A discussion of misconceptions in evolution ... about missing links, or great chains of being, or teleology (the idea that evolution is goal-directed) has got to be the most fun you can have with your pants on. Pursuant to this, let's sharpen and clarify our evolutionary theory mojo by considering the concept of "mosaic evolution" ... what is it, and what isn't it?
Of course, the concept of mosaic evolution, meant to clarify how evolution works, is often itself misunderstood. From Wikipedia:
"Mosaic Evolution is the concept that major evolutionary changes tend to take place in stages, not all at once." .. well, sort of, but that is not the main point and it is not well stated. Continuing, "...It is a pattern in evolution ..." "a" pattern? I wouldn't have said it that way. It is more like a general characteristic of macroevolution ... "... in which the rates of evolution in one functional system vary from those in other systems." .. like, different rates of mutation? of selection? No, no, that's not it, it's about timing, not rates "...For example, in hominid evolution, the dental system, locomotor system, and neurological system, evolved at markedly different rates." Almost. These differed systems have different histories. They do not co-evolve (see below). The word "rate" is ill advised in this context. I'll explain shortly.
The concept of Mosaic Evolution originated with Lemarck, but was not developed in modern evolutionary theory until the middle of the 20th century. Even then, it was not widely heard of until the 1970s and early 80s, Myer and Stanley forced the concept into more general consideration (Stanley in relation to human evolution) so now it is a term that is commonly heard, though often misunderstood. I suspect that part of this confusion comes from the fact that Stephen Jay Gould talked about "bushes" in (human) evolution and "mosaic" in human evolution in the same essay (cited below), because much of the confused rhetoric I run across seems almost plagurized from that essay, and conflates these concepts. Shall we proflate them?
To make the point clear, let's consider mosaic evolution in our own lineage, and in so doing, lets consider only organisms that lived on the historical line from a theoretical common ancestor of chimps and humans (we'll call that the LCA for Last Common Ancestor) to the present day human population. Forget the other branches that certainly existed as part of the human evolutionary "bush." If Australopithecus robustus (to randomly chose a hominid) is not directly ancestral to living humans, then just forget about it. If the Asian Homo erectus left no living progeny but the African Homo "ergaster" (erectus) did, then ignore the Asian ones and include only the African ones. I want to consider only the direct lineal sequence from the chimp-human LCA to me and you. Everyone else go away, this is not "big tent" day in hominid evolution land.
Assume that the human-chimp LCA was very chimp like. This may be wrong but there is good reason to think it, so we'll use that assumption now. In order for humans to exist today, that chimp like form needed to give rise to a new form that looks like us, perhaps all at once or perhaps bit by bit over a long time. Early conceptions of human evolution made the following assumptions:
1) The transition from African non-human ape morphology to human morphology was slow and steady.
2) The transition was unidirectional. Not lineal. Unidirectional. If non-human apes had brow ridges and humans not, then brow ridges slowly went away. They did not become bigger first then go away. If non-human apes had fur and humans did not, there is not an intermediate very furry version, but just a change from non-human ape-level furriness to human style nakedness over time.
3) The set of traits we think of as having changed between the African non-human ape model (as in chimps) and living humans all changed together, possibly causing change in one another. Like what Darwin thought: Uprightedness, freeing of the hands, dexterity, larger brain, all co-evolving.
In other words, the LCA to living human evolutionary "movie" would look like a slow and steady morphing of a chimp-like ancestor to a living human done with modern computer morphing software.
None of these things, it turns out, are true. Enamel thickness and tooth size both went up and then down. Bipedalism ala Australipiths evolved way before the big brain. Brain size seems to have gone up and down, not just up. Other traits seem to have done a two step (or even a three step), to use Glynn Isaac's terminology. The historical timing of the change in tooth size, bipedalism, and brain size were all different and independent. There may be a link between some tooth traits and bipedalism, if canines got small about the same time as bipedalism (conjecture, but it may look that way) but tooth size and enamel thickness is later than those two events, apparently.
Notice how I started out this discussion with what I hope was a slightly over the top but very clear statement that I don't need no stinking side-branches (as it were) to consider mosaic evolution. The fact that paranthropines with whopping big teeth evolved but were not on the direct human lineage (I'm assuming they were not for now, though that could be wrong) is not related to the present discussion. Australopiths were human ancestors, not all of the species that existed, but some of them, perhaps even one or two of the species we know about. The fact that among the many species of Australopiths there were super-giant-toothed forms is a fact, but it is not related to the mosaic thing. All known Australopiths dating to after 3.5 million years or so had bigger teeth and thicker enamel than a presumed chimp-like ancestor, and it does not matter that the biggest-toothed, thickest-enameled ones did not happen to be our direct ancestor. It is still true1 that tooth size went up, then down, and evolved independently from bipedalism and brain size, in our lineage. These traits, collectively, behaving independently as though the scheduling of one change was unrelated to the scheduling of another trait changing, are what you call a mosaic.
