The first large systematic collection of benthic invertebrate megafauna from the Australian continental margin (depths > 100 m) revealed high species richness and novelty on the south‐western continental slope (∼100–1100 m depth; ∼18° S–35° S). A total of 1979 morphologically defined species was discriminated in seven taxa across all samples: Demospongiae, Decapoda, corals (Octocorallia and Antipatharia), Mollusca, Echinodermata, Ascidiacea, and Pycnogonida. Collectively, 59% were estimated to be new or unnamed species. The distribution pattern of megafaunal communities, analysed with a suite of 17 physical covariates, was most influenced at large spatial scales (>100s km) by bottom temperature, oxygen concentration and latitude, whereas at smaller scales (10s of km), seabed type was most influential. Many covariates are driven by the same physical processes and are correlated (e.g. to depth or latitude), thus it is not possible to ascribe causal relationships to fauna distributions. However, their identification highlights the spatial scales that determine the composition of megafaunal communities. Regional‐scale transitions in bottom temperature and oxygen concentration are determined by water masses and currents that interact with the south‐western margin seabed in different ways depending on location. The nested, smaller‐scale heterogeneity of seabed type, classified simply as ‘hard’ or ‘soft’ terrain, differentiates consolidated attachment sites for sessile fauna from sediments suited to mobile and burrowing fauna. Different physical factors affect the distribution of benthic fauna at different scales. Collectively, these patterns of heterogeneity can be represented in a hierarchical framework that consists of biogeographic provinces, biomes, biogeomorphic features, terrains, and finer scales. The Australian government is using a hierarchical approach to identify bioregions for management purposes; a key aim is to ensure that a National Representative System of Marine Protected Areas (NRSMPA) will meet the requirement of comprehensiveness, adequacy and representativeness. Important findings from this study are that the provincial structure of invertebrate megabenthos broadly aligns with the provincial structure derived earlier from the distribution of fishes, but there are differences in the distribution of individual major taxa at both provincial and megahabitat scales. Representative coverage of rarer taxa or narrowly distributed taxa might not be feasible at the same time as ensuring main fauna groups are adequately represented. The hierarchical scales of heterogeneity of the megabenthos in this area, the differences between taxa, and the high proportion of apparently rare species make it clear that it will be as important to manage the area outside the NRSMPA as to manage the NRSMPA itself. Management will be required at different scales that correspond to the multiscale spatial heterogeneity of continental margin fauna.

Williams, A., Bax, N. J., Kloser, R. J., Althaus, F., Barker, B., and Keith G. 2009. Australia’s deep-water reserve network: implications of false homogeneity for classifying abiotic surrogates of biodiversity. – ICES Journal of Marine Science, 66: 214–224. Australia’s southeast network of deep-water marine reserves, declared in July 2007, was designed using a hierarchy that represented the distribution of marine biodiversity as a nested set of bioregions. In this hierarchy, geomorphic units, individual or aggregations of seabed geomorphic features, are the finest scale used in the design process. We evaluated the interaction between two hierarchical levels (depth and geomorphic features), using video survey data on seamounts and submarine canyons. False within-class homogeneity indicated that depth, size, complexity, configuration, and anthropogenic impact need to be added as modifiers to allow geomorphic features to act as surrogates for biodiversity distribution. A consequence of using unmodified geomorphic surrogates, and of not correctly nesting geomorphic features within depth, is the diminished recognition of the importance and comparative rarity of megafaunal biodiversity of the continental margin (<1500-m depths). We call this area the zone of importance, because it is where targeted marine impacts coincide with the greatest megafaunal biodiversity. Refining the geomorphic classification is desirable for future biodiversity characterization, but an alternative approach is to define patterns in biodiversity and abiotic variables jointly, and to utilize finer scale information and provide a classification that preserves the maximum information of both datasets.

