Seagrass ecosystems are intense carbon sinks, but they can also emit greenhouse gases (GHG), such as carbon dioxide (CO2) and methane (CH4), to the atmosphere. Yet, GHG emissions by seagrasses are not considered when estimating global CH4 production rates by natural sources, although these estimations will help predict future scenarios and potential changes in CH4 emissions. In addition, the effect of warming on GHG emissions by seagrasses has not yet been reported. The present study aims to assess the CO2 and CH4 production rates by vegetated and adjacent bare sediment of a monospecific seagrass meadow (Halophila stipulacea) located in the central Red Sea. We measured CH4 and CO2 fluxes and their isotopic signatures by cavity ringdown spectroscopy on chambers containing vegetated and bare sediment. The fluxes were measured at temperatures from 25 °C (winter seawater temperature) to 37 °C to cover the natural thermal range and future seawater temperatures in the Red Sea. Additional parameters analyzed included changes in the sediment microbial community composition, sediment organic matter, organic carbon, nitrogen, and phosphorus concentration. We detected up to 100-fold higher CH4 (up tp 571.65 µmol CH4 m−2 d−1) and up to six-fold higher CO2 (up to 13,930.18 µmol CO2 m−2 d−1) fluxes in vegetated sediment compared to bare sediment, and an increase in CH4 and CO2 production with increasing temperature. In contrast, CH4 and CO2 production rates decreased in communities that were maintained at 25 °C, while communities that were exposed to prolonged darkness showed a decrease in CH4 and an increase in CO2 production rates. However, only minor changes were seen in the microbial community composition with increasing temperatures. These results show that GHG emissions by seagrasses might be affected by natural temperature extremes and warming due to climate change in the Red Sea. The findings will have critical implications for the estimation of natural GHG sources, especially when predicting future changes in the global CH4 budget.

Seagrass meadows, like coral reefs, are in decline globally but are often neglected in marine policy and conservation despite their equally critical ecosystem services. Both habitats can be heavily impacted by wave surges, rainfall-induced earth movement and flooding, changes to water temperature, salinity, and acidity, and increased levels of turbidity—all occurring at increased rates due to a changing global climate. We demonstrate that multispectral satellite imagery, geospatial tools, and classification techniques can be used to inform management by identifying and quantifying changes in seagrass distribution and the presence of sediment-related threats. Results from Dominica indicate near-shore seagrass habitat area increased by 195.7 hectares between 2016 and 2019, suggesting a continued expansion of Halophila stipulacea. Further analysis showed 22.4 hectares of accreted coastal sediment and 1362.2 hectares of suspended sediment captured, placing 424.4 hectares of sensitive reef area at risk of experiencing tissue abrasion or reduced photosynthetic activity. Our methods can be used by marine resource managers and policy makers to inform decisions relating to fisheries production, emissions trading, disaster risk mitigation, and invasive species monitoring, facilitating sustainable growth in the blue economy.

