EDP Sciences logo
Submit your paper
Search
Menu
All issues No 424 (2023) Knowl. Manag. Aquat. Ecosyst., 424 (2023) 19 Full HTML
Issue
Knowl. Manag. Aquat. Ecosyst.
Number 424, 2023
Anthropogenic impact on freshwater habitats, communities and ecosystem functioning
Article Number 19
Number of page(s) 16
DOI https://doi.org/10.1051/kmae/2023014
Published online 20 July 2023

© R.M. Hughes et al., Published by EDP Sciences 2023

Licence Creative CommonsThis is an Open Access article distributed under the terms of the Creative Commons Attribution License CC-BY-ND (https://creativecommons.org/licenses/by-nd/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. If you remix, transform, or build upon the material, you may not distribute the modified material.

1 Introduction

Both native and non-native taxa richness patterns are useful for evaluating areas of greatest conservation concern whether the area is a basin, hydrologic unit, ecoregion, state, or nation. A relatively straightforward way of examining those patterns is by assessing richness determined at site and regional geographic extents (Casarim et al., 2020; Ferreira et al., 2017; Leal et al., 2018; Whittaker, 1960). Studies of the spatial components of richness for fish and macroinvertebrates are widespread. Erös (2007) determined that regional richness was responsible for most of the species richness of six different stream types in Hungary. Ferreira et al. (2017) and Ligeiro et al. (2010) reported that sampling section (typically, <15 m) richness was approximately half that of the entire site richness in Brazilian Cerrado (tropical savanna) streams. Maloney et al. (2011) found that the major drivers for local and regional richness differed, and they differed between ecoregions and the Patuxent basin, Maryland. Leal et al. (2018) determined that the specific fish species contributing to richness were distinctly different among river basins and among sites within those basins in Brazilian Amazonia streams. Casarim et al. (2020) found that fish species regional richness was approximately nine times that of mean site richness in Brazilian Cerrado stream sites occurring across four Minas Gerais national parks. Thus, focusing on site richness, whether native or non-native, in bioassessment surveys misses the importance of regional taxa richness.

Although site and regional taxonomic richness are important indicators of ecosystem condition, they vary for both natural and anthropogenic reasons. (Valdez et al., 2023) They can be increased by the introduction of non-native tolerant species as well as by moderate nutrient and temperature increases in cold-water streams (Davies and Jackson, 2006; Hughes et al., 1998; Lomnicky et al., 2007; McCormick et al., 2001; Mebane et al., 2003; Vadas et al., 2022), warm water streams (Davies et al., 2008; Oliveira-Junior et al., 2015), and lakes (Kaufmann et al., 2014; Whittier et al., 2002). Furthermore, taxonomic richness estimates are increased by increased sampling effort (Cao et al., 2002; Hughes et al., 2012; Kanno et al., 2009; Ligeiro et al., 2013b; Pompeu et al., 2021; Silva et al., 2016; Terra et al., 2013) and increased water body size (Fausch et al., 1984; McGarvey and Hughes, 2008; McGarvey and Ward, 2008). Smith and Jones (2005) estimated that up to 50 sites per watershed must be sampled to obtain 90% of the fish species occurring in Wisconsin (USA) watersheds, which drained 23.6 to 432.8 km2. Clearly, it is important to use standard sampling methods and levels of sampling effort when conducting comparative studies among landscape units (Bonar et al., 2009; Hughes and Peck, 2008). But it is also important to calibrate regional richness against the number of sites sampled through use of sampling effort curves and residual richness estimates (Shurin et al., 2000).

Recent assessments of the spatial partitioning of richness in aquatic ecosystems have focused on stream fishes or macroinvertebrates in relatively small basins or hydrologic units because data are more frequently available for such areas (Erös, 2007; Ferreira et al., 2017; Leal et al., 2018; Ligeiro et al., 2010). However, the USEPA's National Rivers and Streams Assessment (NRSA) offers a fish and macroinvertebrate database for thousands of stream and river sites sampled via standard methods across the entire conterminous United States (CONUS; Herlihy et al., 2020; USEPA, 2015b; USEPA, 2020). The NRSA is based on a probability survey design, which enables inferring the results to over a million kilometers of stream and river length represented by the sample sites. By partitioning the site selection by stream size and ecoregion, the sampling ensured a wide range of natural environmental and anthropogenic disturbance conditions (Cao and Wang, 2023). Also, we were interested in assessing whether, and which, of several candidate natural or anthropogenic predictor variables were most strongly associated with regional richness estimates across the CONUS.

Therefore, in this study we determined fish and macroinvertebrate richness at site, hydrologic unit (HUC), and ecoregion spatial extents. We also determined the major environmental variables explaining residual fish and macroinvertebrate beta taxonomic richness for level-III Omernik (1987) ecoregions (USEPA, 2021). Based on previous research, we had two objectives for this manuscript. (1) Quantify the degree to which taxonomic richness for both fish and macroinvertebrates increases with increased number of sites sampled and increased spatial extents of geographic units. Based on previous studies (Cao et al., 2002; Hughes et al., 2021; Li et al., 2001; Valdez et al., 2023), we predicted that our regional sampling effort would not attain an asymptote for either assemblage. (2) Determine the similarities and differences in environmental predictors of regional richness for both macroinvertebrates and fish among different level-III ecoregions. Based on previous results (Cao and Wang, 2023; Herlihy et al., 2001, Herlihy et al., 2019; Herlihy et al., 2020; Hughes et al., 2021; USEPA, 2021), we expected that the most significant regional richness predictors would differ between fish and macroinvertebrates and that they would include both natural and anthropogenic variables.

2 Methods

2.1 Study design

The NRSA used a probability-based design to select sample sites (Olsen and Peck, 2008; Stevens and Olsen, 2004; USEPA, 2016a). The target population included all streams and rivers with flowing water during the June-September index period based on the coverage in the National Hydrography Dataset (USGS, 2013b), which generally reflects the blue-line stream network at the 1:100,000 map scale. The NRSA is representative of a target population of 1,231,000 km of lotic systems ranging from the Mississippi River to headwater streams. The design was spatially balanced and stratified by state, ecoregion, and stream order to even out the sample site distribution across areas and stream sizes, but site densities were naturally greater in humid than xeric regions. We compiled NRSA data from surveys in the summers of 2008, 2009, 2013 and 2014. When a site was sampled multiple times, we only used data from the latest sample year. All told, NRSA crews sampled 3475 unique stream and river sites across the CONUS (Fig. 1) that contained fish and benthic macroinvertebrate assemblage data. Sites were sampled in one-day field visits (Hughes and Peck, 2008).

thumbnail Fig. 1

Locations of the NRSA sample sites and the nine aggregate ecoregions used for coarse-grain regional richness analysis.

2.2  Fish and macroinvertebrate sampling

Fish assemblages were sampled as described in Hughes and Peck (2008), USEPA (2009), USEPA (2013a); USEPA (2013b), nearly all by backpack or boat electrofishing. In wadeable sites <13 m wide, site lengths equaled 40 wetted channel widths, or a minimum of 150 m. At wadeable sites >13 m wide and boatable sites, minimum site length was the greater of 500 m or 20 wetted channel widths. Large wadeable and boatable sites were sampled beyond the minimum site length until 500 individuals were obtained or a site length of 40 wetted channel widths was reached. Individuals were identified at the site, counted then released alive. Taxonomic names were based on Nelson et al. (2004) and Page et al. (2013). Fish data were absent from 921 sites because of absence of fish (small streams), gear failure (relatively rare), or inability to obtain sampling permits because of endangered or threatened species protections, leaving 2554 sites for fish data analysis (Tab. 1).

