Behaviour-Manipulating Parasites

Summary

Five parasites are famous for changing how their hosts behave.  The evidence behind them is uneven, and the popular stories run ahead of it.

• The zombie ant is the best-measured case. Ants infected by Ophiocordyceps fungi leave the canopy, bite a leaf vein near solar noon, and die locked in place. The fungus wraps the jaw muscles in linked cells but does not enter the brain.¹²
• Sacculina's effects on crabs are well measured, and its behavioural effect uses a program the crab already has. The parasitic barnacle castrates its crab and gives males a female-shaped abdomen. Male and female hosts then clean the parasite's egg sac and release its larvae the way a female crab releases her own young.³⁴
• Toxoplasma's effect on rodents is real but not as specific as advertised. Infected rats stop avoiding cat urine. Later work in mice found the lost fear extends to non-cat odours and tracks brain inflammation. No study has shown that infected rodents are eaten by cats more often in the wild.⁵⁶⁷
• Claims that Toxoplasma shapes human personality have not held up. In a New Zealand birth cohort of 837 people, infected and uninfected personalities were "indistinguishable."⁸
• Leucochloridium's snail behaviour was measured once, and bird predation on its broodsacs never. The evidence that birds attack the pulsing eyestalks is an 1874 study of captive birds.⁹
• Hairworms make crickets more likely to enter water, not seek it. The "suicide" story is unsupported.¹⁰⁷

The young-earth answers, from Answers in Genesis (AiG), the Institute for Creation Research (ICR) and the Creation Research Society Quarterly (CRSQ), are general.  None names a harmless original job for any of the five.  AiG says natural selection built them after the Fall;  ICR says it cannot have.

Verdict:  behaviour does test the design answer harder than anatomy, but only in the cases where the behaviour is precise and coordinated.  The ant fungus and the crab barnacle are those cases.  Toxoplasma and the hairworm need much less to be specified than their reputations suggest.  The young-earth answer is weakest at the zombie ant, where nobody has proposed what the fungus was for.

Part 1: The problem, in its makers' own words

Brian Thomas of ICR stated the problem in 2014.  He listed the ant fungus, the hairworm and the crab barnacle, then wrote:  "These parasites' complicated life cycles and surgically-precise host interactions leave virtually no doubt that they were intentionally fashioned. But how, when, and why could God's originally "very good" creation accommodate such morbid features?"¹¹  He saw both horns:  "If these parasites originated by creation, then they seem to imply an evil Creator that Scripture does not describe."¹¹

The mainstream framing puts the behaviour in the parasite's genes.  David Hughes and colleagues call these systems "extended phenotypes: parasite genomes controlling host behavior."¹  If that is right, the behaviour is not an accident of anatomy.  It is written in the parasite, and it has one visible effect:  moving the parasite to its next host at the host's expense.

That is why these cases are sharper than claws or fangs.  Predation, Venom, and Carnivory found that a claw can dig and a canine can grip fruit.  A program that makes an ant clamp a leaf and die has no obvious second use.  Parasitoid Wasps covers hosts eaten alive.

The test comes from the parent article:  a corruption account needs evidence of an earlier useful activity and of the change that disrupted it.

Part 2: Five systems, measured

Toxoplasma and the rat that stops fearing cats

Toxoplasma gondii reproduces sexually only in cats.  Rodents are intermediate hosts, so a rodent eaten by a cat carries the parasite back.

Measured.  In 2000 Manuel Berdoy, Joanne Webster and David Macdonald at Oxford put rats in outdoor pens with four scented corners:  the rat's own bedding, water, rabbit urine and cat urine.  The rats were laboratory-wild hybrids;  23 confirmed infected, 32 controls.  Controls avoided the cat corner.  Infected rats "showed no overall avoidance of areas with signs of cat presence."⁵  The other three odours were treated alike by both groups.  The paper's title word, attraction, rests on a subset:  the parasite changed behaviour, "in some cases turning their innate aversion into an imprudent attraction."⁵  The clear attraction was in the 14 most active rats, seven per group.

