Information Entropy Measures for Stand Structural Diversity:Joint Entropy

Structural diversity is the key attribute of a stand. A set of biodiversity measures in ecology was introduced in forest management for describing stand structure, of which Shannon information entropy (Shannon index) has been the most widely used measure of species diversity. It is generally thought that tree size diversity could serve as a good proxy for height diversity. However, tree size diversity and height diversity for stand structure is not completely consistent. Stand diameter cannot reflect height information completely. Either tree size diversity or height diversity is one-dimensional information entropy measure. This paper discussed the method of multiple-dimensional information entropy measure with the concept of joint entropy. It is suggested that joint entropy is a good measure for describing overall stand structural diversity.

Impact of cattle density on the structure and natural regeneration of a turkey oak stand on an agrosilvopastoral farm in central Italy

On an agrosilvopastoral farm in central Italy where Maremmana cattle graze in Turkey oak forests,we evaluated the impact of different livestock densities on stand structure,tree diversity and natural regeneration in four types of grazed areas based on the grazing regime adopted:calf-grazed,high-intensity-grazed,low-intensity-grazed,ungrazed control.For each area,we set up three permanent circular plots(radius of 15 m)to survey the structural and dasometric characteristics of the overstorey,understorey,and regeneration layer.The results showed that grazing negatively affected the complexity of the forest structure and its potential to regenerate and maintain a high level of biodiversity.The differences in stand structure observed between the grazing areas were closely related to livestock density.The most sensitive components of the system were the understorey and the regeneration layers.Contrarily,the current grazing management did not affect the dominant tree structure or its composition.Our findings identified medium-term monitoring and regeneration management as the two significant aspects to consider when assessing sustainable livestock.New forests can be established by excluding graz-ing for about 20–25 years.

Plant functional trait diversity and structural diversity co-underpin ecosystem multifunctionality in subtropical forests

Tree species diversity is assumed to be an important component in managing forest ecosystems because of effects on multiple functions or ecosystem multifunctionality.However,the importance of tree diversity in determining multifunctionality in structurally complex subtropical forests relative to other regulators(e.g.,soil microbial diversity,stand structure,and environmental conditions)remains uncertain.In this study,effects of aboveground(species richness and functional and structural diversity)and belowground(bacterial and fungal diversity)biodiversity,functional composition(community-weighted means of species traits),stand structure(diameter at breast height and stand density),and soil factors(pH and bulk density)on multifunctionality(including biomass production,carbon stock,and nutrient cycling)were examined along a tree diversity gradient in subtropical forests.The community-weighted mean of tree maximum height was the best predictor of ecosystem multifunctionality.Functional diversity explained a higher proportion of the variation in multifunctionality than that of species richness and fungal diversity.Stand structure-played an important role in modulating the effects of tree diversity on multifunctionality.The work highlights that species composition and maximizing forest structural complexity are effective strategies to increase forest multifunctionality while also conserving biodiversity in the management of multifunctional forests under global environmental changes.

Tree species identity and interaction determine vertical forest structure in young planted forests measured by terrestrial laser scanning

Vertical forest structure is closely linked to multiple ecosystem characteristics,such as biodiversity,habitat,and productivity.Mixing tree species in planted forests has the potential to create diverse vertical forest structures due to the different physiological and morphological traits of the composing tree species.However,the relative importance of species richness,species identity and species interactions for the variation in vertical forest structure remains unclear,mainly because traditional forest inventories do not observe vertical stand structure in detail.Terrestrial laser scanning(TLS),however,allows to study vertical forest structure in an unprecedented way.Therefore,we used TLS single scan data from 126 plots across three experimental planted forests of a largescale tree diversity experiment in Belgium to study the drivers of vertical forest structure.These plots were 9–11years old young pure and mixed forests,characterized by four levels of tree species richness ranging from monocultures to four-species mixtures,across twenty composition levels.We generated vertical plant profiles from the TLS data and derived six stand structural variables.Linear mixed models were used to test the effect of species richness on structural variables.Employing a hierarchical diversity interaction modelling framework,we further assessed species identity effect and various species interaction effects on the six stand structural variables.Our results showed that species richness did not significantly influence most of the stand structure variables,except for canopy height and foliage height diversity.Species identity on the other hand exhibited a significant impact on vertical forest structure across all sites.Species interaction effects were observed to be site-dependent due to varying site conditions and species pools,and rapidly growing tree species tend to dominate these interactions.Overall,our results highlighted the importance of considering both species identity and interaction effects in choosing suitable species combinations for forest management practices aimed at enhancing vertical forest structure.

