| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Developmental Biology and Genetics |
32Department of Genetics and Evolution, Max Planck Institute for Chemical Ecology, Hans Knöll Strasse 6, D-07745 Jena, Germany; 3Heidelberg Institute of Plant Sciences, University of Heidelberg, Im Neuenheimer Feld 345, D-69120 Heidelberg, Germany
Received for publication April 8, 2005. Accepted for publication July 31, 2005.
ABSTRACT
Of the 340 genera in the Brassicaceae, apomictic reproduction is found only in the North American genus Boechera. We investigated phylogenetic relationships, ability to hybridize, mating system, and ploidy levels of 92 lines sampled from 85 populations and representing 19 Boechera species. Phylogenetic analyses based on chloroplast DNA sequences identified three lineages in the genus. Reciprocal crosses of each line were made to a common sexual diploid B. stricta tester. The resulting F1 progeny were analyzed for the inheritance of polymorphic microsatellite loci, genome size, and seed production. Intraspecific B. stricta crosses confirmed that this species is mostly diploid and sexual. Interspecific crosses revealed many other species were diploid and sexual and could be successfully hybridized with the tester. We also found obligate and facultative apomictic diploid and triploid lines. De novo F1 polyploids (either triploids or tetraploids) were derived from the union of nonreduced (from an apomictic parent) and reduced (from the tester) gametes. However, seed production of these F1 plants was generally low, suggesting a failure in the transmission of apomixis. The creation of a wide array of segregating genetic populations will facilitate future research on the evolution and inheritance of quantitative variation in Boechera.
Key Words: apomixis • Arabis • Boechera • Brassicaceae • genetic crossing • hybridization • microsatellites • polyploidy
The cruciferous genus Boechera contains an array of morphologically and ecologically diverse taxa that have mainly radiated in alpine, montane, and desert regions of western North America during the last two million years. The group is emerging as an important system to address a variety of biological phenomena (reviewed by Mitchell-Olds, 2001
; Dobe et al., 2005
, in press), including the interrelated topics of speciation via hybridization, evolution of apomixis, and the inheritance of quantitative and genetic variation. The goal of this study is the creation and analysis of genetic segregating populations derived from a wide array of Boechera germplasm to facilitate research of these three areas.
Al-Shehbaz (2003)
has recognized 62 species in the North American genus Boechera. The taxa now circumscribed as Boechera have traditionally been classified as members of the cosmopolitan genus Arabis based on morphological characters (Rollins, 1983
, 1993
; Mulligan, 1995
). However, both chromosome number (x = 7) (Löve and Löve, 1976
) and molecular marker studies (Koch et al., 1999
, 2000
) suggest that the group is an independent monophyletic assemblage and is more closely related to other North American genera such as Pennellia and Haliomolobus (Bailey et al., 2002
; Al-Shehbaz, 2003
). From molecular clock calculations based on sequence evolution, the radiation of Boechera species has been estimated to have possibly predated the pleistocenic glaciation and deglaciation cycles (Dobe et al., 2004a
, b
), including the divergence of an estimated forty diploid sexual lineages (Windham et al., 2004
). However, the reproductive isolation of these taxa is incomplete, as evidenced by the frequent occurrence of hybrid and allopolyploid lineages (Böcher, 1951
; Rollins, 1983
, 1993
; Mulligan, 1995
; Sharbel and Mitchell-Olds, 2001
; Dobe et al., 2004b
).
Hybridization, including allopolyploidization, is known to be an important mechanism for speciation and phenotypic diversification in flowering plants (reviewed by Arnold, 1992
; Rieseberg, 1997
; Ramsey and Schemske, 1998
; Soltis and Soltis, 1999
). Phenotypic analysis has been used to characterize several hybrid lineages and their diploid progenitors in Boechera. For example, Rollins (1983)
documented sympatric populations of B. holboellii, B. williamsii, and the phenotypically intermediate hybrids. Further support for hybridization has come from allozyme analysis, in which extensive heterozygosity was found in populations of B. holboellii, B. gunnisoniana, and B. lignifera (Roy, 1995
). Most studies of hybrid speciation have focused on the interspecific hybrid B. divaricarpa derived from the crossing of B. stricta (formerly named Arabis drummondii) and B. holboellii (Böcher, 1951
; Rollins, 1983
; Sharbel and Mitchell-Olds, 2001
; Koch et al., 2003
; Dobe et al., 2004a
, b
). All three taxa are widespread and are often sympatric with one another. Molecular analyses, such as the detection of intraindividual ITS polymorphism (Koch et al., 2003
), the distribution of parental chloroplast haplotypes (Dobe et al., 2004a
), and the combination of parental microsatellite alleles (Dobe et al., 2004b
), have shown that B. divaricarpa has been derived repeatedly.
The extensive hybridization found in Boechera may be fundamental to the occurrence of apomictic reproduction. Analyses of Boechera hybrids have found that many are triploid, produce unreduced gametes, and their progeny arrays show fixed heterozygosity (Böcher, 1951
; Roy, 1995
; Koch et al., 2003
; Dobe et al., 2004b
; Sharbel et al., 2005
), suggesting that they reproduce apomictically (asexual reproduction by seed). Cytological and embryological observations in B. gunnisoniana and B. holboellii have confirmed the occurrence of the Taraxacum-type diplosporous apomixis (Böcher, 1951
; Naumova et al., 2001
; Taskin, 2004
). In many other well-characterized apomictic species complexes, such as Poa, Panicum, Tripsacum, Hieracium, Alchemilla, and Potentilla, apomixis tends to occur in polyploids, and then most often at the tetraploid or greater levels (reviewed in Asker and Jerling, 1992
). While there is a strong link between polyploidy and apomictic reproduction, apomixis has been documented among hybrid diploid cytotypes (Asker and Jerling, 1992
), including some Boechera taxa (Böcher, 1951
).
The incidence of apomixis in diploids suggests that hybridization, rather than polyploidy, could be causal (Mogie, 1992
). A possible explanation is that hybridization interferes with the temporal regulation of normal sexual reproductive pathways (Koltunow and Grossniklaus, 2003
). This putative hypothesis has been formulated as the hybridization-derived floral asynchrony (HFA) hypothesis (Carman, 1997
). The unstable regulation of sexual pathways would explain the frequent occurrence of incomplete apomixis (facultative apomixis) observed in many apomictic lineages. The occasional sexual reproduction of mostly apomictic lineages by hybridization, or backcrossing, with diploid sexual lineages would then lead to the establishment of new recombinant genotypes.
