Wednesday, February 24, 2016

Response to Judge Starling on Novelty

I've just experienced an interesting scientific exchange through social media. I believe much of the exchange boils down to "novelty" - an obsession of mine.

Here is my version of what happened:

1. I saw Dan Graur's post on "All of Evolutionary Biology in 12 Paragraphs" linked here, a provocative claim to distill all of evolutionary biology into 237 words. Very interesting! Do I agree?

Well, #10 stood out to me as false, especially the claim that "there is no true novelty in evolution"

Update: The full statement is here "10. Evolution cannot create something out of nothing; there is no true novelty in evolution."

I am confident that Professor Graur is not a creationist, but this sounds like the creationist logic that goes, "there is only microevolution, but no macroevolution". I am sensitive to these things, having been quoted out of context by creationists. I could imagine creationists using this quote to argue things like 'evolutionists agree that evolution cannot produce new information - only destroy it'. Well, to my mind all of this is about equating processes that occur in populations with macroevolution. Macroevolution and microevolution are not the same -- even if microevolution is involved in every step along the way.

My first response to the 12 paragraphs was this tweet:
If there is no true novelty, evolution cannot happen. Life could not evolve, photosynthesis could not evolve, eyes could not evolve. As I tweeted, macroevolution made ALL of biodiversity. Every species, every trait, every gene of every species. It simply does not follow that evolution could even occur without novelty. Evolution is a tinkerer, and biological entities get copied (at all levels of organization) and they diverge from each other. This is the source of novelty. True novelty. Before the origin of Pax genes, there may have been Paired domains, and there may have been Hox domains. But these domains came together anew (yes, by a single mutation in a single individual, originally, Prof Graur's #5) - but that WAS a novelty, a true novelty. Once that first Pax gene was fixed in that first population by (say) natural selection, evolution created a novelty. Evolution creates novelty like this every day, and has for billions of years to give us slime molds and sloths.

As far as I can tell, my original tweet went mainly unnoticed. I think because I replied to a particular tweet in a thread discussing the "12 paragraphs" tweet.

Next, someone else had the same critique I did about #10 of the 12 paragraphs. Here is what bluebear tweeted:



Dan responded, and I replied to that response as follows:



Well, there is the source of the disagreement/argument. Dan says there is no creation from nothing (creatio ex nihilo) in nature. I agree this is true IN POPULATIONS but as stated above, I do not agree this can be true for evolution AS A WHOLE.  In the tweet, I alluded to this difference by writing "don't submit to popgen hubris of no macroevo".

By that I meant I think there are aspects of macroevolution that are not explained by population genetics alone. For example, I believe the evolution of complexity and diversity are not understood by thinking only about populations. Eyes and species and photosynthesis cannot originate within a single population (usually). These are composites of multiple changes that happen across time, in multiple different populations/generations. Sure, populations change at every step by processes like natural selection. But we can't know about natural selection in that population way back that first had a Pax gene, or that first added a pigment to a photoreceptor cell in the early evolution of eyes. But we still can ask questions about the timing and order of these macroevolutionary events. These are questions outside the scope of population genetics, and within the realm of macroevolution. I personally am not that interested in documenting some change in allele frequencies. But how did complex features, like eyes evolve? How did animals come to be able to *produce* light. These are the big questions that interest me, so I am sensitive to people dismissing these questions by equating microevolution and macroevolution. Such as this tweet:
Dan then tweeted:

It is true that study of macroevolution is a lot about studies of patterns. I am fine with that - we can learn a lot from pattern. But it is an interesting question - what are the rules of macroevolution. To the question, I fairly quickly responded:


Professor Graur then asked to what I was referring, and I answered as follows:

This led to his reading the book in an impressively fast amount of time. He just as quickly dismissed the book, writing:


A bit later, he wrote a longer critique of the book on tumblr, linked here.

Now things get a bit complicated, because the tumblr post is a critique of the book, and a statement that I suck at judging books because the book sucks.

I will discuss a few of the critiques of the book, where I can. But most of the statements are just subjective, so I cannot really comment on opinions of "pretentious" or even "self-contradictory" when no specific instances are stated.

