Literaturdatenbank
Literaturdatenbank |
|
 |
|
Tingley, R. (2008). Using spatially explicit regression models to predict the geographic distributions of species: case studies from the herpetofauna of nova scotia, canada. Unpublished thesis MS, Acadia University.
Added by: Sarina Wunderlich (12 Dec 2010 20:43:16 UTC) |
| Resource type: Thesis/Dissertation BibTeX citation key: Tingley2008a View all bibliographic details  |
Categories: General Keywords: Emydidae, Glyptemys, Glyptemys insculpta, Habitat = habitat, Nordamerika = North America, Schildkröten = turtles + tortoises Creators: Tingley Publisher: Acadia University |
Views: 8/692
Views index: 14% Popularity index: 3.5% |
| Abstract |
|
I provide two case studies from the herpetofauna of Nova Scotia, Canada to illustrate the effects of using different types of explanatory variables in the development of predictive habitat models. In chapter two, I used radio-telemetry data on a declining species of turtle (Glyptemys insculpta) to compare models built using GIS data (indirect predictors) to models built using habitat data collected in the field (direct predictors). Spatio-temporal niche partitioning between males and female turtles confounded the importance of field-based variables. The occurrence of males was successfully predicted using GIS data whereas the occurrence of females was best predicted using field-based variables. These results demonstrate the importance of taking a more sexually explicit and temporally dynamic view of the environmental niche. In chapter three, I used data from a recent herpetofaunal atlas to investigate the relative roles of climate, land-cover and spatial autocorrelation in determining the distributions of eight anurans and three freshwater turtles at a 10-km resolution. The inclusion of land-cover significantly increased the performance of bioclimatic models for the majority of species. Accounting for spatial autocorrelation improved model fit for rare species but generally did not improve prediction success. Although the integration of climate and land-cover data is likely to produce more accurate spatial predictions of contemporary biodiversity, further research is needed to determine whether incorporating land-cover and spatial autocorrelation in climate-induced range shift projections is merited. Researchers should more carefully consider the effects of using different types of explanatory variables when making correlative predictions of species distributions at local and regional scales.
Added by: Sarina Wunderlich |