Search

1. Open-source biodiversity databases contain a large number of species occurrence records but are often spatially biased; which affects the reliability of species distribution models based on these records. Sample bias correction techniques require data filtering which comes at the cost of record numbers, or require considerable additional sampling effort. Since independent data is rarely available, assessment of the correction technique often relies solely on performance metrics computed using subsets of the available – biased – data, which may prove misleading.2. Here, we assess the extent to which an acknowledged sample bias correction technique is likely to improve models’ ability to predict species distributions in the absence of independent data. We assessed variation in model predictions induced by the aforementioned correction and model stochasticity; the variability between model replicates related to a random component (pseudo-absences sets and cross-validation subsets). We present, then, an index of the effect of correction relative to model stochasticity; the Relative Overlap Index (ROI). We investigated whether the ROI better represented the effect of correction than classic performance metrics (Boyce index, cAUC, AUC and TSS) and absolute overlap metrics (Schoener’s D, Pearson’s and Spearman’s correlation coefficients) when considering data related to 64 vertebrate species and 21 virtual species with a generated sample bias.3. When based on absolute overlaps and cross-validation performance metrics, we found that correction produced no significant effects. When considering its effect relative to model stochasticity, the effect of correction was strong for most species at one of the three sites. The use of virtual species enabled us to verify that the correction technique improved both distribution predictions and the biological relevance of the selected variables at the specific site, when these were not correlated with sample bias patterns.4. In the absence of additional independent data, the assessment of sample bias correction based on subsample data may be misleading. We propose to investigate both the biological relevance of environmental variables selected, and, the effect of sample bias correction based on its effect relative to model stochasticity.

Earth observation satellite imagery is increasingly accessible, and has become a key component for vegetation mapping and monitoring. Sentinel-2 satellites acquire optical images with five days’ revisit frequency, which is an important feature to increase the probability of acquisition with reasonable cloud cover in tropical regions. Regular and reliable satellite observations open perspectives for the monitoring of vegetation properties and biodiversity. Atmospheric correction methods (ACMs) producing bottom-of-atmosphere (BOA) reflectance are critical to ensure temporal consistency of higher-level products and optimal sensitivity to changes in vegetation properties. Still their application in tropical regions remains challenging due to complex atmospheric issues. This study aims at performing ACM inter-comparison in the context of tropical forest monitoring. We produced BOA reflectance for a set of Sentinel-2 acquisitions corresponding to a forested area in Cameroon, using four atmospheric correction methods: Sen2cor, MAJA, Overland and LaSRC. We selected five successive acquisitions with moderate to no cloud cover, and computed a set of spectral indices and spectral diversity metrics in order to compare the consistency of these products through time, under the hypothesis that they should remain stable over a short period. We also assessed the agreement between atmospheric correction methods. Two spatial extents were used for the computation of spectral diversity metrics to assess the robustness of the data-driven processes applied to compute spectral diversity. We found that the choice of an ACM did have a significant impact on BOA reflectance and higher-level products. In the visible domain, Overland and LaSRC produced consistent BOA reflectance values, while MAJA and Sen2Cor showed strong variability which could not be explained by changes in surface properties. This directly influenced the temporal consistency of NDVI. Yet, the influence on the temporal consistency for EVI and NDWI was moderate. Spectral diversity metrics were consistent through time for all methods, but to a lesser degree than vegetation indices. When comparing the mean values over the period considered, vegetation indices were stable across methods, but not diversity metrics. Spatial context changes had an impact on the Shannon index, but not on Bray-Curtis dissimilarity. These results suggest that the choice of ACM has major potential implications for tropical forest monitoring.