Evaluation of Multiple Forest Service Based on the Integration of Stand Structural Attributes in Mixed Oak Forests

: In order to understand forest services at stand level through the integration of structural attributes, forest structures in three main stand types were analyzed based on various structural attributes relating to the services of habitat conservation, timber production and soil water conservation in Loess Plateau, China. Forty sample plots with similar site and environment conditions were established in three types of oak stands. Twenty-two indexes such as stand density, mean DBH, mean height, etc., were selected to analyze the relationship between structural attributes and forest service. With a core set of structural attributes selected by principal component analysis, the link between the service and structural attributes and the compatibility between each service was analyzed using correlation analysis. The results show that the oak–broadleaf mixed forest has the highest structural complexity index (SCI), indicating the best habitat conservation among others, while the oak–pine mixed forest and pure oak forest have the highest timber value in different scenarios of timber price, indicating the largest potential for timber production. Little differences were found between stand types regarding the indicators for soil water conservation. The structural attributes that had a positive correlation with habitat conservation (e.g., number of shrub species, species richness of canopy layer, the proportion of broadleaves and snag density) were identiﬁed to be negative for timber production; while the attributes that had a positive correlation with timber production (e.g., stand density and proportion of pine and oaks) were found to be negative for habitat conservation. The results of the trade-off analysis showed that timber production tends to be conﬂicting with the other two non-timber forest services. In order to enhance the provisioning of multiple services, it was suggested to implement the interventions that could balance these services, such as structural retention and single-tree selective logging. This study could contribute to the theoretical base for the decision making in the multi-purpose sustainable management of oak forests in China.


Introduction
As the human population grows, demands for ecosystem services in general, and thus also for forest goods and services, are increasing [1][2][3].This requires maintaining a high standard of forest management and a flexible adaptation of the multiple uses of forests to the complex interactions between the private and public sectors [4,5].Multiple service forestry (MSF) is considered to be a relatively balanced and sustainable strategy to satisfy these diverse demands for forests from multiple stakeholders under the sustainable forest management paradigm [6][7][8].This could contribute to mitigating forest degradation as well as social conflicts on forest resources by improving the productivity of forest lands and diversifying forest uses [4].
Sustainability 2022, 14, 8228 2 of 19 In order to implement the concept and strategy of MSF in practice, some hold that it is achieved usually by segregating productive and protective oriented forest management units at a landscape level, specializing in single dominant uses such as timber production or nature reserve, while others argue that integration of different services into each forest unit at stand level is also possible, such as retention of habitat trees in the managed-forest matrix at a small scale [9].In a segregation system, forest types are either isolated small areas for the preservation of primary forests (e.g., national parks and forest reserves) or intensively managed forests for production objectives (e.g., plantations), where the habitat quality is usually poor.In an integration system, structural retention and restoration measures are an integral part of sustainable forest practices [10].The forest ecosystem provides a variety of services and goods, while there could be conflicts between multiple services in one particular forest, especially between the services of timber production and nature conservation.In order to put MSF into practice, it is essential and effective to identify and work with structural elements within forest ecosystems.For further assessment and management practices, it is fundamental to have a good understanding of the concept of structural elements and their relevance to ecosystem services [5].
Increasing attention has been paid to developing multiple service forestry in China in recent decades [11].In China, the traditional system of forest function division, in which the forest lands are categorized into ecological forests and commercial forests, has been criticized for being unable to adapt to the multiple demands of different sectors in modern society.Hence, it is particularly necessary to develop the management theories and techniques of multi-functional forestry and put them into practice in order to improve the quality of forests and meet the diverse demand for forest services with the rapid development of the economy and society as a whole [7,[12][13][14].
The Loess Plateau is a typical ecologically fragile area with severe soil erosion problems throughout its history in northwest China.Soil water conservation is the main management goal, with fewer values placed on the other forest services, such as timber production and biodiversity conservation [15,16].In the past 40-50 years, a large number of plantations (Oak, Pine and others) have been afforested to reduce soil erosion in the Loess Plateau.Under the Natural Forest Protection Project (NFPP) in 2000, these plantations were all regarded as the same management mode as natural forests, which was strict protection.Most plantations underwent the transition from young to middle-aged [17].The lack of effective management activities (thinning, pruning) aggravated the individual competition and reduced the growth of stands, made these forests unable to have the same strong ecological protection function as natural forests, and also affected their other functions of wood production and diversity maintenance.
Oak forest is one of the main forest types in the Loess Plateau region [18]; like other artificial forests, strict forest protection destroys its stable forest structure and hinders its multiple functions.It is of great importance to investigate the multiple service management linking to stand structural attributes.The overall objective of this research was to study how to combine the three desired services-habitat conservation, soil water conservation and timber production-at the stand-level through the integration of service-supporting structural attributes [1], and they were determined according to the current management goal and potential or actual demands for the forests in the study area.

