Successful Wide Hybridization and Introgression Breeding in a Diverse Set of Common Peppers (Capsicum annuum) Using Different Cultivated Ají (C. baccatum) Accessions as Donor Parents

Capsicum baccatum, commonly known as ají, has been reported as a source of variation for many different traits to improve common pepper (C. annuum), one of the most important vegetables in the world. However, strong interspecific hybridization barriers exist between them. A comparative study of two wide hybridization approaches for introgressing C. baccatum genes into C. annuum was performed: i) genetic bridge (GB) using C. chinense and C. frutescens as bridge species; and, ii) direct cross between C. annuum and C. baccatum combined with in vitro embryo rescue (ER). A diverse and representative collection of 18 accessions from four cultivated species of Capsicum was used, including C. annuum (12), C. baccatum (3), C. chinense (2), and C. frutescens (1). More than 5000 crosses were made and over 1000 embryos were rescued in the present study. C. chinense performed as a good bridge species between C. annuum and C. baccatum, with the best results being obtained with the cross combination [C. baccatum (♀) × C. chinense (♂)] (♀) × C. annuum (♂), while C. frutescens gave poor results as bridge species due to strong prezygotic and postzygotic barriers. Virus-like-syndrome or dwarfism was observed in F1 hybrids when both C. chinense and C. frutescens were used as female parents. Regarding the ER strategy, the best response was found in C. annuum (♀) × C. baccatum (♂) crosses. First backcrosses to C. annuum (BC1s) were obtained according to the crossing scheme [C. annuum (♀) × C. baccatum (♂)] (♀) × C. annuum (♂) using ER. Advantages and disadvantages of each strategy are discussed in relation to their application to breeding programmes. These results provide breeders with useful practical information for the regular utilization of the C. baccatum gene pool in C. annuum breeding.

Introduction approach is technically more complex as it requires embryo excision and in vitro culture. Also, the stage at which embryo abortion occurs after hybridization may depend on the specific genotypes involved in the cross. Thus, for example within Solanaceae, while some authors could rescue interspecific embryos at the latest immature stages [29], there are also examples on which embryos had to be rescued at the earliest stages [30,31]. However, the earlier the stage at which embryo rescue is done, the more difficult is the procedure and the lower the efficiency [32].
Furthermore, even though hybrid materials could be achieved, hybrid sterility must be also considered as an important postzygotic barrier. Full sterility or different degrees of fertility of the interspecific hybrids may vary depending on the parent genotypes. In extreme cases, when sterility is complete due to the lack of chromosome pairing during meiosis, fertility may be restored by poliploidization, enabling pairing of homologous chromosomes in the allopolyploid hybrid [33].
Unfortunately, studies on wide hybridization between C. baccatum and C. annuum and the overcoming of their compatibility barriers are very scarce [29,34], especially regarding the range of diversity encompassed in materials used for crosses. Consequently, there is a lack of detailed practical information about the breeding process and levels of sexual compatibility among species, which might also depend on the genotypes involved or the direction of the crosses, among other factors.
Therefore, the development of approaches which allow overcoming these barriers, assessing all the difficulties which can appear during their application, will provide breeders with useful tools, practical information and a comprehensive perspective for the introgression of genes of interest from C. baccatum to C. annuum. Moreover, the use of a wide genetic diversity will contribute to offer a more complete view of these barriers.
The aim of this work was to compare comprehensively two approaches for the achievement of wide hybridization between C. annuum and C. baccatum: i) genetic bridge using C. chinense and C. frutescens as bridge species; and, ii) direct hybridization between C. annuum and C. baccatum in combination with embryo rescue. In both strategies full diallel interspecific crosses were planned and a range of genetically different genotypes were used. The effects of the direction of the crosses, cross compatibility at different levels among the species involved and hybrid viability and fertility are also discussed.

