Journal of the History of Biology  Springer Science+Business Media Dordrecht 2015 DOI 10.1007/s10739-015-9400-0

Systems Thinking Versus Population Thinking: Genotype Integration and Chromosomal Organization 1930s–1950s

EHUD LAMM The Cohn Institute for the History and Philosophy of Sciences and Ideas Tel Aviv University 69978 Tel Aviv Israel E-mail: [email protected]

Abstract. This article describes how empirical discoveries in the 1930s–1950s regarding population variation for chromosomal inversions affected Theodosius Dobzhansky and Richard Goldschmidt. A significant fraction of the empirical work I discuss was done by Dobzhansky and his coworkers; Goldschmidt was an astute interpreter, with strong and unusual commitments. I argue that both belong to a mechanistic tradition in genetics, concerned with the effects of chromosomal organization and systems on the inheritance patterns of species. Their different trajectories illustrate how scientists’ commitments affect how they interpret new evidence and adjust to it. Dobzhansky was moved to revised views about selection, while Goldschmidt moved his attention to different genetic phenomena. However different, there are significant connections between the two that enrich our understanding of their views. I focus on two: the role of developmental considerations in Dobzhansky’s thought and the role of neutrality and drift in Goldschmidt’s evolutionary account. Dobzhansky’s struggle with chromosomal variation is not solely about competing schools of thought within the selectionist camp, as insightfully articulated by John Beatty, but also a story of competition between selectionist thinking and developmental perspectives. In contraposition, Goldschmidt emphasized the role of low penetrance mutations that spread neutrally and pointed out that drift could result from developmental canalization. This account adds to the dominant story about Goldschmidt’s resistance to the splitting of development from genetics, as told by Garland Allen and Michael Dietrich. The story I tell illustrates how developmental thinking and genetic thinking conflicted and influenced researchers with different convictions about the significance of chromosomal organization. Keywords: Chromosomal inversions, Coadaptation, Heterosis, Systemic mutations, Levels of selection, Evolution of Genetic Systems, Neutral evolution

Introduction Chromosomal inversions were a highly studied topic in the 1930s and 1940s that held implications for many evolutionary and genetic ques-

EHUD LAMM

tions. Among them were: selection versus drift, types of selection (e.g., stabilizing versus directional), adaptation, variation and polymorphisms in natural populations, heterosis, the mapping of genes to chromosomes and to chromosomal locations, chromosomal dynamics, gene action, the structure and chemical basis of genes, and the physiological effects (if any) of chromosomal organization. In this article I contrast the ways in which Theodosius Dobzhansky and Richard Goldschmidt adjusted their views to incorporate new evidence on inversions. Dobzhansky and his coworkers provided increasingly fine grained understanding of the fitness effects of inversions; however, the interpretation of the evidence was not straightforward. Both Dobzhansky and Goldschmidt paid attention to the developmental implications of genetic organization and were on the lookout for chromosomal systems that transcended genes. However, Dobzhansky’s views focused on population dynamics whereas Goldschmidt’s guiding question throughout was the chromosome and its role in the development of the individual organism. Feeling different pressures, the two geneticists, who shared many interests, assimilated the new discoveries in radically different ways. Seeing these developments through the eyes of Goldschmidt offers a new perspective on an important period in the development of population genetics. I will also argue that both Dobzhansky and Goldschmidt espoused a perspective on genetics that can be termed ‘‘systems thinking’’ and that was gradually marginalized. The period of the evolutionary synthesis has been studied extensively by historians (see Smocovitis, 1996). Major controversies that involve both Dobzhansky and Goldschmidt separately have been studied extensively (Beatty, 1987a, b, 1994; Crow, 1987; Dietrich, 2000a, 2008, 2011). So is the fate of development during the evolutionary synthesis, and in particular the role of development in the views of both (Davis et al. 2009; Dietrich, 2000b; Richmond, 2007; Kohler, 1994). Among the factors that that been analyzed are struggles for authority (Sapp, 1987; Davis et al., 2009) and boundary crossing (Kohler, 1994). These, together with social/political implications, that Beatty in particular emphasized in his work on Dobzhansky, are among the driving forces that have been considered in the literature concerning the work I discuss. My focus here is on how a particular series of studies that analyzed chromosomal inversions affected how Goldschmidt and Dobzhansky conceived the organization of the chromosome and the role, if any, of intrachromosomal systems. These issues touch both on the relation between development and evolution and on the gene concept but are distinct from them. Identifying commonalities between Goldschmidt and Dobzhansky

SYSTEMS THINKING VERSUS POPULATION THINKING

is no less instructive than elucidating the controversies they took part in. It allows me to explore the significance of a strand of thought that is not emphasized by synoptic historical accounts (see Smocovitis, 1996, p. 13) but that was a backdrop to major developments in the period. Heterozygosity for chromosomal inversions was considered as a case of hybrid vigor or heterosis. The original observation of hybrid vigor, that is that hybrids often exhibit increased size or strength, is lost in time, and was already noted by both Darwin and Mendel. Two competing explanations for heterosis are the dominance view according to which dominant alleles from one parent mask inferior alleles from the other and the overdominance view according to which hybrids have intrinsic advantages, emanating from traits of the two parental strains or from their interaction (see Crow, 1987). Once heterosis due to heterozygosity for inversions was observed they were used to try and adjudicate between these two explanations. Since 1955, this debate coincided with the Muller–Dobzhansky classical-balance controversy which revolved around the possible value of variation in populations (Beatty, 1987b, 1994; Crow, 1987). At first, though, inversion were thought to be selectively neutral, that is not to directly affect fitness. Under this assumption it was a conundrum why natural populations were polymorphic for specific inversions, in stable frequencies, when this fact was established in the 1930s. Dobzhansky and his colleagues have identified some fifteen different inversions in natural populations of D. pseudoobscura each with a large though limited geographical distribution. Two or three different inversions were typically represented in each locality.1 Various explanations for these observations were discussed and analyzed. Eventually, in the mid 1940s it became evident that different inversions can have different fitness, and that a particular inversion may be more suited to a particular environment, for example to a particular season, than another. Dobzhansky’s results in the mid 1940s indicated that individuals in a population that are heterozygous for naturally occurring inversions generally had a fitness advantage (heterosis) over individuals in the population homozygous for the inversions (Wright and Dobzhansky, 1946). This meant that the hypothesis that polymorphism for inversions was the result of selection favoring a specific set of optimal genes in each environment had to be reconsidered (see Wallace, 1994). More fundamentally, however, the discovery that inversions affect fitness held implications for the conceptualization of the genome. Roughly, the 1

Sturtevant to Wright, March 18, 1936. Sewall Wright Papers, American Philosophical Society.

EHUD LAMM

situation can be described as follows. Inversions were clearly genomic variations of some size. If they were neutral, this level of organization was presumably causally inert in development; if non-neutral, the converse. Depending on the relationship between genes and chromosomal inversions, which can extend beyond single genes, the conclusions about the causal power of inversions held implications for the conceptualization of the role of genes in development and to the question whether the chromosome had a segmental structure.2 Richard Goldschmidt is a particularly interesting case here. He held strong and radical views about all these issues. He rejected the notion of atomic genes and tried to develop a chromosomal perspective that was at once, to use his terms, hierarchical and dynamic (i.e., developmental). His evolutionary account, which he originally supported with data on inversions, assumed their spread was not generally the result of selection yet that they could lead the way to macro-mutations. Goldschmidt was aware of the research done by Dobzhansky and his coworkers and by other groups on the fitness effects of changes in chromosomal organization. Understanding how he justified his non-mainstream views, and how he differed from others who were preoccupied with the same issues, helps us clarify the conceptual landscape: what was considered as understood, what questions remained, and what questions researchers were actively trying to answer. A contrasting case is that of Theodosius Dobzhansky himself who modified his views on the origin and role of chromosomal polymorphisms in natural populations during the same period. The trajectories of Goldschmidt and Dobzhansky are worth comparing and illustrate how scientists’ high-level commitments, possibly even their explanatory styles, affect how they interpret new evidence and adjust to it. In particular, both Dobzhansky and Goldschmidt appealed to developmental considerations; however, as Beatty (1987a, b, 1994) showed, Dobzhansky was mostly concerned with establishing a role for variation in populations and the series of explanations for the data on inversions that he proposed over the years focused on different kinds of selection, attempting to identify a form of selection that was variation-preserving. This concern is most starkly manifested in his debate with Muller concerning the role of variation, the so called classical-balance controversy (Beatty, 1987a, b, 1994; Crow, 1987). This debate, however, happened after 1955 for the most part whereas I am mostly interested in his views many years earlier. Dobzhansky was influenced by the Russian school of population genetics originating with the work of Sergei 2

Muller to Goldschmidt, June 19, 1939. H. J. Muller Papers, Indiana University.

