Submarine canyons are the primary conduits for particulate transport from the continents to the deep sea, including sediments, organic carbon, and pollutants. Previously, focus has primarily been on transport due to downslope gravity currents. There is increasing recognition of the role internal tides may play, however, a lack of long-term seabed monitoring limits understanding of their significance. Here, we present novel near-seafloor monitoring data of breaking internal tides ('bores'), spanning one year within Whittard Canyon, a land-detached submarine canyon on the NE Atlantic margin. Throughout the year, these bores are energetic, with flow speeds regularly exceeding 0.5 m/s, and capable of resuspending seafloor sediments. Bores were observed for similar to 80 % of tidal cycles during the deployment. The shape and structure of bores varies throughout the year, in relation to spring-neap tidal cycles and longer-term seasonal changes. During the period of November-March, these bores generally attain higher speeds, show more abrupt changes in flow speed, and hold the greatest potential for up-canyon sediment transport. While fine-grained material may be moved throughout the year, this stronger winter activity has the greatest capacity to mobilize larger grains up-canyon. We show that bores, driven by internal tides, are an important, but previously poorly-observed agent for particulate transport. Bores occur frequently, yet exhibit variability at both seasonal and shorter time scales that may shape particulate transport within submarine canyons. These findings change our view of the general significance of internal tides in submarine canyons, with wide implications for hazards, tempo and fluxes of particulate transport.

Cabo Verde hosts unique, highly biodiverse marine ecosystems that thrive on volcanic seamounts and island slopes. These ecosystems are shaped by distinct oceanographic dynamics, influenced by the southeastern edge of the North Atlantic Subtropical Gyre (NASTG) and by seasonal upwelling. To explore regional oceanographic variability over time, this study investigates Holocene (last 11.7 ka) sediments using multi-proxy palaeoenvironmental reconstructions from a short core retrieved from ∼ 4,400 m water depth off Cabo Verde.During the Early Holocene, year-round upwelling, or an intensified Guinea Dome, may have inhibited the development of the strong summer stratification characteristic of the modern regional non-upwelling season. Despite humid conditions over the continent, sea surface temperatures (SSTs) remained relatively low during this subepoch, diverging from the present-day pattern in Northwest Africa, where the wet season is marked by weaker upwelling and higher SSTs. This oceanographic state was likely driven by precession-induced insolation changes associated with the precession minimum, which may have modified seasonal regional wind regimes and influenced broader atmospheric processes. Teleconnections related to transitional postglacial conditions and/or continental climate feedbacks, may also have played a role. The Middle Holocene, corresponding to the most humid conditions of this epoch in Northwest Africa, is characterized by reduced upwelling and an eastward expansion of the NASTG, inferred from warmer subsurface conditions at our study site. This interval also provides tentative evidence for enhanced input of North Atlantic Deep Water (NADW) into the Northeast Atlantic Bottom Water (NEABW). During the Late Holocene, intensified upwelling and a reduced influence of the NASTG, possibly due to a westward retraction of its eastern boundary, are suggested at our site, occurring under arid conditions in Northwest Africa.These results highlight that, despite the overall climatic stability of the Holocene, oceanographic conditions off Cabo Verde experienced significant changes in seasonal upper ocean stratification, upwelling, subtropical gyre influence, and deep-water structure. Such insights improve our understanding of regional climate-ocean interactions, helping to refine climate models and improve predictions of ecosystem responses in this sensitive marine region.

