Age trends in genetic parameters and genetic gains of growth traits in multiple progeny test sites of hinoki cypress (Chamaecyparis obtusa)

ABSTRACT Growth data obtained from Chamaecyparis obtusa growing at 28 progeny test sites in the Kanto breeding region of Japan were analyzed to estimate the genetic parameters and genetic gains. Specifically, the data were used to estimate stem diameter at breast height (DBH), tree height (TH), and stem volume in order to assess the feasibility of breeding, and differences in genetic parameters and genetic gains associated with growth traits based on age and climatic conditions were clarified. The median values of narrow-sense heritability for DBH and TH across all ages ranged from 0.229 to 0.263 and from 0.201 to 0.433, respectively. Based on the efficiency of improvement (genetic gain per year), it appears that the most efficient selection for DBH and stem volume would be at one-third of the improvement target age (set to 30 years in the present study). We analyzed the variation of narrow-sense heritability and age trends of those among three groups of test sites exposed to different climatic conditions. Age and group × age interaction were shown to have significant effects on the narrow-sense heritability of TH and volume and that of DBH and volume, respectively. However, no significant effects of group × age interaction were detected on the efficiencies for gain per year of all traits, implying that this parameter might not be affected by climate conditions. Therefore, it was suggested that early selection is possible with almost equal efficiency in all climate groups.


Introduction
The genetic value of families, mating parents, and individuals has been estimated in forest tree improvement programs based on the performance of progeny test sites (White et al. 2007).These datasets have also been used to estimate genetic parameters, such as the heritability of various target traits (e.g.stem diameter, tree height (TH), and wood properties) in many coniferous tree species, including Pinus taeda (Gwaze et al. 1997), Pinus contorta (Ye and Ying 1996;Rweyongeza 2016), Picea glauca (Rweyongeza 2016), Pseudotsuga menziesii (Ye and Jayawickrama 2012), Cryptomeria japonica (Yomogida 1999;Hiraoka et al. 2019), and Larix kaempferi (Dong et al. 2020).Heritability is one of the most important genetic parameters considered in breeding programs, as it indicates to what degree a trait is controlled by genetic factors and how easily it can be improved (Falconer and Mackay 1996).However, heritability is influenced by various factors, such as genetic background, environmental conditions, and experimental design (Hannrup et al. 1998;Kusnandar et al. 1998;Kroon et al. 2011).Additionally, it can vary with tree age; therefore, age trends in heritability have been investigated in many coniferous species (Franklin 1979;Hodge and White 1992;Paul et al. 1997;Jansson et al. 2003;Weng et al. 2007;Hiraoka et al. 2019).Overall, clarifying the variations in heritability among sites and different age classes is essential for effective tree breeding programs.
Genetic correlation is another important genetic parameter to describe the relationship between traits and is used to calculate indirect genetic gain (Falconer and Mackay 1996;White et al. 2007).As forest trees have a longer maturation period than crop plants, trait evaluation for breeding purposes is more lengthy and expensive (Zobel and Talbert 1984).Therefore, early selection based on the estimates of genetic gain is essential to increase the effectiveness of forest tree breeding programs.Specifically, in several programs targeting coniferous species, the ideal ages for indirect early selection have been identified by examining the age trends of genetic gain, which were derived from age -age correlations (Lambeth 1980;McKeand 1988;Jansson et al. 2003;Weng et al. 2007;Hiraoka et al. 2019).
The current nationwide tree breeding program in Japan was established in 1952 for major plantation species, such as C. japonica, Chamaecyparis obtusa, and L. kaempferi, with plustree selection beginning in 1954 to improve forest function and make better use of forestry-derived revenues (Forest Tree Breeding Association 2004).One of the most important goals of breeding programs is to improve growth performance, that is, the volume of produced wood, which can also promote carbon storage by forests (Fukatsu et al. 2013).Wood volume can be increased in terms of both diameter at breast height (DBH) and tree height (TH); however, the responses of these two parameters to environmental conditions may differ between test sites and genotypes (White et al. 2007).
Chamaecyparis is a genus of the Cupressaceae family distributed in the coastal areas of the Pacific and Atlantic Oceans in the northern hemisphere and is native to eastern Asia and the western and eastern margins of North America (Farjon 2005).In Japan, C. obtusa (hinoki in Japanese, hinoki cypress in English) is one of the most important silvicultural species, accounting for 25% of the total plantation area.Since the beginning of breeding programs in Japan, over 1,000 first-generation C. obtusa plus trees have been selected for breeding purposes from forests located in this species' natural distribution range (Forest Tree Breeding Center 2020).In the Kanto, Kansai, and Kyushu breeding regions, breeding populations have been developed using offspring of the firstgeneration plus-tree clones, and progeny test sites have been established for the evaluation of those plus trees.Previous studies have analyzed the genetic parameters associated with growth traits in C. obtusa growing in the three above mentioned regions.In Kyushu, Fukatsu et al. (2013) reported that the narrow-sense heritabilities of DBH, TH, and volume based on a single test site were 0.165, 0.010, and 0.235, respectively.In Kansai, Nasu et al. (2006) reported the genetic parameters of C. obtusa using data obtained from multiple test sites.However, compared with these two regions, the information on genetic parameters in Kanto is still limited.McKeand et al. (2008) revealed that there were moderate regional differences in the heritability of growth traits in P. taeda.On the other hand, Jansson et al. (2003) examined the age trends in the heritability of TH in 13 Pinus sylvestris progeny test sites grouped into northern and southern regions located in southern Sweden and found that it increased with age in both regions, but at different rates.The study suggested that these regional variations in heritability with age might be caused by environmental factors, such as climatic conditions.The heritability of growth traits in C. obtusa, as in other coniferous species, may also be affected by environmental factors.A more accurate selection at a younger age is preferable to improve a target trait to achieve a considerably higher genetic gain per year and effectively promote breeding programs.However, early selection can be associated with difficulties in balancing its effectiveness and uncertainty based on the trait dataset at young ages.In view of this, it is beneficial to investigate age trends in heritability and in genetic gains associated with growth traits.In the present study, we analyzed a multiage dataset of growth traits obtained from C. obtusa trees located at 28 progeny test sites in the Kanto breeding region of Japan.For each progeny test site and age, we aimed to estimate narrow-sense heritability, genetic correlation, genetic gain, and genetic gain per year.Furthermore, we investigated whether the age trends in heritability and genetic gain per year differed across different groups of test sites classified based on climatic variation.

