Molecular markers to study genetic drift and selection in wheat populations

Journal of Experimental Botany, Mar 1999

Studying the heterogeneity in variation of gene frequency among populations or between generations may be a possible way to detect genomic regions experiencing selection. In order to evaluate this approach, RFLP markers were used to compare the allelic frequencies in wheat populations that had been submitted to natural selection. In 1984, samples of two composite cross populations were distributed in the French network for dynamic management of genetic resources. Since then, all the sub-populations have been cultivated in the same sites with no human selection. The strong differentiation between populations found for agro-morphological traits (earliness, resistance to pathogens, …) provided evidence of their adaptation to local conditions. The two initial populations and six derived sub-populations cultivated for 10 years in four contrasted sites were studied with RFLP markers. Differentiation between sub-populations based on RFLP diversity was highly significant. Variations of allelic frequencies of the 30 loci scored were found to be much greater than expected under genetic drift only. This led us to conclude that selection greatly influenced the evolution of the populations. Some of the loci clearly presented a higher differentiation than the others. This might indicate that they were genetically linked to other loci polymorphic in the populations and involved in adaptation. However, the effect of one selected gene on a marker, even located very close to the gene, could not be predicted with certainty. Hence, though the populations were predominantly selfing, it seems that initial linkage disequilibriums between markers and selected genes were not strong enough to control closely the evolution of allelic fequencies at the markers.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

https://jxb.oxfordjournals.org/content/50/332/283.full.pdf

Molecular markers to study genetic drift and selection in wheat populations

S. Paillard 0 P. Brabant 0 0 Station de G en etique Ve ge tale, INRA INA-PG UPS, Ferme du Moulon, F-91190 Gif-sur-Yvette, France INA-PG, 16 rue Claude Bernard, F-75005 Paris, France Studying the heterogeneity in variation of gene frequency among populations or between generations may be a possible way to detect genomic regions experiencing selection. In order to evaluate this approach, RFLP markers were used to compare the allelic frequencies in wheat populations that had been submitted to natural selection. In 1984, samples of two composite cross populations were distributed in the French network for dynamic management of genetic resources. Since then, all the sub-populations have been cultivated in the same sites with no human selection. The strong differentiation between populations found for agro-morphological traits (earliness, resistance to pathogens, ...) provided evidence of their adaptation to local conditions. The two initial populations and six derived sub-populations cultivated for 10 years in four contrasted sites were studied with RFLP markers. Differentiation between sub-populations based on RFLP diversity was highly significant. Variations of allelic frequencies of the 30 loci scored were found to be much greater than expected under genetic drift only. This led us to conclude that selection greatly influenced the evolution of the populations. Some of the loci clearly presented a higher differentiation than the others. This might indicate that they were genetically linked to other loci polymorphic in the populations and involved in adaptation. However, the effect of one selected gene on a marker, even located very close to the gene, could not be predicted with certainty. Hence, though the populations were predominantly selfing, it seems that initial linkage disequilibriums between markers and selected genes were not strong enough to control closely the evolution of allelic fequencies at the markers. Introduction Molecular markers are a useful tool for studying the genetic control of complex traits. For the last decade, they have been largely used to map the loci (QTL) underlying these traits in populations for which they were specifically designed (Stuber et al., 1987; Beavis et al., 1991; Paterson et al., 1991 ). The principle is to create strong linkage disequilibriums (defined as the deviation from random association of the diVerent alleles at diVerent loci) in populations derived from the cross between two inbred lines and then to search for associations between marker loci and complex traits (for details of the methods see Kearsey and Pooni, 1996 ). In these populations, the linkage disequilibrium between two loci is related to the genetic distance on the chromosome between these loci. QTL mapping has allowed the plant geneticists to locate chromosomal segments involved in the genetic control of traits such as disease resistances, yield, morphological traits, and responses to environmental stresses. Hence, when a line, a family or a population is identified for its high level of expression for an interesting complex trait, the understanding of the genetic basis of the trait in this genotype requires the production of a specific recombinant population between the genotype and another one