Unravelling the effects of environmental variation on the population dynamics of structured populations

Complex environmental effects, combined with little temporal replication in most data sets, make investigating the ecological consequences of rapid climate change difficult with current tools. Structured population models are widely used to explore population responses to environmental variation. I develop and apply new statistical methods to parameterise such models. First I describe a structural equation model (SEM) approach for capturing temporal covariation among demographic rates via latent variable(s). When rates are positively correlated the latent variable(s) act as axes of ‘environmental quality’. This provides a simpler target for identifying the drivers of variation, than treating each process independently. Where drivers cannot be identified perturbing the latent variable(s) may represent the best alternative for exploring population-level responses to environmental change. Quantifying the effects of underlying drivers allows population viability under different management strategies to be predicted. Such studies frequently assume a stationary environment, despite rapid climate change. Where climatic drivers are included, single temporal windows of influence are typically chosen a priori. I show forecasted climate change alters predicted population viability under different management regimes in a rare fire-adapted herb. I illustrate that the effect of a single climatic variable may differ over time, suggesting a priori selection of single temporal windows can decrease predictive performance. I use the SEM approach to show that most (co)variation in survival and fecundity across different age-sex classes in a Soay sheep population is driven by a single environmental axis. I show climatic conditions during the energetically expensive autumn rut are nearly as important for overwinter mortality as the winter periods focused on in previous studies. I explore how density dependence, a temporal trend, population structure, and environmental variation interact to drive dynamics in this population. Throughout this thesis I apply novel methods that increase our ability to accurately forecast population dynamics under environmental change.