Watermelons Have Been A Delicious Treat For 4000 Years

DISCUSSION

Domesticated plants are derived from small samples of wild source populations, and thus are themselves founder populations that contain only a small fraction of the genetic diversity in their wild ancestors (Ladizinsky, 1985). Cultigens have various traits that were selected early and continually in the domestication process, such as lack of bitterness, increased size of the harvested parts, increased yield and novel coloration (Heslop-Harrison and Schwarzacher, 2012). The sweet dessert watermelon, Citrullus lanatus, which has relatively little genetic diversity (Levi et al., 2000, 2001; Dane et al., 2004; Nimmakayala et al., 2014a, b; Reddy et al., 2015), follows this general pattern of crop-plant domestication. The fruit flesh of wild and primitive Citrullus is bitter or insipid, hard and pale-coloured (Wehner, 2008). Non-bitterness of the fruits was probably the first and most important trait to be selected in the process of watermelon domestication and, in effect, success in its selection acted as a preadaptive evolutionary event (Cohen et al., 2014) for subsequent selection for tender and sweet fruit flesh. As non-bitterness is conferred by a single recessive gene (Chambliss et al., 1968; Navot et al., 1990; Wehner, 2007), this trait should have been relatively easy to maintain if isolation from neighbouring wild populations was feasible. The modes of inheritance of hard versus tender and of insipid versus sweet flesh in Citrullus have not been illuminated.

The present investigation, which has focused mainly on ancient literature, has attempted to define a narrower time frame and geographic range within which the dessert watermelons having non-bitter, tender, highly coloured, sweet flesh were developed. The latest possible date for this time frame is provided by illustrations of both red-flesh sweet dessert watermelons and white-flesh citron watermelons in illustrated manuscripts prepared in northern Italy dating to the end of the 14th century (Paris et al., 2009, 2013). Botanists of the European Renaissance, such as Fuchs (1542), Gerard (1597), Dalechamps (1613) and Chabrey (1666), observed various cultivars of watermelons that had red and sweet or white and insipid flesh, and noted that the finest watermelons were found in warm regions. The adaptation of watermelons to warm, sunny climates indicates that much of their history lies to the south of Europe (Paris et al., 2013).

Although it is agreed that the genus Citrullus is of African origin (Whitaker and Davis, 1962; Maynard and Maynard, 2000; Zohary et al., 2012), there has been major disagreement concerning where in Africa the dessert watermelon originated. Much of the controversy stems from phenotypic variations that overlap among citron, egusi and dessert watermelons and colocynths. Wild and primitive Citrullus fruits, regardless of taxon, typically have hard, bitter or bland, weakly coloured flesh and, as indicated by Wehner (2008), this situation has repeatedly been a source of incorrect taxonomic identifications. For example, both the citron watermelon PI 296341 and the egusi watermelon PI 457916 have small, spherical, striped fruits with hard, bitter flesh (Fig. 1). As discussed by Chomicki and Renner (2015), in but few cases have voucher specimens been preserved and thus taxonomic identifications often cannot be verified. Accounts of seed sample origins and germplasm descriptions that could help assign taxonomic identity are sometimes lacking, for example a case involving a cross between a dessert watermelon and a purported colocynth (Shimotsuma, 1958). Further confusion has been generated by the inconsistent use of nomenclature; for example, Pangalo (1930) applied the name colocynthoides to the citron watermelon even though he knew that name had been given a half-century earlier by Schweinfurth (1883) to the wild watermelons of the Nile Valley. Thus, some conclusions drawn from crossing experiments within Citrullus have been based on questionable or incorrect taxonomic identities of the parents. Confused taxonomy, together with the weakness of crossability barriers among species of Citrullus, has resulted in several hypotheses for the origin of the sweet dessert watermelon.

