Ecological theory predicts that disease incidence increases with increasing density of host networks, yet evolutionary theory suggests that host resistance increases accordingly. To test the combined effects of ecological and evolutionary forces on host-pathogen systems, we analyzed the spatiotemporal dynamics of a plant (Plantago lanceolata)-fungal pathogen (Podosphaera plantaginis)relationship for 12 years in over 4000 host populations. Disease prevalence at the metapopulation level was low, with high annual pathogen extinction rates balanced by frequent (re-)colonizations. Highly connected host populations experienced less pathogen colonization and higher pathogen extinction rates than expected; a laboratory assay confirmed that this phenomenon was caused by higher levels of disease resistance in highly connected host populations.

Spatial synchrony between populations emerges from endogenous and exogenous processes, such as intra-and interspecific interactions and abiotic factors. Understanding factors contributing to synchronous population dynamics help to better understand what determines abundance of a species. This study focuses on spatial and temporal dynamics in the Eurasian red squirrel (Sciurus vulgaris) using snow-track data from Finland from 29 years. We disentangled the effects of bottom-up and top-down forces as well as environmental factors on population dynamics with a spatiotemporally explicit Bayesian hierarchical approach. We found red squirrel abundance to be positively associated with both the abundance of Norway spruce (Picea abies) cones and the predators, the pine marten (Martes martes) and the northern goshawk (Accipiter gentilis), probably due to shared habitat preferences. The results suggest that red squirrel populations are synchronized over remarkably large distances, on a scale of hundreds of kilometres, and that this synchrony is mainly driven by similarly spatially autocorrelated spruce cone crop. Our research demonstrates how a bottom-up effect can drive spatial synchrony in consumer populations on a very large scale of hundreds of kilometres, and also how an explicit spatiotemporal approach can improve model performance for fluctuating populations.

Random encounter models can be used to estimate population abundance from indirect data collected by non-invasive sampling methods, such as track counts or camera-trap data. The classical Formozov–Malyshev–Pereleshin (FMP) estimator converts track counts into an estimate of mean population density, assuming that data on the daily movement distances of the animals are available. We utilize generalized linear models with spatio-temporal error structures to extend the FMP estimator into a flexible Bayesian modelling approach that estimates not only total population size, but also spatio-temporal variation in population density. We also introduce a weighting scheme to estimate density on habitats that are not covered by survey transects, assuming that movement data on a subset of individuals is available. We test the performance of spatio-temporal and temporal approaches by a simulation study mimicking the Finnish winter track count survey. The results illustrate how the spatio-temporal modelling approach is able to borrow information from observations made on neighboring locations and times when estimating population density, and that spatio-temporal and temporal smoothing models can provide improved estimates of total population size compared to the FMP method.