Informing wild population monitoring through the use of mixed models
Jansch, Cassandra Rae
MetadataShow full item record
Mixed models were described in the context of two monitoring programs. First, counts of blackside dace provided by the U.S. Fish and Wildlife Service for the upper Cumberland River basin, an area with surface coal mines. Data were analyzed using a negative binomial mixed model that included water conductivity as a predictor variable and spatiotemporal variability as random effect terms. Conductivity was inversely related to blackside dace counts. Second, amphibian counts were recorded in the Jean Lafitte National Historic Park by the National Park Service. A negative binomial mixed model was used to estimate temporal trends in amphibian counts over the duration of the pilot study using all available data and for two reduced-effort data sets. Similar temporal trends were estimated for all three data sets. The results presented here highlight the potential practicality of mixed modeling for estimating trends and relationships to inform design and analysis of monitoring programs.