Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system

The poor understanding of changes in mollusc ecology along rivers, especially in West Africa, hampers the implementation of management measures. We used a self–organizing map, indicator species analysis, linear discriminant analysis and a random forest model to distinguish mollusc assemblages, to de...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Zinsou Cosme Koudenoukpo, Olaniran Hamed Odountan, Prudenciène Ablawa Agboho, Tatenda Dalu, Bert Van Bocxlaer, Luc Janssens de Bistoven, Antoine Chikou, Thierry Backeljau
Formato:	article
Lenguaje:	EN
Publicado:	Elsevier 2021
Materias:	Artificial neural network Ecology Freshwater biodiversity Modelling Mollusc community Tropical river systems QH540-549.5
Acceso en línea:	https://doaj.org/article/9c8f5d7d50b7465e8f3cf376f3c50c83
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:9c8f5d7d50b7465e8f3cf376f3c50c83
record_format	dspace
spelling	oai:doaj.org-article:9c8f5d7d50b7465e8f3cf376f3c50c832021-12-01T04:51:47ZUsing self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system1470-160X10.1016/j.ecolind.2021.107706https://doaj.org/article/9c8f5d7d50b7465e8f3cf376f3c50c832021-07-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S1470160X2100371Xhttps://doaj.org/toc/1470-160XThe poor understanding of changes in mollusc ecology along rivers, especially in West Africa, hampers the implementation of management measures. We used a self–organizing map, indicator species analysis, linear discriminant analysis and a random forest model to distinguish mollusc assemblages, to determine the ecological preferences of individual mollusc species and to associate major physicochemical variables with mollusc assemblages and occurrences in the Sô River Basin, Benin. We identified four mollusc assemblages along an upstream–downstream gradient. Dissolved oxygen (DO), biochemical oxygen demand (BOD), salinity, calcium (Ca), total nitrogen (TN), copper (Cu), lead (Pb), nickel (Ni), cadmium (Cd) and mercury (Hg) were the major physicochemical variables responsible for structuring these mollusc assemblages. However, the physicochemical factors responsible for shaping the distribution of individual species varied per species. Upstream sites (assemblage I) showed high DO and low BOD and mineral compounds (i.e., TN, salinity, and Ca), which are primarily responsible for structuring the occurrences of bivalves (Afropisidium pirothi, Etheria elliptica, Sphaerium hartmanni) and the gastropod Lanistes varicus. Sites along the middle reach (assemblage II) were characterised by a high degree of organic pollution but low heavy metal pollution; we detected no specific mollusc indicator species. Downstream sites (assemblage III) displayed high mineral and heavy metal concentrations and a fauna without specific indicator species. Finally, downstream sites associated with brackish water (assemblage IV) displayed important levels of organic and heavy metal pollution. These sites are dominated by diverse gastropods (i.e., Bulinus spp., Gabbiella africana, Indoplanorbis exustus, Pachymelania fusca, Radix natalensis, Stenophysa marmorata and Tympanotonos fuscatus). Our results highlight that mollusc communities in the Sô River Basin are structured by key physicochemical variables related to the river–estuary continuum. Habitats that are progressively more downstream are confronted with increasing anthropogenic stress. Conservation and management plans should focus on downstream habitats.Zinsou Cosme KoudenoukpoOlaniran Hamed OdountanPrudenciène Ablawa AgbohoTatenda DaluBert Van BocxlaerLuc Janssens de BistovenAntoine ChikouThierry BackeljauElsevierarticleArtificial neural networkEcologyFreshwater biodiversityModellingMollusc communityTropical river systemsEcologyQH540-549.5ENEcological Indicators, Vol 126, Iss , Pp 107706- (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Artificial neural network Ecology Freshwater biodiversity Modelling Mollusc community Tropical river systems Ecology QH540-549.5
spellingShingle	Artificial neural network Ecology Freshwater biodiversity Modelling Mollusc community Tropical river systems Ecology QH540-549.5 Zinsou Cosme Koudenoukpo Olaniran Hamed Odountan Prudenciène Ablawa Agboho Tatenda Dalu Bert Van Bocxlaer Luc Janssens de Bistoven Antoine Chikou Thierry Backeljau Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system
description	The poor understanding of changes in mollusc ecology along rivers, especially in West Africa, hampers the implementation of management measures. We used a self–organizing map, indicator species analysis, linear discriminant analysis and a random forest model to distinguish mollusc assemblages, to determine the ecological preferences of individual mollusc species and to associate major physicochemical variables with mollusc assemblages and occurrences in the Sô River Basin, Benin. We identified four mollusc assemblages along an upstream–downstream gradient. Dissolved oxygen (DO), biochemical oxygen demand (BOD), salinity, calcium (Ca), total nitrogen (TN), copper (Cu), lead (Pb), nickel (Ni), cadmium (Cd) and mercury (Hg) were the major physicochemical variables responsible for structuring these mollusc assemblages. However, the physicochemical factors responsible for shaping the distribution of individual species varied per species. Upstream sites (assemblage I) showed high DO and low BOD and mineral compounds (i.e., TN, salinity, and Ca), which are primarily responsible for structuring the occurrences of bivalves (Afropisidium pirothi, Etheria elliptica, Sphaerium hartmanni) and the gastropod Lanistes varicus. Sites along the middle reach (assemblage II) were characterised by a high degree of organic pollution but low heavy metal pollution; we detected no specific mollusc indicator species. Downstream sites (assemblage III) displayed high mineral and heavy metal concentrations and a fauna without specific indicator species. Finally, downstream sites associated with brackish water (assemblage IV) displayed important levels of organic and heavy metal pollution. These sites are dominated by diverse gastropods (i.e., Bulinus spp., Gabbiella africana, Indoplanorbis exustus, Pachymelania fusca, Radix natalensis, Stenophysa marmorata and Tympanotonos fuscatus). Our results highlight that mollusc communities in the Sô River Basin are structured by key physicochemical variables related to the river–estuary continuum. Habitats that are progressively more downstream are confronted with increasing anthropogenic stress. Conservation and management plans should focus on downstream habitats.
format	article
author	Zinsou Cosme Koudenoukpo Olaniran Hamed Odountan Prudenciène Ablawa Agboho Tatenda Dalu Bert Van Bocxlaer Luc Janssens de Bistoven Antoine Chikou Thierry Backeljau
author_facet	Zinsou Cosme Koudenoukpo Olaniran Hamed Odountan Prudenciène Ablawa Agboho Tatenda Dalu Bert Van Bocxlaer Luc Janssens de Bistoven Antoine Chikou Thierry Backeljau
author_sort	Zinsou Cosme Koudenoukpo
title	Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system
title_short	Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system
title_full	Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system
title_fullStr	Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system
title_full_unstemmed	Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system
title_sort	using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a west africa river–estuary system
publisher	Elsevier
publishDate	2021
url	https://doaj.org/article/9c8f5d7d50b7465e8f3cf376f3c50c83
work_keys_str_mv	AT zinsoucosmekoudenoukpo usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT olaniranhamedodountan usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT prudencieneablawaagboho usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT tatendadalu usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT bertvanbocxlaer usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT lucjanssensdebistoven usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT antoinechikou usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem AT thierrybackeljau usingselforganizingmapsandmachinelearningmodelstoassessmollusccommunitystructureinrelationtophysicochemicalvariablesinawestafricariverestuarysystem
_version_	1718405739225022464

Using self–organizing maps and machine learning models to assess mollusc community structure in relation to physicochemical variables in a West Africa river–estuary system

Ejemplares similares