Estimating extinction rate in the presence of observation uncertainty: Occupancy modeling of a long-term, large-scale fragmentation experiment at Wog Wog, Australia

Background/Question/Methods

Determining the true occupancy patterns of a species in space and time is a longstanding problem in ecology. In this study, we addressed this problem by developing a stochastic, spatially explicit occupancy model to assess the long-term occupancy patterns of species in different sized habitat fragments. Sampling began in the Wog Wog fragmentation experiment in southeastern Australia in 1985, with the fragmentation event occurring in 1987. A section of continuous, native Eucalyptus forest was removed and replaced with a plantation of exotic Pinus radiata. Experimental Eucalyptus fragments of sizes 0.25 ha, 0.875 ha, and 3 ha were left within the pine matrix as well as control fragments in continuous Eucalyptus forest. Using a Bayesian approach to fit the model to long-term data on ground dwelling beetles from pitfall traps in the Wog Wog experiment, we estimated the probability of observing a species at an occupied site through time and the probability that a species went extinct at different times. We assessed differences in the extinction and observation probability of species in continuous forest versus fragmented patches, and between small, medium, and large fragments.

Results/Conclusions

As expected, observation probability increases with species abundance for all species under investigation. Interestingly, however, this relationship appears to level off at approximately 40% probability of observation even as abundance continues to increase. This could reflect overall sampling effort in this experiment or a fundamental limit to the effectiveness of the pit fall traps used at Wog Wog. The effect of fragmentation and fragment size on the probability of extinction is highly species-specific. Roughly a third of the species under investigation face a heightened probability of extinction within fragments. Additionally, the same species seem to exhibit an inverse correlation between fragment size and extinction probability.  The probability of extinction for the remaining species seems unaffected or even decreases with fragmentation, possibly as a result of reduced competition with the species most affected by fragmentation.