Next Article in Journal
Spatial Distribution and Migration Mechanisms of Toxic Elements in Farmland Soil at Nonferrous Metal Smelting Site
Next Article in Special Issue
Assessing Impacts of Human Stressors on Stream Fish Habitats across the Mississippi River Basin
Previous Article in Journal
Comparison of the Applicability of J-M Distance Feature Selection Methods for Coastal Wetland Classification
Previous Article in Special Issue
Regional Trends of Biodiversity Indices in the Temperate Mesic United States: Testing for Influences of Anthropogenic Land Use on Stream Fish while Controlling for Natural Landscape Variables
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 15
Issue 12
10.3390/w15122213
water-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Using Multiscale Environmental and Spatial Analyses to Understand Natural and Anthropogenic Influence on Fish Communities in Four Canadian Rivers

by
Beth L. Sparks-Jackson
1,*,
Peter C. Esselman
2,
Chris Wilson
3 and
Leon M. Carl
4
1
Akima Systems Engineering, Contractor to the U.S. Geological Survey, 1451 Green Road, Ann Arbor, MI 48105, USA
2
Great Lakes Science Center, U.S. Geological Survey, 1451 Green Road, Ann Arbor, MI 48105, USA
3
Ontario Ministry of Natural Resources and Forestry, Trent University, Symons Campus, 1600 West Bank Drive, Peterborough, ON K9J 7B8, Canada
4
U.S. Geological Survey, Midwest Region, 1451 Green Road, Ann Arbor, MI 48105, USA
*
Author to whom correspondence should be addressed.
Water 2023, 15(12), 2213; https://doi.org/10.3390/w15122213
Submission received: 30 December 2022 / Revised: 9 May 2023 / Accepted: 16 May 2023 / Published: 12 June 2023
(This article belongs to the Special Issue Disentangling Influences of Natural and Human Factors on Aquatic Ecosystems)
Browse Figures
Review Reports Versions Notes

Abstract

:
Science-based conservation of riverine fishes can be best targeted with specific information about spatial-ecological controls on the community, including anthropogenic stressors. Because anthropogenic stressors can originate at multiple spatial scales, we investigated the influence of natural and anthropogenic variables summarized within the reach, valley, and catchment on fish community composition along four river mainstems in Ontario, Canada. We used Redundancy Analyses (RDA) to explore models with multi- and single-scale variables on fish community composition. We used partial RDAs to differentiate the relative effects of variable types in multiscale models and to determine if spatial variables explained additional variation in fish community composition. Catchment variables accounted for the majority of explained variation in fish community composition in three of the four rivers, but instream habitat variables accounted for considerable variability in fish community composition in the two rivers that are highly fragmented by dams or naturally occurring rapids. Natural and human-derived fragmentation in rivers may reduce the influence of catchment controls, disrupt longitudinal gradients, and increase the influence of local instream habitat. Environmental variables that explained fish distribution had longitudinal or patchy spatial pattern within rivers, but spatial variables representing impediments to fish dispersal and proximity to receiving waterbodies failed to explain additional variation in fish community composition.

1. Introduction

Science-based conservation of riverine fishes can best be targeted with information about spatial-ecological controls on the community, including anthropogenic stressors. Natural and anthropogenic influences can originate from multiple spatial scales, so characterizing their influence requires a multiscale approach. Hierarchy theory [1,2,3] suggests that conditions at the catchment scale will constrain habitats and biota at finer scales such as the individual reach. Catchment level controls such as climate, geology, soils, and topography have indeed been shown to affect valley and reach scale geomorphic, hydrologic, thermal, and chemical conditions that have direct effects on fishes [4,5,6,7,8]. Catchment network structure has also been shown to create strong longitudinal gradients in physical habitat and biology, and associations between fish assemblages and stream size are pervasive [4,9,10,11,12]. The River Continuum Concept (RCC) hypothesized changes in species functional traits and ecosystem processes in response to predictable upstream–downstream changes in physical processes and geomorphology [13]. Meso-scale valley habitat features (e.g., sinuosity, tributary effects, riparian land use, etc.) are strongly controlled by geomorphic processes, proximal landscape features, and network configuration [14,15,16]. Local physical, chemical, and thermal river habitats may influence fish distributions through direct proximity to organisms, and associations between fish communities and instream measures of temperature, substrate, vegetation, hydraulics, and channel shape are common [4,17].
Despite accumulating evidence that catchment controls correspond to fish distribution in many riverine systems [6,18,19,20], some studies have shown local factors to be more important [7,21,22]. This importance of local factors suggests that there are conditions under which catchment controls can be disrupted or absent. Identifying conditions under which local controls become dominant (or vice versa) could help explain contradictory results among previous multiscale studies and provide insights into mechanisms for fish community assembly. Although differences in study design (e.g., variables used, spatial extent, sampling regime, etc.) may contribute to varying conclusions, complex relationships between environmental predictors and biota are likely also at play. In addition, ecologists also recognize that spatial pattern in fish distribution can be unrelated to physical habitat conditions, instead resulting from faunal exclusion by competition, predation, metapopulation dynamics, limits on dispersal, or demographic stochasticity. These factors can be especially influential in large rivers [23] and rivers with dams [24,25].
Anthropogenic influences can have strong effects on fish community composition and structure [26,27], and can presumably also affect hierarchical controls on local communities. For instance, anthropogenic features such as dams have been shown to disrupt relationships between fish and a river’s catchment and result in idiosyncratic spatial pattern in physical habitat and fish distribution. In a modification to the RCC, the Serial Discontinuity Concept [28,29] postulates that dams along river mainstems disrupt the river continuum through changes in flow, temperature, substrate, and lateral and vertical connectivity. Dams can also affect fish distributions through eliminating or reducing access to suitable habitats within a river network [30,31]. Reduced connectivity in rivers can force fishes to exist in locally isolated metapopulations [32,33] where niches are limited by local habitat conditions decoupled from catchment controls. Land cover conversion to anthropogenic use is another pervasive influence on the landscape. Unlike changes to rivers caused by dams, effects of land cover change tend to propagate downstream through river networks [27], potentially reinforcing the importance of catchment controls on local conditions. Other activities such as mining are also known to propagate downstream through the river network to affect biota [34]. Thus, anthropogenic factors may either disrupt or reinforce landscape control on local assemblages, and understanding specifically how these factors affect fish assemblages within a river basin may help to explain contradictory results of multiscale studies.
In this study, we used multiscale and spatial variables in redundancy analyses (RDA) to help disentangle and understand natural and anthropogenic influences on fish community composition in four river basins in Ontario, Canada. Fish community composition was determined through extensive electrofishing. Two of the study rivers had minimal human development in their catchments; one river was moderately fragmented by dams; and one was highly fragmented by dams. We evaluated several research questions within three general themes: (1) multiscale influence—do fish communities respond in similar ways to multiscale influences across basins? If not, does understanding the relative influence of the different scales help to disentangle and understand natural and anthropogenic effects? (2) Variables associated with fish community composition—which environmental variables best explain patterns in fish community composition in the four river mainstems in this study? Can insights be gained from comparing results among different scales and river basins? (3) Spatial effects—do variables that account for spatial effects caused by impediments to dispersal due to dams, longitudinal position, and other factors help explain additional variation in fish community composition along river mainstems? If not, are spatially structured abiotic conditions the primary way that spatial effects act to structure fish communities in these rivers?
We hypothesized that differences in human development and fragmentation would affect the relative importance of catchment, valley, and instream habitat variables when explaining fish community composition. In rivers with minimal human development in the catchment and no dams along the river mainstem, we hypothesized that catchment environmental variables would explain the most variation in fish community composition, and instream habitat variables would explain the least. Conversely, we expected local instream habitats to have stronger influences on fish communities in rivers fragmented by numerous dams. We hypothesized that river size would be an important driver of community composition in all river basins, but that multiscale analyses would help to identify the other natural and anthropogenic drivers acting in each river basin. Although we expected some environmental variables to be spatially structured, we also expected spatial factors such as impediments to dispersal by dams and/or proximity to receiving lake or river to help explain additional variation in the fish community composition.

2. Materials and Methods

2.1. Study Area

The study included four rivers in Ontario, Canada (Figure 1) with varied natural character, anthropogenic development, and fish communities. All four basins are part of the St. Lawrence River watershed and were colonized from Mississippian and Atlantic Coastal refugia as the Laurentian ice sheet retreated ~10,000 years ago [35]. These rivers are strongly influenced by stream thermal regimes that are controlled by climate and glacial deposits [36].
The Grand River (GD) has a large (6800 km2), predominately agricultural catchment draining southern Ontario and discharging into eastern Lake Erie. The river mainstem is moderately fragmented by five low-head dams in the study area. These dams are small but do alter the habitat immediately upstream of the dam [37] and can affect fish movement. Three of the dams are not tall enough to be barriers to jumping migratory fishes; two dams have fish passages of limited efficacy [38]; and one tall dam without fish passage is a barrier to all fishes. The Grand River has a diverse fish community supporting sport fisheries for rainbow trout (Oncorhynchus mykiss), brown trout (Salmo trutta), walleye (Sander vitreus), and bass (Micropterus sp.).
The Ganaraska (GA) and Trent (T) rivers are neighboring basins in southern Ontario that drain into Lake Ontario and the Bay of Quinte, respectively. Despite their proximity, they contrast in many ways. The Ganaraska’s catchment is small (280 km2) and includes diverse surficial geologies and land uses, coldwater and coolwater thermal regimes, and a fish community dominated by salmonids. The Ganaraska has a 4 m-high dam with a fish passage structure located downstream of the study area near the rivermouth. The Trent River (T) is the largest river in this study (12,600 km2) and is highly fragmented; the study area alone includes 24 navigational dam and lock structures and a large in-line lake. The dam and lock system is considered to be semipermeable as it appears to allow for some genetic exchange among mobile river fishes such as redhorses (Moxostoma spp.) [39]. The Trent River fauna is dominated by warmwater fish species and supports highly valued sport fisheries for walleye, muskellunge (Esox masquinongy), and bass [40].
The Petawawa River (P) in northeastern Ontario is moderately sized (4177 km2) and has the least anthropogenic development of the basins in our study. The Petawawa has numerous high-gradient whitewater rapids that may cause natural discontinuities in longitudinal connectivity. The Petawawa River drains into the Ottawa River, which itself is a tributary to the St. Lawrence River. The Petawawa River is undammed and supports a predominantly warmwater fauna including walleye and several species of Centrarchidae. Of the four rivers, the Trent has the highest dam density (0.15 dams/km channel length), followed by the Grand (0.06 dams/km), the Ganaraska (0.03 dams/km), and the Petawawa (0 dams/km).

2.2. Study Sites

A total of 112 sites were sampled for fishes and local habitat along the mainstems of the rivers (Figure 1), including 27 sites along the Grand River, 15 along the Ganaraska River, 54 sites along the Trent River, and 16 sites along the Petawawa River. These same sites were attributed with a common set of variables characterizing natural and anthropogenic factors in catchment and valley spatial units using GIS (see Section 2.4). Sites were numbered consecutively in each basin beginning with the upstream-most site. Lacustrine sections of the rivers (e.g., reservoirs above dams, in-line lakes) were not sampled, as the goal was to characterize riverine fish-habitat relationships. Two sites on the Petawawa River were removed prior to analyses because one had incomplete fish data (site 8), and the other was not on the mainstem (site 13). Sampling on the Grand, Trent, and Petawawa Rivers focused on non-wadeable, lower river mainstems, whereas all sites in the Ganaraska River were wadeable and incorporated a wider range of stream sizes.

