Cartographie de l'habitat de reproduction du tétras-lyre (Lyrurus tetrix) dans les Alpes françaises

Alexandre Défossez, Samuel Alleaume, Marc Montadert, Dino Ienco, Sandra LuquearXiv 2024 · DOI · arXiv

Abstract

The Black Grouse (Lyrurus tetrix) is an emblematic alpine species with high conservation importance. The population size of these mountain birds tends to decline on reference sites and shows differences according to changes in local landscape characteristics. Habitat changes are at the center of the identified pressures impacting part or all of its life cycle. Hence, an approach to monitor population dynamics is through modeling the favorable habitats of Black Grouse breeding (nesting sites). Coupling modeling with multi-source remote sensing data (medium and very high spatial resolution) allowed the implementation of a spatial distribution model of the species. The extraction of variables from remote sensing helped describe the area studied at appropriate spatial and temporal scales: horizontal and vertical structure, functioning, phenology, and biodiversity. An annual time series of radiometric indices from Sentinel-2 generated Dynamic Habitat Indices (DHIs) to derive phenological indications. In addition, very high-resolution images (SPOT 6) provided access to the fine structure of natural habitats. A deep learning model was used to classify elementary strata. Finally, the Biomod2 R platform was applied to model the favorable habitat of Black Grouse reproduction. Of all the models, Random Forest and Extreme Boosted Gradient were the best performing. In this ensemble model, the most important explanatory variables are altitude, the proportion of heathland, and the DHI. Results from the habitat model can be used as an operational tool for monitoring forest landscape shifts and changes, and for delimiting potential areas to protect the species habitat.

Introduction

Illustration of the Black Grouse (Lyrurus tetrix), an emblematic alpine species. Source, p. 1

Illustration of the Black Grouse (Lyrurus tetrix), an emblematic alpine species.

The Black Grouse is an emblematic but vulnerable galliform species in the Alps, currently threatened by climate change, habitat degradation (milieu closure), and increasing human disturbance from mountain sports. Effective conservation requires precise knowledge of where these birds breed. However, mapping "favorable" habitat across the vast, rugged terrain of the Alps is a massive logistical challenge. Traditional methods rely on localized field observations, which lack the scale needed for regional management. This study addresses the gap by using remote sensing to create a predictive Species Distribution Model (SDM) that can identify potential breeding sites across the entire French Alpine range.

Methodology

Methodological workflow for mapping breeding habitats using multi-source remote sensing and ensemble modeling. Source, p. 6

Methodological workflow for mapping breeding habitats using multi-source remote sensing and ensemble modeling.

The authors employed a sophisticated "ensemble modeling" approach, integrating several advanced geospatial techniques:

  • Multi-Source Remote Sensing: They combined Sentinel-2 imagery (10m resolution) to capture vegetation phenology (the "timing" of life cycles) with SPOT 6/7 imagery (1.5m resolution) to analyze fine-scale landscape structure.
  • Deep Learning Classification: A two-stage Convolutional Neural Network (CNN) was used to perform semantic segmentation on the SPOT imagery. This allowed the researchers to distinguish between specific vegetation strata, such as grasslands, forests, and—crucially—ericaceous heathlands (shrubs like blueberries and rhododendrons).
  • Dynamic Habitat Indices (DHI): Using the iota2 pipeline, they calculated DHI components (specifically NDVI Max and NDWI Max) to serve as proxies for peak primary productivity and vegetation water content.
  • Ensemble SDM: Using the biomod2 R package, they tested 12 different algorithms (including GLM, Random Forest, and XGBOOST). They ultimately selected Random Forest (RF) for the final ensemble due to its robustness against overfitting and its ability to handle complex ecological interactions.

Datasets

  • Biological Data: 1,453 validated brood observations (2015–2021) provided by the Observatory of Mountain Galliformes (OGM).
  • Satellite Imagery: Sentinel-2 (Copernicus/ESA) and SPOT 6/7 (CNES/DINAMIS) mosaics from 2020.
  • Topographic Data: BD ALTI (Digital Elevation Model) from the French National Geographic Institute (IGN).
  • Reference Data: BD Forêt V2 (IGN) used to train the vegetation classification models.
  • Pseudo-Absences: A generated set of 4,359 random points used to train the SDM in the absence of "true" absence data.

