Pearl millet landraces display an important variation in their cycle duration. This diversity contributes to the stability of crop production in the Sahel despite inter-annual rainfall fluctuation. Conservation of phenological diversity is important for the future of pearl millet improvement and sustainable use. Identification of genes contributing to flowering time variation is therefore relevant. In this study we focused on three flowering candidate genes, PgHd3a, PgDwarf8 and PgPHYC. We tested for signatures of past selective events within polymorphism patterns of these three genes that could have been associated with pearl millet domestication and/or landraces differentiation. In order to implement ad hoc neutrality tests, a plausible demographic history of pearl millet domestication was inferred through Approximate Bayesian Computation by using eight neutral STS loci.
Domesticated pearl millet exhibited 84% of the nucleotide diversity level found in the wild population. No specific polymorphisms were found either in the wild or in the domestic populations. The Bayesian approach and previous studies suggest that gene flow between wild relatives and domesticated pearl millets is a main factor explaining these results. Early and late landraces did not show significant genetic differentiation at both the neutral and the candidate loci. A positive selection was evidenced in PgHd3a and PgDwarf8 genes of domestic forms but not in the wild population.
Our results strongly suggest that PgHd3a and PgDwarf8 were likely targeted by selection during domestication. However, a potential role of any of the three candidate genes in the phenological differentiation between early and late landraces was not supported by our data. Reasons why these results contrast with previous results that have shown a slight but significant association between PgPHYC polymorphisms and variation in flowering time in pearl millet are discussed.
Citation: Lakis G, Navascués M, Rekima S, Simon M, Remigereau M-S, Leveugle M, et al. (2012) Evolution of Neutral and Flowering Genes along Pearl Millet (Pennisetum glaucum) Domestication. PLoS ONE 7(5): e36642. https://doi.org/10.1371/journal.pone.0036642
Editor: Nadia Singh, North Carolina State University, United States of America
Received: December 5, 2011; Accepted: April 4, 2012; Published: May 14, 2012
Copyright: © 2012 Lakis et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by Agence Nationale pour la Recherche Scientifique (ANR-07Biodiv-005-04) - project PLANTADIV, and (ANR-06-BDIV-003-01) - project Transbiodiv. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Plant domestication is often considered as a two step evolutionary process. The first step corresponds to the evolution of characteristic traits of domestic phenotypes under the selective pressures that have occurred under the conditions of cultivation of wild populations by the first farmers, as early as the beginning of the domestication . These genetically modified traits define the so-called “domestication syndrome” which is generally shared by all the components of the domestic gene pool of the same species , .
The second step, concerns the diversification of the domestic gene pool as a byproduct of local adaptations to new environments and to human needs and tastes. This further evolution has produced the morphological and physiological diversity of the domestic gene pool which largely exceeds what is usually observed in their wild counterparts.
Among the traits usually considered as targets of this second step of the domestication process, cycle length has been crucial for domestic populations to cope with new environmental conditions encountered during the expansion of the cultivation areas , .
As for many other cereals, pearl millet domestication has produced many landraces displaying a strong diversity of their cycle length. This diversity is partly due to photoperiod sensitivity variation among landraces. Pearl millet landraces can be classified according to their cycle length and photoperiod sensitivity. The early varieties (45–70 days between sowing and the beginning of flowering), abundant in the northern dry regions of the Sahel , , are mostly facultative short-day plants but photoperiod insensitive genotypes have also been reported , , . On the other hand the semi-late (70–100 days to flowering) and late landraces (100–150 days to flowering), are more abundant in the wetter southern regions. They are considered as strictly sensitive to day length (absolute short-day plants) , , . Although there is an important variation in flowering time within each group of landraces , we have recently shown, in a case study, that the distribution of flowering times among early and late landraces are clearly distinct .
Up to date, the history of pearl millet domestication is poorly known , . Several scenarios have been proposed on the basis of the geographical patterns of molecular marker polymorphism but no consensus has yet been reached (see a brief review in ). In particular, the origin of the diversity in cycle length (i.e. the origin of early and late pearl millet) has been poorly discussed. Tostain et al.  and Tostain and Marchais  suggested that the domestication of pearl millet in West Africa generated essentially early ﬂowering landraces. This step would have been followed at the west side of the actual Lake Chad by a secondary diversification, which would be at the origin of late flowering landraces and new early flowering landraces of East and South Africa and of India.
The diversity in cycle length plays a major role in Sahelian agrosystems where farmers generally grow both early and late (or semi-late) landraces to cope with the uncertainty of the rainy season. This diversity contributes therefore to ensure regularity in grain production . Other reasons are also invoked by farmers to explain their practice of growing these two distinct varietal types, such as the adaptation of early and late landraces to different type of soils and differences in culinary uses. Also, the position of these two varietal types in the cropping calendar and in the management of agricultural lands allows farmers to cope with the seasonal wanderings of transhumant herds (fields grown with early pearl millet are released when transhumant herds arrive from northern regions). The farmers’ knowledge and practices including classification and careful choice of these two varietal types correspond actually to a functional subdivision and a smart biodiversity management tool in the agrosystems. However, it has recently been stated that an evolution to earliness of pearl millet landraces due to adaptation to recurrent droughts would have occurred during the past decades, at least in some parts of the Sahel , . The occurrence of gene flow between early and late landraces has been reported despite differences in mean flowering time . This gene flow is driven by recent changes in farmer’s practices due to ecological and social changes in the Sahel of Africa . Furthermore Lakis et al.  have proposed that, this gene flow and the subsequent introgression could result in a drastic erosion of cycle length diversity in pearl millet. Conservation of this cycle length diversity is of primary importance for the future of pearl millet improvement and sustainable use. The identification of genes contributing in cycle diversity is consequently relevant to this goal.
The genetic and molecular factors underlying the diversity of cycle length in pearl millet are very poorly known since only a few and very recent studies have been devoted to this topic , . However in Arabidopsis thaliana the flowering pathways and gene networks involved in the floral transition are well described. Four pathways have been shown to be responsible for the control of this process. The autonomous pathway responds to internal signals (such as the developmental stage) independently of environmental signals to initiate flowering . The light dependent pathway is responsible for the perception and integration of changes in light quantity and quality . The Gibberellic Acid (GA) pathway promotes flowering through hormonal signals . The fourth pathway is the vernalization pathway, associated with cold periods to initiate the flowering process , which is a characteristic of species living in the temperate zones. A well known feature of the flowering network is that signals from different pathways are integrated to initiate flowering in the shoot apical meristem (SAM) cells by the protein coded by the FLOWERING LOCUS T gene , traveling from the leaves .
It is now well established that grasses and A. thaliana share most of the genes involved in flowering time controlling pathways. For example, potential functional orthologous of the FT Arabidopsis gene have been identified in rice (Hd3a), maize (ZCN8), wheat (VRN-B3/TaFT) and barley (VRN-H3/HvFT) , , . Potential orthologous of CONSTANS (CO), an integrator of the light dependent pathway in Arabidopsis , have also been identified in those species: Hd1 (rice), conz1 (maize), TaHd1-1(wheat), and HvCO1 (barley) , , , . Moreover, it is also now known that grasses have developed original mechanisms involved in the floral transition . For example, an alternative rice specific, light dependent pathway is controlled by a gene with no identified orthologous in Arabidopsis, Ehd1 .
In this study we used a candidate gene approach in order to test whether three orthologous in pearl millet of flowering genes already identified in other species have been targeted by selection during domestication. The three genes were PgHd3a, PgDwarf8 and PgPHYC. They are respectively orthologous of the Arabidopsis FT and the rice Hd3a, of the maize Dwarf8, and of the Arabidopsis and Sorghum PHYC.
These three genes have been reported to be involved in cycle length diversity in crops. In rice, the Hd3a gene was identified as a QTL which control heading time . Like FT, Hd3a integrates signals from the different flowering pathways. It has been shown that Hd3a haplotypic diversity is associated with flowering time variation of rice cultivars . Similarly, orthologous of FT in wheat seem to be involved in the heading date variation observed within a collection of wheat inbred lines from diverse geographical origins . Dwarf8 was one of the firstly discovered flowering time QTLs in maize , . Dwarf8 is the orthologous of the Arabidopsis Gibberellin Insensitive (GAI) gene and of the Rht gene in wheat . Polymorphisms in the Dwarf8 gene have been shown to be associated with flowering time variation observed among maize inbred lines . It was also suggested by Camus-Kulandaivelu et al.  that Dwarf8 may have been involved in the adaptation of maize to new latitudes through diversifying selection on the heading date trait. The PHYC gene encodes for a photoreceptor protein. It was related to flowering time variation and local adaptation among Arabidopsis accessions . PHYC has also been shown to contribute to variation for cycle length in pearl millet. Indeed Saïdou et al.  found a weak but significant association between polymorphisms in the PgPHYC gene and flowering time variation among a collection of several inbred lines originated from India, West and East Africa.
In this study we first cloned the coding sequences of PgHd3a and PgDwarf8. We also cloned a region of the PgPHYC gene which corresponds to the third intron and part of the fourth exon of the maize PHYC gene. Secondly, we evaluated the level of genetic differentiation between wild and domestic pearl millets, and between early and late landraces on the basis of nucleotide polymorphism in a collection of domestic landraces (early and late) and wild accessions for both candidate genes and control loci. These domestic and wild pearl millets originated from the whole geographical distribution area of pearl millet in the Sahel of Africa. Finally, we tested whether nucleotide polymorphism pattern at these three candidate genes is consistent with a recent selection event and a potential role in the differentiation of early and late landraces. It is well known that the effects of selection and demographic history on nucleotide polymorphisms patterns are difficult to disentangle . To circumvent this problem we implemented ad hoc neutrality tests. We used polymorphism data at eight presumably neutral loci to infer the demographic parameters of a plausible history of the pearl millet domestic populations. Approximate Bayesian Computation approach  and coalescent simulations were used to generate the expected distribution of neutrality tests that fitted the observed data better than the strict Wright-Fisher neutrality model.
Molecular Polymorphism Analyses: PCR, Cloning and Sequencing
The primers used for this study are listed in Table S1. Primers used for the candidate gene PgHd3a were designed using the rice Hd3a and the Arabidopsis FT orthologous sequences (respectively [BD169090.1] and [AB027504]). The primers used for the PgDwarf8 gene were designed on the basis of its orthologous sequence in maize [AF413202.1]. Primers of the PgPHYC gene were designed from the PgPHYC fragment isolated by Saïdou et al. .
