National Forest Inventory (NFI) data are typically limited to sparse networks of sample locations due to cost constraints. While design-based estimators provide reliable forest parameter estimates for large areas, there is increasing interest in model-based small area estimation (SAE) methods to improve precision for smaller spatial, temporal, or biophysical domains. SAE methods can be broadly categorized into area- and unit-level models, with unit-level models offering greater flexibility, making them the focus of this study. Ensuring valid inference requires satisfying model distributional assumptions, which is particularly challenging for NFI variables that exhibit positive support and zero-inflation, such as forest biomass, carbon, and volume. Here, we evaluate nine candidate estimators, including two-stage unit-level hierarchical Bayesian models, single-stage Bayesian models, and two-stage frequentist models, for estimating forest biomass at the county level in Nevada and Washington, United States. Estimator performance is assessed using repeated sampling from simulated populations and unit-level cross-validation with FIA data. Results show that small area estimators incorporating a two-stage approach to account for zero-inflation, county-specific random intercepts and residual variances, and spatial random effects yield the most accurate and well-calibrated county-level estimates, with spatial effects providing the greatest benefits when spatial autocorrelation is present in the underlying population.

A major challenge for accurate characterization of tree architecture with LiDAR is distinguishing leaves from woody parts. Leaves may block scans of woody parts, or be confused with them, when quantitative structure models (QSMs) are applied. Leaf–wood separation algorithms (LWSAs) can be used to simulate ‘leaf-off’ conditions before applying QSMs but need further study. We analyzed 110 deciduous trees of 25 species, scanned in leaf-on and leaf-off conditions, with a Faro Focus3D X330, to examine the influence of leaves-on errors in tree woody volume estimation from QSMs. TreeQSM (v2.4.1) was applied, with and without applying the GBSeparation LWSA, and Real Twig (rTwig) was used to create leaf-off woody tree ‘skeletons’ for reference. The results showed greatly overestimated volume with TreeQSM, because it constructed additional branches out of leaves and the noise generated by them, adding large amounts of non-existent branch volume. GBSeparation allowed for accurate total woody volume estimation, but via compensating errors; overallocation of volume to stems and lower-order branches and omission of smaller, higher-order ones. We tested a novel field-calibrated allometric model for predicting leaf-off tree volume from leaf-on tree volume, which showed promise as an accurate, time-efficient alternative pathway for processing leaf-on TLS data.

Key message Terrestrial laser scanning data of trees combined with models of heartwood content proportion of woody disks can provide precise characterization of total aboveground tree sapwood and heartwood volume. Abstract Quantifying sapwood and heartwood content of trees is challenging. Previous studies have primarily characterized main stem wood composition, while branches have rarely been studied. Terrestrial Laser Scanning (TLS) can provide precise representations of the entire above-ground tree structure, non-destructively, to help estimate total tree sapwood and heartwood volume. In this study, we used TLS to scan above-ground portions of twenty-four open-grown, urban Gleditsia triacanthos trees on Michigan State University campus. TLS data were used to generate quantitative structure models that provided comprehensive characterizations of the total tree woody surface area (WSA) and volume. A subsample of trees was harvested (after scanning) and main stem and branch woody disks were collected to build models of heartwood content proportion. Models were applied to measurements from TLS to quantify complete heartwood and sapwood volume of each tree, including main stem and branches. From the base to the top of the trees, the largest portion of stem vertical cumulative volume was heartwood, whereas vertical cumulative volume of branches showed the opposite pattern. Absolute heartwood volume declined monotonically toward zero from stem base to stem top, while absolute sapwood volume declined sharply from stem base up to near the crown base and then remained relatively constant within crown. We also found that tree WSA increased with sapwood volume for both branches and main stem. This study developed a novel, general method for quantifying total aboveground sapwood and heartwood volume of trees and provided new insights into urban tree growth and structure.