Plant material and growing conditions
A total of 18 accessions from four cultivated species of Capsicum were utilized in the present study: C. annuum (12 accessions), C. baccatum (3 accessions), and the bridge species C. chinense (2 accessions) and C. frutescens (1 accession). This collection encompassed a comprehensive range of geographical origins and fruit morphological traits (Table 1). New materials (i.e. interspecific hybrids and backcrosses) were included progressively in the experiments as they were obtained during the wide hybridization programme. Plants from parent accessions, and subsequent hybrids and backcrosses were transplanted at the four-leaf stage (about 50-60 days after sowing) to glasshouses at the Universitat Politècnica de València (UPV, Valencia, Spain) along three years (2010, 2011 and 2012). Plants were grown in 10 L pots (coconut coir as substrate), under the spring-summer growing cycle as it provides the most favourable conditions for the development and fruit set of Capsicum in the Mediterranean region [8]. Natural illumination was used for this experiment and temperature control systems (heating and cooling) were activated when temperatures dropped below 18°C (night) or rose above 25°C (day). Plants were pruned to four stems and trained with vertical strings. They were drip irrigated every 8 h for 3 min (4 L/h). Fertilizer was applied with the irrigation water, at a rate of 1 g/L of a commercial 15N-2.2P-24.9K water soluble fertilizer (BASF, Barcelona, Spain).

Hybridization technique
Previously to hybridization, female flowers were emasculated and pollen was extracted from male flowers and released carefully on the stigma. To prevent uncontrolled pollination, female flowers were covered with Scotch tape after hybridization. Each cross was tagged with the genotypes involved in the hybridization and the date at which it was performed. Overall, more than 5000 hybridizations were performed in this study.

Genetic bridge approach
Within the genetic bridge strategy (GB), all possible combinations of full (reciprocal) diallel crosses between the 12 C. annuum and the three C. baccatum accessions on one side and the three accessions of the bridge species C. chinense (2 accessions) and C. frutescens (1 accession) on the other were tested (i.e. a total of 45 hybrid combinations and their reciprocals). F 1 hybrid materials obtained were further used for completing the bridge cross (i.e., crossing to C. annuum if C. baccatum had been used to obtain the F 1 , and viceversa). Between 10 and 30 hybridizations were done for each cross combination and direction of the cross. Two strategies were performed to achieve the bridge cross (Fig 1): i) alternative 1, which firstly consisted in crossing C. annuum with the bridge species and, later the obtained hybrids were crossed with C. baccatum; and, ii) alternative 2, which firstly consisted in crossing C. baccatum with the bridge species and, later the obtained hybrids with C. annuum. As a result, a total of over 2800 hybridizations were done in this GB approach. Fully ripe fruits from each successful cross were harvested and their seeds were extracted and dried. After that, seeds were sown in seedling trays (7×12 cells, 480×300×55 mm) containing cultivation substrate (Humin-substrat N3, Klasmann-Deilmann, Germany). Once plantlets reached the four-leaf stage, they were transferred to glasshouses to perform the second cross (hybrid × parental cross) in order to obtain three-way hybrids (Fig 1).

Embryo rescue
Within the embryo rescue strategy (ER), a minimum of 10 hybridizations (and up to 47 in some cross combinations) between the 12 C. annuum accessions on one side and the three C. baccatum accessions on the other were performed using a full diallel scheme (i.e., 36 hybrid combinations) with reciprocals. Also several BC 1 crosses towards C. annuum were later tried on the basis of available hybrids. As a result, a total of 2500 hybridizations were done in the present ER experiment. About 25-30 days after pollination immature fruits were harvested to perform the embryo rescue following the protocol and culture medium optimized by us in previous studies [35,36]. The medium included: agar (7 g/L), indole-3-acetic acid (IAA, 0.01 mg/ L), gibberellic acid (GA 3 , 0.01 mg/L), zeatin (0.01 mg/L), sucrose (40 g/L), and MS (2.2 g/L), at pH 5.7. The basal medium (agar, sugars and MS salts) was sterilized at 121°C for 20 min. Hormones were sterilized by microfiltration (0.20 μm Minisart1 filters) and were added to the warm (35-40°C) autoclaved medium before solidifying. Embryos were excised carefully and immediately cultured in 90×15 mm Petri dishes, which were sealed with Parafilm1 and incubated in a growth chamber (25±10°C; 70% HR; full darkness the first 5 days and a 16h/8h (light/dark) photoperiod since day 6). About 1000 immature seeds were dissected for the ER approach. After 30 days of incubation, seedlings which showed a clear development of root and shoot were transferred to seedling trays and were covered with perforated plastic glasses to prevent dehydration. After one week, plastic glasses were removed. Once F 1 plantlets reached the four-leaf stage, they were transferred to 10 L pots at glasshouses for evaluation and to obtain subsequent BC 1 generations (Fig 1).

