Mapping aboveground forest biomass is central for assessing the global carbon balance. However, current large-scale maps show strong disparities, despite good validation statistics of their underlying models. Here, we attribute this contradiction to a flaw in the validation methods, which ignore spatial autocorrelation (SAC) in data, leading to overoptimistic assessment of model predictive power. To illustrate this issue, we reproduce the approach of large-scale mapping studies using a massive forest inventory dataset of 11.8 million trees in central Africa to train and validate a random forest model based on multispectral and environmental variables. A standard nonspatial validation method suggests that the model predicts more than half of the forest biomass variation, while spatial validation methods accounting for SAC reveal quasi-null predictive power. This study underscores how a common practice in big data mapping studies shows an apparent high predictive power, even when predictors have poor relationships with the ecological variable of interest, thus possibly leading to erroneous maps and interpretations.

Monitoring forest–agriculture mosaics is crucial for understanding landscape heterogeneity and managing biodiversity. Mapping these mosaics from remotely sensed imagery remains challenging, since ecological gradients from forested to agricultural areas make characterizing vegetation more difficult. The recent synthetic aperture radar (SAR) Sentinel-1 (S-1) and optical Sentinel-2 (S-2) time series provide a great opportunity to monitor forest–agriculture mosaics due to their high spatial and temporal resolutions. However, while a few studies have used the temporal resolution of S-2 time series alone to map land cover and land use in cropland and/or forested areas, S-1 time series have not yet been investigated alone for this purpose. The combined use of S-1 & S-2 time series has been assessed for only one or a few land cover classes. In this study, we assessed the potential of S-1 data alone, S-2 data alone, and their combined use for mapping forest–agriculture mosaics over two study areas: a temperate mountainous landscape in the Cantabrian Range (Spain) and a tropical forested landscape in Paragominas (Brazil). Satellite images were classified using an incremental procedure based on an importance rank of the input features. The classifications obtained with S-2 data alone (mean kappa index = 0.59–0.83) were more accurate than those obtained with S-1 data alone (mean kappa index = 0.28–0.72). Accuracy increased when combining S-1 and 2 data (mean kappa index = 0.55–0.85). The method enables defining the number and type of features that discriminate land cover classes in an optimal manner according to the type of landscape considered. The best configuration for the Spanish and Brazilian study areas included 5 and 10 features, respectively, for S-2 data alone and 10 and 20 features, respectively, for S-1 data alone. Short-wave infrared and VV and VH polarizations were key features of S-2 and S-1 data, respectively. In addition, the method enables defining key periods that discriminate land cover classes according to the type of images used. For example, in the Cantabrian Range, winter and summer were key for S-2 time series, while spring and winter were key for S-1 time series.

Understanding the factors that shape the distribution of tropical tree species at large scales is a central issue in ecology, conservation and forest management. The aims of this study were to (i) assess the importance of environmental factors relative to historical factors for tree species distributions in the semi-evergreen forests of the northern Congo basin; and to (ii) identify potential mechanisms explaining distribution patterns through a trait-based approach. We analyzed the distribution patterns of 31 common tree species in an area of more than 700,000 km² spanning the borders of Cameroon, the Central African Republic, and the Republic of Congo using forest inventory data from 56,445 0.5-ha plots. Spatial variation of environmental (climate, topography and geology) and historical factors (human disturbance) were quantified from maps and satellite records. Four key functional traits (leaf phenology, shade tolerance, wood density, and maximum growth rate) were extracted from the literature. The geological substrate was of major importance for the distribution of the focal species, while climate and past human disturbances had a significant but lesser impact. Species distribution patterns were significantly related to functional traits. Species associated with sandy soils typical of sandstone and alluvium were characterized by slow growth rates, shade tolerance, evergreen leaves, and high wood density, traits allowing persistence on resource-poor soils. In contrast, fast-growing pioneer species rarely occurred on sandy soils, except for Lophira alata. The results indicate strong environmental filtering due to differential soil resource availability across geological substrates. Additionally, long-term human disturbances in resource-rich areas may have accentuated the observed patterns of species and trait distributions. Trait differences across geological substrates imply pronounced differences in population and ecosystem processes, and call for different conservation and management strategies.

African forests within the Congo Basin are generally mapped at a regional scale as broad-leaved evergreen forests, with the main distinction being between terra-firme and swamp forest types. At the same time, commercial forest inventories, as well as national maps, have highlighted a strong spatial heterogeneity of forest types. A detailed vegetation map generated using consistent methods is needed to inform decision makers about spatial forest organization and their relationships with environmental drivers in the context of global change. We propose a multi-temporal remotely sensed data approach to characterize vegetation types using vegetation index annual profiles. The classifications identified 22 vegetation types (six savannas, two swamp forests, 14 forest types) improving existing vegetation maps. Among forest types, we showed strong variations in stand structure and deciduousness, identifying (i) two blocks of dense evergreen forests located in the western part of the study area and in the central part on sandy soils; (ii) semi-deciduous forests are located in the Sangha River interval which has experienced past fragmentation and human activities. For all vegetation types enhanced vegetation index profiles were highly seasonal and strongly correlated to rainfall and to a lesser extent, to light regimes. These results are of importance to predict spatial variations of carbon stocks and fluxes, because evergreen/deciduous forests (i) have contrasted annual dynamics of photosynthetic activity and foliar water content and (ii) differ in community dynamics and ecosystem processes.