Quantitative structure models (QSM) provide topological models of trees as geometric primitives created from LiDAR (light detection and ranging)-derived tree point clouds. QSMs allow for efficient estimation of tree metrics traditionally difficult or impossible to measure directly, such as total volume and surface area. However, QSMs are limited by their input point clouds. Technological limitations of current laser scanning sensors severely overestimate the diameters of small branches and twigs, inflating their size in the QSM. Here, we present an R package, rTwig, to correct branch volume overestimation in QSMs with the Real Twig method. Real Twig modifies QSMs, resulting in realistic branch tapering constrained by the species' actual twig measurement and precise and accurate tree metrics. The rTwig package contains fast and efficient tools for QSM and point cloud manipulation and visualization within the R programming environment, and a novel database of twig diameter measurements for a wide range of tree genera and species. We expect rTwig will improve confidence in using LiDAR for forest inventory, monitoring, above ground volume and biomass estimation, and carbon accounting.

The accurate estimation of tree volume and biomass is necessary for forest ecosystems management. However, traditional estimation methods are expensive, require a large amount of labor and materials, and may involve destructive sampling. In this study, a terrestrial laser scanner (TLS) and free software were used to estimate the volume and biomass of the stems of individual trees from two coniferous species, Abies religiosa (Kunth) Schltdl. & Cham and Pinus pseudostrobus Lindl, in the Monarch Butterfly Biosphere Reserve (MBBR), Michoacan, Mexico. TLS is an alternative to traditional measurement methods, which allows tree measurements to be extracted from a point cloud, opening up new opportunities to characterize the volume and biomass of standing trees. The simple linear regression analysis comparing stem volume and biomass estimates from different methods shows that the Vol_TLS and Vol_TModel relationship provides a better fit (R2 = 0.97, RMSE = 0.351 m3) compared to the Vol_TLS and Vol_OModel relationship (R2 = 0.93, RMSE = 0.537 m3). However, comparisons between measurements with predictive models (from destructive methods) and TLS (a non-destructive method) did not show significant differences. The results show that this increasingly accessible technology can be used to adequately estimate forest biomass and volume in a non-destructive manner, which is particularly important in places such as the MMBR.

The critical role of forest carbon modeling in climate change mitigation and adaptation has brought it to the forefront of natural climate solutions (NCS) discussions. To facilitate scientific inquiry related to forest carbon and its climate change mitigation potential, we synthesize current challenges and recommend strategic advancements for quantifying and projecting forest carbon dynamics. A national collaborative effort, engaging stakeholders from academia, industry, and policy sectors, has identified key six areas for scientific development, including tree growth, mortality, and regeneration models, genetics and silviculture, forest floor and belowground carbon, natural disturbances, carbon reporting, and carbon tools and applications. We provide actionable recommendations, such as creating nationally consistent frameworks for forest dynamics models, improving model-data integration for disturbance effects, establishing best practices for uncertainty in carbon reporting, and expanding online tools for spatially explicit carbon estimates. Some of these recommendations are achievable in the short term (3–5 years) and others require longer-term investments (10 + years). Emphasizing a multi-phase strategy that builds on established modeling frameworks such as the Forest Vegetation Simulator (FVS), we articulate how these priorities can support policy and management decisions. Integrating these efforts can advance collective knowledge and enhance our forecasting capabilities of forest carbon in response to climate, disturbances, industrial needs, and management practices, thereby aiding natural climate objectives and the broader ecosystem services provided by forests.