Site and Data Description
The study was conducted in Shuanglong forest farm, Qiaoshan Forest Bureau, Huangling County (latitude:   12 E).This region belongs to the zone of warm temperate climate zone, and the altitude is between 900 and 1500 m.The monthly average temperature ranges from −5.5 • C in January to 23.1 • C in July, and the annual average precipitation is about 683.2 mm.The soil types are mainly forest cinnamon soil and grey cinnamon soil.The total management area of Shuanglong forest farm is 33,333 ha, standing stock volume 1,720,000 cubic meters and forest cover 95.5%.Quercus acutissima forest, which is a zonal vegetation type, pure Q. acutissima forest; Q. acutissima-broadleaf mixed forest; and Q. acutissima-Pinus tabulaeformis mixed forest are the most important three forest types.Based on the similar afforestation and silviculture history, the growth stage of the above three types of oak forests was nearly the same in plantations.
In order to avoid the impact of the environment on the relationship between stand structural attributes and service functions, 40 sample plots were set in areas with similar site conditions (similar aspect and slope position, altitude variation within 100 m, gradient variation within 5 • ) as far as possible.We established 40 sample plots in the oak forest.These sample plots were randomly distributed within each stand type (Table 1 and Appendix A Table A1).These plots were of rectangular shape with size of 20 m × 30 m.There were three 5 m × 5 m subplots systematically established at the upper-right corner, middle and lower-left corner in each plot, along with one 1 m × 1 m quadrat at the lower-left corner per subplot [19].In each plot, tree species name, diameter at breast height (DBH, breast height = 1.3 m) and height (for trees with a DBH ≥ 5 cm) was measured using a tape meter and ultrasonic altimeter (Vertex III).Canopy density was estimated by means of the proportion of the sample points covered by the forest canopy in the total sample points (100 sample point systematic distribution in sample plot).The number of standing deadwood (snags, with DBH ≥ 5 cm) except for plant diseases and insect pests was counted and converted to the number per hectare.The species names of plants in both shrub layer and herb layer were identified and recorded in subplots, and indexes of species cover and abundance of understory vegetation were also recorded.
Soil samples for analyzing the water content and nutrient availability were collected from five points around the four corners and the center of the plot (but outside the subplots) at a depth of 0-20 cm by using a soil auger.Then these samples were merged evenly as one mixed soil sample for a plot.The soil nutrient availability of each mixed sample, including organic matter content, available and total nitrogen, phosphorus and potassium content, was analyzed with potassium dichromate oxidation, the Kjeldahl method and the colorimetric method [20].The soil water content was measured by the weight difference between wet weight and dry weight of soil samples [20].The depth of the litter and humus layer was measured with a steel ruler.
The quantity of coarse woody debris (CWD) is an important structural element on the ground layer.Due to the complexity of the formation of CWD (species, time, external environment, etc.), the CWD with a similar decomposition state, which was partly solid and breakable, were selected in this study.The number of coarse woody debris was counted in the entire plot according to three diameter classes: small (7.5-20 cm), medium (20-40 cm) and large (>40 cm) [21], and then the number of these CWDs was converted to number per hectare.

Conceptual Framework
The conceptual framework demonstrates the methodological pattern of this study, as shown in Figure 1.It addresses the whole process (bottom-up) of the forest management loop from (1) data collection of inventory parameters based on the main forest types in the study area to (2) comparative analysis of the forest structure by stand types; (3) assessment of the three determined forest services by stand types to (4) analysis of the link between structural attributes and the services and trade-off analysis between each service; finally, to (5) recommendation on silvicultural measures based on the integration of structural elements for the optimization of multiple services in a forest according to the implication from analysis and assessment.

