Explicitly integrating a third dimension in marine species distribution modelling

Abstract

Marine species distribution modelling (SDM) studies frequently pair surface environmental variables with deep-water occurrence records de - spite the disparity of environmental conditions at the surface and at depth. Straightforward methods, which require only simple modification of existing SDM techniques, have been recently proposed as a means to overcome this problem. Nevertheless, subsequent marine SDM studies have not adopted these modified techniques, continuing to draw conclusions from analyses that associate 2-dimensional (2D; i.e. surface) environmental data with occurrence points distributed in a 3-dimensional (3D) space (i.e. across a range of depths). These include studies involving multi-species macroecological assessments, where SDMs are rarely tailored to suit each species modelled. Here, we use traditional (2D) and modified (2.5D) SDM methods to emphasise the importance of incorporating depth in marine ecological modelling. We demonstrate that ignoring depth may lead to misrepresentative and potentially misleading results, and show that relatively straightforward 2.5D techniques provide an enhanced representation of the environmental niche and, therefore, produce more representative range estimates than their 2D counterparts. Explicit 3D approaches need to be developed to better represent marine diversity and distributions and to test macroecological generalisations.