Biased selection of predictor variables used in climate forecast species distribution models.
Dr Jeremy Ringma1, Dr Ascelin Gordon1, Ms Stephanie Hogg1, Dr Yan Wang1
1RMIT, St Kilda, Australia
Predicting the likely shift in the distribution of species as a result of climate change has become fundamental research question in the field of conservation biology. A climate forecasted species distribution model (SDM) typically functions by creating a model for a species current distribution which is then projected to future climate scenarios. However, forecasting distributions shifts requires a series of assumptions about predictor variables which define a species ecological niche and how relationships can be extrapolated. Climate SDM’s, despite being widely used, have been criticised for their overreliance on correlative, bioclimatic envelope modelling. We conducted a systematic review randomly selecting 380 SDM papers from the years 2000-2018 (20 per year) to determine the formulation of predictor sets used in climate SDM’s. We found that different predictor sets and modelling approaches where used depending on whether papers aimed to model the current distribution, future distribution under climate change, or potential invasive range of a species. Compared to models which aim to solely estimate the current distribution, climate SDM’s typically adopt a correlative approach to predictor selection, and utilise a large number of climatic predictors, and relatively few non-climatic predictors. As a result, climate SDM’s are likely to overstate the impact of shifting climatic space in estimating species distributions due to the omission on non-climatic variables in model configuration. Their findings may also be overly precise in their estimation of range shifts as they lack many of the niche elements which differentiate between currently realised niche and fundamental niche.
Biography:
eremy is a quantitative ecologist and conservation scientist specialising in terrestrial vertebrates. Jeremy completed his PhD in Australian mammal conservation in 2016 and has since held postdoctoral position at the University of Hawaii, investigating the impact of feral pigs on threatened plant communities, and the NESP threatened species hub 4.1 project, prioritising conservation action for introduced predator affected Australian mammals. In Jeremy’s current position at RMIT he is investigating the use of climatic predictor variables in species distribution modelling.