Section: Ecology
Topic:
Biophysics and computational biology,
Ecology
Predicting species distributions in the open ocean with convolutional neural networks
Corresponding author(s): Morand, Gaétan (gaetan.morand@umontpellier.fr)
10.24072/pcjournal.471 - Peer Community Journal, Volume 4 (2024), article no. e93.
Get full text PDF Peer reviewed and recommended by PCIAs biodiversity plummets due to anthropogenic disturbances, the conservation of oceanic species is made harder by limited knowledge of their distributions and migrations. Indeed, tracking species distributions in the open ocean is particularly challenging due to the scarcity of observations and the complex and variable nature of the ocean system. In this study, we propose a new method that leverages deep learning, specifically convolutional neural networks (CNNs), to capture spatial features of environmental variables. This novelty eliminates the need to predefine these features before modelling and creates opportunities to discover unexpected correlations. Our aim is to present the results of the first trial of this method in the open ocean, discuss limitations and provide feedback for future improvements or adjustments. In this case study, we considered 38 taxa comprising pelagic fishes, elasmobranchs, marine mammals, marine turtles and birds. We trained a model to predict probabilities from the environmental conditions at any specific point in space and time, using species occurrence data from the Global Biodiversity Information Facility (GBIF) and environmental data from various sources. These variables included sea surface temperature, chlorophyll concentration, salinity and fifteen others. During the testing phase, the model was applied to environmental data at locations where species occurrences were recorded. The classifier accurately predicted the observed taxon as the most likely taxon in 69% of cases and included the observed taxon among the top three most likely predictions in 89% of cases. These findings show the adequacy of deep learning for species distribution modelling in the open ocean. Additionally, this purely correlative model was then analysed with explicability tools to understand which variables had an influence on the model’s predictions. While variable importance was species-dependent, we identified finite-size Lyapunov exponents (FSLEs), sea surface temperature, pH and salinity as the most influential variables, in that order. These insights can prove valuable for future species-specific ecology studies.
Type: Research article
Morand, Gaétan 1; Joly, Alexis 2; Rouyer, Tristan 1; Lorieul, Titouan 2; Barde, Julien 1
@article{10_24072_pcjournal_471, author = {Morand, Ga\'etan and Joly, Alexis and Rouyer, Tristan and Lorieul, Titouan and Barde, Julien}, title = {Predicting species distributions in the open ocean with convolutional neural networks}, journal = {Peer Community Journal}, eid = {e93}, publisher = {Peer Community In}, volume = {4}, year = {2024}, doi = {10.24072/pcjournal.471}, language = {en}, url = {https://peercommunityjournal.org/articles/10.24072/pcjournal.471/} }
TY - JOUR AU - Morand, Gaétan AU - Joly, Alexis AU - Rouyer, Tristan AU - Lorieul, Titouan AU - Barde, Julien TI - Predicting species distributions in the open ocean with convolutional neural networks JO - Peer Community Journal PY - 2024 VL - 4 PB - Peer Community In UR - https://peercommunityjournal.org/articles/10.24072/pcjournal.471/ DO - 10.24072/pcjournal.471 LA - en ID - 10_24072_pcjournal_471 ER -
%0 Journal Article %A Morand, Gaétan %A Joly, Alexis %A Rouyer, Tristan %A Lorieul, Titouan %A Barde, Julien %T Predicting species distributions in the open ocean with convolutional neural networks %J Peer Community Journal %D 2024 %V 4 %I Peer Community In %U https://peercommunityjournal.org/articles/10.24072/pcjournal.471/ %R 10.24072/pcjournal.471 %G en %F 10_24072_pcjournal_471
Morand, Gaétan; Joly, Alexis; Rouyer, Tristan; Lorieul, Titouan; Barde, Julien. Predicting species distributions in the open ocean with convolutional neural networks. Peer Community Journal, Volume 4 (2024), article no. e93. doi : 10.24072/pcjournal.471. https://peercommunityjournal.org/articles/10.24072/pcjournal.471/
PCI peer reviews and recommendation, and links to data, scripts, code and supplementary information: 10.24072/pci.ecology.100584
Conflict of interest of the recommender and peer reviewers:
The recommender in charge of the evaluation of the article and the reviewers declared that they have no conflict of interest (as defined in the code of conduct of PCI) with the authors or with the content of the article.
