On June 19, 2018, a collection trip was undertaken to replenish the colony of Paratelenomus saccharalis at the Florida Department of Agriculture and Consumer services – Division of Plant Industry (FDACS-DPI) in Gainesville, Florida. A patch of kudzu, Pueraria montana var. lobata (Lour.) Merr. (Fabaceae) located in Alachua County, Florida (29.8055°N, 82.5301°W) was chosen as the collection site due to its ease of access and favorable conditions for P. saccharalis development. At the site, kudzu foliage bearing M. cribraria eggs was carefully cut and collected to be evaluated for parasitism at the lab.

Of these M. cribraria egg masses, 47 were identified as having an irregular, ash-grey colouration, a common sign of parasitism (Gardner et al. 2013). These egg masses were separated from the others and placed in a 40 cm x 30 cm x 30 cm plexiglass growth chamber supplied with honey-soaked strips of paper and water for the adult parasitoids to access ad libitum. The growth chamber was maintained at 25–27°C, 70–82% humidity and on a 16L:8D light cycle.

Thirty Ooencyrtus nezarae adults were separated into three glass vials (V = 31 cm³ ; 10 parasitoids in each with an unknown sex ratio), each containing one M. cribraria egg mass and a honey-soaked strip of paper. Egg masses were obtained from captive colonies and stored in a freezer for 6 – 14 months before exposure. These vials were then sealed and maintained at 25 – 27°C, 60 – 78% humidity with a 16L:8D light cycle. The remaining parasitoids in the growth chamber were provided with four previously frozen, lab-reared M. cribraria egg masses and were maintained under the conditions described above.

Adult O. nezarae emergence in the glass vials was first observed 14 days later and continued for several more days. Once it was determined that adult parasitoids were unlikely to continue emerging, the 122 adults that emerged were individually placed in separate gelatine capsules and sorted by sex, resulting in 36 males and 86 females. Of these individuals, 32 males and 72 females were released into the eight glass vials, with each vial containing four males and nine females. Each new glass vial was provided with a new, previously frozen, lab-reared M. cribraria egg mass, to further evaluate rates of parasitism and suitability for rearing. The remaining adults were returned to the plexiglass growth chamber to be used as a maintenance colony.

Specimens were point-mounted and deposited in the Florida State Collection of Arthropods (FDACS-DPI, E2018-4411-1). Two specimens (one male and one female) were cleared in potassium hydroxide and slide-mounted in Canada balsam, according to a protocol modified from Noyes (1982). Whole-body images were taken with a Macropod® imaging system using a Canon EOS 6D Mark II camera, a EF 70-200mm lens and a 20X M Plan APO Mitutoyo objective lens. Image stacks were rendered using Helicon Focus® and edited in Adobe Photoshop®. Mandible images were taken on a Zeiss Axio Imager M2 Microscope with Axiocam 503 colour camera and Zen 2 software, using differential interference contrast lighting. Mandible image stacks were rendered using Zerene Stacker.

Six O. nezarae individuals from the DPI colony were selected for DNA extraction and COI barcoding. Samples were collected directly into 70% EtOH and were air-dried two hours before proceeding to DNA extraction. DNA extractions of entire individuals were performed using DNeasy Blood and Tissue Kits (Qiagen). Extraction protocols followed the manufacturer’s recommendations with one minor adjustment: samples were not incubated for 10 minutes following tissue lysis and addition of Buffer AL. DNA extracts were quantified using a NanoDrop 2000 spectrophotometer (Thermo Scientific). At least 20 ng of template DNA was used per PCR.

The 5’-COI barcode region was PCR-amplified using the primers LEP-F1 and LEP-R1 (Hebert et al. 2004). PCRs were performed at 25 µl volumes using HiFi HotStart DNA Polymerase (Kapa Biosystems). PCR thermocycle conditions were: 1) initial denaturing at 95°C for 2:00 minutes followed by 32 cycles of steps 2–4, 2) 98°C for 20 seconds, 3) 40°C for 30 seconds, 4) 72°C for 30 seconds and 5) final extension at 72°C for 7:00 minutes. PCR products were verified by gel electrophoresis and cleaned for sequencing with QIAquick Gel Extraction Kits (Qiagen). Purified PCR products were Sanger sequenced in both directions using BigDye Terminator v3.1 (Applied Biosystems) chemistry on a SeqStudio Genetic Analyzer (Applied Biosystems). Sequence reads were trimmed and sequence contigs were assembled in Sequencher 5.4.6 (Gene Codes Corporation). COI barcodes generated during this study were deposited in GenBank (accessions MH996994-MH996999).

