Aim Species distribution modelling was used to investigate relationships between benthic environmental variables and patterns of demersal fish species distributions using fish occurrence data collected from spatially sparse point sampling. Our goals were to determine: (1) the relative importance of different environmental variables, (2) whether accurate predictions can be developed from these models, and (3) how a comparison of the models against known species ecology can be used to assess model relevance and to provide further insights that could drive future studies of these species.
Location The Recherche Archipelago, southern Western Australia.
Methods Fish distribution data were collected using baited remote underwater video systems (BRUVS): environmental characteristics (substrate type, macroalgal type and presence of sessile biota) were derived from the video footage; and water depth was measured with a depth sounder. Two species distribution modelling techniques were used to explore and quantify the contribution of environmental characteristics to distribution models of 10 temperate marine fish species.
Results Substrate type (reef, sand, cobble) was the most influential variable, and water depth and macroalgal type influenced the probable occurrence of species even over the same substrate type. The probable occurrences of all but one fish species were predicted very successfully, with observed presences being predicted correctly with accuracies > 76%. These predictions were possible due to strong associations between these species and benthic features recorded in this study (substrate type, depth and macroalgal type).
Main conclusions The results demonstrate that the combined influence of multiple environmental gradients must be considered to further develop our understanding of how the environment structures demersal fish distributions. The species distribution models not only agreed with the known ecology of the species examined in detail, but also provided more information on the strength of each environmental attribute and interactions between attributes. These interactions may have previously been misinterpreted as, for example, a depth-driven response. Additional drivers of distribution that were previously not considered, such as the presence of macroalgae, were also found to have influence. This information thus provides a better description of a species’ niche, which is of value to both species management and conservation.