SYSTEMS THINKING VERSUS POPULATION THINKING

Sergeevich Chetverikov (1880–1959; see Kohler 1994, p. 269). In an influential 1926 paper, Chetverikov noted that recessive mutations necessarily lead to cryptic variability, i.e. genetic variability that is not visible phenotypically and hence is hidden from natural selection (Chetverikov, 1961[1926]). Equilibrium frequencies of homozygotes and heterozygotes are established in one generation of free interbreeding, or ‘‘stabilizing crossing.’’ Chetverikov set out to confirm this theoretical result experimentally, an endeavor that continued with the work of Dobzhansky and Nikolay Petrovich Dubinin on inversions in Drosophila. The architecture of the genetic system has direct implications for cryptic variability that presumably should influence the evolution of genetic systems. Chetverikov devoted a large part of his 1926 paper to reflecting on the implications for natural selection of cryptic variation and genotypic complexity more generally. He noted an ongoing bitter dispute concerning the creative force of natural selection. Was natural selection merely a sieve disposing of less fit genes or was it able to direct variability, ‘‘actively entering into the evolutionary process’’ (Chetverikov, 1961, p. 189)? Accepting as definitive the then recent results indicating the constancy of the gene, Chetverikov emphasized the importance of the interactions between the genes of the organism, all of which together form what he called ‘‘the genotypic milieu.’’ Since the interactions are constitutive properties of the genes, they are themselves hereditary. Natural selection can thus create gene combinations that work well together. It is in this way, Chetverikov thought, that natural selection ‘‘actively participates in the evolutionary process.’’ Dobzhansky did not turn his attention to population genetics until 1935, but when he did his work was within this tradition, even if it was more quantitative than the work done in Russia (Provine, 1981, pp. 37, 75). His analysis of inversions during most of this period emphasized the coadaptation of gene complexes and gene arrangements in homologous chromosomes (see Beatty, 1987a, b, 1994; Wallace, 1994). Goldschmidt felt different pressures than Dobzhansky: the burgeoning theory of the gene, which he felt was deeply flawed in its conception of atomic genes and genic action, on the one hand, and his rejection of non-hierarchical and non-developmental views of the genome and its function in development, on the other. Goldschmidt and Dobzhansky differed on multiple dimensions. Goldschmidt rejected the Synthesis, and stressed that macroevolution was not continuous with microevolution; he thought population genetics had to be complemented with physiological genetics, that is with developmental considerations, and that the latter were the more evolutionary significant; and

EHUD LAMM

he rejected the discrete, atomic, gene. Dobzhansky’s commitments were the opposite on all these dimensions. The rest of this article is organized as follows. I begin with the history of the study of chromosomal inversions and heterosis in the 1930s to 1950s. I focus on how Goldschmidt and Dobzhansky adjusted their views to incorporate new evidence. As this story shows, Dobzhansky who did many of the fundamental studies on chromosomal variations in populations of Drosophila amended his views of selection to reflect the growing knowledge of the fitness effects of inversions. Goldschmidt resisted the prevailing accounts of inversions by Dobzhansky and his coworkers in his 1940 book, The Material Basis of Evolution, focusing his attention on the chromosomal basis of genic action. Subsequently he moved his attention to other genetic phenomena, in particular to homeotic mutations, that could be better used to support his convictions about the importance of development and on how developmental integration of new mutants affects evolutionary dynamics. I then discuss Goldschmidt’s hierarchical view of the genome and relate it to views of chromosomal systems affecting patterns of inheritance and explaining population stability from Cyril Darlington, Kenneth Mather, and Dobzhansky’s group. Finally, I discuss the role played in these debates by notions of developmental robustness and selective neutrality. I conclude with some general remarks about hierarchical-systemic views in genetics.

Inversions – Genome Dynamics and Selective Neutrality In 1933, Richard Goldschmidt first presented his unconventional macro-evolutionary views about hopeful monsters (Goldschmidt, 1933). Remarkably, for an article proclaiming a geneticist’s view of evolution and promising to present the evolutionary aspects of genetic research, this talk delivered at the AAAS general meeting has very little genetics in it and scarcely a hint of the heretical views in genetics Goldschmidt is now famous for. These were published some 5 years later. Before getting to them it is helpful to recall Goldschmidt’s early theory of gene action that was published many years earlier in the 1910s, since it continued to shape his thoughts and was considered by his contemporaries to be an important contribution. Goldschmidt studied the production of intersex individuals that presented a mixture of female and male characters, when different varieties of Gypsy moths were crossed. Based on his observations Goldschmidt suggested his balance

SYSTEMS THINKING VERSUS POPULATION THINKING

theory of sex determination, according to which the phenotype depends on a balance between male and female factors. Further observations led him to conclude that different organs could express different sexual characteristics depending, so he argued, on whether their time of differentiation was before or after a turning point in development from male to female or vice versa (Dietrich, 2000b). Goldschmidt argued that genes controlled the velocity of developmental reactions and that normal development depended on the production of the required amount of material at the correct time (Davis et al., 2009). He used this analysis to explain the existence of sensitive periods in development. This work was in physiological genetics, a field to which Goldschmidt made other important contributions since the beginning of his career in Germany (see Richmond, 2007). As we will see, he resisted the change in the focus in genetics from genes in development to transmission genetics and population dynamics. This resistance pushed him in a different direction than the mainstream of the field and was the backdrop to the debates recounted below. Already in his 1933 talk and very clearly in his 1939 Silliman lectures published as The Material Basis of Evolution in 1940 (MBE), Goldschmidt stressed that evolutionary change was built upon the potentialities of the developmental system of the organism. Speciation, in particular, requires that a ‘‘considerable number of developmental processes between egg and adult have to be changed’’ (1933, p. 543). Macro-evolution was not the result of selective accumulation of mutations, according to Goldschmidt, rather it was fundamentally the result of the way the developmental system produced novel phenotypes. Goldschmidt was happy to acknowledge that this seemed to suggest a refined, and legitimate, form of orthogenesis or evolutionary directionality, independent of selection. In 1937/1938 Goldschmidt began voicing his rejection of the particulate gene.3 His main argument at this time was based on position effects. The position effect was discovered in 1925 by Alfred Sturtevant in the Bar allele in Drosophila. It is manifested when what is supposedly the same gene produces different effects depending on the chromosomal environment it is in. Goldschmidt’s argument, in a nutshell, was that 3

According to Dietrich (2011), Goldschmidt’s doubts about the classical gene began in 1932. Goldschmidt reports in his autobiography (Goldschmidt, 1960, p. 322) that the earliest presentation of his non-corpuscular views was published in his book Physiological Genetics (1938); earlier hints of this view can be found in (Goldschmidt, 1937a, b; Goldschmidt 1938). Muller was keenly aware of the possible implications of these and subsequent discoveries about chromosomal rearrangements, yet remained cautious (see Muller to Goldschmidt, June 19, 1939).

EHUD LAMM

this indicated that genes were not independent units of function. In the language of the time: genes were not corpuscular or atomic. Goldschmidt was well aware that these views were ‘‘heretical,’’ hence, he said, his reluctance to voice them in public until 1938, when he felt he gathered sufficient data to support his claims. Some of the difficulties Goldschmidt would go on to champion, in particular chromosomal variation between races and species, were already noted and acknowledged by Dobzhansky who in a 1935 study of the Y chromosome in Drosophila pseudoobscura noted that his data ‘‘indicate that the process of race and species formation may be resolved into at least two components: the genic differentiation and the differentiation in the chromosome structure. These two components seem virtually independent, but the phenomena of position effects may lead us to a recognition of an ultimate connection between the two’’ (Dobzhansky, 1935, p. 374). Goldschmidt will invoke remarks such as this in years to come to suggest that his view offered a more consistent or parsimonious interpretation of the empirical data (see below). Back in 1932, Dobzhansky tried to account for position effects in Bar mutations using D. H. Thompson’s Episome-Protosome model of the gene, which also goes beyond the notion of genes as atomic units (Dobzhansky, 1932). Goldschmidt’s view, which he developed and adjusted over time, was that chromosomal segments, rather than genes, have functional consequences. There are no structural units that have genic action; rather, it is the chromosome as a whole or (later) segments of increasing length that are responsible for genic action. In 1940 Goldschmidt published MBE. Here he presented his evolutionary views in more detail and surveyed a large and varied body of evidence that was meant to support them. In addition to arguing for the role of macro-mutations in speciation he suggested a mechanism underlying these macro-mutations: chromosomal repatterning. This mechanism relied on his view of genic action which, in turn, was founded on the rejection of the particulate gene (I elaborate on the relations between these strands in Goldschmidt thought below). His arguments were meant to support rejecting genes as units of action not as units of transmission; transmission as such was not at the focus of his attention at this point. An important source of evidence at this time, in addition to position effects, was the growing knowledge of chromosomal inversions, especially their variation in natural populations. Inversions were an important topic of empirical and theoretical research in the 1930s and 1940s. During most the 1930s, chromosomal inversions in Drosophila were assumed to be selectively neutral. The

SYSTEMS THINKING VERSUS POPULATION THINKING

neutrality or near neutrality of inversions, since Goldschmidt acknowledged they could have minor fitness affects, played an important role in Goldschmidt’s MBE arguments about the spread of chromosomal variations prior to an evolutionary saltation. Being largely neutral, their spread could not be the result of selection. Goldschmidt was nonetheless convinced that chromosomal patterns were strictly genetically causal, a view that was not shared by researchers who studied the neutrality of inversions and later established that they could in fact affect fitness, in particular Sturtevant and Dobzhansky and Nikolai Dubinin and his colleagues in the USSR. If inversions are merely a reordering of a set of discrete genes whose genic function is independent from one another, as most researchers believed, they should have no functional or fitness effect. Any such effect would be an exception (position effects, for example, could be such exceptional cases). Goldschmidt, in contrast, supporting his view with position effects, viewed as paradigmatic rather than as exceptional cases, regarded chromosomal organization as what causally grounds genic action. In other words he thought chromosomal organization had a direct physiological effect. Chromosomes were not chains of independent genes or ‘‘beads-on-astring.’’ This explained why the accumulation of enough changes in chromosomal organization could lead to a major change in phenotype, and an evolutionary saltation. In the late 1930s Dobzhansky gathered substantial evidence supporting the conclusion that different inversions differed in gene content (see Beatty, 1987a). Inversions were methodologically useful for studying the genetic differences among populations of Drosophila, and were known to have a significant effect in terms of chromosomal dynamics by reducing crossing over. However, the physiological effects they had were supposedly grounded in their differing gene content. Distinguishing between the value of genic and chromosomal heterozygosity became a major goal in the last 25 years of Dobzhansky’s Genetics of Natural Populations, but the question was not resolved successfully (Lewontin, 1981, p. 113). Attributing physiological affects to gene content was exactly what Goldschmidt disputed. Goldschmidt is not unique in interpreting and weighing evidence based on his overall theoretical outlook and the totality of his commitments. We will see the same with Dobzhansky’s changing explanations of inversion polymorphisms in populations, once it was discovered that they affect fitness. What were unique were Goldschmidt’s commitments. In particular, his steadfast loyalty to physiological genetics over transmission genetics, and hence to developmental considerations,

