Abstract:
|
Ecologists, population biologists, and conservation agencies often study population dynamics to evaluate carrying capacities and population growth rates. Because it is usually infeasible to count all individuals within a population, abundance models are used to analyze the investigator's count data throughout a study period. However, if count data are collected seasonally in a way that depends on population dynamics, a traditional abundance model may overestimate or underestimate abundance during periods with less count data. We developed an abundance model that accounts for preferential sampling in time, which produces less biased inference for the true abundance during periods when less count data are available. We extend this method to the multi-species context, where abundance among species may be correlated and sampling frequency may not be tied to a single species. Our proposed method accommodates multiple species and provides inference that is less biased by irregular sampling patterns. We illustrate our method through simulation and a case study of mosquito abundance across the United States, and we demonstrate the improvement of temporal abundance forecasts.
|