Use of AFLP and RAPD molecular genetic markers and cytogenetic analysis to explore relationships among taxa of the Patagonian Bromus setifolius complex

Bromus setifolius var. pictus (Hook) Skottsb., B. setifolius var. setifolius Presl. and B.setifolius var. brevifolius Ness are three native Patagonian taxa in the section Pnigma Dumort of the genus Bromus L. AFLP and RAPD analysis, in conjunction with genetic distance measurements and statistical techniques, revealed variation within this group and indicated that B. setifolius var. brevifolius was closely related to B. setifolius var. pictus, with both taxa being more distantly related to B. setifolius var. setifolius. Cytogenetic analysis confirmed the chromosomal number of B. setifolius var. pictus (2n = 70) and B. setifolius var. setifolius (2n = 28) and showed for the first time that B. setifolius var. brevifolius had 2n = 70. The combination of molecular genetic and cytogenetic evidence supported a species status for two of the three taxa and suggested hypotheses for the evolutionary origin of these complex taxa. Species status was also indicated for B. setifolius var. setifolius. Based on these findings, we suggest that B. setifolius var. pictus be referred to as B. pictus Hook var. pictus, and B. setifolius var brevifolius as B. pictus Hook var brevifolius. The correlation between AFLP diversity and variation in ecological parameters suggested that this marker system could be used to assess breeding progress and to monitor the domestication of Patagonian Bromus species for agronomic use.


Introduction
The genus Bromus L. contains approximately 150 species grouped into six taxonomic sections that are distributed throughout temperate regions of the world (Clayton and Renvoize, 1986). The section Pnigma includes about 60 species that are disseminated across Eurasia and North and South America (Smith, 1970). In Patagonia of South America, the section Pnigma is represented by B. setifolius Presl., which has been considered as a complex taxon because of its extensive morphological variation (Camara Hernandez, 1978;Matthei, 1986;Gutierrez and Pensiero, 1998). Three varieties of this species are recognized, namely, B. setifolius var. setifolius, B. setifolius var. pictus (Hook. F.) Skottsberg and B. setifolius var. brevifolius Ness.
The semi-desert region of Patagonia is a large environmental niche that has been subjected to substantial degradation (Aguiar and Sala, 1998). Bare soils represent 60% of the total area, and a recent history of over-grazing has accelerated the process of desertification. Members of the B. setifolius complex could play an important role in the cessation and reversal of this environmental degradation (Aguiar and Sala, 1997). This is particularly true of B. setifolius var. pictus, which shows a favorable combination of drought resistance, high recruitment capacity and forage nutritive value. Bromus setifolius var. pictus and B. setifolius var. setifolius are sympatric inhabitants of the Patagonian semi-desert, whereas B. setifolius var. brevifolius is found in wetter grass steppe environments of the sub-Andean region (Naranjo et al., 1990;Gutierrez and Pensiero, 1998). Naranjo et al. (1990) performed a biosystematic study of the two sympatric taxa B. setifolius var. setifolius and B. setifolius var. pictus. Numerical taxonomy based on cluster analysis and principal components analysis (PCA) of 35 exomorphological characters detected two phenetically distinct groups that corresponded to B. setifolius var. setifolius and B. setifolius var. pictus. Cytological examination of mitosis demonstrated that B. setifolius var. setifolius plants had a chromosomal composition of 2n = 4x = 28, whereas B. setifolius var. pictus plants had a chromosomal composition of 2n = 10x = 70. Other cytogenetic differences between the two taxa included the karyotypic formula, type of satellite chromosome, basic genome length, asymmetry, C-banding patterns, and meiotic chromosomal behavior.
Meiotic chromosomal pairing in B. setifolius var. setifolius indicated the formation of 14 bivalents, typical of an allotetraploid. In contrast, chromosomal pairing in B. setifolius var. pictus suggested segmental allodecaploid behavior, with the formation of 35 bivalents being the most frequent configuration. With regard to the breeding systems, cleistogamous and chasmogamous flowers were found in B. setifolius var. setifolius whereas only cleistogamous flowers were observed in B.setifolius var. pictus; morphological and cytological analyses revealed no hybrid individuals. The stable morphological features of both taxa, which paralleled the karyotypic differences between them, suggested that they should be recognized as separate species (Naranjo et al., 1990). Although several cytogenetic studies have been reported for taxa of the B. setifolius complex (Covas and Schnack, 1946;Stebbins, 1947;Covas and Hunziker, 1954;Naranjo et al., 1990), the chromosome number of B. setifolius var. brevifolius has not as yet been determined.
Robust delineation among taxa within B. setifolius is problematic (Gutierrez and Pensiero, 1998). The morphological traits that distinguish these taxa are not randomly selected and may consequently bias estimates of genetic distance among groups (Lewontin, 1974). In addition, the discontinuity of morphological characteristics is not always observed. Although the potential influence of environmental selection on adaptive traits is more readily understood for phenotypically-apparent traits than for cryptic markers such as DNA-based polymorphisms, genetic polymorphisms may nevertheless provide complementary data for detailed analysis of closely related genotypes. Molecular genetic methods have been applied to plant biosystematics with considerable success (Doebley, 1992;Soltis and Soltis, 1998). Random amplified polymorphic DNA (RAPD; Williams et al., 1990) and amplified fragment length polymorphism (AFLP; Vos et al., 1995) are two PCR-based techniques that have been successfully applied to the identification and estimation of molecular genetic diversity in various species (Newbury and Ford-Lloyd, 1993;Paran et al., 1998;Kolliker et al., 2001;Massa et al., 2001;Puecher et al., 2001). These techniques provide major advantages compared to other molecular marker methods such as restriction fragment length polymorphism (RFLP) because of their reduced requirement for large quantities of genomic DNA, their ability to amplify DNA from preserved tissues, and the smaller amount of time, labor and cost involved. In addition, unlike the PCRbased simple sequence repeat (SSR) and single nucleotide polymorphism (SNP) techniques, no prior sequence knowledge is required, and both RAPD and AFLP generate multiple locus information in a single assay (Wolfe and Liston, 1998). However, problems with reproducibility, particularly for RAPD, and the generation of dominant markers that are not locus-specific, limit the applicability of these methods, especially for genetic mapping studies compared to simple DNA profiling.
The aim of this study was to use molecular genetic and cytological approaches to examine the relationships among taxa of the B. setifolius complex. The results described here should be useful in elucidating the evolutionary processes that led to the origin of these taxa and should facilitate the domestication of forage grass species with potential agronomic value.