Submarine canyons are ubiquitous geomorphic features found intercepting the continental margins. As such, they provide environmental conditions suitable for many suspension feeding organisms, as they settle on steep rocky canyon walls, whilst taking advantage of increased currents that bring suspended organic matter and food. Additionally, demersal fishing grounds can be found surrounding submarine canyons where it can negatively affect species inhabiting these environments, including vulnerable ecosystems such as cold-water corals (CWCs). In order to understand the impacts of demersal fisheries in CWC communities, we first need to understand their distribution, species composition and vulnerability. Blanes Canyon is an example of a submarine canyon surrounded by demersal fishing grounds, where baseline knowledge on CWCs currently lacks. This study contributes to filling these knowledge gaps by using a dense grid of ROV transects along the canyon, high resolution bathymetry data and CTD surveys, which altogether provide a quantitative description of megabenthic assemblages. Blanes Canyon hosts at least 12 CWC species within 450-1300 m depth range, mainly inhabiting the steep canyon walls. Different assemblages of CWC species were found. Desmophyllum dianthus was the most abundant species, found throughout the entire canyon. Colonial scleractinian species (Desmophyllum pertusum and Madrepora oculata) were found in the canyon head but were lacking in the eastern canyon branch, where octocorals (Muriceides lepida) and black corals (Leiopathes glaberrima) were prevailing. Detailed CTD survey indicated that nepheloid layers (bottom and intermediate) were found at the same depth range as the megabenthic communities, since they provide suspension feeders with particulate organic matter (POM). Overall, this study confirms Blanes Canyon as a CWC habitat, providing densities and spatial distribution of different megabenthic species, along with information of their environmental niches.

Submarine canyons are important conduits for microplastic transport to the deep sea, but the processes involved in that transport and how faithfully seafloor deposits record trends in pollution remain unclear. We use sediment push-cores for microplastic and sediment grain-size analysis from two transects across the Whittard Canyon, UK, to investigate the roles of near-bed flows and sub-seafloor processes in the transport and burial of microplastics and semi-synthetic microfibres. Microplastic and microfibre pollution is pervasive across the canyon at both transects, from the thalweg and from 500 m higher on the flanks, despite turbidity currents being confined to the canyon thalweg. We calculate sediment accumulation rates from 210 Pb dating and show that microplastic concentrations remain similar at sediment depths down to 10 cm. Throughout the Whittard Canyon there is an observed uniformity in the gradual decline in microfibre concentration with sediment depth, despite the variable sample locations and marked variations in sediment accumulation rates. Furthermore, the huge global increase in plastic production rates over time is not recorded, and microplastics are present in sediments that pre-date the mass-production of plastic. The interaction of turbidity currents, deep tidally-driven currents, and sub-seafloor processes affects microfibre burial processes in the deep sea and shreds any potential signal that microplastics may provide as indicators of historical plastic production rates; complicating the use of microplastics as fully-reliable markers of Anthropocene onset.

Benthic communities, such as those dominated by cold-water corals (CWC) and sponges, contribute to the habitat complexity and diversity of deep-sea ecosystems, with characteristic taxa serving as indicators of Vulnerable Marine Ecosystems (VMEs). Many areas on the West African margin remain under-surveyed for the presence of deep-sea benthic communities and, hence, VMEs, limiting the implementation of marine spatial plans. In this study, the benthic communities of SW Cabo Verde (NW Africa), in particular Cadamosto Seamount (SW Brava Island) and the slopes of the islands of Fogo and Brava, were characterized and mapped, providing an assessment of the potential differences in benthic community composition between the two systems (seamount vs island slopes). We employed machine learning approaches (multivariate regression trees and a Random Forest classification) using data on morphospecies composition and densities retrieved from Remotely Operated Vehicle (ROV) video data collected from 2,100 to 1,450 m depth, as well as environmental data on substrate type, terrain, and water column parameters. Ten different benthic communities were identified in the study area, with cnidarians, porifera, and echinoderms as the dominant taxa. Higher densities of CWCs and communities with higher species richness were observed on the seamount compared to the adjacent island slopes. Depth, substrate type, pH and dissolved oxygen were identified as the key environmental variables driving habitat heterogeneity. Additionally, communities composed of characteristic taxa with VME attributes were observed, including (i) sponge aggregations, (ii) scleractinian, (iii) octocoral and black CWC gardens, and (iv) sea pen fields. Despite the apparent low impact of fishing activities in the study area, lost fishing gear was noted in areas where CWCs were observed in higher densities. This study provides important information for the implementation of area-based conservation measures in Cabo Verde, following the application of the precautionary principle.