EHUD LAMM

and his willingness to reject the atomic gene. Rejecting the gene was not an easy thing to do in the late 1930s, let alone in the 1950s, and required Goldschmidt to come up with alternative interpretations to growing amounts of evidence that supported the gene view. Through this period he moved his attention to other genetic phenomena, in particular to homeotic mutations, to provide support for his evolutionary views while remaining committed to his views regarding the gene. Staining of salivary chromosomes in Drosophila (Painter, 1934) made the identification of chromosomal inversions routine. The inversions were interpreted as being rotated blocks of atomic genes, without the need to prove this using the much harder to obtain genetic evidence.4 Observations and analyses on the level of chromosomes, not genes, were used to establish that inversions lead to reduced recombination. Based on this fact, blocks of genes identified by observing inversions were taken to be linkage groups, or high-order genetic units (Sturtevant and Mather, 1938, p. 448). Because they were easy to observe, inversions were used to identify specific blocks of genes. For example, a notation such as II-1 might refer to a specific inversion marked as number 1 on chromosome II, but at the same time it was understood as referring to a specific set of genes (e.g., Dubinin and Tiniakov, 1946b). Most often the individual genes identified by an inversion were unspecified. While the bands on salivary chromosomes could be used to map the location of genes on chromosomes, it was not known if, on a biochemical level, the deeply staining bands were the genes per se. As Painter put it, it was not known ‘‘where are the genes’’ relative to the staining (Painter, 1934). That was a separate empirical question. In addition, inversions were studied as evolutionary events, and the staining results produced by heterozygosity to inversions were used as tools for phylogenetic analysis (see Dobzhansky and Sturtevant, 1938), this being another use that sidesteps the issue of the exact relation between genes and chromosomes and does not depend on gene-level modeling. We see that there were several levels of interpretation of empirical observations that were going on simultaneously, each guided by different theoretical concerns and assumptions. Those, such as Painter, studying the structure of the chromosome, debated which physical features of chromosomes accounted for the different bands created by staining (Painter, 1934). At the same time, staining, whose exact relation to genes was another open question, was used as a tool to determine the location of genes along the chromosome, and to identify inversions, 4

See for example Dobzhansky and Sturtevant’s (1938) remarks about the labor involved in producing the work presented in Sturtevant (1917).

SYSTEMS THINKING VERSUS POPULATION THINKING

which in turn could be analyzed both as chromosomal features (e.g., for explaining reduction in crossing-over), and as indicating blocks of particulate genes (i.e., linkage groups). All these levels of analysis, not known at the time whether complementary or eventually to prove mutually exclusive, could be employed by the same researchers, either when working on different problems, or even as two types of analyses used to attack the very same question. This is illustrated nicely when we contrast Dobzhansky and Sturtevant (1938), which is mostly concerned with inversions as features of chromosomes, with Sturtevant and Mather (1938) which explains the heterosis effect of inversions by explicitly appealing to gene based modeling. Goldschmidt paid particular attention to the work of Dobzhansky and Dubinin (Goldschmidt, 1940, pp. 204, 245, respectively), both of whom studied the distribution of inversions in natural populations. He mentioned Dubinin, a disciple of Chetverikov, as representing what he called ‘‘the population problem’’ approach to explaining variation in populations. We would now call this approach population genetics. To his own account Goldschmidt referred as the ‘‘genetical problem’’. These terms illustrate nicely that for Goldschmidt the term genetic referred primarily to gene action and not to transmission dynamics in populations. Since 1939, Dobzhansky accepted that there were regular seasonal changes in the prevalence of different inversions in natural populations. The frequencies were not fluctuating randomly. He concluded from this that the selective difference between them had to be large since in small populations only strong selection can overcome drift (Wright and Dobzhansky, 1946; see also Beatty, 1987a). Dobzhansky’s initial explanation for this, in 1939, was that inversions may be associated with different sets of alleles, and so may be more or less adapted to particular conditions. The genes comprising each inversion he did not consider to be necessarily adapted to one another. His explanation why an inversion which is superior in one season does not overtake the population was that inversions probably also carry recessive deleterious or lethal alleles, and thus do not fare well in homozygous form (Dobzhansky, 1939). The observation that inversions reduce the amount of recombination, rather than Goldschmidt’s chromosomal patterns, allowed researchers to explain the genetic effect of inversions, their population dynamics, as well as the phenomenon of heterosis, that is that individuals heterozygous for specific inversions often had a fitness advantage over homozygous. The explanation of heterosis was based on the observation that due to reduced recombination an inversion will lead to

EHUD LAMM

the establishment of two separate lines of descent. Heterozygosity for the inversions will then probably mask deleterious mutations, as I explain below (Sturtevant and Mather, 1938). This argument, to which Goldschmidt did not respond in MBE, belongs with the work Goldschmidt classified there as dealing with the ‘‘population problem’’, that is to population genetics. A central part in Goldschmidt’s argument in MBE tied together two seemingly conflicting bits of evidence regarding inversions. Intraspecific variation in chromosomal pattern was considered at the time to indicate that pattern changes may be neutral. However, inversions were known to vary between races, and ultimately between species that are distinguishable by chromosomal pattern. Goldschmidt also emphasized that chromosomal differences between species in the same genus were prevalent. Chromosomal incompatibility moreover explained sterility of hybrids (Goldschmidt, 1940, pp. 189–197). Goldschmidt argued that these facts indicated that inversions could be phenotypically neutral, while their accumulation past some threshold can ultimately lead to speciation (i.e., an evolutionary saltation). Clearly, if chromosomal pattern changes are viewed ‘‘genetically’’ rather than from the perspective of the ‘‘population problem,’’ explaining their apparent neutrality as well as their ultimate evolutionary role becomes more involved than when these phenomena are explained by their effects on chromosomal incompatibility (p. 245). What makes Goldschmidt’s view in MBE particularly interesting is that he tried to combine the common belief, later rescinded, that inversions were selectively neutral, with a chromosomal view of genic action, according to which chromosomal inversions were presumed to have genetic effect rather than be inherently neutral. Goldschmidt’s chromosomal pattern theory of genic action, which he used to develop an hierarchical conception of the chromosome (see below) was aimed at providing such an explanation. In MBE, Goldschmidt took Dobzhansky to task on his views regarding the atomicity of genes. Somewhat sarcastically, Goldschmidt quoted two passages from Dobzhansky supposedly indicating that Dobzhansky felt a tension within his views. The quote from Dobzhansky ends with his claim that ‘‘position effects show that gene mutations and chromosomal changes are not necessarily as fundamentally distinct phenomena as they first appear’’ (Dobzhansky, 1937, quoted in Goldschmidt, 1940, p. 204). Goldschmidt dryly commented that ‘‘the author [namely, Dobzhansky] realized, more or less, that a decision has to be made between the mutationist, neo-Darwinian view of the accumulation of gene mutations and a repatterning of the

SYSTEMS THINKING VERSUS POPULATION THINKING

chromosomes independent of the assumption of genes as units. But he has not yet been willing to cut the Gordian knot’’ (Goldschmidt, 1940, p. 204). Position effects suggested how inversions could produce effects ranging from selectively neutral to strongly affecting fitness. Goldschmidt took this as indication that all mutations boil down to chromosomal rearrangements, that is to changes in intrachromsomal pattern. For him, this was tantamount to rejecting the existence of the particulate gene. Dobzhansky and Goldschmidt both had to accommodate tensions between their commitments, but the commitments were different. Dobzhansky, said Goldschmidt, shied away from ‘‘cutting the Gordian knot’’, since he remained faithful to the notion of units of action even though he acknowledged phenomena such as position effects that indicated that function depends on chromosomal context. Seen from Goldschmidt’s perspective his jab at Dobzhansky seems justified. So how did Dobzhansky handle this tension? How did he interpret the growing evidence concerning the fitness effects of inversions and of heterosis? His accounts all took gene content to be the explanation of fitness effects and selection or drift to be the explanation of population dynamics. Extensive work by him and his coworkers, as we will see, was aimed at studying and explaining the effects of heterozygosity for inversions and for genes. Dobzhansky and Sturtevant (1938), cited in MBE, were concerned with inversions in natural populations of D. pseudoobscura. Two methodological aspects of this work are worth noting. First, they were interested in using inversions to recreate the phylogenetic tree (pp. 32– 34; see Figure 1). As they already noted in a letter to Wright in 1936, it was possible to identify that two chromosomal patterns were related only through a third using staining of salivary chromosomes. In this way it was possible to draw a phylogenetic tree, though this method alone could not identify which was the original form of the chromosome (i.e., where to root the tree).5 They were thus concerned with chromosomal events not simply as events which might have evolutionary significance, but as events leaving footprints that were methodologically useful. Using inversions for this purpose does not depend on assumptions about selection, only on chromosomal similarities. Indeed, on the topic of selection, Dobzhansky and Sturtevant explicitly expressed the view that ‘‘inversions seem to come closest to a neutral condition’’ 5

Sturtevant to Wright, March 18, 1936. Sewall Wright Papers. For background on Sturtevant and Dobzhansky’s early collaboration on evolutionary issues see Kohler (1994, Chap. 8).

EHUD LAMM

Figure 1. From Dobzhansky and Sturtevant (1938).  Genetics Society of America

(p. 61), a view that as mentioned earlier survived till the mid 1940s. Dobzhansky and Sturtevant continued, however, by noting that when non-neutral, the selective value is that of heterosis, whose magnitude they could not evaluate at the time. This brings us directly to the set of concerns of Sturtevant and Mather (1938), which is not cited in MBE. This is a theoretical paper that makes use of a population genetics model to address a question raised by the previous paper, namely that ‘‘heterozygosis for inversions decreases the amount of crossing over, and this may be of selective value in connection with heterosis effects’’ (Dobzhansky and Sturtevant, 1938, p. 61; quoted by Sturtevant and Mather, 1938, p. 447). Heterosis is explained by the fact that in chromosomes with two forms, differing by an inversion, the inversion segment will evolve separately in each, creating two lines of descent. The reason for this is that there is no effective crossing over in individuals heterozygous for an inversion unless the inversion is a very long one.

