Accurate and timely detection of incipient forest fire smoke is essential for forest fire prevention and ecosystem protection. However, existing image-based methods are often limited by irregular smoke appearance, complex forest backgrounds, and limited computational capacity on edge devices. This study proposes LCA-YOLO, a lightweight detection framework for edge-based forest fire smoke monitoring. The model incorporates Linear Deformable Convolution (LDConv) to improve adaptability to irregular smoke morphology, Content-Guided Attention Fusion (CGA-Fusion) to enhance feature representation under cluttered backgrounds, and Adaptive Threshold Focal Loss (ATFL) to increase the optimization emphasis on low-confidence hard samples, which may include faint early-stage smoke. Experimental results on a self-built forest fire smoke dataset and multiple public benchmarks demonstrated a favorable balance between detection accuracy and computational efficiency. With 2.39 M parameters, LCA-YOLO achieved an mAP@0.50 of 0.887 in the seed-0 run and an average mAP@0.50 of 0.8844 ± 0.0052 across five independent runs. Evaluation on an NVIDIA Jetson Nano at an input resolution of 480 × 480 pixels yielded an average inference latency of 78.6 ms per image, corresponding to approximately 12.7 FPS and demonstrating the feasibility of continuous edge-side smoke inference under the tested hardware configuration. These results support the computational feasibility of deploying LCA-YOLO on resource-constrained edge hardware.
Drought is a negative factor limiting tree growth and wood productivity of Catalpa bungei, a native species of the Catalpa genus in China that is characterized by high-quality timber and significant economic value. In the present study, stability analysis and comprehensive selection of elite clones of C. bungei were conducted based on gas exchange, wood growth, and hydraulic architecture at two sites with different rainfall levels. Tree growth and wood productivity of C. bungei were largely inhibited by rainfall shortage and the incongruity between rainfall and temperature. Some clones showed prominent wood productivity and fine wood properties at the Heze site, which experienced water deficit and seasonal drought. The high clone repeatability (R > 0.65) of these traits implies great potential for selecting drought-adaptive varieties. Specific clones with high stability in wood productivity and basic wood properties were selected based on BLUP-GGE biplot analysis. Elite clones (2–8, 2–6, 19–01, 1–3, 6–7, 008–1, and 22–03) with favorite wood growth and hydraulic safety were selected based on comprehensive assessment, and these clones are proposed for extension and cultivation in climate regions with limited rainfall. The outstanding growth performance of elite clones was attributed to their larger leaf area and great photosynthesis capacity per plant, rather than to a high photosynthetic rate per unit leaf area. Most of the elite clones exhibited timely adjustments in hydraulic architecture (vessel diameter and vessel density) between the drought season and the rainy season, which contributed to an effective trade-off between wood growth and hydraulic safety.
Under climate change, increased inputs of coarse woody debris (CWD) have heightened the need for reliable estimates of CWD carbon storage to support forest carbon management and accounting. Conventional field-based approaches are often constrained by high labor demands and substantial time costs, while standard remote sensing faces challenges from canopy occlusion, limiting accurate CWD quantification in forests with complex stand structures. Here, we developed a stratified representation framework for identifying spatial patterns of CWD carbon storage and evaluated it in representative subtropical forests of Mount Wuyi, China. Within this framework, operational CWD carbon strata were first delineated from a feature-selected subset and then recognized using spatially available UAV-derived and topographic predictors. The results showed that (1) remote-sensing-only schemes performed poorly for direct continuous estimation in these complex stands, yielding negative cross-validated R2 values (–0.134 to –0.049). By contrast, the feature-selected subset (S8), driven primarily by CWD physical attributes and elevation, best represented the observed gradient in CWD carbon storage (R2 = 0.713). (2) The strata-based recognition achieved moderate overall performance (mean accuracy = 67.80
