What Was I Thinking? Remembrances of My Dissertation

In 1987, I submitted my Ph.D. thesis, “Environmental Variability and Phenotypic Flexibility in Plants,” at the University of Illinois. What was I thinking? I was dealing with three variables: plants, the growth flexibility of those plants, and the variability of the environment—in this case, of sunlight intensity. I intended to write about all three of these things in this essay, but it quickly got out of hand, so I will just tell you about how I dealt with the complex problem of measuring environmental variability in three habitats: abandoned agricultural fields, tallgrass prairies, and forest floors.

One of the major problems with measuring environmental variability is construct validity—that is, how valid is a way of measuring something so that it is a believable construct of what you claim to be measuring? I explain this concept in my book Scientifically Thinking. There is no single valid way of measuring environmental variability. What I did instead was to measure it three different ways. If these three ways all gave the same results, then they were probably valid, and I could believe the results.

First, I made instantaneous measurements of light (that is, the colors of light that stimulate photosynthesis) with a “quantum sensor” designed to do exactly this thing. The problem is, this kind of light intensity varies from one moment to the next.

So I used a second method. What I really wanted to know was the spatial patchiness of shade caused by leaves. So I made three-dimensional profiles of leaf area in the three habitat types. To do this, I used nine pieces of wood, a string, and a nail. But neither of these methods could prove that variability of sunlight intensity would actually have an effect on plants.

So I used a third method as well. I used phytometers, that is, actual potted plants. I measured the weight of the plants (plus the soil, or course) in lots of different places in the three habitats in the morning and the afternoon. I measured environmental variability as the differences in how much water the plants transpired.

This was how I tackled the problem of construct validity. I measured what I wanted to know three different ways. I was taking a chance. What if the three ways gave three different answers? But I got lucky. All three methods agreed. Chalk one up for construct validity.

I had expected that weedy fields would be the most variable, the forest floor the least variable, and prairies intermediate between them. And this is what I found. I still marvel that it came out that way.

I announced my results triumphantly: weeds had the most phenotypic flexibility and lived in the most variable environment; forest floor plants had the least flexibility, and lived in the least variable environment; and prairie plants and their environment were intermediate. The results all lined up. Despite all of its limitations, this quixotic quest was successful. I believe the results, largely because I took a risk with construct validity, and it paid off. There is no way I could have gotten these results by sheer luck. Here is the plain English summary of my thesis.

A lot of the scientific method is about validity. You have to measure it, and every time you do, you are taking a risk. But if it works, you are closer to understanding the world.