Major challenges for correlational ecological niche model projections to future climate conditions

• Published in: Annals of the New York Academy of Sciences Vol. 1429; no. 1; pp. 66 - 77
• Main Authors: Peterson, A. Townsend, Cobos, Marlon E., Jiménez‐García, Daniel
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 01.10.2018
• Subjects: biodiversity; Climate; Climate change; Climate effects; Climate models; Climatic conditions; Confidence; Ecological effects; ecological niche model; Ecological niches; General circulation models; Geographical distribution; Methodology; niche truncation; uncertainty; niche truncation; ecological niche model; biodiversity; uncertainty; climate change
• ISSN: 0077-8923; 1749-6632
• DOI: 10.1111/nyas.13873

Species‐level forecasts of distributional potential and likely distributional shifts, in the face of changing climates, have become popular in the literature in the past 20 years. Many refinements have been made to the methodology over the years, and the result has been an approach that considers multiple sources of variation in geographic predictions, and how that variation translates into both specific predictions and uncertainty in those predictions. Although numerous previous reviews and overviews of this field have pointed out a series of assumptions and caveats associated with the methodology, three aspects of the methodology have important impacts but have not been treated previously in detail. Here, we assess those three aspects: (1) effects of niche truncation on model transfers to future climate conditions, (2) effects of model selection procedures on future‐climate transfers of ecological niche models, and (3) relative contributions of several factors (replicate samples of point data, general circulation models, representative concentration pathways, and alternative model parameterizations) to overall variance in model outcomes. Overall, the view is one of caution: although resulting predictions are fascinating and attractive, this paradigm has pitfalls that may bias and limit confidence in niche model outputs as regards the implications of climate change for species’ geographic distributions. Species‐level forecasts of distributional potential and likely distributional shifts, in the face of changing climates, have become popular in the literature in the past 20 years. Here, we depict geographic dimensions of effects of (1) specific point occurrence data used, (2) representative concentration pathway, (3) general circulation model, and (4) parameter settings, as they affect future model transfer results.