Climate-correlative modeling of phytogeography at the watershed scale.

Abstract

The goal of this research was to develop a watershed-scale model for predicting changes in plant species distribution and abundance (phytogeography) that might occur as a result of changes in climatic factors with global warming. The model was designed: 1) to be spatially explicit and applicable across the entire watershed; 2) to apply to a number of particular species rather than general vegetation types; 3) to predict abundance as well as presence/absence; and 4) to work with simple environmental data, but reflect a biological rationale. Correlations were sought between current phytogeography in the watershed and the synoptic climate variables mean annual temperature, total annual precipitation and cool-/warm-season precipitation ratio. The contribution of edaphic and topographic variables to correlative models was examined and found to be negligible. The correlations established for current conditions were extended to hypothetical future conditions of changed climate in which the values of the variables were manipulated and the model run to produce predictions of altered future phytogeographies. Twenty-seven different hypothetical climate scenarios were modeled, incorporating a 1°C or 2°C rise in temperature with as much as a 10% increase or decrease in seasonal precipitation. Spatial articulation of the model was achieved through raster analysis of gridcell based data layers in a geographic information system. Primary input layers were a series of high-resolution (360x360m) interpolated climate-variable surfaces and a geographically referenced database of plant species presence and abundance derived from an aerial videography sample of the watershed. Logistic regression analysis was used to calculate, for a given set of conditions, the most probable state (present/absent) and abundance class for ten plant species at each grid-cell location in the watershed. Fragmentation of species' distributions before and after change was examined. Results for all studied species showed marked changes in distribution and abundance with temperature rise. Desert species will likely increase in abundance and occupiable area as forest and woodland species decrease, but much depends on the interaction of precipitation with temperature. Model predictions are conservative compared with paleoecological evidence of past changes.