Can Species Distribution Modelling be Predictive?
Friday, September 10th, 2010Because Species Distribution Modelling (SDM) is an important tool for generating policy relevant information and an instantiation of fundamental ecology, we might hope the answer was YES. But the Thematic revealed more NO answers.
Jorge Soberon’s keynote address explored the ecological concepts and data essential to prediction. Many models are available but they address a restricted set of concepts. What can we do better?
Novel data sources may open new doors model building. Paleo data may allow better reconstruction of climate niches (David Nogues-Bravo) and allow better model validation (Frederik Saltre). With more detailed data trait based SDM may also be possible (Xavier Morin, Frederik Saltre) allowing independent model validation. Without this we may be building models on severely biased data (Jaime Garcia-Marquez).
What if the answer is really NO? Including ecological details may be more difficult than ecological interpretations. By analysing climate data we may be able to prioritise management efforts (Ralf Ohlemuller). Complimentarily, strategic modelling may help develop tactics for management and address evolutionary issues in ecological management (Justin Travis).
So where did the YES answers come from?
Novel methods could come from a Bayesian framework (Greg McInerny, Bob O’Hara, Drew Purves), addressing the ecological features that SDM lacks (Jorge Soberon). Many approaches exist for sampling errors (Bob O’Hara) and we can correct for fine scale heterogeneity (Greg McInerny). These approaches could be combined with mechanistic SDM (Greg McInerny, Drew Purves).
Whilst not a YES, we might have a MAYBE.
