Zhangqi Yang · Hui Xia · Jianhui Tan · Yuanheng Feng · Yongli Huang

Abstract This study addresses the increasing demand for large-diameter production timber, and considers the time and space variability of half-sib families of Pinus massoniana. Height, diameter at breast height (DBH) and timber volume of 440 open-pollinated half-sib progeny families were investigated in 14 progeny trials in different years and production regions. An evaluation of the genetic variation of all half-sib families was carried out during the sustainable rapid growth period and individual volumes were characterized as a major index. ANOVA analysis showed that there was considerable variance in the growth traits of most families in different years and on different sites. The variations caused by temporal and spatial changes of the mating system required three selection methods for analysis. The results show that there were differences among the heritabilities of different growth traits by different halfsib progenies. Average heritability values of height, DBH and volume were 0.33, 0.34 and 0.36, respectively. Fortyfive superior families were selected in every progeny test,12 were selected in progeny trials by different years and five in different habitat progeny trials. Three superior families (Gui GC553A, Gui GC414A and Gui GC431A)were selected, although in different years and production regions. The genetic gains of timber volume of these selected r families ranged from 1.20 to 47.00%, which could provide a foundation for superior wood property selection and serve as material for seed improvement and extension in surrounding areas.

Keywords Pinus massoniana · Open-pollinated progeny ·Large-diameter timber · Superior family selection · Mating system

Introduction

Due to its adaptability and rapid growth, Pinus massoniana is a unique native species of southern China (Li 1990; Zhen et al. 2012) widely distributed in the 17 provinces and cities east of the Yunnan-Guizhou Plateau with a cultivated area of about 200 million km2. P. massoniana, Chinese red pine, has become one of the most significant coniferous species with large stand volumes, and economic and ecological values (Li 1990). Owing to its high cellulose content, long fibre length and diameters and other excellent wood properties, P. massoniana has a variety of uses and application, especially large-diameter timber for pulp and paper production and has also become a main construction timber species (Zhen et al. 2002; Yang et al. 2003).

However, with limited suppplies and increasing demands for timber for building construction, there is considerable potential for a large-diameter P. massoniana timber market. The selection of superior families for largediameter timber is important for improving wood properties. Several studies have reported on P. massoniana for improving pulp and paper yields based on provenance trials(Xu and Wang 2001), on superior families or clone selection and genetic variances in growth increments (Yang 2006), for biomass modelling (Fu et al. 2014), resin yields(Zeng et al. 2013) and chemical composition(Zhou et al.2000). Previous studies noted that the correlation between wood properties and growth traits was weak (Weber and Montes 2005), and therefore the breeding for building timber mainly focused on growth increments and wood densities. However, Liu et al. (2013) found that the growth traits of P. massoniana families were mainly controlled by genetic factors, and differences in wood densities among families was insignificant. In addition, selection for largediameter timber would be more reliable after the trees entered their initial rapid growth period (Macdonald and Hubert 2002). Thus the selection for large-diameter timber traits with growth increment as the main index is the basis of genetic improvement in this study. However, although the genetic improvement of P. massoniana from a focus only on wood yield gradually turned to timber production and quality improvement, previous studies also only concentrated on a single production region for 1 year, which could not determine or guarantee progeny performance stability (Ablett et al. 1989). Hence, multi-year and multisite superior family selection became more acceptable.

Two factors need to be given attention for the optimal selection of superior families for large-diameter construction timber of P. massoniana Selection should proceed after the stand volume has entered a sustainable, rapid growth period. Male parents of open-pollinated progeny are diverse and poorly understood, and therefore the time and spatial changes of their mating system and offspring should be considered (Yuan et al. 2014). As it is more accurate to select individuals during period of rapid volume growth which is 15-25 years in P. massoniana, 14 openpollinated progeny stands, which had been investigated for several years, were identified to carry out the family selection in different years and production regions. This could provide a theoretical basis for genetic improvement and breeding of P. massoniana. The selected families could also be used as materials for wood properties selection and widely applied.

