Evo-devo: evolutionary developmental biology, deep homology, and the genetic tool-kit

Intuition Beginner

Every animal starts as a single cell. As it develops, its genes turn on and off in patterns that build the body — head, eyes, limbs, organs. For most of the twentieth century, embryology (how a single cell becomes an organism) and evolution (how species change over time) were studied separately. "Evo-devo" — evolutionary developmental biology — brought them back together.

The surprise: the same genes control eye development in flies and humans. The fly's compound eye and the human's camera eye look completely different and were thought to have evolved independently. Yet a single conserved gene, called Pax6, sits at the top of the eye-building programme in both. This shared ancestry of the machinery — not of the eyes themselves — is what evo-devo calls "deep homology".

The conclusion, in the words of Sean Carroll: evolution works mostly by changing where and when a small set of conserved master genes is switched on, not by inventing new genes. This conserved set is the "genetic tool-kit". Evo-devo reframes the modern synthesis (the 1930s-1940s fusion of Darwin with genetics): development matters for evolution, because the developmental system biases which forms evolution can reach.

Visual Beginner

The diagram shows the genetic tool-kit. On the left, the Hox gene clusters of Drosophila and the four vertebrate Hox clusters (HoxA–D), colour-coded by anterior–posterior position: genes at one end of the cluster pattern the head end of the embryo, genes at the other end pattern the tail. The order of genes on the chromosome matches the order of body regions they specify — collinearity, established by Akam in 1987. On the right, four master regulators and the structures they pattern across Bilateria: Pax6 for eyes, Nkx2-5 (Tinman in the fly) for hearts, Distal-less for limbs, Hedgehog for segment polarity and signalling. The arrow at the bottom reads "deep homology" — conserved genes deployed to build diverse structures.

The picture captures what evo-devo adds to the modern synthesis: a finite, conserved tool-kit, deployed differently across the animal kingdom, generates the morphological diversity that selection then sorts.

Worked example Beginner

In 1994, Walter Gehring and colleagues at the University of Basel ran a now-classic experiment on the Pax6 gene. Step 1: they took the mouse Pax6 gene and artificially activated it in cells of a Drosophila (fruit fly) larva that normally become a leg.

Step 2: the fly grew an eye on its leg. Not a mouse eye — a complete compound fly eye, with lens, photoreceptors, and the wiring to the fly's nervous system. The mouse gene had directed the construction of a fly eye.

Step 3: the symmetric experiment works in reverse. The fly's version of Pax6 (called eyeless), when artificially activated in mouse tissue, can induce mouse-eye-like structures. The instruction "build an eye" is conserved across roughly 600 million years of independent evolution.

What this tells us: morphological evolution is largely about changing where and when master genes are switched on, not about inventing new genes. The eye, the heart, the limb — each is built by a conserved master regulator that evolution redeploys.

Check your understanding Beginner

Formal definition Intermediate+

The evo-devo synthesis unifies evolutionary biology (selection + genetics) with developmental biology (embryology). The modern synthesis of the 1930s–1940s (Fisher, Wright, Haldane, Dobzhansky, Mayr) largely ignored development: development entered only as a black box producing the phenotypes on which selection acts. Evo-devo restores development to evolutionary theory. Its core notions are the genetic tool-kit, deep homology, the regulatory hypothesis, and modularity.

Definition (the genetic tool-kit). A small set of conserved master-regulatory genes — the Hox cluster (anterior–posterior body-plan patterning), Pax6 (eye), Nkx2-5/Tinman (heart), Distal-less/Dlx (limbs), and the BMP, Wnt, Hedgehog, and Notch signalling pathways — that development deploys to build animal body plans. Tool-kit genes are conserved across Bilateria. Their protein-coding sequences change slowly across hundreds of millions of years of independent evolution.

Definition (deep homology, Carroll 2005). Two structures in different lineages are deeply homologous when the same tool-kit gene — or, more precisely, the same regulatory network of tool-kit genes — patterns both, even when the adult structures look different and were thought on anatomical grounds to have evolved independently. The compound eye of Drosophila and the camera eye of vertebrates are deeply homologous via Pax6.

Definition (the regulatory hypothesis, King-Wilson 1975). Morphological evolution is driven primarily by changes in gene regulation — when, where, and at what level each gene is expressed — rather than by changes in the protein-coding sequences of the genes themselves. The supporting argument: humans and chimpanzees share roughly 98.7% of their DNA sequence identity, yet differ markedly in morphology and cognition; the gap is held to be regulatory.