The fact that tooth size went up then down is NOT related to mosaic evolution. The fact that tooth size and bipedlaism (and brain size and body size) evolved at different times IS mosaic. Also, these things did not evolve at "different rates." Well, maybe they did, but the simple fact that they did not evolve together is the point, that's what makes it mosaic.
Mosaic evolution is a simple concept: That different traits do their own thing, over evolutionary time. They may in fact be related or influence each other, but the timing of the appearance of key traits, or of key changes, is not in lock step in the fossil record of a given lineage.
The fact that tooth size went up then down IS related to the concept of directionally. People often use the word "linearity" in this context, but I find that misleading. Evolution is linear, folks, because it is change over time and it, time, only runs in one direction as far a we are concerned. Yes, a trait can change back and forth, but it is not really going back in time, it is just wafting back and forth across some range of variability as time marches on. When the tide comes in then goes out and then comes in again, that second tide coming in is not the earlier tide repeating itself, it's a different event. Evolution is linear in that it flows one way through time.
But linear implies to many, and justifiably so, straight or unidirectional. Strictly speaking, a "line" is straight, and evolution is in fact straight in the sense that time is an invariant property. But now we are getting metaphysical. Throwing out "linearity" for this reason is misleading. Rather, we should be throwing out "directionality" to make that particular point. The chimp-human last common ancestor did not evolve in the "human direction" ... the directionality of various traits was wiggly. That is not mosaic, but it is non-directional. "Human" is not a direction towards which evolution went, even though it happens to be a place where evolution ended up at a certain moment in time, and the same can be said of every species that ever existed.
Which brings us to the concept of teleology. Teleology is a sharpened directional evolution. Not only does evolution go in one direction, but that direction is oriented towards some goal that makes sense in reference to some larger scale totally obvious pattern of how life works, like all chimp-like creatures should want to evolve into humans, because then, well, they get to be humans! This is, of course, absurd. Who would want to do that.
Having said that, there certainly are instances of directionality in evolution. There are long term patterns that indicate a bias over time. For instance, some sets of lineages seem to get larger over time. Mammals, from the Paleocene to the present, have on average gotten larger brains, though many lineages have not had any brain size change at all. These trends probably make sense as the likely but not inevitable outcome of some relationship between species or species and their biotic or abiotic context. I would predict that you would find, if you looked for it, an increase in secondary compound production in plants and ability to digest or sequester such compounds in herbivores over time, because once this evolutionary "idea" is on the table, various arms races and opportunistic evolutionary events are likely to happen. But even if that pattern does happen, it can be suddenly reset by some major catastrophe, or by some competing evolutionary pattern that overrides it. Directionality in the evolution of neural systems, body size, complex molecules, etc. is likely to happen, but of what form, when, where, and among what organisms, is not predictable because evolution is not teleological.
So let's look at bushes and speciation in multi-celled organisms, especially animals, then we'll discover the true meaning of "transitional fossils," "missing links," and "beer." Well, maybe not beer, but these other two, yes.
As Gould says in the aforementioned essay, evolution generally does not involve the steady change of organisms in a given lineage, but rather, the branching off of new species (speciation). Thus, we get bushes, with the "branching off" event being (... wait for it ...) the branches! But, despite this bushy branchiness, the tip of a given bush can be traced back through one single line of sub-branches to the trunk (which itself may be a tip on some other bush, depending on what we are looking at taxonmically and in terms of scale ... we might be looking at the "bush of primates" instead of the entire "tree of life" for instance). So, the "lineage" of this branch tip, going back to some arbitrarily chosen starting point, is really a series of more or less invariant populations, populations that are represented in the fossil record as the same-ol' same-ol' for some thousands of years, then giving rise to another similarly boring lineage of a different population which we might think of as a different species, then another, and so on.
We hear sometimes that a "link" (missing or otherwise)2 or "transitional form" is a faulty idea because once we identify the transitional form, then there are now two new unknown transitional forms -- two new gaps -- that must also be addressed, one before and one after the newly discovered link, formerly known as missing. But this is NOT what is meant by "transitional form" or "missing link," at least when the terms are being used sensibly. What IS meant is simply one of those twigs I was just talking about. One of the long-term more or less invariant species that makes up one of the branch segments in the lineage from starting point to a particular twig at the outer edge of the bush.
An example: In the Paleocene and before (the Cretaceous) there were many itty bitty mammals running around, some of which are ancestral to living mammals. So, go back in time and find one of these critters that is related as a direct lineal ancestor to the modern Impala. Now, show me all the different species that existed on that specific linage, from itty bitty pre-bovid of the age of the Dinosaurs, to the modern Impala. I don't want you to leave any out. I want fossil representatives of every one of the branch segments, the different species that branched off from an earlier species and hung around for a while, on that lineage, but do leave out the side branches. Only by having each and every one of these species, each having existed for some time in the past, some still existing today (but only an earlier segment of that branch being relevant to the lineage under consideration), each representing one segment of the pre-bovid species I'm thinking of and the modern impala lineage, do we have the whole evolutionary story of the impala.