Cold-water corals and sponges form iconic and globally occurring benthic communities, provide important habitats for a diverse associated fauna and thrive in environmental conditions with often large temporal and spatial variations in near-bottom currents, food availability and other environmental parameters. We investigate the variability of near‐bottom currents and physical processes from simulations with a nested hydrodynamic modelling framework at two seamounts rich in benthic fauna along the Northeast Walvis Ridge, Valdivia Bank and Ewing Seamount. Our aim is to obtain new insights on physical drivers of observed occurrences and distribution of benthic suspension feeders (cnidarians and sponges) in this data‐poor area. We use dynamic downscaling of high-resolution implementations of the ROMS-AGRIF model in combination with high-resolution bathymetry and open boundary forcing from the basin-scale model INALT20 and the OSU inverse tidal model to explore the fine-scale physical processes and mechanisms that potentially drive a continuous or episodic food supply to the benthic communities. Over a three-year period, we analysed how near-bottom currents vary in space and time and assess potential connections between the distribution of filter-feeding fauna and the surrounding physical marine environment. We identified a close link between flow dynamics, internal tide dynamics and faunal species distributions. We propose that physical processes such as kinetic energy dissipation and internal wave dynamics could serve as functional indicators of food supply and particle encounter rates in future species distribution and habitat suitability models for important deep-sea taxa, such as those that represent vulnerable marine ecosystems. Our results also show little impact of mesoscale eddies from the Agulhas Leakage as they propagate north-westward into the southeast Atlantic along a well-defined corridor, which only occasionally extends as far north as the Valdivia Bank and Ewing Seamount.

Following a number of meetings devoted to knowledge sharing, identification of key issues, and discussing the best ways to move forward, a wide international expert community is now able to provide recommendations regarding the monitoring of seafloor macrolitter through observation and imaging. As the seafloor constitutes a major sink for marine litter including plastics, it is important to acquire robust and extensive data on litter distribution, abundance, types and size ranges across marine habitats. This should be done through widely agreed, harmonised, and non-destructive methods encompassing advanced technologies. Training and capacity building are essential elements in this endeavour. Both new and legacy imagery are needed to establish baseline assessments and trends. Informing policy-making is indispensable for effective action through upstream and targeted measures, with seafloor macrolitter (and megalitter) being a vital part of the evidence base for global mitigation measures.

Sea pens are colonial octocorals inhabiting mostly muddy and sandy soft sediments, globally. Despite their cosmopolitan distribution, evidence of high functional value, and susceptibility to external stressors, sea pens remain poorly investigated relative to other coral species. Here, a quantitative analysis of the fine-scale spatial patterns of deep-sea sea pens (Octocorallia: Pennatuloidea) within a submarine canyon conservation area in English waters is conducted using Remotely Operated Vehicle (ROV) seabed images, three-dimensional photogrammetry models, and Spatial Point Pattern Analysis (SPPA). The density distribution of sea pens at the canyon flank scale (100s of metres) is also reported for context. At the canyon flank scale, sea pen density decreases with increasing depth, with colonies virtually absent from waters deeper than ca. 1100 m. At the fine scale, sea pens exhibit a tendency to cluster relative to sea pens and tube-dwelling anemones (Cerianthidae sp.), suggesting mutual habitat associations. Conversely, overdispersion best characterises the nature of sea pen distribution relative to stalked sponges (Hyalonema sp.), suggesting occurrence of competition or, possibly, allelopathy. This study demonstrates that high-resolution spatial analysis techniques can be employed to disentangle biotic interactions among a set of taxa, revealing key processes and structures underlying the ecology of benthic fauna.

Deep-sea polymetallic nodule mining is in the exploration phase at present with some groups proposing a move towards extraction within years1. Management of this industry requires evidence of the long-term effects on deep-sea ecosystems2, but the ability of seafloor ecosystems to recover from impacts over decadal scales is poorly understood3. Here we show that, four decades after a test mining experiment that removed nodules, the biological impacts in many groups of organisms are persistent, although populations of several organisms, including sediment macrofauna, mobile deposit feeders and even large-sized sessile fauna, have begun to re-establish despite persistent physical changes at the seafloor. We also reveal that areas affected by plumes from this small-scale test have limited detectable residual sedimentation impacts with some biological assemblages similar in abundance compared to control areas after 44 years. Although some aspects of the modern collector design may cause reduced physical impact compared to this test mining experiment, our results show that mining impacts in the abyssal ocean will be persistent over at least decadal timeframes and communities will remain altered in directly disturbed areas, despite some recolonization. The long-term effects seen in our study provide critical data for effective management of mining activities, if they occur, including minimizing direct impacts and setting aside an effective network of protected areas4,5.