Cross compatibility evaluation and confirmation of hybridity
In both strategies the compatibility of each cross was evaluated at four levels: i) number of fruits set per number of hybridizations performed, ii) percentage of germinated seeds (or rescued embryos in the case of ER) 1 month after sowing, iii) plant hybrid appearance, and iv) pollen viability (n = 5 mesurements), estimated according to Rodríguez-Riaño and Dafni [37] with 1% (w/v) of thiazolyl blue tetrazolium bromide (MTT) dye (M-2128, Sigma-Aldrich). To confirm the hybrid nature of new materials, we used a set of six polymorphic SSRs ( Table 2). These SSRs markers were scored according to the method described by Minamiyama et al. [38].

Results and Discussion
Both approaches (GB and ER) enabled to successfully achieve the wide hybridization between C. annuum and C. baccatum, including obtaining F 1 hybrids, three-way-hybrids, and even backcross generations within the ER strategy. On the basis of the SSRs analysis, all new materials were confirmed as hybrids of their corresponding parents and, therefore, no uncontrolled pollination was present in the experiment. This is an important step for C. annuum breeding using C. baccatum as donor species for traits present in the latter. In the case of the ER approach first backcross (BC 1 ) generations towards C. annuum were obtained; and, in the case of the GB approach three-way hybrids using C. chinense as bridge species from both cross alternatives (i.e. starting with crosses between C. annuum and C. chinense and between C. baccatum and C. chinense) were achieved. However, a range of results, efficiency rates and crossability barriers was found in both strategies depending on the direction and the genotypes involved in the crosses. Consequently, all these factors must be considered to plan breeding programs which include interspecific crosses between these four species.
Genetic bridge approach C. chinense as genetic bridge species. Despite a high crossability between the species of the annuum complex has been widely assumed [6,8,9], many crosses of C. annuum accessions with both C. chinense or C. frutescens were unsuccessful (Fig 2). Thus, in the C. annuum (♀) × C. chinense (♂) cross scheme, fruit set occurred in 10 of the 24 possible combinations and only three C. annuum accessions (California Wr. red, Bola and Guindilla) were able to set fruit with any of the C. chinense accessions used (Fig 2). Furthermore, only the seeds of four combinations germinated and developed in normal hybrids, which also showed pollen viability estimates comprised between 17-31% (Table 3), considerably lower than those from their parent accessions, always higher than 80% (Table 1). Capsicum annuum accession Bola was the only one which produced hybrid plants with both C. chinense accessions (Table 3). Moreover, one of these (Bola × P15) was the hybrid with the highest pollen viability (31%). Therefore, this accession appears as suitable germplasm for the GB strategy, as well as for the transfer of genes from C. chinense to commercial peppers. Although Zijlstra et al. [11] discarded prezygotic barriers between species belonging to the annuum complex, the low or even nil fruit set rates obtained in our experiment suggest strong prezygotic incompatibility between C. annuum and C. chinense. Even more, as observed in some cases by Inai et al. [39], our low germination rates could be due to embryo abortion, which also suggest postzygotic barriers.
