Does heterogeneity in regenerating secondary forests affect mean throughfall?

Abstract

As secondary tropical forests grow, their canopy structure and density change. This affects the canopy storage and aerodynamic roughness, and thus the amount of water that is lost to interception. Because interception is a considerable part of total evapotranspiration, it is important to assess how interception changes as secondary forests mature, and how this is affected by forest structure. However, the effects of tropical forest regeneration, and in particular changes in forest structure, on mean throughfall are so far poorly studied. This hampers the estimation of the interception loss, and thus the water balance, for regenerating forests. Therefore, we monitored throughfall for twelve regenerating, logged-over forest plots in Sabah, Malaysian Borneo over a 7-month period to determine the effects of forest regeneration on mean throughfall and tested if inclusion of measures of forest heterogeneity improves the prediction of mean throughfall compared to estimates based on tree height or density alone. Mean throughfall varied between 74% and 89% (average: 84%) of precipitation and was lowest in regenerating forest plots with a longer time since logging. There was a significant negative relationship between mean throughfall and tree density or basal area, as well as variables reflecting forest heterogeneity (i.e., the Shannon Diversity Index and the coefficient of variation of the diameter at breast height). Nevertheless, the inclusion of these indicators of heterogeneity did not improve model performance substantially; the best model was a linear relation with tree density alone. These results suggest that in the context of logged and regenerating forests in Sabah, mean throughfall depends mainly on tree density and is not substantially affected by species diversity or structural heterogeneity. To see if mean throughfall could be estimated over larger spatial scales based on LiDAR data, we also tested the relation between mean throughfall and LiDAR-derived Top of Canopy (TCH) but this relation was not significant for our study plots. A more in-depth analysis of LiDAR-products, such as point clouds, may be needed to estimate mean throughfall over large areas in tropical rainforests.