Effetti del cambiamento cimatico sul ecosistema della costa finlandese: un'applicazione di modelli di distribuzione delle specie

The Baltic Sea is an inland sea with brackish water, located in the Northern part of the Atlantic Ocean, between Sweden and Finland. In this sea, and in the surrounding coastal areas, effects of climate change and euthrophication are particularly pronounced. This is because Baltic Sea's extreme location and conditions have caused many changes to be concentrated and to have happened more quickly than in many other regions of the Earth. Furthermore, its unique biodiversity makes this sea a crucial area of interest for the scientific community, and past changes have been object of accurate studies. Nevertheless, the lack of understanding on the expected future changes in these areas is still high. In this master's thesis we apply state-of-the-art statistical species distribution models to predict the likely effects of climate change to biota in the Finnish coastal region. The core part of the thesis is devoted to modelling the distribution and the extent of sea-spawning whitefish (Coregonus lavaretus) and vendace (Coregonus albula) larval in the Gulf of Bothnia, the northern part of the Baltic Sea. Whitefishes have a central role in Gulf of Bothnia ecosystem, other than having a relevant economical and social value in Finnish society. Compared to previous studies from 1990s, the extent of whitefish larval areas has decreased. This suggests that the distribution of the reproduction areas could be a good indicator of the health of the Baltic Sea shores. In this work we discuss the possibility that long-term changes in the environment, such as increasing temperature, decreasing salinity, more frequent iceless winters and increasing eutrophication, have reduced the reproductive success of sea-spawning coregonids. The outputs of this study include larval distribution maps, useful monitoring tools, that may assist integrated coastal zone management and environmental protection, e.g. by focusing conservation measures in the most appropriate place. To conduct the analysis, extensive sampling data on larval occurrences were combined with GIS (geographic information system) raster layers on a selection of relevant environmental variables. The resulting data were used to implement regression models known, in ecology, as species distribution models. These are hierarchical models with additional Gaussian random effect, which predict the spatial probability of species occurrence or abundance. In our study, we represent larvae abundances distributions through counting processes (Poisson or Negative-Binomial), so that we could further model the average species log-density as a linear function of a set of environmental covariates and a Gaussian spatial random effect. After fitting several models we predict larvae abundances along the whole Gulf of Bothnia coastline, at both current and future climate conditions. Finally, results and potential improvements are discussed.