CROTCHETS & QUIDDITIES

How the Eye Got Its BrainKENNETH WEISS

Evolution provides an explanatoryframework for biology. Our usual for-mula is to explain a trait by equatingits present function with its adaptivereason for being, invoking natural se-lection as the causal force. But sincethere is no trait that we won’t try toexplain by such necessarily post hocscenarios of deterministic molding byselection, evolutionary explanationsare often derided by critics as Just Sostories. This is irritating, because wethink we are going Kipling one betterby saying How the Leopard Really GotHis Spots (see Murray, 1997). But ifwe examine our stories more closely,it is difficult to know when we havethe true ones right. The evolution ofeyes illustrates this in many ways.

A WORLD OF EYES

Eyes have attracted evolutionary in-terest for a long time. They are a clas-sic case of improbability cited to sup-port arguments from design. Thereappear to be 65 or so phylogeneticallyindependent forms of eyes (e.g.,Salvini-Plawen and Mayr, 1961). Thisdiversity is exemplified in the outercircle of Figure 1. Systematics is a tan-gled bank on which I’m happy to letothers fathom the diversity of nature(e.g., Wagner, 1989; Lockwood and

Fleagle, 1999) but one doesn’t have tobe a cladist to see why eyes have be-come classic instances of evolutionaryanalogy rather than homology. In-deed, every reader of this column hasprobably answered a question aboutthis correctly on an exam. Or was thatanswer wrong?

On his tangled bank, Darwin shud-dered that the independent evolutionof such remarkably complex struc-tures as eyes would be so unlikely asto present a serious challenge to thecredibility of natural selection. He hy-pothesized an original prototype eyeconsisting of a photoreceptor, asym-metric light shield, and a covering,from which “the difficulty ceases to bevery great in believing that natural se-lection may have converted the simpleapparatus” into the diversity of “opti-cal instruments” seen today. It is notclear how widely he thought this ap-plied, but seems to have been suggest-ing homology as an escape valve fromimplausibly improbable parallelism.So the recent discovery of a geneticmechanism widely shared in animaleye development had a striking im-pact on evolutionary Just So stories,by appearing to be remarkably consis-tent with a homology scenario. Maybethe Just So story about eyes beinganalogous not homologous struc-tures, isn’t. You should have beenmarked off on your exam!

The location where a structure willform in an embryo can be specified bythe action of “selector” genes that in-duce cascades of gene expression anddifferentiation. A gene called pax6 isinvolved in the cascade that initiateseye development. Initially called eye-less in flies (because it explains How a

Fly Lost Its Way), homologous geneswere discovered and found to play asimilar inductive role for eyesthroughout the animal kingdom.Trans-species experiments show theprotein and its gene expression prop-erties retain significant functionalsimilarity between vertebrates and in-vertebrates. Pax6 also is expressed inmany tissues of developing eyes, in-cluding lens placode, cornea, iris, ret-inal and/or photoreceptor cells in atleast some taxa, the crystallin geneproducts that give lens and corneatheir optical properties, and pax6 mayinduce some opsin (photoreceptor)genes.

Yet how can we reconcile this factwith the clear evidence from phylog-eny and morphology that diverse eyesdeveloped independently? Basicallyfollowing Darwin’s logic, Walter Gehr-ing (2002) hypothesizes a Precam-brian proto-eye (Figure 1, center) thatevolved out of basic photoreceptor ca-pability in ancestral cells, and thendiversified through gain or loss ofstructures the diverse eyes that selec-tion demanded for each circumstance(focusing, following motion, etc.). Hecalls this “intercalary” evolution be-cause his idea is that in the evolutionof eyes various already-existing geneswere recruited in different lineages atdifferent stages of eye developmentbetween induction by pax6 at the top,and photoreceptor differentiation atthe bottom of the cascade.

Illustrating this kind of evolu-tion are the crystallins (Piatigorsky,1998a,b). Despite their name, theseproteins did not evolve specificallyas window panes the way that, say,hemoglobins evolved to carry oxy-gen (but see below!). Instead, crys-tallins typically have evolutionarilyprior functions unrelated to vision,but their properties made them suit-

Ken Weiss is Evan Pugh Professor of An-thropology and Genetics at Penn Univer-sity.

Evolutionary Anthropology 11:215–219 (2002)DOI 10.1002/evan.10095Published online in Wiley InterScience(www.interscience.wiley.com).

