A review of the current knowledge on Zeugodacus cucurbitae (Coquillett) (Diptera, Tephritidae) in Africa, with a list of species included in Zeugodacus

Abstract This paper reviews all available information regarding the occurrence and biology of the melon fly, Zeugodacus cucurbitae (Coquillett), in the Afrotropical Region, including data on invasion history, distribution patterns, population genetics, host range, and interspecific competition. Although limited intraspecific variability has been observed within the region regarding the above mentioned aspects, there seems to be no indication that Zeugodacus cucurbitae represents a species complex. A checklist of all of the species included in Zeugodacus as recently proposed by Virgilio et al. (2015) is provided.


Introduction
The melon fly, Zeugodacus cucurbitae (Coquillett) is a major agricultural pest of Asian origin. Despite the vernacular English name and the species-group name, it is reported from a series of unrelated host families in addition to the vast host range within Cucurbitaceae (White and Elson-Harris 1994). The fact that a number of populations of Z. cucurbitae differ in their reported host plants, morphology, etc. from region to region, resulted in the species being included in the Coordinated Research Project on "Resolution of cryptic species complexes of tephritid pests to overcome constraints to SIT application and international trade", initiated by the Joint FAO/IAEA Programme in 2010. This paper reviews the taxonomic position and history of the species within the Tephritidae, provides information on its worldwide distribution and genetic diversity, summarizes the current knowledge regarding the species in Africa, and provides a checklist of all of the species included in Zeugodacus as recently proposed by Virgilio et al. (2015).

Classification and taxonomic history
Zeugodacus cucurbitae (Figure 1) was originally described as Dacus cucurbitae by Coquillett (1899) from two males and two females bred from larvae found in green cucumbers in Honolulu, Hawaii (USA). Bactrocera was considered a subgenus of Dacus until Drew (1989) proposed a classification recognizing both taxa as genera, based upon the abdominal tergites being fused, (in Dacus), or not (in Bactrocera). Drew placed Z. cucurbitae in the subgenus Zeugodacus, first under Dacus following previous authors (Drew 1973), and later under Bactrocera (Drew 1989). The subgenus Zeugodacus belongs to a group of subgenera, characterized by the posterior lobe of the male lateral surstylus being long and the male abdominal sternite 5 being slightly concave along the posterior margin (rather than having a deep V shaped indentation) (Drew and Hancock 1999). At least 50% of the species included in the Zeugodacus group, for which host plant records are available, are cucurbit feeders. Recently the systematic position of Zeugodacus was revised as Bactrocera, Dacus and the subgenera of the Zeugodacus group have different evolutionary histories (Krosch et al. 2012, Virgilio et al. 2015. The molecular data provided support the hypothesis of White (2006) who suggested a common ancestry for Zeugodacus and Dacus (but see Hancock and Drew 2015 for a different hypothesis). Here we refer to the classification proposed by Virgilio et al. (2015) by using the new generic combination Zeugodacus (Zeugodacus) cucurbitae for the melon fly, although most existing literature refer to it under the former combination, Bactrocera (Zeugodacus) cucurbitae.
The genus Zeugodacus currently includes 192 species (see list in Supplementary material 1). Most species within this genus are restricted to the Oriental and Australasian Regions, with a few species reaching into the eastern Palearctic in China and Japan, except for Z. cucurbitae which was introduced into other parts of the world.
Zeugodacus cucurbitae is rather distinctive in adult morphology and can be differentiated from other related species by the following combination of characters: scutum red-brown, with medial and lateral yellow postsutural vittae; large apical spot on the wing with posterior margin reaching about halfway between vein R 4+5 and vein M; infuscation present over crossvein dm-cu and usually also crossvein r-m, wing cells bc and c hyaline, abdomen with a narrow transverse black band across basal margin of tergite 3 and a medial longitudinal black stripe over tergites 3-5 (White 2006, Drew andRomig 2013).
