Origins of Biological Complexity

Carl Zimmer has a great article in Quanta about the origins of biological complexity. Quanta, in case you’re wondering, is the new name for the Simons Foundation’s online science magazine, which is certainly going to be a go-to resource for reliable science stories much of the media would consider to be too subtle or insufficiently newsworthy.

Defining “complexity” is a notoriously tricky business, although we tend to think we know it when we see it. The quest for the One True Definition is a red herring, as what’s interesting is to see what patterns and laws we can associate with different kinds of complexity. In the biological realm, it seems natural to give at least some credit for the development of complexity to the pressures of natural selection. Having more highly-developed sensory apparatus or higher intelligence naturally goes along with greater complexity (or so we tend to think), and there can be obvious evolutionary advantages to these traits.

Carl looks at a new paper by Leonore Fleming and Daniel McShea that looks at the number of different part types, shapes, and colors in everyone’s favorite biological test subject, the Drosophila fruit fly. They argue that complexity increases even in the absence of any evolutionary pressure at all. That’s consistent with a proposal called the “Zero-Force Evolutionary Law,” which says that complexity and diversity simply tend to increase naturally, apart from any nudges evolution might provide. Fleming and McShea looked at the evolution of Drosophila raised in comfortable laboratory environments, where they were provided with unlimited food and perfectly livable conditions, and compared them to wild fruit flies. They conclude that, indeed, the absence of pressures led to increased measures of complexity in the population. Roughly speaking, there was less reason for crazy mutations to die off, so the genome could go galloping freely through the fitness landscape.

Other biologists are skeptical of this way of looking at things. I think the basic point is that it’s easy to see how diversity will increase in the absence of evolutionary pressures, but much harder to useful complexity (like eyes and brains) would develop. To which I imagine the appropriate response is “it depends on the conditions.” If we imagine that offspring survive and reproduce equally well no matter what kinds of mutations they undergo, and there are truly unlimited resources, then I would predict that some descendants would have just as many usefully complex outcomes as they would in the presence of selection pressures. But that’s only because there would be a ginormous number of descendants, and most of them would be utterly unviable in the real world. The fraction of descendants with useful complex features in the selection-free world would doubtless be much lower than in a world with natural selection.

Evolution is able to make some wonderful things, but it really is a blind watchmaker. Mutations and sexual shuffling of genes happen, and then natural selection culls away the less successful outcomes. It doesn’t actually accelerate the production of useful outcomes. So complexity happens naturally (at least, starting from simple states in open systems very far from equilibrium), but evolution brings it into focus.

This entry was posted in Science. Bookmark the permalink.

21 Responses to Origins of Biological Complexity

  1. Robert says:

    I think of a world without natural selection as a Markov Chain with each vector entry as the proportion of organisms with a trait. Natural selection is like multiplying some of the vector entries. With infinite populations, therefore, anything could evolve without natural selection (but only with small proportions). But in the real world with finite populations you need natural selection to get past lower limits (low population and low probability).

  2. mks says:


    neet :3

    Howard Bloom tries to deal with that (among other things) in his book, “The God problem: how a godless cosmos creates”

  3. Rick says:

    Is a truly selection-free environment really possible? The article/study suggests that simply because there is an abundance of resources, there is a lack of selection. Certainly competition is reduced or eliminated, but selection and competition aren’t inextricable. The new environment simply selects for different traits. And while there might initially be an abundance of mutants surviving when they previously couldn’t, it would seem to me that eventually the population will re-stabilize after certain factors kick in. It’s, in a sense, analogous to some current human circumstances. Many Americans have a similar abundance of resources, but not all are equally fit (read: obesity). Really what I’m arguing is the definition of a selective environment; just because it’s a non-competitive one with plenty of resources, doesn’t make it a non-selective one, which questions the reality of an environment free of evolutionary pressures and the validity of the study itself.

