Discussion
Here, for the first time we quantitatively explored energy fluxes of canopy arthropod communities of tropical rainforest transformation systems. Conversion of rainforest into plantations drastically altered abundance, biomass and trophic structure of canopy arthropods, resulting not only in substantially decreased energy fluxes, but also in shifts in the relative importance of ecological functions such as algae-microbivory, herbivory and predation. The observed shifts may have far-reaching consequences for ecosystem functioning and underline the importance to counteract the rapidly accelerating land-use changes in tropical regions.
Total abundance and biomass of canopy arthropods strongly decreased from rainforest and jungle rubber towards monoculture plantation systems of rubber and oil palm, as has been shown previously for ants, spiders and parasitoid wasps (Nazarreta et al. 2020; Azhar et al.2022; Ramos et al. 2022). The decrease of combined average arthropod abundance across 12 major taxa by ca. 70 % from rainforest to monoculture plantations of rubber and oil palm was even more severe than reported for the abundance of soil invertebrates/arthropods at the same study sites of ca. 60 % (Potapov et al. 2019a), suggesting more pronounced effects of land-use changes on aboveground compared to belowground invertebrates, as has been shown for temperate ecosystems (Le Provost et al. 2021). Notably, differences in biomass of canopy arthropods varied between the two landscapes studied, with generally higher abundance and biomass in rainforests of Bukit Duabelas than in Harapan. This was mainly due to high biomasses of large herbivorous/algae-microbivorous Orthoptera and Blattodea in the former and resulted in a more drastic decrease in biomass of up to 87 % from rainforests to plantations in the Bukit Duabelas than the Harapan landscape (ca. 70 %). The regional differences are presumably due to differences in fertility and landscape heterogeneity, but in particular the longer history of rainforest transformation in the Harapan than the Bukit Duabelas landscape (Allen et al. 2015; Harrison & Swinfield 2015; Guillaume et al. 2018; Sibhatu et al.2022). By contrast, rubber and oil palm plantations featured similarly low abundance and biomass of canopy arthropods in both regions, indicating that regional differences are levelled out by strong disturbance such as the conversion of rainforest into plantation systems. This highlights the importance of protecting intact tropical ecosystems at landscape scale.
Taxonomic groups responded differentially to land-use changes. The abundance and biomass of most arthropod groups decreased strongly from rainforest towards plantation systems; loss of Collembola abundance was most pronounced, with a loss of > 85 % from rainforest and jungle rubber to oil palm plantations. Abundance of Formicidae was also reduced by ca. 75 % from rainforest to plantations and biomass of Blattodea even decreased by > 90 %. The strong reduction in abundance and biomass of these taxa may be related to reduced habitat complexity, aboveground plant biomass and diversity, and available food sources (Novotny et al. 2006; Kotowska et al. 2015; Drescher et al. 2016; Zemp et al. 2019). On the other hand, abundance and biomass of some herbivorous taxa, such as Lepidoptera and Curculionidae, were not reduced in oil palm plantations. This is likely due to specialized herbivore species that thrive in oil palm plantations, such as red palm weevils, bagworms and nettle caterpillars, and to the introduced curculionid pollinatorElaeidobius kamerunicus , which is the dominating curculionid species in oil palm plantations in our study region (Greathead 1983; Kasmiatun et al. 2022).
Land-use changes also altered the trophic structure of canopy arthropod communities. Biomass-weighted mean of trophic positions as indicated by Δ15N values were significantly higher in rubber plantations, indicating a higher proportion of predators. Low minimum Δ15N values, with δ15N values being in some cases more than 6 ‰ below those of canopy leaves, suggest that some canopy arthropod taxa, such as Psocoptera, Collembola and Blattodea, predominantly feed on algae or lichens which utilize atmospheric nitrogen and typically are depleted in δ15N compared to vascular plants (Chahartaghi et al. 2005; Erdmann et al. 2007; Maraun et al. 2011). Δ15N values of omnivores such as Hemiptera and Formicidae, but also of higher trophic level taxa such as Diptera, Staphylinidae and Araneae, were remarkably constant across land-use systems. This suggests that, despite high diversity within taxonomic groups (Basset 2001; Nazarreta et al. 2020; Ramos et al.2022), overall trophic niches and therefore ecological functions of these supra-specific taxa bear phylogenetic signal, which was also postulated for belowground systems (Potapov et al. 2019b). In contrast to the relatively low Δ13C values in plantation systems, in rainforest and jungle rubber canopy arthropods were on average 5-6 ‰ enriched in 13C compared to canopy leaves as potential basal resource; Δ13C maxima in rainforest were even close to 9 ‰ above those of leaves. Part of the13C enrichment may be attributed to variations in13C signatures between leaves high and low in the canopy, with the latter used for calibration in this study (van der Merwe & Medina 1991; Hyodo et al. 2010). An additional cause may be selective utilization of only specific plant/leaf components, such as water-soluble amino acids and carbohydrates that are more easily accessible to digestion and are typically enriched in13C compared to bulk tissue of leaves. Tropical leaves feature high amounts of structural and hard-to-digest compounds, such as lignin and cuticular waxes, that are more depleted in13C than other less condensed carbon compounds (Pollierer et al. 2009). Given the diversity of herbivore feeding strategies, including leaf chewing and mining, fruit eating as well as xylem/phloem sucking (Novotny et al. 2010), selective utilization of specific components is likely.