Macroinvertebrate assemblages were sampled as described in Hughes and Peck (2008), USEPA (2009), USEPA (2013a) and USEPA (2013b). At each site, a sampling reach of 150 m or 40 channel widths was established. At wadeable sites, 11 subsamples were collected in a zig-zag pattern at each of 11 equidistant transects by using a D-frame kick net (0.09 m2 area, 500-um (mesh; Hughes and Peck 2008). At non-wadeable sites, samples were collected from the nearshore area of each transect (alternating between banks). The subsamples were combined into a single composite sample for the site, placed in ethanol, and sent to a laboratory, where a fixed count of 500 individuals were identified to specified taxonomic levels (usually genus) by using local, regional, and national keys (USEPA, 2012). The 500-individual count goal in the laboratory was not always achieved. Thus, for consistency, a random subset of 300 individuals was selected from each sample and used for all subsequent data analyses. Very few sites lacked sufficient macroinvertebrate data, 3358 sites were available for analysis (Tab. 1).

Table 1

Codes, names, sample sizes and composite richness for coarse-grain NRSA study units.

2.3 Data analyses

Ambiguous taxa identifications create a problem for calculating richness when dealing with a variety of national, regional, and site-level composite richness estimates. To make things consistent, we used the species level for fish and the genus level for macroinvertebrates in calculating richness. We considered total fish richness (both native and non-native species) because the same species may be native or non-native in some regions. Any fish not identified to species was removed from the analysis (e.g., all hybrids, unknown sculpin, etc.). Similarly, for macroinvertebrates, any individual not identified to genus was removed and any species-level identifications were grouped up to genus.

Site richness was calculated as the number of unique species (fish) or genera (macroinvertebrates) at each sample site. The conterminous USA richness was just the total richness in the entire dataset. Regional richness was calculated for two different, but commonly used, landscape classifications (ecoregion and hydrologic unit, HUC) at both coarse and fine grains. We used Omernik's 85 level-III ecoregions (Omernik, 1987; USEPA, 2021) as the basis for the fine-grained ecoregion composite. For the coarse-grained ecoregion composite, we used the nine aggregate ecoregions used in NRSA analysis (Fig. 1). We defined those nine ecoregions by aggregating the 85 individual level-III Omernik ecoregions as described in Herlihy et al. (2008) and listed in Table 1. For the HUC landscape classification, we used USGS (2013a). The coarse-grained HUCs are the 18 water resource regions (HUC2) in the CONUS (listed in Tab. 1). The fine-grained HUCs are the 391 next finer HUC level (HUC4). Thus, we used four measures of regional richness calculated as the sum of the unique fish species and macroinvertebrate genera in each of the two landscape classification types at both fine and coarse grains.

Richness increases with sampling effort, so the more sites that are sampled in an ecoregion or HUC the larger taxa richness will be. To adjust for this sample size effect, we also calculated regional richness as a residual richness. Using the approach of Shurin et al. (2000), we calculated residual richness as the deviation of observed richness from predicted richness given the number of sample sites in the composite ecoregion or HUC. Predicted regional richness was calculated using an equation of the form R = a*Nb where R is richness, N is the number of sites in the composite, and a and b are fitted regression coefficients. We used non-linear regression and all the regional richness data to fit the model coefficients using SAS PROC NLIN. Residual richness is a residual so it can be positive or negative. High positive residual richness in a region indicates a higher taxa richness than predicted given the number of sample sites in that region and is indicative of taxa-rich regions. On the other hand, negative residual richness indicates that a region has fewer taxa than predicted given the number of samples in the region and indicates taxa-poor regions. Residual richness is not the same as absolute richness (number of taxa in region), similar regional residual richness values may have different observed total regional richness. Using residual richness is necessary to directly compare regions because absolute richness is strongly related to the number of sample sites in a region, and the different regions had different numbers of sample sites. Those site numbers differed because of regional differences in total surface area, drainage density, and vagaries in the stratification associated with the NRSA probability design.

We assessed the relationships between regional-scale environmental variables and the residual regional richness at the level-III ecoregion scale. We removed ecoregions or HUCs that lacked at least 10 sample sites from the analysis. As a result, we did not conduct environment-richness analyses for fine-grain HUCs because nearly half those HUCs lacked sufficient data. We used available GIS data layers to quantify level-III ecoregion-scale road, human population, and mine density, mean air temperature, mean annual precipitation and elevation, ecoregion percentage land cover categories (agriculture, developed, wetland, bare ground, open water, forest, grass+shrub), and historic glaciation within an ecoregion or HUC. GIS data layers for road, human population, and mine densities were taken from ESRI (2006), Falcone (2016), and Mason and Arndt (1996), respectively. We used Daly et al. (2008) for temperature and precipitation data and Danielson and Gesch (2011) for elevation data. Because of the effects of historical glaciation on fish distributions (Smith et al., 2010), we used the extent of the last glacial maximum from Ehlers et al. (2011). Determinations of the various amounts of catchment land uses were based on Yang et al. (2018) and the HUC delineations are from USGS (2013a).

Residual regional richness was then related to the environmental data using correlation, multiple regression, and random forest modeling to determine the degree to which the differing analytical approaches affect environment-richness relationships. As recommended by Cao and Wang (2023) and Mostafavi et al. (2019), we used two different statistical analyses to assess common predictor variables. Density variables, precipitation, and elevation were log10 transformed before analysis. We quantified correlations with Pearson correlation coefficients using SAS PROC CORR, and we used all-subset multiple regression to identify the most important predictor variables (Burnham and Anderson, 1988). We examined all possible model combinations and those with an adjusted Akaike information criterion (AIC) within 5 of the top models were evaluated. Variables that consistently appeared in the top models, weighted by AIC values, were selected for inclusion in the final multiple regression models. The regressions were all conduced in SAS using PROC REG.

Random Forest analysis (Breiman, 2001) was conducted by using the randomForest package in R version 4.0.3 (Liaw and Wiener, 2002; R Core Team, 2020). We set the number of trees to 1000 and the number of variables randomly sampled as predictors at each split (mtry) to the default for regression models, which is the number of predictor variables divided by three. Additionally, predictor variables were evaluated for redundancy through use of Pearson correlation coefficient analysis, with an r > 0.7 as an indicator of redundancy. All correlations > 0.7 occurred between a natural driver and a human pressure (e.g., elevation and human population density). When that occurred, we retained the natural driver and excluded the human pressure variable from the random forest analysis.

3 Results

3.1 Raw taxa richness

NRSA collected a total of 813 distinct macroinvertebrate genera at 3358 sites across the CONUS; Twardochleb et al. (2021) reported a potential total richness of 932 insect genera in the CONUS. At individual sites in our study, richness ranged from 1 to 65 genera per site. Across all sites, mean site macroinvertebrate genus richness was 27.4 (SD = 11.5). Coarse-grain regional macroinvertebrate richness among ecoregions ranged from 357 in the Southern Plains to 484 in the Western Mountains (Tab. 1). Among coarse-grain HUCs, regional richness ranged from 250 in the Lower Colorado to 476 in the Missouri (Tab. 1). Fine-grain macroinvertebrate regional richness was much lower, usually less than half, of the coarse-grain richness in both ecoregion and HUC classes (Fig. 2) because of smaller spatial extents and sample sizes. Macroinvertebrate regional richness is also available in map form (Figures S1 and S2).

For fish, based on 2554 sites across the CONUS, NRSA collected a total of 582 distinct fish species out of a potential 863 species nationally (Jenkins et al., 2015). Individual site richness ranged from 1-54 species with an average of 12.5 (SD=8.6). Coarse-grain ecoregional fish richness had a much wider range than macroinvertebrates, ranging from 84 in the Northern Plains to 350 in the Southern Appalachians (Tab. 1). Among coarse-grain HUCs, regional fish richness ranged from 29 in the Great Basin to 294 in the South Atlantic-Gulf (Tab. 1). Fine-grain fish regional richness was also lower than coarse-grain richness in both ecoregion and HUC classes (Fig. 2) because of smaller spatial extents and sample sizes. Fish regional richness is also available in map form (Figures S3 and S4).

thumbnail Fig. 2

Fish species and macroinvertebrate genera richness at the individual site and various regional scales. Fine-grained regional richness is based on Omernik Level-III ecoregions and level-4 Hydrologic Units (HUCs). Coarse-grained regional richness is based on nine aggregated Omernik ecoregions and level-2 HUCs. Boxes are the interquartile ranges, lines in the boxes are medians, and the whiskers are 5th/95th percentiles.

