A model of fragmentation in the Canadian boreal forest

Ecological studies have generally examined forest fragmentation in terms of descriptive metrics or simulation using Monte Carlo or percolation processes that assume fragmentation is a random process. However, most fragmentation results from human decisions on agricultural settlement. This study used a previously tested rulebased agricultural settlement process model (GEOMOD2) to describe which parts of a boreal forest landscape are selectively cleared for agriculture. Nearness to neighbors, amount of stoniness, soil type, and soil texture best explained the fragmentation process. To compare settlerʹs decisions on the productivity of the landscape with moisture-nutrient gradients, we used a hydrological topographic index to capture the variability of wetness according to hillslope position. Results showed that settlers were selecting higher hillslope positions irrespective of substrate (glaciolacustrine or glacial till); i.e., they appear to have used observable attributes such as stoniness, soil texture, and hillslope position rather than soil productivity in making settlement decisions. Thus, the species richer upper hillslopes of aspen parkland (glaciolacustrine) and aspen and white spruce forest (glacial till) were settled first, while the species poorer lower hillslopes of aspen forest (glaciolacustrine) and white spruce and balsam fir forest (glacial till) were settled later.