Conceptual Framework
The conceptual framework demonstrates the methodological pattern of this study, as shown in Figure 1.It addresses the whole process (bottom-up) of the forest management loop from (1) data collection of inventory parameters based on the main forest types in the study area to (2) comparative analysis of the forest structure by stand types; (3) assessment of the three determined forest services by stand types to (4) analysis of the link between structural attributes and the services and trade-off analysis between each service; finally, to (5) recommendation on silvicultural measures based on the integration of structural elements for the optimization of multiple services in a forest according to the implication from analysis and assessment.It is an integrated assessment of forest ecosystem services based on the analysis of a set of service-supporting structural attributes of the three main stand types-pure oak stand, oak-broadleaf mixed stand and oak-pine mixed stand.The forest structure was analyzed using a set of structural attributes (secondary indicators), and the objective services were assessed using specific service indicators (primary indicators).As shown in Figure 1, a variety of structural attributes that were relevant for supporting the objective services according to the literature review were selected and calculated based on a wide range of inventory parameters measured in the field.Then with a core set of structural attributes selected by principal component analysis, the interrelationship between structural attributes and service indicators was analyzed.

Determination and Calculation of Structural Attributes
A wide variety of structural attributes that were considered to be relevant for supporting the particular objective services were selected for analyzing the forest structure and the link between structure and services.These attributes consist of stand density, mean DBH, mean height, the height of dominant trees, tree size diversity, tree height diversity, snags density, vegetation cover, number of species, species richness, relative abundance (proportion) of dominant species for all layers and relative abundance (proportion) of conifers and other broadleaf species, CWD density and litter depth.They are all considered to be relevant for supporting the objective services in this study (habitat It is an integrated assessment of forest ecosystem services based on the analysis of a set of service-supporting structural attributes of the three main stand types-pure oak stand, oak-broadleaf mixed stand and oak-pine mixed stand.The forest structure was analyzed using a set of structural attributes (secondary indicators), and the objective services were assessed using specific service indicators (primary indicators).As shown in Figure 1, a variety of structural attributes that were relevant for supporting the objective services according to the literature review were selected and calculated based on a wide range of inventory parameters measured in the field.Then with a core set of structural attributes selected by principal component analysis, the interrelationship between structural attributes and service indicators was analyzed.

Data Analysis 2.3.1. Determination and Calculation of Structural Attributes
A wide variety of structural attributes that were considered to be relevant for supporting the particular objective services were selected for analyzing the forest structure and the link between structure and services.These attributes consist of stand density, mean DBH, mean height, the height of dominant trees, tree size diversity, tree height diversity, snags density, vegetation cover, number of species, species richness, relative abundance (proportion) of dominant species for all layers and relative abundance (proportion) of conifers and other broadleaf species, CWD density and litter depth.They are all considered to be relevant for supporting the objective services in this study (habitat conservation, soil water conservation and timber production).Some of the structural attributes were calculated by the corresponding equations listed in Table 2.

Assessment of Forest Services Based on Service Indicators
The three determined objective services-habitat conservation, timber production and soil water conservation-were assessed in a comparative way by stand type, using a set of service indicators.Structural complexity index (SCI) is a key factor in ecology, often positively linked to biodiversity and the carrying capacity of habitats [24,25].It was selected as the evaluating indicator of habitat conservation in this study.In addition, timber value for assessing the service of timber production and soil properties for assessing the service of soil water conservation [26].These service indicators were determined by different methods: Structural Complexity Index: the structural complexity index was calculated by means of average on the basis of the leading variables that were indicated to be important for habitat diversity according to the results of PCA [25].

