This section contains a report on progress with the herpetofauna biodiversity survey (16.2), habitat usage by two forest skinks (16.3) and some frog species (16.4)

Introduction

Sulawesi, Indonesia, is a large equatorial island with a rich and diverse herpetofauna assemblage. Over 100 reptile and 46 frog species have been recorded from the island to date, many of which are endemic. However, despite this high level of endemicity and the biogeographic significance of the region, relatively few surveys have been undertaken of the herpetofauna of Sulawesi and associated smaller islands. These figures are therefore likely to be conservative estimates of the species richness of Sulawesi. Approximately half of the frog taxa known from Sulawesi have been discovered in the past 15 years, many of which have not yet been formerly described. New reptile species have also been discovered in recent years. In addition, the systematics of several other reptile groups requires further resolution, and the status of a number of species in the region requires clarification. The conservation status of many poorly known taxa is also unclear. Consequently there is a great need for further surveys of the herpetofauna of the region.

In 2000 Operation Wallacea commenced a herpetofauna survey in south-east Sulawesi. This survey was conducted as part of a general baseline biodiversity survey of fauna. This year, surveys were conducted on Buton Island, with a limited survey of Hoga Island in the Wakatobi Marine National Park, where Operation Wallacea’s marine projects are based. The primary aims of this study were to:

• refine sampling strategies for herpetofauna surveys in the region
• generate species lists for specific areas/regions
• collect voucher species of any new or poorly-known taxa
• collect baseline information on relative abundance and habitat associations of species
• provide a basis for further expanded surveys and a herpetological research program, to further advance knowledge of the status and biology of the herpetofauna, and ultimately contribute to biodiversity conservation programs in the region.

The survey was based in the vicinity of Labundo bundo in central Buton Island. Sampling was conducted between early July and late September 2000. A small and limited survey was also conducted on Hoga Island in the Wakatobi Marine Park. A variety of methods were employed to sample the herpetofauna.

Pitfall traps were established at 16 sites in a variety of habitats. Each trap comprised five 40 L buckets imbedded in the ground, spaced 5 m apart, with a 1 mm gauge wire mesh drift fence 30 cm high and 20 m long, passing over the centre of each bucket. Pitfall trap sites were chosen to sample the range of broadly-distinct habitat types within the region. These included terraces and mid-slopes in rainforest with varying levels of man-made disturbance from selective logging and rattan harvesting, and several sites in highly disturbed habitats, such as regenerating cleared forest and plantations. Pitfall traps were checked every morning and operated for varying lengths of time; from two to four weeks.

Transects 100 m long were established at 30 sites, including all pitfall sites. Diurnal censuses were conducted for active reptiles along each transect. Each transect was censused five times during different day time intervals: early morning, late morning, midday, early afternoon and late afternoon.

Active opportunistic searches were conducted for reptiles over a wider area. These searches generally comprised walking slowly through various habitats, such as forests, along streams, mangroves, agricultural areas or swamps, looking for active reptiles and investigating under logs, rocks or other ground debris for sheltering animals.

Nocturnal searches were conducted for frogs, geckos and snakes. These searches were mostly targeted at, or near, aquatic environments, such as streams, swamps or rice paddies, but nocturnal searches were also conducted in other forest habitats and caves, specifically targeting geckos and snakes.

Stream transects were established at 11 sites in the Labundo bundo area for standardized censuses of the relative abundance of riverine frog species.

Voucher specimens were collected of most taxa detected to confirm field identification, to aid identification of unknown species or suspected new species, and to collect material for taxonomic groups requiring further systematic resolution. These specimens will be lodged with the National Museum in Bogor.

On Hoga Island, only diurnal and nocturnal opportunistic searches were conducted.

A total of 41 species were detected in the Labundo Bundo region. This comprised 11 species of frog (Ranidae, Bufonidae, Microhylidae & Rhacophoridae), 17 species of snake (Colubridae, Crotalidae, Cylindrophidae, Pythonidae, Typhlopidae), 1 turtle species (Bataguridae) and 12 lizard species (Agamidae, Gekkonidae, Scincidae, Varanidae). Eight reptile species (Colubridae, Hydrophidae, Typhlopidae, Pythonidae, Gekkonidae & Varanidae) and one frog species (Ranidae) were located on Hoga Island. The combined total number of species encountered in the region during this expedition was 44.

