Effects of the spatial pattern of disturbance on the patch-occupancy dynamics of juniper–pine open woodland

Typically, studies of the disturbance effect on metapopulation dynamics are limited to understanding the effect of habitat loss although, recently, the spatial pattern of the disturbance has been shown to influence dynamics. In this study, we used a stochastic patch-dynamic model to investigate the effects of spatial disturbance patterns on the persistence of an open woodland community of Juniperus spp. and Pinus spp. First, we estimated patch-occupancy dynamics by using the coefficients that best predicted the occupancy observed in 1998 based on occupancy data from 1957. Next, we evaluated the effects of the rate and pattern of the disturbance on the extinction probability. In modeling the disturbance, we considered (1) the degree of disturbance produced by scenarios of complete destruction or degradation (with the potential for recolonization), (2) the overall rate of disturbance, and (3) the spatial autocorrelation of habitat destruction. Twenty 40-year simulations predicted a 25% increase in the number of patches, and when 50% of the habitat was removed, the impact was more pronounced after complete destruction than it was after degradation of the area. Predictions based on scenarios of complete destruction, including random, contiguous, Brownian, and autoregressive noise, demonstrated that the impact of disturbance depends upon the spatial structure of the disturbance regimen. The autocorrelated structure of the disturbance regimen had the greatest impact on patch persistence. Patch-occupancy was higher after 20 40-year simulations when habitat loss was randomly distributed than when it followed an autocorrelated patch destruction, which was simulated using autoregressive noise to produce 50% habitat destruction. In addition, while habitat loss was negatively linearly correlated with patch persistence when habitat destruction was randomly distributed, a dramatic transition shift occurred when habitat destruction was simulated following an autoregressive spatial distribution after a certain threshold of habitat destruction (40% of the actual open woodland habitat). Our study suggests that the spatial patterns of the disturbance should be considered when predicting the consequences of fragmentation and improving management strategies.