Abstract
Animals serve as hosts for complex communities of microorganisms, including endosymbionts that live inside their cells. Wolbachia bacteria are perhaps the most common endosymbionts, manipulating host reproduction to propagate. Many Wolbachia cause intense cytoplasmic incompatibility (CI) that promotes their spread to high and relatively stable frequencies. Wolbachia that cause weak or no CI tend to persist at intermediate, often variable, frequencies. Wolbachia could also contribute to host reproductive isolation (RI), although current support for such contributions is limited to a few systems. To test for Wolbachia frequency variation and effects on host RI, we sampled several local Prosapia ignipectus (Fitch)(Hemiptera: Cercopidae) spittlebug populations in the northeastern USA over two years, including closely juxtaposed Maine populations with different monomorphic color forms, “black” and “lined”. We discovered a group-B Wolbachia (wPig) infecting P. ignipectus that diverged from group-A Wolbachia—like model wMel and wRi strains in Drosophila—6 to 46 MYA. Populations of the sister species Prosapia bicincta (Say) from Hawaii and Florida are uninfected, suggesting that P. ignipectus acquired wPig after their initial divergence. wPig frequencies were generally high and variable among sites and between years. While phenotyping wPig effects on host reproduction is not currently feasible, the wPig genome contains three divergent sets of CI loci, consistent with high wPig frequencies. Finally, Maine monomorphic black and monomorphic lined populations of P. ignipectus share both wPig and mtDNA haplotypes, implying no apparent effect of wPig on the maintenance of this morphological contact zone. We hypothesize P. ignipectus acquired wPig horizontally as observed for many Drosophila species, and that significant CI and variable transmission produce high but variable wPig frequencies.
Introduction
Animals interact with microorganisms that influence their behavior, physiology, and fitness (Hurst and Jiggins, 2000; Brownlie et al., 2009; McFall-Ngai et al., 2013; Fredericksen et al., 2017; Gould et al., 2018; Hague, Caldwell and Cooper, 2020). These include associations between hosts and vertically transmitted endosymbionts that live inside their cells (McCutcheon, Boyd and Dale, 2019). Hosts may acquire endosymbionts cladogenically from common ancestors (Raychoudhury et al., 2009; Koga et al., 2013; Toju et al., 2013), from sister species via hybridization and introgression (Turelli et al., 2018; Cooper et al., 2019), or horizontally in ways that are not fully understood (O’Neill et al., 1992; Huigens et al., 2000; Ahmed et al., 2015). While few examples exist, endosymbionts can contribute to host reproductive isolation (RI) and speciation (Coyne and Orr, 2004; Matute and Cooper, 2021), highlighting the importance of discovering and characterizing endosymbiont-host associations.
Maternally transmitted Wolbachia bacteria are widely distributed (Werren, Baldo and Clark, 2008; Zug and Hammerstein, 2012; Weinert et al., 2015), infecting many arthropods and two groups of parasitic nematodes (Bandi et al., 1998), making Wolbachia perhaps the most common endosymbiont in nature. In Drosophila, introgressive and horizontal Wolbachia acquisition seem to predominate (Conner et al., 2017; Turelli et al., 2018; Cooper et al., 2019), but cladogenic acquisition during host speciation has been observed in other taxa (Raychoudhury et al., 2009; Gerth and Bleidorn, 2017). Many Wolbachia manipulate host reproduction to propagate in host populations. For example, many strains cause cytoplasmic incompatibility (CI) that reduces the egg hatch of uninfected embryos fertilized by Wolbachia-infected sperm (Hoffmann and Turelli, 1997). However, if females are also infected, the embryos survive, “rescuing” CI and promoting Wolbachia spread to high frequencies (Hoffmann, Turelli and Harshman, 1990; Turelli and Hoffmann, 1995; Barton and Turelli, 2011; Kriesner et al., 2013).
Wolbachia may contribute to host RI (Coyne and Orr, 2004; Matute and Cooper, 2021), with the best evidence coming from Drosophila. Wolbachia contribute to assortative mating and postzygotic isolation between co-occurring D. paulistorum semi-species (Miller, Ehrman and Schneider, 2010), and to reinforcement of isolation between uninfected D. subquinaria and Wolbachia-infected D. recens (Shoemaker, Katju and Jaenike, 1999; Jaenike et al., 2006). In contrast, Wolbachia do not contribute to RI in the D. yakuba clade, which includes wYak- infected D. yakuba, wSan-infected D. santomea, and wTei-infected D. teissieri (Cooper et al., 2017). Thus, while some results from Drosophila strongly support contributions of Wolbachia to RI, and interest in the possibility of such effects remains high, it is unknown whether Wolbachia effects on RI are common in nature.
Wolbachia frequencies differ significantly among infected host taxa, ranging from very low to obligately fixed infections (Bandi et al., 1998; Kriesner et al., 2013; Cooper et al., 2017; Miller, Ehrman and Schneider, 2010). Wolbachia effects on reproduction (e.g., CI) and fitness, in combination with imperfect maternal transmission, govern its frequencies in host populations (Caspari and Watson, 1959; Hoffmann, Turelli and Harshman, 1990). Intensive sampling of a few systems has revealed both stable and variable Wolbachia frequencies within host populations. Wolbachia that cause intense CI, like wRi in Drosophila simulans, persist at high and relatively stable frequencies, balanced by imperfect maternal transmission (Kriesner et al., 2013; Turelli et al., 2018). In contrast, Wolbachia that cause weak or no CI tend to occur at variable intermediate frequencies (Hoffmann, Clancy and Duncan, 1996; Hamm et al., 2014; Kriesner et al., 2016; Cooper et al., 2017; Meany et al., 2019). These include wMel-like Wolbachia frequencies that vary spatially in D. melanogaster and D. yakuba (Kriesner et al., 2016; Hague et al., 2020), and temporally in D. yakuba and D. santomea (Cooper et al., 2017; Hague, Caldwell and Cooper, 2020). In all but a few systems, limited sampling has left a gap in knowledge about whether Wolbachia frequency variation is common (Hughes et al., 2011; Hamm et al., 2014; Cattel et al., 2016; Schuler et al., 2016; Ross et al., 2020).
Prosapia ignipectus (Fitch) (Hemiptera: Cercopidae) is one of about 14 species of Prosapia and one of two commonly found in the USA, the other being its sister species P. bicincta (Say)(Hamilton, 1977). P. ignipectus occurs in southern Ontario, Canada and the northeastern USA from Minnesota to Maine (Hamilton, 1977, 1982; Peck, 1999; Carvalho and Webb, 2005; Thompson and Carvalho, 2016). These species vary in male genital morphology and in associations with host plants, with P. ignipectus monophagous on the late season C4 perennial grass Schizachyrium scoparium (Little bluestem) (Hamilton, 1982; Thompson, 2004) and P. bicincta polyphagous on a variety of C4 grasses, but not including Little bluestem (Fagan and Kuitert, 1969; Thompson, 2004). Both species have conspicuous dorsal coloration, standing out against their respective host plants. All P. bicincta individuals have a single narrow transverse orange line across the widest part of the pronotum and a pair of narrow orange lines across the elytra. Most P. ignipectus individuals have a solid black dorsal surface, but in Maine some P. ignipectus have P. bicincta-like coloration (Figure 1). Notably, only 10 km separate monomorphic black and monomorphic lined P. ignipectus populations in western Maine, with little evidence of a hybrid zone and no obvious physical barriers to mixing across the boundary (Thompson and Carvalho, 2016). This morphological contact zone has persisted for at least 90 years. About 45 km southwest of this abrupt transition between aposematic color forms, three other P. ignipectus populations were found to be polymorphic with both black and lined forms— these populations are surrounded by monomorphic black populations. It has been hypothesized that Wolbachia determined RI may contribute to preservation of the sharp Maine morphological contact zone (Thompson and Carvalho, 2016).
Here, we use collections of P. ignipectus from several sites in the northeastern USA across two years, in combination with collections of P. bicincta from Hawaii and Florida, USA, to assess modes of Wolbachia acquisition and to test for Wolbachia frequency variation through space and time. By sampling monomorphic black and lined populations and typing both Wolbachia and mtDNA haplotypes, we also test for contributions of Wolbachia to host RI. Finally, we generate whole genome Wolbachia data for phylogenetic analysis and to search for loci associated with inducing and rescuing CI (Beckmann, Ronau and Hochstrasser, 2017; LePage et al., 2017; Shropshire et al., 2018).
Methods
Sampling
We netted specimens from Little bluestem; sorted them by species, sex, and color form; and preserved them in 95% ethanol. The 2019 specimens (N = 4 sites) were collected on August 23. The 2020 specimens (N = 9 sites) were collected on August 9 (Silver Lake, NH), August 17 (Wonalancet, NH), and August 20 (all Maine localities) (Supplemental Table 1). Collection sites were on the verges of public rights of way or privately owned land. In two cases (New Vineyard and New Portland) they correspond to sites reported in Thompson and Carvalho (2016). Specimens were collected near the height of abundance for P. ignipectus, which starts to emerge in adult form in late July and early August. We also sampled three additional spittlebug species at these sites: Lepyronia quadrangularis (Say) (N = 25), Philaenus spumarius (L.) (N = 5), and Philaenarcys killa (Hamilton)(N = 24), all of the family Aphrophoridae. Like, P. ignipectus, P. killa a is monophage on Little bluestem. L. quadrangularis is a polyphage but often abundant on Little bluestem. P. spumarius is an extreme polyphage, with a preference for forbes (herbaceous perennial dicots) but is occasionally collected from Little bluestem in the company of P. ignipectus. By screening them for Wolbachia we tested for the possibility of horizontal Wolbachia transfer through plant interactions (Chrostek et al., 2017). Lastly, because identification of infections in sister hosts enables formal analysis of modes of Wolbachia acquisition (Turelli et al., 2018; Conner et al., 2017; Cooper et al., 2019; Raychoudhury et al., 2009), we also obtained samples of the sister species P. bicincta from Hawaii (N = 60) and Florida (N = 40) to screen for infections. P. bicincta is native to the southeastern USA (Fagan and Kuitert, 1969; Thompson and Carvalho, 2016), but has recently been introduced into the Kona Region of Big Island, Hawaii (Thorne et al., 2018).
Wolbachia typing
We generated whole genome Wolbachia data to type the Wolbachia infecting P. ignipectus and to search for loci associated with CI. We extracted 800ng of high molecular weight DNA (Qiagen Genomic-tip 20/G; Qiagen, Germany) from one black New Vineyard female (see below), and then input and sequenced it (Ligation Sequencing Kit, SQK-LSK109; FLO-MIN106 flow cell) for 48 hours (Oxford Nanopore Technologies). We mapped raw nanopore reads (5.8Gb of data) to all known Wolbachia sequences (NCBI taxid 953) with BLASTn and extracted reads where at least 60% of their length mapped (qcovs >= 60). We then corrected and assembled reads using canu 2.1.1 (Koren et al., 2017, 2018; Nurk et al., 2020) and polished the Wolbachia assembly using nanopolish 0.13.2 (Loman, Quick and Simpson, 2015). We annotated our Wolbachia assembly plus the genomes of model group-A (wMel, Wu et al., 2004; and wRi, Klasson et al., 2009) and group-B (wPip-Pel, Klasson et al., 2008; and wMau, Meany et al., 2019) strains using Prokka v.1.11 (Seemann, 2014). We used only genes present in single copy and with identical lengths in all genomes. To assess the quality of our assembly, we excluded wPig and repeated this with only wMel, wRi, wPip, and wMau.
Preliminary analysis of a few loci placed the P. ignipectus Wolbachia in group-B (see below), but we performed Bayesian analyses using the GTR + Γ + I model for sequence evolution using whole genome data to confirm this (Höhna et al., 2016). Genes were concatenated and partitioned by codon position, with a rate multiplier, σ, assigned to each partition to accommodate variable substitution rates. We used flat, symmetrical Dirichlet priors on the stationary base frequencies, π, and the relative-rate parameters, η, of the GTR model (i.e., Dirichlet(1,1,1...)). As in Turelli et al. (2018), we used a Γ(2,1) hyperprior on the shape parameter, α, of the discrete-Γ model (adopting the conventional assumption that the β rate parameter equals α, so that the mean rate is 1; (Yang, 1994). The Γ model for rate variation assigns significant probability near zero when the α < 1 (accommodating invariant sites). The Γ(2,1) hyperprior on α assigns 95% probability to the interval (0.36, 4.74), allowing for small and large values. Four independent runs for each gene set produced concordant topologies. We diagnosed MCMC performance using Tracer 1.7 (Rambaut et al., 2014).
Wolbachia and mtDNA haplotyping of black and lined color morphs
To confirm that the same Wolbachia strain infects different P. ignipectus populations and color morphs, we amplified and Sanger sequenced five protein-coding Wolbachia genes (coxA, hcpA, fbpA, ftsZ, and wsp) in both directions (Eurofins Genomics LLC, Louisville, Kentucky)(see below, Supplemental Table 2). We also amplified and Sanger sequenced gatb, but sequence quality was consistently too low to include in our analyses. Samples included one infected female of each color form (black or lined), from each of the four populations (Carthage, New Portland, New Vineyard, and Strong) sampled in both years (Supplemental Table 1).
To specifically assess if Wolbachia might contribute to the morphological contact zone between New Vineyard (monomorphic black) and New Portland (monomorphic lined) P. ignipectus, we also we amplified and Sanger sequenced the cytochrome C oxidase I (CoI) mitochondrial locus from one male and one female from these populations, with the exception of one (New Vineyard black male) that did not produce usable sequence. We also produced CoI sequences for one black and one lined female from the polymorphic Strong population.
We visually inspected each sequence for quality and ambiguities, and consensus sequences were used as queries for a BLASTn search and the NCBI “nr” database to confirm that orthologous genes were amplified (Altschul et al., 1990). We then used the “multiple locus query” function of the multi locus sequence typing (MLST) database to type Wolbachia (Baldo et al., 2006). Together these data enable us to test for differentiation in Wolbachia and mtDNA between populations and color forms, including between populations monomorphic for different color forms separated by only 10 km in Maine.
Analysis of CI loci
Recent work has identified CI-causing factors (cifs) associated with WO prophage in Wolbachia genomes (Beckmann, Ronau and Hochstrasser, 2017; LePage et al., 2017; Shropshire et al., 2018; Shropshire and Bordenstein, 2019; Shropshire, Leigh and Bordenstein, 2020). Two genes (cifA/B) transgenically expressed in male D. melanogaster induce CI, while one gene (cifA) expressed in females rescues it. To identify cif loci, we used BLASTn to search for cif homologs in our whole genome raw reads, querying the Type 1 cif pair in wMel, the Type 2 pair in wRi, the Type 3 pair in wNo, the Type 4 pair in wPip, and the Type 5 pair in wStri (Lindsey et al., 2018; Bing et al., 2020; Martinez et al., 2020). We later broadened our search for Type 1 pairs by querying wPip and wNPa pairs (Klasson et al., 2008; Gerth and Bleidorn, 2017). For each Type, we extracted raw reads that covered at least 40% of the genes. We then corrected and assembled the reads with canu 2.1.1 (Koren et al., 2017, 2018; Nurk et al., 2020), producing sequences with about a 1% error rate. We limit our analyses to the discovery of cif types, since we did not generate additional sequence data to further correct the long reads. The assembled genes were compared to those in Martinez et al. (2020).
Analysis of Wolbachia frequency variation
To test for Wolbachia frequency variation, we extracted DNA from many individuals from each collection using a standard squish buffer protocol and identified Wolbachia infections using polymerase chain reaction (PCR) (Simpliamp ThermoCycler; Applied Biosystems, Singapore) (Meany et al., 2019). We amplified the Wolbachia surface protein (wsp) (Braig et al., 1998) and arthropod-specific 28S rDNA, which served as a positive control (Baldo et al., 2006) (Supplemental Table 2). PCR products were visualized using 1% agarose gels. Assuming a binomial distribution, we estimated exact 95% confidence intervals for Wolbachia frequencies for each collection. We used Fisher’s exact test (FET) to determine differences in frequencies among sites, between years, between sexes, and between color forms.
Results
P. ignipectus likely acquired its group-B Wolbachia following initial divergence from P. bicincta
Across all samples, Wolbachia infection frequency (p) in P. ignipectus is high (p = 0.93 [0.90, 0.95]; N = 486). Based on five Sanger sequenced loci, the multiple sequence query of the MLST database supports that a group-B strain, most closely related to Wolbachia in Chloropidae (Diptera) (ID 93, ST 104), infects our P. ignipectus samples—we call this strain wPig. Preliminary phylogenetic analyses using only our five Sanger sequenced genes also placed wPig in group B. Our draft wPig assembly size (1.32Mb, N50 = 91,011) falls in the range of complete Wolbachia genomes (e.g., wMel at 1.26Mb and wRi at 1.44Mb), despite its fragmentation (50 contigs). In total, we extracted 65 single-copy homologs of equal length (43,473 total bp) for our phylogenetic analysis, which also places wPig in group B (Figure 2). When excluding the wPig genome, we were able to extract an additional 135 homologs (16,7241 bp) from wMel, wRi, wPip, and wMau. This indicates that significant residual error in the wPig assembly reduces the number of homologs meeting our equal length criteria for inclusion. Finer placement of wPig among group-B strains will require the generation of short-read data to further correct our draft wPig assembly. Thus, we do not attempt to place wPig precisely among group-B strains.
None of the P. bicincta samples from Hawaii and Florida were Wolbachia infected. Even if some P. bicincta are Wolbachia infected, as previously reported for one individual used as a PCR control in another study (Anderson, Rustin and Eremeeva, 2019), Wolbachia infection frequency (p) must be very low across the P. bicincta range, given our species estimate and credible interval (p = 0.0 [0.0, 0.04]; N = 100), keeping in mind the possibility that the Hawaiian population may have experienced a recent bottleneck during introduction and may not be representative of the species in the native range. Very low frequency Wolbachia infections in global P. bicincta populations, in combination with generally high wPig frequencies in P. ignipectus, indicates that P. ignipectus likely acquired wPig after its initial divergence from P. bicincta. Because testing predictions about modes of Wolbachia acquisition requires formal analysis of Wolbachia, host nuclear, and host mtDNA phylograms and chronograms, we are unable to distinguish between introgressive and horizontal wPig transfer (Raychoudhury et al., 2009; Conner et al., 2017; Gerth and Bleidorn, 2017; Turelli et al., 2018; Cooper et al., 2019). We discuss this further below.
Of the additional species we netted from Little bluestem, all L. quadrangularis were uninfected (p = 0.0 [0.0, 0.14]; N = 25), all P. spumarius were infected (p = 1.0 [0.48, 1.0]; N = 5), and only one P. killa individual was infected (p = 0.04 [0.001,0.21]; N = 24). Because Wolbachia that infect P. spumarius and wPig in P. ignipectus are both at high frequency, we also typed the Wolbachia infecting P. spumarius to determine if a wPig-like variant also infects this host species. The multiple sequence query in the MLST database supports that a different group- B strain, most closely related to the thrip species Aptinothrips rufus (ID 1945, ST 509) infects P. spumarius. Generating more sequence data will be required to resolve the phylogenetically relationships of these and other group-B strains, including Wolbachia in P. spumarius (Lis, Maryańska-Nadachowska and Kajtoch, 2015).
No apparent effect of wPig on the maintenance of the morphological P. ignipectus contact zone
The Strong, Carthage, and Dixfield P. ignipectus populations (Figure 3) were polymorphic for the black and lined forms (Figure 1, Supplemental Table 1), like three populations close to Rumford, Maine sampled in earlier work (Thompson and Carvalho, 2016). This set of mixed color-form populations runs roughly from Rumford northeast to Strong, but not to the sharp boundary dividing the monomorphic black New Vineyard population form the monomorphic lined New Portland population. It has the appearance of a hybrid zone, but one that does not reach the definitive boundary between the forms. The existence of distinct color forms both within and between the populations sampled facilitated investigation of the relationship, if any, between Wolbachia infection and patterns of color form occurrence.
We found no evidence for wPig genetic differentiation between P. ignipectus populations or color forms. Regions of the five wPig genes we sequenced were identical, except for a single nucleotide position in wsp, where the Strong lined sample differed from all others. In addition to populations sharing wPig type based on MLST loci, wPig frequency did not vary between color forms (black: p = 0.93 [0.90, 0.95], N = 338; lined: p = 0.92 [0.86, 0.96], N = 123; FET, P = 0.69), among only males (black: p = 0.84 [0.75, 0.90], N = 98; lined: p = 0.90 [0.79, 0.97], N = 51; FET, P = 0.33), or among females (black: p = 0.97 [0.94, 0.99], N = 240; lined: p = 0.93 [0.85, 0.98], N = 72; FET, P = 0.19), across all samples. wPig frequency also did not differ between New Vineyard (monomorphic black) and New Portland (monomorphic lined) populations (FET, P = 0.16).
We found no evidence for differentiation in CoI mtDNA haplotype between the New Vineyard and New Portland P. ignipectus populations, where all samples were identical across the 680 bp that we recovered. The black and lined females from the polymorphic Strong population also did not differ from each other, or from other populations, across this region. Thus, wPig and mtDNA haplotypes were not differentiated between populations or color forms..
Our mtDNA haplotypes are also very similar to ten P. ignipectus samples included in the Barcode of Life Database (BOLD) (Foottit, Maw and Hebert, 2014). A single base-pair insertion present in all of our samples is absent from all ten BOLD samples. Four other sites in CoI that are polymorphic among the BOLD samples are fixed in our samples for one of the BOLD alleles. mtDNA haplotypes of P. ignipectus and P. bicincta also differ by less than 2 percent (Foottit, Maw and Hebert, 2014).
The wPig genome contains three divergent types of CI loci
We identified Type 1, 3, and 4 cifs in the wPig genome (Martinez et al., 2020). This specific complement of cifs is not found in any other published Wolbachia genomes, but close relatives to each wPig cif Type are. For instance, the wPig Type 1 genes are 99% identical to those in the genome of the Wolbachia infecting the gall-inducing wasp Diplolepis spinosa (Cynipidae), but less than 90% similar to any others (Martinez et al., 2020). The Type 3 wPig genes are 99% identical to those in the genome of the Wolbachia infecting D. spinosa, the Staphylinid beetle Diploeciton nevermanni, and the water strider Gerris buenoi. The wPig Type 4 genes are 99% identical to those in Wolbachia infecting Nomada bees (wNLeu, wNFla, and wNPa), but less than 95% identical to other Type 4 cifs. The Wolbachia infecting D. spinosa does not have Type 4 cifs, distinguishing it from wPig. None of the wPig cifs are truncated relative to copies with 99% identity. Additional sequencing is required to make more detailed cif comparisons.
Pervasive wPig frequency variation
wPig varied in frequency in several ways. First, frequency varied spatially among all samples (FET, P = 0.001)(Table 1), among sites in 2019 (FET, P < 0.0001), and 2020 (FET, P = 0.033). This variation occurred over a geographic radius of only 20 km in 2019 and 70 km in 2020 (Figure 3). Second, frequency varied across all samples between 2019 (p = 0.88 [0.82, 0.92]; N = 169) and 2020 (p = 0.95 [0.92, 0.97]; N = 317) (FET, P = 0.003). For the four sites we sampled in both years, frequencies were only significantly different between 2019 (p = 0.73 [0.56, 0.86]; N = 37) and 2020 (p = 1.0 [0.91, 1.0]; N = 40) in New Vineyard (FET, P < 0.001). Third, across all samples wPig frequency was higher in females (p = 0.95 [0.93, 0.97]; N = 332) than males (p = 0.86 [0.80, 0.91]; N = 154) (FET, P = 0.001). However, this was driven mostly by a paucity of infected males in New Vineyard (males: p = 0.69 [0.50, 0.84], N = 32; females: p = 1.0 [0.92, 1.0], N = 45; FET, P < 0.0001), with no differences in wPig frequency between males and females in other populations. wPig frequency in males was relatively low in 2019 (p = 0.17 [0.02, 0.48]; N = 12), but fixed in 2020 (p = 1.0 [0.83, 1.0]; N = 20). We interpret these results as pervasive spatial, and rare temporal and sex-specific, variation in wPig frequency.
Discussion
Our results suggest that wPig is a group-B Wolbachia acquired after the initial divergence of P. ignipectus from P. bicincta. Analysis of Wolbachia and mtDNA haplotypes indicates that wPig has no apparent effect on the P. ignipectus morphological contact zone in Maine. Across all samples, wPig occurs at very high frequencies, consistent with our discovery of three divergent sets of CI loci in the wPig genome. Finally, we document pervasive spatial, and rare temporal, wPig frequency variation. We discuss this in more detail below.
Wolbachia acquisition in spittlebugs
In contrast to very high wPig frequencies in P. ignipectus, we found no evidence of Wolbachia in our sample of 100 P. bicincta. A prior report of one infected P. bicincta sample indicates that Wolbachia could infect this species (Anderson, Rustin and Eremeeva, 2019). If so, it must be at very low frequencies, given our credible interval here (p = 0.0 [0.0, 0.04]; N = 100). Mathematical models predict that intense CI drives Wolbachia to high frequencies, balanced by imperfect maternal transmission (Hoffmann, Turelli and Harshman, 1990; Turelli and Hoffmann, 1995); conversely, Wolbachia that do not cause strong CI tend to occur at much lower frequencies (Hamm et al., 2014; Kriesner et al., 2016; Cooper et al., 2017; Hague et al., 2020). While crossing to test for CI in the laboratory is not currently feasible in this system, the presence of three sets of CI loci in the wPig genome, combined with its very high frequencies, suggests that wPig causes intense CI.
How did P. ignipectus acquire wPig? There are three possibilities: cladogenic transmission from its most recent common ancestor with its sister species, presumably P. bicincta or a close relative; by introgression from P. bicincta or another close relative; or by horizontal transmission (O’Neill et al., 1992). Given that we find no evidence for a high frequency Wolbachia in P. bicincta, cladogenic acquisition seems implausible. Without more extensive analysis of close relatives, we cannot rule out introgression. However, opportunities for introgression with species other than P. bicincta have likely been limited. Other species of the genus Prosapia or family Cercopidae occur no further north than the USA-Mexico border region, about 1,400 km from the nearest P. ignipectus populations and 3,000 km from the populations studied here.
Overall, the limited data are consistent with relatively recent non-cladogenic transmission, a process that seems to be common among Drosophila species (Turelli et al., 2018). It may also be common among spittlebugs. This would be in stark contrast to obligate transovarial endosymbionts associated with amino acid nutrition in spittlebugs and other hemipterans (Koga et al., 2013). In addition to the thrip-related Wolbachia found in P. spumarius in this study, Nakabachi et al. (2020) report that two spittlebug species, Aphrophora quadrinotata Say and Philaenus maghresignus Drosopoulos & Remane (both Aphrophoridae’), harbor Wolbachia with 16S rRNA sequence that is identical to Wolbachia in two psyllid species, two whiteflies, an aphid, a planthopper, two leafhoppers, two grasshoppers, a mosquito and a weevil. Likewise, Lis et al. (2015) report that Wolbachia they studied in P. spumarius is closely related to strains in vespids, drosophilids, whiteflies, chrysomelid beetles and weevils based on five MLST loci. Kapantaidaki et al. (2021) also report Wolbachia infections at low levels in P. spumarius, as well as higher frequencies in Neophilaenus campestris (Fallen) (Aphrophoridae). Based on five MLST loci, their N. campestris strain is closely related to Wolbachia found in a leafhopper (Hemiptera) and cluster with Wolbachia from a planthopper, a scale insect and a psyllid (all Hemiptera), as well as two chrysomelid beetles, two butterflies, a parasitic wasp and a mosquito. Koga et al. (2013, table S3) report the presence of Wolbachia in the spittlebug Cosmoscarta heros (F.) (Cercopidae), in addition to A. quadrinotata and P. maghresignus.
In contrast, five specimens of Poophilus costalis (Walker) (Aphrophoridae) (Wiwatanaratanabutr, 2015), six specimens of Philaenus tesselatus Melichar (Lis, Maryańska- Nadachowska and Kajtoch, 2015), 37 specimens of Philaenus signatus Melichar (Lis, Maryańska-Nadachowska and Kajtoch, 2015; Kapantaidaki et al., 2021), and single specimens of Philaenus arslani Abdul-Nour & Lahoud, Philaenus loukasi Drosopoulos & Asche, and Philaenus tarifa Remane & Drosopoulos (Lis, Maryańska-Nadachowska and Kajtoch, 2015) were not infected. Based on limited sequence data, the emerging pattern suggests that Wolbachia infection is widespread, but far from ubiquitous among spittlebugs, and that when it does occur, it often involves Wolbachia strains similar to those infecting distantly related insects. Whole Wolbachia and host genomic data is sorely needed to test our hypothesis that horizontal Wolbachia acquisition might be common in spittlebugs.
Little contribution of wPig to the P. ignipectus morphological contact zone
We find no evidence for differentiation in wPig or mtDNA haplotypes among P. ignipectus color forms. This includes the monomorphic black (New Vineyard) and lined (New Portland) populations that are separated by only 10 km in Maine, with no obvious barriers to dispersal or reproduction (Thompson and Carvalho, 2016). We also found no variation in wPig or mtDNA haplotypes between black and lined individuals in the polymorphic Strong population. wPig frequency also did not vary between color forms. These data indicate that wPig is unlikely to significantly contribute to the maintenance of the P. ignipectus morphological contact zone.
How common are Wolbachia effects on host RI? Obligate Wolbachia infections in cooccurring D. paulistorum semi-species contribute to assortative mating and generate hybrid inviability and male sterility (Miller, Ehrman and Schneider, 2010). Wolbachia also contribute to reinforcement between Wolbachia-infected D. recens and uninfected D. subquinaria (Shoemaker, Katju and Jaenike, 1999; Jaenike et al., 2006). In contrast, Wolbachia do not contribute to premating, gametic, or postzygotic RI among the three D. yakuba-clade host species (Cooper et al., 2017). While the crossing schemes used in these Drosophila studies to dissect Wolbachia contributions to RI are not feasible in P. ignipectus and many other systems, our genetic data here lend support to our prior conjecture that Wolbachia contributions to RI observed in some Drosophila may be the exception rather than the rule (Turelli, Lipkowitz and Brandvain, 2014; Cooper et al., 2017).
Pervasive wPig frequency variation
Mathematical models indicate that imperfect maternal transmission, Wolbachia fitness effects, and the severity of CI govern Wolbachia frequencies in host populations. Wolbachia that cause intense CI tend to occur at high and stable frequencies, balanced by imperfect maternal transmission (Barton and Turelli, 2011; Turelli and Hoffmann, 1995; Hoffmann, Turelli and Harshman, 1990; Turelli and Hoffmann, 1991; Carrington et al., 2011; Kriesner et al., 2013); while Wolbachia that cause weak or no CI tend to persist at intermediate, often variable frequencies (Hamm et al., 2014; Kriesner et al., 2016; Cooper et al., 2017; Hague et al., 2020). Accumulating evidence for variable infection frequencies (Hamm et al., 2014; Kriesner et al., 2016; Schuler et al., 2016; Hughes et al., 2011; Lis, Maryańska-Nadachowska and Kajtoch, 2015; Cooper et al., 2017), including our discovery here, highlights that infection frequencies are not static, even for high frequency variants.
With the exception of model systems like wRi in D. simulans, few estimates of the key parameters required to approximate population frequency dynamics and equilibria of Wolbachia exist (Turelli and Hoffmann, 1995; Carrington et al., 2011). wMel-like Wolbachia frequencies in the D. yakuba clade vary through space and time in west Africa (Cooper et al., 2017), due in part to effects of cold temperatures on wYak titer (Hague et al., 2020). CI strength also varies in the D. yakuba clade, which may influence infection frequencies (Cooper et al., 2017; Hague, Caldwell and Cooper, 2020). wMel frequencies vary with latitude in D. melanogaster populations, potentially due to wMel fitness costs in the cold (Kriesner et al., 2016). Interestingly, hot temperatures reduce wMel CI strength and transmission in transinfected Aedes aegypti used for biocontrol of human disease (Ross et al., 2017, 2020), suggesting that temperature may generally influence key parameters underlying Wolbachia infection frequencies.
What underlies variable wPig frequencies in nature? High wPig frequencies and the presence of three divergent sets of cifs suggest, but do not confirm, that wPig causes strong CI. It seems plausible that some or all of these loci were horizontally acquired (Cooper et al., 2019), but additional sequence data are required to test this. We hypothesize that variable wPig transmission rates contribute to the frequency variation we observe, potentially due to environmental effects on titer, as observed for wYak (Hague et al., 2020). Temporal variation in transmission was also observed for wRi between two samples of D. simulans collected from Ivanhoe, California in April and November of 1993 (Turelli and Hoffmann, 1995; Carrington et al., 2011), although the relative stability of wRi frequencies in global D. simulans populations suggests that its transmission persists across a range of environmental conditions. Additional analyses of Wolbachia titer and transmission in the field, and across environmental contexts, are needed to better understand the causes of Wolbachia frequency variation. Comparing the titer and transmission of Wolbachia that occur at different frequencies in nature—for example, those that do and do not cause intense CI — would be particularly useful.
Data Accessibility Statement
All data will be uploaded to DRYAD or GenBank upon acceptance.
Competing Interests Statement
We declare no competing interests.
Author Contributions Section
Timothy B. Wheeler: Data curation, Investigation, Validation, Visualization, Writing - original draft, Writing - review & editing. Vinton Thompson: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Visualization, Writing - original draft, Writing - review & editing. William R. Conner: Data curation, Formal analysis, Investigation, Writing - original draft, Writing - review & editing. Brandon S. Cooper: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing.
Acknowledgments
We thank M. Thorne and A.G. Dale for P. bicincta collections and D. McVicar and F. Selchin for access to the Wonaloncet site. Michael Turelli provided comments that greatly improved an earlier draft. We also thank M. Hague, D. Shropshire, and K. Van Vaerenberghe for very helpful comments. Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health (NIH) under award number R35GM124701 to B.S.C., and by the University of Montana Genomics Core.