Incorporating species detectability into conservation targets based on the species-area relationship.

Aim The species-area relationship-based method of setting habitat-specific conservation targets assumes that detectability of species is perfect. However, this assumption is often violated. This paper aimed to combine species-area relationship ( SAR)-based target setting with multispecies occupancy models to better represent communities with a large proportion of rare species, such as marine invertebrate macrofauna, include variable species detection and quantify uncertainty around conservation targets. Location South African West Coast continental shelf. Methods The macro-infaunal dataset used in these analyses consisted of 427 species from 42 sites and 220 samples. To examine the effect of variable species detection and include uncertainty around conservation targets, we compared asymptotic estimator-based targets with those based on multispecies site occupancy models ( MSOMs) in a Bayesian framework, which allowed us to account for the observation process and quantify uncertainty more comprehensively. Results Targets based on MSOM were generally higher than those based on asymptotic estimators, with wider credible intervals. Between 10% and 15% (95% credible interval) of the area should be protected to represent 80% of the species, with the exception of one habitat with many common species that had a conservation target of 2%. Main conclusions Detection error-corrected targets based on MSOM are higher than uncorrected targets based on asymptotic estimators for habitats with lower detection and higher proportions of rare species. Our estimated conservation targets are close to the 10% target for marine habitats set by the Convention of Biological Diversity and in line with international studies but should be considered minimum targets as they do not take into account species persistence. Our results affirm the value of the modified SAR approach, which is applicable to any marine or terrestrial ecosystem. Unlike earlier methods, our modification naturally quantifies uncertainty in estimated conservation targets providing a measure of target robustness for decision-makers. [ABSTRACT FROM AUTHOR]