SCI = SW I d + SW I h + SW I canopy + SW I shrub + SW I herb n
where SW I d refers to tree size diversity; SW I h refers to tree height diversity; SW I canopy refers to (Shannon-Wiener Index) species richness in canopy layer; SW I shrub refers to (Shannon-Wiener Index) species richness in shrub layer; SW I herb refers to (Shannon-Wiener Index) species richness in herb layer; n refers to the number of structural attributes used in the index (n = 5).Timber Value: For the calculation of timber value, information on the timber price of oaks and pine in China and Europe was also acquired to estimate the potential timber value of the forests using different scenarios; TV = V pine * P pine + V oak * P oak A where TV refers to timber value per ha (EUR/ha); V pine refers to the total timber volume of pine trees (m 3 ); P pine refers to the timber price of pine (EUR/m 3 ); V oak refers to the total timber volume of oak trees (m 3 ); P oak refers to the timber price of oak (EUR/m 3 ); A refers to plot area (ha).Soil properties: the soil properties fundamental for analyzing soil quality were directly used as the (proxy) indicators for assessing the service of soil water conservation.
The values of the above-mentioned service indicators were calculated for each sample plot for the integrated assessment of forest services.The differences in these service indicators between stand types were depicted by boxplots for each service, and the significance of differences in the service indicators between each stand type was tested using the method of single-factor analysis of variance (one-way ANOVA).The pairwise differences between stand types were then analyzed using Tukey's HSD test.All analyses were implemented in R version 4.1.2(R Foundation for Statistical Computing, Vienna, Austria, 2021) [27].

The Interrelation between Structural Attributes and Service Indicators
The interrelation between structural attributes and service indicators by Principal Component Analysis (PCA).This analysis was conducted in R based on the data of all the original variables related to structural attributes and soil parameters.By this means, the variables that contributed largely to the variance of the system and were independent of one another were selected as the leading variables for further analysis.
The method of correlation analysis was also used for identifying the links between a set of selected structural attributes and each determined service in order to provide some hints for the recommendations on what and how structural elements should be incorporated in a stand for optimizing the multi-functionality of the forests.If the correlation coefficient between a particular structural attribute and a service-service indicator is positive, it indicates that improving this structural attribute to some extent could support the relevant service; while if their coefficient is negative, implying that the reduction in this structural attribute might be necessary for enhancing the desired service.

The Structural Attributes of Three Oak Forest Types
A variety of structural attributes for each stand type is summarized in Table 3.The oak-broadleaf mixed forest with the most tree species number had the highest level of over-story species richness (1.069), followed by the oak-pine mixed forest with a slightly lower value (1.019) and the pure oak forest (0.501).The species richness in oak-broadleaf mixed forests, where there were up to 12 plant species on average, was notably higher (1.873) than in the other two types of forest (1.498 and 1.594, respectively).The proportions of dominant species in the shrub layer in all the stand types were remarkably smaller than in the canopy and herb layer in general, ranging from about 35% to 45%.The smaller the proportion of dominant species, the larger the share of other species in a forest community.By contrast, there was basically less number of plant species in the herb layer, between four and six species, and the proportions of dominant herb species were up to around 80% in all stand types.There was no great difference in species richness of herb layer between different stand types, with around 0.45 on average.In terms of the number of snags (standing dead wood), remarkable differences were found between stand types.The highest snags density of 41 stems per ha was observed in the oak-broadleaf mixed forest, while only seven stems per ha could be found in the oak-pine mixed forest.There was also a notable difference regarding the coarse woody debris (CWD) among stand types.The density of CWD in the pure oak forest (341 pieces per ha) was higher than the other stand types, while the oak-broadleaf had the least amount of lying dead wood (217 pieces per ha).
Other stand structural attributes such as stand density, mean DBH, diversity of DBH, mean height, diversity of height, etc., are listed in Table 3.

Forest Services of Three Oak Forest Types
By comparatively analyzing the average values of the ecosystem goods and services (ESS) indicators between each stand type, an overview of the difference in providing services by these stand types was obtained (Table 4).The oak-broadleaf mixed stand had the highest structural complexity index among all the stand types (1.123), while its timber value was the lowest (4045.72EUR/ha) when the timber price of oak is assumed to be lower than pine in the case of China.In contrast, the oak-pine mixed stand had the highest timber value in the scenario with reference timber prices in China, which was 6563.16EUR/ha, while the soil water content was notably lower (20.06%)than that in the other two stand types.In the pure oak stand, despite the fact that the SCI was the lowest with a subtle difference, the timber value in the scenario with reference timber price in Europe was at the highest level among all the types of forest, as well as the soil water content (Appendix A, Table A2).

