The Historical Importance of Modularity

I am currently sitting in on a graduate philosophy of biology seminar and the theme of this semester’s seminar is evo-devo and we recently discussed the concept of modularity. I’m also sitting in on a history of biology course and we have talked a little about the early 19th century French scientist, Georges Cuvier. While attending the seminar, I was delighted to make a historical link between the two! (And oddly enough, one of the works we read in the seminar was a chapter from a book on modularity co-authored by Gunther Wagner which opens with the same link I had made.)

In my posts about the Spencer-Weismann debates, one of Spencer’s anti-Darwinian arguments I focused on can be called the “correlation of parts.” It was an idea articulated earlier by Cuvier, who said (as quoted by Wikipedia),

Today comparative anatomy has reached such a point of perfection that, after inspecting a single bone, one can often determine the class, and sometimes even the genus of the animal to which it belonged, above all if that bone belonged to the head or the limbs. … This is because the number, direction, and shape of the bones that compose each part of an animal’s body are always in a necessary relation to all the other parts, in such a way that – up to a point – one can infer the whole from any one of them and vice versa.

This article from the Academy of Natural Sciences quotes another passage from Cuvier:

Every organized being forms a whole, a unique and closed system, in which all the parts correspond mutually, and contribute to the same definitive action by a reciprocal reaction. None of its parts can change without the others changing too; and consequently each of them, taken separately, indicates and gives all the others.

The parts of an animal are so thoroughly integrated, that changing a single part – without proportionately modifying everything else involved as well – would produce a non-working animal. The correlation of parts prevented Cuvier from accepting Lamarckian evolution, or transmutation, in principle.

The correlation of parts later became an argument for Lamarckian evolution against Darwinian evolution,* as exemplified by Herbert Spencer. (I wonder if there has been any study on this shift in the principle’s use?) He believed, like Cuvier, that frequently a feature of an organism is tightly linked to a multitude of other parts and it would be very unlikely for all these parts to simultaneously vary to the appropriate magnitudes and directions required by neo-Darwinian evolution that prohibits the inheritance of acquired characteristics.

For example, an elk’s antlers are supported by thick skulls and strong back and neck muscles. If an elk were to gain an extra point to the rack through continued growth, increasing its weight, its muscles would be strained to keep the head up and the animal wouldn’t survive. However, in Lamarckian evolution, the muscles that the buck strengthens over its lifetime are passed on to its offspring. Over generations, the muscles would continue to strengthen to the point where extra growth in the rack could be supported. Thus the correlation of parts showed that pure Darwinian evolution couldn’t work in elk antlers, but Lamarckism could – according to Spencer, anyway.**

So, historically, according to some, evolution (or specifically, Darwinian evolution) couldn’t work because of the correlation of parts. I think the concept of modularity easily resolves this.

A module can be loosely defined as a semi-autonomous individual part or process. (I have written about modularity before in a previous post.) An example of modular part is the assortment of lobster appendages: they act and move mostly separate from the rest of their body. A modular process is exemplified by the development of the appendages themselves: each appendage develops indepently of the others, and these appendages are specified and created by modular genetic networks and pathways. Additionally, these modular structures, processes, and genetic networks can evolve freely from the rest of the organism-at-large.

Modularity answers Cuvier’s and Spencer’s objections to evolution. Yes, an organism may be well-integrated, but it is also built of semi-autonomous modules. A part or process of an organism can vary without the rest of the organism falling apart! Spencer argued that if Lamarckian inheritance, which sidesteps the correlation of parts, weren’t true, then evolution couldn’t happen, but perhaps he was half-right: modularity (which also sidesteps the correlation of parts), not Lamarckian inheritance, is what allows evolution to happen.

* I am obviously omitting Darwin’s views on the topic, but given what I read in Gould’s The Structure of Evolutionary Theory, his views seem a bit muddy and complicated. Perhaps I will investigate Darwin’s views on the correlation of parts at a later date, but here I am focusing on correlation of parts vs. modularity.

