This month, we touch on the always-exciting topic of multilevel selection in our Q&A feature with Dr. Charles Goodnight of the University of Vermont. In addition to his work on multi-level selection, Dr. Goodnight has also studied the effects of founder effects and epistasis on evolution. Below, Dr. Goodnight explains why you should be nice to the “old” scientists you meet at conferences, argues for a phenotype-centric view of evolution, and explains how aliens might perceive human civilization. You can read more of Dr. Goodnight’s thoughts at his blog Evolution in Structured Populations.
1) Tell us a bit about your background – what got you into studying evolution?
Looking back I was really lucky in how many important biologists I met throughout my childhood. My parents were both biologists. My father was on a 9 month appointment we would take vacations that were a month long or more. My parents were Opilionid taxonomists, and I traveled with my parents collecting “daddy longlegs” throughout the Mexico and Central America. By the time I graduated from high school I had been to every state in the continental United States, most of Canada, every state in Mexico, and every country in Central America. I had also tagged along on several graduate courses that my parents taught, and generally spent a lot of time hanging around labs and field stations. By the time I was in high school I was convinced I wanted to be an ecologist. Later in college, at the University of Michigan I was exposed to courses by Warren Wagner, Arnold Kluge and Donald Tinkle, and realized that evolution was much more interesting to me.
I should also mention that over the years I have been extremely lucky in the influential people I have met. Apparently, as a baby I sat on the lap of Victor Shelford, my father’s graduate advisor. I strongly suspect I met Ernst Mayr as a child, but of course I was completely un-impressed at the time, and many others that of course I don’t remember. Later as a first year graduate student my advisor Mike Wade took me to a symposium he was speaking at. While there I sat next to this old guy, and this really old guy. It turns out the old guy was James Crow, and the really old guy was Sewall Wright. That meeting I also got a chance to speak with Richard Lewontin. In any case, if I wasn’t determined to be an evolutionary biologist before that meeting, I certainly was afterwards.
2) You were a grad student at the University of Chicago right around the time molecular methods were starting to be used in evolution and ecology. Can you describe this time and the contributions you think they’ve made since?
When I took evolution from Kluge and Tinkle at Michigan we learned about phylogenies. At that time (must have been around 1976) the phylogenetic methods they championed were basically numerical taxonomy. My teaching assistant for that course was Art Dunham, and I distinctly remember him telling me in one lab that there was no way to polarize traits, so numerical taxonomy was the only possible approach – I am sure he would not agree with that today. Also in that class there was never a discussion of using molecular biology in phylogenetic reconstruction. It is really hard to imagine a better teaching team than Kluge, Tinkle and Dunham, so I have to believe that they were teaching the state of the art in 1976. To say that molecular methods have changed phylogenetics is an incredible understatement. During my tenure as a graduate student at the University of Chicago cladistics was developed, and the widespread use of molecular methods came with the rise of PCR and actually working with DNA. It is important to remember that around the time I started at Chicago the only “molecular” methods readily available were protein enzyme electrophoresis. There are very few measurable character states for enzymes, no real way to know if similar migration rates of an enzyme in different species represented an homology. There were a few attempts to use isozymes in phenetic phylogenetic analyses, but frankly they were not convincing.
Population genetics is a bit of a different story. Molecular methods have had a remarkable effect on our ability to monitor gene flow. Perhaps one of the best examples is what molecular markers has done to our ability to monitor paternity in birds. Prior to relatively inexpensive molecular analysis it was virtually impossible to assess paternity by any method other than behavioral observation, and many birds were considered to be essentially monogamous. Molecular markers have revolutionized our understanding here and we now know that extra-pair paternity is common. Similar statements could be made about our ability to measure gene flow between populations. For example, methods such as private allele analyses simply would not have been possible at the time I entered graduate school.
Interestingly, one thing that molecular biology has not changed is theoretical population genetics. Other than a few tweeks on the edges the models really have not changed. By that, I mean selection models, drift models, and the like are essentially unchanged. What has changed is that we now need to apply the theoretical models to molecular genetic data. Aspects of theoretical population genetics that used to be important have become much less so. When was the last time somebody reported the effective number of alleles for a trait? Others have become much more important. We have always known about silent site mutations, but prior to the advent of molecular data we could not have measured dN/dS ratios, or done a McDonald-Kreitman test. Similarly, prior to the advent of molecular data linkage was one of those novelties you discussed in genetics class, whereas now this is a critical element of any molecular genetic analysis.
My guess is the real lessons of molecular genetics are only now coming to light. In particular, the various genome projects have revealed that less than 2% of our genome is actually exon, and “obeying” the triplicate code. We are discovering that epistasis is wildly common, and that gene expression is heavily modified by everything from methylation to iRNA. This really calls into question what we mean by “gene”, and how we extract information from a genomic sequence.
This is a problem at multiple levels. From a statistical perspective, molecular data now puts us in the unfortunate position of having more variables than observations. That is, each organism or RI line we sequence is one observation, and each SNP is one variable. The old rule of thumb with statistics is that you would like to have 20 times as many observations as you have variables. That gets kind of hard when you have a thousand SNPs. I think there is actually a fundamental flaw in QTL analysis. A good statistical test should become more robust as you measure more things. Instead, with QTL analysis, the more markers in your map, the greater the multiple comparisons problem. I don’t have an answer to this, it just suggests to me that there should be a better approach. From a conceptual perspective this also raises issue of how we should think about evolution. A fairly standard definition of evolution is “change in gene frequency”. Such a definition simply doesn’t work when we don’t know what a gene is. Although still in its infancy I am hoping that taking a more genomic approach will start to provide answers to some of these issues.