But first, I reiterate that my main reason for citing the book is that I think thesis of McShea and Brandon IS in fact a rule of macroevolution. In macroevolution, duplication happens. After duplication** happens - at all levels (protein domains, genes, networks, genomes, cell types, organs, modules, populations). After biological entities duplicate, they go their own way. This is the source of evolutionary novelty, biodiversity, and complexity. I see nothing in the tumblr post that argues against this. The tumblr post says very little of substance, in my opinion (although perhaps there is substance behind the comments; but that is not stated in the post). Even though it is a side track from the reason I cited McShea and Brandon, below, I respond to some of the critiques of the book that are written in the tumblr post.


**duplication is not a very precise word here, but it is easily understood by people who study molecular evolution because of gene duplication. What I really mean is "furcation", a word I have coined to include splitting of lineages at any level of organization. Splitting could be duplication, but it also could be fission.

Below, I respond in line (in black text) to Dan Graur's tumblr post about Biology's First Law... The text from the tumblr post is in in blue.
Todd Oakley recommended the book. I became intrigued. Now, that I finished reading it, I’m not impressed. It is a very pompous philosophical treatise
Pompousness is a rather subjective critique that is unrelated to the content of the arguments in the book.
that attempts to explain everything from the evolution of organisms to the fate of unattended picket fences by using a single law.

The unattended picket fence is just an analogy. The authors are not trying to explain the fate of the pickets with their law, which is a biological law.

It’s a law whose purpose is to find an answer to Life, the Universe, and Everything,
I don't recall that the book claims to answer Life, the Universe, and Everything. But it has been a few years since I read the book.

but as opposed to number 42 in The Hitchhiker’s Guide to the Galaxy by Douglas Adams, McShea & Brandon’s First Law is neither amusing nor original.  

Their “First Law” states “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.”
Yes, I believe this to be a rule, or law, of macroevolution.  This is the reason I cited the book - because Dan Graur asked for a rule of macroevolution after claiming that macroevolution is nothing but microevolution.
The mere fact that for these two authors “diversity” and “complexity” are synonymous should have been a warning. (This synonymization was referred to as asinine in several reviews.)
I believe under the definition that McShea and Brandon use for "complexity" -- which I believe is more clearly communicated as "structural complexity" -- that biodiversity is usefully lumped with complexity. More biodiversity is more different kinds of species; more structural complexity is more different kinds of parts (I like to call them components).

 Professor Graur does not state WHY this line of thought is asinine, so I cannot argue against assertions, nor calls to the "authority" of "several reviews".
However, it took me a while to understand that the two authors are either talking about entropy without mentioning the term entropy,
The authors do distinguish their ideas from entropy. I did a quick Google books search, and they mention the similarity on Page 12. There they refer to another chapter where they discuss the difference. My memory is that the biological law (the ZFEL) varies across levels of biological organization in a way that entropy does not. For example, in animals (with multi-cellular bodies) the complexity within each cell goes down (because of specialization) - even while the complexity of the wholes organisms increases.

In any event, I agree it is a useful analogy to think about entropy - after furcation, shit happens, and biological entities diverge. When divergent copies are maintained, complexity or diversity goes up.

or about variance in a random genetic drift process without saying so.

I believe the authors also equate the ZFEL to increasing variance in drift processes. I think they discussed Brownian Motion, a drift model I know about from phylogenetics.
After deciding for myself that the book is shit, I looked for opinions about the book in the literature. Professionals, it seems were not impressed.
As noted by Mohan Matthen from the University of Toronto, the two parts of the law are equally problematic.
The first part “In any evolutionary system in which there is variation and heredity, there is a tendency for diversity and complexity to increase” can be easily shown to be false.
Ummm, so if this can be easily shown to be false, has it been shown false?

The additional clause “in the absence of constraint,” in which “constraint” is undefined, is untestable and hence unscientific.

Constraint is defined.  McShea and Brandon point out that selection can constrain the ZFEL, such that complexity will not increase. The idea is that often, selection will oppose increases in complexity because more complexity will reduce fitness.

