MARINE ECOLOGY

Marine Ecology

Marine Ecology

Introduction Many discussions of ecological and environmental issues involve evaluating the evidence for or against a temporal trend. For example, is the abundance of a particular population increasing, remaining approximately constant, or declining over time? The data to answer this question are often a sequence of annual counts of individuals (e.g., Houlahan et al. 2000). Do the observed counts represent random fluctuations around no trend or do they provide evidence of some trend? If it is reasonable to assume a linear trend, the usual statistical analysis is to fit a linear regression and test the null hypothesis that the slope is zero.

Discussion

Equivalence tests for trend will be illustrated with four long-term data sets on amphibian (salamander) population sizes. Complete counts of all breeding females of two Ambystoma species, A. talpoideum and

A. tigrinum, have been made at Rainbow Bay, South Carolina, USA since 1979 (Semlitsch et al. 1996). Estimates of abundance of Desmognathus monticola and

D. ochrophaeus at Coweeta Hydrological Laboratory North Carolina, USA have been made by constant-effort searches since 1976 (Hairston 1996). The data used here include the population counts until 2002. The number of searches for Desmognathus varied between one and three per year; for this paper, we consider the average count for each year. Two populations (Ambystoma

spp.) have large annual variation; two (Desmognathus

spp.) have small annual variation (Fig. 1). The four were selected from the larger number of amphibian species monitored in these community surveys.

AIC statistics were used to choose an appropriate model for the variability of observations around the log-linear regression line (Verbeke 1997:113-115). For all four species a first-order autoregressive error model was more appropriate than the independence model.

For the two Desmognathus species, an equal-variance model was more appropriate than a weighted model that assumed the variance was a function of the number ...