The importance of proleptic branch traits in biomass production of poplar in high-density plantations

Changjun Ding·Ningning Wang·Qinjun Huang·Weixi Zhang·

Juan Huang1,2·Suli Yan3,4·Boyi Chen3,4·Dejun Liang5·Yufeng Dong6·Yingbai Shen3,4·Xiaohua Su1,2

Abstract Branch phenotypic traits determine tree crown architecture, which in turn governs leaf display, light interception, and biomass production. Sylleptic and proleptic branches are the obviously different branch phenotypes in the poplar crown. Many studies have focused on the influence of sylleptic branch numbers (SBN) on biomass production, but the research on the influence of proleptic branch phenotypes was only a few. To explore the relationship between proleptic branch traits and biomass generation production in a high-density poplar plantation, we investigated the branch phenotypic traits of three poplar genotypes, all of which have high survival rates in forests (＞ 95%) and significantly different crown architecture and biomass performance in the high-density plantations (1667 stems ha-1). The plantation site was established in 2007. A terrestrial laser scanner was used to measure branch characteristics such as length, angle of origin and termination, and azimuth angle. A hierarchical cluster analysis performed on branch characteristics showed that SBN, crown depth, and proleptic branch curvature (PBC) were clustered with biomass production and leaf area index (LAI). Among all of the monitored traits, PBC played the second most important role in biomass production after SBN and was significantly correlated with SBN, LAI, and biomass production. The positive correlation between PBC and SBN indicated that a larger PBC was associated with more sylleptic branches within the monitored genotypes planted in the high-density plantation, providing greater leaf area and biomass production. The results of this study will improve the identification of high-production poplar varieties for cultivation in highdensity plantations for biofuel production.

Keywords Populous·Crown architecture·Sylleptic branch·Proleptic branch·Biomass production·Terrestrial laser scanner

Introduction

Poplars (Populusspp.) are potential renewable energy trees because of their fast juvenile growth and high productivity (Isebrands and Nelson 1981; Dickmann and Keathley 1996; Tuskan 1998; Moreno-Cortés et al. 2017). Crown architecture and canopy density influence forest productivity through foliage distribution and light interception (Isebrands and Nelson 1981; Ceulemans et al. 1990; Benomar et al. 2012; Pereira et al. 2017; Busov 2018). Interspecific and intraspecific variations in canopy structure within the genusPopulushave been well studied (Ceulemans et al. 1990; Wu and Stettler 1998; Gielen et al. 2002; Zeleznik 2007). Branch characteristics (e.g., branch angle, size, number, and distribution) are considered important factors in crown architecture and canopy density (Isebrands and Nelson 1981; Xu et al. 2017; Busov 2018). Therefore, an in-depth understanding of how branch characteristics influence crown structure and canopy density would inform poplar hybrid selection, breeding, and management to optimize forest yield.

Branches in the poplar crown can be classified as either proleptic or sylleptic, according to the length of the dormancy period. Lateral meristems develop into proleptic branches after a period of rest (winter dormancy). Branches formed within the vegetative period and without dormancy will develop into sylleptic branches (Remphrey and Powell 1985; Benomar et al. 2012). Correlation analysis of branch characteristics and crown architecture has been used to determine ways to increase forest yield. Dickmann and Keathley (1996) provided a detailed description of the bestperforming poplar trees grown for energy or wood fiber in a high-density plantation and found that sylleptic branching was one of the most important factors. Ceulemans et al. (1990) found a broad range of branch numbers among different clones. Sylleptic branch number (SBN) has also been shown to be strongly determined by genetics rather than environmental factors (Wu and Stettler 1996, 1998; Marron et al. 2006). Considering the close relationship between SBN and leaf area index (LAI), SBN is significant factor for biomass production in short-rotation poplar coppice (Ceulemans et al. 1990; Rae et al. 2004; Dillen et al. 2009; Benomar et al. 2012; Broeckx et al. 2012) because of the vital function of LAI in biomass production (Verlinden et al. 2013; Wang et al. 2014; Xie et al. 2017). In addition, sylleptic branching is highly related to wood basal area growth (Wu and Stettler 1996) and contributes to the optimization of crown architecture by efficiently filling available space within the crown, thereby increasing the allocation of biomass to the stem and/or increasing leaf area in the crown (Zeleznik 2007; Benomar et al. 2012).

