Tales from the crypt: a parasitoid manipulates the behaviour of its parasite host

There are many examples of apparent manipulation of host phenotype by parasites, yet few examples of hypermanipulation—where a phenotype-manipulating parasite is itself manipulated by a parasite. Moreover, few studies confirm manipulation is occurring by quantifying whether the host's changed phenotype increases parasite fitness. Here we describe a novel case of hypermanipulation, in which the crypt gall wasp Bassettia pallida (a phenotypic manipulator of its tree host) is manipulated by the parasitoid crypt-keeper wasp Euderus set, and show that the host's changed behaviour increases parasitoid fitness. Bassettia pallida parasitizes sand live oaks and induces the formation of a ‘crypt’ within developing stems. When parasitized by E. set, B. pallida adults excavate an emergence hole in the crypt wall, plug the hole with their head and die. We show experimentally that this phenomenon benefits E. set, as E. set that need to excavate an emergence hole themselves are about three times more likely to die trapped in the crypt. In addition, we discuss museum and field data to explore the distribution of the crypt-keeping phenomena.


(a) Emergence hole diameter analysis
Hole diameter appeared to be associated with the identity of the emergent animal. Two measurements of hole diameter were made perpendicular to each other under a Leica M125 Stereo microscope, and the two measurements were averaged. We attempted to measure all holes on the stems collected during the August harvesting session, and a subset of holes on stems from the October harvesting session (focusing on stems that contained all three hole types). Holes were skipped if stem shape prohibited us from getting the hole in an appropriate plane to permit accurate measurements of hole diameter under the microscope.
We performed all statistical analyses in RStudio Version 0.99.489 [1] running R Version 3.2.2 [2]. Models were created using the lmer function in the lme4 package [3].
Model comparison using AICc, and model averaged beta coefficients with 95% confidence intervals were obtained using the AICmodavg package [4].
To explore how the infection status of B. pallida influenced the diameter of the associated emergence hole, we fit 4 linear mixed effects models. An identification code specific to the stem on which a hole was measured was included as a random intercept in all models, as B. pallida on the same stem may lack independence. The competing models included a null model (random intercept only), and the other 3 models contained either a term for B. pallida infection status, a term indicating whether the stems were harvested from Inlet Beach during the first or second sampling session, or both terms.
Model-averaged predictor estimates and 95% confidence intervals were obtained using the modavg function.

(a) Emergence hole diameter analysis
We measured the diameter of a total of 889 holes (404 holes associated with infected B. pallida (Figure 1, see below). B. pallida head capsule widths were ~0.3 to 0.5mm, so infected B. pallida excavate emergence holes that are about 50% smaller relative to the width of their head. Harvest session likely did not influence emergence hole diameters ( Figure 1).

Discussion
Gall wasps are frequently infected by a diverse community of parasitoids and inquilines [5][6][7]. In our system, we observed 12 natural enemies associated with B. pallida. While E. set accounted for the majority of the parasitoids emerging from the stems we collected, 20% of the emergences were comprised of other community members. While E. set appears to consume the adult stage of B. pallida and leaves evidence of its presence in the form of B. pallida exoskeleton scattered around the crypt, most parasitoids consume the immature host stages [8][9][10], and would be less likely to leave behind evidence of their presence if the parasitoid survived to emergence. This suggests that holes excavated by parasitoids other than E. set were likely mischaracterized as holes associated with uninfected B. pallida. As parasitoids are typically equal in size or smaller than their hosts [11], these parasitoids likely excavated smaller emergence holes, driving down our estimate of hole diameter for crypts associated with uninfected B. pallida. These mischaracterized holes may explain the small average diameter outliers in our dataset associated with uninfected B. pallida. Thus, our conclusion that infected B. pallida excavate smaller emergence holes than their uninfected conspecifics is likely sound, and the effect is greater than what we report here.