spiders from mars?

there were a handful of science news stories out this past week about how a couple of researchers reportedly discovered a case of “group selection” in certain spiders (Anelosimus studiosus or tangle web spiders): for example, see Proving ‘group selection’: Spider colonies need the correct mix of personalities to survive and Elusive Form of Evolution Seen in Spiders. a bunch of people on twitter got all excited about this finding, because they wonder if (some of them i think hope that) group selection might also apply to groups of humans. i agree: that would be very interesting to know one way or the other. so i went and read the original paper — Site-specific group selection drives locally adapted group compositions — to see what these guys had found.

before i offer up my admittedly layman’s thoughts on this paper, let me first say what a really neat piece of research this was! if there existed a nobel prize for geeky dedication and sheer nerdiness, these guys would’ve won it! — and i mean that as a compliment! the researchers, pruitt and goodnight, studied groups of tangle web spiders in the wild, captured some and brought them back to their lab, conducted personality tests on the spiders (yes! there are apparently personality tests for spiders!), painstakingly painted those little dots on the backs of individuals to keep track of them (you know, like how they sometimes do with bees), bred the spiders, released new groups of them back into the wild, and checked up on them one and two generations later to see how they fared. this is some really cool research! nerds ftw! (^_^)

but did they find evidence for group selection?

weeeellll, no, i don’t think so.

to begin with, right at the start of the paper pruitt and goodnight (p&g) define group selection as “selection caused by the differential extinction or proliferation of groups.” eeeehhhh, as far as i understand it, that’s not really the definition of group selection, and even the authors admit that their definition is a “broad” one.

group selection is more accurately defined as when “natural selection [operates] between groups of organisms, rather than between individuals.” in contrast, p&g’s broad definition could theoretically include cases in which natural selection worked between individuals (individual selection) which also just incidentally happened to result in the proliferation of the group to which the lucky selected individuals belonged. an example of this is the selection for lactase persistence in some humans in which those individuals who could drink milk as adults were able to leave behind more descendants than those individuals who could not. while lactase persistence might indeed have benefitted groups of milk-drinking individuals, natural selection did not act on the group, but rather on the individuals in that group. (pretty sure i stole this example from @supermisdreavus, but i can’t find where he said that right now.)

in other words, you always need to work out what the target of selection is: the group or the individuals that make up the group. (really it’s ultimately the genes, but — oh, nevermind.) remember that “‘a fleet herd of deer’ is really just a herd of fleet deer.”

so, really, the discussion could end right here, because i don’t think the authors are talking about group selection proper. but, since i’ve read the whole paper, i’ll carry on. (yes, i’m one of those people who’s never learned to quit while they’re ahead!)

a. studiosus spiders live either as solitary individuals or in groups where they cooperate on tasks like hunting and the raising of young. the individuals that live together in groups are, on average, more closely related to one another than those that live alone [pdf] — they’re generally as related to one another as though they were half-siblings. one reason why they’re probably not more related to one another in these groups — like to the degree that ants or bees in colonies often are — is that the males move between groups. remember that.

the personality types of the individual spiders in a. studiosus groups come in two sorts: docile and aggressive. the docile spiders are typically pretty laid back and aren’t much bothered by the presence of other spiders (even spiders from other species), whereas the aggressive individuals like their space — they’ll chase off other individuals. individuals of both types are found in groups of a. studiosus, but the frequencies vary. from the paper:

“At…high-resource sites, small colonies were dominated by docile females and the frequency of aggressive individuals increased with colony size. By contrast, at low-resources sites, small colonies were dominated by the aggressive phenotype and the frequency of the docile phenotype increased with colony size.”

well, that doesn’t sound too surprising at all. in locales where there is plenty of resources, there are more laid back individuals in the colonies, prolly ’cause being laid back works just fine. in areas where resources are lacking, more aggressive individuals do better. btw, they found that the heritability of these personality types in the spiders is 0.66.

groups that have more docile individuals (i.e. the ones in high-resource areas) are at a greater risk of invasion by other types of spiders which, over the long-term, tends to be a really bad thing for an a. studiosus colony (i.e. it’s usually destroyed). groups that have more aggressive individuals (i.e. the ones in low-resource areas) tend in bad times to experience too much “egg case cannibalism.” needless to say, that’s not a good thing over the long-term either.

what p&g did in their study was to introduce into the wild — into differing environments — groups having varying frequencies of these personality types [source]:

“He [pruitt] took spiders from warrior-heavy colonies and used them to assemble new groups that were heavy on the nannies. He also used spiders from mostly docile colonies to create warrior-laden groups. In addition, he assembled control groups that matched the composition of their original groups.”

what they found was that after three generations:

“60 percent of the colonies were extinct. Control groups that returned to their ancestral homes tended to do well, and those that were transplanted into a new environment generally died. Neither of these outcomes was much of a surprise.