De la profondeursur 2 sites en Guadeloupe pendant la saison sèche et la saison humide (Chapitre 3). A u cours de cette étude,les genres Ostreopsis et Prorocentrum étaient les plus abondants. Aucune influence de la profondeur n'a é;étrouvée sur l'abondance totale des dinoflagellés benthiques ; cependant les genres Ostreopsis et GambiericLscusétaient principalement abondants en surface, alors que le genre Coolia était présent plus en profondeur.L'approche d'écologie trophique concernait le lien trophique entre dinoflagellés benthiques toxiqueset méiofaune (Chapitre 1). Les transferts au sein des réseaux trophiques de phycotoxines synthétisées par lesdinoflagellés benthiques toxiques sont généralement étudiés chez les organismes de grande taille, délaissantles organismes de μeLile Laille qui sulll pourtant en contact direct avec les microalgues. Au cours de cetteétude, une attention particulière a été portée aux copépodes harpacticoïdes qui composent ia méiofaune.Cette étude expérimentale, par marquage des microalgues aux isotopes stables, s'est focalisée sur lescopépodes harpacticoïdes et a permis de mesurer le taux d'ingestion d'Amphidinium sp. et d'Ostreopsis sp ..Cette étude démontre que les organismes de la méiofaune peuvent constituer une voie d'entrée des toxinessynthétisées par les dinoflagellés benthiques dans les réseaux trophiques. / The distribution of benthic toxiï rlinofüeefü1tes is 1111known in (Juadeloupe and Martinique even if at theCaribbean basin spatial scale, those microalgae are responsible for serious poisoning such as ciguatera.During this thesis, the ecology of benthic toxic dinoflagellates μreseul was sLudieJ using: spatio-temporalstudies (Chapters 1, 2, 3) and a trophic ecology approach (Chapter 4).First, 27 sites were explored to describe the spatial distribution of benthic dinoflagellatf's prPsf'nt inGuadeloupe and Martinique (Chapter 1). Only the most abundant macrophytes (biological substrates ofbenthic dinoflagellates) at the different sites were considered. ln this study, 161 samples were analyzed and 7times more taxie benthic dinoflagellates were found in Guadeloupe than in Martinique. The genus Ostreopsisdominated the benthic dinoflagellate community in both islands, although this trend has only rarely beenfound in the Caribbean basin.Then, 3 sites per department were chosen in order to set up a monthly monitoring of benthicdinoflagellates abundances duringl8 months (Chapter 2). The sites selected i) had a high abundance ofbenthic toxic dinoflagellates and ii) were identified as potentially dangerous by the Agence Régionale de laSanté. During this study, 927 samples of macrophytes were analyzed and 5 times more benthic toxicdinoflagellates were found in Guadeloupe than in Martinique. Ostreopsis and Prorocentrum genera dominatedrespectively in Guadeloupe and in Martinique. Two times more Gambierdiscus were found in Martinique thanin Guadeloupe even if Guadeloupe is located in the high prevalence area and this genus being responsible forciguatera. This result suggested that species with different toxicities structured the benthic toxicdinoflagellates community in this area. The abundances of benthic dinoflagellates were little influenced byabiotic parameters (temperature and salinity) but structured by biotic parameters (host macrophytes).Halophila stipulacea an invasive angiosperm in the Lesser Antilles has been identified as promoting thedevelopment of the genus Gambierdiscus.. The distribution of benthic toxic dinoflagellates was also studied according to the depth at 2 sites inGuadeloupe during the dry and the wet seasons (Chapter 3). ln order to avoid bias due to the presence ofdifferent macrophytes, this experiment was conducted only on H. stipulacea constituting mono-specificmeadows along a strong depth gradient. ln this study, the Ostreopsis and Prorocentrum genera were the mostabundant. No influence of the depth was found on total abundance of benthic dinoflagellates; however,Ostreopsis and Gambierdiscus genera were mainly abundant near the surface while the genus Coolia was presentdeeper.The trophic ecology approach focused on the trophic link between taxie benthic dinoflagellates andmeiofauna (Chapter 4). Transfers within the food webs of phycotoxins synthesized by benthic dinoflagellatesare mainly studied in large organisms neglecting smaller ones in direct contact with microalgae. Thisexperimental study, by labeling the microalgae with stable isotopes, focused on harpacticoides copepod andmeasured for the first time their ingestion rates of Amphidinium sp. and Ostreopsis sp.. This studydemonstrated that meiofauna organisms can constitute an input channel of phycotoxins from benthicdinoflagellates in food webs.

The hierarchical levels at which resource selection occurs can have important consequences for individual and population energy budgets and structure the impacts of a forager on its ecosystem. Assessing factors affecting resource selection of large marine herbivores across scales is important because of their potentially large impacts on seagrass community dynamics and historical and current changes in their population sizes and those of their potential predators. I explored the factors (predation risk, resource abundance, quality and identity) affecting resource use of large marine herbivores (green turtles, Chelonia mydas) from the scale of habitat patches to forage species within patches. I used a combination of in-water surveys, aerial drone video transects, baited camera surveys, and seagrass community and nutrient content analyses to provide insights into resource use by turtles in multiple ecological contexts. In Abaco, The Bahamas I found relatively intact shark populations, including apex predators, relative to other parts of the Caribbean. In the context of healthy predator populations in Abaco, I tested a priori predictions rooted in Ideal Free Distribution (IFD) theory. Green turtles off Abaco deviated from predictions of an IFD determined by the standing stocks of seagrass. Instead, distributions are consistent with predictions of the foraging arena hypothesis with turtles largely restricted to safe habitat patches and selecting locations within these where seagrass N content is relatively high. Marine invasive species can have detrimental effects on coastal ecosystems and economies. Therefore, understanding the effects of, and factors influencing the rate of spread of the invasive seagrass Halophila stipulacea in the Caribbean is important. In the French West Indies (Guadeloupe, Martinique and St. Martin), I investigated foraging preferences for native versus invasive seagrass species and whether green turtles might facilitate or attenuate the invasion through their choice of habitats and feeding patterns. Green turtle distributions were correlated with native seagrass distributions. Also, despite similar nutrient contents, turtles preferred feeding on native seagrasses irrespective of their relative abundance within a patch. These results suggest that, as predicted by the Enemy Release Hypothesis, green turtles likely facilitate the invasion and spread of the invasive seagrass that may reduce energy flow into turtle populations.