HAL Portal UG Universite de Guyane
Not a member yet
4796 research outputs found
Sort by
Convergent iridescence and divergent chemical signals in sympatric sister-species of Amazonian butterflies
No full textInternational audienceClosely-related species are often partitioned into different geographic areas or micro-habitats as a result of competition and reproductive interference. Here, we investigate how the evolution of shared adaptive traits may shape ecological interactions and genetic divergence, and facilitate coexistence in sympatry. In closely-related Morpho butterfly species living in the understory of the neo-tropical rainforest, the blue iridescent coloration of the wings is likely involved in predation evasion as well as in mating recognition and courtship. These contrasted selective pressures acting on this visual trait shared between closely-related species has likely shaped their coexistence. We used spectrophotometry, behavioral experiments, visual modeling and chemical analyses applied to samples from allopatric and sympatric populations of Morpho helenor and Morpho achilles to characterize how the evolution of visual and chemical traits might favor their coexistence in sympatry: we quantified the differences in wing iridescence and tested for variations in the sexual preference for this trait between allopatric vs. sympatric populations. We found a strong similarity in iridescence among species in sympatry, suggesting convergence driven by predation. Although behavioral results suggest that iridescent signals could also be used as visual cues during mate choice, convergent iridescent signals may impair the visual recognition of sympatric Morpho species. In contrast, divergent chemical bouquets among species suggest that the visual similarity of sympatric Morpho species might have favored the divergence of alternative traits involved in species recognition such as chemical cues. This study underlines how ecological interactions and trait evolution can shape species coexistence in sympatry. Significance StatementBy changing the visual aspect of animals in movement, iridescence likely impacts multiple visual interactions within and among species, and might therefore shape ecological interactions and species diversification in sympatry. Here we study the relative importance of predation-driven mimicry and sexual selection on the evolution of iridescence on the wings of sister-species of Morpho butterflies. We finely quantify this conspicuous wing phenotype and used behavioral experiments to test whether the sexual preference for this trait can vary between allopatric and sympatric populations of sister-species. We find that convergent iridescence in sympatric Morpho species, putatively driven by predators and supporting the hypothesis of evasive mimicry, impairs species recognition and could promote the divergence in alternative chemical cues involved in mate discrimination.</div
MLKAPS: Machine Learning and Adaptive Sampling for HPC Kernel Auto-tuning
No full textInternational audienceMany High-Performance Computing (HPC) libraries rely on decision trees to select the best kernel hyperparameters at runtime,depending on the input and environment. However, finding optimized configurations for each input and environment is challengingand requires significant manual effort and computational resources. This paper presents MLKAPS, a tool that automates this task usingmachine learning and adaptive sampling techniques. MLKAPS generates decision trees that tune HPC kernels’ design parameters toachieve efficient performance for any user input. MLKAPS scales to large input and design spaces, outperforming similar state-of-the-artauto-tuning tools in tuning time and mean speedup. We demonstrate the benefits of MLKAPS on the highly optimized Intel MKLdgetrf LU kernel and show that MLKAPS finds blindspots in the manual tuning of HPC experts. It improves over 85% of the inputswith a geomean speedup of ×1.30. On the Intel MKL dgeqrf QR kernel, MLKAPS improves performance on 85% of the inputs with ageomean speedup of ×1.18
Response of anoxygenic and oxygenic phototrophs to an environmental gradient reveals unusual structure of freshwater microbial communities in the bromeliad ecosystem
No full textInternational audienceAnoxygenic phototrophic bacteria have recently been recognized as a ubiquitous component of microbial communities in lakes and marine environments, but studies of the ecological factors that control their significance are scarce. We conducted a manipulative field experiment using natural freshwater microcosms, the tank bromeliad ecosystem, to test the response of anoxygenic and oxygenic phototrophic microorganisms to an environmental gradient across the forest edge. We assessed the biomass of these photosynthetic communities by their pigment content and used structural equation modeling to evaluate the importance of different habitat variables as ecological drivers. We show that anoxygenic phototrophic bacteria are primarily driven by small detrital organic particles rather than directly by light. In contrast, light and habitat size were the main factors controlling the biomass of oxygenic phototrophic microorganisms. Anoxygenic phototrophic bacteria inhabiting the bromeliad ecosystem represent huge concentrations of bacteriochlorophyll a relative to large pelagic environments and form an essential and dominant part of photosynthetic biomass across a wide range of ecological conditions. These freshwater photoheterotrophs are likely to play a pivotal and unsuspected role in energy flow and nutrient cycling in neotropical forests