** I don’t understand why Spencer excludes the possibility of later fluctuations or mutations that could arise after the weight increase in the antlers. He also doesn’t mention how the skull would thicken as a result of Lamarckian processes. I don’t think Spencer’s argument works at all given his framework, but it still illustrates the use of the principle of correlation of parts as an argument against Darwinian evolution.

Ants and Their Castes in the Spencer-Weismann Controversy

Ant (Wikipedia)

Ants are evolutionarily weird and are quickly rising in my favorite organisms list.  The same evolutionary principles apply to ants as they apply to us, of course, but because ants are haplodiploid, live in large colonies, and have a caste system, biologists have to apply the same principles differently – it isn’t exactly intuitive. Ants (and other insects such as bees and termites) are frequently the subjects of hot debate when it comes to kin selection, but their role in evolutionary disputes is over a century old. Charles Darwin discussed them in The Origin of Species, but they were later the center of the controversy between Herbert Spencer and August Weismann.

As discussed in my last post, Herbert Spencer was a Lamarckian who believed in “inheritance of acquired characteristics” and thought natural selection was “inadequate” for explaining how organisms have evolved. August Weismann contested this; he was known as an “ultra-Darwinian” who believed natural selection explained all biological traits and thought Lamarckism was dead wrong. He believed this because of his idea, called the “Weismann barrier,” which postulated an organism’s hereditary material is contained exclusively in reproductive cells (sperm and eggs) called the germ line. With the barrier, there would be no way for the environment or acquired traits developed in the soma (any cell not sperm/egg) to directly influence the germ line because they were totally isolated from each other. Because their respective biological theories were being questioned in their entirety by the opponent and because neither theory was completely accepted by the scientific community at the time, the stakes were high. Were acquired characteristics actually inherited? Could natural selection explain all traits? What was the role of the environment in modifying traits? Was the Weismann barrier even real?

The “Controversy” began with Spencer’s “The Inadequacy of Natural Selection” and “Prof. Weismann’s Theories” in 1893, followed by Weismann’s response, “The All-Sufficiency of Natural Selection.” Spencer then published “A Rejoineder to Prof. Weismann” and a brief follow-up called “Weismannism Once More.”

The ants were not initially discussed by Spencer, but brought up by Weismann as an example of where he thought Lamarckism faltered, specifically in the evolution and maintenance of their famous caste system. In the caste system, individuals are divided into classes within the colony: typical castes are the queen along with various workers such as soldiers and caretakers (of the larvae). A special quality of the worker class is that they do not reproduce – only the queen and the males do. Weismann exploited this feature to attack Lamarckism for all it was worth.

Weismann observed that worker ants have reduced reproductive organs, eyes, and wings (“retrogressive”). Because they don’t need them, these reductions may have occurred because of inherited disuse (a Lamarckian argument). As Weismann points out, the argument has one problem: the workers don’t reproduce! They can’t directly transmit their reduced organs (315-316)! Such morphological (and also instinctual) changes occurred after the evolution of the castes themselves, he argues, and thus cannot explain the caste system (328-331).

August Weismann (Wikipedia)

Weismann also makes a point regarding reduced wings: even if workers reproduced, wings can’t degenerate through disuse because wings are perfectly formed before use and deteriorate because of use. Additionally, eyes in the workers can’t have degenerated through disuse because workers still use their eyes! Instead, Weismann argues, the atrophy of worker ant eyes favors his theory of panmixia: highly developed eyes are superfluous for the worker ants and are not maintained by selection (317). [2] [3]

In “The Inadequacy of Natural Selection,” Spencer argued that selection could not produce complex morphological changes because such complexity would require all of the involved traits to vary to similar degrees and in parallel directions. For example, evolving the ability to jump would require many concurrent variations in the skeletal, muscular, and nervous/instinctual systems. In Paleyesque fashion, Spencer lists over two pages of required variations, arguing that the probability of having dozens of traits vary in the right degrees and in the right directions is extremely low. Natural selection wouldn’t even be able to act on the ability to jump. Instead, “using” one’s jump ability strengthens the muscles together – no variation in the isolated germ plasm is necessary – and these acquired changes are what gets passed on to the offspring. (I don’t know how skeletal changes would work though…)

Pheidole megacephela - the big-headed ant referenced by Weismann. Yeah, weird as hell. (Photo by Hirotami T. Imai and Masao Kubota, from Japanese Ant Image Database.)

