The adaptation to novel host plants is an important factor that facilitates the geographic expansion of tephritid flies (Diptera: Tephritidae) (Ju et al. 2012, Wan 2009). Many studies have shown that geographic distribution of tephritids is closely related to host plant diversity (Malacrida et al. 2007, Nardi et al. 2005). As fruit flies expand their geographic range, they will encounter novel hosts with selection pressures possibly affecting their growth, development, reproduction, and, ultimately, their genetic composition (Faucci et al. 2007, Schwarz et al. 2005).

There are several examples where tephritid flies have adapted to new hosts with changes in their genetic composition. In the case of the apple maggot, Rhagoletis pomonella (Walsh), populations adapted to its traditional host apple (Malus) and those adapted to the novel host hawthorn (Crateagus) formed two sympatric host races (Han and McPheron 1994). Similarly, U.S. populations of R. pomonella that adapted to honeysuckle (Caprifoliaceae: Lonicera), a different plant family than the more typical hosts (e.g., Rosaceae), are considered a new hybrid species and often termed the “Lonicera fly” (Martel et al. 2003).

Bactrocera tau (Walker) (Diptera: Tephritidae), is a major economic pest that is listed as a quarantine pest in many countries (Ooi and Wee 2016). This fly was first reported in 1894 in Fujian, a southeast coastal province of China (Walker 1849). The current geographic range of B. tau includes several countries in tropical and subtropical Asia and the South Pacific (Singh et al. 2010), such as China, Thailand, India, Cambodia, Indonesia, and Pakistan (Hasyim et al. 2008, Huang et al. 2005, Huque 2006, Ohno et al. 2008). The geographic range of B. tau is still expanding in many regions. For example, in China, B. tau has recently spread to two northern provinces (i.e., Shanxi and Shaanxi) (Wan et al. 2010). Similarly, in Japan, B. tau has recently been reported for the first time in Okinawa and Ishigaki Islands (Ohno et al. 2008).

Bactrocera tau is polyphagous and mainly infests plants in the Cucurbitaceae family (Christenson and Foote 1960). However, the range of host plants utilized by B. tau has expanded dramatically over the last century (Allwood et al. 1999) and now includes more than 80 plant species (Huang et al. 2005), including species in the plant families Leguminosae (e.g., Phaseolus vulgaris L.), Moraceae (e.g., Ficus racemosa L.) (Sumrandee et al. 2011), Myrtaceae (e.g., Psidium guajava L.) (Hasyim et al. 2008), and Rutaceae (e.g., Citrus) (Zhang and Chen 2012).

Development and reproduction rates, as well as genetic composition of B. tau populations, vary considerably among geographic regions (Sumrandee et al. 2011, Dujardin and Kitthawee 2013, Jamnongluk et al. 2003a, 2003b, Saelee et al. 2006, Thanaphum and Thaenkham 2003). Such variation has been attributed to environmental factors including air temperature, and host species (Dujardin and Kitthawee 2013). Among these factors, host species is considered to be one of the critical factors that determine the geographical distribution of B. tau (Sumrandee et al. 2011). In order to enhance understanding of tephritid distribution and expansion mechanisms, it is necessary to determine the direct impact of host plants on these flies.

The aim of the present study was to examine the effect of host plants on genetic variation in B. tau. Our experiment was conducted under laboratory conditions where factors such as natural enemies, climatic conditions, and human activities could be excluded or dramatically reduced. The flies were fed with both traditional and novel host foods for 15 consecutive generations, and microsatellite markers were used to evaluate the genetic variation among different fly populations. The main objective was to elucidate any genetic variation in B. tau during development on different host plants.

Materials and Methods

Summer squash (Cucurbita pepo var. fastigata L.) infested with B. tau in the field were transported to our entomology laboratory at Yunnan University, China. These infested summer squash were placed in insect-rearing cages (60 × 40 × 45 cm) until B. tau adults emerged, and all emerging flies were subsequently maintained on summer squash. After 2 weeks of rearing, a large number of emerging adult flies were used for this study.

