Deep Learning Based Characterization of Cold-Water Coral Habitat at Central Cantabrian Natura 2000 Sites Using YOLOv8

Abstract

Cold-water coral (CWC) reefs, such as those formed by Desmophyllum pertusum and Madrepora oculata, are vital yet vulnerable marine ecosystems (VMEs). The need for accurate and efficient monitoring of these habitats has driven the exploration of innovative approaches. This study presents a novel application of the YOLOv8l-seg deep learning model for the automated detection and segmentation of these key CWC species in underwater imagery. The model was trained and validated on images collected at two Natura 2000 sites in the Cantabrian Sea: the Avilés Canyon System (ACS) and El Cachucho Seamount (CSM). Results demonstrate the model’s high accuracy in identifying and delineating individual coral colonies, enabling the assessment of coral cover and spatial distribution. The study revealed significant variability in coral cover between and within the study areas, highlighting the patchy nature of CWC habitats. Three distinct coral community groups were identified based on percentage coverage composition and abundance, with the highest coral cover group being located exclusively in the La Gaviera canyon head within the ACS. This research underscores the potential of deep learning models for efficient and accurate monitoring of VMEs, facilitating the acquisition of high-resolution data essential for understanding CWC distribution, abundance, and community structure, and ultimately contributing to the development of effective conservation strategies.

1. Introduction

Cold-water coral (CWC) reefs, such as those formed by the framework-building scleractinians Desmophyllum pertusum (Linneus, 1758) and Madrepora oculata (Linneus, 1758), are vital yet vulnerable marine ecosystems (VMEs) renowned for their biodiversity and crucial role in deep-sea environments [1]. The ecological significance of CWC reefs lies in their ability to create complex three-dimensional structures that provide habitat [2], feeding grounds [3], and nursery areas for a diverse array of marine organisms [4,5], thereby enhancing overall biodiversity and biomass in the deep sea. The intricate framework of these reefs also influences critical ecological processes, including larval dispersal, retention, and feeding efficiency, further underscoring their importance in maintaining the health and productivity of deep-sea ecosystems [6,7,8].

These reefs are found in diverse locations, including continental slopes, seamounts, fjords, and submarine canyons [9,10]. The unique geomorphological characteristics of submarine canyons and seamounts, with their complex topography and steep slopes, offer natural refuges for CWC reefs, shielding them from destructive fishing practices and other anthropogenic disturbances [11,12]. The varied terrain and hydrodynamic conditions associated with these geological formations create a mosaic of habitats that support a rich tapestry of benthic communities, including the iconic CWC reefs.

The scleractinian corals, D. pertusum and M. oculata, are key ecosystem engineers in the deep sea, constructing the framework of CWC reefs that provide essential habitat for numerous associated species. These corals, often referred to as ‘white corals’ due to their ahermatypic nature, do not rely on symbiotic algae for nutrition and can thrive in the cold, dark depths of the ocean. The presence of these corals fosters a diverse assemblage of fauna, including bivalves, gastropods, echinoderms, sponges, and worms, many of which utilize the coral skeletons for attachment or as a source of food [13]. The structural complexity of CWC reefs, with their numerous crevices and overhangs, creates microhabitats that support a wide range of ecological niches, further contributing to the high biodiversity associated with these ecosystems [14].

Despite their ecological importance, CWC reefs face numerous threats, including bottom trawling, deep-sea mining, and climate change [15,16]. These anthropogenic pressures pose a significant risk to the integrity and persistence of CWC reefs, potentially leading to habitat degradation, loss of biodiversity, and disruption of ecosystem functions [17]. The slow growth rates and fragile nature of many CWC species make them particularly vulnerable to disturbance, and recovery from damage can take decades or even centuries [18]. The increasing recognition of the threats facing CWC reefs has led to their designation as VMEs by the United Nations General Assembly (Resolution 61/105), highlighting the urgent need for their protection and conservation [1].

Accurate identification and delineation of CWC reefs, particularly from underwater imagery, remains a challenge due to the lack of precise information on coral cover and the associated structural complexity. Traditional methods of monitoring and assessing CWC reefs often rely on labor-intensive manual annotation of images or video footage, which can be time-consuming and prone to subjective interpretation [19]. The vastness and remoteness of deep-sea environments further complicate efforts to obtain comprehensive and representative data on CWC distribution and abundance.

