Forest surveying and inspection face significant challenges due to unstructured environments, variable terrain conditions, and the high costs of manual data collection. Although mobile robotics and artificial intelligence offer promising solutions, reliable autonomous navigation in forest, terrain-aware path planning, and tree parameter estimation remain open challenges. In this paper, we present the results of the AI4FOREST project, which addresses these issues through three main contributions. First, we develop an autonomous mobile robot, integrating SLAM-based navigation, 3D point cloud reconstruction, and a vision-based deep learning architecture to enable tree detection and diameter estimation. This system demonstrates the feasibility of generating a digital twin of forest while operating autonomously. Second, to overcome the limitations of classical navigation approaches in heterogeneous natural terrains, we introduce a machine learning-based surrogate model of wheel–soil interaction, trained on a large synthetic dataset derived from classical terramechanics. Compared to purely geometric planners, the proposed model enables realistic dynamics simulation and improves navigation robustness by accounting for terrain–vehicle interactions. Finally, we investigate the impact of point cloud density on the accuracy of forest parameter estimation, identifying the minimum sampling requirements needed to extract tree diameters and heights. This analysis provides support to balance sensor performance, robot speed, and operational costs. Overall, the AI4FOREST project advances the state of the art in autonomous forest monitoring by jointly addressing SLAM-based mapping, terrain-aware navigation, and tree parameter estimation.

ABSTRACT Natural vegetation recovery of abandoned agricultural lands could play a significant role in combating climate change by enhancing carbon (C) sequestration over time in a passive and cost‐effective manner. After agricultural land is abandoned, natural regeneration processes, such as the regrowth of tree cover and increased microbial activity, typically restore soil organic carbon (SOC) sinks, thereby capturing atmospheric carbon dioxide (CO 2 ). However, understanding its contribution to CO 2 removal, particularly, on a large, continental scale, remains limited. In this systematic review, we aimed to gather studies investigating soil C sequestration following the abandonment of agricultural activities across Europe. By conducting an integrative analysis of data from 36 studies focusing on natural vegetation recovery after agricultural abandonment, we assessed the relative changes in SOC over time in relation to various environmental factors. Our findings revealed that SOC dynamics are influenced by management, soil reference group, exposition (aspect), and forest type, with remarkable increases found in Mediterranean regions, certain soil groups (Regosols, Cambisols, Calcisols), cropland‐converted broadleaf and mixed forests, and specific aspects (North, South, and South‐West facing sites). However, the results were strongly influenced by the uneven geographical and altitudinal distribution of study sites, which varied in terms of previous land use management, significantly affecting sequestration models. Most studies concentrated on the Mediterranean region, with grassland data predominantly coming from higher elevations. As a result, we call attention to a pressing need for broader research across Europe and present results of a gap analysis of recently abandoned croplands, highlighting especially underrepresented regions such as Northern Spain, Central France, the United Kingdom, Germany, Poland, the Baltic states, Hungary, and the Balkans, where SOC dynamics remain poorly documented. These findings provide a roadmap for researchers and policymakers to prioritize future SOC monitoring and natural vegetation recovery initiatives to enhance soil C sequestration across Europe.

Proforestation, defined as the long-term protection of existing forests to allow the self-development of old-growth attributes, is increasingly promoted as a nature-based solution for biodiversity conservation. However, its effects on tree-related microhabitats (TreMs), key indicators of forest structural complexity, remain unevenly documented across forest types. Most existing studies have largely examined TreM richness and abundance, while effects on TreM composition across contrasting forest contexts have rarely been addressed.We investigated how time since abandonment (TSA), as a proxy for proforestation duration, influences TreM richness, abundance, and composition across three different forest categories spanning Mediterranean to Alpine environments. Within each category, we compared actively managed forests with stands under proforestation for more than 20 and more than 60 years. TSA generally promoted TreM richness and abundance, albeit to a lesser extent in the Mediterranean forest site. In contrast, responses of individual TreM groups were strongly context-dependent, with some groups increasing and others decreasing along the proforestation gradient.TreM composition differed significantly among abandonment stages in all forest categories, although the magnitude was generally modest, with long-term proforested stands supporting distinct assemblages primarily driven by turnover and associated with increasing deadwood availability. The pattern of these compositional shifts varied among forest categories, reflecting differences in environmental conditions and management history.Our findings demonstrate that proforestation could not only enhance TreM availability but also influence TreM assemblages through long-term structural dynamics. Integrating long-term proforested stands into forest planning can represent an effective conservation strategy to promote biodiversity-related forest structures.