SYSTEMS THINKING VERSUS POPULATION THINKING

Since the two chromosomal variants evolve separately, Sturtevant and Mather argued, individuals heterozygous for an inversion are likely to be heterozygous for recessive deleterious mutations, and hence enjoy a selective advantage. This elegant argument relied on a combination of gene and chromosomal perspectives, but, significantly, the chromosomal organization was not seen as having direct physiological consequences, of the sort Goldschmidt argued for. To explain why chromosomal inversions are common in the third chromosome of D. pseudoobscura (an observation that launched the Genetic of Natural Population series; Provine, 1981), yet rare in other chromosomes, Sturtevant and Mather (1938) revisited the question of the advantages of recombination, a question that was prominent at the time. They mentioned that two types of explanation are found in the literature, namely the explanation endorsed by Fisher (1930) and Muller (1932) according to which recombination allows independent favorable mutations to be combined (see Felsenstein, 1974) and the argument that recombination makes the species more ‘‘flexible’’ to changing environmental demands. Sturtevant and Mather reject the former explanation as assigning too small a selective advantage to crossing over, since crossing over has to occur only once in order to combine beneficial mutations. Goldschmidt’s discussion of the chromosome as unit of genetic action and of chromosomal inversions, did not address these issues. In a generally favorable but critical review of MBE in Nature, Conrad Waddington called it ‘‘one of the most important of recent contributions to the theory of evolution’’ (Waddington, 1941). At the same time he rejected Goldschmidt’s repudiation of the gene, noting in particular that the unbridgeable gaps between species could be explained by an appeal to genes. Genes, Waddington argued contra Goldschmidt, depend on highly calibrated developmental paths, and if these are disrupted by a change in an early acting gene, significant recalibration of the genetic background will be needed. This will lead to the isolation of the old and new groups, and explains the unbridgeable gaps in form that Goldschmidt classified as macro-evolutionary changes. What Waddington did, in effect, is to propose a developmental model. As a form of explanation, this model looks very different than population genetics models; in particular it foregrounds the organism at the expense of the genes. As described by Davis et al. (2009), Waddington’s work on homeotic mutants led him to propose a branching track model of development in his 1940 book Organisers and Genes. According to this model, major genes determine which of several distinct developmental paths are taken, with other genes leading to

EHUD LAMM

various alterations. This structure explained the lack of intermediate forms at the tissue level which Waddington observed. Waddington described this alternative as an elaboration of Goldschmidt’s early theory of the gene. Waddington’s analysis of the developmental process, or ‘‘the causal structure of the animal,’’ as stemming from branching developmental paths (later to become his epigenetic landscape model) differed from Goldschmidt’s early theory of the gene which merely argued that development was controlled by genes altering reaction velocities; but both of scientists shared cybernetic views about the interaction between development and evolution (Goldschmidt, 1954; Waddington, 1961). Davis et al. (2009) argue convincingly that Waddington’s work, as well as Goldschmidt’s own work on homeotic mutants that I discuss below and Dobzhansy’s ideas on the effect of homeotic mutants on ‘‘fundamental’’ characters of organisms, show that during the period of the evolutionary synthesis there were research programs attempting to integrate genetics and experimental embryology, and that what happened with development in the synthesis was not a simple story of exclusion. The picture becomes even richer when we consider Waddington’s further remarks about MBE. In his review, Waddington was critical of Goldschmidt’s view about the importance of chromosomal patterns, and his appeal to ‘‘rearrangements of the chromatin’’ as underlying change in the general aspects of the organism, and described it as a ‘‘long shot in the dark’’ (Waddington, 1941). Wright (1941, p. 168) made essentially the same points in his review of MBE in The Scientific Monthly. Thus, while Waddington and Goldschmidt were thinking along somewhat similar lines, Waddington did not repudiate the gene, and was not interested in rearrangements and locality in the chromatin. His concern was with changes in the developmental landscape, not the underlying genetical system. Goldschmidt tried to give a unified account of both. As Dietrich (2000b) is right to emphasize, developmental macromutations were accepted as potentially significant by a wide range of biologists before and after the advent of molecular biology, including Curt Stern, G.G. Simpson, and Sewall Wright – while Goldschmidt’s ideas about chromatin organization were not. The issue of chromatin organization and function in addition to the genetics of development, however, was a concern for both the physiological and the transmission geneticists whose work is discussed here. Goldschmidt also argued that the spread of inversions, thought to be quasi-neutral, was unrelated to selection (Goldschmidt, 1940, p. 191).

SYSTEMS THINKING VERSUS POPULATION THINKING

He thus relied on a notion of drift and assumed the neutrality of chromosomal changes ‘‘until a new reaction system is built.’’ As we will see, neutrality was not just a result of chromosomal genic action; it was also made possible by developmental robustness.6 In the mid 1940s, the work of Sewall Wright with Dobzhansky and of Nikolai Dubinin and his colleagues in the USSR made it clear that chromosomal inversions could create large fitness differences (Dubinin and Tiniakov, 1945, 1946a, b; Wright and Dobzhansky, 1946; see Ives, 1947 for a contemporary review). They studied these effects in natural populations and in the lab. One of the most important observations was that of heterosis. Dubinin and Tiniakov (1946a, b) studied chromosomal inversions in natural populations of D. funebris, and the selective advantages enjoyed by heterozygotes for certain inversions. They noted that their results, which were the first attempt to study the selective effect of the karyotype directly, led to similar conclusions as those of Wright and Dobzhansky (1946) on D. pseudoobscura, who in turn cited Dubinin and Tiniakov (1945). All these were primarily observational studies, whether in the field or in the lab. Dubinin and Tiniakov also cited the more general or theoretical analyses of Sturtevant and Mather (1938) and Dobzhansky and Sturtevant (1938) which I discussed earlier. The fitness differences between individuals with different inversions could be taken as evidence in support of the view that chromosomal organization affects genic action – as Goldschmidt thought. This was not the prevailing view, however. Wallace (1994, p. 156) notes that a crucial observation was that the relative fitness of the same gene arrangement was not always the same but depended on the source population of the chromosomes. This, he writes, was ‘‘regarded as proof that the gene content of chromosomes with the same gene arrangement varied, thus ruling out position effect as the only factor governing the retention of a particular gene arrangement.’’ These results posed two challenges for Goldschmidt. The previously dominant view that inversions were neutral or quasi-neutral played an important role in Goldschmidt’s MBE argument about the spread of chromosomal variations prior to an evolutionary saltation, which he argued was not the result of selective accumulation of (gene) mutations. Moreover, chromosomal organization and gene content were understood to be distinct phenomena. To summarize that story so far: By the mid-1940s it became evident that inversions were not neutral and that hybrids for inversions could 6

For more on the interesting relationship between Goldschmidt and Waddington see Richmond (2007).

EHUD LAMM

have hybrid vigor (heterosis). Extensive research was conducted to explain why heterozygosity could be advantageous. However, all this work accepted the reality of the gene and the distinction between chromosome organization and its gene content. Goldschmidt followed this work, but was an outsider to this consensus. His interest lay in the organism and its development, on the one hand, and the physiological consequences of the organization of the chromosome rather than its gene content, on the other.

Goldschmidt’s Hierarchical View of the Genome From 1944 until his death in 1958 Goldschmidt advanced a hierarchical view of the chromosome (or chromatin). He formulated several versions of this view over the years (e.g., Goldschmidt, 1954; see Dietrich, 2000a). What is common to all these accounts is that chromosomal segments of increasing size were considered as significant levels of organization, each with specific kinds of genetic/physiological effects. The segments are nested, with longer segments surrounding and encompassing smaller ones (see Figure 2). Clearly, the implication was that chromosomal segments larger than a single locus should have specific kinds of actions. Goldschmidt suggested, as an example, that a section of the third chromosome of Drosophila was responsible for segmental determination. The evidence for this was that multiple mutations in segmentation were mapped to this region. Nonetheless, Goldschmidt argued that when not mutated, the control of development was not by individual loci but by the section as a whole (Goldschmidt, 1955, p. 182). He suggested that the larger the section the earlier the

Figure 2. Goldschmidt’s hierarchical view of the chromosome and Mather’s visualization of genes’ ‘‘fields of cooperation’’ which Goldschmidt took to be in agreement with his own view (both from Goldschmidt, 1955, pp. 180–181)

SYSTEMS THINKING VERSUS POPULATION THINKING

developmental processes it controls. The hierarchical structure of the chromosome thus reflects, to an extent, the hierarchical nature of the developmental processes of the organism. Even setting aside for a moment the reality of atomic genes, this type of account entails that the organization of the chromosome is physiologically significant and this has evolutionary implications. In particular, this would clearly impose significant evolutionary constraints. Since 1945, Goldschmidt also emphasized the importance of a different class of empirical phenomena, homeotic mutations, rather than inversions. These are mutations that lead to the transformation of one body part into another. The paradigmatic example is a Drosophila mutation causing a leg to develop in place of an antenna. Homeotic mutations can lead to major changes in the phenotype and provided support for Goldschmidt’s insistence that macro-mutations have a critical role in evolution. However, as in the discussion of macro-mutations in 1933, homeotic mutations were not claimed to necessarily be chromosomal reorganizations (i.e., ‘‘systemic mutations’’). Homeotic mutations, like macro-mutations more generally, showed the possibility of evolutionary saltations and pointed to the importance of development for understanding mutations. Using homeotic mutations to support these conclusions did not rest on an account of the nature of the gene; though for Goldschmidt a genic change implied chromosomal repatterning. What remained critical throughout was that developmental considerations were paramount in Goldschmidt’s account of evolution, while population genetics were at the most playing second fiddle. So, though Goldschmidt grounded his conclusions in different empirical cases, now that inversions were no longer considered neutral and heterosis was explained by appeal to gene complexes maintained by reduced crossing-over, Goldschmidt’s basic line of evolutionary argument remained similar.