Growth data
The growth dataset used in our analysis was obtained from 28 progeny test sites of C. obtusa located in the Kanto breeding region in Japan (Figure 1).These progeny test sites consist of open-pollinated or full-sib families (seedlings) of firstgeneration C. obtusa plus trees (Table 1).In total, 241 plustree clones were used as mating parents for all families.Several parents were common as parents of families in some progeny test sites.Almost all sites were planted using an incomplete random block design with three replications or more.Each replication consisted of plots of approximately 10 trees of a certain family.The principal measures of the growth of individual trees were DBH (cm) and TH (m).These variables were measured via periodical surveys at 5, 10, 20, and 30 years after the establishment of each test site.However, because some progeny test sites were not measured in all of these years, the number of sites for analysis differed among ages (Table 1).DBH was measured using calipers and TH was determined using measuring rods or digital measuring devices (Vertex, Haglof).The mean values of DBH and TH at the age of 20 years at each test site ranged from 7.0 to 15.0 cm, and from 5.6 to 10.8 m, respectively (Table 1).The age trend of the mean of DBH and TH were illustrated in Fig. S1.Using DBH and TH data, the stem volume for all individuals at all measurement ages was calculated using a stem-volume calculation program created at the Forestry and Forest Products Research Institute in Japan (Hosoda et al. 2010).