Due to the exacerbating impacts of global warming, extreme weather events have become increasingly extreme and frequent in the recent decades. The summer drought in 2022 that occurred in the Yangtze River Basin (YRB), China stands as an unprecedented occurrence, significantly threatening the regional ecosystem carbon fixation. However, the effect of such event on vegetation productivity (GPP) as well as the associated feedbacks of diverse vegetation types remain unclear. Therefore, this study used the time-series satellite-based GOSIF GPP data across the YRB to reveal the impact of the extreme 2022 summer drought on vegetation productivity, and explore the vegetation responses with the derived resistance, resilience and elasticity. Our findings showed that the drought severity intensified eastward, with the most severe impacts concentrated in the lower reaches. During the drought period, approximately 71.97
Chromium (Cr) and nickel (Ni) are different from many other elements, having a strong and consistent association in the xylem of Scots pine. On average, one atom of Ni corresponds to 2–3 atoms of Cr. This association cannot be explained by the content of the elements in mobile chemical forms in the soil. This study examined possible explanations of the observed phenomenon. The research materials included soil samples in the research area and at the places of wood sampling in two populations of Scots pine. Chemical analysis was performed with the help of atomic emission spectral analysis. The results show no anomalies in Cr and Ni contents in the soils. Despite an even level of Cr and Ni, their variations among trees are quite wide. The association of Cr and Ni in xylem is reversed with respect to their content in mobile (bioavailable) forms. The association is consistent between the pine populations. Two hypotheses evaluated included: (1) a possible negative correlation between the rate of stem growth and the lignin content, and (2) an ability of Cr and Ni atoms to bond in the xylem. As of yet, no single plausible explanation can fully account for the observation. Disentangling this association will require more penetrating approaches such as electron microscopy and structural analysis.
Forest management typically focuses on assessments of aboveground structure and yield, often overlooking the belowground nutrient reservoirs that sustain long-term ecosystem function. While soil nutrient richness is known to influence forest productivity and resilience, the role of base cation availability in shaping aboveground nutrient dynamics and merchantable wood value remains poorly understood in glaciated hardwood forests. This study investigated how soil nutrient richness, tree size class, and tree genus influence aboveground nutrient concentrations, pools, and merchantable wood volume and value across three managed northern hardwood forests in Vermont and New Hampshire, USA, spanning Poor, Moderate, and Rich soil nutrient richness classes. We measured foliar and wood nutrient concentrations (Ca, Mg, K, P) and estimated biomass and timber value across 45 forest plots. Nutrient-rich soils supported the highest foliar and woody nutrient concentrations, particularly for Ca and P, whereas total aboveground nutrient pools were driven primarily by forest structure rather than soil fertility alone. Acer and Fraxinus dominated aboveground nutrient pools, with Fraxinus having the highest wood concentrations of Ca, Mg, K, and P among all genera. Poletimber trees contained the largest wood nutrient pools due to their high collective basal area. Contrary to expectations, merchantable wood volume did not increase with increasing soil nutrient richness, likely reflecting differences in stand age and composition, indicating that soil fertility assessments alone are not sufficient to predict timber yield or stand value. However, merchantable wood value varied across the richness gradient, with the Moderate Forest consistently yielding the highest simulated returns due to a high abundance of Acer. These findings highlight the importance of soil fertility in sustaining long-term forest productivity, and suggest that standard pre-harvest assessments of species composition and stand structure should be paired with soil fertility surveys to maintain economic value in nutrient-limited northern hardwood forests.