2.3. Instream Habit and Fish Sampling

The sampling objectives were to measure comparable characteristics of instream habitat and to document the fish species present at each site. Wadable and non-wadable sites were defined and sampled differently due to different operating constraints caused by water depth. All sites in the Ganaraska River were wadeable, and site lengths varied by stream width. Sites were defined as either a length of river 20 times the mean stream width or a maximum length of 300 m. Habitat and fishes were sampled across the entire area of a site. All sites in the Grand, Trent, and Petawawa rivers were non-wadeable and defined as 600 m lengths of river with channel sections in the middle and shoreline sections near the river’s edge. The 600 m site was further subdivided (see [40] and Appendix A for details), and multiple locations within a site were sampled to characterize the site. Sampling in the Grand, Ganaraska, and Trent Rivers occurred over multiple, consecutive years, while sampling in the Petawawa River was limited to one year (2001). Sites in the Grand River were sampled one or two times in 2000 and 2001; sites in the Ganaraska River were sampled two or three times between 1997 and 1999; and sites in the Trent River were sampled two or three times between 1999 and 2001. If a site was sampled in multiple years, fish occurrence data were pooled across years to create a presence–absence matrix for each river.
Instream habitat (ISH) variables were calculated from field data collected using standard aquatic habitat sampling techniques. Slight differences in sampling methods among rivers led to differences in ISH variables collected, but measures of channel shape and hydraulic energy were available for all rivers. Measures of water temperature, discharge, conductivity, substrate, and proportions of aquatic vegetation were limited to select rivers (Table 1). Water depths were measured with measuring poles or depth sounders; velocity was measured using a floating object method; hydraulic head was measured with a ruler; substrate and vegetation were directly observed or recorded with an underwater video camera; and temperature and conductivity were measured with hand-held devices. Details of field data collection have been previously published [41] with additional details herein (e.g., sampling dates, expanded descriptions of sampling techniques, etc.). Shoreline and channel habitats were summarized separately at non-wadeable sites, as well as across the entire site by using weighted averages based on the proportion of shoreline and channel habitat at each site. Coefficients of variation (CV) were calculated for most ISH variables to capture habitat complexity. If instream habitat was measured in multiple years, averages and CVs were calculated from pooled measurements across years.
All fish sampling occurred in the summer or early fall. In the Ganaraska River, each site was electrofished along its entire length and width using a towboat or shore unit with one or two dip netters, depending on channel width. In the Grand and the Trent rivers, sites were sampled by boat electrofishing downstream along the 12 channel transects and upstream along the 8 shoreline transects. All boat electrofishing was performed in a single pass using a 5 KW pulsed DC electrofisher with boom anode and one dip netter. Power was maintained at 1000 Watts for all sites to minimize error associated with differential catch efficiency. Fish were typically identified to species in the field and returned to the river unless the fish could not be definitively identified in the field, or a voucher specimen was required.
Because of restrictions on electrofishing within the Canadian Forces Base Petawawa, sites in the Petawawa River were sampled with electrofishing (sites 1–4, 14–16), gillnet and minnow trapping (sites 6–10, 12), or a combination of both methods (sites 5 and 11). Electrofished sites in the Petawawa River were sampled with the same methods as used in the Grand and Trent Rivers. At sites where electrofishing was prohibited, gillnets were set for 1.5 h in 12 channel transects and baited minnow traps were set for 12 h at 8 shoreline transects. A study comparing efficiency of the gears used in the Petawawa River [51] suggested that gear had minimal effect on species presence-absence data and that most fish species present at a site were detected regardless of sample gear used. When gear type was added as a factor into statistical models, gear type did not explain additional variation in fish community composition, suggesting that these differences had minimal influence on the study. Therefore, we felt justified in combining fish presence-absence data collected by all methods in the dataset for the Petawawa River.

2.4. Catchment and Valley Variables

The same set of catchment (CAT) and valley (VAL) variables were calculated for all sites in all river basins (Table 1). The term “catchment” (CAT) is used here to describe the entire upstream land area that drains to a particular study reach (the word “basin” is used to define the catchment that drains to the outlet with the Great Lakes or Ottawa River). We delineated catchment polygons for each site using GIS (including the Hydrology toolset in the Spatial Analyst toolbox in ArcGIS [52]) and the Ontario Integrated Hydrography Data [42]. Catchment polygons were used to aggregate CAT variables that served as proxies for stream size (catchment area and link) and upstream land cover, road density, geology, soils, and climate for each site. Land cover maps compiled from satellite images between 1999 and 2002 [45] were matched as closely as possible to fish sampling timeframes, and climate data were summarized for the period from 1960 to 1990 [49]. We also included the total number of dams upstream or downstream within the river basin to reflect the connectivity of each site to upstream habitat or Great Lake/Ottawa River habitat, respectively [31].
The term “valley” (VAL) describes a broad set of variables that quantify habitat features that are scaled between CAT and instream habitats. Most VAL variables (e.g., riparian land use, channel slope, and channel sinuosity) were summarized within a 1 km radius circular buffer centered on each site. We also counted the total number of tributaries and major tributaries entering the buffer because tributaries have been shown to affect habitat and fish in river mainstems [53,54,55,56]. To measure the size of accessible habitat patches, we calculated the total length of tributary and mainstem habitat (TotHabPtch) and the total length of only mainstem habitat (MSHabPtch) accessible between barriers for each site. The sizes of habitat patches were included as VAL variables because they represent meso-scale habitat availability (for example, see the ranges of values of MSHabPtch in Table 2).

2.5. Spatial Variable Development

In addition to CAT, VAL, and ISH variables, we calculated three spatial variables to assess how spatial characteristics may affect the distribution of fishes in each river. We used these analyses to explore whether the effects of “space” were primarily through a spatially structured physical environment or unrelated to environmental factors. The four rivers varied in outlet connectivity (Lakes Erie or Ontario, Ottawa River), network and in-line lake configuration, and fragmentation by dams. We quantified (1) distance to rivermouth (defined as the “swim” distance up the river network from the rivermouth) as a measure of the potential influence that proximity to the larger species pool present in the Great Lakes or Ottawa River may have on site-level assemblage; (2) Principal Coordinates of Neighbour Matrices (PCNM) as a measure of spatial structure at various scales (see below); and (3) in the Grand and Trent rivers, dam distance, a measure of possible impediments to dispersal from human-made barriers.
PCNM is a spatial eigenfunction analysis used to identify multiscale spatial structure that is not predefined with a specific hypothesis in mind [57,58,59]. We applied PCNM on a matrix of watercourse distances between each pair of sites [60] to derive a new set of independent eigenvectors that represent spatial patterns present at different spatial scales. A second PCNM variable was calculated on a matrix of dam counts between all sites (rather than distances) to represent possible multiscale effects of dams in the Grand and Trent rivers.

2.6. Data Analyses

The overall goal of the analyses was to identify and understand key drivers of fish community composition in the four study rivers. Although disentangling natural and anthropogenic influences in these rivers is a part of that goal, we did not feel that an a priori designation of each environmental variable as either anthropogenic or natural was appropriate. While a designation for some variables is obvious (e.g., urban land use in the catchment), for other variables, understanding whether an influence is natural or anthropogenic is only possible in the context of covarying environmental influences. For example, in a system with dams, a high proportion of hard substrate may result from scour in the tailwaters of the dam itself, while in a river without dams, a high proportion of hard substrate may be driven by natural geomorphic controls.
We used RDA to quantify the proportion of variation in fish community composition in each river basin explained by environmental and spatial variables. RDA is a direct extension of multiple regression that uses matrices of explanatory and response variables rather than a single response variable. RDA provides both descriptive statistics (e.g., identifying environmental factors that explain variability in fish community composition at sites) and inferential statistics (e.g., whether RDA solutions explain more variability in community composition than expected by chance). Although the sampling regime in this study can describe detailed spatial patterns of fish along river mainstems, it likely results in spatial dependence among sites, violating the assumption of independence among observations required for statistical inference using RDA. Therefore, we assessed statistical significance using a cyclical shift permutation procedure wherein the spatial structure of the linear sampling design was maintained by “joining” the upstream-most and downstream-most sites to form a loop [61,62]. Permutations were achieved by rotating the loops, and the permutations were used to estimate a random null distribution to which the empirical test statistic is compared. Since the maximum number of possible permutations was small (i.e., Grand = 53, Ganaraska = 29, Trent = 107, and Petawawa = 27), we assessed statistical significance using the full set of permutations. We also set the significance cutoff at p = 0.1 because restricted permutations on small sample sizes can make it difficult to find statistical significance and greatly increases the risk of type II statistical errors.
RDA solutions were developed within each river basin using all environmental variables combined (ALL = CAT + VAL + ISH) and only variables from each scale (CAT or VAL or ISH). Basin-specific matrices of fish species presence or absence (with all species included) were used as the response variables. Rare species were included in our response matrices because their removal in exploratory analyses resulted in similar statistical significance and minimal changes in variance explained. A stepwise variable selection procedure was used to develop RDA solutions with a goal to maximize variance explained (R2) with a small number of ecologically relevant, statistically unrelated variables. Variables were selected by first including the variable that explained the most variability in fish community composition, then excluding highly correlated (|r| > 0.70) variables from the remaining variable pool. Variables were added only if they were statistically significant (p < 0.10) or if they explained >5% additional variance, even if they were not statistically significant under cyclic permutations. Finally, variables in each RDA solution were double checked for collinearly using variance inflation factors (VIFs). If models with similar amounts of variance explained were identified, the model with lower VIFs was selected. To explore the explanatory power of each included variable, we calculated their marginal and conditional effects. Marginal effects represent variance explained by each variable as the sole predictor while conditional effects represent variance explained by each environmental variable with the variables already selected treated as covariables.
We identified the “best” solution as the RDA solution that explained the most variation in fish community composition. We compared the explained variance for solutions across river basins and variable scales using an adjusted explained variance (AdjR2) that accounts for differences in the number of included variables and sample sizes [63]. We developed an RDA triplot for the best solution for each river basin. These triplots were used to understand species associations to multiscale environmental constraints by showing the variables included in each solution relative to sampling sites and individual fish species. Fish species were classified by tolerance to anthropogenic stress as designated in the Ontario Freshwater Fishes Life History Database [64] and displayed in these triplots by tolerance class. This database defines three tolerance classes: (1) Intolerant: Species that is sensitive to environmental or anthropogenic stresses; (2) Intermediate: Species that is neither particularly sensitive nor insensitive to environmental or anthropogenic stresses; or (3) Tolerant: Species that is fairly insensitive or adaptive to environmental or anthropogenic stresses. We also used partial RDA analyses to assess the relative explanatory power of CAT, VAL, and ISH variables on the best RDA solution developed for each basin. Variation partitioning [65,66] can be used to evaluate the unique and shared contributions of different groups of explanatory variables to explaining variability in the fish presence-absence matrix.
We explored whether adding spatial variables to the best solutions explained additional variation in fish community composition. These partial RDA analyses evaluated the relative contribution of environmental and spatial factors on fish community composition by partitioning explained variance between pure environment, pure space, and shared environment/space components. Up to five positive eigenvectors that were most strongly associated with the fish presence matrix were included in partial analyses as the PCNM or DAMS spatial variables. The values of individual eigenvectors included in analyses were also plotted against site number and visually examined to interpret spatial patterns represented in the individual eigenvectors. All analyses were performed using statistical software (R, [67]; RDAs were performed with the package vegan: Community Ecology Package [68]).

