The genetic basis of complex trait variation is a major challenge for modern biology. There are long-standing debates over whether alleles are common or rare, loci few or many, effects small or large, interactions additive or epistatic, and so on—questions of practical importance for agriculture and human health, and of fundamental importance for evolutionary biology. Lately, the discourse has been heavily influenced by the mass of human genome-wide association studies (GWAS), which have generally revealed a highly polygenic basis for complex traits, with numerous causative polymorphisms of small effect distributed across the genome (Visscher et al., 2012). However, the relevance of these studies for understanding adaptive variation is not clear, as they have almost exclusively involved traits likely to have been under stabilizing or purifying selection. Indeed, among the few GWAS of adaptive variation that have been carried out, some have identified common alleles of major effect (including for human traits such as pigmentation) (Atwell et al., 2010; Beleza et al., 2013).

Here we consider seed dormancy, a complex life history trait that plays a major role in local adaptation for many plant species (Finch-Savage and Leubner-Metzger, 2006). For a weedy annual like Arabidopsis thaliana, the subject of our study, germinating at the right time is crucial for early survival, and also influences other life history traits, such as flowering (Postma and Ågren, 2016; Huang et al., 2010a; Chiang et al., 2013; Donohue et al., 2010; Springthorpe and Penfield, 2015). Seed dormancy varies geographically, with high dormancy being associated with long, dry summers, and low dormancy being more prevalent where summers are short and wet (Chiang et al., 2011; Debieu et al., 2013; Kronholm et al., 2012). Analysis of a diallel cross between a highly dormant line from the Cape Verde Islands and a non-dormant line from Poland revealed several quantitative trait loci (QTL), one of which, DELAY OF GERMINATION1 (DOG1), has been cloned and molecularly characterized (Alonso-Blanco et al., 2003; Bentsink et al., 2006), although its function is poorly understood. It appears to be tightly regulated by a complex array of mechanisms that include alternative splicing, alternative polyadenylation, and a cis-acting antisense non-coding transcript (asDOG1) (Bentsink et al., 2006; Nakabayashi et al., 2015; Cyrek et al., 2016; Fedak et al., 2016), but, although DOG1 mRNA and protein abundance generally correlate with seed dormancy, this relationship is not observed in all accessions (Bentsink et al., 2006; Chiang et al., 2011; Nakabayashi et al., 2012; Fedak et al., 2016). In these cases, epistatic interactions with the genetic background, or other mechanisms, such as the critical self-dimerization of the DOG1 protein, might come into play (Nakabayashi et al., 2015). The DOG1 locus harbors a series of functionally distinct alleles segregating in wild populations, and exhibits a pattern of genetic differentiation suggestive of local adaptation (Kronholm et al., 2012). Moreover, field experiments using a diallel cross between accessions from Sweden and Italy identified a major QTL associated with total fitness that spanned the DOG1 region, and also demonstrated that genetic differences in seedling establishment—and most particularly in seed dormancy—are one of the main components of local adaptation (Postma and Ågren, 2016; Postma et al., 2016; Huang et al., 2010a). Finally, extensive natural variation for seed dormancy is not unique to the DOG1 locus and many other QTLs have been mapped in various populations (Alonso-Blanco et al., 2003; Bentsink et al., 2006, 2010; Amiguet-Vercher et al., 2015; Xiang et al., 2016). The recent cloning of the DELAY OF GERMINATION18 (DOG18)/IBO QTLs in two independent diallel crosses revealed the existence of numerous loss-of-function alleles—mainly caused by nonsense mutations and structural variants—at the REDUCED DORMANCY5 (RDO5) locus in wild A. thaliana populations (Xiang et al., 2016; Amiguet-Vercher et al., 2015; Xiang et al., 2014).

To investigate the genetic basis for seed dormancy variation in a natural population, we carried out a GWAS using 161 lines from Sweden (Figure 1—source data 1), where seeds are normally set in early summer and seedling establishment mostly occurs in fall. We focused on the germination rate of seeds after-ripened for 21 days (GR21), which showed dramatic variation with 46% of lines being dormant (GR21 < 20%) and 35% being effectively non-dormant (GR21 > 80%) (Figure 1A). We observed a strong correlation between GR21 and latitude (Pearson’s r = 0.62) (Figure 1B), largely due to the fact that northern Swedish lines were mostly non-dormant whereas the majority of southern Swedish lines was dormant (Figure 1C and online: https://goo.gl/3HeqX5). GR21 was also strongly correlated with both temperature and length of growing season, consistent with genetic variation for seed dormancy being adaptive (Supplementary file 1).

Figure 1

Download asset Open asset

Figure 1—source data 1

The 161 Swedish accessions used in this study.

Geographic information and GR21 values used to generate Figure 1 and perform GWAS.

https://doi.org/10.7554/eLife.22502.003

Download elife-22502-fig1-data1-v3.xlsx

GWAS for GR21 identified a single highly significant peak spanning almost 50 kb at about 18.6 Mb on chromosome 5 (Figure 2A and online: https://goo.gl/30EPt3). The region includes DOG1, but the significantly associated SNPs were located 21 kb away from this obvious candidate, and several other genes in the region could possibly be involved in dormancy (Figure 2B and D, Figure 2—figure supplement 1 and Materials and methods). Analysis of tDNA insertion lines (in the Col-0 background) for three of these genes revealed that CIPK20—the first gene downstream of DOG1—could potentially affect dormancy as well (although it could also be the case that the cipk20 tDNA insertion directly affects regulation of DOG1; see Figure 2—figure supplement 1). Thus, although DOG1 seemed an excellent candidate for explaining dormancy variation (Alonso-Blanco et al., 2003; Bentsink et al., 2006, 2010; Kronholm et al., 2012), we could not exclude that neighbouring genes also contribute.

Figure 2 with 3 supplements see all

Download asset Open asset

However, a simpler reason for the distant association is that allelic heterogeneity at the DOG1 locus (previously described by Kronholm et al., 2012) gives rise to ‘synthetic’, or ‘ghost’ associations in the region. This phenomenon, which is well-documented in both A. thaliana (Atwell et al., 2010) and rice (Huang et al., 2010b), is a simple statistical artifact that arises from fitting a single-locus, two-allele model when the true cause of phenotypic variation is multiple correlated factors (Atwell et al., 2010; Dickson et al., 2010; Platt et al., 2010).