In fact, most of these species are actually missing from our actual fossil record, but let's say you did a great job with your fossil hunting and found every single distinct lineage that could be called a different species, each giving rise to a branch that became the next on that specific lineage. But one. You are missing one. It is a gap, a missing bit of the record. Let's further assume that the species that predates the missing one, that gave rise to it, had a significant evolutionary difference (number of stomach chambers, ability to produce some cool digestive enzyme, number of bones in the foot, major feature of the dentition, whatever) from the species that came immediately after that missing one.
When did that evolutionary change happen? And if there are two or more such changes, were they in sync, or not?
That missing branch, the one you failed to find despite your Herculean efforts, is a "transitional" form between the two forms that we do have examples of before and after it. It is a missing link. For each evolutionary difference between the pre and post "gap" species, we have a question: Did the shift happen at the beginning of the missing form, so the missing form has it? Or did it happen at beginning of the post-gap species, so the missing one does not have it? Or, is this a rare case of phyletic gradualism where it appears slowly over time in the species that is missing? (Let's assume not.)
Whatever. The point is that when you find me good fossil evidence of that missing species, so we can describe it, then we're done. Finding it does not make two new distinct species that didn't exist before suddenly spring into reality. This is not Zeno's paradox but with mammalian morphology. (In fact, if there is a "gap" there may be two or more branch segments from a given lineage in that gap, but since this was my thought experiment, I decided that there was only one in this case.)
This discussion relates to the pattern of evolution at the level above (just above) the species. At a finer level, there is not likely to be a simple sequence of branches or lineages. Things that may look at first glance to be different sub species might be different populations that later merge (or partially merge). Even what seems like distinct species can exchange subparts of their populations, merge, redivide in some quirky way, spawn hybrids that then become new species or not, and so on. Below a certain taxonomic level (perhaps, better stated, below THE taxonomic level) the whole thing is much messier, which is why seeing the term "Missing Link" in reference to a population (see this post) is annoying. But at the larger scale of taxa, above species or genus, most of the time, in animals and to a lesser extent across multi-celled organisms generally, the branches can be thought of as forming distinct lineages. Lineages made up of distinct branches that happen to be plants (our metaphor runs smack into the subject matter, I'm afraid) are much more likely to recombine or hybridize than with animals, and all sorts of other complexities can happen, but the fact that those complexities are there (and they are very interesting) does not make species go away as a real biological entity, and it certainly does not make lineages go away ... it just makes them more complex. Oh, and if you've ever seen a strangler fig, you would not worry that branches are a poor metaphor just because when they tangle they can combine.
1And shall remain true right up until the day that we discover a more plausible stand-in for an Australopith human ancestor that does not exhibit megadonty, indicating that thew whole two-step thing with teeth ends up being wrong, but that is very unlikely.
2I quickly add that you should never, ever, use the term "missing link" because it carries so much idiotic baggage that it has become meaningless at best, misleading in most cases. Of course, it still may be useful as a literary reference. See this.
SJ Gould, 1977. Bushes and Ladders in Human Evolution. Reprinted in Ever Since Darwin: Reflections in Natural History. Norton Publishing, New York
Excellent post! (And this time my re-post will be done correctly! :) )
I agree that a stem in a lineage will cover the entire transition, but you did gloss the case of phyletic evolution, in which case it will be hard to draw the line between one transitional form and another.
Thank you for that. I had heard the term "mosaic" and for some reason thought it was about bushiness. Now, I have dusted off my Gould and looked up the chapter you mention (I had not realized it was a reprint) and read it again. Your post is a little more clear but the distinction you make is the same as the distinction SJG made.
Vance, it does not even work with phyletic evolution, but yes, I did gloss it because the explanation is a little off track.
If we have a series of discrete entities, than one embedded in the series is flanked by two other discrete entities. BUt, if we are seeking a list of discrete entities (things that are like species) that this will be covered. If we have a spectrum, and you pick a point near the middle, it is not flanked by two discrete entities.
If "mosaic evolution" is real, does that not mean that multiregional evolution is not dead?
And why do we believe that the ancestor of humans and chips is a chimp and not something unlike either?
And why do we believe that the ancestor of humans and chips is a chimp and not something unlike either?
Because everything found around the time of the LCA is chimp-like?
Chimp, not chip.
You mean "mosaic evolution" is not change over time in the patterns of little bits of tile in ancient Roman ruins?