A mature forest remnant in the Amazon was sampled with 22 0.4-ha clusters, which means a sampling effort of 8.8 ha. From this complete sample, nine subdatasets reduced in cluster size and other nine reduced in sample size were created, totaling thus 18 subdatasets. The aim of this study was to dissociate the effects of sample size and cluster size on Hill’s numbers. A cluster consisted of a sample unit composed by four crosswise sub-units of 1,000 m² (20m×50m) each. The 18 subdatasets were aggregated into nine pairs equivalent in sampled area (SA), which decreased from 8.8 ha (complete sample) to successive reductions from 4.4 ha (first pair of subdataset) to 0.8 ha (nineth pair). Hill’s numbers were calculated for every subdataset and then sample-based rarefaction curves were constructed to generate diversity profiles. As a result, species richness was abruptly underestimated in approx. 40% for the first reduction in sample size (SA = 4.4 ha). Significant underestimation of species richness occurred when the SA is below 2.4 ha, and accuracy of the other diversity profiles was severely compromised below 1.6 ha. We concluded that species richness is more sensitive to reductions in SA than Shannon diversity and Pielou’s evenness indices. For a same sampling effort, the choice in installing more small units demonstrated to better capturate the diversity profiles than installing less large units, especially when estimating species richness. Although the patterns of under or overestimates of the other diversity profiles were not exactly clear, the accuracy is substantially hampered when the SA is less than 1.6 ha.

ABSTRACT This paper reviews important processes that drive deforestation and its control in the Brazilian Amazon. Governmental programs decreased the rate of deforestation in the Amazon by 70% from 2004 to 2015. This large reduction was the result of the Action Plan to Prevent and Control Deforestation in Legal Amazonia (PPCDAm) - a task force having the ‘Arc of Deforestation’ as target. During the PPCDAm’s course, the creation of protected areas (PAs), punishment for illegal deforestation, and a soy moratorium were among the most important measures to reduce deforestation rates. Brazil’s 2020 end goal, which was not reached, was to reduce the rate to 80% in relation to the 1996-2005 average. The current goal is to have no (0%) illegal deforestation through 2030. Our review shows both beneficial environmental policies that helped to reduce deforestation, e.g. soy and cattle moratoriums and creation of PAs, as well as threats to Brazil’s Amazon Forest which came from anti-environmental policies between 2014-2022. Considering the main drivers of deforestation so far, we suggest that Brazil can reach the 2030 goal of zero illegal deforestation through (i) the creation and inspection of PAs to avoid illegal logging, (ii) maintenance and strengthening the soy moratorium, (iii) an improved law enforcement related to illegal deforestation, and (iv) a stronger forest concession system.

Despite substantial investments to curb tropical deforestation, effective conservation incentives for Indigenous peoples and local communities is not well-defined and generally under-researched. This study assessed an incentive mechanism in Peru for Indigenous communities that protect enrolled forests to explore whether the stated program goals are actualized in programmatic elements like the requirements, monitoring, and assessment of prioritized outcomes. The research team worked with Indigenous partners to develop key questions regarding how the mechanism could better support their values of conservation and development. Data were sourced from interviews with implementation experts and participants in eight Indigenous communities, a review of programmatic documents, and an assessment of nationally aggregated community data. The results revealed challenges in program capacity, a lack of cultural awareness, and a reliance on capitalistic economic indicators that exclude other aspects of well-being important for Indigenous peoples. We find that the program’s success indicators do not adequately align with conservation or social realities on the ground and that enhanced indicators are needed to ensure success and avoid negative unintended consequences. We demonstrate that enhancing the assessment of governance, economics, engagement, and social inclusion can improve the design, implementation, and monitoring in this and similar programming. We conclude with generalizable recommendations for establishing requirements and monitoring in existing and future conservation incentive programs that target Indigenous communities.