To look for evidence of allelic heterogeneity and synthetic associations, we examined linkage disequilibrium (LD) between the most strongly associated SNP (SNP1, −log₁₀p = 8.73) and the other markers in the region. As is typical in A. thaliana (Kim et al., 2007; Long et al., 2013), LD decayed within 10 kb of SNP1, however, we identified one marker, SNP2, in moderate LD (r² = 0.61) with SNP1 despite a separation of almost 21 kb (Figure 2—figure supplement 2). SNP2 was strongly, albeit non-significantly, associated with GR21 (−log₁₀p = 5.80) and is located in the promoter region of DOG1. As DOG1 expression is often correlated with seed dormancy (Bentsink et al., 2006; Chiang et al., 2011; Nakabayashi et al., 2012; Fedak et al., 2016), we hypothesized that SNP2 could contribute to GR21 variation, and used it as starting point for a multi-locus GWAS based on stepwise inclusion of markers (Segura et al., 2012). The optimal model resulting from starting with SNP2 explained 57% of the variation and consisted of two markers in addition to SNP2: SNP3, also in the DOG1 promoter; and SNP4, a rare allele [minor allele frequency (MAF) = 0.018] in the first exon of DOG1 (Figure 2C–D and Figure 2—figure supplement 3B and E). In contrast, the optimal model resulting from starting with SNP1 explained only 46% of the variation, and identified only one additional marker: SNP4 (Figure 2—figure supplement 3A and D). Starting from SNP3, on the other hand, yielded the same model as starting from SNP2, and it is clear that the combined effect of SNP2 and SNP3 effectively eliminates the signal at SNP1 (Figure 2—figure supplement 3B–C and E–F), suggesting that it (and the small peak around it) is a synthetic association that serves as a proxy for SNP2 and SNP3 (or closely linked variants: SNP3 is in perfect LD with two other nearby SNPs that can explain the variation just as well; see Materials and methods). The rare SNP4, on the other hand, appears to be independent of the other SNPs. In line with these stepwise results, we found that SNP2 and SNP3 were the best pairwise predictors of the dormancy variation (see Materials and methods and Supplementary file 2). Taken together, these GWAS results strongly suggest that variants in the DOG1 promoter region underlie dormancy variation in the Swedish population (but say nothing about the molecular mechanism, a fact to which we will return below).

To confirm the hypotheses generated by GWAS, we turned to transgenics and experimental crosses. Linkage analysis in six different F2 populations confirmed the importance of the DOG1 region (Figure 3A and Figure 3—source data 1). Two of the crosses (TFÄ 07 x Stenk-3 and Eden-2 x ÖMö1-7) demonstrated that the dormant SNP2 allele does not require SNP1, as expected if the association observed at the latter is synthetic. However, due to the strong linkage disequilibrium between SNP3 and SNP1, no cross let us similarly demonstrate that the dormant SNP3 has an effect independently of SNP1. To overcome these limitations, we complemented the non-dormant dog1-2 mutant with genomic fragments—promoter and gene—corresponding to different DOG1 alleles. Lines transformed with predicted dormant alleles carrying either SNP2 or SNP3 (but not SNP1, which is not part of the transformed fragment) were significantly more dormant than those transformed with a predicted non-dormant allele (Figure 3B and Figure 3—source data 2). It is thus clear that DOG1 alleles tagged by SNP2 and SNP3 are capable of producing a dormant phenotype independently of SNP1 (and other closely linked genes, like CIPK20). Finally, two additional F2 populations (TEDEN 03 x TÅL 07 and TAA 14 x TEDEN 03) supported the association between dormancy and SNP4, thus bringing the number of distinct dormant DOG1 alleles to three (Figure 3A).

Figure 3

Download asset Open asset

Figure 3—source data 1

QTL mapping results.

LOD scores and positions of 95% confidence interval of the major QTLs presented in Figure 3A.

https://doi.org/10.7554/eLife.22502.009

Download elife-22502-fig3-data1-v3.xlsx

Figure 3—source data 2

Phenotypes of the DOG1 transgenic lines.

GR21 values of the transgenic lines used to generate Figure 3B.

https://doi.org/10.7554/eLife.22502.010

Download elife-22502-fig3-data2-v3.xlsx

Figure 3—source data 3

Information about Bar 1, an extra Swedish accession used as parent in one of the F2 mapping population.

https://doi.org/10.7554/eLife.22502.011

Download elife-22502-fig3-data3-v3.xlsx

Figure 3—source data 4

SNPs and haplotypes at the DOG1 locus for the 161 Swedish accessions used in this study.

We provide the genotype of the accessions at SNP1-SNP4 as well as their DOG1 haplotype.

https://doi.org/10.7554/eLife.22502.012

Download elife-22502-fig3-data4-v3.xlsx

Figure 3—source data 5

Phenotypes and genotypes of the F2 individuals used in the QTL mapping approach.

This file also contains the NCBI SRA BioProject and BioSamples accession numbers.

https://doi.org/10.7554/eLife.22502.013

Download elife-22502-fig3-data5-v3.xlsx

Figure 3—source data 6

DOG1 PCR primers used in the complementation experiment.