Chromosomal Systems and Population Stability In 1950, Dobzhansky’s preferred explanation for variations in the prevalence of inversions in natural populations was that inversions were linkage groups of genes, which worked well with a complementary linkage group in the homologous chromosome (Beatty, 1987a). These he called ‘‘co-adapted’’ gene complexes (Dobzhansky, 1950, p. 288). Coadaption is a population level notion and is manifested in the frequencies of the various inversions in the population. According to this view on the distribution of inversions, heterosis is the result of selection,

EHUD LAMM

which is responsible for building complementary gene complexes and maintaining co-adaptation. This conclusion was reached by crossing flies from different natural populations and noting the frequency of heterozygous adults. It turned out based on these frequencies that the heterozygotes were not superior to homozygotes. This indicated that intrapopulation superiority of heterozygotes was the result of selection working within interbreeding populations. These results excited Dobzhansky tremendously and became the focus of further studies for several years.7 Dobzhansky happily noted that if these results stood to scrutiny they had devastating implications for the classic view of heterosis being the straightforward result of the masking of deleterious recessives. Rather it was a property of evolved and balanced (i.e., co-adapted) heterozygosity. The results also suggested an explanation of speciation. Border territories between separate intrabreeding populations would give rise to inferior heterozygotes, favoring the evolution of isolating mechanisms.8 In the summer of 1950 a five-week symposium on heterosis was held in Iowa State College (see Crow, 1987). Crow recalls that the conference was well attended, especially by plant and animal breeders. He notes that it was generally understood that the two competing explanations of heterosis, dominance and overdominance, were statistical concepts pertaining to the aggregate effect of genes on the variance of traits. In his short contribution on ‘‘The Nature and Origins of Heterosis’’, which according to Crow was well-received, Dobzhansky maintained that true heterozygote advantage (overdominance) was not the property of a single locus, nor of chromosomal structure, but rather the property of ‘‘integrated systems of polygenes’’ maintained by inversions (1952, p. 222). In 1954, according to Beatty (1987a), Dobzhansky came to the view that heterozygosity can be advantageous in and of itself and that that this may extend even to single genes. This conclusion was based on the evidence found in Vetukhiv (1953) and Brncic (1954) but went significantly beyond it. Vetukhiv (1953) studied heterozygosity that was not necessarily visible in chromosomal morphology. The results suggested that heterozygosity as such can be advantageous only ‘‘up to a certain point,’’ since between-populations F1 hybrids were superior to parental strains, while the F2 generation, affected by crossing over and recombination, was in fact inferior to F1 and often to the parental

7 Dobzhansky to Dunn, 26 April 1947; Dobzhansky to Dunn, 19 May 1947. Dunn correspondence. 8 Ibid.

SYSTEMS THINKING VERSUS POPULATION THINKING

strains as well (see Brncic, 1954). This, it was argued, meant that population genotypes were ‘‘integrated’’. This integration is the result of selection. The superiority of the F1 hybrids was harder to explain and suggested that perhaps some level of heterozygosity can give a limited advantage to the organism. Brncic emphasized that any advantage of heterozygosity as such is ‘‘more than offset by the disintegration of the internally balanced combinations of genes carried in the chromosomes of every population’’ (1954, p. 87). Brncic showed that the heterozygote advantage is lost when the gene complexes are broken up by crossing over (in the F2 generation). This was evidence in favor of ‘‘internally balanced gene systems’’ in the chromosomes, an idea that several workers have previously suggested. As defined by Mather (1943), internal balance referred to chromosomes that are optimal when homozygous whereas he used the term ‘‘relational balance’’ to refer to the relation between pairs of different homologous combinations, namely chromosome pairs that are advantageous when heterozygous. Dobzhansky and his coworkers in this period (in particular Vetukhiv, Brncic, and Spassky) used the term co-adaptation, introduced by Dobzhansky in 1950, to refer to polygene complexes found on homologous chromosomes. The following from Vetukhiv (1953) gives the gist of their view at the time: ‘‘The genotype of each local population or race evidently represents an integrated adaptive system, the different parts of which are mutually adjusted or coadapted by natural selection.’’ (Vetukhiv, 1953, p. 33) The notions mentioned in this quote are all inherently population-level notions. The term genotype in particular refers here to the genetic profile of a local population, not that of an individual. Natural selection drives towards optimal gene frequencies, not simply a set of optimal genes. The optimal frequencies are affected by the cycle of homozygosity–heterozygosity already mentioned by Chetverikov. The evolutionarily significant population balance thus depends on the breeding system of the species (i.e., inbreeding versus outbreeding). The following quotes illustrate Dobzhansky’s thinking through this period, which took populations to be integrated systems calibrated by natural selection (my italics throughout): ‘‘Sexual reproduction has brought about a new form of biological integration. Individuals are combined into reproductive communities, Mendelian populations…. they owe their cohesion… not only to common descent, but, and primarily, to mating and to parentage

EHUD LAMM

bonds. The sexual unions and the gene segregations occur in every generation in Mendelian populations, and determine both the continuity and the changeability of their collective genotypes, gene pools…. No less important is the fact that, in sexual organisms, Mendelian populations, rather than individuals, have become the units of the adaptively most decisive forms of natural selection.’’ (Dobzhansky, 1951) ‘‘individuals are integral parts of the collective genotype, the gene pool, of the population. The gene pool is not an accidental conglomeration of individual genotypes. It is rather an organized system, a system so contrived as to yield the highest mean level of adaptedness in the individuals which compose the population.’’ (Dobzhansky, 1953) ‘‘In recent years it is becoming increasingly appreciated that Mendelian populations, though, of course, composed of individuals, have an internal genetic cohesion which is a property of a population, not that of any individual…. The properties of a population transcend those of the individuals.’’ (Dobzhansky, 1955)9 The sort of systems thinking about the individual and populational genotype exemplified by these quotes was already apparent in Cyril Darlington’s discussion in The Evolution of Genetic Systems (1939), the core ideas of which he already published in the last chapter of Advances in Cytology (1932) as well as in Chetverikov’s 1926 discussion mentioned earlier. Darlington’s seminal ideas were at first vehemently rejected by the proponents of the Synthesis as well as by cytogeneticists, but as these quotes indicate similar ideas were subsequently accepted, in particular by Dobzhansky (for more on their reception see Harman, 2004, pp. 117–118; Smocovitis, 1996, p. 136). Following this critical strand of the story would lead us too far afield and take us to the uneasy integration of cytology and genetics, which preceded the integration of cytogenetics into the synthetic theory by Dobzhanksy in his 1936 lectures that led to Genetics and the Origin of Species in 1937 (Carson, 1980). Mather (1943) studied how population balance (frequency equilibrium) can result from selection on individuals as a question of Mendelian theory and the issue is presented in Darlington and Mather (1950) who emphasized the cycle between ‘‘free’’ and ‘‘potential’’ vari9

I thank Lisa Gannett for sharing these quotes from Dobzhansky with me. For a nuanced analysis of Dobzhansky’s views in this period see Gannett (2013).

SYSTEMS THINKING VERSUS POPULATION THINKING

ability, that is between homozygosity and heterozygosity (see Figure 3). A particular case that called for explanation involved the stable frequency of lethals and of inversions carrying lethals in natural populations, which suggested equilibrium. The stable frequencies had to reflect the fitness of individuals heterozygous for the lethal mutations. This issue already intrigued Dobzhansky and Sturtevant in 1936 who brought it to the attention of Sewall Wright.10 However, equilibrium is not necessarily a result of specific mechanisms that were selected for maintaining it. Dobzhansky’s research that I described earlier was partly an attempt to establish whether such mechanisms existed. Dobzhansky was influenced by Wright’s shifting-balance model, and in the 1951 edition of Genetics and the Origin of Species applied it to explaining stable polymorphisms in populations. His views were also influenced by the Chicago ecologists, notably Alfred Emerson, who applied group-selectionist reasoning. Cain and Shepard objected to this line of reasoning, using the example of sickle-cell anemia. This example showed that heterozygosity can be maintained due to its positive effects on heterozygote individuals, rather than because of the advantages of variation to a population (see Mitman, 1992, p. 120). While related, the system-perspective on the genetic system of populations is, however, distinct from the issue of level of selection. Dobzhansky’s research in this period focused on further elucidating the level at which heterozygosity is advantageous, in particular whether its advantages to the individual stem from single-locus heterozygosity or heterozygosity to inversions and whether any heterozygosity suffices for having this advantage or only one between calibrated sets of genes produced by selection. This work was concerned with the hereditary mechanisms involved, specifically with how co-adapted gene systems are maintained through inheritance. Based on his results, Brncic (1954) suggested that ‘‘Specialized chromosomal mechanisms, such as inversions found in many natural populations, have as their biological function the preservation of such balanced gene systems.’’ (p. 87).The article did not offer an account of what this means or what it entailed more generally about the organization of the chromosome or genome. Dobzhansky (1952) elaborated more on this issue and noted that ‘‘Any factor which restricts or prevents crossing over in chromosomes, or parts of chromosomes, can accomplish the same biological function. Localization of chiasmata may be such a factor. If, for example, chiasmata are found chiefly or 10

Sturtevant to Wright, March 1936; Sturtevant to Wright, Sep. 1938; Wright to Sturtevant, Oct. 1938. Sewall Wright Papers, APS.

EHUD LAMM

Figure 3. From Darlington and Mather (1950)

exclusively at some definite points in a chromosome, the genes carried in the sections which intervene between these points are inherited in blocks. Such gene blocks may act exactly as gene complexes bound together by inversions.’’ The heterozygote advantage observed in F1 by Brncic and Vetukhiv, however, could not be the result of selection building up complementary co-adapted gene complexes, since the parents were from different populations not subject to interbreeding. They did not hypothesize a physiological explanation for these facts, either at the level of the chromosome or the individual. Dobzhansky thus had to amend his entrenched view that heterosis was the result of selection and accept generalized heterosis. This was not an easy move for him to make, we are told (Lewontin, 1981, p. 103; Crow, 1987). For Dobzhansky, however, there remained a clear distinction between chromosome organization and gene content. His work was concerned with the physiological effects of genes on the individual, not on the way chromosomes produce gene function. One of the fundamental things these works

