Host Plants and Host Plant Preference Studies for Bactrocera invadens (Diptera: Tephritidae) in Kenya, a New Invasive Fruit Fly Species in Africa

Abstract

Bactrocera invadens Drew, Tsuruta & White (Diptera: Tephritidae), an invasive fruit fly species of Asian origin, was detected in Kenya in 2003, and is now well established in several parts of the country. We assessed the host range of this major quarantine pest in Kenya by collecting a wide range of cultivated and wild host plants from December 2004 to April 2006. Fruit were collected from 90 plant species representing 40 families from the Coast, Eastern, and Rift Valley provinces of the country where the fly population had been observed to occur in large numbers and where fruit and vegetable production is predominant. Fourteen plant species, among them cultivated and wild fruiting species, were found to be hosts of B. invadens. Fruit of mango, Mangifera indica L. (Anacardiaceae); banana Musa sp. AAA (Musaceae); and citrus [Citrus limon (L.) Burm.f. (lemon), Citrus reticulata Blanco (tangerine), and Citrus sinensis (sweet orange) (all Rutaceae)], were among the cultivated species heavily infested by B. invadens. Marula Sclerocarya birrea (A.Rich) Hochst. (Anacardiaceae) and Terminalia catappa L. (Combretaceae) were found to be the most infested noncultivated plants. These wild plants evidently ensure that sufficient reproductive bases exist for B. invadens during the off-season when the cultivated hosts are not in fruiting. In laboratory host preference studies, mango and banana were found to be the most preferred host plants among the nine cultivated plant species tested.

In March 2003, an invasive species of fruit fly from the genus Bactrocera Macquart was detected in Kenya (Lux et al. 2003) and most recently described as Bactrocera invadens Drew, Tsuruta & White (Diptera: Tephritidae) (Drew et al. 2005). Since the first report, the insect has rapidly spread across the African continent and in addition to Kenya it is now known from 20 other countries, including Angola, Benin, Burkina Faso, Cameroon, Comoros Island, Congo, DR Congo, Equatorial Guinea, Ghana, Guinea, Ivory Coast, Mali, Niger, Nigeria, Senegal, Sierra Leone, Sudan, Tanzania, Togo, and Uganda (Drew et al. 2005, French 2005, Vayssières et al. 2005, Ekesi et al. 2006). B. invadens is thought to have invaded Africa from the Indian subcontinent, and it was discovered in Sri Lanka after it was first reported from Africa (Drew et al. 2005), where it has become a significant pest of quarantine and economic importance (Mwatawala et al. 2004; Vayssières et al. 2005; Ekesi et al. 2006; Ekesi 2006). B. invadens belongs to the Bactrocera dorsalis Hendel complex of tropical fruit flies (French 2005), which comprises >75 species largely endemic to South East Asia (Drew and Hancock 1994; Tsuruta and White 2001; Clarke et al. 2004, 2005,), with undescribed species remaining in collections (Lawson et al. 2003). The group is arguably regarded as one of the most destructive to fruits and vegetables worldwide (White and Elson-Harris 1992; Clarke et al. 2005; Drew et al. 2005). Most recently, B. dorsalis was accidentally introduced into French Polynesia where it has displaced two other tephritid species, spread to five different Society Islands, and it infests 29 different host fruit so far (Vargas et al. 2007). In Kenya, trap catches using methyl eugenol (ME) as attractant revealed that B. invadens is now widely distributed and prevalent in several cultivated and forested areas, particularly in the warm humid lowland areas (S.E. et al., unpublished data). Previously documented species of Bactrocera in Kenya include melon fly Bactrocera cucurbitae Coquillett; olive fruit fly, Bactrocera oleae (Gmelin), Bactrocera biguttula (Bezzi); and Bactrocera munroi White (White and Elson-Harris 1992, Copeland et al. 2004). Among the four species, B. cucurbitae was the most prevalent and destructive, attacking both cultivated and wild cucurbit plants (Ekesi 2006). Other Bactrocera species within the B. dorsalis complex in Africa include Bactrocera zonata (Saunders), which is also an invasive species now established in Egypt (Anonymous 2005) and known to be highly destructive attacking 13 plant species in Mauritius and Réunion (Quilici and Jeuffrault 2001). The arrival of B. invadens adds to the list of Bactrocera species on the continent and neighboring Islands and compounds the fruit fly problems in the region.