The underlined sequences correspond to the flanking sequences of the EcoRI restriction site in the pGreenII 0029 that was used for transformation.

https://doi.org/10.7554/eLife.22502.014

Download elife-22502-fig3-data6-v3.docx

In summary, GWAS identified four DOG1 haplotypes, three associated with dormancy (D2-D4, tagged by the late alleles of SNP2-SNP4, respectively), and one associated with non-dormancy (ND, tagged by early alleles at all three SNP loci) (Figure 3C). These SNPs may be causative, but we emphasize that SNP3 is in perfect LD with two other SNPs a few bp away that can explain the variation just as well (see Materials and methods and Supplementary file 2), and that there may also be structural variants not present in the data we used to carry out GWAS (Long et al., 2013). Furthermore, it is difficult to completely rule out that these SNPs are themselves tagging rare alleles, just like SNP1 is tagging them (Figure 3C). Fully establishing causality would require extensive experiments using lines engineered to test all polymorphisms independently of each other, and would be almost impossible without techniques for in situ gene replacement given the variation in gene expression that results from transgene insertion (Figure 3B). It is also crucial to distinguish between causality in the genetic sense (the focus of this paper) and molecular mechanism: we emphasize that we do not know whether these SNPs affect dormancy by regulating DOG1, asDOG1 or even CIPK20. Given the apparent complexity of DOG1 regulation, elucidating the causative mechanism will be a non-trivial task. It should also be noted that further alleles that we do not have the power to detect, either because they are rare or have small effect, may exist. However, while it is impossible to exclude the possibility that there are alleles we did not detect, a variance-components analysis (Sasaki et al., 2015) estimated that 51% of the dormancy variation is due to genetic variation in the 4.8 kb surrounding DOG1, which suggests that most of the variation at DOG1 has been accounted for by the identified haplotypes (see Materials and methods). The variation in dormancy between lines carrying the same DOG1 haplotype (Figure 3C) is probably due to variation at other loci (the existence of which is evident from GWAS and the F2 crosses, see Figure 2A, Figure 3A and online: https://goo.gl/30EPt3). DOG18/RDO5/IBO, because of its recently characterized multiple natural alleles, was a good candidate to explain part of the remaining dormancy variation. However, no association was observed at this gene, presumably because the four distinct loss-of-function alleles (carried by Vår2-6, Löv-1, ÖMö2-1 and Hov3-2) segregating in our population appear to be singletons and thus cannot be detected using GWAS (and they do not segregate in our six F2 populations).

The geographic distribution of the four DOG1 haplotypes mirrors the GR21 distribution strikingly well (Figures 1C and 3D and online: https://goo.gl/3HeqX5). The majority of northern lines carry the non-dormant haplotype ND, with a small number carrying D4. Southern lines are much more diverse, with about half carrying ND, and the rest carrying either D2 or (much more rarely) D3.

To confirm that this striking variation is not an artifact of laboratory conditions, we performed field experiments in southern Sweden, utilizing two of our previous mapping populations: T880 (D2) x Grön 14 (ND) and Bar 1 (ND) x Ull2-5 (D3) (Figure 3A). Our goal was to track changes in DOG1 haplotype frequency across time, from seed release to the end of the germination window. Because timing of germination partly depends on the interaction between genes and seed maturation environment (Postma and Ågren, 2015; Chiang et al., 2011; Donohue et al., 2005), we first grew a set of F2 individuals for each population in a common garden experiment, from germination to ripening. The progeny of these plants (F3 seeds) was then mixed and dispersed in experimental plots in early summer, using a two-block design for replication. We sampled all visible seedlings (destructive sampling) in September and November and also sampled soil shortly before winter in order to gain access to the live seeds present in newly-formed seed bank (see Materials and methods and Supplementary file 3A).

The results were consistent across replicates, and demonstrated that the DOG1 haplotypes can have a major effect on survival. For both populations, the frequency of the dormant haplotype increased over time (as expected given delayed germination), but the details varied greatly (Figure 4, Supplementary file 3B–C, Figure 4—figure supplement 1 and Figure 4—source data 1). In the D2 x ND population, the frequency of D2 increased moderately between seed dispersal and September, and appeared to continue increasing throughout fall (it was slightly lower in the seed bank) (Figure 4A). This is consistent with D2 delaying germination, and delayed germination increasing seedling survival. In contrast, the frequency of D3 in the ND x D3 population decreased dramatically between seed dispersal and September, but recovered rapidly by November, indicating a dramatic delay in germination (Figure 4B). Furthermore, the D3 frequency in the seed bank was very high, demonstrating that this haplotype not only caused a delay in seed germination but also prevented seeds from germinating at all (Figure 4B).

Figure 4 with 1 supplement see all

Download asset Open asset

Figure 4—source data 1

Genome-wide genotypes of the field experiments samples.

This file also contains sampling information (block, frame, sampling point etc.) and the NCBI SRA BioProject and BioSamples accession numbers.

https://doi.org/10.7554/eLife.22502.016

Download elife-22502-fig4-data1-v3.xlsx

With all live individuals at the end of the experiment counted (i.e., with the seed banks included), the proportion of dormant alleles increased massively in both populations, presumably because non-dormant seeds germinated under unfavorable conditions, and died before they could be sampled (Figure 4C–D). The implied selection strengths are enormous: the viability of homozygous non-dormant DOG1 genotypes relative to homozygous dormant ones until winter was 50–60% depending on population (Supplementary file 3D). Of course there will be further mortality in the seed bank and not all established seedlings will survive and reproduce, but the results serve to illustrate that selection on DOG1 could be very strong indeed, and lend credence to the notion that the observed distribution of dormancy and DOG1 alleles (Figures 1C and 3D) reflects local adaptation. Although our experiments have limitations and do not allow us to decompose total fitness, our results agree with those from previous studies, and it becomes very clear that germination timing is one of the main (if not the main) fitness components (Postma and Ågren, 2016; Huang et al., 2010a). To fully understand how selection acts will require multi-generational field experiments as selection pressures are likely to vary from year to year (and are unlikely to be at equilibrium, because of large-scale climate variations). Under the conditions of our experiment, D2 (the most common dormant haplotype) appears to have had a clear advantage over ND, whereas D3 may have been too dormant (Figure 4).

Furthermore, strong selection could also lead to adaptation on a very local scale. Generally speaking, deep dormancy may allow plants from the south to avoid periodic drought and high temperatures in summer, whereas northern populations may benefit from a shallower dormancy to cope with a short vegetative growth window. However, the fact that southern Sweden harbors large numbers of both dormant and non-dormant DOG1 haplotypes demonstrates that variation for seed dormancy can be found at a very local scale, presumably due to microenvironmental variation, and the distribution of SNP4, which segregates in northern Sweden only, suggests that moderate levels of dormancy could also be beneficial at this latitude.

