Riffle Algal and Macroinvertebrate Recovery From Wet-Seasonflood Disturbance in A Tropical Savannah River

Received: 20 April 2023 Revised: 25 September 2023 Accepted: 8 January 2024
DOI: 10.1002/eco.2617
RESEARCH ARTICLE
Riffle algal and macroinvertebrate recovery from wet-season

flood disturbance in a tropical savannah river, Northern
Australia
Simon Townsend 1,2
1
Faculty of Science and Technology, Charles
Darwin University, Casuarina, Northern Abstract
Territory, Australia
Succession occurs when river biota recovers from the loss of biomass and diversity
2
Water Resources, Department of
Environment, Parks and Water Security,
caused by flood disturbance, and has rarely been examined in the tropical savannah
Palmerston, Northern Territory, Australia biome. Hypotheses are tested which address the environmental context and recovery
Correspondence of riffle macroinvertebrates in an Australian perennial, tropical savannah river, located

Simon Townsend, Faculty of Science and in a catchment with low anthropogenic impact. Recovery occurred over a 32-week
Technology, Charles Darwin University,
Lakeside Drive, Casuarina, 0909, Northern
dry-season period of gradually declining discharge with near-stable water quality, after
Territory, Australia. a 23-week wet-season flood-disturbance period, which was typical of the river's flow
Email: dwntraveller58@hotmail.com
regime. Prolonged, benign environmental conditions during recovery facilitated the
Funding information accrual of macroalgae. Macroinvertebrate abundance, mediated by macroalgal bio-
Department of Environment, Parks and Water
Security
mass, and diversity increased at a logarithmic, asymptotic rate over the recovery
period. Macroinvertebrate assemblages and functional feeding groups (FFGs) tended
to become more similar between consecutive sample dates as recovery progressed,
though with variable abundances of core taxa (Orthocladiinae, Hydropsychidae,
Orthotrichia, Chironominae and Pyralidae) and prominence of Hydroptila late in the
recovery period. Filterer abundances fluctuated most, whilst the relative abundance
of herbivorous shredders and predators generally increased. The return of macroin-
vertebrate abundance and diversity to its pre-disturbance state is attributed to the
similar pre- and post-disturbance environmental conditions. Recovery broadly agreed
with the secondary succession paradigm for macroinvertebrates. Macroinvertebrate
succession was probably driven mainly by biotic factors, after the initial colonisation
period. However, this requires elucidation and expansion of the current paradigm of
river benthic succession from flood disturbance to address macroinvertebrate popula-
tion dynamics under benign abiotic environmental conditions.
KEYWORDS
environmental filter, functional feeding group, hypoxia, macroalgae, predation, Spirogyra,
succession, trajectory
1 | I N T RO DU CT I O N central role in structuring riverine ecosystems (Lake, 2000). Floods

reduce riffle biodiversity and remove algal (Biggs, 2000) and macroin-
Flood disturbances are an important component of the natural dis- vertebrate (Death, 2008) biomass. Floods also have wider ecological
charge regime (Bunn & Arthington, 2002; Poff et al., 1997) and have a implications by interrupting processes such as primary production
Ecohydrology. 2024;e2617. wileyonlinelibrary.com/journal/eco © 2024 John Wiley & Sons Ltd. 1 of 18

https://doi.org/10.1002/eco.2617
19360592, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eco.2617 by Anderson Ferreira - UFGD - Universidade Federal da Grande Dourados , Wiley Online Library on [17/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 of 18 TOWNSEND
(Uehlinger, 2000) and altering food webs (Power et al., 2008). Benthic paradigm: for example, the continual increase in macroinvertebrate
biota typically has low resistance (capacity to withstand disturbance) abundance until the next flood-disturbance event in a New Zealand
but high resilience (capacity to recover) to flood disturbance (Boulton river (Scrimgeour et al., 1988). A temporal sequence of macroinverte-
et al., 1992; Death, 2008; Lake, 2000). brate functional feeding groups (FFGs) can be broadly predicted based
Recovery from flood disturbance, an example of secondary suc- on food resource availability (Boulton et al., 1992; Lepori &
cession, occurs during the period of slowly declining discharge that Malmqvist, 2007; MacKay, 1992). Recovery produces benthic com-
follows a disturbance event. Recovery is dependent on the character- munities that typically return to their pre-disturbance state, unless
istics of the flood disturbance: its magnitude, duration, timing, fre- interrupted by another disturbance event (Death, 2008; Lake, 2013)
quency, and rate of change (Poff et al., 1997). For example, small or factors conspire to produce a different recovery trajectory, for
floods create mosaics of disturbed and undisturbed patches of sub- example, in response to an extreme flood (Woodward et al., 2015).
strate (Lake, 2000), whereas larger floods have more extensive Macroinvertebrate recovery can also be influenced by seasonal fac-
impacts on a river's physical habitat (riverbed scouring, channel tors, such as temperature (Boulton et al., 1992; Woodward
reshaping, and woody debris accumulation and removal). Unusually et al., 2015), highlighting the importance of the timing of a
large floods can leave an imprint on the benthic biota for several years disturbance.
(Eveleens et al., 2019; Haghkerdar et al., 2019). The timing of flood In this paper, I examine the recovery of riffle epilithic algae
disturbances, whether predictable (e.g. every winter) or unseasonal and macroinvertebrates from a flood disturbance in the Roper
and less predictable (e.g. summer events in a temperate climate) River, a perennial river in Australia's tropical savannah, which
affects benthic community resilience (Giller et al., 1991; Woodward occupy an area of 1.3 million square kilometres (Warfe
et al., 2015). et al., 2011; Figure 1). Tropical savannahs cover about 20% of the
Recovery comprises several processes (HilleRisLambers earth's land surface and are poorly researched compared to forests
et al., 2012; Lake et al., 2007; Schuwirth et al., 2016), which vary in the wet tropics (Williams et al., 2022). The Roper River region's
in relative strength as recovery progresses (Chang & monsoonal climate comprises annual periods of wet-season flood
HilleRisLambers, 2016). Species dispersion from a regional pool disturbance of 3–4 months (Petheram et al., 2008;
(Chase, 2003) and colonisation have a prominent role immediately Townsend, 2019; Warfe et al., 2011), followed by an 8–9 month
following disturbance (Connell & Slatyer, 1977; Patrick & dry-season recovery period of slowly declining, groundwater-fed
Swan, 2011). Abiotic, environmental filters (water quality, substrate, discharge. This long recovery period provides an opportunity to
hydraulics and canopy cover) preclude the establishment and persis- examine secondary succession over several months after the typi-
tence of some species, and are likely to play a role early in recov- cally rapid, initial phase of colonisation and growth of resistant taxa
ery (Weiher & Keddy, 1995). The temporal sequence and (Death, 2008).
availability of food resources (seston, fine and coarse organic mat- The Roper River is in one of the few global regions with relatively
ter, and benthic algae) have a critical role throughout recovery low anthropogenic impact on river biodiversity (Vörösmarty
(MacKay, 1992). Successional trajectories of species may be influ- et al., 2010; Williams et al., 2022). Consequently, the regional species
enced by local internal dynamics, referred to as assembly rules pool can be expected to be near-natural, unlike heavily impacted riv-
(Lake et al., 2007), with biotic interactions (intraspecific and inter- ers where the species pool can be greatly diminished (Lake
specific competition, predation and facilitation) probably becoming et al., 2007). The water resources of the Roper River catchment are
more important as recovery progresses (Chang & currently near natural but under increasing development pressure.
HilleRisLambers, 2016). The relative roles of biotic and abiotic pro- Groundwater extraction for agriculture will decrease recovery flows
cesses in shaping secondary succession, however, are unclear with potentially wide-ranging ecological implications, notably for riffle
(Chang & Turner, 2019; Schuwirth et al., 2016). habitats, resulting in losses of habitat area and connectivity, poor
Some broad statements of riffle benthic recovery can be made. water quality, and a transition from lotic to lentic conditions (King
Algal biomass increases following flood disturbance (Biggs, 1996; et al., 2015; Rolls et al., 2012).
Biggs, 2000; Peterson, 1996) to reach a maximum carrying capacity This research complements landscape-scale river studies con-
and then can shift to a loss phase driven by senescence, sloughing, ducted in the Australian tropical savannah (Blanchette &
and grazing (Biggs, 1996). The rate of biomass accrual is primarily Pearson, 2013; Kay et al., 1999; Leigh, 2013; Leigh & Sheldon, 2009).
dependent on optical, nutrient, substrate resources and the The primary findings of this research, which mostly examined macro-
hydraulic environment (Stevenson, 1997). A successional sequence phyte, bank edge, and sandy riverbed habitats, have been the impor-
from diatoms to macroalgae is common (Larned, 2010; tance of river connectivity along the river channel and with floodplain
Peterson, 1996). lakes, and antecedent flood characteristics in shaping benthic macro-
Rates of macroinvertebrate abundance and diversity recovery are invertebrate assemblages. Blanchette and Pearson (2013) also con-
initially rapid (Death, 2008), but then commonly slow, typically fitting cluded wet-season flooding did not appear to reset macroinvertebrate
a logarithmic, asymptotic relationship (Boulton et al., 1992; Flecker & assemblages to a pre-disturbance state and found no consistent pat-
Feifarek, 1994; Greenwood & Booker, 2015; Grimm & Fisher, 1989). tern in taxonomic richness or assembly over the dry-season between
Not all riverine macroinvertebrate successions, however, follow this rivers and sites within rivers. Fewer temporally intensive studies have
TOWNSEND 3 of 18
F I G U R E 1 The Roper River study

reach is in the Australian tropical
savannah defined by vegetation
community (Source: Bushfire and Natural
Hazards Collaborative Research Centre at
Charles Darwin University, Darwin).
been conducted. Over a 1-year period, Marchant (1982) and Outridge 2 | M A T E R I A L S A N D M ET H O D S