3.2 Residual taxa richness

To adjust for the variation in sample size among all the different measures of regional richness, we calculated residual richness following Shurin et al. (2000) by subtracting observed richness from predicted richness. For macroinvertebrates, the fitted predicted richness regression equation (Fig. 3) was R= 55.8*N0.352 (RMSE=31.0). For fish, the fitted predicted richness regression equation (Fig. 3) was R = 14.9*N0.441 (RMSE = 29.7).

thumbnail Fig. 3

Composite fish and macroinvertebrate regional richness versus number of sample sites in the composite for all spatial extents. The red line is fitted from nonlinear regression. Residual regional richness was calculated as the deviation from the fitted line.

3.3 Ecoregional residual taxa richness

Maps of coarse-grain ecoregion residual richness showed the highest regional richness in the Southern Appalachians and Coastal Plains for fish and in the Western Mountains and Xeric West ecoregions for macroinvertebrates (Fig. 4). The lowest residual richness ecoregions were in the Northern Plains for fish and in the Temperate Plains for macroinvertebrates. Regional residual richness patterns in the fine-grain ecoregion maps show more spatial detail (Fig. 5). The highest residual fish richness was in the Southeastern Plains. For macroinvertebrates, residual richness was highest in the Southern Blue Ridge and North Central Appalachians.

thumbnail Fig. 4

Coarse-grain residual regional richness of fish species and macroinvertebrate genera among ecoregions. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.
thumbnail Fig. 5

Fine-grain residual regional richness of fish species and macroinvertebrate genera among fine-grain ecoregions. Ecoregions with <10 sample sites were deemed to have insufficient data. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue-green colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.

3.4 HUC residual taxa richness

HUC coarse-grain residual richness maps resemble the coarse-grain ecoregion maps for fish but not for macroinvertebrates (Fig. 6). Macroinvertebrates have low residual regional richness in the Missouri River HUC but several high residual richness areas throughout the U.S. The HUC fine-grain residual richness maps (Fig. 7) bear a strong resemblance to the coarse-grain HUC maps but contain considerable white space because many fine-grain HUCs lacked enough sample sites to assess, especially for fish.

thumbnail Fig. 6

Coarse-grain residual regional richness of fish species and macroinvertebrate genera among hydrological units. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue-green colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.
thumbnail Fig. 7

Fine-grained residual regional richness of fish species and macroinvertebrate genera among hydrologic units. Units with <10 sample sites were deemed to have insufficient data. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue-green colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.

3.5 Major drivers of macroinvertebrate ecoregional residual richness

For macroinvertebrates, 74 of the 85 level-III ecoregions had enough sample sites to use in this analysis (n ≥ 10). Percent forested land cover was the variable most strongly correlated with macroinvertebrate regional residual richness (r=0.584, Tab. 2). All-subset multiple regression analysis identified four variables as being most important (%forest, %agriculture, air temperature, and road density). The final multiple regression model with these variables had an R2=0.512 and RMSE of 22.4 (Table 2). The random forest analysis also identified %forest, %agriculture and air temperature as being the most important predictors of regional residual richness of macroinvertebrate genera (Table 2).

3.6 Major drivers of fish ecoregional residual richness

For fish, 61 of the 85 level-III ecoregions had sufficient samples (n ≥ 10) for analysis. Road density had the highest correlation with regional residual fish richness (r=0.611) but mean elevation, %grass+shrub, human population density, mean air temperature and mean precipitation were also highly correlated with regional residual richness of fish species (|r| > 0.5, Tab. 3). The all-subset multiple regression analysis identified four variables as being the most important predictors of fish regional residual richness (%grass+shrub, mean air temperature, mean precipitation, and %open water). The final multiple regression model with those variables had an R2=0.631 and RMSE of 21.8 (Tab. 3). The random forest analysis identified %developed, %agriculture, mean elevation, and mean precipitation as being the most important predictors of regional residual fish species richness (Tab. 3).

4 Discussion

4.1 Usefulness of a probability study design and standard methods

To our knowledge, this study is the first to determine mean site and regional richness of both fish and aquatic macroinvertebrates across the CONUS. We could do this because we had access to a dataset based on a probability sample of the USA's entire mapped stream network (USEPA, 2016), together with data collected through use of standard methods at all the sites (Hughes and Peck, 2008; USEPA, 2013a; USEPA, 2013b). Such spatially extensive studies of regional taxa richness based on standard sampling methods are exceedingly rare. Therefore, both the study design and sampling methods make our results representative of the regional patterns in CONUS streams, unlike studies based on ad hoc site selection, large data compilations, and variable sampling methods. In addition, NRSA collected 87% of the CONUS macroinvertebrate genera recorded by Twardochleb et al. (2021) and 67% of the CONUS fish species recorded by Jenkins et al. (2015).

4.2 Limitations of our macroinvertebrate richness estimates

Although Twardochleb et al. (2021) reported a potential total richness of 932 insect genera (versus our 813 total macroinvertebrate genera) across the CONUS, both are likely substantial underestimates of reality. For example, Dieterich and Anderson (2000) and Dieterich (1992) reported collecting 225 genera from six summer-dry Oregon stream sites versus 74 in a neighboring permanent headwater stream. Meyer et al. (2007) reported that 145 aquatic insect genera were collected from eight unmapped headwater streams over 30 y at the Coweeta Hydrologic Laboratory, North Carolina. Underestimates of true site and landscape-scale taxa richness are driven by multiple factors. These may include incomplete taxonomic identification (Hering et al., 2004; Lenat and Resh, 2001), insufficient sampling effort (Cao et al., 2002; Hughes et al., 2012; Li et al., 2001; Li et al., 2014), under-sampling of non-permanent and isolated waters (Anderson and Anderson, 1995; Dieterich and Anderson, 2000; Meyer et al., 2007), seasonal differences in taxa observations (Chen et al., 2014; Dieterich and Anderson, 2000; Fierro et al., 2021), and rare species (Anderson and Anderson, 1995; Hughes et al., 2021; Kanno et al., 2009; Li et al., 2014; Pompeu et al., 2021; Silva et al., 2016; Terra et al., 2013). Therefore, taxa richness estimates should include clear descriptions and limitations of study design, sampling effort, season, and level of taxonomic identifications (Valdez et al., 2023).

4.3 Regional vs. site macroinvertebrate genera richness

Regional richness markedly exceeded mean site richness for macroinvertebrates, as we hypothesized (Tab. 1). For macroinvertebrate assemblages, the small-grain regions incorporated at least seven times as many genera as the average site, and the large-grain regions supported at least fifteen times as many genera. These are greater differences than those reported by Ferreira et al. (2017) for Cerrado stream macroinvertebrates, but our ecoregions and HUCs were larger than their HUCs.

4.4 Regional vs. site fish species richness

Regional richness also markedly exceeded mean site richness for fish, as we hypothesized (Tab. 1). For fish assemblages, small-grain and large-grain regional richness were five and eleven times greater than mean site species richness. Leal et al. (2018) found that large- and small-grain fish regional species richness were eight and four times those of average site species richness in eastern Amazonian streams. Although their basins and regions were smaller than ours, they supported 60 to 134 fish species.

4.5 Effects of sampling effort on taxa richness

As expected, regional macroinvertebrate genus richness and fish species richness increased with the number of sites sampled (Figs 2 & 3), presumably because the pool of available species becomes greater with an increase in the geographic area represented by the sample sites. We observed greater ranges (both negative and positive) in fish composite richness than macroinvertebrate composite richness, presumably because of the wider range in fish species richness than macroinvertebrate genus richness at sites (Hughes et al., 2012; Macedo et al., 2014). However, our sampling efforts failed to attain asymptotes in taxa richness versus sample size (Fig. 3). Therefore, biodiversity surveys based on more extensive sampling may be warranted should the USA wish to more accurately document status and trends in regional macroinvertebrate and fish taxa richness. The NRSA tracks biological status and trends nationally and regionally through use of multimetric indices and observed/expected taxa richness (Hill et al., 2017; USEPA, 2016) but such indicators tend to omit uncommon, rare and particularly sensitive taxa (Brito et al., 2020; Firmiano et al., 2017; Kanno et al., 2009; Martins et al., 2021). Clearly, markedly greater sampling effort than that practiced by NRSA is needed for accurate estimates of regional fish and macroinvertebrate taxa richness.

4.6 Biodiversity hotspots

Although biodiversity assessment is not a stated NRSA goal, it nonetheless indicated USA biodiversity hotspots and cool spots for fish species and macroinvertebrate genera richness (Figs 4–7). Such patterns were also reported by Hocutt and Wiley (1986) and Hughes et al. (2005) for fish. If the USA chooses to protect 30% of its land areas by 2030 (The White House, 2021) and 50% by 2050, it might be wise to focus some of those protections on biodiversity hotspots, as proposed by Leal et al. (2020) for Amazonia. Regional hotspots are indicated by the outliers above the regression lines in Figure 3 and in blue on Figs 4–7. Protected areas might be further focused on regions where both fish and macroinvertebrate richness can be maximized, together with those riverscapes and landscapes where terrestrial protections are projected (Fausch et al., 2002; Hughes et al., 2019; Leal et al., 2020; Su et al., 2021).

4.7 Major predictors of fish & macroinvertebrate taxa richness

Small-grain ecoregions showed different predictors of macroinvertebrate genus and fish species residual regional richness as hypothesized (Tab 2 and 3). Air temperature was an important predictor in both cases but was moderately positive for fish and weakly negative for macroinvertebrates. Cold, high elevation and upper Midwest ecoregions generally support greater residual macroinvertebrate genera richness (Fig. 7), but 40% of the aquatic insect fauna reportedly occurs in the southeastern USA (Morse et al., 1997). However, that estimate was not based on probability sampling nor corrected for sampling effort. On the other hand, generally warmer, lower-elevation southeastern ecoregions support the greatest residual fish species richness (Fig. 7). This is likely associated with the evolutionary and biogeographic drivers of fish species richness, whereby continental glaciation produced refugia, greater ecological stability, speciation, and species richness in the southeastern USA. But tectonism and greater droughts and aridity tended to limit speciation and species richness in the western USA (Hocutt and Wiley, 1986; Leroy et al., 2019; Oberdorff et al., 1995; Smith et al., 2010; Tedesco et al., 2012; Tonn, 1990). On the other hand, in temperate regions, aquatic macroinvertebrate taxa richness typically increases with greater daily, seasonal, and annual air temperature ranges. The opposite occurs at lower daily, seasonal, and annual temperature ranges (Bonacina et al., 2022; Garcia-Giron et al., 2023). Also, Vinson and Hawkins (2003) reported that local contemporary factors were more important drivers of insect taxa richness than historical biogeographic factors globally.

Similarly, % agriculture was negatively correlated with residual regional macroinvertebrate richness, but positively associated with regional residual fish richness (along with % developed land, road density, population density, and precipitation). Presumably, those different relationships are associated with the greater residual regional fish richness in the southeastern USA, where anthropogenic and natural driver variables co-vary (Tab. 3, Smith et al., 2010). On the other hand, residual regional macroinvertebrate richness was strongly correlated with % forested land, which widely occurs in the Western Mountains, Upper Midwest, and Northern and Central Appalachians. Also, Vinson and Hawkins (2003) reported highest EPT (Ephemeroptera, Plecoptera, Trichoptera) genera richness in forested streams globally and Rumschlag et al. (2023) reported declines in sensitive macroinvertebrate taxa over recent decades in conterminous USA agricultural stream sites.

Table 2

Pearson correlation, multiple regression, and random forest results for relating residual regional macroinvertebrate genera richness to regional environmental data composited at the fine-grain ecoregion extent. Variables sorted in order of decreasing correlation strength.

Table 3

Pearson correlation, multiple regression, and random forest results for relating residual regional fish species richness to regional environmental data composited at the fine-grain ecoregion extent. Variables sorted in order of decreasing correlation strength.

4.8 Regression vs. random forest macroinvertebrate richness predictors

All-subsets multiple regression identified similar important predictors of regional richness as random forest (RF) analyses. All-subset multiple regression analysis identified %forest, %agriculture, air temperature, and road density as being the most important predictors of macroinvertebrate regional residual richness (Tab. 2). RF analysis also identified %forest, %agriculture and air temperature as being the top three predictors of macroinvertebrate genera residual richness (Tab. 2), but road density was highly correlated (r > 0.7) with elevation, so it was dropped from the RF analysis. Researchers frequently have associated macroinvertebrate taxa richness with catchment forest and catchment agriculture in temperate streams (Gerth et al., 2016; Rumschlag et al., 2023; Vinson and Hawkins, 2003), as well as in tropical and subtropical streams (Brito et al., 2020; Ligeiro et al., 2013a; Macedo et al., 2014; Martins et al., 2021; Silva et al., 2018). Forests have a positive effect by mitigating many anthropogenic disturbances whereas agriculture leads to increased levels of fine sediments, nutrients, biocides, and physical habitat impairment (Allan, 2004; Allan et al., 1987; Hughes and Vadas, 2021; Kaufmann et al., 2022a; Kaufmann et al., 2022b), thereby reducing the number of sensitive taxa.

4.9 Regression vs. random forest fish richness predictors

As with macroinvertebrates, regression identified different predictors of regional fish richness than RF analyses. All-subset multiple regression analysis identified %grass+shrub, mean air temperature, mean precipitation, and %open water as being the most important predictors of fish regional residual richness (Tab. 3). But RF identified %developed, %agriculture, mean elevation and mean precipitation as being the most important predictors (Tab. 3). Because %grass+shrub and %open water were strongly correlated with more natural predictor variables (elevation, precipitation, air temperature), we dropped the former from RF analysis. Thus, RF and all-subset multiple regression analysis both indicated the importance of both natural and anthropogenic predictors of regional residual taxa richness across the CONUS, as predicted (Herlihy et al., 2019; Herlihy et al., 2020; Hughes et al., 2019). Increased elevation and reduced temperature and precipitation limit the niche space for fish species (McGarvey and Hughes, 2008; McGarvey and Terra, 2016). On the other hand, lower elevations and higher temperatures and precipitation increase the fish niche space (McGarvey and Hughes, 2008; McGarvey and Terra, 2016), as well as the niche space for agriculture and urban development.

5 Conclusions

We recommend increased biodiversity monitoring through use of standard methods and probability survey designs to better determine the status and trends in regional taxa richness in the CONUS and globally (Feio et al., 2022; Rumschlag et al., 2023;). Ideally in the USA, this would be accomplished via collaborative federal, state, and local efforts based on the Clean Water Act, Endangered Species Act, National Environmental Policy Act, and include citizen scientists. Similar policy drivers exist in Australia (National River Health Program, Australian River Assessment System), Europe (Water Framework Directive), Japan (National Census on the River Environment), South Africa (National Water Act, National Environmental Management: Biodiversity Act, River Ecosystem Monitoring Programme), and South Korea (Water Environment Conservation Act). But all these laws and programs currently lack sufficient funding for implementing accurate national or continental estimates of aquatic taxa richness (Feio et al., 2022).

Data availability

NRSA data are publicly available online on the EPA NARS website or from the authors upon reasonable request.

Conflict of interest

The authors declare no competing interests.

Author contributions

CRediT: RMH conceptualization, methodology, writing original draft & revisions. ATH: conceptualization, data curation, methodology, formal analysis, methodology, validation, visualization, writing original draft & revisions. RC: data curation, methodology, validation & visualization. DVP: data curation, methodology & validation. RMM: funding acquisition, resources, formal analysis, writing & editing. SGP: funding acquisition, project administration, resources & supervision.