One hypothesis for the origin of the dessert watermelon is that it is descended from the colocynth of northern Africa (Singh, 1978; Sain et al., 2002; McCreight et al., 2013). Citrullus lanatus and C. colocynthis have been observed to cross spontaneously in the field (Fulks et al., 1979). However, the hybrid plants exhibit chromosome irregularities and lower fertility (Whitaker and Davis, 1962; Shimotsuma, 1963; Sain et al., 2002; Wehner, 2008). Results of investigations using techniques of molecular genetics and genomics indicate that colocynths are more distant from dessert watermelons than are citron watermelons (Jarret and Newman, 2000; Levi et al., 2000, 2001, 2013; Dane et al., 2004; Dane and Lang, 2004; Dane and Liu, 2007; Chomicki and Renner, 2015).

A second, widely accepted hypothesis for the origin of the dessert watermelon is that it is descended from the citron watermelon of southern Africa (Robinson and Decker-Walters, 1997; Maynard and Maynard, 2000; Rubatsky, 2001). However, the findings of 4000-year-old artefacts of watermelon in Egypt are contradictory, as farming was not yet under way in southern Africa at that time (Zohary et al., 2012; Mead and Martens, 2013). Massive preferential segregation (Levi et al., 2003) and reduced pollen fertility (McGregor and Waters, 2013) occur in progenies of crosses between citron and dessert watermelons. Moreover, there are major differences in genome organization between the two (Guo et al., 2013; Reddy et al., 2013; Chomicki and Renner, 2015).

A third, newer hypothesis is that ‘the recent progenitor of modern cultivated watermelon’ is the egusi watermelon of western Africa (Guo et al., 2013). Egusi and dessert watermelons are sister species that diverged only 3.1 million years ago (Chomicki and Renner, 2015). However, the egusi watermelons have peculiar soft, thick, moist, mucilaginous seed coats not reported in any other Citrullus. The wild-type PI 457916 has this trait (Fig. 1) and egusis were domesticated for the consumption of their seeds, not their flesh. Besides the unidirectional crossing ability between dessert and egusi watermelons (Fursa, 1983; Gusmini et al., 2004), egusi watermelons have high nucleotide divergence of reproductive barrier genes from the dessert watermelons, for which has been offered the tenuous explanation that ‘the domestication of watermelon could be a possible force responsible for the rapid evolution of reproductive barriers’ (Guo et al., 2013).

A fourth hypothesis is that the dessert watermelon originated and was first domesticated in northeastern Africa (Fig. 3). Accordingly, the wild, often hard, bitter or insipid watermelons given the botanical variety name colocynthoides by Schweinfurth (1883) and subspecies cordophanus by Ter-Avanesyn (1966) (Table 1) are living representatives of the ancestor of the sweet dessert watermelon (Fursa and Gavrilyuk, 1990). Wild watermelons are widespread in Sudan (Andrews, 1950) and occur in areas of savannah and desert (Bebawi and Neugebohrn, 1991; Hassan et al., 2009), centred in the Nile Valley (Schweinfurth, 1873, 1883; Gouda, 2007; Mariod et al., 2009), to the west in Kordofan (Ter-Avanesyn, 1966) and Darfur (Robinson, 2005) and to the east in Shaqadud (Abdel-Magid, 1989). Their distribution extends north into Egypt (Muschler, 1912; Shimotsuma, 1963), and perhaps south to Kenya (Jeffrey, 1967, 2001) and east to Ethiopia (Fig. 1). The largest extant population of wild dessert watermelons may be one that has been reported adjacent to the Nile River in Sudan (Mariod et al., 2009). The wild population at Shaqadud, in the eastern desert some 140 km northeast of Khartoum, is exploited by local residents for water and sustenance (Abdel-Magid, 1989). Landraces in Darfur are used as a source of water (Keith-Roach, 1924). Wild watermelons are known in Sudan as gurum (Ziyada and Elhussien, 2008; Mariod et al., 2009) and in central and southern Egypt as gurma (Schweinfurth, 1883; Issa Bey, 1930; Manniche, 1989). Unfortunately, Citrullus samples derived from Sudan and Egypt have not been subjected to extensive molecular-genetic scrutiny. Results from DNA-sequence-based, systematic investigations of Citrullus have been presented for only a few accessions from Egypt and none from Sudan (Levi et al., 2013; Nimmakayala et al., 2014b). Two of the Egyptian accessions, listed as US plant introductions PI 525081 and PI 525083, which had been previously thought, based on phenotype, to be a colocynth and a citron, respectively, were observed by Levi et al. (2013) to cluster with dessert watermelons. Next-generation sequencing of ancient DNA (Brown et al., 2014), if applied to comparing watermelon artefacts with modern Citrullus, can be expected to identify the artefacts taxonomically. The latest sequencing and genomics technologies (Guo et al., 2013; Levi et al., 2013; Nimmakayala et al., 2014a, b; Chomicki and Renner, 2015) also should be directed to comparing the wild and primitive watermelons from northeastern Africa with modern, sweet dessert watermelons as well as other Citrullus taxa. If applied to such germplasm, the results can be expected to determine whether the dessert watermelon, Citrullus lanatus, indeed exhibits a clear wild/domesticated dimorphism or merely a continuum from wild to domesticated gene pools reflective of continual selection for improved horticultural traits (Abbo et al., 2014).