Estimation of each parameter
All statistical analyses were conducted using the open-source statistical software R 4.1.2(R Core Team 2021).An estimation of the genetic variance for each trait at each site was obtained through a linear mixed-effects model with restricted maximum likelihood method: where y ij is the measured value for the jth individual in the ith block, μ is the general mean, B i is a fixed effect of the ith block, A ij is the random effect of the additive genetic effect of the jth individual in the ith block, and e ij is the residual.The random factors and individual-tree breeding values were obtained using an individual-tree model of the best linear unbiased prediction.As the volume did not follow a normal distribution, the values were log-transformed.After the obtained values, including breeding values for volume, had been back-transformed, relative genetic gain, relative indirect genetic gain, and efficiencies for gain per year were estimated.The narrow-sense heritability (h 2 ) was estimated using the following formula: where σ 2 A and σ 2 e are additive genetic variance and residual variance, respectively.Genetic variance, breeding value, and narrow-sense heritability were estimated for each progeny test site using the breedR package in R (Munoz and Sanchez 2020).Additionally, the age -age genetic correlation (r age -age ) was calculated based on the correlation coefficient of the breeding value of parents obtained in different years for all pairs at the same sites (Hiraoka et al. 2019).
The direct and indirect genetic gains for each trait were calculated for each site at each measurement age based on the breeding values of the selected parents.The following formula was used to calculate the relative direct genetic gain (G r ) at a selection intensity of the 33.3% uppermost plus-tree clones as mating parents: where ΔG is the genetic gain based on the average breeding value of the selected parents and m is the mean of each site, which is calculated from the sum of μ and the mean of best linear unbiased estimators of block effects in formula (1).The following formula was used to calculate the indirect genetic gain relative to direct selection (GI r_xy ): where ΔGI xy is the average breeding value of parents selected indirectly at age y (fixed to 30 years) at the selection intensity based on the breeding value of age x, and ΔG y is the genetic gain of direct selection at age y.
The following formula (Lambeth 1980;McKeand 1988;Leksono et al. 2006) was used to calculate the efficiencies for gain per year (E): where x and y are the ages of indirect and direct selection conveniently set to y = 30 as the improvement target age in the present study.To account for the time spent on seed orchard establishment and improvement, as well as seed production, 5 years were added per generation.For all growth traits, the Wilcoxon rank-sum test was used to investigate the variation across ages in h 2 , r age -age , G r , GI r_xy , and E.

Grouping of the progeny test sites by difference of climatic condition
We used principal component analysis (PCA) and cluster analysis to categorize the 28 test sites based on the regional variation in climatic conditions (Tables 1 and 2).For the climate variables used in the PCA, in accordance with a report of studies that similarly studied the trends in genetic parameters of C. japonica in the Kanto breeding region (Miura et al. 2005), we selected the following seven climate variables: annual mean temperature, mean temperature in August, mean temperature in February, annual total precipitation, total precipitation from May to September, total precipitation from December to March, and snow accumulation from December to March.Annual maximum and minimum temperatures often occur in August and February, respectively.The climate dataset was based on a 1-km- mesh scale, via the digital national land information of Japan (The Ministry of Land, Infrastructure, Transport and Tourism of Japan 2012).The first and second principal component scores for each site obtained from PCA were used as variables in the cluster analysis.The optimal number of clusters was determined using the upper tail method (Mojena 1977).

Age trends in heritability in relation to climate groups
Based on the datasets of progeny test sites separated into three climate groups, which reflected the climatic variation in the Kanto breeding region.The age trends of h 2 and E for each trait were analyzed using a linear mixed-effects model via the following formula: where y lmn is the h 2 or E of the mth age of the nth progeny test site in the lth climate group, μ is the general mean, Group l is the fixed effect of the lth climate group as a factor, Age m is the fixed effect of mth age as a covariate, Group l × Age m is the fixed effect of interaction between the lth climate group and the mth age, Progeny ln is the random effect of the nth progeny test site nested by the climate group, and e lmn is the residual.The significance of the fixed-effects variables in the linear mixed-effect model was tested by Type III analysis of variance (ANOVA) for h 2 and E. Based on the regression coefficients estimated by the linear mixed-effects model, the regression lines were drawn for each group.When the fixed effect of Group l × Age m interaction is significant, the age trend of h 2 or E differs among climate groups.