Wu and Stettler (1996) showed that proleptic branches are denser, longer, and have more and larger leaves than contemporaneous sylleptic branches. The number of proleptic branches is more positively correlated with tree height growth than is the number of sylleptic branches (Wu and Stettler 1996). Compared with sylleptic traits, proleptic traits (length, number, and leaf area in the branch) can better predict the growth of stem height and basal area (Wu and Stettler 1998). However, the influence of proleptic branches on crown architecture, especially in high-density plantations, remains unclear.

Branch length (BL) and branch diameter (BD), which are tightly correlated with the crown of a single poplar genotype, strongly control wood production and LAI (Ceulemans et al. 1990; Broeckx et al. 2012). Branch angle and branch curvature are significantly different among genotypes (Ceulemans et al. 1990). Nelson et al. (1981) showed that genotypes with the smallest angle of origin (AO) also had the smallest angle of termination (AT), and vice versa. The correlation between AO and AT for a given genotype differs with respect to height and available space (Isebrands and Nelson 1981). It is necessary to develop biologically meaningful variables to evaluate branch angle in intensive poplar cultivation with short rotation (Dickmann and Keathley 1996).

It would be beneficial to understand the correlations between branch phenotypic traits with crown architecture and biomass production. However, taking these measurements is destructive in high-density plantations. Therefore, many existing data sets of these variables are discontinuous. Moreover, traditional measurements of tree height and branch angle are time-consuming, and parameters such as branch azimuth angle (AA) cannot be measured directly (Wang et al. 2015, 2016). Recently developed terrestrial laser scanner could improve the precision and efficiency for the measurements of crown branch characteristics (Wang et al. 2015, 2016; Tao et al. 2015; Woodgate et al. 2015; Xie et al. 2017). The accuracy and feasibility of this scanner in measuring branch characteristics has been demonstrated in several studies (Teobaldelli et al. 2008; Ga?tan et al. 2012; Wang et al. 2015, 2016; Tao et al. 2015).

We hypothesized that varying proleptic branch curvatures (PBCs) for different genotypes would have different effects on crown architecture and biomass production in high-density plantations and that proleptic branch traits would correlate with branch traits of sylleptic branches, leading to optimal crown architecture. We used a terrestrial laser scanner to quantify first-order branch characteristics for threePopulusgenotypes in a high-density plantation (1667 stems ha-1) using 6-year-old trees (2007-2013) under short-rotation intensive cultivation in southern Jinzhou, China. The correlation between branch characteristics and biomass production was analyzed. We anticipated that our results will inform the selection of suitable and high-yield genotypes and development of management strategies for short-rotation, high-density poplar plantations, particularly in northern China.

