental flows will depend on an interaction between the
type of ecosystem to which they are being applied and the initial condition of that
ecosystem. The natural resilience of ecosystems, which is a function of historical water
regimes (reflecting climate), over evolutionary time will determine how sensitive systems
are to drying over various timescales, and how responsive they will be to environmental
flows after dry periods. Many Australian ecosystems are highly adapted to prolonged dry
178 | Flow dependent ecological responses
spells with intermittent water. Systems which have evolved with long term drying may still
be in good condition and able to respond to environmental watering in situations even
when antecedent conditions are very dry (Figure 1). Systems that have evolved with regular
wetting are likely to lose ecological condition as the length of the dry period increases, and
be less able to respond to watering (Figure 1). At an extreme, naturally perennial wetlands,
waterholes or streams which have dried completely will respond differently to the provision
of flows than will systems that have retained some water. For example, a wetland system
adapted to long dry periods with occasional wet periods may respond to the provision of
environmental flows with very large ecological responses that rapidly restore condition.
Systems which are adapted to more regular watering may show large ecological responses
to water if they are already in good condition due to previous watering events, but much
smaller and slower responses if environmental water is provided after unusually long dry
periods. In studying responses of riverine and wetland ecosystems to environmental
watering it is important to take into account natural spatial variability in environmental
conditions as a factor in predicting likely ecological responses to water.
Figure 1. Responses to environmental flows depending on the natural flow characteristics of the ecosystem
and sensitivity to the length of time since last flows.
Flow dependent ecological responses | 179
There may be a range of different conditions for any particular site which will be dependent
on the site’s natural flow regime and effects of different stressors, including river regulation.
For river regulation, sites may be broadly categorised into those where the flow regime is
considerably reduced leading to increased drying and a few other areas where water is kept
artificially and often constantly high (Kingsford 2000; Kingsford et al. 2004). Sites where
there is decreased flooding compared to natural (Kingsford and Thomas 1995, 2004) are
likely to be in poor ecological condition and when provided with environmental flows may
show a slow response, compared to sites where this impact is not so dramatic (Figure 2).
Importantly, this will likely also affect the effects of the flow in terms of duration and
magnitude of flooding. Parched soils will take considerably longer and require more flow to
become saturated and inundated. The actual magnitude of the response will also be
influenced by initial ecological condition. Sites in good condition may have limited capacity
to improve further, but can maintain their condition (Figure 2a). Sites in poor condition may
respond slowly, but may have greater capacity to show improvement in condition (Figure
2b). Sites in very poor condition, compared to natural, may be very slow and have limited
capacity to respond at annual time scales (Figure 2c). Responses will also vary in terms of life
history. A key aim of restoration should be to establish self-sustaining populations which can
grow, survive, reproduce and recruit. It will not be sufficient to simply maintain except if
managing for particular dry periods. The actual nature of these relationships will depend on
the traits of the constituent biota, the landscape context (e.g. relationships to nearby
refugia to provide colonists) and the time scales involved (Bond et al. 2012). Environmental
flows should provide the greatest responses in ecological condition where sites are in poor
or declining condition, relative to natural but extreme dry conditions.
180 | Flow dependent ecological responses
Figure 2. Responses to environmental flows depending on antecedent condition of the ecological
community when the flow is delivered.
THE ROLE OF SHORT TERM DISTURBANCE HISTORY
Effects of recent flow conditions (within the year), in determining the condition of ecological
communities, and thus their potential to respond to environmental flows are poorly known.
Flows supplemented by a high degree of local variability in terms of flow and cease-to-flow
events explained regional ecological condition of macroinvertebrate communities in
Victorian streams (Bond et al. 2012). The recent history of cease-to-flow events strongly
causes distribution of macroinvertebrate taxa to decline (Boulton and Lake 1992; Chessman
2009; Bond et al. 2012). In delivering environmental flows, it is critical to know the natural
flooding regimes of sites to understand any response to environmental flows.
