The importance of vertebrates in regulating insect herbivory pressure along a gradient of logging intensity in Sabah, Borneo

Project Number 186 Back to projects

Start date


End date


Research areas


Data use

Masters Project


Top-down control of invertebrate populations by vertebrates is known to play a vital role in regulating herbivory rates, and can have important consequences for whole forest dynamics. However, the strength and relevance of this ecosystem service can vary greatly between different habitat types, and is expected to further change with human modification of natural landscapes. Rainforests in Borneo, Malaysia, are currently experiencing some of the highest rates of such modification on the globe, and as secondary and logged forests become an ever more common feature of these tropical landscapes, it is crucial to evaluate their conservation value before they are lost entirely. So far attempts to do this have focused chiefly on assessing the responses of specific taxa to logging, and whilst logging is now known to have a significant effect on the abundance and diversity of vertebrates and invertebrates alike, how this translates into the functioning of ecosystem processes such as herbivory is only recently receiving more attention. Recent studies in Borneo found that most invertebrate functional groups that contribute to seed dispersal and invertebrate predation decline in abundance in logged relative to primary forest. However, the ecosystem processes themselves remain quite resilient, thanks to a concurrent and compensating increase in the abundance of vertebrate taxa with functionally similar traits. Considering these findings, and the fact that herbivorous invertebrates had been found to increase in abundance in response to logging, I expect the importance of vertebrates in regulating insect herbivory to increase from primary to logged sites. I aim to test this prediction experimentally, by setting-up vertebrate exclusion plots to determine the effect they have both on herbivory rates and on the abundance and composition of arthropod communities within them, and how this effect varies between sites along a gradient of forest quality.


Study design:

The study will be conducted in East Sabah, Malaysia, on the island of Borneo. Data will be collected from 5 locations spanning a gradient of forest quality (based on above ground biomass measures): from old growth forest at the Maliau Basin Conservation, to varying quality patches of logged forest within the experimental landscape of the SAFE project (Ewers et al., 2011).

Experimental setup and data collection:

Each experimental plot will compare two treatments:

1. Vertebrate exclusion: 2x2x1.5m cages with 0.2cm mesh (allowing for the entry of invertebrates) wrapped around a fixed framework of metal poles.

2. Control: no exclusion, a 2x2m area delineated by a nylon thread stretching around 4 poles.

A procedural control will not be included, based on the results of previous studies which showed cage material itself did not affect herbivory rates (Koh et al., 2008).

Within each treatment, 10 saplings/seedlings will be chosen for the assessment of herbivory rates. Eight newly emergent leaves will be marked and photographed on each seedling, and at the end of three weeks (the time-frame within which shade tolerant tropical species have been found to experience 70% of total herbivory (Coley and Barone, 1996)), they will be photographed again and then changed to new leaves. ImageJ software will be used to quantify herbivory rates as the average percentage of leaf area lost per seedling per day. For consistent measures of herbivory across sites, I will attempt as much as possible to select plant species based on their common occurrence at sites along the gradient.

Each treatment will also be used to assess invertebrate predation rate using the sampling protocol from Michael Boyle’s exploratory project at SAFE (2012). A live mealworm is placed as bait on an acrylic “leaf”, and its predation is recorded over a 24 hour period. This way data collected will be a direct extension of previous research in the area.

Information regarding other explanatory variables relevant to this project such as the composition of vertebrate communities, or environmental variables, are already available for many sites within the SAFE framework, and the positioning of data collection points will be selected to maximize overlap with these.


Invertebrates sampled over the 4 month period will be categorized into functional groups based on feeding guild in order to aid comparison between invertebrate community responses to exclusion.

I will fit linear regression models to assess the relationship between the effect size (on herbivory, predation and invertebrate community measures) of vertebrate exclusion as the response variable, and forest quality as the predictor variable. Once basic relationships are identified, I will conduct a series of mixed effect analyses to include interactions with other potential explanatory variables (environmental and biodiversity related). I will also use path models to assess changes in trophic cascades.

Sample size:

Previous studies found large effect sizes on herbivory between controls and exclusions using even small sample sizes ( e.g. Koh et al., 2008: n=8 , effect size= from 1.2 to 17.2; Kalka et al., 2007: n=43, effect size= 3). Thus, 6 replicates at each site (n= 30) should be sufficient.

Project members
ResearcherProject roleProject contact
Magdalena KwasnickaLead Researcher
Arthur ChungMalaysian Collaborator
Cristina Banks-LeiteSupervisor