Materials and methods

Experimental sites

The experiments were carried out on two sites, one at the Nanning Institute of Forestry Science located in Nanning City, Guangxi province, (108°00′E, 23°10′N) on the northern margin of a tropical monsoon climate. The annual average temperature is 21.5 °C, the frost season 23 days,and precipitation 1246 mm. The altitude of the progeny trials is about 120 m and the terrain is flat.

The other site is the Teng County Forestry Institute in Wuzhou City, Guangxi province (110°00′E, 23°24′N)which has a subtropical humid monsoon climate. Annual average temperature and precipitation are 21.3 °C and 1250 mm, respectively. The annual frost-free season is 305-330 days. The progeny trial sites are hilly and altitudes about 100 m. The soil of the two progeny trials is latosolic red soil suitable for the growth of P. massoniana.

Experimental materials and designs

There were 440 families which came from the first generation seed orchard of P. massoniana in the Fourteen open-pollinated half-sib progeny trials, eight progeny trials(N88, N92A, N92B, N92C, N94A, N94B, N94C, N94D)from the Nanning Institute of Forestry Science and six(D87, D90A, D90B, D90C, D92, D93) from the Teng County Forestry Institute. All progeny trials had different controls in a randomized complete block design. The experimental design, the general afforestation situation and the sources of the 14 progeny trials are shown in Table 1.

Statistical analysis

In 2010, height and DBH of all trees of different families of the N88 progeny trial were measured before thinning,while height and DBH of other progeny trials were measured in the winter of 2008 during dormancy. Survival rates of all families were higher than 90%. Volumes were calculated by:

Statistical analyses used SAS 8.1 software. A linear model was used for joint analyses of the families together by blocks to carry out the ANOVA analysis using Eq. 2(Alwi et al. 1986; Xu 2006):

where Yijkis the family performance in the kth block, u the overall mean, Fithe effect of families, Bjthe effect of blocks, FBijthe interactive effect of the ith family and jth block, Eijkthe random error.

Table 1 General situation of the progeny tests

Family heritability was calculated according to Xu(2006):

The individual heritability of a character was calculated according to Hansen and Roulund (1997) and Xu (2006):

The coefficient of variation (CV) was calculated as:

where SD is the standard deviation of a character,the group average value of a character.

Genetic gain was calculated using the following formula(Xiang et al. 2003):

where ΔG is the estimated genetic gain of a character, S the selection difference,the average value of a character.

Selection of superior families

The growth performance of families and family stability brought about by temporal and spatial changes of the mating system in seed orchard should be considered. Three methods from three different trials were used to select superior families: (1) the selection of growth performance stability within families using the Francis and Kannenberg model analysis according to Wang et al. (2015), which should meet the following conditions, that the variation coefficient was in the top 30% of the minimum values, and the average volume was in the top 10% of the maximum;(2) the selection of stable performance in different years. If the number of times the average timber volume of the family ranked in the top 10% accounted for more than 50%of the number of tests, the family was identified as superior with good growth performance and stable production in different years; and, (3) the selection of performance stability in different production regions. With the comparison of different progeny trials from two different seed orchards,the family whose average timber volume ranked in the top 10% in two production regions could be characterized as superior with good growth performance and stable production quality in different production regions.

Results

ANOVA analysis

The ANOVA analysis is shown in Table 2. Regardless of the insignificant differences between DBH and volume of the families in the D90B progeny test, and except for the significant differences among family volumes in the D92A progeny test, the differences in height, DBH, and volume among families were highly significant, which suggests that there exists considerable genetic variation in growth traits of the half-sib families and could be valuable for genetic improvement of the species.