Spontaneous forest expansion following land abandonment can play a key role in achieving European targets for climate-change mitigation and biodiversity conservation. Understanding how biodiversity relates to carbon (C) stocks across successional stages can inform management strategies that simultaneously promote species diversity and C sequestration, thereby optimizing land use for ecosystem multifunctionality. We analysed 16 chronosequences spanning five successional stages, from meadows and pastures to mature forests (up to similar to 75 years since abandonment), organized into four clusters along a latitudinal gradient in Italy, encompassing the Alpine, Continental, and Mediterranean biogeographical regions. We quantified vegetation, deadwood, and soil C pools and calculated diversity indices for herbaceous plant species. Linear and generalized linear mixed models were used to assess successional stage and site effects on C stocks and diversity indices. Total ecosystem C increased along succession, driven primarily by tree biomass, reaching 195-289 Mg C ha(-1) in late-successional forests. Soil C showed no clear successional trend, with weak or site-specific patterns. Herbaceous species richness and diversity peaked in managed meadows/pastures and early encroachment stages but declined towards closed-canopy forests in three sites. By contrast, a U-shaped pattern emerged in the southernmost site. Consequently, the C-diversity relationship was predominantly negative, except for the non-linear response observed in the Mediterranean site. Overall, spontaneous reforestation promotes C storage but often reduce herbaceous plant diversity, revealing potential trade-offs between climate mitigation and plant diversity. However, under favourable environmental conditions, partial recovery of plant diversity in late-successional forests may occur, suggesting for win-win management policies.

Forest structural heterogeneity is widely recognized as a key driver of biodiversity, yet its effects on different components of understory diversity across spatial scales remain insufficiently understood, particularly when assessed using high-resolution remote sensing techniques. In this study, we examined the relationship between forest structure and understory plant diversity along secondary forest succession using high-density uncrewed laser scanning (ULS) data. We derived a suite of LiDAR-based structural metrics describing canopy height, vertical heterogeneity and canopy openness, and evaluated their association with both taxonomic and functional diversity at alpha- and beta-scales, through linear mixed models and distance-based redundancy analysis. Structural metrics showed a limited explanatory power for local (alpha) diversity, with generally weak relationships across taxonomic and functional indices. In contrast, forest structure consistently explained variation in beta diversity, with canopy openness and vertical heterogeneity emerging as the strongest predictors of community dissimilarity. These findings suggest that forest structure plays a greater role in shaping differences in species composition among stands than in determining within-stand diversity. Overall, our findings highlight the potential of ULS-derived structural metrics to capture spatial patterns of understory diversity, while underscoring the importance of accounting for multiple environmental drivers and scale-dependent processes.

Old-growth tropical forests store vast amounts of carbon in their aboveground biomass (AGB), yet the relative roles of abiotic factors such as climate, soil, and topography in governing its spatial distribution remain poorly understood. In particular, the degree to which climate acts on AGB through forest structure is still poorly quantified at the pantropical scale. Using a pantropical dataset of more than 2,000 old-growth forest plots and a structure-explicit framework, we assess how climate influences AGB through its effects on four structural attributes: basal area, mean diameter, stem density, and basal area-weighted wood density. We find that climate shapes AGB primarily through its effects on forest structure. However, structural attributes respond to climate in opposite directions, so climate’s net effect on AGB largely cancels out, and no clear climate-AGB relationship emerges across tropical regions. Moreover, only wood density responds consistently, decreasing with annual precipitation and increasing with precipitation seasonality, whereas all other attributes respond to climate differently from one region to another. This geographical variation further obscures any global climatic signal on AGB and points to the role of biogeographic history in shaping forest structure. Our findings highlight the central role of the climate-structure nexus in explaining AGB variation, and call for structure-explicit models to improve carbon stock predictions and inform climate adaptation strategies.

