Downscaling future land cover scenarios for freshwater fish distribution models under climate change

Abstract

The decreasing freshwater biodiversity trend can be attributed to anthropogenic impacts in terms of climate and land cover change. For targeted conservation efforts, mapping and understanding the distribution of freshwater organisms consists of an important knowledge gap. Spatial modelling approaches offer valuable insights into present-day biodiversity patterns and potential future trajectories, however methodological constraints still hamper the applicability of addressing future climate and land cover change concurrently in one modelling workflow. Compared to climate-only projections, spatially explicit and high-resolution land cover projections have seen less attention, and the lack of such data challenges modelling efforts to predict the possible future effects of land cover change especially on freshwater organisms. Here we demonstrate a workflow where we downscale future land cover projection data from the Shared Socioeconomic Pathway (SSP) scenarios for South America at 1 km² spatial resolution, to then predict the future habitat suitability patterns of the Colombian fish fauna. Specifically, we show how the land cover data can be converted from plain numbers into a spatially explicit representation for multiple SSP scenarios and at high spatial resolution, employing freshwater-specific downscaling aspects when spatially allocating the land cover category grid cells, and how it can be fitted into an ensemble species distribution modelling approach of 1209 fish species. Our toolbox consists of a suite of open-source tools, including Dinamica EGO, R, GRASS GIS and GDAL, and we provide the code and necessary steps to reproduce the workflow for other study areas. We highlight the feasibility of the downscaling, but also underline the potential challenges regarding the spatial scale and the size of the spatial units of analysis.