COS 308-1 - A comparison of different approximation methods for estimating population growth

Abstract: Estimating the demographic parameters of populations is key to characterizing the Hutchinsonian niche and forecasting geographic range shifts and population persistence in a changing climate. However, estimating growth rate and crowding effects from empirical data can be challenging. One reason is that techniques used to estimate population growth typically require monitoring individuals long enough to observe multiple life stages, which can be daunting when the goal is to model many species or populations. These context-specific population estimates may also yield limited applicability outside the data source domain. In this study, we compared the performance of four procedures for estimating population growth parameters. Three of the procedures we consider assumed density dependence and varied in how the density effects were estimated. The fourth approach was based on trait information and was assumed to be each species' physiological maximum growth rate. To compare the four techniques, we approximated each population growth parameter using annual survey information from the North American Breeding Bird Survey or readily available trait information from the AVONET and Amniote databases for 400 bird species. We then compared how the estimates related to each other and population trends estimated from the Breeding Bird Survey. Here we show that the density-dependent exponential scalar of population growth rates that included environmental stochasticity had a strong positive correlation with survey-wide trend estimates (rho = 0.26) and all of the maximum growth rates approximated from the survey data (rho_x̄ = 0.26). The estimation technique that used trait information produced the lowest estimates on average (x̄ = 0.00, SD = 0.57) and did not correlate with the survey-wide trends across the 400 species. By comparing these techniques, this study provides insight into best practices for approximating population growth parameters using more readily available species data.