The establishment of diploid hybrid apomictic haplotypes in Boechera could be due to homoploid hybridization or alternatively may occur by base chromosome number reductions (Asker and Jerling, 1992
). The "diploid-tetraploid-dihaploid cycle" described by de Wet (1968)
was used to explain the occurrence of diploid apomictic lineages of Dichanthium. In this theory, the union of nonreduced gametes produced by diploid (or triploid) lines establish tetraploid lineages. These tetraploid lineages then give rise to new diploid lineages via the production of reduced egg cells (2n) that develop pseudogamously. Interestingly, few Boechera lineages are tetraploid (reviewed in Dobe et al., 2005
), and when they do occur, they may be sexual (Böcher, 1969
; Johnson, 1970
). The lack of tetraploids could either be explained by their selective disadvantage or alternatively, when they do form, they have a propensity to produce dihaploid offspring. The historical occurrence of hybridization, shifts in breeding systems, and alterations of ploidy have profound implications for our understanding of the inheritance, transmission, and evolution of genetic and quantitative variation for Boechera.
Numerous studies have analyzed molecular and phenotypic diversity of Boechera species. This includes the molecular evolutionary analysis of gene families (Bishop et al., 2000
; Schein et al., 2004
), the phylogeography of haplotypes (Dobe, 2004a
, b
), the occurrence of supernumary B-chromosomes (Böcher, 1951
; Sharbel et al., 2004
, 2005
), and variation in ploidy (Sharbel and Mitchell-Olds, 2001
). Several studies have focused on the evolved responses to pathogen or insect pests, including the interactions of Boechera species with the flower-mimicking fungal rust pathogens of the genus Puccinia (Roy, 1993
, 1995
, 2001
; Roy and Bierzychudek, 1993
), and defenses against herbivorous insects (Siemens et al., 2003
; Windsor et al., 2005
). The further elucidation of these and other patterns of variation would be greatly aided by the creation and analysis of segregating genetic stocks.
Our goal is to facilitate future genetic studies in Boechera and ultimately to study the mechanisms of speciation, the inheritance of apomixis, and the evolution of quantitative genetic variation. In this study, we present our initial work on (1) hybridization between several selected species, (2) the testing of mode of reproduction, and (3) the creation of segregating genetic populations. These three objectives were met by making reciprocal crosses of a wide range of Boechera germplasm with a common diploid sexual B. stricta tester. First, 19 different species were tested for their ability to hybridize with B. stricta. Second, we used the analysis of both ploidy and the inheritance of polymorphic microsatellite loci to determine the breeding systems of the successful crosses in the F1 progeny. Because we were interested in the transmission of apomixis, we analyzed many crosses that involved likely apomicts from B. divaricarpa and B. holboellii. The seed set from the resulting F1 plants was evaluated to determine which populations could be utilized for future studies of the inheritance of quantitative variation, such as genetic mapping and QTL analyses. For this objective, we extensively sampled intraspecific variation of B. stricta. Finally, selected species were analyzed for chloroplast haplotype variation to integrate them into an already existing large-scale, continental phylogeographic framework.
MATERIALS AND METHODS
Plant material
Seeds were collected from up to 31 individual plants (families) from 85 populations representing 19 different Boechera species growing in western North America (Table 1, Fig. 1; see Appendix S1 for geographical information in Supplemental Data accompanying online version of this article). The species were identified using keys from Dorn (1984)
and Rollins (1981). Boechera stricta (48), B. divaricarpa or other hybrids (19), and B. holboellii (9) were sampled from multiple sites. The other 16 species, selected because they were suspected to be diploid, were sampled from only single localities.
|
|
A random individual from each family in each population was selected for genome size analysis (described later). A total of 761 individuals from the 85 populations were analyzed. Single lines were randomly selected from most populations for crossing experiments (described later). Two lines were selected from the Vipond Park site. In several instances, two collections were made from the same approximate site. Each collection was treated as its own population and a line for crossing was selected for each one (Twin Camp, Twin Saddle, and Parker Meadow). If a population contained mixed ploidies or species, then a line from the most prevalent ploidy type was selected for crossing. However, in three instances an individual of each ploidy type or species was selected (Lost Trail Meadow, Ruby Creek, and Parker Meadow). Herbarium specimens for each of the selected lines were made and deposited in the Herbarium of the University of Heidelberg [HEID] (Appendix). Herbarium specimens were examined to confirm species identifications (M. Windham and I. Al-Shehbaz, Missouri Botanical Garden, personal communication).
Genome size measurements
Ploidy analyses were performed on a PARTEC (PARTEC GmbH, Münster, Germany) CCA-II flow cytometer using their CyStain UV precise P nuclei extract and staining kit (PARTEC GmbH, Münster, Germany) according to the manufacturer's protocol. Sample leaf material was measured in combination with an internal size standard (leaf material from Brassica rapa for all samples and confirmed with Matthiola incana for tetraploid individuals) (Sharbel et al., 2005
). The F1 plants (described next) were analyzed by flow cytometry on pooled samples of 1–8 siblings obtained from a single cross. If multiple peaks were detected, each F1 plant was then analyzed separately with flow cytometry.
Genetic crossing and growth of F1 plants
A total of 92 experimental lines, representing all 19 species, was selected for crossing experiments (Table 1, Fig. 1). This included 76 diploids, 14 triploids, one tetraploid, and one hexaploid line. A single diploid genotype of B. stricta was used as the common genetic tester. Dr. Barbara Roy originally collected the tester line, SAD12, from the Taylor River, Gunnison County, Colorado, USA (Ad29T). Microsatellite analysis has shown that this line is an inbred, self-pollinating line that reproduces sexually (unpublished results). In addition, this line has undergone two rounds of self-seed increase by single-seed descent (S2 generation) in the greenhouse prior to its use in this experiment. Eight replicates of the tester S2 plants were used.
Each of the experimental lines was reciprocally crossed to the tester line. For the crosses, 3–5 flower buds with nondehisced anthers were stripped of petals, sepals, and stamens, then pollen from the reciprocal line was deposited on the exposed stigma surface. Several buds from each line were stripped of petals, sepals, and stamens, but no pollen was deposited on the stigma to test for accidental self-pollination or obligate apomixis. Resulting F1 seed was collected when the siliques were matured.
