The genus Pennisetum (Richard, 1805) is one of the most important in family Poaceae family. It comprises about 140 species, distributed in five sections (Penicillaria, Brevivalvula, Gymnothrix, Heterostachya and Eu-Pennisetum) based on morphological characteristics (Stapf and Hubbard 1934). The section Penicillaria includes the economically most important species: elephant grass [Pennisetum purpureum (Schumacher, 1827)], used as forage, and pearl millet [Pennisetum glaucum (Linnaeus, 1753)], used as cereal and forage (Martel et al. 2004).

Molecular analyses based on mitochondrial DNA (Chowdhury and Smith 1988), chloroplast DNA (Renno et al. 2001) and repetitive DNA sequences (Ingham et al. 1993) have revealed significant relationship between the genomes of Pennisetum glaucum, Pennisetum purpureum and Pennisteum squamulatum Fresen., suggesting that these three species may have a common origin (Martel et al. 2004). Among these, the cultivated species Pennisetum glaucum and Pennisetum purpureum are phylogenetically related, possessing a close resemblance between their genomes, constituting a monophyletic group with recent divergence (Martel et al. 2004). In this sense, Pennisetum glaucum is an annual, alogamous, diploid species (2n = 2x = 14, genomes AA), with genome DNA content of 4.72 pg, and constitutes the primary genic pool of this genus. In turn, Pennisetum purpureum belongs to the secondary genic group and is a perennial, alogamous, tetraploid species (2n = 4x = 28, genomes A'A'BB), with genome DNA content of 4.60 pg (Martel et al. 1997). Both species have approximately the same DNA content (pg), but differ with regards to the monoploid size genome (basic number, x). Pennisetum purpureum is a tetraploid and have about half the DNA content (1.15 pg) of the Pennisetum glaucum monoploid genome (2.36 pg). Further, Pennisetum pupureum have smaller chromosomes than Pennisetum glaucum (Martel et al. 2004). This shows that important chromosome changes may be linked to the evolution and divergence among these species (Andrade-Vieira 2010; Barbosa et al. 2003; Martel et al. 2004; Robert et al. 2011).

In spite of their integrating distinct genic groups and differing as to ploidy level, the genetic proximity between these two species becomes evident when the occurrence of natural hybridization is observed. This sexual compatibility is partial, and results in sterile triploid hybrids (2n = 3x = 21, genome AA’B) (Hanna 1987; Martel et al. 2004; Robert et al. 2011; Techio et al. 2006). Cytologically, the genomic proximity has been demonstrated by meiotic analyses of triploids hybrids obtained in breeding programs. During diakinesis and metaphase I in this hybrid, the formation of seven bivalents is frequently observed, resulting from the pairing among chromosomes of genomes A and A’ of Pennisetum glaucum and Pennisetum purpureum, respectively, as well as seven univalents of genome B of Pennisetum purpureum (Jauhar 1968, 1981; Pantulu 1967; Sree Rangasamy 1972; Techio et al. 2005, 2006). The occurrence, even at low frequency, of trivalents and numbers of bivalents above seven suggests both allo- as well as autosindetic pairing among the genomes A, A’ and B (Sethi et al. 1970; Techio et al. 2005, 2006). These observations on the configurations of bivalents and univalents during meiosis, as well as the morphology of metaphase chromosomes, constitute the pioneering works demonstrating homeology among the genomes A, A’ and B. The obtained results suggest that the genome A of Pennisetum glaucum has larger homeology with genome A’ of Pennisetum purpureum, whereas the origin of genome B has not been defined (Jauhar 1981; Techio et al. 2005, 2006).

In this sense, despite evidence for a common evolutionary origin between Pennisetum glaucum and Pennisetum purpureum and the economic importance of these species, there are no studies providing more conclusive data with respect to the homeology among genomes A, A’ and B. Thus, the objective of this work is to describe the proportion and distribution of the homologous regions present in genomes A of Pennisetum glaucum and A’B of Pennisetum purpureum, by cytomolecular analyses using genomic in situ hybridization (GISH).

