Two nonoverlapping autosomal inversions defined unusual neo-sex chromosomes in the Hessian fly (Mayetiola destructor). Like other neo-sex chromosomes, these were normally heterozygous, present only in one sex, and suppressed recombination around a sex-determining master switch. Their unusual properties originated from the anomalous Hessian fly sex determination system in which postzygotic chromosome elimination is used to establish the sex-determining karyotypes. This system permitted the evolution of a master switch (Chromosome maintenance, Cm) that acts maternally. All of the offspring of females that carry Cm-associated neo-sex chromosomes attain a female-determining somatic karyotype and develop as females. Thus, the chromosomes act as maternal effect neo-W's, or W-prime (W′) chromosomes, where ZW′ females mate with ZZ males to engender female-producing (ZW′) and male-producing (ZZ) females in equal numbers. Genetic mapping and physical mapping identified the inversions. Their distribution was determined in nine populations. Experimental matings established the association of the inversions with Cm and measured their recombination suppression. The inversions are the functional equivalent of the sciarid X-prime chromosomes. We speculate that W′ chromosomes exist in a variety of species that produce unisexual broods.
SEX chromosomes are usually classified as X, Y, Z, or W on the basis of their pattern of segregation and the gender of the heterogametic sex (Ohno 1967). However, when chromosome-based sex determination occurs postzygotically, the same nomenclature confounds important distinctions and may hide interesting evolutionary phenomena. The Hessian fly (Mayetiola destructor), a gall midge (Diptera: Cecidomyiidae) and an important insect pest of wheat, presents an excellent example (Stuart and Hatchett 1988, 1991). In this insect, all of the female gametes and all of the male gametes have the same number of X chromosomes (Figure 1A); no heterogametic sex exists. Nevertheless, Hessian fly sex determination is chromosome based; postzygotic chromosome elimination produces different X chromosome to autosome ratios in somatic cells (male A1A2X1X2/A1A2OO and female A1A2X1X2/A1A2X1X2, where A1 and A2 are the autosomes, X1 and X2 are the X chromosomes, and the paternally derived chromosomes follow the slash) (Stuart and Hatchett 1991; Marin and Baker 1998). Thus, Hessian fly “X” chromosomes are defined by their haploid condition in males, rather than by their segregation in the gametes.
An autosomal, dominant, genetic factor called Chromosome maintenance (Cm) complicates Hessian fly sex determination further (Stuart and Hatchett 1991). Cm has a maternal effect that acts upstream of X chromosome elimination during embryogenesis (Figure 1A). It prevents X chromosome elimination so that all of the offspring of Cm-bearing mothers obtain a female-determining karyotype. Cm-bearing females produce only female offspring and are therefore thelygenic. The absence of Cm usually has the opposite effect; all of the offspring of most Cm-lacking females obtain a male-determining karyotype. These Cm-lacking females produce only male offspring and are therefore arrhenogenic. Like a sex-determining master switch, Cm is usually heterozygous and present in only one sex (Figure 1B). Thus, thelygenic females (Cm/−) are “heterogametic,” as their Cm-containing gametes and Cm-lacking gametes produce thelygenic (Cm/−) and arrhenogenic (−/−) females in a 1:1 ratio. Collectively, thelygenic and arrhenogenic females are called monogenic because they produce unisexual families. However, some Hessian fly females produce broods of both sexes and are called amphigenic. No mating barrier between monogenic and amphigenic families exists (Figure 1C), but amphigenic females have always been found in lower abundance (Painter 1930; Gallun et al. 1961; Stuart and Hatchett 1991). In experimental matings, the inheritance of maternal phenotype was consistent with the segregation of three Cm alleles (Figure 1C): a dominant thelygenic allele, a hypomorphic amphigenic allele, and a null arrhenogenic allele (Stuart and Hatchett 1991).
Here we report the genetic and physical mapping of Cm on Hessian fly autosome 1 (A1). Two nonoverlapping inversions were identified that segregated perfectly with Cm. The most distal inversion was present in all thelygenic females examined. The more proximal inversion extended recombination suppression. These observations suggested that successive inversions evolved to suppress recombination around Cm after it arose. The inversions therefore appear to have evolved in response to the forces that shaped vertebrate Y and W chromosomes (Charlesworth 1996; Graves and Shetty 2001; Rice and Chippindale 2001; Carvalho and Clark 2005). We therefore believe the inversion-bearing chromosomes may be classified as maternal effect neo-W's.
