Scale-dependent patterns of soil enzyme activity in a forested landscape

We evaluated the patterns of variation in the activity of four soil enzymes in oak forests soils at spatial scales from 10s of km to <3 m. The four enzymes (b-glucosidase, chitinase, phenol oxidase, and acid phosphatase) are specific for substrates that vary widely in lability or recalcitrance. Significant variations in enzyme activity were observed at the regional (among forested areas), topographic (along elevation gradients within a watershed), and single-tree (1 m upslope and downslope of an individual tree) scales, but not at the local scale (contiguous watersheds within forested areas). However, the specific patterns of variation in relation to spatial scale were unique to each enzyme system. Detrended correspondence analysis (DCA) ordination of the activity of the four enzymes in each soil sample suggested a strong nutrient availability gradient underlying these spatial scale differences. Exploratory path analysis produced relatively strong predictive models based on soil nutrients, organic matter, and moisture for all the individual enzymes except phenol oxidase. However, path analysis produced an even stronger model for the activity of all four enzymes together, using the DCA axis scores as the dependent variable. The results indicate that the four enzyme systems could help resolve spatial dependencies at a range of scales and could also be used to develop a scale-independent metric to be used for regional analyses in a geographic information system environment.