PS 20-30
Optimizing spatial monitoring for rare plants: An example from the California Channel Islands
Spatial surveys generating species presence/absence data are among the most common monitoring tools in conservation management. Given time and cost constraints, maximizing the effectiveness of such surveys for detecting responses to environmental change is critical. However, large data sets that allow comparison of survey strategies are scarce. In this study, we used a database of rare plant occurrences on Santa Rosa Island (SRI; California Channel Islands, USA) to simulate different potential monitoring approaches. Data originally were collected to assess responses of 55 rare and endemic plants after removal of cattle, deer and elk from SRI. From 1994-96, as herbivore control began, nearly 30% of canyon habitat on SRI was systematically searched, the locations of 442 rare plant occurrences mapped and abundances recorded. In 2010-12, more than 90% percent of this canyon habitat was resurveyed. We developed simulation models to subsample our data under different assumptions about survey strategy, testing three main questions: 1) How much sampling effort was necessary to capture changes in species richness and abundance? 2) How did varying the spatial range of sampling relative to local intensity alter results? 3) Could baseline data from 1994-6 be used to design a more effective follow up survey?
Results/Conclusions
Total observed changes in species richness did not saturate with increasing effort for any of our simulated strategies, suggesting that a smaller survey would not have captured the responses measured by our full data set. However, we also found evidence that some strategies were more effective than others. Randomly drawing individual survey days led to better estimates of total richness than selecting whole canyons. We also found that designed strategies were more effective than random sampling. Interestingly, weighting surveys to maximize spatial coverage based on distance between canyons had no effect on detection of species richness. In contrast, weighting sampling by observed richness in the baseline survey led to improved estimates; prioritizing the most species rich canyons while minimizing species overlap between canyons led to the highest detection of final richness. These patterns may be caused in part by a high level of beta diversity between canyons; 20 of the focal species found in the seven best sampled canyons were isolated to just one of those canyons. Given that endemic plant species richness has been correlated with high beta diversity in a number of landscape studies, these results could be of broad value in designing monitoring programs for other ecosystems.