Cruzabilidad entre Arachis gregoryi (Fabaceae) y especies silvestres de Arachis con distintos genomas

1 Embrapa Genetic Resources and Biotechnology, Parque Estação Biológica, PqEB, Av. W5 Norte (final). Caixa Postal 02372, Brasília, DF, Brazil, CEP 70770-917. E-mail: jose.valls@embrapa.br 2 Texas A&M AgriLife Research, Stephenville, Texas, USA. Custodio, A. R., C. E. Simpson & J. F. Montenegro Valls. 2021. Crossability between Arachis gregoryi (Fabaceae) and wild Arachis species with distinct genomes. Bonplandia 30(1): 91-106. http://dx.doi.org/10.30972/bon.3014669 Recibido 7 Febrero 2020. Aceptado 22 Abril 2020. Publicado en línea: 23 de diciembre 2020. Publicado impreso: 15 de febrero 2021. ISSN 0524-0476 impreso. ISSN 1853-8460 en línea.


Introduction
The genus Arachis encompasses 82 species described based on the integration of varied criteria, such as morphological, cytogenetic, and molecular characterization, geographic distribution, and interspecific crosscompatibility (Gregory et Gregory, 1979;Krapovickas et Gregory, 1994;Valls et Simpson, 2005Valls et al., 2013). It is typified by the common peanut, an allotetraploid cultigen with a genomic constitution AABB (Singh et Moss, 1984), that is well diploidized genetically (Bertioli et al., 2011).
Production of synthetic allotetraploids from sterile hybrids between diploid wild species related to the A and B peanut genomes often restores fertility, removing the reproductive barriers, and increasing the possibility of successful construction of pre-breeding lines derived from wild species and A. hypogaea (Simpson, 1991;Fávero et al., 2006;Foncéka et al., 2009). Therefore, knowledge on the crossing behavior of wild diploids most closely related to the peanut, those with the A and B genomes, will help to understand species relationships in the genus, and to produce amphidiploid lines of potentially successful use in peanut breeding programs.
In the framework of peanut breeding based on the incorporation of related wild species germplasm, research on crossability of wild Arachis species with A. hypogaea has been concentrated for decades on species that share the A genome with the crop, and with assorted species then thought to be associated to the B genome (Simpson et Starr, 2001;Singh et Moss, 1984). A review of the genomic characteristics of species in the taxonomic sect. Arachis (Robledo et Seijo, 2010) established a group of true B genome species, gathering A. ipaënsis and A. williamsii, exclusive to Bolivia, and each represented in genebanks by a single available accession, as well as three species occurring in Brazil, A. magna (also occurring in Bolivia), A. gregoryi, and A. valida. The restricted diversity of A. ipaënsis demands further exploration of the closely related species sharing the B genome, for further progress in peanut breeding.
This paper analyses crossing relationships of A. gregoryi with members of section Arachis, including representatives of the A, B, D, F, G, and K genomes, as well as the AB configuration, and with A. paraguariensis, of the taxonomic section Erectoides.

Material and Methods
Crosses were made during two consecutive spring/summer periods, under screenhouse conditions, from October 2006 to late May 2007, and from September 2007to April 2008 The female parent in all crosses was A. gregoryi, accession number VS 14957 (Table1).
It was collected in the wild, in 2004, at the same original site as VSGr 6389, this one collected in 1981, and is a member of the same natural population already represented in crossing experiments (Valls et Simpson, 2005;Fávero et al., 2015a, b).
Male parents represent the genomic and cytological variation so far detected in the taxonomic section Arachis and one species of section Erectoides (Table 1). Single plants of the female parent were cultivated in rectangular pots, each receiving pollen from a single and specific male donor.
Emasculation of the A. gregoryi 14957 flower buds was conducted each day, from 4 to 7 PM. The flower keel was carefully taken out with tweezers, for full anther exposure and removal, and the standard petal was folded again over the exposed stigma, for protection. Pollination of the emasculated flowers was conducted from 7 to 11 AM, the next morning. Pollen grains were transferred from the specific donor with tweezers, which were washed in alcohol between crosses. Emasculated flowers were misted with a fine water spray to facilitate pollen grain adherence to the stigma. Crosses were made during the flowering period of each pair of parents.