The three candidate genes and eight STS single copy loci (≈9800 bp), were PCR amplified using the Invitrogen “Platinum Taq DNA Polymerase High Fidelity” on a collection of wild and domestic pearl millet accessions (in average 22 early and 20 late) originated from the whole geographic distribution area of this species in Africa (Table S2). Six accessions from Asia were also included (Table S2). Three of these eight STS loci were previously sequenced on a subset of accessions .
Each locus was amplified and sequenced from one individual per accession. In some cases the STS loci and the candidate genes could not be amplified in a given individual for technical reasons. Therefore, PCR amplification was done on another individual from the same population, or, when this was not possible, on another accession corresponding to a geographically close population (Table S2). The PCR products of the candidate genes and of the STS 738 were cloned and sequenced using the TOPO-TA (Invitrogen). For each of the other STS locus, PCR products from at least 10 individuals were cloned and sequenced in order to obtain the true haplotypes (see below). These latter were used as references to infer the most likely haplotypes from other sequences obtained from direct sequencing of the PCR products. PCR products were sequenced in forward and reverse directions in order to exclude sequencing errors. To test for Taq polymerase errors in the sequence data, PCR amplification and sequencing were carried out two times for 10 individuals and for each of the PgHd3a gene and the STS 713 loci. No errors were found on the 20 re-sequenced fragments (19830 bp). In addition the STS 306 showed only one SNP out of 38248 sequenced bp. If this SNP was a Taq polymerase error, this result would imply an error rate of <3 errors for every 105 bp. This value is congruent with estimations of the error rate (2×10−5 errors per bp) for the High Fidelity (Invitrogen) Taq DNA Polymerase . All sequences were deposited in Genbank under accession numbers [JQ001940-JQ002518].
Isolation of the PgHd3a cDNA
23DB inbred line plants were grown in 16 h day and 8 h night (long day conditions) during one month. Plants were transferred in short day (SD) conditions: 12 h day and night, to induce floral transition. Total RNA was extracted from leaves collected in the morning from plants grown for 2 weeks in SD conditions using Trizol (Invitrogen). The PgHd3a forward primer was used to perform the 3′ Rapid Amplification of cDNA Ends (RACE) following recommendations for the RACE kit (Clonetech). cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen). The amplification products were cloned in TOPO-TA vector (Invitrogen) in order to obtain the sequence of the PgHd3a full transcript.
Statistical Analyses of Molecular Polymorphisms
Base calling, quality assessment, and sequence assembly were conducted using CodonCode Aligner (V.3.5.7). ClustalW implemented in the Bioedit  software was used to perform the multiple sequences alignment of each locus. Accessions that were undetermined for their cycle length (see table S2) were not included in analyses conducted on early and late accessions separately. Polymorphism and molecular evolution analyses were performed using DnaSP v5.00  and Fabsim software .
In order to infer the allelic phase of the STS sequences obtained from direct sequencing of PCR products, the algorithms PHASE v2.1 ,  implemented in DnaSP  was used. One of the two inferred haplotypes was randomly chosen for further statistical analyses.
Assessment of the genetic structure within the collection of accessions.
The Arlequin software (version 3.5)  was used to estimate F-statistics. The significance of each pairwise Fst value was assessed by performing 10000 permutations. The SNP data generated from the concatenated STS fragments was used to evaluate the genetic structure within the collection of accessions by using the Bayesian method implemented in the Structure software version 2.3.3 . 500,000 iterations were carried out for each run after a burn-in period of 100,000 iterations. The model allowing for admixture and correlated allele frequencies between populations was used for this analysis. Five independent replicates for each value of K (the number of a priori clusters) were performed. This procedure was repeated for K varying from 1 to 10. The optimal number of clusters (K) was identified by using the ad hoc statistics based on the second order change of the likelihood function with respect to K (ΔK) developed by Evanno et al. .
Bayesian Inference of the Demographic History of Pearl Millet Domestic Population
A plausible demographic history of domestication was inferred using an approximate Bayesian computation (ABC) analysis  on the basis of summary statistics estimated from the molecular polymorphism data at the eight STS loci. Many simulated data sets were generated from the model with parameter values taken from prior distributions. The posterior probabilities distributions were estimated on the basis of the most likely simulated data. The latter was identified through estimation of distances between summary statistics of the simulated data and the real data set (see  for a review on ABC).
The demographic model simulated in this study consisted of a constant sized wild population from which an exponentially expanding (domestic) population originates at domestication time. Two versions of this model were used. In the first one, gene flow occurs between wild and domestic populations. The second version did not include the occurrence of gene flow. The parameters of this model were: the scaled (i.e. scaled to coalescent units) mutation rate for the wild population (per bp, θW = 4NeWµ); the scaled mutation rate for the domestic population at present (θD1 = 4Ne1µ) and at the domestication time (θD0 = 4Ne0µ); the scaled time of domestication (T = t/4NeW); the scaled recombination rate (for consecutive bp, ρ = 4NeWr. ); the scaled migration rates for gene flow in both directions: migration from wild to domestic (4NemW→D) and migration from domestic to wild (4NemD→W). µ is the mutation rate per generation per bp, t is the time of domestication in generations unit (1 year per generation for annual plants as pearl millet), and r is the recombination rate (for consecutive bp per generation). It must be noticed that eventual variations of the neutral mutation rate and of the recombination rate among loci are neglected in this model.
Uninformative prior probability distributions were used for θW, log-uniform (min = 10−3, max = 1), θD1, log-uniform (min = 10−8, max = 1), ρ, log-uniform (min = 10−50, max = 1); and the migration rate, uniform (min = 0, max = 100). On the basis of archeological records, pearl millet domestication is believed to have been achieved between 3,000 to 4500 years ago , , . However, it is very likely that agrarian societies on the African continent had a Neolithic way of life since at least the eighth millennium BC , . In consequence, a uniform prior was set (min = 3000; max = 12000) for the parameter t (time of domestication in generations unit). The prior for θD0 was set to be conditional to the values of θD1 and θW so that it always took a value lower to both of them. This was done to ensure the simulated domestication scenario includes a demographic bottleneck with respect to the wild population immediately followed by a demographic expansion. The prior distribution for θD0 was log-uniform with maximum set to the lowest value between θD1 and θW and the minimum set to 10−5 times that value. In order to scale domestication time in generations to coalescent time (4NeW generations unit), the mutation rate was also considered. Mutation rates for maize are estimated to be between 3×10−9 and 1.5×10−7 per bp per year (with point estimates around 3×10−8) . Thus a log-normal prior (log-mean = −17.7, log-s.d. = 1) on µ was chosen in order to cover the range of these estimated values and their uncertainty. Simulations (106 for each of the two scenarios and each locus) were performed with ms . Each locus was simulated separately to take into account the differences in sample sizes and in the length of these sequences.
Intra-population and inter-population summary statistics were computed for real and simulated data in order to compute a distance measure between real and simulated data for the ABC analysis. The best fitting model was chosen based on the posterior probabilities for the two demographic scenarios (absence or presence of gene flow between wild and domestic millet). These posterior probabilities were estimated by the logistic regression approach  from the closest 2×105 simulations from both scenarios. For the chosen model, posterior probability density distributions for the parameter values were estimated by the non-linear regression ABC approach  from the closest 2×105 simulations. Point estimates (median) and 95% highest posterior density (HPD) intervals were estimated from these posterior distributions.
Single-population summary statistics used to describe diversity in both the wild and the domestic populations were mean and variance among loci for: expected heterozygosity (He), nucleotide diversity (π), number of exclusive polymorphisms, variance of pairwise differences and Tajima’s D . Because the domestic population undergoes a more complex demographic history, additional summary statistics were computed on that population: mean and variance among loci for the frequency of the most common haplotype, number of singletons and raggedness index . Two-population statistics included the mean and variance across loci for: total expected heterozygosity, minimal pairwise genetic distance between populations, mean pairwise genetic distance between populations, variance of pairwise genetic distance between populations, number of fixed differences and number of exclusive and shared polymorphisms . In addition, three statistics (RS, Wx2s1 and Wx1s2, unpublished data of Navascués et al.), based on the spatial distribution along the nucleotide sequence of fixed differences and exclusive and shared polymorphism between the wild and the domestic population were used. These statistics are sensitive to the presence and direction of gene flow. For these statistics, the mean across loci and the Kolmogorov-Smirnov D statistic of the comparison between the distribution of the normalized statistics among loci and the standard normal distribution were used. The behavior of these statistics in respect to differentiation and gene flow will be published elsewhere.
Sampling parameter values from prior probability distributions, running ms, calculation of summary statistics and estimation of posterior probabilities were performed in R (R Development Core Team 2009); a script is available from M. Navascués on request. Posterior probabilities were estimated with the R package “abc” (http://cran.r-project.org/web/packages/abc/index.html).
Tests for Selective Neutrality of Candidate Genes
Genetic diversity in candidate genes (PgHd3a, PgDwarf8 and PgPHYC) were tested against the neutral hypothesis including the demographic history of domestic populations as inferred from the ABC analysis. Null distributions of summary statistics were estimated by coalescent simulations (104) using parameters values randomly drawn from the joint posterior probability distributions. The number of polymorphic sites (S) in each simulation was fixed to the observed value. Tajima’s D  and Fu & Li F*  statistics and their associated p-values were calculated for each pseudo-sample. Simulations were performed with ms  and summary statistics were calculated from the ms output files with Fabsim .
Nucleotide Diversity and Polymorphism within the STS Fragments
Identification of sequences similar to the Sequenced Tagged Sites in grasses.
A search of Genbank using BLASTn indicated that the STS 344, the STS 359 and the STS 521 have no similarities with other known nucleotides sequences. The STS 870 shared similarity with an mRNA of a hypothetical protein of sorghum (e-values 4e-42) and maize (e-values 2e-40). The STS 306 shared also similarities with a hypothetical sorghum protein (e-value 9e-68). Remigereau et al.  showed that the STS 713 shared high similarity with multiple plant protein kinases from the RLG family in maize, sorghum and rice (e-values 3e-69 to 4e-04) and that the STS 476 shared high similarity with an mRNA of a hypothetical protein in Sorghum (e-value 3e-165). They also revealed that the STS 738 showed similarity with a single rice BAC clone (e-value 6e-07).
Pattern of STS polymorphism within the domestic and wild populations.