3. Results

3.1. Data Summary

A total of 43,930 fish of 69 species were caught across the 110 study sites over the sample years. The Grand River had the greatest species richness with 47 species, followed by the Trent (38 spp.), the Ganaraska (26 spp.), and the Petawawa (25 spp.). Each basin also had somewhat distinctive fish communities that were dominated by different species (Table 2; Appendix B). The Ganaraska had the highest percentage intolerant species (26%), followed by the Grand and Petawawa (13%), and then the Trent (11%). The pattern in tolerant species was almost inverse to this with the highest percentage in the Trent (27%), followed by the Grand (26%), the Ganaraska (18%), and the Petawawa (17%). The four study rivers also differed in measures of channel size and shape, degree of connectivity, and catchment characteristics (land use, surficial geology, and soils) (Table 2). Notably, the Grand River had the highest CAT percentages of agriculture and urban land cover, but the Trent River had slightly higher percent urban cover in VAL. Sites on the Petawawa River had the least urban and agricultural land use in the CAT and VAL.
Although the total number of variables we collected for CAT, VAL, and ISH scales varied (Table 1), there are several reasons why these differences do not limit our ability to interpret the variance explained portion of the RDA results. Strong correlations (|r| > 0.7) between variable pairs variably limited the suite of variables available for inclusion in RDA analyses. Strong correlations were common between catchment variable pairs (e.g., pairs of land use, geology, and soils variables) and effectively reduced the number of CAT variables by 37%–63% depending on the river basin. Strong correlations were rare between pairs of VAL or ISH variables and extremely rare between variables of different scales of summary. Our use of stepwise regression and inclusion of only two to four explanatory variables further negated the effects of differences in the original number of variables for each scale.

3.2. Environment and Fish Communities

Statistically significant RDA solutions were identified for all study basins both within and across scales of summary except the ISH variable group in the Petawawa River (Table 3). All RDA solutions included four or fewer minimally related explanatory variables (average VIF across all models = 1.8 with a maximum VIF of 3.9 within models). Although not statistically significant (p = 0.11), the Petawawa model with two ISH variables explained nearly 25% of the variability in the fish community composition. Because the Petawawa River had the smallest number of sites and a limited number of permutations for hypothesis testing (i.e., low statistical power), we included and interpreted this model.
RDA solutions explained 17 to 71% of the variation in fish community composition within river basins (Table 3). Environmental variables explained more variation in the fish community composition in the Ganaraska River than in the other three rivers. The solutions developed from the full multiscale variable set explained more variation in fish community composition than the single scale analyses in every case and always included variables from at least two variable scales (Table 3, Figure 2). In the Trent and Petawawa rivers, the solution that explained the most variability in fish community composition included CAT and ISH variables, in the Ganaraska it included CAT and VAL variables, and in the Grand it included variables from all three scales.
Partial analyses on the best RDA models for the Ganaraska, Trent, and Petawawa Rivers indicated that joint effects of variables from different scales were typically negligible or small compared with the independent effects of variables at each scale. CAT variables explained most of the variation in the fish community composition in the Ganaraska River (scales separated by a forward slash indicate joint effects; CAT 45.4%, VAL 12.2%, CAT/VAL 0%;) and Grand River (CAT 12.4%, VAL 6.4%, ISH 3.6%, CAT/VAL 2%, CAT/ISH 1%, VAL/ISH 0%, CAT/VAL/ISH 6.6%). CAT and ISH variables explained similar amounts of variation in the Trent and Petawawa Rivers (Trent: CAT 12.4%, ISH 13.4%, CAT/ISH 2.8%; Petawawa: CAT 15.7%, ISH 14.6%, CAT/ISH 0%).
Limiting the pool of variables to only one scale of summary reduced the amount of variation in the fish community composition explained (Figure 2). In the Grand and Ganaraska rivers, the explained variability was highest for the CAT-only solutions and lowest for ISH-only solutions. In the Trent River, ISH variables alone explained more of the variability in the fish community composition, and VAL explained the least. In the Petawawa River, CAT and VAL variables explained nearly equal amounts of variability, and ISH variables explained the least.
RDA triplots for the best solution for each river basin allowed for a more in-depth evaluation of the relationships between environmental variables, sites, and fish species (Figure 3). In the Ganaraska River, intolerant fish species were associated with higher proportions of forest in the riparian area, higher SoilAWHC, and lower link. In the Ganaraska River, fish community composition was also affected by longitudinal position, with fewer species upstream (sites 1–3), moderate diversity mid-river (sites 4–7), and high diversity downstream (sites 8–15; Figure 3a). In the Grand River, the presence of typically lentic and tolerant channel catfish (Ictalurus punctatus), freshwater drum (Aplodinotus grunniens), alewife (Alosa pseudoharengus), and gizzard shad (Dorosoma cepedianum) distinguished downstream sites and showed marked effects of slower channel velocity and high area-weighted average of soil water holding capacity (SoilAWHC) nearer to Lake Erie. There were three distinct groups of sites along an upstream to downstream longitudinal gradient in the Grand River, each with characteristic fish species and similar catchment characteristics (Figure 3b). In contrast, fish in the Trent River did not appear to be strongly influenced by longitudinal gradients; instead, sites were differentiated by the presence of lotic habitats (hard substrate and lack of vegetation along shoreline) containing mottled sculpin (Cottus bairdii) and darters or lentic habitats (soft substrate with vegetation along shoreline) with brown bullhead (Ameiurus nebulosus), common shiner (Luxilus cornutus), yellow perch (Perca flavescens), and largemouth bass (Figure 3c). A fish community of exclusively intermediate and tolerant species was associated with a higher proportion of urban land use in the catchment and soft substrate with vegetation. In the Petawawa River there was not a strong longitudinal gradient, and only one group of sites (sites 1–4) were distinguishable from the others (due to slightly higher fish diversity including walleye and mimic shiners (Notropis volucellus). There was no distinct pattern in tolerance or strong associations between tolerance and environmental characteristics (Figure 3d).

3.3. Environment, Space, and Fish Communities

Adding spatial variables to the best basin-specific solution improved the amount of explained variance in fish community composition by an average of only 1.9% (range of 0 to 10.9%; Figure 4). Our results indicated that independent effects of environmental variables and joint effects of space and environment (i.e., a spatially structured environment) were the primary influences on fish community composition. Joint effects of space and environment corresponded with variables in the models that represent spatial position (e.g., Link, Catchment Area) and fragmentation by dams (MSHabPtch). For example, the DAMS eigenvector in the Grand River grouped sites between each set of dams, essentially recreating the spatial structure already inherent in the MSHabPtch variable in the model.
Variance in the fish community composition explained purely by a spatial variable was statistically significant in two solutions and provided additional insight into spatial pattern in fish communities in two rivers. In the Grand and the Trent rivers, including the PCNM spatial eigenvectors explained an additional 10.9% (F = 1.8, p < 0.02) and 2.5% (F = 1.5, p < 0.01) of the variance in fish community composition, respectively. In the Grand, the PCNM eigenvectors identified broad spatial patterns in three groups of sites along the mainstem. These included a grouping of the six downstream-most sites, a group of the next seven sites, and then the remaining sites near Brantford and upstream. One PCNM eigenvector for the Trent River differentiated sites upstream and downstream of a very large in-line lake (Rice Lake), while another captured local effects of Rice Lake by differentiating sites immediately upstream and immediately downstream of Rice Lake from sites further away from the lake.

4. Discussion

Disentangling natural and anthropogenic influences on riverine biological communities is made difficult by the complexity of multiscale influences on these systems. Here, we related a large fish community dataset to an extensive environmental and spatial dataset in four river basins in southern Ontario, Canada. The included river basins cover a wide range of anthropogenic and natural conditions. Below, we discuss three major conclusions that can be drawn from the study.

4.1. Anthropogenic Influences Are Pervasive and Affect Fish Community Composition

In all study rivers, associations between catchment and riparian agricultural and urban land cover or their inverse, percent forest cover, accounted for significant variation in fish community composition. Human-made dams also had clear influences on fish communities in the Grand and Trent rivers (discussed in detail in Section 4.2). In the Ganaraska River, pollution intolerant cool/coldwater fish species including Coho Salmon, Slimy Sculpin, Brook Stickleback, and Blacknose Shiner showed strong associations with higher forest cover in the riparian area. Even low levels of development in riparian areas [69] or the catchment overall [70] can result in major shifts in fish communities. Local shading of smaller channels in the Ganaraska River may be essential to protect the cooler thermal regime and high-quality habitat required by these intolerant species.
Higher percentages of CAT urban land use in the Trent River corresponded to communities with no intolerant species. In river catchments with widespread agricultural and urban development, degradation of physical and chemical habitat can eliminate fish species requiring narrow physio-chemical habitats, resulting in a homogenized fish community [27,71,72]. The Trent and Grand rivers have a long history of disturbance and a high proportion of intermediate and tolerant fish species. Despite the largely homogenized fish communities in these basins, we still observed significant evidence of anthropogenic influence in both basins. However, despite a wealth of data about these four rivers, it was still not possible to fully disentangle anthropogenic influence driven through land use versus dams as the two rivers with dams also had the most developed catchments and numerous developed areas along the study areas.
In addition to expected explanatory effects of environmental variables, we hypothesized that including spatial variables accounting for anthropogenic effects of impediments to dispersal by dams and natural effects of distance to source populations would help explain variation in fish community composition. Although the DAMS and distance to rivermouth variables did not explain additional variation in fish community composition in the four basins, spatial variables that accounted for spatial pattern at a variety of scales (i.e., PCNM variables) did. In the Trent River, the primary PCNM spatial variable distinguished sites near both the inlet and outlet of Rice Lake, a large, naturally occurring in-line lake. It is logical that Rice Lake may act as a barrier to upstream movement of lotic fish and a colonizing source of lentic/lotic generalists in downstream areas. In the Grand River, the primary PCNM spatial variable distinguished patches of sites along the river network, suggesting that the environmental variables included in the model failed to fully account for patchiness in fish community composition.
Although purely spatial components typically did not explain additional variation in fish community composition, about half of the explained variance in community composition was accounted for by the joint environment/space component. The large joint environment/space component implies that many of the natural and anthropogenic variables that explain fish community composition in the four rivers had spatial patterns that were longitudinally structured, related to dams, or included spatial pattern accounted for by the PCNM variables. Although a large joint environment/space component limits our ability to fully disentangle the effects of dams and access to species pools in downstream receiving waters, such spatial patterns in environmental variables were expected. Rivers by nature are aggregating, longitudinally structured systems that tend to also have longitudinally structured and patchy habitats [13,28]. The RDA triplots developed in this study clearly illustrated such longitudinal pattern and patchiness in the four study rivers. The high degree of spatial pattern in many environmental variables makes disentangling environmentally and spatially derived processes in river systems more difficult [73]. As metapopulation models are increasingly being used to explore how dams may constrain fish dispersal and persistence [32,74], a better understanding of these spatial processes becomes even more essential.