Some interesting Just So stories about evolution that, O Best Beloved, may notbe just so.

Evolutionary Anthropology 215

able to be opportunistically re-cruited by selection for intercalationinto eye development. In its searchfor panes, different proteins werepicked for the lenses and corneas ofdifferent lineages. Pax6 proteinbinds the regulatory region and mayinduce the expression of most crys-tallin genes in which this has beentested. In this sense crystallins haveclear-cut developmental and func-tional homology, but the genesthemselves are analogous, not ho-mologous.

NOT-SO, OR AT LEASTNOT-SO-FAST

This makes a nice story but thereare several cautionary elements. Gehr-ing suggests intercalation by recruit-ment of available genes into variousstages of eye development, but that’s afar cry from intercalating all the indi-vidual complex morphological bitsand pieces of compound eyes, eyeswith focusing lenses, enclosed orbit,and the like. How evolution could dothis over and over again is what per-

plexed Darwin. Is this merely anotherJust So story dressed in modern ge-netic clothing, or is it really plausible?

If we think of patterning processesrather than genes in the abstract, thestory indeed seems plausible. An or-ganism can be viewed as a set of de-velopmental gene-regulating circuitsrather than of genes each evolved for aspecific use (e.g., see my earlier col-umn, Weiss, Volume 11:3, 2002). Pax6is but one example. Other regulatoryor inductive genetic mechanismswidely shared across the animal spec-trum are responsible for gene-expres-sion cascades that serve functions thatare relevant to eye development: local-izing structures, inducing placode for-mation (lens, cornea), tissue invagina-tion (optic cup, retina), or repetitivepatterning (ommatidia, the 100� eyesalong a scallop’s mantle). These cir-cuits are parts of an ancient animaldevelopmental genetic Toolbox thatcan be induced under various circum-stances, of which eyes are but one.The same pathway is used in makingfly bristles, bird feathers, and yourteeth. There is almost every imagin-able kind of eye, but they are assem-bled by mixing—intercalating—theseready made processes into eye devel-opment.

Such mechanisms may not be sodifficult to induce rapidly in a newevolutionary context, as shown withinspecies by the occurrence of sponta-neous ectopic wings, antennae, legs,eyes, nipples, teeth, or hair. This mayjust require the mutational produc-tion of short regulatory-gene en-hancer sequences in the DNA up-stream of a gene that then becomesrecruited. As an example, the proteincoded by pax6 recognizes the se-quence TAATGCGATTA (with sometolerance for variation), and this se-quence can be found in variable num-ber and location flanking crystallingenes in the chick and various mam-mals. There are enhancer sequencesfor numerous other regulatory factorsin the same region, that vary fromcrystallin to crystallin (Cvekl and Pi-atigorsky, 1996). The number and lo-cation of these enhancers could affecttiming or quantity of expression. Inthis way eye homology may extendwell beyond pax6 itself, but it is anAmish quilt homology. When it was

Figure 1. Intercalary evolution of the eye. A primitive photoreceptor pathway cell leads toa prototype eye from which today’s eyes evolved. A range of very different types of eyesare found within mollusks alone. (modified from Gehring 2002, by permission from Int. J. Dev.Biol.)

216 Evolutionary Anthropology CROTCHETS & QUIDDITIES

time to see the light, each creaturegrabbed different bits off the develop-mental shelf, even if they all used pax6to make it happen.

This explanation makes a plausiblestory for eye evolution. But such sto-ries must be viewed with circumspec-tion. For example, when crystallingenes have been experimentally inac-tivated in mice, there have been littleor no changes in their eyes. If eyesdon’t need crystallins, or don’t needany particular ones, then perhapspax6 enhancers arose in the DNAflanking many genes, by chance or forunrelated cellular functions, generat-ing a statistical distribution of relativeexpression (and hence concentrationlevels of proteins) in eye tissue. Wecall the most abundant of these “crys-tallins” and feel obligated to explaintheir presence in terms of opportunis-tic recruitment by natural selectionfor their refractive properties.

We should at least entertain thepossibility that the random intercala-tion of developmental cascades ex-plains the repeated independent evo-lution of diversity of complex eyes asmuch as specific adaptive forces. Aneye that was good enough, was goodenough.