The most interesting results came from colonies made up of spiders that had been forced into a composition different from the one they grew up in — warrior-majority colonies containing spiders from mostly docile groups, for example. The colonies whose composition fit the new environment tended to survive. But over time, surviving colonies reverted to their members’ original group composition. The warrior-majority colonies went back to having more nannies, for example. On the face of it, this is bizarre behavior; if the colonies are well-suited to their environment, why not maintain that ratio? It seems that some innate sense, perhaps encoded in the spiders’ genes, pulled the colony back to its original configuration, even though this change meant the colony would perish.”

well, i dunno. is that really “bizarre behavior?” i mean, if the personality types of a. studiosus are really highly heritable (0.66), is it strange that a population having come from a bunch of docile individuals should regress toward a docile mean? and vice versa? don’t forget, too, that the individuals in these groups are all related to one another as though they were half-siblings, so presumably individuals of either personality type might carry a great many genes of the other type in their genomes. (don’t know about that — i’m just guessing here, tbh.)

what really made me question whether or not this is “bizarre behavior” is the way in which the researchers bred the spiders when they had them in captivity [from the methods section at the end of the paper]:

“Females were mated randomly to a male of like behaviour type from their same source population, but which was collected from a source colony >5m distance.”

hmmmm. i dunno about that. they mated all the females with males of the same personality types, docile or aggressive? i’m guessing that they did this in order to reduce the number of possible confounding factors in the study, but i’m afraid they might’ve added something to the mix here that wouldn’t be found in nature, i.e. a 100% assortative mating rate (for personality type). mightn’t this almost guarantee that individual spider lineages would regress to their original personality-type means? docile females always mated with docile males and aggressive females always mated with aggressive males? that seems unlikely to happen in nature, especially given the fact that the males normally leave their colonies and move to others. (btw, male a. studiosus spiders prefer moving into colonies over mating with lone females. typical males, favoring harems! (~_^) )

p&g offer a number of explanations for how the frequencies of personalities in the groups might change over time:

“How native spiders are actually able to adjust their composition is unknown, but plausible regulatory mechanisms include developmental plasticity in the docile:aggressive phenotypes, policing of group membership, phenotype-biased dispersal, and/or selective cessation of reproduction.”

they reject the first explanation (the plasticity one) on the basis (in part) of the rather high heritability of spider personality types which they found. i’m inclined to agree with them on that.

out of their other reasons, policing of group membership and selective cessation of reproduction are behaviors that can be easily explained by natural selection between individuals, especially in populations that have rather highly related individuals so that levels of altruism are pretty high. the selective cessation of reproduction occurs, for instance, in some ant colonies since, due to the really high degrees of relatedness between individuals, the inclusive fitness payoffs are really large (eg. if you share three-quarters of your dna with your sister’s offspring, there’ll be a greater genetic payoff in helping her to reproduce rather than reproducing yourself, since you’d only share half of your genome with your offspring). that’s individual selection, not group selection. h*ck! both behaviors also occur in meerkat groups, although they, of course, show much less specialization of individuals than ants or bees. the policing of group membership can also be plausibly explained by natural selection between individuals — for example, aggressive individuals keep at bay all sorts individuals because that’s good for aggressive individuals (who are typically found in sparse environments).

so, i’m not at all convinced that pruitt and goodnight have found an example of group selection. i think they’ve found that genetics (as indicated by the heritability of the spiders’ personality traits) and natural selection certainly shape the average characteristics of groups, but it looks to me as though the seemingly “bizarre behaviors” that they found can easily be explained by individual selection. in fact, i’m more than a little concerned that due to the way they bred the spiders, p&g may have affected the outcomes of the reintroduced groups.

see also: The False Allure of Group Selection from steven pinker.

(note: comments do not require an email. a. studiosus group web!)

Advertisements

16 Comments

  1. Here’s the misdreavus tweet:

    http://twitter.com/SuperMisdreavus/status/477954228002910208

    I’d define group selection as selection that increases the frequency of a trait despite the fact that it is deleterious to individuals (or at least kin groups). I don’t buy for a second that this is what happened here.

    “well, i dunno. is that really ‘bizarre behavior?’ i mean, if the personality types of a. studiosus are really highly heritable (0.66), is it strange that a population having come from a bunch of docile individuals should regress toward a docile mean?”

    Regression to the mean is a one-time effect only. If you took a colony and altered its composition such that it went from being composed of docile to being primarily aggressive spiders, the next generation would regress part of the way towards its original mean, but then it would stop. It wouldn’t regress further. You’d need a continued flow of new genes from the outside to keep the regression going.

    Of course, if males move around, this would do the trick.

    It is non-assortative mating that causes regression to the mean (in humans). If cross-class mating was kept at a bare minimum (less than 3%) castes would eventually form.

