HAL Portal UPF (Université de la Polynésie française)
Not a member yet
3802 research outputs found
Sort by
Increase of coastal shark bite frequency linked to the COVID-19 lockdown reveals a territoriality-dominance behaviour toward humans
No full textInternational audienceThe effects of the COVID-19 lockdown on wildlife aggression toward humans have received little attention. Records from French Polynesia show that the annual average of about five shark bites on humans from 2009 to 2019 increased significantly to 15 in 2020, despite the virtual absence of humans from the marine environment during a six-week curfew in April–May 2020. Bites then returned to baseline levels in 2021 ( N = 4), 2022 ( N = 5), and 2023 ( N = 2). Most bites in 2020 occurred just after the lockdown and were attributed to gray reef sharks, Carcharhinus amblyrhynchos , displaying territoriality rather than self-defense or predatory behaviours. This temporary increase in shark bites suggests that natural territorial behavior, typically suppressed by continued human presence, reemerged during the Anthropause, providing new insights into shark risks for ocean users and the management of human–predator interactions
Publications mathématiques de Besançon: Actes du colloque international « Géométrie algébrique, Théorie des nombres et Applications » (GTA 2021)
No full textInternational audienc
Acoustic mobbing behaviour: vocal fish responses to predation risk through sound communication
No full textInternational audienceIn both terrestrial and marine ecosystems, prey species exhibit various reactions to predators, including avoidance, fighting and freezing. Acoustic communication associated with mobbing has been depicted in numerous tetrapod species. This sonic behaviour remains however largely unknown in fish although mobbing has been shown in different coral reef fish as a response to predation risk. This study focused on the communication strategy of prey species in the presence of predators. We tested the reaction of the holocentrid Sargocentron caudimaculatum when facing a moray eel and compared the sounds they produced against the predator with sounds produced when the fish are hand-held, recorded in standardized conditions. When introduced in mesocosms to a moray eel, S. caudimaculatum produced distinct distress sounds, resembling staccato calls, and swam towards the predator with their heads pointed in its direction and their dorsal fin erected. This observation supports a mobbing behaviour with specimens shifting from an escape behaviour to an aggressive response in presence of predator. Moreover, these sounds are different from those emitted during manual handling. This difference in acoustic signal suggests the coexistence of multiple warning/alarm sounds in this species. Holocentrids can produce different kinds of sounds witnessing their ability to provide graded information based on the perceived predation risk. We suggest that species of this family would be a suitable and promising group with which to test the hypothesis of acoustic coordination where individuals could inform their conspecifics about an increase in the immediate predation risk through appropriate alarm calls and trigger an escape response of the group if needed
The South Pacific: a unique geological and oceanographic region of freshwater island-oasis habitats for anguillid eel population interactions
No full textInternational audienceZoogeography and ecology of 3 widespread anguillid species (Anguilla marmorata, A. megastoma, A. obscura) in the western South Pacific are reviewed in relation to the freshwater habitats and communities of the many islands of Oceania (island geological ages, from > 40 Ma to < 1 Ma. Anguillid eels show habitat segregation among A. obscura (lower reaches-brackish water), A. marmorata (habitat generalist) and A. megastoma (upper reaches), where they feed on amphidromous snails, Macrobrachium shrimp and gobies, and freshwater insect larvae. Some information has been obtained about possible anguillid spawning areas in the region near New Caledonia and Fiji from larval surveys and satellite transmitting tagging studies. Larval drift simulations suggest separate spawning areas occur in the eastern region of French Polynesia. Little is known about how geography might influence the spawning ecologies and population structures of South Pacific anguillid eels that live on the many tropical islands that are spread across a wide region of ocean in the western Indo-Pacific
Translation and grammatical issues in the PISA items. Potential biases in the interpretation of students' answers
No full textInternational audienceThe PISA items are first written in French or English, and then translated into the language of instruction in the different countries where they are submitted. In this respect, translation issues are crucial, whether lexical or grammatical. A translation guide is provided by the OECD. However, in the section of the guide devoted to mathematics, there is no mention of grammatical issues, despite their critical nature for the teaching of mathematics. We hypothesize that this lack of interest is likely to lead to biases in the interpretation of students' answers that might skew the statistical results provided by the OECD, in particularly for those students whose mother tongue is not the language of schooling. The aim of the workshop is twofold. Firstly, we wish to share our methodology and our initial results involving the analysis of a few items and a small number of languages spoken in Europe. This has enabled us to identify some tensions resulting from the legitimate desire to remain close to the s
Environmental DNA for the surveillance of biosecurity threats in Mediterranean lagoons