4.2. Dams and Other Discontinuities Influence Fish Community Composition

This study suggests that both natural features and human-made barriers may have the potential to decouple local fish assemblages from catchment influences by creating discontinuities in habitat. This finding was indicated by the greater relative importance of instream habitat conditions in the Trent and Petawawa rivers. The Trent River is highly fragmented with frequent but semi-permeable dams and interspersed lotic and lentic habitat. We observed community composition in the Trent River to be structured by the local availability of soft substrate and vegetation, in addition to catchment size and urban land cover. Our previous work in the Trent River revealed a similar finding—higher fish species richness in larger fragments results from addition of five minnow species with strong affiliations to aquatic vegetation and lentic conditions [40]. The slope of the species-area relationship in that study indicated a high degree of isolation between populations in different fragments, despite the semi-permeability of the dam and lock system. It may be that dispersal of less mobile species in the Trent River is limited by high-energy habitats found in dam tailraces.
The Petawawa River, by contrast, is interspersed by naturally occurring high-energy rapids. Stronger correlations between instream habitat and community composition (vs. CAT and VAL) in the Petawawa River suggest that a dynamic similar to that observed in the Trent River may be at play. The importance of instream habitat in the Petawawa River indicated that the presence of soft substrate and deeper water, both proxies for large, slow-moving pools, were most strongly linked to compositional patterns. Therefore, it appears plausible that the high gradient cascades separated by large pools present in the Petawawa River may act as natural analogs to reservoir-dam systems. High-gradient stream reaches with turbulent rapids have been shown to limit fish distribution in large rivers [75,76]. The effects of rapids on fish movement in the Petawawa River have not specifically been studied, and this study did not show an obvious restriction of lentic fish to areas downstream of the rapids. This limits our ability to conclude whether the rapids primarily affect fish in this system by physically blocking upstream movement of some fish species or as discontinuities in physical habitat.
In contrast, our results suggest that catchment controls combined with longitudinal barriers caused by impassible dams resulted in three distinct faunal groupings in the Grand River mainstem. This study joins others in suggesting that even small dams and dams with fish passage can limit dispersal sufficiently to structure whole fish communities. For example, in this study, alewife, gizzard shad, and freshwater drum were limited to the section of the Grand River downstream of the second most downstream barrier, Caledonia Dam. Similarly, Reid et al. [77] found that river redhorse (Moxostoma carinatum) were limited to areas downstream of the Caledonia Dam. Bunt et al. [38] and MacDougall et al. [78] found that walleye were largely ineffective at using the fish ladder at the most downstream dam on the system, Dunnville Dam, severely limiting access to upstream spawning habitat. Many river fishes require access to spatially disparate habitats for reproduction, growth, and survival. Therefore, riverine fish communities are particularly prone to faunal insularization from habitat fragmentation [79] which can lead to deleterious genetic effects and local extinction.

4.3. Multiscale Control on Fish Communities

For all rivers, the RDA solutions that explained the most variation in fish community composition included variables from multiple scales of summary, suggesting multiscale control of fish distribution is common in the study rivers. Although the primary aim of many multiscale studies has been to identify the scale with the most influence on riverine biota, we suggest that allowing predictors of fish community composition to be drawn from multiple scales can help fishery managers to identify the specific processes that create spatial pattern in fish within a river. This, in turn, may allow managers to develop targeted, river-specific conservation activities. For instance, in the Trent River, certain small minnows were only associated with the presence of soft substrate and vegetation cover, suggesting that a viable strategy for promoting species diversity may be to encourage activities that enhance substrate diversity and aquatic vegetation cover.
Wang et al. [7,80] postulated that fish communities responded most strongly to catchment variables in basins with extensive anthropogenic land cover while local factors had more influence in largely undisturbed catchments. In this study, instream habitat variables explained much of the variation in both a river basin with the most urban development along the river mainstem and in a river basin with the least urban development. Although our study is limited to four river basins, which in turn limits our ability to generalize results to other rivers, this study suggests that factors other than anthropogenic land cover may affect the relative influence of different scales of environmental variables on riverine fishes.
This study may provide insight into causes for the lack of consensus about the relative importance of local versus landscape scale controls on aquatic communities. We hypothesize that that differences in study design (e.g., variables used and spatial scale assigned, spatial extent, gradient length, sampling regime, etc.) could lead to varying conclusions. By using consistently developed environmental and fish metrics in multiple river basins, differences in study design can be minimized, and a better understanding of how hierarchical relationships may vary among basins may be gained. We know of only one other study that has evaluated the differences in relationships between multiscale environmental variables and riverine fish in multiple drainage basins. Troia and Gido [12] found that the relative importance of local habitat versus catchment characteristics changed with the position of a basin along an east–west aridity gradient. While having a much higher numbers of river basins would be necessary to formulate firm conclusions about the mechanisms underpinning local versus landscape control, our study reinforces the potential value of developing consistent landscape datasets for basins spanning gradients of natural and anthropogenic influence.
In addition to differences between the ability of catchment, valley, and instream habitat variables to explain variation in fish community composition within river basins, there were marked differences in the percent of variation in fish community composition explained in the four rivers. Environmental variables explained less than half of the variation in the fish community composition except in the Ganaraska River. Such results are typical. In a multiscale comparison for multiple river basins, Troia and Gido [12] found that the adjusted proportion of inertia constrained by explanatory variables in canonical correspondence analyses ranged from 0.13 to 0.36 among 13 river sub-basins. Likewise, in a meta-analysis of 158 ecological datasets, Lawler and Torgersen [81] found that the variation in community structure explained by environmental and spatial variables averaged only 50% and was sometimes much lower.
The range of ecological conditions sampled can influence the amount of heterogeneity in environmental variables and biological communities [12,82,83,84]. For instance, sample sites in the Ganaraska River encompassed a comparatively larger range of stream sizes (Strahler order 3 to 5) than the other rivers, likely increasing environmental and faunal heterogeneity and our ability to account for variation in fish community composition. In contrast, sites in the Grand, Trent, and Petawawa rivers were all in downstream, high-order portions of the basin where stream order was nearly invariant across all study sites (i.e., order 7 in the Grand and Petawawa rivers and 7–8 in the Trent River). In the Petawawa River, the proportion of variance in the fish community composition explained by environmental variables was smaller than for the other rivers. The Petawawa River was the most species-poor system, possibly because of slightly more recent glacial retreat and more difficult post-glacial recolonization routes than the other rivers [35]. Thus, our inability to statistically explain very small differences in fish community composition (e.g., 1 or 2 species) between sites in the Petawawa River is not surprising.

5. Conclusions

We used an extensive multiscale landscape and fish dataset in four river basins in Ontario, Canada with strongly contrasting anthropogenic conditions to explore correlates to fish community composition arising from catchment, valley, and instream scales of influence. Like many “whole-ecosystem” studies [85,86], we traded replication for the realism of studying four uniquely composed aquatic landscapes. The contrast in variance explained in the Ganaraska River assemblage (63%) versus the other three rivers (<31%) illustrates how wider ecological gradients might be well quantified by catchment factors. Our results suggest potential mechanisms by which local habitats and biota may become decoupled from hierarchical landscape controls. The Trent and Petawawa results suggest that serial discontinuities within the river network, whether driven by anthropogenic or natural geologic fragmentation, may increase the importance of local habitat controls on communities. Our findings reinforce a growing body of literature suggesting strong effects of spatially structured environmental variables on river fish assemblages. Carefully designed observational studies will be necessary to reinforce our proposed mechanism for decoupling local assemblages from their catchments. Such studies are warranted as their results may indicate conditions under which managers are justified in focusing their energies on local rather than catchment scale data collection and management interventions.

Author Contributions

B.L.S.-J.: conceptualization, writing—original draft preparation, writing—review and editing, data curation, formal analysis, and visualization. P.C.E.: supervision, conceptualization, writing—review and editing, funding acquisition. C.W.: data curation, methodology, investigation. L.M.C.: funding acquisition, project administration, methodology, investigation. All authors have read and agreed to the published version of the manuscript.

Funding

Most of the funding for the large river sampling was provided by the Ontario Ministry of Natural Resources. Additional funds for sampling the Trent River were provided by the Trent Severn Waterway to the Watershed Science Centre at Trent University, Peterborough, ON.

Data Availability Statement

Data are available in a publicly accessible repository. The data presented in this study are openly available in ScienceBase at [https://www.sciencebase.gov/catalog/item/5afc91e1e4b0da30c1bc20ca].

Acknowledgments

We are grateful to Scott Reid, Sarah Crabbe, Jason Barnucz, Steve Chong, Phil Anderson, and J.D. Whall Environmental Consultants for assisting with data collection, preparation, and preliminary analysis. Arthur Cooper calculated the CAT and VAL dam fragmentation metrics. The comments of J. David Allan, Ralph Tingley, and four anonymous reviewers improved the manuscript. Although no referenceable animal care and use protocols were in place during electrofishing, all fish were treated humanely. Electrofishing temporarily stuns fish, and most fish were returned to the river after processing. Fish that could not be identified in the field were humanely euthanized to be identified in the laboratory. The analyses and manuscript development were completed under contract to the U.S. Geological Survey. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Conflicts of Interest

The authors declare no conflict of interest.

Appendix A. Sampling Design

A two-tier sampling design was used in the Ganaraska River (sites within valley segments; not illustrated), and a three-tier nested sampling design was used in the Grand, Trent, and Petawawa rivers (Figure A1a). Each river was first delineated into valley segments with each valley segment defined as a section of river that is relatively uniform in character (Figure A1b). Valley segments may begin or end at a dam, a confluence with a major tributary, a widening of the river into a more lacustrine environment, or a change in surficial geology.
After valley segments were delineated in the Grand, Trent, and Petawawa rivers, each valley segment was divided into 600 m-long sites, with a minimum of 100 m between sites. Within each valley segment, 5 sites were randomly chosen for sampling, and if the segment could not accommodate 5 sites, all sites within the valley segment were sampled (Figure A1b).
Once sites were established, each site was divided into five 100 m blocks separated by four 25 m no-sampling zones (Figure A1c). Each 100 m block was further divided in half, creating ten 50 m sampling zones. Within each site, twelve 50 m channel transects and eight 50 m shoreline transects were sampled. Channel transects were sampled moving downstream, and shoreline transects were sampled moving upstream. Channel transects were randomly chosen throughout the channel section of the study site, and the randomization process continued until 12 unique channel transects were selected for sampling. The site was divided into a grid within each of the 50 m sampling zones by dividing the width of the river at each sampling zone into 10 m wide sections, or if the river was small (i.e., mean width < 50 m), divided into left, middle, and right sections. Each river width section was assigned a number and, using a random number generator, combinations of sampling zones and river width sections were selected. Randomly selected channel transects less than 10 m from shore were excluded to prevent overlapping with nearshore habitats, and transects less than 10 m apart were excluded to minimize the potential of pseudo-replication. Shoreline transects were chosen with a stratified random design. Every 100 m block had 4 possible shoreline options (e.g., two 50 m sections on each bank). Using a random number generator, one section in each of the five blocks was chosen. Then, 3 more were chosen randomly from the remaining 15 shoreline areas. C# and S# labelled arrows in Figure A1c illustrate how this sampling design may be applied at a site.
Water 15 02213 g0a1 550
Figure A1. Visualization of the three-tier nested sampling design used in the Grand, Trent, and Petawawa rivers (a). Each river was delineated into valley segments and 600 m-long sites were identified within each valley segment (b). Each site was divided into blocks of shoreline and channel sampling units (c). Eight shoreline units and twelve channel units were selected for sampling using a stratified random selection process (c).
Figure A1. Visualization of the three-tier nested sampling design used in the Grand, Trent, and Petawawa rivers (a). Each river was delineated into valley segments and 600 m-long sites were identified within each valley segment (b). Each site was divided into blocks of shoreline and channel sampling units (c). Eight shoreline units and twelve channel units were selected for sampling using a stratified random selection process (c).
Water 15 02213 g0a1

