I am quite surprised that Carl Zimmer, in research for his book Parasite Rex, did not encounter the fascinating case of the Ampulex compressa (Emerald Cockroach Wasp) and its prey/host the American Cockroach (Periplaneta americana, see also comments on Aetiology and Ocellated).
In 1999, I went to Oxford, UK, to the inaugural Gordon Conference in Neuroethology and one of the many exciting speakers I was looking forward to seeing was Fred Libersat. The talk was half-hot half-cold. To be precise, the first half was hot and the second half was not.
In the first half, he not just introduced the whole behavior, he also showed us a longish movie, showing in high magnification and high resolution all steps of this complex behavior (you can see a cool picture of the wasp's head here).
First, the wasp gives the roach a quick hit-and-run stab with its stinger into the body (thorax) and flies away. After a while, the roach starts grooming itself furiously for some time, followed by complete stillness. Once the roach becomes still, the wasp comes back, positions itself quite carefully on top of the raoch and injects its venom very precisely into the subesophageal ganglion in the head of the roach. The venom is a cocktail of dopamine and protein toxins so the effect is behavioral modification instead of paralysis.
Apparently, the wasp's stinger has receptors that guide it to its precise target:
"To investigate what guides the sting, Ram Gal and Frederic Libersat of Ben-Gurion University in Beer-Sheva, Israel, first introduced the wasp to roaches whose brains had been removed. Normally, it takes about a minute for the wasp to find its target, sting, and fly off. But in the brainless roaches, the wasps searched the empty head cavity for an average of 10 minutes. A radioactive tracer injected into the wasps revealed that when they finally did sting, they used about 1/6 the usual amount of venom. The wasps knew something was amiss."
The wasp then saws off the tips of the roach's antennae and drinks the hemolymph from them. It builds a nest - just a little funnel made of soil and pebbles and leads the roach, by pulling at its anteanna as if it was a dog-leash, into the funnel. It then lays an egg onto the leg of the roach, closes off the antrance to the funnel with a rock and leaves. The roach remains alive, but completely still in the nest for quite some time (around five weeks). The venom, apart from eliminating all defence behaviors of the roach, also slows the metabolism of the cockroach, allowing it to live longer without food and water. After a while, the wasp egg hatches, eats its way into the body of the roach, eats the internal organs of the roach, then pupates and hatches. What comes out of the (now dead) cockroach is not a larva (as usually happens with insect parasitoids) but an adult wasp, ready to mate and deposit eggs on new cockroaches.
Why was the second half of the talk a disappointment? I know for a fact I was not the only one there who expected a deeper look into evolutionary aspects of this highly complex set of behaviors. However, the talk went into a different direction - interesting in itself, for sure, but not as much as an evolutionary story would have been. Libersat described in nitty-gritty detail experiments that uncovered, one by one, secrets of the neuroanatomy, neurophysiology and neurochemistry of the cockroach escape behavior - the one supressed by toxin - as well as the chemistry of the toxin cocktail. Ganglion after ganglion, neuron after neuron, neurotransmitter after neurotransmitter, the whole behavior was charted for us on the screen. An impressive feat, but disappointing when we were all salivating at a prospect of a cool evolutionary story.
He did not say, for instance, what is the geographic overlap between the two species. I had to look it up myself afterwards. American cockroach can be found pretty much everywhere in the world. The wasp also has a broad geographical range from Africa to New Caledonia (located almost directly between Australia and Fiji) and, since 1941, Hawaii (another example of a non-native species wreacking havoc on the islands), but not everywhere in the world, especially not outside the tropics - there are most definitely parts of the planet where there are roaches but no Ampulex compressa.
In most cases in which one species is suspectible to the venom or toxin of another species, the populations which share the geography are also engaged in an evolutionary arms-race. The victim of the venom evolves both behavioral defenses against the attack of the other species and biochemical resistance to the venom. In turn, the venom evolves to be more and more potent and the animal more and more sneaky or camouflaged or fast in order to bypass behavioral defenses.
There are many examples of such evolutionary arms-races in which one of the species is venomous/toxic and the other one evolves resistance. For instance, garter snakes on the West Coast like to eat rough-side newts. But these newts secrete tetrodotoxin in their skins. The predator is not venomous, but it has to deal with dangerous prey. Thus, in sympatry (in places where the two species co-exist) snakes have evolved a different version of a sodium channel. This version makes the channel less susceptible to tetrodotoxin, but there is a downside - the snake is slower and more lethargic overall. In the same region, the salamanders appear to be evolving ever more potent skin toxin coctails.
Similar examples are those of desert ground squirrels and rattlesnakes (both behavioral and biochemical innovations in squirrels), desert mice (Southwest USA) and scorpions (again it is the prey which is venomous), and honeybees and Death's-Head sphinx-moths (moths come into the hives and steal honey and get stung by bees after a while).
But Libersat never wondered if cockroaches in sympatry with Emerald wasps evolved any type of resistance, either behavioral or physiological. Perhaps the overwhelming number of roaches in comparison with the wasps makes any selective pressure too weak for evolution of defenses. But that needs to be tested. He also never stated if the attack by the wasp happens during the day or during the night. Roaches are nocturnal and shy away from light. The movie he showed was from the lab under full illumination. Is it more difficult for the wasp to find and attack the roach at night? Is it more difficult for the roach to run away or defend itself during the day? Those questions need to be asked.
Another piece of information that is missing is a survey of parasitizing behaviors of species of wasps most closely related to Ampulex compressa. Can we identify, or at least speculate about, the steps in the evolution of this complex set of behaviors (and the venom itself)? What is the precursor of this behavior: laying eggs on found roach carcasses, killing roaches before laying eggs on their carcasses, laying eggs on other hosts? We do not know. I hope someone is working on those questions as we speak and will soon surprise us with a publication.
But let me finish with a witty comment on Zimmer's blog, by a commenter who, for this occasion, identified as "Kafka":
"I had a dream that I was a cockroach, and that wasp Ann Coulter stuck me with her stinger, zombified my brain, led me by pulling my antenna into her nest at Fox News, and laid her Neocon eggs on me. Soon a fresh baby College Republican hatched out, burrowed into my body, and devoured me from the inside. Ann Coulter's designs may be intelligent, but she's one cruel god."
Update: That post on The Loom attracted tons of comments. Unfortunately, most of them had nothing to do with the cockroaches and wasps - Carl's blog, naturally, attracts a lot of Creationists so much of the thread is a debate over IDC. However, Carl is happy to report that a grad student who actually worked on this wasp/cockroach pair, appeared in the thread and left a comment that, among else, answers several of the behavioral and evolutionary questions that I asked in this post.
Update 2: You can watch some movies linked here and here.