HAL-IRD
Not a member yet
83947 research outputs found
Sort by
Parental phenotypes and breeding performance: a review of non-experimental investigation in well-studied Western Palearctic tits and flycatchers
No full textInternational audienceNumerous hypotheses predict associations between phenotype-associated breeder traits and reproductive success. In secondary cavity-nesting passerine birds, which have been investigated most, up to more than 90 parental characteristics have been linked to at least one field measure of breeding performance. However, within study populations, different parental traits, such as clutch size, first-egg date, egg size, nest architecture, plumage colour, adult morphometry, or song performance, are often weakly correlated across female and/or male breeders. Furthermore, many of these studies have been conducted outside the theoretical framework of the 'Individual Optimisation Hypothesis' of clutch size (IOH). IOH predicts that: (i) females with larger clutches will have larger broods at hatching and fledging compared to those with smaller clutches; and (ii) clutch size, which is adaptively adjusted to the parents' ability to rear nestlings, should always maximize the percentage of eggs producing fledglings so that variation in clutch size becomes disconnected from variation in breeding success. In this paper, we present the first detailed review of implications of IOH for parental characters other than clutch size. Our review covered 188 non-experimental studies and 1074 statistical results that examined how parental traits influence breeding success in Western Palearctic Great Tits (Parus major), Blue Tits (Cyanistes spp.), and flycatchers (Ficedula spp.). Clutch size explained one third of the variation in brood size at hatching and fledging within study populations. However, most parental characteristics associated weakly with the number of hatchlings or fledglings, likely because they did not correlate with clutch size. Overall, parental traits were poorly correlated with the proportion of hatchlings and fledglings per egg. We discuss why intraspecific variation in phenotype-associated breeder traits is often disconnected from interindividual differences in breeding success, and highlight the importance of underexplored research problems in avian breeding biology
Estimer et expliquer l'incertitude prédictive en apprentissage automatique
No full textAlthough Machine Learning (ML) models have proved to be useful in key applications, they still have significant limitations, such as sensitivity to distribution shifts. Therefore, it is essential to provide trustworthy monitoring methods when ML models make predictions on new unlabeled data. For an ML model, predictive uncertainty is defined as the uncertainty of predictions, which can be caused by the data or the model. As they identify insufficient knowledge, uncertainty estimates can thus be employed to monitor ML models' predictions and make decision-making safer and more reliable in applications such as financial risk assessment or medical diagnosis. In this thesis, we investigate the problem of estimating and explaining predictive uncertainty in ML. We introduce various methods to improve the quantification and explanation of predictive uncertainty in classification or regression tasks and through different settings such as multimodality or distribution shifts. In particular, we introduce a method to produce adaptive prediction intervals in regression tasks. To address classification tasks under distribution shifts, we first propose a method to quantify and visualize predictive uncertainty with simple representations. We then extend the research to a multimodal Transformer which quantifies uncertainty by design. Also, this architecture makes it possible to explain predictive uncertainty in terms of feature values. As uncertainty quantification is closely related to unsupervised performance estimation, we then introduce a method for practitioners to assess the model's performance in out-of-distribution contexts. Lastly and as a first step toward trustworthy ML, we introduce a risk-based approach and a reporting tool which aim to produce safe machine learning lifecycles.Bien que les modèles de machine learning se soient avérés utiles dans des applications essentielles, ils ont néanmoins des limites importantes, telles que la sensibilité aux dérives de distributions. Ainsi, il est essentiel de proposer des méthodes fiables de contrôle lorsque les modèles de machine learning font des prédictions avec de nouvelles données non labellisées. Pour un modèle, l’incertitude prédictive est définie comme l’incertitude des prédictions, causée par le modèle ou les données. Comme elles identifient un manque de connaissance, les