Improved estimation of forestry edge effects accounting for detection probability

Background/Question/Methods

It is well understood that the effects of even-aged timber harvest extend beyond the edge of the harvest. However, most studies have used nonparametic methods to estimate an edge effect (e.g. Loess, splines). These methods require a subjective evaluation of the asymptote to estimate the edge effect. We used a 4-point logistic function with random site effects in a hierarchical model to estimate abundance as a function of distance from the edge, while accounting for detection probability. Beyond accounting for imperfect detection, this method has distinct parametric advantages, such as being able to quantify the total effect of timber harvesting and exact distances where the population reaches set percentages of carrying capacity. Large harvests with low edge/area ratios may have large effects on a population, despite the small amount of edge relative to the harvest size, because there is a large area far from forested habitat with few if any animals. Many small harvests may also have a large effect on the total population because of the high edge/area ratio. We applied this model to red-backed salamander captures along transects from harvests into forests and hypothesized that mid-sized harvests would have the least total effect on the landscape-scale salamander population.

Results/Conclusions

We found that red-backed salamander abundance reached 95% of carrying capacity 34 m into the forest. We also used this model to estimate the total impact of timber harvests on salamander populations depending on size and shape of the harvests. Assuming circular harvests, a 100 ha harvest reduces the population 20.8% over an area of 460 ha (22% of total area logged), whereas ten 10-ha harvests reduce the population by 21.8% within the same area, and 100 1-ha harvests reduce the population by 26.2%. In contrast to our hypothesis, the population decreases exponentially with decreasing harvest size, with very small harvests (< 1 ha) having the greatest effect. Previous estimates of total effects from harvesting are overly simplistic by just adding the effected edge area to the harvest area. This assumes one effect for the harvest and one for the edge. However, we know that abundance along disturbance gradients often exhibit incremental change. Using our formulation, practitioners can calculate the total effect of harvesting on populations rather than simply the maximal area effected to any degree. This can be done from repeated counts without needing to employ intensive mark-recapture methods.