World Vegetable Center Eggplant Collection: Origin, Composition, Seed Dissemination and Utilization in Breeding

Eggplant is the fifth most economically important solanaceous crop after potato, tomato, pepper, and tobacco. Apart from the well-known brinjal eggplant (Solanum melongena L.), two other under-utilized eggplant species, the scarlet eggplant (S. aethiopicum L.) and the gboma eggplant (S. macrocarpon L.) are also cultivated. The taxonomy and identification of eggplant wild relatives is challenging for breeders due to the large number of related species, but recent phenotypic and genetic data and classification in primary, secondary, and tertiary genepools, as well as information on the domestication process and wild progenitors, facilitates their utilization in breeding. The World Vegetable Center (WorldVeg) holds a large public germplasm collection of eggplant, which includes the three cultivated species and more than 30 eggplant wild relatives, with more than 3,200 accessions collected from 90 countries. Over the last 15 years, more than 10,000 seed samples from the Center's eggplant collection have been shared with public and private sector entities, including other genebanks. An analysis of the global occurrences and genebank holdings of cultivated eggplants and their wild relatives reveals that the WorldVeg genebank holds the world's largest public collection of the three cultivated eggplant species. The composition, seed dissemination and utilization of germplasm from the Center's collection are highlighted. In recent years more than 1,300 accessions of eggplant have been characterized for yield and fruit quality parameters. Further screening for biotic and abiotic stresses in eggplant wild relatives is a priority, as is the need to amass more comprehensive knowledge regarding wild relatives' potential for use in breeding. However, as is the case for many other crops, wild relatives are highly under-represented in the global conservation system of eggplant genetic resources.

INTRODUCTION are incompletely understood (Levin et al., 2006;Meyer et al., 2012). Here, we limit ourselves to identify global occurrences and regions of diversity. A key section is the overview of global genebank holdings of cultivated eggplant and their wild relatives. As we shall demonstrate, for such plants the collection at the WorldVeg is of paramount importance. Composition, seed dissemination and utilization of germplasm from this collection are presented and discussed. The importance of safeguarding and evaluating wild relatives is highlighted, as crop wild relatives are highly under-represented in the global conservation system of plant genetic resources and may harbor important genes for resistance or tolerance to biotic and abiotic stresses.

TAXONOMY, WILD RELATIVES, AND DOMESTICATION OF EGGPLANT
Eggplants are berry-producing vegetables belonging to the large Solanaceae family (nightshade family), which contains ∼3,000 species distributed in some 90 genera (Vorontsova and Knapp, 2012). Out of these Solanum L. is the largest one, with around 1,500 species (Frodin, 2004) including globally important crops such as potato (Solanum tuberosum L.) and tomato (Solanum lycopersicum L.), as well as many other minor crops. Most taxa of Solanum genus have a basic chromosome number of n = 12 (Chiarini et al., 2010).
The Solanum genus is mega-diverse and can be divided into 13 clades, where eggplant is the member of the large and taxonomically challenging Leptostemonum clade (subgenus Leptostemonum Bitter; Knapp et al., 2013), which is commonly known as the "spiny Solanum" group due to the presence of sharp epidermal prickles on stems and leaves . The subgenus Leptostemonum contains around 450 currently recognized species distributed worldwide , many of which originated in the New World (Vorontsova and Knapp, 2012). All three cultivated eggplant species have the Old World in origin (Figure 1). The Old World (Africa and Eurasia) and Australia, are home to more than 300 Solanum species (Levin et al., 2006;Vorontsova and Knapp, 2016). Solanum melongena and S. macrocarpon are usually included in section Melongena Dunal (Lester and Daunay, 2003;Lester et al., 2011), whereas S. aethiopicum is assigned to section Oliganthes (Dunal) Bitter (Lester, 1986).
Solanum melongena is characterized by large morphological diversity, and frequently it has been considered as the same taxonomic species than its wild ancestor S. insanum L. (Ranil et al., 2017). Four taxonomically informal groups, labeled E-H, were considered by Lester and Hasan (1991) to describe the different types of wild and weedy eggplant as well as their distribution (Table 1). However, these four groups are presently considered as representing two different species: the cultivated eggplant S. melongena and its wild ancestor S. insanum . In this way, groups E and F corresponding to extremely prickly and plants that grow wild or weedy in India and Southeast Asia are now included within S. insanum (Ranil et al., 2017). The plants of group G correspond to primitive eggplant cultivars, with small fruits, while the plants of group H are less FIGURE 1 | Schematic representation of taxonomic relationships between the cultivated brinjal eggplant (Solanum melongena) and other cultivated (scarlet eggplant, S. aethiopicum; and gboma eggplant, S. macrocarpon) and wild relatives from the genus Solanum based on Nee (1999), Levin et al. (2006), Weese and Bohs (2010), Stern et al. (2011), Knapp et al. (2013), Syfert et al. (2016), and Vorontsova and Knapp (2016). For each of the species and groups it is indicated if they are part of the primary (GP1), secondary (GP2), or tertiary (GP3) brinjal eggplant genepools. The three cultivated species are indicated with an asterisk.

