Hydrobiologia (2009) 623:87–103

DOI 10.1007/s10750-008-9650-3

PRIMARY RESEARCH PAPER

Relationship of fish and macroinvertebrate assemblages

to environmental factors: implications for community
concordance
Dana M. Infante Æ J. David Allan Æ Simon Linke Æ
Richard H. Norris

Received: 26 March 2008 / Revised: 28 October 2008 / Accepted: 8 November 2008 / Published online: 2 December 2008
Ó Springer Science+Business Media B.V. 2008

Abstract Community concordance describes simi- macroinvertebrates from tributaries of two catch-
larity in distributions and abundances of organisms ments in southeastern Michigan having varied
from different taxonomic groups across a region of landscape characteristics. Classifications of fish and
interest, with highly concordant communities macroinvertebrate assemblages resulted in groups
assumed to respond similarly to major environmental distinguished by differences in taxonomic character-
gradients, including anthropogenic stressors. While istics, functional traits, and stressor tolerance of their
few studies have explicitly tested for concordance respective dominant taxa. Biological groups were
among stream-dwelling organisms, it frequently is associated with principal landscape gradients of the
assumed that both macroinvertebrates and fish study region, which ranged from forests and wet-
respond in concert to environmental factors, an lands on coarse surficial geology to agricultural lands
assumption that has implications for their manage- on finer, more impervious surficial geology.
ment. We investigated concordance among fish and Measures of stream habitat indicated more stable
stream flows and greater heterogeneity of conditions
at site groups with catchments comprising forests and
wetlands on the coarsest geology, but did not
distinguish well among remaining site groups, sug-
Handling editor: Robert Bailey
gesting that habitat degradation may not be the
D. M. Infante (&) driving mechanism leading to differences in groups.
Department of Fisheries and Wildlife, Michigan State Despite broadly similar interpretations of relation-
University, East Lansing, MI 48824, USA ships of site groups with landscape characteristics for
e-mail: infanted@msu.edu
both fish and macroinvertebrates, examination of
J. David Allan site representation within groups indicated weak
School of Natural Resources and Environment, community concordance. Our results suggest that
The University of Michigan, Ann Arbor, MI 48109, USA explicit responses of fish and macroinvertebrates to
landscape factors vary, due to potential differences in
S. Linke
The Ecology Centre, School of Integrative Biology, their susceptibility to controls or to differences in the
University of Queensland, St. Lucia, QLD 4072, scale at which landscape factors influence these
Australia organisms.
R. H. Norris
Cooperative Research Centre for Freshwater Ecology, Keywords Stream
Land use
Geology

University of Canberra, Belconnen, ACT 2601, Australia Habitat
Scale
Classification

123
88 Hydrobiologia (2009) 623:87–103

Introduction followed by local abiotic and biotic variables effec-

tively predicted local assemblages of river systems
The distribution and abundance of organisms in (Quist et al., 2005). In the Taieri River basin, South
fluvial systems are influenced by numerous drivers Island, New Zealand, environmental variables that best
operating at multiple spatial scales (Vannote et al., accounted for variation in macroinvertebrate assem-
1980; Tonn et al., 1990; Poff, 1997). Large-scale blages were primarily those at the catchment scale,
controls include biogeographical constraints, which although the presence of pasture land in the riparian
determine the potential species pool for a given zone was also influential (Townsend et al., 2003).
region, as well as landscape features including While the studies above suggest how fluvial fish and
topography, geology, and climate, which can influ- macroinvertebrate assemblages may be responding to
ence smaller-scale factors important to biota such as hierarchical filters, a question that few studies have
type and availability of energy sources, water quality, addressed is whether or not different groups of
and physical habitat (Poff, 1997; Sneldor & Biggs, organisms respond in parallel to these filters. Commu-
2002). Increasingly, researchers recognize that nity concordance describes the degree to which patterns
human activities at the landscape scale can impact in assemblage structure across a given region are
biological assemblages of streams and that this similar among different taxonomic groups (Paszkowski
control results from landscape’s indirect effects on & Tonn, 2000). Concordance would be expected if
factors that influence biota at smaller spatial scales different taxa exhibit similar responses to major
(Allan, 2004; King et al., 2005; Infante et al., 2006). environmental gradients, including anthropogenic
Consequently, protecting aquatic systems from stressors, and would support the use of one taxonomic
anthropogenic impacts depends not only on revealing group as a surrogate for others. While few studies have
causes of impairment, but also on anticipating the explicitly tested for concordance among stream-dwell-
stream biota’s specific responses to stressors operat- ing organisms, it frequently is assumed that both
ing at multiple scales. Recent attempts to meet these macroinvertebrates and fish respond similarly to envi-
needs include studies identifying ecological indicator ronmental factors operating at various scales (e.g.,
metrics (e.g., Karr, 1991; Wright, 1995; Barbour Kilgour & Barton, 1999). Evidence in support of this
et al., 1999; Norris & Hawkins, 2000) and efforts expectation, however, is contradictory. In the case of
comparing effects of anthropogenic land use at the Taieri, strongly dispersing invertebrates showed a
various scales from the stream reach to the entire weaker relationship to large-scale landscape variables
catchment (e.g., Roth et al., 1996; Lammert & Allan, than did invertebrates with poor dispersal abilities,
1999; Sponseller et al., 2001; Wang et al., 2001). potentially due to their greater ability to overcome
Whether large-scale or local drivers are found to geographic barriers (Townsend et al., 2003). Also,
have greater influence over the distribution and community classifications were not found to be
abundance of fluvial organisms varies with the scale concordant across headwater streams in a near-pristine
of investigation. In studies across large regions, factors area of Finland (Paavola et al., 2003). Patterns in
such as climate, geology, and topography may be macroinvertebrate distributions were related mainly to
primary (Legendre & Legendre, 1984; Whittier et al., stream size and pH; bryophytes responded to nutrient
1988; Van Sickle & Hughes, 2000). However, in content and physical habitat variability; and fish were
studies at smaller scales, local conditions may be best correlated with stream depth, substrate size, and
paramount (Townsend et al., 2003; Wang et al., 2003) water oxygen concentration. A final example comes
due in part to relatively less variation in key landscape from a recent study by Paavola et al. (2006) that
drivers across small regions. This accords with the explicitly addressed the role of spatial scale in
view that large-scale and then smaller-scale factors perceived degrees of concordance. Their results related
operate as hierarchical filters in determining assem- concordance among fish, macroinvertebrates, and bry-
blage composition at a site (Tonn et al., 1990; Poff, ophytes to the scale of investigation: across sites from
1997; Sneldor & Biggs, 2002). For example, a model multiple drainages, organism groups showed similar
developed for native fishes of the North Platte River, responses to environmental variables, while within
USA that used large scale biogeographical variables drainages, groups showed less concordance. They

123
Hydrobiologia (2009) 623:87–103 89

attributed this to the fact that organisms may be

responding similarly to regional environmental factors
but differently to local conditions. Further, they
suggested that for studies occurring at smaller spatial
scales, concordance may be observed if organisms
respond similarly to a dominant environmental gradi-
ent, such as a disturbance like anthropogenic land use.
Because environmental assessments are often
based on the response of a single taxonomic group
to anthropogenic stressors and because actions taken
to protect or restore fluvial systems often occur at
relatively small spatial scales (i.e., individual sub-
catchments or reaches), understanding potential Fig. 1 Location of the Huron and Raisin River Basins and 46
study catchments in southeastern Michigan, USA
differences in the response of fish vs. macroinverte-
brates to stressors, as well as the scale at which
stressors originate, is critical for improved assess- areas of 2,330 and 2,780 km2, respectively. The upper
ment and management. Based on those needs, this portions of both basins lie in an area of relatively coarse
study will evaluate the response of fish and macro- surficial geology shaped by glacial activity and include
invertebrates to dominant landscape gradients across ground and end moraines, outwash plains, and areas of
a relatively small region of southeastern Michigan, ice contact. Toward their outlets, the basins drop onto
comparing their responses to environmental factors to the impervious sand and clay lakeplains adjacent
explore concordance among these taxonomic groups. to Lake Erie (Farrand & Bell, 1982). Approximately
Our study has three objectives. First, we determine 28% of the Huron drainage is urbanized, while only
whether classifications of fish vs. macroinvertebrate 12% of the Raisin drainage is urban land (Cifaldi et al.,
assemblages yield groups of sites with distinct, 2004). The Huron also contains almost twice the area
biologically interpretable characteristics by consider- of forest and wetlands as the Raisin (40 vs. 22%).
ing dominant organisms and their functional traits for However, agriculture dominates the Raisin’s land-
each resulting group. We next consider whether the scape, with 63% of the basin devoted to agriculture
separate classifications of fish and macroinvertebrates compared to 25% of the Huron’s landscape.
yield similar groups of sites. Analogous interpreta-
tions of fish and macroinvertebrate site groups having Subcatchment and site selection
similar site membership would lend support to the
question of their concordance in the study region. Forty-six subcatchments draining primarily second-
Finally, we consider relationships of fish and macr- order streams were selected for study from within the
oinvertebrate site groups to catchment and buffer historically glaciated portions of the Huron and Raisin
landscape characteristics and physical habitat of River basins. Site selection emphasized variation in
study sites. Considering potential differences in land cover within the region; however, we excluded
associations between biological groups and environ- subcatchments with high levels of commercial or
mental variables provides additional insights into industrial development. Within each subcatchment,
concordance within the region and also into the role we selected a single site for study at least 1 km above
of landscape factors in structuring these relationships. the stream tributary’s confluence with a larger tribu-
tary. Biological assemblages and physical habitat
were characterized by sampling a 100 m reach of each
Methods study stream.

Study region Biological data

Located in southeastern Michigan (Fig. 1), the Huron Fish were collected during the summers of 1999 and
and Raisin River basins are of similar size, draining 2000 using three-pass electrofishing depletion

123
90 Hydrobiologia (2009) 623:87–103

surveys in blocked reaches. All fish except for two reflecting different hydraulic conductivities. The
native lamprey were identified to species and measure referred to as ‘‘coarse geology’’ includes
recorded as total number captured per 100 m reach. the sum of coarse end and ground moraine, ice
Macroinvertebrates were collected during the spring contact, and outwash—all geologic types that pro-
of 2000 and 2001 with 250-lm mesh D-nets by mote a stable supply of groundwater to streams. The
sampling all habitat types present in proportion to variable referred to as ‘‘fine geology’’ is the sum of
their relative abundance. A subsample of 500–700 proportions of fine end and ground moraines, and
macroinvertebrates from each site was identified to clay and sand lakeplain, geologic types that tend to
the lowest practical taxonomic level. Non-insect taxa have low permeability and thus encourage greater
were typically identified to order or family, and amounts of surface runoff.
remaining insect taxa were identified to genus, with Land use was quantified for each study subcatch-
exceptions including Chironomidae and Ceratopog- ment and for 200 m wide buffers for the length of the
onidae which were identified to family (Wood, 2002). streams based primarily on 1978 data (Michigan
Resource Information System (MIRIS), 1978).
Physical habitat Inspection of 2001 aerial photography verified the
accuracy of the earlier land cover classification for
Physical habitat features of the study reaches were agricultural and natural (forest, wetland) categories.
summarized by 22 variables grouped into four cate- Urban land use was updated with aerial photos from
gories. Channel shape descriptors, including reach 1995 for the majority of the study region and from
averages for bankfull and low flow channel width, 1998 for Lenawee County (Cifaldi et al., 2004).
depth, and cross-sectional area were determined
following methods described in Infante et al. (2006). Data analysis
Pebble counts were performed through the thalweg
channel, and quantitative measures of substrate con- Analyses were based on relative abundances of fish
sidered in analysis included median particle size and and macroinvertebrates. Rare taxa, defined as those
proportions of fines less than 2 mm in diameter, that occurred at fewer than 10% of sites sampled
gravel, cobble, and coarse substrate greater than (four study sites), were excluded from analysis
180 mm in diameter. Frequency of occurrence of because their presence can result in excessive split-
riffle habitat and wood were summarized from counts ting in cluster analysis (Hawkins et al., 2000). This
at 5 m intervals through stream reaches. Finally, resulted in inclusion of 74 macroinvertebrate taxa
physical habitat was assessed using a visual habitat (out of 182 total) and 26 fish species (out of 41 total).
assessment protocol (Michigan Department of Envi- Before analysis, physical data were log-transformed,
ronmental Quality, 1997). This protocol scores overall as were biological data to down-weight the influence
habitat quality by combining scores from nine indi- of highly abundant taxa (Norris & Georges, 1993).
vidual components including amount and diversity of To consider biological characteristics shared by
available cover; substrate embeddedness; variability groups of sites, we classified sites based on homo-
in velocity and depth; flow stability; bottom deposition geneity of their faunal composition. Classification
and sedimentation; diversity of pools, riffles, runs and was performed in PATN (Belbin, 1992) using the
bends; bank stability; bank vegetation stability; and flexible unweighted pair-groups and arithmetic aver-
streamside cover. For analysis, we considered both the ages (UPGMA) fusion strategy recommended by
total habitat score and scores for individual metrics. Belbin & McDonald (1993), with flexible-Beta set at
-0.1 to optimize the hierarchy. This approach
Landscape variables equally weights site groups during the agglomerative
clustering process, regardless of their size (Lance &
Data describing the landscape within subcatchments Williams, 1967). The Bray-Curtis associating mea-
were obtained using a Geographic Information Sys- sure was used for the dissimilarity matrix because it
tem (GIS). Surficial geology of each subcatchment, ignores joint absences, which are predominant in taxa
determined from maps compiled by Farrand & Bell matrices (Faith et al., 1987). Classifications were
(1982), was summarized into two categories performed separately for fish and macroinvertebrates.