Analysis of growth trait heritabilities

The heritabilities of different progenies were different with height, DBH and volume of all progeny tests being 0.33,0.34 and 0.36, respectively (Table 3). The average individual heritabilities of height, DBH and volume were 0.22,0.18 and 0.19, respectively. For the different traits, family heritabilities were higher than individual heritabilities in most progeny tests. The average family heritability for volume was the highest. On the other hand, except for the family heritability of volume in D87 (0.04) and D90A(0.16), progeny tests t were lower. Family heritability of most other progenies were higher than 0.20 and had moderate or higher levels heritability, which suggests that it was beneficial to carry out the selection for superior families.

Superior family selection

Using the first method mentioned previously, the selection of performance stability within families in all single progeny tests except for D90B, D87 and D90A, identified 45 superior families with good, stable growth (Table 4). The average genetic gain in height, DBH and volume of these families was 2.6, 7.6 and 20.6%, respectively. The highest genetic gain in height (7.1%) was by the Gui GC542A family, while the Gui GC420A family had the highest genetic gains in DBH and volume (33.9% and 69.8%respectively). The Gui GC433A family was selected as a superior family twice. Because of insignificant differences in volume among families of D90B and lower family heritabilities in volume of D87 and D90A, these three were not included in the superior family selection.

Table 2 ANOVA analysis of H, DBH and V of tested families in all the progeny tests

Table 3 Heritability analysis of growth traits

Table 4 Superior family selection in different progeny tests

Superior family selection from the same seed orchard in different years

Comparing nine progeny tests from the Nanning Institute of Forestry Science seed orchard and five from the Damangjie seed orchard in different years, and using the second method, 12 families were selected as superior with good growth performances and stable progeny in different years, eight families were also selected using the first method (Table 5). In the 14 progeny tests, different families were selected at different times, the average was 2.6 times; only 12 families from different years were selected as superior several times, which suggests that there was a large variation among the half-sib progeny from the same clones in different years, and that the progeny quality would vary with changes in reproductive patterns. The ranges of genetic gains in volume for different families from the same seed orchard were different. For the Damangjie seed orchard, the range in volume gain varied from 1.2 to 20.0% and the highest ranges in volume gain was by the Gui GC462A family (from 12.3 to 20.0%).However, the range in volume gains for selected families of the Nanning seed orchard varied from 6.6 to 30.7%; the largest range in volume gain was by the Gui GC416A family (from 8.8 to 30.7%).

Table 5 Superior family selection in multiyear progeny tests

Superior family selection of progeny from different seed orchards

Selection on half-sib progeny families from the same parent clones grown in both the Nanning Institute of Forestry Science seed orchard and the Damangjie seed orchard using the third method. Five families with good growth and stable seed production quality were selected in these two production regions (Table 6). The Gui GC553A, Gui GC414A and Gui GC431A families were selected in all the three methods noted previously. The range of genetic gains in volume varied from 1.7 to 14.5% and the largest range was with Gui GC414A (4.4-14.5%). There were 106 families owned by the two seed orchards but only five were selected from different production regions, indicating that there were large variations among the half-sib progenies from the same parental clone of different seed orchards.

The performance and quality of progeny would varied with the parental mating system.

Discussion

ANOVA analysis

Genetics and variation in economic characters are the basis of selection. ANOVA analysis is an important method to evaluate the degree of mutation of a breeding population(Zhao et al. 2014). In this study, ANOVA analysis of the growth traits show significant differences (P ＜0.01)among families of most progenies. Similar results have been reported for larch (Xia et al. 2016). The results of the study on P. massoniana by Yang et al. (2003) also showed that the variations in wood properties of families were statistically significant and varied, and larger variancescould be applied to the selection of superior families. This suggests that wood properties were important for the selection of superior families for genetic improvement.