MATERIALS AND METHODS
Hessian fly populations:
Hessian fly strains were maintained and experimental matings were performed as previously described (Behura et al. 2004). Because Cm's effect is maternal, the Cm genotype of each female is manifested as the phenotype (sex) of her offspring. Therefore, individual females were genotyped as thelygenic (Cm/−, arrhenogenic (−/−), or amphigenic by sexing their offspring. Individual females were first permitted to mate and deposit their eggs on “Blueboy” wheat seedlings in separate caged pots. In genetic mapping experiments, females were collected separately after oviposition and their DNA was extracted for molecular genotyping. The offspring of these females were allowed to develop and were sexed as adults. When examining the offspring of females for the presence of inversions, chromosome preparations were made of at least 10 offspring of each female. These offspring were sexed cytologically. The remaining offspring were allowed to develop in the caged pots and were sexed as adults.
DNA was extracted from individuals using the DNeasy Blood and Tissue kit (QIAGEN, Valencia, CA) according to the manufacturer's recommendations. AFLP–polymerase chain reaction (PCR) fingerprinting of individual Hessian flies was performed as previously described (Behura et al. 2004), using the AFLP-based Genome Typing Kit for small plant genomes (Invitrogen, Carlsbad, CA). The AFLP band that cosegregated with the thelygenic Hessian fly genotype was amplified using the selective primers EcoRI-AC and MseI-CTG. To determine the DNA sequences of that AFLP band, it was excised from an acrylamide gel and cloned into the pCRII TOPO cloning vector (Invitrogen). Seven plasmids containing the band were then sequenced using a BigDye Terminator Sequencing kit (Applied Biosystems, Carlsbad, CA) and an ABI 3730 DNA analyzer.
To determine recombination frequencies along the long arm of chromosome A1, individual Hessian flies were genotyped for alleles segregating for three sequence tagged sites (STSs) and two simple sequence repeat (SSR) markers (supporting information, Figure S1) that had been discovered previously in BAC sequence data (Aggarwal et al. 2009; Schemerhorn et al. 2009). Each individual was genotyped using its DNA as a template in separate PCRs. Primer pairs used to amplify each marker were as follows: SSR 71 (GenBank accession no. EU340414), GAAGCCATTGCTCTCTCAGG and ATACGGCAAATCGAAGCATT; BAC-end STS CL27b3b (GenBank Trace Archive no. 2136886008), TCGTGAGTTTTACTTGTATG and GTATCTTCGTCGTAACTCATC; BAC-end STS 4e18g (GenBank Trace Archive no. 2136867722), CGCTTGCTCAGGTTGTG and CGAACACTGTGGAAATGC; SSR 113 (GenBank accession no. GQ257854), CCTGAATATGTGCACCGGATA and TGGGAATTTTCTGGTTTTTCA; and BAC-end STS 3L9b (GenBank Trace Archive no. 2136867352), CGAAGACATAAAATATCCACG and GTGGATTTGGAGCGAAGC. PCR was performed in 15 μl containing 10 mm Tris-Cl (pH 8.0), 50 mm KCl, 1.5 mm MgCl2, 0.01% gelatin, 200 μm each of the four dNTPs, 200 nm of each primer, 5 ng of genomic DNA, and 2.0 units of Taq DNA polymerase. Each PCR was performed for 35 cycles of 94° for 30 sec, 52° for 30 sec, and 72° for 1 min and given a final extension at 72° for 10 min. The alleles of all markers, except SSR 71, were visually scored using ethidium bromide staining after 2% agarose gel electrophoresis (Figure S1). Marker SSR 71 was scored using a Beckman Coulter CEQ 8000 DNA analyzer according to the manufacturer's recommendations (Figure S1).
