The genetic architecture of growth traits in Salix matsudana under salt stress

Willow (Salix) is one of the most important ornamental tree species in landscape plants. One species, Salix matsudana, is widely used as a shade tree and border tree because of its soft branches and plump crown. Some varieties of S. matsudana were salt tolerant and could grow normally in coastal regions. However, the molecular mechanisms of salt tolerance for S. matsudana have been less clear. Here, we addressed this issue by performing a mapping experiment containing 195 intraspecific F1 progeny of S. matsudana, derived from salt-sensitive ‘yanjiang’ and salt-tolerant ‘9901’, grown by cuttings in a 100 mM NaCl solution. Growth performance of these progeny under salt stress was investigated, displaying marked genotypic variability with the coefficients of variance of 28.64–86.11% in shoot and root growth traits. We further mapped specific QTLs contributing to these differences to the Salix genome. Of the 204 QTLs identified, a few were detected to explain a remarkably larger portion of the phenotypic variation than many others. Many detected QTLs were found to reside in the region of candidate genes of known biological function. The discovery of growth QTLs expressed under salt stress provides important information for marker-assisted breeding of salt tolerant Salix varieties and founds the basis for the application of S. matsudana in coastal afforestation.


INTRODUCTION
Willows (Salix), originated from China, have a wide distribution and strong adaptability. 1 Willows are one of the major landscaping greening tree species and have played important roles in the ornamental of roads, courtyards, gardens and lakes. To date, some varieties of willows are widely planted in ecological environment construction of coastal areas and have showed good growth traits.
Because of their greening roles, willows have received considerable attention of genetic research. QTL mapping has been used to map specific loci that control biomass growth, developmental, physiological and biochemical traits. For example, Berlin et al. detected 60 QTLs associated with biomass and nitrogen economy traits in the hybrid population of S. viminalis and S. schwerinii using SNP markers, and each of these QTLs could explain 7.7-41.9% of the phenotypic variation. 2 Ghelardini et al. further identified 80 QTLs responsible for growth arrest, leaf senescence and bud burst in the same population, and the most important QTL could explain 34% of the phenotypic variation for leaf senescence. 3 Brereton et al. found QTLs related with enzymolysis saccharify on the 5th, 10th, and 11th linkage groups of 'Björn' (S. viminalis × S. schwerinii) and '78183' (S. viminalis) hybrid population, and the highest contribution of QTL was 56.2%. 4 In a mapping study of S. purpurea L. variety using RAPD markers, major QTLs for salicin were characterized, with the total contribution of 88.8%. 5 As an important trait, pest resistance has been studied extensively in willow. Hoglund et al. have detected major QTLs for insect resistance in S. viminalis using SNP markers, with the contribution of 18%. 6 One major QTL and 14 small QTLs have been found in the backcross population of S. viminalis × S. schwerinii and S. viminalis using SSR and SNP markers. 7 All of the above studies have helped to reveal the genomic locations of genes controlling important traits of willow, and provide important references for marker-assisted breeding of willow.
S. matsudana is one of the most willow species that belongs mainly to allotetraploid. 8 S. matsudana, originated from Northeast China, is named Chinese willow. Because of its strong adaptability and stress tolerance, S. matsudana had been introduced to Australia, Europe and North America 9 used for greening and afforestation. 10 S. matsudana plays important roles in saline land greening and soil heavy metal repair. 11 Some varieties of S. matsudana, such as S. matsudana 'Hailiu 1' and S. matsudana 'Yanjiang', have become species with great potential in the greening of coastal city development of China. 12,13 Because of the irreplaceable ecological status of willow in coastal areas, we have made tremendous efforts to study its genetic basis of complex traits. A SNP marker-based high density genetic map of tetraploid Salix matsudana has been constructed using its F 1 hybrid population derived from 'Yanjiang' (female parent, a salt-sensitive variety in Jiangsu riverine regions of China) and '9901' (male parent, a salt-tolerant variety selected in Shandong coastal regions of China). 8 In this study, we report on the genetic architecture of growth traits for this mapping population grown under salt stress. A number of QTLs have been identified and mapped to the Salix genome, whose function was analyzed by comparing the corresponding candidate homologous genes in Populus trichocarpa. Results from this study will potentially provide guidance for marker-assisted breeding and cloning of genes controlling salt tolerant growth traits of S. matsudana and also have significance for environmental construction in coastal areas.

Plant material
S. matsudana 'Yanjiang' (female parent, a salt-sensitive variety of S. matsudana in Jiangsu riverine regions of China) and S. matsudana '9901' (a salt-tolerant variety of S. matsudana selected in Shandong coastal regions of China) were artificially crossed in the greenhouse in the winter of 2014. Hybrid seeds of F 1 generation were collected and sowed early spring in 2015. A total of 195 hybrid seedlings of F 1 generation were selected randomly and used as a population to construct a high-density genetic map. 8 Cuttings, with diameters 2-3 cm and lengths 8-10 cm, were collected from the seedlings of the mapping population in August 2016 and cultured in the solution containing 100 mM NaCl. Each seedling has 30 cuttings, cultured in 3 transparent cups, with 10 cuttings in each cup. The cuttings were cultured in the illumination incubator with the photoperiod of 12 h day/12 h night, light intensity of 3300 Lux, temperature of 25°C (the above culture method has been granted a patent, with the patent No. of ZL201210160422.6).

Trait measurement
A total of 15 growth traits of the cuttings were measured, which can be divided into aboveground traits and underground traits. The 11 aboveground traits included start sprout days (SSD), branching angle of the first branch (AN), total sprout length (TSL), total leaf number (TLN), expand leaf number (ELN), nonexpand leaf number (NELN), expand leaf length (ELL), expand leaf width (ELW), leaf fresh weight (LFW), leaf dry weight (LDW) and content of chlorophyll (CH); the four underground traits include start root days (SRD), total root number (TRN), longest root length (LRL) and total root length (TRL). Figure 1. Histogram of frequency distribution for 15 growth traits of S. matsudana offsprings under salt stress. SSD and TSL represent start sprout days and total sprout length, respectively; AN represents branching angle of the first branch; TLN, ELN and NELN represent total leaf number; expand leaf number and non-expand leaf number, respectively; ELL, ELW, LFW, LDW and CH represent expand leaf length, expand leaf width, leaf fresh weight, leaf day weight and content of chlorophyll, respectively. SRD, TRN, LRL and TRL represent start root days, total root number, longest root length and total root length, respectively. These comments apply to other figures and tables as well.
SSD, TLN, ELN, NELN, SRD and TRN were observed and recorded directly, among which SSD and SRD were observed initially after 3 days of cultivation and observed once a day in the following days. AN was measured using protractor; TSL, ELL, ELW, LRL and TRL were measured using a tape; LFW was measured using electronic scales; the fresh leaves were dried in the drying oven at 105°for 3 h and then at 80°for 48 h, the dried leaves were collected and measured using electronic scales; CH was measured using SPAD-502 plus chlorophyll meter made in Japan. All the traits were measured within 20 days.
Statistical analysis SPSS 19.0 software was used to analyze the phenotypic variation of 15 traits and their phenotypic correlations. QTL mapping and contribution calculation were conducted using MapQTL6 software. The method of QTL mapping was Interval Mapping.