Results
The median value of h 2 for DBH across all ages ranged from 0.229 to 0.263 (Figure 2).In the case of TH, the median h 2 gradually increased from 0.201 (5 years) to 0.433 (30 years).The median h 2 values for volume across all ages ranged from 0.195 to 0.262.
Table 3 shows the median age -age genetic correlation (r age -age ) values for all traits.In all traits, the correlations became weaker with increasing age differences: r age -age decreased from around 0.8 between 5 and 10 years old to around 0.5 between 5 and 30 years old.For age of 30 years as target age, r age -age with the age of 20 years as selection age were significantly higher values than ages of 5 years and 10 years in DBH and volume.
Figure 2 shows the age trends of the relative direct genetic gain (G r ), the relative indirect genetic gain (GI r_xy ), and the efficiencies for gain per year (E) when the target selection age was set to 30 years for all traits.In the case of DBH, the median G r dramatically decreased from the ages of 5 (30.6%) to 10 years (14.9%) and became almost stable at the ages of 20 (10.6%) and 30 years (10.5%).Substantial differences were found in all age classes except for pairs at the ages of 20 and 30 years.The median G r of TH was approximately 10% at all ages, with no significant difference (p > 0.05).Concerning G r for volume, the median values at the ages of 20 and 30 years were 34.3% and 29.1%, respectively.For all traits, the GI r_xy became higher at more advanced ages.The median GI r_xy of DBH increased from 45.0% in 5-year-old to 87.5% at the age of 20 years and significantly differed among all ages.For TH, a significant difference in median GI r_xy was found between the ages of 5 (44.9%) and 20 years (71.1%).Meanwhile, the median GI r_xy for volume increased from 51.4% to 82.6%.In volume, the median GI r_xy at the age of 20 years was significantly larger than that at the ages of 5 and 10 years.For DBH, TH, and volume, the median E was larger at younger ages.In E for all traits, wider variations were found at younger ages than at the age of 20 years.Concerning DBH, the median at the age of 5 years (1.576) was significantly larger than that at the age of 20 years (1.225).In the case of TH, significant differences in the median value of E were found between the age of 5 and ages of ≥10 years.In the case of E of volume, the median values at the ages of 5 (1.797) and 10 years (1.651) were significantly larger than that at the age of 20 years (1.156).
Furthermore, using DBH and TH at each age as selection traits, the relative indirect gain and efficiencies for gain per year for volume at the age of 30 were estimated (Figure 3).Both the relative indirect gain and efficiencies for gain per year in DBH at ages of 20 and 30 years were larger than those in TH.
Based on the PCA and cluster analysis results, 28 progeny test sites were classified into three climate groups (Group I, 7 progeny test sites; Group II, 9 progeny test sites; and Group III, 12 progeny test sites) (Table 1, Fig. S2).The climatic conditions for each group are shown in Table 2. Group I had a warm and rainy climate, whereas Group II had a cold, relatively rainy summer and snowy winter climate.Group III conditions were typically cold and dry.
To evaluate the variation of h 2 for all growth traits among the different ages, linear mixed-effects models were used (Table 4, Figure 4).According to the ANOVA for the linear mixed-effects models, the regression slopes of h 2 with age were significant, except for in the case of DBH (Table 4).The regression slopes for TH and volume had positive trends (Table 4, Figure 4).Moreover, the interaction of group × age was substantial for DBH and volume (Table 4), while that for TH was marginally significant (p = 0.053).Based on the regression coefficients estimated by the linear mixed-effects models, the regression lines were drawn for each group (Figure 4).According to the group × age interaction, the slopes varied among the groups.In Group I, age trends of h 2 for all traits were consistently positive across the ages (Figure 4); meanwhile, in Group III, the regression slopes were generally flat (Figure 4).We also analyzed the variation of E among the different ages; while the age effects on E were negatively found, no significant group × age interaction was identified for all traits (Table 4, Fig. S3).

Narrow-sense heritability
The heritability of growth traits in various conifer species has been reported as follows: 0.063-0.158for DBH,  and TH in C. japonica (Hiraoka et al. 2019).The h 2 values in the present study were estimated separately for each site as per the usual selection process followed during our tree breeding programs.In general, the heritability of growth traits in conifer species ranges from 0.1 to 0.5; therefore, the h 2 values for C. obtusa in the present study were expected to be within the same range.
However, the estimates of heritability have been shown to be sometimes overestimated if genotypes are tested only in a single environment (White et al. 2007), and this may be the case as well for the values obtained in the present study.

Age -age genetic correlation
In general, r age -age decreases as age differences increase (Hodge and White 1992;Kremer 1992;Gwaze et al. 1997;Rweyongeza 2016;Hiraoka et al. 2019).In Pinus pinaster, the r age -age between height at age 10 and age 50 varies from − 0.136 to 0.764 depending on the different combinations (Kremer 1992).Gwaze et al. (1997) reported that all r age -age values for TH among the ages of 1.5, 9.5, 13.5, and 22.5 years were positive and high (0.76-0.96) in P. teada.In C. japonica, the r age -age values for DBH were 0.847-0.870and 0.765-0.772for the 5-year and 10-year differential pairs, respectively; and 0.531-0.596for the 15-or 20-year differential pairs (Hiraoka et al. 2019).The r age -age and its variability among the ages examined in the present study for C. obtusa were in accordance with the values reported in the above-mentioned studies.