Definition (modularity and evolvability). Animal body plans are modular: limbs, eyes, segments, and organs can change semi-independently of one another. Cis-regulatory elements (enhancers) control the spatial and temporal deployment of tool-kit genes in each module. Modularity makes a developmental system evolvable: a lineage can generate heritable phenotypic variation in one module without disrupting the rest of the organism.

Counterexamples to common slips Intermediate+

• Hox genes determine body plan. Overstated. Hox genes pattern along the anterior–posterior axis; other tool-kit genes pattern other axes (dorsal–ventral, left–right) and organs (Pax6 for eyes, Nkx2-5 for hearts). The Hox cluster is one component of the tool-kit, not the whole of it.

• Pax6 = eye. Oversimplified. Pax6 is necessary but not sufficient: loss of Pax6 blocks eye development, but eye formation requires a downstream network of dozens of genes that Pax6 helps switch on. Pax6 sits near the top of a regulatory cascade, but the cascade itself — not Pax6 alone — is the eye-building machinery.

• Evo-devo refutes Darwin. It does not. Evo-devo refines Darwinism by supplying developmental mechanisms for the generation of the variation on which selection acts. Selection still sorts the variation; the modern synthesis is not false, it is (on the evo-devo view) incomplete.

• Haeckel was right. No. "Ontogeny recapitulates phylogeny" (Haeckel 1866) is rejected. Von Baer's 1828 alternative is correct: embryos specialise from general to particular forms; they do not pass through adult stages of their ancestors.

• Deep homology means common descent of eyes. Contested but the consensus favours common descent. Pax6's common role in eye development across Bilateria could reflect common descent of eyes from a single ancestral eye OR convergent recruitment of Pax6 into independently evolved eyes. Most researchers now favour common descent (Gehring 1994; Carroll 2005), though the question is not fully closed.

• The genetic tool-kit is fixed. It is not. Tool-kit genes themselves evolve, just slowly. New tool-kit components arise over geological time via gene duplication and divergence; the pax, hox, and nkx families are themselves products of ancient duplications.

• Cis-regulatory mutations are the sole driver of morphological evolution. Carroll (2005) argues that cis-regulatory changes dominate; Hoekstra and Coyne (2007) and others argue for a balance between coding and regulatory changes. The empirical question is open, and the answer is likely to vary across traits and lineages.

Key argument: deep homology requires revising the modern synthesis Intermediate+

Argument.

1. Premise 1 (the modern synthesis treats evolution as allele-frequency change). On the standard modern-synthesis view, evolution is the change in allele frequencies in a population over generations, driven by mutation, selection, drift, and migration. Development enters only as the generator of the phenotype — treated as a black box that turns genotypes into phenotypes, with the details relegated to "embryology".

2. Premise 2 (deep homology is real). The same conserved master-regulatory genes — Hox, Pax6, Nkx2-5, Distal-less, and the BMP/Wnt/Hedgehog/Notch pathways — pattern homologous body parts across phyla separated by hundreds of millions of years of independent evolution ^{[Gehring1994]}; ^{[Carroll2005]}. The compound eye of Drosophila and the camera eye of vertebrates share Pax6.

3. Premise 3 (the accessible phenotype space is constrained by the developmental system). Because the tool-kit is finite and deployed through cis-regulatory elements that bind a limited set of transcription factors, only some phenotypic variations are developmentally reachable in a single mutational step. The developmental system is not a black box: it actively biases which phenotypes selection can act on.

4. Conclusion. The modern synthesis requires revision. The textbook picture (mutation → selection → adapted phenotype) must be augmented with a developmental component (developmental tool-kit → biased generation of phenotypic variation → selection). This is the thesis of the "extended synthesis" ^{[PigliucciMuller2010]}.

