classical Mendelian genetics and important results obtained by mathematical population geneticists (especially Ronald Fisher and Sewall Wright), Dobzhansky attempted to pro- vide a genetic analysis of evolutionary problems, most notably the problem of the origin of discontinuity in nature. In doing so, he helped to articulate the theoretical framework around which disparate biological disciplines could be unified.
Dobzhansky’s accomplishments were conceptual as much as empirical. Though he provided important answers to many of the questions he raised, the real novelty of his book lies in the way he framed those questions, drawing on the resources of Mendelian and population genetics. For present purposes, the most important conceptual division of the work was a threefold division of the science of genetics into transmission, devel- opmental, and population genetics (Dobzhansky 1982, 8–12). Dobzhansky placed this di- vision within a larger framework that is worth presenting briefly.
Genetics and the Origin of Species begins by establishing the central problem Dobzhan- sky wished to solve: “the nature and the origin of the discrete groups into which the liv- ing world is differentiated” (Dobzhansky 1982, p. 5).9 The diversity of life is not a contin- uous array; rather, it forms nested clusters. Dobzhansky’s goal was to provide a causal analysis of the origin of these clusters that could explain why they were clusters at all, rather than a continuous array. Dobzhansky differentiated between causal and
9 Treating this as the most pressing problem in evolutionary biology was not at all a trivial move on Dobzhansky’s part. Two decades earlier, Henry Fairfield Osborn (1916, p. 504) had argued that the problem of the origin of species had become “an incidental issue” and that the central problem was “the origin and history of single characters.” Dobzhan- sky’s position marks a complete reversal of this: the problem of the origin of individual
descriptive approaches to evolution. Descriptive disciplines such as systematics and mor- phology were tasked with illuminating the historical course of life (Dobzhansky 1982, p. 6). By contrast, genetics was an experimental, “nomothetic” science. A genetic study of evolution promised to provide “an analysis of the conflicting forces tending to increase or to level off the differences between organisms,” and in doing so would furnish an ex- planation of organic diversity (Dobzhansky 1982, p. 7).10
Each of the three types of genetic study had a role to play in this analysis. The first subdivision, transmission genetics, studied “the mechanisms of transmission of hereditary characteristics from parents to offspring, that is, on the architectonics of the germ plasm of the sex cells.” Transmission genetics offered a genetic theory of heredity, showing how “the character of organisms are determined by the genes carried in the sex cells” (Dobzhansky 1982, 9). By 1937, Dobzhansky could confidently state that “the genetics of the transmission of hereditary characters is, by and large, understood now” (Dobzhansky 1982, 10).
Dobzhansky explicitly admitted, however, that transmission genetics was limited. It did not explain “the mechanisms of gene action in development” (Dobzhansky 1982, 10). This formed the subject matter of developmental genetics, also called “physiological genet- ics” (Goldschmidt 1938). Transmission genetics established correlations between
10 It is worth noting that this distinction between causal and descriptive sciences, with morphology falling on the descriptive side, contributes to what Scott Gilbert (1998) has called the “supersessionist” rhetoric with which geneticists portrayed their science as the completion and perfection of the allegedly dead-end science of embryology. I do not
particular genetic mutations and particular phenotypic effects. Developmental genetics sought to explain the complex process by which the genes produced these effects. In stark contrast to his happy assessment of transmission genetics, Dobzhansky lamented that developmental genetic study had thus far furnished little by way of established knowledge and, even worse, that “no reliable methods have been devised for investiga- tions in this field” (Dobzhansky 1982, 10–11). This pessimism was widely shared (Morgan 1926, 290; Sinnott 1937, 62; Goldschmidt 1938, v–vi).
Both transmission and developmental genetics were “concerned with individuals as units” (Dobzhansky 1982, 11). By contrast, population genetics studied how the genetic composition of a population changes over time. Population genetics, as its name suggests, concerned itself with genetic phenomena at the population level, in particular the dy- namics of changes in allele frequencies. Mathematical models of these dynamics eluci- dated the forces (selection, drift, mutation, migration) that drove these changes, though population genetics was not, for Dobzhansky, limited to these models.
Relying on this threefold division of genetics, Dobhzansky (1982, 11–12) defined evo- lution as “a change in the genetic composition of populations,” with the result that the problems of evolution—in particular the problem of organic diversity—were population genetic problems. Dobzhansky (1982, 13) recognized three basic types of population ge- netic problem: (a) the problem of how mutations enter a population (mutation, gene flow); (b) the problem of how mutations spread or fail to spread once present in a population; and (c) the problem of how the diversity “attained on the preceding two levels” is fixed. This last level concerned itself with the processes that create barriers between populations,
barriers that cause two populations to proceed upon independent evolutionary trajecto- ries. Whereas the first two levels dealt with phenomena occurring within given popula- tions, the third level explained how those populations became units in the first place.