123
Hydrobiologia (2009) 623:87–103 91

To identify dominant environmental variables and 1997; Linke et al., 2005), the corroboration of results
taxa associated with sites, we first used hybrid by multiple analysis techniques lends confidence to
multidimensional scaling (HMDS, Faith et al., the conclusions. To gain additional insights into
1987) in PATN (Belbin, 1992) to ordinate sites based concordance of fish and macroinvertebrates, we
on their fish and macroinvertebrate assemblages. A compared dissimilarity of taxa matrices using a
Monte Carlo simulation (MCSS with 100 permuta- Mantel’s test in PC-ORD 4.0 (McCune & Mefford,
tions, Belbin, 1992) was performed on the resulting 1997). We also considered patterns in site inclusion
ordinations to determine the probability that ordina- in fish vs. macroinvertebrate groupings with contin-
tions explained structure in the data that might have gency analysis performed in SPSS 15.0.
occurred by chance alone (Faith, 1990). Relationships
between individual taxa and positions of sites in the
respective ordination spaces generated for fish and Results
macroinvertebrates were then determined using prin-
cipal axis correlation (PCC, Belbin, 1992, see also Site groupings based on macroinvertebrate
Faith & Norris, 1989). Using PCC, we also deter- assemblages
mined relationships between environmental variables
and positions of sites. Monte Carlo significance tests Sites were clustered into four groups based on
(MCAO with 100 permutations, Belbin, 1992) were similarity in relative abundances of macroinverte-
performed to test the significance of correlation brate taxa, and 23 taxa were identified by PCC as
values obtained in the PCC procedure so that those being significantly associated with variation in
variables most closely associated with the structure of macroinvertebrate assemblages of the study sites
the biological assemblages could be selected. Only (Table 1). Taxa with highest average relative abun-
those environmental variables and fish species with a dances for sites in Group 1 included Gammarus
significance of 0.05 or less were used for interpreta- (Amphipoda) and oligochaetes. Chironomids, which
tion. For macroinvertebrate taxa, those with a were the most abundant invertebrate collected from
significance of 0.005 or less were considered due to all study sites, and gastropods reached their highest
the large number of taxa selected at the lower level of average relative abundances at Group 2 sites, and
significance. none of the Group 2 sites supported either of the two
We next used multiple strategies to determine genera of Plecoptera identified by PCC. Group 3 had
biological and physical characteristics of individual higher median taxa richness than sites of any other
site groups. First, we identified biological character- group (Fig. 2), and 13 taxa reached their highest
istics of groups by comparing average relative average relative abundances at Group 3 sites, includ-
abundances of taxa identified by the PCC for each ing eight genera of Ephemeroptera, Plecoptera, and
group. Also, relative abundances of common taxa Trichoptera (EPT taxa). Only one taxon, Simulium
were summarized by their functional traits and (Simuliidae), was most abundant at Group 4 sites.
averaged across site groups to lend additional support Comparisons of functional traits of all common
to biological interpretations. For this step, macroin- macroinvertebrate taxa including tolerance to stress-
vertebrate trait assignments were made from ors and habit and trophic preferences provide
information summarized by Vieira et al. (2006) and additional insights into group differences. With
fish trait assignments were made from information greater taxonomic richness, Group 3 sites also had
summarized by Goldstein & Meador (2004) and Poff higher median relative abundance of macroinverte-
& Allan (1995). Finally, we determined dominant brates with low tolerance to stressors than other site
physical characteristics by comparing average values groups (Figs. 2 and 3, respectively). Also, sprawlers,
of PCC-identified habitat and landscape variables swimmers, and clingers were more abundant at
across groups. Group 3 sites, as were shredders, scraper-grazers,
Because classification analysis has limitations and predators (Fig. 4). In contrast, organisms that
linked to offsets, chaining effects, and input order, were classified as burrowers and collector-gatherers
as well as theoretical concerns about the occurrence (including chironomids) were most abundant at
of biological organisms in distinct groupings (Podani, Group 2 sites where they averaged more than 80%

123
92 Hydrobiologia (2009) 623:87–103

Table 1 Taxa identified by principal axis correlation as significantly (P B 0.005) associated with variation in macroinvertebrate
assemblages
Order Family Genus Group average relative abundance (%)
1 2 3 4

Amphipoda Gammaridae Gammarus 15.65 0.04 5.89 0.30

Bivalva 0.89 0.24 0.04
Odonata Coenagrionidae Enallagma 0.14 0.13 0.03
Oligochaeta 10.61 4.14 2.04 1.30
Coleoptera Elmidae Dubiraphia 0.81 4.26 0.35 0.07
Diptera Ceratopogonidae 0.31 1.20 0.20 0.11
Diptera Chironomidae 57.71 82.37 55.34 74.43
Gastropoda 0.75 1.17 0.10 0.19
Odonata Calopterygidae Colopteryx 0.38 0.59 0.19 0.17
Coleoptera Elmidae Stenelmis 0.27 0.08 2.11 0.53
Coleoptera Scirtidae Scirtes 0.13 0.24
Diptera Tabanidae Crysops 0.41 0.24 1.39 0.36
Diptera Tipulidae Pseudolimnophila 0.05 0.81 0.14
Diptera Tipulidae Tipula 0.04 0.09 0.01
Ephemeroptera Baetidae Baetis 0.84 0.22 8.16 3.38
Ephemeroptera Heptageniidae Stenonema 0.02 0.24 0.55 0.12
Plecoptera Nemouridae Amphinemura 0.08 0.93 0.09
Plecoptera Perlidae Perlesta 0.07 1.24 0.07
Trichoptera Hydropsychidae Ceratopsyche 0.07 0.70
Trichoptera Hydropsychidae Cheumatopsyche 0.64 0.98 0.94
Trichoptera Hydropsychidae Hydropsyche 0.15 0.08 0.56 0.16
Trichoptera Philopotamidae Chimarra 0.01 0.29
Diptera Simuliidae Simulium 4.12 0.38 4.11 12.03
Included are each taxon’s average relative abundance in each site group, and taxa are sorted to indicate sets of organisms with highest
average abundances per group

of organisms collected; sites in Group 2 also had the through sample reaches), descriptors of hydrology
lowest abundances of low-tolerance organisms (channel incision, metrics scoring flow and bank
(Fig. 3). Sites in Group 1 had high abundances of stability), and embeddedness of stream bed substrate.
crawlers and omnivores, reflecting the dominance of Average scores for habitat assessment metrics were
amphipods at the sites (Fig. 4). Similarly, sites in highest across Group 3 sites, and channel incision
Group 4 were dominated by collector-filterers, which was lowest, together indicating more complex hab-
include simuliids. itat, more stable hydrology, and lower embeddedness
at the Group 3 sites compared to other groups. Sites
Environmental variables characterizing in Group 2 had lowest average scores for all of the
macroinvertebrate clusters visual metrics estimating habitat quality, while scores
for sites in Groups 1 and 4 were mid-range, with
The PCC identified seven habitat variables that were Group 1 sites generally having slightly higher scores
significantly related to variation in macroinvertebrate than Group 4 sites.
assemblages (Table 2). These included descriptors of Seven landscape variables were also identified by
habitat complexity (variability in velocity and depth PCC that were significantly related to macroinverte-
and a diversity of pools, riffles, runs, and bends brate assemblages, and trends across site groups