Mongolian pine (Pinus sylvestris var. mongolica), the most important afforestation species in semi-arid sandy lands, is generally planted at high initial densities to ensure survival. As stands develop, increasing competition for limited water and nutrients drives growth decline and mortality, leading to density regulation through self-thinning. However, how initial density shapes long-term self-thinning trajectories remains unclear, limiting effective density management. Using a 42-year long-term dataset covering a broad initial density gradient (1,111–10,000 trees ha−1), we characterized self-thinning trajectories. Results showed that stand density declined, and mortality ranged from 42 to 79
Understanding the response of soil microbial communities to different plantation reforestation strategies is crucial for tree species selection and successful forest restoration. In this study, we compared soil bacterial and fungal communities in topsoil (0–10 cm) and subsoil (10–30 cm) layers among three plantation restoration strategies, including Cunninghamia lanceolata plantation (CL), Liquidambar formosana plantation (LF), and the mixture plantation of L. formosana and Schima superba (L. formosana-S. superba mixture forest, MIX), through high-throughput sequencing in a subtropical forest. Compared with the LF and MIX strategies, the CL restoration strategy showed the greatest recovery of bacterial richness, reaching 94.88
Optimizing the trade-off between wood quality and tree vigor is a central challenge in silviculture, particularly for Pinus koraiensis plantations where weak natural pruning leads to severe knot defects. While artificial pruning is essential, current guidelines rely on static morphological ratios that ignore the complex interplay between crown physiology and environmental variability. To address this, we integrated retrospective knot analysis with a nonlinear seemingly unrelated regression (NSUR) system to reconstruct and model the dynamic trajectories of three hierarchical metrics: total height ( HT , representing vertical growth potential), height to crown base ( HCB , defining the upper limit of knot-free timber), and height to effective crown ( HEC , a physiological threshold for functional vigor). We then employed a nonlinear seemingly unrelated regression (NSUR) system to integrate these parameters with environmental variables. Hierarchical partitioning revealed that crown recession dynamics are driven by distinct, size-mediated mechanisms. Large trees were predominantly governed by ontogeny and site quality, whereas small trees exhibited high sensitivity to direct climate stress. Crucially, medium-sized trees displayed strong competition-climate interactions, where competitive dominance (larger relative diameter) amplified the benefits of spring warming on height growth while buffering the adverse effects of drought on crown recession. The NSUR system captured these interdependent dynamics with high precision ( R^2 : 0.84–0.90; RMSE : 1.10–1.23 m). Translating these ecological insights into practice, we developed a climate-responsive pruning decision framework covering nine site-climate scenarios. By identifying HEC as the critical biological boundary, this tool enables forest managers to transition from fixed schedules to adaptive interventions, ensuring premium timber production under an uncertain future climate.
Soil microbial CO2 fixation plays a pivotal role in the global terrestrial carbon cycle. In forest ecosystems, the changes in the pathways and potential of microbial CO2 fixation as well as their underlying mechanisms along environmental gradients remain unclear, particularly considering the effects of variations in forest microclimates. To address this gap, this study employed metagenomic sequencing and 13C stable isotope labeling to investigate CO2-fixing microorganisms, pathways, potential, and their relationships with environmental factors across elevations from 894 to 2200 m. The results showed that the dominant CO2-fixing microorganisms were Proteobacteria, Actinobacteria and Acidobacteria in this study, while their compositional proportions showing clear disparities, ranging from 30.88 to 48.17
Leaf geometry mediates the formal and functional integration of plant organs, yet how different geometric traits vary along environmental gradients and how they coordinate lamina and petiole traits, remains poorly understood. The study analyzed 1794 individuals spanning 294 woody plant species along a 4000-km latitudinal transect across China. We quantified the distribution and drivers of two key leaf geometric traits: leaf centroid ratio (LCR) and length to width ratio (LWR). In addition, we assessed how the coupling between geometry and lamina–petiole traits varied across spatial scales. We found that both geometric traits showed consistent declines with increasing latitude, primarily shaped by mean annual temperature and precipitation. LCR, a key regulator of biomechanical stability, showed the strongest connectivity in the lamina and petiole trait network at higher latitudes. LWR, reflecting efficiency in resource acquisition and conservation, maintained high and stable trait associations across all regions. In summary, our results suggest leaf geometric traits capture distinct functional axes of the lamina–petiole complex. This multidimensional framework advances our understanding of plant adaptive strategies and trait integration along environmental gradients.