  4. Joe Dickinson says:

    I think that the idea that a “cushy” lab environment reduces the force of natural selection relative to what is experienced by wild populations is total nonsense. Whatever the environment, individuals compete against one another for survival and reproduction and mutations that are favored in that environment will prevail. So the basic premise of the supposed “test” of the “zero force evolutionary law” is simply invalid.

  5. FrankL says:

    I agree with Rick and Joe, selection is probably not absent. I have heard (not sure if its true – no references) that peacocks went through an evolutionary bottleneck battling some sort of eye disease. Females “learned” ( in the genetic sense) to choose males with healthy looking eyes. Males learned to fool them with large eyes on their tailfeather displays. Such a display must have a cost, but maybe not too bad if times are good. I wonder if drosphilia are complex enough to have an analogous mechanism going on. If so, selection would still be present in the test case, and would be free of any associated cost, so the test case would amplify such selection criteria.

  6. john thompson says:

    This kinda makes punctuated equilibrium make more sense. So imagine a population where there are abundant resources and no predators for a long period of time — you have a bunch randomly distributed mutations floating around the population that increase over time — the range of survivable mutations increases. Then you have a stressor — say a drought or a new predator. Suddenly a huge percentage of the population dies off, and you’re left with a smaller set which happen to have the mutations that encourage survival, and the population becomes more homogeneous, but with a different range of traits than it had before.

  7. Peter says:

    Hmm… I did not understood the same thing…

    From what I understand, they are simply suggesting that although the potential surface into which we evolve may seems bumpy with high cliffs at our scale, at the scale of our component cells, some micro spots on this surface may be rather flat allowing for free lunch evolution. Hence, random mutations are occurring in these particular flat spots without affecting the survival of the cell and species hosting the cell (of course mutations that impact one or the other survival just disappears by natural selection).

    It may then occur that these random mutations hit the jackpot by accident creating a mutation that favors the cell and/or the host species. But it would seem more probable to me that the potential surface suddenly change (a change in the environment) and the micro flat spot on the potential surface is no more flat and hence some of the free lunch mutations are favored over others. Mother Nature claims her due.

    That’s really what I understand it has been suggested. The fruit flies mutations review was just there to show that indeed when you have a flat potential surface a lot of mutations occur. (Although in that case, contrary to what one generally finds in nature, it was a large scale flat potential surface rather than a tiny micro flat spot on the potential surface.)

  8. vmarko says:

    I think that the “complexity from randomness, given unlimited resources” idea is the same as the old monkey-typewriter-Shakespeare thing. The keyword here is “unlimited”, which is nowhere to be found in the real world.

    The question can be phrased more precisely, i.e. in a more quantifiable way. Say we are given some complex system (say, a brain of some defined large size), and some population of organisms that do not have it (say, fruit flies). Assume that the DNA of fruit flies must have finitely many genes at any given point in time. Let the number of members in this population be N at initial time. How much time, and how much life-supporting-resources does this population need in order to generate at least one member with a big brain by random mutations?

    When answering, if either the amount of time or the amount of resources turns out to be “infinite”, the “complexity from randomness” idea is in trouble.

    As soon as you allow yourself to have unlimited resources (or unlimited anything, for that matter), your question is ill-posed, can have multiple answers, and the answers cannot be trusted. Everyone who ever uses the word “infinite” or “unlimited” should go and learn some math first, in order to be able to rephrase the question without those words, and in order to understand the concept of “taking a limit”. Especially if one is dealing with probability and randomness.


  9. David says:

    Increases in complexity are a thermodynamic issue. Complexity arises when a system has a net input of energy and, for biological complexity, sufficient chemical diversity. IOW, the Earth’s biosphere promotes complexity because the sun and geothermal sources are pumping energy into the open system. Add to this energy sink a terrestrial planet’s diversity of elements, and biological complexity inevitably ensues.

    Hence, if you put a bunch of fruit flies in a controlled environment and pump a bunch of energy into it, e.g. food, light and heat, and there’s little surprise that complexity increases faster than for fruit flies scratching out a meager existence in nature, where that net input of energy is harder to come by.