Archaeological artefacts, mostly seeds, and images found in tombs of ancient Egyptian nobility provide important evidence for the presence of Citrullus in northeastern Africa >4000 years ago (Schweinfurth, 1883; Keimer, 1924; Darby et al., 1977; Germer, 1985; de Vartavan and Asensi Amoros, 1997; Zohary et al., 2012). However, the Arabist Watson (1983) has questioned whether the various findings are of watermelon, C. lanatus, or colocynth, C. colocynthis. Both C. lanatus var. colocynthoides and C. colocynthis have been recorded as wild plants in Egypt (Muschler, 1912) and these two taxa are easily confused. Besides taking no account of the Hebrew-language evidence, Watson dismissed all of the archaeological, iconographic and other literary evidence from pre-Islamic times as indicating colocynth or melon rather than watermelon, and promulgated that sweet watermelons originated in the Indian subcontinent and diffused westward as a result of Islamic conquests. There can be no mistaking, though, that seeds >10 mm long, with bulges adjacent to the hilum and having a rough surface, are not of colocynths. The photographs and drawings of seeds presented by de Vartavan (1999) and Wasylikowa and van der Veen (2004) prove that watermelons were present in Egypt and Libya thousands of years ago. Seeds of Citrullus that date from the Roman period at Qusayr al-Qadim in Egypt (Cox and van der Veen, 2008; van der Veen, 2011) and, several hundred kilometres to the north, near the Dead Sea in Israel (Kislev and Simhoni, 2009), shown in clear photographs, likewise cannot be of colocynths. Hence, there can be no remaining doubt that watermelons were present in Egypt and its immediate neighbours for centuries, even millennia, prior to the rise of Islam.

Although the watermelon must have been a familiar item valued by the Egyptians of pharaonic times, from the remains alone it is not clear whether the fruits were gathered from wild or cultivated plants. However, the image of a large, oblong, striped fruit in an Egyptian tomb, >4000 years old (Manniche, 1989, p. 92), is suggestive of a cultivated watermelon. The wild and primitive watermelons of northeastern Africa are typically spherical (Schweinfurth, 1883; Ter-Avanesyn, 1966) (Fig. 1), and therefore this oblong-shaped specimen obviously deviates from the wild type. This fruit appears on a tray served, undoubtedly, to a royal receiver, surely indicating the esteem given to this item and that it was to be eaten fresh (Chomicki and Renner, 2015). Indeed, this fruit could not have been a citron watermelon because citron fruit flesh is too hard to be enjoyed fresh in large quantity. Although melons, Cucumis melo, are large fruits and can be striped, the striping of mature, ripe melons is rarely as stark as that shown in the image. As this large, striped fruit was enjoyed fresh, it must have been non-bitter and tender-fleshed, and its oblong shape indicates that it was taken from a cultivated plant. However, this fruit was not necessarily sweet, at least not according to modern standards (Mohr, 1986; Maynard, 2001). The longing for avattihim by the Children of Israel (Numbers 11:5) places them, at the time of the Exodus, in the same category as the qishu’im (snake melons) and various Alliaceae. Like the snake melons, leeks, onions and garlics, the avattihim of that early time were probably eaten fresh, pickled or cooked, not better tasting than snake melons and certainly not sweet like modern dessert watermelons.