Finally, our results provide a dramatic illustration that the ‘holy grail’ of molecular ecology — finding ‘the genes that matter’—may sometimes actually be obtainable (Mitchell-Olds, 2001; Rockman, 2012). That a single locus should explain most of the variation for a major life history trait is certainly surprising and arguably unprecedented. Of course we do not know how much of dormancy variation DOG1 would explain under field conditions (and we certainly do not know how much of the variation for total fitness it would explain), but our data suggest that it is a major contributor. Although other loci also appeared to be under selection in our field experiments—as expected given that the F2 results show that other loci also contribute strongly to variation for dormancy—allele frequency changes at DOG1 were among the strongest in both populations (Figure 3A and Figure 4—figure supplement 1). Interestingly, GWAS in a global population also identifies DOG1 as strongly associated with flowering time variation (Atwell et al., 2010; 1001 Genomes Consortium 2016), although it is not yet clear whether this association is real or spurious. If the latter, a likely explanation is that correlated selection pressures on two related life history traits—germination and flowering—has generated cross-trait genotype-phenotype correlations. If the former, DOG1 emerges as an evolutionary ‘master switch’ for life history variation, perhaps similar in importance to the central flowering regulator FLC (Li et al., 2014). Indeed a recent study has demonstrated that DOG1 may affect both dormancy and flowering (Huo et al., 2016), and it is interesting to note that both genes appear to be regulated through similar mechanisms, with chromatin remodelling and antisense transcripts switching off sense transcription (Swiezewski et al., 2009; Footitt et al., 2015; Fedak et al., 2016). Perhaps such genes play general role in regulating and modulating environmentally dependent life history transitions?

In conclusion, the study of adaptive traits variation in A. thaliana is starting to reveal a similar picture: multiple, often complex, alleles at a relatively small number of loci account for much of the variation (Li et al., 2014; Sasaki et al., 2015; Chao et al., 2012; Poormohammad Kiani et al., 2012; Forsberg et al., 2015). The phenomenon is certainly not limited to A. thaliana—pigmentation often appears to behave similarly, for example, in humans (Beleza et al., 2013) as well as Drosophila (Rebeiz et al., 2009)—nor is it universal in this species (Atwell et al., 2010). While it is straightforward to show that local adaptation may lead to such an architecture (Yeaman and Whitlock, 2011), much more data—from a variety of organisms and phenotypes—are needed before we can draw general conclusions about when and why this happens.

1. 1. Kerdaffrec E
   2. Filiault DL
   3. Korte A
   4. Sasaki E
   5. Nizhynska V
   6. Seren Ü
   7. Nordborg M

   (2016) Low coverage sequencing of three A. thaliana F2 populations for QTL mapping

   Publicly available at NCBI SRA.

   https://www.ncbi.nlm.nih.gov/bioproject/PRJNA343766
2. 1. Kerdaffrec E
   2. Filiault DL
   3. Korte A
   4. Sasaki E
   5. Nizhynska V
   6. Seren Ü
   7. Nordborg M

   (2016) Low coverage sequencing of A. thaliana samples from Swedish field experiments

   Publicly available at NCBI SRA.

   https://www.ncbi.nlm.nih.gov/bioproject/PRJNA343768

1. 1. Long Q
   2. Rabanal FA
   3. Meng D
   4. Huber CD
   5. Farlow A
   6. Platzer A
   7. Zhang Q
   8. Vilhjálmsson BJ
   9. Korte A
   10. Nizhynska V
   11. Voronin V
   12. Korte P
   13. Sedman L
   14. Mandáková T
   15. Lysak MA
   16. Hellmann I
   17. Nordborg M

   (2013) Massive genomic variation and strong selection in Arabidopsis thaliana lines from Sweden

   Publicly available at Github.

   https://github.com/Gregor-Mendel-Institute/swedish-genomes

1. 1. Alonso-Blanco C
   2. Bentsink L
   3. Hanhart CJ
   4. Blankestijn-de Vries H
   5. Koornneef M

   (2003)

   Analysis of natural allelic variation at seed dormancy loci of arabidopsis thaliana

   Genetics 164:711–729.

   • PubMed
   • Google Scholar
2. 1. Amiguet-Vercher A
   2. Santuari L
   3. Gonzalez-Guzman M
   4. Depuydt S
   5. Rodriguez PL
   6. Hardtke CS

   (2015) The IBO germination quantitative trait locus encodes a phosphatase 2C-related variant with a nonsynonymous amino acid change that interferes with abscisic acid signaling

   New Phytologist 205:1076–1082.

   https://doi.org/10.1111/nph.13225

   • PubMed
   • Google Scholar
3. 1. Arends D
   2. Prins P
   3. Jansen RC
   4. Broman KW

   (2010) R/qtl: high-throughput multiple QTL mapping

   Bioinformatics 26:2990–2992.

   https://doi.org/10.1093/bioinformatics/btq565

   • PubMed
   • Google Scholar
4. 1. Atwell S
   2. Huang YS
   3. Vilhjálmsson BJ
   4. Willems G
   5. Horton M
   6. Li Y
   7. Meng D
   8. Platt A
   9. Tarone AM
   10. Hu TT
   11. Jiang R
   12. Muliyati NW
   13. Zhang X
   14. Amer MA
   15. Baxter I
   16. Brachi B
   17. Chory J
   18. Dean C
   19. Debieu M
   20. de Meaux J
   21. Ecker JR
   22. Faure N
   23. Kniskern JM
   24. Jones JD
   25. Michael T
   26. Nemri A
   27. Roux F
   28. Salt DE
   29. Tang C
   30. Todesco M
   31. Traw MB
   32. Weigel D
   33. Marjoram P
   34. Borevitz JO
   35. Bergelson J
   36. Nordborg M

   (2010) Genome-wide association study of 107 phenotypes in arabidopsis thaliana inbred lines

   Nature 465:627–631.

   https://doi.org/10.1038/nature08800

   • PubMed
   • Google Scholar
5. 1. Beleza S
   2. Johnson NA
   3. Candille SI
   4. Absher DM
   5. Coram MA
   6. Lopes J
   7. Campos J
   8. Araújo II
   9. Anderson TM
   10. Vilhjálmsson BJ
   11. Nordborg M
   12. Correia E Silva A
   13. Shriver MD
   14. Rocha J
   15. Barsh GS
   16. Tang H

