英文摘要: | Climate change poses new challenges to the conservation of species, which at present requires data-hungry models to meaningfully anticipate future threats. Now a study suggests that species traits may offer a simpler way to help predict future extinction risks.
As biodiversity erosion intensifies worldwide1, 2, threatening our biological heritage, nature management tools are needed more than ever3. Unfortunately, the computer models used to assess population declines are data hungry, and data on endangered species — those most in need of modelling — are usually lacking. Writing in Nature Climate Change, Pearson and co-authors4 show a way to sidestep this data disparity by showing that important, and far easier to gather, life-history traits and spatial distribution characteristics may be used as a surrogate for species extinction risks derived from advanced hybrid species range shift and population decline models. As global climate change is already causing species to move out of their traditional range5, the preservation of biodiversity requires efficient conservation prioritization strategies1. This task is facilitated by the International Union for Conservation of Nature (IUCN) Red List of Threatened Species2, a system — designed before anthropogenic climate change was seen as a major threat — to determine the threat status of species for policy and planning purposes. This tool, if applied to all taxonomic groups, could serve as an efficient 'barometer of life'6. Yet, exhaustively sampling the distribution of all endangered species remains a difficult task7, especially for highly mobile or cryptic species, and even if this were achieved, future distributions are likely to differ from present ones8, 9. Predictive models can be used here to complement observations and forecast future distributions7, 10, if made relevant to conservation objectives3, 7. However, the dynamic models required for population viability and extinction risk analyses8, 9 usually need heavier data input for each target species than simple distribution models7, hampering their application to many groups of organisms11. One remedy for this data limitation is to derive predictive tools for groups of species with similar functions or dynamics instead of individual species, making use of the increasing information available in trait databases11, and use them to derive informative biodiversity forecasts to assist conservation planning. A second remedy is proposed by Pearson and co-authors4, and this is where their findings constitute an important advance. They show that a set of life-history and spatial traits of endangered species — as already used in the IUCN red list assessments2 — can be used to predict extinction risks under climate change. They do this by looking for traits that correlate well with model-based predictions of future extinction risks (Fig. 1). The extinction risks are estimated in their study by combining, for each species, two types of model: a first model predicting the spatial distribution of habitat suitability7, using a range of environmental variables as predictors (such as climate, topography, land use and hydrography); and a second, temporal model simulating population demography and dispersal8, 9, based on the predicted habitat suitability, generic demographic cycles (life history) and documented or estimated dispersal capacity. This combined spatio-temporal model is finally used to run repeated simulations (including stochasticity, that is, random natural variation) of populations and their interconnections in geographical space and across time to assess trends in population sizes. Extinction risks were calculated with this hybrid modelling approach for six generic life-history types, based on data for 36 species of amphibians and reptiles endemic to the United States. A separate machine-learning modelling approach was then used to look for relationships between the candidate traits, characterizing life-history and distribution characteristics, and the previously predicted extinction risks. They found currently occupied area, population size and generation time to be among the most important factors in their analysis, and further identified interactions between them (for example, extinction risk decreasing in smaller occupied areas with increasing generation time) as providing additional predictive power. This suggests that current characteristics of species' range and demography, as well as their interactions, may already provide crucial initial information for classifying species as being at risk of extinction under future climate change.
|