Acknowledgments

We thank the large number of state, federal, and contractor field crew members, data management and laboratory staff, and National Rivers & Streams Assessment (NRSA) team members involved with obtaining and synthesizing the NRSA data. This research was conducted while ATH held a National Research Council Senior Research Associateship at the U.S. Environmental Protection Agency's Center for Public Health and Environmental Assessment, Pacific Ecological Systems Division, Corvallis, Oregon. RMH held a Fulbright Distinguished Scholar position at the Federal University of Para. The manuscript was subjected to Agency review and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency. We appreciate the constructive reviews of an earlier manuscript by Joe Ebersole and Frank McCormick.

References

  • Anderson TM, Anderson NH. 1995. The insect fauna of spring habitats in semiaridrangelands in central Oregon. J Kansas Entomol Soc 68: 65–76. [Google Scholar]
  • Allan JD. 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems. Ann Rev Ecol Sys 35: 257–284. [CrossRef] [Google Scholar]
  • Allan JD, Erickson DL, Fay J. 1997. The influence of catchment land use on stream integrity across multiple scales. Freshwat Biol 37: 149–161. [CrossRef] [Google Scholar]
  • Bonancina L, Fasano F, Mezzanotte V, Fornaroli R. 2022. Effects of water temperature on freshwater macroinvertebrates: a systematic review. Biol Rev doi.org/10.1111/brv.12903. [Google Scholar]
  • Bonar S, Hubert W, Willis D (editors). 2009. Standard methods for sampling North American freshwater fishes. American Fisheries Society, Bethesda. [Google Scholar]
  • Breiman L. 2001. Random forest. Machine Learn 45: 5–32. [CrossRef] [Google Scholar]
  • Brito JG, Roque FO, Martins RT, Hamada N, Nessimian JL, Oliveira VC, Hughes RM, de Paula FR, Ferraz S. 2020. Small forest losses degrade stream macroinvertebrate assemblages in the eastern Brazilian Amazon. Biol Conserv 241: 108263. [CrossRef] [Google Scholar]
  • Burnham KP, Anderson DR. 1998. Model selection and inference: a practical information-theoretic approach. New York: Springer. [CrossRef] [Google Scholar]
  • Cao Y, Wang L. 2023. How to statistically disentangle the effects of environmental factors and human disturbances: a review. Water 15: 734. [CrossRef] [Google Scholar]
  • Cao Y, Larsen DP, Hughes RM, Angermeier PL, Patton TM. 2002. Sampling effort affects multivariate comparisons of stream communities. J N Amer Benthol Soc 21: 701–714. [CrossRef] [Google Scholar]
  • Casarim R, Caldeira YM, Pompeu PS. 2020. Representativeness of national parks in protecting freshwater biodiversity: a case of Brazilian savanna. Ecol Freshwat Fish 29: 705–721. [CrossRef] [Google Scholar]
  • Chen K, Hughes RM, Xu S, Zhang J, Cai D, Wang B. 2014. Evaluating performance of macroinvertebrate-based adjusted and unadjusted multi-metric indices (MMI) using multi-season and multi-year samples. Ecol Indicat 36: 142–151. [CrossRef] [Google Scholar]
  • Daly C, Halbleib M, Smith JI, Gibson WP, Doggett MK, Taylor GH. 2008. Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. Internat J Climat 28: 2031–2064. [CrossRef] [Google Scholar]
  • Danielson JJ, Gesch DB. 2011. Global multi-resolution terrain elevation data 2010 (GMTED2010). U.S. Geological Survey Open-File Report 2011-1073. https://www.usgs.gov/core-science-systems/eros/coastal-changes-and-impacts/gmted2010 [Google Scholar]
  • Davies SP, Jackson SK. 2006. The biological condition gradient: a descriptive model for interpreting change in aquatic ecosystems. Ecol Appl 16: 1251–1266. [CrossRef] [PubMed] [Google Scholar]
  • Davies P, Harris J, Hillman T, Walker K. 2008. A report on the ecological health of rivers in the Murray-Darling basin, 2004–2007. Murray-Darling Ministerial Council, Canberra. [Google Scholar]
  • Dieterich M. 1992. Insect community composition and physico-chemical processes in summer-dry headwater streams of western Oregon. Ph. D. Thesis. Oregon State University, Corvallis.xx [Google Scholar]
  • Dieterich M, Anderson NH. 2000. The invertebrate fauna of summer-dry streams in western Oregon. Archiv Hydrobiol 142: 273–295. [CrossRef] [Google Scholar]
  • Ehlers J, Gibbard PL, Hughes PD. 2011. Quaternary glaciations: extent and chronology. https://crc806db.uni-koeln.de/layer/show/6 [Google Scholar]
  • Erös T. 2007. Partitioning the diversity of riverine fish: the roles of habitat types and non-native species. Freshwat Biol 52: 1400–1415. [CrossRef] [Google Scholar]
  • ESRI. 2006. TeleAtlas 2006. ArcGIS 9.2 file geodatabase. 2006; https://www.roadsbridges.com/tele-atlas-north-america-inc [Google Scholar]
  • Falcone JA. 2016. U.S. block-level population density rasters for 1990, 2000, and 2010. U.S. Geological Survey data release. doi.org/10.5066/F74J0C6M. [Google Scholar]
  • Fausch KD, Karr JR, Yant PR. 1984. Regional application of an index of biotic integrity based on stream fish communities. Trans Amer Fish Soc 113: 39–55. [CrossRef] [Google Scholar]
  • Fausch KD, Torgersen CE, Baxter CV, Li HW. 2002. Landscapes to riverscapes: bridging the gap between research and conservation of stream fishes. BioScience 52: 483–498. [CrossRef] [Google Scholar]
  • Feio MJ, Hughes RM, Serra SRQ, Nichols SJ, Callisto M, Macedo DR, Harding J, Yates AG, Odume ON, Baek MJ, Mercado-Silva N, Nakamura K, Jae Y, Chen K, Campbell I, Martins RT, Arimoro FO, Kefford BJ, Moya N, Devi R, Keke UN, Lintermans M, Alves CBM, Monk W, Mori T, Pompeu PS, Robinson W, Shah DN, Sueyoshi M. 2022. Fish and macroinvertebrate assemblages reveal extensive degradation of the world's rivers. Global Change Biol. doi.org/10.1111/gcb.16439 [Google Scholar]
  • Ferreira WR, Hepp LU, Ligeiro R, Macedo DR, Hughes RM, Kaufmann PR, Callisto M. 2017. Partitioning taxonomic diversity of aquatic insect assemblages and functional feeding groups in neotropical savanna headwater streams. Ecol Indicat 72: 365–373. [CrossRef] [Google Scholar]
  • Fierro P, Hughes RM, Valdovinos C. 2021. Temporal variability of macroinvertebrate assemblages in a Mediterranean coastal stream: implications for bioassessments. Neotrop Entomol 50: 873–885. [CrossRef] [PubMed] [Google Scholar]
  • Firmiano KR, Ligeiro R, Macedo DR, Juen J, Hughes RM, Callisto M. 2017. Mayfly bioindicator thresholds for several anthropogenic disturbances in neotropical savanna streams. Ecol Indicat 74: 276–284. [CrossRef] [Google Scholar]
  • Garcia-Giron J, Bini LM, Heino J. 2023. Shortfalls in our understanding of the causes and consequences of functional and phylogenetic variation of freshwater communities across continents. Biol Conserv. doi.org/10.1016/j.biocon.2023.110082 [Google Scholar]
  • Gerth WJ, Li J, Giannico GR. Agricultural land use and macroinvertebrate assemblages in lowland temporary streams of the Willamette Valley, Oregon, USA. Agric Ecosys Environ 236:xxxx 154–165. [Google Scholar]
  • Hering D, Moog O, Sandin L, Verdonschot PFM. 2004. Overview and application of the AQEM assessment system. Hydrobiology 516: 1–30. [CrossRef] [Google Scholar]
  • Herlihy AT, Paulsen SG, Van Sickle J, Stoddard JL, Hawkins CP, Yuan LL. 2008. Striving for consistency in a national assessment: the challenges of applying a reference-condition approach at a continental scale. J N Amer Benthol Soc. 27: 860–877. [Google Scholar]
  • Herlihy AT, Sifneos JC, Hughes RM, Peck DV, Mitchell RM, 2019. Lotic fish assemblage clusters across the conterminous USA and their associations with local- and catchment-scale landscape variables. Am Fish Soc Symp 90: 385–408. [Google Scholar]
  • Herlihy AT, Sifneos JC, Hughes RM, Peck DV, Mitchell RM. 2020. Relation of lotic fish and benthic macroinvertebrate condition indices to environmental factors across the conterminous USA. Ecol Indicat 112: 105958. [CrossRef] [Google Scholar]
  • Hill RA, Fox EW, Leibowitz SG, Olsen AR, Thornbrugh DJ, Weber MH. 2017. Predictive mapping of the biotic condition of conterminous U.S. rivers and streams. Ecol Appl 27: 2397–2415. [CrossRef] [PubMed] [Google Scholar]
  • Hocutt CH, Wiley EO. 1986. The zoogeography of North American freshwater fishes. New York: Wiley. [Google Scholar]
  • Hughes RM, Peck DV. 2008. Acquiring data for large aquatic resource surveys: the art of compromise among science, logistics, and reality. J N Amer Benthol Soc 27: 837–859. [CrossRef] [Google Scholar]