123
Hydrobiologia (2009) 623:87–103 93

Fig. 2 Box-and-whisker plots showing taxa richness arranged

from highest to lowest median values for each macroinverte- Fig. 3 Box-and-whisker plots showing relative abundance of
brate (a) and fish (b) group organisms with low tolerance to stressors arranged from
highest to lowest median values for each macroinvertebrate
(a) and fish (b) group
emphasized interrelationships in natural and anthro-
pogenic landscape gradients throughout the study Site groupings based on fish assemblages
region. Sites in Group 3, for example, had highest
percentages of both undeveloped lands (including Sites were clustered into five groups based on
wetlands and forests) and coarse surficial geology in similarity in relative abundances of fish species, and
their catchments (Table 2). In contrast, Group 2 sites PCC identified seven species that were significantly
had the finest surficial geology and the least undev- associated with variation among assemblages
eloped lands. For Groups 1 and 4, permeability of (Table 3). Sites of Fish Group 1 supported the
surficial geology and natural landscape cover were highest average relative abundance of Rhinichthys
mid-range compared to Groups 2 and 3. atratulus; while present at Group 2 sites, these fish

123
94 Hydrobiologia (2009) 623:87–103

Fig. 4 Differences in macroinvertebrate habitat preferences (a) and trophic guilds (b) by groups. Y-axes indicate relative abundance
(%) of macroinvertebrates in each functional category averaged across site groups

were not collected from sites in other groups. As with macroinvertebrates, comparisons of func-
Etheostoma nigrum and Campostoma anomalum tional traits of common fish species also suggest
were more abundant at Group 2 sites compared to ecological differences among site groups. Group 2
sites in other groups, and Group 4 sites supported sites supported common assemblages with a wider
higher abundances of two centrarchids, Lepomis range of trophic preferences than assemblages of
macrochirus and L. gibbosus. Cottus bairdi domi- other sites (Fig. 5). Also, Groups 2 and 3 supported
nated the assemblages of the Group 5 sites, averaging assemblages with more diverse substrate preferences
91% of all individuals captured, and Luxilus cornutus than fish from other site groups. In terms of habitat
averaged 9% of individuals collected from Group 3 specialization, Group 3 sites had more fish preferring
sites. Species richness varied across Groups 1, 4, and riffles than other groups, while fish preferring pools
5, but Groups 2 and 3 had similar and higher median and backwater dominated Group 4 sites. Group 4
richness (Fig. 2). sites also supported the most fish preferring

123
Hydrobiologia (2009) 623:87–103 95

Table 2 Site group

Variable category Variable Group average value
average values of physical
variables identified by PCC 1 2 3 4
as significantly associated
with variation in Channel shape Channel incision (m) 0.37 0.48 0.29 0.48
macroinvertebrate Visual habitat assessment Embeddedness/siltation 10.0 7.3 13.6 8.9
assemblages (P B 0.05)
Velocity/depth variability 11.0 9.0 13.6 10.1
Flow stability 7.7 4.7 9.1 7.2
Pools, riffles, runs, bends 6.9 6.3 9.2 7.0
Bank stability 5.9 2.7 6.1 4.2
Bank vegetative stability 7.5 5.0 7.6 6.2
Natural landscape Slope 0.0022 0.0024 0.0030 0.0023
Coarse surficial geology (%) 51.13 4.10 62.98 20.64
Fine surficial geology (%) 39.20 95.90 20.39 47.04
Catchment land use Nonforested wetlands (%) 13.45 9.31 17.93 12.98
Wetlands (%) 10.48 3.95 13.63 5.91
Undeveloped lands (%) 33.98 21.92 43.48 29.58
Buffer land use Undeveloped lands (%) 47.02 38.10 61.62 47.40

Table 3 Species identified by principal axis correlation as significantly (P B 0.05) associated with variation in fish assemblages
Family Genus Species Group average relative abundance (%)
1 2 3 4 5

Cyprinidae Rhinichthys atratulus 20.48 14.44

Percidae Etheostoma nigrum 4.61 6.36 1.60 1.33
Cyprinidae Campostoma anomalum 0.08 5.25 2.66
Cyprinidae Luxilus cornutus 0.92 9.21
Centrarchidae Lepomis macrochirus 2.31 5.34 11.76 23.72 0.40
Centrarchidae Lepomis gibbosus 0.56 0.53 1.01 2.66
Cottidae Cottus bairdi 34.28 16.07 11.34 0.00 91.08
Included are each species’ average relative abundance in each site group, and species are sorted to indicate sets of organisms with
highest average abundances per group

vegetation and the most with variable substrate streams with larger ratios may have more variable
preferences. Finally, in terms of tolerance to stress- flow regimes than similarly sized streams with
ors, sites in Group 3 supported the highest median smaller ratios. Average channel fit values were
abundance of fish with low tolerance for stressors lowest at Group 3 sites and highest at Group 5 sites,
(Fig. 3). indicating that conditions were hydraulically more
favorable at Group 3 sites and least favorable at
Environmental variables characterizing fish Group 5 sites. The range in group averages of the
clusters second variable identified by the PCC, bank stability,
followed the same trend as channel fit; stream banks
PCC identified three habitat variables significantly were most stable for sites of Group 3 and least stable
related to variation in fish assemblages (Table 4). for Group 5 sites. The third variable identified by the
First, channel fit, a variable calculated as the ratio of PCC was a visual metric scoring the diversity of
bankfull to low flow stream widths, is an estimate of pools, riffles, runs, and bends throughout stream
the fit of the low flow hydraulics to the available reaches. Group 3 sites had the highest score for
channel cross section (Infante et al., 2006), and diversity of habitat units, and Group 5 sites, the least.

123
96 Hydrobiologia (2009) 623:87–103

Fig. 5 Differences in fish

trophic guilds (a), channel
unit (b), and substrate
preferences (c) by groups.
Y-axes indicate relative
abundance (%) of fish in
each functional category
averaged across site groups

123
Hydrobiologia (2009) 623:87–103 97

Table 4 Site group average values of physical variables identified by PCC as significantly associated with variation in fish
assemblages (P B 0.05)
Variable category Variable Group average value
1 2 3 4 5

Channel shape Channel fit 1.52 1.42 1.26 1.38 1.99

Visual habitat assessment Pools, riffles, runs, bends 6.8 8.3 9.7 7.3 5.3
Bank stability 5.1 4.9 6.6 6.2 4.3
Natural landscape Coarse geology 36.52 31.98 88.18 61.67 28.33
Catchment land use Agriculture (%) 47.95 55.49 20.94 31.25 51.63
Buffer land use Agriculture (%) 34.29 45.96 9.02 19.28 34.84

As with macroinvertebrates, the landscape vari- macroinvertebrate Group 3, characterized by high

ables identified by PCC related to variation in fish median taxa richness, the greatest number of
assemblages (coarse surficial geology and agriculture, intolerant taxa, and highest site-average percentages
Table 4) show the link between geology and land use of catchment coarse surficial geology and natural
across the study region. Sites in Group 3 had the land, included 19 sites. Only seven of these sites
highest percentages of coarse surficial geology and were included in Fish Group 3 which had the most
the least agriculture in their catchments, while sites in closely matching characteristics (high species rich-
Groups 2 and 5 had comparatively fine surficial ness and numbers of intolerant fish, highest site-
catchment geology and more agriculture. average percentages of catchment coarse surficial
geology, and lowest of agricultural land). While all
Concordance among fish and macroinvertebrate seven sites of Fish Group 3 were included in
assemblages Macroinvertebrate Group 3, the remaining 12 sites
were included in Fish Groups 1, 2, and 4. Similarly,
Mantel’s statistic (r = 0.083) indicated that the the 18 sites of Fish Group 2, with the most
dissimilarity matrix of fish relative abundances agriculture and most fine surficial geology in their
across study sites was not correlated with that of catchments, were contained within all four macro-
macroinvertebrates. Also, using contingency analy- invertebrate groups. Together, Mantel’s test and
sis, site overlap in fish and macroinvertebrate groups consideration of the extent of correspondence of
was not strong, even when considering groups with sites between fish and macroinvertebrate groups
similar characteristics (Table 5). For example, support the interpretation that these two organismal