Abstract. Global forest assessments assist climate policy development, ecosystem science, and conservation planning, yet they rely on biomass and canopy data that do not explicitly represent the stand structural attributes derived from tree diameter measurements. This limits the ability to compare size-related structure and within-stand heterogeneity at large spatial scales. Here we present a global, spatially explicit dataset of stand-level tree diameter structure for forest cover in 2020 at 0.027° (~3 km) resolution, based on 1,203,524 georeferenced forest inventory plots comprising 54.6 million trees (≥10 cm DBH) integrated with more than 50 environmental and satellite-derived covariates into machine learning models. The dataset provides the first globally consistent maps of three complementary diameter-based metrics: arithmetic mean diameter (Dmean), quadratic mean diameter (Dqm), and the coefficient of variation of diameter (Dcv), representing average tree size, large-tree dominance, and within-stand size variability, respectively. Model performance of the ecozone-specific Random Forest framework ranged from R² = 0.41–0.82 (RMSE = 3.91–4.63 cm) for Dmean, R² = 0.43–0.83 (RMSE = 4.38–5.27 cm) for Dqm, and R² = 0.47–0.62 with (RMSE = 0.10–0.13) for Dcv across different forest ecozones. By jointly quantifying central tendency and variability in tree size, the dataset revealed spatial patterns of forest structural organization not captured by existing biomass or canopy-height products. It provides a consistent baseline for cross-biome comparison of forest structure, supporting parameterization and evaluation of vegetation and Earth system models, while offering an independent benchmark for remotely sensed structural proxies. Furthermore, it enables spatial assessment of stand structural attributes, including large-tree dominance and structural complexity, facilitating integration of diameter-based structure into global analyses of carbon dynamics and ecosystem functioning.

Watersheds are natural units and meta-ecosystems of the earth's land surface providing multiple ecological functions. However, little is known about the biodiversity-ecological multifunctionality relationships of watersheds, particularly regarding how these relationships scale to large and complex landscapes. Here, we explore the impact of forest tree species richness on the ecological multifunctionality of watersheds in terms of carbon sequestration, carbon storage, water supply, water regulation, and soil conservation, by utilizing integrated ground-sourced forest inventory datasets comprising 846 forest watersheds from the Global Forest Biodiversity Initiative, the Global Streamflow Indices and Metadata Archive, and remote sensing data products. We find a consistently positive relationship between forest tree species richness and watershed ecological multifunctionality by accounting for factors such as forest structural characteristics and environmental conditions. Furthermore, we find that this biodiversity-multifunctionality link is dependent on spatial scale and climatic context, becoming stronger in larger watersheds but diminishing in arid climatic conditions. These insights enhance our understanding of ecosystem multifunctionality and underscore the importance of considering watershed-scale ecological processes and biodiversity in ecosystem management and conservation strategies.

Plenty of information on evapotranspiration (ET) dynamics and partitioning into nonbiological (evaporation, E) and biological (transpiration, T) components is available in literature. However, in agro‐ecosystems where more than one vegetation group is found, like intercropping or grassed orchards and vineyards, it is of great use to understand the contribution to T due to the single plant type or group of plants. We deployed empirical and modeling methods to study the ecosystem evapotranspiration (ET EC ) components in a grassed vineyard in Caldaro (Italy) aiming to assess (a) which process, E or T, had greater influence on ET EC dynamics; (b) which component among grapevines and understorey portion dominated the ET EC ; and (c) how rainfall influences ET EC components. A top‐down approach combined the eddy covariance method to estimate ET EC , and the Transpiration Estimation Algorithm method to partition it. A bottom‐up approach integrated the understorey evapotranspiration (ETu) with modeled vines transpiration (Tv (mod) ). Measured and modeled fluxes showed high daily variability, consistently with meteorological conditions (vapor pressure deficit, Rn and Tair). The mean daily ET EC integrals were 3.45 and 3.40 mm d −1 (2021 and 2022), being T EC (estimated transpiration fraction of ET EC ) the higher contributor (T EC /ET EC of 0.77 and 0.79, same years). From the bottom‐up approach, ETu assessed during ground flux chamber campaigns (0.74–1.65 mm d −1 ) was lower than Tv (mod) . A high agreement (R 2 = 0.85) was found between the eddy covariance ET hourly values and ET by summing Tv (mod) and ETu. We concluded that the T process represented major fluxes in the agroecosystem during the warm season. Furthermore, the bottom‐up approach indicated the vines as primary contributors to ecosystem T, particularly noticeable after rainfall, as the understorey T fraction (Tu) increased when the system became drier. This study helps disentangling grapevine contribution to evapotranspiration from adjacent herbaceous vegetation in a vineyard, and emphasizes the dominance of biologically mediated transpiration influenced by meteorological conditions. This novel combination of approaches not only enhances understanding of Mediterranean viticulture but also illuminates broader applications in sparsely vegetated environments, such as agroforestry systems and orchards, advancing ecological management practices.

