Background

Because of the ease of determining its optimum time for flowering and maturation, bread wheat (Triticum aestivum L., AABBDD, 2n = 42) is cultivated worldwide. Day length and low temperature act as environmental cues affecting the time to heading and flowering. The ability to perceive and respond to these signals is controlled by molecular pathways that regulate early growth habits in response to abiotic stress (Vrn alleles) and photoperiod (Ppd alleles) [1, 2].

In the past, the characterization of the growth habits of winter wheat in China, especially in the Yellow and Huai wheat production region, which is the largest winter wheat production region, relied on field evaluations or in-house artificial identification [3, 4]. However, these identification procedures are tedious and costly, and thus, the practicality of field evaluation and in-house artificial identification methods is limited. Furthermore, inconsistencies between breeders’ descriptions and government registrations sometimes occur due to inexact phenotypic identification methods. In contrast, molecular identification methods are relatively credible, but more novel types of vernalization alleles must be surveyed, and their interactive effects among each other remain unclear.

Indeed, the molecular basis for flowering time regulation has been extensively studied in wheat and other crops [5]. In hexaploid wheat, vernalization requirements are controlled by three major orthologous Vrnalleles—Vrn-A1, Vrn-B1, and Vrn-D1—which have been mapped onto the long arm of chromosomes 5A, 5B, and 5D, respectively [6,7,8]. Each of these loci encodes a MADS-box transcription factor orthologous to AP1 in Arabidopsis, which is reported to be involved in floral meristem development during the transition from the vegetative phase to the reproductive phase [8]. The VRN-1 gene is dominant for the spring growth habit, and it is upregulated by vernalization in winter lines [9, 10]. A homologue of the Arabidopsis FT gene, the Vrn-3 gene, has been mapped to the short arm of chromosome 7 in wheat; this gene upregulated the Vrn-1 genes and thus accelerated heading and flowering indirectly [11].

The emergence of dominant alleles at the Vrn-A1 locus is a result of insertions and deletions within the promoter or a deletion within intron 1, which have been designated Vrn-A1a, Vrn-A1b, and Vrn-A1c, respectively [9, 10, 12, 13]. Spring growth habits can also be attributed to deletions at the Vrn-B1 and Vrn-D1 loci, which have been classified as insensitive vernalization types, and have been designated Vrn-B1a [10, 14, 15], Vrn-D1a, and Vrn-D1b [16]. The Vrn-B1c (novel) allele, which is due to the deletion of 0.8 kb and the duplication of 0.4 kb within intron 1, differs from Vrn-B1a [17, 18]. Another spring allele, Vrn-B1b, has also been described; this allele contains two deletions in the promoter region and is present in spring variety ‘Alpowa’ [14]. The various vernalization requirements of the Vrn-1 alleles or combinations can result in variations in flowering time and spring growth habit [19]. In wheat and other temperate grasses, VRN1 is also expressed in the leaves, where it acts as a repressor of VRN2 [20, 21]. The detailed pathway of the vernalization genes involved in controlling wheat flowering was reviewed by Chen and Dubcovsky (2012) [20].

Photoperiod response is another vital factor affecting flowering time under long-day conditions. For wheat, photoperiod insensitivity (Ppd-1a) is widespread and especially prevalent in regions where crops grow during short days or when the crops mature before the onset of high summer temperatures [22]. Three semi-dominant orthologous Ppd-1 loci—Ppd-A1, Ppd-B1, and Ppd-D1—have been mapped onto the short arm of chromosomes 2A, 2B, and 2D, respectively [23, 24]; these loci are all members of the Pseudo-Response Regulator (PRR) gene family, which is orthologous to the Ppd-H1 gene family in barley [25]. A series of diagnostic markers have been used to efficiently screen for several variants [10, 13, 16, 24, 26].

The Yellow and Huai valley wheat production region (YHW) covers 45% of China’s total cultivation area but contributes 60–70% of the country’s wheat production. Varieties that flower and mature early are helpful in sustaining China’s double-harvest cropping system. In this study, we collected and identified a total of 198 popular varieties, elite lines, and landraces from China to (i) accurately identify the growth habits of the varieties via the field spring sowing method and evaluate their association with heading date (HD; growth period) and flowering date (FD; growth period) at three locations within the YHW during a two-year period (Zhengzhou, Zhumadian and Shangqiu in 2014 and 2015); (ii) use diagnostic molecular markers to determine the main allelic frequencies of Vrn-1 and Ppd-1; and (iii) specifically determine the interactive effects between Vrn-1 and Ppd-1 allelic combinations on heading and flowering times. This study contributes knowledge concerning the effective selection of various types of growth habits of varieties and will be of service to the selection of early-maturation cultivars at the molecular level.

Results

Semi-winter varieties were predominant in the YHW according to the field spring sowing method.

The results of the two-year growth habits were very similar, and the order ranks recorded in 2015 strongly correlated with those recorded in 2016 (Pearson coefficient = 0.96). In general, the ranks of two accessions (Xinong979 and Yumai47) were inconsistent between years, but the discrepancies were only 1–2 grades. Our method separated 10 accessions (Yannong19, Beijing841, etc.) into winterness (grade 0) in 2015 and 2016, which accounted for 5.05%. One hundred and forty-seven accessions (74.24%) were identified as semi-winter types in 2015, and 145 (73.23%) were identified in 2016. In contrast, 41 (20.71%) and 43 (21.72%) grade 5 accessions belonged to the spring type in 2015 and 2016. Overall, the semi-winter varieties (grades1–4) were predominant in the YHW (Table 1; Fig. 1).

Table 1 The phenotypic variants in winter/spring growth habits and allelic variants on Vrn-1 alleles
Full size table
Fig. 1
figure 1

Distribution rate in each grade of winter/spring growth habits among 198 wheat materials in 2015 and 2016

Full size image

In comparison with the data on regional trials (registered, Table 1), our data showed that the consistency was 89.39% when the accessions were divided into winter and spring groups, although 21 varieties need to be re-examined. In detail, of the varieties identified in the winter group, 10 varieties, including Huangming116, Lankaoaizao8, Yumai4, Taikong6, Zhengmai101, Zhongchuang805, Zhoumai23, Zhongyanmai 0708, Huaimai 19 and Xiaoyan 22, qualified as week spring (spring) type according to the registered results. The 11 spring growth habit accessions, which included Jihan2, Hengguan35, Zhengyou6, Luohan6, Luomai23, Pu2056, Shanyou225, Xinong9871, Yunong416, Zhoumai26 and Huarui 00712, were misclassified as accessions having winter (or semi-winter) growth habits. The reason is that these varieties were registered ten years ago, and winter-spring identification was not evaluated during registration tests at that time.

Growth habits were highly associated with growth periods in six environments

The results of the joint ANOVA analysis revealed that significant differences in the mean values of the HD or FD grouped by grades 0–5 in all six environments (Table 2). Briefly, grade 5 exhibited the shortest length of the growth period, while grade 0 exhibited the greatest length. In detail, significant differences in the average values of growth period data were also found between different levels of each trait. The trend was similar to that revealed by the joint ANOVA results. Generally, the smaller the value of the HD or FD is, the greater the value of the grade (Fig. 2).

Table 2 The joint ANOVA analysis in HD and FD grouped by growth habit in six environments
Full size table
Fig. 2
figure 2

Correlations between growth habits and heading and flowering date (HD and FD) in six environments

Full size image

Significant negative correlations were detected between growth habit and growth period in six environments (Table 2, p < 0.01). According to the results of joint variance analysis, the mean values of HD and FD were also correlated with growth habit; the Pearson correlation coefficients were − 0.915 and − 0.886, respectively. Generally, these results also indicated that HD were more tightly, though negatively, related with growth habits. Furthermore, the range of correlation coefficients in six environments were from −0.813 (FD_15_ZMD) to −0.938 (HD_15_ZZ). Thus, we could conclude that the duration of the heading and flowering time of cultivars was tightly associated with growth habits (Additional file 1: Table S2).

Distribution frequency of Vrn-1 alleles in varieties

Because no polymorphisms were found in the Vrn-A1 and Vrn-B3 alleles, we focused on Vrn-B1 and Vrn-D1. The distribution frequency order of the dominant alleles was Vrn-D1a (23.70%) > Vrn-D1b (8.10%) > Vrn-B1a (2.50%) > Vrn-B1b (2.00%) (Table 3). Only one accession was found to carry Vrn-B1b + Vrn-D1a, and 125 accessions presented no dominant alleles. We also used the “consistency index” to evaluate the reliability between the allelic detection and speculated results as described by Stelmakh [19]. According to Stelmakh’s report, the accessions that contained at least one dominant allele were classified as spring types, whereas they were classified as winter types if they had three recessive alleles. Then, we found that nine accessions harbouring dominant Vrn-B1a or Vrn-B1b alleles as well as one accession harbouring Vrn-B1b + Vrn-D1a exhibited the highest rate of consistency (100%). Therefore, these accessions were classified as spring types, which were identical to the results of identification of growth habit in this research. The genotype rate of vrn-B1 + vrn-D1 (63.1%) dominated in all tested panels, and its consistency (96.8% or 95.2%) was also higher than that of Vrn-D1a (48.9%) and Vrn-D1b (25.0%). Therefore, Vrn-D1, especially Vrn-D1b, could not accurately estimate the growth habit.

Table 3 Distribution rates of Vrn-1 and their consistency with results of identification of W/S growth habit
Full size table

Effects of Vrn-1 combinations on HD and FD

The effects of Vrn-B1 + Vrn-D1 combinations concerning HD and FD were examined. In total, there were 6 different types of genotypes grouped by combinations. Among them, 125 accessions had double-recessive vrn-B1 + vrn-D1 alleles. However, Vrn-B1a + vrn-D1, Vrn-B1b + vrn-D1, vrn-B1 + Vrn-D1a, vrn-B1 + Vrn-D1b alleles were carried by 5, 4, 47 and 16 varieties, respectively. Only 1 accession harboured double-dominant Vrn-B1b + Vrn-D1a alleles. Least significant range (LSR, a method of multiple comparison) tests revealed significant differences among the six groups with respect to the mean values of HD and FD in almost all environments (P < 0.05), with the exception that LSR tests for FD_15_ZMD were not significant (highlightedwith yellow). However, no significant differences in the mean values of each group across environments were revealed by the joint ANOVA results (highlighted with green) (Table 4).

Table 4 LSR method of multiple comparison for the effects of Vrn-B1 combined with Vrn-D1
Full size table

Because the low frequency of the Vrn-B1b + Vrn-D1a type (0.5%) made it difficult to exactly compare this allelic combination with other genotypes, we focused on the other five combinations. With respect to their effects, we found that accessions with the vrn-B1 + vrn-D1 genotype presented the latest HD and FD (178.9 d), while varieties that harboured the Vrn-B1b + vrn-D1 allelic combination presented the earliest HD and FD (174.3 d) as well as the shortest growth habit. This method could be useful for precisely identifying the differences in growth habits in each group individually. Then, analyses of Vrn-1 combinations revealed that the effects of the dominant Vrn-B1 genotype on HD and FD were stronger than those of the dominant Vrn-D1 genotype. Finally, we concluded that the rank order of the effects on the growth period was as follows: Vrn-B1b > Vrn-B1a > Vrn-D1b > Vrn-D1a > vrn-D1 = vrn-B1 (Table 4).

Allelic variations of Ppd-1 alleles

No polymorphisms were found in the promoter of Ppd-A1 or Ppd-B1; thus, we focused on their internal variants. Here, variations in the junction sequences of Ppd-B1 were investigated to analyse their allelic variations and effects, which were considered copy number variations (CNVs) [28]. For Ppd-B1, the Ppd-B1a gene has three types in terms of CNV, accounting for 33.8% (Truncated CS type), 8.6% (Intact CS type) and 35.3% (Sonora 64 type) (Table 5). According to a previous report, the first two genotypes were named Ppd-B1c and the Sonora 64 type was named Ppd-B1a [30]. For variations within Ppd-B1, 6 types of genotypic combinations were all detected because of their different types of combination. As to percentage, the “S: N: N” type constituted the largest proportion (34.34%, “Sonora 64 type” only), while the percentages of “N: I: N” (1.01%, “Intact CS type” only) and “S: N: T” (1.01%, “Sonora 64 type + Truncated CS type” for short) were the lowest (Additional file 1: Table S2; Table 5).

Table 5 Allelic variants of Ppd-B1
Full size table

For Ppd-D1, the haplotypes identified among the materials were divided into three types in accordance with the reports of Guo et al. (2010) [31] and Chen et al. (2013) [13]: Hapl I (34.85%, sensitive), Hapl II (0.5%, Chinese Spring, insensitive), and Hapl VII (64.6%, sensitive). Only one variety (Chinese Spring) had a 2.0-kb deletion in the promoter region and thus should be designated Ppd-D1a (insensitive, theoretical relatively short HD and FD). Thus, Hapl I and Hapl VII could be considered recessive alleles in this study (Table 6).

Table 6 Allelic variants of Ppd-D1
Full size table

Effects of single Ppd-1 alleles on HD and FD

One hundred and twenty-five materials that had double-recessive Vrn alleles (vrn-B1 + vrn-D1) were selected to evaluate the influence of Ppd-B1 or Ppd-D1 on plant traits. There were four genotypes, but no significant differences were found among groups according to ANOVA results. Because inconsistencies between the ANOVA and LSR test results were sometimes detected, multiple comparisons were subsequently performed. The average values for phenotypes among the four groups differed significantly for only five traits (three traits in Zhumadian). Furthermore, varieties that harboured “Sonora 64 type” (“N: N: S”) showed the shortest HD and FD. This result was consistent with those reported by Díaz et al. (2012) [28] (Table 7).

Table 7 The effects of single Ppd-B1 and Ppd-D1alleles on HD and FD
Full size table

Regarding Ppd-D1, two genotypes were found, and no significant differences were detected via the LSR method between groups. We predicted that two recessive types of alleles did not contribute to the advancement of heading and flowering time. With respect to the comparisons of the mean values of Ppd-B1 and Ppd-D1 on HD and FD, the effects of Ppd-B1 were somewhat stronger than those of Ppd-D1 (Table 7).

Interactive effects of Ppd-1 combinations on HD and FD

We also examined and assessed the interactive effects of Ppd-B1 and Ppd-D1 combinations. Similarly, no significant differences were found according to the ANOVA results. A total of eight genotypes that contained two Ppd alleles were surveyed, in which the type “N: N: S + Hapl VII” constituted the largest proportion (29, 23.6%), while the percentages of “T: I: N + Hapl I or Hapl VII” were the lowest (Table 8). With respect to their effects, the LSR method revealed significant differences in the mean values among groups for four traits (three in Zhumadian and one in Zhengzhou). We suspected that Ppd-1, especially Ppd-B1, functioned only in specific environments. Generally, the mean values of HD and FD were advanced by the Ppd-1 combinations by only 0.2–0.5 d. Therefore, the effects of Ppd-1 combinations were stronger than those of single Ppd-B1 or Ppd-D1 alleles but were far weaker than those of Vrn alleles (5–7 d earlier heading or flowering in Zhengzhou and Zhumadian). With respect to the individual genotypes, the rank order of their effects on growth period was as follows: Ppd-1 > Ppd-B1 > Ppd-D1.

Table 8 Interactive effects of Ppd-1 combinations on growth period and growth habit
Full size table

Discussion

Consistency of marker analysis and growth habit identification

Wheat is the major crop in the YHW in terms of yield and area in China. This region is located in the transition zone of winter and spring wheat cultivation, where semi-winter varieties and weak spring cultivars are also planted. However, inconsistencies sometimes occur between registered and empirical results. Hence, the precise identification of winter/spring growth habits for newly registered varieties is necessary and helpful not only for the rational use of varieties but also for the provision of vital information for breeders in the YHW. Here, growth habits were examined during a two-year period via a novel field spring sowing identification method and materials along with marker-assisted selection (MAS).

We believe that our identification method in the field is more practical than that conducted in greenhouses, where the materials are grown under conditions closely related to those in the field. Furthermore, correlation analysis revealed that the phenotypic data during two years were also consistent between years. In comparison with registered information, 10 of 155 (6.45%, 2016) winter wheat varieties were inconsistent (Zhongchuang805, Taikong6, Xiaoyan22, etc.), and similar situations were observed in other groups—in particular, 43 spring wheat varieties containing 13 (25.6%, 2016) inconsistent samples (Bainong3217, Luohan6, Xinong979, etc.). We doubted that the reason for this situation was due to their early registration before the materials were rigorously identified. In total, the consistency was approximately 90%, although some varieties presented discrepancies, and our method was more convenient than the report of Gardener and Barnett [32].

More vital clues that we wanted to examine included the consistency between vernalization alleles and growth habit. The results indicated that all ten Vrn-B1 genotypes of spring wheat varieties presented a value of 100%, whereas Vrn-D1 exhibited lower results. Among the 43 cultivars ranked as grade 5 (data from2016, Xinzheng), 37 (86.04%) carried at least one of the tested dominant vernalization alleles and were classified as spring varieties; the other 6 varieties carried the recessive alleles at the three vernalization loci. For winter types, 131 of 155 (84.51%) accessions presented similar consistency. We predicted that there are two main factors that could be responsible for this phenomenon. First, a single individual is genotyped, whereas the phenotype is assessed on a plot scale of multiple individuals, and there may be some variation among individual seeds. Second, two other major pathways also control heading and flowering dates in plants, i.e., the phytohormone gibberellic acid (GA) and the autonomous pathways, in addition to the vernalization and photoperiod pathways [33, 34].

Allelic distributions of Vrn-1 revealed trends and orientations in the YHW.

As no dominant allele of Vrn-A1 or Vrn-B3 was detected, which are probably the two genes that have the strongest effects of those examined, the allelic distributions of the dominant Vrn-B1 and Vrn-D1 alleles are likely responsible for the spring genotypes of wheat varieties. Indeed, the scarcity and decreasing frequency of the Vrn-A1 and Vrn-B3 loci in the YHW have been previously discussed by Zhang et al. (2008) and Chen et al. (2013) [4, 13]. We suspected that the frequencies of recessive vrn-B1 and vrn-D1 likely increased via direct selection due to their contributions to yield traits because of the long maturation period. This inference was in agreement with that reported in the literature [14, 35]. Furthermore, the frequency of dominant Vrn-D1 was higher than that of Vrn-B1 in our tested materials. These results were also consistent with the previous report of Zhang et al. (2015) [36].

Although only 10 dominant Vrn-B1 alleles (5.0%) were discovered in the spring genotypes, the consistency of the marker-growth habit (100%) was better than that of Vrn-D1. Additionally, we found that accessions with single Vrn-B1b alleles exhibited the earliest HD and FD in the six environments; these effects were stronger than those forVrn-B1a, Vrn-D1b and Vrn-D1a. These results were consistent with those of previous studies in which the rank order was Vrn-A1 > Vrn-B1 > Vrn-D1 [37]. Santra et al. (2009) reported that a novel Vrn-B1b allele that resulted from a 36-bp deletion within intron 1, which is referred to as ‘Alpowa’ and carries the winter growth habit alleles vrn-A1 and vrn-B1, was likely to cause a spring growth habit; however, the authors did not provide sufficient evidence [14].

The dominant Vrn-D1 locus occurs in the most popular types and is distributed throughout nearly the entire wheat production region [4, 13]. The previous data also established that carriers of Vrn-A1 or Vrn-D1 tend to produce longer spikes than do carriers ofVrn-B1. As a result, the Vrn-D1 genotypes were prevalent in China [38, 39]. In the present study, in terms of Vrn-D1, the allelic frequency reached 31.8%, which represented the most dominant allele distribution in the population. Thus, the proportion was similar to previous reports; however, poor consistency in the growth habit of Zhengzhou was observed. Thus, it would be interesting to test whether Vrn-D1 and other alleles interact to influence growth habits and period (HD and FD).

The results of the Vrn-1 combination analysis revealed that only one accession carried Vrn-B1b + Vrn-D1a alleles; thus, the samples were so limited that phenotypic data were not statistically representative. For genotypic analysis, in combination with growth habit identification, 63.1% of the materials that had three recessive alleles all belonged to winter or semi-winter wheat (grade 0–4). This tendency was in accordance with that reported by Sun et al. (2009, 61.1%). Moreover, the ANOVA and LSR tests revealed that the mean values of the HD, FD and growth habit among six genotypes differed significantly, which also indicated that the Vrn-1 combinations were tightly associated with phenotypes. Although the effects of these combinations on HD and FD were not definitively stronger than those of single Vrn-B1 alleles, the growth habit level was divided in greater detail. This division will enable a more precise identification of vernalization requirements for accessions at molecular levels.

Interactive effects detected between the Ppd-B1 and Ppd-D1 alleles

Photoperiod responses are controlled by members of the pseudo-response regulator (PRR) gene family in plants. In general, the potential of Ppd-1 alleles to affect insensitivity has been ranked as Ppd-D1 > Ppd-B1 > Ppd-A1 [17, 38]. However, in the present study, only one accession (Chinese spring) was found to carry a 415-bp band that indicated a genotype of Ppd-D1a. Thus, it was difficult to precisely evaluate its effect on phenotype statistically. According to the criterion of previous research, two sensitive haplotypes (Hapl I and Hapl II) of Ppd-D1 were used for the evaluation for their distribution and effects [13, 31].

For Ppd-B1, we only examined the polymorphisms of CNVs of the Ppd-B1 locus because of its tight association with heading and flowering time. Indeed, Zhang et al. (2015) designated eight haplotypes according to the combinations of CNVs of Ppd-B1 and found that the cultivar with Ppd-B1_hapl-VI demonstrated the earliest heading and flowering times [36]. However, the results were not consistent with those of Díaz et al. (2012) [28]. In the present study, 125 accessions carrying recessive vrn-B1 and vrn-D1 alleles were selected. With respect to Ppd-B1, we found that wheat cultivars with “Sonora 64” Ppd-B1a alleles flower earlier than those with “Chinese Spring” alleles, which was in accordance with Díaz et al. (2010). Furthermore, we found that six types of combinations emerged, and the “Truncated CS type” and “Intact CS type” did not simultaneously emerge for Ppd-B1. These results were also not the same as those reported by Chen et al. (2013). We suspected that the complex genetic background (genotypes mixed with Vrn-1 genes) would hinder us from providing definitive results. Thus, we believed that our method was possibly more reliable than previous methods because of the uniform background [30]. With respect to Ppd-D1a, the rare diversity of the Ppd-D1 allele could not be used to exactly evaluate the effects of the variants, and no significant differences were observed between the two haplotypes (Hapl I and Hapl II).

Comparison of effects on growth period (HD and FD) between the Vrn-1 and Ppd-1 alleles

Moreover, from a comprehensive perspective, we concluded that, compared with Ppd-1, Vrn-1 played a major role in regulating heading and flowering traits as well as growth habit. At the Vrn-1 locus, cultivars with the Vrn-B1b + vrn-D1 (174.3 d for HD, 183.2 d for FD) allele both headed and flowered earlier by approximately 4 days than did cultivars with the vrn-B1 + vrn-D1 (178.9 d for HD, 187.9 d for FD) allele (Table 4). Whereas at the Ppd-1 locus, cultivars with the “N: N: S” allele combination (178.6 d for HD, 187.7 d for FD) both headed and flowered approximately 1 day earlier than did cultivars with the “T: I: N” allele combination (179.9 d for HD, 188.5 d for FD) (Table 7). Indeed, the interactive effects of Vrn-1 and Ppd-1 gene combinations were also detected in our research. However, the results of ANOVA and LSR tests revealed weak interactions between Vrn-B1 and Ppd-B1, Vrn-D1 and Ppd-D1, Vrn-B1and Ppd-D1, Vrn-D1and Ppd-B1 (data not shown). We suspected that the complex genetic background in natural populations would make it difficult to reveal this interaction. In a previous study, Shcherban et al. (2014) also found that the haplotypes Ppd-D1a/Vrn-B1a or Ppd-D1a/Vrn-B1a did not differ significantly in heading time from the respective Vrn-1haplotypesharbouring the sensitive allele Ppd-D1b [40]. This finding suggests that it is better for us to examine the interaction between Ppd-1 and Vrn-1 in biparental populations.

Conclusion

In the present study, we dissected the Vrn-1 and Ppd-1 gene composition and found that Vrn-1, rather than Ppd-1, played a major role in controlling vernalization and photoperiod responses in this region. The work will be helpful for guiding the breeding of wheat in the Yellow and Huai wheat production region.

Methods

We tested 198 cultivars (lines) including historic varieties, commercial varieties, and newly bred varieties originating from the YHW. Among them, 159 accessions were from Henan, 10 accessions were from Shandong, 10 accessions were from Shaanxi, 8 accessions were from Jiangsu,4 accessions were from Hebei, 4 accessions were from Beijing, 1 accession was from Shanxi, 1 accession was from Anhui, and 1 accession (Chinese Spring) was from Sichuan (Table 1). The entire original source of the plant materials used in our study was kindly provided by other labs. We complied with the Convention on the Trade in Endangered Species of Wild Fauna and Flora: https://www.cites.org.

Characterization of winter/spring growth habits

Although the growth habit for assessing vernalization is already well established, identification of the exact materials involved is necessary because of differences in environmental conditions. The tested materials were planted at the Zhengzhou Scientific Research and Education Center of Henan Agriculture University (113.7°E, 34.7°N) on 12 March 2015 and at another test site [ZhengHan Seed Technology Co. Ltd., XinZheng (113.7°E, 34.4°N)] on 12 March 2016. Seeds were sown in 1.0-m rows, and individual seeds were spaced 6.67 cm apart; 15 seedlings were reserved per row after wheat seedling emergence. Two replications were planted for reliable data collection. The stage of maturity and percentage of headed spikes were recorded on 25 June in the same year; we repeated these measurements one week later. The growth habit of the materials was divided into grades numbered 0 to 5. The criteria were as follows: 0, no jointing and booting; 1, partial main stem headed; 2, main stem and a few tillers headed; 3, normal heading but abnormal grain filling and immature; 4, normal heading and grain filling but premature; 5, normal maturity.

Identification of HD and FD

The varieties used to assess agronomic traits were planted on 9 October 2013 and 2014 in Zhengzhou (113.7°E, 34.7°N), on 15 October 2013 and 17 October 2014 in Shangqiu (115.7°E, 34.5°N), and on 19 October 2013 and 5 November 2014 in Zhumadian (114.0°E, 32.9°N). All of these locations differed significantly in day length and climatic factors in Henan Province. Each material was planted in two 1.5-m rows; there were 110 seeds per row, and the rows were spaced 23 cm apart. Two replications were planted at each location. Field management practices during our experiments were in accordance with agronomic practices commonly used in the area. The HD and FD were assessed on a plot scale of multiple individuals when more than half of the individual seedlings exhibited classic morphological traits for these events.

DNA extraction and diagnostic markers for Vrn-1 and Ppd-1

DNA was extracted from the seedlings in accordance with a modified SDS-phenol-chloroform method [27]. The primers used were based on those described in many previous reports and were synthesized by Sangon Biotech Co., Ltd. (Shanghai) (Additional file 2: Table S1). To recognize the segments amplified from Ppd-B1, accessions harbouring 994 bp, 425 bp, and 223 bp were designated intact Chinese Spring type (I), truncated Chinese Spring type (T), and Sonora 64 type (S), respectively. If no bands were amplified in the materials, the genotypes were referred to as null (N) [28].

PCR amplification and electrophoresis

PCR amplification reactions were conducted in a 12-μL reaction flask containing 40 ng of genomic DNA, each primer at 2.5 μM, each dNTP at 200 μM, 1× buffer containing1.5 μM MgCl2, and 0.5 units of Taq polymerase. We used a Bio-Rad thermocycler with the following PCR conditions: 94 °C for 3 min; 34 cycles of 94 °C for 30 s, 50 °C to 65 °C for 30 s (annealing temperatures for each primer pair are listed in Additional file 2: Table S1), and 72 °C for 1 min; and a final 10-min extension at 72 °C for preservation. The PCR products were separated by electrophoresis either on a 0.8–1.2% agarose gel stained with ethidium bromide (EB) or an 8% nondenaturing polyacrylamide gel and visualized with silver staining [29].

Statistical analysis

A consistency index (%) was used; this index indicated the number of accessions that originated from the consistent results of both genotype and field identifications divided by the total number of materials. The phenotypic data were imported into R software (R 3.4.1) for analysis via ANOVA, Student’s t-tests and correlation; we used the “reshape” and “agricolae” packages to perform these analyses and the “ggplot2” package for graphical construction.