Monthly Archives: October 2011

Got questions about inclusive fitness?

Over at his blog, Andrew Gelman briefly mentions the recent profile of E. O. Wilson in the Atlantic, and along the way mentions the dustup over inclusive fitness that I may have mentioned here before  (did I? It’s hard to remember).   At the end, he makes a throw-away comment which drove me nuts:

The article also discusses Wilson’s recent crusade against selfish-gene-style simplifications of human and animal nature. I’m with Wilson 100% on this one. “Two brothers or eight cousins” is a cute line but it doesn’t seem to come close to describing how species or societies work, and it’s always seemed a bit silly to me when people try to loop everything back to a selfish-gene story.

I’ve been trying to think of a similarly aggravating comment to make about statistics in return;  maybe “lies, damned lies, and statistics”?  “You can prove anything with statistics”?  “Bayesian statistics suck because I don’t understand where priors come from?”  It bugged me enough that I left this comment:

It doesn’t seem like you know much about inclusive fitness, a theory has been massively successful in evolutionary biology. Despite the odd and unsupported comments made by Nowak et al., it stands firm as a well-supported and useful body of theory. Here’s a link to the letter published in response to Nowak et al.’s original article, signed by 137 authors including most of the field’s brightest minds:

http://www.nature.com/nature/journal/v471/n7339/full/nature09831.html?WT.ec_id=NATURE-20110324

The appeal to authority doesn’t mean that they’re right, of course, but extraordinary claims require extraordinary evidence; Nowak et al. have done nothing but ignore the entire published literature on inclusive fitness spanning decades and comprised of hundreds if not thousands of studies, while proposing a mathematical model that adds nothing to our understanding beyond what current theory already provides.

I respect your work on statistics, have always enjoyed reading your blog, and your book (BDA) is sitting on my shelf right now, but your offhand comment above is uninformed and very aggravating; I’d like to deal with that aggravation by offering to assist you in understanding one of the most powerful explanatory mechanisms in evolutionary biology. The letter above provides a succinct summary of the evidence that Nowak et al. ignore, but it might be a bit much for a non-technical audience; I haven’t published directly in this field, but I do work in evolutionary biology and I should be able to answer any specific questions you may have if you would like to pose them. If I can’t answer them myself, I will find people who can.

I’m not going to go into a full blown recapitulation of inclusive fitness theory and then defend it, because I’d have to write several inconveniently long books to do so.  But since I made the offer over there, I’ll make it here too for any interested readers:  if you have questions burning you up about this whole “inclusive fitness” thing, ask them here in the comments and I will do my best to answer them for you.  And if I don’t know what the answer is, I’ll find it.  No question is too small, though I make no promises on how long or short my answers will be!

I’ll leave off with a quotation from a fantastic book by Andrew Bourke that I’m reading right now, Principles of Social Evolution:

Like any large and active field of investigation, the theoretical study of social evolution is not free from disagreements and unresolved issues (e.g. Taylor and Nowak, 2007; West et al. 2007a).  Paradoxically, while the potential richness of inclusive fitness theory as a general theory of social evolution is still underappreciated, the theory is sometimes perceived as an entrenched orthodoxy. A tendency therefore exists for iconoclastically-minded theoreticians to derive models of cooperation in novel ways and then announce them to be fundamental additions to existing theory (e.g. Killingback et al. 2006; Nowak 2006; Ohtsuki et al. 2006; Traulsen and Nowak 2006).  It is healthy for orthodoxies to be continually challenged by new theories and new data.  However, to date, these models have fallen short of true novelty, as other authors have shown that their results are capable of being derived from inclusive fitness theory (e.g. Grafen 2007a, 2007b; Lehmann et al. 2007a, 2007b; West et al. 2007a).  Indeed, inclusive fitness theory has a long history of successfully assimilating apparent challenges and alternatives (Grafen 1974; Queller 1992; Lehmann and Keller 2006a).  This is not surprising when one considers its deep foundations in the theory of natural selection.  Although it is premature to declare a consensus, a substantial body of opinion therefore holds that claims of fundamental extensions to inclusive fitness theory will have to be radically innovative, as well as robust, to be accepted as such (e.g. Lehmann and Keller 2006a; West et al. 2007a).  For all these reasons, Hamilton’s (1964) inclusive fitness theory will underpin the conceptual reasoning employed throughout this book (pp. 22-23).