Material and methods

Experimental site and plant materials

In an experimental forest in Linghai, Jinzhou, Liaoning province (41°17′ N, 121°36′ E; 17 m above sea level) in the spring of 2007, we established a plantation on a site that was a former maize cropland. In the temperate monsoon climate, the range in annual temperatures is large (average in approximately - 6 to - 8 °C in winter and 22.5-23 °C in summer). From 1951 to 2010, the mean annual temperature was 8-9 °C and mean annual precipitation was 564 mm. The forest soil is black loam with naturally good drainage.After site preparation, 1-year-old (in 2007) root systems (stem removed) of 22 clonal poplar hybrids (genotypes) were planted, one hybrid in each 3 × 2 m plot, 9 root systems of each hybrid as replicates. The site was divided into three blocks, representing three biological replicates, each block contained 22 genotypes which placements were random. The blocks were aligned north-south. In 2013, the 22 poplar hybrids were classified into three groups based on crown architecture. Three genotypes (genotype 6, genotype 14, and genotype 171) with significantly different crown architectures and biomass production were chosen for observation and measurement in this study. Details on the plantation and sampling layout were described by Wang et al. (2014, 2016).Genotype 6 is a hybrid ofP.deltoidesfrom Romania (maternal) andP.nigra‘Vereecken’ from the Netherlands (paternal). It was planted in six rows and five columns in the northern block, six rows and four columns in the middle and southern block. Genotype 14 had the same paternal ancestor as genotype 6, and its maternal ancestor was a hybrid offspring ofP.deltoidesBartr. cv. ‘Shanhaiguanensis’ from China andP.deltoidesBartr. cl. ‘Harvard’ (I-63/51) from Italy. Genotype 14 was planted in four rows and four columns in all three replicates. Genotype 171, a hybrid of hybrids (P.deltoidescl. ‘55/65’ ×P.deltoidescl. ‘2KEN8’) × (P.nigra‘Brummen’ ×P.nigra‘Piccarolo’), was planted in three rows and four columns in all three biological replicates. Genotypes 6 and 14 were generated in 2002, whereas genotype 171 was generated in 2003. All three genotypes had higher survival rates (＞ 95%) in highdensity plantations.

Crown architecture variables

Diameter at breast height (DBH) of trees in the nine plots was measured at a height of 1.3 m with a diameter-measuring caliper. One standard tree (closest to the mean DBH of each plot) was selected from the center of each plot to avoid possible edge effects. The crown architecture was scanned in February 2013 before bud germination (just after the sixth growing season in 2012). Therefore, all measurements were performed on dormant and leafless trees in winter. Trees were scanned with the terrestrial laser scanner and then cut down. The proleptic and sylleptic branches were counted and marked after scanning. For each tree, the total branch number (TBN) of the crown was the sum of the PBN and SBN. The ratio of SBN to PBN was also calculated. All of the crown architecture variables used in this work were measured from three-dimensional (3D) point cloud files, which were scanned with a 3D laser scanner (Faro Laser Scanner Focus3D, focus 3D 120, FARO Technologies, Korntal Münchingen, Germany). The scanner software, SCENE 5.0, was used to combine the scanned points into a 3D image (Wang et al. 2015). Crown depth (CD), crown width (CW), and tree height were then measured using Geomagic Studio 2012 (3D Systems, Rock Hill, SC, USA). Tree height was defined as the distance from the ground to the top of the tree, CW was calculated by averaging the projected crown diameter, and CD was measured as the distance from the basal fork to the top of the tree. We used Geomagic Spark (2012, 3D Systems, USA) to measure proleptic branch and sylleptic branch characteristics including BL, AO, AT, and AA. The BL was defined as the length of the line along the branch from the connection point between the branch and stem (starting point) to its tip (end point). Branch AO was defined as the angle between the stem and the tangent of the branch at the starting point; branch AT as the angle between the stem and the line connecting the starting point and the end point of the branch, and branch AA as the angle between adjacent branches on the stem. BC was calculated as the ratio between branch AO and branch AT. Total BL, AO, AT, AA, and BC were averages of all values for the respective branch traits in the crown. All branches in the crown were measured.