The Darwin Mounds marine protected area has been afforded protection from bottom trawl fishing since August 2003, following the discovery of cold-water coral (CWC) communities in June 1998. Surveys of the area in 2000 provided evidence that deep-water trawling activities were impacting the corals, prompting the development of a conservation response. Here we report the most recent survey of these CWCs in 2019, contrasting those data with a prior survey in 2011, and the earliest observations in the area (1998-2000). Our assessment is focussed on the colonial scleractinian corals Desmophyllum pertusum and Madrepora oculata. The status of the CWCs was determined using seafloor visual imagery from a remotely operated vehicle (2011), off-bottom towed cameras (1998-2000, 2019) and additionally draws on images from an autonomous underwater vehicle seabed survey (2019). Considering the numerical density, seabed cover, and size distribution of living CWCs, no evidence was detected that the previously impacted corals had recovered. The order of magnitude reduction in live coral abundance in impacted areas remained evident in the 2019 survey after 16-years of protection. Given the likely growth rates of D. pertusum and M. oculata, we suggest that a multi-decadal recovery period should be expected. Our interpretation of long-term change was complicated by the evolving monitoring methodology employed, a common problem and tension in the development of long-term offshore ecosystem monitoring programmes. We further consider a prospectus for effective and efficient future monitoring, noting that autonomous systems and computer vision techniques are likely to play an increasingly important role.

This study investigates near-bottom currents and physical processes from simulations with the hydrodynamic model ROMS-AGRIF at two seamounts of the northeast Walvis Ridge to obtain valuable insights about drivers of observed occurrences of benthic suspension feeders (cnidarians and sponges) in this data-poor area. The spatial resolution in each model area was increased across two levels of nested grids from 1,500 m to 500 m resolution with 32 stretched terrain-following (s-) layers in the vertical with high resolution close to the bottom. The parent grids receive initial and boundary conditions from the basin-scale model INALT20 and from solutions of the OTIS inverse tidal model. The model topography is based on GEBCO data with local refinements from multi-beam data collected during different surveys in 2008, 2009, and 2010. Increasing model resolution is an important advancement for precisely evaluating the intrinsic dynamics within challenging rough terrain. We examined how near-bottom currents vary over space and time and investigated potential links between observed Cnidarian and Porifera occurrences and ranges of physical variables and processes. We identified a close link between physical processes and species distributions and suggested that physical processes such as kinetic energy dissipation and internal wave dynamics may be considered in future research as proxies of food supply to benthic suspension feeders. Such mechanistic variables may also be used to supplement more traditional descriptors such as water mass and terrain properties in species distribution models, thus enhancing our ability to predict the occurrence of benthic communities characterized by cnidarians and sponges.

Seafloor surveys often gather multiple modes of remote sensed mapping and sampling data to infer kilo- to mega-hectare scale seafloor habitat distributions. However, efforts to extract information from multimodal data are complicated by inconsistencies between measurement modes (e.g., resolution, positional offsets, geometric distortions) and different acquisition periods for dynamically changing environments. In this study, we investigate the use of location information during multimodal feature learning and its impact on habitat classification. Experiments on multimodal datasets gathered from three Marine Protected Areas (MPAs) showed improved robustness and performance when using location-based regularisation terms compared to equivalent autoencoder-based and contrastive self-supervised feature learners. Location-guiding improved F1 scores by 7.7% for autoencoder-based and 28.8% for contrastive feature learners averaged across 78 experiments on datasets spanning three distinct sites and 18 data modes. Location-guiding enhances performance when combining multimodal data, increasing F1 scores by an average of 8.8% and 37.8% compared to the best-performing individual mode being combined for autoencoder-based and contrastive self-supervised models, respectively. Performance gains are maintained over a large range of location-guiding distance hyperparameters, where improvements of 5.3% and 29.4% are achieved on average over an order-of-magnitude range of hyperparameters for the autoencoder and contrastive learners, respectively, both comparing favourably with optimally tuned conditions. Location-guiding also exhibits robustness to position inconsistencies between combined data modes, still achieving an average of 3.0% and 30.4% increase in performance compared to equivalent feature learners without location regularisation when position offsets of up to 10 m are artificially introduced to the remote sensed data. Our results show that the classifier used to delineate the learned feature spaces has less impact on performance than the feature learner, with probabilistic classifiers averaging 3.4% higher F1 scores than non-probabilistic classifiers.