To evaluate our new sequences in the context of the subfamily, all 5’-COI barcodes for Ooencyrtus and Encyrtinae were mined from the Barcode of Life Database (BOLD) (Ratnasingham and Hebert 2007). The tetracnemine species Aenasius bambawalei Hayat was selected as an outgroup based on the sister relationship between Tetracneminae and Encyrtinae recovered by Munro et al. (2011). The COI barcode dataset contained twenty Ooencyrtus sequences representing at least three identified species: O. kuvanae (Howard), O. ferdowsii Ebrahimi and Noyes and O. nezarae from Alabama, USA. The remaining nine Ooencyrtus sequences were determined only to the level of genus.

Sequences were aligned using ClustalW with default settings (Thompson et al. 1994) as implemented in MEGA7 (Kumar et al. 2016). The sequence alignment was trimmed on both ends to optimise for completeness of data coverage. Short COI sequences and sequences with many ambiguous bases were excluded from analysis. In total, the aligned COI barcode dataset was 562 bp and contained 204 encyrtine sequences with 99% data coverage (Table 1; Suppl. material 1). To visualise the clustering of the sequences in MEGA7, a neighbour-joining tree was inferred using distances computed by the Kimura 2-parameter model (K2P) with complete deletion of missing or ambiguous data (Kimura 1980, Kumar et al. 2016). Computing the K2P distances for the Ooencyrtus data was complicated by the lack of species-level identifications (i.e. sequences need to be grouped by the user to perform the calculations). To overcome this impediment, K2P distances for Ooencyrtus “species” were calculated using molecular operational taxonomic units (OTUs) recovered by species delimitation procedures.

The mPTP species delimitation programme requires a rooted, binary Newick tree for input (Kapli et al. 2017). Maximum likelihood analysis of the encyrtine COI matrix was conducted in W-IQ-TREE (Trifinopoulos et al. 2016). The matrix was partitioned by codon position. The best-fit model of sequence evolution for each partition was selected by ModelFinder (Kalyaanamoorthy et al. 2017), as implemented in W-IQ-TREE, using the Bayesian information criterion (TN+F+G4 first position; GTR+F+I+G4 second position; TIM+F+G4 third position). Bootstrap support values for the most likely tree were calculated using 10,000 ultrafast bootstrap replicates (Hoang et al. 2017). The bootstrap consensus tree was converted to Newick format in FigTree 1.4.0 (Rambaut 2012) for importation into the mPTP programme.

Delimitations computed in the mPTP programme used the minimum branch length threshold calculated with the “--minbr_auto” option. Heuristic maximum likelihood species delimitations were performed with a single coalescent rate averaged over all species. The confidence of the delimitations was assessed using the Markov Chain Monte Carlo (MCMC) sampling method. MCMC parameters included two independent runs with each run starting with the most likely delimitation, 100,000,000 MCMC iterations, and sampling of log-likelihoods every 10,000 MCMC iterations. The convergence of independent runs was evaluated by examining the sampled log-likelihood plots. The two runs converged within the first 1% of MCMC iterations.

Species delimitations were also produced using the server version of Automatic Barcode Gap Discovery (ABGD) (Puillandre et al. 2012). The results of ABGD depend heavily on parameters set by the user. We generally followed the ABGD methods of Pentinsaari et al. (2017), with the prior upper limit to intraspecific divergence (P) set to 0.01 for all tests. We also varied the relative gap width (X) from 0.1 to 2.0 at 0.1 intervals. The ABGD distance matrix was calculated using the K2P model with the transition/transversion ratio set to 1.05, as was calculated in MEGA7 from the alignment.

The three previously frozen M. cribraria egg masses used to rear the first generation comprised 235 eggs. Of these, 65 were deemed unviable because they had desiccated prior to being frozen or during the freezing process, resulting in 170 available eggs. Of the available eggs, 122 individuals emerged: 36 males and 86 females. The second generation provided a total of 751 eggs, with 521 eggs deemed viable. Of the viable eggs, 345 were parasitised (66.2%) and 314 adult O. nezarae emerged: 119 males and 195 females. Rates of parasitism and emergence for each of the vials are provided in Table 2.

Laboratory reared specimens were compared to descriptions in Ishii (1928), Zhang et al. (2005) and Ademokoya et al. (2018). The morphology of the specimens was consistent with the diagnoses given in these references (see Figs 1, 2, 3): frontovertex approximately 1/3 of head width; ocelli forming an obtuse triangle; posterior ocelli closer to compound eye than to occipital margin; mandible with crenulated truncation and distal tooth; stigmal vein of wings longer than postmarginal vein; linea calva open posteriorly; dark colouration on coxae, mesal femora, sub-basal tibiae and apical tarsi; the rest of the legs whitish-yellow; body dark with bluish metallic sheen; reticulation present on most of body; apex of scutellum smooth.

Figure 1.  

Ooencyrtus nezarae, female. A. Dorsal habitus; B. Ventral habitus; C. Lateral habitus.

Figure 2.  

Ooencyrtus nezarae, male. A. Dorsal habitus; B. Ventral habitus; C. Lateral habitus.

Figure 3.  

Ooencyrtus nezarae male mandibles.

Ademokoya et al. (2018) noted that the Alabama population of O. nezarae exhibited differences in overall size and antennal morphometrics from the holotype. It was later confirmed via correspondence with J.S. Noyes at The Natural History Museum, London, that this discrepancy could be explained by allometric scaling of funicular segments (Ademokoya et al. 2018). The Alabama specimens were small (body length 0.70–0.77 mm) and the first two funicular segments of the female antenna were quadrate or slightly wider than long, in contrast to the descriptions of Ishii (1928) and Zhang et al. (2005), which described them as longer than wide. The female O. nezarae from Florida are approximately 0.8 mm in length and the first two funicular segments are roughly quadrate. These observations are consistent with the pattern of allometric scaling in these segments; i.e. larger specimens have proportionally longer funicular segments.

There is some disagreement between Ishii (1928) and Zhang et al. (2005) regarding the proportions of the third funicular segment in O. nezarae. Ishii (1928) describes its length as equal to its width, while Zhang et al. (2005) state that all funicular segments are longer than wide. In this respect, the Florida specimens more closely match the description given by Ishii (1928).

Neighbour-joining analysis excluded all positions with ambiguous data, leaving a total of 487 positions for inclusion (out of 562 possible positions) and recovered an Ooencyrtus group comprised of eight clusters (Fig. 4) (see Suppl. material 2 for groupings used for K2P distance calculations). Mean between-group K2P distances ranged from 5.4% to 13.7%. Mean within-group sequence divergences were low when comparisons were possible. The group containing sequences KF724500 and KF724501 had a mean within-group K2P distance of 0.8%. For O. nezarae, the Alabama and Florida samples had a mean between-group K2P distance of 1.3%. The sequences that form the two O. nezarae groups (Alabama and Florida) have zero within-group divergence. When taken together, the O. nezarae sequences ranged from 8.2% to 13.0% different from the other OTUs.

Figure 4.  

Kimura 2-parameter neighbour-joining tree showing the clustering of similar sequences in the Encyrtinae 5’-COI barcode dataset. Red branches indicate the Ooencyrtus cluster. Blue and fuchsia branches indicate O. nezarae from Alabama and Florida, respectively. Numbers in parentheses after the taxon name indicate how many sequences are represented in that cluster.

W-IQ-TREE analyses found the most likely tree with a log likelihood score of -12138.342. ML bootstrap support for the recovered Ooencyrtus clade was strong (98%; Fig. 5). The internal nodes of the Ooencyrtus clade were also strongly supported, ranging from 92% to 100% ML bootstrap support (Fig. 5). Heuristic maximum likelihood species delimitations using PTP delimited 74 total species from the Encyrtinae COI barcode dataset and eight Ooencyrtus species (Fig. 5). This delimitation scheme supported the split of the Alabama and Florida O. nezarae into reciprocally monophyletic OTUs. Runs of the MCMC PTP analysis returned an average ML support of approximately 0.954 and an average standard deviation of support values amongst these runs of 6.9*10^-6. Within the Ooencyrtus clade, the fraction of MCMC sampled delimitations, in which a node was part of the speciation process, was 1.00 in all cases except for O. nezarae, which was 0.91 (Fig. 5). This indicates that the delimitation that split Alabama and Florida O. nezarae was not always recovered by this analysis.

Figure 5.  

Transformed Ooencyrtus branch from the most likely tree for the Encyrtinae COI dataset. Each coloured bar represents an OTU delimited by the ABGD (initial and recursive partitions) and PTP procedures. Support values are ML bootstraps and the fraction of MCMC sampled delimitations in which a node was part of the speciation process in the PTP procedure, respectively. Numbers in parentheses after the taxon name indicate how many sequences are represented in that OTU.

In ABGD, all tested minimum relative gap widths returned initial delimitations of 68 species from the Encyrtinae dataset at P = 0.01. In this same range of minimum relative gap widths, ABGD returned recursive delimitations of 71 or 72 from the encyrtine dataset at P = 0.01. Across all tests, the initial delimitations recognised seven Ooencyrtus species and lumped O. nezarae from Alabama and Florida into a single OTU (Fig. 5). Across all tests, the recursive delimitations recognised eight Ooencyrtus species (matching the PTP delimitation) and split O. nezarae from Alabama and Florida into two OTUs (Fig. 5).