54 years of microboring community history explored by machine learning in a massive coral from Mayotte (Indian Ocean) - Archive ouverte HAL
Article Dans Une Revue Frontiers in Marine Science Année : 2022

54 years of microboring community history explored by machine learning in a massive coral from Mayotte (Indian Ocean)

Résumé

Coral reefs are increasingly in jeopardy due to global changes affecting both reef accretion and bioerosion processes. Bioerosion processes dynamics in dead reef carbonates under various environmental conditions are relatively well understood but only over a short-term limiting projections of coral reef evolution by 2100. It is thus essential to monitor and understand bioerosion processes over the long term. Here we studied the assemblage of traces of microborers in a coral core of a massive Diploastrea sp. from Mayotte, allowing us to explore the variability of its specific composition, distribution, and abundance between 1964 and 2018. Observations of microborer traces were realized under a scanning electron microscope (SEM). The area of coral skeleton sections colonized by microborers (a proxy of their abundance) was estimated based on an innovative machine learning approach. This new method with 93% accuracy allowed analyzing rapidly more than a thousand SEM images. Our results showed an important shift in the trace assemblage composition that occurred in 1985, and a loss of 90% of microborer traces over the last five decades. Our data also showed a strong positive correlation between microborer trace abundance and the coral bulk density, this latter being particularly affected by the interannual variation of temperature and cumulative insolation. Although various combined environmental factors certainly had direct and/or indirect effects on microboring species before and after the breakpoint in 1985, we suggest that rising sea surface temperature, rainfall, and the loss of light over time were the main factors driving the observed trace assemblage change and decline in microborer abundance. In addition, the interannual variability of sea surface temperature and instantaneous maximum wind speed appeared to influence greatly the occurrence of green bands. We thus stress the importance to study more coral cores to confirm the decadal trends observed in the Diploastrea sp. from Mayotte and to better identify the main factors influencing microboring Frontiers in Marine Science frontiersin.org 01
Fichier principal
Vignette du fichier
fmars-09-899398.pdf (9.65 Mo) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-03938202 , version 1 (13-01-2023)

Identifiants

Citer

Diego Alaguarda, Julien Brajard, G. Coulibaly, M. Canesi, Éric Douville, et al.. 54 years of microboring community history explored by machine learning in a massive coral from Mayotte (Indian Ocean). Frontiers in Marine Science, 2022, 9, ⟨10.3389/fmars.2022.899398⟩. ⟨hal-03938202⟩
187 Consultations
49 Téléchargements

Altmetric

Partager

More