The reciprocal cross C. chinense (♀) × C. annuum (♂) enabled a higher number of successful combinations in terms of fruit set, with 16 out of the 24 possible combinations (Fig 2). However, only hybrid seeds from four combinations germinated (Table 3) and, moreover, hybrids showed stunted growth, filiform leaves, short internodes and did not enter into the reproductive phase (Fig 3). These symptoms were not observed in other plants grown in the same trial and all samples were negative to ELISA tests for Tobamoviruses or tomato spotted wilt virus (TSWV). Thus, the most likely reason for this abnormal development was the virus-like syndrome (VLS), which might be due to the interaction between cytoplasm and nuclear genes of C. chinense and C. annuum respectively [6,39]. Therefore, the results suggest that C. chinense should always be utilised as male parent to achieve viable hybrids with C. annuum.
The four viable C. annuum (♀) × C. chinense (♂) hybrids available were then utilized to complete the bridge cross with C. baccatum. Crosses involving C. baccatum Aji amarillo did not set any fruit (Fig 2), suggesting a low cross compatibility of this accession. Then, regarding the rest of combinations, the best results were obtained using hybrids as female parents, with  Table 3). All these materials showed a normal appearance, Table 3. Descriptive results of the genetic bridge (GB) strategy using the alternative of obtaining hybrids between C. annuum (Ca) and C. chinense (Cch) for being subsequently crossed with C. baccatum to obtain three-way hybrids. Cross combinations that did not set fruit or set fruit but seeds did not germinate are not included in the  although a wide range of pollen viability was observed, which was comprised between 4% and 83% in (Pas × PI15) (♀) × BrP (♀) and (Bola × AjíP) (♀) × AjíR (♂) respectively (Table 3). On the contrary, when C. annuum (♀) × C. chinense (♂) hybrids were utilized as pollen donors, fruit set and successful hybrid plants were only achieved in AjíR (♀) × (Bola × AjíP) (♂) (Fig 2). Surprisingly, this hybrid showed the highest pollen viability (95%) among all hybrid materials of the GB approach involving C. chinense as bridge species, which suggests that the involved genotypes Bola (C. annuum), AjíP (C. chinense), and AjíR (C. baccatum) have a high compatibility. Consequently, given the number of successful combinations (6/8), the following scheme [C. annuum (♀) × C. chinense (♂)] (♀) × C. baccatum (♂) is recommended when using the GB approach. Regarding the bridge cross strategy that involved C. baccatum to obtain the first set of interspecific hybrids, it was possible to achieve normal C. baccatum (♀) × C. chinense (♂) hybrids from all combinations, with the only exception of those combinations involving the Ají Amarillo accession (C. baccatum), which did not to set any fruit (Fig 4) and confirmed its low aptitude for interspecific crosses. Despite these hybrids had low pollen viability (19-33%)  (Table 4), similarly to C. annuum (♀) × C. chinense (♂) materials, it was possible to obtain subsequent crosses with C. annuum.
By contrast, within the reciprocal crosses C. chinense (♀) × C. baccatum (♂), fruit set and viable seeds were observed in all possible combinations (Fig 4). Unfortunately, as observed in C. chinense (♀) × C. annuum (♂) plants, all these hybrids also showed VLS, indicating the presence of detrimental genes in C. chinense cytoplasm that may also induces detrimental effects in hybrids with C. baccatum (Table 4). This is consistent with other works related to interspecific hybridization in pepper [6,7,39,40]. Therefore, as suggested for crosses with C. annuum, Table 4. Descriptive results of the genetic bridge (GB) strategy using the alternative of obtaining hybrids between C. baccatum (Cb) and C. chinense (Cch) for being subsequently crossed with C. annuum to obtain three-way hybrids. Cross combinations that did not set fruit or set fruit but seeds did not germinate are not included in the Wide Hybridization Capsicum annum and baccatum C. chinense should be also utilized as pollen donor in crosses with C. baccatum to prevent detrimental effects in the hybrid offspring. Finally, the use of C. baccatum (♀) × C. chinense (♂) hybrids as male parents for finishing the GB approach only succeed in 1 out of 48 possible combinations (and in one single plant), which also showed a 3% pollen viability (Fig 4, Table 4). Such findings are in agreement with those observed in the C. baccatum (♀) × [C. annuum (♀) × C. chinense (♂)](♂) scheme, confirming that due to the frequently low pollen viability of both interspecific hybrids, C. annuum (♀) × C. chinense (♂) and C. baccatum (♀) × C. chinense (♂), they should not be used as male parents. On the contrary, the best results were achieved utilizing C. baccatum (♀) × C. chinense (♂) hybrids as pistillate parent, which allowed finishing the GB in 15 out of the 48 possible combinations (Fig 4 and Table 4). In addition, many of these hybrids showed high levels of pollen viability (>50%). Therefore, this alternative is more efficient for the GB when starting the cross scheme crossing C. baccatum with C. chinense.