 

Advertisements
Tagged , , , , , ,

Death in the nest: trade-offs rule the day.

Underlying many research programs in biology is the meta question: why is there more than one type of X?  (In continuous form, why is there variation in X?)  This question recurs in many areas of animal behaviour, and indeed in the entirety of the study of evolution itself.  Some examples include:

  • Why do animals show variation in “personality” – why are some consistently more aggressive, more exploratory, bolder, etc.
  • Why are there more than one type of male that females select between?  Why are some “attractive” and others “unattractive” – why aren’t they all attractive? (Sexual selection).
  • If aggressive signals like roaring can make other animals give up a resource or back down from a fight, why don’t all animals use the aggressive signal?  Why is there variation in signal type when all animals should use the same signal, which would then lose all meaning and be ignored?
  • Why do some animals invest heavily in each offspring while others produce as many as they can and invest very little in each?
The general response to most of these questions hinges on the idea of a trade-off.  In its most basic form, a trade-off involves giving up one thing to get (or avoid) another.  In particular, animal behaviour often hinges on cost-benefit tradeoffs.  It is desirable to have some trait or perform some behaviour, but doing so may come with a cost if we have too much of the trait or perform the behaviour too often or at all.   Examples of this litter the pages of any textbook in the biological sciences, from molecular biology up to zoology and ecology;  in particular, we can begin to address the questions I listed above by appealing to trade-offs:
  • Some personality types, like aggressive or exploratory, can confer benefits – such as always winning fights or being the first to find food – but also come with costs – such as the injuries from always fighting or the cost of being eaten while you try to be the first to eat.  Some individuals will be willing to make this trade-off, others will not.
  • The answer to this question has filled entire bookshelves, but here’s one tiny example:  in 1975, Amotz Zahavi published a landmark paper proposing that attractive males are “handicapped”;  they willingly trade off the cost of the handicap for the increased number of matings of come with it.  Zahavi’s “handicap principle”  suggested that this was a reliable indicator of quality to females because only some males would have the required quality (be strong enough, fast enough, etc) to bear the cost of the handicap in order to reap the benefit.
  • One of the most well-known answers to this question began the field known as evolutionary game theory;  at the end of the 1970s, the tragic figure of George Price and the eminent John Maynard Smith answered the question by showing mathematically how frequency-dependence could lead to a trade-off between Hawks, who are aggressive, and Doves, who back down at the first sign of trouble;  when Hawks are extremely common, their aggression leads them into costly fights against each other, which reduces the benefit of aggressiveness and makes Dove-ish behaviour more attractive.  But when Doves are common, Hawks get immense benefit with no cost by bullying Doves around.  (There’s actually significant overlap between this point and the previous, but that’s a topic for another blog post!)
  • An entire branch of evolutionary biology, life history theory, deals with questions like this:  in the face of limited resources, how do individuals make choices about the timing and sequence of events in their life to maximize their fitness?

This general pattern underlies the story behind a neat new advance-access paper from the groups of Alex Kacelnik and Juan Reboreda that manages to give away the good stuff in the title:

Ros Gloag, Diego T. Tuero, Vanina D. Fiorini, Juan C. Reboreda, and Alex Kacelnik. The economics of nestmate killing in avian brood parasites: a provisions trade-off. Behavioral Ecology, 2011.