Species
Groups prickly than other groups and consist of large-fruited landraces and modern cultivars (Daunay et al., 2001;Weese and Bohs, 2010; Table 1). Both groups, G and H, constitute S. melongena . Some studies (Hurtado et al., 2012;Vilanova et al., 2012;Cericola et al., 2013) have also pointed to a genetic and morphological differentiation between Occidental (eggplants from the Mediterranean area, North of Africa, and Middle East) and Oriental (from southeast and eastern Asia).
Solanum aethiopicum is also hyper-variable and is classified into four cultivar groups (Gilo, Shum, Kumba, and Aculeatum; Table 1) based on morphological characteristics and use (Lester, 1986). The Gilo group has edible fruits with different shapes, color, and size, and hairy, inedible leaves; the Shum group has glabrous and small leaves that are eaten as a green vegetable but the fruits are inedible; the Kumba group has glabrous leaves and flattened large fruits, which are edible; the Aculeatum group, on the other hand, has more prickliness than other groups with flatshaped fruit, and are used as ornamentals (Lester, 1986;Prohens et al., 2012;Plazas et al., 2014a).
Although, recent information exists on domestication of eggplants, there are still many unanswered questions about this process. Vavilov (1951) considered S. melongena as being native to the "Indo-Chinese center of origin." However, recent evidence suggests that brinjal eggplant had a multiple independent domestication from S. insanum, which is naturally distributed in tropical Asia from Madagascar to the Philippines  in several centers of domestication (Meyer et al., 2012). Although, the evidence of cultivation of eggplant in both India and China is equally old, archeological evidence suggests that utilization of wild eggplants may have started earlier in India than China, with a subsequent additional and independent center of domestication in the Philippines (Meyer et al., 2012). Around the eighth century, eggplant spread eastward to Japan and then westward along the Silk Road into Western Asia, Europe, and Africa by Arab traders during the fourteenth century, then it was introduced into America soon after Europeans arrived there (Prohens et al., 2005) and later expanded into other parts of world. Much less is known on the domestication of the scarlet and gboma eggplants. Both species were domesticated in Africa, from its respective wild ancestors, which are S. anguivi Lam. in the case of S. aethiopicum (Lester and Niakan, 1986) and S. dasyphyllum Schumach. and Thonn. in the case of S. macrocarpon (Bukenya and Carasco, 1994). Hybrids between cultivated eggplants and their respective wild ancestors are fully fertile (Lester and Thitai, 1989;Bukenya and Carasco, 1994;Plazas et al., 2016).
Solanum melongena and the two other cultivated eggplants are related to a large number of wild species Syfert et al., 2016) that may serve as sources of variation for breeding programs, in particular for traits related to adaptation to climate change but also pest and disease resistance (Rotino et al., 2014). Some of these species are listed in Table 2. Although, the brinjal eggplant is considered to be a vegetable of Asian origin, most eggplant wild relatives are from Africa (Weese and Bohs, 2010). Wild eggplants produce small, bitter, multi-seeded fruits, almost always inedible, and the plant is generally very spiny. Some of them possess high levels of chlorogenic acid and other bioactive compounds, which may have potential interest for human health (Meyer et al., 2015). The wild relatives of eggplant are one of the most variable and intricate groups, in regards to their taxonomic and phylogenetic relationships . Based on crossing and biosystematics data, nine wild species, together with S. melongena, form the "eggplant complex, " which includes the cultivated brinjal eggplant and its closest eggplant wild relatives . Wild relatives can be classified based on their crossability with cultivated species (genepool concept) into primary, secondary, and tertiary genepools (Harlan and de Wet, 1971). The primary genepool (GP1) of brinjal eggplant consists of cultivated eggplant and its wild ancestor S. insanum (Ranil et al., 2017) which can be crossed easily and produce normal fertile hybrids (Plazas et al., 2016). The secondary genepool (GP2) includes a large number (over 40) wild relatives that can be crossed or are phylogenetically close to brinjal eggplant, but the success of the crosses and the viability or fertility of the hybrids with the brinjal eggplant may be reduced. For example, some interspecific hybrids derived from GP2 are partly sterile or weak due to reproductive barriers such as S. dasyphyllum, S. linnaeanum Hepper & P.-M. L. Jaeger or S. tomentosum L. (Rotino et al., 2014;Kouassi et al., 2016). The tertiary genepool (GP3) includes more distantly related species, including New World species, which are used in breeding programs for their resistance features, but crossing needs specific breeding techniques to succeed (e.g., S. torvum Sw., S. elaeagnifolium Cav., and S. sisymbriifolium Lam.; Kouassi et al., 2016;Plazas et al., 2016;Syfert et al., 2016).

GLOBAL OCCURRENCES AND GENEBANK CONSERVATION OF EGGPLANT AND WILD RELATIVES
In the following section we review the current status of eggplant genetic resources including the cultivated species and their most recognized wild relatives using information collected from biodiversity, herbarium, and genebank databases. The Global Biodiversity Information Facility (GBIF) was applied to review the number of recorded occurrences, which can be natural populations, herbarium samples, or other biodiversity records (GBIF, 2017). Scientific names were used as a filter in the search function. The total numbers of records per species were noted, as were clusters of occurrences that were identified visually by applying the database map function. The main cluster of S. melongena was in India, with more than 5,000 of the total number of around 18,000 occurrences. Other clusters were in Turkey, Southeast Asia, and Spain, while the main cluster of occurrences of S. aethiopicum and S. macrocarpon was in West Africa, with a total of 1,288 and 443 occurrences, respectively. Based on the literature of previous studies and characterization data available at the WorldVeg, a list of 35 crop wild relatives was included in this review, which had ∼100 (S. repandum G. Forst.) to more than 7,000 occurrences (S. torvum) on a global scale recorded by GBIF ( Table 2). Important regions for wild relatives vary depending on the species, but include all continents; Latin America, Asia, and Africa are the most common areas for wild relatives.
The Global Gateway to Genetic Resources (GENESYS, 2017) was applied to review the number of conserved genebank accessions. The database includes more than 3 million accessions, which is less than half of the estimated number of more than 7 million accessions that are conserved globally (FAO, 2010). Although, not all national genebanks report to Genesys, we still used the information for reviewing global holdings. Scientific names were used as a filter in the search function of the database, and the most important holding institutions were identified from the summary function of the database. Additional sources were reviewed to try to capture important collections outside Genesys, including national genebank databases and the database for  Ranil et al., 2017). This is probably caused by the fact that many S. insanum accessions are probably conserved as S. melongena, as both species have often been considered as being a single species (S. melongena; e.g., Lester and Hasan, 1991). Also, the correct classification of accessions under "S. indicum L." should be determined, as this name was rejected in 1978 as it was used to refer to two clearly distinct species, the African S. anguivi and the Asian S. violaceum (Vorontsova and Knapp, 2016).
According to our analysis, wild eggplants are greatly underrepresented in ex situ repositories. Such findings are also reported by Castañeda-Álvarez et al. (2016), where eggplants were among the crops whose wild genepools are highly under-represented. Indeed, there is a need for conducting collection missions and conservation actions for eggplant wild relatives (Conservation gaps, http://www.cwrdiversity.org/conservation-gaps/, Accessed February 30, 2017).

EGGPLANT GERMPLASM DISSEMINATION FROM THE WORLD VEGETABLE CENTER
As demonstrated in the previous section, the collection at the WorldVeg is the most significant eggplant collection worldwide. Eggplant is the Center's third most widely distributed vegetable crop after pepper and tomato. A total of 11,383 germplasm samples were distributed from WorldVeg headquarters to 90 countries from the period 2000 to 2017. Most of these were of S. melongena (10,519 samples; 92.4%), followed by S. aethiopicum (738 samples; 6.4%) and S. macrocarpon (126 samples; 2.2%; Table 3). These accessions correspond to landraces and traditional cultivars with significant diversity in plant morphology, fruit types and colors, and resistance to biotic and abiotic stresses. The largest share of germplasm samples went to other genebanks (7,042 samples; 61.8%), followed by The large morphological diversity of the WorldVeg collection is matched by the identification of traits of significant agronomic interest. WorldVeg has compiled and maintained the world's largest germplasm collection of eggplant, and national genebanks and institutions from around the globe have requested and received many samples. A significant number of accessions are internal distributions to WorldVeg regional offices, and in collaboration with partner institutions, the material has been used in breeding programs. New open-pollinated varieties have been released in Uzbekistan, Tanzania, and Mali through selection based on local trait preferences ( Table 4).

UTILIZATION OF EGGPLANT GERMPLASM IN BREEDING
Screening of available accessions for targeted traits (evaluation) and morphological description of the accessions (characterization) are key issues for the breeding process. At the WorldVeg a large number of commercial cultivars, landraces, and germplasm have thus been examined to identify desired genotypes for use in eggplant breeding programs or for recommending to private sector seed companies and other partner institutions. Standardized descriptors included characters both for germination, the vegetative phase, inflorescence descriptors, and fruit and seed traits, respectively ( Table 5).
Large variation in yield parameters and in fruit quality parameters have been documented in the collection (Figures 2, 3). Such data have been compiled over years and can be retrieved from AVGRIS, the World Vegetable Center genebank database system (2017). Among the 1,308 accessions of S. melongena that have been characterized, green and purple fruits were predominant, and could be found in 47 and 38% of the total number of accessions, respectively. Slightly longer than broad, and as long as broad, were the prevalent shapes of the accessions, with 31.1 and 18.7%, respectively. Similarly, huge diversity was found among 98 accessions belong to S. melongena, S. aethiopicum, and S. macrocarpon for 16 morpho-agronomic and fruit traits including plant height, flowering time, flower/inflorescence, fruit length and fruit acidity, but weak association was found between among morpho-agronomic and fruit quality descriptors (Polignano et al., 2010). In terms of fruit taste, 26.8% of accessions had a sweet taste, 53.2% had an intermediate taste and some accessions had bitter taste (6.1%). Large variations in fruit dry matter content, total sugar content, and fiber content of the fruit have been determined in a study of 90 selected eggplant genotypes (AVRDC, 1996). The distribution of dry matter, total sugar, and fiber contents ranged from 5.5 to 10.1, 7.0 to 40.1, and 4.7 to 18.1%, respectively. In another study conducted at the WorldVeg, 33 S. melongena accessions and two S. aethiopicum accessions were evaluated for superoxide scavenging and content of total phenolics and ascorbic acid (Hanson et al., 2006). Solanum melongena accessions S00062, S00022, and S. aethiopicum accession S00197 exhibited high antioxidant activity (Hanson et al., 2006).
Accessions with important traits such as early maturity, high yielding, and resistance to biotic stresses have been identified in the WorldVeg germplasm collection ( Table 6). Based on data from Chen (1998) and the examination of 40 accessions from the WorldVeg collection, among long fruit genotypes, VI045551, VI047333, VI046110, and VI037736 were identified as stable and high yielding (>40 tons per hectare) over spring, summer, and autumn seasons. Accession VI046110 had the highest average yield and the earliest maturing genotype across the three seasons (AVRDC, 1999). In round fruit type, VI046097, VI047332, VI44067, EG233, and EG235 produced the high yields in all three seasons.
Based on data from AVGRIS (2017) compiled over the years and including 1,300 accessions, only 90 accessions (6.8%) had more than 5,000 g of fruit yield per plant (Figure 2). Marketable yields were highly associated with fruit weight and number of fruits per plant. Large diversity in the WorldVeg germplasm collections enabled us to develop several improved eggplant and African eggplant cultivars (Table 3). A total of three eggplant varieties have been commercialized in Uzbekistan and three African eggplant varieties have been released in Tanzania and Mali.
More than 200 accessions have been evaluated for resistance to bacterial wilt (Ralstonia solanacearum) at the WorldVeg under both greenhouse and field conditions (AVRDC, 1999). Among these, 38 accessions were identified with high levels of resistance. These accessions were retested using root wounding and soil drenching inoculation methods in the greenhouse. Data were summarized from the screening and retest studies, and the most resistant accessions were TS3, VI034885, and TS47 from Malaysia; and TS69, TS87, and TS90 from Indonesia with disease indices <10% under both greenhouse and field conditions.
Resistance to eggplant fruit and shoot borer (Leucinodes orbonalis Guenee), leafhopper (Amrasca devastans Distant), and aphids (Aphis gossypii Glover) have been identified at WorldVeg in separate trials (AVRDC, 1999). Leafhoppers and aphids have piercing mouthparts and suck the sap, especially from the leaves, which leads to yellow spots on the leaves, followed by crinkling, curling, bronzing, and drying (or "hopper burn" from leafhopper), but severe aphid infestations cause young leaves to curl and become deformed (AVRDC, 1999;Ramasamy, 2009). Like whiteflies, aphids also produce honeydew, which leads to the development of sooty mold (Ramasamy, 2009). Accessions VI034971, VI035822, and VI035835 were found promising accessions against leafhopper and aphids. Eggplant fruit and shoot borer is an extremely destructive pest, especially in South and Southeast Asia (Ramasamy, 2009). It lays eggs on the foliage and neonate larvae feeds on the tender shoots, boring into the

Code
Definition Scale S110 Germination period Number of days from sowing until first germination S120 Cotyledonous leaf length mm (N = 10) S130 Cotyledonous leaf width mm (N = 10) S140 Cotyledonous leaf color 3 = Green, 5 = Light violet, 7 = Violet, X = Mixture     shoots and fruits, resulting in wilting of young shoots, followed by drying; the fruit becomes unfit for marketing and consumption. Total resistance was not found and moderate resistance was found only in one accession, VI047451 (AVRDC, 1999). This was based on typical damage symptoms, wilting of shoots and feeding holes in a wilted shoot, as well as damaged fruit. Overall, these results show that very promising materials for breeding pest tolerant or resistant varieties can be found in the WorldVeg eggplant collection. However, additional race specific screening is needed to find resistant sources for pests where no resistance or limited resistance has been found.

THE WAY FORWARD
The food security of many countries relies on crops bred from genetic resources outside their region . Therefore, plant genetic resources are a global concern where access and benefit sharing is of paramount importance. Eggplant is an important vegetable crop with a global cultivation area.
From the current study we have confirmed that there are critical gaps in global eggplant collections, especially related to crop wild relatives (Syfert et al., 2016). We have listed more than 35 wild species conserved in germplasm collections, but for many other eggplant wild relatives no accessions are conserved in genebanks; in addition, there still might be undiscovered crop wild relatives. Genetic diversity in wild relatives is much higher than in cultivated eggplant  and could be valuable sources for resistance to biotic and abiotic stresses (Daunay and Hazra, 2012). To date, a limited number of wild relatives have used in eggplant breeding (Rotino et al., 2014) and commercial varieties containing wild relative introgressions are not yet available. To move forward, screening for abiotic and biotic stresses in wild relatives should be intensified and broadened for identification of valuable germplasm accessions for breeding improved eggplant varieties. This information, combined with genomics studies for the detection of genes and QTLs of agronomic importance and their associated markers, will be of great utility in eggplant breeding, as has been demonstrated in some association mapping studies (Cericola et al., 2014;Portis et al., 2015). Recent reviews of the development in eggplant is provided by Frary and Doganlar (2013) and Gramazio et al. (in press). From a utilization point of view, core collections could be established and stakeholders should work together for the development of the next generation of eggplant varieties that can meet the challenges of the present and the future.

AUTHOR CONTRIBUTIONS
DT compiled the major parts of the text; SS contributed with text on genetic resources; JP contributed with text on eggplant wild relatives; YC contributed with reviewing databases; MR and TW contributed with inputs on eggplant taxonomy and breeding.