BD cannot be measured directly from cloud files. However, considering its importance to branch biomass (BB), it was calculated from the linear relationship between BL and BD. Several branches of each genotype with different lengths and diameters were measured. BD was determined at 1 cm above the fork, and BL was measured along the branch as defined by the cloud files. A linear relationship was determined between BL and BD (Fig. S1). A single BD was calculated based on the linear relationship between BD and BL. The BB of a separate branch was calculated as:

whereρis the stem wood density (0.36 g cm-3) for poplar according to the literature (Calfapietra et al. 2003; Huang et al. 2008). The average BB for a single branch was obtained by dividing total BB by the BN. A single proleptic or sylleptic BB was calculated as the ratio of total proleptic or sylleptic biomass to the PBN or SBN. We estimated individual tree stem volume using the binary volume table provided in the technical regulations for poplar cultivation: LY/T 1716-2007 (Chen et al. 2007). The total aboveground biomass per tree (TBP) was calculated as the sum of biomass of all branches and stems, which was calculated as:

whereDis DBH,His tree height,ρis stem wood density, BB is the average single branch biomass in the crown, and TN is the total branch number in the crown. The total biomass per unit land area (TBS, Mg ha-1) was calculated as described by Wang et al. (2014, 2016) as:

LAI measurement

LAI was measured in August 2012. The WinScanopy canopy analyzer (WinSCANOP, Regent Instruments, Quebec City, QC, Canada) was used to calculate the LAI in each of the monoclonal plots. Five random spots were selected in each plot to obtain hemispherical images required to determine the average LAI.

Data analyses

Analysis of variance (ANOVA) was used to calculate the relationship between branch characteristics with crown architecture, biomass production, and LAI among the different genotypes. Multiple comparison analyses were used to determine the genotypic differences in BP, BS, LAI, CD, BN, and BC. Pearson’s correlation coefficients were calculated to evaluate any correlations between branch characteristics and biomass production or LAI. A linear regression was performed between BD and BL. Principal component analysis (PCA) and partial least squares regression (PLS) were used to quantify the importance of the measured variables to biomass production, LAI, crown depth, and PBC. Hierarchical cluster analyses were performed on the variables of branch characteristics, biomass production, and LAI to determine the most relevant variables. ANOVA, regression analyses, Pearson correlation coefficients, and hierarchical cluster analysis were performed using SPSS 18.0 (SPSS Inc., Chicago, IL, USA). The PCA and PLS were conducted using Simca-P 11.0 (Umetrics AB, Umea, Sweden). A total of 200 permutation tests were used in the PLS to prevent over-fitting in the models.

Results

Differences in branch characteristics among genotypes

Three randomly selected representative crown architectures for each genotype are shown in Fig. 1. Images were created from data gotten utilizing the 3D laser scanner. Differences in crown architecture traits (e.g., BN and CD) are easily discerned.

Differences in biomass production, LAI and CD among the three investigated genotypes are shown in Fig. 2. Among in the three genotypes, the TBP and TBS of genotype 171 were highest, whereas biomass production of genotype 14 was lowest (Fig. 2A, B). Of the three genotypes, genotype 171 had significantly higher biomass production, LAI and CD than genotypes 6 and 14 (Fig. 2C, D). The differences in LAI and CD between genotypes 6 and genotype 14 were not significant (Fig. 2C, D).

Fig. 1 Representative crown architectures for the three genotypes generated from the point cloud data scanned using a terrestrial laser scanner. The perspective is from the west, perpendicular to the rows

Neither TBN nor PBN differed significantly among the three genotypes (Fig. 3A). However, there was a significant difference in SBN (Fig. 3A), with the highest SBN in genotype 171 and the lowest in genotype 14. The SBN of genotype 6 was intermediate between genotypes 171 and 14, which was significantly higher than that of genotype 14. Genotype 6 had a shorter average BL than genotypes 14 and 171 (P＜ 0.05); however, their proleptic BLs were similar. Genotype 14 had the longest sylleptic branch length, and genotype 6 had the shortest (P＜ 0.05) (Fig. 3B). The branch AO of the entire crown, the proleptic branches, and the sylleptic branches each differed significantly among the three genotypes (Fig. 3C). Crown branch AO was largest in genotype 6, but lowest in genotype 14. Genotype 14 had the smallest branch AO. The proleptic branch AO was significantly smaller in genotypes 171 and 14 than in genotype 6 (Fig. 3C). Genotype 14 had the smallest sylleptic branch AO (P＜ 0.05). There was no significant difference in sylleptic branch AO between genotypes 6 and 171 (Fig. 3C). There was a similar trend between branch AT and branch AO. The highest average branch AT was found in genotype 6, and the lowest in genotype 14 (P＜ 0.05). Proleptic branch AT and sylleptic branch AT were larger in genotype 6 than in genotypes 14 and 171 (Fig. 3D). AA did not differ significantly among the three genotypes (Fig. 3E). Although there were no significant differences in average BC and SBN, PBC was significantly different among the genotypes (P＜ 0.05), and was highest in genotype 171 and lowest in genotype 14 (Fig. 3F).

