Forecasting ecosystem response to climate change is critical for guiding policy-making but challenging due to: complicated relationships between microclimates and regional climates; species’ responses that are driven by extremes rather than averages; the multifaceted nature of species’ interactions; and the lack of historical analogs to future climates. Given these challenges, even model systems such as the Galapagos Islands, a world-famous biodiversity hotspot and World Heritage Site, lack clear forecasts for future environmental change. Here, we use the Galapagos to illustrate how different conventional approaches can yield opposite forecasts due to the lack of climate analogs in the fitting model. To address the conflict, we developed a novel non-parametric method for simulating the future based on observed vegetation productivity (NDVI) during analogous weather in the present time period.
Results/Conclusions
Using whole satellite images taken from the past to piece together a simulated future, we projected that vegetation productivity will increase over the next century by approximately one standard deviation archipelago-wide, with increases largest during the wet season (January to June), and in the arid zones. This greening may impact a variety of ecological processes, species of conservation concern, and agricultural practices. Applying a similar framework to sparse data on Galapagos Giant Tortoise egg survival, we find that the future has the potential to bring increases in future egg mortality that could potentially be large enough to substantially impact population growth rates, despite the relative low sensitivity of tortoise demographic models to egg survival. Our straightforward approach can be applied to many other regions, particularly those with rapid ecosystem responses to stochastic inter-annual climatic fluctuations that provide appropriate climate analogs for forecasting.