| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systemics and Phytogeography |
22Heidelberg Institute of Plant Sciences, Dept. of Biodiversity and Plant Systematics, Im Neuenheimer Feld 345, D-69120 Heidelberg, Germany; 3Max Planck Institute of Chemical Ecology, Beutenberg Campus, Winzerlaer Str. 10, D-07745 Jena, Germany
Received for publication January 21, 2004. Accepted for publication August 26, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Arabis • Boechera • Boechera stricta • biogeography • Brassicaceae • hybridization • microsatellites • North America • pollen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Al-Shehbaz, 2003
) strongly supported by molecular data (Koch et al., 1999
, 2000
, 2001
). We use the name B. stricta instead of Boechera drummondii as the latter is an illegitime combination (Al-Shehbaz, 2003
). Earlier studies focused on the taxonomy (Rollins, 1983
, 1993
; Mulligan, 1996
), breeding system and embryology (Böcher, 1951
, 1954
, 1969
; Roy and Rieseberg, 1989
; Roy, 1995
), origin of polyploidy and the genetic basis of apomixis (Sharbel and Mitchell-Olds, 2001
; Sharbel et al., 2004b
), hybridization (Koch et al., 2003
), and phylogeography (Dobe et al., 2004
) of these species. These investigations characterized B. stricta as morphologically and genetically variable, predominantly sexual, and diploid, but revealed B. holboellii to be highly polymorphic with respect to morphology, pollen size, and quality, cpDNA and nrDNA variation, chromosome numbers, and breeding system. Their hybrid, B. xdivaricarpa, showed similar degrees of variability as B. holboellii, but was thought to be of recent polyphyletic as well as polytopic origin.
Boechera holboellii can reproduce sexually or via apomixis. Asexual genotypes of B. holboellii may be triploid, aneuploid, or diploid (Böcher, 1951
, 1954
, 1969
; Sharbel et al., 2004a
, b
), hence agamospermy is not necessarily associated with polyploidy, as is usually the case in plants (Gustafsson, 1946
, 1947; Mogie, 1986
; Richards, 1996
). However, as demonstrated by the investigations of Böcher (1951
, 1954
, 1969
), apomictic reproduction is mostly associated with asyndetic pollen meiosis and macrosporogenesis, respectively. Furthermore, Roy (1995)
found substantial allozyme variation and fixed heterozygosity in progeny arrays for diploid and polyploid individuals, which were assumed to be apomicts. The above finding suggests that these apomicts are hybrids between different genotypes. Crosses may occur between closely related genotypes as well as distinct lineages, as both autopolyploidization and hybridization between morphologically distinct taxa were supposed as a mode of origin of apomictic B. holboellii samples (Böcher, 1954
, 1969
). Analysis of nrDNA ITS polymorphism has recently shown that extensive reticulation has played an important role in the evolution of this polymorphic taxon (Koch et al., 2003
). However, that study did not focus to differentiate between the potential roles of intra- and interspecific crosses to explain the observed intraindividual ITS polymorphisms.
Although nrDNA ITS sequences have identified hybrid accessions in several cases (e.g., hybrid B. xdivaricarpa resulting from introgression of B. stricta into B. holboellii), the ITS marker has limited resolution to estimate genetic similarities of hybrids and their parents (Koch et al., 2003
), because parental sequence types may become fixed in hybrids via concerted evolution. For instance, 57 (50%) B. xdivaricarpa accessions in that study carried B. stricta-like ITS types, although they were in fact morphologically intermediate between their parents. Therefore, we employed microsatellites (Goldstein and Schlötterer, 1999
; Balloux and Lugon-Moulin, 2002
) as biparentally inherited markers to estimate genetic distances among and within B. xdivaricarpa, B. stricta, and B. holboellii. This analysis is combined with data from the maternally inherited chloroplast genome, which can follow maternal lineages and trace migration in space and time (Avise, 2000
; Hare, 2001
; Hewitt, 2001
; Hewitt and Ibrahim, 2001
).
For this purpose a phylogeographic model using cpDNA markers has been elaborated (Dobe et al., 2004
) and several post-glacial recolonization events for B. stricta and B. holboellii have been inferred for areas formerly covered by the Cordilleran and Laurentide ice shield of North America (Richmond and Fullerton, 1986
; Brouillet and Whetstone, 1993
). The phylogeographic analysis revealed geographic separation of derived B. holboellii-like chloroplast haplotypes in the Sierra Nevada and Southern Rocky Mountains from their ancestors in the Central and Northern Rocky Mountains and adjacent Cascades. The Rocky Mountains also showed highest values of genetic variation, and, therefore, were assumed to represent a center of diversity. In addition, the study proposed further evidence for glacial refugia in boreal and arctic regions of Canada and the adjacent United States. Here we interpret nuclear DNA variation (nrDNA ITS sequences, microsatellite length polymorphism) and pollen size data in the context of these inferred historic events and geographic patterns.
For practical reasons it is not possible to estimate ploidy levels or chromosome numbers of herbarium specimens by karyological means. However, in the past indirect methods such as correlation of pollen size with ploidy were successfully applied to circumvent this problem (e.g., Ehrendorfer, 1949
; Stepánková, 1993
; Vilhar et al., 2002
; Lihová and Marhold, 2003
). Sharbel et al. (2004b)
and Koch et al. (2003)
demonstrated that pollen of these three Boechera taxa occur in two main size classes, apparently representing somatic diploids and polyploids, respectively. As considerable intraindividual variation in pollen size was found by Böcher (1951)
and Roy (1995)
, we compared pollen size with the maximum number of homologous alleles observed at seven microsatellite loci, in order to determine whether allele number is correlated with inferred ploidy levels.
Herein we focus on several questions: (1) Can pollen size data be used to estimate ploidy levels of herbarium specimens? (2) Do nuclear microsatellites allow to discriminate between B. stricta and B. holboellii? (3) Do nuclear microsatellites support the polyphyletic hybrid origin of B. xdivaricarpa? (4) What is the potential role of hybridization in the origin of B. holboellii? (5) Are phylogeographic inferences based upon cpDNA polymorphism in accord with microsatellite variation?
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Dobe et al., 2004
) and future studies. For computation purposes and presentation of data, origins noted on the voucher labels were translated into geographic coordinates. The majority of the material was either revised or collected by the monographers of the genus, Reed C. Rollins, and Gerald A. Mulligan..
|
(AY165313-AY165439). However, additional sequences from six accessions (Arab0105, -109, -196, -315, -392, -703) were added in this study (403 accessions in total). For DNA extraction, amplification and sequencing of the nuclear ribosomal DNA (ITS1, 5.8 S rDNA gene, ITS2), and cloning of primary PCR products into the pGEM-T-easy cloning vector (PROMEGA) we applied protocols of Koch et al. (2003)
. Sequences were processed by hand and added to the alignment performed by Koch et al. (2003)
(accessible via the Molecular Biology and Evolution homepage: http:// mbe.oupjournals.org). For comparison of the nrDNA-based phylogeny with phylogenetic relationships inferred from chloroplast markers (trnL intron and trnL-trnF intergenic spacer) we used data published in Dobe et al. 2004
.
Analysis of microsatellite length polymorphism, polymerase chain reaction, and allele calling
70 B. stricta, 97 B. xdivaricarpa, and 131 B. holboellii accessions were analyzed for length polymorphism at eight microsatellite loci (Table 1). Specific primers were either taken from Clauss et al. (2002)
or designed in this study using Primer3 software (Rozen and Skaletsky, 2000
) for sequences of microsatellite containing nuclear DNA clones isolated from Arabis fecunda (A. Pepper, M. Koch and T. Mitchell-Olds, unpublished data). For visualization of fragments IRD700 and IRD800 5'-end-labeled forward primers were used (MWG Biotech). We performed 50 µL PCR reactions for each locus separately in a master mix containing 1x PCR buffer (10 mmol/L TRIS/50 mmol/L KCl buffer, pH 8.0), 3 mmol/L MgCl, 0.4 µmol/L of each primer, 0.2 mmol/L of each dNTP, 1 unit Taq DNA polymerase (Schott-Eppendorf), and approximately 1 ng of template DNA using an ABI 9700 (Applied Biosystems) or PTC 200 (Biozym) thermal cycler. Thermal cycling started with a denaturation step at 95°C lasting 5 min followed by 35 cycles each comprising 60 s denaturation at 95°C, 45 s annealing at 46°C (locus d3)/50°C (c8, e9)/54°C (a1, a3, b6, ICE4, SLL2), 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. In order to confirm these inferred genotypes, reactions were rerun by independent PCR amplification 2–3 times. IRD700 and IRD800 labeled PCR products were multiplexed for fragment analysis on a LI-COR L4200S–2 automated sequencer. Gel images were analyzed using Cross Checker 2.91 software (http://www.dpw.wau.nl/pv/pub/CrossCheck/index.html) for estimation of allele sizes. In order to determine the location of the analyzed microsatellite loci in the Arabidopsis thaliana genome, sequence comparisons were performed at TAIR (The Arabidopsis Information Resource; http://www.arabidopsis.org/).