Most frugivorous tephritids within the B. dorsalis group are known to attack a wide range of fruit, vegetable, and wild plant species. For example, three species within the B. dorsalis complex: Bactrocera papayae (Drew & Hancock) has 209 recorded hosts across 51 plant families, B. dorsalis has 124 host species across 42 families in tropical Asia, and Bactrocera carambolae (Drew & Hancock) has 77 host species across 27 families (Drew 1989; Hollingsworth et al. 2003; Clarke et al. 2005). Of the Bactrocera species reported in Kenya, the host range of B. cucurbitae are primarily cucurbits, but it has been recorded from a few noncucurbit hosts (White and Elson-Harris 1992). The olive fly exclusively infests fruit of Olea europaea L., B. munroi infests Olea europaea and Olea welwitschii Gilg & Schnellenberg, and B. biguttula the coastal olive, Olea woodiana Knobl. (Copeland et al. 2004). In Benin, Vayssières et al. (2005) reported 10 plant species as host of B. invadens. In Tanzania, Mwatawala et al. (2006) identified 15 fruit species as host to B. invadens. The only published host record from Kenya is its attack on Strychnos mellodora S. Moore and mango, Mangifera indica L. (Lux et al. 2003, Drew et al. 2005, Ekesi et al. 2006). Host status is a dynamic phenomenon and this list is by no means exhaustive, and given that the B. dorsalis complex to which B. invadens belongs attack several host plant species, it is envisaged that this list is likely to increase.

Because of the “novelty status” of B. invadens, very little is known about the ecology of this pest, and the need to document the host plants of this important quarantine pest becomes crucial. Therefore, the main objective of this study was to catalog the host plants of B. invadens in Kenya, given its importance as a major quarantine pest, to provide necessary information that may be useful for management of the pest. We also conducted host preference studies in the laboratory in a choice and no-choice tests that included nine of the major export fruits and vegetables that were either infested or not infested in the field survey to ascertain the most preferred host plant of the insect.

Materials and Methods

Field Surveys.

Sampling Sites.

Host fruit survey was carried out from December 2004 to April 2006 in three provinces in Kenya where B. invadens had been previously confirmed with ME-baited traps to be in high abundance (Lux et al. 2003; S.E. et al., unpublished data). Priority was particularly given to locations with large diversity of fruit, spread across the Coast, Eastern, and Rift Valley provinces of the country (Fig. 1).

Fig. 1.

Map of Kenya showing fruit collection sites.

At the Rift Valley Province, surveys were concentrated at Nguruman division. At the Coast Province, sampling locations included forested areas on the fringes of the Indian Ocean and high altitude areas in the Taita hills. In the Eastern Province (representing the highland region of the country), sampling locations were varied up to the fringes of Mt. Kenya forest. At each location, approximate latitude, longitude and altitude were taken using a global positioning system device (Table 1).

Table 1.

Fruit sampling sites with approximate georeferenced positions and altitude

Fruit Collection, Handling, and Processing.

Fruit were collected from cultivated fields, backyard gardens, woodlands, roadside shrubs, forested areas, and protected reserves. Often, a few fruit not encountered from sampling sites were purchased from roadside markets, and whenever possible attempts were made to establish the place of origin. Fruit samples collected included ripe to overripe fruit, including those with visible symptoms of fruit fly damage both from the tree and from the ground as “windfalls.” Attempts were made to sample large quantity of fruit with a minimum of 15 fruit per fruiting species although in some cases this sample size could not be maintained due to unavailability of fruit. Fruit collections of the different plant species were separately placed in perforated polyethylene bags in the field for transport to the rearing facility. The rearing facilities were located in each ecozone where fruit were collected and included the International Centre of Insect Physiology and Ecology (icipe)-Muhaka field station for Coast, icipe field station at Nguruman for Rift Valley, and icipe headquarters in Nairobi for Eastern Province samples.