estimations d’incertitude peuvent donc être utilisées pour surveiller les prédictions d’un modèle et rendre le processus de décision plus sûr et plus fiable pour des applications telles que l’évaluation des risques financiers ou le diagnostic médical. Dans cette thèse, nous examinons le problème d’estimation et d’explication de l’incertitude prédictive en machine learning. Nous présentons différentes méthodes qui visent à améliorer la quantification et l’explication de l’incertitude prédictive pour des tâches de classification et régression et dans des configurations variées telles que la multimodalité et les dérives de distributions. En particulier, nous présentons une méthode qui permet de générer des intervalles de prédiction adaptables en régression. Pour les tâches de classification avec dérive de distributions, nous proposons tout d’abord une méthode permettant de quantifier et visualiser l’incertitude prédictive avec de simples représentations. Puis nous étendons l’étude à un Transformer multimodal qui quantifie l’incertitude de manière intrinsèque. Cette architecture permet également d’expliquer l’incertitude prédictive en termes de valeurs des variables en entrée. Puisque la quantification d’incertitude est étroitement liée à l’estimation non supervisée de la performance d’un modèle, nous présentons une méthode permettant d’évaluer la performance d’un modèle dans des contextes de dérives de distributions. Enfin, nous proposons une approche visant une IA responsable. Dans ce cadre, nous présentons une approche par les risques ainsi qu’un outil de reporting qui visent à produire des cycles de vie de machine learning plus sûrs
Systèmes alimentaires et changements climatiques : atténuation et adaptation dans les chaînes agri-alimentaires et dans la consommation
No full textSource Agritrop Cirad (https://agritrop.cirad.fr/614794/)International audienc
Enhancing the reconstruction of the Gabasa Neandertal's diet using Ca and Sr stable isotopes
No full textInternational audienceNeandertals are known to have occupied Eurasia for over 250,000 years and were well adapted to the environmental conditions of the last ice ages. However, the dietary habits of these ancient humans remain debated, with conflicting evidence suggesting that they may have been primarily carnivorous, omnivorous, or even herbivorous. Traditional isotope analyses have provided some insights into Neandertal diets, but their limitations (preservation and baseline effect) have sparked the need for new approaches. These limitations are well known on the Iberian Peninsula, and while recent results of carbon, oxygen, radiogenic strontium and zinc isotopes and trace element ratios allow the reconstruction of the Gabasa (Spain) Neandertal diet, some questions still remain unanswered. Our study explores the potential of using calcium and stable strontium isotopes (d44Ca and d88Sr, respectively) to supplement previous analyses performed on the Gabasa Neandertal. Based on the low d44Ca and d88Sr values observed for the Neandertal specimen, our results suggest a hyper-carnivorous diet that included low but non-negligible quantities of bone, although it is not possible to rule out the possible impact of milk consumption. Overall, our work argues that Ca, Sr and Zn stable isotopes can supplement conventional isotope studies and offer a more comprehensive picture of human diets, including Neandertal
FAIRTraits: an enriched, FAIR‐compliant database of plant traits from Mediterranean populations of 240 species
No full textInternational audienceMany existing datasets also lack the environmental details necessary to investigate trait–environment relationships at local scales. Here, we present FAIRTraits, a comprehensive dataset that addresses these limitations by compiling 189,452 records of quantitative trait measurements collected between 1997 and 2023 from 1955 populations of 240 vascular plant species in the Northern Mediterranean Basin, a region known both for its exceptional biodiversity and as a climate change hotspot. All data were collected by a single research group using consistent and well-documented field and laboratory protocols, ensuring internal consistency across traits, species, sites, and years. FAIRTraits includes 180 traits measured at the individual or replicate level, with no aggregation. It features an unprecedented diversity of traits spanning all major plant organs—leaves, stems, roots, and reproductive parts. These include widely used traits such as specific leaf area and plant height, but also traits that are rarely reported, especially below-ground traits related to root morphology, as well as mechanical properties, phenology, and microbial associations. In addition to raw measurements, species are annotated with