Flow dependent ecological responses | 181
High flow disturbance similarly may negatively influence community response to
environmental flows at one point but be positively impacted downstream. Sites in poor
ecological condition may respond slowly to environmental flows (similar to Figure 2b).
Alternately, sites which have been subjected to high flows and may have responded to
those flows may have limited capacity to respond further to any additional flows (similar to
Figure 2a). Thus it is important to have an understanding of recent flow history in predicting
likely responses to environmental flows.
THE ROLE OF MEDIUM TERM DISTURBANCE HISTORY
Short term event-based antecedent conditions such as drying may be nested within medium
term (years) flow trends that impact on the initial condition of communities and their
potential to respond to environmental flows. Over the period 2000-2010, south eastern
Australia was subjected to intensifying drying due to the Millenium drought. This
contributed to long-term declines in condition of freshwater ecosystems across the region
(Chessman et al. 2009; Kingsford et al. 2011; Thomson et al. 2012) until the drought broke
with extremely wet years in 2010 and 2011. This backdrop of increasing flow stress (Bond et
al. 2012) may have affected the potential for communities at a location to respond to the
provision of environmental flows. Hypothetically, a site one year into the drought may have
remained in relatively good condition with potential to respond to the provision of flows
(Figure 2a), but as the drought progressed, condition and potential to respond to flows may
have reduced (moving from Figure 2b to 2c). The time lags involved in these declines in
ability to respond are likely to be taxa dependent, and are currently little understood.
In providing environmental flows against a backdrop of downwards trending condition due
to medium term factors such as drought or climate-change induced drying, managers need
to be aware of the rate of decline in condition of target taxa, and the ability of target taxa to
respond to provision of flows. Very resilient taxa (such as short lived invertebrates) may
require the provision of single flows at relatively long intervals to restore condition to levels
which are consistent with those that may be predicted in the absence of water abstraction.
However, long-lived, slow responding taxa such as River Red Gums may need provision of
water more regularly.
THE ROLE OF LONG TERM DISTURBANCES
The provision of environmental flows against a background trend of drying is particularly
challenging, but represents an important challenge for environmental flow management.
Long term changes in ecological condition as a result of drying are will occur at scales of
decades and more. Aridification of south-eastern Australia has been occurred over the last
several thousand years, more recently compounded by additional drying generated by
climate change. The ecological response to environmental flows will depend on the
182 | Flow dependent ecological responses
interaction between the rate of decline in ecological condition due to medium term drying
trends, the rate of decline due to anthropogenic causes such as abstraction, and the ability
of taxa to respond to any provision of environmental water. Systems where ecological
condition would be declining due to medium term trends in climate, may respond to the
provision of water and thus on average, exhibit similar trends to those that might have been
expected in the absence of water abstractions (Figure 3a). However, systems where trends
in condition are very rapid and responses of biota to environmental flows are relatively
small or slow may exhibit steep declines in condition even when environmental flows are
provided (Figure 3b).
Figure 3. Responses to environmental flows depending on background trends in condition caused by
medium term (years) processes such as drought.
Similar patterns may be expected for other, long term trends in stressors, even when they
do not directly influence flow. Land use impacts on stream ecosystems can be pervasive and
play out over many decades (Thomson et al. 2012). These existing stressors can result in
biota which have less capacity to respond to changes in hydrologic regimes (Palmer et al.
2008; Mac Nally et al. 2011). Land use change may also be having long term negative
Flow dependent ecological responses | 183
impacts on ecological condition similar to those seen due to drying in Figure 3a, and with
similar effects on responses to provision of flows. For example in central Victoria, sand slugs
as a result of land clearance reduce the potential for stream biota to respond to flows, even
if a fully natural flow regime was to be restored (Atkinson et al. 2008). This is made more
complex by the fact the responses of many taxa are lagged by varying degrees from months
to decades, and that these responses are also influenced by superimposed lagged responses
to a variety of other stressors. Recent studies of urban streams have shown that biota
responses to urbanisation of agricultural land are strongly influenced by lagged responses to
initial land clearance from forested to agricultural landscapes, even though this took place
decades earlier (Roy et al. 2003). In an environmental flows context, responses to flow
provision will occur against a shifting baseline of condition responses to a variety of
stressors that have occurred over decades.