Phosphorus additions increase microbial phosphorus accumulation and carbon turnover in tropical soils in French Guiana
No full textInternational audienceTropical forests often grow on highly weathered soils with rather high nitrogen (N), but low rock-derived phosphorus (P) (and base cation) availability. While the role of P limitation in constraining plant productivity is well established, its impact on heterotrophic microbial communities remains less clear. Specifically, it is crucial to understand how P availability shapes microbial activity, physiology and resource acquisition strategies, but also potential repercussions on organic matter decomposition, nutrient mineralization, and long-term carbon (C) sequestration.To address this knowledge gap, we studied soil microbial communities in tropical lowland forest soils located in the north-eastern Amazon in French Guiana following three years of N and P additions. We assessed soil microbial biomass, stoichiometry, extracellular enzyme activity potential, and respiration rates. Additionally, we quantified soil microbial growth using a substrate-independent method based on the incorporation of 18O from labelled water into microbial DNA.Our results showed that soil microbial biomass slightly increased in response to N, but remained unaffected by P additions. In contrast, P additions increased microbial P content (and reduced associated C:P ratios), suggesting that microbes are highly competitive for P and can act as a significant P sink in these soils. Additionally, P additions also increased total and available soil P pools, indicating that both plant and microbial communities are well adapted to naturally occurring low P availability, and may have reached P saturation after multiple years of nutrient enrichment. Despite these changes, microbial biomass-normalized specific respiration- and growth-rates increased with both N and P fertilization, with a stronger response to P, while overall, the C use efficiency of the microbial communities remained unaffected by both.Our results highlight (i) the pivotal role of soil microbes in C, N and P cycling in tropical forest soil and (ii) the remarkable P storage capacity of microbial communities in highly weathered soils. While microbial C and N dynamics appear tightly coupled, likely due to the similar composition of microbial cell walls, our data demonstrate non-homoeostatic stoichiometric behavior of microbial communities. This underscores the importance of reconsidering assumptions about strict stoichiometric relationships in soil and ecosystem models.
Jean-Marie Brohm. Sociologie politique du sport. Une vision totalitaire du monde. Alboussière, QS ? Éditions, 2025, 422 p
No full textInternational audienc
Restitução e rapatriação de coleções indígenas ao redor do mundo: O direito, amigo o inimigo dos povos ?
No full textInternational audienc
Leptospirosis seroprevalence and exposure factors in three informal settlements of French Guiana: An opportunistic survey
No full textInternational audienceBackgroundLeptospirosis is a zoonotic disease of increasing importance in French Guiana. It particularly affects subjects living in precarious conditions. We aimed to determine the seroprevalence and the risk of exposure to leptospirosis among inhabitants of three informal settlements in French Guiana.MethodsA serological investigation was conducted in 2022 in three informal settlements in the area of Cayenne, the main city of French Guiana. Leptospirosis exposure factors were assessed in volunteers aged > 15 through a standardized questionnaire. Leptospirosis seroprevalence was evaluated with Microscopic Agglutination Test (MAT) using 17 pathogenic Leptospira antigens with a reactivity threshold of 1:100.ResultsIn 266 participants, median [IQR] age was 42 [34–52] and male to female sex ratio was 0.9. Most participants were migrants (96%), mainly from Haiti (83%), and lived in the study area for at least 2 years (82%). Household rodent exposure (89%) and use of other water sources than collective standpoint (92%) were common. An at-risk occupation was reported for 68% of working participants. Leptospirosis seroprevalence was 7.5% (95% CI [4.7-11.4]) with Ballum and Icterohaemorrhagiae as the main serogroups. Foot skin exposure in wet environments was associated with reactive serum (OR 7.6, 95% CI [1.1 - 326.7]).ConclusionDespite a high theoretical risk of leptospirosis exposure among informal settlements inhabitants, only a few participants were seroreactive for Leptospira. This may suggest that despite at-risk exposures the effective transmission of leptospirosis remains limited within the study area. Broader serological surveys and environmental studies should clarify the areas of at-risk leptospirosis transmission in French Guiana
