Reports by previous Varley Gradwell Fellows


Sarah Benyon: 'Dung beetles (Scarabaeinae): Diel flight activity, response to bait size, bait type, and habitat structure in central Zambia'


Robert M. Ewers, Institute of Zoology and University of Cambridge: 'Why do similar species have opposite responses to habitat edges?'

This project represented what I believe is the first experimental manipulation of species to try and elucidate the mechanisms underlying species responses to habitat edges. The study was carried out at the Hope River Forest Fragmentation Project in the Southern Alps of New Zealand, where myself and Raphael Didham have an extensive dataset detailing the response of ca.900 beetle species to forest-grassland edges.

I focused my study on a pair of Mecodema species in the family Carabidae; M. rugiceps and M. fulgidum. These species are relatively abundant, morphologically very similar, but have contrasting response to habitat edges. M. rugiceps is restricted to forest, and at the site I selected for experimental work, was restricted to deep forest (never encountered closer tlian 250 m to the forest edge). By contrast, M. fulgidum is a habitat generalist that is present in both the forest and in the surrounding grassland matrix.

I was able to perform the translocations as planned and described in my proposal. However, the results were not at all what I had expected. In large part, I believe this is because New Zealand experienced one of the coldest summers on record. This project was designed primarily to detect differential mortality rates between the two species across a forest-grassland edge, with the expectation that the two species would vary in their ability to survive the variation in microclimatic conditions across the edge. The study was planned and executed in the austral summer (December 2006 - April 2007) when conditions are hottest and standing microclimate gradients are at their strongest. But the persistent cold weather during the summer resulted in low, or even completely absent, standing microclimatic gradients (analogous to those expected during winter months). To work around this problem I extended my initial planned fieldwork period of one month, to four and a half months, waiting and hoping for settled summer conditions to arrive, but was mostly disappointed.

The results from my planned experiment are not striking, but there are some interesting patterns; there was significant mortality in the forest specialist M. rugiceps but not for the habitat generalist M. fuligdum However, there was no significant correlation between mortality rate and distance to edge for M. rugiceps. The mortality occurred almost exclusively at the very end of the summer, when conditions were at their hottest. To examine this trend in more detail, I was able to purchase thirty years of climate data from a nearby weather station. Although not yet formally analysed, the data indicate that mortality began to occur when soil moisture became much dryer than average at the very end of my field season. I am investigating whether this trend will be strong enough to justify publication. If nothing more, these data serve as a strong indication that M. rutgiceps, which is weakly water-loving and burrows underground during the day, may have its spatial distribution limited by seasonal changes in soil moisture. Although the patterns I have observed are not what

I had originally set out to detect (temporal rather than spatial variation in mortality rate), they do lead to the specific hypothesis that temporal variation in soil moisture is one ecological mechanism that may underlie species responses to habitat edges. Generating an explicit hypothesis such as this was one of the primary purposes of this experiment.


Alex Hayward


Dr. Eleanor Slade, Dept of Zoology, Oxford: 'Effects of habitat modification and fragmentation on dung beetle biodiversity and associated ecosystem functioning' 

Increasingly, secondary forests, plantations and other human-modified habitats will dominate tropical landscapes and there is an urgent need to assess the capacity of these habitats to maintain biodiversity and ecosystem functioning in the long-term. Dung beetles provide an ideal focal guild for such studies as they are sensitive to habitat degradation and fragmentation and play a pivotal and easily-quantified role in tropical forests as nutrient recyclers and secondary seed dispersers.

This study aims to answer two questions: 1) How does the diversity and composition of dung beetle communities differ across habitats that have been modified to different degrees (primary forests, secondary forest of different fragment sizes, and oil palm plantations)? 2) How do rates of dung processing differ across these habitats and how do they relate to any observed biodiversity changes?

Study site and Methods

The study was conducted in Sabah, Malaysian Borneo where there is a gradient of forest types around Maliau Basin that are being intensively studied as part of the Stability of Altered Forest Ecosystems (SAFE) project ( In July 2011 the secondary forest landscape will be experimentally fragmented by oil palm plantations, generating one of the world’s largest ecological experiments to quantify changes in ecological processes across a gradient of habitat modification and fragment The SAFE project has a quasi-experimental design with large-scale spatial replication of sites in each of three forest types: oil palm (OP), logged forest (LF) and old growth forest (OG), as well as the experimental fragmentation plots. The fragmentation of the forest by oil palm follows a split-plot design with six experimental blocks (A-F), each containing four plots: (1) 1 x 100 ha fragment, (2) 2 x 10 ha fragments, (3) 4 x 1 ha fragments; and (4) forest that will be converted to oil palm. Across all blocks, there will be a total of 42 fragments. Within the blocks plots have been marked out following a fractal design, which allows data to be studied across different spatial scales. The basic design consists of sets of three sampling points arranged in a triangle.