The test flies were divided into four groups, and each group was fed one of the following: cucumber (Cucumis sativus L.), pumpkin (Cucurbita pepo var. pepo L.), orange (Citrus), or banana (Musa paradisiaca Colla). Sixty adult flies including 30 males and 30 females were used in each group. The flies were reared continuously on each of the four host plants. However, the flies fed on orange died within three generations; therefore, tests with that host plant were discontinued. Flies on the other three host plants were reared for 15 continuous generations. After 10 d of adult emergence in each generation, we kept 30 male adults and 30 female adults to initiate the next generation. Rearing was conducted in the insect-rearing cages described above, which were placed in an environmental chamber maintained at 27 ± 1°C, 70 ± 5% relative humidity, and a photoperiod of L12:D12 h (lights 0700–1900). Each treatment was replicated three times. Survival rates were calculated for each host group based on the 5^th, 10^th and 15^th generation as described by Vargas et al. (1984). Mean survival rates of the three host plant treatment groups were compared using one-way analysis of variance (Snedecor and Cochran 1989). Means that were significantly different (P < 0.05) were separated using á posteriori Tukey's honestly significant difference test (Zar 1999).

Twenty B. tau adults were selected from each of the 5^th, 10^th, and 15^th generations for each host plant group and used for molecular analysis. These adults were preserved in 95% ethanol and stored at 4°C. The nine populations were named as follows: C5 (the 5^th generation feeding on cucumber), B5 (the 5^th generation feeding on banana), P5 (the 5^th generation feeding on pumpkin), and similarly C10, B10, and P10, and C15, B15, and P15.

In total, 180 adults were used in the molecular analyses. DNA samples of 20 adults per generation per population were extracted using a DNeasy Blood and Tissue Kit (Qiagen, Boston, MA). Seven pairs of highly polymorphic microsatellite primers were used for PCR amplification (Table 1). The amplification schedule was 40 cycles of 15 s each at 94°C, 15 s at 55°C, and 30 s at 72°C, with an initial denaturation step of 10 s at 95°C and a final extension step of 30 min at 72°C, followed by storage at 4°C. PCR products were electrophoresed using an automated ABI PRISM 310 Genetic Analyzer, and allele calling was performed using GeneMapper. An individual allele was declared null for a given locus only after at least two amplification failures.

Table 1 The seven pairs of microsatellite primers for the nine populations of B. tau feeding on the three plant hosts.

View Fullscreen

FreeNA software was used to evaluate the frequency of null alleles for each microsatellite loci (Chapuis and Estoup 2007). GENEPOP4.5 (Rousset 2008) was used to estimate the linkage disequilibrium between pairs of seven microsatellite loci in each B. tau population and deviation from Hardy–Weinberg equilibrium based on Fisher's method. The four genetic diversity indices of each B. tau population, including number of alleles (N_A), number of private alleles (N_P), frequency of private alleles (A_P), observed heterozygosity (H_O), expected heterozygosity (H_E) and gene diversity (H_S), were calculated using FSTAT2.9.3.2 (Goudet 2001).

In order to infer the genetic structure of B. tau populations feeding on different hosts, the Bayesian clustering method implemented in STRUCTURE2.3.4 (Pritchard et al. 2000) was used to first determine whether the nine B. tau populations could be subdivided into different groups. STRUCTURE software runs a model in which K is assigned to various clusters (K is known), and each K is marked with a collection of allele frequencies at each locus. Individuals from different samples are assigned to one assumed cluster, or together to two or more clusters if they shared admixed genotypes (Aketarawong et al. 2007). To select the most likely number of K based on our samples, we arranged our data on various Ks ranging from 1 to 10 and performed ten dependent runs for each K (Shi et al. 2012). Then, the estimated log probability of the data for the different Ks was compared. The proportion of individuals assigned into different K clusters was displayed as Q-matrices. When running STRUCTURE2.3.4, we set a model of admixture with 100,000 burn-in steps followed by 100,000 MCMC simulation steps. Structure groups were visualized using the Distruct software (Rosenberg 2004).