chosen for its diVerent lineage and diVerent level of expression of the trait. Then for QTL mapping, it is necessary to produce a dense genetic map of the population and to evaluate each individual for the trait under consideration. The question of using biochemical or molecular markers for studying the genetics of complex traits in non-structured, multi-parental segregating populations is important, though much more hypothetical. In such populations, linkage disequilibrium between markers and QTL will depend on the mating system, the selection intensity and genetic drift (i.e. the loss of alleles due to the random sampling of a finite number of progenitors at each generation). Molecular markers can be used in a very diVerent way for studying the evolution of such populations. Even if the linkage disequilibrium is not known and is not easily related to physical linkage on the chromosome, the variation of allelic frequencies at the markers can be interpreted as a response to selection or drift. If the population mating system is outcrossing with random mating, the probability is low for a disequilibrium between a marker locus and a QTL to be maintained from one generation to the other. The probability is higher when individuals are preferential selfers. Consistently, in natural populations diVerent authors have found a positive correlation between distance based on markers and distance based on phenotypic quantitative traits between populations of self-pollinated species whereas they found no relation for out-crossing species ( Hamrick and Allard, 1975; Price et al., 1984). An experiment for quantifying changes in allelic frequencies at molecular markers in a population under recurrent selection ( RS ) was conducted at CIMMYT ( Ribaut et al., 1995). They studied cycles 0, 4 and 8 of a full-sib RS experiment for drought resistance in maize where phenotypic selection had significantly reduced anthesis-silking interval (ASI ). Major changes in allelic frequencies were detected for 4 of the 5 loci located near genomic regions which had been previously proved to be responsible for the expression of ASI in an F2 population ( Ribaut et al., 1995). However, the RFLP alleles that increased in the RS population were not systematically those associated with the favourable alleles at QTLs in the F2. These results show that in spite of frequent recombinations, strong and eYcient human phenotypic selection may lead to important hitch-hiking or drift eVects in large genomic regions around the selected QTLs. Hence, strong variations in allelic frequencies at marker loci may implicate the chromosomal regions around the loci in the control of a trait submitted to selection. Specific genetic parameters such as the Wright Fst, a parameter that measures the between populations diVerentiation (for definitions see Hartl and Clark, 1989 ), have been developed to quantify and test the changes in allelic frequencies at marker loci in order to describe genetic diversity in populations under natural selection. From the distribution of allelic frequencies of sampled loci in a population ( Watterson, 1977 ) or between populations (Lewontin and Krakauer, 1973), neutrality tests indicate if the observed polymorphisms correspond to what is expected under the eVects of genetic drift and mutation only (driftmutation equilibrium). As developed by Lewontin and Krakauer (1973 ), markers can be used to detect unexpected shifts of allelic frequencies over time in a population or to reveal non-neutral patterns of diVerentiation between isolated populations. Using hordein loci to study temporal evolution of a barley composite population submitted to 9 years of selection, Ibrahim et al. ( 1996) found directional changes of allelic frequencies at two loci. The authors argued that linkage between some major mildew resistance genes and these hordein loci, together with the selective advantage of resistance to powdery mildew and the inbreeding nature of barley could have led to the observed shifts. On the other hand, few experiments were able to reveal selection using spatial structure of genetic diversity because of the lack of power of the method. In the following, the possible contributions of molecular markers to study genetic drift and selection in a series of wheat populations that have been submitted to natural selection for 10 years in diVerent French locations are considered. The idea is to use the Fst genetic parameter to detect variation in allelic frequencies at marker loci that appear to be outside the expected distribution for neutral loci, and then to connect this with the information available on adaptive traits. Materials and methods The dynamic management programme of wheat populations The aim of a dynamic management (DM ) approach is to maintain or mimic natural processes responsible for the diversification and conservation of genetic variability. The principle is to introduce a polymorphic composite cross population in contrasted environments and to let the subpopulations evolve under the eVects of natural selection, genetic drift and mutation. While each sub-population is supposed to lose some of its initial variability due to drift and selection in a particular environment, diversity is expected to be maintained at the level of all the populations (metapopulation) due