The relationship between crown architecture parameters and biomass production

PCA results showed that measured crown architecture traits differed significantly between genotypes 171 and 14, whereas branch characteristics of genotype 6 did not differ significantly from genotypes 171 and 14 (Fig. 4A).

For the PLS analysis, values larger than 1 were considered significant (Table 1). The factors in the same direction as they-variables showed a positive correlation, and in the opposite direction showed a negative correlation (Fig. 4). Components closer to the origin point showed a lower correlation with they-variable (Fig. 4). In the PLS analysis, in addition to crown characteristics (e.g., LAI, CLA, CD, and CW), the SBN and PBC also greatly contributed to biomass production (Fig. 4B, C; Table 1). We also analyzed the contributions of branch characteristics to LAI and found that SBN and PBC play important roles in determining LAI (Fig. 4D; Table 1).

Fig. 2 Genotypic differences in phenotypes of A biomass per plant, B biomass per unit land area, C leaf area index (LAI), and D crown depth. Genotypes that do not share a letter are significantly different (P ＜ 0.05, n = 3)

The relationships of PBC and SBN with LAI

Significant correlations were found among TBP, TBS, CD, CW, CLA, and LAI. In addition, PBC and SBN showed significant correlations with TBP, TBS, CD, CW, CLA, and LAI (Table 2). PBC was significantly correlated with SBN and RPS.

We performed linear regressions among PBC, SBN, sylleptic branch probability (RPS), and LAI. A significant linear relationship was found among PBC, SBN, RPS, and LAI (Fig. 5).

We determined the branch number by parsing the tree according to the rescaled cluster distance (Fig. 6). TBS, TBP, LAI, crown leaf area (CLA), CD, CW, SBN, PBC, and ratio of sylleptic branch number to total branch number clustered together (cluster 1). Geometric traits of sylleptic branches (sylleptic BL, sylleptic BC, and sylleptic branch AA) clustered together with proleptic BL to form cluster 2. Angle traits (sylleptic AO, sylleptic AT, proleptic AO, proleptic AT, and proleptic AA) formed cluster 3.

Fig. 3 Comparison among the three genotypes of crown architecture variables: A branch number, B branch length, C branch origin angle, D branch termination angle, E branch azimuth angle, and F branch curvature. Genotypes that do not share a letter are significantly different for a variable (P ＜ 0.05, n = 3)

Discussion

PBC significantly affected biomass production

As hypothesized, the three genotypes that differed in biomass production had significantly different crown architectures (CD, LAI, and CLA) (Figs. 1, 2, 3) and branch characteristics (SBN and PBC) (Fig. 3). The PCA and PLS analyses clearly showed a correlation between LAI, CLA, CD, SBN, PBC, CW, and RPS withTBS and TBP (Fig. 4; Table 1). We applied PLS to determine which factors had the most significant effect on biomass production and LAI because PLS is appropriate for numerous colinear variables (Rae et al. 2004; Eriksson et al. 2006). The important role of SBN in poplar biomass production in high-density plantations has been previously reported and it is caused by a close correlation between SBN and LAI (Ceulemans et al. 1990; Rae et al. 2004; Dillen et al. 2009; Broeckx et al. 2012). LAI has been considered a simple and reliable predictor for forest biomass production (Waring et al. 1977; Ceulemans et al. 1990; Verlinden et al. 2015). PLS analysis and Pearson’s correlation analysis also showed that SBN had a significant impact on biomass production, LAI, and CLA (P＜ 0.01) (Fig. 4B, D; Tables 1, 2), consistent with previous findings (Broeckx et al. 2012). The PLS analysis also indicated that sylleptic branch probability (RPS) had a significant effect on biomass production (TBP and TBS) and LAI (Fig. 4; Table 1). These results are consistent with their known effects on crown architecture (e.g., filling available space in the crown and increasing CLA) (Benomar et al. 2012).