Phylogenetic inference
In order to resolve genealogical relationships between the two parental species and their hybrid, a phylogenetic analysis was performed on all nrDNA ITS types (N = 45) from individuals showing no evidence of intraindividual ITS polymorphism. NrDNA ITS sequences from Cusickiella douglasii (AF146515), Cusickiella quadricostata (AF146514), Halimolobos perplexa ssp. perplexa (AJ232926), H. perplexa ssp. lemhiensis (AJ232927), and Polyctenium fremontii (AF183109) served as outgroups. Sequence types reported in this study for the first time (AY165367, AY165372, AY165373, AY165408, AY457928-AY457932), were submitted to GenBank. The maximum parsimony optimality criterion was applied for reconstruction of the phylogeny using PAUP version 4.0b10 (Swofford, 2002
). The heuristic search algorithm was chosen, using the RANDOM ADDITION of taxa and the TBR option (tree bisection-reconnection) for branch swapping, which was restricted to 10 000 retained trees. A strict consensus tree was constructed from the 10 000 shortest trees. Bootstrapping was carried out on 1000 replicates using the HEURISTIC search option.
Assessment of species-specificity of microsatellite alleles
Alleles were considered to be species-specific based upon a frequency of 90% in a single taxon. Singletons were not evaluated. specificstric = allelesstric*1.42*100/(allelesstric*1.42 + allelesholb*0.76) 
90; specificholb = allelesholb*0.76*100/(allelesstric*1.42 + allelesholb*0.76) 90; and specificdiva = allelesdiva*1.02*100/ (allelesstric*1.42 + allelesdiva*1.02 + allelesholb*0.76) 90 (allelesxy = number of alleles observed for species xy).
Genetic distances based on ITS polymorphism
Considering only single nucleotide polymorphisms (SNPs), pairwise uncorrected P-distances were calculated for all nrDNA ITS types (N = 136) using PAUP version 4.0b10, and subsequently processed in a principal coordinate analysis (PCoA) using SYNTAX 5.02 software (Podani, 1994
).
Genetic distances based on microsatellites
Pairwise genetic distances between individuals as well as among major land features of southwestern North America were calculated based on allelic microsatellite variation using the (
µ)2 measure proposed by Goldstein et al. (1995)
. Based on the observed dissimilarity matrix of individuals, a PCoA using SYNTAX 5.02 was performed for all three species investigated. In order to demonstrate genetic similarity among 66 B. holboellii accessions of known ploidy level, a neighbor-joining analysis using MEGA 2.1 (Kumar et al., 2001
) was carried out on a reduced data matrix. Six geographic regions corresponding to major land features south of the last glacial maximum (LGM) of North America (Cascades, Sierra Nevada, Great Basin, Rocky Mountains [North, Middle, and South]; cf. Dobe et al., 2004
) were defined to analyze geographic patterns of genetic variation in B. holboellii. In order to obtain geographic units comprising approximately equal numbers of individuals, major regions were further divided into 12 subregions (finally comprising between 7 and 13 accessions each). Relationships between geographic regions were resolved running a neighbor joining analysis on pairwise (µ)2 distances between subregions using MEGA 2.1 (Kumar et al., 2001
).
Genetic diversity measures
Geographic distribution of microsatellite diversity was estimated for B. holboellii based on the 12 subregions described above. Effective diversity of alleles (va) was calculated for individual loci for each (sub)region according to (Gregorius, 1978
, 1987
). In order to test for an autopolyploid vs. allopolyploid origin of B. holboellii, all accessions were compared for nucleotide diversity (
; Nei, 1987
) of the nrDNA ITS based upon their ploidy levels. Intraindividual diversity was estimated for all three species in calculating the mean number of observed allelic variants per locus over all eight microsatellite loci investigated.
Relationship of ploidy level and pollen size
Correspondence of ploidy level and pollen size was estimated from the Pearson correlation of mean number of observed allelic microsatellite alleles per locus and the canonical discriminant function 1 of pollen length and width using SPSS for Windows (Release 11.0.1. 2001, SPSS Inc., Chicago, Illinois, USA)
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, this makes 136 variants, which were taxonomically distributed as follows: B. stricta with 16, B. holboellii with 70, and B. xdivaricarpa with 80 ITS types (see Table 2 in Supplemental Data accompanying the online version of this article).