Pegs were visible approximately 15 days after pollination and were individually monitored until ripening, or to the end of the life cycle of the individual female parent. Spontaneous seedlings were carefully transferred to smaller, individual plant pots for further development. Harvested fruit segments were dried at room temperature and shelled seeds were later put to germinate under laboratory conditions.
Pollen viability of the parents was estimated both by staining and in vitro germination. Pollen of F 1 hybrids was analyzed by staining only. Eight flower buds were collected from each parent at weekly intervals during the crossing period. Pollen grains were set on a slide, stained with 2% aceto-carmine:glycerine, covered with a cover slip, and allowed to stain for 30 minutes in a moist chamber. Counting of well-formed and ill-formed grains was made under the microscope, including five samples of 100 grains each per slide. The analysis by in vitro germination followed the protocol of Niles et Quesenberry (1992). A basic solution of Medium 11 (10 mg H 3 BO 3 , 30 mg de Ca (NO 3 ) 4 H 2 O, 20 mg MgSO 4 + 7 H 2 O, 10 ml KNO 3 , make up to 100 ml of water), with subsequent addition of 1.5 g sucrose to each 10 ml of basic solution at the time of preparation of the slides. The slides with the pollen grains plus a drop of basic solution are stored in humid chamber for 2 h. The number of pollen tubes larger than the pollen grain was counted. Average counts per flower bud were calculated, and the compiled results were expressed in percentage. The percentage of successful pollinations was calculated by division of the number of hybrids obtained by the number of pollinations performed. These results were multiplied by 100 (Table 2).  (Table 2). Some crossing combinations were repeated in the second season, due to low flower production, lack of flower synchrony between parents, or temporary low pollen estimates of the male parent in the first season.

Pollen viability estimates
Several seedlings voluntarily produced in the pots of their female parent died at early stages, so that it was impossible to determine whether they were hybrids or just maternal offspring. According to the percent of successful pollinations, crossing results were classified into four heterogeneous groups (Table 2). Six crossing combinations, which failed or did not perform well in the first crossing season, for reasons beyond our control, were repeated in the second season, this time showing positive results. Such combinations conducted in both seasons are reported individually within the framework of the four groups, and the results of the second attempt obviously better reflect their crossing potential.
The first group encompasses crossing combinations that produced the highest percent of hybrids per pollination events (17.1 to 11.9%), but most of the hybrids produced did not flower. Group 1 involves vallsii (both 11.9%). Five of the male parents are diploid species associated to the A peanut genome, one to the B genome. The genomes of A. hoehnei 13985 and of A. vallsii are yet to be determined (Table 2).
Concerning male parents with the A genome, the cross of A. gregoryi with A. kuhlmannii produced 24 fruit segments, which generated 12 hybrid plants with pollen viability estimates of 4.8%, in average. Hybrids of A. gregoryi × A. kempff-mercadoi produced quite normal plants, with a typical main axis and ascending lateral branches, but no flowers. Hybrids with A. microsperma also were quite normal plants, but did not flower. Showing excellent survival, hybrids with A. schininii produced highly proliferated lateral branches, but an abnormally short main axis, and no flowers. Hybrids of A. gregoryi × A. stenosperma showed slow growth and early loss of the main axis, although the lateral branches remained alive for several months.
Hybrids with the B genome species A. williamsii, that stood out as a good result obtained in the first season, showed very intensive vegetative growth, with lateral branches extending longer than 2.5 meters, intensive flowering, and average pollen viability estimate of 2.6%, but no peg formation.
The hybrid of A. gregoryi × A. hoehnei 13985 showed a fast growing rate, erect habit, and intensive lateral branching. However, the plants did not flower, so that pollen viability estimates are not available.
Crosses of A. gregoryi × A. vallsii produced 22 fruit segments, which resulted in ten seedlings. Although their hybrid nature could not be confirmed, due to very slow growth and no flowers, the leaflets of the hybrids showed the paternal parent trait, narrow and lanceolate, not of maternal origin.