AimSeamounts are conspicuous geological features with an important ecological role and can be considered vulnerable marine ecosystems (VMEs). Since many deep-sea regions remain largely unexplored, investigating the occurrence of VME taxa on seamounts is challenging. Our study aimed to predict the distribution of four cold-water coral (CWC) taxa, indicators for VMEs, in a region where occurrence data are scarce.LocationSeamounts around the Cabo Verde archipelago (NW Africa).MethodsWe used species presence-absence data obtained from remotely operated vehicle (ROV) footage collected during two research expeditions. Terrain variables calculated using a multiscale approach from a 100-m-resolution bathymetry grid, as well as physical oceanographical data from the VIKING20X model, at a native resolution of 1/20 degrees, were used as environmental predictors. Two modelling techniques (generalized additive model and random forest) were employed and single-model predictions were combined into a final weighted-average ensemble model. Model performance was validated using different metrics through cross-validation.ResultsTerrain orientation, at broad scale, presented one of the highest relative variable contributions to the distribution models of all CWC taxa, suggesting that hydrodynamic-topographic interactions on the seamounts could benefit CWCs by maximizing food supply. However, changes at finer scales in terrain morphology and bottom salinity were important for driving differences in the distribution of specific CWCs. The ensemble model predicted the presence of VME taxa on all seamounts and consistently achieved the highest performance metrics, outperforming individual models. Nonetheless, model extrapolation and uncertainty, measured as the coefficient of variation, were high, particularly, in least surveyed areas across seamounts, highlighting the need to collect more data in future surveys.Main ConclusionsOur study shows how data-poor areas may be assessed for the likelihood of VMEs and provides important information to guide future research in Cabo Verde, which is fundamental to advise ongoing conservation planning. ObjetivoMontes submarinos s & atilde;o importantes forma & ccedil;& otilde;es geol & oacute;gicas com um not & aacute;vel papel ecol & oacute;gico e podem ser considerados Ecossistemas Marinhos Vulner & aacute;veis (VMEs). Dado que muitas regi & otilde;es do mar profundo permanecem inexploradas, investigar a ocorr & ecirc;ncia de esp & eacute;cies indicadoras de VMEs & eacute; um desafio. O nosso estudo teve como objetivo prever a distribui & ccedil;& atilde;o de quatro taxa de corais de & aacute;gua fria (CWC), indicadores de VMEs, numa regi & atilde;o onde dados de ocorr & ecirc;ncia s & atilde;o escassos.Localiza & ccedil;& atilde;oMontes submarinos no Arquip & eacute;lago de Cabo Verde (NO & Aacute;frica).M & eacute;todosUtilizamos dados de presen & ccedil;a-aus & ecirc;ncia de CWC obtidos a partir de imagens de um Ve & iacute;culo de Opera & ccedil;& atilde;o Remota (ROV) durante duas expedi & ccedil;& otilde;es cient & iacute;ficas. Como dados ambientais foram utilizados vari & aacute;veis de terreno calculadas com uma abordagem multi-escala a partir de uma grelha de batimetria com 100 m de resolu & ccedil;& atilde;o, e dados de oceanografia f & iacute;sica obtidos com o modelo VIKING20X, a uma resolu & ccedil;& atilde;o nativa de 1/20 degrees. Duas t & eacute;cnicas de modela & ccedil;& atilde;o (Generalized Additive Models (GAM) e Random Forest) foram usadas e resultados de modelos individuais foram combinados, atrav & eacute;s da m & eacute;dia ponderada, num modelo final Ensemble. O desempenho dos modelos foi validado usando diferentes m & eacute;tricas atrav & eacute;s de t & eacute;cnicas de valida & ccedil;& atilde;o cruzada.ResultadosA orienta & ccedil;& atilde;o do terreno, a larga escala, apresentou uma das maiores contribui & ccedil;& otilde;es relativas para os modelos de distribui & ccedil;& atilde;o de todos os CWCs, sugerindo que intera & ccedil;& otilde;es de hidrodin & acirc;mica com topografia beneficiam os corais, possivelmente pelo aumento da disponibilidade de alimento. No entanto, mudan & ccedil;as na morfologia de terreno a escalas mais finas e salinidade foram importantes para diferen & ccedil;as entre a distribui & ccedil;& atilde;o de esp & eacute;cies espec & iacute;ficas. O modelo Ensemble projetou a presen & ccedil;a de indicadores de VMEs em todos os montes submarinos e, consistentemente, apresentou m & eacute;tricas de desempenho mais altas, superando modelos individuais. No entanto, medidas de extrapola & ccedil;& atilde;o e incerteza foram elevadas, especialmente em & aacute;reas menos estudadas, destacando claramente a necessidade de recolher mais dados.Conclus & atilde;oO nosso estudo mostra como & aacute;reas com poucos dados podem ser avaliadas quanto & agrave; probabilidade de VMEs e fornece informa & ccedil;& otilde;es importantes para guiar futuras investiga & ccedil;& otilde;es em Cabo Verde, sendo fundamental para aconselhar planos de conserva & ccedil;& atilde;o em curso.

