Wolbachia in butterflies and moths: geographic structure in infection frequency

Abstract

Introduction: Butterflies and moths (Lepidoptera) constitute one of the most diverse insect orders, and play an important role in ecosystem function. However, little is known in terms of their bacterial communities. Wolbachia, perhaps the most common and widespread intracellular bacterium on Earth, can manipulate the physiology and reproduction of its hosts, and is transmitted vertically from mother to offspring, or sometimes horizontally between species. While its role in some hosts has been studied extensively, its incidence across Lepidoptera is poorly understood. A recent analysis using a beta-binomial model to infer the between-species distribution of prevalence estimated that approximately 40 % of arthropod species are infected with Wolbachia, but particular taxonomic groups and ecological niches seem to display substantially higher or lower incidences. In this study, we took an initial step and applied a similar, maximum likelihood approach to 300 species of Lepidoptera (7604 individuals from 660 populations) belonging to 17 families and 10 superfamilies, and sampled from 36 countries, representing all continents excluding Antarctica.

Results: Approximately a quarter to a third of individuals appear to be infected with Wolbachia, and around 80 % of Lepidoptera species are infected at a non-negligible frequency. This incidence estimate is very high compared to arthropods in general. Wolbachia infection in Lepidoptera is shown to vary between families, but there is no evidence for closely related groups to show similar infection levels. True butterflies (Papilionoidea) are overrepresented in our data, however, our estimates show this group can be taken as a representative for the other major lepidopteran superfamilies. We also show substantial variation in infection level according to geography - closer locations tend to show similar infection levels. We further show that variation in geography is due to a latitudinal gradient in Wolbachia infection, with lower frequencies towards higher latitudes.

Conclusions: Our comprehensive survey of Wolbachia infection in Lepidoptera suggests that infection incidence is very high, and provides evidence that climate and geography are strong predictors of infection frequency.

Keywords: Bacteria; Butterfly; Latitudinal gradient; Moth.

Fig. 1

Proportion of species and infection frequencies binned in 10 % intervals. Black bars describe the observed infection frequencies within samples from each of the 312 species. For these data, the bin boundaries are treated as upper bounds (so a sample prevalence of exactly 10 % would be placed in the 0-10 % category). Dark grey bars describe the expected proportion of species infected under best-fit beta distribution as estimated by Maximum Likelihood (Table 2), and the best-fit pdf (eq. 3, with ML parameter estimates α = 0.24 and β = 0.63, scaled for visualization) is also shown for comparison. The light grey bars show the expected proportion of species under the parameter estimates of Hilgenboecker et al. [11] for their ‘B(iii)’ arthropod data set

Fig. 2

Comparison of incidence of Wolbachia infection and geography. a-d Scatter plots and Spearman’s rank correlation tests between moment-based estimators of the mean prevalence and incidence [11] and the absolute latitude and longitude of sampling locations (negative latitudes are shown as grey points). e The distribution of Wolbachia infection in Lepidoptera worldwide, based on a survey of countries with at least three screened populations. Countries/territories include: American Samoa, Belarus, China, Croatia, UK, Fiji, France, Germany, India, Japan, Kazakhstan, Malaysia, Mongolia, Poland, Russia, Slovenia, South Africa, Spain, Thailand, Uganda, Ukraine and USA. The green portion of each pie chart represents the mean prevalence (μ) in each country

Fig. 3

Distribution of Wolbachia mapped on the Lepidoptera phylogeny of Regier et al. [33] Moment-based estimators of prevalence (μ), and incidence (x_c), are shown in green and grey respectively. For Gracillarioidea and Gracillariidae, the moment-based estimation method rendered nonsensical negative parameter estimates, and so these are not shown