Introduction
One of the biggest challenges to characterising parasitoid communities is the existence of cryptic species. Many congeneric parasitoid species are morphologically similar or identical (Hayward et al. Reference Hayward, McMahon and Kathyrithamby2011; Zhang et al. Reference Zhang, Gates and Shorthouse2014; Hall et al. Reference Hall, Steinbauer, Taylor, Johnson, Cook and Riegler2017). Consequently, community characterisations are, of necessity, often limited to genus or morphological species-group levels of accuracy. Such lumping can collapse species with different life histories into the same category and confound attempts to elucidate factors that affect community dynamics (Hrček and Godfray Reference Hrček and Godfray2015). The development of inexpensive and user-friendly techniques for extracting and amplifying DNA and the ready availability of published gene sequences at such sites as the Barcode of Life Data System (Canadian Centre for DNA Barcoding, Biodiversity Institute of Ontario, University of Guelph, Guelph, Ontario, Canada) have greatly increased the accuracy and ease of species identification (Hayward et al. Reference Hayward, McMahon and Kathyrithamby2011; Zhang et al. Reference Zhang, Gates and Shorthouse2014; Hrček and Godfray Reference Hrček and Godfray2015; Hall et al. Reference Hall, Steinbauer, Taylor, Johnson, Cook and Riegler2017), although care must be taken to ensure the accuracy of the original identification for published barcodes.
Cynipid wasps of the genus Diplolepis Geoffroy (Hymenoptera: Cynipidae) induce galls on various tissues of host plants in the genus Rosa Linnaeus (Rosaceae) (Shorthouse Reference Shorthouse, Shorthouse and Floate2010). In the Okanagan Valley of south–central British Columbia, Canada, Rosa woodsii is attacked by seven relatively common Diplolepis species (Lalonde and Shorthouse Reference Lalonde and Shorthouse2000; Shorthouse Reference Shorthouse, Shorthouse and Floate2010). Two such species, D. rosaefolii Cockerell and D. variabilis Bassett, are among the most common and develop at more or less the same time during the late summer and early fall in the area. Both induce galls in leaf tissue and often co-occur on the same plant (Fig. 1). Because of this, both species are available to the same guild of parasitoids at more or less the same time. Although these galls are induced by congenerically close relatives, their shapes are markedly different. Diplolepis variabilis induces a highly polymorphic gall composed of soft corky material and ranges from single-inhabitant forms that are the size of a pea to large, amorphous structures formed from the coalesced galls of many individuals (Fig. 1). In contrast, D. rosaefolii induces much smaller, single-inhabitant, lenticular galls composed of highly sclerotised walls that have a uniform shape and rarely coalesce (Fig. 1).
Fig. 1. Galls induced by A, Diplolepis variabilis and B, D. rosaefolii on the same leaf of a Rosa woodsia ramet.
Cynipid-induced galls, in general, are host to a large diversity of hymenopterous inquilines – strategists that exploit the gall but not the gall inducer as a food source – and parasitoids (Shorthouse Reference Shorthouse1973; Stone and Cook Reference Stone and Cook1998). Galls of Diplolepis are no exception (Brooks and Shorthouse Reference Brooks and Shorthouse1997; Shorthouse Reference Shorthouse, Shorthouse and Floate2010; Bannerman et al. Reference Bannerman, Shorthouse, Pither and Lalonde2012). All Diplolepis species in the region induce galls that support at least one species of parasitoid wasp in the genus Eurytoma Illiger (Hymenoptera, Chalcidoidea) (Shorthouse Reference Shorthouse, Shorthouse and Floate2010). Parasitoids in the genus Eurytoma are fairly flexible in their feeding habits and will feed on larval Diplolepis, gall nutritive tissue, and also on other gall inhabitants such as larvae of the inquiline species in the genus Periclistus Förster (Hymenoptera: Cynipidae) (Brooks and Shorthouse Reference Brooks and Shorthouse1997; Leggo and Shorthouse Reference Leggo and Shorthouse2006). Accurate identification of Eurytoma species using classical morphological characters is challenging, making it historically difficult to determine the degree to which these wasps specialise on particular hosts (Zhang et al. Reference Zhang, Gates and Shorthouse2014). However, a recent study that combines classical morphology and molecular characterisation has demonstrated that at least four closely related species, E. calcarea Bugbee, E. iniquus Bugbee, E. longavena Bugbee, and E. imminuta Bugbee (= E. spongiosa 1; Zhang et al. Reference Zhang, Gates and Shorthouse2017), inhabit D. variabilis galls in the Okanagan Valley. Two of these, E. iniquus and E. longavena, also are known to inhabit D. rosaefolii galls in Ontario, but the species of Eurytoma attacking galls of this species in British Columbia have not been investigated (Zhang et al. Reference Zhang, Gates and Shorthouse2014).