Optimization models of stand structure and selective cutting cycle for large diameter trees of broadleaved forest in Changbai Mountain

The optimum models of harvesting yield and net profits of large diameter trees for broadleaved forest were developed, of which include matrix growth sub-model, harvesting cost and wood price sub-models, based on the data from Hongshi Forestry Bureau, in Changbai Mountain region, Jilin Province, China. The data were measured in 232 permanent sample plots. With the data of permanent sample plots, the parameters of transition probability and ingrowth models were estimated, and some models were compared and partly modified. During the simulation of stand structure, four factors such as largest diameter residual tree （LDT）, the ratio of the number of trees in a given diameter class to those in the next larger diameter class （q）, residual basal area （RBA） and selective cutting cycle （C） were considered. The simulation results showed that the optimum stand structure parameters for large diameter trees are as follows： q is 1.2, LDT is 46cm, RBA is larger than 26 m^2 and selective cutting cycle time （C） is between 10 and 20 years.

What diameter?What height?Influence of measures of average tree size on area-based allometric volume relationships

Volume is an important attribute used in many forest management decisions.Data from 83 fixed-area plots located in central New Brunswick,Canada,are used to examine how different measures of stand-level diameter and height influence volume prediction using a stand-level variant of Honer's(1967)volume equation.When density was included in the models(Volume=f(Diameter,Height,Density))choice of diameter measure was more important than choice of height measure.When density was not included(Volume=f(Diameter,Height)),the opposite was true.For models with density included,moment-based estimators of stand diameter and height performed better than all other measures.For models without density,largest tree estimators of stand diameter and height performed better than other measures.The overall best equation used quadratic mean diameter,Lorey's height,and density(root mean square error=5.26 m^3·ha^(-1);1.9%relative error).The best equation without density used mean diameter of the largest trees needed to calculate a stand density index of 400 and the mean height of the tallest 400 trees per ha(root mean square error=32.08 m^(3)·ha^(-1);11.8%relative error).The results of this study have some important implications for height subsampling and LiDAR-derived forest inventory analyses.

Study on Stand Structure Visualization Based on GDI～+

As the premise and foundation conditions of stand growth and management visualization,stand structure visualization is a key part of stand visualization.Considering the particularity of stand structure data,we filter and standardize the necessary factors.To intuitively simulate the spatial and attributive structure of the stand,our objective was to transfer the field coordinates into screen coordinates and to render two-dimensional abridged general view of every single tree one by one.This essay takes GDI ＋ technology as basis and uses C# in.NET Framework to generate two-dimensional spatial structure diagram which can show the stand density,crown density,spatial distribution pattern and so on.Meanwhile,it generates the histogram to reflect the distribution of stand structure attributes and finally achieves the goal of the visualization of stand structure.

Stand structure and height-diameter relationship of a degraded Populus euphratica forest in the lower reaches of the Tarim River, Northwest China

Understanding stand structure and height-diameter relationship of trees provides very useful information to establish appropriate countermeasures for sustainable management of endangered forests. Populus euphratica, a dominant tree species along the Tarim River watershed, plays an irreplaceable role in the sustainable development of regional ecology, economy and society. However, as the result of climate changes and human activities, the natural riparian ecosystems within the whole river basin were degraded enormously, particularly in the lower reaches of the river where about 320 km of the riparian forests were either highly degraded or dead. In this study, we presented one of the main criteria for the assessment of vitality of P. euphrafica forests by estimating the defoliation level, and analyzed forest structure and determined the height-diameter （height means the height of a tree and diameter means the diameter at breast height （DBH） of a tree） relationship of trees in different vitality classes （i.e. healthy, good, medium, senesced, dying, dead and fallen）. Trees classified as healthy and good ac- counted for approximately 40% of all sample trees, while slightly and highly degraded trees took up nearly 60% of total sample trees. The values of TH （tree height） and DBH ranged from 0-19 m and 0-125 cm, respectively. Trees more than 15 m in TH and 60 cm in DBH appeared sporadically. Trees in different vitality classes had different distribution patterns. Healthy trees were mainly composed more of relatively younger trees than of degraded tress. The height-diameter relationships differed greatly among tress in different vitality classes, with the coefficients ranging from 0.1653 to 0.6942. Correlation coefficients of TH and DBH in healthy and good trees were higher than those in trees of other vitality classes. The correlation between TH and DBH decreased with the decline of tree vitality. Our results suggested that it might be able to differentiate degraded P. euphratica trees from healthy trees by determining the height-diameter correlation coefficient, and the coefficient would be a new parameter for detecting degradation and assessing sustainable management of floodplain forests in arid regions. In addition, tree vitality should be taken into account to make an accurate height-diameter model for tree height prediction.