    Reply

  2. @jayman – “I’d define group selection as selection that increases the frequency of a trait despite the fact that it is deleterious to individuals….”

    eh. i’m not sure that the “deleterious to individuals” part is necessary. shouldn’t the fact that the selection targeted the group and not individuals be enough? (altho how you’d ever show that…?)

    @jayman – “Regression to the mean is a one-time effect only.”

    right. but that may have happened here — in fact, i don’t see why it shouldn’t have — and they don’t consider it at all. -?- which i thought was strange.

    @jayman – “It is non-assortative mating that causes regression to the mean (in humans).”

    yes. but also repeated mating of like with like is how you breed for certain traits, no? if the authors bred docile w/docile and aggressive w/aggressive in first generation, you ought to get higher numbers of docile in docile groups and aggressive in aggressive groups even in next generation, no? esp. given high relatedness.

    so there’s likely: regression to the mean + the effects of assortative mating + the high relatedness compounding the assortative mating [edit:] going forward.

    Reply

  3. @hbd chick:

    “yes. but also repeated mating of like with like is how you breed for certain traits, no? if the authors bred docile w/docile and aggressive w/aggressive in first generation, you ought to get higher numbers of docile in docile groups and aggressive in aggressive groups even in next generation, no? esp. given high relatedness.”

    Yes. But left to their own devices, they wouldn’t regress after the first time unless new genes were coming in. After the first regression, they would have reached their new mean, and would stay there (selection/in and out-migration notwithstanding).

    Reply

  4. @jayman – “But left to their own devices, they wouldn’t regress after the first time unless new genes were coming in. After the first regression, they would have reached their new mean, and would stay there”

    yes. but there would be that initial regression to the mean (you would think!). and the researchers don’t seem to have considered it. at least not in the article. -?-

    and then, on top of that, there are also the poss effects of the assortative mating and high relatedness.

    Reply

  5. @jayman – i guess i’m thinking that there’s possibly multiple things going on here (edit: or just one or two of them):

    1) regression to the mean to the averages of the original populations, eg. the descendents of docile individuals taken from majority aggressive groups regressing towards aggressiveness (remembering that “groups” are just big families) — in first generation like you pointed out;

    2) the 100% assortative mating of docile w/docile and aggressive w/aggressive possibly screwing the effects, esp. since this wouldn’t happen in nature;

    and 3) the high relatedness of the individuals w/in groups amplifying the outcomes.

    Reply

  6. group selection is more accurately defined as when “natural selection [operates] between groups of organisms, rather than between individuals.”

    But I thought that individuals were groups — groups of genes that is. I really don’t see the problem with group selection among humans. Darwin was ok with it, wasn’t he? Groups with certain moral characteristics, with presumably a genetic basis, might prevail over groups without those characteristics. I know I am growing feeble, but somebody please explain why group selection is a problem.

    Reply

  7. Regression to the mean is a one-time effect only. If you took a colony and altered its composition such that it went from being composed of docile to being primarily aggressive spiders, the next generation would regress part of the way towards its original mean, but then it would stop. It wouldn’t regress further. You’d need a continued flow of new genes from the outside to keep the regression going.

    I don’t believe this is what Morgan’s research on bristle number of fruit flies showed. He selected for high bristle number for many generations until he had a generation with much higher number than the wild type. Then he let this population mate with itself but with no further selection. It gradually returned the to bristle number of the wild type. I read about this experiment in The Eighth Day of Creation, a great history of genetic research before the genomic revolution.

    Reply

  8. No matter if group selection actually exists or not, it certainly has an interesting history. For example, Dawkins claimed in “The Selfish Gene” that Robert Ardrey, Konrad Lorenz and Irenäus Eibl-Eibesfeldt were “totally and utterly wrong” for supporting group selection. Pinker then seized on this phrase to dismiss Ardrey as “totally and utterly wrong,” period, in “The Blank Slate.” He then managed the ludicrous feat of writing a whole tome about the affair that dismissed Ardrey, the most influential and significant opponent of the Blank Slate orthodoxy in its heyday, in a single paragraph.

    History was subsequently “rearranged” to make E. O. Wilson the knight in shining armor who defeated the Blank Slate single-handedly and became the “father of evolutionary psychology” with the publication of “Sociobiology.” Only two chapters of that book dealt with human nature, and they were both merely repetitions what Robert Ardrey and Konrad Lorenz had written a decade and more earlier. Ardrey’s significance is documented in “Man and Aggression,” a remarkably useful piece of historical source material published in 1968, consisting of a collection of Blank Slater essays, mainly aimed at Ardrey. See, in particular, the essays by Geoffrey Gorer in which he specifically identifies Ardrey as the most influential opponent of the Blank Slate. If more evidence were needed, one need only look at the book review of the “Sociobiology” that appeared in the “New York Times Review of Books.” It was submitted by a gang of pious academics exactly like the one that attacked Nicholas Wade after the publication of “A Troubled Inheritance,” and it condemned Wilson because his ideas on human nature were similar to those of Ardrey!

