INTRODUCTION

Livestock production is a key sector in the economies of Latin American countries, occupying a large area of land with productive potential (FAO 2009). Among the forage plants used, the genus Brachiaria, which is characterized by having cultivated forage pastures, stands out (^{Do Valle and Miles et al. 2001}). The success of these pastures is due to the fact that many species have a wide natural variation in agronomic traits, such as biomass production, nutrient quality, drought tolerance, tolerance to flooded soils and high seed yields (Miles et al. 2004). The qualities of these forage species have been the reason for developing hybrids with resistance to cuckoo spit, the frothed-up sap caused by the spittlebug (Aeneolamia postica), which limits the passage of water and nutrients in the plant (^{De la Cruz-Llanas et al. 2005}), and to foliar fungi such as Rhizotocnia; hybrids have also been developed that are adapted to acid soils with high leveis of aluminum (^{Rao et al. 2006}) and that have higher production of high digestibility forage (^{Lascano 2002}). Most commercially cultivated Brachiaria species are tetraploid apomicts; only B. ruzizienzis reproduces sexually and its recombination with apomictic genotypes can generate hybrids with desirable characteristics (Miles et al. 2004).

Apomixis is present in some Brachiaria species, which implies that the progeny are genetically identical to the mother plant, so they do not present segregation, a phenomenon that is often of importance in plant breeding (^{Grimanelli et al. 2001}). This character is considered relevant in agriculture, due to the possibility of transforming crops of worldwide importance from sexual reproduction to apomictic clones (^{Bicknell and Koltunow 2004}). Apomixis has great agricultural interest, as its fixa-tion would allow maintaining the genetic stability of hybrids and varieties for an indefinite period, while the progeny of sexual reproduction maintain their genetic variation (Koltunow et al. 1995). B. decumbens, B. brizantha and B. ruziziensis are considered the most important species in this regard, since they belong to the same agomictic complex, with the first two being apomictic and the last one sexual (^{Do Valle and Miles 2001}, ^{Risso-Pascotto et al. 2005}).

Traditionally in Brachiaria, morphological descriptors and embryo sac analysis with pistil thinning techniques are used to differentiate progenyvor sexual (^{Savidan 2000}), while genetic-molecular marker techniques have been used successfully since the 1990s, with the advantage that they provide results in a short time from any plant tissue and any plant age; they are efficient because they are not affected by the environment. In most cases, the markers complement the information from the morphological markers (Azofeifa-Delgado 2006). In particular, the Amplified Fragment Length Polymorphism (AFLP) and Cleaved Amplified Polymorphic Sequence (CAPS) techniques are used to sample the entire genome (^{Zorzatto et al. 2010}). There-fore, the objective of the present study was to apply the AFLP and CAPS techniques to differentiate interspecific and sexual apomictic genotypes of Brachiaria spp, selected as apomictic in the field.

MATERIALS AND METHODS

DNA extraction

From 1 g of lyophilized tissue obtained from the mixture of the genotypes listed in Table 1, the DNA was obtained with the cetyltrimethylammonium bromide (CTAB) method (^{Dellaporta et al. 1983}, ^{Zhang and Stewart 2000}).

Table 1 Genetic material and keys used in molecular characterization. 

*ln the hybrids (9-30) the number after the hyphen corresponds to the number of plants resulting from the cross with apomictic characteristics considered in the study. ** In the hybrids (9-30) the numbers correspond to the parents involved in the crossing.

AFLP

Digestión, adapter ligation, pre-amplification reactions and selective amplification were per formed as indicated by GIBCO-BRL Life Technolo gies AFLP™. For the selective amplification, the following primer combinations were evaluated: E-AAC + M-CAC, E-AAC + M-CTG, E-ACC + M-CAC, E-ACC + M-CTG, E-AGG + M-CTC, E-AGG + M-CTA, E-ACT + M-CTC y E-ACT + M-CTA and those which produced clear differences between parents and progeny were selected (Table 2). The fragments obtained in the selective amplification were separated on denaturing 6% polyacrylamide gels (19:1), 7.5 M urea and IX TBE buffer in a BioRAD® vertical sequencing chamber and an E-C Apparatus Corporati ^® model 3000PW power source. The reagents required for electrophoresis and staining were prepared according to ^{Sambrook et al. (1989}). Prior to running the polymerase chain reaction (PCR) products, the gel was pre-run for 30 min at 55 °C. Subsequently, 10 µL of the samples previously mixed and heated with a run buffer at Electrophoresis was run at 80 watts (1000 volts) for 3 h; the gel was then stained in 0.2% silver nitrate solution for 30 min and developed with sodium carbonate solution, made with 30 g of sodium carbonate (Na₂CO₃) per L of distilled water, which formaldehyde and 400 µL of sodium thiosulfate (10 mg mL^-1) were added. When the first bands were developed, after approximately 4 min, the solution was discarded and replaced with fresh solution to continue the development for another 15 min. Finally, the gels were photodocumented for analysis.

Table 2 Primers, number of fragments and level of polymorphism detected with the Amplified Fragment Length Polymorphism (AFLP) and Cleaved Amplified Polymorphic Sequences (CAPS) techniques in parents and progeny of the genus Brachiaria. 

RESULTS AND DISCUSSION

The number of markers obtained was acceptable and informative; 147 corresponded to AFLP, of which 91.3% were polymorphic, while 87 were obtained with CAPS with 79.7% polymorphism. Both types of markers differentiated each of the genotypes studied (Table 2). In the consensus phenogram, constructed with the AFLP markers (Figure 1), three main groups are distinguished: group A comprised by the hybrid 143A-18-3; group B defined by the ♂ 94 and distant from the other two males; and group H with genotypes 4171-9, 11094-1 and 11094-16. Group B was related to the rest of the genotypes, which were clustered based on close genetic similarities. Groups C and G show groupings of full siblings. The C group concentrates genotypes with greater influence of ♂ 18 [10018-1, 10018-16 and 10018-6] and [4118-12, 4118-10 and 4118-6], which is a dominant apomict, and ♀ 110, 29 and 41; this group also includes the genotype 4171-20, which must share characters with the parent ♀ 41 and the indicated half-siblings. Group G contains full siblings originating from ♀ 143A and ♂ 94 [143A94-11, 143A94-3 and 143A94-4, in addition to genotype 143A18-23, which apparently with ♂18. On the other hand, the genotype 143A18-3 is found as an independent group (group A) when it was expected to be grouped with its genetic similarities. The discrepancy in this behavior may be due to mutations in the genome that generate polymorphisms in the profiles obtained. These alterations are common in plants of asexual/vegetative reproduction and it is relatively easy to detect them with molecular markers. Group D contains ♂ 71 and some of its progeny, group E has ♂ 18, and finally group F constains ♀ 100 and 143A. The fact that all the females were distributed throughout the phenogram suggests the genetic contribution of B. ruziziensis used as a female parent. The AFLP markers were able to differentiate all the genotypes and to show the close genomic relationship between some of them; however, they were insufficient to mark the locus or loci responsible for the apomixis (^{Hand and Koltunow 2014}).

Figure 1 Consensus phenogram of apomictic and sexual genotypes of Brachiaria spp., constructed with Amplified Fragments Length Polymorphism (AFLP) genetic markers. Un-weighted Pair Group Method with Arithmetic Mean (UPGMA) clustering and random sam-pling with 500 replicates (bootstrapping). 

Figure 2 shows the grouping formed with CAPS data; ♂ 18 remained independent of the two main groupings, group I and group II, suggesting less genetic similarity according to these markers. Group I includes the parent ♂ 94, the genotypes with greater genetic influence of this parent, as well as some genotypes related to the parents ♂ 18 and ♂71. Group II comprises the females 29, 110 and 41, as well as the hybrids that shared more characters with them, but also ♂ 71, probably due to its facultative character. The other two females, 100 and 143A, were grouped independently. As in the phenogram with AFLP data, the groupings with CAPS were also made up of subgroups of full siblings; forexample, group I included 4: [10018-16,10018-6]; [143A94-3, 143A94-4]; [11094-16, 110941, 11094-7] and [2918-18, 2818-14], while group II comprised 2 subgroups, [4118-10 and 4118-12] and [4194-19, 4194-24], mainly. There are several matches between the two phenograms; there were groupings of full siblings, as well as females 100 and 143A or 29 and 41, which stayed together but in sepárate groups, which explains their close ge netic resemblance. On the other hand, in both phenograms it was possible to sepárate the females from the males; however, in the case of CAPS, ♂71 was not separated from the group of females 29, 110 and 41, perhaps due to their particular qualities. The distribution behavior observed in both phenograms can be useful, especially when considering the alternative of selecting any of the females used in the present study due to their degree of genetic similarity, or ♂18 and 94 that showed less relatedness between them in both marker systems. Another observation was that the progeny was mainly influenced by the genetic characters of the parents ♂18 and ♂94, and to a lesser degree by ♂71.

Figure 2 Consensus phenogram of apomictic and sexual genotypes of Brachiaria spp., constructed with Cleaved Amplified Polymorphic Sequences (CAPS) genetic markers, Un-weighted Pair Group Method with Arithmetic Mean (UPGMA) clustering and random sam-pling with 500 replicates (bootstrapping). 

The search for characters associated with apomixis or sexuality with techniques that detect anonymous markers such as those used in this study has proved difficult, unless sufficient molecular data are considered; this probability increases if precise information is available for the two loci that are assumed to control apomixis in plants (^{Noyes and Rieseberg 2000}). DNA markers associated with apomixis (AFLPs and SCARs) have been re-ported for the genera Paspalum (^{Labombarda et al. 2002}), Pennisetum (^{Ozias-Akins et al. 1998}), Hieriacium (^{Catanach et al. 2006}) and Trichloris crinite (^{Cavagnaro et al. 2006}). In this study, the CAPS markers detected a DNA fragment called C15-8 that differentiated the sexual parents (♀) from the apomictic ones (♂) and that could be associated with the sexuality of B. ruziziensis; this may indicate the absence of this distinctive character in the apomictic genotypes used as males. This fragment was inherited from hybrids such as 9, 10, 12, 14, 15, 16, 18, 19, 22, 26 and 27 (Figure 3) that were selected at the field level as apomicts, one (♂) and that could be associated which may indicate that the B. ruziziensis line used as female is not the same as that of the tetraploid males. This observation can also be supported by separating the females in the two phenograms obtained. An alternative way of determining the effect of this marker is to observe the behavior at field level with respect to sexuality in different environments, to confirm it in a traditional way. The informative capacity of this CAPS technique has also been shown by authors such as ^{Móhring et al. (2005)} who detected self-incompatibility characters in Brassica species. The relative ease of differentiating sexual genotypes (Brachiaria ruziziensis) from interspecific apomicts (B. brizantha and B. ruziziensis) in the present study could be favored by the genomic contrast of the species types involved in the crosses.

Figure 3 Cleaved Amplified Polymorphic Sequences (CAPS) profiles of 30 genotypes of Brachiaria spp., obtained with the ROTH C-15 primer. Lanes 1-3 correspond to 4X males, 4-8 to 2X females and 9-30 to hybrids. The ellipses indicate a distinctive DNA fragment in the females (lanes 4-8) that were inherited by some hybrids and named C15-8. Lane M indicates the molecular weight marker øX174 DNA /Haelll. 

The simple correspondence factor analyses (Table 3) performed for the AFLP and CAPS data allowed weighting the value contributed by each component (amplicon) (^{Dore and Ojasso 2001}). With both techniques, the first six principal components explain 56.47% for AFLP and 54.27% for CAPS, which suggests the effectiveness of the markers used to differentiate each of the genomes. The AFLP and CAPS techniques were appropriate to estimate the genetic relationship of parents and hybrids of the genus Brachiaria. The genetic contribution of male parents 18 and 94 was shown in the progeny, compared to that provided by parent ♂71. The CAPS technique was able to clearly differentiate the diploid genotypes of B. ruziziensis (female) from the tetraploid genotypes of B. brizantha X B. ruziziensis used as males. The CAPS factorial analysis allowed correlating the DNA fragment called C15-8 as a distinctive character of the female sexual parents, unlike that of the male apomictic parents.

Table 3 Eigenvalues of genotypes of the genus Brachiaria, product of simple correspondence