The second group of crossing combinations gathers four crosses with a percent of success from 9.6 to 5.4%, involving A. gregoryi × A. valida (9.6%), A. gregoryi × A. magna 30097 (9.5%), A. gregoryi × A. villosa (8.2%) and A. gregoryi × A. diogoi (5.4%), the first two male parents associated to the peanut B genome, and the last two to the A genome.
In hybrids of A. gregoryi × A. valida, lateral branches were well developed and over 2.5 m long. Flowering was intensive, with an average pollen viability estimate of 5.45%. In spite of this low figure, one peg was produced, showing a fruit segment that, although well formed, did not complete its development and deteriorated in the soil.
The hybrid of A. gregoryi × A. magna 30097 presented average pollen viability estimate of 7.12%. With a rate of success of 8.2%, crosses of A. gregoryi × A. villosa produced four plants, three of which survived well, and were additionally multiplied by cuttings, but never flowered. The hybrid status of the seedlings was easily confirmed by the dense pubescence on the adaxial leaflet surface, a paternal feature.
Hybridization of A. gregoryi × A. diogoi produced three small, but vigorous plants, which did not flower.
The third group of hybrids, with a low rate of success between 3.6 and 0.6%, encompasses 15 crossing combinations, Male parents in this group include representatives of three botanical varieties of A. hypogaea, as well as species either associated to the A or B genomes, besides representatives of the D, F and K genomes, additionally described for species of the Arachis section (Stalker, 1991;Robledo et Seijo, 2008, 2010.
In the intraspecific cross A. gregoryi 14957 × 14753, 24 fruit segments were formed, but only two hybrid plants survived. Pollen viability estimates varied from 36-55.6%. Although the anthers did not open properly at anthesis, and, consequently, release of pollen-grain was restricted, the two hybrid plants produced many pegs with average thickness of 0.68mm, about half the peg thickness measured on the female parent. This F 1 plant behaved as typically annual, dying out at the end of the reproductive period. Seeds germinated voluntarily in the pot, generating ten F 2 plants.
Crosses involving A. gregoryi with A. ipaënsis produced 39 fruit segments, that had no spontaneous germination, with the hybrid pollen viability estimate of 5.4%.
Seedlings resulting from two crossing combinations, involving A. gregoryi × A. batizocoi, showed abnormal growth from the first stages. Fruit segments produced in the respective female plant pots were harvested, some already starting germination. Seedlings and seeds were grown in vitro. The 82 pollinations made in the cross of A. gregoryi × A. batizocoi resulted in the production of 61 fruit segments, which would imply a very high rate of success, if they developed normal seeds and then plants. However, 46 seeds germinated spontaneously producing abnormal seedlings, lacking the aerial organs above the cotyledons. So, the seeds from the 15 remaining fruit segments were set to germinate in vitro. Only a single hybrid plant developed in a greenhouse, but it did not survive, when later transferred to screenhouse conditions. Embryos rescued in vitro still persist but need to be acclimatized.
The crossing of A. gregoryi × A. cruziana resulted in 36 fruit segments. Seedlings were vigorous, forming adult plants with lateral branches longer than 2.5 m, and intensive flowering. The average pollen viability estimate was 5.22%.
Concerning A. gregoryi × A. krapovickasii, 19 fruit segments were formed. The seeds germinated spontaneously, generating vigorous seedlings, but did not continue the development after showing the first leaves. Only one plant survived and flowered, allowing the pollen viability estimate of 0.4%.
The crossing combination A. gregoryi × A. glandulifera, representing the B and D genomes, produced 27 well-formed fruit segments, but 25 had aborted embryos, and one seed did not reach full maturity. The single well-formed seed produced a normal, flowering plant, with an estimated pollen viability of 0.3%. The hybrid status was easily confirmed by the presence of glands in the abaxial leaflet surface, a typical paternal feature.
In a cross involving A. gregoryi × A. benensis, which represents the F genome (Robledo et Seijo, 2010), 25 fruit segments were formed, with spontaneous germination of five seeds. Pollen viability estimates were quite low, 0.5% in average.