Recent advances in computer vision techniques, particularly deep learning-based object detection and segmentation models, have offered promising solutions to these challenges. These models, trained on large labeled datasets, can automatically identify and delineate coral colonies in underwater images, thereby enabling accurate estimation of coral cover and spatial distribution [20,21]. The application of deep learning in marine ecological research has gained significant traction in recent years, demonstrating its potential to revolutionize the way we study and monitor marine ecosystems. In the realm of underwater object detection, this task is particularly challenging due to the unique characteristics of the marine environment, such as poor visibility, light attenuation, and complex backgrounds. Traditional object detection methods, including the Region-based Convolutional Neural Network (R-CNN) series [22,23], have been explored for underwater applications, but often face limitations in computational efficiency and real-time performance. The emergence of single-stage detectors, such as the You Only Look Once (YOLO) series [24,25,26], has addressed these limitations, offering a faster and more streamlined approach. YOLO models have demonstrated remarkable success in various domains, including underwater object detection [21]. Their ability to simultaneously predict bounding boxes and class probabilities in a single pass contributes to their efficiency and suitability for real-time applications. Among these models, YOLOv8 has emerged as a powerful tool, demonstrating effectiveness in detecting various marine organisms, including corals in diverse environments [27].

In this study, we leveraged the power of YOLOv8 for the automated detection and segmentation of coral species in remotely operated towed vehicle (ROTV) imagery collected at two Natura 2000 sites in the Cantabrian Sea. Our objectives are threefold: (1) to develop novel methodologies for monitoring CWC VMEs, (2) to assess variability in coral cover across geographically proximate areas and among transects within each area, and (3) to characterize these CWC communities in terms of CWC coverage. By analyzing ROTV imagery, we aim to overcome the challenges associated with manual annotation and obtain accurate quantitative data for improved understanding and management of these ecosystems.

2. Materials and Methods

2.1. Study Area

This study, framed within the INTEMARES project, focuses on two regions of the bathyal rocky outcrops in the Cantabrian Sea, south of the Bay of Biscay (Figure 1): the Avilés Canyon System (ACS) and El Cachucho Seamount (CSM). These areas were selected due to their designation as vulnerable marine ecosystems (VMEs) and their harboring of benthic communities classified as habitat 1170 (Reefs) under the European Union Habitats Directive (92/43/EEC). Of particular interest within these communities are the cold-water coral reefs, which are a focal point of this research.

El Cachucho Seamount, designated a Marine Protected Area (MPA) and Special Area of Conservation (SAC) in 2011, is characterized by its complex geomorphology, featuring rocky outcrops and steep slopes [28]. Its summit, known as Le Danois Bank, lies at 425 m depth and predominantly consists of rocky outcrops with sparse sediment cover, contrasting with its inner basin (800–1000 m), where sediment accumulation is higher [12].

The ACS, a Site of Community Importance (SCI) and potential SAC within the Natura 2000 network, extends from the continental shelf to bathyal depths. It is characterized by rocky outcrops with diverse morphologies and relief, some of which exhibit tectonic activity [29]. This complex geomorphology creates a diverse habitat that supports rich benthic biodiversity [14,30].

The presence of key reef-building species, such as D. pertusum and M. oculata, highlights their ecological importance [31]. These white corals harbor a suite of associated fauna species such as bivalves, gastropods, echinoderms, sponges, and worms and host breeding grounds for fished species. These species do not depend on living corals, but use the skeletal remains as a substrate for fixation or grazing on sessile invertebrates [30]. The primary role of CWC reefs is to function as feeding grounds, refuges, and as substrata for larval settlement, juvenile growth, and as nursery areas. Furthermore, they contribute to the goods and services of the deep sea. Finally, the three-dimensional nature of these reefs increases the structural complexity of these ecosystems, making them particularly vulnerable and, therefore, deserving of special attention in terms of conservation [32]. In the study area, both species co-occur at different depths. D. pertusum has been recorded in the Avilés Canyon System (ACS) in a bathymetric distribution range of 342–1473 m; in the NW Atlantic, this is the most abundant and widely distributed construction species. M. oculata, in ACS, appears at 342–1660 m and is slow-growing and very vulnerable to trawling; it has construction activity on the continental margins of Europe.