A number of taxa could not be confidently identified in the field, due to the limited information available on species descriptions for the region. The identity of four frog and seven of the lizard taxa are yet to be confirmed.

Pitfall trapping proved to be an effective technique for detecting small, predominantly terrestrial reptile and frog species. Twelve reptile and five frog species were detected by pitfall trapping. Six reptile and two frog species were only detected using this technique during this study. These included two Microhylid frog species, four snake species (Calamaria spp. and Typhlopidae) and three skink species.

Diurnal transect censusing was a relatively ineffective technique. Only five reptile species were located using this approach, all of which were detected using other methods. Opportunistic searches were the most effective technique for detecting larger species, such as snakes, monitor lizards and arboreal reptile species.

Nocturnal censusing was the primary technique for detecting frogs, however, this technique was surprisingly ineffective at detecting nocturnal snakes.

Five species of frog were located that breed in streams. All these species were in the midst of breeding at the commencement of the survey, or commenced breeding around the end of the wet season during the survey period. This provided the opportunity to make preliminary observations of oviposition sites, and conduct surveys for tadpoles. Voucher specimens of tadpoles of these species have been collected.

Data on species community structure is limited at present, but some preliminary observations may be made. There appear to be distinct differences in lizard assemblages between forest habitats and human settlement areas, such as villages and fields. Several species were located in the vicinity of Labundo bundo that were not found in any forest habitats.

Relatively few snake species were found in forest habitats. Several species (Calamaria, Cylindrophis spp., Typhlops ater and Tropidolaemus wagleri). Were restricted to forest habitats, but Colubrids and Python reticularis were more widespread.

The number of frog species detected was lower than expected. This may have been due to the timing of the survey (tail end of the wet season); limited range of habitats or altitudinal variation in central Buton, which would otherwise enhance amphibian species richness; or the effect of isolation from the mainland.

Significant differences were found in the level of activity and detection rates of reptiles and amphibians between the wet and dry season periods sampled this year. All frog breeding activity ceased at the end of the wet season and pitfall trap returns diminished significantly for all taxa except the two most common forest skink species. This observation has significant implications for the timing of future herpetological survey work in the region.

The number of species detected this season is by no means comprehensive, as additional species were still being detected late in the survey period. In addition, only a relatively small area of Buton Island was surveyed.

All specimens collected will be lodged with the museum in Bogor in late September. Confirmation of species I.D.’s is dependent to some extent on examination by Indonesian museum staff and consultation with systematists in Australia and the USA. A final report entitled An Initial Survey of the Herpetofauna of Buton Island was produced in December 2000, which contains a species list, museum specimen list, and analysis of survey data. 

A paper entitled Gillespie, G.R. Howard, S. Lockie, D. & Scroggie, M  A survey of the herpetofauna of offshore islands of south-east Sulawesi, Indonesia and examination of patterns of species assemblages and habitat modification will be submitted to the Journal of Herpetology by February 2003.  

Tropical rainforests generally support diverse assemblages of lizard species, which exploit a wide range of niches. These include both nocturnal and diurnal species, and terrestrial and arboreal species. Such niche partitioning may result in exploitation of a range of different microhabitats within and adjacent to forest environments for thermo-regulation, foraging, shelter and nesting. Knowledge of the habitat requirements and activity patterns of a species, along with life history attributes, is therefore essential to gaining an understanding of the ecology of a species. In addition, knowledge of habitat requirements may provide significant insight into how different species within a community are likely to respond to disturbance processes or other environmental change.