(1988) found macroinvertebrate richness and abundance in floodplain
lakes and a small perennial stream declined to a minimum at the end 2.1 | Climate, catchment description, and
of the dry season in November but recovered with the resumption of study reach
wet-season flows. Relatively little research has been published on
macroinvertebrate ecology in other tropical savanna regions The seventh order (based on a 1:250,000 scale map) Roper River
(e.g. India, South-East Asia, Botswana, Namibia, and Kenya in Africa, drains a catchment of 81,000 km2 (Figure 1) and is one of eight large
and Central and South America), excluding the Brazilian Cerrado perennial rivers in Australia's tropical savannah which collectively
(e.g. Macedo et al., 2014; Moreyra & Padovesi-Fonseca, 2015; Reis drain 500,000 km2 and flow into either the Gulf of Carpentaria or
et al., 2017). Timor Sea (Townsend, 2019). The catchment is highly weathered, with
Based on studies of the Daly River in a neighbouring catchment infertile soils and low topographic relief (maximum elevation 440 m).
(Townsend, 2019), which has a similar hydrology to the Roper River, I Low intensity cattle grazing of the savannah woodland is the principal
hypothesise (1) that recovery from wet-season flood disturbance will land use, though more intensive, small-scale horticulture occurs in the
be underpinned by gradual seasonal recessional discharge with near- upper catchment. The only town in the catchment has a population of
stable water quality to produce a benign environment favourable for 350 (2016 Australian census). Eucalyptus woodland with grass under-
biotic recovery. I hypothesise that (2) benthic algal biomass will storey is the most common vegetation community, whereas Melaleuca
increase to reach a maximum dependent on the days of accrual (paperbark trees) and Pandanus aquaticus dominate the river's riparian
(Biggs, 1996; Townsend & Padovan, 2005). These hypotheses provide vegetation (Faulks, 2001).
a context for the recovery of macroinvertebrate assemblages, which is Annual rainfall averages 843 mm over the Roper River catchment
the focus of this study. (CSIRO, 2009), with 96% falling between November and May to pro-
An additional three hypotheses examine the broad temporal duce a highly predicable seasonal pattern of high wet-season dis-
pattern of macroinvertebrate recovery from flood disturbance. I charges, followed by low dry-season discharges (Figure 2). Monsoonal
hypothesise that (3) abundance and diversity will recover at logarith- and cyclonic rains, rainfall depressions and large storms can produce
mic, asymptotic rates, initially rapid then declining, (4) recovery will several flood peaks during the wet season (Townsend, 2019). Although
follow a sequence of assemblages and FFGs, where filter feeders the seasonal pattern is predictable, high inter-annual variability of wet-
and periphyton grazers of microalgae and fine organic sediment will and dry-season discharges exists (Petheram et al., 2008).
dominate during the initial recovery phase, and shredders and pred- Dry-season discharges in the Roper River originate predominately
ators are the last FFG to become established (MacKay, 1992), and from dolomite aquifers in the upper catchment, and decrease down-
(5) assemblages will return to their pre-disturbance state. The stream by 50–80% due to evapotranspiration losses (Zaar, 2009).
research will contribute to the broader topic of secondary succes- Spring inflows to the river during the dry season have conductivities
sion, and inform the management of the river's flow regime, biodi- (1000–2500 μS cm1; Schult & Novak, 2017) which are at least
versity and ecological services. 10-fold higher than the river's wet-season conductivity and enables
4 of 18 TOWNSEND
F I G U R E 2 Discharge at Red Rock

