Life in the Canopy: Revealing the hidden diversity and ecology of Borneo's arboreal mammals

Tropical rainforest canopies are structurally complex, floristically diverse, and three-dimensionally vast. They play key roles in ecological processes such as nutrient cycling, carbon storage, and plant primary productivity. An ability to utilise the canopy provides wildlife with access to resources and environmental niches not available at ground level, thereby facilitating the co-existence of a very high number of species, and tropical canopies support a substantial proportion of a forest's vertebrate and invertebrate life. Mammals are a biodiverse and functionally important group and within rainforests, more than half of species are arboreal or semi-arboreal, i.e. exclusively or habitually ese the canopy space. However, due to the practical difficulties of sampling at height, tropical forest canopies remain relatively unexplored, and most arboreal taxa, with the exception of some primates, are little known to science.

Arboreal mammal communities consist mainly of species that are small-bodied, cryptically coloured, elusive, and/or nocturnal; yet, traditional ground-based sampling techniques are biased towards larger-bodied, diurnal mammals that do not flee in the presence of people. Camera-trapping, widely used in terrestrial research, has started to be implemented at canopy level with promising initial results, particularly for single-species studies or those focussing on animal use of habitat features such as canopy bridges. However, the implementation of canopy camera-trapping to inventory and study arboreal communities has to date been limited to eight sites, all in the Neotropics or Africa, and with three focussing only on medium- and large-bodied mammals. Furthermore, around half of all arboreal mammal studies utilising camera-traps deployed units at heights of ten metres or less which, in rainforests where canopies reach between 30-60 metres high, misses an extensive portion of the vertical habitat. These factors point to a clear gap in the application of the methodology, and the understanding of arboreal mammal communities, in the extremely tall and hyper-diverse rainforests of Southeast Asia.

The unique height and structure of rainforests on the island of Borneo have given rise to an exceptionally high diversity of canopy-dwelling wildlife; more than half of all mammals are arboreal or semi-arboreal, and the island represents the world epicentre of gliding vertebrate diversity, including 15 gliding mammal species (14 flying squirrels and the colugo, or 'flying lemur'). Meanwhile, logging is a pervasive threat to forests globally, with some of the highest timber extraction rates in Borneo due to the dominance of commercially valuable dipterocarp trees. Many terrestrial taxa are able to persist in recovering-logged forest, but the changes in habitat structure associated with logging activities are likely to have a greater impact on species that directly depend on the complexity and connectivity of the canopy architecture. However, studies explicitly investigating arboreal mammal responses to logging are lacking, and in general, our understanding of rainforest communities is skewed towards terrestrial species and processes. We therefore risk underestimating both the diversity and potential vulnerability of arboreal wildlife, with implications for conservation, and habitat management and restoration.

At our current state of knowledge, it is not clear (i) whether camera-traps set in the forest canopy are an effective sampling method for arboreal mammals in Borneo's immensely tall and biodiverse rainforests; (ii) what the outcome of this sampling would be in terms of community richness and distinctness from terrestrial mammals; and (iii) whether, and to what extent, arboreal mammals are affected by logging.

In this thesis, I document the first community-wide investigation of Borneo's arboreal mammal community using camera-traps. I deployed a network of cameras in the rainforest canopy, paired with units at ground level in both unlogged and recovering-logged forest areas, to test the efficacy and cost-effectiveness of this method in Borneo. Sampling was conducted across fifty locations, divided equally between unlogged forest (N=25) and recovering-logged forest (N=25), with cameras (Reconyx Hyperfire HC500) deployed in a grid formation and spaced on average 1.26 km apart (range = 0.5-4 km). Every sampling location comprised one terrestrial camera-trap and one canopy camera-trap and, to evaluate the utility of placing more than one canopy camera per tree to maximise species detections, experimental second-canopy units were deployed at a subset of twenty locations, selected at random and divided equally between unlogged (N=10) and recovering-logged (N=10) forest areas. Total sampling thus comprised fifty terrestrial cameras and seventy canopy cameras. Terrestrial camera-traps were set approximately 0.3 m above the ground, while canopy units were set between heights of 9.8-52.3 m (mean = 25.9 m). Cameras at each location were deployed for 7-8 months except the experimental second-canopy units, which were in place for a subset of 3 months. After accounting for malfunctions, mammal detection data were obtained from 17,226 camera-trap nights (CTN): 6,661 CTN from terrestrial cameras; 9,156 CTN from canopy cameras; and an additional 1,409 CTN from experimental second-canopy units. Using these data, I quantify species diversity, community composition, and diel activity patterns, and document responses to logging, of both arboreal and terrestrial mammals, finding that differences across strata are much greater than differences between unlogged and recovering-logged forest. I further quantify relationships between mammal occurrence and a suite of remotely-sensed, high-resolution vegetation covariates, including a novel measure of canopy connectivity, and show that this is by far the most important predictor of occupancy for arboreal species among many potential covariates.

My results illustrate the applicability of canopy camera-trapping to study arboreal communities in Borneo, and extend those of terrestrial studies by demonstrating that recovering-logged forests can maintain mammal diversity across strata, underscoring their value for species conservation. However, I also highlight that the arboreal community as a whole, and particular taxa within it, are more vulnerable to the effects of habitat degradation than their terrestrial counterparts.