After obtaining the optimal STRUCTURE groups by using the analysis described above, analysis of molecular variance (AMOVA) was conducted to test molecular variation between these groups by using Arlequin 3.5 (Excoffier and Lischer 2010). The same software was also used to perform an AMOVA to detect the sources of genetic variation based on groups categorized by generations and host fruits.

A neighbor-joining (NJ) tree was constructed with PHYLIP3.69 (Felsenstein 2005) based on the distances of pairwise proportion of shared alleles between populations as the second method to estimate the genetic structure of B. tau populations. The NJ tree was supported by 1,000 bootstrap re-samplings of the original data over loci.

Differentiation among populations, as measured by pairwise F_ST value, was the third approach used to determine the genetic structure of nine B. tau populations using Arleqquin3.5. The same software also was used to estimate the F_ST value among groups defined by STRUCTURE2.3.4.

Gene flow among populations was detected by using GENECLASS 2.0 (Piry et al. 2004). This software estimated the gene flow by calculating the probability of assignment of each individual to other reference populations or assignment in the population itself based on multilocus genotypes (Piry et al. 2004). Values of probability were computed by the standard criterion described by Rannala and Mountain (1997) with 10,000 simulated individuals and a P = 0.01.

Results

Survival

Mean survival among the B. tau populations reared on the three plant hosts for 5, 10, and 15 generations differed significantly (F = 2.78; df = 1,540; P = 0.032 < 0.05), with pumpkin = cucumber > banana (Table 2). Survival rates of flies fed on banana increased with progressive generations.

Table 2 Mean percentage survival rate of the 5^th, 10^th, and 15^th B. tau generations from each of three host plant groups.

View Fullscreen

Genetic diversity

We used five indices, namely N_A, N_P, H_O, H_E, and Hs, to measure genetic diversity within 10 B. tau populations (Table 3). Among the nine B. tau populations, the mean number of N_A, N_P, and Hs of three banana populations were slightly lower than the values of three populations feeding on cucumber and pumpkin, respectively. The value of H_E, H_O, and Hs for the three host populations of the 15th generation (i.e., P15, C15, and B15) decreased slightly compared to the values in populations of the 5th and 10th generations that developed on the same hosts.

Table 3 Genetic diversity index for the nine B. tau host populations.

View Fullscreen

Population group assessment

The STRUCTURE analysis showed that B. tau populations that developed on the three host plants could be subdivided into four genetic clusters (K), as shown by the likelihood curve of STRUCTURE (Fig. 1, 10 runs for each K). The curve reached a plateau when K = 4; therefore, optimal K was set as 4. Each of the 180 flies was subsequently assigned to one of the four clusters with a certain probability Q (Table 4). Flies from populations B5 and B10 were mostly assigned to cluster 1 (Q > 0.45). Individuals from populations P5, P10, and P15 were mostly assigned to cluster 2 (Q > 0.70). Flies from populations C5, C10, and C15 were primarily assigned to cluster 3 (Q > 0.6). Almost all flies of B15 were assigned to cluster 4 (Q = 0.95). Therefore, we defined the four clusters as four groups: G1, G2, G3, and G4 (Fig. 2). The G1 group was composed of populations of the 5^th and 10^th generations that developed on banana (B5 and B10); the G2 group included populations of the 5^th, 10^th, and 15^th generations that developed on pumpkin (P5, P10, and P15); the G3 group was composed of populations of the 5^th, 10^th, and 15^th generations that developed on cucumber (C5, C10, and C15); and the G4 group was only one population, B15.

View Figure

• Download as Powerpoint
• Download Figure

Table 4 Average co-ancestry coefficients for the nine populations of B. tau assigned to four clusters.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Molecular variation analysis

The fixation indices of the AMOVA run using the group of populations assigned by the STRUCTURE results previously described are shown in Table 5. Most of the molecular variation was found within populations (85.05%, P < 0.01), only 8.12% of the variation was found among groups (P < 0.01), and only 6.83% of the variation was found among populations within groups. This partitioning was significant (P < 0.01).

Table 5 The F _ST values for the four groups obtained from STRUCTURE 2.3.4 and analysis of molecular variances based on the four groups.

View Fullscreen

The nine B. tau populations were also grouped by generation and host (Table 6). When populations were grouped by host, most variation was detected within populations (80.85%), 8.76% was among population within groups, and only 10.39% was among groups, with all three values being significant (P < 0.01). When the nine populations were grouped by generation, a substantial proportion of genetic differentiation was within populations (83.64%), 13.68% of the variation was among populations within groups, and only a small proportion (2.68%) of the variation was among groups.

Table 6 The analysis of molecular variance results of the 10 B. tau populations that developed on one of three host plant fruits.

View Fullscreen

NJ tree construction

An unrooted NJ tree constructed with the microsatellite data (based on the pairwise proportion of shared alleles distances) also showed four monophyletic clades, which were similar to the STRUCTURE groups identified for K = 4 (see Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

Pairwise F_ST values

Table 7 lists all microsatellite pairwise F_ST values of the nine B. tau populations, which ranged from 0.043 (C5–C10, P10–P15) to 0.256 (C15––B15). Among 36 F_ST values of the nine populations, we observed that six values were not significant. Slightly high values of differentiation were observed between the B15 population and other populations where nine F_ST values were significant and higher than 0.2. When populations were grouped according to the STRUCTURE results, pairwise F_ST values between the four groups varied from 0.010 to 0.202 (Table 5). All F_ST values among the four groups were significantly different, except for F_ST values among groups G2 and G3, which consisted of the two traditional Cucurbitaceae host feeding populations.

Table 7 Pairwise F _ST values of the nine B. tau populations.

View Fullscreen

Gene flow estimates

GENECLASS 2.0 was used to estimate the gene flow among nine B. tau populations. Gene flow was calculated by probability of individual assignment to different populations (Table 8). Most of the values of gene flow were low and not higher than 0.047. Only five values of gene flow were higher than 0.05, and four values of gene flow were higher than 0.1. The diagonal values of the assignment matrix indicated the average probability with which individuals were assigned to population itself. These probability values ranged from 0.544 (C15) to 0.625 (P5).

Table 8 Mean assignment rate of individuals into (rows) and from (columns) each population based on microsatellite data as estimated by GeneClass 2.

View Fullscreen

Discussion

Survival rate can be used as a direct measure to evaluate the effect of the various hosts on the tephritid flies (Liu et al. 2014). The present study showed that survival rates were higher for the B. tau populations fed on pumpkin and cucumber than those fed on banana, with no differences between the two Cucurbitaceae hosts (Christenson and Foote 1960) (Table 2). Survival rates of flies fed on the two non-Cucurbitaceae hosts (banana and orange) were low and survived only through three generations. Our study further showed that B. tau had higher fitness when feeding on a traditional host compared to the non-traditional hosts, thus, corroborating Liu (2014) who also reported that Bactrocera correcta (Bezzi) (Diptera: Tephritidae) and Bactrocera dorsalis (Hendel) (Diptera: Tephritidae) had high survival rates in their traditional hosts.

The impact of the host plant was also reflected in the genetic variation among the B. tau populations. Molecular variance analysis based on groups clustering by host-feeding populations (i.e., host groups) as well as the various generation populations (i.e., generation groups) demonstrated that significant differences were found among host groups with 10.39% variation; whereas, there were no significant differences among generation groups (Table 6). It appears that host food more than population generations played a critical role in the observed genetic variation of B.tau in our study.

The influence of the host plant on genetic variation in B. tau was primarily related to feeding on the non-traditional hosts. In our study, banana, belonging to Musaceae family, was considered a non-traditional host for B. tau (Huang et al. 2005). Moreover, banana could be considered a novel host for B. tau, given that there have been no reports of B. tau infesting banana under natural conditions (Huang et al. 2005). Compared with the two traditional hosts, the effect of banana on the genetic variation in B. tau was most obvious when considering the frequency balance of gene and genotype, genetic diversity variation, and genetic differences.

In this study, banana-feeding fly populations in the 5th generation as well as the 10th generation had several microsatellite loci that deviated from Hardy–Weinberg equilibrium. Also, in the 15th generation, there remained an individual locus that deviated from equilibrium. By contrast, flies that developed on pumpkin and cucumber deviated from equilibrium only in the 5th generation, with the remaining generations in equilibrium at all loci. Hardy–Weinberg equilibrium could be used to estimate the state of population equilibrium by measuring frequency balance of genes and genotypes (Chen et al. 2005). For example, the 15th generation of the banana-feeding populations had gene and genotype frequencies that did not reach balance, suggesting that it was difficult for B. tau to reach equilibrium when developing on a novel host. The molecular results of this study were consistent with the survival rate results that indicated lower fitness for B. tau on banana, the non-traditional host.

Regarding genetic diversity, three genetic diversity indices (N_A, N_P, and H_S) were somewhat lower in banana-feeding populations compared with populations feeding on the two traditional Cucurbitaceae hosts; however, there were no differences in the three indices between the two Cucurbitaceae host-feeding populations. It should be noticed that a decreasing genetic diversity was also found in B. dorsalis when it infested apple (Rosaceae) and pear (Rosaceae), two non-traditional hosts (Shi et al. 2012).

Furthermore, the nine B. tau populations were divided into four groups in the genetic structure based on NJ tree and STRUCTURE (Figs. 2, 3). Among the four genetic groups, both the G1 and G4 groups were the fly populations feeding on banana, corresponding to the 5^th and 10^th generations, and the 15^th generation, whereas G2 and G3 were the flies feeding on pumpkin and cucumber, respectively. These results revealed that genetic differences were most obvious between banana-feeding populations and populations feeding on the two traditional hosts, with no significant differences between the flies feeding on the two Cucurbitaceae hosts (Table 5). This genetic divergence between banana-feeding populations and the two Cucurbitaceae-feeding groups increased with the increasing number of generations (Tables 5, 7). The most obvious genetic differences observed between the two banana-feeding groups (Table 5) were from different generations of banana-feeding flies. In contrast, the genetic difference between two Cucurbitaceae host-feeding groups did not diverge more with subsequent generations. Perhaps the greater genetic divergence observed in the banana-feeding populations arose from a bottleneck effect while developing on this novel host.

In summary, the results of our study indicate that the novel host banana had a greater impact on genetic variation in B. tau than the two traditional Cucurbitaceae hosts. Genetic variation is considered an important mechanism by which B. tau can adapt to the selective pressures of a novel host. A similar phenomenon also has been found in other insect species. For example, genetic variation was observed between two populations of Ostrinia nubilalis (Hübner) (Lepidoptera: Pyralidae) that were fed on corn and a novel host Artemisia argyi (H. Lév. & Vaniot) (Martel et al. 2003). Significant genetic difference also was observed between butterfly populations of the tribe Nymphalini when feeding on its traditional host Urtica dioica L. and the novel host Ribes uva-crispa L. (gooseberry) (Celorio-Mancera et al. 2013).

Under laboratory conditions, we estimated the impact of various host plants on the genetic variation of B. tau while attempting to exclude the influences of other environmental factors such as natural enemies. We found that B.tau can use a novel food, such as banana, if it enters a new area where traditional hosts are not present. The use of novel host plants appears to be an ecological genetic strategy to improve fitness when dispersing into new regions where traditional hosts may be lacking.