Anticipating establishment cuts in transitional high forests of beech (Fagus sylvatica L.) is a sustainable strategy from both ecological and economic perspectives. It could be scaled over large areas to promote structural heterogeneity while minimizing disturbances. Beech forests, traditionally managed as coppices for firewood, have experienced significant changes in management practices, particularly in Southern Europe. This shift was especially noticeable in the Southern and Eastern Alps, where vast areas of coppice forests were gradually transformed into high forests, as fuelwood demand declined. However, large-scale regeneration cuts at the end of the rotation period can result in economic losses and environmental concerns. We assessed whether applying regeneration cuts at 70 years in transitional high forests can effectively accelerate the coppice-to-high-forest conversion process, relative to the conventional rotation period of 120–140 years. This study examines the possibility to implement regeneration cuts before the common rotation period in temporary high forests by applying four distinct treatments: (1) control—thinning from below; (2) shelterwood system—establishment cut; (3) clear-cut; and (4) crop tree release. We found significant differences in basal area, biomass, leaf area index after tree removal, and harvesting costs/venues, with the shelterwood system being the most economically advantageous treatment. Ten years after the treatment execution, the shelterwood treatment exhibited prompt and widespread regeneration compared to other regeneration treatments, with the highest seedling abundance (12 ± 2 seedlings m−2) and height of the established saplings (93 ± 6 cm). Our findings support the idea of implementing and gradually scaling regeneration cuts in time and space using the group shelterwood system. This approach can increase the structural heterogeneity of forest stands, maintain consistent timber production, and minimize disturbances to fauna and other ecosystem services.

Spontaneous afforestation following land abandonment has been increasingly recognized as a nature-based solution to mitigate climate change and provide measurable benefits to biodiversity. However, afforestation effects on biodiversity, particularly on soil microbial communities, are still poorly characterized, with most previous studies focusing on artificial plantations rather than forest rewilding dynamics. Here, we assessed changes in topsoil physical–chemical properties and related dynamics of bacterial and fungal community composition and structure following spontaneous afforestation of abandoned grasslands in Northeast Italy over the last 70 years. With a space-for-time approach, we selected four chronosequences representing different successional stages: grassland, early (2000–2020), intermediate (1978–2000), and late (1954–1978). Results showed that spontaneous afforestation progressively reduced topsoil pH and total phosphorus (P), while soil organic carbon (SOC), nitrogen (N), and C:N ratio increased. Correspondingly, the overall α-diversity of the fungal community, assessed by ITS DNA metabarcoding, progressively decreased after an initial increase from grassland conditions, following substrate acidification and trophic specialization. Bacterial diversity, assessed by 16S DNA metabarcoding, was highest at the initial stages, then progressively decreased at later stages, likely limited by lower organic matter quality. Shifts of fungal community composition included an increase of ectomycorrhizal Basidiomycota linked to topsoil’s higher SOC, N, and C:N ratio. Differently, bacterial community composition responded substantially to pH, with topsoil acidity favoring Proteobacteria (Pseudomonadota) and Acidobacteria (Acidobacteriota) at the late afforestation stages. Our findings provide a first contribution to clarify how fungi and bacteria respond to spontaneous afforestation. This is particularly relevant in the context of climate change mitigation, considering the fundamental role of microorganisms in shaping soil carbon storage dynamics.