    Fast forward to 2013, and E. O. Wilson published another book; “The Social Conquest of Earth.” It is a ringing endorsement of (you guessed it) group selection! Ever since, the glowing accounts of how Wilson was “the father of evolutionary psychology” have been dropping down the memory hole with alacrity. Of course, the kicker is that the article by Pinker you refer to in your post condemning group selection was a response to Wilson’s book. It’s not hard to imagine why he’s so sensitive about the subject. No matter, he who controls the present controls the past. In fact, the “history” of Ardrey’s legacy has been quite successfully rearranged. And why not? After all, he was a mere playwright who dared to embarrass the whole academic tribe.

    The latest wrinkle is that de Waal has just published “The Bonobo and the Atheist,” in which he slams Dawkins for not being nice to religious people. Like Pinker, he also dismissed Ardrey in a single paragraph as the author of the “Killer Ape Theory,” a straw man invented by his Blank Slate enemies. De Waal then proceeds to channel the real Ardrey! Everything he writes about human evolution reads like a rather less entertaining version of Ardrey’s “The Hunting Hypothesis.”

    Reply

  9. @Helian:

    The situation with Wilson you describe may have something to do with this:

    Math is Hard

    Definitely the case for E. O. Wilson. In a recent essay in the Wall Street Journal, he says that many of the most successful scientists are no more than semiliterate in mathematics, and if anything he’s worse than that.


    E. O. Wilson would have benefited from having that extra sense. If he had it, he might not have suggested that ridiculous “gay uncle” theory … Maybe if Wilson had ever learned to divide by two, he wouldn’t have made this mistake.

    Speaking of which – general intelligence and math ability are fairly well correlated. Maybe a lot of these low-math types just aren’t very smart. I’ve never seen any sign that E. O. Wilson is.

    It’s not enough just to come up with ideas. You need to express them quantitatively, at some point, in order to evaluate them.

    Reply

  10. @Jayman The concocters of the academic “history” of the Blank Slate would have done well to read Wilson more carefully. His dalliance with group selection actually goes back a lot further than “The Social Conquest of Earth.” Of course, he was never so explicit about it in his earlier work. One of the main reasons he is now so open about it is the mathematical cover provided by Harvard biologist and mathematician Martin Nowak and others.

    Reply

  11. “pruitt and goodnight (p&g) define group selection as “selection caused by the differential extinction or proliferation of groups.”

    Actually, not our definition. It is Wade’s definition from 1977. Been around a long time, and it has stood the test of time. That said, I actually prefer to say group selection is occurring when the fitness of an individual is a function of group membership. This is an even broader definition, but it is more in line with the mathematics of group selection, which show that there is no mathematical difference between a lot of frequency dependent selection and multilevel selection (need to go to broader terminology here.)

    “in contrast, p&g’s broad definition could theoretically include cases in which natural selection worked between individuals (individual selection) which also just incidentally happened to result in the proliferation of the group”

    Got that covered too. There is a method of statistical analysis called contextual analysis (CA) that works very nicely to distinguish group selection from indirect selection (changes in group properties due to selection on individuals.) It is technically correlational, but it is a simple extension of standard methods for measuring selection in natural populations. In the cases where it has been confirmed by manipulative studies it has always panned out. By the way, this is one of those cases.

    Also, just to know, in a theoretical paper early this year I was able to use CA to show that multilevel selection and kin selection are based on the same formula. Quit arguing, they are the same thing. If you are invoking kin selection you are invoking group selection, and vice versa. It does turn out they are not identical; however they differ only in the goals of the researchers. Kin selection is an optimality theory in the same sense as optimal foraging theory, and optimal clutch size theory. Multilevel selection and group selection are rate based theories. They tell us how much change occurs as a result of selection, and how fast the population should change in the next few generations. This is why most hard core geneticists and and all agronomists use MLS theory (yup, the eggs and bacon and toast you ate this morning are all the result of group selection). MLS provides rates, and predicted change, which are both hugely more useful then the kin selection result, which is where the population should eventually go — someday — if we are lucky — and give it enough time — and our model was right (better hope the genetics are as simple as you think they are!). This is why kin selection has devolved into the realm of theoretically inclined ivory tower types, while the MLS approach has been embraced by experimentalists who actually need useful results.

    But all that is a side show. The reason we can say unequivocally that this is a group level adaptation is that these spiders were out in the environment for more than a generation, and adjusted their behavior ratios by adding (or removing) individuals of the right (or wrong) behavior. They did this in a manner that was adaptive for their home environment regardless of the actual environment they were in. It is hard to construe this as anything other than a group level adaptation.

    Reply

Leave a Reply to dearieme Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s