Weismann adopts the list-as-many-connected-traits-as-possible argumentation style for the ants, but to argue against Lamarckism. In addition to degenerated organs, worker ants have specialty traits not found in the sexual castes such as thorns and larger heads and jaws (“progressive”). “Many parts must have varied simultaneously and in harmony with one another.” He piles on, in Paleyesque fashion, all of the variations required to produce such morphological changes. But Lamarckism can’t handle all of these necessary variations; because workers are sterile, the acquired characters are not passed on to the offspring! Lamarckism can’t work here.

Instead of using “inheritance of acquired characters” to explain the castes, Weismann believes his germ plasm theory is sufficient, arguing that castes are differentiated by multiple “determinants” (similar to what we would call chromosomes, I presume) within the germ plasm (325-326). Selection then acts on the workers through selection on the queen who carries inactivated worker determinants (i.e., kin selection). Weismann seems prescient, eh?

He admits proving this hypothesis to be true is nearly impossible because the sheer number of ants in a single colony and the nearly infinite amount of minute variations involved would be impossible to study closely. However, according to Weismann, because Lamarckism is inadequate for explaining ant castes, “that it is necessary for us to accept the principle of natural selection. It alone can explain the adaptations of organisms without assuming the help of a principle of design” (319). We know Lamarckism and design are false, so selection, the only alternative, must be the solution to the ant caste system.

Not surprisingly, Spencer doesn’t think Weismann has refuted Lamarckism at all. Instead, Spencer offers his own evidence that selection is inadequate in explaining the ant caste system. He begins his argument with observations of wasps and extrapolating those to ants.

Spencer argues that social insect castes can be explained without selection. [4] He first invokes Haeckel’s concept of heterochrony – social insect development does not recapitulate phylogeny (Haeckel’s biogenetic law [5]), but has instead been reordered to reflect behavioral development. According to Spencer, just as girls play with dolls (maternal instinct) before having children (sexual instinct), queen-wasps progress from “building cells and feeding larvae” (maternal) to exclusively laying eggs (sexual). He argues that worker-wasps are undeveloped queen-wasps – their development was arrested before the sexual stage could arrive. (655). “Thus interpreting the facts, we have no occasion to assume any constitutional difference between the eggs of worker-wasps and the eggs of queens” and that “the larva of a worker-wasp can be changed into the larva of a queen-wasp by special feeding” (656). Additionally, he notes that malnutritioned eggs produce males. Feeding/nutrition – an environmental inductor – controls the caste system in waps, according to Spencer, and not information emanating from Weismann’s germ plasm.

Herbert Spencer (Wikipedia)

Although Spencer has been examining the wasps, he assumes ants develop similarly, applying the “arrested development” (658) argument to the rest of Weismann’s objections. For example, Weismann said that the worker caste has less developed eyes than the queen ants, but because the workers don’t reproduce, the Lamarckian argument of disuse can’t apply to them. Spencer replies that yes, disuse cannot apply here, but because eyes are among the latest organs to develop (larvae are blind), “arrested development” by lack of feeding leaves worker ants with underdeveloped eyes – they were prevented from fully forming (658)! Spencer says “arrested development” takes care of Weismann’s wing argument as well. Furthermore, many species show intermediate classes which under Weismann’s scheme would imply more and more determinants, but with Spencer’s “arrested development,” they are no surprise (659). [4]

He again applies the “arrested development” argument to explain away Weismann’s Amazon ant objection, that larger heads and jaws of the soldier class evolved after the caste system evolved and because they can’t reproduce, those features are not explained by Lamarckism. Spencer simply counters the objection by claiming the Amazon ants are most likely descended from ants that resemble the soldier-caste. He seems to believe that all ants with caste systems derive from an ancestral warrior species as he provides us with some intriguing ant/human sociology:

[The ancestors] must have had marked powers of offence and defence. Of predacious creatures, it is the more powerful which form societies, not the weaker. Instance human races. Nations originate from the relatively warlike tribes, not from the relatively peaceful tribes. Among the several types of individuals forming the existing ant community, to which, then, did the ancestral ants bear the greatest resemblance? They could not have been like the queens, for these, now devoted to egg-laying, are unfitted for conquest. They could not have been like the inferior class of workers, for these, too, are inadequately armed and lack strength. Hence they must have been most like these Amazon-ants or soldier-ants, which now make predatory excursions – which now do, in fact, what their remote ancestors did (663).