Deep-sea currents transfer sediment, nutrients, and pollutants, which drive climatic, ecological and geomorphological variation in the global ocean. The complex interaction of downslope currents and internal tides in submarine canyons has meant that interpreting their stratigraphic record and therefore reconstructing oceanic environments through geological time has proven challenging. We integrate flow measurements with sediment core observations from the Whittard Canyon, to determine whether the stratigraphic signature of turbidity current and internal tide interaction is preserved. Sand is transported by turbidity currents and re-worked by internal tides, forming a suite of characteristic deposits; near-bed flow measurements show that turbidity currents superposed on internal tides collectively exceed a critical bed shear stress for mobilizing fine sand at least 1% of a year, suspending sediment tens of meters above the bed over longer periods. Using these observations, we present a framework to recognize this interaction in the stratigraphic record.

The Confined Morphologies Mapping (CoMMa) Toolbox, a novel ArcGIS Pro python toolbox expressly created for semi-automated seabed morphological mapping, is presented here. The toolbox includes a selection of tools for the pre-processing, delineation and description of confined features on a digital elevation model (DEM) that are either negative or positive. The CoMMa Toolbox addresses the need for a flexible and multi-faceted solution applicable to different mapping problems, also encapsulating and re-interpreting existing methodologies. This study also evaluates, qualitatively and quantitatively, the performance of CoMMa delineations performed on a synthetic bathymetry DEM with 150 coral mounds of known characteristics against manual digitisations completed by five expert geomorphologists. The results show that the best CoMMa delineation falls within the range of competence demonstrated by the expert manual mappers. Edge evaluation metrics and attribute error scores are comparable or often superior to four of the five human delineations, although the Toolbox never reaches the performance of the best expert. Nevertheless, the semi-automated techniques can be of assistance to any user, providing rapid, visually unbiased and consistent delineations, thus saving time: they can then be optimised manually where desirable. Moreover, while the toolbox was created for marine geomorphometry, it can be applied to any DEM, either marine, terrestrial or extra-terrestrial.The CoMMa Toolbox is available in a public GitHub repository with a thorough user guide.

