Assessing populations in complex systems
Loading...
Date
1993
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Surveys of wild populations seldom satisfy assumptions fundamental to the survey techniques used to assess them (i.e., spatial distributions tend to be nonrandom). When sample sizes (i.e., number of observations) are small, or observations are taken without considering independence, assessments of density are likely to be biased. For the purposes of this dissertation, I use the term bias in referring to a systematic distortion of an estimate (operational bias) caused by the sampling procedures that are employed, rather than distortion of the expected value from all possible observations (statistical bias). Bias is exacerbated when the variance-to-mean ratio of the estimated population parameter is scale-sensitive because observations have an inherent scale independent of the population's scale. Density estimates from transects may be biased by differences (1) between the scale of the transect (i.e., a reflection of the area to perimeter ratio) and the scale of animal spacing, or (2) between the scale of the transect (i.e., a reflection of the observed variance-to-mean ratio in a scale-sensitive system) and the scale at which a random distribution of animals can be observed. Each of these sources of bias may be used to remediate (i.e., measure and remove bias) observations and produce unbiased estimates of density. I demonstrate that differences in the behavior of sex-age classes within a population, can lead to class-specific biases with one class being overestimated while another is underestimated. Using aerial surveys of south Texas white-tailed deer populations, I show how these biases can confound assessments of populations in fragmented or disturbed landscapes, or communities where changes in the population of one species occurs independent of other populations. Using simulations and field surveys of desert mule deer in Big Bend National Park, Texas, I show how validation of spatial models can be confounded when biased population estimates fail to match the scale of the modeled population.
Description
Vita.
Keywords
Major wildlife and fisheries sciences