   (2013) Genetic architecture of skin and eye color in an African-European admixed population

   PLoS Genetics 9:e1003372.

   https://doi.org/10.1371/journal.pgen.1003372

   • PubMed
   • Google Scholar
6. 1. Bentsink L
   2. Hanson J
   3. Hanhart CJ
   4. Blankestijn-de Vries H
   5. Coltrane C
   6. Keizer P
   7. El-Lithy M
   8. Alonso-Blanco C
   9. de Andrés MT
   10. Reymond M
   11. van Eeuwijk F
   12. Smeekens S
   13. Koornneef M

   (2010) Natural variation for seed dormancy in Arabidopsis is regulated by additive genetic and molecular pathways

   PNAS 107:4264–4269.

   https://doi.org/10.1073/pnas.1000410107

   • PubMed
   • Google Scholar
7. 1. Bentsink L
   2. Jowett J
   3. Hanhart CJ
   4. Koornneef M

   (2006) Cloning of DOG1, a quantitative trait locus controlling seed dormancy in Arabidopsis

   PNAS 103:17042–17047.

   https://doi.org/10.1073/pnas.0607877103

   • PubMed
   • Google Scholar
8. 1. Broman KW
   2. Wu H
   3. Sen S
   4. Churchill GA

   (2003) R/qtl: QTL mapping in experimental crosses

   Bioinformatics 19:889–890.

   https://doi.org/10.1093/bioinformatics/btg112

   • PubMed
   • Google Scholar
9. 1. Chao DY
   2. Silva A
   3. Baxter I
   4. Huang YS
   5. Nordborg M
   6. Danku J
   7. Lahner B
   8. Yakubova E
   9. Salt DE

   (2012) Genome-wide association studies identify heavy metal ATPase3 as the primary determinant of natural variation in leaf cadmium in Arabidopsis thaliana

   PLoS Genetics 8:e1002923.

   https://doi.org/10.1371/journal.pgen.1002923

   • PubMed
   • Google Scholar
10. 1. Chiang GC
    2. Bartsch M
    3. Barua D
    4. Nakabayashi K
    5. Debieu M
    6. Kronholm I
    7. Koornneef M
    8. Soppe WJ
    9. Donohue K
    10. De Meaux J

    (2011) DOG1 expression is predicted by the seed-maturation environment and contributes to geographical variation in germination in Arabidopsis thaliana

    Molecular Ecology 20:3336–3349.

    https://doi.org/10.1111/j.1365-294X.2011.05181.x

    • PubMed
    • Google Scholar
11. 1. Chiang GCK
    2. Barua D
    3. Dittmar E
    4. Kramer EM
    5. de Casas RR
    6. Donohue K

    (2013) Pleiotropy in the wild: the dormancy gene DOG1 exerts cascading control on life cycles.

    Evolution 67:883–893.

    https://doi.org/10.1111/j.1558-5646.2012.01828.x

    • Google Scholar
12. 1. Cyrek M
    2. Fedak H
    3. Ciesielski A
    4. Guo Y
    5. Sliwa A
    6. Brzezniak L
    7. Krzyczmonik K
    8. Pietras Z
    9. Kaczanowski S
    10. Liu F
    11. Swiezewski S

    (2016) Seed dormancy in arabidopsis is controlled by Alternative polyadenylation of DOG1

    Plant Physiology 170:947–955.

    https://doi.org/10.1104/pp.15.01483

    • Google Scholar
13. 1. Debieu M
    2. Tang C
    3. Stich B
    4. Sikosek T
    5. Effgen S
    6. Josephs E
    7. Schmitt J
    8. Nordborg M
    9. Koornneef M
    10. de Meaux J

    (2013) Co-variation between seed dormancy, growth rate and flowering time changes with latitude in Arabidopsis thaliana

    PLoS One 8:e61075.

    https://doi.org/10.1371/journal.pone.0061075

    • PubMed
    • Google Scholar
14. 1. DePristo MA
    2. Banks E
    3. Poplin R
    4. Garimella KV
    5. Maguire JR
    6. Hartl C
    7. Philippakis AA
    8. del Angel G
    9. Rivas MA
    10. Hanna M
    11. McKenna A
    12. Fennell TJ
    13. Kernytsky AM
    14. Sivachenko AY
    15. Cibulskis K
    16. Gabriel SB
    17. Altshuler D
    18. Daly MJ

    (2011) A framework for variation discovery and genotyping using next-generation DNA sequencing data

    Nature Genetics 43:491–498.

    https://doi.org/10.1038/ng.806

    • Google Scholar
15. 1. Dickson SP
    2. Wang K
    3. Krantz I
    4. Hakonarson H
    5. Goldstein DB

    (2010) Rare variants create synthetic genome-wide associations

    PLoS Biology 8:e1000294.

    https://doi.org/10.1371/journal.pbio.1000294

    • PubMed
    • Google Scholar
16. 1. Donohue K
    2. Dorn L
    3. Griffith C
    4. Kim E
    5. Aguilera A
    6. Polisetty CR
    7. schmitt J

    (2005) Environmental and genetic influences on the germination of arabidopsis thallana in the field

    Evolution 59:740–757.

    https://doi.org/10.1111/j.0014-3820.2005.tb01750.x

    • Google Scholar
17. 1. Donohue K
    2. Rubio de Casas R
    3. Burghardt L
    4. Kovach K
    5. Willis CG

    (2010) Germination, postgermination adaptation, and species ecological ranges

    Annual Review of Ecology, Evolution, and Systematics 41:293–319.

    https://doi.org/10.1146/annurev-ecolsys-102209-144715

    • Google Scholar
18. 1. Fedak H
    2. Palusinska M
    3. Krzyczmonik K
    4. Brzezniak L
    5. Yatusevich R
    6. Pietras Z
    7. Kaczanowski S
    8. Swiezewski S

    (2016) Control of seed dormancy in Arabidopsis by a cis-acting noncoding antisense transcript

    PNAS 113:E7846–E7855.