Planting and growth of F1 plants
A total of 739 F1 seeds were germinated and planted, with 1–8 plants per successful cross (Table 1). Seeds were germinated as already described, then transferred to 96-well flats in a randomized block design. The plants were grown in a controlled growth room (type) in long-day conditions (16 h light, 8 h dark) for 28 d. All the flats were then vernalized at 4°C for 6 weeks to accelerate and synchronize the flowering of the lines. Plants were then transferred back to the growth room, transplanted into 11 x 11 x 13 cm pots and grown until flowering (1–5 weeks after removal from the cold). Tissue for DNA extraction and ploidy analysis was collected 1–3 weeks after plants were removed from 4°C.
One or six F1 lines from each reciprocal cross (a total of 280 lines), were transplanted into 15 x 15 x 15 cm pots and were grown in the greenhouse to generate F2 seed (Appendix S2, see Supplemental Data with online version of this article). Mature siliques were collected during a five-week period. The F2 seeds were cleaned, and the seed yield was scored on a scale from 1 to 4, with 1 = >200 seeds generated, 2 = 200–50 seeds generated, 3 = 50–10 seeds generated, and 4 = 9 or less seeds. Scores of 1 and 2 were taken as evidence of the F1 plant being fertile (comparable to normal Boechera seed yields), and hence either a well-functioning sexual line or apomictic. Scores of 3 and 4 were taken as evidence for abnormal seed production (sterility or poor sexual performance).
DNA extraction, microsatellite amplification and analysis
Parental line DNA isolations were made from freeze-dried tissue according to the CTAB procedure as described in Pires et al. (2004)
. Approximately 0.1 g of fresh leaf material from the F1 progeny was collected, frozen in liquid nitrogen, and genomic DNA was prepared according to DellaPorta et al. (1983)
. Recovered genomic DNA was suspended in 200 µL of 10 mM Tris-Cl, pH 8.0. Microsatellite loci were amplified by the polymerase chain reaction (PCR) as described by Clauss et al. (2001). DNA from the 92 experimental lines were tested using microsatellite primers (Appendix S3, see Supplemental Data with the online version of this article) previously shown to be polymorphic in Boechera (Clauss et al., 2001; Dobe et al., 2004b
) or with several additional primer pairs (B. Song, Max Planck Institute, personal communication). We preferentially used primer pairs that were polymorphic between the experimental lines and the common tester when resolved on 3% Metaphor agarose gels (FMC Bioproducts, Rockland, Maine, USA) (Appendix S3). However, polymorphisms for several lines were detected by analysis of fluorescently labeled microsatellites (as described by Clauss et al., 2001) by TraitGenetics GmbH, Germany (Appendix S3). The presence of both parental alleles (from the tester and the target line) in the F1 was taken as evidence of sexual reproduction, whereas the presence of only maternal alleles was taken as evidence of asexual reproduction.
DNA sequencing and analysis
PCR reactions (30 µL) were performed in a master mix containing 1x PCR buffer (10 mM TRIS/50 mM KCl buffer, pH 8.0), 1.5 mM MgCl, 0.1 µM of each primer, 0.1 mM of each dNTP, half a unit Taq DNA polymerase (Promega, Madison, Wisconsin, USA), and approximately one ng of template DNA using an MJ Research (Bio-Rad., San Francisco, California) PTC-200 thermal cycler. Thermal cycling started with a denaturation step at 95°C lasting 3 min; followed by 30 cycles each comprising 45 s denaturation at 95°C, 30 s annealing at 45°C trnL intron/48°C trnL/F IGS, and 60 s elongation at 72°C. Amplification ended with an elongation phase at 72°C lasting 10 min, and a final hold at 4°C.
The trnL(UUA) intron was amplified using the forward primer 5'-CGA AAT CGG TAG ACG CTA CG-3' and the reverse primer 5'-GGG GAT AGA GGG ACT TGA AC-3' (primer c and d according to Taberlet et al., 1991
). Sequences comprised the complete intron and the second exon of the trnL gene. For amplification of the trnL(UUA)-trnF(UUC) IGS primers 5'-GGT TCA AGT CCC TCT ATC CC-3' (primer e according to Taberlet et al., 1991
) and 5'-GAT TTT CAG TCC TCT GCT CTA C-3' (designed in this study) were used. Amplified sequences included the complete IGS and the first 18 bases of the trnF gene.
PCR products were checked for length and concentrations on 1.5% agarose gels. No purification of PCR products was necessary for subsequent sequence reactions. Cycle sequencing was performed using the TaqDyeDeoxy Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, California) and the original amplification primers. However, the reverse trnL/F IGS primer was modified adding an additional cytosine to its 3'-end. Products were analysed on an ABI 3730XL DNA analyzer automated sequencer (Applied Biosystems). Cycle sequencing was performed on both strands; in the majority of cases each reaction spanned the complete sequence.
Phylogenetic inference
A phylogenetic analysis was performed on trnL intron-trnL/F IGS haplotypes. The trnL intron-trnL/F IGS sequences from Halimolobos parryii (AF307539) and Polyctenium fremontii (AF183043, AF183044) served as outgroups. Analyses were done with new haplotypes (DA-DS; GenBank: DQ013045–DQ013060), in addition to those published in Dobe et al. (2004a)
. The maximum parsimony optimality criterion was applied for reconstruction of phylogenies using PAUP* version 4.0 b10 (Swofford, 2002
). The heuristic search algorithm was chosen, using the RANDOM ADDITION of taxa and the TBR option (tree bisection-reconnection) for branch swapping. GAPMODE was set to MISSING, but a parsimonious informative 5-bp mononucleotide repeat at alignment positions 1038–1041 was coded as additional single binary 0/1 character in the Nexus input file. A strict consensus tree was constructed from 10 000 shortest trees retained. Bootstrapping was carried out on 1000 replicates using the HEURISTIC search option.
RESULTS
Population ploidy analyses and crossing line selection
Ploidy analysis of collection sites (Table 1) found that most (72/85) were of uniform ploidy levels (either diploid or triploid), and 13 were of mixed ploidy. Fifteen of the species sampled were from single diploid-only populations, and individual lines from each were selected for crossing experiments. Triploid populations and/or individuals were identified for B. divaricarpa, B. holboellii, B. stricta, and B. williamsii. A single diploid B. williamsii individual was selected for crossing.