Tree species diversity and stand structure along major community types in lowland primary and secondary moist deciduous forests in Tripura,Northeast India

Tree species diversity and population structure at different community types were described and analyzed for primary and secondary lowland moist deciduous forests in Tripura. Overall 10,957 individual trees belonging to 46 family, 103 genera and 144 species were counted at ≥30 cm DBH （diameter at breast height） using 28 permanent belt transects with a size of 1 ha （10 m × 1000 m）. Four different tree communities were identified. The primary forests was dominated by Shorea robusta （mean density 464.77 trees.ha^-1, 105 species） and Schima wallichii （336.25 trees.ha^-1, 82 species）, while the secondary forests was dominated by Tectona grandis （333.88 trees.ha^-1, 105 species） and Hevea brasiliensis （299.67 trees.ha^-1, 82 species）. Overall mean basal area in this study was 18.01m2.ha^-1; the maximum value was recorded in primary Shorea forest （26.21 m2.ha^-1）. Mean density and diversity indices were differed significantly within four different communities. No significant differences were observed in number of species, genera, family and tree basal cover area. Significant relationships were found between the species richness and different tree population groups across the communities. Results revealed that species diversity and density were increased in those forests due to past disturbances which resulted in slow accumu- lation of native oligarchic small tree species. Seventeen species were recorded with 〈2 individuals of which Saraca asoka （Roxb.） de Wilde and Entada phaseoloides （L.） Men＇. etc. extensively used in local ethnomedicinal formulations. The present S. robusta Gaertn dominated forest was recorded richer （105 species） than other reported studies. Moraceae was found more speciose family instead of Papilionaceae and Euphorbiaceae than other Indian moist deciduous forests. Seasonal phenological gap in such moist deciduous forests influenced the population of Trachypithecus pileatus and capped langur. The analysis of FIV suggested a slow trend of shifting the population of Lamiaceae group by Moraceae species in secondary T. grandis L. dominated community.

Quantifying forest structural diversity based on large-scale inventory data:a new approach to support biodiversity monitoring

Background: The importance of structurally diverse forests for the conservation of biodiversity and provision of a wide range of ecosystem services has been widely recognised. However, tools to quantify structural diversity of forests in an objective and quantitative way across many forest types and sites are still needed, for example to support biodiversity monitoring. The existing approaches to quantify forest structural diversity are based on small geographical regions or single forest types, typically using only small data sets.Results: Here we developed an index of structural diversity based on National Forest Inventory(NFI) data of BadenWurttemberg, Germany, a state with 1.3 million ha of diverse forest types in different ownerships. Based on a literature review, 11 aspects of structural diversity were identified a priori as crucially important to describe structural diversity. An initial comprehensive list of 52 variables derived from National Forest Inventory(NFI) data related to structural diversity was reduced by applying five selection criteria to arrive at one variable for each aspect of structural diversity. These variables comprise 1) quadratic mean diameter at breast height(DBH), 2) standard deviation of DBH, 3) standard deviation of stand height, 4) number of decay classes, 5) bark-diversity index, 6) trees with DBH ≥ 40 cm, 7) diversity of flowering and fructification, 8) average mean diameter of downed deadwood, 9) mean DBH of standing deadwood, 10) tree species richness and 11) tree species richness in the regeneration layer. These variables were combined into a simple,additive index to quantify the level of structural diversity, which assumes values between 0 and 1. We applied this index in an exemplary way to broad forest categories and ownerships to assess its feasibility to analyse structural diversity in large-scale forest inventories.Conclusions: The forest structure index presented here can be derived in a similar way from standard inventory variables for most other large-scale forest inventories to provide important information about biodiversity relevant forest conditions and thus provide an evidence-base for forest management and planning as well as reporting.

