Genetic diversity and sequence variation
Within the genus Citrullus genetic diversity analyses have been investigated since the second middle of the 20thcentury (Hashizume et al. 1996) revealing various trends. Previous knowledge revealed lower genetic diversity inCitrullus for breeding purpose (Levi et al., 2001; Levi et al., 2004). Recent studies shed light on obvious genetic diversity within the genus. For instance, a study using High Frequency Oligonucleotide Target Active Genes (HFO-TAGs) revealed high genetic diversity among Citrullus spp. and highlighted the potential importance of PI accessions as sources of valuable traits like disease resistance (Levi et al., 2013).
Our findings revealed low cpDNA variability among C. lanatus andC. mucosospermus . This was also observed by Dane and Lang (2004) and Dane et al. (2004) who revealed low nucleotide variability based on a low number of parsimony-informative sites within each of the studied species. Most haplotypes were found within non-cultivated (C. colocynthis ) rather than cultivated (C. lanatus and C. mucosospermus ) species. Taxa were highly separated from one another with divergence based mainly on indels and transition events (Dane et al., 2004). However, there was sufficient resolution of the trn T-L and ndh F-rpl 32 non-coding regions to reveal intraspecific variability.
Chloroplast sequence analysis revealed that the ndh F-rpl 32 region exhibits comparatively higher variability within the two cultivated species than the trn T-L region. Dane and Lang (2004) analyzed four chloroplast regions (nhd F, ycf 6-psb M,ycf 9-trn G, and atp A-trn R) and found no variability within cultivated accessions, grouped either by morphological traits or geographical origin. In this study, we used a large number of C. lanatus accessions from a wide geographical range and observed low haplotype diversity within that species, as also revealed by Guo et al. (2013). While many factors can influence sequence diversity, selection is a major contributor via the imposition of bottlenecks that can substantially reduce diversity (Dane & Lang, 2004; Levi et al., 2013). The lack of haplotype divergence within C. lanatus and C. mucosospermus is likely the result of selection or other bottlenecks in the domestication histories of watermelon and egusi melon. Certainly, selection for sweet red-fleshed cultivars with high lycopene content or selection of seed type as source of protein/oil for consumption might contribute to current genetic structure in those cultivated species (Achigan-Dako et al., 2015; Renner et al., 2019).
C. colocynthis exhibited a relatively high number of parsimony-informative characters. Dane et al. (2004) revealed that haplotypes detected within C. colocynthis were associated with geographical origin and that was also confirmed by Levi et al. (2017). The haplotype diversity within C. colocynthis suggests cryptic evolution and calls for a comprehensive morphological comparison of Asian and African colocynths. Such an investigation is exemplified by the recent studies on Cucumis melo that revealed modern melon cultivars go back to two lineages and was domesticated at least twice: in Asia and in Africa (Endl et al., 2018).
Citrullus haplotype evolution
Thirty-eight haplotypes were detected among the cultivated and wildCitrullus accessions used in this study. Dane et al. (2004), found seven haplotypes within the genus, using 55 accessions of C. lanatus,15 accessions of C. colocynthis , and a total of seven cpDNA regions (HinfI, RsaI, TaqI, AluI, HaeIII, MboI, and BglII). With two cpDNA regions and 135 accessions carefully selected to represent a wide geographical region, we detected an even higher haplotype diversity among Citrullus spp. This situation can be expected to continue to evolve as more watermelon accessions from Sudan or northeast Africa are sequenced, particularly, the Sudanese sweet white-fleshed melon. Unfortunately, sampling of C. lanatus from the Darfur region of Sudan has been scarce (Renner et al., 2019).
On average, we observed 9.5 haplotypes per species, varying from 5 to 16. In comparison with other species, Guicking et al. (2011) found 9.8 haplotypes per species in Macaranga and Jakob and Blattner (2006) found 2.83 haplotypes per species in Hordeum . In Citrullus spp., nucleotide substitutions appear to have evolved at different rates, an observation supported by the Fisher’s test for homogeneity of nucleotide substitution. Fu’s test Fs also rejected the null hypothesis of neutrality of evolution of nucleotide substitution, further supporting the hypothesis that the polymorphism pattern observed is non-random. Population expansions tend to produce significantly negative values of D, while population bottlenecks tend to produce significantly positive values of D. In our case the departure from neutrality might indicate that there is a high demographic expansion and a pattern of isolation by distance would be occurred between the continents (Jiang et al., 2016).