gauging station (19 km downstream of
the study reach), showing data collection
dates, median monthly discharge (1972 to
2018), and flood disturbance and
recovery periods. The arrows point to the
start of the 2013/2014 and 2014/2015
disturbance periods defined by the first
major storm runoff event.
conductivity to be used as a tracer of groundwater input to the river litter packs and macrophytes. The sites were on average 75 m apart
(Townsend, 2019). In the dry season, the lower reaches of the Roper and therefore not independent of each other, precluding analyses
River discharge through a series of long, shallow pools which can be such as analysis of variance (ANOVA) of site data.
up to 10-km long and 3-m deep, and connected by riffles and runs of
several kilometres.
2.3 | Data collection dates
2.2 | Study design and site description Data were collected during the later months of the 2014 dry-season
recovery period, and throughout the 2015 recovery period (Table 1).
A single riffle reach (mid-point 134.3350 S 14.567 E) with no tributary To address Hypothesis 1, discharge, velocity, depth and water-quality
inflows was selected in the lower reaches of the Roper River, where data were collected at Site 1 between July and October in 2014, and
anthropogenically reduced discharges are likely to have their greatest in 2015 between May and November at Site 1, and July
ecological impact because discharges are low compared to the upper and November at Site 3. In November 2014, only in situ water-quality
reaches. Six sites were selected in the 450-m long, 15–20-m wide study data were collected.
reach, which confined dry-season baseflows and was located within a Benthic algal and macroinvertebrate, hydraulic and water-quality
larger channel of 40–70 m width which contained most wet-season dis- data were collected to test Hypotheses 2 to 4 (concerning the recov-
charges. Melaleuca trees provided approximately 35% canopy cover ery of benthic biota) for Sites 1 (May to November) and 3 (July to
over the river. The reach is 19 km upstream of the Red Rock gauging November) in 2015. To test for the return to pre-disturbance macro-
station (G9030250) operated by the Northern Territory Department of invertebrate assemblages (Hypothesis 5), data were collected in
Environment, Parks and Water Security, which has monitored discharge October 2015, and September and October 2014, to compare each of
since October 1975. There are no tributary inputs between the gauging the six sites between the 2 years for the same month, and the same
station and the study reach during the dry season. number of days since the commencement of the recovery period.
Benthic biota, water quality and hydraulic data were collected in
2014 and 2015 from the six sites (numbered downstream from 1 to 6)
with a range of depths (0.2–0.7 m) and velocities (0.3–0.9 m s1). 2.4 | Hydrography, and flood-disturbance and
Most recovery data were collected from sites 1 and 3, which had simi- recovery periods
lar depths and flow velocities (0.5 m, 0.80 m s1). Each site com-
prised a 10-m length of embedded cobbles, near-uniform depth and The 2013/2014 and 2014/2015 flood disturbances are compared to
velocity, and absence of other habitats such as woody debris, leaf the 43-year discharge record at Red Rock gauging station to provide
TOWNSEND 5 of 18
TABLE 1 Data collection at the Roper River study reach. current meter (Model C31, OTT Hydromet, Kempten, Germany) when
benthic sample collection was undertaken.
Site
Year, month 1 2 3 4 5 6
2014, Jul
2.5 | Water-quality data collection
2014, Aug
2014, Sep ● ● At each of the six sites, temperature, dissolved oxygen, pH and electri-
2014, Oct ● ● ● ● ● ● cal conductivity were measured between 09:00 and 12:00 h with a
2014, Nov □ Hydrolab Quanta multi-parameter probe (OTT Hydromet, USA) and
2015, May ● turbidity with a Hach 2010Q portable turbidimeter (Hach, Loveland,
2015, Jun ● Co, USA). Average daily temperature and dissolved oxygen at Site
1, excluding November 2014, were determined from 10-min interval
2015, Jul ● ●
monitoring data collected by an EXO2 multi-parameter data sonde
2015, Aug ● ●
(YSI Inc, Ohio, USA).
2015, Sep ● ●
To assess the penetration of light to the riverbed for benthic algal
2015, Oct ● ● ● ● ● ●
photosynthesis, the attenuation of photosynthetically available radia-
2015, Nov ● ●
tion (PAR) was calculated from depth profile measurements made
Note: water-quality and hydraulic data collection. □ in situ water-quality with a LI-192 underwater quantum sensor (Li-Cor, Lincoln, NE, USA)
measurements only, ● water-quality, hydraulic, benthic algal and taken from a boat anchored in the pool immediately upstream of the
macroinvertebrate data collection.
study reach. The vertical attenuation of PAR was calculated from a
regression of natural log-transformed PAR and depth, and is
historical context. The commencement of the disturbance period was expressed as the euphotic depth, which is the depth to which 1% of
defined by the first major runoff event of the wet season with a dis- incident PAR penetrates and indicates the maximum depth of algal
charge maximum at least fourfold higher than the discharge immedi- photosynthesis.
ately preceding the event. Events of this magnitude remove benthic Water samples for dissolved and total nutrients were collected
macroalgae in the neighbouring Daly River (Townsend & from the upstream Site 1 (Table 1). Samples were filtered in the field
Padovan, 2005). through a 0.45-μm Minisart PES syringe filter (Sartorius AG, Germany)
The end of the flood-disturbance period occurred during the wet- then frozen. Chemical analyses were performed in accordance with
to dry-season transition, and was marked by the cessation of major the following APHA (1998) methods: 4500I for nitrite and nitrate,
storm runoff events and greater proportion of groundwater-derived 4500HI for ammonia, 4500G for soluble reactive phosphorus, 4500C
river discharge. This shift was set at 20 m3 s1 based on an examina- for total and soluble nitrogen and 4500B(3)G for total and soluble
tion of the discharge record at Red Rock gauging station. I acknowl- phosphorus. Samples for alkalinity were collected in July and August
edge that the end of the flood-disturbance period is not discrete but each year and analysed by APHA (1998) method 2320B.
rather a transition. Applying these criteria, however, permitted a rela- Water samples collected for chlorophyll a (chla) were filtered
tive and quantitative assessment of the 2013/2014 and 2014/2015 through 0.7-μm Whatman glass fibre filters in the field, placed in a
flood disturbances and their recovery periods in comparison with the small glass bottle, and then frozen. Chlorophyll a was analysed by
historic record. The percentile value of the duration of the flood- fluorometry based on APHA (1998) method 10200H3. In the labora-
disturbance and recovery periods, maximum wet-season discharges, tory, chlorophyll a was extracted by the addition of 90% acetone to
and minimum dry-season discharges was calculated from the Red each bottle and refrigerated for 24 h, and then measured by a non-
Rock discharge record by ranking values from lowest to highest and acidification technique (Welschmeyer, 1994) using a Trilogy® fluo-
converting the rank score to a percentage of the total number of data. rometer (Turner Designs Inc., Sunnyvale, California).
The percentile equates to the percentage of records less than the per-
centile value.
At Site 1, discharge was measured monthly by standard 2.6 | Benthic algal and macroinvertebrate data
hydrographic methods between July and October 2014. Estimates of collection
discharge for the study reach during the 2015 recovery period and
May and June in 2014 were made by linear regression of Red Rock Twelve benthic algal samples were collected randomly at each site
gauging station and Site 1 data (r2 = 0.990, n = 4, p < 0.001). Depth with a periphyton sampler (Davies & Gee, 1993) that scoured an area
over the 2015 flood-disturbance period was monitored at Site 1 every of 8.04 cm2 with a Velcro™ disk attached to a wooden stick. The sam-
15 min with a Diver TD model 11110404 logger (Eijkelkamp Soil & ple method slightly underestimated algal biomass per cobble because
Water, The Netherlands). Ten random measurements of depth and long macroalgal filaments that streamed downstream beyond each
velocity at 40% of the depth (measured from the riverbed and approx- cobble were not sampled. This underestimation would have increased
imating the profile average velocity) were made with an Ott propeller over the dry season as long filaments (greater than 3-cm length)
6 of 18 TOWNSEND
became more numerous. Each disk was placed in a glass bottle and Richness (S), the number of taxa in a sample, can increase as a
frozen within an hour of collection, and then analysed for chlorophyll function of the number of individuals counted (N, abundance) and
a by the method described above for water samples. consequently affect abundance–richness relationships and biodiver-
A composite sample of benthic algae comprising six randomly sity assessments (Gotelli & Colwell, 2001). Rarefied richness was cal-
selected 1-cm2 areas was collected for algal identification and biovo- culated to standardise richness to the minimum average abundance
lume. The sample was preserved in Lugol's iodine and examined in a per cobble for sites 1 (N = 53) and 3 (N = 77) sampled in 2015, and
Lund cell under 400 magnification on a Zeiss Axioscope (Carl Zeiss, for the pre- and post-disturbance comparisons (N = 23).
Göttingen, Germany). Algae were identified to either species or genus Multivariate analyses were undertaken on the average abundance
level. Enumeration was based on one to five vertical transects for the of the five replicates collected at each site. Data were log(x + 1) trans-
most common taxa. Dominant taxa enumerations were supported by formed to down-weight frequently occurring taxa or presence–
the statistical software COUNTEM (Cruz & González, 2018) to ensure absence transformed. FFG relative abundances were also log(x + 1)
errors of less than 10%. The dimensions of each taxon in a sample transformed. Bray–Curtis similarity matrices were computed and dis-
were measured and standard geometric formulae applied to calculate played using nonmetric multi-dimensional scaling (MDS) ordinations.
the biovolume. The total biovolume for each taxon was calculated as Environmental variables were normalised for multivariate analyses.
the product of its cell concentration and biovolume, converted to bio- The similarity of macroinvertebrate assemblages was tested applying
mass assuming 1 cm3 equals 1 g and expressed as a relative abun- a one-way analysis of similarity (ANOSIM; Clarke & Gorley, 2006)
dance of the total sample biomass. based on Bray–Curtis resemblance matrices, which computed similar-
At each site, replicate macroinvertebrate samples were collected ity as an R statistic that ranged between 0 (no difference) and 1 (very
from each of five individual embedded cobbles with similar exposed different).
surface area, typically 20–25 cm in length and half embedded. A fine Temporal sequences of macroinvertebrate assemblage and FFG
mesh net was placed immediately downstream during collection to abundances for sites 1 and 3 were each assessed by a seriation test of
capture any macroinvertebrates disturbed during collection. Each cob- their Bray–Curtis similarity matrix. The null hypothesis for each seria-
ble was gently scrubbed, and the material removed passed through a tion test was the absence of a linear temporal sequence. The test
500-μm sieve. The retained material was then preserved with 70% coefficient assesses the extent to which data points follow a simple
ethanol. In the laboratory, invertebrate specimens were hand-picked linear trend in multi-dimensional ordination space and ranges between
from a petri dish under a dissecting microscope and identified to the zero (no alignment) and one (Clarke & Gorley, 2006).
highest taxonomic level practical, which was generally family but The BEST routine (Clarke & Gorley, 2006) evaluated the environ-
included subfamily and genus for some insects. Abundance was esti- mental variables aligned with the temporal sequences of macroinver-
mated as individuals per cobble. tebrate assemblages for Site 1. Velocity, depth, benthic algal biomass
Taxa were classified into five functional feeding groups (FFGs: and temperature variables were selected for these tests. Time, as days
grazers, gatherers, filterers, predators and shredders; Table S1) since the commencement of the recovery period, was not selected
according to Leigh and Sheldon (2009), who examined the FFGs of because it was correlated to these environmental variables. Water-
two Australian tropical savannah rivers; otherwise, reference was quality variables (nutrient concentrations, water clarity, pH and con-
made to the Australian Centre for Freshwater Ecosystems website ductivity) were not selected because there was no a priori reason for
(last accessed 28 October 2022). For taxa assigned to more than one their selection owing to their small range over the recovery period
FFG, their abundance was shared equally among these groups and, for nutrient concentrations and water clarity, greater relevance
(e.g. one Orthocladiinae individual = 0.5 grazer + 0.5 gatherer) before to benthic algal biomass.
calculating FFG proportions (Dudgeon, 1994). Gatherer and grazer The October 2015 and October 2014 macroinvertebrate assem-
FFGs, however, were combined because the two groups were both blages were compared to test whether post-disturbance assemblages
dominated by Orthocladiinae and Chironominae which have taxa in had returned to their pre-disturbance state. Statistical tests were con-
both FFGs. ducted that compared each site's data between the years, rather than
conducting a reach scale test between the 2 years because the six
sites are not independent due to their close proximity. T-tests were
2.7 | Data analyses conducted for algal biomass (n = 12 for each site and year) and the
following macroinvertebrate variables: abundance, richness, rarefied
Univariate t-test, Pearson correlation and regression analyses were richness, evenness and diversity (n = 5 for each site, each year). Indi-
0
undertaken using SigmaPlot v12. Multivariate analyses, Pielou's J vidual ANISOMs compared assemblage abundances for each of the
evenness, Shannon's H0 diversity metric and rarefied taxa abundances six sites (n = 5 for each site, each year). Three tests also compared
were computed for macroinvertebrate data using PRIMER-E v6.1.16 the algal biomass and macroinvertebrate abundance and biodiversity
(Clarke & Gorley, 2006). The significance levels of multiple compari- metrics for Sites 1 and 2 for dates collected on similar days since the
sons were Bonferroni corrected to an overall level of 5% to reduce commencement of the recovery period in 2014 and 2015. A total of
the chances of type I errors (mistaken rejection of a true null nine tests were conducted for each variable, with a Bonferroni cor-
hypothesis). rected (α = 0.05/9) significance level for each variable.
TOWNSEND 7 of 18
3 | RESULTS soluble form (Table 2). Soluble molar N:P ratios approximated
1, whereas total N:P ratios were at least fivefold higher (Table 2). Sus-
3.1 | Flood-disturbance and recovery periods pended chlorophyll a concentration averaged a low 1.0 ± 0.1 μg L1.
Discharge between November 2013 and November 2015 was typical Hypothesis 2. Benthic algal biomass will increase to
of the river's seasonal pattern (Figure 2). Flood-disturbance periods reach maximum dependent on the days of accrual.
(Figure 2) commenced in November each year and ended in March or
April. The 2013/2014 disturbance period of 127 days approximated Algal biomass exhibited high within-site variability, with a range
the median (130 days; n = 34), whilst the 2014/2015 disturbance that spanned at least an order of magnitude and reflected the
was 35 days longer than the median. Maximum discharges were observed uneven distribution of macroalgae. Biomass increased to
1 1
400 m s 3 3
lower than the median (1260 m s ; n = 38). Over the reach maxima at Sites 1 and 3 (Figure 4; 200 mg chla m2) before
2014/2015 flood-disturbance period, the river's depth at Site the end of the recovery period. Site 1 median biomass fitted a Gauss-
1 reached a maximum of 9.7 m and averaged 2.3 m. ian bell regression (r2 = 0.86; Figure 4), indicating biomass increased
The 2014 and 2015 recovery periods were 214 and 224 days, and then decreased. Site 3 biomass showed a similar trend (Figure 4),
respectively, and about 13 days less than the median (232 days, exhibiting a threefold biomass increase between July and October (t-
n = 32). The 2014 recovery period commenced 22 days earlier than test; p < 0.001), which then halved between October and November
in 2015 (Figure 2). Minimum dry-season discharges, which occur (t-test; p < 0.001). Median algal biomasses at Sites 1 and 3 were not
towards the end of the recovery period, were almost 50% higher than correlated with velocity, depth or water quality (conductivity, temper-
1
the median (1.0 m s 3
, n = 41). ature, turbidity and nutrient concentrations).
Filamentous macroalgae dominated algal biomass (Figure 5). Site
Hypothesis 1. The recovery will occur over seasonal 1 algae shifted from Cladophora spp. dominance in May and June
recessional discharge with near-stable water quality. (Figure 5a) to Spirogyra from mid-July, and was dominated by a Spiro-
gyra fluviatilis complex which is also found in the neighbouring Daly
Discharge, depth and velocity declined gradually over the 2014 River (Townsend et al., 2008). At Site 3 (Figure 5b), July's biomass was
and 2015 recovery periods to reach minima in November each year dominated by Batrachospermum spp. (mainly Barynema australicum)
(Figures 2 and 3a, b). As discharge declined, conductivity increased to and then by Spirogyra. Either diatoms (2–17%) or Cladophora (0–21%)
reflect the greater proportion of groundwater input to the river were second ranked in biomass to Spirogyra from mid-August.
(Figure 3c). The river was alkaline, with pH values that exceeded
8 and bicarbonate alkalinity of 270–310 mg L1. Hypothesis 3. Macroinvertebrate abundance and bio-
Water temperatures increased from a minimum of 22 C in July to diversity will increase at logarithmic, asymptotic rate of
recovery.
a maximum of 30 C in November (Figure 3d). Water clarity, measured
by turbidity (Figure 3) and euphotic depth, also increased over the
recovery period. Euphotic depths (5.2–10.5 m) exceeded the river's Insects dominated the macroinvertebrate fauna at Sites 1 and
depth at all sites (<0.8 m), indicating high incident irradiance at the riv- 3 during 2015, comprising 24 of the 29 taxa identified mostly to fam-
erbed available for photosynthesis. Turbidity (r = 0.75), euphotic ily level and 97.8% of the 6760 individuals counted. The insect order
depth (r = 0.85) and conductivity (r = 0.85) were correlated with day Trichoptera was the most diverse, represented by seven families,
number ( p < 0.02). whereas Diptera and Ephemeroptera comprised, respectively, four
The study reach was well oxygenated, with concentrations of and three families. The most abundant insect orders, present in all
1
7.9–8.6 mg L , excluding November 2014. On 15 November, 7 days samples, were Diptera, Trichoptera, Lepidoptera and Ephemeroptera,
after the peak discharge of a storm runoff event (Figure 2), dissolved which had respective relative abundances of 54%, 29%, 7.4% and
oxygen was 0.3 mg L1, whilst conductivity was 496 μS cm1, pH 6.5%. Core taxa, present throughout the recovery period at Site 1 with
was near-neutral and turbidity exceeded 100 N.T.U. This water qual- an average relative abundance exceeding 5% comprised the following:
ity resulted from a hypoxic runoff event, which originated from a trib- Orthocladiinae, Hydropsychidae, Orthotrichia, Chironominae and Pyr-
utary several kilometres upstream of the study reach, and caused the alidae. Mollusca was the second major group numerically, however,
death of hundreds of fish, Cherabin prawns (Macrobrachium sp.) and constituted a small proportion (1.4%) of individuals and represented
Atyidae shrimps (Fisheries Division, N.T. Department of Industry, by two bivalves (Corbiculidae and Hyriidae) and gastropods
Tourism and Trade; unpublished data). (Viviparidae and Thiaridae).
Nutrient concentrations were consistently low (Table 2), exhibited An effect of abundance on Site 1 richness is suggested from a lin-
no temporal trend (regression analyses not significant) and were simi- ear regression analysis of the richness and rarefied richness. The
lar between years (t-tests for Site 1 July to October data not signifi- regression's slope (0.57 ± 0.13) is significantly different to one
cant). Oxidised nitrogen and ammonia concentrations were generally (no rarefication effect). The effect of abundance is estimated to have
less than 5 μg L1 and comprised 5% of the total nitrogen concen- accounted for two extra taxa at the end of the recovery period. At
tration. Phosphorus concentrations were 2–18 μg L1 and mostly in Site 3, however, the regression slope (0.83 ± 0.27) was not statistically
8 of 18 TOWNSEND
F I G U R E 3 Water quality and

