UW Aquatic & Fishery Sciences Quantitative Seminar

Eric Ward

NOAA Fisheries, Conservation Biology Division

Identifying Spatial Population Structure with Time-Series Models

Abstract

Population dynamics are often structured spatially – neighboring regions may act independently in the classic metapopulation framework, or they may be linked by varying degrees of movement and dispersal. Identifying population structure is important from a management perspective – conservation units may be prioritized depending on whether they are sources or sinks, and how they are connected to neighboring populations. Identifying spatial structure can be problematic for many species, because most existing methods for determining spatial structure require fine scale data, including mtDNA or telemetry data. An alternative approach presented here is to analyze time series of abundance using multivariate state-space models (MSSM). This approach is extremely flexible and can used to model a range of spatial structures, from panmictic populations to metapopulations, even when a large fraction of data is missing.  In this talk, I'll discuss the MSSM framework, including parameter estimation, and model selection techniques.  Two examples will focus on testing hypotheses about geographic divisions, genetic structure (mtDNA), and clustering by environmental covariates.