This is a good discussion, but it is somewhat misleading because the transitions are not continuous and is not a single line. There are a discrete number of ancestors between that ancient mammal and modern extant examples. You actually (in principle) could have a genome of each one.
The problem is that there isn't a single line, it diverges and then converges and after a not so short time, the whole extant population is a descendent of that original ancestor. Not every one of the extant population will be an ancestor of the future population, but a big chunk will be.
In that whole population that will give rise to the new population, you may have both traits. Large brains and small brains, big stomachs and small stomachs. You may even have genes for the traits without the traits themselves being exhibited.
Some of those traits may only be expressed when particular patterns of genes in the population come together. If that trait is good, big brain and big stomach genes for both traits might get conserved. If the trait is bad, big brain and small stomach (or small pelvis), there might be cephalopelvic disproportion and genes don't get conserved.
@Gil: Silly man. Mosaic evolution is the doctrine (nay, teh REVEALED TRUTH!!11!!) that humans descended from Moses. Ignore all that hooey Greg wrote.
(Poor Greg. Tries to write a serious, thoughtful scientific post and the wiseacres take over.)
The problem is that there isn't a single line, it diverges and then converges and after a not so short time, the whole extant population is a descendant of that original ancestor. Not every one of the extant population will be an ancestor of the future population, but a big chunk will be.
That is true, but it is easily managed. Fossil "species" are pretty large entities (because they have the time dimension as well) and even with living species, it isn't really the case that "everything" converges and diverges and so on. As a paleomammalogisgt of sorts, I've worked with fossil australopiths, fossil elephants, and fossil galagos. The possibility of these three ever converging or splitting off in relationship to each other has never been an issue.
It is very important to maintain focus on the scale, and to take that into account, in working with fossils, in working with evolutionary patterns, and in working with these concepts like mosaic or lineages. Everything I'm saying in this post and the preceding post on "missing links" has been explicity at a scale large enough to override those concerns.
Greg Laden @ #12
As a paleomammalogisgt of sorts
Paleomammalogist: from Greek palÃ¤ (old) and Latin mamma (female breast): a scientist who closely studies old female breasts
The idea of a tree of life never made a lot of sense to me. A web always seemed closer to reality. If someone could comment on the following as it sort of smacked into my brain at a hour where thoughts are either stupid or inspired, or both.
A graphing of individuals, one axis representing time,
new branches can start along another branch (parthenogenesis) or at the crossing of 2 branches, at conception, and ends at death. By definition there would be separation at the species level, and by charting boundaries you would see the so called tree of life emerge.
When a fossil is found, the best case is that you find the end of one, single, line. A huge amount has been learned from them, and more still lies in wait, but you cant find a missing link in the tree, as the tree is a convenient fiction. In reality there is a web, and in a web as diverse and interconnected as a global ecosystem, what is one missing link, or for that matter, a single found one.
Robert, at a certain level above the individual the web becomes much less important than the tree. You are thinking at the level of intermingling individuals within overlapping populations. Trees operate at the level of distinct lineages that are no longer interbreeding at all. (Or if they are, only very very rarely).
All known Australopiths dating to after 3.5 million years or so had bigger teeth and thicker enamel than a presumed chimp-like ancestor, and it does not matter that the biggest-toothed, thickest-enameled ones did not happen to be our direct ancestor. It is still true1 that tooth size went up, then down
*Can* we be sure that the ancestors of the austrolopiths increased their tooth size? Could it be, rather, that the LCA had very large teeth, and that *both* the human *and* the chimp lineage have had reductions in tooth size? Not saying that *is* the case, but it seems to me that that's one possibility, or am I missing something?
It is a possibility. But, the evidence for what the LCA looks like is pretty strong Gorillas and chimps have similarly thin enamel, both chomps and bonobos have similar enamel. The earliest australopith remains have thinner enamel than the later ones. But, we've been surprised before.
I have been very interested in functional languages, and I still am, I suppose, but it's pretty obvious to me that to do the work that my company's program does now, functional languages would be a pain. All of them. You'd need extremely skilled people to be able to use them for anything non-trivial, and even then you would have to sweat blood to do things that are easy in the languages made for our kind of software - Java and C#.
Our understanding is that the term "evolution" is misleading. What is more accurate? Descent?
I hate to disagree with you Greg, but Mosaic Evolution is clearly the process in which many many individual tiles slowly come together over millions of years and form a low resolution mural on a wall or floor.
Greg: Feel free to improve the content on Wikipedia anytime. It is beneficial for all of us :)
I don't like messing with Wikipedia directly. It requires skills I don't have, and there is something about the consensus approach that I'm glad to see in operation there but can't participate in. I would only engage with subjects where I'm so much more likely to be right than anyone else involved that I would have to insist that my text not be substantively changed.
Plus, I have this totally obnoxious self aggrandizing and egotistical attitude. I keep it well hidden, but it would be obvious in the Wiki context.