A model for multiple preserve selection for endangered plants

A stochastic simulation model was created to study the persistence of metapopu- lations with emphasis on endangered plants. The objective was to compare the ability of preserve systems (metapopulations) with a variety of spatial configurations and areas to keep endangered species from extinction. The preserve systems consisted of one or two habitat islands corresponding to local areas of good habitat for a species of interest, surrounded by a matrix of poor habitat. In the two island case the distance between the islands was varied.

The model includes environmental stochasticity with spatial and temporal corre- lations. Demographic stochasticity is included in the model by sampling binomial distributions for seed set, seedling survival, adult survival and seed germination from a seed bank. The model does not include size structure but size plasticity was taken into account by using reproductive capacity for islands, as well as number of indivi- duals as system state variables. Maximum reproductive capacities (taken proportional to habitat area) for habitat islands were used to set upper bounds on populations. Thus there was a simple form of population density dependence. A unique feature of the model is that island size and seed dispersal distances are taken into account, along with number of seeds dispersed, to determine the number of seeds sown on the island as opposed to being dispersed beyond the island boundary.

Preservation indices of species were studied. These indices are mean persistence time, probability of persistence for a fixed period of time and mean persistence time relative to a fixed number of years. The mean persistence time relative to a fixed number of years index is original to this work and was emphasized in the study of persistence distributions generated by the model. It was shown that these three indices often agree as to the preferred preservation system design.

Whether a one or two island system leads to higher preservation was shown by the model to depend in a complex manner on a variety of model parameters. Generally if the preserve system area is great enough, then a two island system is preferred. The threshold area at which this occurs is called the splitting area. Splitting areas for the variously simulated species and forms of environmental stochasticity were computed. The results give hope for the development of general guidelines for preserve system design.