No full textInternational audienceInvasive species that outcompete endemic ones and toxic harmful algae that cause algal blooms threaten marine resources like fisheries, aquaculture, and even tourism. Environmental DNA (eDNA) metabarcoding can help as a method for early alert. In this study, we have analyzed communities inhabiting six lagoons within the Gulf of Lion (northwest Mediterranean Sea) with spatial protection as RAMSAR and Natura 2000 sites. Employing the COI gene as the only metabarcode, we found 15 genera that have caused recognized algal bloom outbreaks in the studied lagoons since 2000. In addition, seven alien invasive species that can pose risks to the rich marine resources of the zone and lagoons were also found. The results found from eDNA are consistent with events of toxic algae blooms before and after the sampling moment and with reported occurrences of the invasive species in nearby Mediterranean areas. Multivariate multiple analysis showed the importance of anthropic pressure in the abundance of these nuisance species. Mitigation actions and routine eDNA metabarcoding in zones of special interest like these fragile French Mediterranean lagoons are recommended for early alert of nuisance species in order to plan timely management actions
Lattices in rigid analytic representations
No full textFor a profinite group G and a rigid analytic space X, we study when an OX (X)linear representation V of G admits a lattice, i.e. an OX (X )-linear model for a suitable formal model X of X in the sense of Berthelot. We give a positive answer, under mild assumptions, when X is a "wide open" space. As a consequence, we are able to describe explicit open rational subdomains of X over which V is constant after reduction modulo a power of p. We give applications in two different directions. First, we prove explicit results on the reduction modulo powers of p of sheaves of crystalline and semistable representations of fixed weight. Second, we focus on the sheaves of Galois representations on eigenvarieties, which are important examples of wide open spaces thanks to a result of Bellaïche and Chenevier. We give an application of our main results to the pseudorepresentation carried by the Coleman-Mazur eigencurve, which can be made explicit whenever equations for a rational subdomain of the eigencurve are given
Management des valeurs culturelles : le cas du secteur du tourisme en Polynésie française
No full textInternational audienc
Comparison of in situ black-lipped oyster spat collection and larval dispersal modelling results in semi-closed pearl-farming lagoons of the Tuamotu Archipelago
No full textInternational audienceSpat collection of the pearl oyster Pinctada margaritifera in atoll lagoons of French Polynesia is the fundamental sustain of black pearl farming. Spat collection has always yielded variable results in space and time, but obvious signs of steady decreases, even collapses, have emerged in several lagoons. Spat collection materializes the ecological connectivity pathways between wild spawning populations and the location of artificial larval settlement substrates. To assess if oyster larval dispersal modelling could capture such pathways, we compared four six-week long spat collector deployment periods with dispersal simulations in two different lagoons. Spat collectors displayed wide spatial and temporal variations. Numerical modelling and field experiments were generally not in agreement. Although both methods have limitations, they can still approximate each other. But the accuracy of model simulations cannot be ascertained with spat collection data only. Using a SWOT (StrengthWeakness-Opportunities-Threats) analysis, we emphasize the complementarity of both approaches for management decisions
Régionalisation d’un modèle pluie-débit pour estimer des crues sur l’île de Tahiti, Polynésie Française
No full textThe tropical volcanic Island of Tahiti is subject to flash floods that expose people and infrastructures to regular risks. The typical violence of these floods results from intense convective rainfall events impacting the steep slopes of the catchments. Local engineers expressed their need for adapted flood estimation methods. The choice of an adapted method relied on a review of current flood estimation methods and regionalization strategies. Due to the available data and knowledge about Tahitian hydrology, conceptual and parsimonious event-based hydrological models deriving from the Soil Conservation Service (SCS) model were selected. The SCS-SMA (Soil Conservation Service - Soil Moisture Accounting) function was adopted and applied in a distributed mode to a grid of regular square meshes to represent runoff production. This function was combined with the Lag-andRoute (LR) transfer function to build a complete flood hydrograph model. The LR function is characterised by a combination of translation and storage, which are assumed to be linearly dependent. To evaluate this model and to develop a calibration methodology, SCS-SMA/LR was applied to a series of observed rainfall-flow events in the Hitiaa catchment. The model was efficient according to the Nash and Sutcliffe Efficiency (NSE) criterion, but the initial condition H0 appeared poorly related to the observed initial baseflow Q0, which was supposed to estimate the initial moisture state of the catchment. Using a stochastic rainfall generator to describe the range of possible rainfall fields for a given event, we showed that the weakness of the correlation was due to the spatial uncertainties associated with rainfall input. They propagate through the model as far as they strongly impact the estimation of H0 and thus the potential link between H0 and Q0. Furthermore, the LR transfer function was found variable from one event to another, suggesting