• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Developing allometric equations to estimate forest biomass for tree species categories based on phylogenetic relationships

    2023-10-07 02:54:00MingxiYngXioluZhouChnghuiPengTongLiKexinChenZelinLiuPengLiCichengZhngJiyiTngZiyingZou
    Forest Ecosystems 2023年4期

    Mingxi Yng, Xiolu Zhou,*, Chnghui Peng,b,**, Tong Li, Kexin Chen, Zelin Liu,Peng Li, Cicheng Zhng, Jiyi Tng, Ziying Zou

    a School of Geographic Sciences, Hunan Normal University, Changsha, 410081, China

    b Institute of Environment Sciences,Department of Biology Sciences,University of Quebec at Montreal,Case Postale 8888,Succursale Centre-Ville,Montreal,Quebec,H3C 3P8, Canada

    Keywords:

    ABSTRACT The development of allometric biomass models is important process in biomass estimation because the reliability of forest biomass and carbon estimations largely depends on the accuracy and precision of such models.National Forest Inventories (NFI) are detailed assessments of forest resources at national and regional levels that provide valuable data for forest biomass estimation.However, the lack of biomass allometric equations for each tree species in the NFI currently hampers the estimation of national-scale forest biomass.The main objective of this study was to develop allometric biomass regression equations for each tree species in the NFI of China based on limited biomass observations.These equations optimally grouped NFI and biomass observation species according to their phylogenetic relationships.Significant phylogenetic signals demonstrated phylogenetic conservation of the crown-to-stem biomass ratio.Based on phylogenetic relationships,we grouped and matched NFI and biomass observation species into 22 categories.Allometric biomass regression models were developed for each of these 22 species categories, and the models performed successfully (R2 = 0.97, root mean square error (RMSE) = 12.9 t?ha–1,relative RMSE=11.5%).Furthermore,we found that phylogeny-based models performed more effectively than wood density-based models.The results suggest that grouping species based on their phylogenetic relationships is a reliable approach for the development and selection of accurate allometric equations.

    1.Introduction

    As the main component of natural carbon sequestration in terrestrial ecosystems, forest ecosystems have sequestered more than 80% of all terrestrial aboveground carbon(Richter et al.,1999;Jandl et al.,2007),in which woody plants accounting for approximately 70%of the carbon storage(Laiho and Laine,1997).Forest carbon is the amount calculated based on directly measurable sizes,such as diameter,height,and volume.Forest carbon estimations are predominantly carried out by converting measurable tree variables.These tree variables have been surveyed for more than a century to compile the National Forest Inventory (NFI) in Europe and North America(McRoberts et al., 2012; Gschwantner et al.,2022).In China, the systematic NFI started in 1973 and is repeated at approximately five-year intervals.There are now nine consecutive surveys available.The NFI system has undergone continuous improvements with respect to sampling design,survey methods,and technical standards(Lei et al.,2009;Zeng et al.,2015;Yin et al.,2022).These NFIs provide data on forest area and volume, which have been converted to forest carbon dynamics to estimate the carbon sequestration ability of forests in China during the past half century (Fang et al., 2001; Zeng, 2014).In addition to area and volume,many countries actively use information on diameter at breast height (DBH)and tree height (H) in their NFIs to estimate forest carbon storage and understand the dynamic development trend of forest growth (Joosten et al., 2004; Domke et al., 2020).Consequently,NFIs play a primary role in forest ecosystem management and policy design to achieve global carbon neutrality.

    The conversion of forest volume to biomass using forest volume and area data provided by the NFI has attracted significant attention(Wang et al.,2001;Henry et al.,2015).The approach,commonly known as the volume-derived method, is a non-destructive and easy-to-implement approach when compared with harvesting methods.There are two primary methods of converting volume into stand biomass: allometric biomass equations and biomass expansion factors.Within the allometric biomass equation method, there are two main types of biomass equations.The first type uses stand volume as the predictor variable to directly calculate whole biomass(Muukkonen,2007),i.e.,aboveground biomass (AGB) and belowground biomass (BGB).The second type includes two steps: (1) multiplying stand volume by the wood density to obtain stem biomass;(2) expanding stem biomass to whole biomass via the allometric equation, with stem biomass as the predictor variable(Fang et al.,2002).Alternatively,a simpler conversion involves biomass expansion factors to represent the ratio of biomass to volume for specific forest types or species.The stand biomass can be estimated by multiplying the stand volume and expansion factor.In addition, if the NFI information includes DBH and H, stand biomass can be summed from individual tree biomass estimated by allometric equations using DBH or DBH and H as the predictor variable (Chave et al., 2005; Zianis et al.,2005;Forrester,2017; Luo et al.,2020;Loubota Panzou et al.,2021).

    Previous studies have provided evidence that allometric relationships vary among different tree species(Xiang et al.,2016).However,the high level of accuracy achieved by specific species equations may restrict their applicability to other species.If an equation based on one species is applied to another species, a substantial bias may occur in the biomass estimation.The estimation of forest biomass based on the NFI information may be hindered by the lack of adequate equations for predicting biomass for every species.Such a limitation arises because conducting destructive harvests,which are labour-intensive,time-consuming,costly and sometimes impossible, pose challenges for obtaining the necessary data (Ishihara et al., 2015; Xiang et al., 2016).Consequently, the estimation of forest biomass based on the NFI remains an urgent need,which could facilitate the comprehensive analysis and classification of the species in NFI.

    The NFI of China provides volume information for 57 individual tree species and five mixed forest types derived from the ninth National Forest Inventory of China.However,we found that only 20 specific species(or forest types)have sufficient field data comprising more than 10 samples for parameterizing their regression models.Although it is the largest public dataset of biomass measurements for China’s forest ecosystems(Luo et al.,2013),including 131 species measured over the past decades,many species have only one or a few measured samples.This means that many species in the NFI were not measured or there was a lack of sufficient samples.Parameterizing an allometric equation requires an adequate sample that is representative of each species, and this is a dilemma faced by modellers.This difficulty may also be an issue in estimating forest carbon in other countries.In a previous study in Europe(Zianis et al., 2005), among all the equations reporting sample sizes,more than 16%of the biomass equations were estimated from a number of samples varying from 3 to 10.This limited amount of data samples may result in considerable uncertainties in biomass estimation, highlighting the pressing need for more samples of specific species.However,owing to the substantial cost of conducting biomass measurements,it is impractical to record a large number of samples for each tree species in an NFI.

