The whitefly Bemisia tabaci is an agricultural pest for which multiple biotypes have been described on behavioural, ecological, and genetic bases. However, this genetic and ecological diversity has been difficult to reconcile in a taxonomically informative manner. B. tabaci is also known for its diverse endosymbiont community, including the obligate nutritional symbiont Portiera aleyrodidarum (Thao & Baumann 2004) and numerous facultative endosymbionts. Overall, symbionts from at least six genera are known associates of B. tabaci, and different symbionts colonizing a single individual may be distributed differently among host body tissues (Fig. 1.)
The specific effects of these endosymbionts on B. tabaci and on one another remain to be characterized, but members of endosymbiont communities have the potential to interact in ways that could range from mutualistic to competitively exclusive (Gottlieb et al. 2008; Ros & Breeuwer 2009; Jaenike et al. 2010). Known symbionts of B. tabaci include (but are not limited to) Wolbachia, which has a wide range of reproductive and other effects on insect hosts (Werren 1997); Cardinium, also a reproductive manipulator in some arthropods (Zchori-Fein et al. 2004; Gotoh et al. 2007); Hamiltonella defensa, an inducer of parasitoid resistance in pea aphids (Oliver et al. 2005); and one symbiont, Fritschea bemisi, currently known only from some B. tabaci biotypes (Thao et al. 2003).
A considerable literature already describes the prevalence and effects of the best-studied arthropod reproductive endosymbiont, Wolbachia, but the true diversity of inherited bacteria in arthropods can be much greater (Chiel et al. 2007; Duron et al. 2008) and both vertically and horizontally transmitted symbionts may establish prolonged and specialized associations between strain and host. Any such stable association would seem promising as at least a supporting taxonomic character for both host and symbiont. However, the taxonomic value of a single host–symbiont association has inherent and serious limitations. Without extensively surveying, culturing, or breeding infected hosts, it is nearly impossible to tell whether any host–symbiont association is stable or transient. Infections by multiple strains of the same symbiont are possible within species, or even individuals (Perrot-Minnot et al. 1996; Reuter & Keller 2003; Keller et al. 2004), while distantly related hosts can share highly similar endosymbiont strains (Baldo et al. 2006; Stahlhut et al. 2010). Moreover, symbionts have evolved mechanisms to ensure their own reproductive success. Inherited intracellular symbionts often manipulate host reproduction to favour an infected matriline. This can trigger a mitochondrial sweep, confounding future taxonomic and phylogenetic interpretations of host mitochondrial gene sequence divergence (Shoemaker et al. 2004; Hurst & Jiggins 2005).
Gueguen et al. investigated prevalence and genetic diversity of bacterial symbionts in several previously identified clades of the B. tabaci complex, and compared these community characteristics via sequence data from host internal transcribed spacer (ITS) and the faster-evolving mitochondrial cytochrome oxidase I (COI) gene. They found a significant association between B. tabaci mitochondrial haplotype and composition of the associated endosymbiont community, but this association was weakened or lost when only a portion of the symbiont community was included in analyses. This result suggests that genetic studies of host–symbiont associations may miss important clues to the nature of host population differentiation if only one symbiont at a time is included. Furthermore, while the relative diversity of host nuclear and mtDNA markers among groups appeared consistent with selective sweeps, the presence and prevalence of any given symbiont was variable across groups of hosts. Also, some combinations of endosymbionts (e.g. Cardinium and Rickettsia) were never found in the same host. While other factors such as founder effects or differences in exposure history could explain these phenomena, it is also possible that community interactions can maintain symbionts at intermediate prevalences, and that competitive exclusion can occur among symbionts.
While endosymbiont community composition in the Gueguen et al. study supported the mtDNA phylogeny of subgroups within B. tabaci clades, the taxonomic and ecological value of this approach is not limited to providing static markers of membership in host populations or taxa. Endosymbionts and their hosts form dynamic communities because of their interactions within the host (Oliver et al. 2008; Jaenike et al. 2010) and establishment of novel endosymbionts can be influenced by geography (Russell et al. 2009), host ecology (Stahlhut et al. 2010), and host phylogenetic similarity (Baldo et al. 2008). Although they use some host resources, endosymbionts can in turn confer nutritional, immunity, or reproductive advantages that stabilize the association. Endosymbionts can even reinforce reproductive isolation among hosts and thus contribute to speciation events (Shoemaker et al. 1999). Symbionts therefore carry historic evidence of horizontal and vertical transmission events, selective sweeps, and trade-offs between their benefits and costs to the host. Teasing out the individual and interactive effects of multiple symbionts on a host individual, population or species is arguably a daunting task, but characterizing host–symbiont community composition is a necessary first step, and one that provides information valuable in its own right. An ecologically and behaviourally well-studied host like the B. tabaci complex provides a natural model for examining the intricate roles of endosymbionts in both the practical identification and the natural history of members of a taxonomically difficult group.