discharge (Q) at Site 1 in 2014 and 2015.
(a) Measured and predicted discharge
based on Red Rock gauging station data,
(b) average cross-section depth and
velocity, (c) conductivity (2015 data;
exponential regression curve; r2 = 0.93;
p = 0.013), (d) turbidity (2015 data;
exponential regression curve; r2 = 0.85;
p = 0.005) and (e) euphotic depth (2015
data; linear regression; r2 = 0.85;
p = 0.003).
T A B L E 2 Mean and standard error of

Nutrient concentration (μg L1) Molar ratio
nutrient concentrations (as N and P) at
Year NOx NH3 TN SRP TP Soluble N:P Total N:P Site 1 for 2014 (n = 4; July to October)
and 2015 (n = 7; May to November)
2014 4.9 ± 0.2 5.0 ± 0.3 130 ± 2 10 ± 2 16 ± 2 1 ± 0.3 8 ± 0.2
recovery periods.
2015 3.5 ± 0.2 3.6 ± 0.1 150 ± 4 6±1 8±3 1 ± 0.2 18 ± 0.4
Note: See Figure 2 for sample dates. NOx, nitrite and nitrate; NH3, ammonia, TN, total nitrogen; SRP,
soluble reactive phosphorus; TP, total phosphorus.
TOWNSEND 9 of 18
F I G U R E 4 Median benthic agal

biomass at Sites 1 and 3, and Gaussian
regressions for Sites 1 (r2 = 0.86,
p = 0.016) and 3 (r2 = 0.69, p = 0.082).
F I G U R E 5 Histogram of benthic algal

biovolume relative abundance for (a) Site
1 and (b) Site 3, 2015. The remaining
histogram above the shaded stacks
represents all other algal taxa, mostly
diatoms.
different to one, indicating no detectable effect of abundance on the logarithmic, asymptotic rate at Site 1, but not at Site 3, though the
richness count. small data set (n = 5) produced a low statistical power of detection.
Of the 13 taxa present at Site 1 in May, nine persisted throughout Taxon richness, rarefied taxon richness, evenness and diversity also
the recovery period. Hydroptila was the only taxon not present early increased at logarithmic, asymptotic rates over the recovery period at
in the recovery period, which increased significantly in abundance Site 1 (Figure 6), whereas only rarefied richness, evenness and diversity
during the late recovery phase (Table 3), reaching 24% relative abun- regressions were significant. Macroinvertebrate abundance and diver-
dance in October at Site 1. Turnover was low, comprising rare taxa sity increased proportionally with algal biomass (Figure 7).
(<2% relative abundance), with an average of 85% of taxa common to
consecutive sample dates. Between three and four new taxa were Hypothesis 4. Secondary succession will follow a
added to the cumulative taxa total each month between May and sequence of macroinvertebrate assemblages and func-
September, whilst only zero to two taxa were added each month tional feeding groups.
between September and November.
Macroinvertebrate abundance increased over the recovery period, Underlying the increase in macroinvertebrate abundance over the
threefold between May and November at Site 1, and twofold between recovery period was fluctuating taxon and FFG abundances. Orthocla-
July and November at Site 3 (Figure 6a), and at a statistically significant diinae was the most consistently abundant taxon at both sites (21–
10 of 18 TOWNSEND
TABLE 3 Abundance of the 12 most common taxa at (a) Sites 1 and (b) 3 (not sampled in May and June).
(a) Site 1
Abundance (ind cobble1)

Taxon
Date FFG May 17 Jun 17 Jul 16 Aug 15 Sep 23 Oct 14 Nov 11 Avg RA%
Orthocladiinae Ga/Gr 28.0 31.2 35.6 67.3 43.0 31.0 7.8 35.6
Hydropsychidae F 21.2 4.0 19.0 3.4 2.6 4.5 38.0 13.8
Orthotrichia Gr 4.2 7.2 7.4 7.9 25.4 7.6 7.0 9.0
Chironominae Ga/Gr 2.4 1.0 5.4 20.8 13.0 13.6 13.2 8.0
Pyralidae S 2.2 1.4 4.2 3.0 10.2 18.2 17.6 6.6
Hydroptila Gr 0.0 0.0 0.0 0.4 5.6 33.6 22.0 6.4
Empididae P 0.8 0.8 4.2 12.4 8.5 6.2 5.9 4.5
Pseudocloeon Ga 2.6 0.2 1.8 2.6 7.0 10.2 6.6 3.6
Simuliidae F 0.0 1.0 8.6 5.6 3.5 1.6 5.6 3.4
Tasmanocoenis Ga 3.2 2.2 0.0 1.9 5.9 2.0 1.0 2.4
Tanypodinae P 0.0 0.8 0.6 3.4 6.1 5.8 2.0 2.1
Platybaetis Gr 3.4 2.4 1.4 0.0 0.0 0.0 0.0 1.6
Total abundance 68.8 52.8 89.2 132.4 136.4 144.2 130.2
(a) Site 3
Abundance (ind cobble1)
Taxon FFG May Jun Jul 16 Aug 15 Sep 23 Oct 14 Nov 11 Avg RA%
Orthocladiinae Ga/Gr 38.2 26.6 43.8 33.8 17.4 28.6
Chironominae Ga/Gr 10.2 20.2 12.6 20.2 16.2 14.3
Hydropsychidae F 21.6 6.2 1.6 13.4 18.5 10.9
Empididae P 6.6 10.6 9.6 11.8 13.0 9.0
Hydroptila Gr 0.2 0.4 11.1 24.2 18.9 7.6
Orthotrichia Gr 7.6 4.6 15.2 8.6 5.8 7.1
Pyralidae S 3.2 1.6 7.3 14.5 16.2 6.4
Tanypodinae P 1.6 0.6 10.0 8.0 1.7 3.3
Pseudocloeon Ga 0.4 1.4 5.2 10.8 4.1 3.2
Corbiculidae F 0.0 0.0 0.4 1.5 14.9 2.4
Tasmanocoenis Ga 0.2 0.6 3.5 4.1 0.8 1.4
Simuliidae F 0.4 0.2 1.0 5.9 1.7 1.3
Total abundance 93.0 77.0 131.0 163.2 133.8
Note: Also shown is taxon feeding functional group (FFG), total abundance and the average relative abundance (avg RA) for the 2015 recovery period. The
abundance data are the average of five cobbles. FFGs: G, gatherer; Gr, grazer; F, filterer; S, shredder; P, predator. The 12 most common taxa comprised
93.1–96.9% relative abundance over the recovery period.
59% between May and October), whereas Hydropsychidae (1–31% taxon presence-absence (Rho = 0.75, p = 0.002), indicating both the
relative abundances) and Orthotrichia (5–19%) fluctuated (Table 3). taxon present and their abundance contributed to sequence. Site 3's
Chironominae, Corbiculidae, Pyralidae and Hydroptila became more trajectory closely follows Site 1 in ordination space (Figure 8a); how-
abundant in the late phase of the recovery period, increasing from ever, Site 3 seriation tests for abundances and presence–absence data
approximately 2% to 15% relative abundance at Site 1. Unlike other were not significant, possibly owing to the low statistical power of
taxa, Platybaetis were most abundant early in the recovery period at detection associated with the small data set (n = 5). No difference
Site 1 (May–July; Table 3) and then either not detected or present in was detected between Sites 1 and 3 assemblages during months of
lower abundances. common data collection between July and November (ANISOMs;
A temporal sequence of assemblages is evident for Site 1 data p > 0.01; Bonferroni correction). Assemblage abundances (Site
based on MDS ordinations of similarity matrices and seriation tests of 1, R = 0.72; Site 3, R = 0.61) and taxon presence–absence
taxon abundances (n = 7, Rho = 0.87, p = 0.001; Figure 8a) and (Site 1, R = 0.78; Site 3, R = 0.54) were significantly different
TOWNSEND 11 of 18
F I G U R E 6 Macroinvertebrate assemblage metrics during flood-