    To date, for those NFI species that lacked sufficient measurement samples,several approaches have been generally applied for modelling:(1) collecting field data by empirically grouping species according to their traits,that is shade tolerance,evergreens,softwood,and hardwood,to construct the equations for these species (Jenkins et al., 2003; Weiskittel et al.,2015);(2)using the equations of specific species to directly estimate the biomass for other species that have similar traits, namely wood density,deciduous,or conifer(Fang et al.,2001;Guo et al.,2010;Zhou et al.,2019);and(3)using a general equation for the species(Chave et al., 2014).Overall, little attention has been paid to grouping tree species to estimate forest carbon based on NFI information.Therefore,further studies should be conducted to develop more suitable methods for accurate estimation of biomass.

    The biomass of a species in an NFI can be computed using different biomass equations for different species groups (Akindele and LeMay,2006).However, this potentially increases the complexity of evaluating the amount of carbon stored in forest based on NFI.If the species cannot be closely matched, the equations cannot adequately explain the variation in biomass measurements (Ducey, 2012).Potential errors may be caused by directly applying the model coefficients of one species to another species(Jenkins et al.,2003).Alternatively,if general equations are applied across species, it would be useful to predict the global-scale forest biomass.Nevertheless, for a single country, NFI-based biomass estimation may need a more accurate approach.A general equation may result in large errors at some sites or species (Chave et al., 2014).Therefore, to use existing measurements and reduce inconsistencies between the model and predicted species, there is a need to explore a consistent method for estimating forest biomass at a relatively large scale.This raises the question of how to parameterize the regression model for the species in the NFI using available data from other species,that is, how to solve the tree species grouping problem (observed tree species vs.species counted in the NFI) to address the issue of the inconsistency between there being fewer measurements available and the necessary statistical criteria for estimating the biomass of each tree species (Jenkins et al., 2003; Weiskittel et al., 2015).Modelling by grouping based on similarities and differences between tree species is an urgent problem for an improved understanding of forest carbon sequestration at the national scale.

    In evolutionary biology, a phylogenetic relationship is considered a surrogate for analyzing variations in species traits,under the assumption that closely related species, that is, tree species close to each other in a phylogenetic tree,are more similar than distant ones(Burns and Strauss,2011; Anacker and Strauss, 2016).Phylogeny has been identified as a robust factor controlling biomass allocation patterns across a wide range of plant sizes, particularly at the family level within higher taxonomic levels(McCarthy and Enquist,2007).We hypothesized that tree species with closer phylogenetic relationships would exhibit similar allometric relationships.Following this,the set of biomass equations constructed by grouping tree species according to the phylogenetic relationship and matching with species in NFIs may serve to mitigate the level of uncertainty associated with estimating forest biomass.In this study,we tested the effect of phylogenetic relationships on the traits of crown and stem biomass,grouped the species according to phylogenetic relationship,and then constructed allometric equations based on each species group.We also compared the goodness-of-fit of the models for different species groups,such as wood density-based groupings.Our objectives were(1)to group species according to their phylogenetic relatedness and investigate the relationship between phylogenetic relatedness and biomass allometry;(2)to develop stand biomass allometric models using stem biomass as the predictor variable for each of the phylogeny-based species groups,by taking advantage of the availability of the NFI data.

    2.Materials and methods

    2.1.Overview and framework

    The study encompassed two distinct datasets: field biomass measurements and NFI information.To effectively group the species,a novel approach based on phylogenetic relationship was adopted.The phylogenetic approach facilitated the grouping of species according to evolutionary similarities among the species.Subsequently, the two datasets were carefully divided into the same categories and matched, which finally yielded 22 categories.After grouping, allometric models were developed through nonlinear regression analysis.In the models,the stem biomass served as the predictor variable, while the whole biomass was considered the dependent variable.To determine the stem biomass,the stem volume was multiplied by the corresponding wood density.Overall,by integrating phylogenetic relationships, proper categorization, and accurate estimation, the framework helps to reduce biomass estimation errors and provides a more robust understanding of biomass patterns and allocation strategies(Fig.1).

    Fig.1.Flow chart for modelling China’s forest biomass at the national scale.Bs,stem biomass; Bw, whole biomass; AGB, above-ground biomass; BGB, belowground biomass.

    2.2.Data source

    The data included two parts: (1) The field forest measurements of volume per hectare (m3?ha–1) and dry weight of biomass per hectare(t?ha–1)of each tree organ,that is,foliage,flowers,fruit,branches,stems,coarse roots, and fine roots for 128 species at 994 sites in forest ecosystems of China(Luo et al.,2013,Figs.1 and 2;Table S1).These data were collected from harvested sample plots and used to develop allometric models to describe the relationship between stem and whole biomass(Bw) and analyse the phylogenetic relationships of biomass allocation.Among the data,measurements without any main organ biomass that is,stem biomass, crown biomass, aboveground biomass (AGB), and belowground biomass (BGB) were excluded from the final dataset that included whole biomass.(2) The latest (ninth) NFI for China’s forests from 2014 to 2018,which compiles information on forest area and volume of five age groups,comprising young,middle,near-mature,mature,and over-mature dominant tree species and forest types totalling 62 species types in 31 provinces of the country (Table S2).Among the 62 species types,the specific tree species name was known for 40 species.A total of 16 species were identified and categorized at the genus level,and six forest types were classified as mixed forests.As a compilation of national resource statistics,the NFIs have undergone a systematic sampling design and rigorous statistical tests to provide reliable information at the national scale(Fang et al.,2001;Lei et al.,2009;Zhang and Wang,2021),although errors may still exist.The information released from these NFIs has been used to publish reports on forest biomass and carbon storage(Food and Agriculture Organization,2020).According to these data,we assigned the species in data (1) and (2) as Measured-species and NFI-species, respectively.The two datasets include some established species categories for mixed stands and multi-species, such as “mixed broadleaf–conifer forests”,“mixed broadleaf forest”,and“mixed conifer forests”.However, the categories were not included in the tree species grouping process due to the availability of sufficient measurements for parameterization(refer to Table S4 in Supporting Information).

    2.3.Examination of the effect of phylogenetic relationship

    Fig.2.Field sites for measuring tree biomass and stem volume in China,with a total of 994 sites and 128 tree species.The green colour represents the forest cover,and the blue circles represents the forest plots.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

    To test whether phylogeny influenced tree morphology, we calculated the crown-to-stem biomass ratio(crown biomass:stem biomass)to represent the relative size of different tree organs.We screened for tree species with at least four biomass observations,resulting in a total of 50 species (Table S3).To assess the mean crown-to-stem biomass ratios at the species level, phylogenetic trees of all the species selected were acquired from the largest mega-tree of vascular plants using the V.Phylo-Maker R package (Jin and Qian, 2019).Furthermore, the phylogenetic signal can capture the tendency of closely related species to exhibit similar trait values due to shared evolutionary history.It quantifies the degrees of trait (e.g., crown-to-stem biomass ratio) conservation or divergence along the branches of a phylogenetic tree,providing insights into the role of phylogenetic relationships in shaping trait variation.When there is a high phylogenetic signal,it suggests that closely related species are more similar in their traits, indicating a strong degree of phylogenetic conservatism.Conversely, a low phylogenetic signal implies that traits are influenced more by other factors such as environmental conditions rather than shared ancestry.To determine the phylogenetic signal of the crown-to-stem biomass ratio, we computed Pagel’s λ value, a widely used indicator for quantifying phylogenetic signals.Pagel’s λ is a number between 0 and 1,where λ=1 indicates that the trait variation depends entirely on phylogeny, and λ = 0 indicates that there is no phylogenetic dependence (Pagel, 1999).The statistical measure assumes of a Brownian model of trait evolution and is evaluated using the phylosig function from the phytools package in R(Revell,2012).We obtained a p-value to assess whether λ was significantly different from 0,that is,the validity of a phylogenetic signal.