In a second crossing combination that required in vitro support, involving A. gregoryi × A. hoehnei 30006, 59 seeds were produced, of which 43 germinated spontaneously. Although under intensive attack of mites, just like the male parent, seedlings started in good shape, but soon died. The remaining 16 seeds were set to germinate in vitro. Only a single plant survived in the greenhouse, and later, under screenhouse conditions. It had profuse lateral branching, with ascending-erect habit, and it  flowered. The pollen viability estimate was 12%, with extremes of 9% and 18.2%. The fourth group involves crosses in which the resulting seeds did not become established as plants, so that the hybrid character could not be confirmed ( Factors like low flowering synchrony and low quantity of flowers produced by the male parents, and, most of the time, low vigor of flowers after manipulation for hybridization were significant in the lack of success in the production of hybrid plants.

Pollen viability estimates
No method for estimation of pollen viability is fully satisfactory, as the stains used in chemical tests react with chemical constituents or structures of the grains, and may not reflect the real germinating capacity (Stanley et Linskens, 1974). In spite of this, estimates of pollen viability are important to ensure that the plant materials involved present, at least in theory, the ability to trigger the complex reproductive process that involves morphological, physiological and genetic aspects.
Comparing the pollen staining and germination estimates, the difference of 10 to 20% is considered a common fact, most possibly due to overestimation with the staining technique and underestimation with germination, as concerns the viable pollen grains (Galetta, 1983). Even though some authors do not consider pollen staining as a viable estimate of plant fertility, the present authors are not as skeptical as to the positive aspects of this technique. Pollen staining has been reported in the evaluation of Arachis crossing programs by several authors (Gregory et Gregory, 1979;Simpson, 1991;Krapovickas et Gregory, 1994;Wondracek-Lüdke et al., 2015;Fávero et al., 2015a, b) and is generally accepted by Arachis researchers as a useful estimate of fertility.
Although a few accessions have shown lower pollen estimates for some time along the crossing season, all others consistently presented high estimates, so that all male parents were considered fit to fertilize the flowers of the female parent. It is not uncommon for a species such as A. helodes to have a low pollen count when some strenuous environmental conditions occur. If daytime temperature exceeds 42ºC, fertile plants of many Arachis species may have a pollen stain of < 50% for two or three consecutive days (Simpson CE, unpublished results).

Hybridization
The female parent A. gregoryi 14957, from the same site of 6389, has shown a broad crossing plasticity, as it was successfully pollinated and developed fruit segments in most crossing combinations, involving distinct genomes, ploidy levels, and taxonomic sections. Although these two accessions were assigned distinct collection numbers and germoplasm codes (BRA), due to the long time elapsed between field collections, previous data and the results of the present research confirm the potential value of this species and of the wild population represented by both the 6389 and 14957 accessions for hybridization studies.
The possibility of polyploidization of the hybrid A. gregoryi 6389 × A. linearifolia, producing an amphidiploid, which was then successfully crossed with A. hypogaea, has been demonstrated (Fávero et al., 2015a). So, it is estimated that additional accessions of A. gregoryi, with a B genome somewhat similar to that of A. ipaënsis, may be used as efficient bridge species for the introgression of genes from wild Arachis species into the cultivated peanut.

Group 1
Crossing combinations in this group were the most successful, although they showed differentiated potential. The five hybrid combinations involving B × A genome holders (A. gregoryi × A. kempff-mercadoi, A. kuhlmannii, A. microsperma, A. schininii, or A. stenosperma, the last two with worse results in the first season), all showing low pollen counts, are good candidates for production of colchicine induced highly fertile amphidiploids. Diploid hybrids of A. gregoryi with species showing the A chromosome pair are of special interest to peanut breeding, once their chromosome doubling will produce lines with the AABB genome constitution, for direct crossing with A. hypogaea. Furthermore, several of the A genome species of which hybrids where obtained show variable degrees of resistance/tolerance to foliar diseases caused by Cercospora arachidicola Hori (early leaf spot), Cercosporidium personatum (Berk. & M.A. Curtis) Deighton (late leaf spot), and Puccinia arachidis Speg. (rust) (Fávero et al., , 2015a. It must be noted that hybrids of A. gregoryi × A. stenosperma have already been reported, but they involve different accessions of each parental species (Fávero, 2015b;Valls et Simpson, 2005;Leal-Bertioli et al., 2017). These previous hybrids flowered in both experiments cited, showing low pollen viability estimates, from 0.4 to 4.9%. Besides peculiarities of the distinct accessions involved, environmental differences at the distinct experimental facilities may not be ruled out as responsible for the lack of flowering in the present study.