Table 5 Site overlap between each fish and macroinvertebrate group

Macroinvertebrates Total
Coarsest geology, most natural land cover Finest geology, least natural land cover
Group 3 Group 1 Group 4 Group 2

Fish
Coarsest geology, Group 3 7 0 0 0 7
least agriculture Group 4 2 7 0 0 9
Group 1 4 3 2 0 9
Group 5 0 0 3 0 3
Finest geology, Group 2 6 5 4 3 18
most agriculture Total 19 15 9 3 46
Results are ordered to reflect ranges in group average values for land use and surficial geology for comparison of fish and
macroinvertebrate groups sharing similar landscape characteristics

123
98 Hydrobiologia (2009) 623:87–103

groups, which are widely used to evaluate stream influences or to differences in the scale at which
condition, exhibit at best weak concordance across landscape factors affect these organisms.
our study region.

Response of biological assemblages to landscape

features
Discussion
Although PCC identified different individual land-
Cluster analyses of fish and macroinvertebrate scape variables that explained variation in fish vs.
assemblages from 46 stream sites in southeastern macroinvertebrate assemblages, the variables repre-
Michigan resulted in distinct biological groups char- sented the same general gradient in landscape
acterized by differences in dominant taxa, taxa conditions for both taxonomic groups. The extent of
richness, and functional traits of organisms including undeveloped lands, including forests and wetlands,
their trophic niches, habitat preferences, and toler- and catchment surficial geology were identified by
ance to stressors. Within taxonomic categories, PCC as explaining patterns in macroinvertebrates,
biological differences were related to differences in while variation in extent of agriculture and coarse
natural and anthropogenic landscape factors. For both catchment surficial geology were identified for fish.
fish and macroinvertebrates, biological assemblages However, across our study region, patterns of land
characterizing sites draining catchments with more use and surficial geology are strongly inter-correlated
agricultural land use and finer surficial geology with one another (Diana et al., 2006). Agricultural
generally had less taxonomic diversity, had less land use is the dominant anthropogenic landscape
diversity in functional traits, and were more tolerant gradient and occurs most commonly in catchments
of stressors than assemblages of sites draining comprising fine surficial geology, whereas undevel-
catchments with more natural lands (i.e., forests, oped lands are more common in catchments
wetlands) and coarser surficial geology. Also, PCC comprising coarser geology. This is likely due to
identified patterns in physical habitat suggestive of the fact that finer, less pervious surficial geology is
mechanistic relationships between landscape factors more suitable for agriculture than coarser, more
and biology: habitat conditions at sites draining pervious areas of the study region (Allan, 2004).
catchments with more anthropogenic land use and Consequently, we conclude that fish and macroinver-
fine surficial geology were generally suboptimal, with tebrate assemblages respond to the same gradient in
more variable flow regimes and less habitat hetero- landscape conditions, which ranges from undevel-
geneity, compared to sites in catchments with more oped lands on coarse, well-drained surficial geology
natural lands and coarse geology. to agricultural lands on finer, more impervious
Despite similarities in interpretations of biological surficial geology.
groups derived for both sets of organisms and in their The fish and macroinvertebrate site groups with
relationships with catchment landscape attributes and the least amount of catchment and buffer agriculture
physical habitat of sites, site overlap between fish and (i.e., most forest and wetlands) and the coarsest
macroinvertebrate groups sharing similar interpreta- surficial geology had biological attributes suggestive
tions was not strong. As shown by Mantel’s test, of least impacted environments. Macroinvertebrate
distributions of fish across study sites were not Group 3 and Fish Group 2 both had high median taxa
significantly correlated with distributions of macro- richness compared to sites in other groups, supported
invertebrates. Also, as suggested by the contingency comparatively large numbers of organisms with low
table, the arrangement of sites into groups revealed tolerance to stressors, and generally supported organ-
limited biological correspondence among sites. isms with more varied habit and trophic preferences.
Together, these results indicate weak community This result is consistent with expectations suggested
concordance and suggest that, within our study by various multimetric approaches for assessing
region, explicit responses of fish and macroinverte- biological integrity that least impacted environments
brates to landscape factors differ, indicating potential should support biological assemblages with greater
differences in their susceptibility to landscape taxonomic and functional diversity than those of

123
Hydrobiologia (2009) 623:87–103 99

impacted environments (Hilsenhoff, 1982; Lyons, 2003; Rios & Bailey, 2006). However, the percentage
1992; Lyons et al., 2001). of agricultural land use at which impacts are observed
In contrast, fish and macroinvertebrate site groups varies widely. Negative effects have been associated
with higher amounts of catchment agriculture and with as little as 20% buffer agriculture (Fitzpatrick
finer surficial geology showed signs of faunal degra- et al., 2001), while other studies have found its
dation. For example, sites of Macroinvertebrate effects to be minimal below 30% (Quinn, 2000) or
Group 4, with less than half as much coarse surficial even 50% agricultural land in the catchment (Wang
geology in their catchments as other site groups, had et al., 1997; Meador & Goldstein, 2003). While the
larval blackflies (Simuliidae) as their only dominant influence of agriculture may not be adequately
taxon, and these organisms were three times as captured by a simple percentage, the ecological
common at Group 4 sites as in other site groups. effects of agriculture generally are apparent when it
Chironomids dominated the assemblages of sites in exceeds 50% of catchment area (Allan, 2004).
Macroinvertebrate Group 2 (82%, Table 1) which
had the least natural land and the finest catchment Landscape’s influence on biota via effects
surficial geology; this group also had the lowest on intermediate factors
median macroinvertebrate taxa richness of all groups.
Fish Group 5, with comparatively low amounts of One of the ways in which landscape factors may be
coarse surficial geology and high amounts of agri- influencing fish and macroinvertebrates of our study
culture, also supported the fewest fish species of all region is through effects on the physical stream
groups. However, while having similar landscape environment, including direct effects on flow regimes
characteristics as Group 5, Fish Group 2 had a and related, indirect effects on physical habitat
median species richness equivalent to that of Fish features. For groups of sites with a greater extent of
Group 3, implying that decreased species richness catchment agriculture and fine surficial geology,
may not always be a resulting impact of agricultural measures of channel incision (for macroinvertebrate
land use on fish assemblages. Further, these results groups) and channel fit (for fish groups) indicated
suggest that fish and macroinvertebrate assemblages more variable flow regimes than sites draining natural
may be responding to stressors associated with landscapes with greater amounts of coarse geology.
agricultural land use in different ways. Despite a Also, as shown for both fish and macroinvertebrate
median taxa richness similar to Fish Group 3, groups, sites in catchments with more agriculture and
assemblages of sites in Fish Group 2 generally fine geology had relatively less habitat heterogeneity
supported more fish that were tolerant of stressors as indicated by habitat assessment metrics. However,
(Fig. 3). While leading to a decrease in numbers of differences in habitat features across site groups
intolerant fish species, agricultural land use may have considered degraded based on biological characteris-
resulted in favorable conditions for tolerant fish tics were generally not pronounced. Although Fish
species. A loss of intolerant fish accompanied by a Groups 1, 2, 4, and 5 and Macroinvertebrate Groups
gain in tolerant, native fish has been identified as a 1, 2, and 4 received lower total scores for habitat
potential factor contributing to the homogenization of assessment metrics than their counterparts draining
biotic assemblages in freshwater systems (Scott & the most natural land and coarsest geology, only
Helfman, 2001; Olden & Poff, 2004). This trend has Macroinvertebrate Group 2, with the lowest average
been documented for fish assemblages impacted by amount of coarse surficial geology and a low
urban land use in the Etowah River Basin, Georgia percentage of undeveloped land, received a substan-
(Walters et al., 2003), and described as occurring tially lower score for flow stability; bank stability;
throughout the southeastern United States (Scott & and diversity in pools, riffles, runs, and bends than
Helfman, 2001). other site groups. This broad similarity in habitat may
Numerous studies document the departure of result from the interrelationship between catchment
fluvial biological assemblages from reference condi- agriculture and fine surficial geology in the region
tions with increasing extent of agricultural land and similar effects that these landscape factors can
within the catchment (Richards et al., 1996; Roth have on stream environments. For example, both
et al., 1996; Stewart et al., 2001; Townsend et al., tillage and drainage practices in agricultural regions

