Diversity of landraces and wild forms of watermelon (Citrullus lanatus)

Watermelon (Citrullus lanatus) is commonly grown in traditional agrosystems throughout the drought-prone Southern Africa as a staple food (edible seeds), a dessert food (edible flesh), and for animal feed. Several morphotypes of watermelon are found in this area; sweet watermelon, cooking melon and seed melon landraces of the traditional agrosystems; and possibly introgressed types which are regarded as agronomic weeds. There has been little work on investigating the relationships between wild and cultivated forms, and to study amount and partitioning of genetic variation, to allow for better conservation strategies. Previous studies have reported relatively low levels of genetic diversity in cultivated watermelon but these have been based mainly on US plant introductions and modern watermelon cultivars linked to breeding programmes for disease resistance. By contrast, germplasm maintained in the putative centre of origin in southern Africa, can be expected to display considerably higher variability. Three different sampling strategies were used to collect plant material of both wild and cultivated forms of cow-melons (Citrullus lanatus var. citroides) and of sweet watermelons (C. lanatus var. lanatus, only known from cultivation); (1) in-depth sampling in the fields of one village in Zimbabwe, (2) medium-scale sampling across the watermelon growing districts in Zimbabwe, and (3) broad-scale sampling across Southern Africa (Botswana, Namibia, South Africa, Zambia and Zimbabwe). Two molecular marker methods were used, random amplified polymorphic DNA (RAPD) and simple sequence repeats (SSR) also known as microsatellite DNA. Similarity matrices obtained with RAPD and SSR, respectively, have been highly correlated, suggesting that for some applications, the less demanding RAPD can be a useful alternative, especially in developing countries. Considerable amounts of genetic diversity were found at all levels, including within-accessions (half-sib families). Sweet watermelon accessions appear to contain almost as much variability as cow-melon accessions. A genetic structure analysis divided the wild-weed-landrace complex collected in one village into three groups confirming the existence of three major forms with limited admixture. Defining the major forms into landraces and/or folk varieties was considered critical for identification of proper units for both on-farm and ex-situ conservation. Distribution of most watermelon accessions in Zimbabwe was associated with sandy loam and sand soils