The easy B × B crossing of A. gregoryi × A. williamsii, associated to the very low pollen viability estimates of the hybrid produced, suggests a potential for successful restoration of fertility through chromosome doubling. Such tetraploid line, gathering characteristics of two B genome species, may be useful for gene pyramidation in the B side.
The easiest crossing, that of A. gregoryi × A. hoehnei 13985, the most prolific male parent in this crossing program, may be of high interest for peanut breeding, but further study of the A. hoehnei genome relationships is still on demand.
Seven accessions of A. hoehnei are available in the Embrapa Wild Arachis Genebank, all from the state of Mato Grosso do Sul, Brazil. But they are not uniform, showing differences in their reproductive cycle and crossing behavior, and doubts still persist about the presence or absence of the A chromosome pair in distinct accessions of the species (Fernández et Krapovickas, 1994;Custodio et al., 2013).
Easy crossing of A. gregoryi × A. vallsii, attained in the second season, but by no means in the first, is not only of genetic, but also of taxonomic importance. Arachis vallsii was originally described as a member of the taxonomic section Procumbentes (Krapovickas et Gregory, 1994). Cytogenetic evidence has confirmed its closer relationship to the Arachis section, although it does not show the A chromosome pair (Lavia et al., 2009) nor a typical B genome configuration (Robledo et Seijo, 2010). Confirmation of the crossability of A. gregoryi × A. vallsii suggests a possibility of incorporating some peculiar features of the latter into peanut breeding programs, especially those related to the ecological preferences of A. vallsii, which thrives in periodically flooded areas, producing its underground fruits in muddy soils (Simpson et al., 2018).

Group 2
As well as A. gregoryi, associated to the B genome (Robledo et Seijo, 2010), A. valida has a potential value for peanut breeding, as it shows resistance to leaf spots (Fávero, 2009), and possibly also to flooding, considering the area of occurrence of its natural populations (Krapovickas et Gregory, 1994). In a parallel study (Wondracek-Lüdke et al., 2015), A. valida has shown a similar potential to that of A. gregoryi for producing hybrids with all representatives of the B genome.
Concerning the hybrid of A. gregoryi × A. magna 30097, similar low values for pollen viability estimates were obtained in Texas (Simpson et Faries, 2001), involving the same male parent and the 6389 accession of A. gregoryi, already mentioned as a member of the same natural population of our female parent. This means that, although A. gregoryi and A. magna seem to be closely related species, even showing individual plants somewhat difficult to discriminate on morphological grounds, and have several sympatric populations, they must have developed genetic barriers, which maintain their reproductive isolation and allow for their persistence as distinct species.
As to the A. gregoryi × A. villosa hybridization, the latter, collected in Brazil on the edge of the Uruguay river, is cytogenetically, as well as geographically considered a member of the La Plata River Basin subgroup of A genome species, therefore narrowly related to A. duranensis (Robledo et al., 2009). Due to this close cytological affinity to the A genome donor of A. hypogaea (Seijo et al., 2004), this AB diploid hybrid is of utmost interest for chromosome duplication and incorporation to peanut breeding programs.
Another A. gregoryi × A. villosa hybrid, obtained by using, respectively, the A. gregoryi accession 6389 and the Uruguayan accession of A. villosa VGoMrOv 12812, not tested here, had a much lower percent of success (0.92%), but flowered, with estimated pollen viability of 7.67% (Fávero, 2015b).