Protecting deep‐sea coral‐based vulnerable marine ecosystems (VMEs) from human impacts, particularly bottom trawling, is a major conservation challenge in world oceans. Management processes for these ecosystems are weakened by key uncertainties that could be substantially addressed by having much greater volumes of quantitative image‐derived data that detail the distribution and abundance of coral reefs and the nature of impacts upon them. Considerably greater volumes of data could be available if the resource costs of image annotation are reduced. In this paper we propose a solution: a deep learning system capable of automatically identifying reef‐building stony corals amongst other seabed substrata in much larger volumes of seabed imagery than was previously possible. Using a previously annotated dataset, we trained a convolutional neural network on approximately 70,000 classified images (‘snips’) comprising six benthic substrate classes, including reef‐building stony coral—‘coral matrix’. Model performance improvements, chiefly by dataset cleaning, transfer learning and hyperparameter optimisation, resulted in the final trained model achieving validation accuracy of 98.19%. The classification was robust: benthic substrate types were accurately differentiated, and in some cases more consistently than was achieved by human annotators. Synthesis and applications. The availability of much larger volumes of automatically annotated image‐derived data will improve spatial management of impacts on coral‐based VMEs in the deep sea by (1) improved cross‐validation and performance of spatial models required to predict coral distribution and abundance over the large scales of managed areas, and (2) establishing empirical relationships between coral abundance on the seabed and coral bycatch landed during fishing operations.

Because benthic ecosystems of seamounts support vulnerable biota that are increasingly threatened by mining and fishing, assessments of ecological conditions and impacts are important to balance resource extraction with biological conservation. These assessments rely on ecological metrics generated from physical collections of fauna with sleds or similar gears, or by surveying the seafloor with cameras. If different gears (sled or cameras) “sample” the benthos differently, derived ecological metrics will also differ. Here, we test this hypothesis by comparing data provided by a robust benthic sled and a quantitative towed camera from colocated transects on seamounts off Tasmania (Australia). Our comparison showed these tools provide complementary but not interchangeable metrics of seamount benthic diversity and abundance. Choice of sampler(s) needs to be determined by survey objectives and consider the following trade‐offs: (1) sleds will detect more species than image‐based techniques; (2) specimens obtained by physical sampling (sleds) can considerably extend the scope of ecological analysis through genetic, isotopic, and other analysis of biological material not available from camera surveys; (3) image‐derived data provide significantly and often considerably, higher estimates of megabenthos abundance; and (4) environmental impacts are much lower in camera surveys and therefore are arguably more ethical—an important consideration when repeat surveys are required, especially in areas of high conservation importance or heightened sensitivity. Undersampling by sleds of corals and sponges is an important aspect of gear selectivity because these taxa are consistently prominent indicators in protocols that identify benthic “vulnerable marine ecosystems” in assessments of human impacts.

Environmental harm to deep-sea coral reefs on seamounts is widely attributed to bottom trawl fishing. Yet, accurate diagnoses of impacts truly caused by trawling are surprisingly rare. Similarly, comprehensive regional assessments of fishing damage rarely exist, impeding evaluations of, and improvements to, conservation measures. Here we report on trawling impacts to deep-sea scleractinian coral reefs in a regional (10–100s of km) fishery seascape off Tasmania (Australia). Our study was based on 148 km of towed camera transects (95 transects on 51 different seamounts with 284,660 separate video annotations and 4,674 “on-seamount” images analysed), and commercial trawling logbook data indexing fishing effort on and around seamounts. We detect trawling damage on 88% (45 of 51) of seamounts. Conversely, intact deep-sea coral reefs persist in refuge areas on about 39% (20 of 51) of the seamounts, and extend onto rocky seabed adjacent to seamounts. Depth significantly shapes the severity of trawl damage. The most profound impacts are evident on shallow seamounts (those peaking in < 950 m depths) where recent and repeated trawling reduced reefs built by scleractinian corals to rubble, forming extensive accumulations around seamount peaks and flanks. At intermediate depths (∼950–1,500 m), trawling damage is highly variable on individual seamounts, ranging from substantial impacts to no detection of coral loss. Deep seamounts (summit depth > 1,500 m) are beyond the typical operating depth of the trawl fishery and exceed the depth range of living deep-sea coral reefs in the region. Accurately diagnosing the nature and extent of direct trawling impacts on seamount scleractinian coral reefs must use stringent criteria to guard against false positive identifications of trawl impact stemming from either (1) misidentifying areas that naturally lacked deep-sea coral reef as areas where coral had been removed, or (2) attributing trawling as the cause of natural processes of reef degradation. The existence of sizeable deep-sea coral reef refuges in a complex mosaic of spatially variable fishing effort suggests that more nuanced approaches to conservation may be warranted than simply protecting untrawled areas, especially when the biological resources with conservation value are rare in a broader seascape context.