Melon (Cucumis melo L.) is an extraordinarily polymorphic species, comprising 15 cultivated and wild botanical groups classified into the two subspecies melo and agrestis (Pitrat, 2008). Sweet melons, which now dominate the global market, emerged relatively recently in Europe during the Middle Ages (Paris et al., 2012). In contrast, vegetable melons, which do not accumulate sugars during ripening and are typically consumed at the immature stage, were already popular in ancient Mediterranean civilizations. However, their cultivation is now limited to small, localized areas, making them neglected and underutilized crops (Paris et al., 2012).

Cucumber melon (Cucumis melo var. chate) is a vegetable melon that produces tender, crispy, cucumber-like pepos with an approximate width-to-length ratio of 1:2. Its epicarp, ranging in color from pale to dark green, is often covered with soft hairs. Cucumber melons were already depicted in wall paintings from Ancient Egypt (Keimer, 1924; Sabato et al., 2019). In Biblical times, they were popular in Judea with the name qishu’im (Paris et al., 2008). During the Middle Ages, their presence in Italy is documented through herbal iconography (Paris et al., 2011). Today, cucumber melon cultivation is largely confined to the Salento area of the Apulia Region (Southern Italy), where it is known by the local names of meloncella and spuredda (Pavan et al., 2017; Lija and Beevy, 2021). In fact, the cucumber melon is so popular in Salento that cucumber (Cucumis sativus) is virtually absent from local dietary habits (Laghetti et al., 2008). Recently, the cucumber melon from Salento was recognized as an Italian Traditional Agri-food Product (Didonna et al., 2023). In addition, a population largely cultivated in the municipality of Leverano, within Salento, known as meloncella striata due to its deeply grooved pepos, has been expanding in both cultivation and market presence, reaching large-scale retail distribution. Finally, the cucumber melon has recently gained significant international attention, highlighting its growing appeal beyond its traditional region of cultivation (Roach, 2023).

Snake melon (C. melo var. flexuosus) is another type of vegetable melon, characterized by slender pepos with a width-to-length ratio ranging from 1:3 to 1:9 (Ali-Shtayeh et al., 2017). It is the most frequently mentioned cucurbit in classical literature (Paris et al., 2012), described by the Roman writers Pliny and Columella (Janick et al., 2007) and depicted in mosaics and paintings of the Roman and Byzantine empires (Omari et al., 2018). Nowadays, the snake melon is occasionally cultivated in the Mediterranean region. In Salento, it is typically grown in the municipality of San Donato, where it is known as cucummaru.

To date, limited information is available on genetic and phenotypic variation of vegetable melons. This would be extremely useful to select populations better suited for cultivation and breeding. Additionally, the identification of peculiar genetic features, such as private allelic variants, could be highly valuable for the traceability and protection of specific populations through quality labels.

High-throughput sequencing and genotyping technologies enable the genetic characterization of plant germplasm collections (Pavan et al., 2021; Delvento et al., 2023). However, sequencing several individuals of open-pollinated populations, necessary to obtain accurate estimates of population genetic parameters, is still, in most cases, economically unfeasible (Correa Abondano et al., 2024). Pool-seq, i.e. the sequencing of pooled DNA from several individual, is a cost-effective approach to infer genome-wide population genetic parameters, such as allele frequencies, genetic diversity and selection signatures (Kofler et al., 2011; Schlötterer et al., 2014), and to quantify the extent of genetic differentiation among populations. Pool-seq has been successfully applied to study genetically heterogeneous populations of different crops (e.g. Correa Abondano et al., 2024; Dziurdziak et al., 2021).

Here, we report the establishment of a collection of open-pollinated, farmer-maintained vegetable melon populations cultivated in Salento. Pooled-seq whole-genome resequencing was performed to study inter- and intra-population diversity. Genetic analysis was complemented by replicated field trials, together providing valuable information for the valorization of these crops.

Cucumber melon, once widespread in the Mediterranean Basin, has seen its cultivation to decline drastically, and is now mainly grown in the Salento area of Southern Italy. The germplasm collection presented here is therefore of pivotal importance for preserving the cucumber melon gene pool from further genetic erosion. In addition, it provides a valuable foundation for breeding programs in cucumber melon, which is regaining popularity in both national and international markets, as well as in other C. melo taxonomical groups, including sweet melons.