Effect of shelterwood cutting method on forest regeneration and stand structure in a Hyrcanian forest ecosystem

A study was carried out to evaluate the effect of shelterwood cutting methods on stand structure and regeneration density. Data were collected from a northern hardwood forest stands in Iran with Fagus orientalis Lipsky as dominant species, with/without shelterwood cutting operation. Results clearly demonstrate that the management of Fagus orientalis Lipsky with shelterwood cutting system affected the frequency and diversity of the understory herbaceous species. The frequency of Viola silvestris Lam., Asperula odorata L., Carex spp. and Rubus hyr-canus Juz increased significantly after shelterwood cutting. The DBH （diameter at breast height） of commercial species in control stands （57.50±2.15 cm） was greater than that in treated stands （50.67±1.88 cm）, whereas the total height of trees was similar between treated and control plots （21±0.5 m）. The number of Parrotia persica seedlings increased by 13.2% from 1995 to 2005 whereas the number of Fagus orientalis and Carpinus betulus seedlings significantly decreased from 1995 to 2005. In conclusion, it confirms that instead of shelterwood cutting method other silvicultural practices such as selection cutting method should be applied for the mountainous beech stands of Hyrcanian forests.

Using a stand-level model to predict light absorption in stands with vertically and horizontally heterogeneous canopies

Background： Forest ecosystem functioning is strongly influenced by the absorption of photosynthetically active radiation （APAR）, and therefore, accurate predictions of APAR are critical for many process-based forest growth models. The Lambert-Beer law can be applied to estimate APAR for simple homogeneous canopies composed of one layer, one species, and no canopy gaps. However, the vertical and horizontal structure of forest canopies is rarely homogeneous. Detailed tree-level models can account for this heterogeneity but these often have high input and computational demands and work on finer temporal and spatial resolutions than required by stand-level growth models. The aim of this study was to test a stand-level light absorption model that can estimate APAR by individual species in mixed-species and multi-layered stands with any degree of canopy openness including open-grown trees to closed canopies. Methods： The stand-level model was compared with a detailed tree-level model that has already been tested in mixed-species stands using empirical data. Both models were parameterised for five different forests, including a wide range of species compositions, species proportions, stand densities, crown architectures and canopy structures. Results： The stand-level model performed well in all stands except in the stand where extinction coefficients were unusually variable and it appears unlikely that APAR could be predicted in such stands using （tree- or stand-level） models that do not allow individuals of a given species to have different extinction coefficients, leaf-area density or analogous parameters. Conclusion： This model is parameterised with species-specific information about extinction coefficients and mean crown length, diameter, height and leaf area. It could be used to examine light dynamics in complex canopies and in stand-level growth models.

Effects of thinning on stand structure and tree stability in an afforested oriental beech(Fagus orientalis Lipsky) stand in northeast Turkey

We studied relationships between stand structure and stand stability according to thinning intensity in an afforested oriental beech stand. Various thinning intensities were applied in sample stands. We sampled eight plots in stands that were lightly thinned, eight plots in heavily thinned stands and eight plots in unthinned stands as a control. Height and diameter distributions of the stands were measured to assess stand structure. We quantified individual tree stability and collective stability. Heavy thinning during the first thinning operation damaged the storied structure of the stand in thicket stage and affected collective structuring ability. While most control plots had multi-storied stands, after light and heavy thinning two-storied structure became more common.Large gaps occurred in the canopy after heavy thinning. On average, nine tree collectives were formed per sampling plot in the untreated stand, seven collectives after thinning in 2008 and four collectives after thinning in 2009. Stable trees accounted for 17 % of trees in control plots, 24 % in lightly thinned plots, and 15 % in heavily thinned plots. Collective stability values were 83 % in control plots, 82 % in lightly thinned plots and 36 % in heavily thinned plots. We conclude that it is necessary to retain collective structuring capacity during thinning operations for sustaining stand stability.

How does the stand structure of treeline-forming species shape the treeline ecotone in different regions of the Nepal Himalayas?

Stand structure dynamics are considered as major happenings in any forest as a response to environmental changes.However,this important topic is underrepresented in the treeline studies in the Nepal Himalayas.We aimed to investigate site-as well as species-specific changes in morphometric features(basal diameter,crown cover,density,and height)along the elevational gradient across treeline ecotones in response to recent environmental changes.The stand structure characteristics of Abies spectabilis,Pinus wallichiana,and Betula utilis across the treeline ecotone of three study sites in Eastern(Barun),Central(Manang),and Western(Dhorpatan)Nepal were analyzed to elucidate structural heterogeneities.Altogether,eight transects(20 m×(60–250 m))across the treeline ecotone were established.Trees of all life forms,trees(>2 m),saplings(0.5–2 m),and seedlings(<0.5 m),within each transect were enumerated and sampled for the morphometric features and age.Site-specific and species-specific stand structure dynamics were found.The rate of basal area increment was higher in Barun,but the Manang treeline,despite profound regeneration in recent years,had a low annual basal area increment.Moreover,the altitudinal distribution of age and morphometry were not consistent among those ecotones.Furthermore,intra-specific competition was not significant.The site-specific stand structure dynamics explain why treelines do not respond uniformly to increasing temperature.It invokes,in further studies,the incorporation of the tree’s morphometric adaptation traits,phenotypic plasticity,and interactions between species genotype and the environment.