Direct and indirect gains
In the present study, we found that the G r of volume at the age of 30 years (29.1%) was higher than those of DBH and TH (Figure 2).This indicated that direct selection with 33.3% selection intensity can result in a 29% increase in stem volume.Moreover, we found that age trends in genetic gains varied among growth traits.Because G r was here calculated based on the breeding values of parents, it was expected to be influenced by differences in the composition of parents among sites, as previously reported for C. japonica by Hiraoka et al. (2019).In the present study, GI r_xy decreased for all traits as the paired age differences increased, and the same trend was observed for r age -age (Table 3).For DBH, TH, and volume, the median values of GI r_xy were above 40% at all the selected ages.For DBH and TH, as the direct genetic gain at the age of 30 years was approximately 10% (Figure 2), the improvement by indirect selection was expected to be more than 4% (G r × GI r_xy ).
The results of the present study revealed the efficiencies for gain per year (E), assuming a target age of 30 years.For all traits, E showed significantly higher values at the age of 5 years than at the age of 10 or 20 years.However, the among-site variation of E at a very young age (i.e.age of 5 years) was larger than that at the age of 20 years (see vertical ranges of boxplots in Figure 2).These large variations in E, which can cause   selection uncertainty, suggested that selection at the age of 5 years may be excessively premature, and could result in erroneous selection in some cases.Ye and Jayawickrama (2012) suggested that the optimal selection ages (with rotation ages at 50) for TH, stem diameter, and volume in P. menziesii were 9, 13, and 11 years, respectively.In P. taeda, the greatest efficiency was obtained at the ages of 4-6 years, while the target TH was set at the age of 32 years (Lambeth and Dill 2001).Weng et al. (2007) reported that the optimal selection age for volume in P. banksiana ranged between the age 5 and 7 years for a target age of 20 years.The optimal selection ages for growth traits in most species were one-fifth to one-third of the target harvest age.Based on the results obtained in the present study for E and r age -age , and considering the need to balance the earliness and effectiveness of selection as well as its uncertainty, it was concluded that the optimal age for the early selection of C. obtusa was 10 years.Furthermore, the comparison of GI r_xy and E for volume at the age of 30 years at the same age between DBH and TH, showed that those of TH was more varied among sites than that for DBH (Figure 3).This indicated that DBH might be a more effective criterion for the selection of stem volume with less erroneous by sites than TH (Figure 3).

Age trends of narrow-sense heritability and efficiencies for gain per year in relation to climatic conditions
In the present study, the h 2 values of all traits were not significantly different among ages (Figure 2) but varied  significantly (except for DBH) among different ages, as shown by the linear mixed-effect ANOVA model (Table 4).We also found a significant effect of group × age interaction on the relationship between h 2 values among ages, suggesting that the age trends for the h 2 of these traits (except for TH) varied among climate groups which were classified based on the climatic conditions.In Group I, associated with a wetter climate, the h 2 values of all growth traits increased with age.In contrast, in Group III, associated with a drier climate, the values of DBH and volume, showed either a decreasing trend or no variation with age.In Group II, the h 2 of DBH was almost stable across age, whereas those of TH and volume increased slightly with age compared with the values observed in Group I.These results suggested that age trends in the genetic control of growth traits in C. obtusa may be affected by climatic conditions.Yamaya et al. (1984) maintained that C. obtusa grows better if exposed to limited snow in winter and heavy rain in summer, which suggests that the climatic conditions of Group I in our study were better for the growth of this species than those of the other two groups (Table 2).Overall, the present study supports the hypothesis that the genetic control of growth traits in C. obtusa increases with age at sites with suitable climatic conditions.In contrast, in sites subjected to heavy snow in winter and limited rainfall during the growing season, growth may be suppressed, causing lower genetic variation and/or greater environmental variation in growth traits, which results in lower h 2 .On the other hand, the absence of significant effects of the interaction found for E in this study.The E is calculated using GI r_xy and spent time on seed orchard establishment and improvement, as well as seed production (Formula 5).The GI r_xy of growth traits generally tended to increase with ages (Figure 2).These results suggested that the E were influenced by spent time considered stronger than by GI r_xy , which might be influenced by many factors such h 2 , genetic correlation and site mean.It was thus suggested that early selection is possible with almost equal efficiency in all climate groups with different climatic conditions when time efficiency is considered.However, the variation detected in the present study may not maintain the same trend as climate change is expected to intensify in the future due to global warming.In addition, the growth responses of C. obtusa may be affected not only by macro-environmental factors (i.e.climatic conditions), but also by microenvironmental factors, such as soil type, slope orientation, and slope inclination.Future studies exploring the detailed relationships between genetic parameters and macro-and microenvironmental factors will provide a deeper understanding of the heritability of growth traits and should also be useful for improving the strategies used in breeding programs targeting C. obtusa.

