The evaluation and characterization of the available genetic diversity between and within species are extremely important for conservation and genetic breeding. Data on the genetic diversity and population structure of Urochloa spp. obtained using molecular markers were reported just over 10 years ago (Table 3). The first studies were performed using random amplified polymorphic DNA (RAPD) markers (Chiari et al., 2008; Ambiel et al., 2010) and generated the first information about the genetic diversity present in Urochloa species at the molecular level.

Shortly thereafter, other Urochloa studies began to be developed using another type of marker known as microsatellites or simple sequence repeats (SSRs; Tables 3, 4), which stand out for being multiallelic and more abundant in plant genomes than RAPD markers. SSRs are also more informative because they are codominant but need to be evaluated as dominant markers in Urochloa genotypes, as these species are predominantly polyploids. In this case, each band or allele of an SSR primer is considered a separate dominant marker, like in RAPD genotyping. Jungmann et al. (2010) used SSR markers to analyze the genetic diversity and population structure of 58 germplasm accessions and two cultivars of U. humidicola from Embrapa Beef Cattle. The accessions were highly structured into four major groups, and the coefficients of dissimilarity ranged from 0.02 to 0.89 considering all genotypes. The study observed that the phenotypic divergence of the only sexual accession of U. humidicola (H031) reflects a high genetic divergence because it comprises a mixture of distinct allelic pools not shared with most accessions (Jungmann et al., 2010). This considerable divergence between H031 and the other accessions of U. humidicola has also been observed in later studies and suggests a hybrid origin (Santos et al., 2015a; Vigna et al., 2016a).

The genetic diversity and population structure of 172 germplasm accessions and six cultivars of U. brizantha were also evaluated using microsatellite markers (Vigna et al., 2011a). As a result, the genetic similarity among all genotypes ranged from 0.40 to 1.00, and two duplicates were found in the germplasm. Structural analyses have suggested germplasm separation into three clusters with significant differentiation among the various clusters (Vigna et al., 2011a). For U. ruziziensis, a study was developed to estimate the molecular diversity among 93 genotypes of Embrapa’s collection using inter-simple sequence repeat (ISSR) markers. The genetic similarity ranged from 0.0 to 0.95, revealing clones and highly divergent genotypes (Azevedo et al., 2011). Recently, the genetic diversity and population structure of 112 Ethiopian U. brizantha accessions (International Livestock Research Institute Forage Field Genebank) collected in nine different regions of Ethiopia and six Urochloa cultivars were analyzed using 23 SSR markers. The structural analysis revealed three clusters with distinct gene pools and significant variation among them. This study established a core collection of Ethiopian U. brizantha accessions consisting of 39 accessions (Tegegn et al., 2019).

The above-mentioned studies were performed within a single species of Urochloa genus, but studies of genetic diversity were also performed with more than one species. Using RAPD markers, Chiari et al. (2008) assessed the genetic diversity of four Urochloa spp. and found a genetic similarity ranging from 0.49 to 0.87 considering all genotypes analyzed. The greatest genetic dissimilarity was detected among genotypes of U. ruziziensis (0.73 to 0.83). In addition, the analysis showed that U. brizantha, U. decumbens, and U. ruziziensis were genetically closer to each other than to U. humidicola. This finding has also been observed in later studies using SSR markers (Jungmann et al., 2009a; Silva et al., 2013; Santos et al., 2015a; Ferreira et al., 2016; Triviño et al., 2017; Pessoa-Filho et al., 2017; Souza et al., 2018).

Nitthaisong et al. (2016) assessed the genetic diversity of 11 Urochloa species using 10 ISSR markers and found high polymorphism between the genotypes. The genotypes were grouped into three clusters: one cluster comprised the majority of polyploid plants (apomictic), and the other two clusters were mainly formed of diploid plants (sexual). U. decumbens, U. brizantha, U. humidicola, and U. dictyoneura were grouped into a separated cluster of U. ruziziensis. Using microsatellite markers to assess the genetic diversity among Kenyan Urochloa spp., Ondabu et al. (2017) found an average diversity of 0.623, with a range of 0.317 to 0.802. Triviño et al. (2017) also used SSR markers to analyze genotypes of Urochloa spp., which mainly originated from the CIAT germplasm, at the molecular level. In this study, the genotypes were distributed in four clusters, and one cluster was composed of most genotypes of U. decumbens and U. ruziziensis, as supported by morphological similarities between the two species (Renvoize et al., 1996). In addition, greater genetic variance was observed within the taxonomic groups (57%) than between groups (43%).

Urochloa genotypes from Tanzania were also evaluated in terms of their genetic diversity and population structure using SSR markers, and high genetic variation was found between the Urochloa genotypes. The population of U. brizantha showed the highest level of genetic diversity. All genotypes were divided into three main groups, and the genetic variation was greater within populations than among populations (Kuwi et al., 2018).

A national collection of Urochloa ecotypes from Uganda, possibly represented by different species, was evaluated using 24 SSR markers (Namazzi et al., 2020). These markers had a high discriminating ability with an average polymorphism information content (PIC) of 0.89 and detected 584 alleles in 99 ecotypes. A structural analysis showed the presence of three distinct gene pools in the Ugandan Urochloa ecotypes, in agreement with previous studies performed in Brazil, Tanzania and Ethiopia (Vigna et al., 2011a; Kuwi et al., 2018; Tegegn et al., 2019).

These studies revealed significant genetic diversity between and within the main Urochloa species, and this finding has also been observed phenotypically. Hundreds of SSR markers are available specifically for Urochloa species, and many of these markers are transferable to different species (Jungmann et al., 2009a,b; Vigna et al., 2011b; Silva et al., 2013; Santos et al., 2015a; Ferreira et al., 2016; Souza et al., 2018; Table 4). Therefore, further studies must be carried out with the available molecular markers, mainly involving characterization of the entire germplasms. For instance, the molecular characterization of U. decumbens germplasm accessions still needs to be performed. In addition, molecular studies to characterize the diversity and genetic structure of Urochloa spp. germplasm and hybrids can be performed rapidly and effectively using the latest genotyping technologies, including genotyping-by-sequencing (GBS).

The grass family (Poaceae) comprises subfamilies that are divided into two large clades: the BOP (Bambusoideae, Oryzoideae, and Pooideae) clade and the PACMAD (Panicoideae, Aristidoideae, Chloridoideae, Micrairoideae, Arundinoideae, Danthonioideae) clade (Kellogg, 2015, Soreng et al., 2015, Soreng et al., 2017, Saarela et al., 2017). The subfamily Panicoideae (approximately 3,500 species in 12 tribes; Bhatt and Thaker, 2021) forms a monophyletic group that includes, in addition to Urochloa, other important economic species, such as maize (Zea mays), sugarcane (Saccharum officinarum) and sorghum (Sorghum bicolor; Hodkinson, 2018). This monophyly has been supported by several molecular phylogenetic studies (Vicentini et al., 2008; Sánchez-Ken and Clark, 2010), including chloroplast and nuclear gene analyses, that aimed to establish a robust phylogenetic framework for unraveling the evolution of this ecologically and economically important subfamily (Kellogg, 1988; Soreng and Davis, 1998; Bhatt and Thaker, 2021).

Giussani et al. (2001) analyzed the molecular phylogeny of the subfamily Panicoideae using sequencing data from the chloroplast gene ndhF. Based on this study, the species were divided into three strongly supported clades, and species of the Urochloa genus were included in the Paniceae clade (base chromosome number of x = 9) together with the Setaria species and most of the Panicum and Pennisetum species. This study showed that Urochloa, Setaria, and Pennisetum are paraphyletic genera. Additionally, in Panicea, Kellogg (2015) placed Megathyrsus in Urochloa, and Washburn et al. (2015) found that the Urochloa and Digitaria genera are paraphyletic.

Urochloa belongs to the Melinidinae subtribe, which is monophyletic (Giussani et al., 2001; Vicentini et al., 2008; Salariato et al., 2010). A chloroplast phylogenetic study of this subtribe discussed the paraphyly of Urochloa (Salariato et al., 2009). In addition, a molecular phylogenetic tree found two major clades comprising species of the Melinidinae subtribe, and all Urochloa species were placed in clade II with some species of Eriochloa as well as Chaetium, Megathyrsus and Scutachne (Salariato et al., 2010).

Recently, genomic datasets (particularly plastomes) have been evaluated to investigate the evolutionary relationships in several grass subfamilies, including Panicoideae (Jones et al., 2014; Cotton et al., 2015; Burke et al., 2016; Saarela et al., 2018; Duvall et al., 2019; Orton et al., 2021), and have contributed to greater resolution of and support for relationships within and among grass subfamilies (Saarela et al., 2018). Based on transcriptome-based phylogenetic inference, Washburn et al. (2017) revealed a Paniceae nuclear species tree in which the Anthephorinae subtribe is a direct sister to the MPC clade (subtribes Melinidinae, Panicinae, and Cenchrinae), corroborating the findings reported by Vicentini et al. (2008). In this phylogenetic tree, U. brizantha was placed together with U. plantaginea, and both in the same clade that U. fusca and Megathyrsus maximus, another important tropical forage grass used in pastures.

The phylogenetic studies conducted specifically with Urochloa remain limited. Torres González and Morton (2005) analyzed 22 genotypes of different species of the Brachiaria and Urochloa genera using ribosomal internal transcribed spacer (ITS) sequences as the molecular dataset and eight morphological spikelet characters (out of 17) used by the informal grouping reported by Renvoize et al. (1996). The main objective was to analyze the phylogenetic relationships of these genera. These researchers observed that B. humidicola and B. dictyoneura were grouped in the same clade, whereas B. ruziziensis and B. brizantha were placed together in a different clade than B. decumbens. No relationship was detected between morphological traits and molecular results. In addition, the authors classified Brachiaria as a monophyletic complex within the genus Urochloa.

Using chloroplast DNA (cpDNA), Salariato et al. (2009) found that U. decumbens and U. brizantha were closely related and separate from U. humidicola and U. dictyoneura. Similarly, U. ruziziensis, U. brizantha, and U. decumbens were included in a strongly supported clade separate from U. humidicola, corroborating the results from other molecular studies. Therefore, these results were different from those found by Torres González and Morton (2005), who included U. decumbens in another group based on molecular and morphological data.

Pessoa-Filho et al. (2017) evaluated the phylogenetic divergence between Urochloa species based on complete chloroplast genomes and observed that most of the polymorphisms between the species were located in intergenic regions, reflecting their phylogenetic distances. The phylogenetic analysis of this study yielded well-supported clades, and the results corroborated the findings reported by Salariato et al. (2009, 2010): U. humidicola is a sister taxon of the U. ruziziensis, U. brizantha, and U. decumbens clades. Also, U. decumbens and U. brizantha chloroplast sequences were highly similar and the phylogenetic tree showed that these species are phylogenetically closer to each other compared to U. ruziziensis and U. humidicola. The above-mentioned study is the only analysis of complete chloroplast genome sequences of Urochloa species and generated data useful for genetic analysis.

Future phylogenetic studies with Urochloa species need to include more samples of each species and with different ploidies to provide better representation and thereby improve the understanding of the phylogenetic relationships between these species. With advances in genomics, new tools and resources are available to fill the gaps in the phylogenetic characterization of the Urochloa genus, including next-generation sequencing (NGS) data and bioinformatics methods. This advance has transformed molecular phylogenetics into phylogenomics (Mardis, 2011; Young and Gillung, 2019). As an example, GBS and RADseq (restriction site-associated DNA sequencing) are being used for phylogenomics analyses of important plant species (Girma et al., 2018; Blanco-Pastor et al., 2019), and RNA sequencing (RNA-Seq) is being used for the development of phylogenetic markers, including SSR and targeted enrichment (Vatanparast et al., 2018; Liu et al., 2021).

The field of cytogenetics addresses chromosomes and their inheritance, including their function, movement, numbers and structure. In addition, this field includes modifying the structure and behavior of chromosomes in relation to processes such as the recombination, transmission and expression of genes (Singh, 1993).

Urochloa species have various ploidy levels, ranging from diploid to nonaploid (Table 2), and this variation is characterized by a predominance of polyploidy, particularly tetraploidy (Valle and Pagliarini, 2009). The basic number of chromosomes was established by Darlington and Wylie (1955) as x = 7 or x = 9, with a predominance of the latter (Bernini and Marin-Morales, 2001; Valle and Pagliarini, 2009). However, cytological studies performed with U. dictyoneura and U. humidicola showed evidence for x = 6 as the base chromosome number for these species because of the presence of hexavalents in accessions with 2n = 36 and 2n = 42 chromosomes and octa- and nonavalents in accessions with 2n = 54 chromosomes (Risso-Pascotto et al., 2006; Boldrini et al., 2009).

Regarding the chromosome number, the highest value (2n = 90) was reported in U. bovonei (Spies and Du Plessis, 1986). For U. brizantha, U. decumbens, and U. humidicola, some studies have described plants with different numbers of chromosomes depending on the level of ploidy (Basappa et al., 1987; Bernini and Marin-Morales, 2001; Valle and Pagliarini, 2009; Jungmann et al., 2010). For U. ruziziensis, a diploid species, cytogenetic descriptions have shown 2n = 2x = 18 chromosomes (Bernini and Marin-Morales, 2001), whereas artificially induced tetraploid plants have 2n = 36 chromosomes (Timbó et al., 2014a).

Duplication of the chromosome number using colchicine was achieved for the first time in genotypes of U. ruziziensis (Gobbe et al., 1981; Swenne et al., 1981), later in diploid accessions of U. brizantha (Pinheiro et al., 2000) and U. decumbens (Simioni and Valle, 2009) and again in U. ruziziensis (Ishigaki et al., 2009; Timbó et al., 2014b). Artificial tetraploid plants have been important for the realization of inter- and intracrosses in breeding programs and thus allow exploration of the genetic variability maintained by apomixis (Felismino et al., 2010).

Chromosome research, particularly meiotic analysis, has been performed with Urochloa species and hybrids (Mendes-Bonato et al., 2002a,b, 2006, 2007; Risso-Pascotto et al., 2002, 2003a,b, 2004a,b, 2005; Junqueira-Filho et al., 2003; Boldrini et al., 2006, 2009, 2011a,b; Adamowski et al., 2007, 2008; Felismino et al., 2010, 2012; Ricci et al., 2010, 2011; Fuzinatto et al., 2012, 2017; Pagliarini et al., 2012; Souza, 2013). These studies have revealed several abnormalities typical of allopolyploids and segmental allopolyploids, such as asynchronous chromosome segregation, intragenomic pairing, and the presence of multivalents and micronuclei. More recently, de Paula et al. (2017) confirmed the allopolyploidy of U. decumbens 4x and U. brizantha 4x and Corrêa et al. (2020) observed that the diploid genotypes of these species are potential ancestors of allotetraploids. Both studies were performed via genomic in situ hybridization (GISH).

Karyotypes of Urochloa species were constructed by Bernini and Marin-Morales (2001), who were the first to describe the variations in the relative chromosome length and satellite position among accessions of Urochloa. These differences may be indicative of the coexistence of different genomes or chromosomal rearrangements in the species. Akiyama et al. (2010) reported differences in the number and morphology of ribosomal DNA (rDNA) sites in the accessions of U. brizantha, U. ruziziensis, U. humidicola, and the Urochloa hybrid Basappa & Muniy. cultivar Mulato (originated from the crossing between U. decumbens and U. ruziziensis). More recently, Moraes et al. (2021) characterized the karyotype of a pentaploid cytotype of U. brizantha by the mapping of rDNA sites and nuclear DNA content analysis. The karyotype organization that has been observed suggests an allopolyploid origin for U. brizantha, corroborating the results from a study of the physical mapping of rDNA genes in this species (Nielen et al., 2009). In another study, the occurrence, mapping, and distribution of gypsy retrotransposons in U. decumbens, U. brizantha, U. ruziziensis and U. humidicola showed that mobile elements, in addition to the chromosome number variation, contributed to karyotype evolution (Santos et al., 2015b; Worthington et al., 2021).

Although cytogenetic studies of the Urochloa genus are in progress, there remain many gaps to be filled. For example, the cytogenetic map of the genus is still incomplete, mainly due to the small size and morphological similarity of chromosomes (Nani et al., 2018). Again, the “era of omics” and computational advances can influence and expand the cytogenetics-related knowledge of Urochloa spp. For instance, oligo- and CRISPR-FISH have been used to improve our understanding of the karyotype and chromosomal variation and evolution in important crops, including polyploid species (Braz et al., 2017; Liu et al., 2019; Li et al., 2020). Chromosome-scale assembly based on single-chromosome optical mapping is also a recently developed approach for analyzing structural variations in large regions (Deschamps et al., 2018) that can provide important insights into genetic variation and for guiding genome assemblies (Leinonen and Salmela, 2020).

Genetic maps are models that describe the distances and relative positions of a group of genes or markers along the chromosomes of a given species. These maps estimate the recombination frequencies, and this estimation provides information about the inheritance of loci that control characteristics of interest (Bourke et al., 2018a). For cases in which the genome has not yet been fully sequenced, as is the case for most species of the genus Urochloa spp., genetic maps are valuable tools for understanding the genetic and genomic organization of species (Aitken et al., 2014). Even for species with a sequenced genome, searching for QTLs remains a strategy for maintaining a relationship between the phenotype and the genotype (Lewin et al., 2009). For example, a QTL found in a biparental population identifies genomic regions for which the parents are polymorphic and whose variation can be associated with phenotypic variation, which suggests a link between the genes and the trait in question (Thérèse Navarro et al., 2020).

In the case of polyploid species, as are most of the species of the genus Urochloa spp., the challenge in the construction of genetic maps is greater than that found with diploid species due to a greater number of allele combinations and the markedly higher number of possible genotypes in a segregating progeny (Bourke et al., 2018a; Mollinari and Garcia, 2019; Mollinari et al., 2020). In addition, the type of polyploidy is often not well defined (autopolyploid or segmental allopolyploid), which makes it difficult to determine the expected patterns of segregation assumed during the construction of a genetic map (Bourke et al., 2018a,b). These and other reasons, such as self-incompatibility and high levels of heterozygosity, explain the difficulty in building high-resolution genetic maps for Urochloa spp. Consequently, few genetic maps have been developed (Thaikua et al., 2016; Vigna et al., 2016a; Worthington et al., 2016, 2019; Ferreira et al., 2019; Worthington et al., 2021; Table 5).

In the first study, Thaikua et al. (2016) used the approach of utilizing separate parents to construct an amplified fragment length polymorphism (AFLP)-based linkage map of the apomictic cultivar “Mulato” (Miles et al., 2004). This cultivar was crossed with three plants of sexual U. ruziziensis cv. “Miyaokikoku,” which generated 84 hybrids. The linkage map had a total length of 1423.2 cM, with 272 AFLP markers, and the apomixis locus was assigned to linkage group 2. Additionally, QTLs associated with agriculturally important traits (morphological traits and percentage of filled seeds) were identified by simple interval mapping (SIM) and composite interval mapping (CIM).

Worthington et al. (2016) constructed the first genetic map based on SNP markers for Urochloa spp. from 167 hybrids generated from a cross between the maternal synthetic autotetraploid U. ruziziensis (BRX 44–0) and the paternal U. decumbens cv. Basilisk (CIAT 606). Separate genetic linkage maps were constructed for each parent following the two-way pseudo testcross strategy. For the maternal linkage map, 706 SNP markers were placed in 34 linkage groups, with a total length of 1985 cM. In contrast, 1,210 SNP markers and markers associated with apomixis were assigned to 36 linkage groups in the CIAT 606 genotype, with a total length of 2,693 cM. These maps were used to assess synteny with foxtail millet (Setaria italica (L.) P. Beauv), the closest relative of Urochloa spp. with a publicly available reference chromosome-scale genome. Synteny was highly conserved between U. decumbens, U. ruziziensis, and foxtail millet, with only one major structural rearrangement. This provides evidence for the quality of the genetic maps, enabling the assignment of linkage groups to chromosomes and the identification of homologs. The “apospory-specific genomic region” (ASGR) was mapped to position 42.5 cM of group 5c, a region syntenic with foxtail millet chromosome 5 (Zhang et al., 2012).

Using SSR markers, Vigna et al. (2016a) built the first linkage map to investigate apomixis in the hexaploid U. humidicola. A full-sib F₁ population obtained by crossing the sexual accession H031 (CIAT 26146) and the apomictic cultivar U. humidicola cv. BRS Tupi was used to generate an integrated genetic map via the multipoint approach based on Markov chains. Forty-nine linkage groups (LGs) were created, with a total length of 1702.82 cM. Eight homology groups (HGs) were formed, and the apo-locus that controls apospory was mapped in LG02 (Pessino et al., 1997, 2004).

More recently, another linkage map was obtained for U. humidicola using SNP, AFLP, SSR, insertion–deletion (InDel) and kompetitive allele specific PCR (KASP) markers (Worthington et al., 2019). The sexual accession CIAT 26146 [EMBRAPA Beef Cattle (EBC) H031] was crossed to the apomictic male parent CIAT 16888 (EBC H027), generating 124 hybrids that were used to construct separate parental linkage maps following the two-way pseudo-testcross strategy. As a result, two haplotype maps (each with 36 linkage groups) were generated. The maternal haplotype map had 2,589 markers and a length of 3,558 cM, and the paternal haplotype map had 1,621 markers and 4,363 cM. Additionally, the ASGR, which is known to be a single dominant Mendelian factor of apomixis in Paniceae grasses (Ozias-Akins and van Dijk, 2007), was mapped to position 55.8 cM of CIAT 16888 (paternal haplotype) linkage group 1b. Synteny analysis of the U. humidicola parental maps with the foxtail millet physical map supported the previous cytogenetic evidence for a base chromosome number of x = 6 in the species (Boldrini et al., 2011a). The authors observed a high degree of synteny between some chromosomes of U. humidicola and foxtail millet, U. decumbens and U. ruziziensis. In addition, they accounted for the fusion of three pairs of chromosomes, which is consistent with the large chromosome size of U. humidicola relative to other Urochloa species (Bernini and Marin-Morales (2001)).

Additionally, in 2019, Ferreira et al. developed the first intraspecific genetic map of U. decumbens. Crossbreeding was performed between sexual U. decumbens D24/27 (artificially tetraploidized with colchicine) and apomictic U. decumbens cv. Basilisk, and 217 hybrids were selected. This mapping population was used to construct a consensus linkage map using the additional information provided by the SNP allele dosages in contrast to traditional methods based on allele presence/absence data (Hackett et al., 2013). A total of 1,000 SNP markers were distributed throughout nine homology groups, with a cumulative length of 1,335.09 cM. In addition, the genetic map allowed the identification of genomic regions related to resistance to spittlebug (Notozulia entreriana Berg), which is the main pest insect that attacks U. decumbens. The percentages of phenotypic variation explained by the QTLs ranged from 4.66 to 6.24%, reflecting the median heritability ( H C 2 = 0.37 ) found for spittlebug resistance in the population.

Recently, a new genetic map was constructed for the same interspecific progeny of Worthington et al. (2016), composed for 169 hybrids generated from a cross between a synthetic autotetraploid accession of U. ruziziensis (BRX 44–02) and U. decumbens cv. Basilisk (CIAT 606), used as male parent (Worthington et al., 2021). The analyses were based on the genetic map of the male parent, which included 4,427 markers distributed in 18 linkage groups, as expected for the segmental allotetraploid U. decumbens (2n = 4x = 36). Once again, the authors observed synteny between U. ruziziensis and foxtail millet, with three large translocations between chromosomes 1 and 7, 2 and 6, as well as 3 and 5, and four inversions between tails in chromosomes 1 and 4, 2 and 9, and 2 and 3. This map allowed the mapping of three significant QTLs associated with root length, biomass under both control and Al3+ stress conditions, and root diameter under stress/control conditions, which explained 12.8 to 16.1% of the phenotypic variance (Worthington et al., 2021).

Genetic mapping is a widely used resource to identify genomic regions responsible for important agronomic traits. One of the first steps in MAS is the development of genetic markers and high-resolution linkage maps, and this step is followed by the mapping of loci related to traits of agronomic and commercial importance. However, despite efforts aimed at building genetic maps for species of Urochloa spp., there is a clear need to improve the quality and increase the resolution of these maps, to map target genomic regions, and to build new maps for other species, including U. brizantha. Recently, packages capable to use all possible segregations for autopolyploid and segmental allopolyploid species were implemented in the R software (polymapR - Bourke et al., 2018b; MAPpoly - Mollinari and Garcia, 2019; Mollinari et al., 2020; R Core Team, 2020). These advances may enable the construction of integrated linkage maps with high resolution in polyploid species of Urochloa spp.

In addition, the reference genomes currently available (Pessoa-Filho et al., 2019; Worthington et al., 2021), new statistical and computational resources, and the possibility of large-scale genotyping followed by estimation of multiple dosages of SNP loci may aid the development of genetic maps for Urochloa spp. with high density and resolution (Bourke et al., 2018a). Although these tools do not have an immediate impact on breeding, the scanning of important genetic information from the genome for genes related to target traits provides a crucial basis for MAS, in addition to promoting faster and more effective molecular improvements of Urochloa species.

Although genetic maps of biparental populations provide a framework for detecting rarely occurring alleles that have a large effect on the phenotype in breeding populations, GWASs, also known as association mapping studies, reflect historical recombination events in diversity panels and are an efficient approach for locating common alleles associated with phenotypes with higher mapping resolution than that achieved with linkage mapping. Together, linkage mapping and GWAS can detect QTLs and elucidate the genetic basis of important traits more efficiently (Talukder et al., 2019).

Similar to genetic maps, GWASs have also benefited from high-throughput marker systems, which allow the detection of markers rapidly and at low cost. However, associative mapping studies of polyploid species remain extremely recent, and only one such study has investigated the Urochloa genus (Matias et al., 2019a). This study was performed with a set of 263 tetraploid interspecific Urochloa hybrids using 26,535 SNP markers with different allele dosage configurations obtained by GBS. Also were evaluated scenarios including additive, dominance, and epistatic effects. Associations were studied regarding agronomic and nutritional forage traits, including the regrowth capacity and crude protein. Seven SNP markers associated with the main forage traits were identified and these allow a better understanding of the Urochloa genome for important economic traits. Additionally, these markers can be useful to speed up breeding program selection cycles.

The authors observed that it is possible to identify the same genomic regions using diploid and tetraploid models, although in some cases the allele substitution effect can be biased. Then, the authors suggest the use of markers with both diploid and tetraploid molecular configurations to account for all regions of the genome in allopolyploid species (Matias et al., 2019a).

This study opens new perspectives for understanding the inheritance of complex traits in Urochloa species and segmental allopolyploids in general. GWASs constitute a promising tool for the identification of molecular markers associated with characteristics of interest that can be used in MAS to accelerate breeding cycles. In the genetic improvement of Urochloa species, in addition to enabling the identification of markers associated with agronomic and nutritional characteristics, GWASs are an important resource in the search for markers linked to resistance to insect pests, such as spittlebugs, and other biotic and abiotic stresses that influence forage production, which are characteristics that were identified as research priorities in the next-generation breeding of tropical forages in Brazil by Pereira et al. (2018).

Genomic selection (GS), which was proposed by Meuwissen et al. (2001), is a powerful and modern breeding tool for identifying and exploiting superior genotypes based on the use of thousands of distributed markers across the genome to predict genetic values (Bhat et al., 2016). The improvement in the accuracy of the selected individuals and the reduction in the duration of the selection cycle time reflect the acceleration of genetic gains achieved by GS in breeding programs. At present, the selection process until the release of new perennial tropical forage cultivars takes approximately 10–15 years (Jank et al., 2014; Matias et al., 2019b). Because the genetic gains in yield achieved from the breeding of forage crops based on traditional methods of phenotypic selection are low (Casler and Brummer, 2008), GS has emerged as a potential opportunity to obtain significant gains in target traits and to reduce breeding cycles (Simeão-Resende et al., 2013; Simeão et al., 2021).

The incorporation of GS in breeding programs depends on several factors, such as the reproductive mode of the species, the genetic architecture and heritability of the characteristics of interest, and the total financial budget of the program, because large phenotyped and genotyped populations are necessary to estimate the effects of molecular markers (Yadav et al., 2020). In the case of tropical forage grasses, appropriately sized training populations, large panels of SNP markers and polyploidy are also important factors to be considered because these make GS even more challenging (Simeão et al., 2021). Therefore, although this potential tool has been widely applied in many species of economic importance, it has not yet been used in breeding programs of tropical forage grasses of the Urochloa genus (Simeão-Resende et al., 2013; Pereira et al., 2018; Simeão et al., 2021). Consequently, there is limited information regarding the application of genomic prediction and GS in tropical forage grasses compared with other important crops.

Simeão-Resende et al. (2013) performed a simulation study to compare the gain and accuracy of different GS methods considering the experimental plot designs used in forage breeding. This study concluded that the application of GS methods in tropical forage breeding has potential when the phenotypic evaluation of individual plants is incapable of predicting performance under sward conditions, when within-family selection pressure is difficult or impossible, and/or when phenotypic evaluations require long cycle times.

Using real data from a panel of interspecific hybrids genotyped by GBS, Matias et al. (2019b) evaluated the influence of multiple traits and allele dosage information on the genomic prediction accuracy in Urochloa spp. The authors concluded that genomic prediction should be used in Urochloa spp. genetic improvement with the aim of reducing time and costs because all models of GS evaluated provided greater genetic gains than phenotypic selection. In addition, SNP markers with allele dosages associated with additive, dominant, and multitrait factors have increased the accuracy of genomic prediction models.

A study of GS performed with Megathyrsus maximus (syn. Panicum maximum), which is another important polyploid forage grass used as pasture in tropical areas, resulted in low genomic prediction accuracies for traits such as leaf dry matter and regrowth capacity (de C Lara et al., 2019). Low values of prediction accuracy were also observed by Matias et al. (2019b), although GS allowed greater genetic gains than those obtained with phenotypic selection. Due to the polyploidy and high levels of heterozygosity of tropical forage grasses, there is a need to develop more accurate predictive models for the incorporation of GS in breeding schemes.

Advances in high-throughput technologies can influence the efficient use of GS in Urochloa breeding schemes with increases in the number and density of markers and large-scale phenotyping. Additionally, statistical and computational methods, including machine learning (ML) and simulation studies, can aid the development of predictive models most suitable for polyploids and the identification of the best potential GS-based breeding scheme designs, which would result in genetic gains with greater accuracy and shorter breeding cycles (Montesinos-López et al., 2021; Simeão et al., 2021).

The transcriptome is the set of RNA transcripts contained in a cell or tissue at a specific stage of development and/or under specific physiological conditions (Ozsolak and Milos, 2010). In this context, transcriptome studies allow identification of the molecular constituents of cells and tissues and understanding the functional elements of the genome and the molecular mechanisms involved in each condition, such as biotic stress in plants (Wang et al., 2009). The most effective approach for creating transcriptome profiles is RNA-Seq, which is performed using deep sequencing technologies (Wang et al., 2009; Kukurba and Montgomery, 2015).

Studies of the transcriptome of the Urochloa genus remain limited (Table 6). The first RNA-Seq study was performed with leaves from two genotypes of U. humidicola that are highly divergent in terms of phenotype and reproductive modes: the only sexual accession of the U. humidicola germplasm, known as BH031, and the apomictic cultivar BRS Tupi (Vigna et al., 2016b). The main objectives were to describe and compare the leaf transcriptome of these divergent genotypes, which were collected at the same field condition. As a result, approximately 76,000 transcripts were de novo assembled, which totaled 35,093 nonredundant unigenes. Most of these transcripts were annotated in the Phytozome database (Goodstein et al., 2011). From these unigenes, new EST-SSRs (4,489) and SNP markers (560,298) that may be associated with important traits for forage breeding, such as flood stress tolerance and the C4 pathway, were identified. Moreover, as in previous studies, a significant difference was observed between the analyzed genotypes, but now in relation to gene expression.

The next transcriptome of the genus Urochloa was assembled de novo from root sequences of U. decumbens cv. Basilisk under two controlled conditions: hydroponic solutions with and without aluminum (AlCl₃). The assembly generated 164,930 transcripts and approximately 69% of these transcripts were attributed to a putative function of the protein via comparison with different protein databases. In addition, 13,375 microsatellite markers were identified in the transcripts. Differential expression analysis revealed a great difference in plants exposed to Al stress, and it was possible to identify transcripts with putative functions related to resistance and aluminum exposure. Therefore, this transcriptome represents a great resource for aiding the elucidation of the main molecular mechanisms of aluminum tolerance in forage grass species (Salgado et al., 2017).

Another transcriptome of the Urochloa genus was also assembled to investigate the molecular basis of toxic aluminum tolerance because this condition directly affects forage productivity and is common in tropical soils. Worthington et al. (2021) sequenced root and stem tissue samples of two replicated plants of U. decumbens CIAT 606 (cv. Basilisk) and U. ruziziensis BRX 44–02 under two conditions: growth in a high aluminum concentration (200 μM AlCl₃) and under control conditions (0 μM AlCl₃). As a result, most of the differentially expressed (DE) genes were regulated at the root of U. decumbens, which is a genotype classified as tolerant to aluminum. The authors identified different genes related to the stress response to aluminum, and the evidence shows that the tolerance mechanisms found in Urochloa are the same as those that occur in rice, another grass species.

The most recent transcriptome of the Urochloa genus investigated the transcriptomic profiles of leaves and roots of three Urochloa interspecific hybrid genotypes with the onset of water stress (Jones et al., 2021). According to a previous analysis, the genotypes were characterized as exhibiting good (gt-17), intermediate (gt-9) and poor tolerance (gt-18) to drought. As a result, the genotypes with good and intermediate tolerance to drought showed enriched DE genes that were more similar to each other. In addition, DE genes related to carbohydrate and cell wall metabolism in the leaves were upregulated in gt-9 and gt-18 and downregulated in gt-17. Because the level of water stress increased for all evaluated genotypes, an excess of downregulated putative apoplastic peroxidases was detected in the roots, which suggests that changes in the architecture of the root cell wall occur in response to the stress. As tolerance to abiotic stresses is one of the main objectives of Urochloa breeding, this study generated important information and knowledge about genes and metabolic pathways involved in the response to water stress (Jones et al., 2021).

Although these transcriptomes represent only a part of the transcripts of three species of Urochloa, they constitute important contributions to the scientific community interested in forage grass and breeding programs. The results provide relevant information on the genomic resources available for Urochloa grasses, mainly due to the possibility of identifying new candidate genes potentially related to characteristics of economic interest. These candidate genes can be further investigated by coexpression networks, which provide valuable information on the biological relationships between genes.

Even so, each species of Urochloa has its own peculiarities, and other transcriptome studies in which a specific stage of development and/or a physiological condition is considered must be performed. For instance, genetic information for some important characteristics, such as resistance to spittlebugs, apomictic reproduction, and flood tolerance, is still lacking, and transcriptomic studies are promising for generating new knowledge about these characteristics. The elucidation of different strategies of spittlebug resistance, such as information about the possible genes involved in this resistance and molecular markers in these genes, provides powerful information for breeding strategies and future studies, including GWAS and GS.

NGS has allowed genome sequencing to be a reality not only for model species but also for nonmodel or “orphan” species and polyploid organisms, such as Urochloa spp. However, despite this greater accessibility, the different levels of ploidy and different reproductive modes, in addition to the availability of different species with specific characteristics, are obstacles to obtaining a representative genome for Urochloa pastures.

The fact that breeding programs were recently created and the lack of financial resources (Pereira et al., 2018) also directly influence the availability of fundamental genomic resources. The existence of only two sequenced genomes representing these pasture grasses is a reflection of the above-described finding (Pessoa-Filho et al., 2019; Worthington et al., 2021). The first, the full characterization of which is not yet available, is a diploid chromosome-scale genome of a heterozygous clone of U. ruziziensis, named C6. Sequencing with PacBio Sequel, based in SMRT (Single Molecule Real Time) technology, generated more than 13.3 million long reads (~142x coverage) and allowed assembly of a genome covering 98.2% of the estimated haploid genome size of 615 Mbp for ruzigrass. Most of the assembled genome (94%) was included in nine scaffolds that correspond to the nine chromosomes of U. ruziziensis, with an N50 of 66 Mpb. Other metrics regarding genome completeness were satisfactory and included an NG50 of 286 kbp, 95.2% BUSCO (Benchmarking Universal Single-Copy Orthologs) complete matches and 83.3% complete single copies (Pessoa-Filho et al., 2019).

Additionally, using a diploid U. ruziziensis genotype (CIAT 26162) as a DNA source, a whole-genome assembly (WGA) was obtained by sequencing with short reads (approximately 100 × coverage) and presented satisfactory completeness (Worthington et al., 2021). The repetition content representing 51% of the total genome and the large number of Gypsy and copia long terminal repeats (LTRs) found were similar to those observed in foxtail millet (Setaria italica (L.) P. Beauv; Zhang et al., 2012), a species that is widely used as a reference in genomic studies of tropical forage grasses. Moreover, approximately 42,000 coding genes have been annotated, and most of these genes encode a protein showing homology to a protein in S. italica.

Worthington et al. (2021) prioritized the assembly of a diploid genome rather than a tetraploid genome because Urochloa grasses are heterozygous outcrossing species. Even so, a polyploid genome is essential for better representing the main accesses used for breeding and reflecting their genetic information, particularly the heterozygosity present. The recent emergence of third-generation sequencing (TGS) technologies can aid the complex and challenging assembly of these polyploid plant genomes because it produces high-quality genome assemblies with high resolution due to the longer length of the reads (Kyriakidou et al., 2018). Consequently, more genomic information is provided, and more genomic variants will be identified to aid genomic studies, such as those that address the genotype–phenotype–environment relationship, which will provide a great resource for improving Urochloa breeding programs.

We also emphasize that these available genomes, despite being diploid, represent an invaluable resource for orphan species because they aid the development and application of genomic tools for breeding and genetic studies of Urochloa grasses (Tang, 2017; Kyriakidou et al., 2018). Until then, our understanding of genomic organization was based on genetic maps, but it is now possible to go even further. The analyses that require the identification of variants, such as GS, GWAS, and genetic mapping, will benefit greatly from this genome because the use of genomes of species genetically close to Urochloa spp. as a reference is associated with a substantial loss of data. Therefore, exclusive information on Urochloa grasses may be revealed, and new approaches can be devised now that this fundamental genomic resource is available.

A polyploid individual has a genome with more than two homologous sets of chromosomes, which is a condition that is quite common in all groups of plants and one of the major forces in plant evolution and speciation (Hegarty et al., 2013; Mason and Wendel, 2020). Polyploids often have larger cells in relation to their ancestral diploids, which can cause an increase in the whole size of the plant (“giga effect”; Stebbins, 1950), in addition to other advantages in genome buffering, vigorousness, and robust adaptation to environmental changes (Tamayo-Ordóñez et al., 2016). From a genetic point of view, polyploids have multiple alleles associated with a single locus, which implies high heterozygosity and more complex segregation compared with that observed with diploids (Schaart et al., 2021), and thus, the traditional breeding of polyploid crops is markedly more challenging.

Many important forage grasses are natural polyploid species, and others have chromosomes artificially duplicated by the use of colchicine, a chemical agent that can be used to produce artificial polyploids on demand. The induction of polyploidy can be an important tool in the genetic improvement programs of some species, which have three basic purposes: polyploidization of a hybrid to restore its fertility or synthesize a new species; polyploidization to obtain a larger and better plant (“giga effect”); and polyploidization to overcome the barrier imposed by different ploidy levels and allow intra- or interspecific crossing (Dewey, 1980; Sattler et al., 2016).

In the U. ruziziensis/U. decumbens/U. brizantha agamic complex, even though sexual individuals are known only in the wild as diploids, they have artificially duplicated chromosomes, which results in the induction of autotetraploids (Gobbe et al., 1981; Swenne et al., 1981; Pinheiro et al., 2000; Simioni and Valle, 2009; Timbó et al., 2014b). These polyploidized sexual individuals are essential for hybridization with apomictic individuals in breeding programs, which allows intra- and interspecific crossings and result in increased genetic variability and enhanced selection.

Natural polyploid species may originate from genomic duplication within a single species, which leads to autopolyploids, or by a hybridization of two closely related species (“subgenomes”), which results in allopolyploids. The identification of each class is frequently accomplished by analyzing meiotic pairing and inheritance patterns: autopolyploids exhibit polysomic inheritance, with pairing and recombination between all homologous copies of each chromosome, whereas allopolyploids display disomic inheritance, in which more-related chromosomes (“homologs”) may pair and recombine more frequently than less-related chromosomes (“homeologs”). However, in many cases, the evolutionary history is more complex, and pairing between related chromosomes from different subgenomes (“homeologous”) occurs to a limited extent. In such cases, individuals are classified as segmental allopolyploids (Bourke et al., 2018a; Mason and Wendel, 2020; Schaart et al., 2021).

Meiotic studies and molecular karyotyping of cultivar and hybrid genera have revealed several abnormalities typical of autopolyploid and segmental allopolyploid in U. decumbens (Mendes-Bonato et al., 2001, 2002b; Worthington et al., 2016), U. brizantha (Mendes-Bonato et al., 2002a, 2006; Mendes et al., 2006), U. ruziziensis (Mendes-Bonato et al., 2006), and U. humidicola (Vigna et al., 2016a; Worthington et al., 2019). These abnormalities, such as distorted chromosome segregation, asynchrony during microsporogenesis, preferential (intragenomic) pairing, and the presence of multivalent and micronuclei, demonstrate the segmental allopolyploid composition of these genomes.

Based on genomic in situ hybridization, de Paula et al. (2017) recently suggested the allopolyploid origin of U. brizantha and U. decumbens, and such results were validated by additional GISH analysis (Corrêa et al., 2020). The authors proposed genomic compositions of BBB¹B¹ for U. brizantha, B¹B¹B²B² for U. decumbens, and B²B² for the diploid U. ruziziensis. The findings indicate U. ruziziensis as the ancestral donor of genome B², while the origin of genomes B and B¹ remains unknown, but eligible candidates would be diploid genotypes of U. brizantha and U. decumbens. The probes used in GISH experiments reveal the homology between the three genomes (B, B¹ and B²) and highlight the greater proximity between all three species, as previously suggested (Pessoa-Filho et al., 2017; Triviño et al., 2017).

An allopolyploid is not strictly the sum of two different genotypes, and most studies that investigated the process of polyploidization observed that genomes usually undergo a series of evolutionary processes after polyploidization, such as natural chromosome duplication, genome reorganization, changes in the gene expression pattern, and sub- and/or neofunctionalization of duplicate genes (Chen and Ni, 2006; Renny-Byfield and Wendel, 2014). A study performed using meiotic analyses and SSR amplification patterns exemplified the occurrence of such processes in U. humidicola (Vigna et al., 2016a). The authors suggested the allopolyploid origin of the genotypes H031 and cv. BRS Tupi from a cross between two different genomes, resulting in a triploid parent (2n = 3x = 18, ABB) that, after natural chromosome duplication, formed an allohexaploid (2n = 6x = 36, AABBBB).

Worthington et al. (2019) reinforced the thesis of AABBBB constitution for the apomictic genotype CIAT 16888; however, linkage and molecular karyotyping results showed no evidence of subgenome differentiation in the H031 genotype (CIAT 26146). More studies on the genomic composition of the sexual genotype H031 are needed. As noted by Worthington et al. (2019), if the sexual and other apomictic accessions of U. humidicola do indeed have different genomic compositions, H031 may need to be classified as a separate species or subspecies.

Despite the widespread occurrence of polyploidy in the Urochloa genus, many questions remain unanswered, which demonstrates the need to conduct additional research to understand the genome composition and evolution in tropical forage grasses. A high-quality reference genome is also fundamental to capture the variation and to better understand these economically important genomes (Tang, 2017; Kyriakidou et al., 2018). The last decades have seen remarkable advances in the field of polyploidy, particularly in molecular breeding approaches that combine QTL, GWAS, and GS models with allele dosage information (Sattler et al., 2016) and in tools for improving the genome sequencing and assembly of polyploid plant crops (Jiao and Schneeberger, 2017). We hope that these tools can aid a deeper understanding of the complexities of genetic variation that underlie the phenotype, which would allow plant breeders to precisely manipulate the genomes of Urochloa species to achieve outstanding results.

Apomixis is defined as an asexual reproductive mode in which the formation of seeds occurs without fertilization, which results in exact genetic replicas (clones) of the female parent (Asker and Jerling, 1992; Hand and Koltunow, 2014). This condition allows us to capture, fix, and propagate the hybrid vigor of superior heterozygous genotypes per seed (Sailer et al., 2016) in a simpler manner compared with vegetative propagation or the production of hybrid seeds from inbred parental lines (Worthington et al., 2016). Several studies have focused on investigating the genetic control of this reproductive mode for practical applications, particularly in agriculture, but much knowledge remains unknown (Hand and Koltunow, 2014; Kaushal et al., 2019; Schmidt, 2020).

In the Urochloa genus, the type of apomixis is aposporic gametophytic, in which fertilization of the endosperm (pseudogamy) is necessary (do Valle, 1990; Lutts et al., 1994). Traditionally, the characterization of the reproductive mode of Urochloa species is conducted by embryo sac analysis, which is a very expensive and laborious method that requires adult plants for evaluation. Sexual and apomictic embryo sacs are often found on the same plant or even in the same ovule, and the observation of at least one aposporic embryonic sac indicates facultative apomixis (do Valle, 1990).

Most studies that investigated the inheritance of apomixis concluded that it is simply inherited and that the trait is conferred by a single dominant Mendelian factor (Valle et al., 1994; Miles and Escandón, 1997; Pessino et al., 1997, 2004), denoted as the ASGR (Ozias-Akins and van Dijk, 2007). Therefore, in crossbreeding, progeny segregates to the reproductive mode on a 1:1 basis (Valle and Savidan, 1996; Barrios et al., 2013; Worthington and Miles, 2015; Vigna et al., 2016a).

Studies were performed to investigate the chromosomal duplication of sexual accessions in some Urochloa species (Swenne et al., 1981; Pinheiro et al., 2000; Simioni and Valle, 2009). In addition, the genetic diversity maintained in apomictic accessions has recently started to be explored in breeding programs through crosses between apomictic (used as pollen donors) and sexual genotypes (Miles, 2007). In addition to increasing genetic variability, these crosses allow the selection of superior genotypes (Simeão et al., 2021). The progeny obtained from these crosses need to be phenotyped with respect to the reproductive mode with the aim of the next hybridizations. Sexual offspring can be used as female parents in future crosses, and apomictic offspring are potential cultivars as well as possible male parents. Therefore, the search for markers linked to apomixis is one of the main demands of breeding programs for the MAS of hybrids, mainly in Brazilian programs. The use of molecular markers that cosegregate with apomixis represents enormous potential and enables the rapid screening of thousands of progenies at the seedling stage (Worthington and Miles, 2015).

In the CIAT breeding program, MAS for apomixis has been performed since 2009 based on a SCAR marker called “N14.” This marker was developed from a RAPD marker linked to apomixis in U. decumbens cv. Basilisk and U. brizantha cv. La Libertad (Pedraza Garcia, 1995) and is able to detect apomixis mainly when cv. Basilisk is used as the pollen donor parent. For more genetically distant species, such as U. humidicola, this marker is not effective (Worthington et al., 2016).

Comparative molecular analyses and cytological studies have revealed that the region that controls apomixis (ACR) is apparently located in a chromosomal region with no recombination events documented to date (Ortiz et al., 2013). Therefore, this genomic structure exhibits a high level of conservation among apomictic species. For instance, the synteny of this region has already been found for Urochloa with rice and corn (Ozias-Akins and van Dijk, 2007).

To date, genetic mapping is one of the most frequent strategies used in investigating apomixis at the molecular level and in the search for markers intrinsically linked to this characteristic in Urochloa spp. Pessino et al. (1997, 1998) genotyped Urochloa hybrids with AFLP and RFLP markers identified in maize and rice and observed a total of nine markers linked to apomixis in U. brizantha cv. Marandu and the close synteny of some markers to the region of maize chromosome 5 and rice chromosome 2. A RAPD marker was identified as linked to apomixis in U. humidicola (Zorzatto et al., 2010). Additionally, in U. humidicola, Vigna et al. (2016a) mapped the locus that controls apospory in LG2, which is located 19.4 cM from the nearest SSR marker, and this finding corroborates the results reported by Pessino et al. (1998).

The ASGR was mapped to U. decumbens chromosome 5 in a region showing synteny with foxtail millet chromosome 5. In addition, the PsASGR-BABY BOOM-like (psASGR–BBML)-specific primer (p779/p780), whose candidate gene for the apomixis parthenogenesis component was first described in Pennisetum squamulatum (Ozias-Akins et al., 1998), was linked with ASGR, which indicates that this region is highly conserved across Paniceae species (Worthington et al., 2016). Following this same strategy, Worthington et al. (2019) mapped the ASGR in U. humidicola chromosome 1, a region syntenic with chromosomes 1 and 7 of foxtail millet. Although the ASGR was mapped to a different carrier chromosome than previously identified in U. decumbens, this study reinforced the postulation of ASGR-BBML as candidate genes for the parthenogenesis component of apomixis.

Some studies have suggested that events related to apomixis occur due to the dysregulation or suppression of genes involved in sexual reproduction (Ozias-Akins and van Dijk, 2007) or to rearrangements in the expression of genes that constitute the sexual pathway (Grimanelli et al., 2001). Thus, genetic studies have improved the understanding of the molecular mechanism of the different modes of reproduction in Urochloa species, which have particularly focused on apomixis (Rodrigues et al., 2003; Alves et al., 2007; Silveira et al., 2009, 2012; Lacerda et al., 2012; Guimarães et al., 2013; Ferreira et al., 2017; Koehler et al., 2020). Most of these studies were performed with U. brizantha, and the most recent study concluded that somatic embryogenesis receptor-like kinase (SERK) genes present different expression during somatic embryogenesis and sporogenesis of sexual and apomictic genotypes, which suggests a possible downregulation associated with apomictic development (Koehler et al., 2020).

These studies at the gene level open new perspectives for the transfer of apomixis to sexual plants in the future via gene editing, but other genomic approaches, including bacterial artificial chromosomes (BACs), should be applied for the identification and characterization of the genomic regions that control apomixis in Urochloa spp. In addition, a greater understanding of the genetic basis and regulatory mechanisms underlying apomictic seed development, including epigenetic mechanisms, is needed. In addition, efforts to develop a diagnostic marker test for apomixis are still highly desirable for the early identification of apomixis because it can shorten selection cycles and reduce the costs of breeding programs.