Plant material
Samples from different accessions of the section Pnigma were obtained from Patagonia, in Argentina ( Table 1). The morphological variation of dried tussocks was used to classify each accession. Seeds obtained from different tussocks were planted in pots to produce plantlets García et al. 313

RAPD and AFLP assays
Young leaves from 20 plantlets of each accession were combined, snap frozen in liquid nitrogen and lyophilized for 48 h. Subsequently, aluminum oxide crystals were added and the samples were ground to a fine powder with a pre-cooled pestle and mortar, prior to DNA extraction by the method of Saghai-Maroof et al. (1984).
The amplification reactions for RAPD analysis was done using different decamer oligonucleotide primers (Operon Technologies, Almeda, USA) in a DNA thermal cycler (Perkin Elmer). After an initial denaturation step at 94°C, 42 cycles of 45 s at 94°C (denaturation), 60 s at 55°C (annealing), and 90 s at 72°C (extension) were done prior to a final extension of 5 min at 72°C and subsequent cooling to 4°C. The amplified products were visualized by staining with ethidium bromide (0.5 mg/mL) after electrophoresis at constant voltage (60 V) in 0.9% (w/v) agarose gels in TAE (40 mM Tris-acetate plus 1 mM EDTA) buffer. Seven microliters of the reaction mixture was added to 3 mL of 1 x gel loading buffer (Maniatis et al., 1982).
AFLP reactions were done as described by Vos et al. (1995) with appropriate modifications. For digestion, 2.0-3.0 mg of the target DNA, 1.25 mL of EcoRI enzyme (12.5 units, 10 U/mL) and 1.25 mL of MseI enzyme (12.5 units, 10 U/mL) were used in a reaction volume of 50 mL. Adaptors specific for each restriction enzyme terminus were ligated to the digested DNA fragments in a final volume of 60 mL followed by incubation at 37°C for 5-6 h. Primary amplification by PCR was done using a +0/+0 primer combination and 30 cycles of 94°C for 30 s, 56°C for 60 s and 72°C for 60 s. Diluted primary amplification product was selectively amplified by using 1 mL aliquots of the selected primer specific for the frequent-cutter restriction enzyme MseI (30 ng/mL) and the rare-cutter restriction enzyme (EcoRI) (5 ng/mL). The primer for EcoRl was end-labelled with g 33 P-ATP to allow autoradiographic detection of AFLP products. The primer combinations used for EcoRI/MseI were: (1) AAA/GGGT, (2) AAG/GGGT, (3) AAG/CCA, (4) AAC/CCA, (5) AAC/GGG, (6) AAC/AAA, (7) AAA/CCA and (8) AAG/CAA. Selective amplification was done using the following touchdown PCR conditions: cycle 1: 94°C for 30 s, 65°C for 30 s and 72°C for 60 s; cycles 2-13: 94°C for 30 s, 65°C for 30 s, with a 0.7°C decrease in each cycle, and 72°C for 60 s; cycles 14-36: 94°C for 30 s, 56°C for 30 s and 72°C for 60 s. After amplification, the sample was mixed with one volume of sequencing loading buffer, heated to 95°C for 5 min and briefly quenched on ice before the loading onto the gel. Samples were run on a 6% (w/v) denaturing polyacrylamide gel (19:1 acrylamide:bis, w/w) in 7.6 M urea and 1x TBE buffer. Electrophoresis was done at constant power (110 W) for~2 h, after which the gel was dried at 80°C for 40 min and analyzed by overnight exposure to a storage phosphor screen for phosphorimaging analysis (Molecular Dynamics).

