Chili pepper (Capsicum sp.) is one of the oldest domesticated crops in the Americas (Davenport, 1970). Currently, chili pepper production exceeds 60 million tons annually (FAOSTAT, 2020), with cultivation in more than 140 countries and an annual revenue of 50 billion dollars. The Capsicum genus includes more than 30 species, from which only C. annuum, C. frutescens, C. chinense, C. baccatum and C. pubescens are considered domesticated. C. annuum is the most phenotypically diverse and broadly distributed species in the genus (Tripodi and Kumar, 2019). Genetic and anthropological evidence suggest C. annuum was domesticated from the wild chiltepin pepper (Capsicum annuum var. glabriusculum) about 8,000 years ago in a region spanning tropical and subtropical America (Davenport, 1970; Hunziker, 2001; Perry et al., 2007). The major differences between cultivated and wild chili peppers are related to fruit morphological and metabolomic traits (Kumar et al., 2018). The chiltepin parent has small, round fruits which are erect, highly pungent, and dehiscent, and chiltepin fruit are primarily eaten by birds and picked from wild stands by humans (Tewksbury and Nabhan, 2001; Kumar et al., 2018). In contrast, cultivated peppers vary in their degree of pungency and external appearance, although they are typically large, elongated, pendant and non-dehiscent, remaining on the plant until harvest. Additionally, domesticated and wild accessions of Capsicum differ in other traits that could play an important role in their fitness and local adaptation such as plant architecture, phenology and physiology. Typically, chiltepin plants are highly branched perennial shrubs with small, round leaves that can become climbing plants (Hernandez-Verdugo et al., 1999; Hayano-Kanashiro et al., 2016). In the wild, these plants can be found in arid regions and their immature stems and fruits produce pigments as protection against UV radiation (Borovsky et al., 2004). On the other hand, due to their adaptation to agricultural production systems, specially those in greenhouse conditions, domesticated chili plants are compact, annual in habit, with large leaves and have little production of pigments, with likely an array of additional adaptations to the conditions of greenhouses and monocultures.

Previous genetic mapping studies of crop domestication traits have identified genomic regions that govern important differences between wild and cultivated plants (Paterson et al., 1988; Doebley et al., 1995; Li et al., 2006; Simons et al., 2006; Dempewolf et al., 2012). Based on the results of these studies, Ross-Ibarra (2005) proposed the generalization that crop domestication is driven by a low number of loci with relatively large effects and a preponderance of recessive action. Although these studies have focused on the major traits associated with the species domestication syndrome, the consequences of domestication, ongoing improvement and dispersal to new environments have been reported to have broader impacts on cultivated plants (Hernández-Terán et al., 2017).

In the current study, a quantitative trait locus (QTL) mapping analysis was conducted in an F₂:₃ population derived from Puya, a domesticated accession, and Chiltepin, a wild ancestor of cultivated pepper, to identify genomic regions associated with traits of agronomic importance or related to the domestication syndrome. We identified novel marker-trait associations by combining standard field evaluation and computer vision strategies using 19 phenotypic traits related with fruit and leaf morphology, phenology and plant architecture. In addition, our QTL analysis showed that most of the traits analyzed share patterns such as few loci of relatively large effect, preponderance of domesticated recessive loci, clustering of QTLs in certain genomic regions or pleiotropic loci, and biased effect of QTLs toward the domesticated phenotype for strongly selected traits.

Continued selection and use during the domestication of Capsicum have produced a broad spectrum of evolutionary changes. These changes have led to the divergence of cultivated chili peppers from wild accessions. In this study, a QTL mapping study using wild and domesticated chili pepper germplasm was conducted to map domestication syndrome phenotypes as well other traits of agronomic interest. Variation in some of the evaluated traits was derived from direct selection during the domestication process (e.g., fruit size/shape, fruit orientation and deciduous fruit), However, for some other traits, the observed variation may have been produced indirectly (e.g., organ pigmentation, plant architecture and leaf size/shape traits) (Ross-Ibarra et al., 2007; Kumar et al., 2018). In our mapping population the greatest variation was observed for traits of both categories such as plant architecture (plant height and main stem length), fruit size and shape, organ pigmentation and fruit orientation.

