Still a Host of Hosts for WolbachiaAnalysis of Recent Data Suggests That 40% of Terrestrial Arthropod Species Are Infected

Wolbachia. Most people have never heard of it. Yet this bacterium lives inside roughly four out of every ten insect species on Earth — maybe more or maybe fewer, depending on how you count. For a long time, scientists kept getting the count badly wrong. That counting problem, and what it took to fix it, is what this lecture is about. Wolbachia is an intracellular bacterium, meaning it lives inside the cells of its hosts rather than between them or in the bloodstream. Its hosts are arthropods, including insects, spiders, mites, and crustaceans. This group makes up roughly 80 percent of all animal species on the planet. Wolbachia spreads primarily through vertical transmission, passing from an infected mother to her offspring. That single fact explains a lot about its biology, because vertical transmission means Wolbachia's evolutionary interests are tied directly to the reproductive success of females. Males are a dead end for the bacterium, so Wolbachia has evolved a remarkable set of tricks to tilt reproduction in favor of females and infected individuals. It can cause cytoplasmic incompatibility, where crosses between infected males and uninfected females produce no viable offspring, giving infected females a mating advantage. It can induce parthenogenesis, or reproduction without males, in some wasp species. It can kill male embryos outright or feminize genetic males into functional females. These are not subtle effects; they reshape the population genetics of every species in which they operate. Given all that, you can see why biologists wanted to know: how widespread is this bacterium? In 2008, Hilgenboecker and colleagues published a meta-analysis that pooled screening results from twenty separate Wolbachia surveys, covering more than nine hundred arthropod species, and analyzed the combined data with a beta-binomial statistical model. Their answer was sixty-six percent. More than two-thirds of arthropod species carry Wolbachia. That figure was striking — much higher than earlier direct surveys, which had reported around twenty percent — and it became widely cited. However, Hilgenboecker and colleagues were careful to flag two problems with their own estimate. First, many of the contributing studies tested only one or a handful of individuals per species. That creates a systematic bias: if a species has a low-prevalence infection — say, only five percent of individuals carry the bacterium — testing just two or three individuals gives you a high chance of drawing only uninfected ones. You would classify that species as uninfected when it isn't. Small sample sizes push toward false negatives, and false negatives drag the incidence estimate down. The second problem ran in the opposite direction. Some studies contributed very large samples, over one hundred individuals per species, and those studies tended to focus on species already suspected or known to be infected. Pooling those in with everything else inflates the estimate. Two biases are working in opposite directions, both introduced by non-random, uneven sampling. Hilgenboecker and colleagues tried to correct for this, but ultimately concluded that better data were needed: more individuals per species, and species chosen more randomly. That's where Roman Zug and Peter Hammerstein entered the picture. Rather than collect new field data, they identified an existing survey that happened to satisfy both criteria and applied the same beta-binomial framework to it. The dataset came from Duron and colleagues, who had screened one hundred thirty-six terrestrial arthropod species, totaling two thousand fifty-two individuals, for seven reproductive parasites. The species spanned fifteen orders and three classes — insects, arachnids, and crustaceans. No species had more than forty individuals sampled, and only twenty-five species had fewer than ten. The median sample size per species was fifteen, and the mode was twenty. That's the sweet spot; enough individuals per species to actually detect low-prevalence infections, but not so many that you're clearly targeting species you already know are infected. The sampling was broad and relatively unbiased. It addressed exactly the weaknesses Hilgenboecker's team had identified. The statistical framework Zug and Hammerstein used is worth understanding because the whole argument depends on it. The beta-binomial model works on two levels. At the lower level, within any given species, the number of infected individuals follows a binomial distribution — think of it as repeated coin flips where the probability of heads is that species' true infection rate. At the upper level, those per-species infection rates themselves follow a beta distribution across species. The model asks: given all the observed infection counts across all species, what is the most likely shape of that beta distribution? Then, what fraction of that distribution sits above a minimal threshold — the cutoff for calling a species truly infected? Zug and Hammerstein used a threshold of c equal to 0.001, meaning a species is counted as infected if at least one in a thousand individuals is expected to carry Wolbachia. The incidence x is then the integral of the prevalence distribution from that threshold up to one — the total probability mass above the cutoff. Their result: x equals 0.41. About forty percent of terrestrial arthropod species carry Wolbachia. Compare that to three numbers: the sixty-six percent from Hilgenboecker's pooled meta-analysis, the naive count from the Duron data itself which yields only twenty-two point eight percent, and the new model-based estimate sitting between them at forty percent. The model does real work here — it recovers infections that raw counting misses, but it doesn't inflate the estimate the way biased sampling did before. Zug and Hammerstein also confirmed a within-species pattern that Hilgenboecker's team had first identified: what the authors call the "most or few" distribution. Within any infected species, you almost never find a middling proportion of infected individuals. Either most individuals in the species carry Wolbachia or only a few do. The distribution is strongly bimodal. Biologically, this makes sense. When Wolbachia first arrives in a new host species — likely through occasional horizontal transfer between species — it starts at low frequency. If the bacterium's reproductive manipulation gives it a selective edge, it sweeps through the population and reaches high frequency. If it doesn't get that foothold, it stays rare or disappears. The middle ground is unstable. You see the outcome of that dynamic, not the process: high or low, almost nothing in between. Zug and Hammerstein also extended the analysis to the other reproductive parasites screened in the Duron survey: Arsenophonus, Cardinium, Rickettsia, Spiroplasma ixodetis, and Spiroplasma poulsonii. Flavobacterium was never observed in the dataset and was excluded. The variation in estimated incidences is large. Spiroplasma ixodetis comes in at about twenty-two percent, Cardinium at about sixteen percent, Arsenophonus at roughly seven percent, Rickettsia at about six percent, and Spiroplasma poulsonii at just three percent. Wolbachia at forty percent is more than twice the next competitor. The paper's conclusion is unambiguous: Wolbachia is the most abundant endosymbiont among arthropod species — and by a wide margin. Why does Wolbachia dominate? The paper points to its combination of traits: a diverse toolkit of reproductive manipulations, predominantly maternal transmission, and occasional horizontal transfers that let it jump into new host lineages. Each of those properties helps it both invade new populations and persist once established. The other symbionts have subsets of those tools, but Wolbachia has the full set. Step back and look at what this paper actually accomplished. The forty percent estimate changes the number that appears in textbooks and review articles — down from sixty-six percent to forty percent, a difference of twenty-six percentage points. That matters for any calculation of Wolbachia's impact on arthropod diversity. But the more durable contribution is methodological. Zug and Hammerstein didn't collect new organisms; they identified a better-designed existing dataset and showed that the choice of dataset changes the answer dramatically. Better sampling design — avoiding single-individual samples and biased large samples, covering a taxonomically diverse set of species — and consistent detection sensitivity together account for most of that twenty-six-point gap. Forty percent of terrestrial arthropod species. Arthropods are the dominant animals on Earth by species count. So, a bacterium that's established itself in four out of every ten of those species has had a hand in shaping the evolutionary history of an enormous fraction of all animal life. Wolbachia is not a curiosity; it's a force. And we're only now getting the count right. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.