Table 6 Superior family selection of the different production regions

Heritabilities of growth traits

Heritability is an important parameter in predicting genetic gain which can reflect the relative genetic variance of a certain trait in a population (Xu 2006). In this study, family and individual heritabilities of various growth traits varied differently. The average family heritabilities, which were moderate or higher (≥0.2), were higher than individual heritabilities. Individual selection of P. massoniana family was vulnerable to environmental factors. Similar results have also been reported for Fraser fir (Emerson 2005) and Populus nigra L. (Lsik and Toplu 2004), implying that considerable gain may be realized by selective improvement methods. The higher the family heritability, the larger genetic differences, heterogeneity and genotypic differences, and genetic improvement potential would be revealed in the half-sib family groups (Lshaq et al. 2015).Heritability may vary for different growth traits. In this study, the average family heritability for volume was the highest, which was more stable and less affected in response to environment changes. Similar results have been reported for Larix gmelinii (Rupr.) Kuzen. (Li et al. 2012).There may be more scope for improving volume than for other traits, and previous estimates by Zhen et al. (2002) on P. massoniana achieved greater improvement with selecting for volume.

Spatiotemporal and reproductive variations of halfsib families

According to production requirements, P. massoniana varieties must have good growth characteristics, stability and ecological adaptability. Compared to mixed families,half-sib families had smaller variabilities and production difficulties were less than for full-sib families, and had better ecological adaptability (Shoemake and Arnold 1995). This indicates that half-sib families were more suitable for the production of superior seed supplies.

In this study, the differences in family traits on different sites varied and levels of heritability were also different,suggesting that site differences had an effect on mutation and heritability (Zhao et al. 2014). Male parents were irregular in half-sib families, and spatial and temporal variations of different seed orchards, ramets and even different years of the mating system could be significantly varied (Erickson and Adams 1990). Zhang et al. (2004)study on the reproduction of Pinus tabulaeformis Carr.found that the outcrossing rate varied greatly before and after intermediate felling, and for different ramets in different years. The differences of outcrossing rate and pollen contamination were also significant. Moreover, Jin et al. (2008)’s report on the combining ability of full-sib family growth traits of P. massoniana showed that male parental combining abilities were greatly different, suggesting that the genetic quality of the progeny would be affected as male parent alters. In addition, with previous studies mainly focused on a particular age in 1 year or site(Ji et al. 2005), variation could not be accurately estimated due to parental reproduction changes and spatiotemporal variance. Furtini et al. (2012) found that the origin of the individual had a small effect on the performance of the clone when evaluated in other environments. These observations suggest that it is important to ensure progeny quality in different years and sites. The selected families were not suitable for mass production and use as mature superior seed sources (Zhao et al. 2014). Therefore, on the basis of each progeny test, conducting the associated and separate selection in terms of years and production regions would benefit superior family selection.

Superior family selection of single progeny trials

Progeny determination plays a crucial role in tree genetic improvement. Fries et al. (2000) showed that wood extractives with high heritability from Pinus sylvestris L.could be selected by progeny tests after genetic improvement had been carried out, and Zhao et al. (2013) found that growth and quality traits of Juglans nigra L. (black walnut) could be improved by progeny trials. In our study,the first method of selecting growth performance stability within families was applied in all progeny tests. Forty-five families were selected from 440 open-pollinated half-sib families in 14 progeny trials. The number of excellent families selected from different progenies was different,which may be related to the number of families and trial conditions. In addition, it may also be associated with seed sources of various female parent clones (Arnold and Cuevas 2003), the significance being that they could show good growth performance and stable offspring under a specific mating pattern. Similar observations are found with research on lodgepole pine (Xie et al. 2007). Female parental clones had a certain advantage in combining ability which could serve for material for improved genetic seed orchards and for the generation of superior seeds under a modified and more stable mating system (Yuan et al.2014). The parent clones of selected families could also be selected into a core breeding group as the next generation parents of genetic improvement.