Greek, Latin and Hebrew writers of the first centuries ce clearly distinguished watermelons, melons and colocynths. Physicians writing in Greek, from Hippocrates (∼400 bce) (Jones, 1967) to Dioscorides (first century) (Osbaldeston and Wood, 2000; Beck, 2005), Galen (second century) (Grant, 2000; Powell and Wilkins, 2003) and Oribasius (fourth century) (Grant, 1997), indicated that the pepon was wet, abundantly cooling, easily digested and diuretic. The pepon must have been large enough to be placed on the head of a child and quite wet and cooling so as to reduce fever, as instructed by Dioscorides. To Pliny, the pepo was a food that was most cooling (Rackham, 1950; Jones, 1951). The Jerusalem Talmud (Kil’ayim 1:8, p. 4a) indicates that the avattiah was a watery fruit. The pepon of the Greek physicians, the pepo of the Latin encyclopaedist Pliny and the avattiah of the Hebrew scholars were one and the same large, watery fruit – the watermelon. The melopepon, the supposed hybrid of an apple with a pepon, had a less moisturizing and cooling effect, according to Galen and Oribasius. The melopepo may indeed have been a new introduction, as stated by Pliny (first century), because it is not mentioned by his contemporary, Dioscorides. According to Pliny, the melopepo became yellow and aromatic and spontaneously detached from the plant when it ripened. Thus the melopepo is without question the melon, Cucumis melo, an edible cucurbit that typically yellows and becomes aromatic, and is the only one that abscises from the plant upon ripening (Whitaker and Davis, 1962; Robinson and Decker-Walters, 1997). The same taxonomic identity of the Hebrew rendition, melafefon, is established by the identical mythical perception of its being an apple-pepon admixture (Jerusalem Talmud, Kil’ayim 1:2, p. 2a). The kolokyntha agria or colocynthis, colocynth, described by the Greek physicians and by Pliny, was a familiar, highly regarded commodity for medicinal preparations and as such could not possibly be undistinguished from the watermelon (Amar and Lev, 2011). To some of the Greek writers quoted by Athenaeus (Olson, 2006), the edible, cultivated kolokunte were large and could be sweet, and thus distinguished from the inedible, wild kolokyntha agria, the colocynth. Indeed, the extremely bitter paqqu‘ot (2 Kings 4:39–40), colocynths, were already clearly distinguished in biblical times from the cultivated, non-sweet but pleasant-tasting avattihim (Feliks, 1968; Janick et al., 2007). The paqqu‘ot, like the yeroqat hamor (squirting cucumber, Ecballium elaterium), are absent from the tractate on tithing, Ma‘asrot, and thus they were not food items, although they were used for other purposes (Feliks, 1968; Janick et al., 2007). Moreover, the Latin recipe book of Apicius distinguished between two kinds of watermelons, using the word pepones for fruits that were dressed and eaten fresh and the word citrium for a fruit that was cooked (Flower and Rosenbaum, 1958).

The Hebrew-language literature provides considerably more information concerning the cucurbit food crops than does the contemporary and near-contemporary Greek and Latin literature. Four cucurbits, the qishu’im, delu‘im, avattihim and melafefonot, are discussed together in some passages. Each of these four must have been widely cultivated, often in the same field, because there are prohibitions on planting them too close to one another (Mishna, Kil’ayim 1:5; Tosefta, Kil’ayim 1:4). Likewise, each must have been grown for the consumption of its fruits because the tractate on tithing contains a number of pertinent instructions concerning each of them. Prior to eating, the qishu’im (snake melons) and the delu‘im (bottle gourds) were to undergo piqqus, rubbing off of the hairs, and thus these fruits were consumed when young (Feliks, 1968; Janick et al., 2007). In contrast, the avattihim and the melafefonot were to undergo shilluq, smoothing of the fruit surface to remove accumulated dust (Qafah, 1963), and thus these fruits were to be eaten mature, upon ripening (Mishna, Tosefta and Jerusalem Talmud, Ma‘asrot 1:4–6). The use of the avattihim when they were ripe, like pomegranates, is confirmed in the tractate on fruit stems (Mishna, ‘Oqazin 2:3).

The descriptions of the pepon/pepo/avattiah as being watery with a thick rind and the melopepon/melopepo/melafefon as becoming yellow and aromatic and spontaneously detaching from the plant when ripe, are sufficient to identify them as watermelons and melons, respectively. Other traits mentioned in the Hebrew literature reinforce these identifications. The melafefonot, but not the avattihim, ripened in two main waves over the course of the growing season (Tosefta, Ma‘asrot 1:6). Ripening in waves is a well-known characteristic of melon crops (Rosa, 1924; McGlasson and Pratt, 1963; Pratt et al., 1977). For tithing, the avattihim were to be laid out individually rather than stacked (Mishna, Ma‘asrot 1:5). Indeed, the watermelons of yesteryear had fragile rinds and thus were highly subject to splitting. They were replaced during the mid-20th century by newly bred cultivars having tough-rinded fruits (Parris, 1949; Ivanoff, 1954). After harvest, the avattihim were fit for contribution for an indefinite period, but the melafefonot could be contributed only during the first 3 d after harvest (Tosefta, Terumot 4:5). Whilst dessert watermelons have a shelf-life of several weeks or months (Keith-Roach, 1924; Rushing et al., 2001), muskmelons and cantaloupes, if not refrigerated, have a shelf-life of only 3–4 d (Burger et al., 2010).

In the tractate on tithing, watermelons are discussed with figs, table grapes and pomegranates (Mishna, Ma‘asrot 2:6, 3:9; Jerusalem Talmud, Ma‘asrot 2:4, 3:4), fruits which are juicy and sweet, and therefore the watermelons of the time must have been likewise. Although the watermelons and melons are usually mentioned together in the Jewish literature, only the watermelons are mentioned with the other three sweet fruits. Sweet melons are thought to have originated in Khorasan, Central Asia, reaching the Mediterranean Basin much later, in the 11th century, and these are the battikh that diffused westward after the rise of Islam (Paris et al., 2012b).

Diocles was quoted by Athenaeus (second century) as saying that the best kolokyntas are glukeian, sweet (Olson, 2006). The early sixth-century De Observantia Ciborum lists the pepone among pomegranates, grapes and figs, confirming the presence of sweet watermelons in what is now Italy (Mazzini, 1984). The early sixth-century writing of Anthimus (Grant, 2007) indicates that the melones were eaten when they were ripe and freshly cut, the seeds still mixed in the flesh. Anthimus, evidently, used the word melones for watermelons, the same word used for them in the 14th- and 15th-century Tacuinum Sanitatis (Paris et al., 2009). Hence, by the second decade of the sixth century, at the very latest, the eating of sweet, ripe watermelons had spread beyond Israel and Greece to Italy and possibly France (Fig. 3). Sweet watermelons had diffused to Europe centuries before their earliest recorded presence in Moorish Spain, in the Cordoban Calendar of 961 ce (Pellat, 1961).

Athenaeus (late second century) quoted Phaenias as stating that the pepon is edible, except for the seeds, when the flesh becomes soft (Olson, 2006) and Quintus Gargilius Martilus (mid-third century) wrote that the pepone are good to eat after the rind and seeds have been removed (Maire, 2007). The Jerusalem Talmud (Ma‘asrot 1:2, 2b) instructs that the seeds of the melafefonot are tithed and those of the avattiah are not, because the former were eaten and the latter were not (Feliks, 2005). Thus, in Roman times around the Mediterranean, the tender, sweet, juicy ripe fruit flesh of the watermelons was consumed and the seeds were not. In Byzantine-era mosaics (350–600 ce) in Israel, watermelons of a variety of sizes and rind colour patterns are depicted. One mosaic has what appears to be a cut fruit with yellow-orange flesh and white seeds (Avital and Paris, 2014).

Watson (1983) cited the medieval Arabic names battikh hindi and battikh sindi as evidence for the derivation of the sweet dessert watermelon in the Indian subcontinent. The former name was first recorded in the Cordoban Calendar of 961 ce (Pellat, 1961) and the latter as the source of the modern Spanish word for watermelon, sandia, Sind being a province in Pakistan. Although Citrullus lanatus is native to Africa, it is conceivable that a cultivar of sweet watermelon could have emerged elsewhere, isolated from local wild or cultivated plants having inferior horticultural traits. Indeed, contamination of an elite watermelon cultivar was described by the late 15th-century writer Ibn Shihna (Sarkis, 1909). Seeds of an excellent watermelon, ‘Raqqi’ (from the environs of Raqqa, adjacent to the Euphrates River in Syria), were imported annually by Gazans because seeds taken from ‘Raqqi’ fruits grown in Gaza and planted the following year produced inferior fruits. Undoubtedly, growing in or near Gaza were local cultivars or wild populations of insipid or bitter Citrullus; some such populations still exist (Shemida‘ and Danin, 1983). To the present, watermelons in Iraq are called raqqi (Chakravarty, 1966; Perry, 2005; Nasrallah, 2007) and therefore it is possible that, similarly, the names battikh hindi and battikh sindi reflected the introduction of a superior cultivar from the Indian sub-continent or, at least, from a land to the east. Such a watermelon was grown in Khwarizm, near the Aral Sea in Central Asia, and its fruits were carefully packed and shipped to the caliphs Wathiq and Ma’mun of Baghdad during the first half of the ninth century (Bosworth, 1968; Said, 1973). On the other hand, there is also much precedent for the use of exotic names for cucurbits as a marketing ploy. The Cucurbita moschata Duchesne ‘Tahitian’ squash has an exotic name, but it is not proof of its geographic origin (Robinson, 1980). Indeed, the familiar dessert watermelon ‘Congo’ was bred in the USA (Levi et al., 2004). The Indian or Sindian battikh would have sounded exotic in medieval Andalusia, as Spain was at the far western end and Pakistan was at the far eastern end of the Islamic Empire. Significantly, battikh hindi and battikh sindi were given as synonyms of dulla‘ in the medieval lexicon of northern Africa and Andalusia (Meyerhof and Sobhy, 1938; Dozy and Engelmann, 1915; Pellat, 1961; Amar, 2000). Dulla‘ is a Berber word for sweet dessert watermelon which appears to share the same Semitic root, d-l-‘, as the Hebrew word for bottle gourd, dela‘at. The existence of the Berber word indicates an established regional familiarity with the sweet watermelon. Indeed, Ysaac Judaeus, a physician of Qayrawan, Tunisia, in his Book of Particulars in Diet (∼ 920) (Sabbah, 1992), mentioned the dulla‘ four decades prior to the first recording of the battikh hindi and battikh sindi in Spain (Pellat, 1961). Also, it is tempting to parallel the juggling of bottle gourd with sweet watermelon between Hebrew and Berber with that of kolokynte and pepon by various Greek writers, as related by Atheneaus (Olson, 2006). Whitaker and Davis (1962) believed, based on historical records and indigenous names for watermelon of Egyptians and Berbers, that the sweet dessert watermelon has its origins in Africa.

The combination of literary evidence, especially that in Hebrew, with the archaeological and iconographic evidence from Egypt and Sudan, and the modern distribution of wild and primitive watermelons in Sudan and neighbouring countries, is overwhelming. Dessert watermelons are native to northeastern Africa and sweet dessert watermelons were selected in Mediterranean lands by no later than the second century ce.