Thèse de doctorat
Résumé : Since the “Great Acceleration” in 1950, starting point of the Anthropocene Epoch, the influence of human activities over all biomes have been unprecedented. The sharp increase of greenhouse gas released in the atmosphere acts jointly with the concomitant rise of temperature at the global scale to destabilise every components of the Earth system (atmosphere, oceans, coastlines, lands). These components are however in constant interaction through complex processes, feedback and thresholds and, as a result, these changes lead to altered atmospheric and thermohaline circulation, more frequent and intense climatic events, droughts and floods, ice melting, sea level rising and ocean acidification, among others. Along with the destructive activities performed by human societies on all biomes for various land-use purposes, the main consequence for ecosystems is habitat loss at the global scale. In this context, a tremendous scientific effort has aimed to evaluate the influence of such changes on Earth biodiversity, eventually demonstrating the disastrous consequences of human activities and evidencing the ongoing 6th mass extinction. However, studies have focused mainly on species diversity, and neglected another key component of biodiversity: ecological interactions. These interactions among species, through predation, parasitism or symbiosis for instance, are as much threatened as species biodiversity. They are however crucial as they determine ecosystem stability, by balancing and dampening perturbations, and drive ecosystem functioning. The study of network of ecological interactions has thus emerged as a key tool to study ecosystems, providing a better understanding of their processes and their overarching structure promoting stability. Tropical coral reefs, being among the most complex, hyperdiverse ecosystems on Earth, are also among the most sensitive and threatened by global changes. Rising sea surface temperatures and increased CO2 concentrations promote mass bleaching events and the bioerosion of coral reef colonies, ultimately leading to a substantial decrease of reef structural complexity. In turn, this loss of habitat profoundly alter coral reef trophic networks and jeopardise the functioning of these ecosystems. While some studies have aimed to reconstruct coral reef trophic network from empiric observations or from theoretical models, we capitalise here on the work of Parravicini et al. (2020) to infer trophic interactions by combining the accuracy of empiric visual gut content data and the predictive power of machine learning algorithms. This PhD thesis therefore aims to evaluate the influence of habitat loss on coral reef ecosystems over different scales, by assembling coral reef trophic networks following this new approach. First, we evaluate the spatial distribution of key topology metrics and energy fluxes across a global biogeographical range and assess the influence of various human- induced stressors on coral reef trophic network structure and functioning. Then, we assess the temporal variations of coral reef trophic network topology over a 19-years time series in the island of Mo’orea, French Polynesia, including large perturbations promoting habitat loss, such as a tropical cyclone and bleaching events. Eventually, we focus on a key component of coral reefs trophic networks, benthic, mobile invertebrates, and observe the community changes in various habitat conditions among three Polynesian islands (Mo’orea, Tahiti and Tetiaroa).