The pool-seq approach applied in this study has confirmed to be a valuable and cost-effective approach for generating population-level genetic data, despite intrinsic limitations such as lack of individual genotype resolution and potential allele frequency biases. Estimation of heterozygosity indicated a highly variable level of genetic diversity within the populations under study ( Figure 3 ). The populations UBGCMC111 and UBGCMC124, characterized by clearly distinguishable pepos (deeply grooved and pale green, respectively, Figure 1 ), were among those with the lowest heterozygosity, likely reflecting stronger phenotypic selection exerted by farmers.

Analysis of genetic diversity by principal component analysis (PCA) revealed a relatively low amount (about 20%) of total genetic variation explained by the first two components, suggesting that genetic differentiation among the studied populations is driven by many loci with small differences in allele frequency. Nonetheless, the PCA biplot highlighted specific patterns of variation for the two snake melon populations UBGCMC106 and UBGCMC107, consistent with their distinct morphological features.

Notably, hierarchical clustering using Wright’s fixation index (F_ST) revealed no genetic divergence between snake melon and cucumber melon populations, as they clustered together ( Figure 4 ). This result aligns with previous reports of genetic and physico-chemical similarity between the two botanical varieties chate and flexuosus (Ali-Shtayeh et al., 2017; Flores-León et al., 2022; Branca et al., 2023). Given their shared cultivation history, dating back to Ancient Egypt, and their mention as qishu’im in Biblical texts (Janick et al., 2007), our evidence supports the hypothesis that snake melon originated from cucumber melon or vice versa. However, as we only analyzed two snake melon populations, we cannot exclude the possibility that these represent hybrids between the chate and flexuosus groups. The pepos of UBGCMC106 and UBGCMC107 share striking similarity with snake melons traditionally cultivated in Israel and the Palestinian territories, known as Green Baladi or Baladi Akhder (Ali-Shtayeh et al., 2017; Omari et al., 2018).

According to F_ST-based hierarchical clustering, the most divergent population is UBGCMC053, originating from the Municipality of Corigliano d’Otranto ( Figure 2 ). Remarkably, Corigliano d’Otranto locates in a culturally and linguistically unique area in Salento known as Grecìa Salentina, which has deep historical ties to Greek culture, dating back to ancient Magna Graecia and the Byzantine period (Pantaleo, 2006; Guardiano and Stavrou, 2021). It might therefore be speculated that this population derives from a distinct introduction from the Greek Peninsula. UBGCMC111, the second-most divergent population in the panel, might also originate from a different gene pool, as it is characterized by unique pepo morphological features.

The identification of fixed and private DNA variants in most of the populations under study represents a valuable resource for their traceability and valorization. In the population UBGCMC111, currently the most widely cultivated and marketed, one fixed private variant was associated with a 3’-UTR mutation in a gene of the ethylene-insensitive 3 (EIN3) family, involved in ethylene signaling. Another variant was annotated in the downstream region of a gene belonging to the Lateral Organ Boundaries (LOB) family, which plays a major role in melon fruit development and morphogenesis (Tufekci, 2023). In UBGCMC124, fixed private intronic and downstream gene mutations were associated with genes putatively involved in photosynthesis, encoding a chloroplast pheophytinase and a photosystem II D1 precursor processing protein (PSB27-H2), respectively. Interestingly, pheophytinases are involved in chlorophyll breakdown (Schelbert et al., 2009), which could explain the distinctive pale green pepo phenotype observed in this population ( Figure 1 ). Further functional studies, based on targeted mutagenesis or transgenic expression, may prove causal association between mutations in the candidate genes above mentioned and the observed phenotypic variation.

With respect to bio-agronomic traits, significant differences were observed among the populations, in line with the results of the genetic analyses. In addition, year appeared as a significant source of variation, likely due to peculiar climatic conditions ( Supplementary Figures 2 , 3 ). In contrast, the population:year interaction resulted to be non-significant in most cases,5 suggesting that all populations may exhibit a relatively stable response to different environmental conditions. Plant yield was significantly associated with the number of fruits, but not with fruit weight ( Table 3 ). As expected, the population UBGCMC111, currently the most popular among farmers, also exhibited good agronomic performance in terms of yield and earliness. However, several other populations were also ranked among the best performing ( Table 2 ). Overall, the cucumber melon and snake melon populations described in this study are the result of farmer selection under local conditions and are, in most cases, genetically heterogeneous. We therefore suggest that formal breeding programs, leveraging the genetic and phenotypic information presented here, could achieve significant genetic gains in the short term.