Modeling temperature gradients across edges over time in a managed landscape

Landscape management requires an understanding of the distribution of habit at patches in space and time. Regions of edge influence can form dominant c omponents of both managed and naturally patchy ecosystems. However, the bou ndaries of these regions are spatially and temporally dynamic. Further, are as of edge influence can be defined by either biotic (e.g. overstory cover) vs. abiotic (e.g. microclimate) characteristics, or structural (e.g. veget ation height) vs. functional (e.g. decomposition rates) features. Edges def ined by different characteristics are not always concordant; the degree of spatial concurrence varies with time. Thus, edge effects are difficult to g eneralize or quantify across a landscape. We examined temperature at eight times of the day across the edge between a clearing and a 50-year-old pine stand. We used simple, nonlinear equations to model and predict temperature gradients across this edge over time. The depth of edge influence (DEI) on temperature varied from 0 to 40 m, depending on the patch type and time of day. Two equations were required to model adequately (r(2)>0.50) patterns of temperature at all eight times of die day. Model fit was best at night ( r(2)=0.97) and lowest in the afternoon (r(2)=0.50). Parameters for the mode ls could be predicted from local, reference weather conditions. However, th ese linear relationships varied among parameters and with time of day (0.29 less than or equal to r(2)less than or equal to 0.99). Model validation wa s weak, with mean absolute percent error >10% for all day-time combinations . The models tended to underestimate DEI for both patch types, though edge depth was more accurately predicted in the closed-canopy stand than in the clearing. The difference between observed and predicted edge effects was hi ghest at midday in the clearing and during the morning under closed canopy. The models predicted the location of peak temperature and the slope of tem perature change (i.e. pattern of temperature variation) across the edge and the range of temperature better than actual values. We suggest that this a pproach may, therefore, be useful for characterizing edge dynamics if a wid er range of local weather conditions could be monitored during initial data collection. The empirical evidence for temporal changes in position and in tensity of abiotic edge effects emphasized the need to quantify these dynam ics across time and space for sound planning at the landscape scale. (C) 19 99 Elsevier Science B.V. All rights reserved.