  • Hughes RM, Vadas RL. 2021. Agricultural effects on streams and rivers: a western USA focus. Water 13. doi.org/10.3390/w13141901 [Google Scholar]
  • Hughes RM, Kaufmann PR, Herlihy A, Kincaid TM, Reynolds L, Larsen DP. 1988. A process for developing and evaluating indices of fish assemblage integrity. Can J Fish Aquat Sci 55: 1618–1631. [CrossRef] [Google Scholar]
  • Hughes RM, Rinne JN, Calamusso B. 2005. Historical changes in large river fish assemblages of the Americas: a synthesis. Am Fish Soc Symp 45: 603–612. [Google Scholar]
  • Hughes RM, Herlihy AT, Gerth WJ, Pan Y. 2012. Estimating vertebrate, benthic macroinvertebrate and diatom taxa richness in raftable Pacific Northwest rivers for bioassessment purposes. Environ Monitor Assess 184: 3185–3198. [CrossRef] [PubMed] [Google Scholar]
  • Hughes RM, Infante DM, Wang L, Chen K, Terra BF, editors. 2019. Advances in Understanding Landscape Influences on Freshwater Habitats and Biological Assemblages. American Fisheries Society, Bethesda. [Google Scholar]
  • Hughes RM, Herlihy AT, Peck DV. 2021. Sampling effort for estimating fish species richness in western USA river sites. Limnologica 87. doi:10.1016/j.limno.2021.125859 [CrossRef] [PubMed] [Google Scholar]
  • Hughes RM, Herlihy AT, Gerth WJ, Pan Y. 2012. Estimating vertebrate, benthic macroinvertebrate and diatom taxa richness in raftable Pacific Northwest rivers for bioassessment purposes. Environ Monitor Assess 184: 3185–3198. [CrossRef] [PubMed] [Google Scholar]
  • Jenkins C, Van Houtan K, Pimm S, Sexton J. 2015. US protected lands mismatch biodiversity priorities. Proc Nat Acad Sci 112: 5081–5086. [CrossRef] [PubMed] [Google Scholar]
  • Kanno Y, Vokoun JC, Dauwalter D, Hughes RM, Herlihy AT., Maret TR, Patton TM. 2009. Influence of rare species on electrofishing distance-species richness relationships at stream sites. Trans Am Fish Soc 138: 1240–1251. [CrossRef] [Google Scholar]
  • Kaufmann PR, Hughes RM, Whittier TR, Bryce SA, Paulsen SG. 2014. Relevance of lake physical habitat assessment indices to fish and riparian birds. Lake Reserv Manage 30: 177–191. [CrossRef] [Google Scholar]
  • Kaufmann PR, Hughes RM, Paulsen SG, Peck DV, Seeliger C, Weber M, Mitchell RM. 2022a. Physical habitat in conterminous US streams and rivers, part 1: Geoclimatic controls and anthropogenic alteration. Ecol Indicat. doi: 10.1016/j.ecolind.2022.109046 [Google Scholar]
  • Kaufmann PR, Hughes RM, Paulsen SG, Peck DV, Seeliger C, Kincaid T, Mitchell RM. 2022b. Physical habitat in conterminous US streams and rivers, part 2: Quantitative assessment of condition. Ecol Indicat. doi: 10.1016/j.ecolind.2022.109047 [Google Scholar]
  • Leal CG, Barlow J, Gardner T, Hughes RM, Leitão RP, MacNally R, Kaufmann P, Ferraz SFB, Zuanon J, de Paula FR, Ferreira J, Thomson JR, Lennox GD, Dary EP, Röpke CP, Pompeu PS. 2018. Is environmental legislation conserving tropical stream faunas? A large-scale assessment of local, riparian and catchment-scale influences on Amazonian fish. J Appl Ecol 55: 1312–1326. [CrossRef] [PubMed] [Google Scholar]
  • Leal CG, Lennox GD, Ferraz SFB, Ferreira J, Gardner TA, Thomson JR, Berenguer E, Lees AC, Hughes RM, MacNally R, Aragão LEOC, de Brito JG, Castello L, Garrett RD, Hamada N, Juen L, Leitão RP, Louzada J, Morello TF, Moura NG, Nessimian JL, Oliveira-Junior JMB, de Oliveira VHF, de Oliveira VC, Parry L, Pompeu PS, Solar RRC, Zuanon J, Barlow J. 2020. Integrated terrestrial-freshwater planning doubles tropical aquatic species conservation. Science 370: 117–121. [CrossRef] [PubMed] [Google Scholar]
  • Lenat DR, Resh VH. 2001. Taxonomy and stream ecology — the benefits of genus- and species-level identifications. J N Amer Benthol Soc 20: 287–298. [CrossRef] [Google Scholar]
  • Leroy B, Dias MS, Giraud E, Hugueny B, Jézéquel C, Leprieur F, Oberdorff T, Tedesco PA. 2019. Global biogeographical regions of freshwater fish species. J Biogeogr 46: 2407–2419. [CrossRef] [Google Scholar]
  • Li J, Herlihy A, Gerth W, Kaufmann P, Gregory S, Urquhart S, Larsen DP. 2001. Variability in stream macroinvertebrates at multiple spatial scales. Freshwat Biol 46: 87–97. [CrossRef] [Google Scholar]
  • Li L, Liu L, Hughes RM, Cao Y, Wang X. 2014. Towards a protocol for stream macroinvertebrate sampling in China. Environ Monitor Assess 186: 469–479. [CrossRef] [PubMed] [Google Scholar]
  • Liaw A, Wiener M. 2002. Classification and regression by randomForest. R News 2: 18–22. [Google Scholar]
  • Ligeiro R, Melo A, Callisto M. 2010. Spatial scale and the diversity of macroinvertebrates in a Neotropical catchment. Freshw Biol 55: 424–435. [CrossRef] [Google Scholar]
  • Ligeiro R, Hughes RM, Kaufmann PR, Macedo DR, Firmiano KR, Ferreira WR, Oliveira D, Melo AS, Callisto M. 2013a. Defining quantitative stream disturbance gradients and the additive role of habitat variation to explain macroinvertebrate taxa richness. Ecol Indicat 25: 45–57. [CrossRef] [Google Scholar]
  • Ligeiro R, Ferreira W, Hughes RM, Callisto M. 2013b. One caveat when using fixed-area subsampling methods to estimate macroinvertebrate richness: a case study with Neotropical stream data. Environ Monitor Assess 185: 4077–4085. [CrossRef] [PubMed] [Google Scholar]
  • Lomnicky GA, Whittier TR, Hughes RM, Peck DV. 2007. Distribution of nonnative aquatic vertebrates in western U.S. streams and rivers. N Amer J Fish Manage. 27: 1082–1093. [Google Scholar]
  • Macedo DR, Hughes RM, Ligeiro R, Ferreira WR, Castro M, Junqueira NT, Silva DRO, Firmiano KR, Kaufmann PR, Pompeu PS, Callisto M. 2014. The relative influence of multiple spatial scale environmental predictors on fish and macroinvertebrate assemblage richness in cerrado ecoregion streams, Brazil. Landsc Ecol 29: 1001–1016. [CrossRef] [Google Scholar]
  • Maloney KO, Manguia P, Mitchell RM. 2011. Anthropogenic disturbance and landscape patterns affect diversity patterns of aquatic benthic macroinvertebrates. J N Amer Benthol Soc 30: 284–295. [CrossRef] [Google Scholar]
  • Martins RT, Brito J, Dias-Silva K, Leal CG, Leitao RP, Oliveira VC, de Oliveira-Junior JMB, Ferraz SFB, de Paula FR, Roque FO, Hamada N, Juen L, Nessimian JL, Pompeu PS, Hughes RM. 2021. Low forest-loss thresholds threaten Amazonian fish and macroinvertebrate assemblage integrity. Ecol Indicat. doi: 10.1016/j.ecolind.2021.107773 [Google Scholar]
  • Mason GT, Arndt RE. 1996. Mineral resources data system (MRDS). USGS Data Series 20. doi: 10.3133/ds20 [Google Scholar]
  • McCormick FH, Hughes RM, Kaufmann PR, Herlihy AT, Peck DV. 2001. Development of an index of biotic integrity for the Mid-Atlantic Highlands Region. Trans Amer Fish Soc. 130: 857–877. [Google Scholar]
  • McGarvey DJ, Hughes RM. 2008. Longitudinal zonation of Pacific Northwest (U.S.A.) fish assemblages and the species-discharge relationship. Copeia 2008: 311–321. [CrossRef] [Google Scholar]
  • McGarvey DJ, Terra BF. 2016. Using river discharge to model and deconstruct the latitudinal diversity gradient for fishes of the Western Hemisphere. J Biogeogr 43: 1436–1449. [Google Scholar]
  • McGarvey DJ, Ward M. 2008. Scale dependence in the species-discharge relationship for fishes of the southeastern U.S.A. Freshwat Biol 53: 2206–2219. [Google Scholar]
  • Mebane CA, Maret TR, Hughes RM. 2003. An index of biological integrity (IBI) for Pacific Northwest rivers. Trans Amer Fish Soc 132: 239–261. [CrossRef] [Google Scholar]
  • Meyer J, Strayer DL, Wallace JB, Eggert SL, Helfman GS, Leonard NE 2007. The contribution of headwater streams to biodiversity in river networks. J Am Wat Resour Assoc 43: 86–103. [CrossRef] [Google Scholar]
  • Morse JC, Stark BP, McCafferty WP, Tennessen KJ. 1997. Southern Appalachian and other southeastern streams at risk: implications for mayflies, dragonflies, damselflies, stoneflies and caddisflies. In Benz GW, Collins DE eds. Aquatic Fauna in Peril: the Southeastern Perspective, Southeastern Aquatic Re search Institute, Special Publication 1. Lenz Design and Communications, Decatur, Georgia, 17–42. [Google Scholar]
  • Mostafavi H, Kordajazi Z, Valavi R, Shafizadeh-Moghadam H, Kambouzia J, Infante DM. 2019. Ensemble modeling of sensitive stream fish species distributions in Iran: expanding knowledge to aid species conservation. Amer Fish Soc Symp 90: 441–462. [Google Scholar]