123
100 Hydrobiologia (2009) 623:87–103

and greater imperviousness associated with finer larger scale is a common inference in stream ecology
surficial geology can encourage surface runoff. (e.g., Kilgour & Barton, 1999), arising from key
Consequently, storm flows may be more powerful differences in characteristics of these two groups of
and more erosive, encouraging siltation and poten- organisms. Compared to fish, macroinvertebrates
tially less habitat heterogeneity in agricultural may respond primarily to localized stressors because
catchments or catchments with finer surficial of shorter life cycles and more limited mobility
geology. during stages when they are restricted to stream
Comparatively stable flows and greater physical environments. Fish may have a greater capacity to
habitat heterogeneity help to distinguish the two site avoid localized disturbances, moving out of impacted
groups with the most natural land and coarse surficial areas or into refuges of good habitat for certain
geology from remaining site groups; however, dif- portions of their life cycle (Schlosser, 1995; Fausch
ferences in habitat features across these remaining et al., 2002).
groups appear insufficient to fully account for the Several studies support the view that fish and
differences in biological assemblages that result from macroinvertebrates are influenced by environmental
the classification. Other factors not accounted for by factors operating at different scales. In an investigation
this study may be contributing to differences in of lakes in the northeastern USA, Allen et al. (1999)
assemblages, perhaps including altered thermal found that birds, fish, and macroinvertebrates were
regimes, elevated nutrient levels, or contamination associated with large-scale measures including cli-
by pesticides or organic matter resulting from agri- mate and land cover but that diatoms and zooplankton
cultural landscapes (Lenat, 1984; Osborne & Wiley, responded more strongly to local-scale variation
1988; Johnson et al., 1997). Yoder & Rankin (2004) including pH and lake depth. Townsend et al. (2003)
report an intriguing paradox from similar studies in showed that macroinvertebrate assemblages in New
Ohio that support this conclusion. While habitat Zealand streams, while influenced by catchment scale
measures were the best predictors of spatial variation measures, were more strongly affected by anthropo-
in an Index of Biotic Integrity (IBI) for fish assem- genic land uses measured at smaller scales, and that
blages within their study region, sites with geographical location was most strongly correlated
*20 years of monitoring experienced improvement with fish assemblage composition throughout the
in the IBI but not in habitat quality, suggesting that region. Johnson & Goedkoop (2002) showed that,
improvements in water quality may have benefited while large-scale factors were important, local-scale
the biota of the region even when habitat remained measures explained more variance in littoral macro-
suboptimal. invertebrate assemblages. In contrast, the distribution
of stream fish in Portugal indicated the greater
Community concordance importance of large-scale environmental factors com-
pared with microhabitat characteristics (Magalhaes
Although the biological assemblages of the Huron et al., 2002). Recently, a study by Burcher et al.
and Raisin formed distinct groups that could be (2007) showed that fish in North Carolina streams
related to ranges in landscape condition, results from were more closely associated with alterations in
the contingency analysis indicated weak concordance stream reach morphology resulting from landscape
between site groups formed from the two sets of disturbance than macroinvertebrates which were asso-
organisms, suggesting that within this study region, ciated with alterations in erosional substrate measured
fish and macroinvertebrate assemblages are respond- at the habitat patch scale. In the present study, PCC
ing to landscape factors in discernibly different ways. analysis showed macroinvertebrate assemblages to be
This weak concordance could reflect scale-dependent significantly associated with three times as many site-
differences in the way that fish and macroinverte- specific local scale habitat metrics as were fish.
brates are influenced by disturbances resulting from We suggest that the scale of this study influenced
agricultural landscapes or by the effects of fine the outcome of our investigation of community
surficial geology. That macroinvertebrates may concordance as well as the degree to which we
respond to local-scale habitat conditions while fish associated biological groups with gradients in land-
are more strongly influenced by impairment at a scape factors. This finding is supported by a recent