|
|
|
|
|
µ)2 distances between both parental species and their hybrid explained 41.6%, 24.5%, and 16.3% of the total variance. The analysis discriminated B. holboellii from the majority of B. stricta accessions, but did not separate B. xdivaricarpa from either parent (Fig. 5). However, eight B. stricta accessions (Arab0008, -248, -299, -334, -339, -730, -778, -803) did not group within the "B. stricta-cluster." Interestingly, all these accessions (except Arab803) were characterized by mean numbers of alleles per locus higher than average (1.14 to 2.25, x = 1.12; Fig. 3). Coordinates 1 and 2 distinguished a subset of B. stricta and B. xdivaricarpa accessions which correlated with the northeastern range distribution of these taxa (ranging from Saskatchewan to the Atlantic coast). The first three coordinates of the PCoA using pairwise p-distances between nrDNA ITS types explained 23.03%, 11.28%, and 6.74% of the total variance (Fig. 6). All B. stricta accessions grouped within one cluster, which was set apart from a second cluster to which the majority of B. holboellii accessions were assigned (Fig. 6a, c). However, several B. holboellii ITS sequences belonged to the "B. stricta-cluster." In contrast, B. xdivaricarpa comprised the whole genetic variation found in both parental species. Plotting pollen size classes onto the PCoA showed that small-pollen (diploid) B. holboellii were restricted to one pole of the "B. holboellii-cluster" (Fig. 6c). All ITS types falling within the range of genetic variation of these diploids (except type BT and BX; see Fig. 1 in Supplemental Data accompanying the online version of this article) can be distinguished by a thymine at alignment position 408 (Fig. 7; cf. Koch et al., 2003
) from all remaining ITS types of the "B. holboellii-cluster." However, most of those ITS types were not unique to diploid B. holboellii, but occurred also in polyploids (see Table 2 in Supplemental Data accompanying the online version of this article).
|
|
|
µ)2 distances between 12 geographical subregions corresponding to major land features south of the last glacial maximum (LGM) in western North America identified partitioning of genetic variation between the northern and southern United States (Fig. 8). Effective diversity of microsatellite alleles (va, sensu Gregorius, 1978
) within subregions ranged between 13.6 and 21.1 in the northern area, and between 19.6 and 29.3 in the south. An obvious geographic trend existed between the two regions with respect to intraindividual genetic diversity, as mean numbers of allelic variants per locus increased toward the south (Fig. 9). For example, all individuals showing values higher than 2.5 were restricted to the south, especially to the Sierra Nevada, while individuals with values below 1.5 predominated in the north. Expressed in terms of ploidy levels, the majority of inferred polyploid accessions occurred in the southern region. In contrast, diploids (as inferred from pollen size) prevailed in the northern United States, but became increasingly rare towards the south.
|
|
; Nei, 1987
) of nrDNA ITS sequences was compared for diploid and polyploid B. holboellii and B. stricta. Diversity was lowest for diploid ( = 0.020 ± 0.014) and approximately twice as high in polyploid B. holboellii ( = 0.048 ± 0.028). In comparison was 0.034 ± 0.022 in B. stricta. Out of 36 ITS types found in B. holboellii, for which ploidy could be inferred, 10 and 24 types were specific to diploids and polyploids, respectively (Fig. 7). In terms of mutations, 10/5 and 15/2 SNPs/insertions, respectively, characterized these to classes of cytotypes. For example, an adenosine at alignment position 184 was found in 14 (58%) ITS types of polyploids, but in no diploid individual. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. Assuming outbreeding and allelic polymorphism within populations, one would expect a correlation between number of alleles per individual and pollen size. Observed pollen sizes occur in two main classes corresponding to mean pollen widths of 14–18 µm (somatic diploid) and 18–26 µm (somatic polyploid), respectively (Koch et al., 2003
; Sharbel et al., 2004b
). A third size class (8–15 µm) represented abnormal, degraded pollen grains, which were excluded from these analyses. Comparison of observed maximum number of allelic variants per locus and pollen size is summarized in Table 3. All individuals producing small pollen expressed at most one or two allelic variants per locus, while individuals with three or four allelic variants at a locus were characterized by large pollen. However, several individuals of all three species producing large pollen were not inferred to be polyploids according to microsatellite data. Although inbreeding is a likely explanation, alternatively this observation may indicate unreduced pollen, as Sharbel et al. (2004b)
recently demonstrated that apomictic diploid B. holboellii produces large pollen. This latter finding is in agreement with the observation by Böcher (1951
, 1954
) and Roy (1995)
that some diploid B. holboellii had dyadic pollen formation. It is interesting in this context that all B. stricta and B. holboellii accessions producing large pollen were heterozygous for at least one locus, while all homozygotes were characterized by small pollen. Roy (1995)
found fixed heterozygosity in progeny arrays of diploid B. holboellii, which is a strong indication of apomixis. These results indicate that small pollen is associated with diploidy, but large pollen does not provide sufficient evidence to judge the ploidy level of its bearer. However, individuals characterized by three or four alleles per locus are likely polyploids, as they produce only large pollen. The relationship between mean number of allelic variants per locus and pollen size is displayed in Fig. 10 and shows a statistically significant correlation (r2 = 0.66, N = 182, P <0.01).
|
|
, that B. stricta is almost exclusively diploid, as only three individuals (out of 42) produced large pollen (Fig. 10). Furthermore, the high percentage of homozygotes (69% = 48/70; Fig. 3) also suggests that B. stricta reproduces predominantly via self-fertilization (Roy, 1995
). In contrast, only one B. xdivaricarpa accession (out of 97) was homozygous for all loci. Furthermore, this taxon produced only large pollen. Both results are in accordance with the predominance of polyploidy (triploidy, tetraploidy) in this species, and its assumed apomictic mode of reproduction (Rollins, 1983
, 1993
). The pattern for B. holboellii was intermediate between the two former species. Individuals with both small and large pollen were detected with high frequencies, as well as homozygotes (N = 22) and heterozygotes (N = 109). The observed frequency of homozygous individuals is substantial, as Roy (1995)
found homozygosity to be extremely rare in B. holboellii. 84% (65/ 77; Table 3) accessions produced pollen as expected from the maximum number of allelic microsatellite variants per locus, confirming the existence of diploid and polyploid cytotypes within this species (Böcher, 1951
, 1969
; Roy and Rieseberg, 1989
; Roy, 1995
; Sharbel and Mitchell-Olds, 2001
; Sharbel et al., 2004b
). Microsatellite data did not provide evidence for tetraploid B. holboellii, in accordance with the results of Sharbel and Mitchell-Olds (2001)
and Sharbel et al. (2004b)
, showing that polyploidy is predominantly expressed on the triploid level in this species.
Genetic differentiation between B. stricta and B. holboellii
Based upon microsatellite variation (Fig. 5) and nrDNA sequences (Figs. 4 and 6) the majority of B. stricta and B. holboellii accessions are genetically distinct from each other. Differentiation was also supported by cpDNA polymorphism (Dobe et al., 2004
), which was largely partitioned between the two taxa. Analogous findings were reported by Sharbel et al. (2004a)
for sequence variation on two nuclear loci and for genome size differences (Sharbel and Mitchell-Olds, 2001
). The discrimination of B. stricta and B. holboellii by genetic markers is also in accordance with their marked morphological divergence (Rollins, 1993
; Mulligan, 1996
). However, the preceding studies also indicate occasional bi-directional introgression between these two species. Interestingly, the majority of data favor gene flow from B. holboellii into B. stricta rather than vice versa. Evidence comes from (1) B. stricta accessions Arab0248, -299, and -803 (Fig. 5) bearing B. stricta-like chloroplast haplotypes, but possessing also several microsatellite alleles (a1_237, b6_308, c8_238, d3_206; Fig. 1) typical for B. holboellii. (2) One B. stricta accession (Arab0728) out of 173 was found by Dobe et al. (2004)
to bear a B. holboellii-like chloroplast haplotype (haplotype CW), while several B. holboellii individuals carried chloroplast types typical for B. stricta. (3) Superiority of B. holboellii as pollen donor in crosses with B. stricta seems likely, as B. holboellii was father of 73% (149/205) of all investigated B. xdivaricarpa accessions (Dobe et al., 2004
). These findings were unexpected as pollen quality is often reduced in B. holboellii (Koch et al., 2003
), and self-fertilization at an early stage of flowering was suggested to predominate in B. stricta (Roy, 1995
).
Nine out of 188 B. holboellii accessions (this study and Koch et al., 2003
; Arab0046, -87, -89, -237, -317, -360, -678, -695, -749) carried typical B. stricta nrDNA ITS types (types e, l, r, v, w, x, az, and gb) and were accordingly grouped with this species in the PCoA (Fig. 6c). However, this marker is of little reliability to estimate overall genetic constitution of these individuals, as any of the original parental alleles or their recombinants may become fixed in hybrids via concerted evolution (Arnheim et al., 1980
; Dover, 1982
; Avise, 1994
; Koch and Al-Shehbaz, 2000
; Koch et al., 2003
). Accordingly, the only two accessions investigated for microsatellite polymorphism (Arab0678, -749) did not join B. stricta in the (µ)2 based PCoA but clustered with B. holboellii. However, the occurrence of B. stricta-like ITS types in B. holboellii clearly indicates reticulation, possibly via backcrosses with B. xdivaricarpa. ITS data did not favor any direction of introgression, as both B. holboellii, carrying B. stricta-like (AS, AT, BJ) and B. holboellii-like chloroplast haplotypes (BY, CC, CG) were found.
East-west geographic differentiation of B. stricta
The microsatellite PCoA distinguished B. stricta (Arab0008, -334, -339, -730, -778) and B. xdivaricarpa (except Arab0367, -368) of northeastern North America (Saskatchewan to the Atlantic coast) from all remaining individuals (Fig. 5). Interestingly, B. stricta sampled from this area exhibited an unusually high mean number of alleles per locus (1.50–1.88, x = 1.12; Fig. 3) and two of these individuals may be triploids. Although these individuals shared alleles commonly found in this species, additional variants were detected that otherwise only occurred in B. xdivaricarpa or B. holboellii (Fig. 1; a3-290*, -291; b6-305, -328*, -330*, -336; c8-233*, -234*; d3-191*, -206,-220*, -221; ICE4-176*). Furthermore, all alleles indicated by an asterisk were geographically restricted to northeastern North America. This observation is of special interest, as three out of four individuals investigated for nrDNA ITS sequence variation carried ITS type e, which had a geographic distribution (estimated from a total of 41 samples) largely confined to this area. Analogously, only one chloroplast haplotype (AS) was found in B. stricta, which dominated (aside from a few singletons) the whole northeast of North America. Furthermore, all alleles confined to the area (except c8-234) were, aside from B. stricta, almost exclusively found in B. xdivaricarpa.
One possible explanation for the origin of these aberrant genotypes could be hybridization of B. stricta with a third, unknown species. However, such a scenario is unlikely due to lack of morphologically distinct features and the derived phylogenetic positions of ITS type e (Fig. 4) and chloroplast haplotype AS (Dobe et al., 2004
) within well-characterized "B. stricta"-clades. Alternatively, the new alleles may have evolved after isolation of B. stricta populations during the Pleistocene. This process which might have been accelerated by genetic drift and apomixis (often accompanied by polyploidization), leading to the fixation of heterozygotes, and possibly increased mutation rates due to heterozygote instability (Amos, 1999
). However, the extent to which heterozygosity increases genetic variation in parthenogenetic populations is poorly understood (Simon et al., 2003
). Nevertheless, the observed homogeneity of the area and the genetic similarities between B. stricta and B. xdivaricarpa with respect to nuclear and chloroplast DNA markers may be explained by post-glacial recolonization from refugia in the northeast, and a recent hybrid origin. Glacial survival of Boechera along the southeastern margin of the continuous ice shield may also apply to the only two B. xdivaricarpa accessions found in this area (Arab0367, -368; Great Lakes), which showed an allelic microsatellite constitution typical for their western relatives. Both possess chloroplast haplotypes suggestive of ice age relicts (Dobe et al., 2004
). Alternatively, recent long-distance dispersal might explain their distribution.
In accordance with our results, Böcher (1969) counted tetraploid B. stricta from Massachusettes, whereas the rest of his counts all are diploids. Therefore, it should be also considered that the northeastern North American populations may represent a different taxon (or even a different species).
Evolution of B. holboellii
The gene phylogeny of B. holboellii using nuclear ribosomal ITS sequences (Fig. 4) was incongruent with phylogenetic relationships inferred from cpDNA markers (trnL intron/trnL-trnF intergenic spacer; Dobe et al., 2004
). Thus, B. holboellii-like haplotypes of the three evolutionarily distinct cpDNA lineages occurred interspersed among the five major clades of the ITS tree (Fig. 4), which suggests extensive reticulation between lineages. The importance of hybridization has already been demonstrated by the frequent occurrence of ITS polymorphism within individuals (Koch et al., 2003
) and may explain the poor resolution of the ITS tree. With respect to ploidy levels, the phylogenetic analysis revealed that diploids share all common sequence types (types h, ac, bt) with polyploids (Fig. 4), indicating that they repeatedly gave rise to polyploid descendants. An analogous picture results from the microsatellite neighbor-joining analysis (Fig. 11). However, considering the whole data set, diploid and polyploid B. holboellii were substantially differentiated based on both ITS types and nucleotide variation. Thus, 10 and 24 ITS types out of 36, characterized by 10/5 and 15/2 SNPs/insertions, were specific to diploids and polyploids, respectively (Fig. 7). However, all "diploid-specific" nucleotides were observed at low frequencies (N = 1–3), while the only frequent ITS type, type h, lacked any unique mutations. These differences result in substantial division of diploid from the polyploid B. holboellii accessions in the PCoA (Fig. 6c). These findings suggest a common evolutionary factor involved in the origin of polyploids. Sharbel and Mitchell-Olds (2001)
inferred repeated origin of triploid B. holboellii, possibly triggered by environmental influence (e.g., cold-shock). Although this hypothesis is suitable to explain the establishment of polyploids with ITS similar to diploid relatives, it does not apply to ITS sequences with polyploid-specific mutations (e.g., an adenosine or guanosine at alignment positions 184 and 408, respectively). This argument becomes more important considering the occurrence of these specific nucleotides among all three major cpDNA lineages. Two scenarios might explain the observed genetic pattern: (1) An unknown sexual, tetraploid lineage(s) within B. holboellii serving as pollen donor, or (2) repeated origin of polyploids resulting from hybridization with a third taxon. The first explanation is not well supported, as most B. holboellii polyploids are triploid apomicts (Böcher, 1951
, 1954
, 1969
; Roy, 1995
; Sharbel and Mitchell-Olds, 2001
; Sharbel et al., 2004b
, this study). However, some tetraploids are known from karyological investigations (Rollins, 1941
; Roy, 1995
). However, at this time there is no evidence for such an independent evolutionary lineage within B. holboellii. The second scenario is likewise speculative, and has to be considered on a broader taxonomic scale. However, nucleotide diversity () was twice as high in polyploid B. holboellii, and this indicates the involvement of distinct genetic lineages in their establishment. It is interesting in this context, that 36 out of 41 ITS sequences, respectively 29 out of 30 copies of the most common ITS type specific to diploid B. holboellii, type h, were detected in individuals carrying haplotypes of chloroplast lineage III (cf. Dobe et al., 2004
; cpDNA haplotypes BU, BX, BY, CB, CC, CF, CG, CI, CR, CU). This significant correlation may be considered as evidence, that these accessions represent a basal evolutionary unit involved in the diversification of the taxon. This view is supported by morphology as 36 out of the 41 ITS sequences were carried by accessions classified as B. holboellii var. retrofracta.
|
et al. (2004)
allocated the majority of ancestral haplotypes in the Central and Northern Rocky Mountains of the United States and adjacent Cascades, while their derivatives (haplotypes D, F, S, U + descendants) prevailed in the Southern Rockies and the Sierra Nevada. This pattern suggests a more recent colonization of the Sierra Nevada. Remarkably, the majority of polyploid accessions fall within the combined distribution area of these derived chloroplast haplotypes. This pattern may indicate relatively recent establishment, perhaps associated with apomixis, of polyploids in the Sierra Nevada and Southern Rocky Mountains. As with ribosomal DNA, genetic diversity based on allelic microsatellite variation was higher in polyploids than in diploids, which might suggests hybridization in the establishment of polyploids. However, this result should be treated with care, as mutation processes in microsatellites are complex (Treuren et al., 1997
; Estoup et al., 2002
; Vigouroux et al., 2003
), and increased length polymorphism may alternatively result from heterozygote instability (Amos, 1999
; Hancock, 1999
) or be associated with directional evolution due to increased mutation rates (Vigouroux et al., 2003
).
Hybrid origin of B. xdivaricarpa based upon microsatellite markers
Both ITS sequences (Fig. 5) and microsatellites (Fig. 6b) demonstrated that B. xdivaricarpa closely resamples genetic variation of its parental taxa, B. stricta and B. holboellii, as previously shown by Dobe et al. (2004)
using cpDNA markers. Furthermore, recent origin of the hybrid, as suggested by the same authors, is in accordance with the low number of B. xdivaricarpa-like microsatellite alleles (Fig. 2). Therefore, a hybrid origin of B. xdivaricarpa was unequivocally supported by three independent molecular marker systems. However, given the high morphological variability of the hybrid (Rollins, 1993
) it seems possible that it has evolved from parent(s) other than B. holboellii, in addition to the B. stricta, which seems to be always one of the parents. Therefore, in future the origin of B. xdivaricarpa should be addressed on a broader taxonomic scale.
Conclusions
Comparison of pollen size data and information on the maximum number of microsatellite alleles per locus indicates that small pollen is associated with diploidy, while large-pollen individuals show heterogeneous ploidy levels. However, individuals assumed to be polyploids from the observed maximum number of homologous alleles produced large pollen as expected. These data show a significant correlation between ploidy and pollen size, which can be used to estimate ploidy levels of herbarium specimens. Allelic microsatellite length variation proved a valuable tool for deeper understanding of evolutionary processes underlying the diversification of the model species B. stricta, B. xdivaricarpa, and B. holboellii. The data suggest occasional introgression of B. holboellii into B. stricta, which was not apparent from nrDNA ITS sequence information. We also found evidence for distinct B. stricta genotypes in northeastern North America, probably representing isolated glacial refugia. Aside from occasional hybridization, both nuclear markers indicate that B. stricta and B. holboellii are clearly distinct taxa. As expected from nrDNA ITS and cpDNA (trnL intron, trnF-trnL IGS) sequences, microsatellite data supported a recent and polyphyletic origin of B. xdivaricarpa. Combined data on pollen size, microsatellites, nDNA, and cpDNA sequences provided evidence for a basal evolutionary unit within B. holboellii represented by diploid, presumably sexual individuals. Hybridization of diploids with genetically distinct lineage(s) is likely to explain the high nucleotide diversity found for polyploid B. holboellii. Finally, geographical correspondence of phylogenetically derived chloroplast haplotype lineages with higher ploidy levels in the southern United States provides an interesting model to study the effects of hybridization and apomixis on the radiation of B. holboellii.
|
FOOTNOTES |
|---|
4 marcus.koch{at}urz.uni-heidelberg.de
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Amos W. 1999 A comparative approach to the study of microsatellite evolution. In D. B. Goldstein and C. Schlötterer [eds.], Microsatellites: evolution and applications, 66–79. University Press, Oxford, UK
Arnheim N. R. Krystal R. Schmickel G. Wilson O. Ryder E. Zimmer 1980 Molecular evidence for genetic exchanges among ribosomal genes on nonhomologous chromosomes in man and apes. Proceedings of the National Academy of Sciences, USA 77: 7323-7327
Avise J. C. 1994 Molecular markers, natural history and evolution. Chapman and Hall, New York, USA
Avise J. C. 2000 Phylogeography: the history and formation of species. Harvard University Press, Cambridge, UK
Balloux F. N. Lugon-Moulin 2002 The estimation of population differentiation with microsatellite markers. Molecular Ecology 11: 155-165[CrossRef][Medline]
Böcher T. W. 1951 Cytological and embryological studies in the amphi-apomictic Arabis holboellii complex. Det Kongelige Danske Videnskabernes Selskab Biologiske skrifter 6: 1-59
Böcher T. W. 1954 Experimental taxonomical studies in the Arabis holboellii complex. Svensk Botanisk Tidskrift 48: 31-44
Böcher T. W. 1969 Further studies in Arabis holboellii and allied species. Saertryk Botanisk Tidskrift 64: 141-161
Brouillet L. R. D. Whetstone 1993 Climate and physiography. In Flora of North America Editorial Committee [ed.], Flora of North America, vol. 1, 15–46. Oxford University Press, Oxford, UK
Clauss M. J. H. Cobban T. Mitchell-Olds 2002 Cross-species microsatellite markers for elucidating population genetic structure in Arabidopsis and Arabis (Brassicaceae). Molecular Ecology 11: 591-601[CrossRef][Medline]
Dobe C. T. Mitchell-Olds M. Koch 2004 Extensive chloroplast haplotype variation indicates Pleistocene hybridization and radiation of North American Arabis drummondii, A. xdivaricarpa, and A. holboellii (Brassicaceae). Molecular Ecology 13: 349-370[CrossRef][Medline]
Dover G. 1982 Molecular drive: a cohesive mode of species evolution. Nature 299: 111-117[CrossRef][Medline]
Ehrendorfer F. 1949 Zur Phylogenie der Gattung Galium. I. Polyploidie und geographisch-ökologische Einheiten in der Gruppe des Galium pumilum Murray (Sekt. Leptogalium Lange sensu Rouy) im österreichischen Alpenraum. Österreichisch Botanische Zeitschrift 96: 109-138[CrossRef]
Estoup A. P. Jarne J.-M. Cornuet 2002 Homoplasy and mutation model at microsatellite loci and their consequences for population genetics analysis. Molecular Ecology 11: 1591-1604[CrossRef][Medline]
Goldstein D. B. A. Ruiz Linares L. L. Cavalli-Sforza M. W. Feldman 1995 Genetic absolute dating based on microsatellites and the origin of modern humans. Proceedings of the National Academy of Sciences, USA 92: 6723-6727
Goldstein D. B. C. Schlötterer 1999 Microsatellites: evolution and applications. Oxford University Press, New York, USA
Gregorius H.-R. 1978 The concept of genetic diversity and its formal relationship to heterozygosity and genetic distance. Mathematical Biosciences 41: 253-271[CrossRef][ISI]
Gregorius H.-R. 1987 The relationship between the concepts of genetic diversity and differentiation. Theoretical and Applied Genetics 74: 397-401[CrossRef][ISI]
Gustafsson Å. 1946 Apomixis in higher plants. Part I. The mechanism of apomixis. Lunds Universitets Arsskrift N.F. Avdelningen. 2 42: 1-67
Gustafsson Å. 1946 Apomixis in higher plants. Part II. The causal aspect of apomixis. Lunds Universitets Arsskrift N.F. Avdelningen. 2 43: 69-179
Hancock J. M. 1999 Microsatellites and other simple sequences: genomic context and mutational mechanisms. In D. B. Goldstein and C. Schlötterer [eds.], Microsatellites: evolution and applications, 1–9. University Press, Oxford, UK
Hare P. H. 2001 Prospects for nuclear gene phylogeography. Trends in Ecology and Evolution 16: 700-706[CrossRef]
Hewitt G. M. 2001 Speciation, hybrid zones and phylogeography—or seeing genes in space and time. Molecular Ecology 10: 537-549[CrossRef][Medline]
va = sum of effective allelic diversities sensu Gregorius (1978)