Here, the question of types and the answer of trade-offs arises in the context of brood parasitism.  Brood parasites are organisms – birds, fish, insects – that relieve themselves of the responsibility of parenthood by tricking other organisms into doing it for them.  In birds, this usually takes the form of brood parasites laying their eggs in other species’ nests, where the enterprising young tykes then pretend to be the offspring of the unlucky suckers who are to play host.  Brood parasites can be specialists that only parasitize the nests of a target host species (or small group of species); an example of this is village indigobirds, who generally parasitise fire-finches (and who also display an interesting mechanism where the young copy the songs of the host species).  Generalists, on the other hand, will parasitise a range of host species;  cowbirds, for instance, are generalists.  Brood parasites can also vary in whether they eliminate the other offspring of the host that they have colonized (nestmate killing) or whether they attempt to blend into the crowd (nestmate tolerant).  To make this more concrete, take a look at this short video showing a newly-hatched cuckoo ejecting a reed warbler chick from the host nest:

The paper I’m talking about here explores an interesting question about brood parasites, namely:  why are some brood parasites nestmate tolerant while others are nestmate killers? Gloag et al. propose a mathematical model that explains this in terms of a “provision trade-off”.  Host nestlings can help the newborn parasite by stimulating the host parents to bring more food than the parasite could solicit alone, and if the parasite can outcompete its nestmates for that additional food, then it does better to let them live.  Thus the trade-off:  when the host offspring increase the fitness of the parasite, it lets them stay, but otherwise it kills its flatmates.  Gloag et al. take the time to break this trade-off down into its constituent parts, namely (in their words, p. 2):

  • The total provisioning rate stimulated by the whole brood, and
  • The share of the provisions received by a parasite nestling.
The simple model they derive shows that when the ability of a parasite to stimulate food provisioning by the host parents is greater than its ability to compete for food with its nestmates, the parasite will do best if it is reared alone and the murder spree begins.  This relationship depends on the interaction between these two variables;  in other words, “[i]f each host nestling causes a greater increase in provisioning than the amount it consumes, then the presence of host chicks would result in higher consumption for the parasite, even if a host chick takes a bigger fraction of the extra food than the parasite.”  The model helps to predict where each scenario – nestmate killing or tolerance – is plausible as a function of this intuitive trade-off.
VIRA-BOSTA (Molothrus bonariensis)

VIRA-BOSTA (Molothrus bonariensis) by Dario Sanches, on Flickr

Gloag et al. then use this model to explain differences not only intra-specific differences between specialist species in their level of nestmate tolerance, but also inter-specific differences within generalist species as well.  This would have been a good paper even if they had stopped there, but they then go on to test their ideas in the field using a generalist parasite, the shiny cowbird (Molothrus bonariensis). Working in South America, they searched for the nests of two types of shiny cowbird hosts, chalk-browed mockingbirds and house wrens, and set up two experimental conditions.   In the “mixed group”, the a single cowbird egg was placed among host eggs, and in the “alone” group, the cowbird eggs were placed in the host nest with dummy eggs so that the cowbird young would be reared alone.  They measured the food amount and quality brought to the nest from video recordings, and measured the physical quality of the resulting offspring (weight and tarsus length).  They also compared the mortality rates of the cowbird chicks to see if there was a difference between the conditions.  Their findings?

In our field study, nestmate tolerant shiny cowbirds encountered both sides of a provisions trade-off depending on the host used. When reared by chalk-browed mockingbirds, nestling cowbirds had higher food consumption, mass gain, and survival when alone in the nest than when sharing with 2 mockingbird young. In contrast, cowbirds reared in the nests of house wrens had higher food intake and growth when reared alongside 3 or 4 host young than when reared alone. (p. 7).

The results of their work suggest strongly that there is a trade-off at work here, and that the virulence of parasite offspring will be affected by the provisioning characteristics of the host environment.  Of course, they are quick to suggest that there are other factors potentially at work in differential growth rates, such as thermoregulation (larger broods can help each other thermoregulate) or size of the nestlings.  Nestling size is an interesting issue, because as the authors mention, cowbird young are larger than house wren nestmates but equal in size to or smaller than their mockingbird counterparts.  This may the competitive ability of the young either through physical competition between nestlings where size would be important, or because parents preferentially feed larger offspring.  (As a by-product, this also raises the longer-standing question of why host parents don’t do a better job at discriminating among their young for parasites in the first place;  for an explanation in terms of yet another trade-off, I’d refer you to this letter to Nature by Arnon Lotem as a possibility).