Appendix B

A total of 69 species of fish were identified during this study (Table A1). Young of year fish and a small number of individual fish that could not be identified to species were not included in analyses. Fish species were classified by tolerance to anthropogenic stress as designated in the Ontario Freshwater Fishes Life History Database [64]. This database defines three tolerance classes: (1) Intolerant: Species that is sensitive to environmental or anthropogenic stresses; (2) Intermediate: Species that is neither particularly sensitive nor insensitive to environmental or anthropogenic stresses; or (3) Tolerant: Species that is fairly insensitive or adaptive to environmental or anthropogenic stresses.
Table
Table A1. Fish species caught in the study rivers with four-letter species codes and tolerance classes used in Table 2 and Figure 3. An X indicates the species was caught in that river basin (GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa).
Table A1. Fish species caught in the study rivers with four-letter species codes and tolerance classes used in Table 2 and Figure 3. An X indicates the species was caught in that river basin (GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa).
FamilyScientific NameCommon NameCodeToleranceGDGATP
AtherinopsidaeLabidesthes sicculusBrook SilversideSILVIntermediateX X
CatostomidaeCarpiodes cyprinusQuillbackQUILIntermediateXXXX
Catostomus commersoniiWhite SuckerWSUKTolerantXXXX
Hypentelium nigricansNorthern Hog SuckerHOGSIntermediateX
Moxostoma anisurumSilver RedhorseSRDHIntermediateX XX
Moxostoma carinatumRiver RedhorseRVRHIntolerant X
Moxostoma erythrurumGolden RedhorseGREDIntermediateX
Moxostoma macrolepidotumShorthead RedhorseSHRHIntermediateX XX
Moxostoma valenciennesiGreater RedhorseGRDHIntolerantX X
CentrarchidaeAmbloplites rupestrisRock BassROCKIntermediateX XX
Lepomis gibbosusPumpkinseedPUNKIntermediateXXXX
Lepomis macrochirusBluegillBLUEIntermediateX X
Micropterus dolomieuSmallmouth BassSBASIntermediateX XX
Micropterus salmoidesLargemouth BassLBASTolerantXXX
Pomoxis annularisWhite CrappieWCRPTolerantX
Pomoxis nigromaculatusBlack CrappieCRAPTolerantX X
ClupeidaeAlosa pseudoharengusAlewifeALEWIntermediateX
Dorosoma cepedianumGizzard ShadGIZZTolerantX
CottidaeCottus bairdiiMottled SculpinMOTTIntermediate X
Cottus cognatusSlimy SculpinSLIMIntolerant XX
CyprinidaeCarassius auratusGoldfishGLDFTolerantX
Cyprinella spilopteraSpotfin ShinerSFSHIntermediateX
Cyprinus carpioCommon CarpCARPTolerantX X
Hybognathus hankinsoniBrassy MinnowBRSSIntermediate X
Luxilus chrysocephalusStriped ShinerSTSHIntermediateX
Luxilus cornutusCommon ShinerCOMMIntermediateXXX
Margariscus margaritaPearl DacePERLIntermediate X
Nocomis biguttatusHornyhead ChubHRNYIntermediateXX
Notemigonus crysoleucasGolden ShinerGOLDIntermediate XXX
Notropis atherinoidesEmerald ShinerEMRLIntermediateXXXX
Notropis heterodonBlackchin ShinerCHINIntolerant XX
Notropis heterolepisBlacknose ShinerBLAKIntolerant X
Notropis hudsoniusSpottail ShinerSPOTIntermediateXXXX
Notropis photogenisSilver ShinerSILSIntolerantX
Notropis rubellusRosyface ShinerROSEIntermediateX XX
Notropis volucellusMimic ShinerMIMCIntermediateX XX
Phoxinus eosNorthern Redbelly DaceNRBDIntermediate X
Pimephales notatusBluntnose MinnowBNOSIntermediateXXXX
Pimephales promelasFathead MinnowFATHTolerant XXX
Rhinichthys atratulusBlacknose DaceBNDCIntermediate X
Rhinichthys cataractaeLongnose DaceLONGIntermediateXX X
Semotilus atromaculatusCreek ChubCHUBIntermediateXXX
Semotilus corporalisFallfishFALLIntermediate XX
EsocidaeEsox luciusNorthern PikePIKEIntermediateX X
Esox masquinongyMuskellungeMUSKIntermediate XX
FundulidaeFundulus diaphanusBanded KillifishKILLTolerant X
GasterosteidaeCulaea inconstansBrook SticklebackSTIKIntermediate X
HiodontidaeHiodon tergisusMooneyeMOONIntolerantX
IctaluridaeAmeiurus nebulosusBrown BullheadBBULIntermediateXXXX
Ictalurus punctatusChannel CatfishCCATTolerantX X
Noturus flavusStonecatSTONTolerantX
LepisosteidaeLepisosteus osseusLongnose GarLGARTolerantX XX
PercidaeEtheostoma blennioidesGreenside DarterGSIDIntermediateX
Etheostoma caeruleumRainbow DarterRDRTIntolerantX
Etheostoma exileIowa DarterIOWAIntermediate XX
Etheostoma flabellareFantail DarterFANTIntolerantX
Etheostoma nigrumJohnny DarterJOHNTolerantXXXX
Perca flavescensYellow PerchPRCHIntermediate XX
Percina caprodesLogperchLOGPIntolerantX XX
Percina maculataBlackside DarterBDRTIntermediateX
Sander vitreusWalleyeWALLIntermediateX XX
PercopsidaePercopsis omiscomaycusTrout-PerchTRPRIntermediateX
SalmonidaeOncorhynchus kisutchCoho SalmonCOHOIntolerant X
Oncorhynchus mykissRainbow TroutRAINIntolerantXX
Oncorhynchus tshawytschaChinook SalmonKINGIntolerant X
Salmo truttaBrown TroutBTRTIntolerant X
Salvelinus fontinalisBrook TroutBROKIntolerant X
SciaenidaeAplodinotus grunniensFreshwater DrumDRUMTolerantX X
UmbridaeUmbra limiCentral MudminnowMUDMTolerant XX