The Interrelation between Structural Attributes and Service Indicators
The Core Structural Attributes by Principal Component Analysis The summary table of PCA (Table 5) shows the first nine principal components, which have cumulative variance beyond 80%, with eigenvalues above 1 (except PC9).Among them, the first two principal components, PC1 and PC2, have proportions of variance (eigenvalue) greater than 15% (20.1% and 15.6%, respectively), and their eigenvalues are above 2.This means that the top two principal components contributed most to the variation in the system, so basically, they can provide a simpler description of the system based on the combined set of variables.Hence, the following analysis of PCA is mainly focused on these two components.In order to reduce the number of variables for further analysis, those variables that have high loadings (above 0.2) in relation to the principal components 1 and 2 in PCA were selected.Some variables were still retained if they could characterize the forest structure in different aspects (e.g., stand volume and dominant height), even though they are correlated.Therefore, based on these standards, 10 leading variables were selected, including: (1) species richness (SWI) in the canopy layer, (2) species richness (SWI) in the shrub layer, (3) species richness (SWI) in the herb layer, (4) snags density, (5) mean DBH, (6) dominant tree height, (7) stand density, (8) canopy cover, (9) shrub cover, (10) litter depth.
The biplot of PCA (see Figure 2) visualizes the interrelationship among all the original structural attributes and between these attributes and the principal components 1 and 2. Principal component 1 (PC1) is mainly a combination of seven variables (indicated in the green circle in the graph), including the stand density, canopy cover, proportion of pine, litter depth, available N, available K and organic matter content.
The matrix of loadings of PCA (Table 6) displays the corresponding loadings of these variables in relation to the PC1, which are 0.275, 0.206, 0.241, 0.183, 0.267, 0.273 and 0.244, respectively.The graph also shows that on the left-hand side (negative side of PC1), there is a group of variables mainly related to the horizontal and vertical structure of forests (indicated in the green circle in the graph), including mean DBH, diversity of DBH, mean height, diversity of height, dominant tree height, and stand volume, etc. Apart from these, the principal component 2 (PC2) is highly associated with the variables related to habitat diversity (indicated in the yellow circle in the graph), which are the snags density (with loadings of 0.282), the proportion of broadleaf (0.281), species richness (SWI) in canopy layer (0.220), shrub layer (0.376) and herb layer (0.262), and the number of species in each layer (0.204, 0.362, 0.293, respectively).They provide a basis for the selection of parameters for establishing the index for assessing the habitat service, namely the structural complexity index (SCI) (Appendix A Table A3).

Correlation between Structural Attributes and the Service Indicators
As is shown in the correlation matrix below (Table 7), the variables that had the most significant (p < 0.01) positive correlation with SCI were species richness of shrub layer (0.7522), species number of shrub layer (0.5942), species richness of herb layer (0.5848), the proportion of broadleaf species (0.5414) and species richness of canopy layer (0.5330).The variables, which include species number of herb layer (0.4642), tree size diversity (0.4577), snag density (0.4564) and mean DBH (0.4444), had a less significant (p < 0.05) positive relationship with SCI.Only a few structural attributes had a negative correlation with SCI.The most significant variables were the proportion of dominant shrub (−0.6310) and the proportion of dominant herb (−0.5036).The stand density (−0.3526) and proportion of pine (−0.2828) had a less negative correlation with SCI.The correlation of other structural indexes with SCI was not significant.