Analysis of molecular marker data
RAPD and AFLP bands were treated as dominant genetic markers and scored as polymorphic features if present in some but not all accessions, and as monomorphic features when the band was present in all of the accessions evaluated. To avoid underestimating the genetic similarity, the monomorphic and polymorphic bands were both included in the analysis. For RAPD data, genetic distance was estimated using the index of Balakrishnan and Sanghvi (1968). Matrices generated by scoring the AFLP bands were used as input data for the application NTSYS-pc version 2.01 (Rohlf, 1988). The Jaccard similarity index was applied using the SAHN program of NTSYS, and cluster analysis by UPGMA (unweighted pair-group method with arithmetic averages) was used done to obtain dendrograms. The data was also used for multidimensional principal coordinate analysis (PCOA).

Cytogenetic analysis
Chromosomal number and morphology were determined in mitotic metaphases of 20 plants from each accession (B. setifolius var. setifolius, B. setifolius var. brevifolius and B. setifolius var. pictus; Bs1, Bb1 and Bp1, respectively). Seeds from accessions Bs1, Bb1 and Bp1 were placed on wet filter paper in Petri dishes. Root tips were incubated in ice water for 36 h, fixed in ethanol-glacial acetic acid (3:1, v/v) for 2 days at 4°C, and then transferred to 70% ethanol and refrigerated until analyzed. Squashed preparations were prepared in 2% hematoxylin in acetic acid and examined with a Zeiss Axiophot microscope. Photographs were taken with Kodak T-Max-100 film. 314 Molecular and cytogenetic analysis of Patagonian Bromus setifolius

RAPD analysis
Of the 20 decamer RAPD primers that were tested for PCR amplification, 11 yielded clear, reproducible amplification profiles with the three pooled accessions. Thirtythree of 41 clearly distinguished bands were polymorphic (Figure 1). Calculation of the genetic distances showed that B. setifolius var. pictus and B. setifolius var. brevifolius were closer to each other than either was to B. setifolius var. setifolius (Table 2).

AFLP analysis
The amplification profiles obtained with different AFLP primer combinations in the seven samples corresponding to different accessions were used to score 692 fragments. The number of bands in each accession differed among taxa, with B setifolius var. setifolius having a lower number of bands (439)  Within B. setifolius var. pictus, 429 bands were monomorphic and 84 were polymorphic. A comparison of the amplification profiles among accessions in this taxon allowed the identification of accession-prevalent and species-prevalent markers (Figure 2).
Samples BpRMg1 and Bs1 were from the same location and, in some cases, the same tussock included both species. In contrast, in the other species, individual tussocks contained only one species, but generally included plants with more than one genotype for the species. The genetic distances among the samples analyzed by AFLP were calculated by using the Jaccard similarity coefficient ( Table 3). The degree of similarity among the five accessions of B. setifolius var. pictus was high (0.834-0.988). Figure 3 shows the dendrogram obtained by cluster analysis of the similarity indexes using the UPGMA method ( Figure 3).
The pooled sample of B. catharticus genotypes was clearly separated from accessions of the section Pnigma, thus confirming that B. catharticus was an outgroup in this analysis. The three taxa within section Pnigma were grouped in a single large cluster. Within B. setifolius var. pictus, accessions from the same geographical area (Rio Mayo) were grouped in the same sub-cluster. Two additional sub-clusters were identified in accessions from this geographical location and differed in their grazing history. Figure 4 shows a 3-dimensional plot of the AFLP data based on multidimensional PCOA. The first principal coordinate (depicted by the length of the rectangle in Figure 4) explained 51.58% of the total variation and separated B. setifolius var. setifolius from B. setifolius var. pictus and B. setifolius var. brevifolius. The second principal coordinate (depicted by the width of the rectangle) explained 21.52% of the variation and separated B. setifolius var. brevifolius García et al. 315 (Table 3).

Discussion
Molecular marker-based genetic diversity analysis RAPD and AFLP analyses of taxa in the genus Bromus (section Pnigma) yielded similar results. Although difficulties associated with the reproducibility of RAPD limit the applicability of this technique (Perez et al., 1998), RAPD may nevertheless be useful for preliminary assessments of intraspecific, interspecific and intergeneric genetic variation in grasses (Wang et al., 2001;Garcia et al., 2004). In contrast, the greater accuracy and information content of AFLP make this technique more useful for detailed investigations such as in molecular breeding.
The variation in the number of bands among accessions was relatively small when compared to the large differences in chromosome numbers among taxa. Despite its lower chromosome number the tetraploid taxon B. setifolius var. setifolius had only 6.7% fewer bands than the decaploid taxon B. setifolius var. pictus. The higher content of genomic DNA in B. setifolius var. setifolius (1 Cx = 2.06 pg) compared to B. setifolius var. pictus (1 Cx = 1.63 pg) (Naranjo et al., 1990) may compensate for the lower chromosome number and account for the complexity of its AFLP profile. Fay et al. (2005) reported that the number of AFLP bands tended to increase up to a DNA content of 1C = 8.43 pg. 316 Molecular and cytogenetic analysis of Patagonian Bromus setifolius   Bromus setifolius var. pictus and B. setifolius var. brevifolius were more similar to each other than either was to B. setifolius var. setifolius. This observation is consistent with studies of morphological variation in which B. setifolius var. pictus and B.setifolius var. brevifolius showed a high degree of similarity and were difficult to distinguish taxonomically (Camara Hernandez, 1978). Al-though variation in the morphological traits did not permit a clear delineation between the taxa (Gutiérrez and Pensiero, 1998), both RAPD and AFLP analysis revealed bands that may be considered as taxon-specific discriminatory markers. This was particularly the case in the AFLP profiles. However, these results must be treated as preliminary until a larger number of accessions have been analyzed. AFLP analysis showed that very low frequencies in one taxon and very high frequencies in another may be readily misclassified as diagnostic bands because of the small sample sizes involved. Consequently, such features may be more accurately described as taxon-prevalent and require more intensive analysis.
Analysis of the genetic structure of plant populations requires the sampling of individual plants. In grasses of the genus Bromus, it is conventional to consider each tussock as an individual genotype, but it is also possible for individual tillers in a single tussock to be derived from different individuals of the same species, or even from different species. Wind damage on the Patagonian steppe produces high mortality during seedling recruitment (Aguiar and Sala, 1997). Consequenly, long-established tussocks may provide a 'nurse effect' by protecting the seedlings against wind. The composite content of individual tussocks is a potentially important confounding factor in the study of genetic diversity and may also provide an explanation for some of the difficulties previously encountered in the taxonomic discrimination of related species. The use of individual inflorescences as the unit of study, rather than the entire tussock, provides a way to address this problem.

Relationships between genetic variation and ecological parameters
In grass species there is not always a correlation between variation in morphological traits, geographical distribution, life history traits and biochemical or molecular genetic data. The lack of such correlations for B. catharticus was explained in terms of phenotypic plasticity (Puecher et al., 2001). In the grass Schizachyrium scoparium (Carman and Briske, 1985) morphological differences were detected between populations that had been subjected to long-term grazing and those that had not been grazed, but isoenzyme variation was not related to grazing history. Schrauf and Naranjo (1988) attributed the observed differences in isozenzyme frequencies between grazed and non-grazed populations of Pappophorum caespitosum and P. philippianum to the random effects of genetic drift.
As shown here, accessions of B. setifolius var. pictus with different grazing histories were resolved into different sub-clusters based on AFLP data. In addition, non-grazed accessions had more AFLP bands than those that had been grazed. These results may have several explanations. Sheep, as the primary grazing animals in this study, may have a more selective grazing behavior than bison (Carman García et al. 317 and Briske, 1985) and cows (Schrauf and Naranjo, 1988), as found for Danthonia species (Scott and Whalley, 1984). Alternatively, the wider range of markers analyzed by AFLP makes this system more capable of assessing a more representative sample of genomic variation than isoenzymes, thereby allowing finer discrimination among accessions.
Cytogenetic analysis and evolutionary relationships Kozuharov et al. (1981) showed that diploid, tetraploid and higher polyploid level chromosomal races within the annual Bromus species B. secalinus and B. lanceolatus were morphologically similar and that there was little justification for their taxonomic separation. In contrast, the perennial octoploid species B. lacmonicus and the diploid species B. parilicus are morphologically similar but show features that reflect their cytological differences and consequently justify their species status. Similarly, B. setifolius var. setifolius (2n = 28) and B. setifolius var. pictus (2n = 70) are perennial taxa with stable morphological differences that parallel their karyotypic divergence and support their classification as separate species. Naranjo et al. (1990) proposed that B. setifolius var. setifolius (sensu Naranjo et al. (1990): B. setifolius) may be a progenitor taxon of the polyploid B. setifolius var pictus (sensu Naranjo et al. (1990): B. pictus) on their morphological similarity, karyotypic structure, growth pattern, breeding system and presence of reproductive isolation. Recent work using fluorescence in situ hybridization with total genomic DNA as a probe (GISH) has shown that B. setifolius var. setifolius or a closely related taxon is one of the progenitors of B. setifolius var. pictus, thus confirming the allopolyploid nature of this species (Garcia et al., 2005).
This report is the first to describe the chromosomal number of B. setifolius var. brevifolius (2n = 70). As shown here, molecular genetic data indicated a close relationship between B. setifolius var. brevifolius and B. setifolius var. pictus, and AFLP-based PCOA suggested that B. setifolius var. brevifolius may be directly derived from B. setifolius var. pictus, rather than the second allopolyploid progenitor, in agreement with the morphophysiological reticular pattern (Garcia et al., 2001). Bromus setifolius var. brevifolius has wide leaves and a fast initial growth, B. setifolius var. setifolius has short internodes and high drought resistance, and B. setifolius var. pictus has wide leaves, short internodes, fast initial growth and high drought resistance. The putative progenitor of B. setifolius var. pictus may have had a haploid chromosomal complement of n = 21, assuming that B. setifolius var. pictus has retained the genetic complement of both ancestral species (some chromosomal rearrangements may have occurred subsequent to polyploid formation). Additional studies are required to determine the identity of this progenitor.

Conclusions and agronomic implications
A taxonomic realignment of Bromus species within section Ceratochloa has been proposed based on morphological, cytological and molecular genetic data (Massa et al., 2004). The molecular genetic and cytogenetic data described here have clarified the relationships among taxa of the Patagonian Bromus section Pnigma. These data agree with the findings of Naranjo et al. (1990). Taxonomically, B setifolius var. setifolius and B. setifolius var. pictus should be recognized as distinct species under the names B. setifolius Presl. and B. pictus Hook, respectively. The cytological and molecular genetic similarity between B. setifolius var. pictus and B. setifolius var. brevifolius suggests that these taxa be considered as varieties within the same species, i.e., B. pictus var. pictus and B. pictus var. brevifolius, respectively.
Bromus pictus var. pictus is currently the most attractive option for a cultivated species to mediate the reversal of environmental degradation in Patagonia because of its combination of drought tolerance, high recruitment capacity and forage quality. Biodiversity conservation issues may also be addressed through domestication and improved breeding of B. pictus var. pictus, particularly since genetic erosion through sheep grazing has pushed species of the section Pnigma to the brink of extinction. The genetic conservation of B. pictus var. pictus will prove effective for the section as a whole, and the initial data described here for molecular marker variation will provide valuable information for marker-assisted breeding and future evolutionary studies.