Introduction

Masson pine (Pinus massoniana) is an important afforestation and timber species, particularly in China, where it is widely distributed in all regions south of the Yangtze River Basin. Additionally, this species occupies the most extensive afforestation area in southern China. The breeding of Masson pine has always been a key emphasis in the work of Chinese tree breeders. Seed orchards are the main source of excellent seeds of P. massoniana and are thus established with the purpose of sustained and efficient production of seeds with high genetic quality. Parental selection is an important process that should not be overlooked during the construction of seed orchards. This process can be approached in two ways, namely, a priori and a posteriori selection. A posteriori selection is always used to assess the genetic quality of parents based on progeny performance, but it is a long and tedious process for perennial woody plants. By contrast, a priori selection based on parent–offspring relationships may simplify and shorten the process of parental selection (Dias et al. 2004).

Based on molecular data, some previous studies have observed significant positive correlations between parental genetic distance (GD) and the growth performance of hybrid offspring (Tian et al. 2016; Zhang et al. 2013). Furthermore, these correlations are known to differ among plant species, but few studies on tree species have been reported. Although relationships between parental GD and offspring performance have been reported in trees such as Liriodendron (Yao et al. 2015) and Eucalyptus grandis (Abad et al. 2005), nonsignificant or weak correlations were observed between parental GD and offspring performance because of the limited number and type of molecular markers used. In some previous studies, random amplified polymorphic DNA (RAPD) (Krystkowiak et al. 2009; Kuczyńska et al. 2007), amplified fragment length polymorphism (AFLP) (Legesse et al. 2008) and sequence-related amplified polymorphism (SRAP) (Tian et al. 2016) markers were used to evaluate GDs and predict the performance of offspring. With the development of molecular markers and high-throughput sequencing technology, simple sequence repeat (SSR) (Yao et al. 2015; Xu et al. 2017; Li et al. 2018) and single nucleotide polymorphism (SNP) (Luo et al. 2016; Ndhlela et al. 2015) markers have gradually become mainstream tools used in related research. SSR markers in particular are widely used in such studies due to their advantages of codominance, high polymorphism and low cost (Xing et al. 2014; Wegary et al. 2013; Yao et al. 2015; Grattapaglia et al. 2014).

Compared with crops, trees have a longer life history and breeding cycle, making the evaluation of offspring performance in the process of tree breeding extremely time consuming. Additionally, a shorter receptive period and high rainfall in pollination-time, cone maturation in late autumn next year make artificial controlled pollination of P. massoniana difficult. Moreover, manual controlled pollination of P. massoniana is cumbersome and numerous. All the factors mentioned above limit studies on the relationships between parental GD and offspring performance through artificially controlled hybridization. Hence, using half-sib families of P. massoniana is particularly important in related research. However, what effect does the parental combination have on an offspring’s performance? Is the genetic quality of seeds in seed orchards consistent among years? What are the main factors that affect the stability of seed genetic quality among years? This study aimed to investigate the relationships of parental GD with the stability of seed genetic quality among years and with offspring performance in open-pollination (OP) seed orchards of P. massoniana to answer the questions presented above and to provide guidance for parental selection during the construction of seed orchards.

Materials and methods

Parents, progeny and field trials

The clonal seed orchard of P. massoniana was built in 1985 with rootstocks and is located in the Wuyi Forest Farm in Zhangping, Fujiang Province (24° 58.072′ N, 117° 27.509′ E). A total of 218 clones were grafted in 1987, with a planting density of 4 m × 4 m. The parental clones originated from some provinces in which the species is naturally distributed in China. OP seeds were harvested in 1993, 1994 and 1995, and the progeny tests were established in 1995, 1996 and 1997. Seedlings of half-sib progeny were planted in a randomized complete block design with 8 replicates and 5 seedlings in each plot. The growth traits [including height (H) and diameter at breast height (DBH)] were measured when the progeny were 14 years old. The wood volume (V) was estimated with the following formula (Zhang et al. 2013):

$$V = 0.000062341803 \times {\text{DBH}}^{1.8561497} \times {\text{H}}^{0.9568492} .$$