In comparison with other outcrossing wild relatives of crops, wild pearl millet showed a lower level of nucleotide diversity. Indeed, the nucleotide diversity (π) of the wild pearl millet was 0.0062 on average whereas it is 0.0095 in teosinte  and 0.012 in H. annuus  who are also outcrossing species. However it is higher than the nucleotide diversity of the highly selfing Glycine soja (0.0022)  and Triticum turgidum ssp. dicoccoides (0.0027) . Interpretation of this result is not straightforward, in particular because the level of neutral genetic diversity is not only influenced by the mating system, but also by the demographic history of populations, which is largely unknown for wild relatives of crops. For example, the various climatic periods during the last 20000 years (the maximal age of the last glacial era) in Africa , including the recent drought episodes, could have strongly modified the distribution and the demography of grasses, among which the wild Pennisetum glaucum, and therefore their genetic diversity.
Our results also showed that the nucleotide diversity of the domesticated pearl millet is relatively high compared to other major crops. Indeed the nucleotide diversity of the domestic samples estimated by π ranges from 0.0021 to 0.0101 (depending on the STS locus considered) and its average value was 0.0054. This last value is very close to the maize nucleotide diversity (0.0063)  and almost identical to the nucleotide diversity of barley (0.0051) , but higher than the nucleotide diversity found in sorghum (0.0022)  and rice (0.0023) . However, these estimations were obtained on non homologous loci. This hinders comparison of the level of nucleotide diversity among species since it is possible that the evolutionary rates of the loci studied in the different species are very different due to differences in the evolutionary constraints suffered by these non homologous loci. This is especially true for pearl millet because it is not known whether the STS loci are coding sequences.
The pearl millet average ratio πdomestic/πwild for the STS loci (Table 1) showed that the domestic populations of pearl millet are 16% less polymorphic in average than the wild populations. This was expected since plant domestication is generally associated with a loss of diversity due to the contribution of only a subset of the wild populations to the domestic gene pool. It is noticeable that no specific sites and haplotypes were found either in the wild or in the domestic populations. As far as we know, the loss of diversity we observed on the domesticated pearl millet is one of the smallest among all studied cereals .
Our data also showed that the polymorphism revealed by the STS loci can be considered as generally neutral in both the wild and in the domestic populations when tested against the standard Wright-Fisher neutral model. The wild and the domestic pearl millet showed a slight excess of rare polymorphism as shown by negative values of Tajima’s D. This result was unchanged when the six Asian accessions were removed (data not shown). The singletons were distributed on a high number of lineages suggesting that the STS loci for both the domestic and wild populations had a star shaped genealogy (Table 1). However, Tajima’s D values were only slightly significant for the STS 713 in the domestic populations and the STS 521 in the wild and domestic populations (Table 1).
Comparisons of nucleotide diversity between early and late landraces on the basis of STS locus.
The early and the late landraces revealed almost the same nucleotide diversity level (Table S3). In fact, the average values of π for the early landraces and for the late landraces were almost identical (0.0055 and 0.0051, respectively). Finally, neutrality tests implemented separately on early and late pearl millet landraces gave similar results in the two groups of plants (Table S3).
Very few studies have compared the genetic diversity of different phenological groups in cereals. Results similar to ours were obtained for maize inbred lines on the basis of 1095 sequenced genes. The early temperate populations and the late tropical populations were found to have almost equal values of nucleotide diversity (0.0065 and 0.0061) .
Genetic differentiation between Wild, Early and Late Pearl Millets and Levels of Genome Admixture
Our results showed that the wild and the domestic populations are significantly genetically differentiated, as suggested by Fst values (Table S4) and by the analysis of the genetic structure within the collection of accessions. The average Fst value between the wild and the domestic populations across all STS loci was 0.12. All the Fst values estimated on each STS locus were significant, except for STS 521 and STS 738. The Bayesian analysis of the population genetic structure showed that the most likely number of groups was K = 2 (Figure 1A). Figure 1B shows that each of the two clusters was composed of both domestic and wild individuals. The wild pearl millets showed high level of genome admixture from both clusters. Genome admixture was also found for the domestic individuals, but in a lesser extent. This pattern may be due to either shared ancestral polymorphisms or migration between the two forms of pearl millet. The difference in genome admixture proportion between wild and domestic individuals suggested that the gene flow may be asymmetrical with preferential introgression of wild genotypes by genes from the domestic populations.
A. Values of ΔK calculated by using the Evanno et al. (2005) method according to the number K of clusters B. Representation of the individual assignment probabilities to each of the 2 inferred clusters; individuals were arranged according to the estimated proportion of admixture in their genome.
The STRUCTURE analysis dedicated to the domestic populations only did not reveal any clear clustering of the individuals according to either their phenological type or their geographic origin (data not shown). In addition, Fst values (Table S4) between early and late accessions were not significant across all loci (average Fst = 0.04) except for STS 306 and STS 344. Altogether, these results confirmed that the differentiation between the two groups of landraces was weak.
Bayesian Inference of the Demographic History of Pearl Millet Domestication
The demographic history of pearl millet domestication is largely unknown. As stated above, several scenarios could explain the pattern of genetic diversity observed within the domestic gene pool and its level of genetic differentiation with the wild Pennisetum glaucum. In order to get a better insight of what could have been a plausible history of pearl millet domestication, an Approximate Bayesian approach was developed using the supposedly neutral STS loci. This has also allowed us to distinguish, in a second step, the effect of demography from the effect of directional selection on the polymorphism pattern within the three flowering candidate genes. Indeed modifications of polymorphisms pattern in the genome of domestic plants relatively to their wild ancestor have been driven by both the demographic history of domestic population, which is known to affect the whole genome, as well as by positive selection for adaptive mutations, which should affect only domestication genes and surrounding genome regions.
The posterior density probability distributions, of the parameters included in the model, were different from the prior distributions with clearly identified modes (Figure 2). This allowed point estimations of these parameters except for rho, the effective migration rates (Figure 2) and t, the time of domestication in generation units (not shown). Indeed, the posterior distributions of these parameters were very similar to their prior distributions. Thus, our data was not informative enough to get reliable estimates of these parameters.
Data shown was obtained for the scenario “with migration”. The prior (black line) is used as a reference for the posterior distribution (blue lines) obtained from the rejection followed by the non linear regression. Parameters T, M and rho are scaled to the wild population effective size New.
The “with migration” scenario was only two times more likely (p = 0.637) than the “without migration” scenario (p = 0.363). Neither of the two scenarios could therefore be excluded. Moreover, the estimations of the different parameters from the two models were very close from each other and their confidence intervals broadly overlapped (Table 2, and Table S5). In addition the current occurrence of gene flow between wild and domestic gene pools has been shown in several studies , , a condition that likely existed since the beginning of domestication. Therefore, only results obtained from the “with migration” scenario are extensively shown and discussed hereafter (Table 2 and Figure 2).
The inferences revealed that the estimation of the neutral mutation rate was 6.5×10−8/bp/year which is about ten times higher than the only estimation available up to date for a nucleotide sequence (the Adh1 gene) in pearl millet .
Our results showed that the posterior distribution for the beginning of domestication “t” in years is very similar to the prior distribution. This is not surprising pertaining to the very narrow time window we have defined for the prior. No direct inference about this parameter can therefore be given. Nevertheless, the posterior distribution of “T” can be used to infer a domestication time in years (one generation per year) by using its median and the medians of the posterior distributions for both the mutation rate (µ) and the wild population mutation rate (θW). We estimated, the time of domestication in 4NeW generations, as equal to 0.0704 and therefore a domestication time equal to about 8900 years. This makes sense relatively to what archeological data has suggested about the beginning of agriculture in Sahel. We would like to point that this method of calculus does not take into account the uncertainty of the three estimators. Thus, the estimation of T expressed in years should be taken with a lot of caution.
The simulation data showed that the bottleneck severity at the domestication time might have been strong, the estimation of the θD0/θW ratio being equal to 1.1×10−2. However this conclusion should be taken with cautious because of the large 95% credibility interval of this parameter (7.51×10−5−0.25). The upper bound of this interval means that the diversity of the founder domestic population at the domestication time could represent up to 25% of the genetic diversity of the wild populations.
We estimated the magnitude of the population expansion after the initial bottleneck by the ratio between θD1 and θD0, the population mutation rates of the current and the initial domestic populations respectively. It was equal to 1.83×103 (with a 95% credibility interval of (12.69−6.56×105)). Again, the large credibility interval on this parameter does not allow a precise estimation of the strength of this expansion. Demographic events undergone by domestic pearl millet populations, such as bottlenecks and expansion, could also explain the high variance of the πdomestic/πwild ratio among the STS locus that was observed (Table 1).
Cloning and Characterization of the FT(Hd3a), PHYC and Dwarf8 Orthologous Loci in Pearl Millet
The identification of the true orthologous of Hd3a and FT in pearl millet was not trivial. Indeed, Hd3a (rice), FT and TFL1 (arabidopsis) and their homologous in other species belong to the PEBP (phosphatidylethanolamine-binding protein) super family, a multi-gene highly conserved protein family found in many Eukaryotes species . In the plant model Arabidopsis, some of the PEBP family members could act in redundancy to induce flowering such as FT and MFT , while other members may act antagonistically, such as FT and TFL1 . Two amino acid residues are thought to be the most critical for distinguishing FT and TFL1 activity in Arabidopsis, the Tyr85/His88 and the Gln140/Asp144 (FT/TFL1) . It has been also shown for Arabidopsis that the Tyr(Y85)/His(H88) determines whether the protein act as an activator or a repressor of flowering .
We first isolated a fragment of 918 bp from PCR amplification showing only one band, using primers designed for Hd3a coding regions. We then sequenced the amplicon and determined the structure of this gene fragment by isolating the cDNA (Figure 3A). We believe that the cloned fragment corresponds to the true orthologous of Hd3a in pearl millet (PgHd3a) for several reasons:
- The high similarity of the PgHd3a putative protein with putative proteins in cereals which have been proved to have the same physiological role in promoting flowering, as FT in Arabidopsis: 91% of protein sequence identity with TaFT/VRN-B3, 90% with HvFT/VRN-H3 and 87% with Hd3a (Figure 3B).
- The putative protein of PgHd3a showed in every accession the two amino acids residues that are thought to be the most critical for the flowering promotion activity of FT in Arabidopsis (Figure 3B).
- The sequence of the putative protein encoded by the pearl millet PgHd3a was in the same monophyletic group as TaFT/VRN-B3, HvFT/VRN-H3, Hd3a and FT (Figure 3C).