Contrary to other species like the Bactrocera dorsalis (Hendel) populations found in Africa (see Drew et al. 2005, White 2006, there is little intraspecific variability observed in adult Z. cucurbitae specimens with regard to scutal and abdominal patterns. Drew and Romig (2013) only mention that the fuscous marking on the scutum can be absent or present. White (2006) indicates that the anterior supra-alar and prescutellar acrostical setae can be rarely absent (the latter being one of the main differentiating characters between Dacus and most Bactrocera species), while the basal scutellar seta can be rarely present (hence, four setae in total rather than the usual two which are situated apically on the scutellum). The crossband on r-m is not always distinct. However, these differences do not seem to reflect any particular pattern linked to cryptic speciation but rather phenotypic plasticity. Zeugodacus cucurbitae was not included in the list of the Asian species complexes defined by Drew and Romig (2013). No key is available to differentiate it from all other Zeugodacus species. Drew (1989) provides a general key for Bactrocera of the Australasian and Oceanian regions, including Z. cucurbitae and 19 other Zeugodacus species, while Drew and Romig (2013) provide descriptions and some diagnostic features for 101 species from South-East Asia, but no key. White (2006) and Virgilio et al. (2014) provide a key for African Dacina including Z. cucurbitae.
DNA barcoding shows remarkably low intraspecific variability. A pilot study including COI barcodes of 44 specimens originating from 11 countries along the entire distribution range (Virgilio and De Meyer, unpublished data) revealed an average K2P genetic distance (Kimura 1980) of only 0.02% (Figure 2). Similarly, the concatenation of mitochondrial DNA sequences (COI and ND6 gene fragments) from 100 specimens from Asia, Hawaii, African continent and islands of the Indian Ocean resulted in 22 haplotypes with 21 polymorphic sites and an average p-distance of only 0.003% (Jacquard et al. 2013). Minimum Spanning Network indicated the occurrence of two main haplotype groups corresponding to specimens from (a) Asia and Hawaii, and (b) the African continent including also Reunion Island.

Distribution, origin and population structure
Although Z. cucurbitae was originally described from the Hawaiian Islands, its presence there was the result of accidental human-mediated introduction (Bess 1961). About a decade later the first record from India was published (Froggatt 1909). Since then, it has been reported from multiple countries in the Asian and Australian-Oceanian Regions (Dhillon et al. 2005, Drew 1982, Drew and Romig 2013. It is abundant throughout Central and East Asia (including Pakistan, India, Bangladesh, Nepal, China, Indonesia and the Philippines) and Oceania (including New Guinea and the Mariana Islands). In some of these regions, it has been the subject of a number of introductions, eradication attempts and subsequent re-introductions. This is in particular the case in parts of the Pacific like the Northern Mariana Islands and Nauru (Dhillon et al. 2005), although it has also been successfully eradicated (Suckling et al. 2014) from regions in which it was well established, such as southern Japan in the 1990ies, using Sterile Insect Technique (Koyama et al. 2004). Since 1956 Z. cucurbitae has been detected a number of times in California (Papadopoulos et al. 2013), but its permanent establishment on the North American mainland is not confirmed.
In Africa, the first record dates back to 1936 from Tanzania (based upon a male specimen in the collection of the Natural History Museum in London, collected at Tanga on January 10 th , 1936 by N. Krauss. See Bianchi and Krauss (1937) for report on this expedition, although this record is not specifically mentioned). No other species that are closely related to Z. cucurbitae are found in Africa, and its occurrence on the continent is also attributed to introduction. However, it is unclear whether it was introduced at that time (1936) or whether it was already present for a much longer time. There are historical ties between the eastern coastal area of Africa (dominated by the so-called Swahili culture) and the near East and Indian subcontinent that date back to 100 AD (Gilbert 2004), with movements and shipments of commodities between both regions. The first records from the African mainland were restricted to coastal Tanzania and Kenya (first record 1937) (Table 1).
