Development of edge effects around experimental ecosystem hotspots is affected by hotspot density and matrix type
The simultaneous influence from multiple edges on remnant forest patches (such as wildlife corridors, protection buffers, small unharvested remnants or corners of larger patches) in harvested forest landscapes could impair, or possibly enhance, their effectiveness. When multiple edges are in close proximity, there may be interactions of edge influence such that the observed response is greater or less than would result from the influence of either edge alone. We examined possible ways in which two nearby forest edges of similar or different types might interact in terms of their influence on forest structure. We present an ‘edge interaction’ model for three possible types of interaction of edge influence: (1) no interaction, edge influence is limited to the strongest influence from either edge, (2) positive interaction, observed edge influence is greater than from either edge alone; (3) negative interaction or resistance, the influence from both edges is less than from a single edge (e.g., an older edge is resistant to effects from a younger edge). Empirical data for forest structure at the edges of cutblocks (harvested areas) and water bodies were entered into the models to predict edge influence in narrow forest corridors assuming the null hypothesis of no interaction. Randomization tests were used to compare predictions to observed edge influence on recently-fallen logs and Populus spp. (P. tremuloides Michx. and P. balsamifera L.) sapling density in lakeshore buffers in boreal mixedwood forest as well as on canopy cover and log, tree and snag abundance in riparian buffers and forested corridors separating cutblocks in Picea mariana (Mill.) BSP. forest. In lakeshore buffers, there was evidence of both positive and negative interaction at different locations within the buffer or at different times since buffer creation for both the abundance of logs and Populus sapling density. Trends suggested positive interaction for canopy cover and live tree density in riparian buffers and cutblock separators, and for snag density near the cut edges of riparian buffers. Testing hypotheses arising from our model of interaction of edge influence could lead to a clearer understanding of edge influence in fragmented landscapes.