Reconstruction. The argument turns on whether developmental constraints are scientifically inert (just background conditions, already implicit in the modern synthesis) or scientifically productive (new explanations that the modern synthesis cannot supply). The defenders of revision — Carroll, Raff, Davidson, Pigliucci, Müller — argue that evo-devo supplies novel explanatory resources: constraint, modularity, evolvability, and the regulatory hypothesis, none of which reduces cleanly to mutation-selection-drift. The defenders of the modern synthesis — Futuyma, Wray — hold that developmental constraints are already accommodated as "constraints on selection", and that evo-devo adds mechanism but does not revise structure. The substantive philosophical question is whether evo-devo's additions change the structure of evolutionary theory or merely fill in its mechanistic details. On the revisionary reading, the developmental system is a cause of evolutionary direction; on the conservative reading, it is the substrate on which standard causes act.

Bridge. The deep-homology argument builds toward 20.05.02 the unit-of-selection debate, where the question "what does selection act on?" receives a parallel multi-level treatment — the developmental tool-kit is part of the structure that bears selection, not just the substrate on which selection acts. The foundational reason evo-devo is philosophically load-bearing is that it identifies the developmental system as a cause of evolutionary direction, and this is exactly the bridge from the population-thinking of the modern synthesis to the developmental-systems thinking of the extended synthesis. The central insight generalises across 18.11.01 embryology and morphogenesis, where the tool-kit's molecular mechanics are worked out in detail, and the regulatory hypothesis appears again in 19.01.01 Mendelian genetics as the modern-synthesis foundation that the evo-devo revision refines. Putting these together identifies the accessible phenotype with the output of a constrained developmental system, and the bridge is from the molecular-mechanics discoveries of the 1980s–2000s to the conceptual revision that Pigliucci and Müller crystallise.

Exercises Intermediate+

Interpretive debates Master

Debate 1 (Haeckel 1866 versus von Baer 1828: the embryology framework). Ernst Haeckel's Generelle Morphologie der Organismen (1866) defended the biogenetic law — "ontogeny recapitulates phylogeny" — holding that embryos pass through the adult stages of their evolutionary ancestors ^{[Haeckel1866]}. Karl Ernst von Baer's Über Entwicklungsgeschichte der Thiere (1828) had already given the alternative ^{[VonBaer1828]}: embryos pass from general forms to specialised forms, never through adult stages of other animals. Von Baer's view is correct; Haeckel's is rejected in its literal form. Modern evo-devo retains the kernel of Haeckel's insight — development carries evolutionary history — while rejecting the literal recapitulation claim.

Debate 2 (Lewis 1978: the bithorax complex and Hox genes). Edward B. Lewis's 1978 paper in Nature identified the bithorax complex in Drosophila — a cluster of genes controlling segment identity in the fly's posterior half ^[Lewis1978]. Loss-of-function mutations transform segments into the form of segments ahead of them: a third thoracic segment becomes a second, producing a fly with four wings instead of two. The discovery opened the molecular genetics of body plans and won Lewis a share of the 1995 Nobel Prize.

Debate 3 (Nüsslein-Volhard–Wieschaus 1980: the Heidelberg screen). The saturation-mutagenesis screen of Drosophila embryos by Christiane Nüsslein-Volhard and Eric Wieschaus systematically identified every gene required to establish the larval body plan ^[NVW1980]. Roughly 120 genes set up the embryo's axes, segments, and tissue types. The screen (Heidelberg, 1979–1980) shared the 1995 Nobel with Lewis and inaugurated the systematic molecular-genetic dissection of development.

Debate 4 (McGinnis–Garber–Wieland 1984: the homeobox; Akam 1987: collinearity). The homeobox is a 180-base-pair DNA sequence shared by Drosophila homeotic genes and their vertebrate orthologues, discovered by McGinnis and colleagues in 1984 ^{[McGinnis1984]}. The homeobox encodes a protein domain (the homeodomain) that binds DNA and regulates the transcription of downstream genes. Its discovery was the first molecular evidence that the same machinery patterns bodies across phyla. Michael Akam's 1987 review in Development established collinear Hox expression: the order of Hox genes on the chromosome matches the anterior–posterior order of body regions they pattern, in both insects and vertebrates ^[Akam1987].

Debate 5 (King–Wilson 1975: the regulatory hypothesis). Mary-Claire King and Allan C. Wilson's 1975 paper in Science argued that morphological evolution is driven primarily by changes in gene regulation, not by changes in protein-coding sequence ^{[KingWilson1975]}. The evidence: humans and chimpanzees are approximately 98.7% sequence-identical in alignable regions yet morphologically and cognitively divergent. The hypothesis predates the discovery of the tool-kit but is its conceptual foundation — without a regulatory account, deep homology would be a comparative curiosity rather than an evolutionary mechanism.