Correlation between Structural Attributes and the Service Indicators
As is shown in the correlation matrix below (Table 7), the variables that had the most significant (p < 0.01) positive correlation with SCI were species richness of shrub layer (0.7522), species number of shrub layer (0.5942), species richness of herb layer (0.5848), the proportion of broadleaf species (0.5414) and species richness of canopy layer (0.5330).The variables, which include species number of herb layer (0.4642), tree size diversity (0.4577), snag density (0.4564) and mean DBH (0.4444), had a less significant (p < 0.05) positive relationship with SCI.Only a few structural attributes had a negative correlation with SCI.The most significant variables were the proportion of dominant shrub (−0.6310) and the proportion of dominant herb (−0.5036).The stand density (−0.3526) and proportion of pine (−0.2828) had a less negative correlation with SCI.The correlation of other structural indexes with SCI was not significant.
There were only two variables that had a significant (p < 0.01) positive correlation with timber value, which were stand volume per ha (0.9001) and proportion of pine (0.5288).In addition to these, the stand density (0.4222), mean height (0.4626) and mean height of the dominant tree (0.4845) had a less significant (p < 0.05) positive correlation with timber value.On the other hand, the timber value had a less significant (p < 0.05) negative correlation with the proportion of broadleaf (−0.4516).The correlation of other structural indexes with timber value was not significant.
Herb cover was the structural attribute that had the most significant (p < 0.01) positive correlation with soil water content, with a coefficient of 0.5119, and next was litter depth (0.4247) which had a less significant (p < 0.05) positive correlation.The effects of other attributes contributed to higher soil water content (mean height of dominant trees, shrub cover, diversity of height, etc.) were not significant.
Variables that had negative links with soil water content included the stand density (−0.3593), canopy cover (−0.2519) and proportion of pine (−0.3147) had no significant negative correlation with soil water content.
Moreover, the relations between each ESS indicator were not remarkable, especially the correlation between SCI and soil water content, with a coefficient of 0.0242.Nevertheless, some negative, though not significant, correlations between timber value and the other ESS indicators for habitat service and soil water conservation service.Plots with higher timber values appeared to have relatively lower SCI, soil water content and organic matter, in spite of their low coefficients (−0.1076, −0.0412 and −0.0480, respectively).

Link between Stand Structural Attributes and Forest Services
Our results show the relationship between stand structural attributes and main forest services, which demonstrate the importance of considering multiple forest attributes to understand the drivers of ecosystem service [5,28].Because forest services may be related to the combined effect of stand attributes and environmental factors, and the environmental factors were complicated and less amenable to management [5] the stands with the similar site and environmental conditions were selected in our study to avoid the impact of environmental and site conditions.
Structural complexity index (SCI) is a key factor in ecology, often positively linked to biodiversity and the carrying capacity of habitats [24,25].Our results clearly show that certain forest attributes, especially species richness of arbor, shrub and herb, species number of shrub and thw proportion of broadleaf tree species, had, on average, positive effects on SCI.This implies that more diverse plant species in all the strata, especially in the shrub layer, and a larger proportion of broadleaf trees would significantly contribute to the structurally more complex forests, which could provide better quality conditions for habitat service, especially for some small animals and birds [29][30][31].Moreover, other structural attributes such as species number of canopy and herb layer, tree size diversity, snag density and mean DBH, which had a great correlation with habitat heterogeneity and a noticeable positive association with this index [3,32].Nonetheless, the correlations between some attributes (e.g., litter depth, CWD density, canopy density, etc.) and SCI were not closely related in our results; it does not mean that these indexes are not relevant for wildlife conservation [33][34][35][36].Instead, we can imply from the structural analysis in our case that the oak-pine mixed forest that had the largest amount of coarse woody debris and thickest layer of litter, though not the highest SCI, may offer the resource required for ground foragers.Apart from these positive attributes, this study also found that variables such as the proportion of dominant shrub, dominant herb, stand density, etc., which always decreased the heterogeneity of forest structure, had the most significant negative on SCI.These findings could provide a base for recommending which structural elements should be reduced in order to improve the habitat service in a forest.
It is self-evident that stand volume is highly dependent on the tree diameter, tree height and stand density.Additionally, the height diversity and DBH diversity were also found to be positively linked to timber value, implying that improving the diversity of horizontal and vertical structures could bring about higher timber values by increasing the forest productivity [37,38].On the other hand, the negative correlation between the proportion of broadleaf, oak and timber value may cause by the market acceptance and usage history of these species in china.Other structural attributes such as species richness of arbor, shrub, species number of shrub layer, etc., which make the stand heterogeneity stronger, had a negative association with the timber value.
Plant species diversity can enhance soil conservation, but this effect was comparatively weak relative to the contributions of plant cover and density to soil conservation [39,40].We found that understory cover, especially herb cover, had a positive correlation with soil water conservation, while arbor cover had a negative effect on soil water conservation.The result of correlation analysis on the structure-service link also illustrates that the soil water content had a positive correlation with the proportion of oaks and accordingly had a negative correlation with the proportion of pine, which is similar to others' research [41].As plant litter falls to the ground, it forms an assembly developing a porous barrier that is structured by wind and water [42]; litter acts as a sponge, reducing surface runoff and soil erosion, especially in areas with serious soil erosion [43].Our results show litter depth has a significant positive correlation with soil water conservation.The higher amount of CWD in a forest is supposed to generate more soil water content for the reason that these residuals could facilitate the process of holding back the rainfall [21], while this result is not obvious in our study.