The A. gregoryi × A. diogoi diploid hybrid is another interesting line for chromosome duplication, once the paternal species, then represented by the distinct Paraguayan accession GKP 10602, has been used to produce the amphidiploid line that resulted in the release of the cultivar 'COAN' (Simpson et Starr, 2001), which has near immunity to the root-knot nematode species Meloidogyne arenaria (Neal) Chitwood and M. javanica (Treub) Chitwood. Natural populations of A. diogoi occur along the Paraguay river basin and its tributaries, in areas subject to periodic flooding (Krapovickas et Gregory, 1994), and have shown resistance to early and late leaf spots, thrips, and jassids (Fávero, 2009).

Group 3
In spite of the distinct ploidy levels involved, crosses of A. gregoryi × A. hypogaea did not require special techniques. However, embryo abortion at early stages of seed formation was frequent. Mature seeds were produced in crosses with three varieties (aequatoriana, hirsuta, and fastigiata). None has shown dormancy, and they generally germinated, but most seedlings died at early stages. All four surviving plants of A. gregoryi × A. hypogaea subsp. fastigiata var. aequatoriana were triploid (2n=3x=30) and highly sterile, while the single surviving plant of A. gregoryi × A. hypogaea subsp. hypogaea var. hirsuta was too weak, and roots were not adequate for chromosome counting, but its morphology, quite different from the female parent, led to the assumption of the hybrid state. Also, a single slow growing, confirmed cytologically as a triploid plant was rescued from the combination A. gregoryi × A. hypogaea subsp. fastigiata var. fastigiata, which, although healthy, never flowered.
Manipulation of triploid interspecific hybrids of Arachis to produce useful breeding lines is laborious. It is first necessary to duplicate chromosomes to produce a fertile hexaploid line, which is then backcrossed to A. hypogaea, until it stabilizes at 2n=40, due to chromosome loss (Simpson, 1991(Simpson, , 2001, but with no guarantee of a balanced AABB genome structure. In spite of the more complicated pathway, triploid hybrids produced in this program have a potential use in peanut breeding. Different behavior presented by crossing combinations of A. gregoryi with these three botanical varieties of A. hypogaea would recommend additional trials, involving the remaining varieties, as well as accessions of the "Xingu" type, morphologically divergent, so far unclassified at the subspecific level, and more closely related to A. monticola (Nascimento et al., 2020), of which no hybrids were obtained in this study.
Considering B × B hybrids, it is noteworthy that only one intraspecific combination (A. gregoryi 14957 × 14753) has been successful, out of three attempted. The accessions tested, 14957 and 14753, are from adjacent counties in Mato Grosso State, but were collected in nature 99 km apart. Although pollen viability estimates of this hybrid reached 43.6%, the highest figure for all hybrids obtained in the present study, it is far below estimates of over 90% mentioned for other intraspecific and even some interspecific crosses involving other Arachis species (Krapovickas et Gregory 1994, Simpson et Faries 2001Fávero et al., 2015 a, b;Wondracek-Lüdke et al., 2015). So, crossing data obtained in this study for the specific accession used as the female parent should be taken carefully as concerns other A. gregoryi materials, except in comparisons with the 6389 accession, recognized as a member of the same natural population.
Contrary to the general trend, seeds resulting from the A. gregoryi × A. ipaënsis crossing did not present spontaneous germination. Lack of spontaneous germination is a useful trait, allowing for timely use of the hybrid seed in breeding programs. Due to the relevance of A. ipaënsis in the origin of A. hypogaea (Kochert et al., 1996;Seijo et al., 2004Seijo et al., , 2007Fávero et al., 2006;Bertioli et al., 2016), and the fact that there is only one accession of the former species available in genebanks, hybrids with A. ipaënsis are important for potentially expanding the possibilities of introgression of favorable characteristics associated to the B genome, for the improvement of peanut cultivars.
Similar to the hybrids with A. ipaënsis, and contrary to the behavior of the A. gregoryi × A. magna 30097, seeds of A. gregoryi × A. magna 14724 did not show spontaneous germination. This emphasizes the diversity of the available A. magna accessions, already put in evidence by molecular cytogenetics  and molecular marker analyses (Moretzsohn et al., 2013).
Although at lower rates of success, once again B × A genome crosses produced interesting candidates for induced chromosome doubling for direct crossing with A. hypogaea at the tetraploid level, including the one involving A. duranensis, the A genome donor of A. hypogaea (Kochert et al., 1996;Seijo et al., 2004Seijo et al., , 2007Fávero et al., 2006;Bertioli et al., 2016), besides A. helodes, A. schininii, and A. simpsonii. Crosses of the B genome A. gregoryi with representatives of the D, F and K genomes, though important to the comprehension of genetic and evolutionary aspects of the genus, are not likely to produce amphidiploids as close to A. hypogaea, but that does not exclude their interest for peanut breeding, once such species eventually present favorable attributes, related to biotic and/or abiotic constraints. For instance, the K genome A. batizocoi has been used in the construction of TxAG-6, the breeding line that gave rise to the cultivar 'COAN' (Simpson et Starr, 2001). While, in the present study, hybrid seedlings of A. gregoryi × A. batizocoi produced mostly abnormal seedlings, lacking the aerial organs above the cotyledons, a reverse hybrid involving the same accession of A. batizocoi and A. gregoryi 6389 has shown normal development and flowering, with a pollen viability estimate of 4.5% (Simpson et Faries, 2001).
The low pollen viability estimate of the A. gregoryi × A. glandulifera hybrid possibly reflects a wide genetic distance between the parents involved. Intraspecific hybrids involving distinct accessions of A. glandulifera, all from Bolivia, and interspecific hybrids of A. glandulifera × A. batizocoi and A. duranensis, have been obtained elsewhere (Stalker, 1991). However, despite 835 pollinations in both ways, involving A. glandulifera and two cultivars of A. hypogaea representing subspecies hypogaea and fastigiata, not a single triploid hybrid was obtained (Stalker, 1991). Successful crosses of A. gregoryi × A. glandulifera may open a pathway to bringing desirable genes of the highly prolific A. glandulifera into pre-breeding lines similar to the peanut B genome, eventually conditioned to chromosome duplication.
It is interesting to notice the distinct results shown by the crossing combination A. gregoryi × A. hoehnei. The most successful crossing in this study has been obtained using A. hoehnei 13985 as the male parent, while that with A. hoehnei 30006 lies in the third group. The founder seed of the male parent A. hoehnei 30006 has been collected in the field at the same natural site of the type collection (Krapovickas et Gregory, 1994), so this accession is the most typical of A. hoehnei, while other accessions listed at the moment under the same name, such as 13985, may belong to a distinct taxon, yet to be described .

Group 4
Although this group encompasses unsuccessful crosses or crosses in which the hybrid nature of seed produced could not be confirmed, this cannot be taken as a definitive indication of incompatibility. For example, the unsuccesfull hybrization of A. gregoryi with A. diogoi, A. duranensis, and A. stenosperma (three A genome species) and A. vallsii (undetermined genome) in the first season was superseded by successful crosses in the second season. Such differences, which we cannot easily account for, may be under the influence of several factors, such as environmental factors, vigor of plants, attack of pests, and even the manipulation of the flowers of each parent during emasculation. But it must be considered that they did not affect crosses in the first season so intensely, when involving A. williamsii, A. valida and A. magna 30097 (B genome species), and A. villosa (A genome).
The genetic diversity of accessions of a same species that show different crossability results needs to be evaluated. This refers to the distinct varieties and off-types of A. hypogaea, as well as to species such as A. gregoryi, A. magna, and A. hoehnei. In summary, it has been shown that A. gregoryi is a promising diploid wild species to be incorporated into peanut breeding programs. It has shown plasticity for crosses with species of several genome types present in the taxonomic section Arachis. Its inclusion in breeding programs will expand the diversity of species and breeding lines associated to the B genome, with special attention to the actual availability of a single accession of A. ipaënsis, involved in the origin of A. hypogaea.
Similarly positive crossing results with distinct botanical varieties of A. hypogaea indicate that the use of A. gregoryi in pre-breeding efforts is a possibility, also through the more laborious triploid/hexaploid/backcrossing strategy.
Finally, crossability of A. gregoryi with several representatives of the A genome group, and the general sterility of the AB hybrids, suggests a good potential use of A. gregoryi in pre-breeding efforts through the amphidiploid strategy, which has already produced relevant results in the genetic breeding of the cultivated peanut.