[1] Dissolved organic carbon pools and export from the coastal ocean, Global Biogeochemical Cycles, Volume 29 (2015) no. 10, pp. 1725-1738 | DOI
[2] Nice weather for bettongs: using weather events, not climate means, in species distribution models, Ecography, Volume 35 (2012) no. 4, pp. 306-314 | DOI
[3] Fine-scale structures as spots of increased fish concentration in the open ocean, Scientific Reports, Volume 11 (2021) no. 1, p. 15805 | DOI
[4] In search of relevant predictors for marine species distribution modelling Using the MarineSPEED benchmark dataset, Diversity and Distributions, Volume 24 (2018) no. 2, pp. 144-157 | DOI
[5] A deep learning approach to species distribution modelling, Multimedia Tools and Applications for Environmental & Biodiversity Informatics, Springer International Publishing, Cham, 2018, pp. 169-199 | DOI
[6] Operation of zoogeographic barriers, Systematic Biology, Volume 23 (1974) no. 2, pp. 248-256 | DOI
[7] Modelling the oceanic habitats of two pelagic species using recreational fisheries data, Fisheries Oceanography, Volume 24 (2015) no. 5, pp. 463-477 | DOI
[8] Thermal preference predicts animal personality in Nile Tilapia Reochromis niloticus, Journal of Animal Ecology, Volume 85 (2016) no. 5, pp. 1389-1400 | DOI
[9] Analysis of the role and robustness of artificial intelligence in commodity image recognition under deep learning neural network, PLOS ONE, Volume 15 (2020) no. 7, p. e0235783 | DOI
[10] A simulation approach to assessing bias in a fisheries self-sampling programme, ICES Journal of Marine Science, Volume 79 (2022) no. 1, pp. 76-87 | DOI
[11] Multi-label learning from single positive labels, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 933-942 | DOI
[12] Convolutional neural networks improve species distribution modelling by capturing the spatial structure of the environment, PLOS Computational Biology, Volume 17 (2021) no. 4, p. e1008856 | DOI
[13] Explicitly integrating a third dimension in marine species distribution modelling, Marine Ecology Progress Series, Volume 564 (2017), pp. 1-8 | DOI
[14] Novel methods improve prediction of species’ distributions from occurrence data, Ecography, Volume 29 (2006) no. 2, pp. 129-151 | DOI
[15] AI-based mapping of the conservation status of orchid assemblages at global scale. Version 1, arXiv, 2024 | DOI
[16] Global ocean gridded L4 sea surface heights and derived variables NRT, Mercator Ocean International, 2017 | DOI
[17] Global ocean biogeochemistry hindcast, Mercator Ocean International, 2018 | DOI
[18] Global ocean biogeochemistry analysis and forecast, Mercator Ocean International, 2019 | DOI
[19] Multi observation global ocean 3D temperature salinity height geostrophic current and MLD, Mercator Ocean International, 2020 | DOI
[20] Global ocean gridded L4 sea surface heights and derived variables reprocessed (1993-ongoing), Mercator Ocean International, 2021 | DOI
[21] Global ocean L4 significant wave height from reprocessed satellite measurements, Mercator Ocean International, 2021 | DOI
[22] Global ocean colour (Copernicus-GlobColour), bio-geo-chemical, L4 (monthly and interpolated) from satellite observations (1997-ongoing), Mercator Ocean International, 2022 | DOI
[23] PyTorchLightning/Pytorch-Lightning: 0.7.6 release. Version 0.7.6, Zenodo, 2020 | DOI
[24] The importance of temporal resolution for niche modelling in dynamic marine environments, Journal of Biogeography, Volume 44 (2017) no. 12, pp. 2816-2827 | DOI
[25] Migration, residency, and homing of Bluefin Tuna in the western Mediterranean Sea, ICES Journal of Marine Science, Volume 71 (2014) no. 3, pp. 510-518 | DOI