Selection of half-sib families

A total of 72 shared half-sib families were screened from the 3 progeny tests (2475 individuals). Standard deviations (SDs) of V were used to measure the growth stability of shared half-sib families across the 3 progeny tests. The 9 half-sib families with the lowest SD values were screened as the stable groups (SGs), and the 9 half-sib families with the highest SD values were screened as the unstable groups (USGs) (Table 1). There were 369 individuals of SGs and USGs used in paternity analysis and other further analysis. In addition, all progeny screened as described above were also divided into three groups based on a descending ranking of average V [the top-four half-sib families with the highest values as the excellent group (EG), the last-four half-sib families with the lowest values as the worst group (WG), and the medianfour half-sib families with intermediate values as the middle group (MG)] (Table 1). It should be noted that three less differentiated families were respectively removed between two groups of EG, MG and WG (EG–MG and MG–WG) in order to show significant differences among groups.

Table 1 Half-sib families in groups based on different classification bases
Full size table

Sample collection and PCR amplification

All needles were collected in 2014 and stored at − 70 °C. Total genomic DNA was extracted from fresh leaves using a Plant Genomic DNA Kit (ZomanBio Inc., Beijing, China) and quantified with a NanoDrop 2000c spectrophotometer (ThermoFisher Scientific, Waltham, USA). A total of 27 polymorphic SSR markers were used in this study (Supplementary Table 1) (Bai et al. 2014). SSR-PCR was performed in a 10 μL reaction mixture containing 50 ng of DNA template, 1.0 μL of 10 × PCR buffer (10 mM Tris–HCl), 2.5 mM MgCl2, 0.25 mM dNTPs, 0.3 μM primer, and 0.5 U of Taq polymerase (Takara Bio Inc., Dalian). The PCR program was as follows: an initial denaturation at 94 °C for 5 min; 25 cycles of 30 s at 94 °C, 30 s at the annealing temperature, and 40 s at 72 °C; and a final extension at 72 °C for 5 min. PCR products were tested by using 8% polyacrylamide gels for 2.5 h at 150 V in a vertical gel electrophoresis apparatus (Liuyi Co., Ltd., Beijing). PCR products of 5 SSR markers were measured using a Qsep100 capillary gel electrophoresis system (BiOptic Inc., Taipei).

Data analysis

The genetic parameters, including the number of alleles (Na) per locus, Nei’s gene diversity (h) and the polymorphism information content (PIC), were estimated with POPGENE 32 version 1.32 (Yeh 1999) and PowerMarker version 3.25 (Liu and Muse 2005). Nei’s genetic distance of parents were estimated using PowerMarker version 3.25 (Liu and Muse 2005; Nei 1978). A total of 369 individuals of SGs and USGs were used for paternity analysis with 18 known females and 217 candidate males using the revised likelihood-based paternity inference program CERVUS 3.0.7 (Kalinowski et al. 2007) by using known females and 217 candidate males. Pearson’s correlation coefficients between parental GD and the V of progeny were calculated using a custom script in R 3.4.0. The ggplot2 package in R was used to construct a scatter diagram and other plots (Wickham 2009).

Results

Genetic diversities of parents and progeny

In this study, 27 SSRs were used to determine the genetic diversity of parents and progeny. A total of 164 alleles were detected with SSR markers (2–18 alleles per locus), resulting in 5.86 alleles per locus on average. In addition, the PIC of polymorphic SSRs ranged from 0.269 to 0.869. Botstein et al. (1980) noted that codominant loci may be highly informative (PIC > 0.5), reasonably informative (0.5 > PIC > 0.25) and slightly informative (PIC < 0.25) (Botstein et al. 1980). Most of these polymorphic SSRs were highly or reasonably informative and could be useful in the study of genetic diversity or in parentage analysis. The primers PMa25a (10 alleles), PMa25b (12 alleles), PMa51 (14 alleles), PMa72 (18 alleles), PMa96 (12 alleles) and WX005 (14 alleles) amplified more alleles than the others because the capillary gel electrophoresis system was used for these markers (PMa25a and PMa25b are two loci of the primer P.Ma25) (Table 2).