Superior family selection from the same seed orchard in different years

Growth performance was mainly affected by genetic and environmental factors, and the differences varied with age(Mencuccini et al. 2007). Diao et al. (2016) noted that age could affect the optimum selection period, and that the effects of age changed with different traits. However, it is unreliable to conduct the evaluation in only 1 year (Johnson et al. 1997). In this study, 12 superior families were selected using the second method (the selection of stable performance in different years), and eight families selected by the first method among these 12, indicating that these 8 families could maintain stability in different years and with different progenies. Some families from the same seed orchard showed little change in different years. Xia et al. 2016 reported that superior families of Larix olgensis A. Henry could be selected from the 4th year to the 10th year. Several measurements at different ages should be implemented to enhance reliability while selecting. Early selection of Larix kaempferi (Lamb.) Carr. showed that optimum height and DBH were different at different ages,and the optimum age for height was 4 years and five for DBH (Diao et al. 2016). These different results might be correlated with different experimental ages, planting designs and/or species. The fact that it is valid to determine early selection age over several years was also supported by studies on P. massoniana (Chen 2011) and hybrid larch(Miao et al. 2017). In addition, the results also show that parental clones of superior families have high combining abilities and could ensure stability of their offspring in the dynamic breeding system (Zhang et al. 2004). Further,superior families can be used directly for promotion, but the improved seed orchards established with these parental clones need stability measurements.

Selection of superior families from different production regions

Growth performances of families from different production regions varied (Bian et al. 2014). Regionalization is an important strategy in breeding programs (Diao et al. 2016).Using the third method (selection of performance stability from different production regions), five families were selected from two regions, suggesting that environmental conditions in different regions have an uncertain effect on the performance of most progeny families. The study by Stoehr et al. (1998) of white spruce showed that differences in environmental conditions between seed orchard location and location of origin could affect progeny performance and alter physiological traits. The change of spatial location was contained in the change of production regions,which would influence genetic analysis. Chen et al. (2018)improved the efficiency of genetic improvement by analyzing the spatial variability of Norway spruce, and in the study by Lin et al. (2017), it was noted that using different spatial analysis models could improve the genetic analysis model. Thus, the environmental variation of the production region should be considered in the selection. In this study,the selected superior families are the product of these two production regions and could provide the basis for the study of P. massoniana geographic variations. Moreover,these five families were also all selected by the first method, suggesting that these parental clones could produce progeny with good growth performance and stable variation under two different production regions The Gui GC553A, Gui GC414A and Gui GC431A families were selected by the three methods, indicating that they could be superior families with little variation and environmental impact for the production of large-diameter timber, and could serve as primary seed sources for demonstration trials and extension in cultivated areas.

Because of variations caused by time and spatial changes of the reproduction system (Kassaby and Ritland 1986), growth variation caused by the effects of time and spatial changes should be considered. Multiple sets of progeny from the same female parents over several years and production regions also should be used to carry out progeny tests on several sites in the same year. With the absence of ideal experimental materials at present, in this study we took the progeny from production regions of different years as trial material, which could preliminarily explore the stability of the performance of half-sib families under different mating systems (Chybicki and Burczyk 2013). However, due to the limited material, the degree of variation caused by time and spatial changes of the mating system of every family could not be accurately estimated.In order to ensure reliability only by improving the selection intensity, the number of selected families was less in this study. If more accurate selection criteria could be made for these variances in the future research, more superior families would be selected for production through the genetic improvement research process.

Conclusions

The breeding and selection of superior families for largediameter timber production has considerable potential through the genetic improvement of P. massoniana. This study found that there were abundant phenotypic variations and moderate heritability in volume traits of P. massoniana, which was beneficial to the selection of superior families. Although spatiotemporal changes and reproductive systems had effects on growth traits and genetic variations, 45 superior families were selected from all progeny tests, 12 were selected by different years and five were selected from different production regions. Three families, (Gui GC553A, Gui GC414A and Gui GC431A),were selected using three selection methods. These families could consequently serve as material for seed improvement and for promotion in local areas. Based on this study, goaldirected improved breeding groups and improved seed orchards can be established. Genetic variation and special combining ability of full-sib families, and individual superior tree selection focused on wood density and other timber indices should the focus of future work.