A. structure of the PgHd3a gene, E:exon and I: intron; the 918 bp fragment used in the population genetics analysis is shown between the two dotted red lines; the cDNA sequence was used to complete the 3′ region of the original sequence and to obtain the protein sequence. B. Multiple sequence alignment of the proteins of PgHd3a, the orthologous of FT (wheat (TaFT), barley (HvFT) and rice (Hd3a)) and TFL1; in the black boxes: the Tyr/His(88) and the Gln/Asp(145) residues which are essential to distinguish FT and TFL1. C. A neighborhood joining tree of the PEBP protein family in pearl millet (PgHd3a), maize (ZCN), rice (Os, RFT1, Hd3a, Hd3b), wheat (Ta), barley (Hv) and Arabidopsis (FT). The tree was built with the Poisson corrective distance  model. Robustness of nodes was tested on the basis of 1000 Bootstrap iterations.
The PgDwarf8 cloned fragment was 1233 bp. Compared to the maize Dwarf8, the amplified fragment in pearl millet had no intron as the maize gene . It started 360 bp from the ATG and finished 148 bp before the end of the exon partial sequence published by Thornsberry et al. . The putative protein sequence revealed a high degree of similarity with the maize Dwarf8 protein (93% of amino acids identity).
The PgPHYC fragment cloned in this study was 823 bp, only 43 bp shorter than the sequence amplified by Saïdou et al. .
The structure of PgPHYC is unknown. However the sorghum PHYC gene was proven to have 4 exons and 3 introns. The cloned fragment of PgPHYC included the whole of the third intron and the first 123 bp of the 4th exon (89% of nucleotide identity with the 4th exon of Sorghum propinquum).
Polymorphism Analysis of the Candidate Genes
PgHd3a and PgDwarf8 showed significant differentiation between the domestic and the wild populations (Fst = 0.12; and 0.095 respectively) (Table S4) while PgPHYC did not reveal such a differentiation (Table S4). Also, within the domestic population, the loss of genetic diversity relatively to the wild population was particularly higher for PgDwarf8 than for the STS loci (Table 1, Table 3 and Table 4). On the contrary PgPHYC showed a gain of diversity in the domestic population in comparison with the wild population.
No differentiation between early and late landraces was detected at the three candidate genes (Table S4). It is noticeable that for PgDwarf8, the nucleotide diversity (π) of late landraces was twice the π value of early landraces (Table 4). On the contrary, for PgPHYC, the late landraces showed a lower level of nucleotide diversity (π) than the early landraces. No specific haplotypes fixed in either the early or in the late landraces have been identified in PgHd3a, PgDwarf8 PgPHYC. This was not surprising since the number of singletons was very high for all of these genes.
Evidence for the Fingerprint of a Selective Sweep in Two of the Candidate Genes
It is noteworthy that tests for neutrality used in this study are conservative. This is a direct consequence for having simulating each gene using a broad range of values from the posterior distribution rather than the use of the point estimate values .
The results of the neutrality tests showed significant departures from selective neutrality in both PgHd3a and PgDwarf8 in the domestic population but not in the wild population (Table 3 and Table 4). These two genes showed an excess of rare alleles compared to neutrality as witnessed by negative DT and F* values. Because the distribution of DT and F* values under the neutral hypothesis takes into account the inferred demographic history of the domestic population thanks to the modeling approach, we argue this result could be an evidence of a past selective sweep that has concerned these two genes in the domestic but not in the wild population. It was noticeable that PgHd3a showed a deviation from the neutral expectation in each domestic group (late and early landraces) (Table 3) whereas this was not the case for PgDwarf8 since the neutrality tests were significant only in the late landraces (Table 4). This may suggest that PgDwarf8 was targeted by selection only in the late group and not in the early group of landraces. This result could also be due to the lack of power to detect selection in the early landraces.
The polymorphism pattern in the PgPHYC fragment did not show any significant deviation from the neutral expectation. However, it is noticeable that the different subsamples exhibited contrasted results (Table 4). Indeed the early landraces and the wild populations showed negative values for D and F*, indicating therefore an excess of rare polymorphism. On the other hand the late domestic populations showed positive values of D and F* indicating an excess of allele at intermediate frequencies.
The demographic History of Pearl Millet Domestication
Archeological data for pearl millet and other cereals allowed us to define a narrow window for the time of domestication. Indeed the beginning of domestication has been documented for wheat in the Fertile Crescent (10,500 to 9500 yr BP ), for rice in Asia (9000 to 7000 yr BP , ) and for maize in Meso-america (∼9000 yr BP ). Also for pearl millet, Manning et al.  have revealed that pearl millet could have been domesticated before 4500 years BP. These authors have also hypothesized that pearl millet cultivation could have begun 6000 years ago. It is also likely that wild grass harvesting, including Pennisetum species, was intensive 9000 years ago in the Sahel , . These archeological dates were proved to be very useful to establish a calibration point for the estimation of the mutation rate in our model. This is witnessed by the high differences between the prior and the posterior distributions of µ. This estimation of a neutral mutation rate for pearl millet STS sequences will be very useful for future population genetics simulations.
Our molecular data showed that the domestic and wild pearl millet have similar amounts of nucleotide polymorphism. The domestic forms had seemingly lost only a small fraction (16%) of the neutral genetic diversity which is currently found in the wild populations of Pennisetum glaucum. This result is in accordance with those found in previous studies on pearl millet genetic diversity. Indeed, on the basis of microsatellite markers, Oumar et al.  showed that the domestic populations displayed 83% of the genetic diversity found in the wild populations, while it was 57% (estimation from θs values) in maize compared to teosinte . In addition, Tajima’s D estimated on neutral STS loci were skewed towards negative values. Thus, domestic Pearl millet populations lack evidence of a bottleneck signature contrarily to what has been shown in sorghum  and in maize . The demographic scenario inferred from our data was not fully conclusive in estimating the severity of the bottleneck. However, the upper bound of the credibility interval of the strength of the bottleneck estimation is hardly compatible with the level of genetic diversity we observed in the current domestic population relatively to the wild population. The recovery of genetic diversity in the domestic population could be partly explained by two evolutionary processes. First, the population expansion after the initial step of the domestication, could have given the opportunity for new mutations to accumulate in domestic populations. This expansion may contribute to explain the excess of rare alleles we observed on STS loci. However, this factor could have played only a minor role, taking into account the very short time that has elapsed since the beginning of domestication (only a few thousand of generations), and the mutation rate we estimated (6.5×10−8). Second, the migration of wild lineages into the domestic genetic pool may also explain the recovery of a significant amount of the genetic diversity after the initial bottleneck. The existence of gene flow between wild and domestic forms has been suggested by different studies ,  and was supported by our simulation results. This gene flow may also explain why no fixed differences at the nucleotide level were found in either the wild or the domestic populations.
Genetic diversity within and among early and late landraces at neutral loci.
This study revealed that the early and late landraces did not show significant genetic differentiation at the STS loci. The same level of genetic diversity was observed in both the early and the late landraces. These results could challenge the hypothesis of Tostain et al.  according to which the late landraces have evolved from a secondary diversification of early landraces at the west of the actual Lake Chad region. Other hypotheses on the origin of late landraces can be proposed. For example, they could have been the result of several independent local selection processes from early landraces all along the geographic distribution area of pearl millet. Yet, to our opinion, no strong arguments can actually be put forward against the fact that late landraces could have preceded early landraces. Indeed wild pearl millets are early flowering plants but nonetheless strongly sensitive to photoperiod as most wild plants in tropical areas. This characteristic ensures the coincidence between the rainy season and the life cycle of wild plants. Early landraces of pearl millet are also early flowering but much less photosensitive while late landraces have been shown to be late flowering and highly sensitive to day length , . It is therefore difficult to single out a hypothesis about the evolutionary trajectories of both the cycle duration and the sensitivity to photoperiod, which have occurred within the cultivated gene pool since the beginning of pearl millet domestication.
The existence of a gene flow between early and late landraces could be another explanation for the absence of differentiation between these two phenological groups. Indeed gene flow between early and late pearl millets has been shown to occur in some areas in Niger by Lakis et al., , and was also suggested by other authors , . Lakis et al.,  have shown that this gene flow was driven by changes in farmers’ practices in response to social and environmental modifications. It is however not yet known whether this gene flow occurs all along their common geographic distribution area. Similarly, whether this gene flow is only a recent phenomenon driven by current changes or it has occurred for a long time in the past is also still to be elucidated.
We are conscious that the domestication scenario we simulated may be much simpler than the real domestication history of pearl millet. For example, multiple domestication events or a single domestication followed by a separation of early and late landraces may have occurred. In this study, we took into account the most common and simple hypothesis of a single domestication for pearl millet which has been proposed by several authors , , . Yet, the evolutionary history of the wild population in our model is also likely oversimplified since we have hypothesized a constant wild population size since the domestication time. Major demographic events associated with climatic changes in the Sahelian region since the last millennium may have shaped the polymorphism pattern in the wild pearl millet populations. Thus, much more data and more knowledge on the history of the wild population are needed in order to simulate more complex scenarios of pearl millet domestication and to enhance the accuracy of parameter estimates. Nevertheless, the basic intent of our simulation work was to provide a more adequate model than the Wright-Fisher strict neutral model to test for selective sweeps in candidate genes.
The Selection Regime for the Candidate Genes
Among the three candidate genes, two of them PgHd3a and PgDwarf8 showed polymorphisms patterns that are compatible with a past event of directional selection in the domestic populations but not in the wild population. However we cannot exclude the possibility that the excess of rare alleles in comparison to the neutral expectation could be explained by the presence of a hidden genetic structure and unequal sampling in the domestic populations. Our analyses could have failed in detecting such a structure because of the moderate number of loci used in this study.
Another possibility is that recent gene flow from the wild population to the domestic genetic pool could have also affected the polymorphism frequency spectrum, by increasing the number of rare alleles, and consequently, the significant negative values for DT, and F* neutrality tests. In this case, clusters of singletons, corresponding to rare introgressed fragments, are expected to be found in a limited number of lineages. However, our data showed that singletons at both STS loci and candidate genes are distributed nearly equally among lineages (Table 1, Table 3 and Table 4). Thence, our data are in accordance with the occurrence of regular gene flow between wild and domestic populations since the beginning of domestication rather than rare introgression events.