that its parametrization depends on the intensity of the flood, but such dependence was not incorporated into the function structure. However, the SCS-SMA-LR model remains relevant for estimating the average behavior of rivers, which justified the relevance of its regionalization. The model was then calibrated on a selection of 9 donor catchments that are representative of various physio-climatic environments. Regionalization of the model relied on classifications and relationships between three parameters chosen as regionalization vectors and three descriptors: the median antecedent rainfall index averaged over the surface of the catchment area, the geological nature of the formations and the type of land use. The regional model was considered to be efficient on small catchments for the estimation of project floods. It was then tested on two larger and more heterogeneous catchments. The model performance remained satisfactory, as did the regional interpretations of the model parameters. However, this change of scale revealed a new bias in the parameterization of the transfer function, linked to the assumption of linearity between the propagation time and the diffusion time. This study allowed to identify the factors responsible for hydrological variability at a regional scale, especially the impact of the initial moisture conditions or the one of urbanization. Thanks to its distributed structure, the model is able to integrate any local change in land use or in storage. The model built is adapted and potentially efficient, but its performance remains limited at the event scale mainly due to the lack of knowledge about the spatial variability of rainfall. A generic method was developed to quantify the impact of these uncertainties. The improvement of the model relies on the space refinement of rainfall data acquisition, and also on methodological developments concerning for example the structure of the transfer function.L’île volcanique tropicale de Tahiti est soumise à des crues éclair qui exposent hommes et infrastructures à un risque régulier. La morphologie escarpée des bassins versants et l’amplitude des événements pluvieux expliquent la violence de ces crues et le besoin de méthodes d’estimation adaptées. La revue des méthodes actuelles d’estimation des crues et l’état des connaissances physiques climatiques et hydrologiques sur l’île justifient ensemble le choix d’un modèle hydrologique événementiel conceptuel et parcimonieux comme méthode locale d’estimation des crues. La fonction SCS-SMA (Soil Conservation Service – Soil Moisture Accounting) a été adoptée et appliquée en mode distribué à une grille de mailles carrées régulières pour représenter la production des écoulements. Cette fonction a été associée à la fonction de transfert Lag-and-Route (LR), pour construire un modèle complet de calcul des hydrogrammes de crue. La fonction LR est caractérisée par la combinaison d’une translation et d’un stockage, supposés être linéairement dépendants. Pour évaluer ce modèle et élaborer une méthodologie de calibration, il est d’abord appliqué à une série d’événements pluie-débit observés sur un bassin test, le bassin de Hitiaa. La performance du modèle apparaît limitée en partie par la faible corrélation entre la condition initiale du modèle H0 et le débit initial observé Q0 qui est censé donner accès à l’état initial d’humidité sur le bassin. A l’aide d’un générateur stochastique de pluies qui décrit la gamme des pluies possibles pour un événement donné, on montre que cette faible corrélation trouve son origine dans les incertitudes spatiales sur les pluies qui se propagent dans le modèle jusqu’à perturber l’estimation de H0 et donc son lien supposé avec Q0. Par ailleurs, la fonction de transfert LR montre une variabilité d’un évènement à l’autre, qui suggère que la paramétrisation de la fonction dépend de l’intensité de la crue, ce qui n’a pas été pris en compte ici. Le modèle SCS-SMA-LR reste pertinent pour estimer le comportement moyen des cours d’eau, ce qui justifie la poursuite des travaux de régionalisation. La régionalisation du modèle s’appuie sur une première sélection de 9 bassins versants. Trois paramètres sont choisis comme vecteurs de régionalisation, puis reliés à différents descripteurs : l’indice de pluies antérieures, la nature géologique des formations et l’occupation des sols. Le modèle régional est jugé performant sur les petits bassins versants pour l’estimation des crues de projet. Il a ensuite été testé sur deux bassins de plus grande superficie, et plus hétérogènes. Les performances du modèle sont restées satisfaisantes, ainsi que les interprétations régionales des paramètres du modèle. Ce changement d’échelle a cependant fait apparaître un nouveau biais dans la paramétrisation de la fonction de transfert, lié à l’hypothèse de linéarité entre le temps de propagation et le temps de diffusion. Cette étude a finalement permis de progresser sur la connaissance des facteurs qui expliquent la réponse des bassins à l’échelle régionale, notamment l’impact des conditions initiales d’humidité ou l’urbanisation. De par sa structure distribuée, le modèle est capable d’intégrer tout changement local d’occupation des sols, ainsi que des ouvrages de stockage. Le modèle construit est potentiellement performant, mais ses performances restent limitées à l’échelle évènementielle principalement par manque de connaissances précises de la variabilité spatiale des précipitations. Une méthode générique a été développée pour quantifier l’impact de ces incertitudes. L’amélioration du modèle repose sur l’acquisition de données de précipitations plus précises dans l’espace ainsi que sur certains développements méthodologiques sur la structure du modèle, notamment au niveau de la fonction de transfer