- Focus on connectivity: ignores pool and upland habitat scales
- Assumes field data are representative
- Assumes random dispersal and no habitat selection
- Model is static; based on current land use
- Resistance values are best guesses - empirical data not yet available (but see Rothermel and Semlitsch 2002)
- Photo-interpreted potential vernal pools are not field-validated
Obviously, any modeling effort such as this carries a large number of assumptions. I’ll run through some of the more important ones.
- First, this model focuses on connectivity, and ignores the pool and upland habitat scales. It assumes that all pools are equal, which is certainly not true.
- We assume, of course, that field data are representative.
- We assume that dispersal is random and nondirectional. This is unlikely to be true, but the relative results of the model should be robust to this, as we’re modeling what is available to dispersing animals.
- This model is static, and doesn't account for land use history, nor for future development, although one could do scenario analyses.
- The resistance values we used are based on expert opinion, because empirical data are unavailable, although some work along these lines is being done. My co-author, Sam Cushman, is looking at the sensitivity of this model to a range of scales and sets of resistance values.
- Finally, the potential vernal pool data are not field validated. Errors in these data are a problem for the model, but this also points out a strength of this model: it can be used to prioritize field validation of potential vernal pools, and can be rerun as the potential vernal pools are updated.