SYSTEMS THINKING VERSUS POPULATION THINKING

attempted to assess was the role of selection in producing co-adapted gene complexes, i.e. internally balanced gene systems in populations. By focusing on gene causality at the level individual organisms they come closer to the kind of concerns that motivated Goldschmidt yet their concerns remained very different. Developmental Robustness In 1955, influenced by the work of his friend and colleague I. M. Lerner, another Russian e´migre´, Dobzhansky suggested that heterozygous individuals have an advantage over homozygotes simply due to the fact that they possess two forms of a gene and are hence better protected in fluctuating environments (Dobzhansky and Levene, 1955). As in Goldschmidt’s accounts, causality is at the level of the individual. Unlike Goldschmidt, however, Dobzhansky’s version of Lerner’s views accepts the gene and does not depend on a specific account of genic action, while Goldschmidt’s account was rooted in a developmental account of chromosomal gene action. In contrast to the type of explanation sought by Goldschmidt, Lerner’s developmental homeostasis is concerned with pairs of alleles on homologous chromosomes, not on the internal organization of single chromosomes. Lerner argued that heterozygotes enjoy superior developmental buffering as compared to homozygotes. If this is the case, heterozygosity has a dual role in Mendelian populations: a store of genetic variability in populations and a functional role in the development of individuals (Lerner, 1954; Dobzhansky and Levene, 1955; Hall, 2005). If this explanation is correct, sexual reproduction has physiological advantages for individuals. This view, then, combines both levels of causality: population dynamics and individual development. Heterozygotes are better able to buffer against changes in internal and external environments, including presumably different genetic milieus. Wallace (1994) disputes Lewontin’s claim mentioned above that Dobzhansky dropped the notion of coadaptation after 1955, because heterosis was found in crosses between different populations with no interbreeding. This implied that it was not produced by selection for heterossis. Wallace acknowledges that in this period his own contact with Dobzhansky was more limited than in previous years; however, he disputes this testimony. In his 1962 book Mankind Evolving Dobzhansky presented his views on the benefits of variation to a broader audience. He emphasized the role of the ‘‘genetic environment’’ in which genes have to operate, essentially Chetverikov’s 1926 ‘‘genotypic

EHUD LAMM

milieu’’. Dobzhansky used the wonderful phrase ‘‘adaptively ambivalent genes’’ to refer to genes that are useful in combination with some genes, harmful in combination with others. He claimed that diversifying selection will maintain such genes in populations. Diversifying selection was his term for what Mather (1955) called ‘‘disruptive selection’’ and occurs when a Mendelian population repeatedly experiences different environments or ways of life that favor different genetic combinations (Dobzhansky, 1962, p. 248). This explanation does not involve direct conflict between groups. Dobzhansky highlighted the fact that changing environments pose challenges both to individuals, who need to be developmentally robust, and to populations who must adapt. He, however, referred specifically to the results of Thoday (1955) and to Waddington’s 1957 book. Thoday showed experimentally that heterozygotes produced by crossing flies from different populations did not exhibit a developmental advantage either in the environmental conditions the parents were adapted to, nor in a foreign environment. Heterozygotes produced by flies from the same population were superior to homozygotes. These results indicated that heterozygosity as such cannot explain the developmental advantages exhibited by heterozygotes. Rather, the advantage is due to the adaptedness of the strain to a particular heterozygotic state, namely Dobzhansky’s coadaptation or Mather’s relational balance (see Figure 4). Individuals enjoying relational balance may exhibit developmental flexibility, but heterozygosity alone does not produce developmental flexibility (Thoday, 1955). This agreed with the results of the Vetukhiv and Brncic papers. This brings us in some sense full circle. The work I focused on is often presented as a story about conceptualizing selection, presented following 1955 through the classical-balance debate with Muller (Beatty, 1987a, b, 1994; Crow, 1987), and in Goldschmidt’s case as a story about the gene concept (Carlson, 1966; Dietrich, 2000a, b, 2008). Two other related questions played a role throughout this period: the role of chromosomal organization and mechanisms, and the relationship between genes and development.11 The latter issue has received attention in discussions of the treatment of development in the evolutionary synthesis (Smocovitis, 1996; Dietrich, 2000b; Davis et al., 2009), but I hope I have shown that the former was a point of contact between researchers with diverse agendas and commitments and deserves historical scrutiny. What I referred to above as systems thinking, both 11

While I stress the elements of the story that I think do not receive the attention they deserve in Beatty’s account, I should say that his analysis of Dobzhansky’s views played a critical role in my thinking.

SYSTEMS THINKING VERSUS POPULATION THINKING

Figure 4. Table 1 from Thoday (1955) with annotations. A Heterozygotes produced by flies from the same population have greater developmental stability in both the original environment and a foreign environment. The underlined result indicates that these flies fare better in a foreign environment compared to homozygotes (342 versus 451). B Heterozygosity alone produced bad developmental outcomes in both home and foreign environments.  Cold Spring Harbor Laboratory Press

regarding population variation and regarding organismic development, was part and parcel of how the empirical data were interpreted and shaped the empirical questions that were posed. This is seen most notably in Vetukhiv (1953), Brncic (1954) and Thoday (1955) and in the work of Goldschmidt throughout this period, in particular his interpretation of the results of Dobzhansky, Dubinin and their coworkers and in his hierarchical view of the genome. A different yet related style of system thinking is found in Waddington’s model of organism development, which he explicitly contrasted with Goldschmidt’s views of development. Dobzhansky’s reference to Thoday and Waddington in Mankind Evolving, as well as his comments about Thoday’s paper when it was presented, illustrate nicely how closely population-thinking and developmental considerations were related and how Dobzhansky connected the two (see Table 1, which appears in Thoday’s paper). As I

EHUD LAMM Table 1. Factors affecting fitness (from Thoday, 1955) 1. 2. 3.

4.

Genetic Stability (the stability of the genotype) Genetic Flexibility (the capacity of the genotype to vary) Phenotypic Flexibility (the capacity of the individual to adapt to local conditions), itself conveniently classified into Developmental Flexibility (capacity of the individual to adapt irreversibly to local conditions) and Behaviour Flexibility (capacity of the individual to adapt reversibly to local conditions); and Stability of the Environment

understand it, the story of Dobzhansky’s struggles with chromosomal variation is not solely about competing schools of thought within the selectionist camp (e.g., debates about variation-preserving selection), as articulated by Beatty, but also a story of competition between selectionist thinking and developmental perspectives of the sort we associate with Goldschmidt (Smocovitis, 1996; Dietrich, 2000b; Davis et al., 2009; Richmond, 2007). The ‘‘winning’’ style of thought was not at all evident as of 1962. An important manifestation of the developmental perspective is found in the role attributed to neutrality in Goldschmidt’s account, that I now turn to.

Bridging the Gap between Genotypic and Phenotypic Neutrality Neutrality may be the result of changes that are genetically causally inert (for example synonymous mutations) or of changes that are masked by higher levels of organization (e.g. by developmental canalization; backups genes). When addressed quantitatively two types of averaging complicate attempts to determine if a change is neutral. First, the average effect of multiple loci may mask the effects of each gene and, second, a neutral effect when averaging across different individuals is not conclusive evidence that the genetic change has no causal effect on fitness. You have to experimentally control for both these possibilities which is difficult to do or resort to mechanistic understanding of the processes in which the genes are involved. Goldschmidt consistently expressed the view that the low penetrance and expressivity of major mutations is important for the spread of mutations that may be of low viability, and may be part of the process of preadaptation (Goldschmidt, 1946, 1952, pp. 99–100, 102). Accordingly, Goldschmidt claimed that ‘‘the evolutionary process begins with mutants of low penetrance and expressivity’’ (Goldschmidt, 1946, p. 316). This line of argument already appeared in his discussion of chromosomal inversions in MBE, and again in Goldschmidt’s analysis

SYSTEMS THINKING VERSUS POPULATION THINKING

of homeotic mutants after he stopped using inversions as the primary evidence for his views. Goldschmidt stated that his view about the role of low penetrance mutants was in agreement with Wright’s views, and thus he could be understood as simply talking about the spread of such quasi-neutral mutants via drift. Dietrich discusses the scientific interaction between Goldschmidt and Wright in the mid-1940s and notes the in 1950 Wright included macromutations in his shifting balance theory, to Goldschmidt’s delight (Dietrich, 2000b, 2011). During these years, Wright was also a close collaborator of Dobzhansky’s (Lewontin et al., 1981). Significantly, Goldschmidt stressed that the harmonious integration of mutations into the developmental system was the result of developmental capacities of the organism, not of the selective accumulation of modifiers (Goldschmidt, 1952, pp. 99–100). He admitted that this process was as yet ‘‘mystical,’’ but nonetheless stressed its evolutionary importance. The developmental integration ‘‘relieves the evolutionary processes, in the case of macromutation, of a good deal of work which would [otherwise] be necessary’’ (ibid. p. 103). Such possibilities are among the things experimental studies of the sort done by Thoday (1955) were meant to address. For Goldschmidt, this was of utmost importance: ‘‘This is one of the reasons why I have tried for a long time to convince evolutionists that evolution is not only a statistical genetical problem but also one of the developmental potentialities of the organism’’ (loc. cit.) What Goldschmidt here called the ‘statistical genetical problem’, he simply called ‘the population problem’ in MBE, while his own concern was always with what he refers to here as ‘the developmental potentialities of the organism’; in MBE he characterized this, somewhat confusingly, as the ‘genetical problem’. Concluding Remarks It is sometimes remarked that Richard Goldschmidt had a grand system-building style characteristic of German thought (see Dietrich, 2000a; Richmond, 2007). He even described himself as rushing to place new facts within theoretical wholes due to his ‘‘genetic makeup’’ (Dietrich, 2011, p. 698). Goldschmidt’s renewed reputation is tied to his being championed by Stephen Jay Gould, who portrayed him as a heretic and emphasized Goldschmidt’s belief in evolutionary saltations over Goldschmidt’s other scientific convictions (ibid.) This characterization might suggest that Goldschmidt was different than his contemporaries in the way he privileged his theoretical commitments over the empirical data. I prefer to tell the story of Goldschmidt’s and

EHUD LAMM

Dobzhansky’s coping with the empirical results on inversions differently. Goldschmidt resisted the bifurcation of genetics into a science of transmission and a science of development, remaining loyal to a German physiological genetics tradition (see Richmond, 2007). In that, he was fighting a rearguard fight well after others conceded. However, both he and Dobzhansky struggled with the relationships between chromosomal organization, development and selection. My aim here was to highlight the integration of new evidence into these evolving conceptual schemes as well the similarities in their system-thinking. T. H. Morgan’s resistance to Mendelism, rooted in his preoccupation with how genes control development is somewhat similar (Gilbert, 1978) as are Johannsen’s ‘s holistic views regarding the genotype (Roll-Hansen, 2009, 2014a, b). The renewed interest in these reflects their relevance to the concerns of the so-called ‘‘post-genomic era’’ and probably also the salience of systems biology. John Beatty has argued that Dobzhansky’s struggle with the issue of variation, and his pro-drift views, were driven by his values, namely the benefits he wanted to attribute to variation, more than by matters of fact (Beatty, 1987a, b). I tried to show that a no less significant aspect of this history is the tension between population level, selectionist accounts, and a mechanistic, hierarchical and developmental perspective on genetic phenomena on all levels: the intra-chromosomal (e.g., Goldschmidt’s accounts of genic action), organismic (heterozygosity; Waddington’s model) and population levels. These commitments can be related to the Russian tradition, originating with Chetverikov. These levels were appealed to when thinking about the organization and evolution of genetic systems by several other prominent geneticists: Cyril Darlington, Kenneth Mather, Sewell Wright and Conrad Waddington. Throughout this paper I have used the term mechanistic thinking to capture something that is not identical with the mechanicism/organicism divide. My goal was to group together and emphasize the similarities among causal and mechanistic explanations on potentially multiple levels of the biological hierarchy, in particular those that are concerned with the internal structure of the system, and to distinguish them from statistical evolutionary analysis that focuses on global behavioral features. The former set of views need not be committed to parts being prior to the whole, or that organisms can be decomposed into independent sub-parts. I argued that the distinction between the two modes of explanation cuts across the divide between transmission genetics and physiological genetics. If true, this observation problematizes the use of these terms as historiographical (rather than historical) categories.