An original output of our data is that it suggests that PgHd3a could have been the target of selection during the early phase of domestication before the phenological differentiation between early and late landraces. Indeed, PgHd3a showed a strong signal of a past selective sweep in both the early and the late landraces but not in the wild populations. Our results do not support a role of this gene, or at least of differences in the coding sequence, in the phenological differentiation between early and late pearl millets. This result confirms the results from Saïdou et al.  who have shown there was no correlation between polymorphisms in the coding sequence of PgHd3a and the variation in flowering time within a collection of pearl millet landraces from India, West and East Africa. However, in contrast to our study, they did not reject neutrality at this gene. The difference between the two studies could be a consequence of the very low level of nucleotide diversity  found at this gene (S = 7; θ = 0.16×10−3) in comparison with our data (S = 47; θ = 11.2×10−3), resulting in a lack of power to detect selection. However, the low polymorphism found by Saïdou et al.  could also itself be the consequence of a selective sweep. It is however noticeable that the level of nucleotide polymorphism was clearly lower for all the genes studied by Saïdou et al.  (0.16×10−3<θ<2.07×10−3) than for the three candidate genes in our study. Thus, this may also point at differences in sampling representativeness of the early and late pearl millet gene pools between the two studies.
Why could PgHd3a have been the target of selection in the whole domestic population but not in the wild? This gene plays a major role in the floral transition . It has a well known function of integrating the different signals which promotes flowering  and several studies have demonstrated its implication in flowering time variation. For example, in wheat, Bonnin et al.  have revealed that orthologous of Hd3a could be involved in the variation of heading date within a collection of wheat inbred lines originated from various geographical origins. Also, in rice, Takahashi et al.  have shown that both mRNA expression level of Hd3a and molecular polymorphisms within its promoter region, are significantly associated with flowering time variation in a collection of cultivars. But more interestingly, these last authors have also shown that interaction effects between this gene and other flowering genes, Ehd1 and Hd1, contributed largely to flowering time variation among rice cultivars. As stated above, the wild and the domestic pearl millet, display a different phenological habit. It is therefore possible that PgHd3a was involved in the phenological divergence between wild and domestic populations through selection that drove change in its function. Such a change could have been a prerequisite for phenological differentiation to occur later on within the domestic population through interaction between PgHd3a and other flowering genes. Further functional analyses of PgHd3a and studies of molecular polymorphisms in its promoter and in other genes within the same flowering gene network could provide more insights as to why PgHd3a was the target of selection in the domestic pearl millet as a whole.
In contrast with PgHd3a, PgDwarf8 showed a significant signal of a past selective sweep only in the late landraces although this gene also showed negative DT and F* values in the early landraces. This result could be due to a lack of power to detect selection in the early population. It could also suggest that PgDwarf8 has been the target of human selection during a secondary step of the domesticated pearl millet evolution which could have been more specific to late landraces. Yet, it is noticeable that no specific alleles or haplotypes at PgDwarf8 have been found in the late varietal group in comparison to early landraces and even to the wild pearl millets. Introgression of late landraces by alleles from the early domestic population could explain this last result. It is also possible that cis-regulatory sequences rather than the coding region were directly targeted by the selective event. Recombination between early and late haplotypes would therefore explain the absence of specific haplotypes in the coding region.
In maize, molecular polymorphisms in Dwarf8 have already been shown to be associated with flowering time variation and to display the fingerprint of a selective sweep within a collection of 92 inbred lines . It is noteworthy that Thornsberry et al.  included mainly “late” flowering populations of tropical and semi-tropical inbred lines in their study. The association of Dwarf8 polymorphisms with flowering time variation was also confirmed in the early flowering elite European maize inbred lines . These authors have also found a significant association between Dwarf8 polymorphisms and plant height. This finding has led the authors to reconsider the role of Dwarf8 in maize architecture. Indeed, mutants in orthologous of Dwarf8, in wheat (Rht1) , in barley (Sln1) , and in rice (Slr1)  have been identified as being able to decrease the responsiveness to Gibberellic Acid and accordingly to decrease plant height. It is therefore not possible to exclude the possibility that PgDwarf8 may have been under selection within the domestic pearl millet gene pool because of its role in plant architecture rather than in flowering time variation.
Finally, the polymorphism within the studied region of PgPHYC (third intron) did not depart from strict neutrality in both the wild and the domestic populations. This suggests this gene was not involved in phenological evolution and diversification of pearl millet. However, in an experiment carried out by using a collection of 89 pearl millet lines, Saïdou et al.  have shown that molecular polymorphisms within the third intron of PgPHYC, was associated with variation in flowering time. They identified five haplotypes, among which three were found in very low frequency. The two other haplotypes have been found in intermediate frequencies leading to a deviation of the frequency spectrum from strict neutrality (Tajima’s D = 2,38 ; p<0,05). The authors suggested these two haplotypes could have been maintained by balancing selection targeting this gene through selection on cycle length. The pearl millet sample used in our study was much more genetically diverse at this gene. Indeed, 11 haplotypes were detected within the domestic populations among which the two major haplotypes found by Saïdou et al. . These two haplotypes were also the most frequent in our sample of domestic landraces. As stated above, the difference between results obtained in these two studies could be due to differences in geographical representativeness of the two samples. Thus, it is possible, that PgPHYC has been the target of balancing selection only in some areas in Africa, or in some specific landraces. It was noticeable that, in the study of Saïdou et al. , the variation in flowering time was relatively low (overall mean = 58.8 days to female flowering (SE ± 0.54)). This range of variation is poorly representative of differences between early and late varieties according to farmers’ classification. Also, these two types of pearl millet are well known to differ in photoperiod sensitivity while differences among early landraces could rather mainly correspond to differences in intrinsic earliness . Thence, differences between early and late landraces may not likely rely on differences of PHYC activity though this gene could contribute to variation in intrinsic earliness. Actually, the wider geographical area and the larger phenological variation covered by our sample could have hindered the detection of a selective sweep. Our data may also have revealed a weak fingerprint as a result of a composite sample including landraces that could have experimented different selection events impacting cycle evolution (i.e. different targeted genes or differences in time and/or intensity of the selective sweeps). Finally, in both studies, only the third intron of PgPHYC was studied. It is therefore not possible to exclude that other regions of this gene were targeted by human selection during the different steps of domestication.
List of the primers used in this study for PCR amplification and sequencing. Legend: * Primers designed for this study. The same primers were used for PCR amplification and sequencing except internal primers for STS 738.
List of accessions sequenced for the eight STS loci and the three candidate genes.
Genetic diversity and tests for neutrality for the 8 STS loci. aS: number of segregating sites. bSin: number of singletons. cSpecific sites: sites polymorphic in one population but monomorphic in the other. dθ: Watterson estimator of sequence diversity per site, π: average number of differences per site between two sequences. eNumber of lineages with at least one singleton. fD values were calculated using Fabsim, P-values of D were calculated for the standard coalescent model (*: p<0.05; ns: not significant). P-values were computed without correction for multiple tests.
Fst values between wild and domestic populations and between early and late landraces. Fst values were estimated for each of the STS loci and the candidate genes. aP-values were computed without correction for multiple tests. (***:p<0.001; **: p<0.01; *: p<0.05; ns: not significant).
Estimates of demographic and genetic population parameters obtained from the demographic model “without migration.” aθW is the population mutation rate of the wild population, θD0 is the population mutation rate of the domestic population at domestication time, θD1 is the population mutation rate of the domestic population at present time, T is the time of domestication in units of 4NeW generations, and µ is the mutation rate per bp per generation. bHPD interval is the interval of parameter values with the highest posterior density.
We thank the ICRISAT (Niamey) and the INRAN (Niger) for providing seed samples of some of the accessions used in this study.
Conceived and designed the experiments: GL TR. Performed the experiments: GL SR NT MSR MS. Analyzed the data: GL MN. Wrote the paper: GL MN TR. Performed bioinformatics analysis: ML FL. Critically revised the manuscript: FD.