The phenotypic contrast between the wild and domesticated chili pepper parental lines used in the current study allowed the analysis of a greater variety of traits of agronomic interest than in previous reports (Naegele et al., 2014; Chunthawodtiporn et al., 2018), which emphasizes the importance of using wild germplasm in mapping studies. In total 19 qualitative and quantitative traits related to plant architecture, fruit, and phenology were evaluated. Many of the trait frequency distributions resembled the expected segregation patterns for qualitatively and quantitatively plant traits. In particular, the ratio observed for fruit orientation suggests that the erect fruit trait is inherited as a recessive gene as previously reported in chili peppers (Lee et al., 2008) and other crops (Sun et al., 2019). Similarly, segregation patterns for chili pepper stem pigmentation in our study suggests that this trait is controlled by a single dominant gene, which agrees with previous Capsicum reports for pigmentation in foliage, flowers and fruit (Borovsky et al., 2004). On the other hand, a distortion in the expected segregation ratio (3:1) was observed for some traits such as the form of unripe fruit, unripe fruit pigmentation and growth habit. This distortion is probably due to the lack of phenotypic data for almost a third of the population due to semi-sterility, the (non-random) loss of plants in the F₂ population (Supplementary Table S1), or the bias of qualitative phenotyping. As in most domesticated species (Ross-Ibarra, 2005; Diaz-Valenzuela et al., 2020), the phenotypic data was skewed towards the wild phenotype in both qualitative and quantitative traits. Although this trend has been analyzed directly for target phenotypes during crop domestication, our results show that this trend is also present for non-target traits such as branch angle, leaf traits, plant height, main stem length and stem pigmentation (Figure 2B). These results suggest that the genetic changes produced by domestication affect non-target traits and that the inclusion of these traits is necessary for a better understanding of conscious and unconscious selection.

As expected, significant positive correlations were observed among traits related to leaf, fruit shape/size and organ pigmentation. Interestingly, negative correlations were observed between both fruit shape and size with traits associated with the wild phenotype (e.g., fruit orientation and unripe fruit pigmentation), which may be due to the relatively low rate of recombination in an F₂ population. Notably, the possibility of analyzing traits such as form of unripe fruit, seedless fruit, deciduous fruit and growth habit which have not been analyzed in previous studies highlight the value of using exotic and wild accessions in mapping experiments.

To the best of our knowledge, this is the first genetic map derived from a cross between a wild pepper (Capsicum annuum var. glabriusculum) and domesticated pepper. This genetic map showed a great consistency (r > 0.99) with the cytological characteristics and physical architecture of the chili pepper genome (Kim et al., 2014). Moreover, a previously reported reciprocal translocation between chromosome 1 and 8 (Wu et al., 2009; Park et al., 2014) was confirmed. This major translocation in the domesticated genome differentiates it from its wild ancestor and other related species such as C. frutescens and C. chinense (Wu et al., 2009). It has been suggested that, as in other species (Jáuregui et al., 2001; Farré et al., 2011; Stathos and Fishman, 2014).), this translocation acts as a genetic barrier due to sterility, particularly between domesticated and ancestral forms (Dempewolf et al., 2012; Lee et al., 2016).

In chili pepper, variations in shape and increase in fruit size are the main changes observed in domesticated fruits, which is associated with the change in fruit orientation (from erect fruit in the ancestor to pendant fruit in domesticated forms). In this study, 10 QTLs were identified for fruit size and shape related traits, of which five were located in a common region of chromosome 2 (140.5–163.2 Mb). Several QTLs of fruit size and shape on chromosome 2 have been reported in different genetic backgrounds in chili peppers (Han et al., 2016; Chunthawodtiporn et al., 2018). The QTL on chromosome 2 co-localizes with the ovate gene which has been directly associated with fruit shape and size in pepper (Tsaballa et al., 2011). On the other hand, Individual QTLs for fruit-related traits have been previously reported on chromosome 4 (Lee et al., 2020), however, no cluster of QTLs has been indicated in this region as in this study. Similarly, QTL FWE1.1 on chromosome 1 colocalizes with FW-1 previously reported by Han et al., 2016, while there were no equivalent QTLs identified in chromosome 12 (FS12.1) with those previously reported for the corresponding traits. This may be explained by the differences in parentals, ours including a wild relative. A major QTL (FO12.1) was identified for the fruit orientation trait on chromosome 12, which agrees with previously reported associations (Lee et al., 2008; Han et al., 2016).

The adaptation of domesticated plants to agricultural environments led to the loss of useful alleles for wilder and more severe environments (Barchenger et al., 2019). In particular, organ pigmentation present in wild plants represents a useful protective mechanism against abiotic factors such as radiation (Winkel-Shirley, 2001). Fruit and stem pigmentation were analyzed in this population, and we detected a colocalized QTL on chromosome 10. Only unripe fruit pigmentation has been analyzed in previous studies (Han et al., 2016), in which the A gene has been identified as responsible for this phenotype (Borovsky et al., 2004); therefore, it is likely that both unripe fruit pigmentation and stem pigmentation share the same genetic basis. On the other hand, the perennial growth habit that characterizes wild plants changed to an annual growth habit in cultivated forms, which represents an advantage for their cultivation rather than for their survival in their natural environment. The phenotypic variation for the growth habit trait (GH) allowed the detection of a novel QTL (GH1.1) on chromosome 1. The loss of dispersal mechanisms was an important change that allowed the larger fruits to remain on the plant until they were harvested manually. A novel QTL linked to the deciduous fruit was identified in chromosome 10.

As mentioned previously, a translocation was detected between the wild and domesticated genomes, and to which can be attributed the sterility of some of the progeny observed in the population as in other species (Jáuregui et al., 2001). Two novel QTLs (SF1.1 and SF6.1) for sterility were identified, which might be useful in breeding programs as well as for future evolutionary studies.

The results of the QTL analysis presented herein show that the genetics of Capsicum domestication follows the predominant patterns in most species; few loci of relatively large effect, preponderance of recessive domesticated alleles, clustering of QTLs, pleiotropic loci and effect of QTLs biased toward the domesticated phenotype for strongly selected traits. The number and effect of QTLs identified in this study suggests that Capsicum crop evolution was driven primarily by a few loci of large effect that may have caused major phenotypic leaps, as has been suggested in other species (Doebley and Stec, 1993; Koinange et al., 1996; Xiong et al., 1999). This condition of the phenotypes of interest could have made the selection more efficient and ultimately allowed a faster transformation towards the domesticated forms (Barton and Keightley, 2002). Because random mutations are likely to produce primarily loss of function, most domesticated traits have been reported as recessive traits that usually decrease fitness in the wild (Ladizinsky, 1985; Pitrat, 1986; Sogaard and von Wettstein-Knowles, 1987; Lester, 1989). Our results show that for traits directly related to domestication this pattern is true with few exceptions, 11 of 12 QTLs were recessive to the domesticated phenotype. Furthermore, for most polygenic traits, the gene action of the QTLs corresponded to a skew distribution towards the wild type (towards the lowest value) (Figure 2B). QTL clustering, a popular attribute in the genetics of species domestication, was also detected in this study. This pattern has been mostly interpreted as closely linked loci or as a pleiotropic for several traits. It has been proposed that this condition has been opportune during domestication since the QTL clusters would become fixed faster (Poncet et al., 1998). Among the traits involved in the domestication of Capsicum, fruit shape and size traits were the main objective. For all QTLs identified for these traits the effects are biased towards the domesticated phenotype presumably due to the strong directional selection exerted on these traits.

Our phenotypic and QTL information, specifically for fruit traits, follow the same patterns reported by Diaz-Valenzuela et al., 2020 at the transcriptome level for the F₁ hybrid of our same population. The general observation of recessiveness biased towards the domesticated phenotype in our study is consistent with the bias of the entire transcriptome towards the wild parent. In addition, the authors propose that the phenotypic divergence is mainly influenced by trans regulation, which complements the findings of the genetic architecture of the traits analyzed in this study. The relevance of trans-acting regulation variation in the phenotypic transformation of this species could also partially explain the pleiotropic effects of selection during domestication on both target and non-target traits. Thus, the findings of both studies carried out on the domestication of Capsicum provide relevant information to support the role of the omnigenic model in this process.

This study provides novel insights on the genetic control of both agronomic traits and traits of evolutionary interest in Capsicum. Our results demonstrate that the genetics of Capsicum domestication follows the predominant patterns in most species. The genetic map revealed a reciprocal translocation which is a genetic barrier mechanism that commonly accompanies domestication and contributes to differentiation between crops and their ancestors on short evolutionary scales. The analysis of target and non-target traits suggests that their consideration will allow a better understanding of crop evolution and will facilitate the design of strategies for the use of wild material in breeding processes. Additionally, the phenotypic variation observed in our population allowed the detection of QTLs for a wide variety of traits, including several that were explored for the first time (FUF, SF, DF and GH). Also, due to the role of structural variations in environmental adaptation (Huang et al., 2021; Songsomboon et al., 2021), the observed translocation between chromosome 1 and 8 may have been involved in the dispersal of Capsicum accessions to different agroecosystems. The confirmation of this hypothesis and experimental validation of genes identified within our QTLs will provide valuable information in the history of the less-known Capsicum domestication compared to other Mesoamerican crops.