RESULTS
Variation and correlation in growth traits under salt stress All 15 growth traits expressed under salt stress showed approximately normal distributions ( Figure 1). Skewness of these traits ranged from 0.0738 to 2.383 ( Table 1), meaning that all the traits belong to a positively skewed normal distribution. Results from variation analysis showed that 15 growth traits had varying degrees of variation, with the coefficients of variation of 28.64-86.11%. Among the 15 growth traits measured, SSD and SRD had similar coefficients of variation of about 30%, demonstrating that the time differences of rooting and sprouting under salt stress were relatively small. The coefficient of variation for TRL was the highest (86.11%) among the 15 traits, meaning that TRL might be strongly influenced by salt stress.
Results from correlation analysis of the growth traits of S. matsudana offsprings under 100 mM NaCl treatment (Table 2) showed that SSD and SRD were positively correlated; LRL, TRN and TRL were positively correlated with each other; TSL, LFW, LDW and CH were positively correlated with each other; ELL, ELW, LFW, LDW and CH were positively correlated with each other.
QTL mapping for growth traits under salt stress On the basis of the phenotypic data of the 15 growth traits of S. matsudana offsprings, genes controlling these traits were mapped to the previously constructed intraspecific genetic map of tetraploid S. matsudana. A total of 204 QTLs were detected, with the LOD thresholds of 3.00-5.98, with the contributions ranging from 9.90 to 55.70% ( Table 3). The QTLs were named under the following roles: start with letter 'q' which represents QTL; followed by the abbreviation of the corresponding trait; end with the number of the linkage group(s) on which the trait was mapped and the marker No. on the linkage group (s). QTLs of the 15 growth traits were summarized as follows: • SSD: A total of four QTLs were detected, among which one was mapped on the 17th linkage group and the other three mapped on the 33rd linkage group. The maximum contribution was 11.00% for qSSD31. SD, SK and CV represent standard deviation, skewedness and coefficient of variation, respectively. Explanations of the trait abbreviations are shown in the comments of Figure 1.
The genetic architecture in Salix matsudana J Zhang et al. Detailed QTL mapping results of the 15 growth traits are listed in Table 3.
S. matsudana belongs to a willow species of allotetraploid (2n = 4x = 76), the genetic map of which consisted of 38 linkage groups. 8 Locations of QTLs for 15 growth traits of S. matsudana offsprings under salt stress are shown in Figure 2. As can be seen, the 204 QTLs detected were located on 25  In addition, the genetic map was constructed using SLAF-seq technique, whose length of sequence tags was about 200 bp. 14 As most genes had the length of about or 41 kb, adjacent QTLs detected in this study might belong to one or more genes. Consequently, we tested segregation profiles of QTLs controlling the same trait on the same linkage group in the F 1 offspring of S. matsudana. Results showed that three QTLs on linkage group 6 (qCH6.1, qCH6.2 and qCH6.3) are segregating among the 195 offspring in the same fashion, suggesting that these three QTLs may represent a single locus. A similar result was obtained for two QTLs on linkage group 15 (qELW15.5 and qELW15.6), two QTLs on linkage group 28 (qLRL28.1 and qLRL 28.2), and two QTLs on linkage group 31 (qSSD31.2 and qSSD31.3).
Functional analysis of candidate homologous genes of S. matsudana QTLs in Populus trichocarpa Willow and poplar belong to Salicaceae. They had evolved from the same ancestor and had many homologous fragments in their genomes. As the genomes of willow have not been released to public, while the genome of P. trichocarpa has been well studied, 15 we aligned the high density genetic map of S. matsudana with the genome of P. trichocarpa. Results showed that the 38 linkage groups of S. matsudana had different correspondence with the chromosomes of P. trichocarpa. The homologous fragments for all 38 linkage groups of S. matsudana could be found in the genome of P. trichocarpa, demonstrating high homology between the genomes of S. matsudana and P. trichocarpa (Figure 3).
To further predict the function of genes associated with QTLs of the 15 growth traits of S. matsudana offsprings under salt stress, we searched for all genes in the genome regions of P. trichocarpa that are homologous with 204 QTLs in S. matsudana, and analyzed possible functions of these genes through functional annotation based on NR database. Results showed that genes with multiple functions have been found in the genome regions of P. trichocarpa that are homologous with QTLs related with seven out of the 15 growth traits, including AN, ELN, ELL, ELW, LFW, LRL Table 2.    Explanation of the trait abbreviations are shown in the comments of Figure 1.
The genetic architecture in Salix matsudana J Zhang et al.
and TRL, in S. matsudana ( Figure 4). These genes might have important roles in the processes of ion transport, energy conversion, signal transduction, protein metabolism, water transportation, stress resistance and biomembrane repair.