At the rearing facility, fruit were counted, weighed, and secured in well-aerated rectangular plastic containers. Small fruit (<5 cm in diameter) were held together in 1.5-liter rectangular transparent plastic containers (20 by 12.5 by 8 cm) (Kenpoly, Nairobi, Kenya). Larger fruit (>5 cm in diameter) were held in groups of two or three in 3-liter rectangular plastic containers (20 by 12.5 by 15 cm) (Kenpoly). Fruit >10 cm diameter in were held in cylindrical plastic buckets (25 by 30 cm) (no. 20, Nairobi Plastics Limited, Nairobi, Kenya). The rim of the containers was covered with a fine netting material held in place by the perforated cover of the containers that was capable of retaining adult tephritids. The fruit were placed on 40–60 mm of moistened sterilized sand at the bottom of the rearing containers. The sand served both as the pupation medium for the larvae that exited the fruit in addition to soaking up fruit juices (Woods et al. 2005). Fruit were held at ambient conditions for 4–6 wk, depending on the fruit species.

Rearing cages were checked daily, and puparia were picked from the sand with a pair of soft forceps, counted and placed in petri dishes with moistened filter paper. In some large juicy fruit, pupation occurred inside the fruit, and in this case rotting fruit also were dissected to completely recover all remaining puparia. The petri dishes with puparia were then held in small, ventilated, transparent cylindrical plastic cages (5.5 by 12.5 cm) (no. J-12, GP Plastics, Nairobi, Kenya) until eclosion. Emerging tephritids were provided with an artificial diet that consisted of a volumetric mixture of 1:3 enzymatic yeast hydrolysate and sugar, and water was provided in pumice granules. Flies were allowed to feed for 4 d until full adult development and body colorations were attained. They were then killed by placing them in a freezer and later preserved in 70% alcohol. All specimens were shipped to the icipe Biosystematics unit for identification where a reference collection is kept. Samples of flower, fruit (for small fruit), leaf, twig (or both) from unknown plant species also were collected, pressed, and bagged. The collected plant samples were identified using the keys of Kenya trees, shrubs, and lianas (Beentje 1994). Photographs also were taken of each plant or fruit sampled to aid in plant identification, most of which was provided by Dr. R. S. Copeland (icipe). Voucher specimens of all collections of the plant species are maintained at icipe. Plant nomenclature used conforms to the International Plant Names Index database (IPNI 2004) and the Missouri Botanical Garden database W³ TROPICOS (MBOT 2006).

Laboratory Host Preference Studies.

Choice Test.

The experiments were conducted in 90- by 90- by 90-cm Plexiglas cages in a laboratory maintained at ambient conditions. Nine fruit and vegetable species, including mango, papaya (Carica papaya. L.), banana (Musa sp. AAA), guava (Psidium guajava L.), sweet orange [Citrus sinensis (L.) Osbeck], custard apple (Annona squamosa L.), cucumber (Cucumis sativus L.), avocado (Persea americana Miller) and tomato (Lycopersicon esculentum Miller), which were hosts and nonhosts records from the field surveys, were tested. Each experimental cage was divided equally into nine subunits, and each unit held one fruit species supported by a string from the roof of the cage. Distance between each fruit species was 30 cm. All fruit and vegetable species were tested when they were either fully ripe (mango, papaya, banana, orange, guava, tomato, and custard apple) or at mature green stage (avocado and cucumber). It is generally well established that host plants are most susceptible to fruit flies at these stages. One hundred adult B. invadens (consisting of 50 females and 50 males) at 2–3 wk old were then released inside the cages for a period of 24 h. Flies were fed on 1:3 volumetric mixture of enzymatic yeast hydrolysate and sugar. Water also was provided on pumice granules. After 24 h, all fruit species were removed and incubated individually as described for field surveys. Records were kept for pupal recovery and percentage of adult emergence from the total puparia recovered. Four replicated cages were maintained and the experiment was repeated twice.