BAC library screening and polytene chromosome preparations were performed as previously described (Behura et al. 2004). BAC clones were derived from three BAC libraries, the Hf library (X. M. Liu et al. 2004), the Mde library (Behura et al. 2004), and the CL library (Aggarwal et al. 2009). Fluorescence in situ hybridization (FISH) to Hessian fly polytene chromosomes was performed using BAC DNA probes. BAC DNA (1 μg) was nick translated with biotin (Roche), digoxigenin (Roche), or dinitrophenyl-11-dUTP (DNP, Invitrogen). Hybridizations were performed using 20 ng of each labeled BAC per polytene chromosome preparation. Biotin-labeled probes were detected using AlexaFluor 488-conjugated streptavidin (Molecular Probes, Eugene, OR). Digoxigenin-labeled probes were detected using anti-digoxigenin rhodamine (Roche). DNP-labeled probes were detected using rabbit anti-dinitrophenyl KLH IgG (Molecular Probes) followed by chicken anti-rabbit IgG conjugated with AlexaFluor 647 (Molecular Probes). Microscopy was performed using a Nikon Eclipse 80i epifluorescence microscope. Images were captured with a Photometrics Cool Snap HG CCD camera controlled with Metamorph software (Universal Imaging).
To characterize rearrangements and order contigs on the A1 chromosome, a series of FISH experiments was performed in which a single BAC from two or three different contigs was used as a probe. Digital images from these experiments were then examined to determine BAC (contig) order. To determine the orientation of a contig on the chromosome, two BACs from each of the 74 A1 contigs were used as probes in separate FISH experiments. To detect inversions in Hessian fly populations, BACs Mde37d21 and Hf5d5 were used as probes. To estimate the length of the inversions, the relative physical length of each inversion, compared to the total length of all chromosomes, was determined in 10 images using the Metamorph software. Those estimates were then multiplied by the size of the Hessian fly genome (158 Mb) and the means and standard errors were calculated.
Genetic and physical mapping of Cm:
To identify a Cm-linked DNA marker, we genotyped females in a single Indiana L strain family as thelygenic (n = 35) or arrhenogenic (n = 48) and then subjected the DNA of each genotyped female to AFLP–PCR DNA fingerprinting. One AFLP band that failed to recombine with the thelygenic phenotype was discovered (Figure S1). Upon determining the sequence of this band, we recognized that it was identical to marker 2709 (GenBank accession no. BV079629), an AFLP that had been mapped previously within a cluster of 12 AFLPs near the telomere of the long arm of Hessian fly polytene chromosome A1 (Behura et al. 2004). In that study, we noted the long arm of A1 (A1q) appeared to have a relatively low rate of genetic recombination.
The Mde BAC library was screened using AFLP 2709 as a probe to identify a BAC clone (Mde37d21) containing marker 2709. Mde37d21 was then used as a probe in a FISH experiment to determine its position on the polytene chromosomes. As expected, its position was near the end of A1q. However, the hybridization signal was more distal on some chromosomes than others (Figure 2A). This suggested that, among some individuals, a chromosome rearrangement was present near the end of A1q. The putative rearrangement was observed only in females that were apparently heterozygous for the rearrangement. Polytene chromosome morphology in those individuals was also consistent with the presence of a chromosome rearrangement; the ends of the polytene A1q homologs were unpaired in putative rearrangement heterozygotes (Figure 2A), but paired in females lacking the putative rearrangement (Figure 2B).
Contig mapping of chromosome A1:
To verify the presence of a rearrangement and determine its structure, BACs from each of 74 A1 BAC contigs (Aggarwal et al. 2009) were used as FISH probes to determine the order of the contigs on putatively rearranged and nonrearranged A1 chromosomes. These experiments revealed that near the Mde37d21 position, three contigs (367, 1288, and 1530) had an inverted order on approximately one-quarter of the chromosomes examined (data not shown). This indicated that the putative rearrangement was a small inversion (Figure 3). BACs Mde37d21 and Hf7b16 confirmed this hypothesis by clearly defining inversion breakpoints (Figure 2C). The relative length of the inversion, hereafter called In(A1q1), indicated it was only 2.1 ± 0.2 Mb long (1.3% of the genome). Interestingly, a second group of six contigs (1267, 94, 252, 302, 643, and 360) also had an inverted order in approximately one-quarter of the chromosomes in several preparations (data not shown). This suggested the existence of a second, more proximal, and nonoverlapping inversion (Figure 3). This inversion's breakpoints were apparent in FISH experiments that used both BAC Hf5d5 and BAC CL1c14 as probes (Figure 2D). The relative length of this inversion, hereafter called In(A1q2), indicated that it was 3.2 ± 0.6 Mb long (2% of the genome). We used Mde37d21 as a diagnostic FISH marker for In(A1q1) and Hf5d5 as a diagnostic FISH marker for In(A1q2) in all subsequent experiments.