  • Nelson JS, Crossman EJ, Espinosa-Pérez H, Findley LT, Gilbert CR, Lea RN, Williams JD. 2004. Common and Scientific Names of Fishes from the United States Canada and Mexico. Bethesda: American Fisheries Society. [Google Scholar]
  • Oberdorff T, Guégan J-F., Hugueny B. 1995. Global scale patterns of fish species richness in rivers. Ecography 18: 345–352. [CrossRef] [Google Scholar]
  • Oliveira-Junior JMB, Shimano Y, Gardner TA,Hughes RM., de Marco Junior P, Juen L. 2015. Neotropical dragonflies (Insecta: Odonata) as indicators of ecological condition of small streams in the Eastern Amazon. Austral Ecol 40: 733–744. [CrossRef] [Google Scholar]
  • Olsen AR, Peck DV. 2008. Survey design and extent estimates for the Wadeable Streams Assessment. J N Amer Benthol Soc 27: 822–836. [CrossRef] [Google Scholar]
  • Omernik JM. 1987. Ecoregions of the conterminous United States. Ann Assoc Amer Geogr 77: 118–125. [CrossRef] [Google Scholar]
  • Page LM, Espinosa-Pérez H, Findley LT, Gilbert CR, Lea RN, Mandrak NE, Mayden RL, Nelson JS. 2013. Common and Scientific Names of Fishes from the United States Canada and Mexico. American Fisheries Society, Bethesda. [Google Scholar]
  • Pompeu PS, Carvalho DR, Leal CG, Leitão RP, Alves CBM, Braga DF, Castro MA, Junqueira NT, Zuanon J, Hughes RM. 2021. Sampling efforts for determining fish species richness in megadiverse tropical regions. Environ Biol Fish. doi: 10.1007/s10641-021-01184-7 [Google Scholar]
  • R Core Team, 2020. R: a Language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. https://www.R-project.org/ [Google Scholar]
  • Rumschlag, SL, Mahon MB, Jones DK, Battaglin W, Behrens J, Bernhardt ES, Bradley P, Brown E, de Laender F, Hill R, Kunz S, Lee S, Rosi E, Schafer R, Schmidt TS, Simonin M, Smalling K, Voss K, Rohr JR. 2023. Density declines, richness increases, and composition shifts in stream macroinvertebrates. Sci Adv 9. doi:10.1126/sciadv.adf4896 [CrossRef] [PubMed] [Google Scholar]
  • Shurin JB, Havel JE, Leibold MA, Pinel-Alloul B. 2000. Local and regional zooplankton species richness: a scale-independent test for saturation. Ecology 81: 3062–3073. [CrossRef] [Google Scholar]
  • Silva DRO, Ligeiro R, Hughes RM, Callisto M. 2016. The role of physical habitat and sampling effort on estimates of benthic macroinvertebrate taxonomic richness at basin and site scales Environ Monitor Assess 188: 340. [CrossRef] [PubMed] [Google Scholar]
  • Silva DRO, Herlihy AT, Hughes RM, Macedo, DR, Callisto M. 2018. Assessing the extent and relative risk of aquatic stressors on stream macroinvertebrate assemblages in the neotropical savanna. Sci Tot Environ 633: 179–188. [CrossRef] [Google Scholar]
  • Smith KL, Jones ML 2005. Watershed-level sampling effort requirements for determining riverine fish species composition. Can J Fish Aquat Sci 62: 1580–1588. [CrossRef] [Google Scholar]
  • Smith GR, Badgley C, Eiting TP, Larson PS. 2010. Species diversity gradients in relation to geological history in North American freshwater fishes. Evolut Ecol Res 12: 693–726. [Google Scholar]
  • Stevens DL, Olsen AR. 2004. Spatially balanced sampling of natural resources. J Am Stat Assoc 99: 262–278. [CrossRef] [Google Scholar]
  • Su G, Logez M, Xu J, Tao S, Villéger S, Brosse S. 2021. Human impacts on global freshwater fish biodiversity. Science 371: 835–838. [Google Scholar]
  • Tedesco PA, Leprieur F, Hugueny B, Brosse S, Durr HH, Beauchard O, Busson F, Oberdorff T. 2012. Patterns and processes of global riverine fish endemism. Global Ecol Biogeogr 21: 977–987. [CrossRef] [Google Scholar]
  • Terra BDF, Hughes RM, Araujo FG. 2013. Sampling sufficiency for fish assemblage surveys of Atlantic Forest streams, southeastern Brazil. Fisheries 38: 150–158. [CrossRef] [Google Scholar]
  • The White House. 2021. Executive order on tackling the climate crisis at home and abroad. Section 216. https://www.whitehouse.gov/briefing-room/presidential-actions/2021/01/27/executive-order-on-tackling-the-climate-crisis-at-home-and-abroad/ [Google Scholar]
  • Tonn WM. 1990. Climate change and fish communities: a conceptual framework. Trans Am Fish Soc 119: 337–352. [CrossRef] [Google Scholar]
  • Twardochleb L, Hiltner E, Pyne M, Sarnetske P. 2021 Freshwater insects CONUS: a database of freshwater insect occurrences and traits for the contiguous United States. Global Ecol Biogeogr 30: 826–841. [CrossRef] [PubMed] [Google Scholar]
  • USEPA (United States Environmental Protection Agency). 2009. National Rivers and Streams Assessment: field operations manual. EPA 841/B-04/004, Office of Water and Office of Environmental Information, U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2012. National Rivers and Streams Assessment 2013‐2014: Laboratory Operations Manual. EPA‐841‐B‐12‐010. Office of Water. U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2013a. National Rivers and Streams Assessment 2013/14: field operations manual − wadeable. EPA 841/B-12/009b, Office of Water and Office of Environmental Information, U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2013b. National Rivers and Streams Assessment 2013/14: field operations manual-non-wadeable. EPA 841/B-12/009a, Office of Water and Office of Environmental Information, U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2016a. National Rivers and Streams Assessment 2008–2009 technical report. EPA 841/R-16/008, Office of Water and Office of Research and Development. U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2016b. National Rivers and Streams Assessment 2008-2009: a collaborative survey. EPA/841/R-16/007. Office of Water and Office of Research and Development, U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2020. National Rivers and Streams Assessment 2013-2014: a collaborative survey. EPA841-R- 19–001. Office of Water and Office of Research and Development, U.S. Environmental Protection Agency, Washington. [Google Scholar]
  • USEPA (United States Environmental Protection Agency), 2021. Level III and IV ecoregions of the continental United States. https://www.epa.gov/eco-research/level-iii-and-iv-ecoregions-continental-united-states [Google Scholar]
  • USGS (United States Geological Survey), 2013a. National hydrography geodatabase: the national map viewer. https://viewer.nationalmap.gov/viewer/nhd.html?p=nhd [Google Scholar]
  • USGS (U.S. Geological Survey, U.S. Department of Agriculture, & Natural Resources Conservation Service), 2013b. Federal standards and procedures for the national watershed boundary dataset (WBD). TM 11-A3 Section A: Federal Standards in Book 11 Collection and Delineation of Spatial Data. https://www.usgs.gov/core-science-systems/ngp/national-hydrography/watershed-boundary-dataset [Google Scholar]
  • Vadas RLJr, Hughes RM, Bello-Gonzales O, Callisto M, Carvalho D, Chen K, Davies PE, Ferreira MT, Fierro P, Harding JS, Kleynhans CJ, Macedo DR, Mercado-Silva N, Moya N, Nichols SJ, Pompeu PS, Ruaro R, Stevenson RJ, Terra BF, Thirion C, Ticiani D, Yoder CO. 2022. Assemblage-based biomonitoring of freshwater ecosystem health via multimetric indices: a critical review and suggestions for improving their applicability. Wat Biol Secur. doi.org/10.1016/j.watbs.2022.100054 [Google Scholar]
  • Valdez JW, Callaghan CT, Junker J, Purvis A, Hill SLL, Pereira HM. 2023. The undetectability of global biodiversity trends using local species richness. Ecography. doi: 10.1111/ecog.06604 [Google Scholar]
  • Vinson MR, Hawkins CP. 2003. Broad-scale geographical patterns in local stream insect genera richness. Ecography 26: 751–767. [CrossRef] [Google Scholar]
  • Whittaker RH. 1960. Vegetation of the Siskiyou Mountains, Oregon and California. Ecol Monogr 30: 279–338. [Google Scholar]
  • Whittier TR, Paulsen SG, Larsen DP, Peterson SA, Herlihy AT, Kaufmann PR. 2002. Indicators of ecological stress and their extent in the population of northeastern lakes: a regional assessment. BioScience 52: 235–247. [CrossRef] [Google Scholar]
  • Yang L, Jin S, Danielson P, Homer C, Gass L, Bender SM, Case A, Costello C, Dewitz J, Fry J, Funk M, Granneman B, Liknes GC, Rigge M, Xian G. 2018. A new generation of the United States national land cover database: requirements, research priorities, design, and implementation strategies. J Photogram Remote Sens 146: 108–123. [CrossRef] [Google Scholar]

