Scale-dependent relationships between landscape structure and microclimate

Abstract

This paper examined the scale-dependent relationships between landscape structure (e.g., slope, elevation, and overstory canopy coverage) and microclimate (e.g., air and soil temperatures and soil moisture) at different spatial scales along a 10050 m transect in the Southeastern Missouri Ozarks, USA. The landscape structure and microclimate variables were measured every 10 m along the transect during the growing season, June to September, 1996. We used the simple correlation analysis and "moving window" technique (no overlap between two adjacent windows) to examine the correlation coefficients between landscape structure and microclimate variables at scales or window sizes from 10 m to 2000 m. Because the sample size decreased rapidly in the "moving window" method, we also used the standardized cross-variogram to investigate the relationships between landscape structure and microclimate variables at a larger range of scales from 10 m to 8000 m. We found that the relationships between landscape structure and microclimate were apparently scale-dependent along the transect, suggesting the interactions between landscape structure and microclimate were stronger at some scales than others. The landscape structure variables were poor in explaining the variation of each microclimate variable at fine scales (e.g., 10 m). The correlations between elevation and microclimate variables were, in general, significantly improved with the increase of scales, while the improvement was less significant for slope and canopy coverage. Of the landscape structure variables, elevation, in general, had a higher correlation with the microclimate variables than slope and overstory canopy coverage at most scales examined. Our results suggest that small scales (e.g., 500 m) are more appropriate though the relationships vary at the larger scales. Both the simple correlation analysis and standardized cross-variogram analysis were effective and, in general, consistent in characterizing the scale-dependent relationships between landscape structure and microclimate. Meanwhile, the standardized cross-variogram had the advantage to examine the relationships at large scales over the correlation analysis because the sample size reduced rapidly in the correlation analysis.