One last point: I started as an undergraduate at a time when we knew essentially nothing about molecular genetics, indeed Kimura’s neutral model came out around the time I entered high school. In college I learned a lot of facts and a lot of concepts about biology. Most of the concepts, in some form or another have served me well. Most of the facts have changed entirely. Over the years I have extracted isozymes for electrophoresis; the details are useless, but the underlying ideas about neutral variation are still useful. I have done PCR on only vaguely programmable machines. Again the details are now useless, but the concepts of how PCR works are quite valuable. My point it is clear that teaching and learning facts is mostly a waste of time. Teaching and learning concepts are very valuable.
3) You’ve recently started a blog Evolution in Structured Populations. What do you write about? What’s the most interesting aspect of writing this blog?
I am aiming my blog at biologists and particularly evolutionary biologists. Over the years I have developed some strong opinions on how we should think about evolution, and I have come to think that some of the core tenets of our evolutionary thinking are basically unworkable. In standard evolutionary thinking we put the gene as being of foremost importance. Dawkin’s caricatures it best (although I think he wanted to characterize it!) when he argues that DNA are “immortal coils” that use phenotypes as “vehicles” to take them forward to the next generation. In this blog I am asking how would our understanding change if we reversed this idea, and thought of phenotypes as using genes to create new phenotypes. Obviously evolution does not change, but our interpretation of evolution very much changes. As I have told friends: It changes nothing and everything. The nothing part is that it absolutely is not a call to abandon molecular genetics, or frankly any areas of research we are doing now. The everything part is that the “gene” loses its centrality, and it becomes much easier to incorporate everything from epigenetics to culture into our evolutionary theory. I haven’t gotten there yet, but I also think this simple change in perspective has the potential to change our views on such fundamental questions in biology as the evolution of sex – for phenotypes shaping the distribution of offspring phenotypes is important — and the origins of life – inheritance may have come second and have been driven by selection.
4) Why do you think the debate between group selection and multilevel selection has been so vociferous?
It is actually a bit confusing why people are so argumentative. Some of it dates back to the early days of group selection. There was a period in the 1960s when there was a great deal of naïve group selection thinking. It was probably for the best that there was a revolt against the sloppy thinking that gave rise to the “good of the species” thinking. The people who were most vocally anti-group selection were also very good writers, and that poisoned the world for group selection for a very long time. Why the controversy continues today is not so clear. It is interesting that it is mostly very one sided. Those who champion group selection tend to understand kin selection, and dismiss it because it is not useful to them. Those who tend to champion kin selection tend to not understand group selection and dismiss it because it is a priori wrong.
You might ask why those using a group selection approach would find kin selection not useful. It turns out that kin selection and multilevel selection (we have part ways with group selection here) are similar, but not identical approaches. In the kin selection approach you solve for the optimal genotype or mix of behaviors. In the multilevel selection approach you solve for the strength and direction of selection. Many multilevel selection advocates are active experimentalists. For such people the kin selection approach is very unsatisfying. The kin selection approach may provide insights into where a population should evolve towards, but really does not provide the kinds of measures that a field experimentalist would find terribly useful. Further, as often as not these people are not interested in “altruism” per se. A case in point are agronomists that are interested in improving crops, be it egg yield in chickens, or meat production of pigs. Deciding whether or not pigs and chickens are altruistic does not put eggs and bacon on the breakfast table.
On the other side of the coin I think for students of kin selection it is hard to see the relevance of much of multilevel selection work. In half of the treatments in my group selection studies group and individual selection are going in the same direction. If your focus is on altruism, what is the point of those treatments? Another interesting difference is that because of Hamilton’s bias kin selection has been developed very much from a gene’s eye point of view. An individual is altruistic towards those with whom they might share genes. Multilevel selection, on the other hand developed very much from a quantitative genetics perspective, which is intrinsically phenotypically based. Thus, again, I suspect students of kin selection do not understand the utility of an approach such as contextual analysis that measure phenotypic selection with no reference to whether or not that selection will cause any genetic changes.
I should emphasize that I am firmly in the multilevel selection camp, so my answer here is undoubtedly biased.
5) If an alien species is observing human society on Earth, what do you think they would find most remarkable?
Interesting question. First off, I actually think that there is pretty strong evidence that interstellar space travel is impossible. My reasoning: If it was possible, and at all common, we would have been visited, and if our extraterrestrial visitors were anything like humans we would have been completely exploited, and probably eaten.
I would like to say that what would most impress an alien would be some aspect of our art or technology, but the truth is I don’t think that would be true. We also tend to forget old technologies that are incredibly important, but have been around for more than a lifetime. So, while I might be impressed with modern computers, the truth is sewers and modern sanitation are probably a more important inventions.
That said, I think the most remarkable thing we have done is that we as a single species have occupied every corner of this planet except Antarctica (we have stations, but no permanent residences). In the process we have coopted a huge fraction of the worlds available energy, and managed to suppress nearly all of our predators and parasites. This is amazing. We are a huge resource for anything that could eat us, and yet we have managed to suppress all of those predators and parasites. We also seem to have found methods to avoid killing each other. Yes, there have been horrible wars, but there are also 7.5 billion people on earth. Every day of my life, except one, there have been more people alive at the end of the day than at the beginning of the day. Our wars may be vicious, but at least in my life, they have never been enough to decrease the population size even momentarily. By the way, the day that the population size went down was December 26th 2004 when a tsunami hit Indonesia 230,000 people. So, I think our aliens would be most impressed with our ability to exploit this planet for our own ends, and to pack so many people into this small space with relatively little fighting.