    2.4.Species grouping

    The two species data sets, Measured-species and NFI-species, were classified into an equal number of groups (22 groups) based on their phylogenetic relationship, number of measured biomass samples (primarily for Measured-species), and wood density (primary for NFIspecies).Subsequently, the species groups of the two species datasets were paired according to the phylogenetic relationship (Fig.1).The average trait values of closely related phylogenetic groups are closer than those of distantly related groups (Losos, 2008; Crisp et al., 2009).Nevertheless, some species could be exceptions.A few NFI-species that do not belong to any genus or family of Measure-species could not be grouped based on phylogenetic relationship.The level of phylogenetic closeness among the hardwood species in Group#9(Model 9)exhibits a wide range of variation.They could first be classified by class of seed plants into gymnosperm and angiosperm and then grouped based on similar wood density.

    Another modelling requirement is to ensure that there are sufficient measurements for each group.For developing a biomass allometric model, it is necessary to have a wide range of data that encompasses small to large trees.We grouped tree species with a suitable number of biomass samples measured, which allowed for reliable model parameterization.Although the plant classification system is still under development,there is a universal consensus on the genus-level classification of dominant tree species in national forests.If the tree species in NFIs do not have sufficient observed biomass data for common species,observations from the same genus can be used to determine the parameters of the biomass equation by regression fitting.Using the V.PhyloMaker package,we generated two phylogenetic trees for the 128 Measured-species and 59 NFI-species in the ninth NFI of China (Fig.1).The NFI-species and Measured-species were matched according to two phylogenetic trees and the number of biomass samples observed for each species.

    2.5.Modelling

    We used a conventional allometric equation (power-law function;Huxley, 1924) to model the allometric relationship between stem biomass and whole biomass.Stem biomass was calculated by multiplying the volume with wood density,which can be represented by the slope of the linear regression equation for the dependent variable stem biomass and the independent variable wood volume (V) (Liu et al., 2019; Zhou et al.,2019).Therefore,the biomass equation used at the national scale was constructed as follows:

    where y is the whole biomass (Bw= AGB + BGB) per hectare, x is the stem biomass per hectare(Bs),ρ is the wood density,V is the stem volume per hectare, α and β are two undetermined parameters, and β is the allometric scaling coefficient.This biomass equation was used to estimate the whole biomass per unit area.Given that the information released in China’s NFI only included the forest volume and area data for dominant tree species and age groups corresponding to each province,Eqs.1 and 2 were parameterized based on the grouped species and field measurements at the plot level(Fig.1).The wood density ρ of a species can be estimated based on wood industry standards, literature, public wood density databases, and field measurements of stem biomass and volume (Chinese Academy of Forestry Sciences, Research Institute of Wood Industry,1982;Zanne et al.,2009;National National Forestry and Grassland Administration, 2017).For reference, we listed the wood density values for each NFI-species(Fig.4).

    2.6.Modelling comparison

    In the Measured-species dataset consisting of 128 species and 994 measurements, 63 species had a small number of measurements, only 165 in total.The number of these species accounted for a large proportion of Measured-species, reaching 49.2%.We compared these species using different grouping methods parameterized in Eq.1 by grouping the species based on phylogenetic relationship (Grouping I) and different wood densities (Grouping II).Grouping II included two different classifications of wood density values using different initial values to avoid coincidences in the results.The two groups were compared according to residual error,root mean square error(RMSE),relative RMSE(%RMSE),percentage error (error), and coefficient of determination (R2) for the species rarely measured.We calculated the RMSE and %RMSE for the whole biomass for plot i,as follows:

    where N denotes the number of plots of measured biomass in each equation, and yobs_iand ypred_iare the measured whole biomass and estimated whole biomass of plot i in each equation, respectively.RMSE and%RMSE were used to assess the prediction errors and performance of models.A lower RMSE value signifies higher predictive accuracy,while a lower %RMSE value indicates increased stability in the model’s performance across different data subsets.In addition, the R2value, which ranges from 0 to 1, provides information on the goodness-of-fit of the model.

    Furthermore, the model test was designed using training and test datasets.We selected four categories,namely Models 12,13,17,and 22,each having a large sample size(greater than 50).The data were divided into training and test datasets in a 7:3 ratio.We compared the model performance on training and test datasets separately using two distinct grouping methods based on phylogeny and wood density.

    3.Results

    3.1.Biomass allocation phylogeny and grouped species

    Fig.3.Phylogenetic trees of selected species and the ratio of crown biomass to stem biomass (CB/SB).Purple denotes gymnosperms, and dark blue denotes angiosperms.Available data from 50 species were sifted through measurements for the analysis.The phylogenetic tree was constructed using R package V.PhyloMaker.Pagel’s λ is a phylogenetic signal with a value between 0 and 1.The significant level is*p<0.05.(For interpretation of the references to colour in this figure legend,the reader is referred to the Web version of this article.)

    Tree crown-to-stem biomass ratio is a phylogenetically conserved characteristic.We detected a significant phylogenetic signal(Pagel’s λ=0.359, p < 0.05; Fig.3), indicating that the close tree species tended to have similar morphology.Based on the phylogenetic relationship and number of measured data, the NFI-species were grouped into 22 categories (Fig.4).There were 10 categories of angiosperms and 12 categories of gymnosperms.A total of 22 species-specific allometric models were developed, and each model corresponded to a category of NFI-species.Twelve models corresponded to a single species, and 10 models corresponded to species groups, which included a total of 59 species.Corresponding to these 22 categories of NFI-species, each Measured-species was also merged into 22 categories.This ensured that the model could be parameterized based on the samples measured ranging from 9 to 266.The number of species measured included in the models ranged from 1 to 24,with a median of 4.

