Influence of Sampling Extent on the Relative Importance of Biotic and Abiotic Factors in Explaining Variation in Stream Fish Density
Troy G. Zorn and Michael J. Wiley
Abstract.—In regional survey studies of habitat and fish assemblages, potentially important biological interactions can be masked by strong gradients in habitat variables and associated collinearities among biological variables. We used structural equation modeling to compare the causal influences of local habitat and biotic factors on fish density in rivers and to determine the extent to which the set of sites chosen for analysis influenced their apparent importance. When all sites in our Michigan data set were used, spatial patterns in brook trout Salvelinus fontinalis biomass were 28 times more sensitive to habitat variables than brown trout Salmo trutta biomass. However, when the sample was restricted to trout streams, then brook trout biomass patterns were twice as sensitive to brown trout biomass as habitat variables. In a similar analysis for smallmouth bass Micropterus dolomieu, habitat factors had the strongest effects on fish densities when the analysis was based on all samples available. However, when the sample was limited to steams in which smallmouth bass actually occurred, direct effects of forage fish abundance and indirect effects of habitat via forage fish abundance were more prominent. In both the trout and smallmouth bass analyses, regional data sets (which included sites where the species of interest was absent) overemphasized the importance of habitat factors on fish abundance, but restricting the sample to sites having the species of interest elevated the importance of biotic factors. In reality, both habitat and biotic factors are important to these species, but the variance structure of the sample being analyzed had an overriding influence on the statistical importance of one versus the other. These findings help to resolve apparently conflicting results of previous studies assessing the relative influence of habitat and biotic factors on population abundance.