An accurate estimation of organic carbon (OC) in forest ecosystems is essential for understanding carbon dynamics and informing climate change mitigation strategies. This study presents a novel, explainable machine learning framework to estimate two key carbon pools: carbon sequestration in living trees (CSE) and carbon storage in standing deadwood (SDC). The methodology is structured into five key steps. First, we extract Gray-Level Co-occurrence Matrix (GLCM) texture features from LiDAR-derived canopy height models to quantify spatial heterogeneity in forest structure. Second, we integrate these GLCM metrics with vegetation indices (VIs), geomorphological variables, and weather data to create six distinct input configurations. Third, we train and evaluate teen models on each configuration to assess model performance and feature synergy. Fourth, we apply SHapley Additive exPlanations (SHAP) to the three models to transform them into an interpretable white-box model, identifying key predictors such as AVG_mean, SD_entropy, and SD_homogeneity. Finally, we assess model uncertainty using jackknife resampling and error bar analysis. The results indicate that CatBoost and Random Forest models deliver the highest performance for OC estimation. This study is the first to apply GLCM features for the joint estimation of CSE and SDC at a regional scale and to integrate explainable AI into forest carbon modelling. The framework provides a practical, transparent tool for forest managers, policymakers, and carbon monitoring systems, supporting high-resolution, scalable, and interpretable OC assessments.

Climate change undermines forests' health, vitality, and, as a consequence, tree functionality, productivity, and resilience to biotic disturbances. Mountain and sub-alpine forests are particularly susceptible to climate extremes and are showing signs of degradation in Europe. Warmer temperatures, drought, higher frequency and intensity of natural disturbances increasingly alter species distribution and survival, their growing capacity, reproduction, establishment, as well as their potential adaptation to climate change. Real-time monitoring of trees' and stands' responses to such events provides an effective way to better understand and even foresee the adverse side effects of climate change. The use of advanced and innovative monitoring tools and devices is required for ensuring long-term, large-scale, and real-time monitoring of forest dynamics. Here, we present the TreeTalker Italia Network (TTIN), i.e., the first largescale network of tree-proximal sensors (TreeTalkers (c)) at a national scale in Italy. We describe the recent advances, innovations, and potential of such devices for continuous monitoring and research. As a primer, we argue that TTIN will provide effective support to ongoing science and policy efforts for monitoring natural resources' dynamics on a large scale (e.g., forest inventory, climate impacts), including their effects on human well-being.

Species' traits and environmental conditions determine the abundance of tree species across the globe. The extent to which traits of dominant and rare tree species differ remains untested across a broad environmental range, limiting our understanding of how species traits and the environment shape forest functional composition. We use a global dataset of tree composition of >22,000 forest plots and 11 traits of 1663 tree species to ask how locally dominant and rare species differ in their trait values, and how these differences are driven by climatic gradients in temperature and water availability in forest biomes across the globe. We find three consistent trait differences between locally dominant and rare species across all biomes; dominant species are taller, have softer wood and higher loading on the multivariate stem strategy axis (related to narrow tracheids and thick bark). The difference between traits of dominant and rare species is more strongly driven by temperature compared to water availability, as temperature might affect a larger number of traits. Therefore, climate change driven global temperature rise may have a strong effect on trait differences between dominant and rare tree species and may lead to changes in species abundances and therefore strong community reassembly.

In this paper, we present the results of the development of an autonomous mobile robot for forest monitoring and mapping within the AI4FOREST project. This research project funded by the Italian National Recovery and Resilience Plan aims to design and implement an autonomous robotic system capable of navigating into a forest to create a digital twin of the environment and estimate the tree parameters. The proposed system is built on a Scout 2.0 wheeled mobile robot by AgileX that integrates a simultaneous localization and mapping approach for autonomous navigation and 3D reconstruction of the forest. Furthermore, the mobile robot is capable of detecting trees and estimating their diameters from point cloud data and a vision-based deep learning architecture. Experimental results in a wooded area demonstrate the capability of the robot for autonomous forest monitoring and mapping.