- 1. Hillman GC, Davies MS (1990) Domestication rates in wild type wheats and barley under primitive cultivation. Biological Journal of the Linnean Society 39: 39–78.GC HillmanMS Davies1990Domestication rates in wild type wheats and barley under primitive cultivation.Biological Journal of the Linnean Society393978
- 2. Harlan JR (1971) Agricultural origins: centers and noncenters. Science 174: 468.JR Harlan1971Agricultural origins: centers and noncenters.Science174468
- 3. Hammer K (1984) Das domestikationssyndrom. Genetic Resources and Crop Evolution 32: 11–34.K. Hammer1984Das domestikationssyndrom.Genetic Resources and Crop Evolution321134
- 4. Jones H, Leigh FJ, Mackay I, Bower MA, Smith LMJ, et al. (2008) Population-based resequencing reveals that the flowering time adaptation of cultivated barley originated east of the Fertile Crescent. Molecular biology and evolution 25: 2211–2219.H. JonesFJ LeighI. MackayMA BowerLMJ Smith2008Population-based resequencing reveals that the flowering time adaptation of cultivated barley originated east of the Fertile Crescent.Molecular biology and evolution2522112219
- 5. Cockram J, Jones H, Leigh FJ, O’Sullivan D, Powell W, et al. (2007) Control of flowering time in temperate cereals: genes, domestication, and sustainable productivity. Journal of Experimental Botany 58: 1231–1244.J. CockramH. JonesFJ LeighD. O’SullivanW. Powell2007Control of flowering time in temperate cereals: genes, domestication, and sustainable productivity.Journal of Experimental Botany5812311244
- 6. Niangado O (2001) The state of millet diversity and its use in West Africa. In: Cooper HD, Spillane C, Hodgkin T, editors. Broadening the genetic base of crop production. pp. 147–157.O. Niangado2001The state of millet diversity and its use in West Africa.HD CooperC. SpillaneT. HodgkinBroadening the genetic base of crop production147157
- 7. Mariac C, Jehin L, SaÏDou AA, Thuillet AC, Couderc M, et al. (2011) Genetic basis of pearl millet adaptation along an environmental gradient investigated by a combination of genome scan and association mapping. Molecular Ecology 20: 80–91.C. MariacL. JehinAA SaÏDouAC ThuilletM. Couderc2011Genetic basis of pearl millet adaptation along an environmental gradient investigated by a combination of genome scan and association mapping.Molecular Ecology208091
- 8. Belliard J, Pernès J (1985) Pennisetum typhoides. CRC Handbook of Flowering 4: 22–37.J. BelliardJ. Pernès1985Pennisetum typhoides.CRC Handbook of Flowering42237
- 9. Haussmann BIG, Boureima SS, Kassari IA, Moumouni KH, Boubacar A (2007) Mechanisms of adaptation to climate variability in West African pearl millet landraces – a preliminary assessment. Journal of SAT Agricultural Research 3: BIG HaussmannSS BoureimaIA KassariKH MoumouniA. Boubacar2007Mechanisms of adaptation to climate variability in West African pearl millet landraces – a preliminary assessment.Journal of SAT Agricultural Research3
- 10. Clerget B, Haussmann BIG, Boureima SS, Weltzien E (2007) Surprising flowering response to photoperiod: Preliminary characterization of West and Central African pearl millet germplasm. Journal of SAT Agricultural Research 5: B. ClergetBIG HaussmannSS BoureimaE. Weltzien2007Surprising flowering response to photoperiod: Preliminary characterization of West and Central African pearl millet germplasm.Journal of SAT Agricultural Research5
- 11. Lakis G, Ousmane AM, Sanoussi D, Habibou A, Badamassi M, et al. (2012) Evolutionary dynamics of cycle length in pearl millet: the role of farmer’s practices and gene flow. Genetica. G. LakisAM OusmaneD. SanoussiA. HabibouM. Badamassi2012Evolutionary dynamics of cycle length in pearl millet: the role of farmer’s practices and gene flow.Genetica
- 12. Oumar I, Mariac C, Pham JL, Vigouroux Y (2008) Phylogeny and origin of pearl millet (Pennisetum glaucum [L.] R. Br) as revealed by microsatellite loci. TAG Theoretical and Applied Genetics 117: 489–497.I. OumarC. MariacJL PhamY. Vigouroux2008Phylogeny and origin of pearl millet (Pennisetum glaucum [L.] R. Br) as revealed by microsatellite loci.TAG Theoretical and Applied Genetics117489497
- 13. Robert T, Khalfallah N, Martel E, Lamy F, Poncet V, et al. (2011) Pennisetum. Wild Crop Relatives: Genomic and Breeding Resources. pp. 217–255.T. RobertN. KhalfallahE. MartelF. LamyV. Poncet2011Pennisetum.Wild Crop RelativesGenomic and Breeding Resources217255
- 14. Tostain S, Riandey MF, Marchais L (1987) Enzyme diversity in pearl millet (Pennisetum glaucum). TAG Theoretical and Applied Genetics 74: 188–193.S. TostainMF RiandeyL. Marchais1987Enzyme diversity in pearl millet (Pennisetum glaucum).TAG Theoretical and Applied Genetics74188193
- 15. Tostain S, Marchais L (1989) Enzyme diversity in pearl millet (Pennisetum glaucum). Africa and India Theor Appl Genet 77: 634–640.S. TostainL. Marchais1989Enzyme diversity in pearl millet (Pennisetum glaucum).Africa and India Theor Appl Genet77634640
- 16. Sivakumar MVK (1992) Climate change and implications for agriculture in Niger. Climatic change 20: 297–312.MVK Sivakumar1992Climate change and implications for agriculture in Niger.Climatic change20297312
- 17. Vigouroux Y, Mariac C, De Mita S, Pham J-L, Gérard B, et al. (2011) Selection for Earlier Flowering Crop Associated with Climatic Variations in the Sahel. PLoS ONE 6: e19563.Y. VigourouxC. MariacS. De MitaJ-L PhamB. Gérard2011Selection for Earlier Flowering Crop Associated with Climatic Variations in the Sahel.PLoS ONE6e19563
- 18. Luxereau A, Roussel B (1997) Changements écologiques et sociaux au Niger. L’Harmattan. Paris, France. A. LuxereauB. Roussel1997Changements écologiques et sociaux au Niger.L’Harmattan. Paris, France
- 19. Saïdou AA, Mariac C, Luong V, Pham JL, Bezançon G, et al. (2009) Association studies identify natural variation at PHYC linked to flowering time and morphological variation in pearl millet. Genetics 182: 899.AA SaïdouC. MariacV. LuongJL PhamG. Bezançon2009Association studies identify natural variation at PHYC linked to flowering time and morphological variation in pearl millet.Genetics182899
- 20. Putterill J, Laurie R, Macknight R (2004) It’s time to flower: the genetic control of flowering time. Bioessays 26: 363–373.J. PutterillR. LaurieR. Macknight2004It’s time to flower: the genetic control of flowering time.Bioessays26363373
- 21. Schepens I, Duek P, Fankhauser C (2004) Phytochrome-mediated light signalling in Arabidopsis. Current opinion in plant biology 7: 564–569.I. SchepensP. DuekC. Fankhauser2004Phytochrome-mediated light signalling in Arabidopsis.Current opinion in plant biology7564569
- 22. Blázquez MA, Green R, Nilsson O, Sussman MR, Weigel D (1998) Gibberellins promote flowering of Arabidopsis by activating the LEAFY promoter. The Plant Cell Online 10: 791.MA BlázquezR. GreenO. NilssonMR SussmanD. Weigel1998Gibberellins promote flowering of Arabidopsis by activating the LEAFY promoter.The Plant Cell Online10791
- 23. Henderson IR, Dean C (2004) Control of Arabidopsis flowering: the chill before the bloom. Development 131: 3829.IR HendersonC. Dean2004Control of Arabidopsis flowering: the chill before the bloom.Development1313829
- 24. Valverde F, Mouradov A, Soppe W, Ravenscroft D, Samach A, et al. (2004) Photoreceptor regulation of CONSTANS protein in photoperiodic flowering. Science 303: 1003.F. ValverdeA. MouradovW. SoppeD. RavenscroftA. Samach2004Photoreceptor regulation of CONSTANS protein in photoperiodic flowering.Science3031003
- 25. Turck F, Fornara F, Coupland G (2008) Regulation and identity of florigen: FLOWERING LOCUS T moves center stage. Annu Rev Plant Biol 59: 573–594.F. TurckF. FornaraG. Coupland2008Regulation and identity of florigen: FLOWERING LOCUS T moves center stage.Annu Rev Plant Biol59573594
- 26. Kojima S, Takahashi Y, Kobayashi Y, Monna L, Sasaki T, et al. (2002) Hd3a, a rice ortholog of the Arabidopsis FT gene, promotes transition to flowering downstream of Hd1 under short-day conditions. Plant and Cell Physiology 43: 1096.S. KojimaY. TakahashiY. KobayashiL. MonnaT. Sasaki2002Hd3a, a rice ortholog of the Arabidopsis FT gene, promotes transition to flowering downstream of Hd1 under short-day conditions.Plant and Cell Physiology431096
- 27. Lazakis CM, Coneva V, Colasanti J (2011) ZCN8 encodes a potential orthologue of Arabidopsis FT florigen that integrates both endogenous and photoperiod flowering signals in maize. Journal of Experimental Botany. CM LazakisV. ConevaJ. Colasanti2011ZCN8 encodes a potential orthologue of Arabidopsis FT florigen that integrates both endogenous and photoperiod flowering signals in maize.Journal of Experimental Botany
- 28. Yan L, Fu D, Li C, Blechl A, Tranquilli G, et al. (2006) The wheat and barley vernalization gene VRN3 is an orthologue of FT. Proceedings of the National Academy of Sciences 103: 19581.L. YanD. FuC. LiA. BlechlG. Tranquilli2006The wheat and barley vernalization gene VRN3 is an orthologue of FT.Proceedings of the National Academy of Sciences10319581
- 29. Suárez-López P, Wheatley K, Robson F, Onouchi H, Valverde F, et al. (2001) CONSTANS mediates between the circadian clock and the control of flowering in Arabidopsis. Nature 410: 1116–1120.P. Suárez-LópezK. WheatleyF. RobsonH. OnouchiF. Valverde2001CONSTANS mediates between the circadian clock and the control of flowering in Arabidopsis.Nature41011161120
- 30. Yano M, Katayose Y, Ashikari M, Yamanouchi U, Monna L, et al. (2000) Hd1, a major photoperiod sensitivity quantitative trait locus in rice, is closely related to the Arabidopsis flowering time gene CONSTANS. The Plant Cell Online 12: 2473.M. YanoY. KatayoseM. AshikariU. YamanouchiL. Monna2000Hd1, a major photoperiod sensitivity quantitative trait locus in rice, is closely related to the Arabidopsis flowering time gene CONSTANS.The Plant Cell Online122473
- 31. Miller TA, Muslin EH, Dorweiler JE (2008) A maize CONSTANS-like gene, conz1, exhibits distinct diurnal expression patterns in varied photoperiods. Planta 227: 1377–1388.TA MillerEH MuslinJE Dorweiler2008A maize CONSTANS-like gene, conz1, exhibits distinct diurnal expression patterns in varied photoperiods.Planta22713771388
- 32. Nemoto Y, Kisaka M, Fuse T, Yano M, Ogihara Y (2003) Characterization and functional analysis of three wheat genes with homology to the CONSTANS flowering time gene in transgenic rice. The Plant Journal 36: 82–93.Y. NemotoM. KisakaT. FuseM. YanoY. Ogihara2003Characterization and functional analysis of three wheat genes with homology to the CONSTANS flowering time gene in transgenic rice.The Plant Journal368293
