Friday, April 27, 2012

The beautiful evidence of science

These days, while I finish writing The Functional Art and recording the videos for the companion DVD, I am also re-reading one of my favorite science books, Why Evolution is True, by Jerry A. Coyne (see his blog). Over the years, I've realized that immersing yourself into something that is unrelated to what you are working on can spark unexpected insights. Besides, works such as Why Evolution is True are nonfiction literature at its finest: they are enlightening not just because of their content, but also because of their style. Or because of the seamless interaction between the two.

I am currently in the section on gradual evolutionary change, that can eventually lead to speciation. Coyne highlights some cases, such as two species of Eucyrtidium, a single-celled plankton. The two species branched out from a common ancestor; one of them became larger than its forebears, the other one slightly smaller. The book illustrates this process with a diagram similar to the one below, which I got from this website. I believe it's a great example of how to make science visually evident. I explain how to read the chart right underneath it.

The scale on the left represents age, or depth at which the fossils depicted were found. Therefore, you need to read this graphic from the bottom up.

The first quadrant uses a scale that goes from 0 to 40, which is repeated on the rest. Those are percentages. The histogram —the black bars— represents the percentage of fossils in that oldest sample that lie on each size group (horizontal scale, below). Notice that the highest bar is more or less in the middle, in the 90-98 range. This is a normal distribution: nearly 40% of the fossils have an average size. There are a few outliers on the extremes (the 72-80 and 99-107 ranges), but not that many.

Something similar happens in the second quadrant: the bars of the histogram cluster around the middle of the distribution. All fossils in the sample have more or less the same size. There's little variation. We still have a single species of Eucyrtidium.

But see what happens on the third quadrant from the bottom: suddenly, the biggest bar is on the far left, and it almost reaches 60% of the sample. Besides, a good amount of fossils begin to be slightly larger (those that are in the center-right of the distribution). This trend continues on quadrant four: the distribution is spreading out. Two new species are slowly starting to separate from each other.

At this point, we enter the top three quadrants. On these distributions, the fossils cluster not around a single tallest bar, but around two. We have now two species branching out from a common ancestor: one of them tends to be slightly smaller than it, on average, and the other one becomes much larger, with a small percentage of individuals reaching the far right side of the size scale.

This is what I would call, following Edward Tufte'beautiful evidence'. As he says in this presentation (which is not just about Medieval drawings), "evidence is evidence, whether words, numbers, images, diagrams, still or moving". Watch it.