Autosome 1 inversions in Hessian fly populations:
To examine the prevalence of the inversions, we examined their distribution in nine Hessian fly populations (Table 1). These populations had been collected independently from wheat fields across the United States from 1955 to 2005 and from Israel in 2008. They were maintained in the greenhouse as large (N > 1 × 106) separate populations. Using FISH, we examined 10 individuals within the broods produced by individual females taken from these populations. In(A1q1) was observed in every population (Table 1). In(A1q2) was observed in all populations except Georgia E, Georgia vH9, Indiana D, and Israel. The inversions were never homozygous, and In(A1q2) was never present unless In(A1q1) was also in cis. Thus, three A1 variants were observed (Table 1, Figure 2). These were named A1-prime-1 (A1′1), which carried only In(A1q1); A1-prime-2 (A1′2), which carried both In(A1q1) and In(A1q2); and A1, which lacked either inversion.
A1 and A1′ chromosome distribution matched the distribution of Cm (Figure 1, B and C, Table 1): Females in each all-female brood (n = 51) were composed of A1′[1 or 2]/A1 (Cm/−) heterozygotes and A1/A1 (−/−) homozygotes. Males in each all-male brood (n = 39) were always A1/A1 (−/−). The chromosomes of both males and females in nine amphigenic broods lacked inversions. The Kansas amphigenic broods were subsequently inbred by full-sib matings for two consecutive generations as amphigenic populations lacking A1 inversions. The presence of In(A1q1) in the Israeli population (Figure 2, E and F) indicated that In(A1q1) evolved before the introduction of the Hessian fly into the new world.
Correspondence between A1 inversions and the thelygenic phenotype:
To test the correspondence between Cm and the inversions further, a single male selected from an all-male A1/A1 family was mated with both a thelygenic female and an arrhenogenic female taken from an [A1/A1; A1′2/A1] all-female family (Figure S2). The F1 female offspring of the thelygenic female were then allowed to freely intermate with the F1 male offspring of the arrhenogenic female. The F2 adults were also permitted to freely intermate. FISH diagnostics were then used to karyotype the polytene chromosomes of 10 individuals within each of the broods produced by 86 individual F2 females. Again, the correspondence between A1′2 and Cm was perfect. Each all-female F3 family (n = 48) contained both A1′2/A1 (Cm/−) and A1/A1 (−/−) females. Therefore, assuming that F2 males (−/−) never transmitted A1′2, thelygenic F2 females (Cm/−) always carried A1′2. Consistent with this observation, each all-male F3 family (n = 38) contained only A1/A1 (−/−) males. Thus, the arrhenogenic F2 females (−/−) that produced the F3 males clearly lacked A1′2. No recombination between In(A1q1) and In(A1q2) was observed among the 860 individuals examined.
A1 inversions in exceptional males:
Chromosome nondisjunction occasionally occurs during Hessian fly oogenesis, producing ova that lack X chromosomes (A1A2OO) (Stuart and Hatchett 1991). After fertilization and chromosome elimination, the somatic cells of the zygotes produced by these ova have a male-determining karyotype (A1A2OO/A1A2X1X2). If their mother is thelygenic, these zygotes develop as “exceptional males” in female broods (Figure S3 A). These exceptional males transmit their maternally derived autosomes and their paternally derived X chromosomes to their offspring (Figure S3 B). We reasoned that because exceptional males can inherit Cm from a thelygenic mother, they could be used to test the correspondence between the inversions and the thelygenic phenotype and determine if the inversions are homozygous viable.
Nine exceptional males were discovered in the brood of a single Missouri A1′2/A1 thelygenic female. To determine whether any of these had inherited A1′2 from their mother, these males were permitted to mate with multiple sisters. Each mated female was then separately caged on wheat seedlings, and the polytene karyotypes of at least 10 of her offspring were examined. Only six males successfully mated. FISH diagnostics determined that four of these transmitted A1 and two transmitted A1′2. As expected (Figure S3 B), the males that transmitted A1′2 (Cm) produced male broods that were A1/A1′2 (−/Cm) (n = 7; Figure 4A), and female broods that were segregating for A1/A1′2 (−/Cm) and A1′2/A1′2 (Cm/Cm) (n = 6; Figure 4, B and C). Males that transmitted A1 (−) produced male broods that were A1/A1 (−/−) (n = 6) and female broods that were segregating for A1/A1 (−/−) and A1′2/A1 (Cm/−) (n = 4, data not shown). No recombination between the inversions on the A1′2 was observed. The discovery of A1′2 homozygous female larvae (Figure 4C) clearly indicated that the A1′2 chromosome was homozygous viable.