categorical descriptors (e.g., life form, photosynthetic pathway, and successional status), and species-level values taken from a Mediterranean flora, for key traits such as reproductive phenology and maximum height. To support analyses that account for environmental variability, each observation is linked to detailed descriptors of the plot where the individual was sampled, including climate data, soil physicochemical properties, and disturbance regime. Full metadata on sampling protocols and measurement methods are provided for every trait and environmental variable. FAIRTraits was built in compliance with the FAIR principles of data management (Findable, Accessible, Interoperable, and Reusable). Metadata are described using the Ecological Metadata Language (EML); trait definitions are standardized using community-endorsed semantic resources. The data are archived across two interoperable repositories: GBIF (via Darwin Core and trait-specific extensions) for taxon–trait associations and InDoRES for environmental and contextual data. These efforts ensure long-term preservation, data traceability, and seamless integration with plant trait databases such as BROT or TRY, and cross-organism initiatives such as the Open Traits Network or the Encyclopedia of Life. FAIRTraits offers a robust, richly documented, and reusable resource for investigating plant functional strategies, trait–environment relationships, and scaling from individuals to communities and ecosystems. It also provides a concrete example of how trait datasets can meet the highest standards of data quality and interoperability—serving as a model for future community-led initiatives in functional ecology. The FAIRTraits database is released under the CC-BY Attribution 4.0 International license
Does index insurance work? Insights from eight experiments in agriculture
No full textVoxDev Bloghttps://voxdev.org/topic/agriculture/does-index-insurance-work-insights-eight-experiments-agricultureIndex insurance can help smallholder farmers take on more productive risks, but its impacts remain modest, uncertain, and highly context dependent
The potential of attractive insecticide-treated nets (ITNs) in reducing malaria transmission: a modeling study
No full textInternational audienceRecent studies suggest that insecticide-treated nets (ITNs) may actively attract malaria vectors, increasing their likelihood of coming into contact with the insecticide while potentially reducing personal protection. The impact of such attractive ITN on malaria transmission at the community level remains unclear. In this study, we developed a model to compare the effectiveness of attractive, inert and deterrent ITNs under varying levels of human usage and different degrees of physiological or behavioral resistance to insecticides in malaria vectors. We developed a model to simulate the host-seeking, feeding and mortality (HSFM) of mosquito vectors facing ITNs. This model allows mosquitoes to choose between two rooms based on the presence and remote influence (attractive, inert or deterrent) of ITNs. The HSFM model was then integrated into a malaria transmission model to compare the Plasmodium transmission potential of mosquitoes exposed to these different type of ITNs under various scenarios of ITN coverage, physiological resistance, quantitative behavioral resistance, and qualitative behavioral resistance. The model predicts that attractive ITNs consistently reduce malaria transmission potential of vectors more effectively than inert or deterrent ITNs, even in the presence of resistant vector phenotypes. For instance, at intermediate use rate (50%), strongly attractive ITNs are expected to reduce transmission by up to 67% compared to deterrent ITNs. In resistant vector populations, attractive ITNs remained more effective overall, though the reduction in transmission were less pronounced. Our findings suggest that both inert and attractive ITNs could enhance malaria control efforts, outperforming current deterrent ITNs, even in resistant vector populations. Shifting from deterrent to inert or attractive ITNs could significantly improve vector control strategies, warranting further research and product development to fully explore and optimize this approach
Rare-metal mineralization and crustal magmatism in a collisional orogen: protracted evolution of the Sioule metallogenic zone (Variscan French Massif Central)