  10. Low Math, Meekly Interacting says:

    Serious question: Does more random = more complex? Also, it is well known that evolution can occur in the absence of any obvious selective pressure (genetic drift), so it’s not clear to me what real advance this study has made. Furthermore, we seem to be considering, if anything, intraspecies complexity, not organismal complexity. One mutant may be different than the other, but it isn’t necessarily more complex. Even with the proliferation rate of fruit flies, it would take a very long time to evolve an appreciably more complex organism, at least in the way I characterize complexity.

  11. Brody Facoum says:

    At the population genetics and quantitative genetics end of systems biology (q-bio.PE and q-bio.QM) there are plenty of people working with models that resemble “zero force laws” (ZFL, to contrast with the ZFEL acronym for the “Zero-Force Evolutionary Law” conjectured in McShea & Brandon, neither of whom is a working systems biologist). In particular, the Hardy-Weinberg equilibrium is effectively a ZFL model.

    There is no real “fight” about ZFL models in the most general sense as they underpin simplifying models that allow one to import mathematics from elsewhere in science, notably perturbation theory and renormalization group analysis. Sober was probably the first English writer to make more than casual use of the comparison between a study of forces and HW, starting in the 1980s.

    In physics, a ZFL is roughly an inertial system in which there are zero external forces acting on the system’s components. If we observe an object accelerating in such a system, we can infer that there is a force acting on it, and that we can make further inferences about the source and magnitude of the force.

    In systems biology, a ZFL is roughly an environment in which there is no bias on allele frequencies for a particular locus where the alleles at that locus have the same average chance of being passed to offspring of the breeding population. If there is a change in allele frequencies from one generation to the next, one can infer that there is a “force” acting on the population, and can make further inferences about the source and magnitude of the “force”. That is likely to help identify possible mechanisms at various scales.

    However, as in physics, force cancellations and the like mean that one cannot conclude that unchanging allele frequencies means that no force is acting on the population. Moreover, as the ZFL is a simplification, one has to be careful about wrongly concluding that there are no net forces acting on a population.

    McShea & Brandon’s ZFEL, on the other hand, says (per Matthen), “In any evolutionary system in which there is variation and heredity, there is, in the absence of constraint, a tendency for diversity and complexity to increase”. Their definition of complexity in context is the “number of part types or degree of differentiation among parts”.

    That is pretty different from a ZFL model in which in the absence of evolutionary forces complexity (in their sense) and diversity may change from one generation to the next only to the extent that there are departures from inherent equal fecundity and equal chance of inclusion of each allele at each locus. It is not at all clear that such “unforced” change should always be in the direction of greater diversity or greater complexity (in their sense).

    There may be a correspondence between some set of ZFL models and ZFEL, but since ZFL models are investigative tools rather than scientific theories in their own right, determining that does not seem like a fruitful endeavour other than as a means of challenging ZFEL by demonstrating that it shares known flaws of generic ZFL models.

    Finally, most people working with ZFL models likely would not use a “zero force” term, even though Sober’s arguments provoked a lot of debate and work on the Hardy-Weinberg principle, and it is normal to talk about “forces” leading to departures from the HW equilibrium, and to describe them as vector quantities.

  12. James Cross says:

    There is another great Zimmer article that complements this one in the NY Times.

    It really shows that organisms evolve not in isolation from other organisms but often in a synergistic relationship with the others.

    My impression is that the evolution of the diverse forms of life is not just a matter of random mutation. Life evolves by any and all means at its disposal. It can evolve by assimilating other organisms as in endosymbiosis. It can evolve by cooperating with both like and unlike organisms. It can re-purpose dormant genetic instructions and make remarkable jumps in a short period of time. Random mutation may still play a big role and, of course, natural selection plays the major in whether the new organism, whether created by mutation or otherwise, continues.

