Insignificant figures

Before writing a paper I make lots of figures. Some will be improved upon and included in the paper. Others just help me understand what the data are telling me. Graphical data analysis.

Still others, like this one, end up being rather pointless exercises. This categorizes those barnacle larvae that died during our study, based on the stage during which they died, and whether they spent abnormally long in that stage and/or the previous stage before dying. The reason this is pointless, despite the nice colors and nested boxitude, is that all the same information can more compactly and clearly be displayed in a table. However, I now know how to use, and when not to use, the R package treemap.

A third way of seeing things

The paper I'm writing now is on sex-biased longevity in primates. I'm trying to understand why in some species the males live longer, and in others the females live longer. The story I'm trying to test goes like this:

The largest investment in primates' offspring is in the form of care, not eggs or sperm or pregnancy. In species where females provide all of the care, males need not stay with a single mate, so the operational sex ratio is male biased (i.e. lots of males are out looking for a mate, while most of the females are pregnant or nursing and therefore not looking to mate), reproductive skew is high for males (some males will father lots of offspring, others none), and the fitness rewards to successful male competitors are great. Larger males are more likely to win these competitions, resulting in an increase in optimal male size. These large belligerent males risk increased mortality through conflict, and through diversion of physiological and developmental resources away from longevity and into competitiveness. Simultaneously, selection for longevity in females may be increased by the need to stay alive until their young become independent (and in the case of humans, to care for grandkids). This leads to males who, relative to females, are non-caring, short-lived, large and conflict prone.

I have data (gleaned from the literature) for a bunch of primate species on how long males and females live, how much care the fathers provide in each species, how big males and females are and how frequently and intensely males fight with each other. Depending on how I analyze these data, I get two very different answers. If I treat each species as an independent sample, and look at the correlations between these variables, the data completely support the story. The species range along a continuum having short-lived belligerent large, uncaring males at one end and long-lived caring low-conflict small males at the other.

But most evolutionary biologists would say that is not the right way to analyze the data. I need to take into account the relationships among the species. Two traits might seem to be correlated not because the one makes the other selectively advantageous, but because a group related species all have the one and all have the second. Take the example of feathers and laying eggs. All birds have feathers and all birds lay eggs. Is this because something about feathers requires egg-laying (probably not) or because all birds are descendents of some ancestral birds that laid eggs and had feathers, and ever since no bird has arisen that didn't do both those things. That all birds have eggs and feathers is an example of what biologists call evolutionary inertia.

The question this poses for me in writing my paper on primates is whether the correspondence between sex-biased longevity and these other variables is because of the adaptive story I told you, or because of evolutionary inertia. So I look at the data a second way, and see that there is a great deal of evolutionary inertia in these traits. Most primate species are somewhere near the middle of that continuum I described. Ever species at one end is from one group of related monkeys. Every species at the other end of the continuum is either a great ape or from a different family of monkeys. Plugging my data into software designed to test for evolutionary inertia, I find that closely related species are very likely to have similar values for all the traits I am measuring, and that inertia is fully sufficient to explain the correlations between these traits. It is like eggs and feathers in bird.

As I am trying to write the paper to present all of this in a scientifically rigorous way, I am struggling with what to say about it. The easiest, and least interesting, conclusion would be to simply say that the correlations are purely illusions conjured by evolutionary inertia. What I'm attempting to find a way to argue it that it could just be inertia, but that the inertia might be because of selective effects. In other words, that both the inertia story and the selective story could simultaneously be true, and closely related species are similar not because they can't change, but because what was adaptive for their common ancestor is still adaptive for them. If a male from a very belligerent species was to grow in a way that made him not a good competitor, but able to live a long time and provide care to his young, this wouldn't fit with the way his species lives, so he wouldn't pass on his novel trait, and his species would stay like their ancestors had been. It is evolutionary inertia caused by adaptive mechanisms. I think my next step is to find out who has already written about adaptive evolutionary inertia, and what they said.

Down to the wire

One of my professors, RL has this habit of being uncannily right most of the time. My student, LZ, and I were working on analyzing our data this afternoon. The first draft of her senior thesis is due tomorrow, and we decided we may as well try one last thing, calculating a variable that RL had suggested. We plugged it into our model and BAM! highly statistically significant, much more so than anything else. And, in retrospect, it makes beautiful biological sense. Of course it all depends on cumulative lay order across generations! Any brilliant biodemographer would have seen that at once.

Now LZ has at least 12 hours to rewrite her paper.