to genetic diVerentiation. The founding experiment of dynamic management has been conducted since 1928 on barley populations in the United States (Harlan and Martini, 1929; Allard, 1988, 1990). Few other experiments are to be found in the literature. A pilot programme for the dynamic management of genetic resources of winter wheat (Triticum aestivum L.) has been conducted at INRA in France since 1984 (Henry et al., 1991). Two initial composite populations (PA, PB) were created by crossing 16 parental lines ( French varieties, INRA lines, exotic lines) ( Thomas et al., 1991; David et al., 1997). The mating system in the populations is predominantly selfing ( Enjalbert et al., 1998). The genetic base of PB was wider than that of PA due to more exotic lines among the parents. Dwarfing genes were segregating in the two initial populations. In 1984, after three years of bulk multiplication, seed samples of the two initial populations were distributed into the sites of a multilocation experimental network composed of INRA stations, agricultural universities and schools throughout France. Each genetic pool was not represented in each locality but there were seven locations for PA and nine for PB ( Fig. 1). Since 1984, each of the local populations has been cultivated Fig. 1. Location in France of the PA and PB pools. every year without any conscious human selection using seeds of the same population harvested the previous year in the same location under the same cultural conditions. A 100 m2 area was recommended so as to obtain around 10 00015 000 plants per population. The populations were isolated from each other and from neighbouring crops to avoid pollen pollution. RFLP (Restriction Fragment Length Polymorphism) markers Total DNA was extracted from lyophilized young leaves following a rapid procedure adapted from Dellaporta et al. (1983). Enzyme restriction, electrophoresis, blotting onto Hybond N+ membranes (Amersham) and non-radioactive hybridizations (DIGBBoehringer-Mannheim) were performed as described by Lu et al. (1994). Sixteen RFLP probes (19 enzyme/probe combinations) that had been found polymorphic within the 32 parental lines of the PA and PB pools were used to study the DM populations (see Enjalbert et al., 1998, for more details). Among the 16 probes, 14 provided by P. Leroy (INRA Clermont-Ferrand ) had been obtained and mapped by the INRA-Genoble group (Nelson et al., 1995). These probes were chosen for their high polymorphism information content and for their balanced distribution over the genome. Two probes provided by K Devos, had been obtained and mapped near the dwarfing genes (Rht1 and Rht2) at the John Innes Centre, Norwich (Gale et al., 1997). Due to the hexaploid genome of wheat, the 19 enzyme/probe combinations correspond to 30 loci. Six populations (that will be referred to as final populations) were studied after 10 years of multiplication: three PA populations cultivated at Le Moulon (48 9 N 2 E ), Rennes (48 1 N 2 W ) and Toulouse (43 6 N 1 E ), and three PB populations cultivated at Le Moulon, Rennes and Venours (45 8 N 1 W ) ( Fig. 1). From the bulk harvest of the 1994 season, a random sample of 78 individuals was analysed. To estimate the initial frequencies of markers in the founding populations, two random samples of inbred lines derived by single seed descent (SSD) from the initial populations PA0 and PB0 were also studied. Analysis of genetic diversity The Wright parameter Fst was computed for each locus as follows: HtHs k m k Fst= Ht with, Ht=1j=1p:j2, Hs=i=1wi(1j=1pi2j) m and p:j= wipij, i=1 where wi is the relative number of individuals (weight) of population i, pij is the frequency of allele j in population i, k the number of alleles and m the number of populations. Fst parameter provides an indication of distance between two or more populations according to the diVerences in their allelic frequencies, and Fst varies from 0, when all populations share the same allelic frequencies, to 1, when each population has fixed one allele. Temporal Fst will be reported when one or more final populations are compared to the initial one, and spatial Fst, when final populations are compared together. Fst was estimated for three kinds of population for each pool (PA and PB): (i) temporal Fst between the three final populations pooled together and the initial one, (ii) spatial Fst between the three final populations, (iii) Fst between the populations taken two by two. As a generalization of this approach, multilocus estimations of the Fst parameter were then computed using Genepop software ( Raymond and Rousset, 1995). The Fst for all pairs of populations, including the initial ones, were estimated as well as the spatial and temporal Fst at the whole pool level. Temporal variations of allelic frequencies at the multilocus level were also studied using Ne, the eVective size of a population. For a given population, this parameter corresponds to the equivalent number of individuals of a theoretical panmictic population that would provide the same variation in allelic frequencies due to genetic drift alone. The greater the variation of frequencies, the lower is the eVective population size Ne. Unlike the temporal Fst, this parameter is corrected for variation due to random sampling of the individuals taken for the study (Nei and Tajima, 1981). Results and discussion Evolution of agro-morphological traits Each year, a sample of the seeds harvested from each population was stored in a centralized cold room. Results for agro-morphological traits have been obtained from diVerent previous experiments all using seeds from some of these samples so as to grow and evaluate plants from the diVerent populations in the same location. The first significant changes appeared after only six years of multiplication. The mean plant height of all the populations of PA and PB pools increased compared to the initial situation (David et al., 1992; Le Boulch et al., 1994 ). This increase was interpreted as the result of the competition for light between plants of a local population. Evolution was rapid because the presence of polymorphic dwarfing loci led to a large and simply inherited genetic variability. A northsouth ( latitudinal ) gradient of earliness was found among the PA and PB populations (David et al., 1992). The populations cultivated in the south of France became genetically earlier than those cultivated in the north. This was interpreted as an adaptation to climatic constraints. In the south of France, water and heat stress appear early during plant development so that seed filling of late plants may be hindered, whereas in the north, these stresses seldom happen and late plants may accumulate more dry matter (Le Boulch et al., 1994). Studying the resistance to powdery mildew (Erysiphe graminis f. sp. tritici ), Le Boulch et al. (1994 ) found significant changes in the frequencies of some specific resistance genes in the PA populations after eight years of multiplication. In the final populations, individuals were found that accumulated more resistance genes than any individual analysed in the initial population. Unlike the case of the resistance to Rhynchosporium secale and to Erysiphe graminis f. sp. hordei in the barley populations (Allard, 1990; Ibrahim and Barrett, 1991 ), no significant increase was found for adult resistance when population means were compared with those of the respective initial population for both pools. However, adult resistance was found to diVer according to the multiplication site and to the initial pool. Significant changes were also found for the frequencies of resistance genes to other diseases (stripe rust, eyespot) (data not published ), giving evidence that fungal diseases were an important selective pressure in the network. David (1992) found a significant correlation (r=0.78) between yield of a given PA population and yield of the corresponding PB population cultivated in the same site and under the same cultural condition. This led us to the conclusion that for this complex trait the populations have evolved according to the local conditions. Hence, except for genes poorly adapted to interindividual competition (such as dwarfing genes), the diversity of the environments in the network enabled the overall variability to be maintained. The diVerentiation between populations seems to have been due to adaptation of each population to its environment, but evolution under the eVect of genetic drift can not be ruled out. However, the evolution in the first generations was so fast that drastic changes in the genetic structure of the populations even for non-adaptive polymorphisms can be expected. Molecular markers Taking the three populations of each pool as a whole, it was found that no allele was lost in PA whereas only 1 was lost in PB. Thus it is likely that if all the populations of each pool had been analysed, all the initial alleles would certainly have been found. Hence, the metapopulation is eYcient to maintain the diversity of RFLP loci. Multilocus estimates of the eVective sizes (Ne) of the six populations ranged from 42 to 208 (data not shown). These estimates were much lower than expected under genetic drift alone (>5000) considering the number of plants cultivated for each population (~10 000). Multilocus estimation of the diVerentiation between populations within pools ( Table 1) was also found to be much higher (Fst=0.039 for PA, Fst=0.112 for PB) than predicted in the case of evolution under genetic drift only (Fst= 0.002). Populations of pool PB appeared more divergent than populations of pool PA and no convergence was found between the PA and the PB populations cultivated at the same site, since the Fst between PA Moulon and PB Moulon and the Fst between PA Rennes and PB Rennes are equal to, or greater than the Fst between PA0 and PB0. However in both pools, it was noted that: (i) the Rennes population appeared to have evolved the most rapidly with regards to the initial situation, and (ii) the Rennes and the South populations ( Toulouse or Venours) were always the most distant populations. This may be related to the contrasting environmental (climatic as well as parasitic) conditions at both kinds of sites. Low Ne means important shifts with regards to the initial population and high within pool Fst means diVerentiation according to the multiplication site. This indicated that temporal and spatial variation of the allelic frequencies of some of these markers were much greater than expected under random sampling of gametes alone. This can be explained either by hitch-hiking due to physical linkage with loci submitted