Most B. stricta populations (40/48) contained only diploid individuals (Table 1), and representatives were selected from each site. Six diploid B. stricta lines (Hessie, Lost Trail Meadow, Sagebrush Meadow, Ruby Creek, Twin Saddle, and Eagle Mountain) were selected from mixed ploidy sites, in which the nondiploids were of hybrid origin (e.g., B. divaricarpa or Boechera spp.) based on trichome morphology. Two triploid and phenotypically B. stricta individuals (both with malpighiaceous trichomes) were selected, one from an all-triploid population (Toiyabe West), and the other from a mixed diploid and triploid population (Kiger Gorge Lookout Ridge).
Seven of the nine populations of B. holboellii sampled from Montana and Idaho were all diploid, and single lines were selected from each site. A diploid individual was selected from the mixed ploidy Bear Ridge site. The Eight Mile Creek site was of mixed ploidy, including a second generation greenhouse-raised hexaploid plant that was selected for crossing (A. Shumate, Fairleigh Dickinson University, unpublished data).
Eight hybrid populations (B. divaricarpa and Boechera spp.) were found to be all triploid, and individual lines from these were selected. Also, triploids from three mixed, but predominately triploid, sites were selected (Slater Ridge, Upper East Creek, and Lima Gulch). Polyploid hybrids were selected from two predominantly diploid B. stricta sites: a triploid was selected from the Lost Trail Meadow site and a tetraploid was selected from the Ruby Creek site. There were six populations from Montana that contained only diploid B. divaricarpa individuals (Vipond Park, Garfield, Bandy Ranch, Deadman Junction, Nicholia Trailhead, and Mule Ranch), and one line from each site, or two for Vipond Park, were selected.
Generation of F1 seed
The amount of F1 seed generated from each cross varied (0–110 seeds). The number of F1 seed generated was not considered as indicative of crossing success because crosses were done nonrandomly due to asynchrony in flowering (the lines range from 65–400 d to flower; E. Schranz, unpublished data). The differences in bud size and timing of anther dehiscence of the lines may also have affected the crossing. However, it can be generalized that most intraspecific reciprocal crosses of B. stricta were successful, and that most interspecific crosses were successful in generating seed on at least one of the two parents. The exception is B. holboellii; of the nine B. holboellii lines tested, crosses involving only four of them generated any seed, and only one of these with B. stricta as the maternal plant.
Crossability and reproductive mode of parental lines
A total of 729 F1 plants were grown and analyzed (Table 1). Of the 19 species tested by crossing with the tester all yielded F1 progeny, but they showed differences in fertility. Seven species (B. stricta, B. divaricarpa, B. holboellii, B. fendleri, B. lasiocarpa, B. pallidifolia, and B. selbyi) had at least one line that yielded fertile reciprocal F1 plants. Five crosses were only successful and fertile when B. stricta tester was the maternal parent (B. crandallii, B. lyallii, B. pendulina, B. pygmaea, and B. schistacea) and two when the tester was the paternal parent (B. lignifera and B. microphylla). Crosses involving five species with the tester produced F1 plants that had little or no seed (B. lemmonii, B. perennans, B. pulchra, B. suffrutescens, and B. williamsii).
The analysis of microsatellite inheritance and ploidy of the crosses was used to assess the breeding system of the parental lines and the production of nonreduced gametes. The F1 plants were analyzed with microsatellite loci that were polymorphic for the parental alleles, as detected by metaphor gels or with fluorescently labeled primer pairs run out on a DNA sequencer (Appendix S3, see Supplemental Data with the online version of this article). All F1 plants with the tester used as the maternal plant were, as expected, heterozygous for the parental alleles and thus were sexually derived.
Of the 16 diploid species represented by only a single line, nine of them were successfully used as female parents (Table 1). The analyses of their resulting F1 offspring found that most were sexually derived (B. lasiocarpa, B. lemmonii, B. lignifera, B. lyallii, B. microphylla, B. perennans, B. selbyi, and B. fendleri). The sexually derived progeny from most of these species were diploid and hence, derived from the union of two reduced gametes. However, B. lignifera produced only nonreduced females gametes (all triploid F1 progeny). Both B. pallidifolia and B. selbyi were found to be facultative apomicts and produced mixed ploidy offspring (both diploid and triploid F1 progeny). In addition, in the ploidy analyses of F1 plants, triploid offspring were produced when B. lignifera, B. pulchra, and B. williamsii were used as male parents, showing that these three species produce nonreduced male gametes.
There were very few successful crosses between the tester and any of the B. holboellii lines (Table 1). Only one line (Ranch Creek-Sliderock) was successfully reciprocally crossed, and it was found to be apomictic and heterozygous for tested microsatellites. When used as a male, it contributed both reduced and nonreduced gametes with the resulting diploid progeny being sterile and the triploids fertile. One line (Lower Hot Springs) was successfully used as a male only and contributed reduced pollen grains. The resulting diploid plants were a mix of sterile and fertile individuals. Two additional lines were only successfully used as females; Bear Ridge was found to be apomictic and Panther Creek was sexual and had reduced egg cells. The resulting sexually derived F1 plants from Panther Creek were infertile.
The results of microsatellite and ploidy analysis of crosses involving the tester and B. divaricarpa and Boechera spp. revealed a high frequency of apomixis and nonreduced gamete production (Table 1). First, the diploid Boechera spp. line from Bandy Ranch was found to be a sexual diploid. Similarly, the tetraploid Boechera spp. line from Ruby was also found to be sexual with the union of the reduced, but 2x, female gamete and the reduced tester pollen producing fertile triploid F1 plants. Second, the diploid B. divaricarpa lines had two alleles at microsatellite loci analyzed and fertilized the tester with nonreduced pollen grains giving rise to infertile triploid F1 individuals. Only one of these five (Vipond Park, ES8) also fertilized the tester with a reduced pollen grain giving rise to a fertile diploid F1. Two of the diploid B. divaricarpa lines (Vipond Park-ES8 and Mule Ranch) were also successfully used as female parents and both were found to be facultative apomicts, resulting in sexually derived triploid F1 individuals. The triploids from Vipond Park (ES 8) were infertile and the triploids from Mule Ranch were fertile.