    https://doi.org/10.1073/pnas.1608827113

    • PubMed
    • Google Scholar
19. 1. Finch-Savage WE
    2. Leubner-Metzger G

    (2006) Seed dormancy and the control of germination

    New Phytologist 171:501–523.

    https://doi.org/10.1111/j.1469-8137.2006.01787.x

    • Google Scholar
20. 1. Footitt S
    2. Müller K
    3. Kermode AR
    4. Finch-Savage WE

    (2015) Seed dormancy cycling in arabidopsis: chromatin remodelling and regulation of DOG1 in response to seasonal environmental signals

    The Plant Journal 81:413–425.

    https://doi.org/10.1111/tpj.12735

    • Google Scholar
21. 1. Forsberg SK
    2. Andreatta ME
    3. Huang XY
    4. Danku J
    5. Salt DE
    6. Carlborg Ö

    (2015) The Multi-allelic genetic architecture of a Variance-Heterogeneity locus for molybdenum concentration in leaves acts as a source of unexplained additive genetic variance

    PLOS Genetics 11:e1005648.

    https://doi.org/10.1371/journal.pgen.1005648

    • PubMed
    • Google Scholar
22. 1. Goslee SC
    2. Urban DL

    (2007) The ecodist package for Dissimilarity-based analysis of ecological data

    Journal of Statistical Software 22:1–19.

    https://doi.org/10.18637/jss.v022.i07

    • Google Scholar
23. 1. Hancock AM
    2. Brachi B
    3. Faure N
    4. Horton MW
    5. Jarymowycz LB
    6. Sperone FG
    7. Toomajian C
    8. Roux F
    9. Bergelson J

    (2011) Adaptation to climate across the Arabidopsis thaliana genome

    Science 334:83–86.

    https://doi.org/10.1126/science.1209244

    • PubMed
    • Google Scholar
24. 1. Hijmans RJ
    2. Cameron SE
    3. Parra JL
    4. Jones PG
    5. Jarvis A

    (2005) Very high resolution interpolated climate surfaces for global land areas

    International Journal of Climatology 25:1965–1978.

    https://doi.org/10.1002/joc.1276

    • Google Scholar
25. 1. Horton MW
    2. Hancock AM
    3. Huang YS
    4. Toomajian C
    5. Atwell S
    6. Auton A
    7. Muliyati NW
    8. Platt A
    9. Sperone FG
    10. Vilhjálmsson BJ
    11. Nordborg M
    12. Borevitz JO
    13. Bergelson J

    (2012) Genome-wide patterns of genetic variation in worldwide Arabidopsis thaliana accessions from the RegMap panel

    Nature Genetics 44:212–216.

    https://doi.org/10.1038/ng.1042

    • PubMed
    • Google Scholar
26. 1. Hrabak EM
    2. Chan CW
    3. Gribskov M
    4. Harper JF
    5. Choi JH
    6. Halford N
    7. Kudla J
    8. Luan S
    9. Nimmo HG
    10. Sussman MR
    11. Thomas M
    12. Walker-Simmons K
    13. Zhu JK
    14. Harmon AC

    (2003) The Arabidopsis CDPK-SnRK superfamily of protein kinases

    Plant Physiology 132:666–680.

    https://doi.org/10.1104/pp.102.011999

    • PubMed
    • Google Scholar
27. 1. Huang X
    2. Schmitt J
    3. Dorn L
    4. Griffith C
    5. Effgen S
    6. Takao S
    7. Koornneef M
    8. Donohue K

    (2010a) The earliest stages of adaptation in an experimental plant population: strong selection on QTLS for seed dormancy

    Molecular Ecology 19:1335–1351.

    https://doi.org/10.1111/j.1365-294X.2010.04557.x

    • PubMed
    • Google Scholar
28. 1. Huang X
    2. Wei X
    3. Sang T
    4. Zhao Q
    5. Feng Q
    6. Zhao Y
    7. Li C
    8. Zhu C
    9. Lu T
    10. Zhang Z
    11. Li M
    12. Fan D
    13. Guo Y
    14. Wang A
    15. Wang L
    16. Deng L
    17. Li W
    18. Lu Y
    19. Weng Q
    20. Liu K
    21. Huang T
    22. Zhou T
    23. Jing Y
    24. Li W
    25. Lin Z
    26. Buckler ES
    27. Qian Q
    28. Zhang QF
    29. Li J
    30. Han B

    (2010b) Genome-wide association studies of 14 agronomic traits in rice landraces

    Nature Genetics 42:961–967.

    https://doi.org/10.1038/ng.695

    • PubMed
    • Google Scholar
29. 1. Huo H
    2. Wei S
    3. Bradford KJ

    (2016) DELAY OF GERMINATION1 (DOG1) regulates both seed dormancy and flowering time through microRNA pathways

    PNAS 113:E2199–E2206.

    https://doi.org/10.1073/pnas.1600558113

    • PubMed
    • Google Scholar
30. 1. Kang HM
    2. Sul JH
    3. Service SK
    4. Zaitlen NA
    5. Kong S-yee
    6. Freimer NB
    7. Sabatti C
    8. Eskin E

    (2010) Variance component model to account for sample structure in genome-wide association studies

    Nature Genetics 42:348–354.

    https://doi.org/10.1038/ng.548

    • Google Scholar
31. 1. Kim S
    2. Plagnol V
    3. Hu TT
    4. Toomajian C
    5. Clark RM
    6. Ossowski S
    7. Ecker JR
    8. Weigel D
    9. Nordborg M

    (2007) Recombination and linkage disequilibrium in Arabidopsis thaliana

    Nature Genetics 39:1151–1155.

    https://doi.org/10.1038/ng2115

    • PubMed
    • Google Scholar
32. 1. Kronholm I
    2. Picó FX
    3. Alonso-Blanco C
    4. Goudet J
    5. Meaux J

    (2012) Genetic basis of adaptation in arabidopsis thaliana: Local Adaptation at the Seed dormancy qtl dog1

    Evolution 66:2287–2302.

    https://doi.org/10.1111/j.1558-5646.2012.01590.x

    • Google Scholar
33. Preprint

    1. Li H

    (2013) Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM

    PREPRINT.