No-Choice Test.

Two sets of experiments were conducted under the no-choice test. In the first experiment, we measured female fecundity on fruit domes of the nine plant species for a period of 10 d. Fruit domes were made by scooping the pulp and seeds out of the fruit. Fruit peel thickness varied slightly among the fruit species and ranged from 2.2 to 2.5 mm in diameter. Each fruit dome was then transferred into a 20- by 20- by 20-cm Plexiglas cage, and a pair of adult B. invadens was released inside the cage. Records were kept of the number of eggs laid daily on the dome by washing the eggs off the underside of the domes. Flies were fed as described previously. To record hatch rate, the eggs were transferred onto strips of moist blotting paper and hatch rate was determined by observing for eclosion under a binocular microscope after 2 d. The cages were arranged in a complete randomized design with three replications.

In the second experiment, whole fruit were exposed to flies in 25- by 25- by 25-cm Plexiglas cages. A single fruit sample from each of the nine plant species listed above was transferred into each cage. Forty 2–3-wk-old adult flies (20 females and 20 males) were released in each cage for a period of 24 h. Flies were fed as in previous experiment. At the end of the exposure period, the fruit was removed and processed as described previously. Five replicates were maintained and the experiment was repeated twice.

Statistical Analysis.

Data for field surveys are presented according to plant species, family, location, number of fruit collected and weight, number of infested fruit, and number of adults. Infestation by B. invadens followed the methodology of Cowley et al. (1992), and it was calculated as the ratio of number of adults per kilogram of fruit collected (infestation index). In the laboratory experiments, data were tested for normality and homogeneity of variance by transforming to natural logarithms (pupal recovery and fecundity) and angular transformation (percentage of egg hatch and adult emergence) before subjecting to analyses. Because the experimental design in the choice experiment did not support the assumption of sample independence for analysis of variance (ANOVA), the nonparametric equivalents, Kruskal–Wallis and chi-square tests, were used to analyze the data. In the no-choice experiment, data were subjected to ANOVA using the generalized linear model (Proc GLM), and means were separated by Tukey honestly significant difference (HSD) test (P = 0.05). All analyses were performed using the SAS package (SAS Institute 2001).

Results

Field Survey.

B. invadens was reared from a total collection of 3,913 fruit from a range of habitats that included 14 plant species and eight families from surveys carried out at the Coast, Eastern, and Rift Valley provinces of Kenya (Tables 1 and 2). Fruit species positive for B. invadens included both cultivated and wild host plants (Table 2).

Table 2.

Host fruit infestation indices for B. invadens in three provinces of Kenya from December 2004 to April 2006

The majority of B. invadens-infested samples were from commercial fruit. Ten of the host plants are new records for B. invadens in Kenya. During the survey, a collection of 4,630 fruit composed of 76 other plant species from 32 families did not yield B. invadens. Our data indicated that B. invadens was capable of infesting fruit over an altitudinal range of 20–1,335 m above sea level (masl), with infestation varying from 1.1 to 652.8 flies per kg of fruit (Tables 1 and 2).