DARKLY THROUGH A LENS

The frequency of the light spectrumthat a given opsin responds to islargely determined by a few criticalamino acids, and many an adaptationtale has been told about how the re-sponse spectrum of the opsins hasbeen mutationally “tuned” by selec-tion to the living circumstances ofvarious creatures (Mollon, 1989;Yokoyama, 1999, 2000). For example,opsins in coelacanths (Latimeria) liv-ing below 200 M under water perceivethe dim, blue-shifted light that pene-trates that deep. This leads us to as-sume this is the result of submarinewarfare in which fish with essentiallyX-ray vision snapped up all the food,starving less optically perceptive com-patriots into extinction. But mightthere be a different Story. . . .

Once upon a most High and Far-Off time, Beloved, all creatures inthe great grey-green, greasy Lim-popo River had eyes like you andme. Alligator dozed on the banksall set about with fever-trees, andFish swam lazily in the light ofthe High Sun, fearing the deeps,where only the dim-light goes, lesthe bump his nose in the dark.One day in that very old time,

Coel-A-Canth strayed down, downin the O So Murky and Lo, hecould see! He and his mates frol-icked in the empty waters. Timepassed, and they were long forgot-ten at the Top. But if you godown today, you’ll find them stillat play, with their Eyes-for-the-deep-deep-blue.

In this alternative tale (Note 1), be-havior sorts existing variation with lit-tle or no need for adaptive selectionfor dim-light vision. Fish go wherethey can see, and stay to fish thosewaters (so to speak), associating andmating with each other. If you can’tsee in the dark you don’t go there. Asmutations conferring better dim-lightvision arose, the fish moved evendeeper until they had basically nocontact with the upside world. Muta-tions that eliminate color vision couldaccumulate because they have noharmful effect.

This scenario would lead today’scoelacanths to be as “adapted” as ifthe telegenic blood-and-guts melo-drama of classical darwinian selectionwere responsible. The genetic signa-ture of reduced variation in the opsingenes relative to general variation inthe coelacanth genome would also befound, because only a subset of thevariation in the population venturesdeep. An inverted alternative could bethat coelacanth vision is primitive invertebrates, but that gets us in toodeep for this column.

An anthropologically interesting op-sin story is that of primate color vi-sion. Old World monkeys, and greatapes (including humans) have two X-linked opsin genes, one sensitive tored and one to green light. This, alongwith blue opsin generates our trichro-matic vision. However, most NewWorld monkeys—and most other pri-mates—have only a single X-linkedopsin gene (see Boissinot et al., 1998).In many species this gene has a mul-tiple-allele polymorphism, so that het-erozygous females are trichromats,but homozygous females and allmales are dichromats. The number ineach category depends on the localallele-frequencies, which vary amongspecies and populations. Did a multi-allelic system evolve many times indifferent species? Perhaps. Howlermonkeys are a New World exception

Figure 2. Coel-A-Canth out of its depth in the great grey-green Limpopo River (Modifiedfrom Kipling, 1902; fish from http://www.unmuseum.org/coelacan.htm, copyright Lee Krys-tek 2001.)

CROTCHETS & QUIDDITIES Evolutionary Anthropology 217

that have true trichromacy (two X-linked opsins). Did that evolve inde-pendently? Probably.

These questions relate to whetherprimates were originally polymorphi-cally trichromatic and diurnal, or di-chromatic and nocturnal (Tan and Li,1999; Heesy and Ross, 2001). Detailedanalyses of opsin light-frequency sen-sitivities have shown the functionalvalue or adaptive use of color visionby various vertebrates, including pri-mates for whom color vision is valu-able for finding fruit in dappled forestvisual backgrounds (Dominy et al.,2001; Heesy and Ross, 2001; Jacobs,1995; Mollon, 1989). However, suchexplanations essentially equate presentfunction with past natural selectionand this can require ad hoc contor-tions. Explaining the strange patternsof polymorphism-based color visionby adaptive scenarios might be a chal-lenge even for Kipling, and simplerexplanations may be tenable.

The diversity of spectral peak sensi-tivities observed in the single X-linkedopsins of many primates suggests thatselective pressure maintains someability to use light in this part of thespectrum, but that what we see inmost primates today may just be therange generated by random geneticdrift in the opsin gene, consistent withthat constraint (Figure 3). It is notnecessary to invoke repeated Hobbes-ian wars of all against all, with specialcolor requirements consistent over

long time periods for each species, toexplain variable patterns of dichromatmales and a subset of trichromat fe-males.