disturbance recovery. Sites 1 (closed circles; n = 7) and 3 (open
circles, n = 5) averages per cobble during the 2015 dry season with
significant logarithmic, asymptotic regressions for (a) abundance (Site
1: r2 = 0.80, p = 0.007), (b) taxon richness (Site 1: r2 = 0.87,
p = 0.002), (c) rarefied taxon richness (Site 1: r2 = 0.71, p = 0.017;
Site 3: r2 = 0.83, p = 0.032), (d) Pielou's (J0 ) evenness index (Site 1:
r2 = 0.77, p = 0.011; Site 3: r2 = 0.96, p = 0.003) and (e) Shannon's
diversity index (Site 1: r2 = 0.88, p = 0.005; Site 3: r2 = 0.76,
p = 0.050).
(ANOSIM tests; p < 0.02) between the beginning and end of the
recovery periods for both sites.
Macroinvertebrate assemblages became more similar over the
recovery period at Site 1, though not at Site 3, increasing from 60% to
70% similarity based on Bray–Curtis similarities between consecutive
sample dates (linear regression; r2 = 0.95, p < 0.001). Site 1's trajec-
tory was correlated (BEST procedure) with benthic algal biomass
(Rho = 0.65; p = 0.015), with velocity second ranked (Rho = 0.08).
Benthic algal biomass, however, was correlated with day number since
the commencement of recovery (Pearson correlation test; r = 0.88;
p = 0.01).
Site 1 FFG abundance assemblages also showed a significant tra-
jectory in ordination space (n = 7; Rho = 0.69, p = 0.007; Figure 8b),
although not at Site 3. Grazers and gatherers dominated the FFG
(65% relative abundance; Figure 9) at both Sites 1 and 3. Absolute
abundances and relative abundances of filterers fluctuated more than
the other FFGs (Figure 9). Shredder abundances, predominantly
Pyralid moths, increased logarithmically at Site 1 between May and
November (r2 = 0.83, p < 0.001) and linearly between July
and November at Site 3, r = 0.86, p = 0.023; Figure S1) to reach a
2
maximum in November. Predator abundances (mostly Empididae and

Tanypodinae) also increased, but reached a maximum in August–
September, and then declined in November (Gaussian bell regressions:
Site 1, r2 = 0.93, Site 3, r2 = 0.96 p < 0.01; Figure S1), coincident with
an almost halving of gatherer/grazer abundances (Figure 9) and
decreased algal biomass (Figure 4). Gatherer/grazer (r = 0.80,
p = 0.0017), predator (r = 0.80, p = 0.0019), and shredder abun-
dances (r = 0.62, p = 0.031), though not filterers, were correlated
with median algal biomass at Sites 1 and 3. A correlation between
predator and gatherer/grazer abundances at Sites 1 (r = 0.92;
p < 0.001) and 3 (r = 0.92; p = 0.020) was underscored by Chirono-
minae (gatherer/grazer) and Empididae (predator) correlations (Site 1:
r = 0.96, p < 0.001; Site 3: r = 0.73, p = 0.16). No statistically signifi-
cant relationships were detected between predator and either filterer
or shredder FFG abundances for either site.
Hypothesis 5. The macroinvertebrate assemblage will

return to its 2014 pre-disturbance state.
12 of 18 TOWNSEND
F I G U R E 7 Relationship between macroinvertebrate assemblage

metrics and algal biomass. Sites 1 (closed circles; n = 7) and 3 (open
circles, n = 5) averages per cobble of algal biomass (x-axis), with Site
1 (dashed line) and Site 3 (dotted line) for significant linear
regressions. (a) abundance (Site 1: r2 = 0.70, p = 0.020; Site 3:
r2 = 0.80, p = 0.039), (b) taxon richness (Site 1: r2 = 0.91, p < 0.001;
Site 3: r2 = 0.89, p = 0.016), (c) rarefied taxon richness (Site 1:
r2 = 0.56, p = 0.048), (d) Pielou's (J0 ) evenness index (Site 1: r2 = 0.60,
p = 0.042) and (e) Shannon (H0 ) diversity index (Site 1: r2 = 0.72,
p = 0.016).
The October 2014 and 2015 hydraulic environments were similar

at each of the six sites (non-significant t-tests for depth and velocity).
Water quality was also similar between years, and with no longitudinal
trend (non-significant regression analyses with distance from Site 1).
Benthic algal biomass, however, was lower at four sites in 2014 (40–
140 chla mg m2) compared to 2015 (site t-tests; each test
p < 0.006). Macroalgae dominated algal biomass in both years (54–
95%), comprising mainly of Spirogyra, whilst second ranked were Cla-
dophora, Batrachospermum and Stigeoclonium. Diatoms were the next
most common algal group based on biovolume.
Abundances, taxon richness, rarefied taxon richness, evenness
and diversity were not statistically different between years at each of
the six sites (t-tests), suggesting no difference between the pre- and
post-disturbance October population metrics. An ordination of taxon
assemblage (Figure 8b) and FFG (Figure 8d) abundances for the six
sites and each year shows data points representing the later part of
the recovery period (September to November) separated from the
earlier recovery period (May to August) data points, and site and date
data occupying similar ordination space. Site 2, the deepest site (aver-
age 0.60 m) with the lowest velocity (average 0.32 m s1), and Site
4, a shallow site (average 0.32 m), tended to separate from the other
four sites (Figure 8b, d). ANOSIM tests of macroinvertebrate assem-
blage and FFG abundances in October between the 2 years for each
site were not significantly different.
The October 2014 and 2015 sample dates, however, were,
respectively, 199 and 177 days after the commencement of the
recovery period, which constitutes a lag of 22 days for the 2014 suc-
cession compared to 2015. Three additional tests were conducted
that compared macroinvertebrate assemblages between the 2 years
collected within 6 days of each other after the commencement of the
recovery period. At Site 1, the November 2015 (day 204) and October
2014 (day 199) abundance, diversity metrics (t-tests) and assemblages
(ANISOM test) were not significantly different. Nor were the October
2015 (day 171) and September 2014 (day 177) assemblages different
at Sites 1 and 2. Environmental parameters were also similar for these
day numbers between the 2 years, except algal biomass. Algal bio-
masses at Site 1 in 2014 were approximately half the 2015 values,
whilst Site 2 algal biomass day on day 171 in 2014 was double the
2015 day 177 algal biomass.
TOWNSEND 13 of 18
F I G U R E 8 Ordination plot of
macroinvertebrate taxon and functional
feeding group (gatherer/grazer; filterer,
shredder, predator) abundance
assemblages using non-metric
multidimensional scaling of Bray–Curtis
similarity matrices for Sites 1–6, month of
sample collection (May to November) and
year (2014 indicated by *; otherwise
2015). Taxon and functional feeding
group (FFG) assemblage ordination points
labelled, respectively, as (a, b) month
number and (c, d) site number.
Continuous (Site 1) and dashed (Site 3)
lines represent 2015 assemblage
trajectories, with arrows pointing to
direction of time. Samples were collected
monthly. Stress values for taxon (a, c) and
FFG (b, d) assemblage plots 0.9 and 0.8,
respectively.
F I G U R E 9 Sites 1 (a, b) and 3 (c, d)