Fieldwork was carried out during February-March 2011 when dung beetle biomass, diversity and function peaks. I sampled the dung beetle community and the associated ecosystem function of dung removal at the second order points, a distance of 178 m between points, as dung beetles are relatively mobile and so this ensures independence between samples. Therefore, 27 points were sampled in each of the forest types (OG, LF, OP) and 16 points in each of the fragment blocks (A-F)

Results: Dung beetle diversity and composition

At each second order point a standard dung-baited pitfall trap was placed. Traps were baited with 25g human dung and beetles collected after 48 hours (Figure 4). The beetles were stored in 75% alcohol in a freezer and are now in the Oxford University Museum of Natural History (OUMNH) awaiting identification. All the necessary permits to carry out the research within Sabah and the SAFE/Maliau sites and export and import permits were obtained. I visited Dr Arthur Chung, my local collaborator, and a reference collection will be deposited in the FRC, Sandakan and OUMNH at the end of the project.

Large numbers of beetles were collected across all sites and in total nearly 10kg of beetles have been brought back for identification. Identification of the samples will begin this month and should be completed by December 2011. The dung beetle community diversity, biomass and composition will then be compared between the sites to assess changes across the land use gradient from old growth forest through secondary forest to oil palm. The data from the fragment plots also serves as the crucial baseline data before fragmentation begins, allowing a comparison of pre-fragmentation communities across a connected gradient of primary forest, logged forest and oil palm plantation and in the future will provide the baseline to map post-fragmentation community disassembly and associated ecosystem functioning changes over time across the different fragment sizes and isolations.

Results: Dung removal

Rates of dung removal were measured by monitoring standard-volume dung piles placed in the field. Fresh cattle dung was collected from a local oil palm plantation, homogenised and frozen for 24 hours to kill any invertebrates. Piles were placed at each of the second order points, with a minimum time between the pitfall trapping and dung removal experiment of two weeks. the dung removal experiment. 700g of dung was placed in the field with a plate as a rain-shield The dung was collected after 24 hours and re-weighed. Three moisture control piles of dung, from which all beetles were excluded using a fine mesh, were placed in each of the sites. This allowed the mass loss due to evaporation or mass gain due to excess rain to be accounted for.This prevented interference between the two components of the study, but was close enough in time that the dung beetle assemblages caught during trapping would be similar to those present during the dung removal experiment. 700g of dung was placed in the field with a plate as a rain-shield. The dung was collected after 24 hours and re-weighed. Three moisture control piles of dung, from which all beetles were excluded using a fine mesh, were placed in each of the sites. This allowed the mass loss due to evaporation or mass gain due to excess rain to be accounted for.

The preliminary results suggest that old growth forest has the highest rates of dung removal, and oil palm the lowest. However, there is a high degree of variability between both OG and OP sites. In the case of the OG forest this may be due to local differencesin soil types. Interestingly, OP3, which has the highest rate of dung removal is an older oil palm site, established in 2000 (OP1 and OP2 were established in 2006), and as such it has a much higher and denser canopy. In the logged forest blocks that are to be fragmented (A-F) there is a lot of variability in dung removal; block A has as much dung removal as old growth forest, while blocks D-F have similar dung removal to oil palm.

Once the dung beetle samples have been sorted and identified I will relate the dung beetle diversity, biomass and composition in each of the sites to the dung removal. Vegetation, soil and micro-habitat data exists for each of the second order points and this will also be examined to see if these may account for differences between sites in dung beetle diversity and composition and amount of dung removal.

Conclusions so far

The study has collected crucial baseline data on dung beetle community composition and associated ecosystem processes before fragmentation begins, allowing a comparison of pre-fragmentation communities across a connected gradient of primary forest, logged forest and oil palm plantation, and in the future will provide the baseline to map post-fragmentation community changes. The chance to collect ‘before and after data’ is rare and there are few studies which have been able to do this, and none from South-East Asia, thus this project provides a unique and exceptional opportunity to enhance our understanding of the consequences of habitat modification and fragmentation and the implications for forest management and conservation. While the results presented so far are very preliminary they suggest that even logged forest, in which most of the large dipterocarp trees have been removed, can maintain a high degree of ecosystem functioning, in some cases as high as in old growth forest. Moreover, although oil palm clearly has lower ecosystem functioning than old growth forest, these results suggest than some functioning is retained and that 10 year old oil palm, which has a higher canopy layer and superficially resembles a forest, can maintain levels of functioning similar to that of some logged forest plots.