    3.2.Allometric relationship between stem biomass and whole biomass

    Using existing data on destructive stem biomass and whole biomass across forest ecosystems in China, we were able to fit 22 allometric models for all species (59) in the ninth NFI of China.The relationships between stem biomass and whole biomass were effectively explained by the allometric models(Fig.5).All of these models,with R2values ranging from 0.901 to 0.997, demonstrate a high level of performance, with an average variance explained of 97.0% (Table 1).The parameters of the biomass models for the 22 models are listed in Table 1.All the models were highly significant(p<0.01)for both parameters α and β.The value of the RMSE ranged from 7.0 to 24.1 t?ha–1, with an average of 12.9 t?ha–1.The mean%RMSE for the whole biomass of the allometric models was 11.5%(ranging from 6.1%to 22.5%).The scaling coefficient(β)was higher for angiosperm species,with an average of 0.90,and significantly lower for gymnosperm species,with an average of 0.84(p <0.01).

    3.3.Comparison of different species groupings

    To compare the fitting accuracy of the allometric models between different species groups,we selected eight models(with*in Table 1)to test the fitting ability of Groupings I and II,including 63 species and 165 field measurements.Depending on the wood density (WD) measured,these species were classified into four groups (Table 2).Their wood densities were distributed at 0.40,0.50,0.60(WD Classification 1;t?m–3)and 0.35, 0.45 and 0.55 (WD Classification 2; t?m–3).These models (in Table 2) can be substituted for the models with asterisks(*) in Table 1.The model number,sample size(N),parameters α and β,and values of R2,RMSE,and%RMSE are listed in Table 2.The error analysis indicated that the residual range for Grouping I (-48% to 48%) was less than that of Grouping II(-95%to 53%for WD Classification 1;-99%to 53%for WD Classification 2;see Fig.6).The slope for Grouping I was 0.01,while for Grouping II, the slopes were -0.07 and -0.13 corresponding to WD Classification 1 and 2, respectively.The significant difference in RMSE and %RMSE between the two groups indicated that the model performance of Grouping I was more effective than that of Grouping II.On average,the grouping based on the phylogenetic relationship presented a higher goodness-of-fit than the models grouped based on wood density.

    The robustness test results showed that both the phylogeny and wood density-based models demonstrated consistent performance between the training and test datasets(Figs.S1 and S2).The model evaluation on the test set suggested that the phylogeny-based model(R2=0.964;RMSE=12.4 t?ha–1; %RMSE = 12.1%) outperforms the wood density-based model(R2=0.964;RMSE= 13.4 t?ha–1;%RMSE=13.0%;Fig.S3).

    Fig.4.Grouping and classification of NFI-species and Measured-species based on the phylogenetic relationships and the number of measured biomass samples recorded.The left(orange)and right(light green)phylogenetic trees are the measured tree species,with angiosperms on the left and gymnosperms on the right.The number in parentheses after the tree species names indicates the number of records measured for that species.The NFI-species at the top are listed in order of phylogenetic relationships.The symbol“√”corresponds to the NFI-species above for grouping to construct biomass models.There were 22 groups in total,that is 22 models could be developed.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

    4.Discussion

    Our results have demonstrated a relatively strong fit for the allometric relationship between stem biomass and whole biomass.The whole biomass (Bw) estimated using the phylogeny model effectively fit the measured values, compared with that estimated by the wood density model (Figs.5 and 6).This suggested that the classification of existing tree species according to the phylogenetic relationship can be used to construct suitable biomass equations.The biological meaning of this approach lies in the fact that the allometric relationships of tree species with close phylogenetic relationships have similar characteristics(Reich et al., 2014; Sun et al., 2020).This result confirms our hypothesis.In addition to analyzing the allometric relationship between the stem and whole-tree biomass, we tested the phylogenetic signal.A significant signal was observed for the crown-to-stem biomass ratio (Fig.3).Phylogenetic analysis suggested that the allometric relationship between stem and crown biomass was generally related to the phylogenetic relationship of the species, showing phylogenetic conservatism in the geometric structural characteristics of trees (Reich et al., 2014; Poorter et al.,2015).Although we only focused on the relationship between the crown and stem biomass in the present study, previous works have reported that phylogeny influences the scaling exponent of allometric relationships of root biomass with DBH(Sun et al.,2020).Consequently,considering phylogeny can facilitate the classification of species into appropriate categories based on their shared evolutionary histories.Biomass allometric equations and estimations of forest carbon can be used for improving the model by considering species classification.Overall, developing biomass models by grouping tree species according to phylogenetic relationships can achieve a high degree of fitting accuracy.

    Fig.5.Power equations(y=αxβ)of the allometric relationship between stem biomass and whole biomass.The equations were developed by grouping 22 tree species in the NFI by phylogenetic relationships.In the regressions,y and x denote whole biomass(t?ha–1)and stem biomass(t?ha–1),respectively;stem biomass is the product of wood density(ρ)and volume(V);α and β represent model regression coefficients.All data were measured in the field by destructive sampling for 128 species.The numbers in parentheses denote the 22 species grouped in the NFI (see Fig.4).R2 means coefficient of determination.

    Table 1 Allometric models for estimation of whole biomass in NFI-species.The model numbers correspond to the ones in Figs.4 and 5.

    Table 2 Allometric models parameterized based on the species wood density.

    Fig.6.Comparison of model fitting between two parameterizations.(a)Grouping I, species were categorized according to their phylogenetic relationships, corresponding to the tree species in the equations with asterisks (*) in Table 1.The biomass of these species was rarely measured or had few data samples.(b) Grouping II, species were classified according to wood density,corresponding to the models and WDs in Table 2.The tree species are the same as Group 1.(c)The values of RMSE and%RMSE are averages of all groups.WD,wood density; RMSE, root mean square error; %RMSE, relative RMSE.

    In some cases, wood density can also be a reference factor for grouping species and modelling allometric relationships (Manuri et al.,2014).From the perspective of stem biomass,this is appropriate because species with the same wood density should have a similar relationship with volume-to-Bs.However, their metabolic patterns and nutrient use efficiency are not necessarily the same.This may result in differences in the ratios of crown-to-stem biomass and the allometric relationships between crown and stem biomass.This implies that the parameters of the biomass equation (Eq.1) would be different.A comparison of grouping approaches confirmed that mixing the measurements of the species that have different allometric relationships may lower the goodness-of-fit.The averages of the evaluation indicators (R2, RMSE, and %RMSE) for the goodness-of-fit indicated clear differences between Grouping I and Grouping II(the last line in Table 2).Fig.6 illustrated that Grouping I had lower residual errors, higher accuracy(RMSE),and higher precision(%RMSE) than Grouping II.Given that the different initial values were designed for Grouping II,the potential similarities in the results could be substantially reduced, which improved the robustness of the analysis.These findings partly confirm our hypotheses.In general,grouping species based on phylogenetic relationships may be a more effective solution.Species and wood density are the most essential dimensions for grouping species and constructing biomass equations.Conducting a comparison between these two dimensions could be beneficial for an improved understanding of the effects of species composition on modelling forest biomass at the national scale.We expect that biomass datasets will be continually updated in the future.These rich field data will ensure accurate model parameterizations.However, we were not able to test the effects of grouping in more dimensions, such as climatic zone,ecoregion,soil texture,or life habit,which may also influence the allometric relationship(Reich et al.,2014;Loubota Panzou et al.,2021).