The crosses involving the 12 triploid B. divaricarpa and Boechera spp. lines were particularly interesting because of the frequent production of tetraploid F1 progeny (Table 1). Five of the lines (Alpine Lake, Slater Ridge, Lost Trail Meadow, Upper East Creek, and Lima Gulch) contributed nonreduced 3n male gametes to the tester, resulting in tetraploid F1 individuals of mixed fertility (some fertile and some infertile). Additionally, five of the lines (Salmon Mountain, Slater Ridge, Toiyabe East, Carson Pass, and Alpine Meadows) produced nonreduced 3n female gametes that when fertilized resulted in tetraploid F1 plants. Four of these were facultative apomicts, but the line from Alpine Meadows produced only sexually derived tetraploid F1 progeny of mixed fertility.
The results of microsatellite and ploidy analyses of F1 plants derived from the intraspecific crossing of B. stricta lines to the tester confirmed that they were mostly diploid, sexual, and produced reduced female and male gametes. Almost all parental lines were homozygous for the tested microsatellite primers (Appendix S3, see Supplemental Data with the online version of this article). However, six B. stricta lines were deviant in some manner. First, the two triploid B. stricta lines (Toiyabe west and Kiger Gorge Lookout Ridge) were found to reproduce apomictically and had two or three alleles for tested microsatellites. Second, two diploid lines (Travis Gulch-west and Hoosier Ridge-low) were found to be apomictic, but contributed reduced male gametes to the tester. These F1 plants were fertile. Both of the parental lines were heterozygous for some tested microsatellite loci. Third, one diploid line (Garfield) reproduced apomictically, donated a nonreduced male gamete to the tester (the resulting triploid F1s were infertile) and was heterozygous for all tested microsatellite loci. Fourth, a diploid line (Morgan Creek) was a facultative apomict with the sexually derived progeny being infertile triploids. However, this line contributed reduced pollen to the tester plant. This line also was heterozygous at tested microsatellite loci.
Sequence analysis and inference of chloroplast haplotypes
Length of the combined trnL intron and trnL-trnF intergenic spacer varied between 851 and 1066 bp. Forty-three variable sites along with five indels (alignment positions 226– 235, 797–1003, 890–981, 982–1078, and 1037–1041) were identified among the 85 analyzed parental lines (Table 1). The observed mutations define 34 chloroplast haplotypes (Appendix S4, see Supplemental Data with the online version of this article). Eighteen out of these (haplotypes C, D, S, U, AH, AK, AR, AS, AV, BF, BH, BI, BR, BS, BU, BY, CI, CL) have been already recognized by Dobe et al. (2004a)
. The remaining ones (haplotypes DA to DS; DQ013045–DQ013060) were found for the first time. Absolute frequencies of particular haplotypes varied from singletons, found only once, to 27 (haplotype AH).
Phylogenetic relationship of chloroplast haplotypes
The sequences were assembled according to the trnL intron/trnL-trnF IGS alignment published in Dobe et al. (2004a)
. The length of that alignment (1125 bp) and position of sites were not changed by the addition of the new haplotypes. Hence, sites in the alignment are directly comparable between the two studies. In B. stricta, 14 different haplotypes were found within the present data set, 5 in B. holboellii, 8 in the hybrid taxa (B. divaricarpa), and 12 in the remaining 13 species included in the phylogenetic analysis (Table 1). The strict consensus tree (Fig. 2; tree length = 96, CI excluding autapomorphies = 0.82) calculated from 10 000 shortest trees reconstructed by the maximum parsimony analysis comprised three major clades of approximately equal length. The topology of the tree and evolutionary position of haplotypes supported the phylogenetic relationships reconstructed by Dobe et al. (2004a)
. Haplotypes found for the first time in this study (haplotypes DA to DS) clustered within those three lineages among the formerly defined ones (haplotypes C to CL). Minimum uncorrected p distances (Swofford, 2002
) between any haplotype of the new data set and those of the previously published set (haplotypes A to CY) ranged between 0.1 and 0.22% (corresponding to 1 to 2 SNPs), while maximum p distances among all haplotypes of both studies reached 1.4%. This indicates that most of the plastid DNA genetic variation was already recovered by Dobe et al. (2004a)
in the three model taxa B. divaricarpa, B. holboellii, and B. stricta investigated by these authors. All haplotypes (except haplotypes BS and DP) carried by B. stricta were restricted to lineage II, therefore, forming a monophyletic group. Those cp types (except one accession of haplotype AH) in B. holboellii (see also Dobe et al., 2004a
) clustered within lineages I and III together with B. crandallii, B. fendleri, B. lasiocarpa, B. lemmonii, B. lignifera, B. microphylla, B. pendulina, B. perennans, B. pulchra, B. pygmaea, B. selbyi, and B. schistacea. Finally, those lines, which were assumed to be of hybrid origin (B. divaricarpa) were scattered all over the tree, i.e., occurring in all three major lineages.
|
The extensive polymorphism found in the Boechera lineages is thought to be the outcome of the combined evolutionary forces of hybridization, polyploidy, apomixis, and natural selection on quantitative variation. The results of our experimental crosses have confirmed the ability of Boechera species to hybridize, the occurrence of diploid and triploid apomicts, and the formation and functionality of nonreduced gametes as detected by the creation of new polyploid lineages. In addition, our results confirm that most lines of the widely distributed species B. stricta and the lines of several other species are sexual diploids and should be genetically tractable systems for analysis of quantitative traits.
Hybridization
The work presented here establishes that wide crosses are possible throughout the genus; 18 species were successfully crossed with B. stricta, a taxon that might represent a more distinct lineage within Boechera as shown by our phylogenetic analysis. The high success rate of crossing suggests that sexual homoploid speciation could have been a critical factor in the radiation of the genus. Crosses involving 13 species with the tester resulted in sexually derived F1 progeny that were fertile. The additional five species were also successfully crossed to the tester, but the resulting sexually derived F1 progeny produced little or no seed. Such low-fertility lineages are not necessarily doomed to extinction. Many Boechera species are perennials, thus low fertility lines could persist for several years. Backcrossing or selection on even a few progeny might allow these lineages to gain renewed fertility. For example, Rieseberg and colleagues have shown that early generation sunflower hybrids have greatly reduced fertility, but after several generations of selection and genetic recombination highly fertile and successful lineages were established that resemble naturally occurring populations (Rieseberg et al., 2000
, 2003
; Lexer et al., 2003
).