123
Hydrobiologia (2009) 623:87–103 101

work by Paavola et al. (2006), who showed that Belbin, L. & C. McDonald, 1993. Comparing three classifi-
concordance among fish, macroinvertebrates, and cation strategies for use in ecology. Journal of Vegetation
Science 4: 341–348.
bryophytes was stronger across multiple river basins Burcher, C. L., H. M. Valett & E. F. Benfield, 2007. The land-
spanning several ecological regions and weaker cover cascade: relationships coupling land and water.
throughout an individual watershed. Based on litera- Ecology 88: 228–242.
ture reviewed above, it appears that concordance Cifaldi, R. L., J. D. Allan, J. D. Duh & D. G. Brown, 2004.
Spatial patterns in land cover of exurbanizing watersheds
between different taxonomic groups in their response in southeastern Michigan. Landscape and Urban Planning
to environmental gradients may be most readily 66: 107–123.
detected in studies that span a broad geographic Diana, M., J. D. Allan & D. M. Infante, 2006. The influence of
region, allowing for strong gradients in climate, physical habitat and land use on stream fish assemblages
in southeast Michigan. In Hughes, R. M., L. Wang & P.
natural vegetation patterns, and geology. Numerous W. Seelbach (eds), Influences of Landscapes on Stream
classifications performed across large regions have Habitats and Biological Assemblages. American Fisheries
identified natural factors important to stream biota that Society, Bethesda, Maryland.
varied little in this study of the Huron and Raisin River Faith, D. P., 1990. Benthic macroinvertebrates in biological
surveillance: Monte Carlo significance tests on functional
basins, including climate and vegetation patterns in groups’ responses to environmental gradients. Environ-
Quebec (Legendre & Legendre, 1984); geology, soils, mental Monitoring and Assessment 14: 247–264.
and vegetation in Oregon (Whittier et al., 1988; Van Faith, D. P. & R. H. Norris, 1989. Correlation of environmental
Sickle & Hughes, 2000); and geology, soils, and variables with patterns of distribution and abundance of
common and rare freshwater macroinvertebrates. Biolog-
vegetation in Kansas (Hawkes et al., 1986). With ical Conservation 50: 77–98.
differences in these large-scale drivers minimized, the Faith, D. P., P. R. Minchin & L. Belbin, 1987. Compositional
greatest source of variation in the present study was dissimilarity as a robust measure of ecological distance.
due to local landscape features, allowing for the close Vegetatio 69: 57–68.
Farrand, W. R. & D. L. Bell, 1982. Quaternary Geology of
association of the biota with these gradients. Southern Michigan (map). Department of Geological
Sciences, University of Michigan, Ann Arbor, Michigan.
Acknowledgments We wish to thank the many individuals Fausch, K. D., C. E. Torgersen, C. V. Baxter & H. W. Li, 2002.
involved with data collection and processing including Matt Landscapes to riverscapes: bridging the gap between
Diana, Janelle Francis, Diana Karwan, Eric Sokol, Jo Wilhelm, research and conservation of stream fishes. BioScience
and Mahya Wood. This research was supported by a grant from 52: 483–498.
the National Science Foundation’s Water and Watersheds Fitzpatrick, F. A., B. C. Scudder, B. N. Lenz & D. J. Sullivan,
Program. 2001. Effects of multi-scale environmental characteristics
on agricultural stream biota in eastern Wisconsin. Journal
of the American Water Resources Association 37: 1489–
1507.
References Goldstein, R. M. & M. R. Meador, 2004. Comparisons of fish
species traits from small streams to large rivers. Trans-
Allan, J. D., 2004. Landscapes and riverscapes: the influence of actions of the American Fisheries Society 133: 971–983.
land use on stream ecosystems. Annual Review of Ecol- Hawkes, C. L., D. L. Miller & W. G. Layher, 1986. Fish
ogy, Evolution, and Systematics 35: 257–284. ecoregions of Kansas: stream fish assemblage patterns and
Allen, A. P., T. R. Whittier, D. P. Larsen, P. R. Kaufmann, R. J. associated environmental correlates. Environmental
O’Connor, R. M. Hughes, R. S. Stemberger, S. S. Dixit, R. Biology of Fishes 17: 267–279.
O. Brinkhurst, A. T. Herlihy & S. G. Paulsen, 1999. Hawkins, C. P., R. H. Norris, J. N. Hogue & J. W. Feminella,
Concordance of taxonomic composition patterns across 2000. Development and use of predictive models for
multiple lake assemblages: effects of scale, body size, and assessing the biological integrity of streams. Ecological
land use. Canadian Journal of Fisheries and Aquatic Sci- Applications 10: 1456–1477.
ences 56: 2029–2040. Hilsenhoff, W. L., 1982. Using a biotic index to evaluate water
Barbour, M. T., B. D. Gerritsen, B. D. Snyder & J. B. Stribling, quality of streams. Technical Bulletin Number 132.
1999. Rapid Bioassessment Protocols for use in Streams Wisconsin Department of Natural Resources, Madison,
and Wadeable Rives: Periphyton, Benthic Macroinverte- Wisconsin.
brates and Fish. Second Edition. EPA 841-8-99-002. US Infante, D. M., M. J. Wiley & P. W. Seelbach, 2006. Rela-
Environmental Protection Agency, Office of Water, tionships among channel shape, catchment characteristics,
Washington D.C. and fish. In Hughes, R. M., L. Wang & P. W. Seelbach
Belbin, L., 1992. PATN (Pattern Analysis Package) Technical (eds), Influences of Landscapes on Stream Habitats and
Reference. Commonwealth Scientific and Industrial Biological Assemblages. American Fisheries Society,
Research Organization, Division of Wildlife and Ecology, Bethesda, Maryland.
Canberra, Australia.

123
102 Hydrobiologia (2009) 623:87–103

Johnson, R. K. & W. Goedkoop, 2002. Littoral macroinverte- Surface Water Quality Division. Great Lakes Environ-
brate communities: spatial scale and ecological mental Assessment Section. Staff Report.
relationships. Freshwater Biology 47: 1840–1854. Michigan Resource Information System (MIRIS), 1978.
Johnson, L. B., C. Richards, G. E. Host & J. W. Arthur, 1997. Michigan Land Use (geographic information system
Landscape influences on water chemistry in Midwestern maps). Michigan Department of Natural Resources, Lan-
stream ecosystems. Freshwater Biology 37: 193–208. sing, Michigan.
Karr, J. R., 1991. Biological integrity: a long-neglected aspect Norris, R. H. & A. Georges, 1993. Analysis and interpretation of
of water resource management. Ecological Applications benthic macroinvertebrate surveys. In Rosenberg, D. M. &
1: 66–84. V. H. Resh (eds), Freshwater Biomonitoring and Benthic
Kilgour, B. W. & D. R. Barton, 1999. Associations between Macroinvertebrates. Chapman and Hall, New York.
stream fish and benthos across environmental gradients in Norris, R. H. & C. P. Hawkins, 2000. Monitoring river health.
southern Ontario, Canada. Freshwater Biology 41: Hydrobiologia 435: 5–17.
553–566. Olden, J. A. & N. L. Poff, 2004. Ecological processes driving
King, R. S., M. E. Baker, D. F. Whigham, D. E. Weller, T. E. biotic homogenization: testing a mechanistic model using
Jordan, P. F. Kazyak & M. K. Hurd, 2005. Spatial con- fish faunas. Ecology 85: 1867–1875.
siderations for linking watershed land cover to ecological Osborne, L. L. & M. J. Wiley, 1988. Empirical relationship
indicators in streams. Ecological Applications 15: between land-use cover and stream water quality in an
137–153. agricultural watershed. Journal of Environmental Man-
Lammert, M. & J. D. Allan, 1999. Assessing biotic integrity of agement 26: 9–27.
streams: effects of scale in measuring the influence of land Paavola, R., T. Muotka, R. Virtanen, J. Heino & P. Kreivi,
use/cover and habitat structure on fish and macroinverte- 2003. Are biological classifications of headwater streams
brates. Environmental Management 23: 257–270. concordant across multiple taxonomic groups? Freshwater
Lance, G. N. & W. T. Williams, 1967. A general theory of Biology 48: 1912–1923.
classificatory sorting strategies: I. Hierarchical Systems. Paavola, R., T. Muotka, R. Virtanen, J. Heino, D. Jackson & A.
The Computer Journal 9: 373–380. Maki-Petays, 2006. Spatial scale affects community con-
Legendre, P. & V. Legendre, 1984. Postglacial dispersal of cordance among fishes, benthic macroinvertebrates, and
freshwater fishes in Quebec Peninsula. Canadian Journal bryophytes in streams. Ecological Applications 16:
of Fisheries and Aquatic Sciences 41: 1781–1802. 368–379.
Lenat, D. R., 1984. Agriculture and stream water quality – a Paszkowski, C. A. & W. M. Tonn, 2000. Community con-
biological evaluation of erosion control practices. Envi- cordance between the fish and aquatic birds of lakes in
ronmental Management 8: 333–343. northern Alberta, Canada: the relative importance of
Linke, S., R. H. Norris, D. P. Faith & D. Stockwell, 2005. environmental and biotic factors. Freshwater Biology 43:
ANNA: a new prediction method for bioassessment pro- 421–437.
grams. Freshwater Biology 50: 147–158. Podani, J., 1997. On the sensitivity of ordination and classifi-
Lyons, J., 1992. Using the index of biotic integrity (IBI) to cation methods to variation in the input order of data.
measure environmental quality in warmwater streams of Journal of Vegetation Science 8: 153–156.
Wisconsin. US Department of Agriculture, Forest Service, Poff, N. L., 1997. Landscape filters and species traits: towards
North Central Forest Experiment Station, General Tech- mechanistic understanding and prediction in stream
nical Report NC-149. St. Paul, Minnesota. ecology. Journal of the North American Benthological
Lyons, J., R. R. Piette & K. W. Niermeyer, 2001. Develop- Society 16: 391–409.
ment, validation, and application of a fish-based index of Poff, N. L. & J. D. Allan, 1995. Functional organization of
biotic integrity for Wisconsin’s large warmwater rivers. stream fish assemblages in relation to hydrologic vari-
Transactions of the American Fisheries Society 130: ability. Ecology 76: 606–627.
1077–1094. Quinn, J. M., 2000. Effects of pastoral development. In Collier,
Magalhaes, M. F., D. C. Batalha & M. J. Collares-Pereira, K. J. & M. J. Winterbourn (eds), New Zealand Stream
2002. Gradients in stream fish assemblages across a Invertebrates: Ecology and Implications for Management.
Mediterranean landscape: contributions of environmental Caxton Press, Christchurch, New Zealand.
factors and spatial structure. Freshwater Biology 47: Quist, M. C., F. J. Rahel & W. A. Hubert, 2005. Hierarchical
1015–1031. faunal filters: an approach to assessing effects of habitat
McCune, B. & J. Mefford, 1997. PC-ORD. Multivariate and nonnative species on native fishes. Ecology of
Analysis of Ecological Data. Mjm Software Design, Freshwater Fish 14: 24–39.
Gleneden Beach, Oregon. Richards, C., L. B. Johnson & G. E. Host, 1996. Landscape-
Meador, M. R. & R. M. Goldstein, 2003. Assessing water scale influences on stream habitats and biota. Canadian
quality at large geographic scales: relations among land Journal of Fisheries and Aquatic Sciences 53: 295–311.
use, water physicochemistry, riparian condition, and fish Rios, S. L. & R. C. Bailey, 2006. Relationship between riparian
community structure. Environmental Management 31: vegetation and stream benthic communities at three spa-
504–517. tial scales. Hydrobiologia 553: 153–160.
Michigan Department of Environmental Quality, 1997. Qual- Roth, N. E., J. D. Allan & D. L. Erickson, 1996. Landscape
itative biological and habitat survey protocols for influences on stream biotic integrity assessed at multiple
wadeable streams and rivers. MI/DEQ/SWQ-96/068. spatial scales. Landscape Ecology 11: 141–156.