Table 2 The polymorphism of SSR markers used in this study
Full size table

High genetic diversity is an important genetic prerequisite for seed breeding that can produce more genetic variation. In this study, an analysis of genetic diversity among parents and progeny (1995, 1996 and 1997) in orchards was performed (Table 3). The Shannon’s information index of genetic diversity among parents and for the progeny tests (1995, 1996 and 1997) was 1.1538, 1.1665, 1.1777 and 1.1683, respectively, and the corresponding values of Nei’s gene diversity were 0.6027, 0.6011, 0.6079 and 0.6051, respectively. All these parameters of genetic diversity indicated that parents and progeny had high genetic diversity and were not significantly different.

Table 3 The genetic diversities of parents and progeny in 1995, 1996 and 1997
Full size table

Paternity analyses of samples

To investigate the parents of individuals in the SGs and USGs in the 3 progeny tests (1995, 1996 and 1997), parental analysis was performed with 217 candidate parents based on 27 pairs of SSR primers. Male parents of progeny in different groups were analyzed statistically under confidence levels of 95% and 80% (Table 4). The progeny of the SGs and USGs had different determination rates ranging from 0.1892 to 0.2429 (confidence level = 95%), while the determination rates were mainly in the range of 0.5405–0.8154 under a confidence level of 80%.

Table 4 Paternity analyses of different groups of progeny in a Pinus massoniana seed orchard
Full size table

Distributions of parental genetic distances among groups

The purpose of seed orchards is to produce seeds with high genetic quality, and parents have important effects on the genetic quality of seeds. To investigate the relationships between parents and the stability of seed genetic quality among years, the relationships between parental GD and the stability of progeny performance among 1995, 1996 and 1997 was assessed by using parents determined in the parental analysis (confidence level = 80%).

In this study, we compared the distributions of parental GDs among groups in different years to assess relationships between parents and the stability of seed genetic quality among years (Fig. 1a). A significant difference in parental GD was found between the SGs and USGs (p < 0.001), and the average parental GD of the SG (mean = 0.4251) was much lower than that of the USGs (mean = 0.6019). Compared with the USGs (the proportion of parents (PPs) = 0.271 ~ 0.358), the SGs had more parent pairs (PPs = 0.702 ~ 0.750) for which the GD was less than 0.5, while the SGs had fewer parent pairs (t PPs = 0.191 ~ 0.225) than the USGs (PPs = 0.396 ~ 0.492) for which the GD was more than 0.6. All of these factors indicated that the distribution of parental GDs in the SGs were narrow, while that in the USGs were more dispersed.

Fig. 1
figure 1

Distribution of parental genetic distances between groups. a Distribution of parental genetic distances between the stable and unstable groups; b distribution of parental genetic distances among the best, middle and worst groups

Full size image

Additionally, the relationships between parental GDs and progeny performance in terms of V were evaluated among the EG, MG and WG. The average GDs among groups were as follows: WG (0.4634) < MG (0.4993) < EG (0.5613). Additionally, the proportion of parent pairs for which the GD was less than 0.5 was highest in the WG (PPs = 0.639), followed by the MG (PPs = 0.549) and the EG (PPs = 0.486), while the proportion of parent pairs for which the GD was greater than 0.6 was larger in the EG (PPs = 0.419) than in the MG (PPs = 0.232) and WG (PPs = 0.301) (Fig. 1b). In total, the EG contained greater and more discrete parental GDs than the WG. Hence, use of the EG makes it possible to produce more genetic variation. A further comparison of the progeny performance in terms of V between the EG and WG was made (Fig. 2). The EG had superior performance in terms of V than the WG, and the maximum V was found in the EG. In summary, parents with distant GDs in the EG can produce more high-quality variation, and their offspring are better able to grow.

Fig. 2
figure 2

Distribution of progeny wood volumes between the best groups and worst groups. EG excellent groups, WG worst groups

Full size image

Correlations between parental genetic distance and progeny growth

Positive and weak correlations were detected between parental GD and progeny performance (H, DBH and V), and the correlation coefficients ranged from 0.146 to 0.225. Parental GD was significantly correlated with V and DBH (p < 0.05 or p < 0.01) among ranges of parental GDs, while no significant correlations with H were detected among most ranges of parental GDs (Table 5). Further regression analysis indicated that differentiation was also more extreme (pros and cons of coexistence) when the parental GDs were greater (especially when the GD > 0.6) (Fig. 3). The parental GDs of offspring with maximum values of V (0.4592 m3) and DBH (0.276 m) were greater than 0.6. Hence, we should focus on choosing parents with greater GDs, which are likely to produce offspring with excellent performance.

