Computer model predicts chameleon distribution in Madagascar: satellite data and museum specimens can accelerate efforts to complete species inventories, aid in the design of future reserves, and help find new species

Christopher J. Raxworthy, Associate Curator in the American Museum of Natural History's Division of Vertebrate Zoology, and six colleagues have developed a modeling approach for studying biodiversity that combines satellite data with specimen locality data from museum collections and which has successfully predicted the geographic distribution of 11 chameleon species in Madagascar. In an unanticipated result especially useful to conservation efforts, the model also correctly predicted the existence of previously unknown areas of chameleon distribution, where 7 new chameleon species were found. This discovery suggests that for poorly explored regions, satellite data and data from museum collections can help identify promising places to survey for new species--an exciting development, especially beneficial to the conservation community.

Understanding the distribution of a species is one of the most important and basic requirements for effective conservation and ecological and evolutionary research. Yet most tropical species have not been surveyed well enough to allow an accurate assessment using conventional methods, in which biologists visit a region in an intensive effort to locate and count species. Dr. Raxworthy and his colleagues have demonstrated that existing museum collections and satellite measurements of Earth's surface and climate can be used to accurately predict species distributions anywhere on Earth--even in poorly explored tropical regions. This study is the first to successfully predict the distribution of any species in Madagascar using satellite imagery and information from museum specimens, and the first to evaluate the predictive usefulness of historical museum specimens in collections (dating back to the 1800s) versus recently collected field data from Madagascar.

The acquisition of detailed descriptions of exactly where Earth's rich biodiversity lives, species by species, requires decades of dedicated fieldwork. In Madagascar, teams of highly trained biologists visit often remote sites to do this work. A paucity of data on the regional distribution of species, especially in more remote areas, is one of the major factors that complicates and potentially delays conservation decisions that could save threatened flora and fauna. This new research by Dr. Raxworthy and his associates, published in a recent issue of the journal Nature, demonstrates conclusively that a technological solution exists that can speed up the process of regional species inventories and thereby prevent unnecessary loss of threatened animals, especially in tropical environments with diverse habitats and climates. The research also shows that both historical and modern field data can be extremely useful for predicting chameleon species distribution in Madagascar, although contemporary field data used in concert with satellite data provides the most accurate biogeographic distribution predictions.

This new chameleon prediction study tested the accuracy of several other distribution models--including one based on information gathered from historical museum specimens, and one based on data from more recently collected specimens--against other locality data that was set aside for testing purposes, and against recent inventories of 11 sites where chameleons were also surveyed. Although the historical data alone prove to be predictive in ways that are useful to conservation decision-makers, and accuracy improved with the combination of modern and historical data, the team found that modern data alone were the best predictor of where the 11 chameleon species live.

All of the models rely on environmental data collected by several satellites and the Space Shuttle, provided by the National Aeronautics and Space Administration (NASA), U.S. Geological Survey, and National Oceanic and Atmospheric Administration (NOAA). Environmental data include land cover (as viewed from space), rainfall, cloud cover, average and seasonal temperatures, and topographic data, which were input into GARP (Genetic Algorithm for Rule-set Prediction), a software package for biodiversity and ecology research that allows users to predict species distributions.

The model's intriguing ability to predict the location of previously undiscovered chameleon species arose unintentionally. When Dr. Raxworthy and his colleagues examined the models, they found overlapping areas of error where the models incorrectly predicted that 4 species lived. Examining their field data collected in two of these regions, they realized that these areas actually contained 7 other closely related species that are new to science. The areas that initially seemed to represent "error" in the models pointed to regions that are of critical conservation importance because they provide habitats for locally confined species that had not been previously recognized. Identifying neglected areas with unique biodiversity currently has enormous value in Madagascar, because the Malagasy Government recently announced plans to expand the protected area network, thus providing a new opportunity to conserve species currently excluded from the island's existing reserves.