References

  1. Allen, T.F.H.; Starr, T.B. Hierarchy: Perspectives for Ecological Complexity, 2nd ed.; University Chicago Press: Chicago, IL, USA, 1982; ISBN 978-0-226-01431-9. [Google Scholar]
  2. O’Neill, R.V.; DeAngelis, D.L.; Waide, J.B.; Allen, T.F.H. A Hierarchical Concept of Ecosystems; Princeton University Press: Princeton, NJ, USA, 1986; ISBN 0-691-08436-X. [Google Scholar]
  3. Parsons, M.; Thoms, M.C.; Norris, R.H. Using Hierarchy to Select Scales of Measurement in Multiscale Studies of Stream Macroinvertebrate Assemblages. J. N. Am. Benthol. Soc. 2004, 23, 157–170. [Google Scholar] [CrossRef]
  4. Hynes, H.B.N. The Ecology of Running Waters; University of Toronto Press: Toronto, ON, Canada, 1970; ISBN 0-8020-1689-8. [Google Scholar]
  5. Frissell, C.A.; Liss, W.J.; Warren, C.E.; Hurley, M.D. A Hierarchical Framework for Stream Habitat Classification—Viewing Streams in a Watershed Context. Environ. Manag. 1986, 10, 199–214. [Google Scholar] [CrossRef]
  6. Johnson, L.B.; Richards, C.; Host, G.E.; Arthur, J.W. Landscape Influences on Water Chemistry in Midwestern Stream Ecosystems. Freshwat. Biol. 1997, 37, 193–208. [Google Scholar] [CrossRef]
  7. Wang, L.; Lyons, J.; Rasmussen, P.; Seelbach, P.; Simon, T.; Wiley, M.; Kanehl, P.; Baker, E.; Niemela, S.; Stewart, P.M. Watershed, Reach, and Riparian Influences on Stream Fish Assemblages in the Northern Lakes and Forest Ecoregion, USA. Can. Fish. Aquat. Sci. 2003, 60, 491–505. [Google Scholar] [CrossRef]
  8. Gordon, N.D.; McMahon, T.A.; Finlayson, B.L.; Gippel, C.J.; Nathan, R.J. Stream Hydrology. An Introduction for Ecologists, 2nd ed.; John Wiley & Sons: Chichester, UK, 2004; ISBN 978-0-470-84358-1. [Google Scholar]
  9. Huet, M. Profiles and Biology of Western European Streams as Related to Fish Management. Trans. Am. Fish. Soc. 1959, 88, 155–163. [Google Scholar] [CrossRef]
  10. Schlosser, I.J. Environmental Variation, Life History Attributes, and Community Structure in Stream Fishes: Implications for Environmental Management and Assessment. Environ. Manag. 1990, 14, 621–628. [Google Scholar] [CrossRef]
  11. Fischer, J.R.; Paukert, C.P. Habitat Relationships with Fish Assemblages in Minimally Disturbed Great Plains Regions. Ecol. Freshw. Fish 2008, 17, 597–609. [Google Scholar] [CrossRef]
  12. Troia, M.J.; Gido, K.B. Predicting Community-Environment Relationships of Stream Fishes across Multiple Drainage Basins: Insights into Model Generality and the Effect of Spatial Extent. J. Environ. Manag. 2013, 128, 313–323. [Google Scholar] [CrossRef]
  13. Vannote, R.L.; Minshall, G.W.; Cummins, K.W.; Sedell, J.R.; Cushing, C.E. The River Continuum Concept. Can. J. Fish. Aquat. Sci. 1980, 37, 130–137. [Google Scholar] [CrossRef]
  14. Hynes, H.B.N. The Stream and Its Valley. Verb. Internat. Verein Limnol. 1975, 19, 1–15. [Google Scholar] [CrossRef]
  15. Seelbach, P.W.; Wiley, M.J.; Baker, M.E.; Wehrly, K.E. Initial Classification of River Valley Segments across Michigan’s Lower Peninsula. In American Fisheries Society Symposium; American Fisheries Society: Bethesda, MD, USA, 2006; Volume 48, pp. 25–48. [Google Scholar]
  16. Jones, N.E.; Schmidt, B.J. Tributary Effects in Rivers: Interactions of Spatial Scale, Network Structure, and Landscape Characteristics. Can. J. Fish. Aquat. Sci. 2016, 74, 503–510. [Google Scholar] [CrossRef]
  17. Statzner, B.; Gore, J.A.; Resh, V.H. Hydraulic Stream Ecology: Observed Patterns and Potential Applications. J. N. Am. Benthol. Soc. 1988, 7, 307–360. [Google Scholar] [CrossRef]
  18. Zorn, T.G.; Seelbach, P.W.; Wiley, M.J. Distributions of Stream Fishes and Their Relationship to Stream Size and Hydrology in Michigan’s Lower Peninsula. Trans. Am. Fish. Soc. 2002, 131, 70–85. [Google Scholar] [CrossRef]
  19. Esselman, P.C.; Allan, J.D. Relative Influences of Catchment- and Reach-Scale Abiotic Factors on Freshwater Fish Communities in Rivers of Northeastern Mesoamerica. Ecol. Freshw. Fish 2010, 19, 439–454. [Google Scholar] [CrossRef]
  20. Pease, A.A.; Taylor, J.M.; Winemiller, K.O.; King, R.S. Multiscale Environmental Influences on Fish Assemblage Structure in Central Texas Streams. T. Am. Fish. Soc. 2011, 140, 1409–1427. [Google Scholar] [CrossRef]
  21. Johnson, R.K.; Furse, M.T.; Hering, D.; Sandin, L. Ecological Relationships between Stream Communities and Spatial Scale: Implications for Designing Catchment-Level Monitoring Programmes. Freshwat. Biol. 2007, 52, 939–958. [Google Scholar] [CrossRef]
  22. Terra, B.d.F.; Hughes, R.M.; Araújo, F.G. Fish Assemblages in Atlantic Forest Streams: The Relative Influence of Local and Catchment Environments on Taxonomic and Functional Species. Ecol. Freshw. Fish 2016, 25, 527–544. [Google Scholar] [CrossRef]
  23. Grenouillet, G.; Pont, D.; Harisse, C. Within-Basin Fish Assemblage Structure: The Relative Influence of Habitat versus Stream Spatial Position on Local Species Richness. Can. J. Fish. Aquat. Sci. 2004, 61, 93–102. [Google Scholar] [CrossRef] [Green Version]
  24. Falke, J.A.; Gido, K.B. Spatial Effects of Reservoirs on Fish Assemblages in Great Plains Streams in Kansas, USA. River Res. Appl. 2006, 22, 55–68. [Google Scholar] [CrossRef]
  25. Zhao, K.; Song, K.; Pan, Y.; Wang, L.; Da, L.; Wang, Q. Metacommunity Structure of Zooplankton in River Networks: Roles of Environmental and Spatial Factors. Ecol. Indic. 2017, 73, 96–104. [Google Scholar] [CrossRef]
  26. Karr, J.R. Biological Integrity—A Long-Neglected Aspect of Water- Resource Management. Ecol. Appl. 1991, 1, 66–84. [Google Scholar] [CrossRef]
  27. Allan, J.D. Landscapes and Riverscapes: The Influence of Land Use on Stream Ecosystems. Annu. Rev. Ecol. Evol. Syst. 2004, 35, 257–284. [Google Scholar] [CrossRef] [Green Version]
  28. Ward, J.V.; Stanford, J.A. The Serial Discontinuity Concept of Lotic Ecosystems. In Dynamics of Lotic Ecosystems; Fontaine, T.D., Bartell, S.M., Eds.; Ann Arbor Science: Ann Arbor, MI, USA, 1983; pp. 29–42. [Google Scholar]
  29. Stanford, J.A.; Ward, J.V. Revisiting the Serial Discontinuity Concept. Regul. Rivers 2001, 17, 303–310. [Google Scholar] [CrossRef]
  30. Rolls, R.J.; Stewart-Koster, B.; Ellison, T.; Faggotter, S.; Roberts, D.T. Multiple Factors Determine the Effect of Anthropogenic Barriers to Connectivity on Riverine Fish. Biodivers. Conserv. 2014, 23, 2201–2220. [Google Scholar] [CrossRef]
  31. Cooper, A.R.; Infante, D.M.; Wehrly, K.E.; Wang, L.; Brenden, T.O. Identifying Indicators and Quantifying Large-Scale Effects of Dams on Fishes. Ecol. Indic. 2016, 61, 646–657. [Google Scholar] [CrossRef]
  32. Fagan, W.F. Connectivity, Fragmentation, and Extinction Risk in Dendritic Metapopulations. Ecology 2002, 83, 3243–3249. [Google Scholar] [CrossRef]
  33. Rodeles, A.A.; Galicia, D.; Miranda, R. A Simple Method to Assess the Fragmentation of Freshwater Fish Meta-Populations: Implications for River Management and Conservation. Ecol. Indic. 2021, 125, 107557. [Google Scholar] [CrossRef]
  34. Daniel, W.M.; Infante, D.M.; Hughes, R.M.; Tsang, Y.; Esselman, P.C.; Wieferich, D.J.; Herreman, K.; Cooper, A.R.; Wang, L.; Taylor, W.W. Characterizing Coal and Mineral Mines as a Regional Source of Stress to Stream Fish Assemblages. Ecol. Indic. 2015, 50, 50–61. [Google Scholar] [CrossRef]
  35. Mandrak, N.E.; Crossman, E.J. Postglacial Dispersal of Freshwater Fishes into Ontario. Can. J. Zool. 1992, 70, 2247–2259. [Google Scholar] [CrossRef]
  36. Chu, C.; Jones, N.E.; Mandrak, N.E.; Piggott, A.R.; Minns, C.K. The Influence of Air Temperature, Groundwater Discharge, and Climate Change on the Thermal Diversity of Stream Fishes in Southern Ontario Watersheds. Can. J. Fish. Aquat. Sci. 2008, 65, 297–308. [Google Scholar] [CrossRef]
  37. Raab, D.; Mandrak, N.E.; Ricciardi, A. Low-Head Dams Facilitate Round Goby Neogobius Melanostomus Invasion. Biol. Invasions 2018, 20, 757–776. [Google Scholar] [CrossRef]
  38. Bunt, C.M.; Cooke, S.J.; Scott McKinley, R. Assessment of the Dunnville Fishway for Passage of Walleyes from Lake Erie to the Grand River, Ontario. J. Great Lakes Res. 2000, 26, 482–488. [Google Scholar] [CrossRef]
  39. Reid, S.M.; Wilson, C.C.; Mandrak, N.E.; Carl, L.M. Population Structure and Genetic Diversity of Black Redhorse (Moxostoma Duquesnei) in a Highly Fragmented Watershed. Conserv. Genet. 2008, 9, 531–546. [Google Scholar] [CrossRef]
  40. Carl, L.M.; Esselman, P.C.; Sparks-Jackson, B.L.; Wilson, C.C. The Species–Area Relationship for a Highly Fragmented Temperate River System. Ecosphere 2021, 12, e03411. [Google Scholar] [CrossRef]
  41. Carl, L.M.; Sparks-Jackson, B.L.; Munn, S.T.; Esselman, P.C.; Wilson, C.C. Habitat and Fish Assemblages along Four River Mainstems in Ontario, Canada, 1997 to 2001, with Supporting Spatial Data. U.S. Geological Survey Data Release, 2021 [Data Set]. Available online: https://www.sciencebase.gov/catalog/item/5afc91e1e4b0da30c1bc20ca (accessed on 13 February 2023).
  42. Ontario Ministry of Natural Resources. Ontario Ministry of Natural Resources Ontario Integrated Hydrology Data (OIHD), Version 2.0 [Data Set]. 2015. Available online: https://geohub.lio.gov.on.ca/maps/mnrf::ontario-integrated-hydrology-oih-data/about (accessed on 13 February 2023).
  43. Ontario Ministry of Natural Resources. Ontario Dam Inventory, Version 1 [Data Set]. 2014. Available online: https://geohub.lio.gov.on.ca/datasets/mnrf::ontario-dam-inventory/explore (accessed on 13 February 2023).
  44. Statistics Canada. Census Road Network File, 2001 [Data Set]. 2001. Available online: https://www12.statcan.gc.ca/census-recensement/2011/geo/RNF-FRR/index-2011-eng.cfm?year=01 (accessed on 13 February 2023).
  45. Wang, L.; Riseng, C.M.; Mason, L.A.; Wehrly, K.E.; Rutherford, E.S.; McKenna, J.E.; Castiglione, C.; Johnson, L.B.; Infante, D.M.; Sowa, S.; et al. A Spatial Classification and Database for Management, Research, and Policy Making: The Great Lakes Aquatic Habitat Framework. J. Great Lakes Res. 2015, 41, 584–596. [Google Scholar] [CrossRef] [Green Version]
  46. Ontario Geological Survey. Quaternary Geology, Seamless Coverage of the Province of Ontario: Ontario Geological Survey, Data Set 14 [Data Set]. 1997. Available online: https://data.ontario.ca/dataset/quaternary-geology-of-ontario (accessed on 13 February 2023).
  47. Ontario Geological Survey. 1:250 000 Scale Bedrock Geology of Ontario Ontario Geological Survey. Miscellaneous Release–Data 126—Revision 1 [Data Set]. 2011. Available online: https://data.ontario.ca/dataset/1250-000-scale-bedrock-geology-of-ontario/resource/8ef6fca2-9fa4-4958-8c6c-a190f639d6a6 (accessed on 13 February 2023).
  48. Soil Landscapes of Canada Working Group. Soil Landscapes of Canada Version 3.2 [Data Set]. 2011. Available online: https://sis.agr.gc.ca/cansis/nsdb/slc/v3.2/index.html (accessed on 13 February 2023).
  49. Hijmans, R.J.; Cameron, S.E.; Parra, J.L.; Jones, P.G.; Jarvis, A. Very High Resolution Interpolated Climate Surfaces for Global Land Areas. Int. J. Climatol. 2005, 25, 1965–1978. [Google Scholar] [CrossRef]
  50. Ontario Ministry of Natural Resources and Forestry. Provincial Digital Elevation Model-South [Data set]. 2016. Available online: https://geohub.lio.gov.on.ca/maps/mnrf::provincial-digital-elevation-model-pdem/about (accessed on 13 February 2023).
