Rainfall partitioning in differently used montane rainforests of Central Sulawesi, Indonesia

Abstract

This study was conducted within the project 'Stability of Rainforest Margins in Indonesia' (STORMA) along the margins of the Lore Lindu National Park in Central Sulawesi, Indonesia. The objectives of this study were (i) to quantify differences in the structure of the investigated forests, which were subject to different forest use; (ii) to determine throughfall, stemflow and interception in those differently used forests and to relate observed patterns in rainfall partitioning also to structural properties of the forest stands; and (iii) to assess the vertical distribution of leaf wetness duration in the forest canopy in relation to temporal patterns of precipitation.Data on the influence of different forest use intensity on stand structure and rainfall partitioning with its temporal and spatial distribution were collected in the vicinity of the village Toro. There, stands of four forest types were analyzed which were under increasingly intensive management practices rooted in local traditions: (i) natural forest (NF), (ii) forest subject to small-diameter timber extraction (STE), (iii) forest subject to large-diameter timber extraction (LTE), and (iv) cacao agroforest (AF) under trees remaining from the natural forest. The study plots of 0.15nha each, with three replicates per use type, were situated between 800 to 1140 m asl within the lower montane rainforest. In the region annual rainfall ranged locally from 2437 to 3424 mm. In the natural forests, Meliaceae, Lauraceae, Sapotaceae, and Fagaceae dominated with decreasing number of species and endemics with use intensity, while cacao dominated the agroforest.Many forest structural parameters reflected the gradient of forest use intensity. Tree basal area (dbh ≥ 10 cm) decreased from relatively high median values in the NF to the AF, which was paralleled by a decrease in tree height. Canopy openness was lowest in STE (7%) and highest in cacao agroforest (16%). Estimated leaf area index (LAI), based on the analysis of hemispherical photos, reached a median of 6.4 m2 m 2 for NF, 5.2 for STE, 4.9 for LTE, and 5.7 for AF.Throughfall (Tf) reached a median of 70% of gross precipitation (Pg) over all NF plots, 79 and 80% for STE and LTE, respectively, and 81% for AF. Stemflow (Sf) was below 1% in all studied use types, although occasional palms provided exceptionally high yields. Thus, rainfall interception (I) was highest on NF plots where 30% of Pg (median) was re-evaporated back into the atmosphere. In a multiple linear regression, the combination of tree height and LAI explained 81% of the variance in interception. A possible reason for this tree height-LAI-interception relationship is that tall trees increase the vertical extension of foliage and other canopy components contributing to the canopy water storage, resulting in a higher canopy roughness and a more effective energy exchange with the atmosphere. Under the prevailing conditions in the study region, this would allow the canopy of taller stands to dry up faster between subsequent rainfall events and provide higher water storage for interception at relatively similar leaf area.A long duration of leaf wetness and insufficient drying of leaves was confirmed by evidence from the leaf wetness experiment. The canopy was wet during an average of 25 to 30% of time during a three month observation period. However, comparing extreme two-week scenarios within that period, surface wetness lasted still for 5% of the time in a relatively dry period, whereas the canopy was wet during 45 to 55% of the time in a rainy period. In the lower shade canopy, continuous surface wetness for periods of up to 22 hours and more existed frequently, although rainfall occurred only during afternoon thunderstorms of limited duration. The long duration of surface wetness has implications for forest interception models which commonly assume a complete drying of the canopy between consecutive rainfall events.