From genetic variation to biotic interactions: towards a multi-scale approach for Species Dsitribution Modelling

Abstract

Biological communities are being affected by anthropogenic driven climate change. One of these effects is a change in the geographical range of species, where species occur in a certain area. Although species occur in a certain geographical range, often they can potentially occur in other areas as well, based on their environmental niche. Species distribution modelling (SDM) is a method of predicting this full possible geographical range. SDM looks at the occurrence of species in a certain area and at the environmental conditions in that area. Based on whether the species does or does not occur for certain values of environmental variables, a prediction is made of the full geographical range. Unfortunately, multiple issues still exist in SDM. For example, SDM does not explicitly consider any variation among individuals or populations. Additionally, the geographical range of a species can also be affected by biotic interactions with other species, which is not considered in regular SDM either. In order to solve these issues, some extension of SDMs have been made; genetically-informed models (GIMs) and community distribution models (CDMs). GIMs are models that explicitly consider genetic variation among populations or individuals of a species. This allows for the consideration of local adaptation in populations. Local adaptation is the concept that populations of a species are adapted to the environmental conditions in their specific location, leading to a higher fitness. Multiple methods for GIMs exits, the most promising of which is a method where genetic clusters are defined. By looking at the genetic data of individuals, clusters of individuals who are genetically similar can be made, on which SDM will be performed. These clusters have been seen to coincide with locally adapted populations of a species. Additionally, studies using this method have shown an improved accuracy of predictions, compared to regular SDMs. CDMs are models that consider multiple species at once. Due to biotic interactions, the distribution of species can be affected by other species and the consideration of multiple species can thus account for these effects. Some earlier methods applied SDM to each species independently after which the results would be combined, a so-called 'predict first, assemble later' approach. However, this assumes that all species exist independently and these methods have shown to be outperformed by other methods of CDMs. These other methods often make use of an 'assemble-and-predict-together' approach. For these methods, co-occurrence of species is considered, based on which biotic interactions can be inferred. Especially, HMSC, which is a type of 'assemble-and-predict-together' approach, can provide an important step towards accurately incorporating multiple species in SDM. For the future, it could be promising if the method of creating genetic clusters from GIMs could be combined with the 'assemble-and-predict-together' approaches from CDMs. This way, multiple clusters from multiple species could be considered at the same time. This would provide a multi-scale approach, incorporating information on both genetic level as well as community level. Future research testing such a method, could prove its functionality.