In molecular evolution, one process that constrains the ZFEL is concerted evolution. After duplication, we expect genes to go there own way - we expect the complexity of that gene family to increase. But concerted evolution keeps them similar or the same.
Other reviews, such as by Noël Bonneuil from the Institut National d'Etudes Demographiques and Kele Cable from the University of Minnesota were also quite lethal in their verdicts on the merits of the book. 
“Biology’s First Law” is clearly neither a first nor a law.
Okay, why?
Following Samuel Johnson, I can state with confidence that McShea & Brandon book is both good and original. Unfortunately, the part that is good is not original, and the part that is original is not good. 
If you don’t believe me that none of the ideas in McShea & Brandon’s book are original, kindly read Brooks & Wiley’s 1986 Evolution as Entropy to judge for yourself. As noted previously, before making extravagant claims, kindly read the literature. 
I think this request to read Brooks and Wiley (it's actually 1988) is directed at McShea and Brandon. From a Google book search, here are a few times that M&B cite B&W:

page 11 "based on what we have said so far, some will be poised and ready to make the leap from the notion of the accumulation of accidents to the second law of thermodynamics (... Brooks and Wiley, 1988).

page 12 - " some work in the past few decades on the application of the second law to biology has been inspirational (especially Wicken, 1987; Brooks and Wiley, 1988; Salthe, 1993), and we gratefully acknowledge the intellectual debt. "

There are other citations too.


So, who needs to kindly read the literature before making extravagant claims?


And back to my reason for citing McShea and Brandon in the first place. A rule of macroevolution is that complexity increases. Perhaps this is in fact quite related to the idea of entropy. So, by stating that M&B is not original, I assume that Prof. Graur accepts the thesis of Brooks and Wiley, 1988. Does that mean he accepts my "rule" of macroevolution, that complexity and diversity happen?

I’ll finish by quoting an anonymous critic on Amazon, “The authors’ shocking ignorance of (or willful disregard of) the history of evolutionary thought is the only surprising thing in this book. I can only hope that they are as embarrassed by their own poor scholarship as they deserve to be.”
Simply asserting and accusing, without backing up those assertions or accusations is the epitome of poor scholarship. The one specific example of poor scholarship above (Brooks and Wiley) did not hold. It was cited multiple times by McShea and Brandon.
There are amazing scientists out there that I don’t trust with books, movie, or restaurant recommendations. Yes, it is difficult to reconcile the Todd Oakley, who conducted such groundbreaking studies in molecular evolution, as his “The origins of novel protein interactions during animal opsin evolution,” with the Todd Oakley who was greatly impressed by a very bad book.
Thank you for the back handed compliment.

I will say in response that I very much respect Professor Graur. I learned molecular evolution from his text book! Once I asked him via Twitter about radical amino acid changes. A couple days later (or less maybe) he responded with a detailed review on his tumblr page. That was an amazingly collegial and scholarly thing to do! I was greatly impressed.

Today, it is hard for me to reconcile that Dan Graur with the Dan Graur who wrote this mainly content free, yet still blustery, critique of a book that I respect, written by authors whom I respect.

Anyway, the latter Todd Oakley owes me $14.04 and half a pound of candy as compensation for his book recommendation. 
I'd rather use my $14.04 to buy each of us a beverage (and a pound of candy), and discuss all this the next time we meet in person.

In the end, I think it ironic that all our disagreement comes down to our view of what 'novelty' means. I believe that novelty must be rampant in evolution - even true novelty. Novelty can come from new combinations of existing biological elements. So to can novelty and originality in scholarship come from new combinations of existing ideas. This is how I read McShea and Brandon. Many of the ideas I had seen or heard before - partly because M&B are professors where I was a graduate student. But by putting many ideas together in a new way,  namely by being bravely and ambitiously general, I found the book Biology's First Law to be a novelty. A true novelty.



Evolution cannot create something out of nothing; there is no
true novelty in
evolution

6 comments:

Thomas Turner said...

Thanks for posting this. I think the specific disagreement about novelty is mainly semantic. If you fuse two domains to create a new type of protein, is that new or recycled? Here is what would help me understand Dan's definition of novelty: in what OTHER processes is there true novelty? In human thought & culture, for example, you might also posit that there are novel ideas, perhaps like the number zero. Alternatively, you might think that all ideas are built on/from other ideas. The number zero was only thought up after the idea of numbers was thought up and people certainly already understood not having any of something, so these ideas were just combined.

Todd Oakley said...

Thanks Tom. Yes, it is mainly a semantic/definition issue about novelty. It is actually a bit of a paradox. I understand the idea that evolution recycles and uses existing parts all the time. But the simple fact that we have biodiversity indicates that there was novelty somewhere along the line. Otherwise, we logically would need to posit a homunculus that had ALL of biodiversity at the origin of life that just got recycled differently in different lineages.

Thomas Turner said...