The samples tested have generallybeen small, and bigger samples mightreveal variation in all these species. Itwould be rather arbitrary to use anallele frequency- or sample-dependentdefinition of a species’ vision, so wemay be over-interpreting the limiteddata. Once gene duplication gener-ated two X-linked opsins in anthro-poid ancestry, selection or behaviorcould play around with the greatersensitivities made possible, with onegene sensitive to red and the other togreen parts of the spectrum, but weshould keep in mind that red-greencolor vision is highly variable in hu-mans, too, and not all of us are truetrichomats.

A VISUAL MIRAGE

There is another way in which ouradaptive scenarios may be literal mi-rages. Humans have trichromaticcolor vision, but in important ways wedon’t need it. The loss of an opsin genecan lead to colorblindness, but JamesMcCann and Edwin Land (inventor ofPolaroid photography) discoveredwith clever filtering experiments, thathumans can perceive colors that arenot there. That and much other workhas shown that what we do is com-pare the signals received by our differ-

ent opsins from the same visual“pixel.” Comparative processing en-ables images to retain their perceivedproperties under varying light inten-sity, shading, and so on. The responsespectra of different opsins also over-lap to some extent, and there are var-ious post-photoreceptor neural pro-cessing mechanisms (that couldthemselves evolve adaptively). Evenflies can adapt to images by takingtheir context into account. We mightsay that color is in the mind ratherthan the eye of the beholder, andshould be careful speculating aboutwhat organisms can see (Figure 4).

JUST SO!

Kipling’s creatures got their traitsby chance and circumstance, andtheir own silly behavior. Biologists arekept busy writing our own storiesabout the Once Upon a Times. In thereal world, behaviorally driven selec-tion of local environments can lead toadaptation just as classical darwinismdoes, but where environment is se-lected by behavior to fit the genes,rather than the darwinian other wayround. Perhaps we should broadenour explanatory search space (Note 2).

Part of the problem is that there isno one mandate: not even all crea-tures in the same environment use vi-sion in the same way. Not all see thesame colors. Not all predators havesimilar vision. Humans are at risk inthe dark but can’t see in the dark. Pi-neal response does not require “eyes.”Even among mollusks there are di-verse types of eyes (Figure 1, left). Me-dian eyes may have evolved multipletimes as a separate visual sense in ar-thropods, but ocelli do not use pax6though they do use (their own) opsins.Some nematodes even use hemoglobinfor light perception (Burr et al., 2000).

Pax6 is highly conserved and impor-tant in eye-precursor tissue. It re-mains expressed in many tissues indeveloping eyes. Its enhancers arefound around many eye-related genes;

(Virtual) Figure 4. Believing is seeing. Seehttp://people.msoe.edu/�taylor/eisl/land.htm

Figure 3. Peak wavelength sensitivites of primate X-linked opsins.

218 Evolutionary Anthropology CROTCHETS & QUIDDITIES

but it is not the universal Monarch ofthe Sees atop all animal vision. At theother developmental end, opsins arealso conserved. But some of the linksin the evolutionary reconstructionshave been challenged. For example,planarians express, but do not re-quire, pax6 in producing eyes. In flies,pax6 regulates the rhodopsin rh1, andpax6 enhancer sequences are foundflanking most vertebrate opsin genes,but that use of pax6 is not necessaryfor eyes in an actual fly. Even the orig-inal eyeless story was not so categori-cal; the flies often had eyes (some-times small or imperfect), and eyescan reappear after some generationsbecause of other genes or pathways.Another gene, eyes absent, is expressedearly in development in flies and mu-tations variously affect either theocelli or the compound eyes.

In the face of this patchwork evolu-tion, Darwin’s queasiness about eyeevolution should sober our efforts toexplain our own favorite traits. Weproperly invoke the contingent natureof evolution in trying to explain adap-tive diversity: creatures evolve to usewhat is there to be used. How, if, orwhy natural selection in a particularsetting chose a particular option wasat the time it happened truly ad hoc,but in our time we must assess it posthoc. Not every possible explanationmay be equally plausible and variouskinds of tests may eliminate somecontenders. But because we can al-ways find a plausible adaptive expla-nation for any observation we maketoday, evolutionary reconstructionsreally are just-so stories in importantways. Trying to understand whenwe’ve got the true one is one of themost serious challenges in evolution-ary biology.