Despite relatively constant trophic levels of taxonomic groups, the biomass distribution among trophic levels, as indicated by Δ15N classes, differed significantly between land-use systems, but this also varied between landscapes. The shift from a close-to-normal biomass distribution in rainforests, in particular in Bukit Duabelas, to more uneven distributions in oil palm and rubber, suggests that, in line with our first hypothesis, land-use intensification decreases trophic redundancy and causes a more scattered distribution of trophic niches, possibly mitigated by reduced resource diversity (Krause et al. 2020). Low trophic redundancy may increase the vulnerability to the loss of trophic functions and to extinction cascades in intensively managed land-use systems (Sanderset al. 2018).
In line with our second hypothesis, energy fluxes were remarkably reduced by rainforest conversion into plantations. Total energy fluxes were reduced by up to 75 % in rubber and oil palm plantations compared to rainforest, and reduced by > 30 % compared to jungle rubber, mirroring the reductions in biomass. As energy fluxes can be used to infer ecosystem multifunctionality and stability (Barneset al. 2018; Manning et al. 2018; Potapov 2022), the strong reduction in energy fluxes in plantation systems points to the loss of ecosystem functions and potentially to the deterioration of ecosystem stability. Significantly lower total energy fluxes in plantations compared to rainforests contrast findings from soil food webs at our study sites, where the presence of earthworms as large detritivores in plantations counterbalanced declined energy fluxes in other trophic groups (Potapov et al. 2019a). However, energy fluxes to canopy arthropods were not only just reduced, but shifted among different trophic groups, such as herbivores, algae-microbivores and predators, representing different ecosystem functions. Importantly, shifts in energy fluxes and respective functions were similar, but not congruent with shifts in biomass among trophic groups. This highlights that energy flux approaches go beyond biomass-based approaches by also considering metabolic rates and assimilation efficiencies of consumers, as well as trophic structure and preferred prey, thereby more realistically reflecting community functioning.
In rainforests, algae-microbivory was the most important feeding strategy, followed by herbivory, while both of these feeding types were equally important in jungle rubber. Such high energy fluxes to algae-microbivores have not been demonstrated before for canopy arthropods and suggest that tropical canopy food webs more heavily rely on resources other than higher plants, with algae and microorganisms playing an even greater role than tissue of higher plants such as leaves. In contrast to vascular plant epiphytes, microbiota in tropical forest canopies are not well-studied (Nakamura et al. 2017). Microbial communities differ strongly with vertical stratification in tropical forests and are responsible for decomposition of suspended detrital substrates (Gora et al. 2019). Fungi, algae and lichens can significantly contribute to the diet of leaf litter and bark-inhabiting Oribatida and Collembola (Chahartaghi et al.2005; Erdmann et al. 2007; Susanti et al. 2019). Potentially, due to the low resource quality and high lignin and wax contents of leaves of tropical trees, these resources are of particular relevance for arthropods in the canopy of tropical rainforests. This calls for further studies to disentangle microbial contributions from fungi, bacteria and algae to the diet of canopy arthropods and how they are influenced by land-use change.
While rainforest and jungle rubber were characterized by pronounced energy fluxes to algae-microbivores and herbivores, predation was the main feeding type in rubber. Food webs with pronounced energy fluxes to lower trophic levels, i.e. bottom heavy food webs, as in rainforest and jungle rubber, are assumed to be more stable than those with higher fluxes to high trophic levels (Rip & McCann 2011; Barnes et al.2018). The strong reduction in energy fluxes to lower trophic levels and the concomitant shift to fluxes at higher trophic levels in rubber plantations are likely related to the conversion from evergreen to deciduous forest, causing temporal gaps in resource availability for herbivores and, due to faster replacement of leaves, lower algae-microbial colonization. As more energy is required to support high trophic levels, there is potentially an overall faster transfer and higher loss of energy along food chains in rubber plantations. This faster transfer may contribute to higher carbon and nutrient losses in rubber plantations (De Ruiter et al. 1993; Barnes et al.2018; Guillaume et al. 2018). On the other hand, a significantly higher ratio of herbivory compared to predation, as in oil palm plantations, may indicate reduced pest control and concomitantly higher losses of oil palm tissue to herbivore pests, necessitating higher use of pesticides (Corley & Tinker 2015).
Here, for the first time we documented the dramatic impacts of tropical land-use change on abundance, biomass, trophic structure and functioning of canopy arthropod communities. We showed that the conversion of rainforest into rubber and oil palm plantations not only strongly reduces energy fluxes, but also alters ecological functions, shifting from stable systems with high energy fluxes at lower trophic levels and high trophic redundancy in rainforest to more contrasting trophic niches in plantation systems. While oil palm plantations were characterized by a high ratio of herbivory to predation suggesting reduced pest control, predation was dominant in rubber plantations, providing the risk of higher energy and nutrient losses. Together, our findings indicate that the conversion of rainforest into plantation systems compromises ecosystem multifunctionality and stability, and suggest high sensitivity of rainforests to even low levels of disturbance such as selective logging. A combined food web approach, not only considering changes in abundance and biomass, but also in trophic structure and energy fluxes is promising to comprehensively trace changes in ecological functions due to land-use change and may provide a reliable basis to foster informed management decisions.