Exploring tree diversity and stand structure of savanna woodlands in southeastern Sudan

Savanna woodlands in Sudan host great biodiversity, provide a plethora of ecosystem goods and services to local communities, and sustain numerous ecological functions. Although the importance of the Acacia trees in these areas is well known, up-to-date information about these woodlands' diversity is limited and changes in their woody vegetation composition, density, diversity and relative frequency are not monitored over time. This study explored tree diversity and stand stage structure in Nuara Reserved Forest, a typical savanna woodland ecosystem in southeastern Sudan. A total of 638 circular sample plots(1000 m^2 for each) were established using a systematic sampling grid method. The distance between plots was 200 m. In each plot, all living trees with diameter at breast height(DBH) ≥5.00 cm were identified and counted, and their DBH values were recorded. From these data, tree composition, diversity, density and stage structure were assessed. There were 12,259 individual trees representing four species(Acacia seyal, Balanites aegyptiaca, Acacia Senegal and Acacia mellifera) that belong to two families. The dominant species was Acacia seyal. Average tree density was 191 trees/hm^2 and the Shannon-Weiner index for trees diversity was 0.204. Overall, young trees comprised 86.30% of the forest. The state of tree richness and density in the study area was low compared to other similar environments in the region and around the world. We recommended adoption of a proper management system that includes monitoring of woody vegetation diversity in this forest, and management actions to enhance tree diversity and sustain ecosystem services to local communities. In addition to care for the dominant Acacia seyal stands, more attention and conservation should be devoted to reestablishing Acacia senegal and Acacia mellifera trees because of their high ecological and economic values for local communities.

Mature stand developmental stage has ceased to constitute the most suitable habitat for the capercaillie in the Augustow Forest,Poland

Background:Forest management affects the habitat conditions for many forest-dwelling species.Among them,the capercaillie(Tetrao urogallus)is a rare forest grouse inhabiting old,mature forests.We compared the structure of forest habitat among 9 active and 9 abandoned leks in the Augustow Forest(North-Eastern Poland),within a radius of 1 km of the leks,defined as the Key Areas for the capercaillie in lowland temperate forest.Habitat measurements were conducted on 1779 circular plots.Assessments made on all plots related to 13 habitat variables measured or noted in the field,including stand structure,canopy closure,stand developmental stage,percentage of Scots pine(Pinus sylvestris),soil fertility and soil moisture,the share of undergrowth,the cover of shrubs,the cover of bilberry(Vaccinium sp.),and the presence of certain habitat elements important to the capercaillie.Results:To compare the still-occupied and the abandoned KAs for the capercaillie,a logistic regression model was developed.The variables best explaining differences between these two categories were:the occurrence of undergrowth layers,canopy closure in the second canopy layer,and stand age.According to the model,with the increase of the shrub-layer cover as well as the density of trees,the probability of the presence of the capercaillie decreased.The capercaillie in the area of the Augustow Forest occupy mainly dry and poor,middle-aged,pinedominated forests,with a moderate extent of stand canopy closure and only weakly-developed layers of undergrowth.Conclusions:The filling-in of mature stands with sub-canopy trees and shrubs(the process which is stimulated by climate change and site eutrophication)causes structural changes,which are unfavourable to the capercaillie.This might explain why in the course of the recent decades the capercaillie has abandoned the oldest stands,distinguished by the presence of bigger shares of undergrowth.The capercaillie has shifted to younger stands,which reveal a lesser extent of canopy closure and a more limited development of understorey vegetation.