123
Hydrobiologia (2009) 623:87–103 103

Schlosser, I. J., 1995. Critical landscape attributes that influ- invertebrate traits for North America: U.S. Geological Sur-
ence fish population dynamics in headwater streams. vey Data Series 187. U.S. Department of the Interior, Reston,
Hydrobiologia 303: 71–81. Virginia (Available from: http://pubs.usgs.gov/ds/ds187/).
Scott, M. C. & G. S. Helfman, 2001. Native invasions, Walters, D. M., D. S. Leigh & A. B. Bearden, 2003. Urbani-
homogenization, and the mismeasure of integrity of fish zation, sedimentation, and the homogenization of fish
assemblages. Fisheries 26: 6–15. assemblages in the Etowah River Basin, USA. Hydrobi-
Sneldor, T. H. & B. J. F. Biggs, 2002. Multiscale river envi- ologia 494: 5–10.
ronment classification for water resources management. Wang, L., J. Lyons, P. Kanehl & R. Gatti, 1997. Influences of
Journal of the American Water Resources Association 38: watershed land use on habitat quality and biotic integrity
1225–1239. in Wisconsin streams. Fisheries 22: 6–12.
Sponseller, R. A., E. F. Benfield & H. M. Valett, 2001. Rela- Wang, L., J. Lyons & P. Kanehl, 2001. Impacts of urbanization
tionships between land use, spatial scale and stream on stream habitat and fish across multiple spatial scales.
macroinvertebrate communities. Freshwater Biology 46: Environmental Management 28: 255–266.
1409–1424. Wang, L., J. Lyons, P. Rasmussen, P. Seelbach, T. Simon, M.
Stewart, J. S., L. Wang, J. Lyons, J. A. Horwatich & R. Wiley, P. Kanehl, E. Baker, S. Niemela & P. M. Stewart,
Bannerman, 2001. Influences of watershed, riparian-cor- 2003. Watershed, reach and riparian influences on stream
ridor, and reach-scale characteristics on aquatic biota in fish assemblages in the Northern Lakes and Forest Eco-
agricultural watersheds. Journal of the American Water region, USA. Canadian Journal of Fisheries and Aquatic
Resources Association 37: 1475–1487. Science 60: 491–505.
Tonn, W. M., J. J. Magnuson, M. Rask & J. Toivonen, 1990. Whittier, T. R., R. M. Hughes & D. P. Larsen, 1988. Corre-
Intercontinental comparison of small-lake fish assem- spondence between ecoregions and spatial patterns in
blages: the balance between local and regional processes. stream ecosystems in Oregon. Canadian Journal of Fish-
The American Naturalist 136: 345–375. eries and Aquatic Sciences 45: 1264–1278.
Townsend, C. R., S. Doledec, R. H. Norris, K. Peacock & C. J. Wood, M. G. S., 2002. The influence of land cover and in-
Arbuckle, 2003. The influence of scale and geography on stream habitat on the biological condition of streams in
relationships between stream community composition and southeastern Michigan. M.S. Thesis, University of Mich-
landscape variables: description and prediction. Fresh- igan, Ann Arbor, Michigan.
water Biology 48: 768–785. Wright, J. F., 1995. Development and use of a system for
Van Sickle, J. & R. M. Hughes, 2000. Classification strengths predicting the macroinvertebrate fauna in flowing waters.
of ecoregions, catchments, and geographic clusters for Australian Journal of Ecology 20: 181–197.
aquatic vertebrates in Oregon. Journal of the North Yoder, C. O. & E. T. Rankin, 2004. Trends in biological
American Benthological Society 19: 370–384. integrity, biodiversity and aquatic habitat in the Eastern
Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell Corn Belt Plains Ecoregion: Implications for the protec-
& C. E. Cushing, 1980. The river continuum concept. tion and restoration of streams in the St. Joseph River
Canadian Journal of Fisheries and Aquatic Sciences 37: Watershed. Executive Summary for the Midwest Biodi-
130–137. versity Institute. Aquatic Life Use Fact Sheet 1-MBI-03.
Vieira, N. K. M., N. L. Poff, D. M. Carlisle, S. R. Moulton, M. L. Columbus, Ohio.
Koski & B. C. Kondratieff, 2006. A database of lotic

123