英文摘要: | Biologists have been conducting surveys aimed at discovering and describing species for centuries. However, it is not known how many species are on the Earth, because discovering species on a one-by-one basis is time consuming and expensive. This lack of knowledge is an impediment to conservation aims and other goals important to society. To address this need, the researchers have developed an approach to identifying groups of populations that are likely to consist of several independent species. Because this approach is based on existing ecological and genetic data, the technique will become more refined as additional data are collected. The proposed research will develop and test this approach using data from an ecosystem in the Pacific Northwest that supports much biodiversity, and where several cryptic species have recently been discovered.
The proposed research will develop a predictive framework for the discovery of cryptic biodiversity that can be applied to all members of an ecosystem. In this two-phase framework, researchers will (i) gather environmental, taxonomic, functional, and genetic data from a reference set of taxa native a model ecosystem and identify which of these species contain cryptic diversity, (ii) conduct statistical analyses to identify habitat features shared by the taxa that harbor cryptic diversity, (iii) collect and analyze environmental data for a second set of species to make predictions about which of these species contain cryptic diversity, and (iv) test these predictions via collecting genetic data for the second set of taxa. The temperate rainforests of the Pacific Northwest of North America will serve as the model system. The disjunction of conspecific populations or putative sister-species pairs between Pacific coastal and interior Rocky Mountain segments presents clear hypotheses regarding potential cryptic diversity: either pre-Pleistocene vicariance, which predicts high cryptic diversity, or post- Pleistocene dispersal where which predicts a lack of cryptic diversity. These predictions will be tested by collection of genomic scale data for 24 endemic disjunct plants and animals, and application of Approximate Bayesian Computation to assess support for each hypothesis. This set of taxa will be then used as a training-set for classifying cryptically diverse species from their occurrence data and climatic variables associated with each. Classical multivariate approaches, such as Discriminant Function Analysis, as well as newer decision-tree approaches (such as RandomForest) will be assessed. |