to selection or by more indirect eVects since strong selection on some loci leads to a high variance of the reproductive contribution of each individual to the next generation and hence to an increase of genetic drift over the whole genome. Since the observed shifts were not evenly distributed Table 1. Multilocus diVerentiation parameters ( Fst) between final populations (populations of the tenth generation of cultivation) of PA and PB pools and between final populations and initial populations (PA0 and PB0) PA (Fst=0.039)a PB (Fst=0.112)a Moulon Rennes Toulouse PB0 Moulon Rennes Rennes 0.037 Toulouse 0.028 PA0 0.023 Moulon 0.120 Rennes 0.171 Venours 0.176 aGlobal Fst within each pool indicating spatial diVerentiation. Expected value for a population of Ne~5000 individuals: 0.002. among loci, the problem concerning the distribution of the monolocus Fst parameter was further investigated. Temporal evolution was studied by comparing, for each pool, the Fst between the initial population and the pooled three final populations. The distributions among the loci are given in Fig. 2a, b. The distributions among the loci of the spatial Fst between the three final populations within each pool are given in Fig. 2c, d. Whereas the shapes of the distributions were rather similar in PA and PB for temporal Fst, the shape of the PB distribution for spatial Fst was much flatter than that of PA distribution. This latter result was consistent with the higher multilocus spatial Fst found in pool PB ( Table 1). In both pools, most of the loci presented low temporal Fst, indicating a good conservation of the initial frequencies. In both pools, loci with the highest temporal Fst were not highly diVerentiated ( low spatial Fst). Hence, this indicated convergent evolution within each pool. However, as the loci with high temporal Fst were not the same in pools PA and PB, evolution at the genetic level was not convergent across the pools. In both pools, some loci clearly revealed more diVerentiation than the others (high spatial Fst, Fig. 2c, d ). Among the loci studied, no correlation was found between the spatial Fst in PA and the one in PB ( Fig. 3). Only two loci ( XFBA152a, XFBA381) showed a large Fst in both pools. The parallel evolutions observed for these two loci ( Table 2) could be explained by the strong linkage disequilibrium found between them in both initial populations PA0 and PB0 as well as in the parental lines. A previous study had shown that both loci were located on the 6B chromosome near the centromere ( Enjalbert et al., 1999 ). Hence, the information provided by the two loci was considered to be redundant and only one was taken into account. The detailed observation of the evolution of allelic frequencies at these loci revealed diVerent patterns in PA and PB. DiVerentiation in pool PB was mainly due to the divergence of the Rennes population, whereas it was due to the divergence of the Toulouse population in PA ( Table 2). Hence even for these two loci, no evidence of convergent evolution of chromosomal regions was found that would be correlated to adaptation to the diVerent conditions in both pools. This can be explained by the great diVerence in the genetic variability available in each pool: QTL involved in local adaptation may be polymorphic in one pool and not in the other, or the polymorphism at the QTL may be diVerent in both pools (this is a possible explanation for Fig. 2. Distributions of spatial and temporal monolocus Fst in PA and in PB pools. Fig. 3. Relations between monolocus spatial Fst estimated in PA and in PB pools. Table 2. Detail of pairwise Fst for two marker loci (XFBA381 and XFBA152a) showing strong spatial structuration in both PA and PB pools Rennes Toulouse Rennes Venours Rennes Rennes the two loci studied above), or disequilibrium between marker and QTL might be found in one pool and not the other. In order to test more precisely if some chromosomal regions were involved in local adaptation, a temporal approach was focused on the changes in allelic frequencies observed in PA and PB, in Le Moulon and Rennes as the two pools were present in both sites. Whereas in Rennes, the loci with strongest temporal Fst were not the same in PA and PB populations, in Le Moulon four of the six loci with the highest Fst were shared by populations PA and PB. Though the changes were not in the same direction, some chromosomal segments involved in the adaptive response to Le Moulon seemed to be polymorphic in both pools PA and PB. The idea of looking for chromosomal regions bearing QTLs involved in local adaptation using the shifts in allelic frequencies could be validated studying the behaviour of marker loci located near major genes polymorphic in the populations and that were submitted to selection in one or more sites. The first example was the XFBA204a locus located on chromosome 7DL. It is loosely linked to one interspecific insertion of an Aegilops ventricosa chromosome segment, that carries the Pch1 gene conferring resistance to eyespot (Pseudocercosporella herpotrichoides, Doussinault et al., 1983). Previous results revealed that the frequency of the resistant allele of Pch1 had strongly increased in the Rennes PA population leading to a significant temporal Fst in this population and to a significant diVerentiation between the