Leaf-wood separation plays an important role in estimating aboveground biomass (AGB) of trees from terrestrial laser scanning (TLS) point clouds. Yet, leaf-wood separation studies have predominantly focused on reporting the accuracy of leaf and wood point separation. Assessments of the impact of these algorithms on the subsequent AGB estimations, based on commonly used quantitative structure models (QSMs), have been limited. Therefore, in this study, we quantified the impact of 11 published leaf-wood separation algorithms on QSM-based tree AGB estimation using an independent benchmarking dataset. The benchmarking dataset consists of AGB measured for 20 destructively harvested trees from a mixed temperate forest in Harvard Forest and AGB estimated from QSMs built on manually segmented tree point clouds of 856 broadleaved trees in Wytham Woods under leaf-off conditions. These benchmarking AGB values were compared to the AGB estimated from QSMs built on the leaf- removed point clouds resulting from the different separation algorithms performed on the leaf-on tree point clouds of the same trees. The results of the study indicated that for most of the algorithms, the leaf-removed AGB estimates for both coniferous and broadleaved trees underestimated the AGB (conifers:-17 % to-3 %, broadleaf:-14 % to-2 %) compared to the destructively measured AGB in Harvard Forest. In Wytham Woods, leaf-removed AGB estimates from all separation algorithms consistently underestimated the AGB (-46 % to-24 %) compared to the AGB from the leaf-off point clouds. Most leaf-wood separation algorithms performed better on broadleaved trees than on coniferous trees. Moreover, significant differences were observed among different algorithms in estimating AGB for trees of the same forest type. For coniferous trees, the relative difference (RD) of leaf-removed AGB estimates from QSMs and separation algorithms ranged from-27 % to 16 %, among which the best performing algorithms demonstrated similar optimal performance, with small RD values of approximately-3 % to 2 %. For broadleaved trees, the leaf-removed AGB estimates from QSMs and eight separation algorithms, as well as leaf-off point cloud estimates (approximately at 10 %), were closely in agreement with the harvested benchmark values, among which the best performing algorithms had a RD value approximately within f2 %. Additionally, most separation algorithms could lead to better estimates of trunk biomass than branch biomass, whereas the estimation for branch biomass consistently exhibited varying degrees of underestimation. These findings provide a timely reference for utilizing leaf-wood separation algorithms for QSM-based AGB estimation.

Vanilla planifolia is native to the Mexican tropics. Despite its worldwide economic importance as a source of vanilla for flavoring and other uses, almost all vanilla is produced by expensive hand-pollination, and minimal documentation exists for its natural pollination and floral visitors. There is a claim that vanilla is pollinated by Melipona stingless bees, but vanilla is more likely pollinated by orchid bees. Natural pollination has not been tested in the Yucatán region of Mexico, where both vanilla and potential native bee pollinators are endemic. We document for the first time the flowering process, nectar production and natural pollination of V. planiflora, using bagged flower experiments in a commercial planting. We also assessed the frequency and visitation rates of stingless bees and orchid bees on flowers. Our results showed low natural pollination rates of V. planifolia (~ 5%). Only small stingless bees (Trigona fulviventris and Nannotrigona perilampoides) were seen on flowers, but no legitimate visits were witnessed. We verified that there were abundant Euglossa and fewer Eulaema male orchid bees around the vanilla plants, but neither visited the flowers. The introduction of a colony of the stingless bee Melipona beecheii and the application of chemical lures to attract orchid bees failed to induce floral visitations. Melipona beecheii, and male orchid bees of Euglossa viridissima and E. dilemma may not be natural pollinators of vanilla, due to lack of attraction to flowers. It seems that the lack of nectar in V. planifolia flowers reduces the spectrum of potential pollinators. In addition, there may be a mismatch between the attractiveness of vanilla floral fragrances to the species of orchid bees registered in the studied area. Chemical studies with controlled experiments in different regions would be important to further elucidate the potential pollinators of vanilla in southern Mexico.