[26] Oceanographic changes and exploitation drive the spatio-temporal dynamics of Atlantic Bluefin Tuna Thunnus thynnus, Fisheries Oceanography, Volume 23 (2014) no. 2, pp. 147-156 | DOI
[27] Spatial and temporal variability in the trans-Pacific migration of Pacific Bluefin Tuna (Thunnus orientalis) revealed by archival tags, Progress in Oceanography, Volume 162 (2018), pp. 52-65 | DOI
[28] Atmosphere‐ocean ozone exchange: a global modeling study of biogeochemical, atmospheric, and waterside turbulence dependencies, Global Biogeochemical Cycles, Volume 23 (2009) no. 4, p. 2008GB003301 | DOI
[29] Data quantity is more important than its spatial bias for predictive species distribution modelling, PeerJ, Volume 8 (2020), p. e10411 | DOI
[30] GBIF, 2023 (https://www.gbif.org/)
[31] The GEBCO_2022 grid - a continuous terrain model of the global oceans and land, Documents,Network common data form. Version 1. NERC EDS British Oceanographic Data Centre NOC, 2022 | DOI
[32] Global trends in ocean phytoplankton: a new assessment using revised ocean colour data, Remote Sensing Letters, Volume 8 (2017) no. 12, pp. 1102-1111 | DOI
[33] Predicting species distribution: offering more than simple habitat models, Ecology Letters, Volume 8 (2005) no. 9, pp. 993-1009 | DOI
[34] Predictive habitat distribution models in ecology, Ecological Modelling, Volume 135 (2000) no. 2-3, pp. 147-186 | DOI
[35] Deep residual learning for image recognition, In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 770-778 | DOI
[36] Summary for policymakers, Special Report on the Ocean and Cryosphere in a Changing Climate, 2019 | DOI
[37] Historical overfishing and the recent collapse of coastal ecosystems, Science, Volume 293 (2001) no. 5530, pp. 629-637 | DOI
[38] Captum: a unified and generic model interpretability library for PyTorch, arXiv, 2020 | DOI
[39] FSLE - Finite-Size Lyapunov Exponents and orientations of the associated eigenvectors (Version DT2021). [Dataset], CNES, 2021 | DOI
[40] Plantnet/Malpolon: Deep-SDM framework, 2023 (https://github.com/plantnet/malpolon)
[41] Chapter 19 - Ocean Pollution, Waste, Academic Press, Boston, 2011, pp. 265-279 | DOI
[42] Temporal resolutions in species distribution models of highly mobile marine animals: recommendations for ecologists and managers, Diversity and Distributions, Volume 23 (2017) no. 10, pp. 1098-1109 | DOI
[43] RSS Cross-Calibrated Multi-Platform (CCMP) 6-hourly ocean vector wind analysis on 0.25 deg grid, Version 3.0, Remote Sensing Systems (RSS) Air-Sea Essential Climate Variables (AS-ECV), 2022 | DOI
[44] Ecological niche models and species distribution models in marine environments: A literature review and spatial analysis of evidence, Ecological Modelling, Volume 415 (2020) | DOI
[45] Integrating dynamic environmental predictors and species occurrences: Toward true dynamic species distribution models, Ecology and Evolution, Volume 10 (2019) no. 2, pp. 1087-1092 | DOI
[46] Species distribution modeling, Geography Compass, Volume 4 (2010) no. 6, pp. 490-509 | DOI
[47] Frequent locations of oceanic fronts as an indicator of pelagic diversity: application to marine protected areas and renewables, Marine Policy, Volume 45 (2014), pp. 318-329 | DOI
[48] Delayed response of fish abundance to environmental changes: a novel multivariate time-lag approach, Marine Ecology Progress Series, Volume 456 (2012), pp. 159-168 | DOI
[49] Deep-SDMs in the open oceans - OUTPUTS - World, Zenodo, 2023 | DOI
[50] Deep-SDMs in the open oceans - OUTPUTS - western Indian Ocean, Zenodo, 2023 | DOI
[51] Deep-SDMs in the open oceans - OUTPUTS - World +2°C, Zenodo, 2023 | DOI
