Conclusions
We demonstrate that a time-series of imagery from an AUV was able to detect a strong and consistent protection effect on the presence of urchin barrens in a study region on the east coast of Tasmania. Although the NTR has lower rates of barrens presence we also found the odds of barrens presence increasing at approximately 31.9% per year in both NTR and reference sites. Therefore, there is currently strong evidence for the resistance of the NTR to barrens formation, but importantly there is also no evidence of recovery. While barrens cover is currently relatively low inside the NTR, a slowing of the rate of increase or a recovery of kelp habitat may be necessary to prevent an eventual phase shift to persistent barrens. With ongoing climate change related warming and incursion of urchins into new areas, monitoring the increase in barrens is critical to informing adaptive management and developing innovative solutions. To support these efforts, we also report that measures of reef complexity and depth from multibeam sonar mapping provide important predictors for the occurrence of barrens. Therefore, detailed seabed mapping and spatial analysis may help in spatial management when targeting efforts to reduce urchin numbers via mechanisms such as spatially targeted fishing incentives for urchin divers. Finally, we advocate the use of spatio-temporal models such as those presented here when assessing time-series of benthic imagery due to their flexible nature and ability to incorporate spatial and temporal correlation inherent in imagery data.