Inferring Ecological Processes from Spatial Data
Jussi Jousimo · Työväentutkimus Vuosikirja · 2018
Environmental and ecological processes drive interactions between organisms in space and time. Inferences from observations, that are dependent and measure the processes of interest indirectly, require an appropriate statistical approach that accounts for the processes' autocorrelation and stochasticity. While acknowledged by ecologists, the presence of spatial dependencies is often ignored. We developed a spatio-temporally explicit statistical framework and considered three case studies in spatial population ecology. In the first study, we conducted a simulation study by mimicking a large-scale and long-term survey of a sparsely populated hypothetic species. The spatio-temporal approach provided density estimates which were found to be unbiased and more precise compared to the baseline models. Furthermore, the model was able to uncover missing observations if there were nearby observations. In addition, we developed a method to estimate population densities for habitat types that were not covered in the study by utilizing individual-level GPS movements data. During winter, the Eurasian red squirrel (Sciurus vulgaris) depends on cone seeds as its main source of nutrition and it is preyed on by several predators. In the second case study, we combined data from surveys of almost 30 years of multiple species in Finland and applied the spatio-temporal framework to find the effect of adjacent trophic interactions on the population dynamics of the red squirrel. The results showed that the population fluctuations of the red squirrel appears to be mostly driven by bottom-up effects such as the cone yield. Positive association with the predator abudance likely results from a shared preference for the habitat rather than predation, but local variation may exists. Powdery mildew (Podosphaera plantaginis) is a fungal disease that infects the plant Plantago lanceolata. In the Åland island of Finland, the plantago is found in a network of meadows and occupancy of the mildew was observed by a ten-year survey. The pathogen follows metapopulation dynamics with occasional colonizations and extinctions of local populations in the host network. Ecological theory predicts that the more isolated populations of a host plant are less susceptible to an infection. On the other hand, higher rates of gene flow in more connected areas facilitate the evolution of resistance. In the last study, by applying the spatio-temporal approach, we found the spatial connectivity to correlate negatively with the occupancies and colonizations and positively with the extinctions. A laboratory experiment confirmed that the plantago is more resistant in more dense sub-networks. We hypothesize this to be a result of a higher gene flow in more dense sub-networks and/or an ability to invest in resistance more in nutritionally rich areas.