Projecting genetic diversity and population viability for the fragmented orang-utan population in the Kinabatangan floodplain, Sabah, Malaysia

Abstract

Genetic management of fragmented populations poses logistical and theoretical challenges to conservation managers. Simulating changes in genetic diversity and differentiation within and among fragmented population units under different management scenarios has until now rarely used molecular marker data collected from present-day populations. Here we examine the genetic implications of management options for the highly fragmented yet globally significant orang-utan population in the Lower Kinabatangan Wildlife Sanctuary, Sabah, Malaysia. We simulated the effects of non-intervention, translocation, corridor establishment and a mixture of the latter 2 approaches on future genetic diversity in this population using the stochastic simulation software VORTEX and a well-described molecular dataset for 200 individuals from within the Sanctuary. We found that nonintervention resulted in high extinction risks for a number of subpopulations over short demographic timescales (<5 generations). Furthermore, the exclusive use of either translocation or corridor establishment as a management tool was insufficient to prevent substantial levels of inbreeding using demographically and logistically feasible translocation rates and was insufficient to prevent inbreeding and extinction in the most isolated subpopulations using conservative corridor establishment rates. Instead, a combination of modest translocation rates (1 ind. every 20 yr) and corridor establishment enabled even the most isolated subpopulations to retain demographic stability and constrain localised inbreeding to levels below a threshold of 0.1. Our simulations suggest that this mixed management approach is both a pragmatic and potentially successful course of action and that this combination may be useful in other species and fragmented populations in the future. The use of present-day molecular data in stochastic simulations requires further development, but here we show that it can aid predictive modelling. © Inter-Research 2010.