Given the high degree of lifecycle synchrony and spatial concordance between Diplolepis variabilis and D. rosaefolii in the Okanagan, we hypothesised that these two species will be attacked by the same species of Eurytoma at a given locality and that locality will be a better predictor of Eurytoma species presence than the species of host. In this study, we therefore sampled co-occurring galls of D. variabilis and D. rosaefolii from a number of sites in the Okanagan Valley, British Columbia and, using molecular tools to identify individuals to species, asked whether or not the assemblage of Eurytoma species present in the samples was affected by host species or by geographic location.
Methods
Collections were made during autumn 2009. Each rose bush located within a radius of 10 m represented one sampling site, and the site’s position was recorded using a geographic positioning system. Thirty-two sites were sampled for galls of both D. variabilis and D. rosaefolii. A sample of 15–30 galls of each species was collected at each site and taken back to the lab. Galls were refrigerated and removed only just before dissection. Material from 5 of the 32 sites (sites 5, 7, 9, 10, and 28), spanning a distance of 100 km from near Osoyoos to the Kelowna airport (Fig. 2), was used in this study. The sites used were those with recognisable Eurytoma larvae in both D. variabilis and D. rosaefolii galls. Eurytoma larvae dissected from galls of D. rosaefoliii and D. variabilis were first identified using published illustrations (Shorthouse Reference Shorthouse1973; Leggo and Shorthouse Reference Leggo and Shorthouse2006), then reserved for DNA extraction. In total, DNA was extracted from each of 56 samples using a DNeasy® Blood and Tissue kit (Qiagen, Inc., Valencia, California, United States of America).
Fig. 2. Map of sampling locations for material used in this study. Galls of both Diplolepis variabilis and D. rosaefolii were collected at sites near Kelowna (sites 5 and 7), Peachland (site 28), and Osoyoos (sites 9 and 10), British Columbia, Canada.
The extracted DNA was used to amplify a 433-bp segment of the cytochrome b mitochondrial gene, and each clean amplified sample was sequenced. The primers used were CB1 (5′–3′TATGTACTACCATGAGGACAAATATC) and CB2 (5′–3′ATTACACCTCCTAATTTATTAGGAAT) (Jermiin and Crozier Reference Jermiin and Crozier1994; Stone and Cook Reference Stone and Cook1998; Hall et al. Reference Hall, Steinbauer, Taylor, Johnson, Cook and Riegler2017). The Go Taq, Green Master Mix (Promega Corporation, Madison, Wisconson, United States of America) procedure was used for polymerase chain reaction (PCR). The PCR cycle consisted of 3 minutes at 94 °C, followed by 35 cycles of denaturation for 1 minute at 94 °C, annealing for 0.5 minute at 45 °C, and extension for 1 minute at 72 °C. At the end of the 35 cycles, a final 10 minutes at 72 °C occurred, and samples were then held at 4 °C until recovery. Gel electrophoresis was used to confirm amplification. Samples showing clear bands at 433 bp were then sequenced in both forwards and reverse directions. Sanger sequencing was performed by the University of British Columbia’s Fragment Analysis and DNA Sequence Service (FADSS; Kelowna, British Columbia, Canada). We then assembled forwards and reverse sequences using Codon Code Aligner (Version 7.0.1; Codon Code Corporation, Centerville, Massachusetts, United States of America) and saved the resulting contigs. Contig sequences that had quality values below 50% were removed from the dataset, and the remaining contigs were aligned using the Muscle alignment option in Seaview 4.7 (Gouy et al. Reference Gouy, Guindon and Gascuel2010). Before tree construction, the Blast function on GenBank was used, and any sequences that did not cluster within the genus Eurytoma were discarded. We then downloaded cytochrome b sequences of E. adleriae Zerova – a parasitoid of various gall-inducing hosts in Europe and Asia – and Bruchophagus caucasicus Zerova – a eurytomid in the same subfamily as the genus Eurytoma – from GenBank to use as outgroups. This entire process resulted in 21 high-quality sequences. These and the outgroup sequences were used to construct a maximum likelihood tree using Seaview 4.7 (Gouy et al. Reference Gouy, Guindon and Gascuel2010), per the steps described here.
The cytochrome b sequences were assigned to described species using published cytochrome oxidase I (COI) barcodes from Zhang et al. (Reference Zhang, Gates and Shorthouse2014). To do this, material sequenced by Earley et al. (unpublished) was referenced. Earley et al. had extracted DNA from a large collection of Okanagan Eurytoma reared from galls of D. variabilis and amplified both the cytochrome b and COI mitochondrial gene regions for each extracted specimen. Earley et al. were thus able to assign their cytochrome b sequences to barcoded species using the associated COI sequences. In the current study, a maximum likelihood tree was constructed using Earley et al.’s (unpublished) cytochrome b-sequenced material, together with the sequences from the current study, using Seaview 4.7 (Gouy et al. Reference Gouy, Guindon and Gascuel2010). With the exception of one sequence, the current study’s material could be assigned to a described and barcoded species with 100% bootstrap confidence.
Results and discussion
Our calibrated cytochrome b tree shows that the valley supports two well-supported clades of Eurytoma that parasitise the inhabitants of galls of both Diplolepis variabilis and D. rosaefolii (Fig. 3), with no evident preference by either parasitoid for a particular host (X 2 = 1.65, 1 df, P > 0.05) or for a particular location (see the second point in the paragraph below).
Fig. 3. Maximum likelihood tree of Eurytoma samples dissected from galls of D. variabilis (light brown circles) and D. rosaefolii (red circles). Numbers in circles indicate the collection site, as mapped in Fig. 2. The tree was constructed with a bootstrapped (1000 iterations) maximum likelihood routine on Seaview (Gouy et al. Reference Gouy, Guindon and Gascuel2010) using the cytochrome b sequences obtained from this study’s samples and outgroups (E. adleriae, Bruchophaga caucasicus) downloaded from GenBank (https://www.ncbi.nlm.nih.gov/nuccore?db=Nucleotide). Values at major nodes indicate percent bootstrap branch support. Species assignments were made by using the published cytochrome oxidase I (CO1) barcodes associated with our cytochrome b sequences per Earley et al.’s unpublished data.
A number of interesting points emerged from the analysis. Firstly, the Eurytoma species that attack D. rosaefolii in the Okanagan Valley differ from those that attack this host in Ontario (Zhang et al. Reference Zhang, Gates and Shorthouse2014, Reference Zhang, Gates and Shorthouse2017). In particular, E. imminuta (Zhang et al.’s (Reference Zhang, Gates and Shorthouse2014) E. spongiosa 1), the most ubiquitous species in the samples collected, was not recorded by Zhang et al. (Reference Zhang, Gates and Shorthouse2014, Reference Zhang, Gates and Shorthouse2017) as attacking the inhabitants of galls of D. rosaefolii. Secondly, although one of the species, Eurytoma longavena, was present only in samples taken from southern and central sites (sites 9, 10, and 28), while the other identified Eurytoma species, E. imminuta, is distributed amongst the galls of both host species across the entire sampled area in the valley, this apparent local restriction of E. longavena does not significantly differ from a random expectation even when we group sites 9 and 10 (southern) and sites 5 and 7 (northern) to reduce the number of categories (X 2 = 3.333, 2 df, P > 0.05). Thirdly, one of our sequences could not be assigned to a species (Fig. 3), suggesting that there is at least one nonbarcoded and possibly undescribed Eurytoma species in the valley. Finally, Zhang et al. (Reference Zhang, Gates and Shorthouse2014, Reference Zhang, Gates and Shorthouse2017) report a number of other Eurytoma species present within galls of both D. variabilis and D. rosaefolii that were not found in our samples.
It should be noted that Zhang et al.’s (Reference Zhang, Gates and Shorthouse2014) sample locations for both D. variabilis (Kelowna Airport, Kelowna, British Columbia) and D. rosaefolii (Ontario) differ from the sample locations used in this study, although their sample of galls of D. variabilis did come from a location that was within 500 m of our sites 5 and 7. The above shows that the composition of the Eurytoma portion of the parasitoid communities associated with Diplolepis gall-formers is diverse and possibly affected by the local pool of available species. No invariant host species-specific assemblage of parasitoids occurs, at least when it comes to Eurytoma. This was a result hinted at by Bannerman et al. (Reference Bannerman, Shorthouse, Pither and Lalonde2012) for the parasitoid assemblage attacking inhabitants of D. variabilis galls in the Okanagan and demonstrated by Aebi et al. (Reference Aebi, Schonrogge, Melika, Alma, Bosio, Quacchia, Picciau, Abe, Moriya, Yara, Seljak, Stone, Ozaki, Yukawa, Ohgushi and Price2006) for invasive populations of the chestnut gall wasp at different global locations. The finding is also consistent with a study in Idaho, United States of America, where the character of the vegetation surrounding a local Diplolepis community affected the diversity of the parasitoid community (Looney and Eigenbrode Reference Looney and Eigenbrode2011).
We found no apparent subdivision of the Eurytoma parasitoid community on the basis of host species in our samples. However, this is only at the level of discrimination afforded by sequencing the cytochrome b region and only across two of the approximately half-dozen galls of species of Diplolepis that can be readily found in the Okanagan Valley (Lalonde and Shorthouse Reference Lalonde and Shorthouse2000). One next step will be to sample Eurytoma from galls of other Diplolepis species to determine whether subdivision of the parasitoid community occurs across other hosts. In addition, the use of more variable genetic materials, such as single-nucleotide polymorphisms, would help to determine whether subdivision occurs over a shorter timescale than can be demonstrated by the variation present in cytochrome b (Hopper et al. Reference Hopper, Oppenheim, Kuhn, Lanier, Hoelmer, Heimpel, Meikle, O’Neil, Voegtlin, Wu, Woolley and Heraty2019).
If further investigation demonstrates that at least some of the species in the Eurytoma complex do not discriminate amongst galls induced by Diplolepis, the Diplolepis–Rosa system could be a useful model for experimentally examining the effects of multiple host–parasitoid dynamics (Holt Reference Holt1977; Morris et al. Reference Morris, Lewis and Godfray2004). In the Okanagan Valley, galls induced by D. variabilis, D. rosaefolii, and some of the other Diplolepis species present in the valley, are convenient subjects for such a study: they are abundant and show high site constancy. Such systems lend themselves to experimentation because of the ease of manipulating and re-visiting individual galls (Fernandes and Price Reference Fernandes and Price1992; Price et al. Reference Price, Abrahamson, Hunter and Melika2004). The marked persistence of populations of some Diplolepis species, such as D. variabilis, and the transience shown by other Diplolepis species (Lalonde and Shorthouse Reference Lalonde and Shorthouse2000) suggest that Diplolepis–Rosa systems may be useful models for investigating factors that affect the stability of host–parasitoid systems in general (Holt Reference Holt1977; Morris et al. Reference Morris, Lewis and Godfray2004; Van Veen et al. Reference Van Veen, Morris and Godfray2006).
Acknowledgements
We thank Michael Russello for advice given at several points along the way, J.D. Shorthouse for help in identifying the larvae, Mahsa Amirabbasi for doing the sequencing, and Morgan Hoffman and Rosemary Garner for help with the extractions. Y.M. Zhang and J.D. Shorthouse kindly provided constructive comments on an earlier draft of this paper. This work was partially funded by an NSERC Discovery Grant to R.G.L.
Open access