Previous analyses of meiotic pairing in the triploid hybrid have showed that the genomes A and A´ are more related. On the other side, between both and the genome B there are affinity/homeology reduced (Jauhar and Hanna 1998, Techio et al. 2005). In this study, hybridization of genomic probes of Pennisetum glaucum and Pennisetum purpureum were used for the first time to demonstrate and to confirm homeology among the three genomes. It was evaluated the distribution and proportion of these homeologous regions in the family constituted by the parental Pennisetum purpureum (BAG 65) and Pennisetum glaucum (BN2), and by the triploid hybrid (BAG 65 × BN2) originating from this crossing.

The higher level of homeology between genomes A and A’ was confirmed because the 14 chromosomes belonging to genome A’ of Pennisetum purpureum were strongly marked and distinguished from the 14 chromosomes from genome B using the genomic DNA of Pennisetum glaucum (genome A) as probe in metaphases of Pennisetum purpureum (Fig. 1a). The chromosomes of genome A’ presented marks in along almost role chromosome length, whereas genome B presented small marks dispersed over the length of its chromosomes (Fig. 2a). Moreover, approximately 29% of Pennisetum purpureum genome (A’B) was hybridized by the genome A of Pennisetum glaucum (Table 1). This percentual represents only the A’ genome since the markers on genome B chromosomes were not record because it were dispersed on chromosomes. The observed homeology was only quantified in the genome A’ of Pennisetum purpureum, due to the difficulty in measuring the small and dispersed marks found in genome B (Fig. 2a).

The homeology between genomes A and A’ was confirmed by the extensive marking of Pennisetum glaucum chromosomes by the probe A’B of Pennisetum purpureum. All 14 chromosomes from genome A of Pennisetum glaucum were almost completely marked, with large blocks of probe signals observed on the chromosomes (Fig. 1b). The markings by the probe of genome A’B observed in the centromeric and pericentric regions represented 63% of the genome of Pennisetum glaucum (Table 1 and Fig. 2b). These marked portions result from hybridization, both between the genomes A and A’ and, in smaller proportion, genomes A and B, observed both in karyograms of Pennisetum purpureum and triploid hybrids (Fig. 2a, c).

In the triploid hybrid (AA’B) the hybridized portion of genome A (Pennisetum glaucum) corresponded to 54%, and the signal of probe A’B (Pennisetum purpureum) to 49% of its total genome (Table 1). Despite the similarity in proportion, the distribution pattern for the probes from the parental individuals was different in the hybrid (Fig. 1c). The seven chromosome of the hybrid were entirely marked with the probe of genomic DNA from Pennisetum glaucum. The remaining chromosomes from Pennisetum purpureum parental (genome A´B) presented marks only in the centromeric and pericentromeric regions (Fig. 2c). However, when the probe with DNA of Pennisetum purpureum was used in chromosomes of the triploid hybrid the marks were observed mainly in centromeric and pericentromeric regions, but some chromosomes appearing almost totally marked (Fig. 1d).

The differences in marking pattern observed in the triploid hybrid, mainly between genomes A and A’, could be explained by the presence of two genomes (A’ and B) in the same probe. The observations evidence the changes arising from interspecific hybridization in Pennisetum purpureum genomes. Once combined in a polyploid hybrid nucleus, extensive reorganization may rapidly occur in the parental diploid genomes, both intra and intergenomically (Soltis and Soltis 1999; Chen et al. 2006). The rapid intergenomic rearrangements in polyploids in relation to the diploid progenitors have been demonstrated in allohexaploid F1 hybrids of Avena sativa (Leitch and Bennett 1997), in hexaploids of wheat (Nelson et al. 1995) and soybean (Shoemaker 1996), as well as in triploid hybrids embryos of Pennisetum (Campos 2007).

Besides the existing homeology among genomes A, A’ and B, the utilization of GISH in the genomes of Pennisetum glaucum, Pennisetum purpureum and interspecific hybrid enable to verify the differences in chromosomes size, and also chromosomes number of these species. Analyzing cells of the interspecific hybrid, the difference in size between the parental chromosomes becomes evident, with those of Pennisetum glaucum being larger (Fig. 2c). It can also be observed that the total length of Pennisetum purpureum chromosomes did not increase proportionally in relation to those of Pennisetum glaucum (Table 1 and Fig. 2a and b), and that the chromosomes of genome B do not differ significantly in size in relation to genome A’ (Table 2 and Fig. 2a). These differences in size and chromosome number between the two species reflect their evolutionary history.

The evolutionary tendency among true grasses, which have a common and recent origin, is that the most derived species have emerged after reduction of the number and increase of the size of chromosomes in relation to the ancestors (Avdulov 1931; Bennetzen 2007; Crepet and Feldman 1991; Martel et al. 2004; Paterson et al. 2004; Stebbins 1956, 1971). This tendency applies to the genus Pennisetum, whose common ancestral pattern presented the basic number of chromosomes x = 9. Furthermore, the evolutionary pattern inside the genus follows the same tendency, this way, it is observed that the ancestor of the species from section Pennicillaria, to which Pennisetum purpureum and Pennisetum glaucum belong, presented basic number of chromosomes x = 7. In this sense, analyzing the phylogeny of the genus Pennisetum presented by Martel et al. (2004) and the information on chromosome size and homeology of genomes A, A’ and B observed in the present work, it can be inferred that the species Pennisetum purpureum and Pennisetum glaucum have concomitantly diverged from the common ancestor. The origin of Pennisetum purpureum occurred at the interspecific hybridization event, combining the genome A of the ancestor with genome B of a second, still unknown. Therefore, genome A’ could be considered a subgenome of the ancestor A due to genomic and structural changes that occurred during evolution. On the other hand, the species from the primary genic pool of Pennisetum glaucum have diverged from the common ancestor through increase of chromosome size, probably by increment in the genic sequences. The increase in chromosome size in this species could be explained, as described by Poncet et al. (2002), by duplication of some genes as consequence of domestication syndrome. The presence of non-homologous recombination observed among the chromosomes of Pennisetum glaucum reinforces the hypothesis of genic duplication as part of the differentiation in this species (Jauhar 1970; Martel et al. 2004).

This hypothesis, presented for the evolution and divergence of Pennisetum purpureum and Pennisetum glaucum from the common ancestor, may be further reinforced by the differences observed in relation to the size of chromosomes from genomes A, A’ and B, as shown in Table 2. Analyzing the size of the monoploid complement of genome A, it can be verified that it is 24% larger in relation to the length of the chromosomes of genome A’ of Pennisetum purpureum. Considering that the genomes A and A’ have evolved from an ancestor genome A, the difference in chromosome size could be related to genic duplication in Pennisetum glaucum and to genomic rearrangements observed in the allotetraploid hybrid Pennisetum purpureum. Rearrangements and loss of genomic sequences are common events after hybridization (Kellis et al. 2004), as observed in this study by comparison between different genomes combined in the triploid hybrid (Table 2). In this case, a reduction of 60% of genome A can be observed in the hybrid in relation to the parental Pennisetum glaucum, along with 44% and 52% reduction of genome A’ and B, respectively, in relation to the parental Pennisetum purpureum.

In this work, GISH confirmed the homeology among genomes A, A’ and B and enabled the identification and distribution of the homeologous regions in the chromosomes. Moreover, the GISH markings were able to separate the different genomes, leading the comparison on the size of the chromosomes in each of these three genomes. This distinction of the different genomes confirmed the occurrence of rearrangements after interspecific hybridization, especially when the synthetic triploid hybrid was analyzed, and prove the allotetraploid origin of Pennisetum purpureum It also show that genomes A’ and B have chromosomes similar in size. In evolutionary terms, the results reinforce that the genomes A and A’ have diverged from an ancestral genome A by increase of chromosome size in Pennisetum glaucum and rearrangements and/or deletions in Pennisetum purpureum. The reorganizations occurring in the ancestral genome A during evolution have generated the subgenome A’ of Pennisetum purpureum.