The roles of dispersal and terrain conditions in determining the abundance and distribution of tree species in post-agricultural forests

Since the peak of agricultural production in 1880, the forest in the Finger Lakes Region of New York, USA, increased from less than 20% to over 50% of land cover. Two factors controlling tree species composition in regenerating forest are a species' (1) ability to disperse into abandoned fields and (2) ability to grow under environmental conditions present there. To assess the influence of these two factors on distribution and abundance, I modeled regeneration of forests on a 10 x 14 km landscape with a history of increasing abandonment from agriculture.; To establish the most probable initial conditions in the pre-agricultural forest, I modeled distribution and abundance for eight tree species ( Acer rubrum, Acer saccharum, Fagus grandifolia , Fraxinus americana, Pius strobes, Quercus rubra, Tilia americana, Tsuga canadensis ) under the assumption that current distribution in remnant forest patches is highly correlated to terrain variables. A regression tree algorithm predicted the relative abundance of each tree species according to variables gathered en situ. Within a geographic information system (GIS), each prediction model was applied to digital maps of the environmental variables, producing a map of potential distribution for each species. These maps were evaluated against the remnant forest data and qualitative descriptions in the literature.; For three wind-dispersed species (Acer saccharum, Fraxinus americana, Tilia americana), I then predicted the distribution and abundance that would have occurred in regenerating forests if they were solely dependent on dispersal and growth. Within a GIS, dispersal into abandoned fields was modeled in two ways: as uniform and as decreasing with distance from seed sources. Growth was spatially modeled in two ways: as varying at each location and as a mean for all areas of the same age. Each prediction was evaluated using mean abundance data and the pattern abundance after five generations.; Environmental conditions had the most influence in predicting overall abundance and the patterns of abundance in the regenerating forests. The primary roles of distance-dependent dispersal were limiting distribution and lowering abundance. Changing dispersal characteristics had little effect on the pattern of abundance within the dispersal distance. Increasing maximum dispersal distance had no effect beyond 300 meters.