    http://arxiv.org/abs/1303.3997v2

    • Google Scholar
34. 1. Li P
    2. Filiault D
    3. Box MS
    4. Kerdaffrec E
    5. van Oosterhout C
    6. Wilczek AM
    7. Schmitt J
    8. McMullan M
    9. Bergelson J
    10. Nordborg M
    11. Dean C

    (2014) Multiple FLC haplotypes defined by independent cis -regulatory variation underpin life history diversity in Arabidopsis thaliana

    Genes & Development 28:1635–1640.

    https://doi.org/10.1101/gad.245993.114

    • Google Scholar
35. 1. Long Q
    2. Rabanal FA
    3. Meng D
    4. Huber CD
    5. Farlow A
    6. Platzer A
    7. Zhang Q
    8. Vilhjálmsson BJ
    9. Korte A
    10. Nizhynska V
    11. Voronin V
    12. Korte P
    13. Sedman L
    14. Mandáková T
    15. Lysak MA
    16. Seren Ü
    17. Hellmann I
    18. Nordborg M

    (2013) Massive genomic variation and strong selection in Arabidopsis thaliana lines from Sweden

    Nature Genetics 45:884–890.

    https://doi.org/10.1038/ng.2678

    • PubMed
    • Google Scholar
36. 1. Mantel N

    (1967)

    The detection of disease clustering and a generalized regression approach

    Cancer Research 27:209–220.

    • PubMed
    • Google Scholar
37. 1. McKenna A
    2. Hanna M
    3. Banks E
    4. Sivachenko A
    5. Cibulskis K
    6. Kernytsky A
    7. Garimella K
    8. Altshuler D
    9. Gabriel S
    10. Daly M
    11. DePristo MA

    (2010) The genome analysis toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data

    Genome Research 20:1297–1303.

    https://doi.org/10.1101/gr.107524.110

    • PubMed
    • Google Scholar
38. 1. Mitchell-Olds T

    (2001) Arabidopsis thaliana and its wild relatives: a model system for ecology and evolution

    Trends in Ecology & Evolution 16:693–700.

    https://doi.org/10.1016/S0169-5347(01)02291-1

    • Google Scholar
39. 1. Nakabayashi K
    2. Bartsch M
    3. Ding J
    4. Soppe WJ

    (2015) Seed dormancy in arabidopsis requires Self-Binding ability of DOG1 protein and the presence of multiple isoforms generated by alternative splicing

    PLoS Genetics 11:e1005737.

    https://doi.org/10.1371/journal.pgen.1005737

    • PubMed
    • Google Scholar
40. 1. Nakabayashi K
    2. Bartsch M
    3. Xiang Y
    4. Miatton E
    5. Pellengahr S
    6. Yano R
    7. Seo M
    8. Soppe WJ

    (2012) The time required for dormancy release in Arabidopsis is determined by DELAY OF GERMINATION1 protein levels in freshly harvested seeds

    The Plant Cell 24:2826–2838.

    https://doi.org/10.1105/tpc.112.100214

    • PubMed
    • Google Scholar
41. 1. Pagnussat GC
    2. Yu HJ
    3. Ngo QA
    4. Rajani S
    5. Mayalagu S
    6. Johnson CS
    7. Capron A
    8. Xie LF
    9. Ye D
    10. Sundaresan V

    (2005) Genetic and molecular identification of genes required for female gametophyte development and function in Arabidopsis

    Development 132:603–614.

    https://doi.org/10.1242/dev.01595

    • PubMed
    • Google Scholar
42. 1. Platt A
    2. Vilhjálmsson BJ
    3. Nordborg M

    (2010) Conditions under which genome-wide association studies will be positively misleading

    Genetics 186:1045–1052.

    https://doi.org/10.1534/genetics.110.121665

    • PubMed
    • Google Scholar
43. 1. Poormohammad Kiani S
    2. Trontin C
    3. Andreatta M
    4. Simon M
    5. Robert T
    6. Salt DE
    7. Loudet O

    (2012) Allelic heterogeneity and trade-off shape natural variation for response to soil micronutrient

    PLoS Genetics 8:e1002814.

    https://doi.org/10.1371/journal.pgen.1002814

    • PubMed
    • Google Scholar
44. 1. Postma FM
    2. Ågren J

    (2015) Maternal environment affects the genetic basis of seed dormancy in Arabidopsis thaliana

    Molecular Ecology 24:785–797.

    https://doi.org/10.1111/mec.13061

    • PubMed
    • Google Scholar
45. 1. Postma FM
    2. Ågren J

    (2016) Early life stages contribute strongly to local adaptation in Arabidopsis thaliana

    PNAS 113:7590–7595.

    https://doi.org/10.1073/pnas.1606303113

    • PubMed
    • Google Scholar
46. 1. Postma FM
    2. Lundemo S
    3. Ågren J

    (2016) Seed dormancy cycling and mortality differ between two locally adapted populations of arabidopsis thaliana

    Annals of Botany 117:249–256.

    https://doi.org/10.1093/aob/mcv171

    • PubMed
    • Google Scholar
47. Book

    1. R Core Team

    (2014)

    R: A Language and Environment for Statistical Computing

    Vienna, Austria: R Foundation for Statistical Computing.