The geomorphology of the Blanes submarine canyon has been characterized based on the analysis of highresolution hull-mounted multibeam bathymetry (up to 5 m grid size). Additionally, the Hybrid Remotely Operated Vehicle (H-ROV) Ariane was used to map with unprecedented detail (80 cm grid size) the morphology of vertical rockwalls on the canyon flanks, and the ROV Liropus was used to acquire high-resolution video footage on the same walls. The Blanes Canyon exposes well stratified Miocene and Plio-Quaternary successions dipping towards a SW direction, which are affected by NE-SW and NW-SE oriented fault systems that played an important role in the canyon development. Its structural character is evidenced by the rectilinear trajectories and sharp bends of the canyon axis and rims, the alignment of pockmark fields and the presence of vertical rockwalls. The canyon transversal profile is markedly asymmetric due to the underlying stratigraphy. The eastern flank, corresponding to the cataclinal slope characterized by strata beds dipping in the same direction as the slope, is smooth and mainly dominated by slides and toe gullies. The western flank, corresponding to the anaclinal slope facing opposite to the dip of the strata, is steeper and dominated by a dendritic network of rim gullies. H-ROV bathymetric maps and video footage of vertical walls display a wide variety of fine-scale morphological elements that evidence ongoing mass wasting and gully development as the main mechanisms involved in canyon wall retrogression. The multi-scale study of the Blanes Canyon has allowed a better characterization of the erosive processes involved in the broadening of submarine canyons and retrogression of the walls, thus contributing to the better understanding of the evolution of submarine canyons developed in tectonically controlled stratigraphic bedded sequences.

Remote sensed mapping data and seafloor in-situ imagery are often gathered to infer benthic habitat distributions. However, leveraging multimodal data is challenging because of inherent inconsistencies between measurement modes (e.g., resolution, positional offsets, shape discrepancies). We investigate the impact of using location metadata in multimodal, self-supervised feature learning on habitat classification. Experiments were carried out on a multimodal dataset gathered using and Autonomous Underwater Vehicle (AUV) at the Darwin Mounds Marine Protected Area (MPA). Introducing location metadata improved F1 classification performance of a Bayesian classifier by an average of 27.7% over all conditions tested in this work, with a larger improvement of 32.9% achieved when multiple remote sensing data modes are combined for the analysis.

Seabed cover of organisms is an established metric for assessing the status of many vulnerable marine ecosystems. When deriving cover estimates from seafloor imagery, a source of uncertainty in capturing the true distribution of organisms is introduced by the inherent variability and bias of the annotation method used to extract ecological data. We investigated variability and bias in two common annotation methods for estimating organism cover, and the role of size selectivity in this variability. Eleven annotators estimated sparse cold-water coral cover in the same 96 images with both grid-based and manual segmentation annotation methods. The standard deviation between annotators was three times greater in the grid-based method compared to segmentation, and grid-based estimates from annotators tended to overestimate coral cover. Size selectivity biased the manual segmentation; the minimum size of colonies segmented varied between annotators fivefold. Two modelling techniques (based on Richard’s selection curves and Gaussian processes) were used to impute areas where annotators identified colonies too small for segmentation. By imputing small coral sizes in segmentation estimates, the coefficient of variation between annotators was reduced by approximately 10%, and method bias (compared to a reference dataset) was reduced by up to 23%. Therefore, for sparse, low cover organisms, manual segmentation of images is recommended to minimise annotator variability and bias. Uncertainty in cover estimates may be further reduced by addressing size selectivity bias when annotating small organisms in images using a data-driven modelling technique.

EDITORIAL article Front. Mar. Sci., 09 November 2023Sec. Deep-Sea Environments and Ecology Volume 10 - 2023 | https://doi.org/10.3389/fmars.2023.1304429

Ocean ecosystems are at the forefront of the climate and biodiversity crises, yet we lack a unified approach to assess their state and inform sustainable policies. This blueprint is designed around research capabilities and cross-sectoral partnerships. We highlight priorities including integrating basin-scale observation, modelling and genomic approaches to understand Atlantic oceanography and ecosystem connectivity; improving ecosystem mapping; identifying potential tipping points in deep and open ocean ecosystems; understanding compound impacts of multiple stressors including warming, acidification and deoxygenation; enhancing spatial and temporal management and protection. We argue that these goals are best achieved through partnerships with policy-makers and community stakeholders, and promoting research groups from the South Atlantic through investment and engagement. Given the high costs of such research (€800k to €1.7M per expedition and €30–40M for a basin-scale programme), international cooperation and funding are integral to supporting science-led policies to conserve ocean ecosystems that transcend jurisdictional borders.