Contrasting the ability of data to make inferences regarding dispersal: case study of the Red-cockaded woodpecker (Picoides borealis)

Abstract

Dispersal is a critical biological process that contributes to the persistence of species in complex and dynamic landscapes. However, little is known about the ability of different types of data to reveal how species interact with landscape patterns during dispersal. Further, application of process-based, landscape-scale models able to capture the influence of land use and climate change are limited by this lack of dispersal knowledge. Here we highlight a method for building such models when dispersal parameters are unknown, but information on the mating system and survival are available. We applied a common statistical framework, rooted in information theory, to contrast the ability of abundance, movement, and genetic data to estimate dispersal parameters for endangered Red-cockaded woodpecker (RCW), using an individual-based, spatially-explicit population model. Dispersal was modeled as a multifaceted process in which foray distance, long-distance dispersal, competition for mates, and landscape permeability were treated as uncertain. We found that movement data are three-times more powerful than abundance data collected at the same spatial and temporal scales. However, habitat occupancy data collected over much a shorter time scale but at regional spatial scales were very effective for estimating dispersal. We also found that one-year of abundance data provided a similar reduction in uncertainty as genetic differences among breeding groups estimated using a 24-year pedigree. Substituting population genetic data for movement and abundance data often led to the same parameter values, but not always. Our study highlights important differences in the information content of data commonly collected in the field.