Tree bark and wood density are highly variable and weakly positively correlated, with species having more or less dense bark than wood to adapt to different environmental stressors. Tree bark is a complex, multifunctional structure and bark density varies widely across species. While wood density is recognized as a fundamental indicator of the functional ecology of trees, bark density has received much less attention as a key functional trait. Theoretically, bark and wood density should co-vary to some degree, but comprehensive examinations of this covariation are scarce. How do key functional traits of individual trees and species relate to bark and wood density variation/covariation? How does a tree’s life history and evolved tolerance to environmental stress shape variation/covariation in bark and wood density? This study draws from published literature and a large database of individual tree measurements of trees of diverse species and growing conditions, from forest ecosystems across the United States and Canada, to understand covariation between bark and wood density and its relationship to life-history traits and evolved tolerances to environmental stressors. The results of this study show a high tree-to-tree variation in both bark density and wood density, with inherited differences in tissue formation constraining the range of bark and wood densities. All analyses show that bark density was weakly, positively correlated with wood density. Mixed effects modeling showed a strong phylogenetic signal in variation in bark and wood density that was partially explained by the need for species to produce more or less -dense bark and wood to adapt to different environmental stressors (tolerance of drought, shade, frost, waterlogging and fire were all examined), with clearly different relationships for angiosperms versus gymnosperms.

Terrestrial laser scanning data can be converted to reliable woody aboveground biomass estimates, but estimation quality is influenced by growing environment, leaf condition, and variation in tree density affecting volume to mass conversion. Both rural and urban forests play an important role in terrestrial carbon cycling. Forest carbon stocks are typically estimated from models predicting the aboveground biomass (AGB) of trees. However, such models are often limited by insufficient data on tree mass, which generally requires felling and weighing parts of trees. In this study, thirty-one trees of both deciduous and evergreen species were destructively sampled in rural and urban forest conditions. Prior to felling, terrestrial laser scanning (TLS) data were used to estimate tree biomass based on volume estimates from quantitative structure models, combined with tree basic density estimates from disks sampled from stems and branches after scanning and felling trees, but also in combination with published basic density values. Reference woody AGB, main stem, and branch biomass were computed from destructive sampling. Trees were scanned in leaf-off conditions, except evergreen and some deciduous trees, to assess effects of a leaf-separation algorithm on TLS-based woody biomass estimates. We found strong agreement between TLS-based and reference woody AGB, main stem, and branch biomass values, using both measured and published basic densities to convert TLS-based volume to biomass, but use of published densities reduced accuracy. Correlations between TLS-based and reference branch biomass were stronger for urban trees, while correlations with stem mass were stronger for rural trees. TLS-based biomass estimates from leaf-off and leaf-removed point clouds strongly agreed with reference biomass data, showing the utility of the leaf-removal algorithm for enhancing AGB estimation.

The Forest Inventory and Analysis (FIA) Program of the U.S. Department of Agriculture, Forest Service conducts the national forest inventory of the United States.Although FIA assembles a myriad of forest resource information, many analyses rely on the fundamental attributes of tree volume, biomass, and carbon content.Due to the chronological development of the FIA Program, numerous models and methods are currently used across the country, contingent upon the tree species and geographic location.Thus, an effort to develop nationally consistent methods for prediction of tree volume, biomass, and carbon content was undertaken.A key component of this study was amassing existing data in conjunction with collection of new data to fill information gaps related to tree size and species frequency and spatial distributions.These data were used in a modeling framework that provides compatible predictions of tree volume, biomass, and carbon content across the entire United States.National-scale comparisons to currently used methods show that only a small increase in volume occurs, but substantial increases in biomass and carbon are realized due to relatively large increases in predicted tree top/limbs biomass and carbon.Changes in tree carbon were also affected by use of newly developed species carbon fractions instead of the current constant conversion factor of 0.5.Examples of the calculations required to predict tree volume, biomass, and carbon content for commonly encountered tree conditions provide step-by-step implementation details.