Sulawesi, Indonesia, is a large equatorial island with a rich and diverse herpetofauna assemblage. Over 30 species of lizard have been recorded from the island to date, many of which are endemic. However, limited surveys have been undertaken in the region and this number is most likely a conservative estimate of the species richness. No research has been conducted to date into the ecology of any of these species

As with elsewhere in Indonesia and other equatorial regions, the forests of Sulawesi have been cleared extensively for human settlement. As most of the island was originally forested, most of the faunal species are adapted to rainforest habitats and are probably displaced by forest clearance. Maintenance of the herpetofaunal biodiversity is therefore dependent to a large extent on conservation and appropriate management of the remaining forest areas. However, the remaining forested areas, including the reserve systems, are subjected to a range of disturbance processes, including timber and rattan harvesting, firewood collection and hunting. The impact of the changes to forest environments brought about by these disturbance processes on the herpetofauna of the region is unknown. Information is required on the habitat requirements and ecology of various species to identify their vulnerability, and gain an understanding of their responses to different levels of disturbance.

The aim of this project is to gain an understanding of the habitat associations, daily activity patterns and niche partitioning of two sympatric terrestrial forest lizard species in Sulawesi. In addition, their responses to differing levels of forest disturbance/degradation will be investigated.

The study was carried out in the forested regions surrounding Labundo Bundo in central Buton Island, Southeast Sulawesi. The region is characterized by undulating hills, with elevations from sea level to 500 m. Much of the region remains forested but river flats and valleys have been cleared for agriculture. Two reserves are present in the region: the Kakinauwe Reserve runs along a ridge system to the northeast of the village, and the Lambusanga Reserve occurs to the south. Selective logging and rattan extraction has occurred extensively throughout the region, including the reserve system. Selective logging appears to occur mostly in relatively accessible areas and evidence of this diminishes with increasing distance from human settlement. Logging activity has not been uniform, resulting in a mosaic of minor and major-disturbed forest areas. However, rattan extraction has been much more widespread.

Two species of terrestrial lizard species were examined in this study: Mabuya rudis and Sphenomorphus textum. Preliminary surveys of the region indicated that these were relatively common forest floor species, occurring in sympatry throughout the region in forested areas. Mabuya rudis also commonly occurs in a range of highly modified habitats, such as fields and around villages (G. Gillespie pers. obs.). Within the forest, M. rudis tends to be observed in open sunny patches, often near fallen trees, while S. textum may be observed in a wider range of microhabitats. These observations suggest that there may be significant microhabitat partitioning between these two species, and differences in their responses to forest clearance or disturbance.

Transects 100 m long were established at 31 sites. Sites were chosen to incorporate the topographic variation in the region (i.e., ridge, slope, gully & flat), and as much as possible of the range of structural variation and disturbance in the forest. This was assessed visually, based upon canopy gap, height and girth of trees present, and understorey structure and species composition. Several sites were also selected in grossly modified habitats, such as cocoa plantation and regenerating forest in old fields.

Transects were marked every 5 m with flagging tape. Visual censuses of lizards were conducted at transects by one person walking slowly along the transect, pausing every 1 - 2 m. Each transect was censused at five different time intervals: early morning, late morning, midday, early afternoon and late afternoon. Weather variables (amount of cloud cover, sunlight, wind & precipitation) were recorded during each census. Only one census was conducted on each transect per day. Transects were censused in blocks of five over five consecutive days, varying the order of census each day. The actual time at which each census started varied by up to an hour, due to variation in the time it took to conduct the previous census, and time taken to travel between transects.

Whenever a lizard was observed on a transect the species and its activity was identified, and a range of micro- and macrohabitat attributes were recorded for the site at which the animal was first observed. These included ground cover (litter, logs, rock & vegetation cover) within a 0.5 m² quadrat and distance to various habitat features, such as trees and logs. At completion of all censuses on a transect an equal number of random sites were generated, using a random number table, and the habitat attributes of these were sampled in exactly the same manner.

The general habitat characteristics of each transect were also recorded. Six 5 x 5 m quadrates were established at 20 m intervals along each transect. Within each of these, measurements were made of the circumference of each tree above 4 m in height and the length and diameter of each log (over 10 cm diameter). The density of vegetation at heights of 0.5 and 1.5 m was estimated with the aid of a graduated pole 1 m long with 2 cm graduations. The pole was held horizontally at the respective height, and the number of graduations visible from a distance of 5 m was counted. Canopy density was estimated by taking a photograph of the canopy with a 50 mm lens from the centre of the quadrat. Canopy density will be scored by overlaying a transparent grid on each photographic print and counting the number of cross lines intersecting vegetation cover. For each transect, the mean for each of these habitat variables will be calculated across the six quadrats.

Comparisons of activity levels of the two species throughout the day will be examined by comparing mean activity levels of each species across all transects at the various time intervals. Weather conditions (cloud cover, temperature, sun light and wind) will be transformed to meet assumptions of normality where necessary, and reduced to independent variables by Principal Components Analysis. A two way ANCOVA will be performed, with species and census time interval as categorical variables, and weather principal components with Eigen values greater than one and time of census commencement as covariates.

Microhabitat use of the two species will be analysed by comparing the habitat characteristics of each site at which lizards were observed to those of the random sites, which represent available habitat. Habitat variables will be examined for normality and equal variance, and where necessary, transformed to meet these assumptions for parametric statistical tests. Differences in the habitat characteristics selected by each species will be assessed by discriminant function analysis.

Broad habitat associations will be examined by multiple regression. Mean habitat variable scores from 5 x 5 m quadrates and 0.5 m² random sites will be reduced by Principal Components Analysis. The relative abundance of each species across all transects will be regressed against those PC’s with Eigen values greater than one.

Data collection was completed on 7^th September and all data has been entered onto a computer database.

Preliminary examination of the data indicates pronounced differences in the daily activity levels of each species. Mabuya rudis appears to be active predominantly in the middle of the day, while S. textum is active all day from early morning through to the evening. Mabuya rudis appears to have a strong association with canopy gaps and logs while the habitat affinities of S. textum are less obvious.

The results of this work will be discussed in the context of existing literature on lizard habitat partitioning in forest ecosystems. It is also expected that this work will provide some insight into the ecological interactions of rainforest disturbances and this lizard community. This work will also provide a foundation for further, more detailed, investigation of the ecology of lizard communities on Buton Island.

Dissertations entitled Habitat Partitioning between Two Species of Skink in the Forests of Buton Island were completed by December 2000 by Emma James from the University of the West of England and Thomas Monteath.

A paper entitled Habitat Partitioning between Two Species of Skink (Mabuya rudis and Sphenomorphus textum) in the Forests of Buton Island will be prepared by December 2003 by Graeme Gillespie, Emma James and Thomas Monteath.

Introduction

Despite the biogeographic significance of Sulawesi and high level of endemicity, the herpetofauna, remains largely unstudied. This is especially the case for the frog fauna. Of the 46 taxa currently listed as occurring in Sulawesi, approximately half have only been discovered in the past 15 years, and many have not yet been formally described. The conservation status of many poorly known taxa is also unclear. Virtually nothing is known about the ecology of these species. As with many equatorial regions of the world, the forests and wetlands of Sulawesi are under continuing pressure from human disturbance and encroachment. Frogs as a group may be particularly sensitive to environmental change, and are likely to be significantly adversely impacted upon by current activities in the region. There is currently concern globally about declines in frog populations, due to various causes, but the reasons for many of these declines are unknown. There is currently no information on the status of amphibian populations in Indonesia, or the extent of population declines in the region. Baseline information is therefore needed on the demography and ecology of species and communities in the region to address these issues.

A primary requirement for gaining an understanding of the ecological relationships of any species with its environment or within a community is to determine its habitat requirements. This information provides the basis for determining ecological requirements for reproduction feeding, physiological maintenance and predator-prey relationships. Knowledge of these relationships may then lead to an understanding of how human disturbance processes may impact upon a species or community and how best to ameliorate these impacts. Knowledge of habitat requirements is also essential in conservation biology and management in order to identify what physical attributes of the landscape need to be conserved to protect certain species or communities.

At the commencement of the present study, no information was available on the habitat requirements or ecological associations of any frogs in south-east Sulawesi. The aim of this project was to conduct a preliminary examination of the community composition of stream-breeding frogs on Buton Island, and identify key habitat associations of these species. Specifically this study will examine the hypothesis that riverine frogs on Buton Island exhibit similar habitat partitioning and distribution as other riverine anuran communities.

