Conversion of forests to pasture has been extensive throughout the tropics and subtropics, and these forests continue to experience deforestation at high levels. Such clearing can alter plant-animal dynamics and thereby degrade ecosystem functioning. It is often assumed that re-establishment of vegetation will recover forest-like communities of animals such as invertebrates, together with the roles they play in important ecosystem processes. However, there has so far been little empirical evidence to support this. This thesis investigated the impact of deforestation on invertebrate-mediated ecosystem functioning, and its recovery during regeneration of rainforests on former pasture. Specifically, it focused on two key ecosystem processes which underly forest regeneration dynamics, while also comparing two different reforestation pathways. The two processes were leaf litter decomposition rates and insect seedling herbivory. For decomposition, functional associations with decomposer invertebrates as well as vegetation structure and floristics were also investigated. Both these ecosystem processes were examined across 25 sites representing different phases of rainforest deforestation and subsequent reforestation.
The study area was the Numinbah Conservation Area and surrounds on the Gold Coast Hinterland, in the eastern Australian subtropics. Here the local council had begun restoration management on disused pasture approximately 10 years prior to the beginning of this thesis. Five study sites were placed in each of five vegetation types, consisting of two reference states of grazed pasture and old-growth rainforest (to represent the deforestation process and to indicate the target of restoration), together with three types of regenerating vegetation: unassisted woody regrowth aged 20-50 years on former pasture, and assisted regeneration aged 1-4 and 5-11 years after human interventions that aimed to accelerate natural regeneration within previously-unassisted regrowth. To measure decomposition rates (mass loss % per unit time), open and closed mesh litter bags (to exclude macro-invertebrates >1mm length and width) were deployed across the 25 sites filled with leaves of five local tree species for five and eight months. Most of the decomposition had already occurred by five months, as only a further 3% decomposition occurred in the following three months. Macro-invertebrates did contribute to mass loss, but their exclusion from litter bags only marginally decreased decomposition rates, by 3%. Decomposer community composition at a broad taxonomic level differed strongly between pasture and all other vegetation types. Decomposition rates in open bags were about 50% slower in pasture than in old-growth forest, but had recovered to 83% of old-growth values in all types of regenerating vegetation. Samples of decomposer invertebrates were also collected from ground litter within each of the 25 sites. Abundances of macro- and meso-decomposer invertebrates decreased by 95% and 77% respectively in pasture. Abundances of invertebrate decomposers had recovered to old-growth forest levels in all restoration types of recovering vegetation. However, decomposition rates in open-bags were significantly correlated (across sites) with abundances of both macro- and meso-decomposers, most strongly so for meso-decomposers.
Investigations were also made into how vegetation structure and floristic composition of trees differed among the five vegetation types, and associations between these differences and variations in decomposition differed among sites. The main changes in vegetation were: (1) after 5-10 years of interventions to assist natural regeneration, abundance and species richness of trees (>1 m height) with stem sizes <10 cm dbh and 10-50 cm dbh had recovered, (2) also after 5-10 years of interventions to assist natural regeneration, canopy cover had recovered, and (3) tree species composition (all stem sizes) differed between old-growth and regenerating areas, and between unassisted and assisted natural regeneration (respectively dominated by Lantana camara and various pioneer tree species). There was a large step-change in decomposition rates associated with the transition from pasture to regenerating woody vegetation (33% difference), during which canopy cover (from 0% in pasture to 51-69% in regenerating areas) was both most important for describing differences in vegetation, and a good predictor of decomposition rates. A second step-change of decomposition rates was associated with the transition from regenerating woody vegetation (of all types) to old-growth rainforest (17% difference), during which both canopy cover and tree species composition were most important in describing differences in vegetation. However, canopy cover no longer explained variation in decomposition rates among woody regeneration areas and old-growth rainforest; in these cases, tree species composition was the better predictor.
To investigate seedling herbivory, 200 seedlings of two tree species (100 Glochidion ferdinandi and 100 Toona ciliata) were planted across the 25 sites (eight seedlings of each species per site), and leaf area loss (%) and the types of damage were recorded on immature and mature tagged leaves at four times during seven months. The percentage of leaf area lost due to herbivory accumulated in leaves over time for both species and at seven months was 25% and 37% respectively for G. ferdinandi and T. ciliata. There was no difference in the accumulated amount of herbivory between leaves which were immature or mature when first tagged, and there was no effect of vegetation type on herbivory rates in either species. Leaf start age (immature vs mature) also had no effect on the type of herbivory. Vegetation type did have an effect on the type of herbivory, which was mostly driven by differences in pasture, but more so for T. ciliata than G. ferdinandi. However, the damage types were driven by relatively uncommon types of leaf damage. On the other hand, the damage types which occurred most frequently, and caused the greatest amount of damage, occurred consistently throughout all the vegetation types. Most leaf damage was caused by various types of Lepidoptera larvae.
The results of this study showed that conversion of forest to pasture halved decomposition rates but had little effect on seedling herbivory. The reduction in decomposition was associated with an effect of deforestation on important drivers of decomposition rates, including the abundance and composition of decomposer invertebrates, and amount and composition of canopy cover. During regeneration, decomposition rates partially recovered, associated with a full recovery of decomposer invertebrates. However, a full recovery of decomposition rates was likely restricted by a difference in litter layer quality, because tree species composition in old-growth rainforest differed greatly from that in all types of regenerating vegetation. Seedling herbivory was unaffected by deforestation, most likely because adults of the most common type of insect herbivores, Lepidoptera, can detect suitable host plants through the volatile chemicals that they emit, and because they are also capable of flying the distances (<2 km) from source populations. Since seedling herbivory was little effected by deforestation, the concept of recovery during regeneration did not apply, and neither did it change during regeneration.
Overall, both types of invertebrate-mediated ecosystem functioning were resilient to deforestation as any decline in functioning quickly recovered (or partially recovered) during reforestation. These results occurred irrespective of whether interventions to assist natural regeneration were used. Decomposition rates had already substantially recovered before the woody regrowth reached 20-50 years of age, and seedling herbivory was unaffected by the types of vegetation within which the host plant was located. Therefore, it is unlikely that any types of interventions used to assist natural regeneration would have a significant effect on invertebrate-mediated ecosystem functioning subsequent to regeneration of regrowth aged 20-50 years. Further, the recovery of the invertebrate communities was likely facilitated by the proximity to an old-growth rainforest source, together with extensive surrounding forest cover of several types. Therefore, a reduced supply of source invertebrate populations in the landscapes that surround regenerating areas may, in poorly forested regions, lead to lower functional resilience. Nevertheless, in moderately forested landscapes (such as in the present study), simply removing barriers to natural regeneration may be all that is required to achieve successful recovery of invertebrate-mediated ecosystem functioning during regeneration.