- 33. Griffiths S, Dunford RP, Coupland G, Laurie DA (2003) The evolution of CONSTANS-like gene families in barley, rice, and Arabidopsis. Plant Physiology 131: 1855.S. GriffithsRP DunfordG. CouplandDA Laurie2003The evolution of CONSTANS-like gene families in barley, rice, and Arabidopsis.Plant Physiology1311855
- 34. Colasanti J, Coneva V (2009) Mechanisms of floral induction in grasses: something borrowed, something new. Plant Physiology 149: 56.J. ColasantiV. Coneva2009Mechanisms of floral induction in grasses: something borrowed, something new.Plant Physiology14956
- 35. Doi K, Izawa T, Fuse T, Yamanouchi U, Kubo T, et al. (2004) Ehd1, a B-type response regulator in rice, confers short-day promotion of flowering and controls FT-like gene expression independently of Hd1. Genes & Development 18: 926.K. DoiT. IzawaT. FuseU. YamanouchiT. Kubo2004Ehd1, a B-type response regulator in rice, confers short-day promotion of flowering and controls FT-like gene expression independently of Hd1.Genes & Development18926
- 36. Monna L, Lin H, Kojima S, Sasaki T, Yano M (2002) Genetic dissection of a genomic region for a quantitative trait locus, Hd3, into two loci, Hd3a and Hd3b, controlling heading date in rice. TAG Theoretical and Applied Genetics 104: 772–778.L. MonnaH. LinS. KojimaT. SasakiM. Yano2002Genetic dissection of a genomic region for a quantitative trait locus, Hd3, into two loci, Hd3a and Hd3b, controlling heading date in rice.TAG Theoretical and Applied Genetics104772778
- 37. Takahashi Y, Teshima KM, Yokoi S, Innan H, Shimamoto K (2009) Variations in Hd1 proteins, Hd3a promoters, and Ehd1 expression levels contribute to diversity of flowering time in cultivated rice. Proceedings of the National Academy of Sciences 106: 4555.Y. TakahashiKM TeshimaS. YokoiH. InnanK. Shimamoto2009Variations in Hd1 proteins, Hd3a promoters, and Ehd1 expression levels contribute to diversity of flowering time in cultivated rice.Proceedings of the National Academy of Sciences1064555
- 38. Bonnin I, Rousset M, Madur D, Sourdille P, Dupuits C, et al. (2008) FT genome A and D polymorphisms are associated with the variation of earliness components in hexaploid wheat. TAG Theoretical and Applied Genetics 116: 383–394.I. BonninM. RoussetD. MadurP. SourdilleC. Dupuits2008FT genome A and D polymorphisms are associated with the variation of earliness components in hexaploid wheat.TAG Theoretical and Applied Genetics116383394
- 39. Thornsberry JM, Goodman MM, Doebley J, Kresovich S, Nielsen D, et al. (2001) Dwarf8 polymorphisms associate with variation in flowering time. Nat Genet 28: 286–289.JM ThornsberryMM GoodmanJ. DoebleyS. KresovichD. Nielsen2001Dwarf8 polymorphisms associate with variation in flowering time.Nat Genet28286289
- 40. Koester RP, Sisco PH, Stuber CW (1993) Identification of quantitative trait loci controlling days to flowering and plant height in two near isogenic lines of maize. Crop science 33: 1209–1216.RP KoesterPH SiscoCW Stuber1993Identification of quantitative trait loci controlling days to flowering and plant height in two near isogenic lines of maize.Crop science3312091216
- 41. Peng J, Richards DE, Hartley NM, Murphy GP, Devos KM, et al. (1999) ‘Green revolution’genes encode mutant gibberellin response modulators. Nature 400: 256–261.J. PengDE RichardsNM HartleyGP MurphyKM Devos1999‘Green revolution’genes encode mutant gibberellin response modulators.Nature400256261
- 42. Camus-Kulandaivelu L, Veyrieras JB, Madur D, Combes V, Fourmann M, et al. (2006) Maize adaptation to temperate climate: relationship between population structure and polymorphism in the Dwarf8 gene. Genetics 172: 2449.L. Camus-KulandaiveluJB VeyrierasD. MadurV. CombesM. Fourmann2006Maize adaptation to temperate climate: relationship between population structure and polymorphism in the Dwarf8 gene.Genetics1722449
- 43. Balasubramanian S, Sureshkumar S, Agrawal M, Michael TP, Wessinger C, et al. (2006) The PHYTOCHROME C photoreceptor gene mediates natural variation in flowering and growth responses of Arabidopsis thaliana. Nature genetics 38: 711.S. BalasubramanianS. SureshkumarM. AgrawalTP MichaelC. Wessinger2006The PHYTOCHROME C photoreceptor gene mediates natural variation in flowering and growth responses of Arabidopsis thaliana.Nature genetics38711
- 44. Charlesworth B, Charlesworth D, Barton NH (2003) The effects of genetic and geographic structure on neutral variation. Annual Review of Ecology, Evolution, and Systematics. pp. 99–125.B. CharlesworthD. CharlesworthNH Barton2003The effects of genetic and geographic structure on neutral variation.Annual Review of Ecology, Evolution, and Systematics99125
- 45. Csilléry K, Blum MGB, Gaggiotti OE, Fran ois O (2010) Approximate Bayesian computation (ABC) in practice. Trends in ecology & evolution 25: 410–418.K. CsilléryMGB BlumOE GaggiottiO. Fran ois2010Approximate Bayesian computation (ABC) in practice.Trends in ecology & evolution25410418
- 46. Remigereau MS, Lakis G, Rekima S, Leveugle M, Fontaine MC, et al. (2011) Cereal Domestication and Evolution of Branching: Evidence for Soft Selection in the Tb1 Orthologue of Pearl Millet (Pennisetum glaucum [L.] R. Br.). PloS one 6: e22404.MS RemigereauG. LakisS. RekimaM. LeveugleMC Fontaine2011Cereal Domestication and Evolution of Branching: Evidence for Soft Selection in the Tb1 Orthologue of Pearl Millet (Pennisetum glaucum [L.] R. Br.).PloS one6e22404
- 47. Belanger AE, Lai A, Brackman MA, LeBlanc DJ (2002) PCR-based ordered genomic libraries: a new approach to drug target identification for Streptococcus pneumoniae. Antimicrobial agents and chemotherapy 46: 2507.AE BelangerA. LaiMA BrackmanDJ LeBlanc2002PCR-based ordered genomic libraries: a new approach to drug target identification for Streptococcus pneumoniae.Antimicrobial agents and chemotherapy462507
- 48. Hall TA (1999) BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. pp. 95–98.TA Hall1999BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT9598
- 49. Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25: 1451.P. LibradoJ. Rozas2009DnaSP v5: a software for comprehensive analysis of DNA polymorphism data.Bioinformatics251451
- 50. Ramírez-Soriano A, Calafell F (2008) FABSIM: a software for generating FST distributions with various ascertainment biases. Bioinformatics 24: 2790.A. Ramírez-SorianoF. Calafell2008FABSIM: a software for generating FST distributions with various ascertainment biases.Bioinformatics242790
- 51. Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for haplotype reconstruction from population data. The American Journal of Human Genetics 68: 978–989.M. StephensNJ SmithP. Donnelly2001A new statistical method for haplotype reconstruction from population data.The American Journal of Human Genetics68978989
- 52. Stephens M, Donnelly P (2003) A comparison of bayesian methods for haplotype reconstruction from population genotype data. The American Journal of Human Genetics 73: 1162–1169.M. StephensP. Donnelly2003A comparison of bayesian methods for haplotype reconstruction from population genotype data.The American Journal of Human Genetics7311621169
- 53. Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10: 564–567.L. ExcoffierHEL Lischer2010Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows.Molecular Ecology Resources10564567
- 54. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155: 945.JK PritchardM. StephensP. Donnelly2000Inference of population structure using multilocus genotype data.Genetics155945
- 55. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14: 2611–2620.G. EvannoS. RegnautJ. Goudet2005Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study.Molecular Ecology1426112620
- 56. Amblard S, Pernès J (1989) The identification of cultivated pearl millet (Pennisetum) amongst plant impressions on pottery from Oued Chebbi (Dhar Oualata, Mauritania). African Archaeological Review 7: 117–126.S. AmblardJ. Pernès1989The identification of cultivated pearl millet (Pennisetum) amongst plant impressions on pottery from Oued Chebbi (Dhar Oualata, Mauritania).African Archaeological Review7117126
- 57. Balter M (2007) Seeking agriculture’s ancient roots. Science 316: 1830.M. Balter2007Seeking agriculture’s ancient roots.Science3161830
- 58. Manning K, Pelling R, Higham T, Schwenniger JL, Fuller DQ (2010) 4500-Year old domesticated pearl millet (Pennisetum glaucum) from the Tilemsi Valley, Mali: new insights into an alternative cereal domestication pathway. Journal of Archaeological Science 38: 312–322.K. ManningR. PellingT. HighamJL SchwennigerDQ Fuller20104500-Year old domesticated pearl millet (Pennisetum glaucum) from the Tilemsi Valley, Mali: new insights into an alternative cereal domestication pathway.Journal of Archaeological Science38312322
- 59. Wendorf F, Close AE, Schild R, Wasylikowa K, Housley RA, et al. (1992) Saharan exploitation of plants 8,000 years BP. Nature 359: 721–724.F. WendorfAE CloseR. SchildK. WasylikowaRA Housley1992Saharan exploitation of plants 8,000 years BP.Nature359721724
- 60. Wendorf F, Schild R (1998) Nabta Playa and Its Role in Northeastern African Prehistory. journal of anthropological archaeology 17: 97–123.F. WendorfR. Schild1998Nabta Playa and Its Role in Northeastern African Prehistory.journal of anthropological archaeology1797123
- 61. Clark RM, Tavaré S, Doebley J (2005) Estimating a nucleotide substitution rate for maize from polymorphism at a major domestication locus. Molecular biology and evolution 22: 2304.RM ClarkS. TavaréJ. Doebley2005Estimating a nucleotide substitution rate for maize from polymorphism at a major domestication locus.Molecular biology and evolution222304
- 62. Hudson RR (2002) Generating samples under a Wright–Fisher neutral model of genetic variation. Bioinformatics 18: 337.RR Hudson2002Generating samples under a Wright–Fisher neutral model of genetic variation.Bioinformatics18337
- 63. Beaumont MA (2008) Joint determination of topology, divergence time, and immigration in population trees. Simulations, Genetics and Human Prehistory. pp. 135–154.MA Beaumont2008Joint determination of topology, divergence time, and immigration in population trees.Simulations, Genetics and Human Prehistory135154