Ajai Vyas and Robert Sapolsky's Stanford group repeated it in 2007 with rats and mice.  Infected rats' share of time near bobcat urine, against bobcat plus rabbit, rose from 0.345 to 0.611, where 0.5 is chance.  They called the loss of fear "remarkably specific, because infection did not diminish learned fear, anxiety-like behavior, olfaction, or nonaversive learning."¹²  Their own data carry a limit:  a towel and collar soaked in cat odour were avoided equally by infected and uninfected rats.¹²

Disputed.  In 2020 Madlaina Boillat and colleagues at Geneva ran mice through a battery of tests.  With fox, bobcat, guinea pig and home odours on offer, infected mice showed "No particular attraction to bobcat odor."  They found that the parasite "lowers general anxiety in infected mice, increases explorative behaviors, and surprisingly alters predator aversion without selectivity toward felids."⁶  The size of the change tracked the number of brain cysts and blood markers of inflammation.  Hannah Johnson and Anita Koshy's 2020 review says negative rodent studies exist, and "Even within those studies reporting positive results, the results are more heterogenous than often recognized."¹³

Asserted.  That the change helps the parasite reach cats.  Jean-François Doherty's 2020 review in Proceedings B:  "there is no sound evidence that the behavioural changes in infected rodents increase the transmission of T. gondii to felids."  He adds that the main lineages in Europe and North America "are essentially clonal and do not need to reproduce sexually in felid hosts."⁷  Mechanism:  unknown.  Vyas found cysts in every brain region tested, with about twice the density in the amygdala.¹²  Boillat's data point to inflammation.

Humans.  Karen Sugden, Terrie Moffitt, Avshalom Caspi and colleagues tested 837 members of the Dunedin birth cohort at age 38.  They found 236 were infected (28.2%).  They tested psychiatric diagnoses, self-harm, crime, personality, IQ and 14 cognitive measures.  Suicide attempt was marginal (p = .06), and one cognitive measure was worse.  Their conclusion:  "there was little evidence that T. gondii was related to increased risk of psychiatric disorder, poor impulse control, personality aberrations or neurocognitive impairment."⁸  They concede low power for rare outcomes such as schizophrenia.⁸  Johnson and Koshy count "well over 200 papers" on human behaviour, and judge that "publication bias is likely a contributing factor, and most studies lack multifactorial analyses."¹³

The zombie ant's death grip

Measured.  Hughes and colleagues followed infected carpenter ants (Camponotus leonardi) in a Thai forest in September 2007.  Then:  "Transitions from erratic wandering to death grips on a leaf vein were abrupt and synchronized around solar noon."¹  Sixteen bites were timed.  The mean sun elevation at the bite was 80.3 degrees, near the day's maximum of 87.3.  Earlier work at the site found 98% of ants bite a major vein.  "Biting leaves is not part of the repertoire of healthy ants of this species."¹  All ten biting ants sectioned had wasted jaw muscles and heads packed with fungal cells.  "This lock-jaw means the ant will remain attached to the leaf after death."¹

Mechanism, partly known.  Maridel Fredericksen and colleagues imaged infected ants cell by cell in 2017.  "Fungal cells were found throughout the host body but not in the brain, implying that behavioral control of the animal body by this microbe occurs peripherally."  The fungal cells "joined together to form networks that encircled the muscles."²  The brain result rests on three ants per group.  In 2014, ICR's Thomas wrote that "the fungal extensions penetrate the insect's brain";  that was not yet tested, and the 2017 imaging found none inside it.¹¹

Unexplained.  Hughes's team wrote that their data "do not explain why infected ants occur on leaves of a distinct NNW orientation [20] or how the fungus causes its hosts to choose distinct parts of the leaf."¹  Their conclusion calls it "a complex example of behavioral manipulation requiring coordinated changes of host behavior and morphology."¹

Sacculina and the castrated crab

Sacculina carcini is a barnacle that lives as a root system inside a crab.  Only its reproductive sac, the externa, shows under the crab's tail flap.

Measured:  castration and feminisation.  Tommy Kristensen, Jens Høeg and colleagues measured green crabs from Danish waters in 2012.  "Infected males acquire a broader, longer and segmented abdomen, fringed with marginal setae."  Their copulatory appendages and legs shorten and their claws shrink.⁴  Males carrying an externa have abdomens 11.1% broader than uninfected males.⁴  The paper states that hosts of this barnacle family "become morphologically modified, sterilized and subjected to physiological and behavioural changes."⁴  The broad flap protects the parasite.  Among more than 17,000 infected crabs, 23.8% of males but only 14.6% of females bore scars of a lost externa, and "Externa loss usually leads to the death of the parasite."⁴

Measured:  brood care, in one small study.  Tohru Takahashi, Akiko Iwashige and Shuhei Matsuura filmed shore crabs infected by Sacculina polygenea in an aquarium in 1997.  When they glued dirt onto three externae, the hosts cleaned it off within about 20 minutes.  They recorded 12 larval releases, 7 by males and 5 by females.  "When the parasite released nauplii, the host assisted the parasite by adopting the stereotypic larval release behavior of female crabs."³  Males did it too.

