Molecular Placement of an Outbreak-Causing Gall Wasp, Zapatella davisae (Hymenoptera: Cynipidae), with Comments on Phylogenetic Arrangements in the Tribe Cynipini

Gall wasps (Hymenoptera: Cynipidae) are a worldwide family with over 1,300 described species that create galls on a variety of plant tissues and plant species (Rokas et al. 2002, Stone et al. 2002). In addition to the diversity of gall morphologies that have evolved within this group, these insects have fascinated researchers due to their multitrophic interactions (e.g., Nicholls et al. 2017). Unfortunately, inconsistencies regarding morphological characteristics used to define genera of gall wasps have complicated the taxonomic placement of species in this group and have obscured our understanding of their evolutionary relationships (see Melika and Abrahamson 2002). Clarification of cynipid taxonomy is particularly important given that both introduced (e.g., Cooper and Rieske 2007; Schönrogge et al. 1995) and native (e.g., Eliason and Potter 2000; Pujade-Villar et al. 2014) gall wasp species separated from their regulating natural enemies have the potential to break out at high densities and cause extensive host plant damage and mortality.

As in many taxonomic groups, the reconstruction of ancestral relationships among gall wasp species has benefited from the sequencing of molecular markers (e.g., Nicholls et al. 2017, 2018a, 2018b; Rokas et al. 2002; Ronquist et al. 2015). One genus of gall wasp absent from previous molecular studies is Zapatella Pujade-Villar & Melika (Pujade-Villar et al. 2012). This genus was erected following the description of a new species of oak gall wasp, Zapatella grahami Pujade-Villar & Melika, collected in Costa Rica, that did not clearly fit the descriptions for existing genera. Included in this genus is the recently described black oak gall wasp, Z. davisae Buffington & Melika, that has caused extensive damage and mortality to black oaks, Quercus velutina Lam. (Fagales: Fagaceae), in the northeastern United States (Buffington et al. 2016; Davis et al. 2017). While DNA sequencing of Z. davisae individuals collected on Long Island, NY, and Cape Cod, MA, highlighted that these individuals are related to unidentified specimens with published DNA sequences from the southeastern and northeastern parts of North America (Davis et al. 2019), it is unknown how this species is related phylogenetically to other described species in the Cynipini. For example, Pujade-Villar et al. (2012) note that Zapatella is closely related to species of Cynipini in the genera Callirhytis Förster, Bassettia Ashmead, and Plagiotrochus Mayr (in fact, they transferred several “problematic” species in each genus to Zapatella in an effort to address some historical confusion related to these species). However, previous molecular analyses (Melika et al. 2010) indicate that Callirhytis and Plagiotrochus belong to separate subgroups within the Cynipini (identified as the “Cerris” and “Quercus” sections by the authors, respectively). Furthermore, Callirhytis appears to be wildly polyphyletic (Nicholls et al. 2017). Therefore, it is still unclear where Zapatella (and by association Z. davisae) fits within the tribe Cynipini.

The purpose of this study was to determine the placement of Z. davisae by reconstructing the phylogenetic relationships of species in the tribe Cynipini. We did this by generating novel sequences from individuals collected on Long Island, NY, and Cape Cod, MA, and comparing them to publicly available sequences. Using multilocus phylogenetic reconstructions, we comment on the relationship of Z. davisae to other sampled species, and on taxonomic relationships within the Cynipini more broadly.

Specimen collection. Specimens of Z. davisae for molecular analysis were collected on Cape Cod, MA, and on Long Island, NY, in early spring 2015 as part of a larger study of factors influencing outbreaks in the northeastern United States (Davis et al. 2017, 2019). To collect specimens, black oak branches with noticeable stem galls were trimmed from their host trees, placed in a 3.8-L (1-gallon) zip-lock bag and stored at 4°C in a growth chamber (Percival Scientific Inc., Perry, IA) for 6 weeks until adult emergence. After emergence, adult gall wasps and associated species (i.e., parasitoids or inquilines) were removed from the bags, placed in separate 1.5-µl microcentrifuge tubes with 99% ethanol, and stored at room temperature. While both parasitoids and inquilines were collected, DNA was subsequently extracted only from specimens suspected to be the primary gall former. Voucher specimens are housed at the National Museum of Natural History, Smithsonian Institution, Washington, DC, under the following accession umbers USNMENT01119082-3, USNMENT01119700-39, and USNMENT01119755-810 (Buffington et al. 2016).

DNA extraction and amplification. DNA was extracted from individuals using the Qiagen DNeasy kit (QIAGEN, Valencia, CA) following the manufacturer's instructions. For each specimen, fragments of the nuclear loci 28S and long-wavelength opsin (LWRh) were amplified. The 28S gene was amplified using two pairs of primers, 28SF (5′–AGTCGTGTTGCTTTGATAGTGCAG–3′) with 28Sbout (5′–CCCACAGCGCCAGTTCTGCTTACC–3′) or 28SFAf (5′–GGTACTTTCAG GACCCGTCTT–3′) with 28Sin1 (5′–ACCTTCACTTTCATTAYGCCTTTA–3′) and the fragment of LWRh was amplified using the forward primer LWRhF (5′–AATTGCTATTAYGARACNTGGGT–3′) and the reverse primer LWRhR (5′–ATATGGAGTCCANGCCATRAACCA–3′), according to the thermocycler protocols described in Rokas et al. (2002). Polymerase chain reaction (PCR) products for each locus were then visualized on 1.5% agarose gels and, prior to sequencing, products were purified using Exonclease 1 (Thermo Scientific) and shrimp alkaline phosphatase (New England BioLabs) according to the Thermo Scientific PCR and purification protocol. Cleaned PCR products were sequenced at the DNA Analysis Facility on Science Hill at Yale University using an ABI 3730 sequencer (Life Technologies). Forward and reverse sequence reads were then aligned and edited using Geneious 11.1.2 (Kearse et al. 2012), and a consensus sequence was generated for each sample. In addition, we obtained a sequence of the mitochondrial locus cytochrome b (cytb) for Z. davisae that was previously published in GenBank (accession: SRX2830002).

Alignment and concatenation. Previously, Davis et al. (2019) sequenced a fragment of the mitochondrial locus cytochrome oxidase I from 54 Z. davisae individuals collected on Cape Cod and 27 Z. davisae individuals collected on Long Island, and found all individuals could be assigned to a single invariant haplotype. Here, we also found that our fragments of 28S and LWRh were invariant across sequenced individuals and thus present a single consensus sequence for Z. davisae for each locus.

To determine the phylogenetic placement of Z. davisae, we queried GenBank to obtain sequences for 28S, LWRh, and cytb from all species of Cynipini for which all three loci were available. Individual locus alignments including downloaded sequences and the representative sequence from Z. davisae were then constructed using MUSCLE (Edgar 2004) as implemented through the EMBL-EBI web service (Li et al. 2015). Each alignment was evaluated to include only a single sequence for each species using the following approach. First, a distance tree was generated using the neighbor-joining distance algorithm implemented in Genenious, and any sequence that did not form a clade with other sequences from the same species was removed. Sequences were then evaluated for length, with the longest sequence(s) for each species being retained and then evaluated for quality, with the sequence(s) with the fewest number of “N” base calls being retained. If multiple sequences for a species were still present, we retained the sequence that came first alphabetically for subsequent analyses. The alignment for each locus was truncated to the length of the shortest sequence, and a concatenated alignment was created using MESQUITE v. 3.04 (Maddison and Maddison 2017). The hypervariable regions of 28S and two introns located within the fragment of LWRh were excluded prior to phylogenetic analyses. For published sequences, species names were updated based on Melika and Abrahamson (2002), Melika et al. (2010), and Pujade-Villar et al. (2014), when necessary.

Phylogenetic analyses. Bayesian phylogenetic reconstructions were then estimated using MrBayes 3.2.7 (Huelsenbeck and Ronquist 2001; Ronquist et al. 2012). A seven-partition scheme was used, with one partition for 28S, and one partition for each codon position for LWRh and cytb, respectively. Markov chain Monte Carlo searches were run for 10,000,000 generations, and a burn-in of 25%, with default heating values as implemented through the CIPRES Science Gateway (Miller et al. 2010), and the phylogenetic reconstruction was then visualized in FigTree version 1.4.2 (Rambaut and Drummond 2009). Maximum likelihood phylogenetic reconstructions were then estimated using RAxML v. 8.2.12 (Stamatakis 2014) using default settings and support was estimated using 100 bootstrap replicates under the GTRCAT model as implemented in the CIPRES Science Gateway.

DNA amplification and alignment. In total, we amplified fragments of 28S and LWRh from 14 Z. davisae specimens, 10 collected on Cape Cod and 4 collected on Long Island. As noted above, all sequences were invariant at each sampled locus. Genotype sequences for 28S and LWRh are published in GenBank with the accession numbers MK089551and MK089552, respectively. In total, we included sequences from an additional 78 species of Cynipini. The alignment for 28S included 451 base pairs (after truncation), the alignment for the LWRh exons included 395 base pairs, and the alignment of cytb included 291 base pairs (after truncation). A complete list of all species, including their GenBank accession numbers, is presented in Table 1.

Phylogenetic analyses. The best tree from the RAxML analysis is presented in Fig. 1, and had a negative log-likelihood score of –11542.64. After 10 million generations, the average standard deviation of the split frequencies for the final analysis was 0.0067 for the MrBayes reconstruction. Both methods of reconstruction (maximum likelihood and Bayesian) converged on similar topologies, with no supported nodes in conflict (Bayesian posterior probability [BPP] and bootstrap percentage of support [BPS] ≥0.75 or 75%, respectively). Support values are presented graphically in Fig. 1 (bold lines indicate support from both analyses, dashed lines indicate support from one analysis, and narrow lines indicate no support from either analysis).

Regarding the placement of Z. davisae, our reconstructions indicate that it is member of the Cynipini section “Quercus” (1.00 BPP and 100% BPS). Within this group, the maximum likelihood reconstruction indicated that Z. davisae is a basal member of this group, most closely related to species of Callirhytis and Neuroterus, though this relationship was not well supported (≤0.50 BPP) as the Bayesian analysis placed Z. davisae in a polytomy that also included most members of this section.

Gall wasps have long fascinated researchers because of the diversity of plant-induced structures they produce and their unique evolutionary adaptations (Mayr 1881; Stone et al. 2002). Consequently, numerous phylogenies for this group have been proposed based on morphological and molecular characters (e.g., Kinsey 1920; Liljebad and Ronquist 1998; Rokas et al. 2002; Ronquist 1994; Ronquist et al. 2015). Here, we build on these previous studies and reconstruct relationships for 79 species of gall wasps in the tribe Cynipini and find that the outbreak-causing black oak gall wasp Z. davisae is placed clearly within the “Quercus” section of the tribe Cynipini (Fig. 1) and appears to be a basal member of this group. This placement of Z. davisae supports the finding of Pujade-Villar et al. (2012) that Zapatella is likely a valid and distinct genus of Cynipini; however, our results are less clear about the authors' statement that members of this genus are closely related to species of Bassetia, Callirhytis, and Plagiotrochus. While our results suggest a relationship between Zapatella and Callirhytis, our analysis and the analysis of Melika et al. (2010) indicate that Plagiotrochus belongs to a distinct and different section of the Cynipini that includes only genera that form galls on oaks in the Quercus subgenus Cerris (1.00 BPP, 100% BPS, our analysis). In contrast, Z. davisae is assigned to the “Quercus” section (following Melika et al. 2010) that includes species that form galls mostly on oak species in the Quercus subgenus Quercus (Csóka et al. 2005). We were unable to include sequences for Bassettia in our analyses, as no public sequences of 28S or LWRh are available; however, a sequence of cytb was recently published in GenBank (Nicholls et al. 2018a) and based on our post-hoc analysis using neighbor-joining for only cytb sequences, these two genera appear to be closely aligned. The inclusion of sequences from additional loci in this analysis, or other species of Bassetia could help to further refine our understanding of the relationship between Zapatella and other “Quercus” section species. Future studies should also seek additional samples from other species of Zapatella and other recently described Nearctic genera from which members of Callirhytis have been reassigned (e.g., Melikaiella: Pujade-Villar 2014) and other New World genera (e.g., Dros Kinsey; Pujade-Villar et al. 2017) to further clarify relationships between members of this confusing group of species of Cynipini.

Comments on generic reclassifications within the Cynipini. While much work has been done recently to address the generic assignments of species within the oak gall wasp tribe Synergini (e.g., Ács et al. 2010; Bozsó et al. 2014, 2015; Pénzes et al. 2009; Schwéger et al. 2015a, 2015b) and the Cynipini (e.g., Melika and Abrahamson 2002; Melika et al. 2010; Pujade-Villar et al. 2012, 2014), generic assignments in the Cynipini continue to be problematic. Based on our analyses, only four genera with multiple species were reconstructed as monophyletic (Besbicus Weld, Cynips L., Heteroecus Kinsey, Plagiotrochus Mayr, and Pseudoneuroterus Kinsey). One additional genus, Amphibolips Reinhard, could also be considered monophyletic if the generic assignment presented in Melika and Abrahamson (2002) is used for Amphibolips quercuspomiformis (Bassett). However, this species was recently transferred from Amphibolips to Callirhytis by Pujade-Villar et al. (2014). Both of our analyses indicate that the assignment of this species to Amphibolips might be the more accurate, as it was reconstructed as sister to Amphibolips quercusjuglans (Osten Sacken). Lastly, our results support the recent reassignment of Andricus spectablis (Kinsey) to the newly described genus Protobalandricus Melika et al. (Nicholls et al. 2018b), as this species is not aligned with other members of the genus Andrius (Hartig).

Conclusions. Our multilocus concatenated phylogenetic reconstructions suggest that Zapatella davisae is a basal member of the “Quercus” section of the tribe Cynipini. Our results also highlight that additional phylogenetic work focused on generic assignments is still required. Lastly, we encourage oak gall wasp researchers to publish DNA sequences in public databases to aid global efforts to resolve species assignments in this important and fascinating group. Throughout our research, we frequently encountered authors discussing the results of unpublished phylogenetic analyses. This lack of public data-sharing is unfortunately preventing further progress on the study of the evolution and life cycles of these fascinating organisms.

This work could not have been possible without the assistance of D. Gilrein from the Cornell Horticultural Station for help finding field sites in Long Island and E. Mooshain, C. Camp, G. Hespeler, E. Lee, E. Kelly, and Q. Duepont for specimen collection and DNA extraction work. We also thank J. Boettner, H. Broadley, B. Griffin, R. Gwiazdowski, and M. Labbé, as well as W.A. Gardner and two anonymous reviewers for constructive feedback on earlier drafts of this manuscript. Funding was provided by grants from the Woodbourne Arboretum and Arborjet Inc.