three PA populations (high spatial Fst) ( Table 3 ). Significant diVerentiation was also found for the XFBA204a locus due to a strong increase of allele 2 in Rennes and to a lesser extent in the Le Moulon population ( Table 3). It is proposed that diVerentiation of XFBA204a diversity ( locus with the 5th highest spatial Fst in PA, Fig. 3 ) was due to the hitch-hiking eVect from the Pch1 gene on the marker. The presence of an initial linkage disequilibrium between Pch1 and XFBA204a in the parental lines of PA (data not shown) supported this assumption. As dwarfing alleles were highly counter-selected in all the populations of the network, the locus Rht1 was the second example to be considered. Two of the markers ( XPSR144 and XPSR622) were genetically linked to Rht1 (respectively with 9 and 4 cM ). However, the initial linkage disequilibrium between Rht1 and both marker loci was not known since information for the genotype at Rht1 was missing for many parental lines. Low but significant diVerentiation was found in pool PA for both Frequency of allele R Fst (P-value)a 0.006 (0.1714) 0.069 (0.0030) XFBA204a locus Frequency of allele 2 Fst (P-value)a 0.070 (<0.0001) 0.080 (<0.0001) aUnbiased estimate of the P-value of the probability test of null hypothesis H0: the allelic distribution is independent across populations (i.e. Fst=0) according to Raymond and Rousset (1995). marker loci (spatial Fst, Table 4 ) due to small shifts of allelic frequencies in one or two populations of each pool. Variations of the allelic frequencies at the two markers were more important in pool PB. For XPSR622, one allele strongly increased in Rennes and Venours while remaining stable in Le Moulon leading to a high global temporal Fst and a high spatial Fst ( Table 4). For XPSR144, evolutions were divergent from one population to the other leading to an even higher spatial Fst ( locus with the fifth highest spatial Fst in PB, Fig. 3) and a nonsignificant temporal Fst. These results showed that the evolution of the polymorphism at a marker locus is not easily predictable even if the locus is genetically linked to a polymorphic locus that has undergone a strong directional selection. Table 4. Structuration parameters for two RFLP loci (XPSR144 and XPSR622) located near the dwarfing gene Rht1 in PA and PB pools: temporal Fst between each final population and the initial population (PA0 or PB0), global temporal Fst between the three PA (respectively PB) pooled populations and PA0 (respectively PB0), spatial Fst between the three final populations of each pool Fst XPSR144 (P-value)a 0.014 (0.0540) 0.000 (0.8090) 0.003 (0.3550) 0.001 (0.7870) 0.021 (0.0050) 0.001 (0.7300) 0.070 (0.0009) 0.136 (<0.0001) 0.007 (0.4770) 0.200 (<0.0001) Fst XPSR622 (P-value)a 0.015 (0.067) 0.013 (0.002) 0.004 (0.004) 0.002 (0.167) 0.036 (<0.001) 0.002 (0.4700) 0.020 (0.0410) 0.122 (<0.0001) 0.016 (0.0220) 0.091 (<0.0001) The multilocus genetic parameters estimated in the DM populations gave evidence that the variations of allelic frequencies observed for the RFLP markers can not be explained by evolution under genetic drift only but by direct or indirect eVects of selection. However, it could not be ruled out that the populations that have most drastically evolved (PA and PB Rennes) had experienced a significant uncontrolled reduction of their reproductive size (i.e. a bottleneck) at any of the generations. Distribution of monolocus Fst revealed some loci that clearly presented unexpectedly large variations of their allelic frequencies. The conclusion was that the chromosomal regions around these loci potentially had a strong eVect in the determinism of an adaptive quantitative trait submitted to selection in the population under consideration. However, even though the populations were predominantly self-pollinating, these results showed that the eVect of one selected gene on a marker, even located very close to the gene, could not be predicted with certainty. Evolution of allelic frequencies at a marker is a stochastic event modified by linkage and linkage disequilibrium with the selected gene. Linkage changes only the magnitude and/or the probability of the diVerent possible evolutions at the marker. Yet, the bigger the linkage disequilibrium the more predictable the evolutions. Hence, the use of highly polymorphic markers such as microsatellites would be more powerful for such studies since the larger number of alleles per locus increases the probability of finding strong or total linkage disequilibria between marker loci and selected genes. However, using markers to control selection in such composite cross populations still appears questionable. aUnbiased estimate of the P-value of the probability test of null hypothesis H0: the allelic distribution is independent across populations (i.e. Fst=0) according to Raymond and Rousset (1995). Allard RW. 1988. Genetic changes associated with the evolution of adaptedness in cultivated plants and their wild progenitors. Journal of Heredity 79, 235238.


This is a preview of a remote PDF: https://jxb.oxfordjournals.org/content/50/332/283.full.pdf

J. Enjalbert, I. Goldringer, S. Paillard, P. Brabant. Molecular markers to study genetic drift and selection in wheat populations, Journal of Experimental Botany, 1999, 283-290, DOI: 10.1093/jxb/50.332.283