Use of multivariate associations among species for detecting changes in native plains fish community structure in Colorado

Science Center Objects

The Colorado Parks and Wildlife (CPW) is responsible for monitoring the status of fishes in Colorado and currently seeks to establish a rigorous, efficient monitoring program for fishes that occur in stream habitats from the mountain-plains interface east to the Colorado state line. The primary goals of this program are to: 

  1. monitor the status of species of concern designated within Colorado
  2. monitor the status of all species including those not currently considered at-risk
  3. monitor these species at several scales—reach, segment, stream, sub-basin, watershed. While each of these goals  could entail different sampling methods or analyses, all require spatially extensive sampling of sites either randomly selected or otherwise demonstrably representative of the waters to which inference is desired.

There are at least 2 issues to address in order for the monitoring program to meet its goals. First, there are no previously tested standardized indicators for monitoring the status of plains stream fishes in Colorado. Second, little is known regarding how comparable plains fish communities collected from sites sampled based on convenience are to those collected from randomly  selected sites.

The U.S. Geological Survey, in cooperation with Colorado Parks and Wildlife, assessed the potential use of combining recently (2007 to 2010) and formerly (1992 to 1996) collected data to compare plains fish assemblages sampled from random and fixed sites located in the South Platte and Arkansas River Basins in Colorado.

The first step was to determine if fish assemblages collected between 1992 and 1996 were comparable to samples collected at the same sites between 2007 and 2010.  If samples from the two time periods were comparable, then it was considered reasonable that the combined time-period data could be used to make comparisons between random and fixed sites. In contrast, if differences were found between the two time periods, then it was considered unreasonable to use these data to make comparisons between random and fixed sites.

One-hundred samples collected during the 1990s and 2000s from 50 sites dispersed among 19 streams in both basins were compiled from a database maintained by Colorado Parks and Wildlife.  Nonparametric multivariate two-way analysis of similarities was used to test for fish-assemblage differences between time periods while accounting for stream-to-stream differences.