  51. Barnucz, J. A Comparative Study of Sampling Efficiency between Boat Electrofishing and Non-Lethal Gillnet/Minnow Trap Sampling Used in Non-Wadeable River Sampling Methodology, Petawawa River, Ontario; Peterborough, ON, Canada, 2002; Unpublished internal Report. [Google Scholar]
  52. ESRI ArcGIS Desktop: Release 10.4.1; Environmental Systems Research Institute: Redlands, CA, USA, 2015.
  53. Osborne, L.L.; Wiley, M.J. Influence of Tributary Spatial Position on the Structure of Warmwater Fish Communities. Can. J. Fish. Aquat. Sci. 1992, 49, 671–681. [Google Scholar] [CrossRef]
  54. Benda, L.; Andras, K.; Miller, D.; Bigelow, P. Confluence Effects in Rivers: Interactions of Basin Scale, Network Geometry, and Disturbance Regimes. Water Resour. Res. 2004, 40, W05402. [Google Scholar] [CrossRef]
  55. Kiffney, P.M.; Greene, C.M.; Hall, J.E.; Davies, J.R. Tributary Streams Create Spatial Discontinuities in Habitat, Biological Productivity, and Diversity in Mainstem Rivers. Can. J. Fish. Aquat. Sci. 2006, 63, 2518–2530. [Google Scholar] [CrossRef]
  56. Hitt, N.P.; Angermeier, P.L. Evidence for Fish Dispersal from Spatial Analysis of Stream Network Topology. J. N. Am. Benthol. Soc. 2008, 27, 304–320. [Google Scholar] [CrossRef]
  57. Borcard, D.; Legendre, P. All-Scale Spatial Analysis of Ecological Data by Means of Principal Coordinates of Neighbour Matrices. Ecol. Model. 2002, 153, 51–68. [Google Scholar] [CrossRef]
  58. Borcard, D.; Legendre, P.; Avois-Jacquet, C.; Tuomisto, H. Dissecting the Spatial Structure of Ecological Data at Multiple Scales. Ecology 2004, 85, 1826–1832. [Google Scholar] [CrossRef] [Green Version]
  59. Dray, S.; Legendre, P.; Peres-Neto, P.R. Spatial Modelling: A Comprehensive Framework for Principal Coordinate Analysis of Neighbour Matrices (PCNM). Ecol. Model. 2006, 196, 483–493. [Google Scholar] [CrossRef]
  60. Landeiro, V.L.; Magnuson, W.E.; Melo, A.S.; Espírito-Santo, H.M.V.; Bini, L.M. Spatial Eigenfunction Analyses in Stream Networks: Do Watercourse and Overland Distances Produce Different Results? Freshwat. Biol. 2011, 56, 1184–1192. [Google Scholar] [CrossRef]
  61. Besag, J.; Clifford, P. Generalized Monte Carlo Significance Tests. Biometrika 1989, 76, 633–642. [Google Scholar] [CrossRef]
  62. Fortin, M.-J.; Drapeau, P.; Jacquez, G.M. Quantification of the Spatial Co-Occurrences of Ecological Boundaries. Oikos 1996, 77, 51–60. [Google Scholar] [CrossRef]
  63. Peres-Neto, P.R.; Legendre, P.; Dray, S.; Borcard, D. Variation Partitioning of Species Data Matrices: Estimation and Comparison of Fractions. Ecology 2006, 87, 2614–2625. [Google Scholar] [CrossRef] [PubMed]
  64. Eakins, R.J. Ontario Freshwater Fishes Life History Database. Version 5.15. Online Database. 2023. Available online: https://www.ontariofishes.ca (accessed on 13 February 2023).
  65. Borcard, D.; Legendre, P.; Drapeau, P. Partialling out the Spatial Component of Ecological Variation. Ecology 1992, 73, 1045–1055. [Google Scholar] [CrossRef] [Green Version]
  66. Borcard, D.; Legendre, P. Environmental Control and Spatial Structure in Ecological Communities: An Example Using Oribatid Mites (Acari, Oribatei). Environ. Ecol. Stat. 1994, 1, 37–61. [Google Scholar] [CrossRef]
  67. R Core Team. R: A Language and Environment for Statistical Computing, Version 3.5.1; R Foundation for Statistical Computing: Vienna, Austria, 2018. [Google Scholar]
  68. Oksanen, J.; Blanchet, F.G.; Friendly, M.; Kindt, R.; Legendre, P.; McGlinn, D.; Minchin, P.R.; O’Hara, R.B.; Simpson, G.L.; Solymos, P.; et al. Vegan: Community Ecology Package Version 2.4-3. 2017. Available online: https://CRAN.R-project.org/package=vegan (accessed on 13 February 2023).
  69. Effert-Fanta, E.L.; Fischer, R.U.; Wahl, D.H. Effects of Riparian Forest Buffers and Agricultural Land Use on Macroinvertebrate and Fish Community Structure. Hydrobiologia 2019, 841, 45–64. [Google Scholar] [CrossRef]
  70. Wang, L.; Lyons, J.; Kanehl, P. Impacts of Urban Land Cover on Trout Streams in Wisconsin and Minnesota. Trans. Am. Fish. Soc. 2003, 132, 825–839. [Google Scholar] [CrossRef]
  71. Allan, J.D.; Erickson, D.L.; Fay, J. The Influence of Catchment Land Use on Stream Integrity across Multiple Spatial Scales. Freshwat. Biol. 1997, 37, 149–161. [Google Scholar] [CrossRef] [Green Version]
  72. Richards, C.; Johnson, L.B.; Host, G.E. Landscape-Scale Influences on Stream Habitats and Biota. Can. J. Fish. Aquat. Sci. 1996, 53, 295–311. [Google Scholar] [CrossRef]
  73. Cottenie, K. Integrating Environmental and Spatial Processes in Ecological Community Dynamics. Ecol. Lett. 2005, 8, 1175–1182. [Google Scholar] [CrossRef] [PubMed]
  74. Jager, H.I.; Chandler, J.A.; Lepla, K.B.; Van Winkle, W. A Theoretical Study of River Fragmentation by Dams and Its Effects on White Sturgeon Populations. Environ. Biol. Fish. 2001, 60, 347–361. [Google Scholar] [CrossRef]
  75. Torrente-Vilara, G.; Zuanon, J.; Leprieur, F.; Oberdorff, T.; Tedesco, P.A. Effects of Natural Rapids and Waterfalls on Fish Assemblage Structure in the Madeira River (Amazon Basin). Ecol. Freshw. Fish 2011, 20, 588–597. [Google Scholar] [CrossRef]
  76. Gibson-Reinemer, D.K.; Rahel, F.J.; Albeke, S.E.; Fitzpatrick, R.M. Natural and Anthropogenic Barriers to Climate Tracking in River Fishes along a Mountain–Plains Transition Zone. Divers. Distrib. 2017, 23, 761–770. [Google Scholar] [CrossRef]
  77. Reid, S.M.; Mandrak, N.E.; Carl, L.M.; Wilson, C.C. Influence of Dams and Habitat Condition on the Distribution of Redhorse (Moxostoma) Species in the Grand River Watershed, Ontario. Environ. Biol. Fish. 2008, 81, 111–125. [Google Scholar] [CrossRef]
  78. MacDougall, T.M.; Wilson, C.C.; Richardson, L.M.; Lavender, M.; Ryan, P.A. Walleye in the Grand River, Ontario: An Overview of Rehabilitation Efforts, Their Effectiveness, and Implications for Eastern Lake Erie Fisheries. J. Great Lakes Res. 2007, 33, 103–117. [Google Scholar] [CrossRef]
  79. Fausch, K.D.; Torgersen, C.E.; Baxter, C.V.; Li, H.W. Landscapes to Riverscapes: Bridging the Gap between Research and Conservation of Stream Fishes. Bioscience 2002, 52, 483–498. [Google Scholar] [CrossRef] [Green Version]
  80. Wang, L.; Seelbach, P.W.; Lyons, J. Effects of Levels of Human Disturbance on the Influence of Catchment, Riparian, and Reach-Scale Factors on Fish Assemblages. Am. Fish. Soc. Symp. 2006, 48, 199–219. [Google Scholar]
  81. Lawler, J.J.; Torgersen, C.E. Assessing the Relative Importance of Factors at Multiple Spatial Scales Affecting Terrestrial and Aquatic Wildlife. Curr. Landsc. Ecol. Rep. 2020, 5, 12–24. [Google Scholar] [CrossRef]
  82. Cooper, S.D.; Diehl, S.; Kratz, K.; Sarnelle, O. Implications of Scale for Patterns and Processes in Stream Ecology. Austral Ecol. 1998, 23, 27–40. [Google Scholar] [CrossRef]
  83. Paavola, R.; Muotka, T.; Virtanen, R.; Heino, J.; Jackson, D.; Mäki-Petäys, A. Spatial Scale Affects Community Concordance among Fishes, Benthic Macroinvertebrates, and Bryophytes in Streams. Ecol. Appl. 2006, 16, 368–379. [Google Scholar] [CrossRef] [PubMed]
  84. Heino, J. A Macroecological Perspective of Diversity Patterns in the Freshwater Realm. Freshwat. Biol. 2011, 56, 1703–1722. [Google Scholar] [CrossRef]
  85. Carpenter, S.R. Microcosm Experiments Have Limited Relevance for Community and Ecosystem Ecology. Ecology 1996, 77, 677–680. [Google Scholar] [CrossRef]
  86. Schindler, D.W. Whole-Ecosystem Experiments: Replication Versus Realism: The Need for Ecosystem-Scale Experiments. Ecosystems 1998, 1, 323–334. [Google Scholar] [CrossRef]
Water 15 02213 g001 550
Figure 1. Map of the four study rivers in Ontario, Canada. In the regional map, the grey outlined boxes designate the sampled portions of each watershed (shown in black). River-specific maps show the distribution of sample sites (blue stars) along the river mainstem, dams (red bars), inline lakes (dark grey), and urban areas (light grey with major cities labeled).
Figure 1. Map of the four study rivers in Ontario, Canada. In the regional map, the grey outlined boxes designate the sampled portions of each watershed (shown in black). River-specific maps show the distribution of sample sites (blue stars) along the river mainstem, dams (red bars), inline lakes (dark grey), and urban areas (light grey with major cities labeled).
Water 15 02213 g001
Water 15 02213 g002 550
Figure 2. The percentage of total variance explained by Redundancy Analysis (RDA) solutions for all possible variables (ALL), only catchment variables (CAT), only valley (VAL), or only instream habitat (ISH) variables. Explained variance is AdjR2, which allows comparison of variance explained across and within river basins. River basin names are abbreviated GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa.
Figure 2. The percentage of total variance explained by Redundancy Analysis (RDA) solutions for all possible variables (ALL), only catchment variables (CAT), only valley (VAL), or only instream habitat (ISH) variables. Explained variance is AdjR2, which allows comparison of variance explained across and within river basins. River basin names are abbreviated GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa.
Water 15 02213 g002
Water 15 02213 g003 550
Figure 3. Redundancy Analysis (RDA) triplots for the (a) Ganaraska River (GA), (b) Grand River (GD), (c) Trent (T), and (d) Petawawa (P) rivers. In all plots, sites are numbered upstream to downstream, fish species are color-coded by tolerance class [64], and explanatory variables are from the best RDA solution for each river. Variable codes are in Table 1, and fish species codes are in Appendix B. Axes labels include the percent variance explained (R2) by each axis.
Figure 3. Redundancy Analysis (RDA) triplots for the (a) Ganaraska River (GA), (b) Grand River (GD), (c) Trent (T), and (d) Petawawa (P) rivers. In all plots, sites are numbered upstream to downstream, fish species are color-coded by tolerance class [64], and explanatory variables are from the best RDA solution for each river. Variable codes are in Table 1, and fish species codes are in Appendix B. Axes labels include the percent variance explained (R2) by each axis.
Water 15 02213 g003
Water 15 02213 g004 550
Figure 4. The percentage of adjusted total variance (AdjR2) in the fish community composition explained by partitioning contributions of non-spatial environment, spatially structured environment, pure spatial, and residual (i.e., unexplained). Spatial components are distance to rivermouth (DRVM), Principal Coordinates of Neighbour Matrices (PCNM), and PCNM on separation by dams (DAMS). River basin names are abbreviated GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa.
Figure 4. The percentage of adjusted total variance (AdjR2) in the fish community composition explained by partitioning contributions of non-spatial environment, spatially structured environment, pure spatial, and residual (i.e., unexplained). Spatial components are distance to rivermouth (DRVM), Principal Coordinates of Neighbour Matrices (PCNM), and PCNM on separation by dams (DAMS). River basin names are abbreviated GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa.
Water 15 02213 g004
Table