Thus Spencer bypasses Weismann’s objection that the soldier caste can’t transmit their acquired characteristics: they didn’t need to do so because they inherited them! “It is not that the soldier-ants have gained these traits,” he argues, but “the other castes have lost them” as “early arrest of development causes absence of [soldier traits] in the inferior workers; and from the queens they have slowly disappeared by inheritance of the effects of disuse” (663).

Spencer provides us with a Lamarckian framework for caste evolution. Not only does he claim that classical Lamarckian use/disuse inheritance is involved, but he also adds direct stimulation by the environment to explain how castes can form. (I think the latter is an underappreciated aspect of Lamarckism that I only just read about in Peter Bowler’s The Eclipse of Darwinism, but Spencer appears to find it an incredibly important aspect.) Weismann, on the other hand, also provides us with a useful framework in which to understand how the germ plasm theory and Lamarckian theory were distinct and at the time, completely opposed. We are left with no clear answer, however. As the paleontologist Henry Fairfield Osborn wrote about the controversy:

In fact, the reason these papers, interesting and able as they are, leave no final verdict in the mind is that neither meets the tests of scientific truth. When we look beneath the surface and recover from the first blinding effects of the brilliant style which characterizes both attack and reply, we see that both set forth mainly the modes in which nature may be supposed to act, rather than the mode in which nature does act. Nature, if anything, is illogical in many of her forms (313).

This discussion, at least in its theoretical phase, has reached its climax in this controversy (315).

In the end, and as with most false dichotomies, the answer to caste differentiation lies somewhere in the middle between Spencer and Weismann. Weismann’s barrier concept turned out to be correct, at least in many animals (to the exclusion of the rest of life… over 99% of it) – the germ cells are isolated from the somatic cells – but the barrier doesn’t rule out Spencer’s environmental influence on development either.

Caste differentiation is still a topic of debate over a century later. A study from just this past April found the protein responsible for queen development in bees that is found in “royal jelly.” When this “royal jelly” containing the protein is given to bee larvae, a queen bee results. Fascinatingly, when this protein is fed to flies – which have no castes – “queen flies” with large body sizes and ovaries come about. However, a neat 2008 review by Anderson et al. (PDF) points out that historically, environmental factors (as argued by Spencer) were shown to be the dominant form of caste determination (and involved more than just feeding, but season, temperature, and others) but that, at least in some ant species, genetic factors (Weismannian) have recently been shown to play a large role as well. Of course, genetics and environment do not act independently, but concurrently, and also act upon each other. Both Weismann and Spencer were right; they just didn’t know it!

_____________

[1] Panmixia is different from the economical hypothesis that traits may be lost because of the excess energy it costs to develop and maintain them. Instead, panmixia holds that selection has been removed allowing the trait to be reduced due to lack of maintenance.

[2] Weismann says Spencer accepts the giraffe’s neck as a result of selection because “the process appears easy to imagine” (320). A topic for future investigation.

[3] Weismann also applies panmixia to caste-specific behaviors (333).

[4] Here is an example of where I have trouble understanding why Spencer believes selection has no role to play in ant castes. Why couldn’t selection have produced the environmentally induced development? Is this an error/oversight on Spencer’s part, or me not properly understanding the debate (especially without the knowledge scientists have developed since)?

[5] So far, Spencer seems to not be applying Haeckel’s biogenetic law – ontogeny recapitulates phylogeny – but alludes to it when discussing ant wings: “Wings are late organs insect phylogeny, and therefore will be among those most likely to abort where development is prematurely arrested” (658).

______________________

History:

Osborn, Henry Fairfield. “The Discussion Between Spencer and Weismann.”Psychological Review 1 (1894): 312-315.

