Introduction
Distributional changes of marine species associated with climate change are being documented worldwide (e.g. Poloczanska et al., 2013, Perry et al., 2005). In cases where newly arriving species play a dominant ecological role, dramatic ecosystem changes or “phase shifts” may occur with major impacts on endemic communities. Understanding how conservation measures may confer resilience, that is, the ability to resist and recover from disturbances (Holling, 1973, Hodgson et al., 2015), is therefore a key conservation priority in the face of climate change (Bates et al., 2019). Conservation measures that may confer resilience, such as the protection of habitats or species from additional human pressures (Carpenter et al., 2001), are typically legislated and managed on smaller scales than those over which climate change impacts occur. Therefore, studies that examine the efficacy of localised conservation measures in protecting habitats and biodiversity are required.
No-take reserves (NTRs) are areas where fishing pressure is reduced/removed. NTRs provide researchers with a means of examining the resilience of ecosystems that may have more “intact” communities compared to fished areas. The greater abundance and size of predators in some NTRs has been linked to increased community stability when compared to adjacent fished areas (e.g. Bates et al., 2017, Bates et al., 2013, Mellin et al., 2016, Strain et al., 2019) as well as possible dampening of marine pest outbreaks (Vanhatalo et al., 2017). Ecological theory developed from invasion ecology suggests that that the top-down control of higher trophic level predators can aid in reducing the establishment and persistence of invasive or range extending species (e.g., Worm et al., 2006, Byrnes et al., 2007, Bulleri et al., 2009, Vanhatalo et al., 2017). However, the number of empirical studies examining the effect of NTRs on local ecosystem resilience to climate-driven invasive species impacts is limited by the technical difficulties associated with acquiring high-quality time-series monitoring data in the marine environment.
Here, autonomously acquired marine imagery is coupled with spatio-temporal statistical analyses to assess how an NTR mediates local ecosystem resilience in a regional hotspot of global climate warming. The marine region of south east of Australia, and in particular the east coast of Tasmania, has been identified as having warming above the global average (Ridgeway, 2007, Dunstan et al., 2018). The strengthening of the East Australian Current (EAC) has resulted in increased ocean temperatures and an extended poleward penetration of the current over the last 60 years. This EAC extension has coincided with the arrival of many new species from sub-tropical latitudes into this temperate mid-latitude region (Robinson et al., 2015). Perhaps the most ecologically important new arrival is the long-spined sea urchinCentrostephanus rodgersii (Agassiz). The first records of this species in northern Tasmania were recorded in the late 1970s, and sightings have been increasingly moving poleward through time (Ling and Keane, 2018). This urchin species is an ecosystem engineer, capable of transforming productive kelp beds and invertebrate covered reefs into bare rock barrens with major impacts on biodiversity and flow-on effects for economically important rock lobster and abalone fisheries (Ling, 2008).
Dive-based and towed video surveys of urchin densities and barrens habitat on the east coast of Tasmania have highlighted concerning trends over the last 15 years, with an estimated increase in C. rodgersii numbers of approximately 50% and an overall four-fold increase in barrens habitat between 2001 and 2016 (Ling and Keane, 2018, Johnson et al., 2005). However, Tasmanian NTRs may offer resilience to barrens formation and expansion due to the higher density of predators able to control urchin densities (Bates et al., 2017, Ling and Johnson, 2012, Ling et al., 2009). The main potential predator of C. rodgersii is the rock lobster Jasus edwardsii , a species that is the target of commercial and recreational fisheries. Higher densities of lobsters inside Tasmanian NTRs have been shown to result in three to seven times higher predation rates, depending on habitat and potential refuge for urchins (Ling and Johnson, 2012). Monitoring the distribution of C. rodgersii across its extended range and the extent and rate of expansion of barrens habitat as driven by climate change, and whether the spread of barrens may be mitigated by NTR establishment, is therefore a high priority management issue from both a conservation and fisheries perspective.
Long term monitoring of urchin benthic habitat over large spatial extents is a challenging problem. The use of imagery as a monitoring tool for the marine benthos has been increasing over recent decades. Sampling platforms such as Autonomous Underwater Vehicles (AUVs), towed and drop camera systems and Remotely Operated Vehicles (ROVs) have allowed collection of large amounts of data over greater areas and deeper depths than have been traditionally surveyed using diver-based approaches (Pizarro et al., 2013, Durden et al., 2016). Imagery or video footage from these platforms can typically be geolocated, allowing for spatial components to be incorporated into subsequent analyses; including the co-location of observations with mapping products such as multibeam sonar (e.g., Hill et al., 2014) and the analysis of spatial patterns in distributions of target species (e.g., Perkins et al., 2018). NTR monitoring programs utilising these platforms such as Australia’s Integrated Marine Observing System (IMOS) AUV program and California’s Marine Protected Area ROV monitoring program now have data spanning over a decade in some reserves. As these programs are located across regions experiencing climate change impacts, an opportunity exists to utilise the available monitoring data to compare NTRs and adjacent fished areas through time.
Here we use a Bayesian hierarchical modelling approach to analyse the presence of C. rodgersii barrens using a time series of marine imagery collected by an AUV from an east coast Tasmanian NTR and nearby control sites. A survey of barrens inside and outside of the long-established (> 25 year) Governor Island NTR using benthic imagery from an AUV revealed that the NTR has a reduced presence of barrens compared to nearby fished areas (Perkins et al., 2015); however, analysis was based on a single set of survey data undertaken in 2010-11. Subsequent AUV surveys have been conducted across the same sites, providing data to examine changes through time. The analysis of this novel dataset of marine imagery examines: (i) the difference in presence of urchin barrens inside the NTR compared to control sites (i.e., the “protection effect”); (ii) whether the NTR mitigates the rate of barrens expansion through time; (iii) whether depth and rugosity are important ecological predictors of the presence of barrens; and (iv) whether including spatial and temporal correlation structures improves predictive capability.