Fig. 4 Principle component analysis (PCA) and partial least squares regression (PLS) scatterplots for crown architecture variables, biomass production, and LAI. A PCA score for branch characteristics and biomass production of the different genotypes. B PLS score for biomass production per tree. Black triangles represent the y-factors of biomass production per tree. C Biomass production per unit land area. Black triangles represent the y-factors of biomass production per unit land area, and D LAI. Black triangles represent the y- factors of total leaf area per unit land area. In A, gray pots represent genotype 6, black triangles represent genotype 14, and stars represent genotype 171. In B-D circles represent the x-factors

The proleptic branches are larger and fewer in number compared to the sylleptic branch in the crown. In a poplar crown, proleptic and sylleptic branches together form the crown architecture. Previous researches have mainly focused on the relationship between sylleptic branch traits and biomass production (Wu and Stettler 1996, 1998; Marron et al. 2006; Broeckx et al. 2012); the larger (proleptic) branch has not received as much attention. Isebrands and Nelson (1981) pointed out the importance of branch angles in response to available space and to increases in height growth among different genotypes. Our results showed that proleptic branches indeed play an important role in biomass production.

Wu and Stettler (1998) reported that proleptic branch traits are better predictors for biomass production than sylleptic branch traits. Our work demonstrates the importance of PBC on biomass production. As shown in Figs. 2 and 3, TBP, TBS, LAI, and PBC values were highest in genotype 171 and lowest in genotype 14. PBC showed a significant correlation with LAI (P＜ 0.01), CLA (P＜ 0.01), TBS (P＜ 0.05), and BP (P＜ 0.05) (Table 2). The PLS allowed an evaluation of the relationship between BS, BP, and LAI with PBC (Fig. 4; Table 1). PBC was positively correlated with LAI (Figs. 4, 5C; Tables 1, 2). LAI was considered a decisive factor for woody biomass (Pellis et al. 2004; Broeckx et al. 2012; Verlinden et al. 2013). Furthermore, PBC clustered with TBP, TBS, and LAI (Fig. 6). According to a study on a 3-year high-density poplar plantation, the high biomass production genotype also had high PBC (Ceulemans et al. 1990). A significant correlation between PBC and LAI was also observed in a 2-year-old high-density poplar plantation (Broeckx et al. 2014). Thus, crown PBC could be useful as a biomass predictor for high-density poplar plantations.

Table 1 Variable importance in projection (VIP) scores for crown architecture traits, total biomass per plant (TBP) and total biomass per unit land area (TBS) in the PLS