    The method presented in this study has broad application prospect.It addresses the issue of lack of measurements to obtain a biomass equation for a species and how the field data for other species are used to construct the biomass equation for this species.Constructing equations by grouping species without an agreed generalizable solution may become an issue because of the large biological differences between tree species.Biologically,allometric relationships reflect the metabolic efficiency of a plant as a whole.Metabolic efficiency is affected not only by climate or site conditions(Lines et al.,2012;Dusenge et al.,2019)but also by plant traits (Mcgill et al., 2006).This is also attributed to species evolution based on the taxonomy(Vasseur et al.,2012;Weng,2014).This implies that the appropriateness of tree species must be stressed equally in the analysis of the allometric relationships of different tree organs.Biomass may differ largely between the forests of the two species (Scheller and Mladenoff, 2005).A pronounced difference in carbon sequestration has been indicated for forest communities with different biodiversity (Wardle et al., 2012).In terms of practical applications, our method has demonstrated a trade-off between measuring costs and large field samples of demand for parameterizing the biomass of a specific species.In some cases,some tree species in the NFIs do not have sufficient biomass observations.Therefore,biomass equations based on fewer samples may have large uncertainties (van Breugel et al., 2011).Our method has provided an alternative option for collecting and grouping the data measured from tree species with similar traits.

    Nevertheless, uncertainty and issues remain.In grouping based on phylogenetic relationship,we found two main issues.One is that species may differ in traits,including if they are phylogenetically close.There are differences in the ratio of crown-to-stem biomass(e.g.,Pinus sylvestris vs.Pinus densiflora) and wood density(e.g., Pinus yunnanensis vs.Pinus dendata;Fig.3).Assigning them into the same group may reduce the model’s goodness-of-fit.In this case,we tend to group the species by wood density, but this may be a challenge if the wood density data are also unsuitable.Second,the phylogenetic trees would be over-pruned because of insufficient data or tree species samples.The clipped phylogenetic tree only contained a relatively small number of tree species,which may have led to two species with different characteristics appearing relatively close in the phylogenetic relationship.This may lead to misclassification of species,the direct effect of which may significantly reduce the fit of the model.During parameterization, for outliers in these species, we considered it necessary to carefully confirm their phylogenetic paths and consider whether grouping based on wood density is appropriate.In addition, some tree species are very different from each other.For example, it is taxonomically controversial whether Ginkgo biloba is a gymnosperm or an angiosperm.The species is difficult to group based on phylogenetic relationship.We tentatively classified this species as Cupressaceae, a group with a long evolutionary history.In addition to phylogeny, temperature, precipitation, soil moisture, and nutrient availability also significantly influence biomass allometric relationships(Chave et al., 2014).If tree species are grouped without considering environmental factors, there may be uncertainty in the modelling.The main reason for these uncertainties is the lack of data,which is difficult to resolve within the short term by expanding biomass datasets using harvesting methods.Our study aimed to maximize the use of measured data and explore improved methods for NFI-based forest biomass modelling.Other uncertainties, such as measurement errors in biomass and wood density, have been discussed in depth in many previous studies (Chave et al.,2009;Henry et al.,2010;Holdaway et al.,2014;Temesgen et al.,2015)and are beyond the scope of this study.

    5.Conclusions

    This study has provided an improved and reliable method for grouping species to establish an allometric biomass model with limited observational data.This grouping was based on the phylogenetic relationship and number of observed biomass records.We matched these observations with those of China’s ninth NFI,and established 22 biomass allometric models.The phylogeny-based biomass models showed a higher model efficiency and lower prediction error(R2=0.97,RMSE=12.9 t?ha–1, %RMSE = 11.5%) than the traditional wood density-based models, demonstrating higher accuracy biomass models for tree species grouping based on phylogenetic relationship.We applied the tree species grouping method to reduce uncertainty in the estimation of forest biomass.The approach presented in this study represents an effective framework for species grouping to develop biomass models that facilitate increasingly accurate national-scale extrapolations and could help reduce the uncertainty in forest carbon budget estimation.We hope that the theoretical approach to tree species grouping presented in this study will contribute to a more quantitative understanding of global forest biomass.

    Funding

    This work was supported by the Science and Technology Innovation Program of Hunan Province (2022RC4027) and the Joint Fund for Regional Innovation and Development of the National Natural Science Foundation of China(U22A20570).

    Ethics approval and consent to participate

    Not applicable.

    Consent for publication

    Not applicable.

    Statement of authorship

    Conceptualization:MY,XZ;Methodology:MY,XZ,TL;Investigation:MY, XZ; Visualization: MY; Funding acquisition: CP; Writing original draft: MY,XZ,TL,KC,CP,ZL, PL,CZ,JT, YZ.

    Declaration of competing interest

    The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

    Acknowledgments

    We thank Profs.Jing Wang and Jiaxiang Li for their constructive suggestions and valuable comments.

    Appendix A.Supplementary data

    Supplementary data to this article can be found online at https://doi.i.org/10.1016/j.fecs.2023.100130.