2.2. Data Acquisition

High-resolution underwater imagery was obtained using two ROTV: Politolana [33] and TASIFE [34]. These vehicles are capable of descending to depths of 2000 m and are equipped with a high-resolution camera, bidirectional telemetry, and an acoustic positioning system (Figure 2). The camera was oriented in a zenith position, capturing images of the seabed at five-second intervals, and synchronized with environmental data to ensure the acquisition of comprehensive datasets during each dive.

A total of 19 transects were conducted on different dates in both study areas: 9 in the CSM and 10 in the ACS. To standardize data acquisition, the ROTV Politolana was maintained at a constant distance of 1.5 m from the seabed during all transects, each lasting 20 min and covering an average distance of 460 m. The samplings were conducted at depths ranging from 450 to 1200 m. The Avilés transects were associated with submarine canyon head areas, whereas most of the El Cachucho transects were located in areas adjacent to the seamount. This sampling approach provided a detailed view of the diversity of habitats and communities present in the study area, as well as an assessment of the influence of geomorphology and other environmental factors on species distribution.

2.3. Data Processing and Analysis

In this study, the YOLOv8l-seg model was employed for the detection and segmentation of the coral species, M. oculata and D. pertusum (Figure 3), due to several key advantages it offers. YOLOv8l-seg is a state-of-the-art model characterized by its unified architecture, capable of performing both object detection and instance segmentation in a single process [26]. This capability is particularly crucial in the analysis of complex underwater imagery, where corals often exhibit irregular shapes and may be partially obscured by other elements in the environment. Moreover, YOLOv8l-seg has consistently demonstrated superior performance compared to previous YOLO versions and other object detection models, such as Faster R-CNN, across a range of computer vision applications [35]. The model’s efficiency, accuracy, and ability to handle instance segmentation make it well-suited for the challenges of automating cold-water coral analysis in underwater images.

The YOLOv8l-seg architecture is based on a deep neural network composed of three main components:

• Backbone (CSPDarknet53): This component is responsible for extracting relevant features from the input images at different scales. The CSPDarknet53 architecture has proven highly effective in feature extraction for object detection tasks.
• Neck (Path Aggregation Network, PAN): This network combines the features extracted by the backbone at different scales, thereby enabling better detection of objects of various sizes. YOLOv8 utilizes a modified PAN structure to optimize this process.
• Head: This component performs final detection and segmentation predictions. In the case of YOLOv8l-seg, the head has two branches: one for object detection, predicting bounding boxes and object classes; and another for instance segmentation, generating accurate segmentation masks for each detected object.

The dataset used for model training and validation consisted of 670 coral images collected during various campaigns (see Acknowledgements) at Natura 2000 sites within the central Cantabrian region, representing a batch from each of the 19 transects conducted in the study areas. These images were manually labeled in YOLO format using the CVAT tool. The labeling process was optimized through an iterative approach that combined the training of an initial model with manual correction of the predictions generated by that model, utilizing the “auto_annotate” function of Ultralytics with a YOLOv8 model and SAM “mobile_sam.pt” [36]. Annotations in COCO format were converted to YOLO PyTorch. Model training was performed for 500 epochs with an initial learning rate of 0.01, applying data augmentation techniques to enhance model generalization.

For validation, 20% of the images (128 images) were reserved as an independent validation dataset, and 5-fold cross-validation (K-Fold) was implemented. Model performance was evaluated using metrics such as precision (B, M), recall (B, M), F1 score (B, M), intersection over union (IoU), mAP50, mAP50-95, and fitness. The complete source code, weights, and example data are available at: https://github.com/AlbertoGaya/cold-water-coral-reef/tree/main (27 August 2024).

Using the data extracted from the YOLOv8l-seg model, a comprehensive analysis was conducted to assess coral cover and species distribution in the study areas. The mean percentage of area covered and the number of individuals per species and transect were calculated. Non-parametric statistical tests (Kruskal–Wallis and Dunn’s test) were applied to compare coral cover between percentage areas and identify significant differences.