Reliability Assessment of 15 Gridded Rainfall Datasets for the Construction of a Daily High-Resolution Reanalysis across Senegal for Agroclimatic Applications
No full textInternational audienceAbstract This study focuses on developing a new high-resolution gridded rainfall dataset for Senegal, essential for supporting rainfed agriculture, which is sensitive to climate variability. Given the limited number of rain gauges, the research evaluates 15 publicly available gridded rainfall datasets (P datasets) against data from 21 stations of the Senegalese National Meteorological Service (ANACIM) over a 17-yr period (2005–21). The evaluation employs several agroclimatic indices, including the onset and cessation of rain, duration of the rainy season, and extreme events. The findings reveal that the reliability of P datasets varies significantly based on the metrics used. For total rainfall, African Rainfall Climatology, version 2 (ARC2), Climate Hazards Infrared Precipitation with Station (CHIRPS), ERA5, and Rainfall Estimation Algorithm, version 2 (RFEv2) emerged as the most reliable datasets, with ERA5 achieving the highest Kling–Gupta efficiency (KGE) value of 0.81 at daily scale. In terms of agroclimatic parameters, ARC2, CHIRPS, and RFEv2 excelled in accurately representing the start (KGE ≥ 0.45) and end (KGE ≥ 0.39) dates of the rainy season. However, P datasets generally overestimate rainfall events and struggle with identifying dry spells. The newly constructed merged dataset (M dataset) demonstrated over 100% improvement in correlation for daily estimates and significant bias reductions: 99.19% for ARC2, 80% for CHIRPS, and 90.57% for RFEv2. This research provides critical insights for selecting appropriate datasets to enhance climate information for agricultural decision-making in Senegal
The value of local allometries from airborne laser scanning for tropical forest biomass estimates
No full textInternational audienceTo accurately assess forest carbon stocks for climate change projections, information on tree height and stem diameter is essential. However, a persistent lack of reliable plot-level inventory data, particularly in carbon-rich tropical forests, leads to local biases in biomass estimates. Pantropical allometries for tree dimensions and biomass have reduced bias at the regional level, but there continue to be inconsistencies and biases at the local level due to reference data quality. For example, classical instruments for measuring tree height such as clinometers and rangefinders have limited accuracy in dense, closed-canopy forests and can only be applied over small scales. The present study seeks to establish the effectiveness of airborne laser scanning data for determining site-specific allometric relationships between stem diameter and tree height, thereby improving the accuracy of above-ground biomass estimations (AGB) at the plot level. We used 118.75 ha of ground inventory data from a vast network of permanent sample plots in a tropical moist forest in French Guiana. The plots covered a range of forest structure and heights as they included undisturbed forests as well as previously logged over plots. Ground data was combined with data derived from airborne laser scanning (ALS) to establish allometric height-diameter (H-DBH) models, both via a Bayesian multilevel modeling approach and an individual-based forest model (Canopy Constructor approach). Our results show that replacing a universal pantropical allometry with a locally derived species specific ALS-based H-DBH relationship nearly halved the mean height prediction error. Incorporating species identity into Bayesian models contributed to more than 50% of the total error reduction, a pattern reliably inferred by Canopy Constructor even without direct crown measurements. Both approaches yielded consistent AGB predictions, which were 11 to 13% higher (40 to 54 t ha -1 ) than those obtained using pantropical allometries. These findings underscore the potential of ALS data to enhance biomass estimations by reducing biases at local scales, providing a more accurate foundation for carbon stock assessments in tropical forests
Observation-only deep learning for gappy satellite-derived ocean colour data using 4DVarNet
No full textInternational audienceMonitoring optical properties of coastal and open ocean waters is crucial to assessing the health of marine ecosystems. Deep learning offers a promising approach to address these ecosystem dynamics, especially in scenarios where gapfree ground-truth data is lacking, which poses a challenge for designing effective training frameworks. Using an advanced neural variational data assimilation scheme (called 4DVarNet), we introduce a comprehensive training framework designed to effectively train directly on gappy data sets. Using the Mediterranean Sea as a case study, our experiments not only highlight the high performance of the chosen neural network in reconstructing gap-free images from gappy datasets but also demonstrate its superior performance over state-of-the-art algorithms such as DInEOF and end-to-end neural mapping schemes based CNN or UNet architectures.</div