Ecosystem functioning in fragmented rainforest

6.1 Abstract

Oil palm companies certified by the Roundtable on Sustainable Palm Oil aim to reduce biodiversity losses by protecting High Conservation Value (HCV) forest, but it is not known whether these forest fragments can

maintain ecosystem functions. I studied 18 forest sites (16 fragments, range 5-3,529 ha, and 2 continuous forest sites) in Sabah, Malaysia and assessed the impacts of fragmentation on ecosystem functions performed by dung beetles: dung removal, seed burial (by tunnellers), and seed dispersal (by rollers). There were only minor differences in dung removal, seed burial and seed dispersal between undisturbed and twice-logged continuous forest sites, but these ecosystem functions were at least 50% lower in forest fragments. This decline was largest for secondary seed dispersal with only 4/16 forest fragments having any seeds rolled at all, whereas 13/16

fragments had some dung removed or seeds buried. Among fragments, functions varied little in response to changing fragment area, forest quality or isolation, although dung removal, seed burial and secondary seed

dispersal were highest in larger fragments. These ecosystem functions are likely to have key roles in maintaining the viability of plant species

populations and so should be considered in HCV management plans.

Continuous forest sites maintained far higher rates of these ecosystem functions than did fragments, even when heavily degraded (twice-logged), and degraded continuous forest should be a high priority for conservation.

However, if only fragments can be maintained, then they need to be at least 100 ha in size to support key dung beetle functions.

6.2 Introduction

The detrimental impacts of deforestation and forest fragmentation on species richness and abundance are now well documented (Nichols et al.

2007; Gibson et al. 2011; Bregman et al. 2014). However, ecosystems consist of more than just static collections of species and individuals. They are dynamic systems reliant on interactions between species and the

transfer of resources and energy (Morris 2010). For example, processes such as seed dispersal and decomposition result from individual species

consuming resources, growing, reproducing and transferring energy (Reiss et al. 2009). These ecosystem processes are vital for the maintenance of the abiotic and biotic conditions essential for longer term species

persistence (Millennium Ecosystem Assessment 2005; Hooper et al. 2012), but species extinctions can lead to declines in key ecosystem processes (Hooper et al. 2012). Therefore, it is vital to understand how deforestation and forest fragmentation influence ecosystem functioning.

Sustainability initiatives such as the Roundtable on Sustainable Palm Oil (RSPO) aim to mitigate the negative environmental impacts of

deforestation and forest fragmentation by protecting HCV areas within plantations. However, methods for conserving forest areas with High Conservation Values (HCVs) do not explicitly consider ecosystem

functioning, except in cases where it provides crucial ecosystem services in critical situations, such as flood and erosion prevention (HCV 4, Brown et al. 2013a). This subset of ecosystem services overlooks the important supporting and regulating services/processes that maintain abiotic and biotic conditions essential for species persistence, which could influence the maintenance of HCVs over time (Millennium Ecosystem Assessment 2005; Hooper et al. 2012). HCV assessors cannot be expected to

comprehensively measure ecosystem functions during brief HCV

assessments (Meijaard & Sheil 2012, Chapter 2), and so research is needed to quantify how the size and quality of HCV areas affects the maintenance of ecosystem functions. This research can be used by HCV assessors to inform management recommendations that ensure HCVs are maintained over time.

HCV areas tend to be isolated forest fragments within oil palm plantations (Edwards et al. 2010, Chapter 4). Differences in forest fragment area, forest quality and isolation can alter ecosystem functions either through impacts on species that mediate the functions (Larsen et al. 2005a), or through changes in abiotic conditions that directly affect functions. Whilst changes in abiotic conditions could directly impact ecosystem functions such as leaf litter decomposition, reliant on chemical decomposition (Meentemeyer 1978), previous studies have shown little change in leaf litter decomposition rates following selective logging or habitat

fragmentation (Vasconcelos & Laurance 2005; Barlow et al. 2007).

However, abiotic changes can have strong indirect effects on ecosystem functions by altering the abundance and behaviour of functionally

important species (e.g. Doube 1990). For example, disturbed riverine and logged forest habitats show increased canopy openness and reduced heterogeneity of canopy habitats that support different assemblages of ants and dung beetles that are better able to tolerate these more exposed microhabitats (Davis et al. 1998; Klimes et al. 2012). Vegetation structure and abiotic conditions can be highly altered in forest fragments in response to edge effects (Laurance et al. 2002; Ewers & Didham 2007; Ewers et al.

2007) and human disturbances, such as logging, that increase canopy openness and so increase temperature and decrease humidity below the canopy (Hamer et al. 2003; Laurance et al. 2011; O’Brien et al. 2013).

Therefore, these abiotic changes in response to fragmentation and logging could alter provision of ecosystem functions.

