Dataset Information

Elucidating ecological complexity: Unsupervised learning determines global marine eco-provinces.

ABSTRACT: An unsupervised learning method is presented for determining global marine ecological provinces (eco-provinces) from plankton community structure and nutrient flux data. The systematic aggregated eco-province (SAGE) method identifies eco-provinces within a highly nonlinear ecosystem model. To accommodate the non-Gaussian covariance of the data, SAGE uses t-stochastic neighbor embedding (t-SNE) to reduce dimensionality. Over a hundred eco-provinces are identified with the density-based spatial clustering of applications with noise (DBSCAN) algorithm. Using a connectivity graph with ecological dissimilarity as the distance metric, robust aggregated eco-provinces (AEPs) are objectively defined by nesting the eco-provinces. Using the AEPs, the control of nutrient supply rates on community structure is explored. Eco-provinces and AEPs are unique and aid model interpretation. They could facilitate model intercomparison and potentially improve understanding and monitoring of marine ecosystems.

SUBMITTER: Sonnewald M

PROVIDER: S-EPMC7259926 | biostudies-literature | 2020 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Elucidating ecological complexity: Unsupervised learning determines global marine eco-provinces.

Sonnewald Maike M Dutkiewicz Stephanie S Hill Christopher C Forget Gael G

Science advances 20200529 22

An unsupervised learning method is presented for determining global marine ecological provinces (eco-provinces) from plankton community structure and nutrient flux data. The systematic aggregated eco-province (SAGE) method identifies eco-provinces within a highly nonlinear ecosystem model. To accommodate the non-Gaussian covariance of the data, SAGE uses t-stochastic neighbor embedding (t-SNE) to reduce dimensionality. Over a hundred eco-provinces are identified with the density-based spatial cl ...[more]

PMID: 32523981

Similar Datasets

Project description:BackgroundEcological attributes estimated from food web models have the potential to be indicators of good environmental status given their capabilities to describe redundancy, food web changes, and sensitivity to fishing. They can be used as a baseline to show how they might be modified in the future with human impacts such as climate change, acidification, eutrophication, or overfishing.MethodologyIn this study ecological network analysis indicators of 105 marine food web models were tested for variation with traits such as ecosystem type, latitude, ocean basin, depth, size, time period, and exploitation state, whilst also considering structural properties of the models such as number of linkages, number of living functional groups or total number of functional groups as covariate factors.Principal findingsEight indicators were robust to model construction: relative ascendency; relative overhead; redundancy; total systems throughput (TST); primary production/TST; consumption/TST; export/TST; and total biomass of the community. Large-scale differences were seen in the ecosystems of the Atlantic and Pacific Oceans, with the Western Atlantic being more complex with an increased ability to mitigate impacts, while the Eastern Atlantic showed lower internal complexity. In addition, the Eastern Pacific was less organised than the Eastern Atlantic although both of these systems had increased primary production as eastern boundary current systems. Differences by ecosystem type highlighted coral reefs as having the largest energy flow and total biomass per unit of surface, while lagoons, estuaries, and bays had lower transfer efficiencies and higher recycling. These differences prevailed over time, although some traits changed with fishing intensity. Keystone groups were mainly higher trophic level species with mostly top-down effects, while structural/dominant groups were mainly lower trophic level groups (benthic primary producers such as seagrass and macroalgae, and invertebrates). Keystone groups were prevalent in estuarine or small/shallow systems, and in systems with reduced fishing pressure. Changes to the abundance of key functional groups might have significant implications for the functioning of ecosystems and should be avoided through management.Conclusion/significanceOur results provide additional understanding of patterns of structural and functional indicators in different ecosystems. Ecosystem traits such as type, size, depth, and location need to be accounted for when setting reference levels as these affect absolute values of ecological indicators. Therefore, establishing absolute reference values for ecosystem indicators may not be suitable to the ecosystem-based, precautionary approach. Reference levels for ecosystem indicators should be developed for individual ecosystems or ecosystems with the same typologies (similar location, ecosystem type, etc.) and not benchmarked against all other ecosystems.

Project description:BackgroundThe backbone of the eukaryotic tree of life contains taxa only found in molecular surveys, of which we still have a limited understanding. Such is the case of Picozoa, an enigmatic lineage of heterotrophic picoeukaryotes within the supergroup Archaeplastida, which has emerged as a significant component of marine microbial planktonic communities. To enhance our understanding of the diversity, distribution, and ecology of Picozoa, we conduct a comprehensive assessment at different levels, from assemblages to taxa, employing phylogenetic analysis, species distribution modeling, and ecological niche characterization.ResultsPicozoa was among the ten most abundant eukaryotic groups, found almost exclusively in marine environments. The phylum was represented by 179 Picozoa's OTU (pOTUs) placed in five phylogenetic clades. Picozoa community structure had a clear latitudinal pattern, with polar assemblages tending to cluster separately from non-polar ones. Based on the abundance and occupancy pattern, the pOTUs were classified into four categories: Low-abundant, Widespread, Polar, and Non-polar. We calculated the ecological niche of each of these categories. Notably, pOTUs sharing similar ecological niches were not closely related species, indicating a phylogenetic overdispersion in Picozoa communities. This could be attributed to competitive exclusion and the strong influence of the seasonal amplitude of variations in environmental factors, such as temperature, shaping physiological and ecological traits.ConclusionsOverall, this work advances our understanding of uncharted protists' evolutionary dynamics and ecological strategies. Our results highlight the importance of understanding the species-level ecology of marine heteroflagellates like Picozoa. The observed phylogenetic overdispersion challenges the concept of phylogenetic niche conservatism in protist communities, suggesting that closely related species do not necessarily share similar ecological niches. Video Abstract.

Dataset Information

Elucidating ecological complexity: Unsupervised learning determines global marine eco-provinces.

Publications

Elucidating ecological complexity: Unsupervised learning determines global marine eco-provinces.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets