Discovery, evaluation and distribution of haplotypes of the wheat Ppd-D1 gene


  • Zhiai Guo,

    1. Key Laboratory of Crop Germplasm and Biotechnology, Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing 100081, China
    2. College of Agriculture, Sichuan Agricultural University, Ya’an 625014, Sichuan, China
    Search for more papers by this author
  • Yanxia Song,

    1. Key Laboratory of Crop Germplasm and Biotechnology, Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing 100081, China
    Search for more papers by this author
  • Ronghua Zhou,

    1. Key Laboratory of Crop Germplasm and Biotechnology, Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing 100081, China
    Search for more papers by this author
  • Zhenglong Ren,

    1. College of Agriculture, Sichuan Agricultural University, Ya’an 625014, Sichuan, China
    Search for more papers by this author
  • Jizeng Jia

    1. Key Laboratory of Crop Germplasm and Biotechnology, Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing 100081, China
    Search for more papers by this author

Author for correspondence:
Jizeng Jia
Tel: +86 10 8210 5831


  • Ppd-D1 is one of the most potent genes affecting the photoperiod response of wheat (Triticum aestivum). Only two alleles, insensitive Ppd-D1a and sensitive Ppd-D1b, were known previously, and these did not adequately explain the broad adaptation of wheat to photoperiod variation.
  • In this study, five diagnostic molecular markers were employed to identify Ppd-D1 haplotypes in 492 wheat varieties from diverse geographic locations and 55 accessions of Aegilops tauschii, the D genome donor species of wheat.
  • Six Ppd-D1 haplotypes, designated I–VI, were identified. Types II, V and VI were considered to be more ancient and types I, III and IV were considered to be derived from type II. The transcript abundances of the Ppd-D1 haplotypes showed continuous variation, being highest for haplotype I, lowest for haplotype III, and correlating negatively with varietal differences in heading time. These haplotypes also significantly affected other agronomic traits. The distribution frequency of Ppd-D1 haplotypes showed partial correlations with both latitudes and altitudes of wheat cultivation regions.
  • The evolution, expression and distribution of Ppd-D1 haplotypes were consistent evidentially with each other. What was regarded as a pair of alleles in the past can now be considered a series of alleles leading to continuous variation.


Photoperiod has an important effect on plant growth and development. Wheat (Triticum aestivum L.) is a quantitative long-day plant, whose flowering is accelerated by long days. Photoperiod insensitive varieties flower under both short- and long-day conditions but photoperiod sensitive types show delayed heading, or do not head at all, if the day length and number of long days do not reach a threshold for floral initiation. Simultaneously, wheat is one of the most adaptable of crops and is planted in virtually all countries where the photoperiod varies dramatically and continually, suggesting there may be corresponding variations in photoperiod response among wheat varieties adapted to different environments.

The genetic control of photoperiod sensitivity in wheat is primarily determined by the Ppd-D1 and Ppd-B1 genes located on the homoeologous chromosomes 2D and 2B, respectively (Welsh et al., 1973; Law et al., 1978; Scarth & Law, 1983; Wilhelm et al., 2009). However, only two alleles, photoperiod insensitive (Ppd-D1a and Ppd-B1a) and sensitive (Ppd-D1b and Ppd-B1b), respectively, at each locus were known previously and, as such, these do not adequately explain the broad adaptation of wheat to photoperiod variation.

The Ppd-D1 gene is an orthologue of Ppd-H1 on barley chromosome 2H (Laurie, 1997; Börner et al., 1998). Ppd-H1 was cloned using fine mapping and was identified as a member of the pseudo-response regulator (PRR) family (Turner et al., 2005). In Arabidopsis, genes in this family are associated with circadian functions. Pseudo-response regulators (PRRs) proteins are characterized by a pseudo-receiver domain near the N-terminus and a CCT (CO, CO-like, and TOC1) domain near the C-terminus (Griffiths et al., 2003; Mizuno & Nakamichi, 2005). The wheat chromosome 2D PRR gene homolog of Ppd-H1 was isolated from a ‘Chinese Spring’ wheat BAC (bacterial artificial chromosome) clone. Sequencing of 2D PRR gene homologs from a number of varieties revealed five polymorphisms (Beales et al., 2007). The first was a 2089 bp deletion upstream of the coding region; this was a polymorphism associated with photoperiod insensitivity. This deletion mutant, Ppd-D1a, in photoperiod insensitive varieties caused mis-expression of the 2D PRR gene and permitted early flowering in both short- and long-day conditions. The second was a mariner-like transposable element (MLE) present in intron 1 of Ppd-D1 in some varieties, such as ‘Mercia’ and ‘Cappelle-Desprez’. The third polymorphism was a 5 bp deletion in exon 7, which created a frame shift and resulted in a nonfunctional protein. The fourth was a SNP (A/G) in exon 7, predicted to cause an Ala to Thr change upstream of the CCT domain. The fifth polymorphism was a 16 bp deletion including the last two bases of the CCT domain in exon 8, which was predicted to change the last amino acid of the CCT domain (Gln to Leu) and to produce a new COOH terminus in the protein.

In the present paper, we report six Ppd-D1 haplotypes, and the development of a haplotype tree. We examined the mRNA expression patterns, relationships with major agronomic traits and worldwide distributions of each haplotype. The results provide a basis for investigating the evolution of the wheat Ppd-D1 gene and are applicable for understanding and improving the adaptation of wheat to specific environments.

Materials and Methods

Plant materials

A total of 492 common wheat varieties (genomes AABBDD) selected from 41 countries on six continents, 25 synthetic wheats derived from crosses of tetraploid wheat (AABB) and Aegilops tauschii (DD), and 30 A. tauschii accessions were surveyed. Assuming 20 different A. tauschii accessions were used as parents of the synthetics and that they were different from the 30 accessions per se, a total of 55 different A. tauschii accessions were examined. The numbers of wheat accessions from each region reflected the world wheat production areas. The Chinese accessions were sampled from core collections of Chinese common wheat germplasm and comprised about half the total. They included landmark landraces and leading modern varieties derived from the 10 major agroecological wheat regions. The non-Chinese accessions represented a worldwide sampling of germplasm and included varieties widely used in Chinese breeding programs. The synthetics were originally produced by the Chinese Academy of Agricultural Sciences (CAAS) or the International Maize and Wheat Improvement Center (CIMMYT). The separate A. tauschii and synthetic accessions represented the ancestral DD genome of hexaploid wheat. The varieties for studying agronomic traits were planted in October, 2006, in two environments with clear differences in daylength during the growth cycle, namely Beijing (116.2°E, 40.2°N) and Luoyang, Henan (111.6° E, 33.8° N).

DNA extraction and molecular markers

DNA was extracted from seedlings using the phenol–chloroform method (Sharp et al., 1988). The primers (see the Supporting Information, Table S1) were synthesized by Augct Biotechnology Co. Ltd, Beijing; A common forward primer Ppd-D1_F combined with two reverse primers, Ppd-D1_R1 and Ppd-D1_R2, were used to detect the 2 kb deletion in the Ppd-D1a allele. If the amplification using primer Ppd-D1_F and Ppd-D1_R1 produced a 414 or 453 bp band, the variety carried the recessive daylength sensitive allele Ppd-D1b. If a 288 bp fragment was amplified using primers Ppd-D1_F and Ppd-D1_R2, the variety carried the dominant insensitive allele Ppd-D1a. Another marker was used to survey the 16 bp insertion in exon 8. Polymerase chain reaction (PCR) products amplified with primer Ppd-D1exon8_F1 and Ppd-D1exon8_R1 were digested by HpaII, leading to bands of 326 bp and 22 bp, or 257, 69 and 22 bp, from the deleted and intact sequences, respectively. The three markers mentioned were based on Beales et al. (2007).

If there was no transposable element present in intron 1 of the Ppd-D1 gene, amplification with primers D520F and D520R gave a 2.6 kb band; in the presence of the element there was no amplification. If either the upstream 2 kb deletion, or the TE insertion in intron 1, was present, the D78 primers gave no product. If the upstream region was intact and the TE was absent the primers gave a 1 kb band.

Marker D5 was used to detect the 5 bp deletion in exon 7. This assay used nested PCR with two pairs of primers. The first primer pair, D5-1F and D5-1R, was used for PCR. The product was then used as a template for amplification with the second pair, D5-2F and D5-2R. This generated a 184 bp product from varieties without the deletion and a 179 bp product from varieties with the deletion.

Real-time quantitative PCR for expression analysis

RNA samples were extracted using the Trizol reagent from seedlings grown for 7 d at 24°C in a short day (8 h light 16 h dark) regime without vernalization. DNA was removed by digestion with DNAse I (Ferments, Ontario, Canada) before reverse transcription. cDNA first strand was synthesized using M-MLV transcriptase (Invitrogen). The Ppd-D1 amplification primers (HvPRR72_F2, GATGAACATGAAACGGG, and TaPRR72_DgR2, GTCTAAATAGTAGGTACTAGG) were from Beales et al. (2007). Amplification of β-tubulin (5′-TGTGCCCCGTGCTGTTCTTATG-3′ and 5′-CCCTTGGCCCAGTTGTTACCC-3′) was used as an internal control to normalize all data. RNA was quantified using an option 2 real-time PCR instrument with a SYBRGreen mastermix kit (Tiangen, Beijing, China). The relative quantification method (ΔΔCT) was used to evaluate quantitative variation between the three replicates (Livak & Schmittgen, 2001).

Phylogenetic and partial correlation analyses

A phylogenetic haplotype network was constructed based on statistical parsimony using the software network (Fluxus Technology Ltd, Suffolk, England). All indels were coded as substitutions (Caicedo & Schaal, 2004). All the statistical analysis including bivariate and partial correlations was done by using the spss15.0 software (SPSS Inc., Chicago, Illinois, US).


Marker development and Ppd-D1 haplotype detection

A series of molecular markers were developed for detecting sequence variations in Ppd-D1 (Fig. 1a,b). Markers D520 and D78 were designed around the mariner-like transposable element (TE) insertion in intron 1. Marker D520 detected the TE insertion, whereas D78 identified two allelic variations at the same time because its forward primer was located in the 2089 bp deletion region and its reverse primer was located downstream of the TE insertion site. Marker D5 was designed to detect the 5 bp deletion in exon 7.

Figure 1.

 Six haplotypes of the Ppd-D1gene were detected. (a) A series of molecular markers based on five polymorphisms were used to identify the gene haplotypes. Tall rectangles represent coding regions, low rectangles represent the introns, 5′UTR and 3′UTR regions. Insertions and deletions are indicated by + and −, respectively. (b) Amplification products of Ppd-D1 markers. M, DNA markers 100 bp or 200 bp; 1, Yanzhan 1; 2, Akagomughi; 3, Chinese Spring; 4, Ningchun 10; 5, Fr81-12; 6, Hussar; 7, Hezuo 2; 8, Early Premium; 9, S231 (synthetic); 10, S213 (synthetic).

Using the above three markers and those previously reported for identifying the upstream 2089 bp deletion and the 16 bp deletion in exon 8 (Beales et al., 2007), the four polymorphisms of the Ppd-D1 gene were studied in all 492 wheat (Table S2 and S3) and 55 A. tauschii accessions (Table S4). With primers Ppd-D1_F and Ppd-D1_R1, some synthetics and A. tauschii accessions amplified a 453 bp fragment, which differed from those produced by common wheat. This was caused by 24 bp and 15 bp insertions, separated by 105 bp, in the 2089 bp intact region. The 5 bp deletion in exon 7 and the TE insertion in intron 1 were present only in a proportion of photoperiod-sensitive ppd-D1 genotypes with an intact upstream region. No variety had the 5 bp deletion and the TE insertion combined. The 16 bp insertion in exon 8 was present in almost all the synthetic wheats except for two accessions and all the A. tauschii accessions.

Thus a total of six Ppd-D1 haplotypes, I–VI (Fig. 1a), was detected. The 2 kb upstream deletion distinguished haplotype I from haplotype II. Haplotypes III and IV carried the TE insertion in intron 1 and the 5 bp deletion in exon 7, respectively. Haplotypes V and VI were characterized by the intact 16 bp insertion in exon 8, with VI carrying the 24 bp plus 15 bp insertions in the 2 kb upstream region.

Relative expression abundance of Ppd-D1 haplotypes

To estimate the effects of the different Ppd-D1 haplotypes, seedlings of each haplotype, represented by two vernalization-insensitive varieties, were grown in a short-day photoperiod (8 h light and 16 h dark), and their relative expression values were measured at 3-h intervals over a continuous 24 h period using the D genomic specific primers. Previous studies on the Ppd-D1 homologues OsPRR37 in rice and AtPRR7 in Arabidopsis showed that the expression of both follow a circadian rhythm with a peak during the day and a decline to very low levels during the dark period (Matsushika et al., 2000; Murakami et al., 2003, 2005). Our results (Fig. 2) showed that haplotypes I–V had similar expression patterns. All expression peaks were at the same time, but their levels were very different. The order of expression peak values of the five haplotypes from high to low was I, IV, V, II and III (Fig. 3a). Among them, haplotype I was significantly higher (< 0.01) than haplotypes II and III, and haplotype IV was higher (< 0.05) than II and III. No significant differences were detected among the remaining haplotypes despite the larger numerical differences between their mean values. These results suggest that the different haplotypes may control the expression of the Ppd-D1 gene.

Figure 2.

 Relative expression levels of Ppd-D1 in seedlings of five haplotypes grown under short days (8 h light and 16 h dark) by real-time quantitative RT-PCR. Each haplotype was represented by two vernalization-insensitive varieties. RNA samples were taken at 3 h intervals over a 24 h period. Sampling was conducted for 7 d after germination. The means ± SD from three repeated experiments are shown. Open bars, light; closed bars, darkness. Expression levels in (a) haplotype I varieties ‘Yanzhan 1’ and ‘Zaosui 30’; (b) haplotype II varieties ‘Chinese Spring’ and ‘Ningchun 10’; (c) haplotype III varieties ‘Kehan 6’ and ‘Fr81-12’; (d) haplotype IV varieties ‘Hezuo 2’ and ‘Thatcher’; and (e) haplotype V synthetic varieties ‘S213’ and ‘S231’.

Figure 3.

 Correlation of relative expressions of Ppd-D1 haplotypes and days to heading. (a) Relative expression in Ppd-D1 haplotypes in the morning (the peak time) in short-day conditions. Capital letters represent significance at = 0.01, and small letters significance at = 0.05; (b) Days to heading of varieties with different Ppd-D1 haplotypes grown in environments i (116.2°E, 40.2°N) and ii (111.6°E, 33.8°N); (c) Correlation of relative expression abundance in Ppd-D1 haplotypes in the morning and days to heading of varieties grown in the environments i (R2 = 0.4619, = 0.207) and ii (R2 = 0.6673, = 0.091).

Association of Ppd-D1 haplotypes with agronomic traits

One of the factors influencing the development of wheat plants in particular environments is the allelic constitution at the Ppd-D1 locus. To assess the effects of the different haplotypes, we analysed differences in certain agronomic traits associated with yield in a range of varieties planted in three environments in China in 2007, including Beijing (116.2° E, 40.2° N), and in rainfed and irrigated conditions in Luoyang, Henan (111.6° E, 33.8° N). Comparing the effects of each haplotype on agronomic traits (Table 1), there were significant effects on three traits, namely days to heading, plant height and 1000-kernel weight. Heading time was the trait most closely associated with haplotypes, with significant differences (< 0.01) between haplotypes I, II and III at both locations. The number of days to heading was largest for haplotype III, smallest for I and intermediate for II, IV and V. The average plant height of haplotype II varieties was higher than those of the other four haplotypes, and the differences between I and V were significant (< 0.01) in all three environments. The mean plant heights of haplotypes III and I were lower than those of II and IV, respectively; the differences were significant (< 0.05) in two environments. Haplotype V made a positive contribution to 1000-kernel weight and its difference from II was significant (< 0.01) in all three environments. The average 1000-kernel weight of haplotype I was intermediate and differences from V and II were significant (< 0.01) in two of the three environments, and significant at = 0.05 in the third. Some traits also showed significant differences among three haplotypes at = 0.05 in at least two environments. For example, the average peduncle length of haplotype IV varieties was longer than those of I and V, and the spike lengths of group I varieties were shorter than those of III and V.

Table 1.   Comparison of agronomic traits associated with Ppd-D1 in five haplotypes (I–IV) grown in different environments
NMean ± SDNMean ± SDNMean ± SDNMean ± SDNMean ± SD
  1. Different letters in column ‘Mean ± SD’ of each haplotype indicate significant differences between haplotypes, capital and small letters indicate significance at < 0.01 and < 0.05, respectively.

  2. N, number of accession; E, environments.

  3. In column ‘E’, i represents for Beijing (116.2° E, 40.2° N), China; ii and iii for rainfed and irrigated, Luoyang (111.6° E, 33.8° N), Henan, China, respectively.

Days to headingi103207.21 ± 3.73a(A)37210.27 ± 3.83b(B)4217.75 ± 2.63c(C)14211.36 ± 4.13b(B) 6208.50 ± 3.15ab(AB)
ii116185.08 ± 4.10a(A)43190.09 ± 5.36b(B)9195.44 ± 5.81c(C)15190.53 ± 6.06b(BC)10189.10 ± 7.82b(B)
Days to maturingi76245.36 ± 2.70a 29245.55 ± 2.85a 1251.00 8247.88 ± 1.73b5246.40 ± 1.67ab
ii117222.89 ± 4.43a(A)42224.07 ± 5.65ac(AC)8230.00 ± 5.29b(B)15226.47 ± 4.75bc(BC)10227.90 ± 5.72b(BC)
Plant heighti8181.99 ± 16.47a(AC)3597.97 ± 13.71b(B)580.48 ± 18.64ac(ABC)1190.28 ± 15.51ab(AB)361.85 ± 2.37c(C)
ii105103.73 ± 20.26a(A)41123.17 ± 19.16b(B)6111.50 ± 30.30ab(AB)12117.00 ± 27.49bc(AB)9103.19 ± 8.14ac(A)
iii110115.70 ± 19.34a(A)42139.51 ± 18.91b(B)9116.28 ± 24.98ac(A)14127.65 ± 25.84bc(AB)8111.70 ± 5.63ac(A)
Peduncle lengthi8128.10 ± 6.90a(AB)3530.82 ± 4.77b(A)526.92 ± 8.71abc(AB)1130.22 ± 8.74ab(AB)320.13 ± 3.36c(B)
ii10831.11 ± 5.53a4132.34 ± 5.21ab632.45 ± 9.12ab1235.74 ± 10.15b930.44 ± 4.80a
iii10336.02 ± 5.89a(A)4239.67 ± 5.97bc(BC)934.36 ± 5.74a(AB)1441.66 ± 9.57c(C)835.64 ± 3.50ab(ABC)
Spike lengthi819.39 ± 1.76a(A)3510.48 ± 1.72b(B)512.54 ± 1.27c(B)1111.15 ± 2.69bc(B)310.77 ± 0.21abc(AB)
ii1089.55 ± 1.70a419.65 ± 2.01ab610.47 ± 0.92ab1210.39 ± 2.79ab910.91 ± 0.77b
iii1109.80 ± 1.92a429.59 ± 2.12a911.42 ± 1.25b1410.76 ± 2.93ab811.56 ± 1.17b
Spikelet number per spikei8118.04 ± 2.64a3518.97 ± 2.25ab520.61 ± 1.57b1119.62 ± 3.32ab318.33 ± 0.42ab
ii10819.09 ± 1.97a4119.50 ± 2.14a620.00 ± 2.62a1219.15 ± 4.07a918.19 ± 1.92a
iii10919.33 ± 2.10a4219.23 ± 2.37a919.64 ± 2.26a1419.46 ± 3.11a819.54 ± 1.72a
Kernel number per spikei8143.19 ± 9.33ab3539.76 ± 7.61a549.07 ± 6.05b1144.16 ± 9.01ab351.93 ± 0.42b
ii10843.95 ± 6.38a(A)4142.36 ± 7.03a(AB)639.42 ± 5.32ab(AB)1241.20 ± 9.78ab(AB)937.32 ± 6.92b(B)
iii10949.08 ± 8.32a4245.42 ± 10.49b942.28 ± 4.56b1447.48 ± 13.47ab847.71 ± 6.18ab
Spike numberi818.30 ± 2.65a348.03 ± 2.70a57.16 ± 1.76a119.61 ± 3.64a38.60 ± 1.74a
ii10810.61 ± 2.06a4110.70 ± 2.48ab610.69 ± 1.71ab1112.12 ± 1.89b910.48 ± 2.89ab
iii10911.98 ± 3.03a4212.17 ± 3.26a710.89 ± 2.50a1412.07 ± 3.66a812.42 ± 2.42a
Thousand kernel weighti8129.52 ± 5.96a(A)3526.83 ± 6.06bd(A)522.64 ± 5.44d(A)1130.56 ± 7.00ab(AB)338.84 ± 4.25c(B)
ii10829.75 ± 5.77a(A)4126.13 ± 5.13b(B)628.88 ± 4.57ab(ABC)1230.59 ± 5.30a(ABC)936.14 ± 8.14c(C)
iii11033.47 ± 6.77a(A)4029.24 ± 5.55bc(B)926.63 ± 4.34b(B)1432.25 ± 4.09ac(AB)838.03 ± 4.79d(A)

Correlations of relative expression of Ppd-D1 haplotypes and days to heading of varieties

Haplotype III expressed at very low levels and showed later heading; haplotype I was highly expressed and showed earlier heading and the others were intermediate. To further understand the role of haplotypes in heading time, we analysed the correlation of relative expression abundance of Ppd-D1 haplotypes and days to heading of varieties (Fig. 3). Linear regression analysis demonstrated that there was a negative, but not significant, correlation (R2 = 0.4619, = 0.207) between relative expression level in the morning and heading date of varieties grown in Beijing (116.2° E, 40.2° N).There was a comparatively stronger correlation (R2 = 0.6673, = 0.091) in environment ii (111.6°E, 33.8°N) where the daylength was shorter than in environment i and where heading time in sensitive varieties was delayed relative to Beijing.

Evolutionary history of Ppd-D1 haplotypes

A total of 547 accessions were scored for Ppd-D1 allelic variation using molecular markers. The six Ppd-D1 haplotypes were differently distributed among common wheat and A. tauschii. Only haplotypes II, V and VI were present in A. tauschii, whereas haplotypes I, II, III and IV were present in common wheat, indicating that haplotypes V and VI were relatively ancient and unselected.

A haplotype tree was constructed using a maximum parsimony criterion. There were four equally parsimonious clades from haplotype II and only one step variations occurred between them (Fig. 4). Haplotype II was very close to haplotype V with a difference of only the 16 bp polymorphic region in exon 8, suggesting that haplotype II originated from haplotype V. Haplotype V was similar to haplotype VI with a difference of 24 bp plus 15 bp deletions in the promoter region. There were also only single polymorphic differences between haplotypes II and I, II and III, and II and IV, which were a 2 kb deletion upstream of the coding sequence, a TE insertion in intron 1 and a 5 bp deletion in exon 7, respectively, These variations suggested that haplotypes I, III and IV were each derived from haplotype II and that the variation permitted adaptation to diverse growing environments. We speculated that V and VI were ancient haplotypes, occurring only in the wild species, whereas haplotype II was shared by wild and cultivated species from which haplotypes I, III and IV were more recently derived.

Figure 4.

 Haplotype tree of Ppd-D1. Circles with Roman numbers denote haplotypes and each branch represents one mutation. Haplotype II occurs in both wild and cultivated species. The arrow indicates the evolutionary direction.

Geographical distribution of Ppd-D1 haplotypes

In order to investigate the worldwide distribution of the Ppd-D1 haplotypes, 492 varieties selected from six continents and 41 countries were surveyed (Fig. 5a, Table S5). Our main focus was on Asian (mostly Chinese) varieties, which comprised 51.6% of the total. Southern, western and eastern Asian varieties were similar, with highest frequencies of haplotype I followed by haplotype II. Haplotype III was not found in any Asian country except for a few accessions in China. The frequency of haplotype I (53.6%) was also higher in Oceania where the second most frequent haplotype was IV (35.7%). Because accessions from Mexico, located in lower latitudes (ranging from c. 16° N to 30° N) was dominated by haplotype I, the overall frequency of haplotype I (55.2%) was higher than those of II, III and IV in North American. Haplotype III occurred in the USA and Canada where the latitude was higher (ranging from c. 30° N to 65° N).

Figure 5.

 Distribution of Ppd-D1 haplotypes I–IV in six continents. Black, white, dark gray and light gray sections of pie charts show the proportions of Ppd-D1 haplotypes I, II, III and IV, respectively. (a) Distribution of Ppd-D1 haplotypes in world wheat growing regions; (b) Distribution of Ppd-D1 haplotypes in 10 Chinese wheat-growing regions.