Wilson's Warbler feeding it's Cowbird chick  "offspring"

Why are you feeding this monster? (by Alan Vernon, on Flickr)

The work on trade-offs in this paper provide a simple and intuitive model for the action of brood parasites across a wide variety of situations, and then back it up with empirical data that demonstrate this trade-off in action.  It’s hard to ask for more from a paper!  Of course, as with every paper you’ll ever read, “more research is needed” (we have to say that, or we’re straight out of a job, aren’t we).  It wil be interesting to see if this trade-off does actually hold in other species, and combining the principles in this paper with a phylogenetic analysis would make for a fascinating approach. In the meantime, though, if you’ve read this far I’d urge you to take the lesson of this paper to hear and learn to look for the trade-offs inherent in many biological systems.  As a guiding principle of biology, I guarantee that you’ll see it almost everywhere you look.

Tagged , , ,

The economics of science blogging.

Well, Christie Wilcox has completed her series on scientists engaging with social media and unsurprisingly, I’ve got some thoughts.  I’m not going to spend time recapping her argument here, because the points I’m going to make are fairly targeted so if you haven’t read what she’s written, you should go do so now.  Done?   Before we go on, let me reiterate what I’ve said before:  I like what she’s written, and I think there’s some great stuff in it.  Don’t mistake my disagreement for belief that there’s nothing valuable in what she says;  I just want to broaden the conversation a bit.  Okay.  So, let’s take a look:

Sure, if you don’t build it they can’t come.  But just because you do build it, doesn’t mean they will.

Christie continues to gloss over the hard work required to build an audience in social media, and I’d like to take the opportunity to point out – again – that it’s not as simple as getting a blog, a Twitter account, or a Facebook page and starting to write.  Her first benefit to blogging is “If It Was Worth Doing, It’s Worth Telling Someone About”, which she illustrates with an encouraging tale of the success of social media:

Just ask Peter Janiszewski, of Obesity Panacea. Last year, he and his colleague published a fascinating paper in the prestigious journal Diabetes Care. The problem was, it went unnoticed. For three months, his study wasn’t blogged about. It wasn’t picked up by the press. No one seemed to care.

But Peter cared. He decided that the paper fit well into his blog’s theme, and wrote a 5-part series on the topic of metabolically-healthy obesity, the final post of which was a discussion of his recently published paper.

The series was a hit. Peter’s blog posts received over 12,000 pageviews and more than 70 comments from readers during the week of the series. As Peter recounts, “Put another way, the same research which I published in a prestigious medical journal and made basically no impact, was then viewed by over 12,000 sets of eyes because I decided to discuss it online.” A few days later, an article about his study was published on MSNBC.com.

Sure, Peter’s tale might be exceptional, but the point is there is a lot of potential to expand the reach of your research over social media. This kind of exposure isn’t just for the sake of communication. As Daniel McArthur noted, “A fairly hefty proportion of the readership of most science blogs consists of other scientists, so having your work disseminated in these forums both increases your profile within the scientific community, promotes thoughtful discussion of your work and can lead to opportunities for collaboration.”.

That’s great!  Well, it’s great if it happens.  But the argument of her post is essentially a cost-benefit trade-off, and I feel that Christie overemphasizes the benefit while ignoring the cost.  As to benefit, I’d like to ask a question here:  how likely is it that any given post on a scientific topic will receive n (here, n = 12,000) pageviews?  To answer that question quantitatively would require access to data that I don’t have – we would need to get traffic logs from multiple blog sources and use advanced statistical methodology to work out the distribution of page views as a function of source, content type, and so on.  But in the absence of such a data source, we can proceed by approximation and see if we can come up with some reasonable thoughts on the subject.