Compatibility between Each Forest Service
Trade-offs and synergies in the supply of forest ecosystem services are common, but the drivers of these relationships are poorly understood [3].According to the findings in this research, trade-offs exist between timber production and the other non-timber services, whereas these two services are relatively compatible with each other at the stand level.In order to optimize the production of marketable timber, management typically focuses on growing even-aged, homogenous forest stands [44][45][46].However, such management can reduce the supply of other ecosystem services, such as carbon sequestration [47], water availability [48,49] and biodiversity conservation [50].Despite that, both competitive and compatible relations are not significant.The results of the statistical analysis show that the timber value was negatively correlated with the other indicators used for the assessment of habitat service and soil water conservation service [3].
In the oak-broadleaf mixed forest, the species richness in the canopy layer is higher than in the oak-pine mixed forest and pure oak forest, which is beneficial for providing habitat for more diverse fauna or birds [29][30][31][32][33], and the considerable mixture of coniferous species contributes to the litter cover that is good for retaining water in the soil [42,43], while the proportion of commercial tree species (pine and oak in this case) is smaller in such stands.According to the study by the United States Department of Agriculture (USDA) on the compatibility of multiple forest uses [51], determining whether, as well as to what extent, timber production may be complementary with other non-timber services remains a challenge for the research of trade-off and compatibility of production in multi-resource forest management.
Nelson et al. found the negative correlation between commodity production and biodiversity conservation as well as ecosystem services as a clear trade-off in the study on modeling the trade-offs among multiple services at a landscape level.This might trigger the risk of shifting land cover towards more production-oriented and less conservation [52].This is the same in our study case, where the results showed that timber value was negatively correlated with structural complexity index (SCI) and also with soil water content and organic matter content, while SCI and the soil indicators were positively correlated [53].

Recommendation for Silvicultural Practices
According to the correlation analysis, plant species richness and species number (especially in shrub layer), mean DBH, diversity of DBH (stand uneven ages),and snag density played a positive role in habitat service provision.It is essential to increase the levels of these structural attributes when the manager aims to enhance the habitat service [54].The diverse shrub species composition and large cover of shrubs in oak-broadleaf mixed forests should be maintained by limiting the shrub clearing activity or thinning some trees to generate canopy gaps, which could facilitate the regeneration of understory vegetation as well as seedlings [55].Enrichment planting of native broadleaf species is recommended for improving the canopy species diversity.Standing dead wood should be retained in order to provide special habitats required by birds (e.g., woodpeckers), beetles, fungi, etc. [34].In order to increase the mean diameter and both vertical and horizontal diversity, the forest manager should not harvest all the trees of large diameter in a short period.Instead, they should selectively cut part of the upper-layer trees and part of the mid-layer trees with a longer rotation time, which allows the regrowth of smaller trees and thus the adjustment of diameter distribution [46].Retention of old-growth forests could also be helpful for promoting the heterogeneity of horizontal and vertical structures of the forest and saving the large trees for some habitat-specific fauna [5,36].
In the case that more weight is given to wood production in the future when the policy of logging ban is loosened, the forest manager is supposed to prepare to promote the timber value in this forest area where soil water conservation has been the priority management goal.As it is suggested by the results, enlarging the share of tree species such as pine and oak as well as increasing the number of trees would raise the income generated from timber harvest.However, if the decision-maker aims at achieving the multi-functionality of a forest, it is not recommended to increase these structural attributes by planting more trees of commercial species for the reason that it would lead to a forest structure that is not beneficial for habitat service or soil water conservation.Instead, it is preferable to maintain and increase the proportions of oak and pine by their basal area by thinning some of the small trees and retaining those of larger diameter until the mature age.