CONCLUSION
Assessing the effects of environmental flows is difficult because responses are likely to be
contingent on antecedent flow conditions, and play out on a complex landscape of
background trends, legacies from past impacts and with lagged responses by some biota.
However, sophisticated analyses are now capable to disentangling these effects (e.g.
Thomson et al. 2012; Bond et al. 2012), but it is important to note that flow regimes do not
operate in isolation from long-term and large scale effects of other impacts. The responses
to environmental flows described in the remainder of this report occur against a range of
hydrologic backgrounds from relatively wet (for studies which took place after the
exceptionally wet years in 2010/2011), to extremely dry (for studies which occurred during
the Millenium Drought. Studies where environmental flows were provided against a
hydrologic background of natural high flows may have reduced the total magnitude of the
responses seen, but may also have ‘primed’ natural systems to be able to respond to
environmental flow provision. Environmental flow responses need to be considered in
terms of the ability of taxa to respond, which is a product of the traits of the target taxa and
the antecedent flow conditions that precede the environmental flow. In addition, it is
important to recognise that longer term trends in ecological condition may underlie and
even obscure effects of environmental flows in some contexts.
REFERENCES
Atkinson, B.L., Grace, M.R., Hart, B.T., and Vanderkruk, K.E.N. (2008) Sediment instability
affects the rate and location of primary production and respiration in a sand-bed
stream. Journal of the North American Benthological Society 27: 581-592.
Bond, N., Thomson, J. and Reich, P. (2012) Macroinvertebrate responses to antecedent flow,
longterm flow regime characteristics and landscape context in Victorian rivers,
National Water Commission, Canberra.
184 | Flow dependent ecological responses
Boulton, A.J. and Lake, P.S. (1992) The ecology of two intermittent streams in Victoria,
Australia, III, temporal changes in faunal composition’, Freshwater Biology 27: 123–
138.
Chessman, B.C. (2009) Climatic changes and 13-year trends in stream macroinvertebrate
assemblages in New South Wales, Australia. Global Change Biology 15: 2791–2802
Gehrke, P.C., Brown, P., Schiller, C.B., Moffatt, D.B. and Bruce, A.M. (1995) River regulation
and fish communities in the Murray-Darling System, Australia. Regulated Rivers:
Research and Management 11: 363–375.
George, R.J., McFarlane, D.J. and Speed, R.S. (1995) The consequences of a changing
hydrological environment for native vegetation in South-western Australia. In: The
Role of Networks- Nature Conservation 4 Edited by: D.A. Saunders, J.L. Craig and
E.M. Mattiske. 9-21 CSIRO Sydney: Surrey-Beatty, Sydney.
Hey, D.L. and Philippi, N.S. (1999) A Case for Wetland Restoration. Wiley-Interscience. 215 p.
Hobbs, R.J. and Norton, D.A. (1996) Towards a conceptual framework for restoration
ecology. Restoration Ecology 4: 93–110
Jordan III, W.R. (2003) The Sunflower Forest: Ecological Restoration and the New
Communion with Nature. University of California Press. 264p
King, J., Brown, C. and Sabet, H. (2003) A scenario-based holistic approach to environmental
flow assessments for rivers. River Research and Applications 19: 619-639.
Kingsford, R.T. (2000) Ecological impacts of dams, water diversions and river management
on floodplain wetlands in Australia. Austral Ecology 25: 109 – 127
Kingsford, R.T. (2011) Conservation management of rivers and wetlands under climate
change - a synthesis. Marine and Freshwater Research 62: 217 - 217
Kingsford, R.T., Curtin, A.L., Porter, J.L. (1999) Water flows on Cooper Creek determine
‘boom’ and ‘bust’ periods for waterbirds of the Paroo and Warrego Rivers. Biological
Conservation 88: 231-48.