PBC was highly correlated with SBN

Branch characteristics determine crown architecture and canopy structure and thus influence light interception and biomass accumulation (Remphrey and Powell 1985; Ceulemans et al. 1990; Casella and Sinoquet 2007; Guisasola et al. 2015). The role that SBN plays in biomass production is a consequence of increasing CLA (Broeckx et al. 2012). Sylleptic branch traits are under genetic control and tend to include more quantitative trait loci that affect biomass and LAI (Wu and Stettler 1998; Wu 1998). Proleptic branch traits predict basal area and stem height growth better than SBN. However, it has remained unclear which proleptic branch traits contribute to biomass production (Wu and Stettler 1998; Wu 1998). In our work, PBN had a significant correlation with biomass production and LAI (Figs. 2, 3, 4; Tables 1, 2), which is consistent with the previous findings (Wu 1998). PBC also had a strong influence on biomass production and LAI (Fig. 4; Table 1). To understand the effects of PBC on biomass production, we analyzed the correlation of biomass production with other crown architecture traits and found that all of the monitored traits were tightly correlated with biomass production. Figures 2 and 3 show that PBC had a similar trend to biomass production, CD, and LAI. PLS analysis showed that PBC had a significant influence on LAI (Fig. 4C; Table 1). Pearson’s correlation coefficients indicated that PBC was significantly correlated with TBP, TBS, CD, LAI, CLA, SN, and RPS (Table 2). Hierarchical cluster analysis showed that PBS clustered with TBP, TBS, LAI, CLA, CD RPS, and SBN (Fig. 6). Together these results suggest that PBS is highly correlated with biomass production and SBN. However, in the PLS analysis, the variable importance in projection (VIP) score for PBC was lower than that for SBN, and Pearson’s correlation coefficient was lower than that of SBN (Fig. 4; Tables 1, 2). We found a significant linear relationship between PBC and SBN (Fig. 5). It has been suggested that SBN optimizes crown architecture by filling crown space, increasing leaf area and light interception, and allocating biomass to the stem (Benomar et al. 2012; Broeckx et al. 2012). In a high-density crown, the space available for sylleptic branches is constructed and provided by PBC. This study reveals that proleptic branches interact with SBN. Together, the two branch types optimize crown structure and increase biomass production.

Fig. 5 Paired correlations among PBC, SBN, RPS, and LAI. A Correlation between PBC and SBN. B Relationship between PBC and RPS. C Relationship between PBC and LAI

Fig. 6 Dendrogram of the hierarchical clustering analysis conducted on branch characteristics measured on three Populus genotypes (left side: abbreviations as defined in the main text). The three clusters are indicated on the dendrogram branches

Table 2 Pearson’s correlation coefficients for crown architecture parameters (n = 3)

Previous studies have demonstrated that leaf shade tolerance (i.e., leaf light compensation point and respiration rate) contributes much to LAI and biomass production in high-density plantations with the same genotypes (Wang et al. 2014). In this work, we found that proleptic branch curvature had positive relationships with SBN, LAI and biomass production. There may be some interactions or tradeoffs between leaf light response traits and branch geometry traits, and these interactions or trade-offs may influence biomass production. These questions can be addressed in future studies.

Conclusions

In this work, we documented the relationships among branch characteristics, crown architecture, and biomass production in three poplar genotypes growing in a high-density, shortrotation plantation. Cluster analyses were performed on all of the variables, and Pearson’s correlation coefficients were used to illustrate their relationships to biomass production. The CD, SBN, and PBC were correlated with biomass production. PLS analysis showed the importance of branch characteristics on biomass production. The BN, particularly the SBN, played the most important role in determining biomass production. The second most important predictor of biomass was the CD, and the third was PBC. Hierarchical cluster analysis also showed that PBC clustered with biomass production, LAI, and SBN. These results show that PBC has a significant impact on biomass production in highdensity poplar plantations. The Pearson’s correlation analysis and linear regression showed that PBC interacts with SBN and that CD and PBC are significantly correlated. We propose that PBC provides the space for sylleptic branches and that together these two types of branches optimize crown structure.

AcknowledgementsWe gratefully acknowledge the technical assistance of Zhiyan Yang, Chengchao Yang, Jiandong Peng, and Shusen Liu. We also thank Wenguang Yin of Beijing Haoyu World Surveying and Mapping Developing Limited for his helpful advice with the terrestrial laser scanner and software for cloud data analysis. Further technical assistance was received from the Liaoning Provincial Poplar Research Institute. We also thank Dr. Liyan Ping and Dr. Eric McLamore for their help during the preparation of this manuscript.