[52] Deep-SDMs in the open oceans - MODEL CHECKPOINT, Zenodo, 2023 | DOI
[53] Deep-SDMs in the open oceans - CODE, Zenodo, 2024 | DOI
[54] GeoEnrich v0.5.8: a new tool for scientists to painlessly enrich species occurrence data with environmental variables, Zenodo, 2023 | DOI
[55] The potential of convolutional neural networks for modeling species distributions., Peer Community in Ecology (2024) | DOI
[56] An approach to consider behavioral plasticity as a source of uncertainty when forecasting species' response to climate change, Ecology and Evolution, Volume 5 (2015) no. 12, pp. 2359-2373 | DOI
[57] GHRSST Level 4 MUR 0.25deg global foundation sea surface temperature analysis (v4.2), NASA Physical Oceanography DAAC, 2019 | DOI
[58] Resampling methods for a reliable validation set in deep learning based point cloud classification, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B2-2022 (2022), pp. 617-624 | DOI
[59] PyTorch: an imperative style, high-performance deep learning library, arXiv, 2019 | DOI
[60] Right Whales, Encyclopedia of Marine Mammals, Academic Press, 2009 (https://books.google.fr/books?id=2rkHQpToi9sC&redir_esc=y)
[61] Do marine and terrestrial ecologists do it differently?, Marine Ecology Progress Series, Volume 304 (2005), pp. 283-289 (https://www.jstor.org/stable/24869863)
[62] An application of satellite-derived sea surface temperature data to the Skipjack (Katsuwonus pelamis Linnaeus, 1758) and Albacore Tuna (Thunnus \emphalalunga Bonaterre, 1788) fisheries in the North-east Atlantic, International Journal of Remote Sensing, Volume 17 (1996) no. 4, pp. 749-759 | DOI
[63] PhytoBase: a global synthesis of open-ocean phytoplankton occurrences, Earth System Science Data, Volume 12 (2020) no. 2, pp. 907-933 | DOI
[64] A review of Humpback Whales' migration patterns worldwide and their consequences to gene flow, Journal of the Marine Biological Association of the United Kingdom, Volume 89 (2009) no. 5, pp. 995-1002 | DOI
[65] Pushing the limits in marine species distribution modelling: lessons from the land present challenges and opportunities: marine species distribution models, Global Ecology and Biogeography, Volume 20 (2011) no. 6, pp. 789-802 | DOI
[66] ESA Ocean Colour Climate Change Initiative: Version 5.0 Data, NERC EDS Centre for Environmental Data Analysis, 2021 | DOI
[67] Mapping global human dependence on marine ecosystems, Conservation Letters, Volume 12 (2019) no. 2, p. e12617 | DOI
[68] Drivers and impacts of the most extreme marine heatwave events, Scientific Reports, Volume 10 (2020) no. 1, 1, p. 19359 | DOI
[69] Axiomatic attribution for deep networks, arXiv, 2017 | DOI
[70] Top marine predators track Lagrangian coherent structures, Proceedings of the National Academy of Sciences, Volume 106 (2009) no. 20, pp. 8245-8250 | DOI
[71] Atlantic Bluefin Tuna (Thunnus Thynnus), Smithsonian Institute, 2009 (https://ocean.si.edu/ocean-life/fish/atlantic-bluefin-tuna-thunnus-thynnus)
[72] Facts and uncertainties about the genetic population structure of Atlantic Bluefin Tuna (Thunnus thynnus) in the Mediterranean. Implications for fishery management, Reviews in Fish Biology and Fisheries, Volume 21 (2011) no. 3, pp. 527-541 | DOI
[73] Biodiversity's big wet secret: the global distribution of marine biological records reveals chronic under-exploration of the deep pelagic ocean, PLOS ONE, Volume 5 (2010) no. 8, p. e10223 | DOI
[74] Wedge-Tailed Shearwater (Ardenna pacifica), Birds of the World, Cornell Lab of Ornithology, 2020 | DOI
[75] A flexible Bayesian framework for unbiased estimation of timescales, Nature Computational Science, Volume 2 (2022) no. 3, pp. 193-204 | DOI
Cited by Sources: