Andreas Wilting, Frauke Fischer, K. Eduard Linsenmair
Address (University of Würzburg, Biocentre, Department of Animal Ecology and Tropical Biology 97074 Würzburg, Germany)
Corresponding author Andreas Wilting
Phone: +49-931-888 4316
Fax: +49-931-888 4352
Email: a.wilting@gmx.de
In our recent paper [2] we proposed the potential of a rigorous track classification method to study secretive carnivores in tropical rainforests. On the basis of six clouded leopard track sets we estimated a rough minimum density of clouded leopards in our small study area and extrapolated our local results to the landscape level.
We are grateful for the critical response by Gordon et al., but would like to emphasise, that we are fully aware of the limitations of the track classification method. Our extrapolated clouded leopard numbers were rather intended to be a first working hypothesis for further research than a reliable estimation of the actual population size in the whole State of Sabah.
We would like to respond to the main concerns of Gordon et al. to clarify our results and help to prevent misinterpretations that might have negative effects on the management of one of the most threatened cat species in Asia.
Correctly Gordon et al. pointed out, that it is extremely difficult to recognise individual felids from their tracks. Therefore we noted the limitations that have to be considered when applying the track classification method. The authors criticised that we applied a principal component analysis (PCA) to separate the track sets, although the tracks of the same clouded leopard were not independent. We disagree that the PCA assumes that the cases are independent, because we used the PCA only as a means of exploratory data analysis to reduce our original 14 variables to a two dimensional graph for better illustration. Because we did not statistically test our data e.g. by calculating the confidence intervals, our tracks do not have to be independent. Furthermore, the assignment of the tracks to the six track sets was part of the analysis. Gordon et al. suggested averaging track measurements for each track set. Applying this reduction we would have eliminated the variance of tracks within one track set. Further, within the scatter plots the track sets could never intersect with each other in space, because each track set would be only represented by one point in the graph. In summary we think that we have applied the PCA correctly and we feel safe to presume a minimum number of four clouded leopards in our study site.
We are aware that it can never be guaranteed by pugmark assessment to track all individuals in a study area and fully agree with Gordon et al. that due to hard and unsuitable substrates the capture probabilities of an individual might be lower than using, for example, a camera-trapping approach. We tried to overcome these uncertainties by applying a capture-recapture analysis, which incorporates the capture probabilities. However, due to the low number of recaptures in our study we emphasized that our calculated density should rather be taken as a rough minimum estimate and not as the true number. Nevertheless, we suppose, fully in line with previous methodological publications [e. g. [9,11][13-15]], that the track classification method will have a high potential for further research activities; presumably not to provide true numbers, but rather as a cheap alternative to estimate rough minimum numbers in a particular site. We totally agree with Gordon et al. that the calculated track-based estimates have to be checked by an independent method like camera-trapping, and we are currently planning to apply both methods in different study sites for such an evaluation.
Furthermore, Gordon et al. criticized that we up-scaled our local results to the landscape level. We are equally concerned and well aware of the fact that without any detailed information about the other areas such extrapolations are based on very weak evidence. We discussed the problems resulting from this approach in our publication (different legal hunting and poaching pressures; different forest structures and protection status of the reserves, and different prey abundances in the reserves). The authors are right that the close proximity of our study site to the delimitated oil palm plantation affect the density of potential prey species, but without any information about the extent of regional differences and without any knowledge about clouded leopards’ preferred prey species in Borneo, we were not able to consider metabolic requirements in our rough calculation.
Gordon et al. are right to criticised that we ignored the smaller isolated populations. However, we did not intend to give an estimation of the total numbers of clouded leopards in Sabah, we rather wanted to locate areas that might be large enough to hold a viable population (defined as > 50 individuals). Therefore, we excluded those smaller populations on purpose. Being aware of all these uncertainties in our rough estimation we still suppose that as a first working hypothesis these figures are of great value for future research. It is a first tentative step to fill a tremendous gap of knowledge. For a species with such limited information on its distribution and status, even very rough estimates, based on limited data, are valuable and important. Gordon et al. are right in stating that these numbers should not become a “quoted fact” in literature. They should rather motivate researches to test these numbers during intensive field studies and help to set priorities for future research plans. We are equally concerned about the “high” population estimates stated in the global media for Borneo and Sumatra. However, we neither gave any estimates for the whole of Borneo nor for Sumatra. If the numbers in the global media are based on our rough estimates for Sabah the discussion in our paper was not sufficiently considered and for the best of this species we hope that more precise data will be provided by further research very soon.
These upcoming research activities are of even greater importance, because recent reclassification of clouded leopards suggests a distinct species (N. diardi) on the Sundaland islands Borneo and Sumatra [8][24,25]. Furthermore a wider genetic sampling by Wilting et al. [25] indicates limited gene flow and population division between the islands of Borneo and Sumatra. This reclassification puts the distinct Sundaland clouded leopards on Borneo at an even greater risk of extinction.
Answers to all your Biology Questions