STUDY ON THE TREE GROWTH, ARCHITECTURE AND STAND STRUCTURE OF KOREAN PINE PLANTATION

The artificial pure and mixed Korean pine (Pinus koraiensis) forests were investigated at Dailing Forestry Bureau in Xiaoxing'an mountains from 1990 to 1992. Depending on the distance between the samplings of Korean pine and their neighbor trees, the neighbor tree height, the size of neighbor tree canopy, and dimension of neighbor tree. The forest structure was classified into three types: (1) prowth of a tree in the light (open), (2) Growth of a tree in the canopy gap (Gap), (3)Growth of a tree under broad-leaved tree canopy. The frequeney, height, and age of stem divergence of Korean pine tree were investigated by sampling trees. The temporal and spatial model of the tree growth was applied on basis of the height of stem divergence, ratio of height and DBH, and character of tree stem.The morphology and growth character of Korean pine trees during different development stage were forecasted.

Floristic Diversity and Stand Structure of the Ngambe-Ndom-Nyanon Communal Forest, Littoral Region of Cameroon

The present study aimed to contribute to the sustainable management of the Ngambe-Ndom-Nyanon communal forest in Cameroon. Thus, an evaluation of the floristic diversity and stand structure of the forest was carried out. By using systematic sampling, a diverse inventory of the plant species was carried out, taking into account the dendrometric measurements of all the species present in the forest (trees, shrubs, herbs). Data were analyzed with Microsoft Excel software to bring out curves and histograms. This software also made it possible to produce tables and also to come up with different diversity indexes. A total of 19,007 individuals from 395 plant species grouped into 99 families were identified in the area. The family Euphorbiaceae was the most represented. Species dominated in this stand were Uapaca guineensis and Pycnanthus angolensis with 9% and 6% of individuals identified. The result of the Shannon index showed that all the strata in the forest are diversified with the highest values in the adult secondary forest (FSA) with indices of 4.74, followed by young secondary forest (FSJ) and the secondary forest (FS) with indice respectively 4.68 and 4.61. The values of the Sorensen index show that 100% of species are common in the young secondary forest (FSJ) and the adult secondary forest (FSA). In the same forest, the FSA and Swamp with FSJ and Swamp have just 15% of species in common respectively. This diversified forest stand projects an inverted J structure, thus testifying to its strong potential for stems of the future. According to the research objectives, some measures were proposed for sustainable management of this forest, such as the reforestation of areas degraded by agricultural activities and the promotion of agroforestry practices by using trees with the high fertilizing potential to improve agricultural yields while preserving the trees in place.

Stand structure and yield of the mixed white poplar and black locust plantations on sandy ridges between the Danube and Tisza rivers in Hungary

The paper deals with the stand structure and yield of black locust (Robinia pseudoacacia L.) forests mixed with white (Populus alba L.) in various proportions, partly applying a new methodological approach. The main stand structure and yield factors were determined separately for each species, measured stem by stem, using the volume functions prepared for each species. The ratio of the volumes of the species (A and B) in mixed and in pure stands (based on volume tables) was determined. A close relationship has been found between the ratio by relative total volume and the proportion (by the number of stems) of the species. The relative surplus in the volume of the mixed stands varied between 1.24-1.55 at the age of 16 compared to the control, i.e. the yield of pure stands of the species concerned. The trial has also proven that if two species have a fast initial growth rate and a similar rotation age, they can be planted in mixed stands resulting in mutual advantages.

Prediction of tree crown width in natural mixed forests using deep learning algorithm

Crown width(CW)is one of the most important tree metrics,but obtaining CW data is laborious and timeconsuming,particularly in natural forests.The Deep Learning(DL)algorithm has been proposed as an alternative to traditional regression,but its performance in predicting CW in natural mixed forests is unclear.The aims of this study were to develop DL models for predicting tree CW of natural spruce-fir-broadleaf mixed forests in northeastern China,to analyse the contribution of tree size,tree species,site quality,stand structure,and competition to tree CW prediction,and to compare DL models with nonlinear mixed effects(NLME)models for their reliability.An amount of total 10,086 individual trees in 192 subplots were employed in this study.The results indicated that all deep neural network(DNN)models were free of overfitting and statistically stable within 10-fold cross-validation,and the best DNN model could explain 69%of the CW variation with no significant heteroskedasticity.In addition to diameter at breast height,stand structure,tree species,and competition showed significant effects on CW.The NLME model(R^(2)=0.63)outperformed the DNN model(R^(2)=0.54)in predicting CW when the six input variables were consistent,but the results were the opposite when the DNN model(R^(2)=0.69)included all 22 input variables.These results demonstrated the great potential of DL in tree CW prediction.