Key Results and Insights

Model Performance

The Random Forest ensemble achieved high predictive accuracy, with an Area Under the Curve (ROC) of 0.96 and a True Skill Statistic (TSS) of 0.82. While XGBOOST showed slightly higher raw performance metrics, the authors opted for Random Forest for the final projection, noting that XGBOOST's near-perfect scores likely indicated overfitting to the local training data.

Variable Importance

  • Altitude: ~24–25% (The most influential variable). The model effectively filters out areas below 1,600m, as observations below this altitude have become anecdotal due to climate-driven habitat shifts.
  • Heathland Rate: ~14–23%. The study confirms that the presence of ericaceous shrubs is a primary driver of habitat suitability, providing essential food and cover for chicks.
  • Phenology (DHI): ~10–15%. By using maximum vegetation indices, the researchers successfully captured the peak biological activity necessary for reproduction, bypassing seasonal snow cover noise.

The resulting maps show a clear "sweet spot" for breeding between 1,600m and 2,300m, with suitability dropping to near zero above 2,400m where rocky mineral surfaces dominate.

Limitations and Future Work

  • Computational Constraints: The authors encountered significant RAM issues when running biomod2 on large datasets, suggesting that future large-scale modeling will require more efficient parallel processing or cloud-based workflows.
  • Regional Variation: The current model was optimized for the Northern Inner Alps. In the Southern Alps, the birds rely more on larch forests (mélézin), which would require a re-calibrated model.
  • Operational Use: The next step is to use these breeding maps to explain the distribution of "singing males" (leks) and to improve sampling designs for future population monitoring. These maps constitute a valuable decision-making tool for the conservation management of mountain open forests.

Summary updated

Abstract

The Black Grouse (Lyrurus tetrix) is an emblematic alpine species with high conservation importance. The population size of these mountain bird tends to decline on the reference sites and shows differences according to changes in local landscape characteristics. Habitat changes are at the centre of the identified pressures impacting part or all of its life cycle, according to experts. Hence, an approach to monitor population dynamics, is trough modelling the favourable habitats of Black Grouse breeding (nesting sites). Then, coupling modelling with multi-source remote sensing data (medium and very high spatial resolution), allowed the implementation of a spatial distribution model of the species. Indeed, the extraction of variables from remote sensing helped to describe the area studied at appropriate spatial and temporal scales: horizontal and vertical structure (heterogeneity), functioning (vegetation indices), phenology (seasonal or inter-annual dynamics) and biodiversity. An annual time series of radiometric indices (NDVI, NDWI, BI {\ldots}) from Sentinel-2 has made it possible to generate Dynamic Habitat Indices (DHIs) to derive phenological indications on the nature and dynamics of natural habitats. In addition, very high resolution images (SPOT6) provided access to the fine structure of natural habitats, i.e. the vertical and horizontal organisation by states identified as elementary (mineral, herbaceous, low and high woody). Indeed, one of the essential limiting factors for brood rearing is the presence of a well-developed herbaceous or ericaceous stratum in the northern Alps and larch forests in the southern region. A deep learning model was used to classify elementary strata. Finally, Biomod2 R platform, using an ensemble approach, was applied to model, the favourable habitat of Black Grouse reproduction. Of all the models, Random Forest and Extreme Boosted Gradient are the best performing, with TSS and ROC scores close to 1. For the SDM, we selected only Random Forest models (ensemble modelling) because of their low susceptibility to overfitting and coherent predictions (after comparing model predictions).In this ensemble model, the most important explanatory variables are altitude, the proportion of heathland, and the DHI (NDVI Max and NDWI Max). Results from the habitat model can be used as an operational tool for monitoring forest landscape shifts and changes. In addition, to delimiting potential areas to protect the species habitat, which constitute a valuable decision-making tool for conservation management of mountain open forest.
BibTeX
@misc{defossez2024cartographie,
  author = {Alexandre Défossez and Samuel Alleaume and Marc Montadert and Dino Ienco and Sandra Luque},
  title = {{Cartographie de l'habitat de reproduction du tétras-lyre (Lyrurus tetrix) dans les Alpes françaises}},
  year = {2024},
  eprint = {2402.18597},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2402.18597},
}
Figure