USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Distribution of Fishes in the Red River of the North Basin on Multivariate Environmental Gradients

The Species-environment Relationship


Species-area

Species richness of streams in the Red River basin ranged from a low of 13 for the Mustinka River to a high of 65 for the Otter Tail River (Table 7). Watershed area (mi2) ranged from 321 for the Rabbit River to 5690 for the Sheyenne River, stream length (mi) ranged from 21 for the Rabbit River to 608 for the Red Lake River, and average annual discharge (cfs) near the mouth ranged from 2 for the Elm River to 1115 for the Red Lake River.

There was an increase in species richness of streams with an increase in watershed area, stream length, and average annual discharge. Fifty percent of the total variability in species richness was accounted for by regression with watershed area (Figure 13), while 57% of variation was accounted for by regression with stream length, and 60% was accounted for by regression with average annual discharge. In all three analyses, the dependence of species richness on the independent axis was significant (ANOVA tests, P=0.0001).

GIF - Relationship among species richness and various river characteristics.(pic)

Figure 13. Relationships among fish species richness and watershed drainage are (mi2), total stream length (mi), and average annual discharge (cfs) of 26 streams in the Red River basin. Lines represent the simple regression models provided.

Environmental characterization of reaches

Water quality parameters of 46 stream reaches are the mean value for each during the open-water period (April-November). Specific conductivity (ÁS/cm) ranged from 357 for the Otter Tail River in the NLF to 3802 for the Turtle River in the RRV (Mean=999, SE=89) (Table A2). Hardness (mg/L CaCO3) ranged from 186 for the Otter Tail River in the NLF to 729 for the Rabbit River in the RRV (Mean=384, SE=21). Alkalinity (mg/L CaCO3) ranged from 192 for the Tongue River in the RRV to 400 for the Wild Rice River, North Dakota, in the RRV (Mean=274, SE=7). Dissolved sulfate (mg/L SO4) ranged from 4 for the Otter Tail River in the NLF and the Red Lake River in the NMW to 626 for the Rabbit River in the RRV (Mean=198, SE=25). Dissolved chloride (mg/L Cl) ranged from 2 for the Otter Tail River in the NLF, the Red Lake River in the NMW, and the Wild Rice River, Minnesota, in the NLF to 741 for the Turtle River in the RRV (Mean=43, SE=16). Residue (mg/L) ranged from 206 for the Otter Tail River in the NLF to 2406 for the Turtle River in the RRV (Mean=647, SE=59). Nitrogen (mg/L N) ranged from 0.03 for the Otter Tail River in the NLF to 2.57 for the Bois de Sioux River (Mean=0.76, SE=0.09). Phosphorus (mg/L P) ranged from 0.03 for the Red Lake River in the NMW to 1.03 for the Middle River in the RRV (Mean=0.35, SE=0.04).

Average annual discharge (cfs) ranged from 3 for the Maple River in the NGP to 1758 for the Red River (Mean=190, SE=47) (Table A2). Average high discharge in May (cfs) ranged from 14 for the Elm River in the RRV to 5905 for the Red River (Mean=697, SE=153). Average low discharge in May (cfs) ranged from 1 for the Elm River in the RRV, the Maple River in the NGP, and the Rabbit River in the RRV to 1737 for the Red River (Mean=207, SE=50). Coefficient of variation of mean monthly discharge ranged from 15 for the Red Lake River in the NMW to 213 for the Goose River in the TRA (Mean=127, SE=7).

Species characterization of reaches

Occurrence of stream fish species varied between 46 stream reaches in the Red River basin. Species richness ranged from 11 for the Mustinka River in the RRV to 55 for the Otter Tail River in the NCH. The rank (frequency of occurrence) of fish species reported from each of the 46 reaches also varied among reaches and ranged from zero (not present at any site within the reach) to five (present at 80.5-100% of the sites within the reach) (Table A3).

Canonical correspondence analysis

The first two axes of the CCA ordination (CC1 and CC2) explained the species-environment relationship well and accounted for 49% of the variation in the weighted averages of the 79 fish species with respect to 12 environmental variables. The "weighted average" indicates the "center" of a species distribution along an environmental variable (ter Braak 1986). For this study, the weighted average of each species was the average value of each environmental variable at reaches where the species occurs; the weighting of each reach was proportional to the species frequency of occurrence. Arrows indicating the relative importance and direction of the 12 environmental variables were placed on the axes by CANOCO (Figure 14). Each arrow points in the direction of maximum variation in value of the corresponding variable, and each may be extended in both directions from the origin of the plot (origin=grand mean of each environmental variable). The length of arrows represents the degree to which each is correlated with CCA axes. Therefore, important environmental variables (in terms of predicting fish assemblage composition) have longer arrows than less important ones.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 14. Canonical correspondence analysis ordination diagram showing environmental variables (arrows) most strongly correlated with axes CC1 and CC2. Variables include specific conductivity (CON), hardness (HARD), alkalinity (ALK), dissolved sulfate (SUL), dissolved chloride (CHL), residue (RES), dissolved nitrogen (NIT), phosphorus (PHOS), average annual discharge (MEANFLO), high discharge during May (HIFLO), low discharge during May (LOWFLO), and the coefficient of variation of mean monthly discharge (CV). In terms of predicting fish assemblage compostion, important environmental variables have longer arrows than less important ones.

