Published: Jan 1976
| ||Format||Pages||Price|| |
|PDF (168K)||12||$25||  ADD TO CART|
|Complete Source PDF (5.1M)||12||$75||  ADD TO CART|
The problem considered here is whether a specified environmental change does, or does not, affect a biological community. Commonly used procedures are often inappropriate for the data or for the questions asked. Such procedures, as well as an appropriate one, are discussed and illustrated using a set of artificial data with properties similar to those of a real benthic community. Ninety-two stations on a kilometre grid are sampled for abundances of six benthic species and four environmental variables, before and after an increased inflow of dissolved salts.
A multiple regression model which assumes that species abundances are related to environmental variables in a linear additive manner yields inefficient and misleading results. Cluster analysis methods efficiently describe natural spatial patterns, but temporal changes are swamped by the spatial patterns and are not detected. A linear additive model which relates changes in species abundances to changes in environmental variables removes the variation due to spatial pattern from the analysis and clearly shows that biological changes related to increased salt concentration have occurred. Methods for describing the changes and testing related hypotheses are presented.
water quality, water pollution, aquatic biology, monitors, water analysis
University of Manitoba, Winnipeg 19,
Paper ID: STP27846S