    日韩欧美在线二视频| 久久久久国产精品人妻aⅴ院| 国产真实乱freesex| 免费人成视频x8x8入口观看| 久久久国产精品麻豆| 欧美黑人欧美精品刺激| 久久精品影院6| 国产高清视频在线观看网站| 19禁男女啪啪无遮挡网站| 淫秽高清视频在线观看| 天天添夜夜摸| 亚洲一码二码三码区别大吗| a在线观看视频网站| 国产成人一区二区三区免费视频网站| 国产成人啪精品午夜网站| xxx96com| 夜夜躁狠狠躁天天躁| 又爽又黄无遮挡网站| 欧美成人免费av一区二区三区| 女人高潮潮喷娇喘18禁视频| 成人三级黄色视频| 久久精品国产清高在天天线| 国产精品日韩av在线免费观看| 亚洲av日韩精品久久久久久密| 亚洲欧美一区二区三区黑人| 青草久久国产| 麻豆成人午夜福利视频| 亚洲人与动物交配视频| 窝窝影院91人妻| 又紧又爽又黄一区二区| 国产野战对白在线观看| 亚洲九九香蕉| e午夜精品久久久久久久| 人妻丰满熟妇av一区二区三区| 国产精品99久久99久久久不卡| 国产亚洲精品第一综合不卡| 午夜亚洲福利在线播放| 首页视频小说图片口味搜索| 欧美绝顶高潮抽搐喷水| 成人三级黄色视频| 久久中文字幕一级| 欧美一区二区国产精品久久精品 | 国产又色又爽无遮挡免费看| 一级黄色大片毛片| 天堂av国产一区二区熟女人妻 | 黑人巨大精品欧美一区二区mp4| 精品少妇一区二区三区视频日本电影| 日本黄色视频三级网站网址| 亚洲av成人不卡在线观看播放网| 中文字幕人成人乱码亚洲影| 亚洲,欧美精品.| 欧美一区二区国产精品久久精品 | 欧美成人免费av一区二区三区| 亚洲欧美日韩高清在线视频| 搡老熟女国产l中国老女人| netflix在线观看网站| 亚洲欧洲精品一区二区精品久久久| 欧美在线黄色| 18禁观看日本| 午夜激情av网站| 欧美精品啪啪一区二区三区| 国产又黄又爽又无遮挡在线| 午夜福利成人在线免费观看| 亚洲第一欧美日韩一区二区三区| 亚洲成人久久爱视频| 亚洲人与动物交配视频| 老司机福利观看| 亚洲欧洲精品一区二区精品久久久| 狂野欧美白嫩少妇大欣赏| 久久九九热精品免费| 母亲3免费完整高清在线观看| 性色av乱码一区二区三区2| 亚洲精品中文字幕一二三四区| 91国产中文字幕| 亚洲第一欧美日韩一区二区三区| 国产av麻豆久久久久久久| 国产在线精品亚洲第一网站| 午夜视频精品福利| 哪里可以看免费的av片| 日韩有码中文字幕| 人人妻人人看人人澡| 成人av一区二区三区在线看| 国产一区二区在线av高清观看| 中国美女看黄片| 又爽又黄无遮挡网站| 特级一级黄色大片| 黄色女人牲交| a级毛片在线看网站| 国产成人啪精品午夜网站| 激情在线观看视频在线高清| 国产高清激情床上av| 母亲3免费完整高清在线观看| 熟女少妇亚洲综合色aaa.| 777久久人妻少妇嫩草av网站| 天天躁狠狠躁夜夜躁狠狠躁| 99国产精品一区二区蜜桃av| 国产高清videossex| 精品不卡国产一区二区三区| 一进一出好大好爽视频| 国产91精品成人一区二区三区| 天天躁夜夜躁狠狠躁躁| 亚洲中文字幕日韩| 日韩欧美 国产精品| 老熟妇仑乱视频hdxx| avwww免费| 免费电影在线观看免费观看| 亚洲av电影在线进入| 禁无遮挡网站| 欧美日韩亚洲国产一区二区在线观看| 黄色a级毛片大全视频| 又黄又爽又免费观看的视频| 麻豆久久精品国产亚洲av| 亚洲成人中文字幕在线播放| 色综合站精品国产| 久久精品成人免费网站| 99久久综合精品五月天人人| 丁香六月欧美| 亚洲美女黄片视频| 精品久久久久久久毛片微露脸| 18禁美女被吸乳视频| 一本久久中文字幕| 禁无遮挡网站| 精品久久蜜臀av无| 日韩大尺度精品在线看网址| 亚洲国产日韩欧美精品在线观看 | 亚洲九九香蕉| 国产精品av视频在线免费观看| 黄色视频,在线免费观看| 亚洲色图 男人天堂 中文字幕| 97超级碰碰碰精品色视频在线观看| 一区二区三区国产精品乱码| 欧美在线黄色| 最近最新中文字幕大全免费视频| 精品日产1卡2卡| 国产伦在线观看视频一区| 人人妻人人澡欧美一区二区| 欧美zozozo另类| 老司机福利观看| 亚洲国产日韩欧美精品在线观看 | 99riav亚洲国产免费| 麻豆av在线久日| 久久久久九九精品影院| 十八禁人妻一区二区| 色av中文字幕| 三级男女做爰猛烈吃奶摸视频| 亚洲美女视频黄频| 一本久久中文字幕| 亚洲国产欧美网| 男人舔奶头视频| 午夜亚洲福利在线播放| 在线播放国产精品三级| 91字幕亚洲| 俄罗斯特黄特色一大片| 欧美一级毛片孕妇| 黄色成人免费大全| 国产黄a三级三级三级人| 欧美日韩一级在线毛片| 亚洲国产欧美网| av中文乱码字幕在线| 国产黄色小视频在线观看| 成人三级做爰电影| 久久香蕉精品热| 欧美黄色片欧美黄色片| 精品无人区乱码1区二区| 91九色精品人成在线观看| 97人妻精品一区二区三区麻豆| 听说在线观看完整版免费高清| 国产欧美日韩一区二区精品| 99国产精品一区二区三区| 国产高清激情床上av| 禁无遮挡网站| 中文字幕高清在线视频| 日韩精品免费视频一区二区三区| 好男人在线观看高清免费视频| 最新美女视频免费是黄的| 国产一区二区三区视频了| 久久久水蜜桃国产精品网| 久久久久性生活片| 国内精品一区二区在线观看| 亚洲男人天堂网一区| 美女 人体艺术 gogo| 无限看片的www在线观看| 亚洲乱码一区二区免费版| 我要搜黄色片| 蜜桃久久精品国产亚洲av| 国产激情欧美一区二区| 久久久精品大字幕| 日韩欧美在线二视频| 日韩大尺度精品在线看网址| 在线a可以看的网站| 12—13女人毛片做爰片一| 天天躁狠狠躁夜夜躁狠狠躁| 丁香欧美五月| 日本五十路高清| 亚洲精品美女久久久久99蜜臀| 色精品久久人妻99蜜桃| 熟女少妇亚洲综合色aaa.