Table 5 Correlations between parental genetic distance and progeny performance
Full size table
Fig. 3
figure 3

Linear correlation between parental genetic distance and progeny performance

Full size image

Discussion

In the process of plant breeding, parental selection plays an important role in producing offspring with excellent performance (Xing et al. 2014; Hallingbäck and Jansson 2013). The relationships between parental GD and progeny performance have always been a key issue in cross-breeding (Legesse et al. 2008; Sang et al. 2015; Xing et al. 2014), and correlations have been detected in some plants (Tambasco-Talhari et al. 2005; Zhang et al. 2013; Hung et al. 2012; Wegary et al. 2013). Undoubtedly, significant and strong correlations between parental GD and progeny performance can provide guidance in predicting the growth performance of offspring (D’Andrea et al. 2013). Although a strong correlation between parental GD and progeny performance has been reported in some plants, such as Salix babylonica (Teklewold and Becker 2006), Triticum aestivum (Krystkowiak et al. 2009), and Zea mays (Makumbi et al. 2011), weak and nonsignificant correlations have also been found in many previous studies (Yao et al. 2015; Luo et al. 2016; Tian et al. 2016). In this study, significant but weak correlations were found between parental GD and the V and DBH of progeny, which was consistent with the results of previous studies (Luo et al. 2016; Ndhlela et al. 2015). However, weak correlations are not powerful in predicting the performance of progeny for two reasons: (1) the molecular markers used in studies are not associated with genes or loci controlling quantitative traits; and (2) genetic and environmental interactions can also result in weak correlations (Yao et al. 2015; Tian et al. 2016). Hence, new polymorphic markers (SNPs) (Sang et al. 2015; Luo et al. 2016; Ndhlela et al. 2015), quantitative trait locus (QTL) mapping (Marcotuli et al. 2017; Cai et al. 2017) and genome-wide association studies (GWAS) (Huang et al. 2012; Weng et al. 2011) may be powerful tools in predicting the correlations between parental GD and progeny performance. Although no significant and strong correlation were detected between parental GD and progeny performance (H, DBH and V), the distributions of parental GDs have influence on the stability of progeny performance among years. The parents with narrow distributions of parental GDs have more stable seed genetic quality among years than those parents with dispersed distributions of parental GDs in seed orchard. Moreover, more variation in progeny performance was also found when parental GD was greater than 0.6. And the parental GDs of the offspring with the maximum values of V and DBH were more than 0.6. Hence, we can assume that the individuals with excellent performance exhibited parental GDs greater than 0.6, which is similar to the result obtained in a study of Liriodendron (Yao et al. 2015).

The main purpose of seed orchards is to continuously and efficiently produce seeds with excellent genetic qualities. However, can the seeds maintain high genetic quality? How can fluctuations in the genetic quality of seeds across years be avoided? Answering these questions will require further studies. In this study, we investigated the relationships between parental GD and the stability of seed genetic quality in P. massoniana across years using 27 SSRs. The distributions of parental GDs of the USGs were more discrete than those of the SGs, which indicated that more discrete distributions of parental GDs may result in fluctuations in seed genetic quality. Relationships between parental GD and the stability of seed genetic quality were not found in previous studies that focused on the correlation between parental GD and combining ability (specific combination ability (SCA) and general combining ability (GCA)) or heterosis (Tian et al. 2016; Luo et al. 2016; Yao et al. 2015).

Conclusions

In this study, significant and weak correlations were detected between parental GD and the growth performance of offspring (V and DBH), which are not sufficient to predict the performance of progeny (Ndhlela et al. 2015). However, we also found that a narrow distribution of parental GDs contributed to the sustainable production of seeds with excellent genetic quality. Additionally, we found some progeny with excellent performance when parental GDs were greater than 0.6. Hence, we should focus on the selection of parent pairs with a narrow range of GDs to ensure the sustainable and stable production of P. massoniana seeds with excellent genetic quality. However, in the breeding process, parental combinations with a high GD should be selected as hybrid parents to obtain offspring with excellent genetic quality.