Phylogenetic Analyses and Species Identification
A total of 67 parasitoids and 72 beetle hosts were successfully sequenced, and of these, there were 51 pairs of parasitoids plus their corresponding beetle hosts that were both successfully sequenced. The final alignments included 80 sequences for the parasitoid alignment, including one outgroup and 8 voucher sequences (Peixoto et al., 2018), and 85 sequences for the host beetle alignment, including one outgroup and 12 voucher specimens (Nahrung et al., 2020; Peixoto et al., 2018). Both alignments consisted of 684 characters.
For the parasitoids, both Bayesian (Fig. 2) and maximum likelihood phylogenetic analyses (SI Fig. 1) recovered five distinct and well supported clades (pp = 1; Bootstrap ≥ 90). Four of these clades corresponded to the Peixoto et al. (2018) voucher specimens. One clade (hereafter referred to as Eadya sp. 1) did not fall with any known vouchers and all three specimens in the clade did not have adult vouchers for identification. As the specimens could be E. duncanRidenbaugh, 2018 or Eadya falcata Huddleston & Short, 1978 which have never been sequenced, or a new yet to be described species, the identity of the parasitoids could not be confirmed. All Eadyaspecimens sequenced for this study were recovered in either the E. daenerys , E. annleckieae , or E. sp. 1 clades. The only difference between the Bayesian (Fig. 2) and maximum likelihood (SI Fig. 1) phylogenies was the placement of the E. sp. 1 clade. The Bayesian analysis recovered E. sp. 1 sister to E. daeneryswith a posterior probability of 0.90, while the maximum likelihood analysis recovered it sister to the clade of E. paropsidis +E. spitzer with poor support (Bootstrap = 58).
For the beetles, both Bayesian (Fig. 3) and maximum likelihood (SI Fig. 2) analyses recovered six distinct and strongly supported clades (pp = 1; bootstrap ≥ 91) with identical relationships. All recovered beetle clades corresponded to UCF voucher specimens from Peixoto et al. (2018) and Nahrung et al. (2020). Host specimens sequenced for this study were recovered in four out of the six beetle species for which we had vouchers: Paropsis aegrota-elliotti Selman, 1983;Paropsisterna m-fuscum (Boheman, 1859); Pst .agricola (Chapuis, 1877); and Pst . cloelia clades. A distinct barcode gap, with interspecific distances greater than intraspecific distances, was observed in both the parasitoid (Table 1) and beetle datasets (Table 2). The largest intraspecific genetic distance for parasitoids was 1.06% in E. spitzer and 0.78% inPst. m-fuscum for the beetle hosts. Interspecific genetic distances in parasitoids ranged from 6.45% between E. daenerysand E. sp. 1, to 31.07% between E. paropsidis andE. annleckieae (Table 1). Large values were also recovered in the host dataset, with the smallest interspecific distance being 9.77% between Pst. agricola and Pst. variicollis* , and the largest 19.07% between P. charybdis and Pst. agricola (Table 2).
Given that all sequenced specimens fell within well supported clades with distinct barcoding gaps, we could identify the unknown sequenced specimens for both parasitoids and hosts as follows: 20 E. daenerys , 44 E. annleckieae (not including the four Tasmanian specimens), three E. sp. 1, one P. aegrota elliotti , fivePst. m-fuscum , 17 Pst. agricola , and 49 Pst. cloelia . These identifications were used for all downstream analyses.