The remaining females in three [A1/A1; A1′2/A1] female broods and four [A1/A1′2; A1′2/A1′2] female broods were allowed to develop, mate, and reproduce on separate caged pots of wheat seedlings. The sexes of the broods produced by these females were then determined. The results confirmed the correspondence between the inversions and the thelygenic phenotype (Table 2): Both thelygenic (Cm/−) and arrhenogenic (−/−) females were present in equal numbers in [A1′2/A1; A1/A1] broods, but only thelygenic females (Cm/−; Cm/Cm) were present in [A1/A1′2; A1′2/A1′2] broods.
Recombination frequencies over the long arm of chromosome A1:
To test for recombination suppression along the long arm of each A1 variant, five polymorphic A1q molecular markers were genetically mapped using three mapping populations (Figure 5). The first population consisted of the male offspring of an arrhenogenic female (−/−) lacking A1 inversions (A1/A1; n = 54). The second population consisted of the female offspring of an A1′1/A1 thelygenic (Cm/−) female (n = 50), and the third population consisted of the female offspring of an A1′2/A1 thelygenic (Cm/−) female (n = 132). To map Cm, the females in the latter two populations were genotyped as Cm/− and −/− by sexing their offspring. In the absence of inversions, only the two most proximal markers failed to recombine. However, in the presence of In(A1q1), there was no recombination between Cm and the two most distal markers, 27b3 and SSR 71. This suggested that In(A1q1) suppresses recombination in the distal region of A1q where Cm resides. Interestingly, when In(A1q1) was present, recombination appeared to increase between the more proximal markers on the chromosome. When both In(A1q1) and In(A1q2) were present, all recombination between all but the two most proximal markers (SSR 113 and 3L9b) was completely eliminated. In(A1q2) therefore appeared to extend recombination suppression with Cm. This was consistent with the absence of recombination we observed between the inversions in previous experiments.
We hypothesize that the A1 inversions are a thelygenic-determining neo-sex chromosome. Other sex chromosomes have evolved when a dominant, sex-determining, genetic factor became the master switch in a sex determination pathway (Charlesworth 1978, 1996; Bull 1983; Rice 1996). An inversion then evolved near this factor to prevent the breakup of allele combinations that had sex-specific benefits. The recombinationally isolated chromosomal segment was then subject to gene erosion (Charlesworth and Charlesworth 2000). This process is visible in the remnants of the genes that were once shared by the proto-sex chromosomes of humans (Lahn and Page 1999; Hughes et al. 2005), other mammals (Graves and Shetty 2001), and birds (Ellegren and Carmichael 2001; Handley et al. 2004). It has also been observed in the chromosomes of snakes (Jones and Singh 1985), plants (Z. Liu et al. 2004; Nicolas et al. 2004; Zluvova et al. 2005), and fungi (Lengeler et al. 2002), as well as in autosome–sex chromosome translocations (Steinemann and Steinemann 1998; McAllister 2003; Bachtrog 2005) and synthetic sex chromosomes (Rice 2002).
We believe the A1′ inversions provide additional evidence of this process. We note that the A1′ chromosomes have three properties that are associated with neo-Y and neo-W chromosomes (Charlesworth 1978): (1) They carry the dominant master switch of sex determination (Cm), (2) they are normally present in a heterozygous state, and (3) they are associated with inversions that suppress genetic recombination. Genetic decay was not observed as the A1q inversions proved to be homozygous viable. However, this may be attributed to either a lack of evolutionary time or the frequency in which exceptional males make these chromosomes homozygous in nature.
A chronology that is consistent with the evolution of a neo-sex chromosome is also evident in the distribution of the A1′ inversions. Before Cm arose, females would have produced offspring of both sexes. Amphigenic females have this primitive condition and lack A1 inversions. After Cm evolved, alleles of genes near Cm might have acquired thelygenic- or female-specific functions, making the evolution of an A1 inversion advantageous as it suppressed the breakup of this gene assembly. Consistent with this possibility, In(A1q1) and In(A1q2) are present on the Cm-bearing chromosome. It is possible that In(A1q1) is the Cm mutation. If so, then In(A1q2) represents the first recombination suppressing inversion. If In(A1q1) is not the Cm mutation, then In(A1q2) represents an extension of recombination suppression, one that is reminiscent of the strata on the human Y (Lahn and Page 1999) and the chicken W (Handley et al. 2004). After the evolution of these chromosomes, we suspect that an arrhenogenic A1 probably evolved to complement its thelygenic A1′ counterparts, and together these chromosomes engendered the monogenic phenotypes that are predominant in Hessian fly populations. However, monogeny is disadvantageous when the family of an individual female becomes geographically isolated from all others. We suspect that this disadvantage has both maintained amphigenic genotypes in Hessian fly populations and promoted alleles that favor the production of exceptional males.