macroinvertebrate functional feeding
group relative abundances and absolute
abundances.
14 of 18 TOWNSEND
Overall, the comparison of site environmental hydraulic condi- et al., 2008). This may have then been followed by a phase when bio-
tions, and macroinvertebrate assemblages and FFGs were not signifi- mass declined, equating to the loss phase described by Biggs (1996).
cantly different based on October and comparable day number data The replacement of Cladophora and Batrachospermum dominance by
for 2014 and 2015, though site algal biomasses did differ. Spirogyra could have been due to their greater resistance to flood dis-
turbance, analogous to the replacement of Rhizoclonium by Spirogyra
in a temperate stream (Power & Stewart, 1987).
4 | DISCUSSION
4.1 | Chemical and physical environment 4.3 | Macroinvertebrate recovery
The long period of gradually declining groundwater-fed discharge dur- Macroinvertebrate abundance and biodiversity metrics fitted a loga-
ing the dry-season recovery period produced a benign hydraulic and rithmic, asymptotic relationship for Site 1 data, and for half the met-
water-quality environment, supporting Hypothesis 1. Nutrient con- rics tested for Site 3 data, broadly supporting Hypothesis 3. These
centrations were consistently low, soluble N:P ratios suggest persis- increases were linked to time and macroalgal biomass, which can pro-
tent nitrogen limitation for algal growth, and the high-water clarity vide not only a habitat but also a food resource (Dudley et al., 1986).
and shallow riffle enabled light penetration for benthic algal photosyn- The latter is supported by the findings of a food web study by Jardine
thesis, in common with the neighbouring Daly River et al. (2012) who concluded Ephemeroptera, Odonata and Pyralidae
(Townsend, 2019; Townsend et al., 2017; Townsend & carbon contents were heavily reliant on Spirogyra dominated biofilm
Padovan, 2005). Declining discharges did not approach lentic condi- in the neighbouring Daly River. Autogenic sloughing of macroalgae
tions, which disadvantage rheophilic taxa in the region (Leigh, 2013). may have limited macroinvertebrate abundance and diversity by
In contrast, rivers with shifting chemical conditions can result in removing individuals and decreasing habitat volume (Dudley
rapid biotic responses which redirect succession; for example, the et al., 1986; Suren & Jowett, 2006). Death and Zimmermann (2005)
macroinvertebrate population crash caused by the onset of nitrogen also found a positive relationship, albeit a weak one (r2 = 0.11),
limitation and low-quality food in a desert stream (Grimm & between algal biomass and macroinvertebrate abundance, and linked
Fisher, 1989). The benign physical and chemical environment of the this to canopy cover, stone substrate movement, and recovery time
Roper River and other northern Australian tropical savannah rivers, since flood disturbance. The positive relationship between taxon rich
however, can be interrupted by storm runoff events, such as the hyp- and diversity with algal biomass in the Roper River is not necessary
oxic event of November 2014. These discrete runoff events are infre- causal, but instead coincidental with other influences such dispersion,
quent in the region and tend to occur most often at the end of the dry colonisation, and habitat availability, although there is some evidence
season (Townsend, 1994; Townsend et al., 1992). Roper River hypoxia that macroalgae facilitate macroinvertebrate colonisation during the
was most likely caused by the deoxygenation of runoff waters by the initial phase of recovery (Downes et al., 2000).
microbial decomposition of organic matter which had accumulated Secondary succession of macroinvertebrates followed a sequence
over the dry season, accentuated by the high-water temperatures, of assemblages and FFGs, defined analytically by a linear projection
which have a low oxygen solubility and support high rates of microbial through multi-dimensional space, supporting Hypothesis 4. The
metabolism, compared to colder waters (Townsend, 1994; Townsend observed succession, however, applies only to the mid (May–July) to
et al., 1992). Anthropogenic factors associated with cattle grazing, late (July to November) phases of recovery and excludes the rapid ini-
however, may also have contributed; for example, through the accu- tial phase of colonisation and growth of resistant biota. There is no
mulation of cattle dung. evidence of a major change in macroinvertebrate successional compo-
sition in this study, such as the population crash observed by Grimm
and Fisher (1989), which would be expected to disrupt succession and
4.2 | Benthic algal recovery redirect a new trajectory. The shift in macroalgal composition to Spiro-
gyra dominance in July did not coincide with a major shift in macroin-
Benthic macroalgae in the region have low resistance to flood distur- vertebrate assemblages, though Dudley et al. (1986) have shown
bance (Townsend & Padovan, 2005) and would have commenced macroinvertebrate taxa can be associated with different macroalgal
recovery before the first sample date in May when biomass was sub- genera. Riffle macroinvertebrate succession in the Roper River con-
stantial (median 38 chla mg m2), akin to the Daly River (Townsend trasts with the indistinguishable assemblages between the beginning
et al., 2017; Townsend & Padovan, 2005). Macroalgal colonisation and end of the recovery period of lentic (Leigh, 2013) and lotic (Kay
may have originated from low order streams where macroalgae are et al., 1999) habitats in the region, although the presence/absence
resilient to frequent, wet-season spates (Townsend & Douglas, 2014). data sets of these studies probably reduced the ability to detect
In agreement with Hypothesis 2, benthic algal biomass increased to assemblage differences. The contrasting decline in macroinvertebrate
reach a maximum, which was probably constrained by shear-stress abundance and richness over the recovery period in a shallow stream
forces (Townsend & Padovan, 2005), autogenic sloughing (see Boulê- in the region (Outridge, 1988), which drained a floodplain billabong,
treau et al., 2006) and nutrient limitation (Biggs, 2000; Townsend was probably associated with changing habitat conditions.
TOWNSEND 15 of 18
At a seasonal scale, the gradually declining baseflow of the natural (Leigh, 2013), drying (Sponseller et al., 2010), habitat modification and
flow regime, associated near-stable water quality, facilitated dry- increased complexity (e.g. growth of macrophytes; Blanchette &
season recovery from flood disturbance. During the mid and late Pearson, 2013), and seasonal impacts (Boulton et al., 1992;
periods of recovery, succession was probably driven mainly by biotic Woodward et al., 2015). Recovery was underpinned by the natural
rather than abiotic factors, as proposed by the harsh-benign hypothe- flow regime of the river and facilitated by a benign abiotic environ-
sis of Peckarsky (1983) and discussed by Chang and HilleRisLambers ment. The logarithmic, asymptotic increase of macroinvertebrate
(2016), and suggested by the gatherer/grazer and predator abundance abundance and diversity concurs with the general model of macroin-
correlation and benign abiotic environment. vertebrate recovery to reflect increasing ecosystem production
Mackay (1992) concluded seston filterers and grazers of fine sedi- (Odum et al., 1971).
ment and bacteria would be amongst the first colonisers to establish, The study was limited to a single river reach and needs to be
followed by algal grazers, gatherers, and then finally predators and placed in a wider spatial and temporal context by comprehensively
shredders. In this study, the relative abundance of filterers did not assessing succession over multiple years and in several catchments to
show an overall decline during mid and late phases of recovery, as capture the natural variability of savannah river flood disturbances
would be predicted based on MacKay (1992), but instead no temporal and the environmental conditions for recovery. Leigh (2013) has
trend was evident. Gatherer and grazer FFGs showed no shown that the macroinvertebrate richness of riverine bank and sand
temporal pattern either, though this was could be due to their colla- habitats at the end of the recovery period was correlated with water
tion as a single group. Shredder abundance increased over the recov- quality, as well as wet-season flood-disturbance timing and magni-
ery period, which can be attributed to increased macroalgal food tude. Moreover, inter-annual shifts of riffle macroinvertebrate assem-
resources (Cummins, 2019) and habitat volume for Pyralidae, rather blages, which coincided with low and high flows (Dostine &
than microbially conditioned coarse particulate organic matter pre- Humphrey, 2012), can occur at the end of the recovery period. Cli-
dicted by MacKay (1992), which was not present in the riffles. Preda- mate change will add to inter-annual variability of flood disturbances
tor abundances increased over the recovery period, as hypothesised in northern Australia by increasing both the year-to-year variability of
by Mackay (1992), though there is evidence that abundances declined cyclonic-driven disturbances and the proportion of severe cyclones
late in the recovery period, possibly associated with the loss of macro- (CSIRO, 2020). At the same time, anthropogenic groundwater and
algal habitat volume. dry-season river extractions will decrease discharges over the recov-
The macroinvertebrate assemblage trajectory progressed towards ery period (King et al., 2015).
greater similarity between sample dates, with fewer new taxa in the Perennial tropical savannah rivers like the Roper River are well
late recovery period and decreasing diversity metrics. This suggests a suited to observational studies into secondary succession, due to the
decrease in the successional rate, defined by Boulton et al. (1992) as long recovery period and near-stable abiotic environment. Inclusion of
the degree of change in assemblage composition and observed for tra- the initial recovery phase, not addressed in this study, could elucidate
jectories in a desert stream. A stable or climax macroinvertebrate the relative role of resistant and colonising taxa in setting succession
community was not attained in the Roper River, which concurs with trajectory. A research priority is to understand the relative roles of
the contention that succession rarely reaches an equilibrium abiotic and biotic influences on succession (see Chang &
(Walker & Del Moral, 2003). The assemblage, however, was domi- Turner, 2019; Schuwirth et al., 2016). More specific ecological ques-
nated by a few core taxa in the later phase of recovery tions of interspecific and intraspecific species interactions, and the
(Orthocladiinae, Hydropsychidae, Orthotrichia, Chironominae, Pyrali- drivers of macroinvertebrate population dynamics under benign envi-
dae, Hydroptila), albeit with fluctuating abundances. ronmental conditions may be best addressed by controlled experi-
Based on macroinvertebrate abundance and diversity metrics, and ments. A greater knowledge of the recovery mechanisms from flood
assemblage similarities, each site returned to its pre-disturbance state, disturbance can inform ecologically sustainable development to better
supporting Hypothesis 5. I surmise the comparable preceding flood preserve river biodiversity, prevent ecological degradation and avoid
disturbance, duration of the recovery period, environmental condi- costly restoration works in the future.
tions, and the accrual of filamentous macroalgal biomass collectively
resulted in similar October and day number (since the commencement AC KNOW LEDG EME NT S
of the recovery period) macroinvertebrate assemblages between the Access to the river was kindly provided by the manager of the Lone-
2 years. The November 2014 hypoxic pollution event seemingly had some Dove Station. Field and laboratory work was conducted by staff
negligible impact on riffle recovery following flood disturbance and of the former Aquatic Health Unit of the Department of Environment,
probably removed biota earlier than would have naturally occurred Parks and Water Security (Northern Territory Government). Dr
through flood disturbance. The same event, however, several weeks S. Skinner (Royal Botanic Gardens of Sydney) and G. Przybylska
earlier would be expected to re-set the recovery trajectory. (AlgaeTest Consulting) provided algal identification and enumerations.
This observational study provides an insight into the recovery of The manuscript was improved by discussions with Dr Peter Dostine
riffle biota from flood disturbance over several months. During the (Department of Environment, Parks and Water Security) and com-
recovery, other potentially confounding influences on secondary suc- ments by Dr C. Leigh (Royal Melbourne Institute of Technology Uni-
cession were absent, such the transition from lotic to lentic conditions versity) and the reviewers.
16 of 18 TOWNSEND
DATA AVAI LAB ILITY S TATEMENT Dostine, P. L., & Humphrey, C. L. (2012). Macroinvertebrate responses to
The data that support the findings of this study are available from the reduced baseflow in a stream in the monsoonal tropics of northern
Australia. National Water Commission.
corresponding author upon reasonable request.
Downes, B. J., Lake, P. S., Schreiber, E. S. G., & Glaister, A. (2000). Habitat
structure, resources and diversity: The separate effects of surface
RE FE R ENC E S roughness and macroalgae on stream invertebrates. Oecologia, 123(4),
APHA. (1998). Standard methods for the examination of water and wastewa- 569–581.
ter (20th ed.). United Book Press Inc. American Public Health Associa- Dudgeon, D. (1994). The influence of riparian vegetation on macroinverte-
tion, American Water Works Association, Water Environment brate community structure and functional organization in six new
Federation Guinea streams. Hydrobiologia, 294(1), 65–85. https://doi.org/10.
Biggs, B. F. (1996). Hydraulic habitat of plants in streams. Regulated Rivers: 1007/BF00017627
Research & Management, 12, 131–141. https://doi.org/10.1002/(SICI) Dudley, T. L., Cooper, S. D., & Hemphill, N. (1986). Effects of macroalgae
1099-1646(199603)12:2/3<131::AID-RRR385>3.0.CO;2-X on a stream invertebrate community. Journal of the North American
Biggs, B. J. F. (2000). Eutrophication of streams and rivers: Dissolved Benthological Society, 5(2), 93–106.
nutrient-chlorophyll relationships for benthic algae. Journal of the Eveleens, R. A., McIntosh, A. R., & Warburton, H. J. (2019). Interactive
North American Benthological Society, 19, 17–31. https://doi.org/10. community responses to disturbance in streams: Disturbance history
2307/1468279 moderates the influence of disturbance types. Oikos, 128(8), 1170–
Blanchette, M. L., & Pearson, R. G. (2013). Dynamics of habitats and macro- 1181. https://doi.org/10.1111/oik.05868