Quantitative Structure Models (QSMs) are fit to tree point clouds to represent the topology of trees as a network of cylinders. QSMs allow for the calculation of metrics difficult to measure without destructive sampling, including total tree volume. Current limitations in terrestrial laser scanning technology make small branches difficult to accurately resolve, causing overestimation of small branch volume in QSMs, which can translate into overestimating tree biomass. We present a new method called Real Twig to correct overestimated small branch and twig cylinders in QSMs. Real Twig differs from current methods by using twig diameters measured directly from corresponding tree species to model a unique taper for every path in the QSM, using the QSM's inherent branching topology, but without relying on predefined mathematical or allometric relationships. To test Real Twig, we generated QSMs for different sets of trees that had detailed dry mass and density measurements obtained via felling after scanning. QSM-based biomass estimates were obtained by multiplying the tree's QSM-based volume estimate by the tree's specific basic density value. We trained our method with high-quality data consisting of five northern red oak (Quercus rubra L.) and five red maple (Acer rubrum L.) trees, using two different versions of TreeQSM, a widely used algorithm for generating QSMs. We further tested our method on three publicly available datasets, including managed forests and large tropical trees, collected with both phase-shift or time-of-flight sensors. QSMs corrected with our Real Twig method showed a very large improvement in tree biomass estimation, with a relative mean error of -1.2%, a relative root mean square error of 10.5%, and a concordance correlation coefficient of 0.999, compared to a relative mean error 76.8%, a relative root mean square error of 48.7%, and a concordance correlation coefficient of 0.982, when using the standard outputs of TreeQSM.

Climate change is presenting a global challenge to society and ecosystems. This is changing long-standing methods to determine the values of forests to include their role in climate mitigation and adaptation, alongside traditional forest products and services. Forests have become increasingly important in climate change dialogues, beyond international climate negotiations, because of their framing as a Natural Climate Solution (NCS) or Nature-Based Solution (NBS). In turn, the term “Climate-Smart Forestry” (CSF) has recently entered the vernacular in myriad disciplines and decision-making circles espousing the linkage between forests and climate. This new emphasis on climate change in forestry has a wide range of interpretations and applications. This review finds that CSF remains loosely defined and inconsistently applied. Adding further confusion, it remains unclear how existing guidance on sustainable forest management (SFM) is relevant or might be enhanced to include CSF principles, including those that strive for demonstrable carbon benefits in terms of sequestration and storage. To contribute to a useful and shared understanding of CSF, this paper (1) assesses current definitions and framing of CSF, (2) explores CSF gaps and potential risks, (3) presents a new definition of CSF to expand and clarify CSF, and (4) explores sources of CSF evidence.

Major climate negotiations prominently feature forest protection as one of the most important mechanisms to reduce emissions from the land sector, specifically by the release of greenhouse gases from deforestation and degradation of forested landscapes. However, despite an estimated USD 12 billion pledged to address tropical deforestation, best practices to distribute incentives for conservation activities by Indigenous peoples and local communities (IPLCs) are not well-defined and are generally under-researched. Peru holds the second largest share of Amazon Forest – locally and globally important for biodiversity, climate regulation, and carbon storage. Communal landholders make decisions over more than 65% of this ecosystem. As part of an ambitious plan to conserve 79% of its tropical forest (54 out of 68 million hectares), Peru’s Ministry of the Environment (MINAM) has pursued a mechanism called Conditional Direct Transfer (TDC, in Spanish), to stop deforestation that uses incentives to shape behavior and decision-making.While some communities conserve forests successfully under the rubric of the program, others decline to participate or are suspended for non-compliance. Considering the persistent pressure anticipated on Amazonian forests, we ask: How can TDCs be improved to achieve long-term conservation and development objectives? In this paper, we share these results to argue that a systematic assessment of the following categories---Governance, Economics, Participation, and Social Impacts—can effectively provide a strong foundation for design, implementation, and subsequent monitoring of incentive programs around the world. The conclusion includes an overview of generalizable recommendations for Program Implementers in consideration of responsibility and best practices.