Debate 6 (Gehring 1994: Pax6 as master regulator). Walter J. Gehring and colleagues showed that ectopic expression of mouse Pax6 in a Drosophila leg directs the formation of a compound fly eye ^{[Gehring1994]}. The Pax6 protein is so conserved across Bilateria that a mammalian version can drive insect eye development, and conversely the fly orthologue (eyeless) can induce mouse-eye-like tissue. The result is the strongest single piece of molecular evidence for deep homology of the eye — and by extension for the deep-homology thesis in general.

Debate 7 (Carroll 1995+, Davidson 2001: the deep-homology thesis and the regulatory network). Sean B. Carroll's work on butterfly wing-pattern eyespots (1995+) and his 2005 textbook (with Grenier and Weatherbee) generalised the evo-devo programme into the deep-homology thesis ^{[Carroll2005]}: morphological evolution acts predominantly on cis-regulatory enhancers that control where and when tool-kit genes are expressed, rather than on the protein-coding regions of those genes. Eric H. Davidson's Genomic Regulatory Systems (2001) ^{[Davidson2001]} formalised the regulatory network as the unit of developmental control, with cis-regulatory elements reading off combinations of transcription factors to specify time and place of gene expression. Together, Carroll and Davidson turned evo-devo from a comparative programme into a mechanistic one.

Debate 8 (Pigliucci–Müller 2010: the extended synthesis). Massimo Pigliucci and Gerd B. Müller's edited volume Evolution: The Extended Synthesis ^{[PigliucciMuller2010]} argues that evo-devo, phenotypic plasticity (West-Eberhard 2003), niche construction, and epigenetic inheritance jointly require a structural revision of the modern synthesis. The volume's defenders hold that the modern synthesis, formulated in the 1930s–1940s before the molecular-genetic revolution, treats development as a black box and is therefore conceptually incomplete. Critics — Futuyma, Wray — reply that the modern synthesis can absorb these additions as mechanistic elaborations without structural revision. The debate continues as the central methodological dispute in contemporary philosophy of evolutionary biology.

Synthesis. The evo-devo programme builds toward 20.05.02 the unit-of-selection debate by adding a developmental axis to the question of what selection acts on — the developmental tool-kit is part of the structure that bears selection, not just the substrate on which selection acts. The central insight is that the genetic tool-kit is a finite, conserved resource that development deploys modularly, and this is exactly what makes evolution evolvable: the same gene can be re-deployed in a new module without disrupting its old role. The foundational reason evo-devo is philosophically load-bearing is that it identifies the developmental system as a cause of evolutionary direction. Putting these together, the regulatory hypothesis (King-Wilson 1975), deep homology (Gehring 1994; Carroll 2005), and modularity jointly reframe the modern synthesis: morphological evolution is the redeployment of conserved machinery through regulatory change, not the generation of new machinery by coding mutation. The pattern generalises: each layer of biological organisation — gene, cis-regulatory element, cell, embryo, population — has its own characteristic causes, and identifying the developmental layer as causally real identifies evo-devo with a research programme distinct from both developmental biology and population genetics. The bridge is from the molecular discoveries of 1978–2005 (bithorax, the homeobox, the Heidelberg screen, Pax6) to the conceptual revision of Pigliucci-Müller 2010, and the pattern recurs as each newly-characterised tool-kit gene supplies a fresh case study in deep homology.

Full argument set Master

Argument 1 (the developmental-constraint argument against the allele-frequency reduction).

Premises. (i) The set $P_{\mathrm{acc}}$ of phenotypes developmentally accessible to a population in a single generation is determined by the structure of the developmental system: the tool-kit, the cis-regulatory architecture, the modular decompositions of body plans. (ii) Selection sorts among phenotypes in $P_{\mathrm{acc}}$ but does not itself generate them; the developmental system does. (iii) The structure of the developmental system is itself the product of evolutionary history and varies in a constrained, lineage-specific way.

Argument. The direction of phenotypic evolution is jointly determined by selection (which sorts) and by development (which generates the sort space). A mutation that would be beneficial if its phenotype were accessible is unavailable if the developmental system cannot produce the phenotype in a single mutational step. Therefore the textbook identity "evolution = change in allele frequencies under selection, drift, and migration" is incomplete: it omits the developmental shaping of $P_{\mathrm{acc}}$.

