November 9 2016

12:00 LSB 2320

Jessica Blois
School of Natural Sciences
University of California, Merced

Impact of climatic novelty: translating insights from the past to projections for the future Impact of climatic novelty: translating insights from the past to projections for the future


Future climates are projected to be highly novel in many places around the world. In response, widespread biodiversity alteration has been predicted, characterized by species extinctions, invasions, and ecosystem transformation. These biodiversity predictions are typically based on algorithms that link the distributions of species and assemblages today wholly or in part with climate, then project them to the future given different climate scenarios. The robustness of these algorithms to novel future climates represents a significant source of uncertainty for the pattern of future biodiversity. Recently, we examined how well models perform when projecting fossil pollen assemblages from eastern North America across periods of past climate change, specifically testing two different groups of modeling algorithms- those that are based only on climate and those that also include species co-occurrences. In this talk, I examine the magnitude and novelty of future climate change relative to the past 21,000 years and use the fossil pollen record of eastern North America to test the robustness of these different algorithms to climate novelty. Even in eastern North America, which represents a conservative region for future climate novelty, the accuracy of models that project changes in species and assemblages is likely to be low in the future. However, some types of algorithmsthe models that incorporate co-occurrences among species and not just climatemay better rise to the challenge of predicting biodiversity into a novel future.















































































































































































































































































































































































































































































































this is idtest: