Use of Capture Data to Quantify Change and Test for Effects on the Abundance of Wild Populations

摘要

Two new abundance estimators for multi-plot population studies are presented. One census model permits animal abundance estimation with as few as 4 trapping periods. The other census model estimates the areal mean abundance among a series of animal populations while providing measures of spatial heterogeneity and the catch-effort relationship. Statistical tests of treatment hypotheses were developed for manipulative experiments. The nature of the test statistics was dependent upon the type and performance of the abundance estimators used in the surveys. The appropriate data transform depends on the type of census technique (a logarithmic transform for tag-recapture and an inverse transform for removal sampling). Simulation studies indicate a 20% increase in the power of tests when proper data transformation is used. When homogeneity is assumed, indices of abundance rather than estimates of absolute abundance form the basis of the statistical tests. Tests of homogeneity are presented for tag-recapture and removal sampling techniques. Assumption of homogeneity when valid increases the power of tests 2 to 20 times. An odd's ratio design is presented which pairs control and treatment sites using abundance estimates prior to and after the treatment application. The observations from each control-treatment pair are used to estimate a contrast for a time-by-treatment interaction which has an expected value of zero. For equal levels of replication, the odd's ratio design improves power over simple paired designs. However, when costs are considered, sample paired designs are more efficient. The advantage of the odd's ratio design is it assesses effects when treatments cannot be randomized to the replicate study plots. (ERA citation 10:003381)