Zeugodacus cucurbitae has also been introduced to several islands in the western Indian Ocean, with the first record in Mauritius in 1942 (Orian and Moutia 1960) and in La Réunion in 1972 (Vayssières 1999, White et al. 2001). More recently (since 1999) it was reported from the island Mahé of the Seychelles (White et al. 2001), where it is now also considered established. Its presence on the Comoro Archipelago is questionable (De Meyer et al. 2012) and so far no records are reported from Madagascar. Despite its longtime occurrence in eastern Africa and the Indian Ocean, Z. cucurbitae apparently did not spread rapidly to other parts of Africa. The first record from Central Africa was a mention in Fontem et al. (1999), where it is reported (as Table 1. First records of Zeugodacus cucurbitae in African countries (based upon records in Orian and Moutia 1960, Vayssières et al. 2007and De Meyer and White 2007 Dacus cucurbitae) as the most prevalent insect pest observed by farmers on tomatoes in Cameroon. No voucher specimens could be traced to any collections in order to confirm this record, and there is the possibility that it was based on a misidentification of another dacine attacking tomatoes. For example, Dacus punctatifrons Karsch has been reported as a major pest of tomato in Cameroon (Okolle and Ntonifor 2005). The first voucher specimens from West Africa that could be confirmed to belong to Z. cucurbitae are from Ivory Coast and the Gambia and were collected in 1999 at Korhogo and Brikama, respectively, while in 2000 one of the authors (JFV) discovered it in Mali in cuelure traps and emerging from young pumpkins. Since the beginning of the 21 st Century, several records of Z. cucurbitae from West and Central Africa became known (Table 1) and it is now well established in most parts of the region (Vayssières et al. 2007; Figure 3a). In eastern Africa, Z. cucurbitae has been reported from a much larger range than just Kenya and Tanzania and it is now found from Ethiopia and the Sudan to Malawi and northern Mozambique (Table 1). It is unclear whether these 21 st century records are a true reflection of a further recent expansion of its geographical range or that they are due to incomplete sampling in preceding decades. However, the currently observed dispersal of this species has also increased the awareness of its economic significance. Zeugodacus cucurbitae has been considered a major pest species of commercially grown cucurbits in large parts of Asia (Kapoor 1989, Koyama 1989 and Hawaii (Harris 1989) for a long time. However, in the Afrotropical region, not much research was devoted to this species in comparison to other cucurbit infesting dacines, except for La Réunion (White and Elson-Harris 1994, Vayssières 1999, Ryckewaert et al. 2010 and Mauritius (Sookar et al. 2012(Sookar et al. , 2013. This is currently changing due to the recent observations on its distribution and dominance in particular crops (see below under 'host range and interspecific competition').
Given the current geographic distribution of other Zeugodacus species (all restricted to the Oriental, Australasian and eastern Palearctic Regions) and the historical data of its occurrences in Africa and Hawaii, it is generally assumed that Z. cucurbitae originated in the Oriental Region and that its current distribution in Africa and in other parts of the world is the result of several invasion events (see Virgilio et al. 2010). The analyses by Jacquard et al. (2013) of sequences obtained from samples from throughout the known distribution range of Z. cucurbitae revealed a main genetic split between samples from (a) Asia and Hawaii, and (b) Sub Saharan Africa and La Réunion Island. The main differences between the African and all other samples suggested a bottleneck(s) following introduction, yet this model was not supported by the studies of Virgilio et al. (2010). Relationships between populations from different geographic areas were further resolved through a macrogeographic population structure analysis based on 25 populations genotyped at 12 microsatellite loci (Virgilio et al. 2010). Populations could be subdivided into five main geographic groups (African continent, Western Indian Ocean islands, Indian Subcontinent, South-East Asia, and Hawaii; Fig 4).
Levels of genetic diversity and individual Bayesian assignments (Virgilio et al. 2010) seem to suggest that Z. cucurbitae originated on the Indian Subcontinent and might have expanded its range to South-East Asia and Hawaii on one hand and to Africa and the Indian Ocean islands on the other (although recent anthropogenic transport might have contributed to inter-regional gene flow). Sookar et al. (2013) looked at the mating compatibility between populations of Mauritius, the Seychelles and Hawaii but only found random, non-assortative mating between the populations. Within La Réunion, Jacquard et al. (2013) also described the occurrence of local genetic clusters with distinct distributions across the eastern and western coast of the island. These clusters have possible African origin and are interconnected by high levels of gene flow both within La Réunion and between La Réunion and the African mainland. Dhillon et al. (2005) list 81 plant species, including several non-cucurbits, as possible hosts for Z. cucurbitae. However, several of these hosts are considered doubtful because they were either based on casual observations (White and Elson-Harris 1994) or they are a result of induced oviposition under laboratory conditions. The latter approach provides unreliable data regarding the true natural host range of any fruit fly and should be considered with caution when determining host status (Aluja and Mangan 2008). De Meyer et al. (2007) list 45 plant species, belonging to 9 different families, that are considered hosts of Z. cucurbitae in Africa (including Indian Ocean islands) ( Table 2). The majority of these records are based on rearing of infested fruits collected in the wild. Twenty-nine of them are Cucurbitaceae. Cucumis spp. (in particular cucumber (C. sativus L.) and melon (C. melo L.)) and Momordica spp. (in particular M. cf trifoliata Hook. f. and bitter gourd (M. charantia L.)) were the preferential hosts both in West and East African studies (western Africa: Vayssières et al. 2007;Tanzania: Mwatawala et al. 2010). These studies have shown that in general cucurbit hosts are preferred over non-cucurbit hosts, with very low infestation rates and incidences in the latter. However, Vayssières et al. (2007) indicated that there are geographical differences with Z. cucurbitae being more oligophagous on La Réunion Island (with no genetic differences between flies infesting wild and cultivated hosts, see Jacquard et al. 2013), while having a broader host range in western Africa. Also, infestations rates can differ according to the region. For example Cucumis melo yielded 26-50 specimens/kg of fruits in West Africa, compared to 51-75 in Réunion, and more than 100 in Tanzania. Lagenaria siceraria (Molina) Standl. yielded very low numbers in West and East Africa but more than 100 specimens/kg in Réunion. These examples are, however, based on too limited a number of samples to draw definite conclusions, and it is not clear what are all of contributing causes of these differences in infestation rates. Seasonal differences (Mwatawala et al. 2009), weather variability, host availability, and interspecific competition could also be factors (Mwatawala et al. 2009, Vayssières et al. 2008). Although the low preference for non-cucurbit hosts has limited impact on actual crop loss, the mere presence in commercial hosts, such as mango (Mangifera indica L.),  (Figure 3b-f) and there is thus, interspecific competition for the same larval food source. Studies on the interspecific interactions between these cucurbit feeders in Africa are, however, scarce. Mwatawala et al. (2010) studied the host range and relative abundance of cucurbit feeders in central Tanzania. They concluded that Z. cucurbitae dominated most cucurbit hosts, in comparison to the indigenous Dacus species. Only Dacus ciliatus was predominant in some hosts like Citrullus lanatus (Thunb.) Matsum. and Nakai (and Momordica charantia to a lesser extent). A pilot study exploring genetic differentiation between 42 Tanzanian Z. cucurbitae specimens  (1996)(1997)(1998)(1999), three species (Z. cucurbitae, Dacus ciliatus, and Dacus demmerezi (Bezzi)) infested a range of 16 cucurbit species (Vayssières and Carel 1998;Vayssières 1999). The altitudinal limits of Z. cucurbitae, D. ciliatus and D. dem- Figure 5. Zeugodacus cucurbitae specimens (n = 42) reared from four different hosts (Cucumis dipsaceus, Cucurbita sp., Luffa sp., Momordica rostrata) at the Sokoine University of Agriculture (Morogoro, Tanzania) and genotyped at 19 microsatellite loci (Mwatawala, Virgilio, De Meyer, unpublished data). merezi were, respectively, 1200m, 1400m, and 1600m during the hot season. These three species have an overlap on all cucurbit crops up to 600m during the cold season and until 1200m during the hot season. At least one abiotic factor (altitude) and two biotic ones (host availability, interspecific competition) are the main screening factors for species-dominance in La Réunion. Among the 16 cucurbit hosts, D. ciliatus dominated in the cultivated hosts Citrullus colocynthis (L.) Schrader, Cyclanthera pedata (L.) Schrader, Secchium edule (Jacq.) Sw., and several cultivars of Cucumis melo and Cucurbita pepo L., which were cultivated above the altitudinal limit of Z. cucurbitae (600m during the cold season and up to 1200 meters during the hot season). Zeugodacus cucurbitae dominated on wild species such as Coccinia grandis (L.) Voigt., Cucumis anguria L., Lagenaria sphaerica (Sond.) Naudin, Momordica charantia, and also cultivated ones such as Citrullus lanatus, Cucumis melo, Cucumis sativus, Curcubita pepo, Luffa acutangula (L.) Roxb., Luffa cylindrica M. Roem., Momordica sp., and Trichosanthes cucumerina L. (Vayssières 1999). Vayssières et al. (2008) compared in detail the demography of Z. cucurbitae and D. ciliatus on La Réunion. They concluded that both species have a distinctly different life-history pattern with Z. cucurbitae being characterized by a later onset of reproduction, a longer oviposition time, longer life span and higher fecundity, while D. ciliatus has earlier reproduction, lower oviposition time, shorter life span and lower fecundity.

Host range and interspecific competition
These differences in demography seem to lead to exploitative and interference competition between the two species (and most likely other cucurbit infesters as well), with Z. cucurbitae having an advantage over Dacus ciliatus. This predominance is suggested by the majority of infestations in wild cucurbit species in the field by Z. cucurbitae. Duyck et al. (2004) reviewed the invasion biology of (polyphagous) fruit flies and demonstrated that presence of several introduced species in areas already occupied by other tephritids, results in a decrease in number and niche shift of the pre-established species. This is largely governed by life-history strategies that species adopt for interactions in near-optimal conditions. Although the review focused on polyphagous species, a similar scenario should be considered for oligophagous pests like Z. cucurbitae. So far, all studies indicate that Bactrocera species are best adapted to exploit and to compete with other species in the same ecological niche (Duyck et al. 2006, Vayssières et al. 2008). It has also been suggested that host-range can allow niche differentiation  and that this could be the reason for the different host ranges observed for Z. cucurbitae in La Réunion versus West Africa (Vayssières et al. 2007), with Z. cucurbitae being more polyphagous in West Africa. While only two indigenous cucurbit-feeding fruit flies are found on La Réunion (Vayssières andCarel 1998, De Meyer et al. 2012), at least nine are reported from West Africa ). This could reflect higher interspecific competition in the latter case, with occasional shifts of Z. cucurbitae to non-cucurbits.
In addition to interspecific competition, the host availability and ecological niches will also affect the occurrence and impact of Z. cucurbitae. Earlier studies in Hawaii have shown that it is a species that is mainly found in warmer areas and that its abundance declines with increasing rainfall and increasing elevation (Vargas et al. 1989). This preference for warmer periods was confirmed in studies in La Réunion (Vayssières 1999). Studies in Tanzania showed that Z. cucurbitae was either absent or relatively less abundant at higher elevations along a transect from approx. 600 masl to 1650 masl. However, the exact relationship between these biotic and abiotic factors that can have an impact on the host range in different African populations, is currently poorly known and requires further investigation.

Conclusion
Morphologically and genetically Zeugodacus cucurbitae shows mating compatibility among test populations and limited intraspecific genetic and morphological variability. It is still not clear if the relatively recent records for this species on the African mainland (1930s in East Africa, beginning of 21 st century in West Africa) are the result of local expansions of already established African populations or of one or more introductions from non-African sources. Regardless differences in host range reported across African populations there is no evidence supporting the existence of genetically isolated host races with specific feeding preferences and the observed host range variability seems more to be related to factors such as interspecific competition, host availability, and ecological niche partitioning. Although our study focused on the African populations, there is no indication that the situation might differ across the distribution of Z. cucurbitae.