Kingsford, R.T., Jenkins, K.M., Porter, J.L. (2004) Imposed hydrological stability on lakes in
arid Australia and effects on waterbirds. Ecology 85: 2478 – 2492.
Kingsford, R.T. and Thomas, R.F. (1995) The Macquarie Marshes and its waterbirds in arid
Australia: A 50-year history of decline. Environmental Management 19: 867-878.
Kingsford, R.T. and Thomas, R.F. (2004) Destruction of wetlands and waterbird populations
by dams and irrigation on the Murrumbidgee River in arid Australia. Environmental
Management 34: 383-396.
Leigh, C., Sheldon, F., Kingsford, R.T., Arthington, A.H. (2010) Sequential floods drive 'booms'
and wetland persistence in dryland rivers: a synthesis. Marine and Freshwater
Research 61: 896-908.
Mac Nally, R., Cunningham, S.C., Baker, P.J., Horner, G.J. and Thomson, J.R. (2011) Dynamics
of Murray-Darling floodplain forests under multiple water stressors - the past,
present and future of an Australian icon. Water Resources Research, 47.
Flow dependent ecological responses | 185
Palmer, M.A., Liermann, C.A.R., Nilsson, C., Florke, M., Alcamo, J., Lake, P.S. and Bond, N.
(2008) Climate change and the world’s river basins, anticipating management
options. Frontiers in Ecology and the Environment, 6, 81–89.
Pickett, S.T.A and Parker, V.T. (1994) Avoiding the Old Pitfalls: Opportunities in a New
Discipline Restoration Ecology 2: 75–79.
Poff, N.L., Allan, J.D., Bain, M.B., Karr, J.R., Prestegaard, K.L., Richter, B.D., Sparks, R.E. and
Stromberg, J.C. (1997) The natural flow regime – a paradigm for river conservation
and restoration. Bioscience 47: 769-784.
Postel, S. and Richter, B. (2003) Rivers of life: Managing water for people and nature.
Washington: Island Press.
Puckridge, J.T., Walker, K.F. and Costelloe, J.F. (2000) Hydrological persistence and the
ecology of dryland rivers. Regulated Rivers: Research & Management 16: 385–
402.Roy, A.H., Rosemond, A.D.,Paul, M.J., Leigh, D.S. and Wallace, J.B. (2003).
Stream macroinvertebrate response to catchment urbanisation. Freshwater Biology,
48, 329 – 346.
Thomson, J.R., Thompson, R.M., Metzeling, L., Reich, P., Bond, N.R., Cunningham, S.C. and
Mac Nally, R. (2012) The influence of vegetation, flows and climate variation on
stream biota: Lessons from the Big Dry in southern Australia. Global Change Biology
18: 1582-1596
186 | Flow dependent ecological responses
3
Summary of research implications
Assessing the responses of ecosystems to in-channel flows can be difficult because there is
limited understanding of flow-ecology relationships for many aquatic species and for
important ecosystem processes a relatively small number of examples studies of in-channel
pulsed flows (Watts et al. 2009). In addition, in-channel flow pulses can be short-lived and
responses to these flows are likely to be influenced by antecedent flow conditions and past
impacts, and there may be lagged responses by some biota (see Section 2). However
controlled field-based studies of environmental flows can account for these complexities
and contribute to improving predictive models.
Field-based experiments assessing ecosystem responses to in-channel flows as well as
analysis and synthesis of existing datasets were undertaken as part of this subproject. We
sought to improve our understanding of ecosystem responses to individual flow events and
improve the capacity to predict the outcomes of environmental flows within in-channel
ecosystems. However, important knowledge gaps remain, and our research has raised
further questions and identified some areas for future research. Addressing some of these
knowledge gaps will help us improve our understanding of responses to in-channel
environmental flows and will assist managers better use environmental water to help
restore riverine ecosystems.