    Of course, the theory of evolution by natural selection doesn’t try to explain the origin of life, but there remains a big question about the point at which natural selection begins to play a major role in the development of new forms of life. In the Life Before Earth hypothesis, Alexei A. Sharov and, Richard Gordon assume that natural selection was at work perhaps for billions of years before the most primitive one celled organisms appear on Earth. From what we know about life now, it seems that the simplest organism is a microbe with 120 genes. For the Sharov and Gordon hypothesis to be correct we would to believe either that viable proto-organisms with much fewer genes than this can exist or evolution by natural selection could take place with molecules that are not true organisms. Bottom line is there is a big jump from inanimate molecules that can copy themselves to full fledged organisms. We will likely discover some other mechanism than random mutation was at work during the development of the first life.

  13. CEB says:

    What about conducting this study on humans, say comparing a population from a highly developed urban area to a group of “wild” humans from New Guinea? Might be interesting.

  14. James Cross says:


    No wild human in New Guinea.

    Biologically and culturally, for that matter, they are just as evolved as we are.

  15. CEB says:

    Sure, they are just as evolved, but the subject of the post was complexity. Aren’t humans in a developed urban environment more removed from day to day struggles for existence than are people in a more natural (wild) environment like New Guinea or maybe rural India? I would guess that the evolution of non-useful traits in urban areas is more common that it is in “wild” areas. In other words, if this study applies to fruit flies, maybe it also applies to humans.

  16. James Cross says:

    You could make an argument that the urban human has a harder struggle for existence than a tribal one in New Guinea. Some one did a study of hunters and gatherers and found they had more leisure time than we do.

    At any rate, with human reproductive rates, you would need to run the experiment for probably about a thousand years or so.

  17. Low Math, Meekly Interacting says:

    There are days when I check out the link, lightly skim the post, then comment. Yesterday was one of those days. Sorry.

  18. James Gallagher says:

    We need more planets for data, this one only had two major evolutionary outcomes – the dinosaurs and us.

    Mind-boggling to imagine what else could have evolved.

  19. Jud says:

    Stephen Jay Gould, a person somewhat familiar with evolution, wrote about something similar to this at length in his book Full House. I’ll try not to mangle his idea too badly in my vague recollection of the amateur understanding I had at the time I read the book many years ago.

    Picture the genealogical “bush” of a random evolutionary process. It would branch out in all directions from its origin, like a bush seen from above. But at its origin, there was little or no space for life to evolve into something simpler, since simpler would be non-life. So our “bush” is up against a wall on one side. The only directions in which it can branch out are equally or more complex. We certainly see all sorts of evolution at the equally complex level – most living things are relatively simple microscopic organisms. And then there are branches toward more complexity.

    So you have a random process with a limitation on evolution toward simplicity producing a pattern that *looks like*, but really is not, evolution toward greater complexity.

  20. Torbjörn Larsson, OM says:

    Good description. And mind that adaptation most often simplifies (as more than half of organisms are parasites, IIRC).

    The generic idea sounds vaguely like the “hopeless monster” idea. But near neutral drift will also destroy what has been complexified just because it is random, unless it is fixated.

    Renaming what is known sounds like what is going on. That can be bad, neutral, or beneficial depending on what ideas the new terms engender…

  21. Torbjörn Larsson, OM says:

    @James Cross:

    “We will likely discover some other mechanism than random mutation was at work during the development of the first life.”

    I don’t think so, since Lane & Martin published the paper where they claimed a homology, in the trait sense of phylogeny, between early chemoautotroph metabolism and pH-modulated alkaline hydrothermal vent chemistry. E.g. we now know that life arose here (or at least on a planet with such vents), and that evolution (most basically as variation and selection on populations, whether chemical or biological) is responsible.

    It also makes sense to include the other known trait homology, between CHNOPS cells and the frequency of elements in protoplanetary systems.

    As for minimal cells, Shostak protocells need one gene (a replicator) given a vent environment. Simulations shows that a replicator strand can crystallize out of a random strand gas in ~ 30 kyr in such environments, driven by the free energy. (Though the L&M result shows that coevolution was what happened, so the pathway taken was likely faster than that.)