As a whole, the comparison between the different alternatives for GB showed that the use of [C. annuum (♀) × C. chinense (♂)] (♀) × C. baccatum (♂) crossing schemes allowed obtaining plants with a range of pollen viability (4-83%) in six cross combinations. The use of the [C. baccatum (♀) × C. chinense (♂)] (♀) × C. annuum (♂) scheme provided 13 fertile combinations with up to 94% pollen viability. Therefore, according to these findings, we recommend to utilize the latter alternative to introgress genes from C. baccatum to C. annuum. Moreover, this strategy provides 50% C. annuum genome, while three-way hybrids from the other scheme will only carry 25% of C. annuum genome. Nevertheless, it should be also considered that this strategy, depending on the character to introgress, may require checking the resulting phenotypes at the first step from crosses between C. baccatum and C. chinense before performing the second cross involved in the bridge cross. In fact, Yoon and Park (2005) [12] utilized the other alternative, despite they also found low pollen viability.
C. frutescens as genetic bridge species. In comparison to C. chinense, C. frutescens showed more difficulties to be used as a bridge species. Thus, although C. frutescens is phylogenetically close to C. annuum and it is generally supposed that hybridization between both species can be done easily [9], results suggest that such compatibility is not so obvious. In this way, when C. frutescens was utilised as pollen donor with C. annuum, hybrid seeds did not germinate (Fig 2), which is consistent with other authors who have reported some cases of embryo abortion and even after using embryo rescue they obtained partially sterile hybrids [41].
Otherwise, when C. frutescens was utilized as pistillate parent, only 4 out of 12 possible combinations were successful, in addition, once germinated they showed dwarfism (Table 5 and Fig 5). In this regard, while some authors found partial embryo abortion [41], Yazawa et al. [40] obtained dwarfism in all hybrids obtained and concluded that dwarfism is controlled by two complementary dominant genes. Thus, C. frutescens (♀) × C. annuum (♂) crosses seem to follow the Bateson-Dobzhansky-Muller (BDM) model, which explains the simplest scenario for generating genetic incompatibilities without detriment to either of two diverging lineages. The 'classical' BDM model involves two epistatic loci aaBB and AAbb wich are innocuous in their native context but interact negatively in hybrids [42].
Regarding the crossing between C. baccatum and C. frutescens, hybrids were only obtained when the latter was utilized as pistillate parent (Fig 4) as was early observed by Rao et al. [43]. However, in agreement with Pickersgill [44] and the results of this experiment with C. chinense, all hybrids were also affected by virus-like syndrome (VLS), probably due to C. frutescens cytoplasmic genes, which interact with C. baccatum nuclear genes. In addition, although Rao et al. [43] were able to recover normal hybrids in appearance, they were sterile. Because of that, despite more C. frutescens genotypes could be tested for compatibility with C. annuum or C. baccatum, in our opinion C. frutescens should not be used as bridge between both species.
In vitro rescue of immature embryos from crosses between C. annuum and C. baccatum In a preliminary experiment, a diallel cross scheme was done between eight accessions of C. annuum (California Wr. both red and yellow, Arnoia, Bierzo, Guindilla, Pasilla, Piquillo, and Serrano) and the three C. baccatum accessions. The few set fruits achieved were hasvested at the fully ripe stage and their seeds were removed and evaluated. All seeds were empty and showed necrotic points in the center (Fig 6), which is frequently related to embryo abortion. In fact, as expected, germination tests demonstrated that these seeds were not viable. Those results were in agreement with other authors [6,23] and confirmed that embryo abortion is an important barrier in crosses between C. annuum and C. baccatum. In addition, further anatomical studies of interespecific embryos close to abortion revealed that the main cause of abortion was an early hardening of the endosperm, which limited the development of the embryos and caused morphological deformities and, eventually, embryo collapse and death (Fig 7). Therefore, embryo rescue was the only alternative to achieve these interspecific hybrids.  Table 5. Descriptive results of the bridge cross technique using C. frutescens as genetic bridge between C. annuum and C. baccatum. Those F 1 combinations which did not set fruit, or set fruit but seeds did not germinate, are not included in the As a whole, the use of C. baccatum accessions as male parents provided the highest efficiency in terms of both fruit set and number of regenerated hybrids. Thus, fruit set was recorded in 67% of C. annuum (♀) × C. baccatum (♂) combinations (26 out of 36), although the efficiency rate ranged from 4% to 30%, depending on the genotypes involved (Table 6). Interspecific embryos could be excised and cultured in vitro from half of these combinations, while no embryos were found in the other half, which might be due to embryo abortion at very early stages (globular or earlier), whose rescue and regeneration would be extremely difficult or unfruitful [29,36]. Finally, plantlets from 10 C. annuum (♀) × C. baccatum (♂) combinations were obtained and reached the adult stage. The efficiency of this in vitro technique varied among combinations, ranging from 6% to 75% in P534 × AjíR and P716 × AjíR, respectively (Table 6). Furthermore, combinations involving Ají Rojo and Brazilian Pumpkin C. baccatum accessions yielded the highest number of C. annuum (♀) × C. baccatum (♂) individuals: 36 and 16 hybrid plants, respectively, while only 3 hybrid plants from a single combination were obtained with Ají Amarillo (Table 6).
On the other hand, a considerably lower efficiency was found in C. baccatum (♀) × C. annuum (♂) hybridizations and fruit set fruit was only observed in 14 out of the 36 possible combinations (Table 6). In addition, only three hybrid plants from two combinations, Ají R × Bie and Aji R × Piq, were eventually obtained (Table 6), consequently, breeders are advised to use C. annuum as female parent in crosses with C. baccatum to achieve the highest efficiency.
In both cases, pollen fertility was relatively low, comprised between 0% and 33% of P716 (♀) × Aji A (♂) and Ají R (♀) × Bie (♂) ( Table 6), which was comparatively lower than the values recorded in the GB strategy. In any case, such values are in agreement with those reported by other authors [24,45].
Finally, also supported by in vitro rescue, BC 1 generations towards C. annuum were attempted. Each hybrid was backcrossed with the corresponding C. annuum parent for compatibility reasons. All BC 1 crosses involving hybrids as male parents failed and no fruit was obtained. As a result, fruit set was only achieved when hybrids were utilised as female parents. Furthermore, BC 1 materials were only achieved with C. annuum (♀) × C. baccatum (♂) hybrids, while all BC 1 crosses using C. baccatum (♀) × C. annuum (♂) hybrids failed (Table 7). Despite fruit set was low and, consequently, only three BC 1 combinations were obtained, in vitro rescue enabled to regenerate a high number of embryos (Table 7).

Conclusions
According to the results, wide hybridization between C. annuum and C. baccatum is possible using both the GB and ER approaches, although the degree of success is highly dependent on the genotype to obtain interspecific hybrids and subsequent generations. The best crossing schemes to obtain successful hybridization and introgression from C. baccatum to C. annuum have been identified (Fig 8), and the genotypes with the best performance in these experiments are good candidates for introgression breeding from C. baccatum to C. annuum. Ultimately, these results provide breeders with relevant information on wide hybridization approaches and on appropriate plant material to be used for successfully incorporate the C. baccatum gene pool as a source of variation for introgression breeding in C. annuum breeding programmes. Additionally, those breeders interested in introgressing genes of interest in C. annuum from C. chinense or C. frutescens should consider the prezygotic and postzygotic barriers between these species identified here in order to achieve a high efficiency.