Such was the conceptual framework of Dobzhansky’s evolutionary theorizing. Noth- ing in this framework precludes the inclusion of developmental genetic discoveries into evolutionary theorizing. Dobzhansky conceived developmental genetics as an important but unfortunately immature part of the broader science of genetics. And it was genetics as a whole, not specifically transmission genetics or population genetics, that was to in- form evolutionary theorizing.
Lying in the background of Dobzhansky’s inclusive attitude toward developmental genetics is his conception of the split between heredity and development, which differed in essential features from Lillie’s.11 For Lillie, as we saw, there was an ontological split between genetics (heredity) and development. For Dobzhansky, in stark contrast, there was no such ontological split. Instead, there was a methodological split between transmission genetics and developmental genetics.12 They were different subfields of genetics, with differ- ent methods (which, by 1937, had reached markedly different stages of development). There was, however, no principled barrier to incorporating their findings into unified explanations, including explanations of evolutionary phenomena.
It is nonetheless true that Genetics and the Origin of Species incorporates very little de- velopmental genetics in its explanations of evolutionary phenomena. Dobzhansky him- self offered a perfectly innocuous reason for this: there was little to include. Later works of the synthesis did begin to include developmental genetic theories in their evolutionary theorizing. For instance, Julian Huxley (2010, 73–74) incorporated some of Goldschmidt’s developmental genetic conclusions in his explanation for why wild type characters are less modifiable by environmental changes than mutant characters. Similarly, Curt Stern (1963, 19–20), in his contribution to a volume edited by Mayr and Simpson, offered a speculative explanation of the evolution of sibling species in terms of the phenotypically silent remodeling of genetic reaction networks. G. Ledyard Stebbins, the foremost plant biologist of the synthesis, developed an entire research program in plant developmental genetics (Smocovitis 2008). Finally, it is worth noting that, after Britten and Davidson (1969) published their landmark paper introducing an early form of GRN theory, both Mayr (1970, 183) and Dobzhansky (1970, 413–14) cited the paper (in particular its evolu- tionary implications) sympathetically: Mayr with considerable enthusiasm, Dobzhansky rather more cautiously but still hopefully. For Mayr, understanding the structure of the “complex epigenetic systems” of organisms was important for understanding the amount and kind of genetic variability they could tolerate (Mayr 1969b, 316). Britten and Davidson’s proposal promised to help elucidate this problem. Mayr thus called for a de- velopmental genetic contribution to answering one of the core problems of the modern synthesis: the origin of discontinuous variation in nature.
The case of Richard Goldschmidt is similarly instructive. Like Dobzhansky, Gold- schmidt believed in the possibility of developmental genetics. Unlike Dobzhansky, Gold- schmidt actually worked out a sophisticated theory of developmental genetics, from which he drew substantial evolutionary conclusions (Goldschmidt 1938, 1982). These conclusions were profoundly at odds with the synthetic theory: Goldschmidt denied that macroevolution is compounded microevolution, denied that geographic races are incipi- ent species, and allowed only a reduced role for natural selection in driving evolutionary change. The architects of the synthesis responded, not by arguing that his work was ir- relevant, but rather by assimilating it where they could (as in the cases of Huxley and Stern, mentioned above), while denying the aspects of his theory that led to his problem- atic evolutionary conclusions (Davis, Dietrich, and Jacobs 2009).
What the case of Goldschmidt shows is that the proponents of the synthesis made quite substantial assumptions about what developmental genetics would look like when mature. Namely, they assumed that it would not have implications like those of Gold- schmidt’s developmental genetics. Were development truly irrelevant to the synthetic the- ory, such assumptions would be unnecessary.
As these examples show, the synthesis included from start what developmental ge- netic information it could. Thus we can see that the conceptual exclusion of development from the synthesis exists only from the developmentalist perspective that regards devel- opmental genetics with suspicion. From such a perspective, developmental genetics can only get at the most superficial part of development: realization. The major phenomena
In this sense, the developmentalists excluded themselves.13 From the vantage of the syn- thesis, which was committed to the relevance of genetics to all of development, there was not and could not be any such exclusion.
My analysis thus far shows that the conceptual exclusion of development from the synthesis, insofar as it was real, rested on an empirical question: how much of develop- ment can be explained genetically? I have shown that participants in the synthesis an- swered this question optimistically, and accordingly included developmental genetic findings in their evolutionary theorizing. But nothing I have presented thus far resolves Lillie’s powerful worries that the question is to be answered pessimistically: very little of development admits of a genetic explanation. Was Lillie right? We will see that neither of his arguments for the impossibility of developmental genetics succeeds. One was prob- lematic from the start (Section 2.5). The other raised a more pressing issue, but even this has recently been resolved (Section 2.7).