A significant proportion of Southern Ocean seafloor biodiversity is thought to be associated with fragile, slow growing, long-lived, and habitat-forming taxa. Minimizing adverse impact to these so-called vulnerable marine ecosystems (VMEs) is a conservation priority that is often managed by relying on fisheries bycatch data, combined with threshold-based conservation rules in which all “indicator” taxa are considered equal. However, VME indicator taxa have different vulnerabilities to fishing disturbance and more consideration needs to be given to how these taxa may combine to form components of ecosystems with high conservation value. Here, we propose a multi-criteria approach to VME identification that explicitly considers multiple taxa identified from imagery as VME indicator morpho-taxa. Each VME indicator morpho-taxon is weighted differently, based on its vulnerability to fishing. Using the “Antarctic Seafloor Annotated Imagery Database”, where 53 VME indicator morpho-taxa were manually annotated generating >40000 annotations, we computed an index of cumulative abundance and overall richness and assigned it to spatial grid cells. Our analysis quantifies the assemblage-level vulnerability to fishing, and allows assemblages to be characterized, e.g. as highly diverse or highly abundant. The implementation of this quantitative method is intended to enhance VME identification and contextualize the bycatch events.
Human activity puts our oceans under multiple stresses, whose impacts are already significantly affecting biodiversity and physicochemical properties. Consequently, there is an increased international focus on the conservation and sustainable use of oceans, including the protection of fragile benthic biodiversity hotspots in the deep sea, identified as vulnerable marine ecosystems (VMEs). International VME risk assessment and conservation efforts are hampered because we largely do not know where VMEs are located. VME distribution modelling has increasingly been recommended to extend our knowledge beyond sparse observations. Nevertheless, the adoption of VME distribution models in spatial management planning and conservation remains limited. This work critically reviews VME distribution modelling studies, and recommends promising avenues to make VME models more relevant and impactful for policy and management decision making. First, there is an important interplay between the type of VME data used to build models and how the generated maps can be used in making management decisions, which is often ignored by model-builders. Overall, there is a need for more precise VME data for production of reliable models. We provide specific guidelines for seven common applications of VME distribution modelling to improve the matching between the modelling and the user need. Second, the current criteria to identify VME often rely on subjective thresholds, which limits the transparency, transferability and effective applicability of distribution models in protection measures. We encourage scientists towards founding their models on: (i) specific and quantitative definitions of what constitute a VME, (ii) site conservation value assessment in relation to VME multi-taxon spatial predictions, and (iii) explicitly mapping vulnerability. Along with the recent increase in both deep-sea biological and environmental data quality and quantity, these modelling recommendations can lead towards more cohesive summaries of VME’s spatial distributions and their relative vulnerability, which should facilitate a more effective protection of these ecosystems, as has been mandated by numerous international agreements.
Marine imagery is a comparatively cost-effective way to collect data on seafloor organisms, biodiversity and habitat morphology. However, annotating these images to extract detailed biological information is time-consuming and expensive, and reference libraries of consistently annotated seafloor images are rarely publicly available. Here, we present the Antarctic Seafloor Annotated Imagery Database (AS-AID), a result of a multinational collaboration to collate and annotate regional seafloor imagery datasets from 19 Antarctic research cruises between 1985 and 2019. AS-AID comprises of 3,599 georeferenced downward facing seafloor images that have been labelled with a total of 615,051 expert annotations. Annotations are based on the CATAMI (Collaborative and Automated Tools for Analysis of Marine Imagery) classification scheme and have been reviewed by experts. In addition, because the pixel location of each annotation within each image is available, annotations can be viewed easily and customised to suit individual research priorities. This dataset can be used to investigate species distributions, community patterns, it provides a reference to assess change through time, and can be used to train algorithms to automatically detect and annotate marine fauna.
Human activity puts our oceans under multiple stresses, whose impacts are already significantly affecting biodiversity and physicochemical properties. Consequently, there is an increased international focus on the conservation and sustainable use of oceans, including the protection of fragile benthic biodiversity hotspots in the deep sea, identified as vulnerable marine ecosystems (VMEs). International VME risk assessment and conservation efforts are hampered because we largely do not know where VMEs are located. VME distribution modelling has increasingly been recommended to extend our knowledge beyond sparse observations. Nevertheless, the adoption of VME distribution models in spatial management planning and conservation remains limited. This work critically reviews VME distribution modelling studies, and recommends promising avenues to make VME models more relevant and impactful for policy and management decision making. First, there is an important interplay between the type of VME data used to build models and how the generated maps can be used in making management decisions, which is often ignored by model-builders. We encourage scientists towards founding their models on: (i) specific and quantitative definitions of what constitute a VME, (ii) site conservation value assessment in relation to VME multi-taxon spatial predictions, and (iii) explicitly mapping vulnerability. Along with the recent increase in both deep-sea biological and environmental data quality and quantity, these modelling recommendations can lead towards more cohesive summaries of VME’s spatial distributions and their relative vulnerability, which should facilitate a more effective protection of these ecosystems, as has been mandated by numerous international agreements.
Medical tasks are prone to inter-rater variability due to multiple factors such as image quality, professional experience and training, or guideline clarity. Training deep learning networks with annotations from multiple raters is a common practice that mitigates the model’s bias towards a single expert. Reliable models generating calibrated outputs and reflecting the inter-rater disagreement are key to the integration of artificial intelligence in clinical practice. Various methods exist to take into account different expert labels. We focus on comparing three label fusion methods: STAPLE, average of the rater’s segmentation, and random sampling of each rater’s segmentation during training. Each label fusion method is studied using both the conventional training framework and the recently published SoftSeg framework that limits information loss by treating the segmentation task as a regression. Our results, across 10 data splittings on two public datasets (spinal cord gray matter challenge, and multiple sclerosis brain lesion segmentation), indicate that SoftSeg models, regardless of the ground truth fusion method, had better calibration and preservation of the inter-rater rater variability compared with their conventional counterparts without impacting the segmentation performance. Conventional models, i.e., trained with a Dice loss, with binary inputs, and sigmoid/softmax final activate, were overconfident and underestimated the uncertainty associated with inter-rater variability. Conversely, fusing labels by averaging with the SoftSeg framework led to underconfident outputs and overestimation of the rater disagreement. In terms of segmentation performance, the best label fusion method was different for the two datasets studied, indicating this parameter might be task-dependent. However, SoftSeg had segmentation performance systematically superior or equal to the conventionally trained models and had the best calibration and preservation of the inter-rater variability. SoftSeg has a low computational cost and performed similarly in terms of uncertainty to ensembles which require multiple models and forward passes. Our code is available at <href='https://ivadomed.org'>https://ivadomed.org</a>.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.