Dr Tom Fayle, Imperial College, London

Habitat fragmentation and the functioning of ant communities in tropical forests The world’s natural habitats are becoming degraded and fragmented, with resulting losses of species, and predicted changes in ecosystem functioning. I was awarded the Varley-Gradwell Travelling Fellowship in Insect Ecology in May 2011, to support an investigation into the way that rates of ant-mediated nitrogen pre-emption change along a gradient of habitat disturbance in Sabah, Malaysia. This gradient comprised old growth lowland dipterocarp rain forest, twice-logged rain forest, and oil palm plantation. Furthermore, collaboration with the SAFE project (Stability of Altered Forest Ecosystems), enabled me to sample in areas of continuous forest that were destined to become fragments of a range of sizes.

During August-September 2011 I conducted pre-fragmentation N pre-emption assays at sampling points across the habitat disturbance gradient. The protocol for the assays was altered slightly from that detailed in my original proposal. In order to understand how heterogeneity in resource distribution affected redistribution rates, and furthermore, how utilisation of a heterogeneous resource changes with disturbance, I used a range of earthworm bait pellet sizes, rather than the single size originally proposed (see below). These were arrayed in a 6 by 5 grid on a laminated card at each site. We conducted over 300 assays, each one lasting for 40 minutes (12,000 minutes of observations) at the sites of the pitfall traps. I have trained three SAFE research assistants in the assay protocol, which will greatly facilitate future work. Two other RAs have also been trained in (lab-based) ant identification. In the process of this training I updated the current key to ant genera for Sabah, and a colleague of mine translated it into Malay. This key is now freely available online and we hope to further refine it in the future (

Entry of the remaining pre-fragmentation removal rate data, identification of the 15% remaining pre-fragmentation ant samples, and full post-fragmentation field assays (I am now in Sabah) are all currently on-going. However, preliminary analyses of a subset of the pre-fragmentation data show that N pre-emption is faster in less disturbed forest, and that this difference is greatest for larger pellet sizes. One possible explanation for this pattern is that less disturbed forest supports higher densities of ant species with large body sizes (e.g. the giant forest ant, Camponotus gigas, which is restricted to old growth forest). If these larger ants preferentially remove larger pellets (which seems likely, given previous work on food item-body size matching in ants (J. Trop. Ecol. 22: 685-693), and can be tested later using my data), then their loss is predicted to impact the size distribution of removed pellets in the manner observed. If this response is also detected in the full dataset, then we will have made some progress towards understanding the links between habitat disturbance, resource heterogeneity and ecosystem functioning.

I would like to take this opportunity to thank the board of the fund for the award, which has enabled me to expand my research in this new direction. In particular, being able to make pre-fragmentation assays has been invaluable, since these will form the baseline for future sampling rounds. The N pre-emption assay work is also forming the basis for a grant application I am making as a PI to the Grant Agency of the Czech Republic (GACR), with a view to understanding how this redistribution of nutrients affects plant communities.


Dr Richard Merrill, University of Cambridge: 'Ecological speciation in tropical butterflies'

Speciation is the process ultimately responsible for generating biodiversity, and understanding its mechanisms remains a fundamental goal. I was awarded the Varly-Gradwell Traveling Fellowship in Insect Ecology in May 2012 to support my work on mimetic Heliconius butterflies. My research focuses on the ecological, genetic and developmental basis of adaptive behaviours that contribute to the evolution of new species. In particular, I am interested in how behavioural isolation evolves, and how genetic architecture and other factors may influence this process.

It is now widely accepted that adaptation to different ecological niches can drive speciation. However, if populations remain in contact, speciation is constrained because recombination breaks down associations between key traits that characterize emerging species. A number of mechanisms may overcome this by ‘coupling’ traits involved in speciation, including for example, pleiotropy and genetic linkage. In Heliconius butterflies colour pattern is both used to warn predators and as a mate recognition cue. I have previously demonstrated that loci underlying differences in colour pattern and the corresponding mate preference, between the sister species H. cydno and H. melpomene, are physically associated in the genome. Thus individuals that have the same colour pattern genes, which are under disruptive selection for mimicry, are strongly predisposed to mate with one another.