My title? Light by itself is useful, but

its pattern can be more So. Single-celled organisms already possessedbasic photoreception Very Long Ago.But though they could detect light,complex organisms (Note 3) had toevolve a brain so they could see. Nowthat is a pretty good Story!

NOTES

I would welcome any comments onthis column: kenweiss@psu.edu. Imaintain Crotchety Comments on myweb page: www.anthro.psu.edu/rsrch/weiss_lab.htm

I thank Anne Buchanan, DeanSnow, John Fleagle, and ChristopherHeesy for critical help on this manu-script, and W.A. Harris, T. Oakley, S.Yokoyama, W.-H. Li, J. Piatigorsky,and J. Neitz for technical discussion(they may not agree with what I say).

1. Do we tend inadvertently to thinklike Kipling? “[My purpose is to] elu-cidate how the coelacanth has modi-fied its visual pigments to adapt to itsnatural habitat.” Yokoyama, 1999.

2. Behavioral adaptation has longbeen considered under sometimes-controversial rubrics such as the Bald-win effect or sympatric speciation.

3. Well, even some single-celled al-gae have eyes that without a braininterpret light in regard (at least) tolocomotion.

TO READ

Many things discussed here can beprofitably explored by web searching.Boissinot S, Tan Y, Shyue S-K, Schneider H,Sampaio I, Neiswanger K, Hewett-Emmett D, LiW-H. 1998. Origins and antiquity of X-linkedtriallelic color vision systems in New World mon-keys. Proc Natl Acad Sci USA 95:13749–13754.Burr A, Hunt P, Wagar D, Dewilde S, Blaxter M,Vanfleteren J, Moens L. 2000. A hemoglobin withan optical function. J Biol Chem 275:4810–4815.Cvekl C, Piatigorsky J. 1996. Lens development

and crystalline gene expression: many roles forPax-6. BioEssays 18:621–630.

Dominy NJ, Lucas PW, Osorio D, Yamashita N.2001. The sensory ecology of primate food per-ception. Evol Anthropol 10:171–186.

Gehring W. 2002. The genetic control of eyedevelopment and its implications for the evo-lution of the various eye-types. Int J Dev Biol46:65–73.

Harris WA. 1997. Pax6: Where to be conserved isnot conservative. Proc Natl Acad Sci USA 94:2098–2100.

Heesy CP, Ross CF. 2001. Evolution of activitypatterns and chromatic vision in primates: Mor-phometrics, genetics and cladistics. J Hum Evol40:111–149.

Jacobs G. 1995. Variations in primate color vi-sion: mechanisms and utility. Evo Anthropol3:196–205.

Kipling R. 1902. Just So Stories. London: Pavil-lion Books.

Lockwood CA, Fleagle JG. 1999. The recogni-tion and evaluation of homoplasy in primateand human evolution. Ybk Phys Anthropol 42:189 –232.

Mollon JD. 1989. “Tho’ she kneel’d in that placewhere they grew . . . ”: the uses and origins ofprimate color vision. J Exp Biol 146:21–38.

Murray J. 1997. Mathematical biology, 2nd ed.Berlin: Springer Verlag.

Oakley T, Cunningham C. 2002. Molecular phy-logenetic evidence for the independent evolution-ary origin of an arthropod compound eye. ProcNatl Acad Sci USA 99:1426–1430.

Piatigorsky J. 1998a. Gene sharing in lens andcornea: facts and implications. Prog Retin EyeRes 17:145–174.

Piatigorsky J. 1998b. Multifunctional lens crys-tallins and corneal enzymes. More than meetsthe eye. Ann N Y Acad Sci 842:7–15.

Salvini-Plawen L., Mayr E. 1961. On the evolu-tion of photoreceptors and eyes. Evol Biol 10:207–263.

Tan Y, Li WH. 1999. Trichomatic vision in pros-imians. Nature 402:36.

Wagner GP. 1989. The biological homology con-cept. Ann Rev Ecol Syst 20:51–69.

Weiss KM. 2002. Is the message the medium?Biological traits and their regulation. Evol An-thropol 11:88–93.

Yokoyama S. 2000. Molecular evolution of verte-brate visual pigments. Prog Retin Eye Res 19:385–419.

Yokoyama S. 1999. Adaptive evolution of colorvision of the Comoran coelacanth (Latimeriachalumnae). Proc Natl Acad Sci USA 96:6279 –6284.

© 2002 Wiley-Liss, Inc.

CROTCHETS & QUIDDITIES Evolutionary Anthropology 219