Eleven stream sites were selected from those in the local area of La Bundo Bundo. A range of stream types and locations were chosen that best represented the diversity of streams present. Site choice was constrained to a small extent by problems of accessibility: a 15 km radius from La Bundo Bundo was the limit of reasonable travel for the purposes of this study.

At each site, a transect of either 100 or 200 m was measured. The length of the transect was dependent on accessibility and heterogeneity of the stream. The transects were marked every 5 m by numbered reflective tape/card to allow easy visibility at night.

In order to characterise the macrohabitat of the streams, readings were made of variables considered to be important to frog distribution at five locations along the length of the transects. These were spaced every 25 or 50 m, depending upon transect length. The macrohabitat data collected characterised the type of stream site and the level of disturbance of the immediate area. Qualitative notes about the streams were also taken to assist in this characterisation.

In order to characterise the microhabitat associations of each frog species along streams, a range of microhabitat variables were chosen, based upon those identified as important in previous studies in other areas, and from examination of structural characteristics of the local streams. Variables selected included: substrate; distance from water’s edge, flow types (pool and riffle); distance from litter, vegetation, log, rock and nearest tree with a diameter greater than 50 cm.

Censuses of the stream sites were carried out by spotlighting for frogs at night between nightfall and 11 pm, or until the census of the stream was complete. Local weather conditions were recorded at the beginning of each census. In order to assist with the spotting of frogs, head torches were used, which allow detection of the orange eye-glow reflected when the frogs were spotlit.

The census itself was carried out by walking down the middle of the stream (where possible) and searching an area including the whole of the stream terrace, two metres either side of the stream terrace and extending two metres above the stream/substrate. Lack of unnecessary noise and light was important to avoid scaring off some of the more skittish species. Search intensity was kept as consistent as possible between the sites.

When a frog was spotted, it was captured, and its exact location along the transect noted. The frog was then identified, measured for length, weighed and sexed where possible. Having released the frog, the distance to various microhabitat variables was estimated, along with the distance along the transect. All estimations were carried out by a single recorder in order to maintain consistency of errors.

In order to compare site selection characteristics of various species of frogs with the actual availability of site selection characteristics along streams, an equal number of ‘random frog’ sites was generated in three dimensional space along the transect using random numbers. The positions of these sites was equivalent to the area of search for actual frogs.

The microhabitat association data for both random and actual frog sites and the macrohabitat characteristics of the streams allowed testing of the habitat partitioning hypothesis.

Other observations on behaviour, life history and morphological variations were also noted, to augment the limited information available on the life history of these species.

The relative abundance of each species will be examined between streams with respect to macrohabitat variability. Macrohabitat characteristics of streams will be reduced to a small number of independent variables by Principal Components Analysis, subject to meeting the assumptions of parametric tests. The relative abundance of each species will then be regressed against principal components with Eigen values greater than 1. Due to the small number of streams sampled in this study, there is likely to be inadequate statistical power to resolve all but the strongest trends, and relative abundance of frogs may not be normally distributed across stream sites. Spearman Rank Correlations will therefore be performed also.

Microhabitat use of frogs will be compared among species and with available habitat by using either Discriminant Function Analyses or Non-linear Multidimensional Scaling, depending upon the characteristics of the data.

Preliminary examination of the data indicates pronounced differences in the microhabitat associations of most frog species along all streams. The results of this work will be discussed in the context of existing literature on stream frog habitat partitioning. It is also expected that this work will provide some insight into the ecological interactions between the physical characteristics of streams and their suitability for each species. This work will also provide a foundation for further, more detailed, investigation of the ecology of this frog community on Buton Island.

A dissertation entitled Habitat Utilisation and Basic Ecology of Rain Forest Stream frogs on Buton Island was produced in December 2000 by David Lockie from the University of Manchester. A copy of this dissertation which was awarded a First is available at the Labundo field centre and from the Op Wall UK offices.

A paper entitled Gillespie, G.R. Lockie, D. Scroggie, M.P. Iskander, T. (2003)  Habitat use of stream breeding frogs in south-easterm Sulawesi and some preliminary observations on community organisation has been submitted to the Journal of Tropical Ecology.