Spectral diversity (SD) in reflectance can be used to estimate plant taxonomic diversity (TD) according to the Spectral Variation Hypothesis (SVH). However, contrasting relationships between SD and TD have been reported by different studies. Indeed, multiple factors may affect SD, including spatial and spectral scales, vegetation characteristics and the adopted SD computational method. Here, we tested the SVH over 171 plots within a large and heterogeneous forest area in North-Eastern Italy using Sentinel-2 data, aiming at identifying possible factors affecting the strength and direction of SD-TD relationship. SD was determined using 'biodivMapR' (BD) and 'rasterdiv' (RD) R packages and 38 possible combinations of SD indices, at both alpha (within a community) and beta (among communities) levels, and computational parameters accounting for spatial and spectral scales. Information on vegetation structure was either retrieved from ground-based or LiDAR data. A Random Forest approach was used to disentangle the relationships between SD, TD and vegetation structure, and to identify the best combination of SD computational parameters. At the alpha-level, we found negative relationship between TD and RD SD indices, which was mainly driven by the presence of gaps within the forest canopy. As regards BD, we found that this algorithm reduced background contribution on SD and was able to differentiate major forest types (broadleaves vs conifers), but derived alpha-SD indices were marginally correlated with alpha-TD. At the beta-level, we observed a statistically significant positive correlation between BD SD indices and TD (maximum r = 0.24). Finally, we found stronger correlations and R2 when SD indices were calculated using smaller computation windows and over a larger pixels extraction area. Our findings suggest that vegetation cover and structure play a major role, with respect to inter-species spectral differences, in determining alpha-SD, and that SD might better capture differences in species composition at the landscape-level rather than the richness of individual communities.

AimTo determine the relationships between the functional trait composition of forest communities and environmental gradients across scales and biomes and the role of species relative abundances in these relationships.LocationGlobal.Time periodRecent.Major taxa studiedTrees.MethodsWe integrated species abundance records from worldwide forest inventories and associated functional traits (wood density, specific leaf area and seed mass) to obtain a data set of 99,953 to 149,285 plots (depending on the trait) spanning all forested continents. We computed community-weighted and unweighted means of trait values for each plot and related them to three broad environmental gradients and their interactions (energy availability, precipitation and soil properties) at two scales (global and biomes).ResultsOur models explained up to 60% of the variance in trait distribution. At global scale, the energy gradient had the strongest influence on traits. However, within-biome models revealed different relationships among biomes. Notably, the functional composition of tropical forests was more influenced by precipitation and soil properties than energy availability, whereas temperate forests showed the opposite pattern. Depending on the trait studied, response to gradients was more variable and proportionally weaker in boreal forests. Community unweighted means were better predicted than weighted means for almost all models.Main conclusionsWorldwide, trees require a large amount of energy (following latitude) to produce dense wood and seeds, while leaves with large surface to weight ratios are concentrated in temperate forests. However, patterns of functional composition within-biome differ from global patterns due to biome specificities such as the presence of conifers or unique combinations of climatic and soil properties. We recommend assessing the sensitivity of tree functional traits to environmental changes in their geographic context. Furthermore, at a given site, the distribution of tree functional traits appears to be driven more by species presence than species abundance.

AimEcological and anthropogenic factors shift the abundances of dominant and rare tree species within local forest communities, thus affecting species composition and ecosystem functioning. To inform forest and conservation management it is important to understand the drivers of dominance and rarity in local tree communities. We answer the following research questions: (1) What are the patterns of dominance and rarity in tree communities? (2) Which ecological and anthropogenic factors predict these patterns? And (3) what is the extinction risk of locally dominant and rare tree species?LocationGlobal.Time period1990-2017.Major taxa studiedTrees.MethodsWe used 1.2 million forest plots and quantified local tree dominance as the relative plot basal area of the single most dominant species and local rarity as the percentage of species that contribute together to the least 10% of plot basal area. We mapped global community dominance and rarity using machine learning models and evaluated the ecological and anthropogenic predictors with linear models. Extinction risk, for example threatened status, of geographically widespread dominant and rare species was evaluated.ResultsCommunity dominance and rarity show contrasting latitudinal trends, with boreal forests having high levels of dominance and tropical forests having high levels of rarity. Increasing annual precipitation reduces community dominance, probably because precipitation is related to an increase in tree density and richness. Additionally, stand age is positively related to community dominance, due to stem diameter increase of the most dominant species. Surprisingly, we find that locally dominant and rare species, which are geographically widespread in our data, have an equally high rate of elevated extinction due to declining populations through large-scale land degradation.Main conclusionsBy linking patterns and predictors of community dominance and rarity to extinction risk, our results suggest that also widespread species should be considered in large-scale management and conservation practices.