Impacts of land-use conversion in Sumatra, Indonesia on soil nitrogen cycling, soil nutrient stocks and ecosystem dynamics

Abstract

Over the last two decades, deforestation rates in Sumatra, Indonesia have rapidly increased resulting in the conversion of large tracts of lowland forest into monoculture plantations of oil palm (Elaeis guineensis) and rubber (Hevea brasiliensis). Land-use conversion to agricultural systems has been found to decrease soil nutrient stocks and soil nutrient cycling rates overtime, which can lead to a dependence on fertilization that only temporarily improves soil nutrient availability. Furthermore, conversion of forest to crop monocultures threatens the high levels of biodiversity present in tropical forested systems, which subsequently influences ecosystem functioning. The focus of this thesis was to determine the impacts of land-use conversion on soil nutrient status and ecosystem dynamics, as well as provide an understanding of the mechanisms driving these changes. All three studies were a part of a large interdisciplinary research project examining the environmental and social effects of tropical land-use change. Sampling for each study took place in Jambi Province, Sumatra, Indonesia—an area that was once heavily forested, but has experienced high forest conversion. Two soil landscapes, defined by their dominant soil texture and type, were selected to represent the region: loam and clay Acrisol soils. In each soil landscape, four land-use systems were examined: lowland rainforest and rubber interspersed in naturally regenerating forest (referred here as “jungle rubber”) and monoculture plantations of rubber (7-17 years old) and oil palm (9-16 years old). The aim of the first study was to assess changes in soil nitrogen (N) cycling rates with conversion of forest to oil palm and rubber plantations. Gross soil-N cycling rates were measured using the 15N pool dilution technique with in-situ incubation of soil cores. In the loam Acrisol soil, where fertility was low, microbial biomass, gross N mineralization and ammonium (NH4+) immobilization were also low and no significant changes were detected with land-use conversion. The clay Acrisol soil, which had higher initial fertility based on the reference land uses had larger microbial biomass and NH4+ transformation rates compared to the loam Acrisol soil. Conversion of forest and jungle rubber to rubber and oil palm in the clay Acrisol soil decreased soil fertility subsequently reducing microbial biomass and decreasing NH4+ transformation rates. Our findings suggest that the larger the initial soil fertility and N availability, the larger the reductions upon land-use conversion. The aim of the second study was to assess changes in soil biochemical characteristics and soil nutrient stocks down to 2 m depth with land-use change, and to determine the proportions of overall variance of soil biochemical characteristics that were accounted by the spatial components within our nested experimental design. Clay content influenced soil fertility and the higher nutrient stocks were found in the clay Acrisol reference land uses. Management practices in the converted land uses exerted the strongest influences on soil pH, base saturation, extractable phosphorus and exchangeable sodium. The majority of the soil biochemical characteristics and nutrient stocks did not exhibit significant effects of land-use change. Based on variance components analysis on the nested spatial structure of our experimental design, the overall variance on many of the soil biochemical characteristics was accounted by the variation amongst replicate plots rather than by land-use types. These results indicated that in order to detect significant effects of land-use change on soil biochemical characteristics in our nested experimental design, more replicate plots per land-use type should be sampled. The aim of the third study was to differentiate direct land-use effects from indirect bottom-up effects on below- and aboveground taxa. Generalized multilevel path models (a form of structural equation modeling) that allowed for direct and interactive effects of land-use with abiotic variables and bottom-up effects among biotic variables were constructed using data collected on plants, microorganisms, litter invertebrates, arboreal ants, birds and environmental parameters (soil and microclimatic properties). Results from the path models demonstrated that land-use change imposed direct effects on plants, belowground taxa at lower trophic levels (i.e., detritivores and herbivores) and arboreal ants, but almost all land-use impacts at the highest trophic levels of invertebrates and birds were bottom-up controlled. This study revealed that land-use change directly and indirectly drives large-scale ecological shifts, but the effects detected at the highest trophic levels were mostly dependent on lower trophic-level organisms. The soil-N cycling rates and N pools measured in the first study were combined with parallel studies on N-oxide emissions and N leaching, to generate a more holistic picture of the general soil-N cycle in this converted landscape. Analysis on sample optimization was conducted on the soil biochemical characteristics in the top 0.5 m depth from the second study, to determine the minimum number of replicates per land-use type needed to detect significant differences between land uses within our experimental design. The soil components incorporated within the multilevel path models from the third study were extracted and direct relationships between these soil properties and ecosystem biodiversity and biomass were examined to better understand the role soil nutrient status plays within these transforming systems. Overall, the results from these three studies illustrate that soil nutrient status is an important ecosystem component, and changes in soil nutrient status due to land-use conversion can potentially affect biodiversity and ecosystem functioning.