Spencer, Herbert. “Appendix B. The Inadequacy of Natural Selection, Etc., Etc.” The Principles of Biology. Revised and Enlarged Edition. Vol. 1. New York: D. Appleton and, 1898. 602-91.

Weismann, August. “The All-Sufficiency of Natural Selection.” Contemporary Review 64 (1893): 309-338.

Science:

Anderson, Kirk E., Timothy A. Linksvayer & Chris R. Smith. “The causes and consequences of genetic caste determination in ants (Hymeonoptera: Formicidae)” (PDF). Myrmecological News 11 (2008): 119-132.

Kamakura, Masaki. “Royalactin induces queen differentiation in honeybees.” Nature 473 (2011): 478–483. doi:10.1038/nature10093

“Spandrels” before “spandrels” were cool

In 1979, Stephen Jay Gould and Richard Lewontin famously attacked what they called the “Adaptationist Program.” They accused evolutionary biologists and sociobiologists of concocting “just-so stories” in which scientists would claim a particular trait, an adaptation, was a result of natural selection without rigorously testing their hypotheses. If they did test the claim and it turned out the claim was false, the scientist would create another just-so story, rarely questioning whether the trait was an adaptation or possibly a byproduct or fixed by non-adaptive processes. Most readers are familiar with this argument, so I won’t expand any further.

Upon reading material for my history major paper, I came across some arguments by the biologists T.H. Morgan and William Bateson that seemed oddly familiar…

T.H. Morgan in Evolution and Adaptation (1903):

They [Darwinians] have contented themselves, as a rule, with pointing out that certain structures are useful, and this has seemed to them sufficient proof that the structures must have been acquired because of their value (462).

Morgan was an experimental biologist which is evident in his argument: speculations about evolution and adaptations weren’t enough. This argument continues over a century later with disputes over evolutionary psychology studies that sometimes make large, sweeping arguments based on poor or little experimental evidence. Sociobiology was one of the targets of Gould and Lewontin, but Morgan was making similar arguments against the natural historians of his day.

Even more eerily though, William Bateson, in “Heredity and Variation in Modern Lights” (1909), wrote

By suggesting that the steps through which an adaptative mechanism arose were indefinite and insensible, all further trouble is spared. While it could be said that species arise by an insensible and imperceptible process of variation, there was clearly no use in tiring ourselves by trying to perceive that process. This labour-saving counsel found great favour. All that had to be done to develop evolution-theory was to discover the good in everything, a task which, in the complete absence of any control or test whereby to check the truth of the discovery, is not very onerous. The doctrine “que tout est au mieux” was therefore preached with fresh vigour, and examples of that illuminating principle were discovered with a facility that Pangloss himself might have envied, till at last even the spectators wearied of such dazzling performances.

I trust that readers familiar with the Spandrels paper will catch the parallel immediately: Bateson accused Darwinian selectionists of being Panglossian in 1909! 70 years before Gould & Lewontin launched their famous attack, Bateson had made an identical argument (although the former put it more eloquently and forcefully).

Thankfully, Gould was aware of this. Biologist Donald Forsdyke notes on his website that in the 1993 book dedicated to the ‘Spandrels’ paper, Understanding Scientific Prose, that

Gould conceded that “I did not know about Bateson’s invocation of Voltaire when I wrote ‘Spandrels,’ but the convergence is scarcely surprising, as Dr. Pangloss is a standard … form of ridicule.” To criticize the adaptionist program Bateson had introduced, not spandrels, but “toolmarks, mere incidents of manufacture, benefiting their possessor not more than the wire-marks on a sheet of paper, or the ribbing on the bottom of an oriental plate renders these objects more attractive to our eyes.” Furthermore, examples of the doctrine that all is for the best “were discovered with a facility that Pangloss himself might have envied.

Not only did Bateson make an argument about Dr. Pangloss, he even had “spandrels”! As is frequently the case, new ideas are rehashes of older ones, whether intentional or not. The famous phrases, “those who cannot remember the past are condemned to repeat it” and “plus ça change, plus c’est la même chose,” even apply to the sciences. It pays off to know your history (which Gould certainly understood).