Forest ecosystems represent the largest terrestrial carbon reservoir globally. Accurate quantification of forest biomass is critical for ecosystem service assessment, carbon sink trading, and supporting China’s dual-carbon climate targets. However, the lack of individual tree biomass models for dominant species in the heterogeneous habitats of the Nanling Mountains has hindered regional carbon accounting and sustainable forest management. Here, we employed cluster analysis to classify major tree species into functional types with consistent allometric growth strategies, enabling unified group-level biomass models that improve the accuracy and efficiency of regional stand biomass estimation. Using stem analysis method, we quantified stem, branch, leaf, and root biomass for each sampled tree. The allometric parameters of 13 widely used models were estimated via ordinary least squares regression. The results showed that: (1) 68 major tree species could be classified into 10 functional types. Leaf shape, taxonomic family, maximum potential height, and wood density were identified as the key traits driving functional type differentiation. (2) Biomass models were developed at the group level rather than for individual species. All 13 allometric models exhibited good performance in predicting organ biomass for each group. (3) Model validation confirmed strong performance across all groups, with adjusted R2 values exceeding 0.90, low root-mean-square errors, and negligible systematic bias. These findings confirm the applicability of the developed models for estimating organ biomass across different tree species in Nanling Mountains. These models provide a regional biomass reference for subtropical natural forests, supporting more accurate forest biomass estimation, carbon stock assessment, carbon accounting, and forest management, thereby contributing to China’s dual-carbon goal and the construction of regional ecological security barriers.
Accurate tree-species semantic segmentation from unmanned aerial vehicle (UAV) RGB imagery is essential for stand-level forest inventory, species composition assessment, and ecological management. Existing UAV RGB tree-species segmentation methods generally rely on discriminative image features, but they often struggle to preserve structural consistency under complex canopy conditions, such as shadow occlusion, scale variation, and subtle inter-species differences. As a result, they tend to produce fragmented crowns, noisy class boundaries, and unstable multi-class predictions. To address these issues, we propose HDP-SAM, a prompt-free and parameter-efficient SAM adaptation framework for UAV forest imagery. HDP-SAM leverages SAM’s pretrained mask priors and structural representations to enhance crown-level coherence. We further incorporate a dynamic prototype guidance (DPG) module into the decoder feature stream, where input-conditioned tree-species prototypes are derived from the current decoder features to recalibrate the dense feature stream before final classification, helping improve class assignment consistency and reduce confusion among visually similar species. We further adopt a hierarchical parameter-efficient fine-tuning (H-PEFT) strategy that combines adapters and LoRA across different stages of the SAM encoder to balance segmentation accuracy and parameter efficiency. Experiments on the Huangshan dataset and the Quebec Trees Zone 1 show that HDP-SAM consistently outperforms existing SAM adaptation methods and competitive remote-sensing baselines. It achieves 59.70
Genomic selection (GS) could be used to reduce the long cycle time for tree breeding. This assumes that marker-based predictions are sufficiently accurate without phenotypes. We evaluated GS across two linked generations of Norway spruce (Picea abies (L.) H. Karst), the first consisting of 954 plus-trees (G0) and the second of 956 progeny trees representing 34 full-sib families (G1), using 16 clonal field trials across mid- and southern Sweden. Both generations were measured for height and genotyped using a 50 K SNP chip array. Cross-validations within and across generations were compared. GS efficiency was evaluated using global prediction accuracy and within-family predictive ability, using GBLUP with phenotypes for independent validation. We used computer simulations to emulate the experimental data and repeat the same analysis under different assumptions of effective population size. Additional simulations were performed to investigate the cross-generation GS accuracy in future generations. Simulations assuming a historical effective population size of 1000 or 5000, together with experimental results, indicate that prediction accuracy decreased by 49–76
Tree‑species composition strongly influences forest‑soil properties, a relationship that becomes critical during forest conversion and soil‑fertility considerations. We employed a transect‑based research design to quantify the spatial effects of 83‑year‑old European beech (Fagus sylvatica L.) “Green‑Eye” (GE) groups on site conditions in the Central European Uplands, focusing on moderately sandy, acidic soils. We also conducted ground vegetation inventories (species composition, cover, and Ellenberg indicator values) to infer site quality with respect to nutrient status, soil acidity, light regime, and moisture availability. The GE act as facilitators and generate a dual buffering effect: (1) a microclimatic buffer that moderates temperature extremes and soil moisture, and (2) a pH buffer that reduces soil acidity in the upper Humipedon (the organic‑organomineral surface horizons). Compared with adjacent coniferous stands (CS) of Norway spruce (Picea abies (L.) H. Karst.) and Scots pine (Pinus sylvestris L.), the GE showed a 28
This study analyzed the effects of even-aged forest management (BAU) and continuous cover forestry (CCF) on waterborne phosphorus and nitrogen loads from peatland and mineral soil forests. The impact of minimizing nutrient loads instead of maximizing economic profit was examined in both silvicultural systems, using models for tree growth, survival, and regeneration, as well as for the effect of different forest treatments on nutrient loads. The data for the calculations were a random sample of forest stands, representing two municipalities in North Karelia, eastern Finland. The results showed that the loads of nitrogen and phosphorus from managed commercial forests can be reduced by increasing the use of continuous cover forestry. Significant reductions were possible in both peatland forests and mineral soils. Nutrient loads could also be reduced in even-aged BAU management by decreasing the use of clear-felling, in both mineral soil and peatland forests. Nutrient loads increased with increasing harvest volumes in both silvicultural systems. Significant reductions in nutrient loads can be achieved with a reasonably small loss in the profitability of forest management. The optimal use of continuous cover management reduces nutrient loads without decreasing the profitability of forest management.