Additionally, non-metric multidimensional scaling (nMDS) was employed to visualize the similarity between transects based on coral reef composition, and hierarchical cluster analysis was used to group the most similar transects, with results presented on a map.

Regarding the experimental setup, all analyses were performed in a Jupyter Lab environment using Python 3.9. The Ultralytics package was employed for model training and inference. The hardware configuration included an Intel Core i7-13700F Processor (16 cores, 2.1 GHz), 16 GB DDR5 RAM (2 × 8 GB, 4800 MHz), and an NVIDIA GeForce RTX 4070 VENTUS 2X E 12G OC GPU.

3. Results

3.1. Model Performance Evaluation

The YOLOv8l-seg model demonstrated robust performance in the detection and segmentation of the target coral species, M. oculata and D. pertusum. Validation results, both in 5-fold cross-validation and independent validation, are summarized in

Table 1. Cross-validation results on bounding boxes (B) and masks (M), and independent validation results.

Validation	Precision (P)	Recall (R)	mAP50	mAP50-95
Cross-validation (B)	0.784	0.703	0.781	0.544
Cross-validation (M)	0.784	0.694	0.769	0.508
Independent validation	0.839	0.749	0.833	0.601

.

In the independent validation, the model achieved even higher performance, highlighting its ability to generalize to unseen data. Notably, the model exhibited slightly superior performance in detecting and segmenting D. pertusum (P = 0.876, R = 0.810) compared to M. oculata (P = 0.804, R = 0.693).

Overall, the validation results support the effectiveness of the YOLOv8l-seg model in the automated detection and segmentation of cold-water corals in underwater imagery, suggesting its potential as a valuable tool for monitoring and assessing these vulnerable ecosystems.

While precise runtime measurements were not collected in this study, the utilization of the YOLOv8l-seg model in conjunction with a dedicated GPU (NVIDIA GeForce RTX 4070) enabled efficient processing of the high-resolution imagery, facilitating the timely completion of the analysis.

3.2. Coral Cover Comparison between Study Areas

Descriptive analysis of the data revealed differences in coral cover among transects and study areas. The mean cover of D. pertusum was 0.56% ± 0.02% (range: 0–13.29%), while that of M. oculata was 0.40% ± 0.01% (range: 0–7.58%). The Kruskal–Wallis test confirmed significant differences in total coral cover both among transects (p < 0.001) and between CSM and ACS (p < 0.001), with higher cover in the latter.

Non-metric multidimensional scaling (nMDS) and hierarchical cluster analysis (Figure 4) identified three distinct groups based on coral species composition, showing an increasing gradient of cover from group yellow to group red.

Group yellow, with the lowest coral cover (maximum 3% for M. oculata), is mainly distributed in CSM. Group blue, with intermediate cover (maximum 6% for D. pertusum), is found in both CSM and ACS. Group red, with the highest cover (maximum 13% for M. oculata), is exclusively located in the La Gaviera canyon head within the ACS (Figure 5).

4. Discussion

The results of this study demonstrate the effectiveness of deep learning models, such as YOLOv8l-seg, in the automated detection and segmentation of cold-water corals in underwater imagery. The high performance of the model in terms of precision, recall, and mAP (Table 1) highlights its potential as a valuable tool for streamlining the monitoring and assessment of vulnerable ecosystems, overcoming the limitations of laborious manual annotation. These results are consistent with other studies that have successfully applied deep learning techniques for the automated identification of benthic fauna [27,35] and highlight the growing potential of these methods in marine ecological research.

Analysis of the data revealed significant variability in coral cover, not only among different transects, but also between the El Cachucho and Avilés areas. This discrepancy suggests that coral distribution is influenced by local factors, such as substrate density, curvature, and rugosity [37], as evidenced in the La Gaviera canyon head, where particular environmental conditions appear to favor higher coral cover. The identification of three distinct groups of transects based on coral species composition (Figure 4 and Figure 5) supports this hypothesis, showing an increasing gradient of cover from group yellow to group red (La Gaviera).