The previous discovery of inversions that control postzygotic sex determination also supports our hypothesis. The X-prime (X′) chromosomes of Sciara coprophila and S. impatiens, discovered in the 1930s, carry an inversion and a maternal effect master switch, the “prime factor” (PF) that controls X chromosome elimination (Metz 1938; Gerbi 1986; Sánchez and Perondini 1999). X linkage obviously distinguishes PF from Cm. However, this distinction is superficial. Because the master switch acts maternally and the maternal karyotypes are diploid for all chromosomes, X′ and A1′ are functionally equivalent. Both chromosomes engender female development and therefore may be classified as maternal effect neo-W's; we suggest the term “W-prime” (W′), where “prime” represents the maternal effect. Unlike a neo-W, these neo-W′'s establish thelygenic females (ZW′) as the heterogametic “sex” and arrhenogenic females (ZZ) and males (ZZ or ZO) as the homogametic “sexes.” Thus, the genetic conflicts that act as a force in the evolution of a sex chromosome (Partridge and Hurst 1998; Werren and Beukeboom 1998; Rice et al. 2008) are probably altered when the chromosome is a neo-W′.
The discovery of the A1′ inversions led us to question whether similar inversions might exist in other monogenic species. Bell (1982) placed these into three catagories: cyclically parthenogenetic taxa (cladocerans, rotifers, and cynipid wasps), ectoparasitic arthropods (e.g., Pediculus), and a “mixed bag” of cecidomyiids, sciarids, isopods, amphipods, and centipedes. More recently, symphypleonan springtails have been shown to belong in the latter category (Dallai et al. 1999), and monogeny has been more firmly established among cyclically parthenogenic cynipids (Stone et al. 2002), certain aphids (Rispe et al. 1999; Miller and Avilés 2000), and Daphnia (Crustacea: Cladocera) (Innes and Dunbrack 1993). At least one cyclically parthenogenetic nematode may also be monogenic (Albertson et al. 1979). Interestingly, it is possible that the anomalous Hessian fly chromosome cycle may have evolved in a cyclically parthenogenetic gall midge (Nicklas 1960) as different groups of primitive gall midges utilize this life cycle (Matuszewski 1982; Normark 2003; Roskam 2005).
Others have thoughtfully modeled the molecular mechanisms that underlie the sciarid X′ chromosome maternal effects (Haig 1993; de Saint Phalle and Sullivan 1996; Sánchez and Perondini 1999). We do not attempt to review those models here. However, while we do not suggest that Cm is necessarily homologous to the sciarid PF, we do note that the relatively small size of In(A1q1) presents an excellent opportunity to identify a candidate PF and examine the veracity of those models. Finally, we note that the A1′2 inversions would make an excellent target for foreign DNA constructs. Because the paternally derived chromosomes are always eliminated from the sperm (Figure 1A), insertions into the other autosomes and X chromosomes would be lost when males carrying the inserted construct mate only with arrhenogenic females. The A1′2 inversions would protect a foreign construct from this type of loss because they ensure maternal transmission.
We thank R. H. Shukle and P. Weintraub for providing Hessian fly populations, and we thank S. A. Jackson, V. Ferris, and two anonymous reviewers for comments that improved the manuscript. We gratefully acknowledge an award from the U.S. Department of Agriculture–National Research Initiative (2004-03099) that supported this work.
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.108589/DC1.
↵1 Present address: EMBRAPA, Rod. MG 424 KM45 Sete Lagoas, MG, Brazil.
↵2 Present address: Department of Horticulture, University of Wisconsin, Madison, WI 53706-1590.
Communicating editor: D. Charlesworth
- Received August 13, 2009.
- Accepted December 11, 2009.
- Copyright © 2010 by the Genetics Society of America