Introduction

Drosophila suzukii (Matsumura) is a pest of soft fruit such as cherry, strawberry and blueberry. Females of this species lay eggs in ripening fruit before harvest and cause serious damage (Kanzawa 1939; Kimura et al. 1977; Nishiharu 1980; Mitsui et al. 2006, 2010; Lee et al. 2011; Walsh et al. 2011; Calabria et al. 2012), unlike many other fruit-feeding Drosophila species that lay eggs in damaged or rotting fruit. This species has its origin in the temperate and subtropical regions of Asia (Kanzawa 1939; Lemeunier et al. 1986), and has recently colonized North America and Europe (Hauser 2009; Walsh et al. 2011; Calabria et al. 2012). To develop an integrated management system for this pest species, information on natural enemies is very important (Walsh et al. 2011; Chabert et al. 2012). So far, Trichopria sp. (Hymenoptera, Diapriidae; cited as Phaenopria sp. in Kanzawa’s paper), Asobara japonica Belokobylskij, A. tabida Nees von Esenbeck (Hymenoptera, Braconidae) and Ganaspis xanthopoda Ashmead (Hymenoptera, Figitidae) have been reported to parasitize some frugivorous Drosophila species including D. suzukii (Kanzawa 1939; Mitsui et al. 2007; Ideo et al. 2008). In our parasitism experiments (Mitsui and Kimura 2010), however, Ganaspis individuals that emerged from D. lutescens Okada (a species breeding on fermenting fruit) that were identified as G. xanthopoda did not parasitize D. suzukii, suggesting that the Ganaspis individuals parasitizing D. suzukii and D. lutescens are different host races or host-specific species. To verify this possibility, we carried out field survey and parasitism experiments on Ganaspis individuals parasitizing D. suzukii and examined the molecular and morphological differences between individuals parasitizing D. suzukii and D. lutescens.

Materials and methods

Field survey

Wild cherry fruit (Prunus donarium Sieb.) is one of the major resources of D. suzukii in late spring and early summer in central Japan (Nishiharu 1980; Mitsui et al. 2006, 2010). This fruit (about 5 mm in diameter) changes color from yellow to black on the tree and then falls to the ground. Drosophila suzukii females mainly lay eggs in black fruits before they fall (Mitsui et al. 2006). In the present survey, black fruits on trees and those recently fallen on the ground were collected in wooded areas in Naganuma park located in the suburbs of Tokyo (35.6°N, 139.4°E) in June, 2010 and 2011. In 2010, fruits were collected from three sites (sites A, B and C) in this park, and from various sites in 2011. Collected fruits were brought back to the laboratory and placed in plastic containers with cloth or paper. When drosophilid larvae in the fruits pupated on the cloth or paper, they were collected, identified to species, placed on Petri dishes with wet filter paper, and examined for the emergence of flies or parasitoids. When parasitoids emerged, some were maintained alive in vials containing Drosophila medium to examine their host use and some were preserved in 100 % alcohol for molecular and morphological studies. In addition, Ganaspis individuals that emerged from D. suzukii pupae collected from wild cherry fruits in Sendai (38.2°N, 140.9°E), and those that emerged from D. lutescens pupae collected from banana baits placed in wooded areas in the suburbs of Tokyo and Sendai were used for molecular and morphological studies.

Parasitism experiments

To examine the host specificity of Ganaspis individuals, parasitism experiments were conducted using the following Drosophila species: D. suzukii, D. lutescens, D. rufa Kikkawa & Peng, D. auraria Peng, D. biauraria Bock & Wheeler, and D. triauraria Bock & Wheeler. These Drosophila species belong to the melanogaster species group (Lemeunier et al. 1986), and are major native fruit-feeding species in central Japan (Nishiharu 1980; Mitsui et al. 2010). Laboratory stocks of these Drosophila species originated from several females collected in and near Tokyo and were maintained on cornmeal-malt medium for a month to several years in the laboratory. Parasitism experiments were performed as follows. Five to 20 females of these Drosophila species were introduced into a vial containing a cherry fruit (Prunus avium L.) and allowed to oviposit for a day. Intact cherry fruit was provided for D. suzukii, whereas fruit pricked by needle was provided for the other five species since they do not oviposit in intact fruits. Two days later, five Ganaspis females and several males were introduced to the vial and allowed to oviposit. After 2 days, parasitoids were removed from the vial and emergence of flies and/or parasitoids from the vial was recorded. In addition, Ganaspis females were examined to see whether they parasitize D. suzukii larvae growing in cornmeal-malt medium (Drosophila medium) or not. Experiments were carried out in the same way as described above, except cornmeal–malt medium was provided rather than cherry fruit. The experiments were done under long-day conditions (15 h light–9 h dark) at 23 °C.

Molecular analysis

Nucleotide sequences of the mitochondrial COI (cytochrome oxidase subunit I) gene and two nuclear DNA regions, ITS1 (intertranscribed spacer sequence I of ribosomal RNA genes) and ITS2, were determined for Ganaspis individuals collected. DNA was extracted from each specimen by a modified phenol–chloroform protocol. All amplifications were preformed in 23 μl reaction volumes containing 1.3 mM MgCl2, 0.042 mM dNTP, 2.6 μM primers, 0.042 U Ampli Taq DNA polymerase, and 2.4 μl 10× PCR buffer. The PCR profile consisted of one cycle of denaturation (94 °C for 10 min), 35 cycles of denaturation (94 °C for 1 min), annealing (50 °C for 1 min) and extension (72 °C for 1.5 min), followed by one cycle of final extension at 72 °C for 12 min. Amplified products were diluted to 1 ng/μl, and used as sequencing templates.

Amplification of ITS1 and ITS2 fragments was performed by following primer pairs: 5′-GCTGCGTTCTTCATCGAC-3′ and 5′-CGTAACAAGGTTTCCGTAGG-3′ for ITS1 (412–647 bp); 5′-TGTCAACTGCAGGACACATG-3′ and 5′-AATGCTTAAATTTAGGGGGTA-3′ for ITS2 (448–564 bp). Amplification of COI fragments was preformed with a pair of primers, 5′-GGTCAACAAATCATAAAGATATTGG-3′ (LCO) and 5′-TAAACTTCAGGGTGACCAAAAAATCA-3′ (HCO) (about 600 bp) (Folmer et al. 1994).

For all sequence reactions, a Big Dye Terminator Cycle Sequencing Kit (ABI) was used. Sequencing was carried out with a 3100 Genetic Analyzer (ABI), utilizing the same primers used for PCR amplification.

Phylogenetic analysis of sequence data was conducted using Mega5 software (Tamura et al. 2011). Sequences were aligned manually and used to construct phylogenetic trees using the neighbor-joining (NJ) method (Saitou and Nei 1987). Nucleotide distances in NJ trees were estimated by Kimura’s two-parameter method (Kimura 1980). In the phylogenetic tree with COI, Leptopilina japonica japonica Novković & Kimura and L. heterotoma Thompson were used as the outgroup. Bootstrap values were obtained after 1,000 replications.

Morphometric analysis

The forewing and antenna were subjected to analysis. For the forewing, geometric morphometric analysis was conducted with 169 females (61 from D. suzukii and 108 from D. lutescens) and 153 males (46 from D. suzukii and 107 from D. lutescens). The right forewing of each specimen was mounted on a slide in Hoyer’s medium. Digitized forewings were assigned seven landmarks, positioned at forewing vein intersections and terminations (Fig. 1) using TpsDig 2.12 software (Rohlf 2008a). To examine forewing size differences among wasps, ANOVA was conducted on the centroid size parameter (the square root of the sum of squared distances between each landmark and the forewing centroid). In order to extract shape variables, raw coordinates were first superimposed by a generalized orthogonal least-square Procrustes (GPA) procedure, standardizing the size of landmark configurations and removing translational and rotational differences (Rohlf and Slice 1990). Next, the partial warps were calculated, and the obtained ‘weight matrix’ (w; Rohlf et al. 1996) was subjected to canonical discriminant analysis. Centroid size and weight matrix were obtained utilizing TpsRelw 1.46 software (Rohlf 2008b).

Fig. 1
figure 1

Position of the seven landmarks used in geometric morphometric analyses on the right forewing of a Ganaspis specimen

Full size image

In the antenna analysis, 140 females (52 from D. suzukii and 88 from D. lutescens) and 131 males (36 from D. suzukii and 95 from D. lutescens) were used. The digitized antenna of each specimen was measured for length of each segment using Photoshop CS5.1 (Adobe systems Incorporated, San Jose, USA). The proportion of each segment to total length was then calculated, and the obtained matrix was subjected to canonical discriminant analysis.

Statistical analyses were performed using Jmp ver. 6.1 (SAS Institute, Cary, USA).

Results

Field survey

In 2010, a total of 3,938 D. suzukii pupae were obtained from 3,892 wild cherry fruits collected; 367 pupae from 319 fruits on trees and 3,571 pupae from 3,573 fallen fruits (Table 1). In 2011, the numbers of fruits and pupae collected were not counted. Drosophila suzukii was the only species that bred on these fruits. In 2010, 26 Ganaspis individuals emerged from 367 D. suzukii pupae collected from fruits (the parasitism rate was about 7 %), and a total of 141 individuals emerged from 3,571 pupae collected from fallen fruits (the parasitism rate was about 4 %). In 2011, 112 Ganaspis individuals emerged. In addition to Ganaspis individuals, Leptopilina japonica japonica and unidentified Asobara species were recorded from D. suzukii pupae collected from fallen fruits.

Table 1 Individual numbers of parasitoids (G: Ganaspis individuals, Lj: Leptopilina japonica, Aspp: Asobara spp.) that emerged from D. suzukii pupae obtained from wild cherry fruits collected in Naganuma park, Tokyo, in June 2010 and 2011
Full size table

Parasitism experiments

Table 2 shows the results of parasitism experiments on Ganaspis individuals that emerged from D. suzukii pupae. In the experiment where D. suzukii was used as host, 129 Ganaspis individuals and 408 host flies emerged when cherry fruits were used, but no Ganaspis individuals emerged when Drosophila medium was used. In the experiment with the other five Drosophila species, no Ganaspis individuals emerged even when cherry fruits were used.

Table 2 Individual numbers of flies and parasitoids that emerged in parasitism experiments
Full size table

Molecular analysis

The ITS1 sequence was successfully obtained from 22 Ganaspis individuals out of 23 individuals used in the molecular analysis, whereas the COI and ITS2 sequences were obtained only from 12 and 15 individuals, respectively. Figure 2 shows the neighbor-joining phylogenetic trees based on these three nucleotide sequences. Irrespective of original locality (i.e., Tokyo or Sendai), Ganaspis individuals from D. suzukii and those from D. lutescens are discriminated by these sequences with high (>97 %) bootstrap support. Nucleotide divergence between them ranged from 4.5 to 5.7 % for COI, from 7.2 to 8.6 % for ITS1, and from 1.7 to 2.6 % for ITS2.

Fig. 2
figure 2

Neighbor-joining trees for Ganaspis individuals that emerged from D. suzukii and D. lutescens collected from Tokyo and Sendai based on COI (573 bp), ITS1 (334 bp) and ITS2 (357 bp) nucleotide sequences. Individuals are mentioned with the localities they are from (Tokyo or Sendai), the host species (in parentheses: D. suzukii or D. lutescens), individual numbers, and accession numbers (in parentheses). Leptopilina japonica japonica and L. heterotoma are used as the outgroup in the COI tree, and no outgroup was used in ITS1 and ITS2 trees. Bootstrap values are given above the branches supporting groups (values below 50 % are omitted)

Full size image

Morphological analysis

Figure 3 shows the distributions of canonical variate scores in the discriminant analyses on forewing shape and relative length of antennal segments of females and males. These traits were significantly different between Ganaspis individuals from D. suzukii and those from D. lutescens (ANOVA, P < 0.001). In particular, the overlap of the canonical variate scores in the antenna analysis was small. However, the overlap in forewing analysis was very large. Forewing centroid size (FC) and antennal length (AL) were significantly (ANOVA, P < 0.001) larger in individuals from D. suzukii [FC = 339.5 ± 17.4 in the female and 316.9 ± 19.9 in the male (mean ± SD); AL = 1.42 ± 0.07 in the female and 2.36 ± 0.16 in the male (mean ± SD in mm)] than in those from D. lutescens (FC = 284.0 ± 16.7 in the female and 272.4 ± 16.0 in the male; AL = 1.11 ± 0.06 in the female and 2.00 ± 0.12 in the male) in both sexes.

Fig. 3
figure 3

Distribution of the first canonical variate scores based on morphometric analyses of forewing and antenna shape

Full size image

Discussion

Ganaspis individuals from Drosophila suzukii parasitized D. suzukii larvae occurring in fresh cherry fruit but did not parasitize those in Drosophila medium, and further they did not parasitize larvae of some other fruit-feeding Drosophila species even if they occurred in cherry fruit. The field survey revealed that they parasitize D. suzukii larvae in wild cherry fruit on trees, but it was not determined whether they also parasitize D. suzukii larvae in fallen fruits or not. In contrast, Ganaspis individuals from D. lutescens parasitize larvae of D. lutescens, D. rufa, D. biauraria, D. triauraria and some other species occurring in Drosophila medium, but they did not parasitize D. suzukii larvae occurring in Drosophila medium (Mitsui and Kimura 2010). Thus, Ganaspis individuals from D. suzukii and D. lutescens do not overlap in host use, and the former seems to specialize on D. suzukii.

The Ganaspis individuals from D. suzukii and D. lutescens were clearly discriminated by nucleotide sequences of COI, ITS1 and ITS2 with high bootstrap value. In addition, they were statistically different in the morphology of forewing and antenna. These results suggest that the gene flow is limited between them. Although they might have diverged as different species, in the present paper they are tentatively assigned as the suzukii-associated and the lutescens-associated types of G. xanthopoda (since D. lutescens is the major host of the latter). To determine their species status, difference from or identity with the holotype specimen must be examined. For this purpose, however, morphological comparisons would not be sufficient, because the two types of G. xanthopoda reported in this study cannot be distinguished only by their morphology. It is necessary to examine the molecular differentiation and reproductive isolation of the two types of G. xanthopoda in this study from individuals occurring at the type locality, i.e., Grenada of the Caribbean region (Ashmead 1896). This remains for future study.

Trichopria sp., Leptopilina japonica and some Asobara species also parasitize D. suzukii as observed in the present and previous studies (Kanzawa 1939; Mitsui et al. 2007; Ideo et al. 2008), but they are not specialists on D. suzukii (van Alphen and Janssen 1982; Carton et al. 1986; Mitsui et al. 2007; Ideo et al. 2008; Novković et al. 2011). Therefore, the suzukii-associated type of G. xanthopoda would be the best agent for biological control of D. suzukii.