DISCUSSION
Growth performance under salt stress As one of the most important willow species, S. matsudana is widely distributed and grown in China. Divergent habitat differences in the distribution of S. matsudana have contributed to extensive adaptabilities in the long term evolution process of this species. To explore the possibility of S. matsudana to be grown in coastal areas, some researchers had investigated its salt tolerance and selected some salt-tolerant varieties of this species. 12,13 However, the molecular mechanisms underlying the salt tolerance of S. matsudana have been less studied, which has restricted efficient selection of more salt tolerant S. matsudana varieties. QTL mapping of important economic traits for S. matsudana is a powerful tool to reveal the molecular mechanisms of salt tolerance for this species, which could provide a basis for map-based cloning and marker-assisted breeding.
QTL mapping is a technique that can find the genome locations of genes controlling quantitative traits by combining phenotypic variation of traits with marker-based genetic map. 16,17 Variation in traits is the basis for successful QTL mapping of the target traits. To efficiently conduct QTL mapping, we measured 15 growth traits of cuttings from the 195 F 1 offspring of salt-sensitive S. matsudana 'Yanjiang' × salt-tolerant S. matsudana '9901' under 100 mM-salt stress, and variation in these traits was analyzed. Results showed that these traits displayed approximately normal distributions (Figure 1), and coefficients of variance for the 15 growth traits ranged from 28.64 to 86.11% (Table 1). According to the previous definition, the variation of a trait is significant if coefficient of variance for the trait is 410%. 18 In this study, all the 15 growth traits had the coefficients of variance of significantly higher than 10%, demonstrating that there were significant variations among different genotypes in these traits. Significant variation in the growth traits of S. matsudana offspring under salt stress has provided the basis for QTL mapping of these traits.
Correlations between the 15 growth traits of S. matsudana offsprings under salt stress showed that SSD and SRD were positively correlated ( Table 2), demonstrating that rooting and sprouting were two mutually promoted growth processes. LRL,  (Table 2), implying that the increase of root number contribute to the rise in root length. TSL, LFW, LDW and CH were significantly related with each other ( Table 2), implying that the growth of sprout and leaves is complementary to each other. There were significant correlations between ELL, ELW, LFW, LDW and CH ( Table 2), showing that the growth of leaf traits promoted each other. These results were in accordance with the previous results about correlations between growth traits in Prunus mume and Portulaca oleracea. 19,20 QTL mapping of growth traits under salt stress Results from QTL mapping for the F 1 hybrid population of S. matsudana under salt stress showed that a total of 204 QTLs were detected for 15 growth traits with the LODs of 43.00 and the contributions of 49.90% (Table 3). These QTLs were located on 25 linkage groups of the genetic map, each of which could explain 9.90-55.70% of the corresponding trait variation (Table 3). Among the 15 growth traits, 3 QTLs on linkage group 18 were related with ELN, with the contributions of 51.20%, 50.40% and 48.70%, respectively (Figure2 and Table 3); 3 QTLs on linkage group 13 were related with NELN, with the contributions of 55.40%, 55.20% and 54.90%, respectively ( Figure 2 and Table 3); 1 QTL related with TRN were found on linkage groups 3 and 16, with the contributions of 55.10% and 55.70%, respectively ( Figure 2 and Table 3). These results showed that there were genes on these linkage groups that control some growth traits of S. matsudana under salt stress.
It has been found that QTLs associated with different traits were located within the same region of the same linkage group. For example, QTLs related with LRL and ELW were located in the same regions between 161.109 and 189.675 on linkage group 28 (Figure 2), and QTLs related with SRD and AN were located in the same regions between 133.302 and 154.899 on linkage group 33 ( Figure 2). It is common that QTLs controlling different traits were located in the same regions. For instance, QTLs controlling the growth and wood quality traits of Eucalyptus 21 were located in the same regions. Such so-called pleiotropic QTLs were also observed for different growth traits in willow, 22,23 different leaf morphology traits in oak 24 and different chemical component related traits in peach and oil palm. 25,26 Pleiotropic effects were believed to be an important component for the genetic architecture of quantitative traits. [27][28][29] Functional analysis of growth QTLs under salt stress Willow and poplar are all species in Salicaceae. They had evolved from the same ancestor and had many homologous fragments in their genomes. 30,31 As the genomes of willow have not been released to public, functions of candidate genes associated with QTLs controlling the growth traits of S. matsudana under salt stress were predicted according to the genome of P. trichocarpa. Results from colinear analysis between the linkage groups of S. matsudana and the chromosomes of P. trichocarpa showed that the 38 linkage groups of S. matsudana had different correspondence with the chromosomes of P. trichocarpa. The homologous fragments for all 38 linkage groups of S. matsudana could be found in the genome of P. trichocarpa (Figure 3), demonstrating high homology between the genomes of S. matsudana and P. trichocarpa and functions of candidate genes associated with QTLs in S. matsudana might be predicted based on the genome information of P. trichocarpa.
In this study, genes in the homologous regions of QTLs controlling the growth traits of S. matsudana were searched in P. trichocarpa. Finally, genes in the homologous regions of QTLs controlling seven growth traits of S. matsudana were found in P. trichocarpa, obtaining functional annotation successfully. For example, QTLs controlling LFW may reside in the homologous regions of genes encoding kinesis-like protein A (Figure 4e).
Similarly, we found that QTLs controlling ELW may be related to genes encoding amino acid transporter family protein and DNA helicase family protein (Figure 4d), QTLs controlling ELN related to genes encoding ATPase family protein and aux/IAA family protein (Figure 4b), and QTLs controlling TRL related to genes encoding callus protein P23 and cell cycle checkpoint control family protein (Figure 4g). These genes might have important roles in the processes of ion transport, energy conversion, signal transduction, protein metabolism, water transportation, stress resistance and biomembrane repair. We predicted therefore that the metabolism was vigorous during the growth of S. matsudana after salt treatment, which might be controlled by multiple genes with varying functions. However, since willow and poplar have undergone divergent evolution processes, there must be great differences between their genomes. The exact function of genes associated with QTLs controlling the growth traits of S. matsudana under salt stress could only be verified based on the genome information of S. matsudana.

CONCLUSIONS
Using cuttings collected from 195 F 1 hybrid offsprings of saltsensitive S. matsudana 'Yanjiang' × salt-tolerant S. matsudana '9901' as a mapping population, variation and correlations among 15 growth traits of the cuttings cultured in 100 mM NaCl solution were analyzed. QTLs controlling these growth traits were mapped on the linkage groups and gene functions located in the QTLs were predicted. We found significant variation among all the traits measured. Some growth traits were positively correlated with each other. A total of 204 QTLs were detected for the 15 growth traits. Genes located in these QTLs might have different functions and participate in different processes of growth and development.