CHANGES IN BENTHIC WETTED AREA AND AVAILABILITY OF SLACKWATER IN
RESPONSE TO IN-CHANNEL FLOWS
We investigated lateral connectivity and availability of slackwater habitat in the EdwardWakool system during in-channel flows using hydrodynamic modelling (Section 5.2). Many
different organisms depend on these habitats. We showed that the relationship between
discharge and wetted surface area and the area of slackwater in rivers in this system was
not linear and most likely strongly influenced by geomorphology. It may be more
appropriate to use area of inundation, instead of daily discharge, to better understand the
relationship between in-channel flows and ecosystem responses. Future research could
focus on examining the nature of this relationship in other river systems to improve models
for predicting the responses of in-channel flows on slackwater dependent biota and
ecosystem functions.
WHOLE STREAM METABOLISM
Our investigation of whole instream metabolism in the Edward-Wakool system (Section 5.3)
demonstrated that small managed flow pulses did not stimulate primary production. This
clearly raised the possibility that larger flows may be needed to test the hypothesis that
more inundation of riverbanks and backwaters may provide greater stimulation of primary
Flow dependent ecological responses | 187
production and respiration. This will increase the biomass of primary producers and provide
important food resources for high trophic levels, including macroinvertebrates and fish.
ORGANIC MATTER BREAKDOWN
There were generally low organic matter (leaf-litter) breakdown rates in relation to changes
in flow regime among four rivers in the Edward-Wakool system (Section 5.4). This implies
that such rates may not be a sensitive indicator for assessing environmental watering,
within in-channel ecosystems where shredder consumers are in low densities. This may not
be a general finding across other river systems. Also, other mechanisms may possibly alter
breakdown rates.
INVERTEBRATE RESPONSE TO IN-CHANNEL FLOW PULSES
We also investigated responses of macroinvertebrate biomass in the Edward-Wakool
system (Section 5.5) following delivery of small environmental flows that occurred after
large magnitude unregulated flow pulses. Zooplankton abundance, biomass and richness
were highest directly after large unregulated flow pulses, whereas small environmental flow
events alone did not initiate zooplankton response, possibly due to minimal inundation of
riverbanks during these small events (see section 5.2). Increased macroinvertebrate
biomass, following unregulated events, provided indirect evidence of benefits for higher
trophic levels, such as fish and frogs. The interlinkages among components of the foodweb
in response to environmental flows require further investigation.
FISH SPAWNING
We investigated fish spawning in two separate studies. In the Edward-Wakool river system,
opportunistic spawning and recruitment of carp gudgeon occurred during and after inchannel environmental flow events (Section 5.6). However, the timing of spawning for
Murray cod, Australian smelt, common carp and native floodplain-spawning species was
independent of these flow events. Widespread abundance of opportunistic spawning
species, such as carp gudgeon, may reflect their ability to successfully spawn and recruit
under variable in-channel flow conditions. In the Murray River at Barmah-Millewa Forest,
temperature was the dominant and overriding environmental parameter influencing
spawning occurrence for all four species (Murray cod, trout cod, silver perch and golden
perch) , although attributes of flow were also important and in particular affected the
spawning intensity of the latter three species (Section 5.7). Our studies demonstrated the
significance of timing of water application in the design and management of environmental
flows and could be used to create a stronger conceptual or analytical framework for flow
management for improved conditions for native fish spawning.