| 国产精品亚洲美女久久久| 成人av一区二区三区在线看| 五月玫瑰六月丁香| 色精品久久人妻99蜜桃| 国产成人啪精品午夜网站| 亚洲色图av天堂| 国产又色又爽无遮挡免费看| 久久精品影院6| 后天国语完整版免费观看| 免费观看精品视频网站| 三级男女做爰猛烈吃奶摸视频| 88av欧美| 久久久国产成人精品二区| 777久久人妻少妇嫩草av网站| 在线永久观看黄色视频| 午夜免费激情av| 久久久精品大字幕| 午夜影院日韩av| 日韩大码丰满熟妇| 日日摸夜夜添夜夜添小说| 国产亚洲精品久久久久久毛片| 美女免费视频网站| 国产私拍福利视频在线观看| 国产精品免费一区二区三区在线| 变态另类丝袜制服| 禁无遮挡网站| 欧美一级a爱片免费观看看 | 最近最新中文字幕大全免费视频| 人妻久久中文字幕网| 在线十欧美十亚洲十日本专区| 美女大奶头视频| av视频在线观看入口| 精品久久久久久成人av| 色老头精品视频在线观看| av在线播放免费不卡| 18美女黄网站色大片免费观看| 精品久久久久久久久久久久久| 日本黄色视频三级网站网址| 久久精品影院6| 免费看十八禁软件| 欧美性长视频在线观看| 久久久国产精品麻豆| 欧美大码av| 美女大奶头视频| 操出白浆在线播放| 黄色成人免费大全| 亚洲aⅴ乱码一区二区在线播放 | 在线观看午夜福利视频| 国产成人精品久久二区二区免费| 波多野结衣高清作品| 蜜桃久久精品国产亚洲av| 久久久久亚洲av毛片大全| 男人的好看免费观看在线视频 | 国产精华一区二区三区| 手机成人av网站| videosex国产| 啦啦啦观看免费观看视频高清| 欧美在线黄色| 欧美中文日本在线观看视频| 国产精品久久久av美女十八| 日本a在线网址| 欧洲精品卡2卡3卡4卡5卡区| 亚洲自拍偷在线| 亚洲成av人片免费观看| 一本综合久久免费| 久久香蕉国产精品| 50天的宝宝边吃奶边哭怎么回事| 男人舔女人下体高潮全视频| 可以免费在线观看a视频的电影网站| 人妻夜夜爽99麻豆av| 亚洲国产欧美一区二区综合| www日本在线高清视频| 国产三级黄色录像| 亚洲欧美日韩高清在线视频| 岛国视频午夜一区免费看| 成人亚洲精品av一区二区| 久久久久久人人人人人| 国产av又大| 精品福利观看| 久久精品综合一区二区三区| 一级毛片女人18水好多| 18禁国产床啪视频网站| 法律面前人人平等表现在哪些方面| 波多野结衣高清无吗| 亚洲精华国产精华精| 国产午夜精品论理片| 女生性感内裤真人,穿戴方法视频| 欧美国产日韩亚洲一区| 亚洲国产看品久久| 老司机福利观看| av中文乱码字幕在线| 成人av在线播放网站| 精品不卡国产一区二区三区| 18美女黄网站色大片免费观看| 不卡av一区二区三区| 久久精品综合一区二区三区| 国产精品野战在线观看| 亚洲精品国产精品久久久不卡| 黄色丝袜av网址大全| 人人妻,人人澡人人爽秒播| 麻豆成人午夜福利视频| 悠悠久久av| 亚洲五月婷婷丁香| 日本黄色视频三级网站网址| 日本一二三区视频观看| 成人欧美大片| 国产成人系列免费观看| 亚洲一码二码三码区别大吗| 欧美成人免费av一区二区三区| 中文亚洲av片在线观看爽| 久久精品综合一区二区三区| 国产成人欧美在线观看| 欧美日本视频| 午夜影院日韩av| 色精品久久人妻99蜜桃| 国产精品电影一区二区三区| 在线视频色国产色| 18禁裸乳无遮挡免费网站照片| 欧美乱码精品一区二区三区| 国内揄拍国产精品人妻在线| 真人一进一出gif抽搐免费| 香蕉久久夜色| 精品国产美女av久久久久小说| 国产熟女午夜一区二区三区| 欧美乱色亚洲激情| 99久久精品热视频| 国产熟女午夜一区二区三区| 国产精品乱码一区二三区的特点| 国产av一区二区精品久久| 在线观看免费午夜福利视频| 老司机深夜福利视频在线观看| 国产精品爽爽va在线观看网站| 成人欧美大片| 亚洲av日韩精品久久久久久密| 脱女人内裤的视频| 久久香蕉国产精品| 男男h啪啪无遮挡| 天堂√8在线中文| 久久人妻av系列| 久久精品国产亚洲av香蕉五月| 欧美一级毛片孕妇| 在线观看一区二区三区| 亚洲男人的天堂狠狠| 国产精品久久久av美女十八| 成熟少妇高潮喷水视频| 国产主播在线观看一区二区| 黄色丝袜av网址大全| 草草在线视频免费看| 精品午夜福利视频在线观看一区| 精华霜和精华液先用哪个| 97超级碰碰碰精品色视频在线观看| 伊人久久大香线蕉亚洲五| 麻豆av在线久日| 搡老熟女国产l中国老女人| 一a级毛片在线观看| 俺也久久电影网| 国产高清videossex| 久久久久久人人人人人| 18禁美女被吸乳视频| 50天的宝宝边吃奶边哭怎么回事| 国产一区在线观看成人免费| 后天国语完整版免费观看| 国产成+人综合+亚洲专区| 欧美午夜高清在线| 香蕉国产在线看| 无限看片的www在线观看| 亚洲美女黄片视频| 黄片大片在线免费观看| 女人爽到高潮嗷嗷叫在线视频| 国内揄拍国产精品人妻在线| 免费在线观看黄色视频的| 欧美日韩精品网址| 精品久久久久久久毛片微露脸| 亚洲国产精品久久男人天堂| 日韩欧美 国产精品| 色播亚洲综合网| 亚洲午夜理论影院| 在线永久观看黄色视频| 97超级碰碰碰精品色视频在线观看| 国产亚洲精品久久久久久毛片| 俄罗斯特黄特色一大片| 91老司机精品| 男女下面进入的视频免费午夜| 日韩有码中文字幕| 亚洲人成网站在线播放欧美日韩| 18禁美女被吸乳视频| 天天一区二区日本电影三级| 久久久久久久精品吃奶| 又紧又爽又黄一区二区| tocl精华| 此物有八面人人有两片| www国产在线视频色| 巨乳人妻的诱惑在线观看| www国产在线视频色| 欧美在线黄色| 级片在线观看| 国产av不卡久久| 精品久久久久久久人妻蜜臀av| 琪琪午夜伦伦电影理论片6080| 国产成人精品久久二区二区免费| 91av网站免费观看| 亚洲一区中文字幕在线| 18禁裸乳无遮挡免费网站照片| 欧美三级亚洲精品| 成人av在线播放网站| 男人舔女人的私密视频| 桃色一区二区三区在线观看| 亚洲国产看品久久| 国产aⅴ精品一区二区三区波| 黄片小视频在线播放| 亚洲欧美日韩高清在线视频| 亚洲无线在线观看| 日韩国内少妇激情av| 