In addition to impacts on microclimate and vegetation structure,

fragmentation affects species richness and abundance by reducing habitat availability and changing the structure of remaining habitat patches (i.e.

number/shape of patches, distance between patches) (Fahrig 2003; Ewers

& Didham 2006; Hanski et al. 2013). Habitat loss leads to local species extinctions in line with Species-Area Relationships (MacArthur & Wilson 1967), and altered patch network structure can change extinction and colonisation dynamics and reduce the viability of species’ populations (Hanski et al. 2013). How these changes in biodiversity affect ecosystem

changes in fragment area, forest quality and isolation. Studies of

functionally important dung beetles, birds, bees, ants and termites have reported decreasing species richness and abundance with declining fragment area (Chapter 5, Klein 1989; Laurance et al. 2002; de Souza &

Brown 2009; Hill et al. 2011; Bregman et al. 2014). However, effects of these species declines on ecosystem functions such as pollination, dung and litter decomposition can only be inferred in the absence of direct

measurements of the functions (Reiss et al. 2009). Slade et al (2011) reported reduced dung and seed removal following high-intensity commercial selective logging, but measurements of the response of multiple ecosystem functions to disturbance and fragmentation gradients are lacking (Loreau et al. 2001; Peh & Lewis 2012).

Dung beetles have critical roles in nutrient cycling and secondary seed dispersal, suppression of mammalian parasites and bioturbation (the mixing of soil and dung particles) (Nichols et al. 2008), making them an excellent taxon for measuring multiple ecosystem functions. Nutrient cycling and secondary seed dispersal influence soil fertility, plant productivity, seedling survival and plant composition and so are likely to be particularly important for overall ecosystem functioning (Nichols et al. 2008). Dung beetles

separate into different functional groups in relation to their nesting behaviour, with tunnellers burying dung directly under the dung pile, rollers moving dung horizontally away from the dung pile and dwellers using dung in situ in the dung pile (Hanski & Cambefort 1991). These functional groups contribute to different ecosystem functions, with

tunnellers being important for dung burial (nutrient cycling) and rollers for secondary seed dispersal (Estrada & Coates-Estrada 1991; Andresen & Feer 2005; Larsen et al. 2005a; Slade et al. 2007b). Dwellers contribute little to dung or seed removal (Slade et al. 2007b). In roller and tunneller groups larger species are especially important for these functions (Slade et al.

2007b; Dangles et al. 2012). Tunnellers’ importance for dung burial is likely to be because they are the most speciose and abundant functional group (~85% of total dung beetle biomass and species richness in continuous forest), and because the group contains some disproportionately efficient

2007). Rollers remove fewer seeds in total than tunnellers, but by moving seeds horizontally away from the dung pile where seed density is high, rollers may reduce density-dependent seedling mortality and, by burying seeds at shallow depths, rollers may promote seedling germination

(Andresen & Feer 2005; Nichols et al. 2008). Reducing density-dependent seed and seedling mortality is crucial for seedling recruitment and the maintenance of seedling diversity (Bagchi et al. 2014). However, secondary seed dispersal by rollers has been largely overlooked in previous studies recording just the proportion of seeds removed from a site without separating those moved by rollers and tunnellers (Andresen & Feer 2005;

Slade et al. 2007b, 2011). These separate measurements can also improve understanding of how different species’ responses to fragmentation can impact different ecosystem functions, given that rollers decline more in abundance that tunnellers following fragmentation (Chapter 5). To address this knowledge gap, seed burial (by tunnellers) and secondary seed

dispersal (by rollers) are measured separately in this chapter.

In Chapter 5, I showed there was little difference in large tunneller and roller biomass between twice-logged and pristine forest, but that biomass of both groups was significantly lower in forest fragments. These results suggest that there will be little difference in dung removal, seed burial or secondary seed dispersal between undisturbed and twice-logged continuous forest but that these ecosystems functions may be reduced in forest

fragments compared to continuous forest sites. Results from Chapter 5 showed that among forest fragments, tunneller biomass showed little further decline with decreasing area, implying that dung removal and seed burial may still be maintained in even the smallest fragments. By contrast, roller biomass declined with decreasing fragment area suggesting that horizontal seed dispersal is likely to decline with decreasing fragment size.

Previous studies have reported declines in seed dispersal and dung removal following forest fragmentation, but most research has been focussed in the Afro- and Neo-tropics in the context of small-scale, low-intensity

agriculture or cattle pasture (Andresen 2003; Chapman et al. 2003). Oil palm agriculture is one of the most dominant crops in tropical landscapes,

needed to assess how ecosystem functions are affected in oil palm

dominated landscapes. This research can inform the management of HCV areas in oil palm plantations, and more widely help to evaluate whether forest fragments in intensive agricultural landscapes can maintain key ecosystem functions.

I measured rates of dung removal, seed burial and horizontal seed dispersal in 16 forest fragments, one heavily disturbed (twice-logged) continuous forest site and one unlogged continuous forest site in Sabah, Malaysia. This allowed me to assess differences in dung removal, seed burial and

horizontal seed dispersal between unlogged and twice-logged continuous forest sites (Hypothesis 1), between continuous forest sites and forest fragments (Hypothesis 2), and among forest fragments in relation to fragment area, forest quality and isolation (Hypothesis 3). These hypotheses were as follows: