Genetic relationships of Arachis (Fabaceae) accessions based on microsatellite markers

The genus Arachis is endemic to South America and contains 83 described species assembled into nine taxonomical sections. The section Arachis is of particular interest because it includes the cultivated peanut (A. hypogaea) and its closely related wild species. In this study, we used 26 microsatellite markers to analyze the genetic variability and relationships of some recently collected germplasm accessions of species in the Arachis section, with emphasis on the B genome species. The knowledge of the genetic relationships among species and accessions is necessary for a more efficient management of germplasm collections and use of wild species for crop improvement. This is especially important for the B genome species, as only one accession of A. ipaënsis, the B genome donor to the allotetraploid A. hypogaea (AABB), is available in germplasm collections worldwide. The results shed more light on the genetic relationships between accessions of A. ipaënsis, A. gregoryi, A. magna, A. valida and A. williamsii, what expands the number of accessions for incorporation of useful genes from the species associated with the peanut B genome. The analyses also showed a generally high level of intraspecific genetic variability, but usually grouped the accessions according to their genome types and species. However, accessions of some species did not group as expected, and these results suggest the need of further taxonomic revision of a few taxa, especially some accessions of A. gregoryi, A. magna and A. kuhlmannii and the circumscriptions of sections Erectoides and Procumbentes.


Introduction
The genus Arachis contains 83 described species, assembled into nine taxonomic sections according to their morphology, cross-compatibility relationships and geographical distribution in South America (Krapovickas and Gregory 1994;Simpson 2005, 2017;Valls et al. 2013;Seijo et al. 2021). Brazil is the largest holder of wild Arachis, as 65 species of all nine sections occur in its territory and 46 are exclusive to the country. Most of the Arachis species are diploid with 2n = 2x = 20, five are tetraploid (2n = 4x = 40), and four are aneuploid or dysploid, with 2n = 2x = 18 chromosomes (Fernández and Krapovickas 1994;Krapovickas and Gregory 1994;Lavia 1998;Peñaloza and Valls 2005;Ortiz et al. 2017;Silvestri et al. 2017).
The section Arachis is the most important because it contains the cultivated A. hypogaea, and its wild progenitors: A. duranensis and A. ipaënsis (Kochert et al. 1996;Seijo et al. 2004;Fávero et al. 2006;Grabiele et al. 2012;Moretzsohn et al. 2013;Bertioli et al. 2016). Within section Arachis, six genome types (A, B, D, F, G, and K) have been described, based on classical and molecular cytogenetic studies (Husted 1933;Stalker 1991;Fernández and Krapovickas 1994;Seijo 2008, 2010;Silvestri et al. 2015). Most species in the Arachis section have an A genome type, characterized by the presence of a so-called A chromosome pair, which has a reduced size (Husted 1936) and a lower level of euchromatin condensation in comparison to the other chromosomes (Seijo et al. 2004). Genome A species were arranged into three subgroups (Chiquitano, La Plata River Basin, and Pantanal) based on the variability observed in the heterochromatin and 18S-26S rRNA loci (Robledo et al. 2009). The remaining diploid species with x = 10 do not have the "A chromosome" pair and have been divided into four genome types: (1) B stricto sensu, without centromeric bands and similar to the B genome of A. hypogaea; (2) K, showing large centromeric bands and similarities with the Chiquitano A genome subgroup; (3) F, with reduced but uniform amount of centromeric heterochromatin, and (4) the D genome, characterized by an asymmetric karyotype with several subtelocentric or submetacentric chromosomes, only found in A. glandulifera (Stalker 1991;Seijo 2008, 2010). Finally, the G genome was described for three species of section Arachis with reduced basic chromosome number (x = 9), and uniform DAPI bands (Silvestri et al. 2015).
In the last decades, some 1250 accessions of 63 described native wild Arachis species have been collected with germplasm in Brazil (http:// alelo bag. cenar gen. embra pa. br/ Alelo Consu ltas). Wild species are highly promising sources of genes for desirable agronomic traits, such as earliness, drought tolerance, resistance to foliar diseases, viruses, and nematodes (Stalker 2017). The characterization of these new germplasm accessions is important for their conservation and use for the improvement of cultivated peanut.
Microsatellite or SSR (Simple sequence repeat) markers have been widely used to analyze the genetic variability in plant species, since they are multiallelic, polymorphic, randomly distributed through plant genomes, and typically codominant markers. In addition, microsatellites have proven to be highly transferable between species of Arachis (Moretzsohn et al. 2004(Moretzsohn et al. , 2013Hoshino et al. 2006;Gimenes et al. 2007;Koppolu et al. 2010).
The objective of this study was to analyze the genetic variability and relationships of some recently collected accessions of species in the Arachis section, highlighting the five previously described species determined to have a B genome stricto sensu (Robledo and Seijo 2010), that is, not yet considering the recently described A. inflata (Seijo et al. 2021), and compare them with accessions of the Erectoides and Procumbentes sections. The genetic variation within and between 31 species of Arachis was analyzed using microsatellite markers, contributing to a more efficient conservation and use of this germplasm. We also took the opportunity to examine some small doubts regarding the identity of a few accessions, as described below, which can become a burden in the management of germplasm.

Plant material
Plants were obtained from the Genebank of Wild Species of Arachis, maintained at Embrapa Genetic Resources and Biotechnology (Cenargen), Brasília-DF, Brazil (http:// alelo bag. cenar gen. embra pa. br/ Alelo Consu ltas, Williams 2022). With the exception of the shy seed producer A. pflugeae, transplanted directly from its natural site, all plants were grown from seeds and maintained in greenhouses. A total of 93 accessions belonging to three taxonomic sections were included. These accessions represent the six varieties of A. hypogaea (Krapovickas and Gregory 1994), three landraces of A. hypogaea from the Xingu Indigenous Park (Freitas et al. 2007) and 30 wild species, being 24 of section Arachis, four of section Procumbentes and two of section Erectoides (Table 1).
Arachis magna accessions include plants representing a single field population but showing three flower colors, V 13761-orange, yellow and cream, as well as the additional sample V 14727, from the same site but collected in the wild six and a half years later. Also from a single location, V 13765 and V 14707 were collected with this same time interval. A similar situation occurs with the accession V 6389 of A. gregoryi, collected again in the wild as V 14957, in what was believed by the same original collectors to be the same geographic location, respectively in 1981 and 2004. This long period of time, associated with obvious changes in the dirt road system and local land use recommended the initial designation of these germplasm samples as distinct accessions.
Arachis batizocoi includes a field accession (K 9484) as well as a mutant of the same observed when cultivated by Walton C. Gregory in Raleigh, at North Carolina State University, showing a thick, ribbed epiphyll, and locally designated "Corduroy". Likewise, the two samples of A. cruziana showed small differences in anthocyanin pigmentation and hairiness, observed during the initial seed multiplication.
The geographic coordinates and altitude of the Brazilian accessions were obtained directly on site with GPS equipment, or accurately verified, based on field notes. Geographic information on accessions from Argentina, Bolivia and Ecuador was extracted from the literature (Krapovickas and Gregory 1994) or from herbarium labels, or retrieved from germplasm databases.
Of the 30 wild species, 14 are represented by plants directly multiplied from seeds rescued from nature at the same site of collection of the herbarium vouchers later designated as type specimens. In addition, A. pflugeae, a species with very limited seed production, was sampled from plants transplanted directly from the wild population documented by the type specimen in its original collection.

DNA extraction and PCR amplification
Total DNA was extracted from young leaflets as described by Grattapaglia and Sederoff (1994) modified by the inclusion of an additional precipitation step with 1.2 M NaCl. The DNA was diluted in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0), and the concentration estimated in 1% agarose gels. PCR products were obtained using microsatellite primer pairs designed for A. hypogaea or A. stenosperma (Moretzsohn et al. 2005). PCRs were performed in 6 μl volumes, containing 0.6 μl 1X PCR buffer (10 mM Tris-HCl pH 8.4, 50 mM KCl, 2 mM MgCl 2 ), 1.2 μl BSA (2.5 mg/ml), 1.0 μl dNTP (2.5 mM), 0.1 μl primer pair (5 μM each), 0.2 μl Taq DNA polymerase (5 units/μl), 0.9 μl ultra-pure water, and 2.0 μl DNA (5 ng/μl). The amplification cycle consisted of heating at 94° C for 5 min, followed by 30 cycles of 94° C for 1 min, annealing temperature specific for each primer pair (48°-60° C) for 1 min, and 72° C for 2 min, and a final extension at 72° C for 7 min. The reactions were performed on ABI 9700 thermocyclers (Applied Biosystems, CA, USA). The forward primer was labeled with one of the fluorescent dyes HEX, 6-FAM or NED (Applied Biosystems). The primers were multiplexed according to the fluorescence, annealing temperature and size of the amplified alleles. PCR products were denatured and size fractioned by electrophoresis on an ABI 3730 DNA Analyzer (Applied Biosystems). Allele sizing of the electrophoretic data was performed using GeneMapper 4.1 software (Applied Biosystems).

Data analysis
A total of 26 microsatellite markers were included in this analysis. Pairwise genetic distances were estimated from the allelic data using the modified Rogers' genetic distance (Goodman and Stuber 1983) and the software BOOD (Coelho 2000). The resulting diagonal matrix was then submitted to cluster analysis using UPGMA (unweighted pair-group method analysis). The reliability of the generated dendrogram was tested by bootstrap analysis also using the BOOD program with 10,000 iterations. The consistency of the resulting dendrogram was also evaluated by the cophenetic correlation coefficient (r), evaluated by Mantel's test at 5% significance level and with 1,000 permutations. All these analyses were performed using NTSYS 2.21 (Rohlf 1993). The total number of amplified alleles (A), observed (Ho) and expected heterozygozities (He) were estimated for each species using GDA (Lewis and Zaykin 2001).    1). Therefore, all the accessions were differentiated using the 26 loci. Based on this data, a dendrogram was constructed using the UPGMA method (Fig. 1). The cophenetic correlation coefficient (r) was 0.84 and significant by the Mantel's test (p < 0.01), indicating that the tree reflected the original genetic distance matrix. Eleven major groups were evident (Fig. 1), although distances between groups were usually very low, with low bootstrap values (Supplementary File 1).
Group 1 consisted of the only know accession of A. ipaënsis (B genome) and the tetraploid accessions, including representatives of the six botanical varieties of A. hypogaea, the landraces cultivated by Brazilian Indians (Xingu), and A. monticola. The accessions of A. hypogaea clustered into two subgroups, separating the two subspecies, with the exceptions of the subsp.   Accessions of the same species tended to group together, with the exception of some few A. magna and A. gregoryi accessions (Fig. 1). All accessions of the three K genome species (A. batizocoi, A. cruziana, and A. krapovickasii) and the only known D genome species (A. glandulifera) formed a clear group (Group 3), closely related to the B genome species group. Group 5 contained the two accessions of A. vallsii, which was originally described as a member of the section Procumbentes by Krapovickas and Gregory (1994). The six accessions of A. hoehnei were joined in Group 6, closely related to a group containing most of the A genome species analyzed, and the only F genome species included in the present study, A. benensis (Group 7). Group 8 contained the six accessions of the three G genome species (2n = 18), while Group 9 had three accessions of A. kuhlmannii and the only A. kempff-mercadoi accession included in this study. Five of the six accessions of sections Procumbentes and Erectoides clustered on Group 10, while A. kretschmeri (section Procumbentes) formed an external group (Group 11).

Discussion
The present analysis of genetic relationships based on microsatellite markers showed, in general, the grouping of accessions according to their assignment to species and genome types, corroborating the taxonomic and cytogenetic classification.
The landraces of A. hypogaea cultivated by the Brazilian Indians in the Xingu Indigenous Park have some unique morphological traits, especially in the pods, that exceed the variation described for the six formal botanical varieties. However, the ten accessions of A. hypogaea grouped together (Group 1), showing the strong affinity of its six varieties, as well as the three Xingu Indigenous Park types, identified as White Kayabi, Nambikwara and Xingu (Freitas et al. 2007). The A. hypogaea accessions were very closely related to A. monticola, a species probably involved in the origin and domestication of peanut (Gregory and Gregory 1976;Raina and Mukai 1999;Lavia et al. 2008;Bertioli et al. 2019), and A. ipaënsis, which shows high cytogenetic homology (Kochert et al. 1996;Seijo et al. 2004Seijo et al. , 2007Fávero et al. 2006;Robledo and Seijo 2010) and an extremely high DNA similarity (99.96%) with the B genome of A. hypogaea (Bertioli et al. 2016). Arachis ipaënsis is considered the donor of the B genome to A. hypogaea (Kochert et al. 1996;Seijo et al. 2004;Fávero et al. 2006;Bertioli et al. 2016). Our results corroborate this assumption, since A. ipaënsis was the only B genome species placed in the group containing the A. hypogaea accessions (Group 1). This group was closely associated to Group 2, which contained most of the B genome accessions. Only a single accession of A. ipaënsis is available in germplasm collections worldwide, which was collected in Bolivia, from where it may now have disappeared (Williams 2022;Supplementary File 2). The genetic similarity of A. ipaënsis and the species of Group 2, composed by a significant number of accessions of A. magna, A. gregoryi, A. valida, and the only known accession of A. williamsii, and the knowledge of their genetic relationships, open new possibilities for the incorporation of useful genes into cultivated peanut. Some of the accessions of A. gregoryi and A. magna included in the present study had never been analyzed by molecular markers.
The five accessions of A. valida grouped together. These same accessions were included in our former study (Moretzsohn et al. 2013) and V 9153 did not cluster as expected. This result raised questions about its taxonomy, but our present study showed that it belongs to A. valida, which is a well-supported species. In contrast, accessions of A. magna and A. gregoryi grouped into different subgroups, and three A. magna accessions, including the two representatives of the easternmost natural population, formed a differentiated group (Group 4). The intraspecific variability of these two species has been shown by the number and distribution of rDNA loci ) and by microsatellite markers and sequences of single-copy genes (Moretzsohn et al. 2013) and their taxonomic status needs further investigation.
Arachis batizocoi, A. cruziana, and A. krapovickasii grouped together (Group 3), associated to the B genome species. The placement of the three K genome species, formerly classified as B genome species, in a consistent group supports the validity of the genome reassignment made by Robledo and Seijo (2010). Arachis glandulifera was also placed in Group 3. This D genome species has a unique karyotype within section Arachis (Stalker 1991;Fernández and Krapovickas 1994;Samoluk et al. 2019), but FISH mapping of rRNA loci and DAPI banding have shown homologies between A. glandulifera and A. batizocoi (Robledo and Seijo 2008). Additionally, A. glandulifera tends to group with the K genome species, when analyzed by molecular markers (Moretzsohn et al. 2004(Moretzsohn et al. , 2013Tallury et al. 2005;Bechara et al. 2010).
These four species are native to Bolivia, where A. glandulifera, the only one also found in Brazil, might have originated from one of the K genome species after fixing chromosomal rearrangements.
The two accessions of A. vallsii grouped together, in Group 5, closely related to the B and K genome species. Arachis vallsii was originally assigned to section Procumbentes (Krapovickas and Gregory 1994). However, its reclassification into section Arachis has been proposed, based on morphological and chromosomal features ), as well as in its annual life cycle, a feature absent in the Procumbentes. Analysis of interspecific and intersectional crossability showed that A. vallsii produces hybrids with representatives of different species of section Arachis, including A. hypogaea and A. monticola (unpublished data from our research group). These results strongly suggest that A. vallsii should be classified in the section Arachis, as proposed by Lavia and coworkers (2009) and corroborated by Moretzsohn and coworkers (2013).
The six accessions of A. hoehnei were joined in a consistent group (Group 6). Most of the genetic studies based on molecular markers have shown the close genetic relationship and clustering of A. hoehnei accessions, corroborating its taxonomic status Cunha et al. 2008;Bechara et al. 2010;Moretzsohn et al. 2013). Despite this, cytogenetic studies of different accessions of A. hoehnei found conflicting results. Accession K 30006, representative of the type collection, was described as not having the small 'A' chromosome pair based on classical cytology (Feulgen staining) (Fernández and Krapovickas 1994), chromosome painting of telomeric repeats, and by GISH/FISH (Du et al. 2016(Du et al. , 2019. The last study also showed the absence of the 'A' pair in the accession V 9094. In contrast, FISH analysis of the accession V 9146 showed the presence of a pair of small chromosomes similar in size to the "A" chromosomes, but with a different chromatin condensation from the other A genome species . The presence of the 'A' pair was also found in the accessions K 30006 and V 9094, in an unpublished study of Germán Robledo, cited by Robledo and Seijo (2010). The great majority of genetic studies also showed that A. hoehnei is more closely associated to the A genome species Gimenes et al. 2007;Cunha et al. 2008;Koppolu et al. 2010;Bechara et al. 2010;Moretzsohn et al. 2013;Vishwakarma et al. 2017), in accordance with our data. Crossings between A. gregoryi, which has a B genome, and the accession K 30006 of A. hoehnei were unsuccessful (Custodio et al., 2021). These results reinforce that A. hoehnei is distant from the B genome species. However, more cytogenetic studies of a diverse panel of accessions are still necessary to determine the genome constitution of A. hoehnei.
Most accessions of A genome species grouped together (Group 7). The A genome species have been previously arranged into three subgroups (La Plata River Basin, Chiquitano, and Pantanal) based on the variability of DAPI heterochromatic bands and 18S-26S rRNA loci (Robledo et al. 2009). A total of 14 out of the 17 accessions of the Pantanal subgroup included in the present study were accordingly located in this subgroup. The three accessions of A. helodes, which occurs naturally in a very restricted area in the surroundings of Cuiabá (Mato Grosso, Brazil) grouped together, closely related to A. diogoi Vp 5000 and one of the two Chiquitano species included (A. cardenasii GKP 10017). Arachis stenosperma, probably cultivated by indigenous peoples in ancient times, and quite possibly subject to human migration (Custodio et al. 2005), has a wide area of occurrence and a long disjunction between Central Brazil and the Atlantic coast. Despite this, all accessions of this species grouped together, within Group 7. Arachis kuhlmannii is found in a more restricted area, the West Central region of Brazil, but four of its accessions were located in Group 7, while three accessions formed a differentiated group with the unique accession of A. kempff-mercadoi (Chiquitano group) included in this study (Group 9). Krapovickas and Gregory (1994) mentioned some morphological differences between accessions of A. kuhlmannii. In addition, studies using different molecular markers have shown the high genetic variability of A. kuhlmannii Koppolu et al. 2010;Moretzsohn et al. 2013;Fávero et al. 2017).
Using RAPD markers and several of the same accessions tested here, Fávero et al. (2017) assigned the accessions of A. kuhlmannii, with minor exceptions, to three geographic groups: group 1-from the state of Mato Grosso, west of the Paraguay river, including V 6352, V 6380, and V 8887, these three accessions gathered here with the Bolivian A. kempffmercadoi in group 9; group 2-also from the state of Mato Grosso, but east of the Paraguay river, including V 13779, here included in group 7; and group 3-from Mato Grosso do Sul, again east of the river, which included V 9235 and V 9479, also gathered here in group 7. Thus, the A. kuhlmannii accessions included in group 9 come from natural sites west of the Paraguay River, while those in group 7, together with the easternmost A. stenosperma, occur east of the river. Accession V 14691, not available for Fávero et al. (2017) survey, fits the pattern, as it was collected in the urban area of Cáceres, very close to the east bank of the Paraguay River. In any case, additional taxonomic studies seem to be necessary for the material currently classified as A. kuhlmannii.
Group 7 also contained the only F genome species included in this study, A. benensis. This species was formerly considered to belong to the B genome group, but it was reassigned to the F genome, due to differences in amount and distribution of heterochromatin and the presence of centromeric bands (Robledo and Seijo 2010). Arachis benensis is a peculiar species, since it has the type 9 satellited chromosome that is typical of section Procumbentes (Fernández and Krapovickas 1994). The genetic relationships of the F genome species (A. benensis and A. trinitensis) with species from the other genome types of section Arachis are not clear, since different studies have found different results (Moretzsohn et al. 2004(Moretzsohn et al. , 2013Milla et al. 2005;Tallury et al. 2005;Friend et al. 2010). However, they are mostly associated with the species with K and D genomes instead of the B genome species (Moretzsohn et al. 2004(Moretzsohn et al. , 2013Friend et al. 2010). Crossings between A. gregoryi (B genome) and A. benensis (F genome) resulted in hybrids with only 1% pollen viability (Custodio et al., 2021). These results show that A. benensis is genetically distant from the B genome species and also support the validity of the genome reassignment made by Robledo and Seijo (2010). The clustering of A. benensis in a group containing the A genome species observed here is surprising and can probably be explained by some contamination of the sample used for genotyping.
The six accessions of the three species with 2n = 2x = 18 chromosomes (A. decora, A. palustris, and A. praecox) grouped together (Group 8). These Fig. 1 Dendrogram based on genetic distances estimated by the modified Rogers' coefficient of 93 accessions belonging to 31 species of the genus Arachis generated by UPGMA ◂ species usually form a well-differentiated group in most genetic relationship studies using molecular markers, what shows they are very closely related. Arachis decora, from Goiás and Tocantins States, and A. palustris, from Tocantins and Maranhão, were more closely related, while A. praecox, from Mato Grosso, was genetically differentiated from the other two species. The origin of these species is controversial, and sometimes they are associated with the A genome group (Tallury et al. 2005;Bravo et al. 2006;Koppolu et al. 2010) or, more often, with the B, D, F and K genome species (Moretzsohn et al. 2004(Moretzsohn et al. , 2013Gimenes et al. 2007;Bechara et al. 2010;Friend et al. 2010). Studies based on rDNA loci and position of heterochromatin suggested these G genome species were derived from an A genome species (Silvestri et al. 2015). In our study, they grouped outside a major group containing the A, B, D, F, and K genome species and thus the result was not informative about the origin of the G genome species.
Group 10 contained three species of section Procumbentes and two of Erectoides. Therefore, the microsatellite markers used here could not separate these species according to their sections, but did show a clear differentiation of A. kretschmeri, of section Procumbentes (Group 11). Genetic studies, based on ITS and the plastid trnT-trnF sequences (Bechara et al. 2010;Friend et al. 2010), and on microsatellite  and InDel (Vishwakarma et al. 2017) markers have shown that species of these two taxonomic sections (and also from section Trierectoides) tend to be placed in a same clade or similarity group. Additionally, species that currently belong to these three sections were formerly classified in the section Erectoides (Krapovickas 1969;Gregory et al. 1973). Their separation into three sections was based on cytogenetic (Fernández and Krapovickas 1994) and hybridization studies (Gregory and Gregory 1979;Krapovickas and Gregory 1994), which included only one (A. lignosa) of the four species of section Procumbentes included in this study. These results evidenced that more studies are needed to understanding the circumscriptions of sections Erectoides, Procumbentes, and Trierectoides.

Conclusions
In summary, the results presented here, based on 26 microsatellite loci, shed more light on the genetic relationships of Arachis species. In general, results were consistent with the current classification, but suggest the need of further taxonomic revision of a few taxa, especially some accessions of A. gregoryi, A. magna and A. kuhlmannii and the circumscriptions of sections Erectoides and Procumbentes. This revision is currently underway by our research group through the analysis of a higher number of accessions, species, and molecular markers associated to botanical descriptions and will certainly contribute to a better understanding of the taxonomy of these taxa.
The similarity groups usually contained species of the same genome types, corroborating the current genome assignments. Arachis ipaënsis was the only B genome species grouped together with all the A. hypogaea accessions confirming this species was the B genome donor to A. hypogaea.
Knowledge of the genetic relationships between accessions of A. ipaënsis, A. gregoryi, A. magna, A. valida and A. williamsii, some of which had never been analyzed by molecular markers, expands the number of available accessions for incorporation of useful genes from species associated with the peanut B genome.
The two A. vallsii accessions grouped together and were more closely related to the Arachis section species than to the Procumbentes, to which it was originally assigned (Krapovickas and Gregory 1994). These results gave additional support for the reclassification of A. vallsii into the section Arachis, as already proposed Moretzsohn et al. 2013).
It is also relevant to note that the SSR profiles were consistent with the proximity of accessions of A. batizocoi, A. cruziana, and some of A. gregoryi and A. magna, currently kept as distinct in gene banks, due to small morphological peculiarities perceived during their initial increase under management, or obtained through recurrent collections in the wild, basically from the same original populations, but after long intervals of time. Development (CNPq), Coordination for the Improvement of Higher Education Personnel (CAPES) and Embrapa Genetic Resources and Biotechnology for the scholarship and grants received through project numbers: 310707/2019-6, 310026/2018310026/ -0, 312215/2013310026/ -4, 313763/2013310026/ -5, 401939/2013310026/ -8, 483860/2012310026/ -3, 561768/2010310026/ -2 and 001/2011 Author Contributions ARC, MCM, and JFMV conceived and planned the study and wrote the manuscript; JFMV provided the germplasm accessions, which were multiplied by ARC for the experiment; ARC and ABS carried out the genotyping; MCM and ARC analyzed the data. All authors read and approved the final manuscript.
Data availability An additional dataset is available as a Supplementary Material. Any other data generated during the current study is available from the corresponding author on reasonable request.

Competing interests
The authors declare no conflict of interest.