SYSTEMS THINKING VERSUS POPULATION THINKING

Evolutionary developmental biology (evo-devo) emphasizes the importance of the relation between selection and development, since it is development that determines the range of variability that is available for selection, and can thus influence evolutionary trajectories. This observation is at the base of Goldschmidt’s remarks about orthogenesis mentioned earlier. However, a more subtle aspect of the relation between development and selection is no less significant. It is not just that development determines the phenotypes that are subject to selection – the environment is fundamentally causally entangled in both. The selecting and inducing aspects of the environment may be one and the same and, more generally, development and selection are not two temporally distinct processes taking place one after the other. This insight is significant for the argument in Goldschmidt (1952) about the developmental integration of mutations, discussed in the previous section, as well as in the work of Lerner and Thoday discussed earlier in relation to Dobzhansky. Goldschmidt’s evolutionary account reflected a fundamental causal relationship between selection and development implicit in his views. His work on the effect of stress on phenotypes and on phenocopies (developmentally induced changes that mimic mutations), in particular, provided support for this observation and grounded his emphasis on the developmental integration of mutations. Goldschmidt’s distinction between the ‘genetical’ and the ‘population’ problems reflects a distinction between the genetic causality of the chromosomal pattern, that is the direct physiological or developmental effect of inversions, on the one hand, and any of the effects that depend on and reflect population dynamics, on the other. For Goldschmidt, a physiological geneticist, the ‘genetical problem’ dealt with the former set of issues. Goldschmidt, however, went further and wanted to replace the very notion of an atomic gene with a developmental account of the process by which chromosomal segments come to have a genic effect. Davis et al. (2009) correctly argued that development was not left out of the synthesis; rather the difficulty was with the way that it was assimilated in it. I would venture a similar claim about the way the gene and the chromosome were assimilated. Indeed, Goldschmidt’s work indicates how closely related, though still distinct, these issues were. They were both related to the causal structure of the theory. For Goldschmidt, not only selection and development were causally entangled so was the genic action of chromosome sections. I presented Goldschmidt’s theory of genic action in the context, first, of the state of empirical knowledge regarding the fitness effects of inversions, second, the theoretical models that were pursued at the time

EHUD LAMM

for explaining the spread of inversions, and third, Goldschmidt’s own theoretical commitments, and in particular his ambivalent views about selection and about the role of development in the spread of major mutations. Goldschmidt’s work is directly concerned with issues that were central to the discussion of chromosomes and chromosomal events, notably inversions. Goldschmidt’s arguments for rejecting the gene should be understood as part of the effort to understand the chromosome as an organized entity (see Huxley, 2010[1942], pp. 85–87), not only in the context of the theory of the gene.12 The same goes for Dobzhansky’s work on chromosomal variation in natural populations. The chromosome, as opposed to the atomic-particulate gene, could be conceptualized as an organized system, and in particular a developmental one. The border between issues relating to chromosomal dynamics and the broader aspects of developmental regulation, however, is not easily drawn. In addition to remaining loyal to his physiological genetics origins, Goldschmidt also exemplified a tradition of system thinking in classical genetics which is manifested just as clearly in Dobzhansky’s attack on the question of chromosomal inversions. Cyril Darlington’s Evolution of Genetic Systems, which Goldschmidt remarkably does not mention anywhere that I could find, with its emphasis on the necessary relationship between the organization of the karyotype and the mating system of the species, is as close to the ideal type of this approach I can think of. The system perspective, already incipient in Chetverikov, is also manifested in the views of Dobzhansky about inversions, in particular in his early views, in Mather’s analysis of internal and relational balance, and in the work of Lerner on genetic and developmental homeostasis. From early on, Mendelian populations were understood as integrated by mating strategies, and could thus have homeostatic properties such as the frequency of lethals.13 Through Sewell Wright, this perspective, championed as late as the 1950s, early 1960s, by Dobzhansky, had a close connection with group-selection and organismic views of populations found in the work of the ecologists Warder Clyde Allee and Alfred Emerson (see Mitman, 1992), though group selection was not an essential component of these accounts. Together, these analyses comprised several levels of biological organization: from that of the single chromosome, to the karyotype, the development of individuals and their mating strategies and up to population level dynamics (on Dobzhansky’s integration of diverse considerations see 12 13

Muller to Goldschmidt, 19 June, 1939. Sturtevant to Wright, March 1936. Sewall Wright Papers, APS.

SYSTEMS THINKING VERSUS POPULATION THINKING

Provine, 1981, pp. 75–76). Lewontin drove this point home when he noted that Dobzhansky consistently worked at the highest available genetic resolution, yet that the genetic load controversy that fueled a large part of his later career would not be resolved until the relation between gene and organism is understood (Lewontin, 1981, pp. 94, 115). Crow (1987) made a similar point when he noted that the problem of genetic variability has by 1987 become a question of ‘‘distinguishing between the roles of chance and deterministic factors and trying to find what function, if any, is played by the enormous amounts of DNA not involved in coding for proteins.’’ The Neo-Darwinian model ‘‘flattened’’ the hierarchical-systemic approach that fueled the work discussed in this paper and marginalized it; organicist theories, another case of system-thinking, that had been highly influential the first half of the century met a similar fate (Gibson, 2013). The focus on controversies such as the classical-balance debate and Goldschmidt’s repudiation of the gene may make it harder to appreciate this transition. And yet, there is a line connecting the early work on conceptualizing the chromosome that I discussed and current work on elucidating and conceptualizing genomic function and mechanisms (i.e., chromatin dynamics; see Lamm, 2014). Looking at this tradition from outside the confining perspective of the history of genetics helps see its influence on the Synthesis, and quite possibly on post-Synthesis developments.

Acknowledgments I thank Oren Harman and Sara Schwartz for comments on an earlier version of the paper. I have benefitted tremendously from discussing these ideas with John Beatty and Michael Dietrich. Lisa Gannett very kindly shared research material and her analysis.

References Beatty, John. 1987a. ‘‘Dobzhansky and Drift: Facts, Values, and Chance in Evolutionary Biology.’’ L. Kru¨ger (ed.), The Probabilistic Revolution. Cambridge, MA: MIT Press, pp. 271–311. ——. 1987b. ‘‘Weighing the Risks: Stalemate in the Classical/Balance Controversy.’’ Journal of the History of Biology 20: 289–319.

EHUD LAMM

——. 1994. ‘‘Dobzhansky and the Biology of Democracy: The Moral and Political Significance of Genetic Variation.’’ M. Adams (ed.), The Evolution of Theodosius Dobzhansky. Princeton: Princeton University Press, pp. 195–218. Brncic, Danko. 1954. ‘‘Heterosis and the Integration of the Genotype in Geographic Populations of Drosophila Pseudoobscura.’’ Genetics 39(1): 77–88. Carlson, Elof Axel. 1966. The Gene: A Critical History. Philadelphia: W.B. Saunders. Carson, Hampton L. 1980. ‘‘Cytogenetics and the Neo-Darwinian Synthesis.’’ Ernst Mayr and William Provine (eds.), The Evolutionary Synthesis. Cambridge: Cambridge University Press. Chetverikov, Sergey S. 1961. ‘‘On Certain Aspects of the Evolutionary Process from the Standpoint of Modern Genetics.’’ Proceedings of the American Philosophical Society 105(2): 167–195. Crow, James. 1987. ‘‘Muller, and Overdominance.’’ Journal of the History of Biology 20: 351–380. Darlington, Cyril Dean and Mather, Kenneth. 1950. The Elements of Genetics. London: Allen & Unwin. Davis, Gregory, Dietrich, Michael, and Jacobs, David. 2009. ‘‘Homeotic Mutants and the Assimilation of Developmental Genetics into the Evolutionary Synthesis.’’ Joe Cain and Michael Ruse (eds.), Descended from Darwin: Insights into American Evolutionary Studies, 1900–1970. Philadelphia: American Philosophical Society. Dietrich, Michael R. 2000a. ‘‘From Gene to Genetic Hierarchy: Richard Goldschmidt and the Problem of the Gene.’’ P. Beurton, R. Falk, and H. Rheinberger (eds.), The Concept of the Gene in Development and Evolution: Historical and Epistemological Perspectives. Cambridge: Cambridge University Press, pp. 91–114. ——. 2000b. ‘‘From Hopeful Monsters to Homeotic Effects: Richard Goldschmidt’s Integration of Development, Evolution, and Genetics.’’ American Zoologist 40(5): 738–747. doi:10.1093/icb/40.5.738. ——. 2008. ‘‘Striking the Hornet’s Nest: Richard Goldschmidt’s Rejection of the Particulate Gene.’’ O. Harman and M.R. Dietrich (eds.), Rebels, Mavericks, and Heretics in Biology. New Haven, CT: Yale University Press. ——. 2011. ‘‘Reinventing Richard Goldschmidt: Reputation, Memory, and Biography.’’ Journal of the History of Biology. doi:10.1007/s10739-011-9271-y. Dobzhansky, Theodosius. 1932. ‘‘The Baroid Mutation in Drosophila Melanogaster.’’ Genetics 17(4): 369–392. Dobzhansky, Th. 1935. ‘‘The Y Chromosome of Drosophila Pseudoobscura.’’ Genetics 20(4): 366–76. ——. 1937. Genetics and the Origin of Species. New York: Columbia University Press. ——. 1939. ‘‘Genetics of Natural Populations IX. Temporal Changes in the Composition of Populations of Drosophila pseudoobscura.’’ Genetics 28(2): 162. ——. 1950. ‘‘Genetics of Natural Populations XIX. Origin of Heterosis through Natural Selection in Populations of Drosophila Pseudoobscura to Natural Selection.’’ Genetics 35(3): 288. ——. 1951. Genetics and the Origin of Species, 3rd ed. New York: Columbia University Press. ——. 1952. ‘‘Nature and Origin of Heterosis.’’ J. Gowen (ed.), Heterosis. Ames, IA: Iowa State College Press, pp. 218–223. ——. 1953. ‘‘Some New Trends in Population Genetics and in Evolutionary Studies.’’ A.A. Buzzati-Traverso (ed.), International Union of Biological Sciences Symposium on Genetics and Population Structure. Pavia: IUBS, pp. 95–97.