The variability in coral cover observed even among nearby transects highlights the inherent challenges associated with sampling these VMEs. Cold-water coral habitats often occur in discontinuous patches or are strongly delimited by specific conditions such as substrate type [38], depth, slope, orientation [39], and currents [40]. The fragmented and localized nature of cold-water coral reefs, coupled with the technological and logistical constraints associated with deep-sea research [41], makes it challenging to obtain a comprehensive and representative picture of the distribution and abundance of these species. In this regard, two transects conducted in La Gaviera, located in areas with conditions distinct from those of the reefs, exhibited very different coral cover values, reinforcing the notion of spatial heterogeneity in these ecosystems (Figure 6). Interestingly, all the transects located below 650 m belong to the yellow group. The red group, exclusively located in La Gaviera canyon head, has a mean depth of 773 m, while the blue group ranges between 750 and 1200 m. This suggests a potential depth gradient in coral cover, although the patchy nature of the transects and the limited sample size in different depth ranges require further research to confirm this hypothesis. However, our findings provide valuable preliminary evidence suggesting a potential depth-related pattern in cold-water coral distribution.

The application of deep learning models like YOLOv8l-seg, in conjunction with the collection of detailed environmental and biological data, presents a promising avenue for enhancing our understanding and capacity to protect cold-water coral reefs. The integration of environmental data, such as temperature, salinity, current velocity, and substrate characteristics, into predictive models could help elucidate the complex relationships between physical factors and coral distribution. Such models could also be used to forecast the potential impacts of climate change and other anthropogenic disturbances on these vulnerable ecosystems, informing the development of adaptive management strategies.

The promising results obtained with YOLOv8 in this study highlight the transformative potential of deep learning in facilitating the automated assessment of vulnerable marine ecosystems. The model’s efficacy in accurately detecting and segmenting cold-water corals, even within the challenging visual conditions of the deep sea, paves the way for more efficient and comprehensive monitoring efforts. However, we recognize that the inherent complexities of underwater imaging, such as low visibility and color distortion, present ongoing challenges [42]. Future research could explore the integration of advanced image enhancement techniques, leveraging innovations in reinforcement learning [43] or metalens technology [44], to further refine the accuracy and robustness of deep-sea object detection models. The expansion of automated detection and segmentation to encompass a wider range of benthic species, including sponges and gorgonians, would also significantly enhance our understanding of cold-water coral ecosystems’ structure and function [32]. Additionally, addressing limitations related to variations in image scale due to fluctuations in ROTV altitude and inconsistencies in data collection arising from disparities in transect design could further improve the precision and comparability of future studies. The continued integration of these advanced techniques with ongoing improvements in data acquisition and processing will undoubtedly enhance our capacity to study, monitor, and ultimately conserve these invaluable ecosystems.

5. Conclusions

The automated analysis of underwater imagery using YOLOv8l-seg has proven to be an effective tool for the detection and segmentation of cold-water coral species, facilitating the assessment and monitoring of these vulnerable ecosystems. Our results reveal significant variability in spatial coral cover, not only among geographically distinct areas but also between nearby transects within the same area, highlighting the inherently patchy and localized nature of these habitats. This heterogeneity underscores the challenges of sampling and monitoring cold-water coral reefs and emphasizes the need for comprehensive, high-resolution surveys to accurately assess their distribution and abundance. The observed depth gradient in coral cover, with a potential optimum range, warrants further investigation to understand the underlying ecological drivers.

The next step in this research involves leveraging the acquired coverage data and associated environmental variables (e.g., temperature, salinity, depth, substrate type, current flow) to develop a predictive model for cold-water coral species distribution. Such a model could help identify key environmental predictors of coral presence and abundance, enabling more targeted and efficient surveys and informing the design of effective conservation and management strategies, particularly in the context of seabed management.

Despite methodological limitations, this study provides valuable insights into the distribution of key species in Natura 2000 sites and lays the groundwork for future research integrating environmental data and expanding the range of species studied, thereby enhancing our ability to understand, manage, and conserve cold-water coral reefs. The continued integration of these advanced techniques with ongoing improvements in data acquisition and processing will undoubtedly enhance our ability to understand, manage, and conserve these invaluable ecosystems.