Table 1. Environmental variables developed to explain variation in fish community composition across and within study rivers. Variable scales include Catchment (CAT), Valley (VAL), and Instream Habitat (ISH). Some variables include multiple, related measures and are indicated by variables with a “Max#” larger than one. X indicates the variable was available for a river basin (GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa).
Table 1. Environmental variables developed to explain variation in fish community composition across and within study rivers. Variable scales include Catchment (CAT), Valley (VAL), and Instream Habitat (ISH). Some variables include multiple, related measures and are indicated by variables with a “Max#” larger than one. X indicates the variable was available for a river basin (GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa).
ScaleCodeMax#DescriptionSource aGDGATP
CATCA1Catchment area (km2)1, 11XXXX
Link1Shreve link number (count of upstream links)1XXXX
DamCtDnM1No. of dams along the downstream mainstem1, 2XX bXX b
DamCtUp1No. of dams upstream along all paths1, 2XXXX
RoadDen1Total length of road\catchment area (per km)3, 11XXXX
WSLU_Dev1Total % of catchment as urban or agriculture (%)4, 11XXXX
WSLU_For1Total % of catchment as forest (%)4, 11XXXX
WSLU_Wet1Total % catchment as wetland (%)4, 11XXXX
WSLUC_type8% of catchment in each land use class (e.g., urban, water, evergreen forest, etc.) (%)4, 11XXXX
SG_BedR1Total % catchment as bedrock surficial geology (%)5, 11XX bXX
SG_Till1Total % catchment as till surficial geology (%)5, 11XXXX
SG_Gfluv1Total % catchment as glaciofluvial deposits surficial geology (%)5, 11X bXXX
SGC_type9% of catchment in each surficial geology class (e.g., Maryhill Till, glaciomarine deposits, organic deposits, etc.) (%)5, 11XXXX
BGC_type8% of catchment in each bedrock geology class (e.g., gneisses of metasedimentary origin, felsic igneous rocks, etc.) (%)6, 11XX bXX
SoilDepth1Area-weighted average of classed depth to bedrock or root restricting layer7, 11X bX bX bX
SoilAWHC1Area-weighted average of soil water holding capacity class7, 11XXX bX
SoilKSAT1Area-weighted average soil hydraulic conductivity7, 11XXXX
MaxTemp1Average maximum temp of the warmest month (°C)8, 11XXXX
TempVar1Average temperature seasonality (StDev × 100) (°C)8, 11XXXX
AnnPrec1Average annual precipitation (mm)8, 11XXXX
PrecVar1Average precipitation seasonality (Coefficient of Variation)8, 11XXXX
VALRipLU_Dev1Total % of local riparian buffer as urban or agriculture (%)4, 11XXXX
RipLU_For1Total % of local riparian buffer as forest (%)4, 11XXXX
RipLU_Wet1Total % of local riparian buffer as wetland (%)4, 11XXXX
RipLUC_type9% of local riparian buffer in each land use class (%)4, 11XXXX
TribCt1Total # of tributaries entering mainstem in buffer1, 11XXXX
MajorTribCt1Total # of major tributaries entering mainstem in buffer1, 11XXXX
Sinuos1Local channel sinuosity in buffer1, 11XXXX
Slope1Local channel slope in buffer1, 9, 11XXXX
MSHabPtch1Total length of mainstem habitat between dams (km)1, 2XX bXX b
TotHabPtch1Total length of all stream habitat between dams (km)1, 2XX bXX b
ISHCDepth1Average depth in channel habitat (m)10X XX
CDepthCV1Coefficient of variation (CV) of depth in channel habitat (%)10X XX
SDepth1Average depth in shoreline habitat (m)10X XX
SDepthCV1CV of depth in shoreline habitat (%)10X XX
Depth1Average depth (m)10XXXX
DepthCV1CV of depth (%)10XXXX
Width1Average river width (m)10XXXX
WidthCV1CV of river width (%)10 XXX
CVel1Average surface velocity in channel habitat (m/s)10X XX
CVelCV1CV of surface velocity in channel habitat (%)10X XX
SVel1Average surface velocity in shoreline habitat (m/s)10X XX
SVelCV1CV of surface velocity in shoreline habitat (%)10X XX
Vel1Average surface velocity (m/s)10X XX
VelCV1CV of surface velocity (%)10X XX
HydHdAve1Average hydraulic head (mm)10 X
HydHdCV1CV of hydraulic head (%)10 X
CSubs1Proportion hard substrate in channel habitat10 XX
CSubsCV1CV of proportion hard substrate in channel habitat (%)10 XX
SSubs1Proportion hard substrate in shoreline habitat10 XX
SSubsCV1CV of proportion hard substrate in shoreline habitat (%)10 XX
Subs1Proportion hard substrate10 XX
SubsCV1CV of proportion hard substrate (%)10 XX
MNPSAve1Mean particle size (cm)10 X
MNPSCV1CV of mean particle size (%)10 X
MXPSAve1Maximum particle size (cm)10 X
MXPSCV1CV of max particle size (%)10 X
CVegProp1Proportion of channel habitat with vegetation 10 X
SVegProp1Proportion of shoreline habitat with vegetation10 X
VegProp1Proportion of river habitat with vegetation10 X
WTemp1Water temperature (°C)10X X
Cond1Conductivity (μS/cm)10 X
Qcms1River discharge (cms)10X
Notes: a Data sources: 1 Ontario Integrated Hydrology Data [42]; 2 Ontario Dam Inventory [43]; 3 Digital Road Network 2001 [44]; 4 GLAHF harmonized land cover 2000/2001 based on a combination of 2000 Provincial Landcover Ontario (PLO) and 2000 Southern Ontario Land Resources Information System (SOLRIS) [45]; 5 Quaternary Geology of Ontario [46]; 6 Bedrock Geology of Ontario [47]; 7 Soil Landscapes of Canada [48]; 8 WorldClim [49]; 9 Provincial Digital Elevation Model [50]; 10 Developed from field measurements; 11 Developed with standard GIS procedures. b Variable not included in analyses; data available, but values did not vary.
Table
Table 2. The four study rivers have notable differences in catchment, valley, instream habitat, and fish community characteristics, as well as variability between sites in each river basin (GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa). Numeric values are average (min, max), except for variables with values that did not vary among sites or variables summarized by basin. Table 1 includes full descriptions of variable codes, and Appendix B includes fish species codes included here.
Table 2. The four study rivers have notable differences in catchment, valley, instream habitat, and fish community characteristics, as well as variability between sites in each river basin (GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa). Numeric values are average (min, max), except for variables with values that did not vary among sites or variables summarized by basin. Table 1 includes full descriptions of variable codes, and Appendix B includes fish species codes included here.
Summary VariableGDGATP
Catchment (CAT):
    CA (km2)5088 (3387, 6645)110 (11,269)10,040 (7318, 12,552)3486 (3164, 4130)
    Link (#)328 (252,434)41 (7, 76)411 (303, 502)274 (261, 289)
    WSLUC_Urban (%)9.1 (8.5, 10.1)2.2 (0.8, 3.2)2.9 (2.6, 3.2)0.1 (0.1, 0.3)
    WSLUC_Agriculture (%)71.8 (70.0, 72.8)51.2 (42.9, 59.1)26.1 (23.2, 29.0)0 (0, 0.1)
    SG_Gfluv (%)071.4 (46.3, 100)7.2 (6.2, 8.1)29.9 (28.8, 30.2)
    DamCtDnM (#)3 (1, 5)112 (0, 24)0
Valley (VAL):
    RipLU_Urban (%)23 (1.0, 94.8)3.4 (0.1, 11.1)23.2 (1.0, 95.1)3.8 (0.0, 52.8)
    RipLU_Wet (%)8.7 (0.0, 29.2)17.0 (7.4, 29.9)25.2 (0.0, 86.9)0.1 (0.0, 0.3)
    Slope (%)0.1 (0.0, 0.3)0.5 (0.1, 1.4)0.1 (0.0, 0.5)0.1 (0.0, 0.5)
    MSHabPtch (river km)36.8 (15.0, 54.7)51.812.6 (0.8, 62.1)72.7
Instream Habitat (ISH):
    Width (m)116 (55, 210)8 (3, 16)188 (69, 604)160 (73, 398)
    Depth (m)1.8 (0.7, 3.6)0.3 (0.1, 0.5)4.0 (1.7, 6.4)4.0 (2, 9.5)
Fish community:
    Species richness (by basin)47263825
    Species richness (by site)11.0 (2,16)10.8 (3, 17)13.2 (4, 23)9 (6,15)
    Intolerant species (% by basin)13231113
    Tolerant species (% by basin)26182717
    Most common fish speciesGREDBTRTROCKSBAS
CARPRAINPUNKPUNK
SBASSLIMSBASROCK
COMMKINGBLUELOGP
GRDHWSUKBNOSBNOS
Table
Table 3. Summary of Redundancy Analysis (RDA) solutions including statistical significance (p) and F-ratio, percent variance explained (R2), and adjusted percent variance explained (AdjR2). Variables are listed in order of inclusion with percentages of marginal and conditional variance explained. Table 1 includes full descriptions of variable codes. Statistical significance is indicated from all possible permutations with cyclic shifts (*** p < 0.01, ** p < 0.05, * p < 0.10). River basin names are abbreviated GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa and variable pools are abbreviated Catchment (CAT), Valley (VAL), Instream Habitat (ISH), or ALL (CAT+VAL+ISH).
Table 3. Summary of Redundancy Analysis (RDA) solutions including statistical significance (p) and F-ratio, percent variance explained (R2), and adjusted percent variance explained (AdjR2). Variables are listed in order of inclusion with percentages of marginal and conditional variance explained. Table 1 includes full descriptions of variable codes. Statistical significance is indicated from all possible permutations with cyclic shifts (*** p < 0.01, ** p < 0.05, * p < 0.10). River basin names are abbreviated GD = Grand, GA = Ganaraska, T = Trent, P = Petawawa and variable pools are abbreviated Catchment (CAT), Valley (VAL), Instream Habitat (ISH), or ALL (CAT+VAL+ISH).
RiverVariable Poolp (F)%ExpAdj %ExpIncluded Variables (Scale)Marg % ExpCond % Exp
GDALL0.019
(4.0)		42.131.6Link (CAT)9.8 **21.0 **
MSHabPtch (VAL)8.3 *9.1
SoilAWHC (CAT)6.26.2
CVel (ISH)5.8 *5.8 *
CAT0.037
(3.8)		33.424.7Link17.7 **21.0 **
SoilAWHC6.96.7
WSLU_Dev5.75.7
VAL0.019
(3.7)		33.424.7MSHabPtch14.9 **18.1 **
RipLUC_Emergent Wetland8.5 **8.6 *
RipLUC_Agriculture6.0 *6.0 *
ISH0.019
(2.9)		19.713.0Vel12.8 **14.3 **
SDepth5.45.4
GAALL0.033
(9.0)		71.063.1Link (CAT)8.3 *39.6 **
RipLU_For (VAL)12.717.4
SoilAWHC (CAT)6.06.0
CAT0.033
(6.9)		65.455.9Link18.8 **41.4 **
MaxTemp11.715.8
SoilAWHC8.28.2
VAL0.067
(7.1)		54.046.3Slope33.1 **32.7 **
RipLUC_Mixed Forest21.321.3
ISH0.067
(2.7)		32.220.9Width20.413.8 **
HydHdAve17.5 *17.5*
TALL0.01
(6.3)		34.128.7SSubs (ISH)6.3 ***13.9 ***
CA (CAT)9.7 **9.8 ***
WSLUC_Urban (CAT)6.5 ***6.6 ***
SVegProp (ISH)3.8 **3.8 **
CAT0.01
(6.4)		20.016.8SoilAWHC12.0 **11.0 ***
WSLUC_Mixed Forest9.0 *9.0 *
VAL0.04
(5.4)		17.314.1RipLU_Wet10.6 ***12.4 **
MSHabPtch5.0 **5.0
ISH0.01
(4.6)		27.221.2SSubs6.4 ***13.9 ***
SVegProp5.7 ***5.6 **
SSubsCV4.9 **4.9 **
SVel2.9 *2.9 *
PALL0.07
(2.4)		41.624.0WSLU_Wet (CAT)19.116.4 **
SSubs (ISH)13.914.3
CDepth (ISH)10.810.8
CAT0.002
(2.4)		30.117.4WSLU_Wet15.516.4
PrecVar13.713.7
VAL0.04
(2.4)		30.417.7RipLUC_Water15.214.5
RipLU_Dev14.9 *11.2
ISH0.11
(1.8)		24.811.1SSubs15.314.4
DepthCV10.410.4
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Sparks-Jackson, B.L.; Esselman, P.C.; Wilson, C.; Carl, L.M. Using Multiscale Environmental and Spatial Analyses to Understand Natural and Anthropogenic Influence on Fish Communities in Four Canadian Rivers. Water 2023, 15, 2213. https://doi.org/10.3390/w15122213

AMA Style

Sparks-Jackson BL, Esselman PC, Wilson C, Carl LM. Using Multiscale Environmental and Spatial Analyses to Understand Natural and Anthropogenic Influence on Fish Communities in Four Canadian Rivers. Water. 2023; 15(12):2213. https://doi.org/10.3390/w15122213

Chicago/Turabian Style

Sparks-Jackson, Beth L., Peter C. Esselman, Chris Wilson, and Leon M. Carl. 2023. "Using Multiscale Environmental and Spatial Analyses to Understand Natural and Anthropogenic Influence on Fish Communities in Four Canadian Rivers" Water 15, no. 12: 2213. https://doi.org/10.3390/w15122213

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop