BOXES
Box 1.
Box 1. Beyond fig wasps: seeking generality across systems. Our insights derive from an unusual system comprising well replicated lineages of the host plant colonised by species-specific insects, where the latter cohabit spaces that serve as both brood and mating sites. However, such peculiarities have previously served as testing ground for what would later become a broader generalisation, e.g., Hamilton’s theory on local mate competition (Hamilton 1967). Here we present a brief outline of how research can be extrapolated to less constrained systems. wolPredictor can easily be brought to bear on data sets which include the following information: i) Wolbachia strain type data at the individual insect level (ideally for multiple loci and individuals per species sampled). ii) A molecular phylogeny including each insect screened for Wolbachia with which to quantify evolutionary distance between infected lineages. iii) Environmental and natural history data with which to estimate or measure physical proximity. This is a user-input decision dependant on whatever binary (contact vs. non-contact) degree of ecological contact is considered pertinent. Several systems would provide suitable opportunities for testing the generality of our theory:Associations through shared-host herbivory: Herbivorous insects typically consume a limited range of closely related plants (e.g., congenerics), and this is determined to a large extent by their natural history and taxonomy (Novotny  et al. 2010). Leaf miners and galling insects are among the most host-specific guilds, while most caterpillars lay in the middle of the host-specificity spectrum, and adult leaf chewers are among the most polyphagous. Given the varying degrees of host-specificity, we may expect a continuum such that systemic patterns of Wolbachia are found among leaf miners which however dissipate as we sample leaf chewing phasmids, under the proviso that contact is not occurring under alternate ecological circumstances. We note that both enclosed (e.g., oak gall wasp) and considerably more open communities (e.g., forest food webs) would represent good points along the continuum at which to measure.  Associations through parasitism: Herbivores themselves are attacked by host-specific parasitoids (typically Hymenoptera or tachinid Diptera), making the latter potential candidates for Wolbachia-mediated RI. Additionally, parasitoids in most systems are relatively host-specific. We expect that haplodiploids, like Thysanoptera and other Hymenoptera, would be good candidates for Wolbachia-mediated RI. Some species of gall thrips (Thysanoptera, Kladothrips) form eusocial societies specialised across specific host plant taxa from the genus Acacia (Leguminosae, Mimosoideae; (Crespi  et al. 2004). There are clearly parallels here with the fig wasp system, prompting further study. Other candidate communities comprise the various plant antagonists such as florivores and other exploiters that share hosts (Bronstein  et al. 2003).  Associations through pollination: Communities of specialist pollinators would make interesting comparative systems where ecological contact is again achieved through use of the same resources. Post-pollination larval development, such as the case of fly pollinators ovipositing in the inflorescence of Artocarpus (Moraceae), might also be relevant for beetle and fly groups (Sakai 2002). Outside Moraceae, obvious candidates are other brood pollination mutualisms such as Yucca (Agavaceae) and the pollinating yucca-moths (genera Parategeticula and Tegeticula). Incidences of phylogenetic incongruence among the host plant and the moths minimises the degree of coevolution, leaving space for alternative mechanisms of divergence (Althoff  et al. 2012; Darwell et al. 2016). The gracillariid pollinators of leafflowers (Phyllanthaceae) and the dipteran pollinators of Trollius (Ranunculaceae) may also serve as systems where alternative mechanisms of RI operate. Non-haplodiploid organisms may prove less prone to Wolbachia induced RI, but many are at risk of chromosomal inversions (Wellenreuther & Bernatchez 2018): these can also be modelled using wolPredictor along with other forms of RI.  Other associations: More generally, sampling of other invertebrate communities previously not considered as candidates for CI mediated restrictions on geneflow can provide insights, given sufficient consideration of sampling strategy and metadata collection. For example, soil insect communities provide frequent opportunity for ecological contact and microclimate formation.