Conclusion. The revised identity is "evolution = change in allele frequencies under selection, drift, migration, and developmental constraint". The developmental system is a cause of evolutionary direction, not merely the substrate on which other causes act. $\square$

Argument 2 (the chimp–human regulatory gap as existence proof).

Premises. (i) Humans (Homo sapiens) and chimpanzees (Pan troglodytes) share approximately 98.7% of their DNA sequence identity in alignable regions, including the protein-coding regions of nearly all transcription factors. (ii) Humans and chimpanzees differ markedly in morphology — brain size (roughly threefold), limb proportions, cranial structure, dentition — and cognition. (iii) The number of protein-altering substitutions that distinguish the two lineages is too small, by itself, to account for the phenotypic gap under any plausible additive model of effect size.

Argument. If morphological divergence were driven primarily by protein-coding sequence divergence, the 1.3% sequence divergence between humans and chimps would have to underwrite a many-fold phenotypic divergence. The arithmetic is implausible on its face: a few thousand protein-changing substitutions, mostly of small effect, do not add up to a threefold brain-size difference without some amplifying mechanism. If, by contrast, morphological divergence is driven primarily by regulatory divergence — changes in when, where, and how much each gene is expressed — the 1.3% sequence divergence can include large regulatory changes in enhancers, promoters, and trans-acting factors, with phenotypic effects far exceeding their sequence length.

Conclusion. The chimp–human gap is an existence proof for the King-Wilson regulatory hypothesis: regulatory divergence can carry phenotypic divergence that protein-coding divergence cannot. The hypothesis does not deny that protein changes matter — only that they are the dominant driver of morphological evolution. $\square$

Connections Master

• The unit of selection 20.05.02. The unit-of-selection debate asks "what does selection act on?"; evo-devo asks "what does selection have to work with?" — the developmental tool-kit biases which phenotypes are accessible, just as population structure biases which levels selection can act on. The two debates are mutually reinforcing: the multi-level-selection framework needs a developmental account of what generates each level's heritable variation, and evo-devo needs a population-genetic account of how that variation is sorted.

• Mendelian genetics — segregation and dominance 19.01.01. Mendelian genetics is the modern-synthesis foundation that the evo-devo revision refines. The King-Wilson regulatory hypothesis does not replace Mendelian inheritance; it adds a layer: the same alleles, inherited by Mendelian rules, have phenotypic effects that depend on the developmental and regulatory context in which they are expressed. Deep homology is therefore consistent with Mendel — it identifies the cis-regulatory architecture that mediates how Mendelian alleles generate phenotypes.

• Embryology and morphogenesis 18.11.01. Embryology and morphogenesis is the developmental-biology side of evo-devo. The tool-kit's molecular mechanics — Hox collinear expression, Pax6 ectopic induction, Spemann's 1924 organiser experiment — are worked out in detail in developmental biology; evo-devo carries these mechanisms into evolutionary time, asking how the same machinery has been redeployed across hundreds of millions of years of divergence.

Historical & philosophical context Master

The nineteenth-century embryology debate set the agenda. Ernst Haeckel's Generelle Morphologie der Organismen (1866) defended the biogenetic law — "ontogeny recapitulates phylogeny" ^{[Haeckel1866]} — holding that embryos pass through the adult forms of their evolutionary ancestors. Karl Ernst von Baer's Über Entwicklungsgeschichte der Thiere (1828) had already given the alternative ^{[VonBaer1828]}: embryos pass from general to specialised forms, never through the adult stages of other animals. Von Baer's view is correct; Haeckel's literal claim is rejected, though the kernel — that development carries evolutionary history — survives in modern evo-devo.

The early twentieth century split embryology from genetics. Thomas Hunt Morgan's Drosophila work established the Mendelian-chromosomal theory of inheritance (Morgan, Sturtevant, Muller, and Bridges, 1915 onwards) but set embryology aside; Hans Spemann's organiser experiment (Spemann and Mangold 1924) founded developmental mechanics but did not connect to genetics. The modern synthesis of the 1930s–1940s (Fisher, Wright, Haldane, Dobzhansky, Mayr) unified Darwinism with Mendelism but largely excluded development, on the grounds that the genetics of the time could not address it. For nearly forty years, the two fields proceeded separately.