188 | Flow dependent ecological responses
PREDICTING THE CONDITION OF YOUNG-OF-YEAR FISH
We also retrospectively investigated juvenile body condition of two small-bodied native
species (Australian smelt and unspecked hardyhead ) in the mid-Murray River, using six
years of data with differing antecedent flow conditions (Section 5.8). Juveniles have limited
capacity to store energy and do not use it for reproduction, and so their condition probably
reflects current system energetics. This study provided insight into the longevity of flowrelated energy pulses, rather than the capacity of organisms to store energy from previous
energy pulses. There was preliminary evidence that antecedent flow conditions altered
system energetics and fish condition, The importance of flows a year before fish were
spawned, suggests that antecedent conditions may affect the energetics of the system in
excess of a year later. The relationship between antecedent flow and juvenile fish body
condition, however, differed markedly between the two species studied. For Australian
smelt, body condition was positively related to short-term flow and negatively related to
long-term flow, while for unspecked hardyhead, body condition was unrelated to short-term
flow, and negatively (but mildly) related to long-term flow. The different responses among
species with similar life histories and habitat preferences, suggests it is unwise to make
generalisations about how the body condition of different species will respond to flow. As
we relied on a relatively small dataset with limited capacity to detect relationships and how
different components of flow interacted, we caution against extrapolation of these findings
and recommend additional investigation in this area to further determine how such
sequencing of flow may benefit native fishes.
HYPOXIC BLACKWATER FISH KILLS
We also focused on the role of hypoxia and tolerances of juvenile stages of four lowland
river fishes native to the Murray-Darling Basin (Section 5.9). Despite the natural occurrence
of hypoxia and blackwater in lowland rivers of the Murray-Darling Basin, large bodied
predatory fish in this ecosystem are vulnerable to acute mortality induced by low oxygen
concentration and associated water chemistry changes. Logistic models predicted that
hypoxic blackwater mortalities of golden perch, silver perch, Murray cod and freshwater
catfish may occur at oxygen concentrations as high as 4.54 mg L-1 and sub-lethal stress, as
indicated by onset of aquatic surface respiration, may occur above 5.00 mg L-1 at 25–26 °C.
The thresholds reported in our study provided a starting point for managers to develop
hypoxia mitigation strategies but must be carefully interpreted, given potential sub-lethal
effects, and ambient and antecedent environmental conditions. Consequently, a better
understanding how these variables interact will be essential in further developing hypoxia
management plans to prevent fish kills and the loss of aquatic biodiversity.
Flow dependent ecological responses | 189
FLOW AND INUNDATION MODELS
A key aspect of understanding the effects of flow on ecosystems in wetlands and floodplains
is understanding the relationship between flow regimes and flooding regimes (i.e. height,
frequency, duration, timing). There are few data available for flooding, but classification of
inundation from satellite imagery can be linked to flows through modelling. This can be
enhanced by improving the availability of flow data for flow gauges near wetlands.
Development of integrated flow and flood models allow sub-regional predictions of area
inundated to be made on a daily basis (Sections 5.10 and 5.11). Models from this project
exist for the Macquarie Marshes, Murrumbidgee River and Lower Murrumbidgee which are
helping link to ecosystem responses. Such linking will greatly increase our understanding of
flood regimes and the link to river flows. By linking flood regimes to floodplain-dependent
biota and ecosystem processes should help guide environmental flow management.
WETLAND AND FLOODPLAIN VEGETATION
Variations in seasonal flow timing, duration, depth and habitat type were all found to affect
riparian understory vegetation production and community composition (Section 5.12). The
responses were, however, complex and no single variable (season, flood duration, depth or
habitat type) can be used to predict production or community structure after inundation.
One of the outcomes of the research was that productivity and diversity were not well
correlated with Black Box wetlands supporting the greatest number of species while
wetland habitats supported the greatest increase in plant biomass. This suggests that
ecosystem water requirements to sustain productivity may differ from the water
requirements to sustain biodiversity. The results clearly demonstrate that flow has a
significant influence on species richness both within wetlands and at a landscape scale. This
suggests that achieving environmental flow objectives may require strategies to create a
habitat mosaic to promote both diversity and productivity.