Among the plant species sampled, B. invadens infestation was recorded from cherimolia, soursop, custard apple (Annonaceae), Citrus limon (L.) Burm.f. (lemon), Citrus reticulata Blanco (tangerine), sweet orange (Rutaceae), Cordia myxa L. (Boraginaceae), tomato (Solanaceae), mango (Anacardiaceae), banana Musa sp. AAA (Musaceae), guava (Myrtaceae), Sclerocarya birrea (A.Rich.) Hochst. (marula), Sorindeia madagascariensis L. (Anacardiaceae), and Terminalia catappa L. (tropical almond) (Combretaceae) (Table 2). The families Anacardiaceae, Annonaceae, and Rutaceae had the highest number of species infested, with B. invadens reared from three species sampled in each family. The most heavily infested Anacardiaceae was M. indica, with infestation reaching 130.3 flies per kg fruit at Muhaka, Coast Province, whereas in the Annonaceae, the wild species, A. cherimola sampled at Nthagaiya, Eastern Province, recorded the highest number of B. invadens (85.0 flies per kg fruit) (Table 2). Among the Rutaceae, C. limon was the most infested at Nguruman, Rift Valley Province (32.3 flies per kg fruit). In the other provinces, members of the Rutaceae sampled were less infested, and C. sinensis and C. reticulata had the highest level of infestation at 2.0–5.6 flies per kg fruit. Of the wild host fruit sampled, the highest level of infestation was recorded on T. catappa (652.8 flies per kg fruit) sampled from Nguruman, Rift Valley Province and S. birrea (238.8 flies per kg fruit) sampled from Muhaka, Coast Province (Table 2). Generally, fruit infestations were higher at low elevations than at the highland areas (Tables 1 and 2). For example, mango fruit infestation varied from 39.2 to 130.3 flies per kg fruit at the low-elevation locales in the Coast Province and Rift Valley compared with 0–29.4 flies per kg fruit in the high elevation areas of the Eastern Province (Table 2).

Other tephritid species also were encountered during the survey period, including B. cucurbitae, Ceratitis anonae (Graham), Ceratitis capitata (Wiedemann), Ceratitis cosyra (Walker), Ceratitis rosa Karsch, Ceratitis pedestris (Bezzi), Ceratitis pinax Munro, Dacus frontalis Becker, Dacus vertebratus (Bezzi), Trirhithrum nigerrimum (Bezzi) and Trirhithrum senex Munro. The host plants data collected for these fruit flies are not presented because they were similar to previously documented information (Liquido et al. 1991; White and Elson-Harris 1992; Mukiama and Muraya 1994; Quilici and Jeuffrault 2001; Copeland et al. 2002, 2006,; De Meyer et al. 2002; Ekesi 2006).

Laboratory Host Preference Studies.

Choice Test.

There was a significant difference in the number of puparia recovered (Kruskal–Wallis: H = 56.2, df = 8, P = 0.0001) and adult emergence (Kruskal–Wallis: H = 36.2, df = 8, P = 0.0001) from the nine fruit species exposed to B. invadens(Fig. 2A).

Fig. 2.

Host preference and performance of B. invadens in terms of pupal recovery, adult emergence, fecundity, and fertility on nine cultivated plant species in a choice (A) and no-choice (B and C) tests in the laboratory. Bars and data points on line with same letter do not differ significantly by Tukey’s HSD test (P = 0.05). Error bars denote SE.

Highest number of puparia was recovered from mango, papaya, and banana, whereas the lowest recovery was observed in tomato, cucumber, and custard apple (Fig. 2A). Percentage adult emergence ranged from 60 to 86% (Fig. 2A).

No-Choice Test.

There was a significant difference among the fruit species in total fecundity over 10 d (F = 12.6; df = 8, 18; P = 0.0034) and fertility (F = 24.5; df = 8, 18; P = 0.0001) (Fig. 2B). The highest number of eggs was recovered from mango, papaya, banana, and cucumber domes compared with the other five fruit species (Fig. 2B). Egg fertility ranged from as low as 21% in tomato to 82% in mango (Fig. 2B).

In the second set of experiments in the no-choice test, pupal recovery and adult emergence were also significantly different among the fruit species (F = 33.4; df = 8, 81; P = 0.0001 and F = 24.0; df = 8, 81; P = 0.0001, respectively) (Fig. 2C). Mango, papaya, and banana recorded the highest number of puparia, whereas the lowest occurred on cucumber (Fig. 2C). Percentage of adult emergence varied from 31 to 86% (Fig. 2C).

Discussion

B. invadens was reared from 14 plant species, representing eight plant families most of which are new host plant family records in Kenya and in Africa. In the first description, Drew et al. (2005) listed four cultivated host plants, namely, guava, mango, citrus, papaya, and some unidentified wild plants as host of B. invadens in Africa. In Benin, West Africa, Vayssières et al. (2005) reported attacks on cashew [Anacardium occidentale L. (Anacardiaceae)], pepper (Capsicum annuum L.), Cucurbita spp., custard apple, guava, mango, papaya, Diospyros montana Roxburgh (Ebenaceae), and Vitellaria paradoxa C.F. Gaertner (Sapotaceae) by B. invadens. In Tanzania, Mwatawala et al. (2006) reported 15 host plants and identified mango, loquat [Eriobotrya japonica (Thunb.) Lindley], guava, and grapefruit (Citrus x paradisi Macfad.) as the favored hosts. In our study, B. invadens was found to infest fruit species within the families Annonaceae, Rutaceae, Boraginaceae, Solanaceae, Anacardiaceae, Musaceae, Myrtaceae, and Combretaceae, suggesting that B. invadens is an emerging polyphagous pest that may be capable of sustaining its population through reproduction on a range of cultivated and wild fruit.

The plant families listed above have been reported to be key host plant families of several Bactrocera species, including members of the B. dorsalis complex of fruit flies (White and Elson-Harris 1992, Tsuruta et al. 1997, Clarke et al. 2005). In Mauritius and Réunion, B. zonata infests A. reticulata and M. indica (Quilici and Jeuffrault 2001), and the relative importance of these plants as hosts of B. invadens also is confirmed by our results. Bactrocera dorsalis infests Annona spp., Citrus spp., M. indica, Musa spp., and P. guajava (Armstrong 1983, Allwood et al. 1999, Clarke et al. 2005), and this is consistent with our findings for B. invadens.

M. indica was the most important host of B. invadens among the fruit sampled within the Anacardiaceae. The data reveals mango infestation in mid- to high-elevation areas of Eastern Province of Kenya. In previous studies by Ekesi et al. (2006), no B. invadens was recovered from mango in this locality. The current record of B. invadens in this locality clearly indicates that the pest is gradually expanding its range and exploiting host fruit at higher elevation areas of the country. S. birrea also seems to be an important reservoir host for B. invadens. High infestation levels were recorded at the Coast and Rift Valley provinces (238.8 and 39.7 B. invadens per kg respectively). This plant generally fruit sporadically throughout the year (Jøker and Erdey 2003), and it may be an important off-season host for B. invadens in the absence of the primary cultivated host plants.

Among the Annonaceae, B. invadens was reared from A. squamosa, A. muricata, and A. cherimola. Mwatawala et al. (2006) showed that A. muricata was a major host of B. invadens in Tanzania. Studies elsewhere have shown that other species of Bactrocera, such as B. carambolae, Bactrocera correcta (Bezzi), B. dorsalis, Bactrocera kandiensis Drew & Hancock, Bactrocera frauenfeldi (Schiner), and B. papayae are frugivorous on this family (Tsuruta et al. 1997, Hadwen et al. 1998, Hollingsworth et al. 2003, Clarke et al. 2005). B. invadens evidently uses these wild and planted cultivars of Annonaceae, and management activities directed at B. invadens should take into account the importance of these plants as hosts of the insect.

C. limon was the most infested of the Rutaceae by B. invadens compared with C. reticulata and C. sinesis. In Tanzania, B. invadens was not found to attack C. limon, although infestations were observed on C. reticulata and C. sinensis (Mwatawala et al. 2006). Among the B. dorsalis complex, C. sinensis seems to be a less preferred host plant. For example in Surinam, South America, Clarke et al. (2005) reported that infestation rates by B. dorsalis on C. sinensis was 1.2% compared with 16.3% for M. indica and 11.3% for P. guajava. In our study, we observed an infestation rate of up to 4.6 and 5.6 B. invadens per kg on C. sinensis and C. reticulata, respectively. The observed high levels of infestation in C. limon were indeed remarkable given the acidic nature of this plant. Vayssières et al. (2005) has reported high infestation of B. invadens in a similarly acidic host plant (A. occidentale), suggesting the pest may be adapted to a wide range of fruit characteristics.

Terminalia catappa (Combretaceae) is generally known to harbor a complex of fruit fly species of the B. dorsalis group, including B. zonata, B. correcta, B. dorsalis, B. kandiensis, B. papayae, B. zonata, and Bactrocera sp. near nigrotibialis (taxon A) (Tsuruta et al. 1997, Quilici and Jeuffrault 2001, Hollingsworth et al. 2003, Quilici et al. 2005, Clarke et al. 2005). The high infestation levels recorded in our study (652.8 per kg B. invadens) confirm the status of T. catappa as an important host plant of B. invadens. This is perhaps not surprising given that the plant species is native to Asia (Styger et al. 1999, Thomson and Evans 2006). T. catappa can flower up to three times a year, producing fruit almost year-round (Thomson and Evans 2006; S.E., unpublished data), and it probably harbors successive generations of this pest, which infest orchards when fruiting begins. In Kenya, T. catappa thrives as an ornamental tree, mostly used as shade trees around the homesteads and sometimes in proximity to mango orchards. Under such systems, management strategies for B. invadens also must take cognizance of the presence of this important wild host in addition to the cultivated plants.

We reared B. invadens from banana (Musaceae), which is known to be a major host of Bactrocera species, the most important being Bactrocera musae (Tryon), and B. papayae (White and Elson-Harris 1992; Clarke et al. 2005). B. invadens can infest green banana both in the laboratory and field (S.E., unpublished data). Because this fruit is largely exported around the world at the mature green stage, it is likely that a strategy exploiting avoidance of B. invadens by harvesting and shipping banana at maturity may be inappropriate for evading infestation by this pest.

The Myrtaceae, guava is known to host a variety of fruit fly species worldwide, including several species of Bactrocera (White and Elson-Harris 1992, De Meyer et al. 2002, Clarke et al. 2005). We were able to rear B. invadens from this plant confirming the status of guava as a major host plant of fruit flies. In West and Central Africa, Vayssières et al. (2005) reported this plant as a major host of B. invadens. Similarly in Tanzania, Mwatawala et al. (2006) showed that guava was highly favored by B. invadens.

Among the Solanaceae sampled in the current survey, B. invadens was recorded from tomato. In other regions of the world, L. esculentum is attacked by other Bactrocera species such as B. carambolae (Clarke et al. 2005), B. papayae (Hadwen et al. 1998, Clarke et al. 2005), B. tryoni (Froggart) (Balagawi et al. 2005), and Bactrocera latifrons (Hendel) (Liquido et al. 1994). However, Mwatawala et al. (2006) did not find infestation on this plant in Tanzania. In our survey, tomato samples from which B. invadens was reared were collected from a backyard garden. No flies have yet been detected from fruit collections at commercial scale where pesticides are regularly used for management of fruitworms and tetranychid mites.

B. invadens also was reared from Boraginaceae in C. myxa collected from Coast Province. Fruit fly records for Cordia species in Africa are scanty; nevertheless, the Boraginaceae are reported to be hosts to some Ceratitis species (Quilici and Jeuffrault 2001, De Meyer et al. 2002). No tephritids were reared from fruit of eight Cordia species collected in Kenya from 1999 to 2001 (Copeland et al. 2002). Our record from Kenya is the first with regard to infestation by Bactrocera species.

Among all the infested plant species, the highest infestation rates were recorded from low elevation areas of the Coast and Rift Valley provinces compared with the higher elevation areas of Eastern Province. In a recent study on mangoes in Kenya, Ekesi et al. (2006) showed a significant inverse relationship between the numbers of B. invadens infestation per kilogram of mango fruit and elevation from which fruit were collected, and they concluded that B. invadens seemed to be a lowland resident pest. Most fruit flies from the genus Bactrocera are considered to be lowland residents. Vargas et al.(1983) demonstrated that fruit infestation by B. dorsalis in native and exotic forests on Kauai Island was moderate at middle (579–800 masl) elevation and low at high (>800 masl) elevation. Generally, elevation by itself does not determine fruit fly distribution, but associated factors such as temperature, rainfall, and host plants at such elevation play a significant role (Nishida et al. 1980).

In the laboratory host preference studies, results of the choice experiment showed that mango and papaya were the most preferred host plants of B. invadens, followed by banana. These laboratory results agree with results obtained from the field on mango and banana, but they sharply contrast with field survey results for papaya where no infestation was recorded. The reason for lack of infestation of papaya in the field is not very clear given that large quantity of papaya (42.3 kg) was sampled from localities that were heavily infested by B. invadens. In Tanzania (neighboring country to Kenya), Mwatawala et al. (2006) did not record any infestation on papaya by B. invadens in the Morogoro region. However, in the Zanzibar region of the country, B. invadens infestation on papaya has been observed in field samples as low as 2 kg of fruit in a locality with a lower prevalence of B. invadens than Nguruman, Kenya (A. I. Ali, unpublished data). This observation largely highlights the need for continuous field survey in different localities in Kenya.

In the no-choice fecundity and fertility studies, highest numbers of eggs were laid on mango, papaya, banana, and cucumber domes compared with the other plant species tested. Pupal recovery from mango, papaya and banana were consistent with results from fecundity and fertility tests, but the level of pupal recovery from cucumber was the lowest among all the fruit tested. In general, fecundity and fertility results for papaya and mango support pupal recovery, but this was not the case with cucumber. The reason for the contrasting results is unclear and warrants further investigation.

In conclusion, within the B. dorsalis complex to which B. invadens belongs, some insects are specialist host range species, whereas others are general polyphagous species. Our result suggests that B. invadens may be an emerging polyphagous species. In general, the host list generated in the current study is unlikely to be exhaustive, and periodic surveys, especially for the B. invadens negative plant species, would be necessary. We also acknowledge that the sample size for some of the fruit species collected in the current study may be low, but results presented here may still be useful in making some phytosanitary and pest management decisions.

Host plant preference studies clearly demonstrated that mango and banana are the most preferred host plants of B. invadens in Kenya; thus, the pest is likely to jeopardize lucrative export of these crops from this region. Indeed, some countries have already banned the importation of these fruit from Kenya and Uganda due to the threat posed by B. invadens (S. Muchemi, personal communication; E. Niyibigira, personal communication). Future studies should now aim at developing effective management strategies that should include baiting techniques, use of biological control agents, sound orchard sanitation, and postharvest treatments for quarantine-sensitive markets.

Acknowledgments

We acknowledge the staff of the Biosystematics Unit at icipe for identification of all insect specimens. We appreciate exchange of ideas with S. A. Lux (icipe), and we also thank R. S. Copeland (icipe) for useful discussion and comments on earlier draft of the manuscript and for help in identification of most of the plant specimens collected. The assistance of P. O. Agolla and P. Nderitu in field collection and rearing activities is gratefully acknowledged. We are thankful to N. J. Liquido (USDA–APHIS–PPQ, Hawaii) for careful review of the manuscript. The research was supported by funds from the International Fund for Agricultural Development and the International Atomic Energy Agency to the African Fruit Fly Initiative and by the German Academic Exchange Service scholarship to I. R. through the African Regional Postgraduate Program in Insect Science.

Author notes

¹

International Centre of Insect Physiology and Ecology (icipe), P.O. Box 30772-00100 GPO, Nairobi, Kenya.

²

Department of Zoological Sciences, Kenyatta University, P.O. Box 43844-00100, Nairobi, Kenya.

³

Corresponding author, e-mail: .