The reunion began in the 1970s. Edward B. Lewis's 1978 Nature paper on the bithorax complex ^[Lewis1978] identified Hox genes and their collinear arrangement; the Nüsslein-Volhard–Wieschaus Heidelberg screen ^[NVW1980] systematically catalogued the pattern-control genes (1995 Nobel Prize, shared with Lewis); McGinnis, Garber, Wieland, and Gehring's 1984 Cell paper ^{[McGinnis1984]} discovered the homeobox shared across insects and vertebrates; Akam's 1987 Development review ^[Akam1987] established collinear Hox expression as a general feature of Bilateria. Mary-Claire King and Allan C. Wilson's 1975 Science paper ^{[KingWilson1975]} had supplied the regulatory hypothesis a decade earlier, identifying gene regulation as the principal substrate of morphological evolution. Walter J. Gehring's 1994 Cell paper on Pax6 ^{[Gehring1994]} demonstrated deep homology at the molecular level: the same master regulator patterns eye development across 600 million years of divergence.

Sean B. Carroll's work on butterfly wing-pattern eyespots (1995 onwards) and his 2005 textbook From DNA to Diversity (with Grenier and Weatherbee) ^{[Carroll2005]} generalised the programme into the deep-homology thesis. Eric H. Davidson's Genomic Regulatory Systems (2001) ^{[Davidson2001]} formalised the regulatory-network framework; Rudolf A. Raff's The Shape of Life (1996) gave the field its canonical textbook; Mary Jane West-Eberhard's Developmental Plasticity and Evolution (2003) extended the framework to phenotypic plasticity. Pigliucci and Müller's edited volume Evolution: The Extended Synthesis (2010) ^{[PigliucciMuller2010]} crystallised the philosophical revision, arguing that evo-devo, plasticity, niche construction, and epigenetic inheritance jointly require a structural revision of the modern synthesis — a claim that remains the central methodological dispute in the philosophy of evolutionary biology.

Bibliography Master

Akam, Michael. "The Molecular Basis for Metameric Pattern in the Drosophila Embryo." Development 101, no. 1 (1987): 1–22.

Carroll, Sean B. Endless Forms Most Beautiful: The New Science of Evo Devo and the Making of the Animal Kingdom. New York: W. W. Norton, 2005.

Carroll, Sean B., Jennifer K. Grenier, and Scott D. Weatherbee. From DNA to Diversity: Molecular Genetics and the Evolution of Animal Design. 2nd ed. Malden, MA: Blackwell Science, 2005.

Davidson, Eric H. Genomic Regulatory Systems: Development and Evolution. San Diego, CA: Academic Press, 2001.

Gehring, Walter J. "Exploring the Homeobox." Cell 78, no. 2 (1994): 191–201.

Haeckel, Ernst. Generelle Morphologie der Organismen: Allgemeine Grundzüge der organischen Formen-Wissenschaft. 2 vols. Berlin: Georg Reimer, 1866.

King, Mary-Claire, and Allan C. Wilson. "Evolution at Two Levels in Humans and Chimpanzees." Science 188, no. 4184 (1975): 107–116.

Lewis, Edward B. "A Gene Complex Controlling Segmentation in Drosophila." Nature 276, no. 5688 (1978): 565–570.

McGinnis, William, Robert L. Garber, Jan Wirz, Akihiro Kuroiwa, and Walter J. Gehring. "A Homologous Protein-Coding Sequence in Drosophila Homeotic Genes and Its Conservation in Metazoans." Cell 37, no. 2 (1984): 403–408.

Nüsslein-Volhard, Christiane, and Eric Wieschaus. "Mutations Affecting Segment Number and Polarity in Drosophila." Nature 287, no. 5785 (1980): 795–801.

Pigliucci, Massimo, and Gerd B. Müller, eds. Evolution: The Extended Synthesis. Cambridge, MA: MIT Press, 2010.

Raff, Rudolf A. The Shape of Life: Genes, Development, and the Evolution of Animal Form. Chicago: University of Chicago Press, 1996.

von Baer, Karl Ernst. Über Entwicklungsgeschichte der Thiere: Beobachtung und Reflexion. Königsberg: Bornträger, 1828.

West-Eberhard, Mary Jane. Developmental Plasticity and Evolution. Oxford: Oxford University Press, 2003.