Haplotypes I, II and IV occurred at similar frequencies (33.3%, 25.6% and 28.2%, respectively) in South America. Haplotype I frequencies for Europe, and especially Africa (4.55%), were lower than in most other areas. In Europe, haplotype distribution differed from south to north, with haplotype I being more frequent in the south (45.5%) than in the north and west (8.0%) where the latitude was higher and where haplotype III predominated (52.0%). African varieties were predominantly haplotype IV (72.7%).

Thus, haplotype I was most common in Asia, Oceania and Mexico (the lower latitude regions), haplotype II was mainly distributed in Asia and haplotype III prevailed in the higher latitude regions of Europe and North America, but was absent in Oceania, Africa and other Asian regions, except China. Haplotype IV was the most universally distributed across all continents, except Asia.

Ten major wheat agroecological regions were defined in China on the basis of environment and wheat growth types (Dong & Zheng, 2000; Zhuang, 2003) (Fig. 5b). With about half of all wheat accessions coming from China, we analysed the geographic distribution of Ppd-D1 haplotypes in that country (Fig. 5b). Haplotype I was the most frequent (57.4%), but its frequency varied across regions, being higher in the Middle and Lower Yangtze Valley Winter Wheat Region (MLYW, 93.1%) and South China Winter Wheat Region (SCW, 91.6%), the lower latitude and altitude regions. It was absent in the Sinkiang Winter–Spring Wheat Region (SKWS, 0%), the higher altitude region. The haplotype I frequency in the other regions changed from high to low in the following pattern: Yellow and Huai Winter Wheat Region (YHW, 70.7%), Northern Winter Wheat Region (NW, 54.2%), Southwestern Winter Wheat Region (SwW, 48.1%), Northeastern Spring Wheat Region (NeS, 37.5%), Northern Spring Wheat Region (NS, 36.3%), Northwestern Spring Wheat Region (NwS, 33.3%) and Qinghai–Tibet Spring–Winter Wheat Region (QTSW, 23.5%). Haplotype II appeared in all the Chinese agroecological regions and its frequency was far higher than those of haplotypes III and IV among Chinese varieties. Haplotypes III and IV were relatively uncommon, and when present, they were found in the NeS and SKWS regions.

The association between Ppd-D1 haplotype frequencies and latitude or altitude of the 10 regions was analysed. Because the expressions of haplotypes II and III were lower and not significantly different from each other, they were pooled for analysis. There was a significant negative correlation between distributional frequencies of haplotype I and latitude as well as altitude (Table 2). The distributional subtotal frequencies of haplotypes II and III also had a significant positive correlation with altitude and latitude. The results showed that photoperiod responsiveness correlated with the latitudes and altitudes of the source regions reflecting the conditions of day length and temperature in their growth and developmental stages. The SKWS, QTSW northeast and northwest regions at higher latitudes and altitudes had relatively longer days at critical stages in their specific growing seasons (Zhuang, 2003), so these were dominated by photoperiod sensitive varieties. The Middle and Lower Yangzte Valley, Yellow and Huai district and South China regions are located in lower latitudes, where the climate is unfavorable for photoperiod sensitive varieties because of comparatively shorter daylengths, and high temperatures in summer, combined with rain and low light intensity. Therefore, many varieties derived from these regions are photoperiod insensitive allowing them to flower under short day conditions with milder temperatures.

Table 2.   Partial correlations of latitude, altitude and distribution of Ppd-D1 haplotypes in China
Haplotype IHaplotypes II +  IIIHaplotype IV
  1. *= 0.05 **= 0.01.


The haplotype I frequency was higher in Chinese cultivars, and it could be further divided into two groups with a lower frequency in landraces (39.7%) compared with modern varieties (77.9%). Hence, there was a clear trend for modern varieties to have a photoperiod insensitive phenotype, consistent with a previous report by Yang et al. (2008).


Five polymorphic sites were detected in the Ppd-D1 gene; two in the promoter region and one in the first intron were close to the 5′ transcription start site. The other two polymorphisms were located in exons 7 and 8, near to the 3′ end of the transcription region. No polymorphic site was detected in the middle region of the gene. The polymorphism distribution in Ppd-D1 was similar to that of polymorphic SNP sites in the human genome in which there were higher frequencies of polymorphic SNPs around the 5′ coding start sites and near the 3′ end of the coding sites (The International HapMap Consortium, 2007). Variable transcription levels seemed to be important for regulating genetic function, and both the promoter region and first intron appeared more likely to be involved. To adapt to diverse environments, more polymorphic variations in these regions were retained during wheat domestication and improvement.