invertebrate assemblages in rivers of the Australian dry tropics. Fresh- Faulks, J. J. (2001). Roper River catchment: An assessment of the physical
water Biology, 58(4), 742–757. https://doi.org/10.1111/fwb.12080 and ecological condition of the Roper River and its major tributaries.
Boulêtreau, S., Garabétian, F., Sauage, S., & Sánchez-Pérez, J. (2006). Department of Lands, Planning and Environment.
Assessing the importance of a self-generated detachment process in Flecker, A. S., & Feifarek, B. (1994). Disturbance and the temporal variabil-
river biofilm models. Freshwater Biology, 51, 901–912. ity of invertebrate assemblages in two Andean streams. Freshwater
Boulton, A. J., Peterson, C. G., Grimm, N. B., & Fisher, S. G. (1992). Stability Biology, 31(2), 131–142. https://doi.org/10.1111/j.1365-2427.1994.
of an aquatic macroinvertebrate community in a multiyear hydrologic tb00847.x
disturbance regime. Ecology, 73(6), 2192–2207. https://doi.org/10. Giller, P. S., Sangpradub, N., & Twomey, H. (1991). Catastrophic flooding
2307/1941467 and macroinvertebrate community structure. Verhandlungen der Inter-
Bunn, S. E., & Arthington, A. H. (2002). Basic principles and ecological nationalen Vereinigung für Theoretische und Angewandte Limnologie, 24,
consequences of altered flow regimes for aquatic biodiversity. Environ- 1724–1729.
mental Management, 30, 492–507. https://doi.org/10.1007/s00267- Gotelli, N. J., & Colwell, R. K. (2001). Quantifying biodiversity: procedures
002-2737-0 and pitfalls in the measurement and comparison of species richness.
Chang, C., & HilleRisLambers, J. (2016). Integrating succession and com- Ecology Letters, 4(4), 379–391. https://doi.org/10.1046/j.1461-0248.
munity assembly perspectives. F1000Res, 5, 2294. https://doi.org/10. 2001.00230.x
12688/f1000research.8973.1 Greenwood, M. J., & Booker, D. J. (2015). The influence of antecedent
Chang, C. C., & Turner, B. L. (2019). Ecological succession in a changing floods on aquatic invertebrate diversity, abundance and community
world. Journal of Ecology, 107(2), 503–509. https://doi.org/10.1111/ composition. Ecohydrology, 8(2), 188–203. https://doi.org/10.1002/
1365-2745.13132 eco.1499
Chase, J. M. (2003). Community assembly: When should history matter? Grimm, N. B., & Fisher, S. G. (1989). Stability of periphyton and macroin-
Oecologia, 136(4), 489–498. https://doi.org/10.1007/s00442-003- vertebrates to disturbance by flash floods in a desert stream. Journal of
1311-7 the North American Benthological Society, 8, 293–307.
Clarke, K. R., & Gorley, R. N. (2006). Primer v6: User manual/Tutorial. Plym- Haghkerdar, J. M., McLachlan, J. R., Ireland, A., & Greig, H. S. (2019).
outh Marine Laboratory. Repeat disturbances have cumulative impacts on stream communities.
Connell, J. H., & Slatyer, R. O. (1977). Mechanisms of succession in natural Ecology and Evolution, 9(5), 2898–2906. https://doi.org/10.1002/
communities and their role in community stability and organization. ece3.4968
The American Naturalist, 111(982), 1119–1144. https://doi.org/10. HilleRisLambers, J., Adler, P. B., Harpole, W. S., Levine, J. M., &
1086/283241 Mayfield, M. M. (2012). Rethinking community assembly through the
Cruz, M., & González, V. (2018). Simplified procedure for efficient and lens of coexistence theory. Annual Review of Ecology, Evolution, and
unbiased population size estimation. Plus One, 13(11), e0208359. Systematics, 43(1), 227–248. https://doi.org/10.1146/annurev-
https://doi.org/10.1371/journal.pone.0208359 ecolsys-110411-160411
CSIRO. (2009). Water in the Roper region. In CSIRO. (Ed.), Water in the Jardine, T. D., Pettit, N. E., Warfe, D. M., Pusey, B. J., Ward, D. P.,
Gulf of Carpentaria drainage division: A report to the Australia Govern- Douglas, M. M., & Bunn, S. E. (2012). Consumer–resource coupling in
ment from CSIRO Northern Australian sustainable yields project (pp. 50– wet–dry tropical rivers. Journal of Animal Ecology, 81(2), 310–322.
120). CSIRO. https://doi.org/10.1111/j.1365-2656.2011.01925.x
CSIRO. (2020). The state of the climate 2020. CSIRO. Kay, W. R., Smith, M. J., Pinder, A. M., Mcrae, J. M., Davis, J. A., &
Cummins, K. W. (2019). Functional analysis of stream macroinvertebrates. Halse, S. A. (1999). Patterns of distribution of macroinvertebrate fami-
In D. Gökçe (Ed.), Limnology - Some new aspects of inland water ecology lies in rivers of north-western Australia. Freshwater Biology, 41(2),
(pp. 63–78). Rijeka: IntechOpen. 299–316. https://doi.org/10.1046/j.1365-2427.1999.00432.x
Davies, A. L., & Gee, J. H. R. (1993). A simple periphyton sampler for algal King, A. J., Townsend, S. A., Douglas, M. M., & Kennard, M. J. (2015). Impli-
biomass estimates in streams. Freshwater Biology, 30, 47–51. cations of water extraction on the low-flow hydrology and ecology of
Death, R. G. (2008). The effect of floods on aquatic invertebrate communi- tropical savannah rivers: An appraisal for northern Australia. Freshwa-
ties. In J. Lancaster & R. A. Briers (Eds.), Aquatic insects: Challenges to ter Science, 34(2), 741–758. https://doi.org/10.1086/681302
populations (pp. 103–121). CAB International. Lake, P. S. (2000). Disturbance, patchiness, and diversity in streams. Jour-
Death, R. G., & Zimmermann, E. M. (2005). Interaction between distur- nal of the North American Benthological Society, 19, 573–592.
bance and primary productivity in determining stream invertebrate Lake, P. S. (2013). Resistance, resilience and restoration. Ecological Man-
diversity. Oikos, 111(2), 392–402. agement & Restoration, 14(1), 20–24.
TOWNSEND 17 of 18
Lake, P. S., Bond, N., & Reich, P. (2007). Linking ecological theory with cerrado streams using biological metrics and the index of habitat integ-
stream restoration. Freshwater Biology, 52(4), 597–615. https://doi. rity. Insects, 8(1), 10.
org/10.1111/j.1365-2427.2006.01709.x Rolls, R. J., Leigh, C., & Sheldon, F. (2012). Mechanistic effects of low-flow
Larned, S. T. (2010). A prospectus for periphyton: Recent and future ecohydrology on riverine ecosystems: Ecological principles and conse-
logical research. Journal of the North American Benthological Society, quences of alteration. Freshwater Science, 31, 1163–1186.
29, 182–206. Schult, J., & Novak, P. (2017). Water quality of the Roper River 2012–2016.
Leigh, C. (2013). Dry-season changes in macroinvertebrate assemblages of Palmerston Northern Territory Department of Environment and Natu-
highly seasonal rivers: Responses to low flow, no flow and antecedent ral Resources.
hydrology. Hydrobiologia, 703(1), 95–112. https://doi.org/10.1007/ Schuwirth, N., Dietzel, A., & Reichert, P. (2016). The importance of biotic
s10750-012-1347-y interactions for the prediction of macroinvertebrate communities
Leigh, C., & Sheldon, F. (2009). Hydrological connectivity drives patterns under multiple stressors. Functional Ecology, 30(6), 974–984. https://
of macroinvertebrate biodiversity in floodplain rivers of the Australian doi.org/10.1111/1365-2435.12605
wet /dry tropics. Freshwater Biology, 54(3), 549–571. https://doi.org/ Scrimgeour, G. J., Davidson, R. J., & Davidson, J. M. (1988). Recovery of
10.1111/j.1365-2427.2008.02130.x benthic macroinvertebrate and epilithic communities following a large
Lepori, F., & Malmqvist, B. (2007). Predictable changes in trophic commu- flood, in an unstable, braided, New Zealand river. New Zealand Journal
nity structure along a spatial disturbance gradient in streams. Freshwa- of Marine and Freshwater Research, 22, 337–344.
ter Biology, 52(11), 2184–2195. https://doi.org/10.1111/j.1365-2427. Sponseller, R. A., Grimm, N. B., Boulton, A. J., & Sabo, J. L. (2010).
2007.01846.x Responses of macroinvertebrate communities to long-term flow vari-
Macedo, D. R., Hughes, R. M., Ligeiro, R., Ferreira, W. R., Castro, M. A., ability in a Sonoran Desert stream. Global Change Biology, 16(10),
Junqueira, N. T., Oliviera, D. R., Firmiano, K. R., Pompeu, P. S., & 2891–2900. https://doi.org/10.1111/j.1365-2486.2010.02200.x
Callisto, M. (2014). The relative influence of catchment and site vari- Stevenson, R. J. (1997). Scale-dependent determinants and consequences
ables on fish and macroinvertebrate richness in Cerrado biome of benthic algal heterogeneity. Journal of the North American Bentholo-
streams. Landscape Ecology, 29(6), 1001–1016. https://doi.org/10. gical Society, 16, 248–262.
1007/s10980-014-0036-9 Suren, A. M., & Jowett, I. G. (2006). Effects of floods versus low flows on
Mackay, R. J. (1992). Colonization by lotic macroinvertebrates: A review of invertebrates in a New Zealand gravel-bed river. Freshwater Biology,
processes and patterns. Canadian Journal of Fisheries and Aquatic Sci- 51(12), 2207–2227. https://doi.org/10.1111/j.1365-2427.2006.
ences, 49, 617–628. 01646.x
Marchant, R. (1982). Seasonal variation in the macroinvertebrate fauna of Townsend, S. A. (1994). The occurrence of natural fish kills, and their
billabongs along Magela Creek, Northern Territory. Australian Journal causes, in the Darwin-Katherine-Jabiru region of northern Australia.
of Marine and Freshwater Research, 33, 329–342. Mitteilungen Internationale Verein Limnologie, 24, 197–205.
Moreyra, A. K., & Padovesi-Fonseca, C. (2015). Environmental effects and Townsend, S. A. (2019). Discharge-driven seasonal pattern of ionic solutes,
urban impacts on aquatic macroinvertebrates in a stream of central suspended sediment and water clarity for a tropical savanna river in
Brazilian Cerrado. Sustainable Water Resources Management, 1(2), 125– northern Australia. Marine and Freshwater Research, 70(11), 1585–
136. https://doi.org/10.1007/s40899-015-0013-8 1602. https://doi.org/10.1071/MF19017
Odum, E. P., Odum, H. T., & Andrews, J. (1971). Fundamentals of ecology. Townsend, S. A., Boland, K. T., & Wrigley, T. J. (1992). Factors contributing
Saunders. to a fish kill in the Australian wet/dry tropics. Water Research, 26,
Outridge, P. M. (1988). Seasonal and spatial variations in benthic macroin- 1039–1044.
vertebrates communities of Magela Creek. Northern Territory Townsend, S. A., & Douglas, M. M. (2014). Benthic algal resilience to fre-
Australian Journal of Marine and Freshwater Research, 39, 211–223. quent wet-season storm flows in low-order streams in the Australian
Patrick, C. J., & Swan, C. M. (2011). Reconstructing the assembly of a tropical savanna. Freshwater Science, 33(4), 1030–1042. https://doi.
stream-insect metacommunity. Journal of the North American Bentholo- org/10.1086/678516
gical Society, 30(1), 259–272. https://doi.org/10.1899/09-169.1 Townsend, S. A., & Padovan, A. V. (2005). The seasonal accrual and loss of
Peckarsky, B. L. (1983). Biotic interactions or abiotic limitations. A model benthic algae (Spirogyra) in the Daly River, an oligotrophic river in trop-
of lotic community structure. In T. D. Fontaine, III & S. M. Bartell ical Australia. Marine and Freshwater Research, 56(3), 317–327.
(Eds.), Dynamics of lotic ecosystems (pp. 303–323). Ann Arbor Science. https://doi.org/10.1071/MF04079
Peterson, C. G. (1996). Response of benthic algal communities to natural Townsend, S. A., Schult, J. H., Douglas, M. M., & Lautenschlager, A. (2017).
physical disturbance. In R. J. Stevenson, M. L. Bothwell, & R. L. Lowe Recovery of benthic primary producers from flood disturbance and its
(Eds.), Algal ecology: Freshwater benthic ecosystems (pp. 375–403). implications for an altered flow regime in a tropical savannah river
Academic Press. (Australia). Aquatic Botany, 136, 9–20. https://doi.org/10.1016/j.
Petheram, C., McMahon, T. A., & Peel, M. C. (2008). Flow characteristics aquabot.2016.08.004
of rivers in northern Australia: Implications for development. Journal of Townsend, S. A., Schult, J. H., Douglas, M. M., & Skinner, S. (2008). Does
Hydrology, 357(1–2), 93–111. https://doi.org/10.1016/j.jhydrol.2008. the Redfield ratio infer nutrient limitation in the macroalga Spirogyra
05.008 fluviatilis? Freshwater Biology, 53, 509–520.
Poff, N. L., Allan, J. D., Bain, M. B., Karr, J. R., Prestegaard, K. L., Uehlinger, U. (2000). Resistance and resilience of ecosystem metabolism
Richter, B. D., Sparks, R. E., & Stromberg, J. C. (1997). The natural flow in a flood-prone river system. Freshwater Biology, 45, 319–332.
regime: a paradigm for river conservation and restoration. Bioscience, Vörösmarty, C. J., McIntyre, P. B., Gessner, M. O., Dudgeon, D.,
47, 769–784. Prusevich, A., Green, P., … Davies, P. M. (2010). Global threats to
Power, M. E., Parker, M. S., & Dietrich, W. E. (2008). Seasonal reassembly human water security and river biodiversity. Nature, 468(7321), 334.
of a river food web: floods, droughts, and the impacts of fish. Ecologi- Walker, L. R., & Del Moral, R. (2003). Primary succession and ecosystem
cal Monographs, 78(2), 263–282. https://doi.org/10.1890/06-0902.1 rehabilitation. Cambridge University Press.
Power, M. E., & Stewart, A. J. (1987). Disturbance and recovery of an algal Warfe, D. M., Pettit, N. E., Davies, P. M., Pusey, B. J., Hamilton, S. K.,
assemblage following flooding in an Oklahoma stream. The American Kennard, M. J., Townsend, S. A., Bayliss, P., Ward, D. P.,
Midland Naturalist, 117, 333–345. Douglas, M. M., Burford, M. A., Fin, M., Bunn, S. E., & Halliday, I. A.
Reis, D. F. d., Salazar, A. E., Machado, M. M. D., Couceiro, S. R. M., & de (2011). The ‘wet–dry’ in the wet–dry tropics drives river ecosystem
Morais, P. B. (2017). Measurement of the ecological integrity of structure and processes in northern Australia. Freshwater Biology,
18 of 18 TOWNSEND
56(11), 2169–2195. https://doi.org/10.1111/j.1365-2427.2011. Zaar, U. (2009). Water Resources of the Roper River region (2009/16D).
02660.x Department of Natural Resources, Environment and the Arts.
Weiher, E., & Keddy, P. A. (1995). Assembly rules, null models, and trait
dispersion: New questions from old patterns. Oikos, 74(1), 159–164.
https://doi.org/10.2307/3545686 SUPPORTING INF ORMATION
Welschmeyer, N. A. (1994). Fluorometric analysis of chlorophyll a in the Additional supporting information can be found online in the Support-
presence of chlorophyll b and pheopigments. Limnology and Oceanog-
ing Information section at the end of this article.
raphy, 39(8), 1985–1992. https://doi.org/10.4319/lo.1994.39.8.1985
Williams, B. A., Watson, J. E. M., Beyer, H. L., Grantham, H. S.,
Simmonds, J. S., Alvarez, S. J., … Runting, R. K. (2022). Global drivers of
change across tropical savannah ecosystems and insights into their How to cite this article: Townsend, S. (2024). Riffle algal and
management and conservation. Biological Conservation, 276, 109786. macroinvertebrate recovery from wet-season flood
https://doi.org/10.1016/j.biocon.2022.109786 disturbance in a tropical savannah river, Northern Australia.
Woodward, G., Bonada, N., Feeley, H. B., & Giller, P. S. (2015). Resilience
Ecohydrology, e2617. https://doi.org/10.1002/eco.2617
of a stream community to extreme climatic events and long-term
recovery from a catastrophic flood. Freshwater Biology, 60(12), 2497–
2510. https://doi.org/10.1111/fwb.12592