Mechanism.  Kristensen's team reads the broad abdomen as the adaptation and the other male changes as "pleiotropic side effects" of "hormonal manipulation."⁴  The behaviour appears to be the crab's own brood-care program.  The parasite switches it on in males, who normally never use it.

Leucochloridium and the pulsing eyestalk

Leucochloridium flukes grow in snails as bright, banded broodsacs that swell into the snail's eyestalks and pulse.  The fluke matures in birds.

Measured, once.  Wanda and Tomasz Wesołowski compared infected and uninfected amber snails in Białowieża National Park, Poland.  Before their work, "The demonstration that the sporocysts also manipulate the behaviour of the snails is still largely missing."  Infected snails "moved farther, positioned themselves in more exposed and better illuminated places, situated higher in the vegetation."¹⁴  This was a field comparison of snails, not an experiment.

Asserted.  That birds take the broodsacs as caterpillars.  Ed Yong's report on the study says "The only evidence that birds are attracted to the broodsacs comes from an 1874 study in which captive birds attacked the sacs. No one has checked that wild birds behave in the same way."  He quotes the Wesołowskis:  "there is not a single study documenting attacks of definite passerine hosts on snails with broodsacs… We think that such a situation is quite embarrassing".⁹  David Attenborough's Trials of Life filmed a flycatcher eating an infected snail.⁹  A film is an anecdote, not a rate.  Mechanism:  unknown.

Hairworms and the cricket that jumps into water

Hairworms (phylum Nematomorpha, not roundworms) grow inside crickets and must reach water as adults.

Measured.  Frédéric Thomas and colleagues watched an open-air swimming pool for two years and saw nine insect species enter water and release hairworms.  In field and lab tests, "crickets infected by the nematomorph Paragordius tricuspidatus are more likely to jump into water than uninfected ones."  But "The idea that this manipulation involved water detection from long distances by infected insects is not supported."  They propose erratic wandering that "brings them sooner or later close to a stream," then a change that makes them enter.¹⁰  At a Japanese stream in 2011, Takuya Sato and colleagues found infected camel crickets and grasshoppers "20 times more likely to enter a stream."¹⁵

Mechanism, suggested.  A 2005 protein study by David Biron and colleagues of infected grasshopper brains "suggest[s] that the adult worm alters the normal functions of the grasshopper's central nervous system."  Worm proteins resembling insect Wnt proteins suggest "a case of molecular mimicry."¹⁶

Asserted.  Suicide and water-seeking.  Doherty:  "there is no empirical evidence supporting the hypothesis that infected hosts actively seek water or that hosts voluntarily jump into water when close to it."  What has support is that "infected hosts tend to move around more, perhaps erratically."⁷

The five together

Part 3: The young-earth answers, sorted

The young-earth writers state three positions.

AiG:  natural selection after the Fall.  Elizabeth Mitchell wrote in 2014:  "we can be confident that fungi, viruses, bacteria, flatworms, protozoans, and insects fulfilled helpful, not harmful, roles in that original world."  Since the Fall, "mutations, horizontally transferred genes, environmental changes, and host changes have produced many harmful organisms."  She concludes:  "Zombie-making parasites show how natural selection has functioned in this fallen world."¹⁷  Her mechanism is that parasites which "impair neural or hormonal control of behavior to serve the needs of the parasite are naturally selected."¹⁷  Her own summary of Toxoplasma, rats that "seek out cats," overstates the rodent data.¹⁷

ICR:  switched on, or broken.  Thomas rejects selection as the builder:  evolutionary origins would "imbue nature with insightful and creative powers that it simply does not have."¹¹  He offers two options.  "First, God may have planted potential parasitic tools and information into organisms at creation, but they were only switched on after the Curse. Second, some parasitic situations likely arose from Curse-caused mutations that damaged once-healthy symbiotic relationships."  He adds that "More research is needed to select from at least two possible answers."¹¹  AiG and ICR therefore disagree on whether natural selection can build a manipulator.

CRSQ:  designed for another job.  Mark Armitage wrote in 2000 on flukes, the group that includes Leucochloridium:  "A caring God may have designed trematodes, now recognized as parasites, to serve other functions before the Fall of man."  He asks, "Were these organisms designed from the start to perform functions they no longer perform?"  He also says the complex life cycles "were planned by the Creator," and that how symbionts became pathogens "is open for speculation."¹⁸

Joseph Francis's model of disease adds displacement:  disease can come from "post-Fall displacement or movement of the microbe from the site where it performed its beneficial function."¹⁹  Warren Shipton allows "front-loaded genes now facilitating pathological life cycles," and suggests "some nematodes were designed to regulate insect populations."²⁰  Neither discusses a behaviour manipulator.