In this work we discovered new Ppd-D1 haplotypes, and investigated their expression, evolution, association with agronomic traits and the distribution, respectively. In fact, all of these features should be closely associated. The haplotypes determined gene expression and function, which presumably limited the genotypic distribution through adaptation and plant ecotype development.

Haplotype I showed the highest expression level among the Ppd-D1 haplotypes and it had a 2 kb deletion in the promoter region relative to haplotype II. We speculated that there was a binding site for its transcription factor, a negative regulator, within this 2 kb region and that loss of the site (or of recognition) promoted the expression of Ppd-D1. Varieties belonging to haplotype I therefore had the earliest heading time and prevailed in regions with short growth periods (i.e. lower latitude and altitude regions as reported by other workers; Jin, 1961; Martiníc, 1975; Hunt, 1979; Worland et al., 1998). Haplotype I is likely a mutant from haplotype II which was distributed mainly in Asia. This suggests that Ppd-D1a originated in Asia. Haplotype I was detected in many Chinese landraces, such as ‘Mazhamai’, ‘Youzimai’, ‘Jiangdongmen’ and ‘Wangshuibai’ in the present study, and in Japanese landraces, such as ‘Akakomugi’ (Borojevic, 2005). These landraces were used widely as parents in Chinese and worldwide wheat-breeding programs.

Haplotype III showed the lowest expression level among the Ppd-D1 haplotypes and differed from II by a TE element inserted in the first intron. The insertion may regulate transcript level through an effect on transcript processing similar to that reported for the FLC gene of Arabidopsis (Liu et al., 2004) and the VRN-H1 and Vrn-1 alleles (Fu et al., 2005), the latter conferring spring growth habit in barley and wheat. This result suggested the first intron of a gene is important for regulating gene expression and that a mutation in the region likely plays a key role in creating variation that has agronomic significance for cereals. This insertion in Ppd-D1 may be responsible for the latest ear emergence and longest growth period, providing genotypes adapted to northern latitudes and higher altitude regions.

The daylength-sensitive haplotype II, which occurs in the wild species as well as in common wheat, appears to be the progenitor of haplotypes I, III and IV. Its expression abundance was close to that of haplotype III and significantly lower than that of haplotypes I and IV. Haplotype II varieties with a heading time intermediate between I and III, are distributed mainly in higher altitude and latitude regions of China, especially in landraces.

Haplotype IV is characterized by a 5 bp deletion in exon 7. Its peak expression value was intermediate, being 5% higher than those of haplotypes II and III, but lower than haplotype I. Varieties with haplotype IV were widely distributed across all continents; their heading times and other agronomic traits were intermediate to haplotypes I and III. All the characteristics of haplotype IV were consistent with the gene expression level.

The existence of at least six haplotypes of Ppd-D1 allows for continuous variation in the action of a single gene. Daylength insensitive haplotype I varieties planted in lower latitudes showed relatively earlier heading, whereas sensitive haplotype III varieties with later heading occurred in higher latitudes. Between the extremes of III and I, haplotypes II, IV, V were intermediate in expression. Thus what was previously regarded as a pair of sensitive and insensitive alleles can now be represented as a series of alleles.

The discovery of a series of haplotypes with apparent adaptive significance suggests many more opportunities for fine tuning genotypes to environmental conditions. Clearly, a situation previously envisaged as two loci, Ppd-D1 and Ppd-B1, each with two alleles is much more complex. In addition to these loci, other loci regulating photoperiod response were reported on chromosomes 6B, 7A, 7B and 7D (Sourdille et al., 2000; Kuchel et al., 2006), where the allelic numbers at each locus might be more than two. The potential of such a large number of genotypes could obviously contribute to the wide adaptability of wheat, especially when other genetic systems, such as response to vernalization are also considered.


The project was sponsored by the Chinese National Basic Research ‘973’ Program (2004CB117200 and 2010CB125900) and Chinese National ‘863’ Program (2006AA10A104). We are grateful to: Dr David A. Laurie, Crop Genetics Department, John Innes Centre, UK; Dr Robert McIntosh, the University of Sydney, Australia; Dr Song Ge, Institute of Botany, Chinese Academy of Sciences; and Dr Yongfu Fu, Institute of Crop Science, CAAS, for help with revising this manuscript.