The environmental variables most important in explaining fish distributions were determined by examining correlations with axes. Correlations of 12 environmental variables and the first two canonical correspondence axes are provided in Table 8. The first axis was most important, explaining 25% of variance, and was most strongly correlated with residue, specific conductivity, and average annual discharge. The second axis explained 24% of variance and was most strongly correlated with the CV of mean monthly discharge, average low discharge during May, and hardness. The third and fourth axes explained only 15% and 8% of variance, respectively, and were not considered further.

Seventy-nine fish species were positioned in relation to 12 environmental variables (Figures 15-21). The arrows corresponding to these variables from Figure 14 can be superimposed onto ordinations of species, and the intersection of a perpendicular line from each species point to an arrow represents the center of the species distribution along that environmental gradient. The weighted average for a given species is higher than average if the intersection of a species lies on the same side of the origin as the head of the arrow of interest does and is lower than average if the origin of the plot lies between the intersection and the head of the arrow. An example is provided on Figure 15 for species of the family Catostomidae along the environmental gradient, residue. It appears that none of the Catostomids preferred waters with high residue, and only the bigmouth buffalo, quillback carpsucker, and white sucker preferred waters with residue levels above the average value for the Red River basin. The shorthead redhorse occurred primarily in waters with residue levels near the basin average, and other species were found mostly in waters with lower than average residue levels. It appears that the northern hogsucker has been taken in only waters with very low residue levels. Superimposing other important environmental variables on the species plot revealed further trends. Most catostomids have preferred waters with a specific conductivity, CV of mean monthly discharge, and hardness near or below basin averages and waters with average annual discharge and average low discharge in May above basin averages.

GIF -- CCA Biplot for Catostomid Species.(pic)

Figure 15. Example of CCA biplot showing the positioning of catostomid species in streams of the Red River basin along the enviromnmental gradient, residue. The intersection of perpendicular lines with the arrow indicated the approximate ranking of weighted averages of taxa with respect to residue. This method was used to analyze all stream fishes with respect to 12 environmental variables.

In similar fashion, all other environmental variables (arrows) can be placed on the species ordinations (Figures 15-21) to create CCA biplots. The effects of multiple environmental variables on species can be assessed simultaneously. This analysis will utilize the six variables most strongly correlated with CC1 and CC2 in discussions which follow.

The sand shiner, bigmouth shiner, and rosyface shiner have preferred waters with values of the 12 environmental variables near the basin averages (Figure 16). The river shiner has occurred primarily in waters with higher than average residue, specific conductivity, hardness, and CV of mean monthly discharge. All other shiners were found in waters with values of these environmental variables far below the basin average. The shiners were separated only slightly by average annual or May low discharges, as all have preferred waters at or above the basin averages for these variables. The flathead chub, common carp, and silver chub have preferred waters with higher than average residue, specific conductivity, and hardness (Figure 17). The largescale stoneroller, fathead minnow, and creek chub have occurred primarily in waters with a higher than average CV of mean monthly discharge and lower than average annual and May low discharges. The northern redbelly dace, finescale dace, hornyhead chub, golden shiner, and central stoneroller have preferred waters with lower than average residue, specific conductivity, and hardness, with the central stoneroller also preferring waters with very low CV of mean monthly discharge and high average annual and May low discharges. The longnose dace and brassy minnow have preferred waters with values of the environmental variables at the basin averages, as they are found near the origin of the plot.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 16. Canonical correspondence analysis ordination of cyprinid species (Notropis) in streams of the Red River basin. Species points can be superimposed on an ordiantion of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 17. Canonical correspondence analysis ordination of cyprinid species in streams of the Red River basin. Species points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

The channel catfish and black bullhead have preferred waters with only slightly higher than average residue, specific conductivity, and hardness (Figure 18). The black bullhead was found in waters higher in CV of mean monthly discharge, while the channel catfish was found in waters with higher than average annual and May low discharges. The yellow bullhead has preferred waters with much lower than average residue, specific conductivity, hardness, and CV of mean monthly discharge and much higher than average annual and May low discharges. The stonecat, tadpole madtom, and brown bullhead have preferred waters with values of the environmental variables at the basin averages, as they are found near the origin of the plot.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 18. Canonical correspondence analysis ordination of ictalurid species in streams of the Red River basin. Species points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