Despite this, we are largely ignorant of the genetic structures underlying these associations. I now am addressing this through multiple approaches. In 2012 and ‘13, supported by the Varly-Gradwell Traveling Fellowship, I spent over 4 months in Panama collecting tissue from H. melpomene and H. cydno across a developmental time course. In total, I collected eye and brain tissue from 226 individuals, including various pupal and adult stages, raised in a common garden. I have now extracted RNA from these samples and a subset of these are awaiting sequencing at a genomics facility in Edinburgh. This will produce vast amounts of data allowing me to examine gene expression differences both across the colour/preference candidate region and the larger transcriptome. This will first allow me to ask whether there a correlated difference in expression between genes associated with mimetic wing pattern shifts and expression in eye and brain tissue. Second, I will be able to combine the resulting expression data with quantitative trait loci analyses. This will hopefully lead to the identification of candidate genes underlying divergent mate preferences in Heliconius. The project is ongoing and I am still awaiting the sequencing data. However, I am confident that this work will produce interesting results and will be published. (Resulting publications will be available online:

I would very much like to thank the board of the fund of the award, which has supported a key aspect of this project. My current position has no associated funds for research, and so funding from the Varley-Gradwell fellowship has been particularly valuable.


Insect pre-dispersal seed predation across a tropical rainfall and plant species richness gradient
Tropical forests support an exceptional richness of flowering plants, many with over 250 tree species per hectare1. But how are so many ecologically similar species competing for the same limiting resources (light, water, nutrients) able to coexist? The Janzen-Connell mechanism proposes that negative density-dependent mortality mediated by natural enemies, such as pathogens and herbivores, prevents a small number of dominant competitors from monopolising space. This prevents the exclusion of weaker and rarer species, thus enhancing diversity2,3,4.

Recent survey work in the lowland tropical forest on Barro Colorado Island, Panama, has shown that insect seed predators consuming seeds pre-dispersal show high levels of host-specificity5. Specialisation is an essential factor in the Janzen-Connell mechanism as it ensures the creation of enemy-free space for other, rarer species to occupy. However, pre-dispersal seed predators are rarely studied, and assessments of their specificity across larger spatial scales and climatic gradients are unknown.

With evidence that pathogen and herbivore activity increases positively with humidity6,7, it is possible that globally observed positive correlations between rainfall and species richness8,9 may be driven by more intense natural enemy pressure as rainfall  increases. The 80 km stretch of rainforest spanning the Isthmus of Panama demonstrates such a steep increase in both humidity and species richness when moving from the drier Pacific to the more humid Atlantic coast10
Between March and June 2014 I travelled to Panama to investigate how seed predators may contribute to the structuring of diverse tropical forest communities. Conducting experiments that contributed to work at Oxford  University led by Dr Owen Lewis, Dr Lars Markesteijn and Patrick Kennedy, we assessed whether:

1) Mean pre-dispersal seed predation rates increase with humidity
2) Prevalence and strength of enemy mediated negative density-dependent mortality increases with humidity
3) Host-specificity of pre-dispersal insect seed predators increases with humidity

Freshly abscised seeds were collected from target tree species and families at eight 1 ha mapped forest plots spanning the rainfall and species richness gradient in order to rear seed predators developing within them. Emerging adults and larvae were pinned or stored in alcohol for later morphological identification and/or DNA barcoding to allow the construction of a plant-predator food web to assess specificity variation across the gradient. After a three month rearing period, seeds were dissected to estimate mean predation rates for focal plant species, families and sites spanning the rainfall gradient.

By the completion of the 2014 field season, 33,393 seed dispersal units were collected from 51 plant species, with a total of 47 seed predator morphospecies reared. It now remains to have all insect material identified, but preliminary assessments appear to corroborate evidence of high host-specificity in pre-dispersal insect seed predator food webs, with a single host plant for 91.5% of morphospecies. Cross-gradient analyses of specificity and predation rates remain to be conducted, but hopefully will yield exciting results.

I would like to thank the board of the Varley-Gradwell fellowship for supporting my research, and would strongly encourage others to apply. I am very grateful for the experience this funding has given me.

1. Valencia R, et al. in Tropical Forest Diversity and Dynamism (eds. Losos, EC, et al.) (University of Chicago Press, 2004).
2. Harms, K.E., et al. Nature 404, 493-495 (2000).
3. Petermann, J.S., et al. Ecology 89, 2399-2406 (2008).
4. Comita, L.S., et al. Science 329, 330-332 (2010).
5. Gripenberg, S., unpublished data.
6. Coley, P.D. Clim. Change 39, 455-472 (1998).

7. Brenes-Arguedas, T., et al. Ecology 90, 1751-1761 (2009).
8. Phillips, O.L., et al. Proc. Natl. Acad. Sci. U. S. A. 91, 2805-2809 (1994).

9. Kreft, H., et al. Proceedings of the National Academy of Sciences 104, 5925-5930 (2007).
10. Engelbrecht, B.M.J., et al. Nature 447, 80-82 (2007).