The phylogenetic distance between taxa could be an important factor influencing their ability to hybridize. Analysis of polymorphisms of the combined region of the trnL intron and trnL-trnF IGS have shown that there are three major chloroplast lineages, with the B. stricta being mostly limited to lineage II, and most B. holboellii together with those 12 additional species (B. crandallii, B. fendleri, B. lasiocarpa, B. lemmonii, B. lignifera, B. microphylla, B. pendulina, B. perennans, B. pulchra, B. pygmaea, B. selbyi, and B. schistacea) included in the phylogenetic analysis for the first time to the I and III lineages (see also Dobe et al., 2004a
). However, the high crossability of those taxa belonging to lineage I and III to the common tester (of lineage II) indicate that there may be only limited barriers to gene flow between lines of even highly diverged genetic backgrounds. We have no experimental estimate of the extent of gene flow between cpDNA lineages, however, incongruencies between phylogenies based on nuclear ribosomal DNA (ITS) and cpDNA sequences of B. holboellii indicated that there was extensive horizontal gene flow between lineage I and III (Dobe et al., 2004b
). Also, the occurrence of a few B. stricta lines in lineages I and III and of a few B. holboellii lines in lineage II provide evidence of past introgression events as discussed in Dobe et al. (2004b)
.
Interpretation of pathways and outcomes of hybridization events has concentrated so far on the establishment of polyploids and apomictic genotypes (Rollins, 1983
; Naumova et al., 2001
; Sharbel and Mitchell-Olds, 2001
; Dobe et al., 2004b
). Nevertheless, the success of many crosses among the diploid sexuals included in this study, both in terms of vitality and fertility of offspring, opens the possibility that homoploid speciation played a more significant role in the radiation of the genus than previously understood. However, to prove that such hybridization events occurred during the natural evolutionary history of Boechera, it will be necessary to integrate our experimental results with field observations, phylogenetic reconstruction, and population genetics. Such integration will facilitate a workable species concept and the development of a tractable species classification. For the most part, few hypotheses exist for past hybridization events, but we can make comparisons between our artificial hybridization results and previous studies of at least one naturally occurring hybrid, B. divaricarpa.
The current hypothesis is that B. divaricarpa is the hybrid between B. stricta and the polymorphic (or even polyphyletic) B. holboellii (Rollins, 1983
, 1993
; Böcher, 1951
; Koch et al., 2003
; Dobe et al., 2004a
, b
). The results of our crossing experiments between nine different B. holboellii lines and the B. stricta mostly failed in obtaining sexually derived fertile F1 offspring. Obviously this could be due to particular lines selected for crossing from this highly polymorphic taxon, and we currently are attempting to cross a larger set of lines from the two species. However, evidence from previous studies have also suggested that B. holboellii in fact might not be one of the parents of B. divaricarpa. Molecular results strongly support B. stricta as one of the parents of B. divaricarpa, most often the maternal parent (71% of B. divaricarpa accessions have a B. stricta chloroplast haplotype), but there is less support for B. holboellii as the other parent (Koch et al., 2003
; Dobe et al., 2004a
, b
). For example, B. divaricarpa has a shared B. stricta ITS-type 41% of the time, a shared unique B. holboellii ITS-type 23% of the time, and a unique (or recombinant) B. divaricarpa ITS-type 36% of the time (Koch et al., 2003
).
The parentage of B. divaricarpa is further complicated by the fact that B. holboellii has been found to be paraphyletic (Dobe et al., 2004a
, b
). In our phylogenetic analyses, B. holboellii does not form a monophyletic group, but rather its haplotypes are interspersed among the other taxa, sharing its genetic variation with other species included in this study. An intriguing possibility is that B. holboellii is itself of hybrid origin. There is a strong possibility that any one, or several, of the species in lineages I and III could be the other parent of B. divaricarpa. The only certainty is that unraveling the parentage of B. divaricarpa (and B. holboellii) lineages will have to be deduced from future population genetic and experimental approaches.
It is also important to note that while B. stricta and B. holboellii may now be reproductively isolated from one another, as suggested by our crossing experiments, there could still be indirect genetic exchange between the two by backcrossing and introgression with intermediate and established B. divaricarpa lineages. Indeed, when we look at the results of our crosses with B. divaricarpa and the B. stricta tester, crosses involving 13 of the 19 lines resulted in sexually derived and fertile F1 lines. The high success of backcrossing involving B. divaricarpa and B. stricta is also significant because it illustrates the successful recombination of asexual and sexual lineages and changes in ploidy.
Apomixis and polyploidy
Of the 92 lines of Boechera tested by crossing, we found evidence for apomixis and/or increases in ploidy for 31 of the lines. Both apomixis and polyploidization are the outcomes of a failure of meiosis (apomeiosis). In apomixis, the nonreduced egg cell develops parthenogenically and maintains the parental genotype and ploidy. Whereas polyploidization occurs because nonreduced gametes, either egg cells or pollen, participate in sexual reproduction and cause an increase in ploidy and genetic recombination in the progeny. The two potential outcomes for nonreduced gametes (apomixis or sex) are by not mutually exclusive and likely often occur in the same plant. Additionally, some Boechera genotypes may produce nonreduced gametes and set apomictic seed at one developmental stage and then produce normal reduced egg and pollen and be sexual at another (Böcher, 1951
). The co-occurrence of sexual reproduction (either with reduced or nonreduced gametes) and asexual (or apomictic) reproduction is referred to as facultative apomixis. Studies in other systems have found that facultative, rather than obligate, apomixis is more often encountered (Asker and Jerling, 1992
).
By examining the F1 plants derived from pollinating our Boechera lines with tester pollen, we could detect cases of apomixis (all progeny had the same genotype and ploidy as the maternal plant), fertilization of nonreduced egg cells (progeny have recombinant genotypes and increased ploidy), and facultative apomixis (some F1 progeny with the same genotype and ploidy as the mother and some with recombinant genotypes and potentially increased ploidy). The findings of our crosses are presented in Table 1, but with increased sample sizes, evidence of facultative apomixis could be found for lines currently classified as sexual or obligate apomicts. The results of our crosses identify two types of individuals that are uncommon among flowering plants, namely apomictic diploids and at least partially sexual triploids.
Triploidy is a rare condition among flowering plants (Ramsey and Shemske, 1998
), because of their low or absent seed set. Triploids have been selected in agriculture, for example, bananas are triploid (and seedless) and triploid "seedless" varieties of watermelons have been developed (Varoquaux et al., 2000
). Not surprisingly then, when triploid lineages are found they are often obligate apomicts, such as dandelions (van Dijk, 2003
) or skeleton weed, Chondrilla juncea (Bergmann, 1950
). Previous studies have made the assumption that triploid Boechera lineages are apomicts, and some individuals produce high fractions of nonfunctional pollen (Koch et al., 2003
; Sharbel et al., 2005
). However, of the 16 triploid lines tested in this study, eight of them parented sexually derived tetraploid F1 progeny with the tester (four lines as the maternal parent and five as the paternal parent, with one overlap). The relatively high proportion of tetraploid offspring produced by crossing triploid and diploid plants begs the question of what the role of such tetraploids could play in the evolution of Boechera, and why few naturally occurring Boechera tetraploid individuals are observed.
While tetraploids are rare in populations of Boechera in western North America, they do exist. For example, Johnson (1970)
and Böcher (1969)
described the observation of such individuals, and we identified a single sexual tetraploid individual from the Ruby site in Montana. Also, tetraploid individuals have been observed, such as a B. holboellii from Michigan (Sharbel and Mitchell-Olds, 2001
) and B. stricta from Massachusetts (Böcher, 1969
). The low frequency of tetraploids suggests that they may be selected against, possibly because of a fitness disadvantage. But sexual tetraploids existing at low frequency in populations of dandelions have been shown to be critical for the creation of new triploid apomictic cytotypes by the production of reduced 2n gametes that recombine with reduced n gametes from sexual diploids (Verduijn et al., 2004
). Another possible mechanism by which tetraploid lineages could contribute to the evolution of Boechera is by the production of reduced 2n egg cells, which then develop parthenogenically. These newly formed diploid lineages could be sexual or apomictic. Such cycles of ploidy have been described for other apomictic complexes, such as Potentilla argentea and in the Bothriochloa-Dichanthium complex (de Wet, 1968
; Asker and Jerling, 1992
). The analysis of the F2 progeny from our newly derived tetraploid lineages could be used to look for evidence of the production of parthenogenically derived diploid lineages.
Intriguingly, we found direct evidence for apomixis and/or nonreduced gamete production for 11 diploid Boechera lines, coming from six species (B. selbyi, B. williamsii, B. pallidifolia, B. lignifera, B. holboellii, and B. divaricarpa). Apomixis at the diploid level (excluding adventitious embryony) has been infrequently observed, but has been reported in Potentilla argentea and Hierochloe australis (Asker and Jerling, 1992
; but see Holm and Ghatnekar, 2004
). These 11 diploid Boechera lines showed some heterozygosity at microsatellite loci tested (Appendix S3), attesting to their likely hybrid origin. Dobe et al. (2004b)
similarly observed that heterozygous individuals tended to produce nonreduced gametes. Some of these lines produced sexually derived triploid lineages when crossed with the tester, attesting to their production of 2n gametes. Most of these new triploid F1 lineages had low fertility, but some F1s from crosses involving B. divaricarpa from Mule Ranch and Vipond Park and from B. lignifera produced fertile, potentially apomictic, triploids.
The finding that most newly formed triploids were not apomictic, as determined by their low levels of seed set, highlights that apomixis in Boechera may not be under simple genetic control by a single "apomixis" causing locus or factor. Rather, apomixis is likely controlled by a number of genes and/or requires certain gene dosage ratios. For example, studies of other systems have found evidence for between two and five genes being involved in the regulation of the trait and that crosses between sexual and apomictic lineages often uncouple the various regulatory components of apomixis (van Dijk et al., 1999
; Noyes and Reiseberg, 2000
; Matzk et al., 2005
). Let us assume, for instance, that at least two doses of "apomixis factor" are needed; this could explain why the diploid offspring of the cross between the Ranch Creek B. holboellii and the tester was stertile (meiosis failed), but triploids F1 plants possessing two B. holboellii-genomes were fertile (apomictic). The need for certain gene dosages could also explain the differential reproductive success of the offspring of the crosses between B. divaricarpa and B. stricta, if we assume that the various B. divaricarpa lines carried different dosages of apomixis factors.
The study of the transmission of apomixis in Boechera and future identification of apomixis genes should be facilitated by the findings from the crosses involving the B. divaricarpa individuals from the Vipond Park site. The two lines tested from this site behaved differently. One line (ES8) was a facultative apomictic with the maternally derived sexual triploid progeny being of low fertility. But, when this line was used as a paternal parent, it fertilized the tester with a mix of reduced and nonreduced pollen. The resulting diploid and triploid lines both were fertile. The other line (ES9) was apomictic and fertilized the tester with only nonreduced pollen giving rise to low-fertility triploid F1. Results of the Vipond Park population are of particular interest because a BAC library was constructed from DNA derived from the same seed stock as these two lines (Shein et al., 2004). Thus, some important molecular resources already exist and, in conjunction with the progeny derived from these crosses, should provide important material for tackling the control and transmission of apomixis in Boechera.
We detected both diploid and triploid B. stricta apomictic lines, but all of these showed some heterozygosity at tested microsatellite loci. These apomictic lines are phenotypically B. stricta, but molecular results support the role of hybridization or introgression in their history (e.g., triploid accession ES37 carried a chloroplast haplotype from lineage III and contains non-stricta microsatellite alleles). Such taxa present particular challenges to the taxonomy of the group. In only two instances were triploid B. stricta identified (Toiyabe west and Kiger Gorge Lookout Ridge). Interestingly, both of these are outside of proposed B. stricta glacial refugia (Dobe et al., 2004a
).
Recombination of sexual diploids
Our crosses and phylogenetic work confirmed that B. stricta is primarily a diploid sexual species and constitutes a rather isolated lineage within Boechera, making it the most tractable group for addressing patterns of genetic variation in the group. Of the 48 B. stricta lines tested, crosses involving 40 produced only sexually derived diploid F1 progeny. Previous crossing studies (Roy, 1995
) and molecular analyses (Sharbel and Mitchell-Olds, 2001
; Dobe et al., 2004a
, b
) have found evidence that B. stricta is a highly selfing sexual species.
The wide distribution of our samples across western North America also makes it useful for addressing patterns of geographic evolution. Our lines of B. stricta cover most of the putative glacial refugia areas of the species. The analysis of chloroplast sequence variation found, along with basal haplotypes, a high proportion of derived types in the Rocky Mountains, from the north in Montana and Idaho to the south into Colorado and Utah (Dobe et al., 2004a
). Two main haplotypes were found to have likely colonized formerly glaciated areas of North America, with the AH haplotype spreading into the Northwest and AS spreading in the Southwest, North Central and Northeast. This genetic variation was fully covered by the 40 sexually reproducing lines crossed to the common tester, indicating the potential of genetic recombination between even the most diverged genotypes present in this species.
Future directions
First, the crossing of a diverse group of germplasm to a common tester will facilitate future studies of speciation and hybridization in the genus Boechera, the successful or failed transmission of apomixis, and the evolution of genetic and quantitative variation. Second, future analyses of crosses from different species will allow us to examine species differences, reproductive barriers, hybridization, and introgression (e.g., Bradshaw et al., 1995
; Reiseberg et al., 2000
). Finally, examination of the progeny resulting from the crosses between asexual and sexual lineages will allow us to study the transmission and control of apomixis (e.g., Noyes and Rieseberg, 2000
; Vijverberg et al., 2004
). Also, we can examine our newly formed triploid and tetraploid lineages to assess their potential role in evolution of the genus, including testing the "diploid-tetraploid-dihaploid" hypothesis and investigating the contribution of rapid non-Mendelian changes to the success of the new polyploids (e.g., Liu and Wendel, 2002
; Schranz and Osborn, 2004
). Finally, we can analyze our available segregating F2 populations derived from these crosses, in conjunction with population genetics, gene cloning, and transformation (Taskin et al., 2003
), to dissect the genetic control of traits of interest. For example, we are examining our lines for variation in flowering time and the production of the glucosinolate secondary compounds. Work is now underway for constructing genetic linkage maps from several F2 populations. These can be used to map quantitative trait loci (QTL) that control local adaptation to biotic and abiotic factors. Fundamental to our future work will be the exploitation of the close phylogenetic relationship between Boechera and Arabidopsis thaliana, both by comparative mapping and genomics approaches. Finally, our use of a common tester in all of our crosses will allow for direct comparisons of analyses from multiple F2 populations.
|
|
|
1 The authors thank S. Neumann, K. Baeuer, J. Schneider, S. Ball, and A. Boerner for technical assistance. We also thank M. Windham, B. Smith, S. Rhodes, R. Ortner, B. Vining, J. Ladyman, A. Shumate, and B. Roy for help with seed collections. Special thanks to B. Song and M. Clauss for help with microsatellite analyses, and M. Windham and I. Al-Shehbaz with species identifications. Support for this research was provided by the U.S. National Science Foundation (DEB 9527725) and the Max Planck Gesellschaft. 
4 Author for correspondence (e-mail: eschranz{at}ice.mpg.de
)
LITERATURE CITED
Al-Shehbaz I. A. 2003 Transfer of most North American species of Arabis to Boechera (Brassicaceae). Novon 13: 381-391[CrossRef][ISI]
Arnold M. L. 1992 Natural hybridization as an evolutionary process. Annual Review of Ecology and Systematics 23: 237-261[CrossRef][ISI]
Asker S. E. L. Jerling 1992 Apomixis in plants. CRC Press, Boca Raton, Florida, USA
Bailey C. D. R. A. Price J. J. Doyle 2002 Systematics of the halimolobine Brassicaceae: evidence from three loci and morphology. Systematic Botany 27: 318-332[ISI]
Bergman B. 1950 Meiosis in two different clones of the apomictic Chondrilla juncea. Hereditas 36: 297-320[ISI]
Bishop J. G. A. M. Dean T. Mitchell-Olds 2000 Rapid evolution in plant chitinases: molecular targets of selection in plant-pathogen coevolution. Proceedings of the National Academy of Sciences, USA 97: 5322-5327
Böcher T. 1951 Cytological and empbryological studies in the amphil-apomictic Arabis holboellii complex. Biologiske Skrifter 6: 1-58
Böcher T. 1969 Further studies in Arabis holboellii and allied species. Saertryk Af Tidsskrift 64: 141-161
Bradshaw H. D. S. M. Wilbert K. G. Otto D. W. Schemske 1995 Genetic-mapping of floral traits associated with reproductive isolation in monkeyflowers (Mimulus). Nature 376: 762-765[CrossRef]
Carman J. 1997 Asynchronous expression of duplicated genes in angiosperms may cause apomixis, bispory, tetraspory, and polyembryony. Biological Journal of the Linnean Society 61: 51-94[CrossRef]
Clauss M. J. H. Cobban T. Mitchell-Olds 2002 Cross-species microsatellite markers for elucidating population genetic structure in Arabidopsis and Arabis (Brassicaeae). Molecular Ecology 11: 591-601[CrossRef][Medline]
DellaPorta S. L. J. Wood J. B. Hicks 1983 A plant DNA minipreparation: version II. Plant Molecular Biology Reports 1: 19-21
DeWet J. 1968 Diploid-tetraploid-haploid cycles and the origin of variability in Dichanthium agamospecies. Evolution 22: 394-397[CrossRef][ISI]
Dobe C. T. Mitchell-Olds M. A. Koch 2004a Extensive chloroplast haplotype variation indicates Pleistocene hybridization and radiation of North American Arabis drummondii, A. x divaricarpa, and A. holboellii (Brassicaceae). Molecular Ecology 13: 349-370[CrossRef][Medline]
Dobe C. T. Mitchell-Olds M. A. Koch 2004b Intraspecific diversification in North American Boechera stricta (= Arabis drummondii), Boechera x divaricarpa, and Boechera holboellii (Brassicaceae) inferred from nuclear and chloroplast molecular markers: an integrative approach. American Journal of Botany 91: 2087-2101
Dobe C. T. Sharbel M. A. Koch 2005 Towards understanding the dynamics of hybridization and apomixis in the evolution of genus Boechera (Brassicaceae). Systematic Biodiversity. In press