SYSTEMS THINKING VERSUS POPULATION THINKING

——. 1955. ‘‘A Review of Some Fundamental Concepts and Problems of Population Genetics.’’ Cold Spring Harbor Symposia on Quantitative Biology 20: 1–15. ——. 1962. Mankind Evolving. New Haven: Yale University Press. Dobzhansky, Theodosius and Levene, Howard. 1955. ‘‘Genetics of Natural Populations. XXIV. Developmental Homeostasis in Natural Populations of Drosophila pseudoobscura.’’ Genetics 40(6): 797–808. Dobzhansky, Theodosius and Sturtevant, Alfred H. 1938. ‘‘Inversions in the Chromosomes of Drosophila pseudoobscura.’’ Genetics 23(1): 28–64. Dubinin, Nikolai P. and Tiniakov, G.G. 1945. ‘‘Seasonal Cycles and the Concentration of Inversions in Populations of Drosophila funebris.’’ American Naturalist 79: 570– 572. ——. 1946a. ‘‘Inversion Gradients and Natural Selection in Ecological Races of Drosophila funebris.’’ Genetics 31(6): 537. ——. 1946b. ‘‘Natural Selection and Chromosomal Variability in Populations of Drosophila funebris.’’ Journal of Heredity 37(2): 39–44. Felsenstein, Joseph. 1974. ‘‘The Evolutionary Advantage of Recombination.’’ Genetics 78(2): 737–756. Fisher, Ronald A. 1930. The Genetical Theory of Natural Selection. Oxford: Clarendon Press. Gannett, Lisa. 2013. ‘‘Theodosius Dobzhansky and the Genetic Race Concept.’’ Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences 44(3): 250–261. Gibson, Abraham H. 2013. ‘‘Edward O. Wilson and the Organicist Tradition.’’ Journal of the History of Biology 46(4): 599–630. Gilbert, Scott F. 1978. ‘‘The Embryological Origins of the Gene Theory.’’ Journal of the History of Biology 11(2): 307–351. doi:10.1007/BF00389303. Goldschmidt, Richard. 1933. ‘‘Some Aspects of Evolution.’’ Science 78(2033): 539–547. ——. 1937a, ‘‘Spontaneous Chromatin Rearrangements and Theory of Gene,’’ Proceedings of the National Academy of Sciences USA, 23: 621–623. ——. 1937b, ‘‘Spontaneous Chromatin Rearrangements in Drosophila,’’ Nature, 140(3548): 767. ——. 1940. The Material Basis of Evolution. New Haven, CT: Yale University Press. ——. 1938. Physiological Genetics. New York: McGraw-Hill. ——. 1946. ‘‘An Empirical Evolutionary Generalization Viewed from the Standpoint of Phenogenetics.’’ American Naturalist 80(792): 305–317. ——. 1952. ‘‘Homoeotic Mutants and Evolution.’’ Acta Biotheoretica 10(1–2): 87–104. ——. 1954. ‘‘Different Philosophies of Genetics.’’ Science 119(3099): 703–710. ——. 1955. Theoretical Genetics. Berkeley: University of California Press. ——. 1960. In and Out of the Ivory Tower: The Autobiography of Richard B. Goldschmidt. Seattle: University of Washington. Hall, Brian K. 2005. ‘‘Fifty Years Later: I. Michael Lerner’s Genetic Homeostasis (1954)—A Valiant Attempt to Integrate Genes, Organisms and Environment.’’ Journal of Experimental Zoology Part B: Molecular and Developmental Evolution 304B(3): 187–197. Harman, Oren S. 2004. The Man Who Invented the Chromosome: A Life of Cyril Darlington. Cambridge, MA: Harvard University Press. Huxley, Julian. 2010[1942]. Evolution: The Modern Synthesis. Cambridge, MA: MIT Press.

EHUD LAMM

Ives, Philip Truman. 1947. ‘‘Second Chromosome Inversions in Wild Populations of Drosophila Melanogaster.’’ Evolution 1(1/2): 42–47. doi:10.2307/2405402. Kohler, R.E. 1994. Lords of the Fly: Drosophila Genetics and the Experimental Life. Chicago: University of Chicago Press. Lamm, Ehud. 2014. ‘‘The Genome as a Developmental Organ.’’ The Journal of Physiology 592(11): 2283–2293. Lerner, I. Michael. 1954. Genetic Homeostasis. London: Oliver and Boyd. Lewontin, Richard C. 1981. ‘‘The Scientific Work of Th. Dobzhansky.’’ R.C. Lewontin, John A. Moore, William B. Provine, and Bruce Wallace (eds.), Dobzhansky’s Genetics of Natural Populations I–XLIII. New York: Columbia University Press. Lewontin, Richard C., Moore, John A., Provine, William B., and Wallace, B. (eds.). 1981. Dobzhansky’s Genetics of Natural Populations I–XLIII. New York: Columbia University Press. Mather, Kenneth. 1943. ‘‘Polygenic Inheritance and Natural Selection.’’ Biological Reviews 18(1): 32–64. doi:10.1111/j.1469-185X.1943.tb00287.x. Mather, Kenneth. 1955. ‘‘Polymorphism as an Outcome of Disruptive Selection.’’ Evolution 9(1): 52–61. Mitman, Greg. 1992. Ecology, Community, and American Social Thought, 1900–1950. Chicago: University of Chicago Press. Muller, Herman J. 1932. ‘‘Some Genetic Aspects of Sex.’’ The American Naturalist 66(703): 118–138. Provine, William B. 1981. ‘‘Origin of the GNP Series.’’ R.C. Lewontin, John A. Moore, William B. Provine, and Bruce Wallace (eds.), Dobzhansky’s Genetics of Natural Populations I–XLIII. New York: Columbia University Press. Painter, Theophilus S. 1934. ‘‘Salivary Chromosomes and the Attack on the Gene.’’ Journal of Heredity 25(12): 465–476. Richmond, Marsha L. 2007. ‘‘The Cell as the Basis for Heredity, Development, and Evolution: Richard Goldschmidt’s Program of Physiological Genetics.’’ M.D. Laubichler and J. Maienschein (eds.), From Embryology to Evo-Devo. Cambridge MA: MIT Press, pp. 169–212. Roll-Hansen, Nils. 2009. ‘‘Sources of Wilhelm Johannsen’s Genotype Theory.’’ Journal of the History of Biology 42(3): 457–493. ——. 2014a. ‘‘Commentary: Wilhelm Johannsen and the Problem of Heredity at the Turn of the 19th Century.’’ International Journal of Epidemiology 43(4): 1007–1013. ——. 2014b. ‘‘The Holist Tradition in Twentieth Century Genetics. Wilhelm Johannsen’s Genotype Concept.’’ The Journal of Physiology 592(11): 2431–2438. Sapp, Jan. 1987. Beyond the Gene: Cytoplasmic Inheritance and the Struggle for Authority in Genetics. New York: Oxford University Press. Smocovitis, V.B. 1996. Unifying Biology: The Evolutionary Synthesis and Evolutionary Biology. Princeton: Princeton University Press. Sturtevant, Alfred H. 1917. ‘‘Genetic Factors Affecting the Strength of Linkage in Drosophila.’’ Proceedings of the National Academy of Sciences of the United States of America 3(9): 555. Sturtevant, Alfred H. and Mather, Kenneth. 1938. ‘‘The Interrelations of Inversions, Heterosis and Recombination.’’ The American Naturalist 72(742): 447–452. Thoday, John M. 1955. ‘‘Balance, Heterozygosity and Developmental Stability.’’ Cold Spring Harbor Symposia on Quantitative Biology 20: 318–326.

SYSTEMS THINKING VERSUS POPULATION THINKING

Vetukhiv, Mikhail. 1953. ‘‘Viability of Hybrids between Local Populations of Drosophila Pseudoobscura.’’ Proceedings of the National Academy of Sciences of the United States of America 39(1): 30. Waddington, Conrad H. 1941. ‘‘Evolution of Developmental Systems.’’ Nature 147(3717): 108–110. ——. 1961. The Nature of Life. London: Allen & Unwin. Wallace, Bruce. 1994. ‘‘Theodosius Dobzhansky Remembered: Genetic Coadaptation.’’ M Adams (ed.), The Evolution of Theodosius Dobzhansky. Princeton, Princeton University Press, pp. 155–162. Wright, Sewall. 1941. ‘‘The Material Basis of Evolution.’’ The Scientific Monthly 53(2): 165–170. Wright, Sewall and Dobzhansky, Theodosius. 1946. ‘‘Genetics of Natural Populations. XII. Experimental Reproduction of Some of the Changes Caused by Natural Selection in Certain Populations of Drosophila pseudoobscura.’’ Genetics 31(2): 12.