Riffle Algal and Macroinvertebrate Recovery From Wet-Seasonflood Disturbance in A Tropical Savannah River

Uploaded by

Copyright:

Available Formats

Riffle Algal and Macroinvertebrate Recovery From Wet-Seasonflood Disturbance in A Tropical Savannah River

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Riffle Algal and Macroinvertebrate Recovery From Wet-Seasonflood Disturbance in A Tropical Savannah River

Uploaded by

Copyright:

Available Formats

Received: 20 April 2023 Revised: 25 September 2023 Accepted: 8 January 2024

Riffle algal and macroinvertebrate recovery from wet-season

Simon Townsend 1,2

Correspondence of riffle macroinvertebrates in an Australian perennial, tropical savannah river, located

1 | I N T RO DU CT I O N central role in structuring riverine ecosystems (Lake, 2000). Floods

Ecohydrology. 2024;e2617. wileyonlinelibrary.com/journal/eco © 2024 John Wiley & Sons Ltd. 1 of 18

F I G U R E 1 The Roper River study

been conducted. Over a 1-year period, Marchant (1982) and Outridge 2 | M A T E R I A L S A N D M ET H O D S

F I G U R E 2 Discharge at Red Rock

F I G U R E 3 Water quality and

T A B L E 2 Mean and standard error of

F I G U R E 4 Median benthic agal

F I G U R E 5 Histogram of benthic algal

Abundance (ind cobble1)

Abundance (ind cobble1)

F I G U R E 6 Macroinvertebrate assemblage metrics during flood-

maximum in November. Predator abundances (mostly Empididae and

Hypothesis 5. The macroinvertebrate assemblage will

F I G U R E 7 Relationship between macroinvertebrate assemblage

The October 2014 and 2015 hydraulic environments were similar

F I G U R E 9 Sites 1 (a, b) and 3 (c, d)

4.1 | Chemical and physical environment 4.3 | Macroinvertebrate recovery

You might also like