| dc.description.abstract |
Although feral goats compete with domestic sheep and contribute to the total grazing impact on rangeland pastures their numbers tend to be severely underestimated, particularly in the wooded habitats they prefer. Dung counts provide a potentially useful indication of animal numbers, but have seldom been used for estimating relative numbers of goats grazing in sheep rangelands, because both sheep and goats produce very similar pellets. From a locally-collected reference collection, we devised a number of simple criteria for distinguishing between some of the dung produced by sheep and goats in the mulga rangelands of north-westem New South Wales, but found that there was substantial overlap: some pellets produced by both sheep and goats (but more commonly by goats) look the same. We developed an equation for estimating the origin of such pellets, from a maximum likelihood solution of appropriate probability functions, determined from the reference collection and from the relative proportions of identifiable pellets. This enabled us to estimate the relative abundance of sheep and goats from counts of their pellets. Animal numbers calculated from derived estimates of the relative abundance of dung were comparable with numbers observed in ground and helicopter surveys. The general approach could be used for any problem involving allocating numbers of similar objects to different sources, provided a reference collection is available from which identification probabilities can be determined. |
|