Changes in flooding frequency can shift vegetation communities, ultimately transitioning
them into different communities, consequently affecting their dependent fauna
communities (Section 5.13). Managers of environmental flows need to understand the
individual nature of vegetation communities and their reliance on different flooding regimes
so that they can ensure resilience across the spectrum of floodplain vegetation communities
(e.g. redgum-wetland, black box habitats). This approach may not maximise overall
vegetation productivity, but will increase diversity and ensure productive patches, reflecting
the gradient of flooding regimes within the managed area. Flow timing is also complex and
amenable to management and variation in timing of inundation may increase long-term
species diversity. From a simple productivity perspective, flows in warm months increase
productivity. Flows delivered in cool months, however, may not result in low rates of
190 | Flow dependent ecological responses
production, if the water temperature or soil moisture remains high with the onset of
warmer conditions.
Knowledge of dependencies between vegetation communities and flooding regimes can
also inform management aimed at restoration to a former state, using flooding or
environmental flows. For aquatic and flood-dependent vegetation, seed bank longevity is a
critical limiting factor, with extended dry periods reducing opportunity for recovery. Both
long-term flood regimes and short-term flood events significantly affect vegetation and
woodland bird communities in floodplain woodlands – therefore both must be managed for
short-term and long-term biodiversity outcomes.
Inundation of the whole site may not always be necessary to maintain site value and
biodiversity during droughts, provided flooding occurs with sufficient frequency over long
periods of time. Floodplain woodlands are important productivity refuges from both
drought and land clearing, and the use of environmental water and natural flooding must be
considered to maintain or restore floodplain woodland condition.
ANTECEDENT FLOWS AND THE FISH COMMUNITY
Relatively few studies have attempted to quantify importance of antecedent conditions
which can alter flow-ecology relationships. Our study of six floodplain wetlands inf the midMurray, revealed that antecedent flows are important in determining the structure of the
wetland fish community (Section 5.14). We investigated the relative importance of shortterm (3 month prior), medium-term (3 month to 1 year prior) and long-term (1 to 5 years
prior) antecedent flow conditions on juvenile fish abundance and developed a Bayesian
hierarchical multiple-species model. We observed that native and non-native fishes tended
to differ in their response to antecedent flows, and that the response of some non-native
species to recent flow was affected by prior medium and longer-term conditions. Native fish
generally responded positively to frequent watering of wetlands, while non-native species
generally responded positively to long dry periods followed by short-term watering events.
For some non-native species (common carp), long dry periods not only generated a positive
response in abundance, but enhanced their positive response to short-term watering
events. This divergence between native and non-native fishes’ response to antecedent flows
provides opportunities for managers to allocate flows to favour native fish assemblages. Our
evaluation of hypothetical flow scenarios suggests that frequent inundation of wetlands will
promote native fish over non-native fish. Future research and management of
environmental flows for fish assemblages should consider the role of antecedent conditions
and be mindful of different species’ responses to varying flow characteristics.
Flow dependent ecological responses | 191
FROGS IN THE MURRAY-DARLING BASIN
For four species of frogs (Limnodynastes fletcheri, Limnodynastes tasmaniensis, Litoria
peronii and Crinia parinsignifera), there were no common drivers of occupancy across the
sites (Section 5.15). Further, significant flow and habitat drivers varied for species across the
basin. In particular, the two hydrological variables of water permanence and flood return
were not identified as particularly informative at a basin-wide level for any species. Of the
variables considered, region was overwhelmingly the best predictor in all models. These
findings indicate that these widespread species are unlikely to be responding consistently to
the same hydrological and biophysical variable across the basin, with the implication that
management for frog species will likely need to be localised.
192 | Flow dependent ecological responses
CONTACT US
FOR FURTHER INFORMATION
t 1300 363 400
+61 3 9545 2176
e enquiries@csiro.au
w www.csiro.au
Land & Water Flagship
Neil Sims
t +61 3 9545 2163
e neil.sims@csiro.au
w www.csiro.au/businessunit-flagshipname
YOUR CSIRO
Australia is founding its future on
science and innovation. Its national
science agency, CSIRO, is a powerhouse
of ideas, technologies and skills for
building prosperity, growth, health and
sustainability. It serves governments,
industries, business and communities
across the nation.
Flow dependent ecological responses | 193