Conclusion
In the present study, we analyzed a multiage dataset of growth traits (DBH, TH, and volume) obtained from C. obtusa growing at 28 progeny test sites covering various climatic conditions in the Kanto breeding region of Japan.The analysis estimated the narrow-sense heritability of these growth traits, and the values obtained were considered to be comparable to those observed in other coniferous species.For all growth traits, the E was significantly higher at the age of 5 years than at the age of 10 or 20 years, although the among-site variation in E was larger at the former age than at the latter ages.Considering the need to balance the earliness, effectiveness, and uncertainty of selection, it was suggested that the optimal age to conduct the early selection of C. obtusa is 10 years.
We then evaluated whether the heritability of growth traits might show increasing or decreasing trends with age, and/or whether these trends varied among different climate groups.The results showed that the heritability of TH and volume varied significantly among ages, and the group × age interaction also had a significant effect on the age trends in heritability.This indicated that regional variations in these trends may be caused by environmental factors, such as climatic conditions.The genetic control of growth traits depending on age was also suggested to differ across climatic conditions, implying the need to investigate the effects of the environment on genetic control and evaluate the characteristics of breeding materials in order to develop more effective breeding strategies for C. obtusa.

Figure 1 .
Figure 1.Map of analyzed progeny test sites in the Kanto breeding region.Note: The numbers from 1 to 28 represent Site ID.The shaded area shows the Kanto breeding region.

Figure 2 .
Figure 2. Age trends of relative direct gain, relative indirect gain, and efficiencies for gain per year in each growth trait.Note: The same letters are given when there is no significant difference based on p-values from Wilcoxon rank-sum test.The relative direct gain for volume in the ages of 5 and 10 years were not shown because the logarithmically transformed values were overstated due to the smaller measured values.
in parentheses represent interquartile ranges.The same letters are given when there is no significant difference at 5% level based on p-values from Wilcoxon rank-sum test.DBH, diameter at breast height; TH, tree height.

Figure 3 .
Figure 3. Age trends of relative indirect gain and efficiencies for gain per year in DBH and TH for 30-year volume as a target trait.Note: The same letters are given when there is no significant difference among ages based on p-values from Wilcoxon rank-sum test.**, significant difference at 1% level between indirect relative gain in DBH and TH at same age.

Figure 4 .
Figure 4. Age trends of narrow-sense heritability of growth traits in each climate group.Note: Open circles in gray color represent narrow-sense heritability in each site.Gray lines connect the measurement year of each progeny test site.The black lines were drawn based on the regression coefficients of fixed-effects parameters of linear mixed-effect model.

Table 1 .
Summary of all progeny test sites.
Note: DBH, diameter at breast height; TH, tree height.Progeny test site ID from 1 to 28 corresponds with the ID illustrated in Figure1.Climate groups (I, II, and III) are grouped by PCA and cluster analysis using climatic conditions as variables.Open and Full indicate that the progeny site consisted of open-and full-sib families, respectively.N p indicates the total number of parents in families.

Table 2 .
Environmental conditions of each climate group classified using clustering analysis.

Table 3 .
Median values of age -age genetic correlation (r age -age ) in growth traits.

Table 4 .
Results of ANOVA for the linear mixed-effect model testing whether the h 2 and E of each trait along ages varies among groups.
Note: The table shows F-values and p-values (in parentheses).The values in bold indicate statistical significance at the level of p < 0.05.DBH, diameter at breast height, TH, tree height.