    • Google Scholar
48. 1. Rebeiz M
    2. Pool JE
    3. Kassner VA
    4. Aquadro CF
    5. Carroll SB

    (2009) Stepwise modification of a modular enhancer underlies adaptation in a Drosophila population

    Science 326:1663–1667.

    https://doi.org/10.1126/science.1178357

    • PubMed
    • Google Scholar
49. 1. Rockman MV

    (2012) The qtn program and the alleles that matter for evolution: ALL that'S gold does not glitter

    Evolution 66:1–17.

    https://doi.org/10.1111/j.1558-5646.2011.01486.x

    • Google Scholar
50. 1. Sasaki E
    2. Zhang P
    3. Atwell S
    4. Meng D
    5. Nordborg M

    (2015) "Missing" G x E Variation Controls Flowering Time in Arabidopsis thaliana

    PLoS Genetics 11:e1005597.

    https://doi.org/10.1371/journal.pgen.1005597

    • PubMed
    • Google Scholar
51. 1. Segura V
    2. Vilhjálmsson BJ
    3. Platt A
    4. Korte A
    5. Seren Ü
    6. Long Q
    7. Nordborg M

    (2012) An efficient multi-locus mixed-model approach for genome-wide association studies in structured populations

    Nature Genetics 44:825–830.

    https://doi.org/10.1038/ng.2314

    • PubMed
    • Google Scholar
52. 1. Smouse PE
    2. Long JC
    3. Sokal RR

    (1986) Multiple regression and correlation extensions of the mantel test of matrix correspondence

    Systematic Zoology 35:627–632.

    https://doi.org/10.2307/2413122

    • Google Scholar
53. 1. Springthorpe V
    2. Penfield S

    (2015) Flowering time and seed dormancy control use external coincidence to generate life history strategy

    eLife 4:e05557.

    https://doi.org/10.7554/eLife.05557

    • Google Scholar
54. 1. Swiezewski S
    2. Liu F
    3. Magusin A
    4. Dean C

    (2009) Cold-induced silencing by long antisense transcripts of an Arabidopsis polycomb target

    Nature 462:799–802.

    https://doi.org/10.1038/nature08618

    • PubMed
    • Google Scholar
55. 1. The 1001 Genomes Consortium

    (2016) 1,135 Genomes Reveal the Global Pattern of Polymorphism in Arabidopsis thaliana

    Cell 166:481–491.

    https://doi.org/10.1016/j.cell.2016.05.063

    • PubMed
    • Google Scholar
56. 1. Van der Auwera GA
    2. Carneiro MO
    3. Hartl C
    4. Poplin R
    5. Del Angel G
    6. Levy-Moonshine A
    7. Jordan T
    8. Shakir K
    9. Roazen D
    10. Thibault J
    11. Banks E
    12. Garimella KV
    13. Altshuler D
    14. Gabriel S
    15. DePristo MA

    (2013) From FastQ data to high confidence variant calls: the genome analysis toolkit best practices pipeline

    Current Protocols in Bioinformatics 43:1–33.

    https://doi.org/10.1002/0471250953.bi1110s43

    • PubMed
    • Google Scholar
57. 1. van Zanten M
    2. Koini MA
    3. Geyer R
    4. Liu Y
    5. Brambilla V
    6. Bartels D
    7. Koornneef M
    8. Fransz P
    9. Soppe WJ

    (2011) Seed maturation in Arabidopsis thaliana is characterized by nuclear size reduction and increased chromatin condensation

    PNAS 108:20219–20224.

    https://doi.org/10.1073/pnas.1117726108

    • PubMed
    • Google Scholar
58. 1. Visscher PM
    2. Brown MA
    3. McCarthy MI
    4. Yang J

    (2012) Five years of GWAS discovery

    The American Journal of Human Genetics 90:7–24.

    https://doi.org/10.1016/j.ajhg.2011.11.029

    • Google Scholar
59. 1. Xiang Y
    2. Nakabayashi K
    3. Ding J
    4. He F
    5. Bentsink L
    6. Soppe WJJ

    (2014) REDUCED DORMANCY5 encodes a protein phosphatase 2c that is required for seed dormancy in Arabidopsis

    The Plant Cell Online 26:4362–4375.

    https://doi.org/10.1105/tpc.114.132811

    • Google Scholar
60. 1. Xiang Y
    2. Song B
    3. Née G
    4. Kramer K
    5. Finkemeier I
    6. Soppe WJ

    (2016) Sequence polymorphisms at the Reduced Dormancy5 pseudophosphatase underlie natural variation in arabidopsis dormancy

    Plant Physiology 171:2659–2670.

    https://doi.org/10.1104/pp.16.00525

    • PubMed
    • Google Scholar
61. 1. Yeaman S
    2. Whitlock MC

    (2011) The genetic architecture of adaptation under migration-selection balance

    Evolution 65:1897–1911.

    https://doi.org/10.1111/j.1558-5646.2011.01269.x

    • Google Scholar
62. 1. Zhang Z
    2. Ersoz E
    3. Lai C-Q
    4. Todhunter RJ
    5. Tiwari HK
    6. Gore MA
    7. Bradbury PJ
    8. Yu J
    9. Arnett DK
    10. Ordovas JM
    11. Buckler ES

    (2010) Mixed linear model approach adapted for genome-wide association studies

    Nature Genetics 42:355–360.

    https://doi.org/10.1038/ng.546

    • Google Scholar

Author details

1. Envel Kerdaffrec

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   EK, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   No competing interests declared.

   "This ORCID iD identifies the author of this article:" 0000-0001-8667-6850

2. Danièle L Filiault

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   DLF, Analysis and interpretation of data, Contributed unpublished essential data or reagents

   Competing interests

   No competing interests declared.

   "This ORCID iD identifies the author of this article:" 0000-0002-2938-3071

3. Arthur Korte

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   AK, Analysis and interpretation of data, Contributed unpublished essential data or reagents

   Competing interests

   No competing interests declared.

   "This ORCID iD identifies the author of this article:" 0000-0003-0831-1463

4. Eriko Sasaki

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   ES, Analysis and interpretation of data, Contributed unpublished essential data or reagents

   Competing interests

   No competing interests declared.

5. Viktoria Nizhynska

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   VN, Acquisition of data, Contributed unpublished essential data or reagents

   Competing interests

   No competing interests declared.

6. Ümit Seren

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   ÜS, Analysis and interpretation of data, Contributed unpublished essential data or reagents

   Competing interests

   No competing interests declared.

   "This ORCID iD identifies the author of this article:" 0000-0003-4031-0428

7. Magnus Nordborg

   Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter, Vienna, Austria

   Contribution

   MN, Conception and design, Analysis and interpretation of data, Drafting or revising the article

   For correspondence

   magnus.nordborg@gmi.oeaw.ac.at

   Competing interests

   MN: Reviewing editor, eLife

   "This ORCID iD identifies the author of this article:" 0000-0001-7178-9748

• 4,975

  views

• 792

  downloads

• 58

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.