The evidenced pile is empty.  No young-earth source found names a harmless function for any of these five, and no study offers one.  The nearest thing is a present benefit to a third party.  Sato found that crickets driven into the stream supplied 60% of the yearly energy intake of an endangered trout.¹⁵  That is a real ecological role.  The cricket still dies.

Part 4: Why behaviour is the harder test

Where the claim holds.  A claw can have a gentle use because it is a general tool.  A manipulation has one output:  the host goes where the parasite's next stage needs it.  The zombie ant shows this most clearly.  The fungus coordinates a climb, a bite timed near noon, a bite on a vein, muscle wasting that locks the jaw, and a spore stalk after death.  Hughes's team call it "requiring coordinated changes of host behavior and morphology."¹  A loss of function cannot produce that list.  Sacculina is similar:  castration, a hormonal change that broadens the male abdomen, and a borrowed brood program all serve the parasite's sac.

Where it overreaches.  The data shrink what the parasite must specify in three of the five cases.

• Toxoplasma's rodent effect may be a side effect of brain inflammation that lowers fear in general.⁶ Most lineages do not need cats at all.⁷ If so, nothing about cats was specified.
• The hairworm's measured effect is more wandering and more water entry, not navigation.¹⁰
• Sacculina writes no new behaviour. It turns on the crab's existing brood care in the wrong sex.³

So a behaviour that must be written into the parasite is shown for the ant fungus and partly for Sacculina.  It is not shown for Toxoplasma, the hairworm or Leucochloridium.

Where the young-earth answer is weakest.  At the zombie ant.  AiG's answer, selection in a few thousand years, is the answer ICR says selection cannot give.  ICR's first option, tools "planted" at creation and "switched on after the Curse," concedes that the program was designed for harm, then moves the harm to a later date.  Its second option, a damaged symbiosis, has no candidate:  no one has proposed what harmless relationship an ant and this fungus once had.  Armitage's "other functions" is unnamed.

Where the mainstream account is also thin.  Hughes's team wrote in 2011:  "Future work should address the genetic basis of such extended phenotypes."¹  The behaviours are measured;  their origins are not reconstructed.  That gap is not evidence for creation, but it means neither side has a step-by-step history.

Objections and responses

"Toxoplasma proves parasites can rewrite a mammal's mind."  The rodent loss of fear is real.  Its cat-specificity is disputed, its link to predation is untested, and in mice it tracks inflammation.⁶⁷  In people the best cohort found nothing in personality.⁸

"These parasites show a cruel designer."  This article tests the biology, not the theodicy.  The biology shows that the ant fungus and the crab barnacle are precise, and that no one has shown an earlier harmless form.  Whether God may permit such systems is outside this article's scope, as the parent article notes.

"God could have front-loaded the genes and switched them on at the Curse."  That is ICR's first option.¹¹  It is testable.  If free-living relatives carry silent copies of the fungal effector genes, the option gains support.  No young-earth source found has looked.

"No evidence of an original role is not evidence that there was none."  True.  An empty pile is a gap, not a disproof.  Mitchell's selection account also shares the mainstream account's gap:  no one has shown the steps.

Issues and unanswered questions

• Full texts not read. Thomas et al. 2002 and the Wesołowskis' paper were read from abstracts only; both publishers returned bot challenges. The Wesołowskis' sentence on bird attacks is quoted through Ed Yong. Andersen et al. 2009, source of the 98% vein figure, was used only through Hughes 2011.
• Small samples. Takahashi's crab behaviour rests on 12 releases and 3 grooming tests. Fredericksen's brain result rests on 3 ants per group. Berdoy's attraction rests on 14 rats.
• No young-earth source on Sacculina or Leucochloridium specifically. Thomas names Sacculina; Armitage writes on other flukes. Web searches of the AiG, CMI and ICR sites found nothing on either, though CMI's own search page did not return results to a script.
• Toxoplasma may not be a manipulator at all. That helps a displacement account, but no creationist has made the argument from the data.

Verdict

The five famous manipulators are not equally real.  The zombie ant's timed death grip and Sacculina's castration and brood care are well measured.  Toxoplasma's rodent effect is real but general, its human personality effect has not survived a good cohort, Leucochloridium's bird link is untested, and the hairworm raises the odds of water entry without steering.

Where the behaviour is precise, it tests the design answer harder than anatomy, because it has no evident second use.  The young-earth literature has not met that test.  AiG hands the work to natural selection, ICR offers two options without choosing, and CRSQ proposes unnamed "other functions."  The zombie ant is where the answer is weakest.

Sources