Measuring and modelling the shape of tree stems is a fundamental component of forest inventory systems for both commercial and biological purposes. The change in diameter of the stem along its length (a.k.a. 'taper') is one of the most important and widely used means of predicting tree stem volume. Until recently, the options for obtaining accurate estimates of stem taper and developing stem taper models have been limited to measurements of felled trees or the use of optical dendrometers on standing live trees. Here, we tested both a tripod -mounted terrestrial laser scanner (TLS; a Focus 3D 120 of FARO Technologies, Inc., Lake Mary, FL, USA), and a mobile laser scanner (MLS; the ZEB1 of the GeoSLAM Ltd, Nottingham, UK) to measure tree diameters at various heights along the stem of 20 destructively harvested broadleaf and needleleaf species using the outer hull modelling method, for the purpose of developing individual -tree and species -specific taper models. Laser scanner specifications were a major factor determining stem taper measurement accuracy. The longer-range, low beam divergence TLS could estimate stem diameter to an average of 15.7 m above ground (about 79 per cent of the canopy height), while the shorter-range high beam divergence MLS could estimate an average of 11.5 m above ground (about 45 per cent of the canopy height). Stem taper error increased with respect to height above ground, with the TLS providing more consistent and reliable diameter measurements (root mean square error (RMSE) = 1.93 cm; 9.57 per cent) compared with the MLS (RMSE = 2.59 cm; 12.84 per cent), but both methods were nearly unbiased. We attribute similar to 60 per cent of the uncertainty in stem measurements to laser beam diameter and point density, showing positive and negative correlations, respectively. MLS was unable to converge on the two tested taper models but was found to be an efficient means of easily sampling diameters at breast height (DBH) and reconstructing stem maps in simple forest stands with trees greater than similar to 10 cm DBH. TLS provided precision stem diameter measurements that allowed for the creation of similar taper models for three out of the four study species. Future work should focus on evaluating MLS systems with improved specifications (e.g. beam divergence and range), since these instruments will likely lead to dramatic improvements in reliable estimates of forest inventory parameters, in line with the current TLS technology.

A healthy urban forest is important to improve both human health and overall environmental quality. There is currently a lack in technology that has the ability to evaluate living trees in their natural environment without invasive destructive sampling. This work presents a cost-effective, real-time microwave tomography system for practical forestry applications. The scattering measurement system is designed using wide band microstrip monopole antennas (1--5 GHz) and coupled to a switching matrix and a controlling software for automated real-time data collection. A time reversal signal processing algorithm is developed for performing imaging, based on the measurements. The imaging system is initially validated by imaging simple cylindrical targets and finally utilized for imaging defects in different tree trunk samples. Preliminary experimental results demonstrate the practicality, novelty and benefits of this approach for forestry imaging applications.

Estimating tree leaf biomass can be challenging in applications where predictions for multiple tree species is required. This is especially evident where there is limited or no data available for some of the species of interest. Here we use an extensive national database of observations (61 species, 3628 trees) and formulate models of varying complexity, ranging from a simple model with diameter at breast height (DBH) as the only predictor to more complex models with up to 8 predictors (DBH, leaf longevity, live crown ratio, wood specific gravity, shade tolerance, mean annual temperature, and mean annual precipitation), to estimate tree leaf biomass for any species across the continental United States. The most complex with all eight predictors was the best and explained 74%-86% of the variation in leaf mass. Consideration was given to the difficulty of measuring all of these predictor variables for model application, but many are easily obtained or already widely collected. Because most of the model variables are independent of species and key species-level variables are available from published values, our results show that leaf biomass can be estimated for new species not included in the data used to fit the model. The latter assertion was evaluated using a novel "leave-one-species-out" cross-validation approach, which showed that our chosen model performs similarly for species used to calibrate the model, as well as those not used to develop it. The models exhibited a strong bias toward overestimation for a relatively small subset of the trees. Despite these limitations, the models presented here can provide leaf biomass estimates for multiple species over large spatial scales and can be applied to new species or species with limited leaf biomass data available.