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

Gradient Analyses

To determine the effects of water quality and flow on the distribution of fish species, streams were divided into "reaches" by ecoregion classification (Omernik and Gallant 1988) and by an ecoregion transition zone (TRA). Forty-six reaches were produced from 27 streams (Table 1), and each reach was characterized both by environmental factors and fish species occurrences.

Environmental characterization of reaches

Information regarding water quality and flow was obtained for each reach from USGS, NDDH, and MPCA monitoring sites for 1962-1994 through a STORET (EPA national water quality data bank) data retrieval and review of published documents (Table 1). The number of years for which data were available for reaches varied greatly. Therefore, the means calculated as described below are from as many as 30 years of data collection for some reaches, or as few as 3 years of data collection for others. Water quality variables were specific conductivity (S/cm), hardness (mg/L as CaCO3), alkalinity (mg/L as CaCO3), dissolved sulfate (mg/L as SO4), dissolved chloride (mg/L as Cl), residue (mg/L at 180C), dissolved nitrogen (NO2 and NO3, mg/L as N), and dissolved phosphorus (mg/L as P), all of which were high values during the open water season (April-November of each year). Flow variables were average annual discharge (cfs), high discharge during May (cfs), low discharge during May (cfs), and the CV of mean monthly discharge (unitless). The latter was a measure of reach flow variability throughout the year. Means and CV were calculated by use of Quattro Pro.

Species characterization of reaches

To coincide with environmental data, occurrences of 79 fish species at sites within 46 stream reaches were recorded for the period 1962-1994. The number of sites used to characterize each reach varied (Table 1). In the original species data base for the Red River basin, only presence or absence information was recorded. Therefore, to obtain a measure of frequency of occurrence of each species in each reach, species were ranked by dividing the number of sites at which each occurred by the total number of sites sampled. A rank of zero was given to species that occurred at 0% of sites within a reach; and ranks of one, two, three, four, and five if at 0.5-20%, 20.5-40%, 40.5-60%, 60.5-80%, or 80.5-100% of sites, respectively.

Species-environment relationship

Canonical correspondence analysis to determine species-environment relationships was conducted using the CANOCO (1988) computer program (ter Braak 1986). The CCA related the relative abundances of species to measured environmental variables. Ordination axes produced by CCA were linear combinations of environmental variables. Input was means for 12 environmental variables (Table A2) and ranks of 79 species (Table A3) as described for each of 46 stream reaches. The environmental variables were log transformed as required to achieve univariate normality of each individually as tested using the PROC UNIVARIATE procedure (SAS Institute 1990a). Even after transformation, the variables sulfate and phosphorus could not be considered as random samples from normal distributions (P<W 0.006 or less).

The CANOCO program produced seven biplots which placed each fish species in multidimensional (environmental) space. Each species was examined in relation to the environmental gradients of most importance in defining fish assemblage composition. Species were also compared to one another to determine which (for example, within a family group) is most tolerant or intolerant to certain factors. The CANOCO program also produced four biplots which placed each reach into multidimensional (environmental) space. By examining these plots, the probability of a given species occurring at each reach could be assessed.

A review of the assumptions of CCA and its usefulness in multivariate direct gradient analysis in ecology is provided by ter Braak (1986) and Palmer (1993). It is the best technique available for relating community composition to known environmental variability.

Previous Section -- Species-area Relationship
Return to Contents
Next Section -- History of Stream Fish Survey Effort In the Red River Basin

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 04:36:34 EST
Sioux Falls, SD [sdww55]