Conclusions
This study addressed the central research question about how to combine multiple forest services based on the integration of structural elements at the stand level.The results demonstrate that timber production appeared to conflict with habitat conservation and soil water conservation.The oak-broadleaf mixed forest was characterized as the stand type with the best habitat conservation while yielding the lowest timber value.The oak-pine mixed forest was identified as the kind of forest that had the most potential for timber harvest (in the scenario where pine is more valuable than oak), while the soil water conservation in this stand type was weakened.Nonetheless, such negative relation was not significant according to correlation analysis, implying that the conflicts between these services were not intense.In other words, the compatibility between multiple services is possible within a stand.Because soil erosion is a very important ecological problem in the study area, soil water conservation is the main management goal; it is suggested that forest management measures should be taken to increase the diversity of broadleaf tree species and the complexity of forest stand.This research could contribute to a basis for the sound decision making in the management planning of multiple-service oak forests, in order to enhance the integrated benefits of the oak forests, especially between the socio-economic services (e.g., timber production) and the ecological services (e.g., soil water conservation and habitat conservation).

Figure 1 .
Figure 1.Conceptual framework of the study.

Figure 1 .
Figure 1.Conceptual framework of the study.

Sustainability 2022 , 21 Figure 2 .
Figure 2. Biplot of PCA displaying the interrelationship among the original variables and between these variables and the principal components, with the grouping of the leading variables.The yellow circle indicates the variables important for assessing the habitat diversity; the blue circle indicates the variables important for assessing the soil nutrients availability; the green circle indicates the variables important for assessing the timber value.

Figure 2 .
Figure 2. Biplot of PCA displaying the interrelationship among the original variables and between these variables and the principal components, with the grouping of the leading variables.The yellow circle indicates the variables important for assessing the habitat diversity; the blue circle indicates the variables important for assessing the soil nutrients availability; the green circle indicates the variables important for assessing the timber value.

Table 1 .
Number of plots allocated to each stand type.

Table 2 .
Calculation equations of some structural attributes.
[22]umber of diameter class; p i : the proportion of individual trees in i-th diameter class[22]Tree height diversity SW I h = − n ∑ i=1 p i lnp i -n: number of height class; p i : the proportion of individual trees in i-th height class [22] Species richness in canopy layer SW I canopy = − n ∑ i=1 p i lnp i -n: number of species in canopy layer; p i : the proportion of basal area of i-th species [3] Under-story Shrub cover/ herb cover C Shrub/herb = ∑ n i=1 SC i n % SC i : total species cover of i-th subplot; n: number of subplots (n = 3) N % N domin : number of individual plants of dominant species in shrub/herb layer; N: total number of individual shrub/herb plant Species richness in shrub layer SW I shrub = − n ∑ i=1 p i lnp i -n: number of species in shrub layer; p i : the proportion of individual plants of i-th species [23] Species richness in herb layer SW I herb = − n ∑ i=1 p i lnp i -n: number of species in herb layer; p i : the proportion of individual plants of i-th species [23]

Table 3 .
Summary of structural attributes by stand types.
Within rows, the number means ± standard deviation of different sample plots.

Table 4 .
Summary of service indicators by stand types.Within rows, the number means ± standard deviation of different sample plots.Means followed by different letters are significant differences (p < 0.05) based on Tukey's test.*means that the parameter is significant at the level of 0.05; ** means that the parameter is ex-tremely significant at the level of 0.01.

Table 5 .
Summary table of PCA.

Table 6 .
Matrix of loadings of all variables to PC1 and PC2.

Table 7 .
The matrix of correlation between structural attributes and forest service indicators.
* means that the parameter is significant at the level of 0.05; ** means that the parameter is extremely significant at the level of 0.01.

Table A2 .
Summary of computed values of service indicators for each plot.