It does seem somewhat of a macroevolutionary pattern, because the sampling period matters. If you look back all the way to the origin of life, it seems right to say something novel has evolved between then and now (life). Intermediate sampling periods seem more ambiguous but still reasonably 'novel" (e.g. eyes). Even 4 billion years ago, however, there were many small steps that were just using what was already available, and the novelty emerges in hindsight as a macroevolutionary pattern. Macroevolutionary process I'm not sure about, but macroevolutionary pattern seems right.

desmondramirez said...

Fun and thought provoking read Todd! Tom, towards your first point, would it be helpful to define a true novelty as something that emerges from the combination of two (or more) pre-existing components that is greater than the sum of those components? Alternatively, novelty is not an extension of a current function.

In your example, if two domains fuse to create a new kind of protein whose function could only come from the fusion of those domains, then that would be a novelty. The domains are not novel, but the new protein itself is.

Re the number zero-- zero as a digit to represent the count of nothing is probably not novel. But maybe we could see it's role in mathematics as novel, since in mathematics zero is not just a placeholder for a null quantity and has unique mathematical properties.

Dan's examples of venoms-- venoms are composed of genes who's relatives had other functions. But venoms are not *just* digestive enzymes. The sum of their activity comes from the unique combinations of digestive enzymes+other stuff that makes them lethal in very specific ways. The function of venoms is greater than the sum of the individual digestive enzymes that go into it. You could throw together a suite of digestive enzymes and inject it into an animal and have nothing happen. And sure, venoms have evolved multiple times independently, maybe multiple times independently from digestive enzymes even, but they are still novelties.

Final example: eyes. Let's define a extremely simple eye as two light sensitive cells next to each other associated with pigment. The phototransduction cascade it uses is pre-existing. The pigment synthesis pathway is pre-existing. Maybe even the cell type these are expressed in is pre-existing. What makes this eye a novelty? It is the combination of expression of the phototransduction cascade in adjacent cells AND the association of those cells with pigment synthesis that allows a new function that was previously unattainable by any of the pieces alone-- directional light sensing.

So then what isn't a novelty? Can novelties even exist at a popgen scale? Based on what I outlined above, new beak morphologies in Darwins finches aren't novelties. The evolution of host preference from hawthorns to apples is probably not a novelty. Still cool, interesting and important, but not evolutionary novelties. We can do popgen analyses on venoms, watch evolutionary arms races between predators and prey. On the other hand, watching the arm race now is watching modifications of current function, not seeing the emergence of an entirely new function from parts that had previous functions. Can we find populations in a non-venomous species with non-venomous ancestors that have become venomous? Could we consider that watching the emergence of an evolutionary novelty at a popgen scale?

I have now officially spent too much time writing up this comment instead of my current manuscript but I'm really interested in this conversation. Never sorry.

Thomas Turner said...

That would be a permissive (but still potentially valid) definition, Des, and would definitely occur in microevolution. The new protein of two domains has to start as a polymorphism in a population. Your host-switch case could also be novel. A single amino acid change in an olfactory receptor may change the perception of an odor but might not change a behavior. A different change might have the same properties, but together they might have a threshold affect and shift host preference to a new fruit. This is literally more than a sum: it is epistasis.

Among the most novel things I know of in microevolution are “novel” genes. As listed in the abstract of this paper (http://www.pnas.org/content/103/26/9935.short), they might arise from “exon shuffling, gene fission/fusion, retrotransposition, duplication-divergence, and lateral gene transfer” — all clearly recycled but maybe “novel” depending on your definition. The most novel case of all, and the focus of that paper, are “de novo” genes. Even these aren't actually de novo, though. They are made up of DNA that was previously not protein-coding, but that DNA is itself made up of a bunch of dead bits of old genes. If you wanted to assemble a new gene from random AGCTs you would never get there, but the base composition of junk DNA is non-random due to its previous functions.

Bogi T said...

Yes, I believe that his disagreement is mainly semantic, and population geneticists tend to see distinctions between character identities and character states as nothing more than semantic quibbling and not something that reflects the way in which development is structured. That's why he lumps evolution of halteres and origin of wings from forelimbs in the same category. If Dan was comparative morphologist he would see that there are certain patterns in morphological evolution, and perhaps the most important one is individuation of serially repeated body parts. But I'd bet you that he wouldn't be so dismissive if he had read Ken Weiss's paper on duplication with variation from 1990; I know that he really has a lot of respect for him.