In light of Canada’s 2050 carbon neutrality target, Parks Canada is implementing strategies to achieve net-zero emissions, with carbon offset projects in national parks playing a key role. This study examines visitors’ willingness to pay (WTP) for carbon offsets in Banff National Park, employing the extended theory of planned behavior to analyze the factors influencing WTP. Contingent valuation methods are used to estimate the price visitors are willing to pay, and results indicate that behavioral attitudes, subjective norms, and perceived behavioral control significantly influence WTP through behavioral intention. Additionally, time preference, defined as the tendency to prioritize short-term benefits over long-term ones, directly impacts WTP, while risk perception has no significant effect. On average, visitors who were willing to pay contributed 16.13 CAD (12.10 USD), which represents approximately 9
Reliable assessment of ecosystem quality is essential for managing plantation systems, yet uncertainty persists about how indicator selection and scoring methods affect ecological interpretation. Here, we evaluated soil and vegetation quality across contrasting plantation types using total data sets (TDS) and reduced minimum data sets (MDS), combined with linear and non-linear scoring methods. Across seasons, soils under Ulmus pumila consistently exhibited higher nutrient availability and carbon stocks than those under Pinus tabuliformis, reflecting contrasting plant functional strategies. Despite substantial data reduction, MDS-based indices closely matched TDS-derived values (R2 = 0.65–0.86), indicating that a limited set of indicators captured the dominant variation in ecosystem quality. Linear scoring methods further showed greater sensitivity in resolving species and seasonal differences, with stronger correlations than non-linear approaches, such as SQI: R = 0.81 vs 0.65 in Pinus; R = 0.93 vs 0.91 in Ulmus. In contrast, non-linear models tended to dampen intermediate variation and reduce discriminatory power. These findings suggest that simplified assessment frameworks can effectively characterize ecosystem quality when variation follows structured, trait-related ecological gradients. This study provides an analytical framework that can be applied and evaluated across broader forest systems to support ecosystem assessment and management.
Although recognized as Asia’s “Water Tower”, our understanding of past natural and recent anthropogenic enforced climate change is limited for much of the Himalayas. The main contributing factors for that are sparse and short meteorological observations, seasonally restricted and often imprecise proxy archives, and spatially heterogeneous and temporally unstable climate dynamics. Here we present a network of five maximum latewood density (MXD) chronologies from 135 living fir (Abies spectabilis) trees from sites near the upper treeline between 3220 and 3750 m a.s.l. in Nepal. Individual series were processed to preserve interannual, decadal, and centennial-scale variability. The resulting composite chronology was calibrated against April–September (AMJJAS) temperatures over the period 1951–2022 CE (rs = 0.64; p < 0.001), enabling the first MXD network-based temperature reconstruction for the Central Himalayas. Despite the relatively short reconstruction coverage (1775–2022 CE), it by far surpasses any regional instrumental record and explains more than 40
