Visual detection based distance sampling offers efficient density estimation for distinctive low abundance tropical forest tree species in complex terrain

Abstract

Good density estimates for low abundance tree species are costly to achieve especially in rugged or disturbed forest landscapes. More efficient methods would be of considerable value to managers and conservationists. Here we assess a method that has been neglected in this context. We examine and compare distance-based visual detection line-transects and conventional fixed-width transects for assessing a distinctive low abundance species of conservation significance, Myrianthus holstii Engl., in three separate areas, within a steep, disturbed mountain rain forest. Precision and implied accuracy appeared substantially better with the visual detection line-transect than with the fixed-width transect for equivalent costs and effort at all three landscapes but as the two methods provide different estimates there are questions of possible bias in both approaches. We discuss the strengths and weaknesses of the distance approach and suggest some recommendations concerning its application. We conclude that the distance method is suited to low density species that are easily identified, even when understorey vegetation and terrain severely impair visibility. However, due to the differences in detection probabilities, populations need to be stratified both by tree size and context.