Spatial prediction of tree and shrub succession in a small watershed in Northern Cape Breton Island, Nova Scotia, Canada

Abstract

This paper describes the development of a spatially-explicit forest transition model for the Clyburn River Valley watershed in northern Cape Breton Island, Nova Scotia, Canada. The model links spatial quantities of available sunlight, degree-day accumulation and soil water content to the establishment and growth of individual tree and shrub species. Environment-species interactions are captured by way of an artificial neural network (ANN) trained to detect temporal patterns produced with a forest gap model (GIZELA) calibrated for environmental and forest species conditions encountered in northern Cape Breton Island. Impact of environmental conditions on forest succession is expressed through numerical adjustments of the ANN-produced forest-transition projections for several representative landscape types. The ANN-transition modelling approach used is largely automated, making it easy to apply at the species level. ANN calculations explain >95% of the variation present in all GIZELA simulations. Forest-transition calculations are subsequently applied to a representative area of the Cape Breton Highlands as a demonstration of the landscape application of the ANN. Forest-transition modelling can aid in the understanding and prediction of natural forest succession at the landscape level, facilitating the development of long-term conservation plans.