- 64. Blum MGB, François O (2010) Non-linear regression models for Approximate Bayesian Computation. Statistics and Computing 20: 63–73.MGB BlumO. François2010Non-linear regression models for Approximate Bayesian Computation.Statistics and Computing206373
- 65. Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123: 585.F. Tajima1989Statistical method for testing the neutral mutation hypothesis by DNA polymorphism.Genetics123585
- 66. Harpending HC (1994) Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution. Human biology; an international record of research 66: 591.HC Harpending1994Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution.Human biology; an international record of research66591
- 67. Wakeley J, Hey J (1997) Estimating ancestral population parameters. Genetics 145: 847.J. WakeleyJ. Hey1997Estimating ancestral population parameters.Genetics145847
- 68. Fu YX, Li WH (1993) Statistical tests of neutrality of mutations. Genetics 133: 693.YX FuWH Li1993Statistical tests of neutrality of mutations.Genetics133693
- 69. Wright SI, Bi IV, Schroeder SG, Yamasaki M, Doebley JF, et al. (2005) The effects of artificial selection on the maize genome. Science 308: 1310.SI WrightIV BiSG SchroederM. YamasakiJF Doebley2005The effects of artificial selection on the maize genome.Science3081310
- 70. Liu A, Burke JM (2006) Patterns of nucleotide diversity in wild and cultivated sunflower. Genetics 173: 321.A. LiuJM Burke2006Patterns of nucleotide diversity in wild and cultivated sunflower.Genetics173321
- 71. Hyten DL, Song Q, Zhu Y, Choi IY, Nelson RL, et al. (2006) Impacts of genetic bottlenecks on soybean genome diversity. Proceedings of the National Academy of Sciences 103: 16666.DL HytenQ. SongY. ZhuIY ChoiRL Nelson2006Impacts of genetic bottlenecks on soybean genome diversity.Proceedings of the National Academy of Sciences10316666
- 72. Haudry A, Cenci A, Ravel C, Bataillon T, Brunel D, et al. (2007) Grinding up wheat: a massive loss of nucleotide diversity since domestication. Molecular biology and evolution 24: 1506.A. HaudryA. CenciC. RavelT. BataillonD. Brunel2007Grinding up wheat: a massive loss of nucleotide diversity since domestication.Molecular biology and evolution241506
- 73. Adams JM, Faure H (1997) Preliminary vegetation maps of the world since the last glacial maximum: an aid to archaeological understanding. Journal of Archaeological Science 24: 623–647.JM AdamsH. Faure1997Preliminary vegetation maps of the world since the last glacial maximum: an aid to archaeological understanding.Journal of Archaeological Science24623647
- 74. Caldwell KS, Russell J, Langridge P, Powell W (2006) Extreme population-dependent linkage disequilibrium detected in an inbreeding plant species, Hordeum vulgare. Genetics 172: 557.KS CaldwellJ. RussellP. LangridgeW. Powell2006Extreme population-dependent linkage disequilibrium detected in an inbreeding plant species, Hordeum vulgare.Genetics172557
- 75. Hamblin MT, Casa AM, Sun H, Murray SC, Paterson AH, et al. (2006) Challenges of detecting directional selection after a bottleneck: lessons from Sorghum bicolor. Genetics 173: 953–964.MT HamblinAM CasaH. SunSC MurrayAH Paterson2006Challenges of detecting directional selection after a bottleneck: lessons from Sorghum bicolor.Genetics173953964
- 76. Caicedo AL, Williamson SH, Hernandez RD, Boyko A, Fledel-Alon A, et al. (2007) Genome-wide patterns of nucleotide polymorphism in domesticated rice. PLoS genetics 3: e163.AL CaicedoSH WilliamsonRD HernandezA. BoykoA. Fledel-Alon2007Genome-wide patterns of nucleotide polymorphism in domesticated rice.PLoS genetics3e163
- 77. Glémin S, Bataillon T (2009) A comparative view of the evolution of grasses under domestication. New phytologist 183: 273–290.S. GléminT. Bataillon2009A comparative view of the evolution of grasses under domestication.New phytologist183273290
- 78. Yamasaki M, Tenaillon MI, Vroh Bi I, Schroeder SG, Sanchez-Villeda H, et al. (2005) A large-scale screen for artificial selection in maize identifies candidate agronomic loci for domestication and crop improvement. The Plant Cell Online 17: 2859.M. YamasakiMI TenaillonI. Vroh BiSG SchroederH. Sanchez-Villeda2005A large-scale screen for artificial selection in maize identifies candidate agronomic loci for domestication and crop improvement.The Plant Cell Online172859
- 79. Pernes J (1986) Outbreeding and domestication process in cereals: the case of maize (Zea mays L. and pearl millet Pennisetum americanum L. K. Schum). Actualités botaniques 133: 27–34.J. Pernes1986Outbreeding and domestication process in cereals: the case of maize (Zea mays L. and pearl millet Pennisetum americanum L. K. Schum).Actualités botaniques1332734
- 80. Mariac C, Robert T, Allinne C, Remigereau MS, Luxereau A, et al. (2006) Genetic diversity and gene flow among pearl millet crop/weed complex: a case study. TAG Theoretical and Applied Genetics 113: 1003–1014.C. MariacT. RobertC. AllinneMS RemigereauA. Luxereau2006Genetic diversity and gene flow among pearl millet crop/weed complex: a case study.TAG Theoretical and Applied Genetics11310031014
- 81. Gaut BS, Clegg M (1993) Nucleotide polymorphism in the Adh1 locus of pearl millet (Pennisetum glaucum)(Poaceae). Genetics 135: 1091.BS GautM. Clegg1993Nucleotide polymorphism in the Adh1 locus of pearl millet (Pennisetum glaucum)(Poaceae).Genetics1351091
- 82. Danilevskaya ON, Meng X, Hou Z, Ananiev EV, Simmons CR (2008) A genomic and expression compendium of the expanded PEBP gene family from maize. Plant physiology 146: 250–264.ON DanilevskayaX. MengZ. HouEV AnanievCR Simmons2008A genomic and expression compendium of the expanded PEBP gene family from maize.Plant physiology146250264
- 83. Yoo SY, Kardailsky I, Lee JS, Weigel D, Ahn JH (2004) Acceleration of flowering by overexpression of MFT (MOTHER OF FT AND TFL1). Molecules and cells 17: 95.SY YooI. KardailskyJS LeeD. WeigelJH Ahn2004Acceleration of flowering by overexpression of MFT (MOTHER OF FT AND TFL1).Molecules and cells1795
- 84. Ahn JH, Miller D, Winter VJ, Banfield MJ, Lee JH, et al. (2006) A divergent external loop confers antagonistic activity on floral regulators FT and TFL1. The EMBO Journal 25: 605–614.JH AhnD. MillerVJ WinterMJ BanfieldJH Lee2006A divergent external loop confers antagonistic activity on floral regulators FT and TFL1.The EMBO Journal25605614
- 85. Hanzawa Y, Money T, Bradley D (2005) A single amino acid converts a repressor to an activator of flowering. Proceedings of the National Academy of Sciences of the United States of America 102: 7748.Y. HanzawaT. MoneyD. Bradley2005A single amino acid converts a repressor to an activator of flowering.Proceedings of the National Academy of Sciences of the United States of America1027748
- 86. Ross-Ibarra J, Wright SI, Foxe JP, Kawabe A, DeRose-Wilson L, et al. (2008) Patterns of polymorphism and demographic history in natural populations of Arabidopsis lyrata. PLoS One 3: e2411.J. Ross-IbarraSI WrightJP FoxeA. KawabeL. DeRose-Wilson2008Patterns of polymorphism and demographic history in natural populations of Arabidopsis lyrata.PLoS One3e2411
- 87. Tanno K, Willcox G (2006) How fast was wild wheat domesticated? Science 311: 1886.K. TannoG. Willcox2006How fast was wild wheat domesticated?Science3111886
- 88. Liu L, Lee GA, Jiang L, Zhang J (2007) Evidence for the early beginning (c. 9000 cal. BP) of rice domestication in China: a response. The Holocene 17: 1059.L. LiuGA LeeL. JiangJ. Zhang2007Evidence for the early beginning (c. 9000 cal. BP) of rice domestication in China: a response.The Holocene171059
- 89. Purugganan MD, Fuller DQ (2009) The nature of selection during plant domestication. Nature 457: 843–848.MD PuruggananDQ Fuller2009The nature of selection during plant domestication.Nature457843848
- 90. Piperno DR, Ranere AJ, Holst I, Iriarte J, Dickau R (2009) Starch grain and phytolith evidence for early ninth millennium BP maize from the Central Balsas River Valley, Mexico. Proceedings of the National Academy of Sciences 106: 5019.DR PipernoAJ RanereI. HolstJ. IriarteR. Dickau2009Starch grain and phytolith evidence for early ninth millennium BP maize from the Central Balsas River Valley, Mexico.Proceedings of the National Academy of Sciences1065019
- 91. Berthaud J, Clément JC, Emperaire L, Louette D, Pinton F, et al. (2001) The role of local-level gene ow in enhancing and maintaining genetic diversity. In: Cooper HD, Spillane C, Hodgkin T, editors. Broadening the genetic base of crop production. 81 p.J. BerthaudJC ClémentL. EmperaireD. LouetteF. Pinton2001The role of local-level gene ow in enhancing and maintaining genetic diversity.HD CooperC. SpillaneT. HodgkinBroadening the genetic base of crop production81
- 92. Tostain S (1990) Rapport de mission ICRISAT/ORSTOM Rapport multigraphié, ORSTOM, Paris. S. Tostain1990Rapport de mission ICRISAT/ORSTOM Rapport multigraphié, ORSTOM, Paris
- 93. Andersen JR, Schrag T, Melchinger AE, Zein I, Lübberstedt T (2005) Validation of Dwarf8 polymorphisms associated with flowering time in elite European inbred lines of maize (Zea mays L.). TAG Theoretical and Applied Genetics 111: 206–217.JR AndersenT. SchragAE MelchingerI. ZeinT. Lübberstedt2005Validation of Dwarf8 polymorphisms associated with flowering time in elite European inbred lines of maize (Zea mays L.).TAG Theoretical and Applied Genetics111206217
- 94. Chandler PM, Marion-Poll A, Ellis M, Gubler F (2002) Mutants at the Slender1 locus of barley cv Himalaya. Molecular and physiological characterization. Plant physiology 129: 181.PM ChandlerA. Marion-PollM. EllisF. Gubler2002Mutants at the Slender1 locus of barley cv Himalaya. Molecular and physiological characterization.Plant physiology129181
- 95. Ikeda A, Ueguchi-Tanaka M, Sonoda Y, Kitano H, Koshioka M, et al. (2001) slender Rice, a Constitutive Gibberellin Response Mutant, Is Caused by a Null Mutation of the SLR1 Gene, an Ortholog of the Height-Regulating Gene GAI/RGA/RHT/D8. The Plant Cell Online 13: 999.A. IkedaM. Ueguchi-TanakaY. SonodaH. KitanoM. Koshioka2001slender Rice, a Constitutive Gibberellin Response Mutant, Is Caused by a Null Mutation of the SLR1 Gene, an Ortholog of the Height-Regulating Gene GAI/RGA/RHT/D8.The Plant Cell Online13999
- 96. Nei M, Kumar S (2000) Molecular evolution and phylogenetics: Oxford University Press, USA. M. NeiS. Kumar2000Molecular evolution and phylogenetics: Oxford University Press, USA