No full textInternational audienceThe West European Variscan Belt (360-300 Ma) is a major metallogenic province for rare metals, hosting both magmatic and hydrothermal deposits. Magmatic mineralization occurs in granitic or pegmatitic intrusions enriched in Li-Sn-Nb-Ta, while hydrothermal deposits are found in quartz veins and metasomatic rocks containing Sn ± W. Although these two mineralization types are often spatially associated within the same deposit particularly at the margins of granitic batholiths, they typically exhibit complex, multiphase evolutionary histories that are difficult to reconcile with a single metallogenic event. This issue is well exemplified in the Sioule region, an archetypal metallogenic zone within the Variscan Belt. There, a large quartz vein system mineralized in wolframite (W) and minor cassiterite (Sn) was later intruded by the Beauvoir rare-metal granite, which notably hosts disseminated Li-micas along with magmatic cassiterite. While the late-orogenic nature of rare-metal magmatism (315-310 Ma) is well constrained, the timing of the hydrothermal W-Sn mineralizing events and their relationship to regional tectonic and magmatic processes remains controversial, with previously proposed ages ranging from 330 to 310 Ma, and even older than 360 Ma. To address this, a petro-geochemical study combined with U-Pb zircon-apatite geochronology was conducted on metasediments and felsic igneous rocks from the Sioule Metallogenic Zone (SMZ), along with direct U-Pb dating of ore minerals including cassiterite and wolframite. Our data indicate that the protoliths of Sioule series migmatites include Late-Ediacarian (< 540 Ma) detrital sediments as well as peraluminous orthogneisses dated at 510-490 Ma. Partial melting likely occurred during or before Early Variscan peraluminous magmatism at 354 ± 6 Ma, marked by the emplacement of now variably deformed biotite ± garnet ± cordierite granitoids (Chantelle and St-Gervais massifs). This magmatic event was subsynchronous with N110-trending dextral shearing, and early W hydrothermalism as confirmed by a wolframite U-Pb age at 351 ± 9 Ma. By 333 ± 6 Ma, the southern part of SMZ was exhumed, and metaluminous to peraluminous felsic magmas were either erupted as tracy-dacite tuffs or crystallized at depth as biotite microgranite (Pouzol-Servant massif). Further north, highly peraluminous and Li-rich granitic magmatism, a precursor of rare-metal granite magmatism, formed rhyolitic dykes and granitic pluton (Colettes massif), possibly linked to N60 senestral shearing. Such event was likely accompanied by a second W mineralizing phase, constrained by apatite U-Pb dating at 329 ± 5 Ma. Between 320 and 310 Ma, the southeastern part of the SMZ experienced a new plutonic-volcanic episode marked by the emplacement of biotite granite (Champs massif) and trachy-dacite tuff. Such volcanism appears geochemically similar to the regionally recognized Visean volcanic event at ca. 330 Ma. At this time, magmatic fluids exsolved from the crystallizing Beauvoir rare-metal granite precipitated hydrothermal cassiterite (319 ± 7 Ma) and a third generation of wolframite.This study highlights the protracted magmatic and hydrothermal history of a metallogenic zone of the Varican belt in relation to episodic peraluminous magmatism. It also supports the common temporal decoupling of Sn and W deposition events, even within the same deposit
Daily and seasonal spatial behaviour of waved whelk Buccinum undatum: implications for fishery management and restoration
No full textInternational audienceMany marine invertebrate fisheries are vulnerable to overexploitation and require conservation measures to ensure their sustainability. In recent years, landings of waved whelk Buccinum undatum (Linnaeus, 1758) have fallen considerably along the coasts of the St. Lawrence (Canada), dropping by up to 76% in some fishing areas. Local overfishing may be explained, in part, by the species’ oviparous reproduction, which limits larval dispersion and population interactions. Using a two year long-term acoustic telemetry study, we tracked 20 individuals to assess their general movement behaviour and potential to disperse within an area. Tagged individuals displayed considerable daily movements (from a mean of 2 to 16 m/h), with no sensible differences between day and night. Both speed and habitat usage potential (HUP) varied among seasons, most likely due to the species' reproductive cycle: HUP was significantly larger during the breeding season of May (4570 m2) and June (2779 m2) than in the preceding winter months (1046 m2). Although some individuals moved over a considerable depth range, we did not observe seasonality over the bathymetric gradient of the area. Overall, the species' restricted HUP limits individual dispersal, resulting in very poor connection between neighbouring populations and making B. undatum vulnerable to local overfishing
Les grandes cultures et le changement climatique : les cas des filières riz, sorgho, canne à sucre et coton
No full textSource Agritrop Cirad (https://agritrop.cirad.fr/614698/)International audienc