The white crappie has preferred waters with much higher than average residue, specific conductivity, hardness, and CV of mean monthly discharge, while all other centrarchids occurred in waters at or below basin averages for these variables (Figure 19). Also, most centrarchids have preferred waters with higher than average annual and May low discharges. The pumpkinseed, bluegill, largemouth bass, and green sunfish have preferred waters with lower than average residue, specific conductivity, hardness, and CV of mean monthly discharge.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 19. Canonical correspondence analysis ordination of centrarchid species in streams of the Red River basin. Species points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

The sauger has preferred waters with higher than average residue, specific conductivity, and hardness, while all other percids occurred in waters at or below basin averages for these variables (Figure 20). The logperch, and particularly the least darter and rainbow darter, have preferred waters with much lower than average residue, specific conductivity, hardness, and CV of mean monthly discharge and much higher than average annual and May low discharges. The johnny darter, Iowa darter, and blackside darter have preferred waters with values of the environmental variables at the basin averages, as they are found near the origin of the plot.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 20. Canonical correspondence analysis ordination of percid species in streams of the Red River basin. Species points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

The goldeye, white bass, mooneye, silver lamprey, and freshwater drum have preferred waters with higher than average residue, specific conductivity, hardness, average annual discharge, and May low discharge (Figure 21). The rainbow trout, brown trout, and brook stickleback have preferred waters with higher than average CV of mean monthly discharge and lower than average annual and May low discharge. The central mudminnow, banded killifish, brook trout, mottled sculpin, and bowfin have preferred waters with lower than average residue, specific conductivity, and hardness. The bowfin prefers waters with considerably higher than average annual and May low discharge.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 21. Canonical correspondence analysis ordination of several fish species in streams of the Red River basin. Species points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

Forty-six stream reaches were positioned in relation to the 12 environmental variables (Figures 22-25). Scores for each reach on CC1 and CC2 were the weighted mean species scores. Reaches located within the NCH ecoregion all plotted in the lower right quadrant of the CCA diagram (Figure 22). Waters at these reaches had lower than average residue, specific conductivity, and hardness. Reaches in the Pelican and Otter Tail Rivers had much higher than average annual and May low discharges and lower than average CV of mean monthly discharge.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 22. Canonical correspondence analysis ordination of reaches in the North Central Hardwoods ecoregion of the Red River basin. Points represent the weighted mean species scores for each reach. Reach points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

Reaches located within the NGP ecoregion plotted in the upper two quadrants of the CCA diagram (Figure 23). Waters of these reaches had slightly higher than average residue, specific conductivity, and hardness. Reaches in the Maple and Forest Rivers had much higher than average CV of mean monthly discharge and much lower than average annual and May low discharges.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 23. Canonical correspondence analysis ordination of reaches in the Northern Glaciated Plains ecoregion of the Red River basin. Points represent the weighted mean species scores for each reach. Reach points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

Reaches located within the NLF and NMW ecoregions plotted in the lower two quadrants of the CCA diagram (Figure 24). Waters of these reaches had lower than average residue, specific conductivity, and hardness. Reaches in the Clearwater, Roseau, and Wild Rice Rivers had waters with near average CV of mean monthly discharge and average annual and May low flows. Reaches in the Red Lake and Otter Tail Rivers had waters with exceptionally high average annual and May low discharges and low CV of mean monthly discharge.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 24. Canonical correspondence analysis ordination of reaches in the Northern Lakes and Forests and Northern Minnesota Wetlands ecoregions of the Red River basin. Points represent the weighted mean species scores for each reach. Reach points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.

Reaches located within the RRV ecoregion plotted in the upper two and the lower left quadrants (Figure 25). Reaches in the Red, Goose, and Rabbit Rivers had waters with higher than average residue, specific conductivity, and hardness. Reaches in the Rabbit and Elm Rivers had waters with exceptionally high CV of mean monthly discharge and low average annual and May low discharges. Reaches in the Otter Tail and Red Lake Rivers had waters with exceptionally high average annual and May low discharges.

GIF - Canonical correspondence analysis ordination diagram.(pic)

Figure 25. Canonical correspondence analysis ordination of reaches in the Red River Valley ecoregion of the Red River basin. Points represent the weighted mean species scores for each reach. Reach points can be superimposed on an ordination of 12 environmental arrows (Figure 14) to interpret variation along environmental gradients.


Previous Section -- Association Analysis
Return to Contents
Next Section -- Discussion

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo USA.gov logo U.S. Department of the Interior | U.S. Geological Survey
URL: http://www.npwrc.usgs.gov/resource/fish/norbasin/specarea.htm
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 04:37:27 EST
Sioux Falls, SD [sdww55]