Research Spotlight: Using Museum Specimens to Refine Models of Species Distribution

Contributed by: Charlotte Germain-Aubrey

PDF version of the article

Using distribution models are crucial for estimating levels of biodiversity at the landscape level. Museum specimens are a significant source of information for these models as they witness current but also past habitats. When coupling this information with historical climate information, museum data become a powerful tool to define a species’ niche, infer its potential changes in the historical past, and project the impact of global change on its distribution.

Museum Specimens for Better Models

In my research, I am developing tools that will enable scientists to construct models based on digitized museum specimen data (and other information). My own research focuses on plants in Florida. Using data from herbaria and other observation sources, I gathered enough high-quality data to build distribution models for 1,548 species and create maps of alpha diversity and endemism hotspots. Associating the date of collection of the specimen with the yearly climate data (from PRISM) proved very useful, especially for narrow endemics, usually limited by subtle variations in climate variables. These methods are still being refined and tested, and Florida offers an unprecedented source of expert knowledge that we can use to ground proof our models. So far, our models have yielded more realistic alpha diversity maps than the typical modeling based on climate variables averaged over several decades (Bioclim), and further work will continue to improve the models (Fig. 1.).

The Future of Florida Plants

We have projected these models into the future to investigate the impact of climate change on the landscape of Florida plants. The most diverse areas will lose diversity (in terms of species number) while the least diverse areas will gain diversity (Fig. 2). We hypothesize that there will be a homogenization of the landscape, with some species becoming more widespread, while others disappear. The mapping of beta diversity over the landscape seems to confirm this hypothesis, with fewer areas showing high geographic species turnover (Fig. 3).

We have also calculated the risk of complete extinction of all species of plants and of endemic species only, using raw models under climate scenarios caused by different levels of CO2 emissions. Gradually, we added layers of complexity, taking sea level rise, dispersal limitation, and projections of urbanization into consideration (Table 1). The main factor affecting the extinction risk in plants in Florida is the level of CO2 emissions!!! The good news is that it is one factor we can do something about…

Biodiversity as a Product of Evolution

Preserving species number is not sufficient for effective conservation. IUCN guidelines explicitly state that a community of species should also be assessed as a product of evolution. For the same number of species, is it better to preserve representatives of 5 or 10 different plant families? In order to estimate the evolutionary diversity of plants in Florida, we reconstructed a phylogenetic tree of the same 1,548 species for which we have distribution models. Harvesting approximately half the matK + rcbL sequences from GenBank and producing the remainder ourselves, we obtained a tree of satisfying resolution and topology (Fig. 4). Combining the information from the species distribution map (e.g., Fig. 1), which encompasses over 8,000 pixels, or communities (list of species), with the phylogenetic tree, we can calculate the accumulated evolutionary history for each pixel (Fig. 5). With a randomization test, we can then evaluate whether each community represents more (overdispersal) or less (clustering) evolutionary history than expected at random, given all species present in all communities of Florida. This approach allows us to reconstruct a map of the evolutionary history of plant communities in Florida (Fig. 6). This map clearly distinguishes the three major EPA ecoregions in Florida, an encouraging sign that this approach successfully reconstructs expert knowledge on biodiversity and can therefore be applied to regions with lesser expert knowledge but important conservation needs.

View this article as a PDF