两个人视频免费观看高清| 成年版毛片免费区| 俺也久久电影网| 色哟哟哟哟哟哟| 久99久视频精品免费| 91字幕亚洲| 夜夜夜夜夜久久久久| 美女免费视频网站| 正在播放国产对白刺激| 99在线人妻在线中文字幕| 中文字幕高清在线视频| 亚洲七黄色美女视频| 亚洲av五月六月丁香网| 久久性视频一级片| 国产三级中文精品| 久久草成人影院| 黑人巨大精品欧美一区二区mp4| 久久久精品欧美日韩精品| 黄色毛片三级朝国网站| 嫩草影院精品99| 亚洲最大成人中文| 俺也久久电影网| 亚洲一区二区三区色噜噜| 操出白浆在线播放| 美女大奶头视频| 国产久久久一区二区三区| 欧美黄色片欧美黄色片| 国产成人精品久久二区二区免费| 国产单亲对白刺激| 一区福利在线观看| 国产成人av激情在线播放| 亚洲中文av在线| 一本综合久久免费| 久久天躁狠狠躁夜夜2o2o| 夜夜看夜夜爽夜夜摸| 99在线视频只有这里精品首页| 亚洲 国产 在线| e午夜精品久久久久久久| 国产一区二区在线观看日韩 | 亚洲av第一区精品v没综合| 亚洲欧美精品综合一区二区三区| 最近在线观看免费完整版| 久久久久亚洲av毛片大全| 亚洲第一欧美日韩一区二区三区| 午夜免费观看网址| 男男h啪啪无遮挡| 一夜夜www| 国产成人一区二区三区免费视频网站| 在线十欧美十亚洲十日本专区| 欧美黑人欧美精品刺激| svipshipincom国产片| 丁香欧美五月| 桃色一区二区三区在线观看| 男人舔奶头视频| 亚洲一卡2卡3卡4卡5卡精品中文| 国产精品一区二区精品视频观看| 18禁美女被吸乳视频| 中文资源天堂在线| 亚洲无线在线观看| 国产av麻豆久久久久久久| 深夜精品福利| 久久午夜综合久久蜜桃| 国产精品久久久人人做人人爽| 可以免费在线观看a视频的电影网站| 男人舔女人的私密视频| 麻豆国产av国片精品| 熟女电影av网| 久久精品91无色码中文字幕| 国产黄片美女视频| 99久久无色码亚洲精品果冻| 熟妇人妻久久中文字幕3abv| 国产精品精品国产色婷婷| 曰老女人黄片| 亚洲色图 男人天堂 中文字幕| 国产男靠女视频免费网站| 欧美中文综合在线视频| 国产黄a三级三级三级人| 搞女人的毛片| 午夜免费激情av| 国产精品久久久久久亚洲av鲁大| 老司机福利观看| 深夜精品福利| 一本大道久久a久久精品| 中国美女看黄片| 天天躁夜夜躁狠狠躁躁| 欧美极品一区二区三区四区| 日本精品一区二区三区蜜桃| 搡老熟女国产l中国老女人| 精品久久久久久久末码| 久久久久久九九精品二区国产 | 日韩欧美一区二区三区在线观看| 可以在线观看毛片的网站| 91av网站免费观看| 精品国产亚洲在线| 成人国语在线视频| 亚洲av美国av| 好看av亚洲va欧美ⅴa在| 久久精品成人免费网站| 久久久久亚洲av毛片大全| 国产久久久一区二区三区| 亚洲精品粉嫩美女一区| 欧美精品啪啪一区二区三区| 最近在线观看免费完整版| 成人三级黄色视频| 黄色a级毛片大全视频| 91av网站免费观看| 久久久久久免费高清国产稀缺| aaaaa片日本免费| 国产精品久久久久久精品电影| 成人精品一区二区免费| 色播亚洲综合网| 国内少妇人妻偷人精品xxx网站 | 变态另类成人亚洲欧美熟女| 99久久国产精品久久久| 亚洲精品久久成人aⅴ小说| 亚洲精品av麻豆狂野| 久久中文字幕一级| 动漫黄色视频在线观看| 亚洲免费av在线视频| 丰满人妻一区二区三区视频av | 可以免费在线观看a视频的电影网站| 美女黄网站色视频| 国产熟女xx| 亚洲精品久久国产高清桃花| 人人妻人人澡欧美一区二区| 精品第一国产精品| 免费看美女性在线毛片视频| 动漫黄色视频在线观看| 少妇的丰满在线观看| 88av欧美| 少妇熟女aⅴ在线视频| 久久久国产成人免费| 日韩 欧美 亚洲 中文字幕| 99精品在免费线老司机午夜| 99国产精品一区二区蜜桃av| 日本免费一区二区三区高清不卡| 欧美日韩亚洲国产一区二区在线观看| 国产精品九九99| 欧美精品亚洲一区二区| 免费在线观看成人毛片| 久久天堂一区二区三区四区| 色综合婷婷激情| 18禁裸乳无遮挡免费网站照片| 午夜影院日韩av| 50天的宝宝边吃奶边哭怎么回事| а√天堂www在线а√下载| 每晚都被弄得嗷嗷叫到高潮| 欧美另类亚洲清纯唯美| videosex国产| 亚洲人成网站在线播放欧美日韩| 国产69精品久久久久777片 | 在线观看免费午夜福利视频| 久久精品国产亚洲av香蕉五月| 亚洲午夜理论影院| 91国产中文字幕| www日本黄色视频网| 国产99久久九九免费精品| 大型av网站在线播放| 午夜福利高清视频| 国产高清videossex| 国内毛片毛片毛片毛片毛片| 国产99白浆流出| 欧美日韩亚洲国产一区二区在线观看| 视频区欧美日本亚洲| 亚洲专区国产一区二区| 成年免费大片在线观看| 亚洲精品色激情综合| 色老头精品视频在线观看| 国产成人av教育| 18禁国产床啪视频网站| 天天一区二区日本电影三级| 国产成人aa在线观看| 香蕉av资源在线| 两性午夜刺激爽爽歪歪视频在线观看 | 欧美最黄视频在线播放免费| 亚洲黑人精品在线| 国产精品久久久久久亚洲av鲁大| 两个人看的免费小视频| 18禁美女被吸乳视频| 精品电影一区二区在线| 中文在线观看免费www的网站 | 妹子高潮喷水视频| 99热这里只有精品一区 | 免费搜索国产男女视频| 日本五十路高清| 欧美一级a爱片免费观看看 | 久久久久久九九精品二区国产 | 一个人免费在线观看的高清视频| 女同久久另类99精品国产91| 欧美黄色淫秽网站| 长腿黑丝高跟| 三级男女做爰猛烈吃奶摸视频| 成人国语在线视频| 老汉色av国产亚洲站长工具| 一二三四在线观看免费中文在| 女人被狂操c到高潮| АⅤ资源中文在线天堂| 国产黄片美女视频| 国产精品亚洲av一区麻豆| 天堂动漫精品| av福利片在线观看| 国产成人欧美在线观看| 欧美一区二区国产精品久久精品 | 一级片免费观看大全| 亚洲成人精品中文字幕电影|