As evidence, I’ll use this post from Nate Silver (who writes the genius FiveThirtyeight blog, and from whom I’ve shamelessly stolen the title of this post) on the traffic at the Huffington Post.  Nate was making a point there that I would like to similarly make here:  most blog posts sink into obscurity.  He shows that page views at the HuffPo can be reasonably assumed to follow a power law, where a few posts get major traffic and most get very little, if any.  (Seriously, read Nate’s post – it’s a great analysis).  I would extend this argument to science blogging and suggest that it’s extremely likely that science blogging follows the same distribution of attention:  most posts get little to no attention.  The same relationship probably carries over to the blog level, where some blogs get a lot of attention and most receive none.  Using Nate’s comments metric, I’d guess that Pharyngula gets an order of magnitude more page views in an hour than I’ve ever gotten here.  That’s not a whine about my traffic, but a comment on building an audience.  I’m enjoying the process of writing and blogging, and I’m slowly starting to get a few regular readers, but it’s been nearly a year for this blog;  unless I’m willing to put a lot more work into it my numbers aren’t going to be anywhere near 12,000 page views for my upcoming seventeen part series on the spatio-temporal dynamics of producer-scrounger games.

You might think that this is sour grapes on my part, because Christie has a much broader reach than I do, but you’d be missing the point on two fronts.  First, I like doing this, and I’d do it no-one was reading (arguably, I have been doing just that for most of this blog’s existence. 🙂 ).  Second, I’m trying to say that a great audience doesn’t just show up.  It requires a lot of smart, interesting posting which requires time and effort punctuated by long periods of little to no extrinsic payoff.  It’s also an unavoidable and often unremarked truth that long, detailed posts on scientific topics receive less attention than short, snappy posts with squid videos.  Does the benefit of the possible wide attention that comes with social media outweigh the cost of building and maintaining that audience?  That’s an empirical question, and I’m going to suggest that for some people it’s possible that the answer will be “no”.

I’d also like to ask two shorter questions here.  First, is it reasonable to assume that anyone who wants to build an audience will be able to build an audience, even assuming that they all put in the required work?  And second, what happens when the market of science bloggers gets flooded with new entrants?  Does it make sense to assume that building an audience will remain as easy or hard as it is now when the room gets more crowded and you have to shout harder to be heard?

And the cost itself?  Well, the cost is variable.  Again, I blog largely because I enjoy it and because of of Christie’s fifth benefit:  practicing my mad skillz.  The opportunity cost in terms of lost research time or reading time is worth it to me, but it’s still a trade-off that I am consciously making.   Academic advancement committees have yet to recognise blogging as a valuable activity, and they are unlikely to reward you for the number of Twitter followers you have;  Christie is asking academics to make what could possibly be a significant sacrifice in terms of career advancement for uncertain returns in the social media space.  I’m willing to roll the dice on that myself, but I’m not willing to claim that everyone else should too.  Will the networking benefits that come from Christie’s second point or the branding benefits that come from her fourth point make up for this?  I don’t know yet, because they haven’t happened to me.  Check back in a year and we’ll see.  In the mean-time, though, Christie has a lot of great sounding benefits with no evidence that they really occur to most people blogging science.

I’m going to make a counter claim, and leave it up to the reader to decide what they believe to be more likely (in the absence of data that I either don’t know about or which doesn’t exist yet).  Christie’s argument carries the implicit assumption that nearly any scientist who blogs can, if they work for it, achieve great results and reap large benefits from the use of social media.  I’m going to claim, instead, that science blogging shares more with most other forms of audience-building activities than we would like to admit:  the outcomes will be highly variable.  Just like musicians and artists, actors and authors, many more will try than will succeed.  Some will gain much glory, while others will toil for months or years and either satisfy themselves with a small audience[1] or give up in frustration.

“All you have to do is…”

Christie goes on to say that social media enhances non-social outreach, which is a point that we argued about on her previous post.  I’m grateful that she’s addressed it, but unconvinced by her comments.  Here’s part of her salvo:

All you have to do is be a little creative. Let’s say your lab currently does community outreach by going into local schools and talking about science, for example. Imagine how many more school kids could be reached if you made the materials you create or your lessons available online, complete with an outline of how the day was structured and reflections afterwards.