Cite this article as: Hughes RM, Herlihy AT, Comeleo R, Peck DV, Mitchell RM, Paulsen SG. 2023. Patterns in and predictors of stream and river macroinvertebrate genera and fish species richness across the conterminous USA Knowl. Manag. Aquat. Ecosyst., 424, 19

Supplementary Material

Figure S1: Composite macroinvertebrate genus richness in coarse- and fine-grain ecoregions.

Figure S2: Composite macroinvertebrate genus richness in coarse- and fine-grain hydrologic units.

Figure S3: Composite fish species richness in coarse- and fine-grain ecoregions.

Figure S4: Composite fish species richness in coarse- and fine-grain hydrologic units.

Access here

All Tables

Table 1

Codes, names, sample sizes and composite richness for coarse-grain NRSA study units.

In the text
Table 2

Pearson correlation, multiple regression, and random forest results for relating residual regional macroinvertebrate genera richness to regional environmental data composited at the fine-grain ecoregion extent. Variables sorted in order of decreasing correlation strength.

In the text
Table 3

Pearson correlation, multiple regression, and random forest results for relating residual regional fish species richness to regional environmental data composited at the fine-grain ecoregion extent. Variables sorted in order of decreasing correlation strength.

In the text

All Figures

thumbnail Fig. 1

Locations of the NRSA sample sites and the nine aggregate ecoregions used for coarse-grain regional richness analysis.
In the text
thumbnail Fig. 2

Fish species and macroinvertebrate genera richness at the individual site and various regional scales. Fine-grained regional richness is based on Omernik Level-III ecoregions and level-4 Hydrologic Units (HUCs). Coarse-grained regional richness is based on nine aggregated Omernik ecoregions and level-2 HUCs. Boxes are the interquartile ranges, lines in the boxes are medians, and the whiskers are 5th/95th percentiles.
In the text
thumbnail Fig. 3

Composite fish and macroinvertebrate regional richness versus number of sample sites in the composite for all spatial extents. The red line is fitted from nonlinear regression. Residual regional richness was calculated as the deviation from the fitted line.
In the text
thumbnail Fig. 4

Coarse-grain residual regional richness of fish species and macroinvertebrate genera among ecoregions. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.
In the text
thumbnail Fig. 5

Fine-grain residual regional richness of fish species and macroinvertebrate genera among fine-grain ecoregions. Ecoregions with <10 sample sites were deemed to have insufficient data. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue-green colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.
In the text
thumbnail Fig. 6

Coarse-grain residual regional richness of fish species and macroinvertebrate genera among hydrological units. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue-green colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.
In the text
thumbnail Fig. 7

Fine-grained residual regional richness of fish species and macroinvertebrate genera among hydrologic units. Units with <10 sample sites were deemed to have insufficient data. Red-orange-yellow colors denote negative residual richness (below the predicted richness line for their number of samples). Blue-green colors denote positive residual richness. Positive residuals indicate greater than expected regional richness given the number of sampled sites; negative residuals indicate lower than expected regional richness given the number of sampled sites.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.