As Molly Wood has taught me, the phrase “all you have to do” and its variants is usually a signal that what follows isn’t going to be easy at all.  And again, there’s a cost-benefit trade-off here.  Packaging such a lesson for wide release and then actually releasing it in a form that will be of use to others requires a fair amount of work.  What poor grad student or honours student gets stuck with that thankless task on top of what else they have to do?  And where is the imaginary army of eager school children waiting around on the internet for additional work from some scientist they’ve never heard of?

Or, let’s say you organize volunteering events which benefit the environment, like beach cleanups or invasive species removal efforts. How many more helping hands do you think you’d get if you posted them as facebook events or developed a network of local tweeps who like to volunteer?

I just checked, and I have right now I have 80 followers on Twitter (huh, more than I thought!).  Most of those don’t live in Sydney.  If I combined my blog and Twitter accounts and organized a beach cleanup for Maroubra Beach, using all of my social media powers, it would still be a pretty sorry affair.  Again, though, this isn’t a whine;  it just shows that receiving the benefits requires a lot of work to build and maintain that network.  All I have to do is … put months or years of effort into getting people to pay attention to me on social media so that I can leverage it occasionally?  If building that network doesn’t come with its own intrinsic benefit, I don’t see this occasional payoff being the clincher.

But maybe my call will go viral, and a thousand people will descend on the beach!  Ah, the promise of social media:  an unexpected ballooning of your audience to hundreds, thousands, or even millions with little to no warning.  That will still be cold comfort to me on the other 999 times out of 1000 that I’m standing on the beach, holding a trash bag and looking sorrowfully at miles of sand that I’ll be cleaning up by myself.

The good and the bad.  There’s no ugly here.

Christie’s benefits – like the value of disseminating your research, networking, practicing skills, enhancing out-reach, and so on – are great things to highlight about social media.  My point here isn’t to dissuade scientists from using blogging or Twitter as a tool for enhancing the communication of science or for their own benefit;  it’s to point out that the relationship between the benefits that come from social media and the costs required to acquire those benefits is complicated and somewhat unclear.  Christie herself is out on the forefront of blogging science, and from her platform at Scientific American she has the potential to get huge benefits from the work that she’s put in.  But I have to ring the bell and point out the elephant in the room:  Christie is pitching this piece to scientists who don’t use social media, and is asking them to take a risk on it.  It’s not something that should be entered into without thinking about it first.

———–

[1].  It’s worth pointing out that some people would be happy to have a small audience of their peers consistently read their work.  Maybe it’s not 12000 page views that are important, but getting 50 views from people in your field. As scientists we’re used to this model, because it’s how much of the science community operates.  My response to this idea is three-fold.  First, you still have to build the audience by reaching through the noise to those 50 people , with even more variance in possible outcome.  With social media, it’s probably easier to get 10000 people you dont’ know than 50 people you respect to read your material.  Second, if you’re publishing in the field, most of the people that you would want to be on that list are probably already aware of you and your work.  Where’s the added value of social media then?  Third, Christie’s piece focuses on big numbers (12000 page views, networking x 1000), and so I will too.

Ironically (since I’m arguing that not everyone should have to use blogs, Twitter, or Facebook) I’ve been having a grand time hashing things out with Christie Wilcox over on her blog about the push to get scientists online.  You can follow her rebuttal to my post, Kevin Zelnio’s post, and the latest developments on her blog here

The conversation continues…

I didn’t even know he was being investigated…

Huh. (h/t Andrew Gelman;  see also discussion here).

The last time I wrote about Kanazawa was here.  I agree with the commentators at scatterplot that the punishment seems odd, though I confess that I don’t know what would be appropriate here.  There’s an under-explored intersection between blogging and academic work that deserves more attention, and Kanazawa’s post and punishment are a perfect test case.