Evolution of a Polyphenism by Genetic Accommodation Yuichiro Suzuki* and H. Frederik Nijhout Polyphenisms are adaptations in which a genome is associated with discrete alternative phenotypes in different environments. Little is known about the mechanism by which polyphenisms originate. We show that a mutation in the juvenile hormone-regulatory pathway in Manduca sexta enables heat stress to reveal a hidden reaction norm of larval coloration. Selection for increased color change in response to heat stress resulted in the evolution of a larval color polyphenism and a corresponding change in hormonal titers through genetic accommodation. Evidently, mechanisms that regulate developmental hormones can mask genetic variation and act as evolutionary capacitors, facilitating the origin of novel adaptive phenotypes. P olyphenisms, such as the castes of social insects, the solitary and gregarious phases of migratory locusts, and the winged and wingless forms of aphids, are evolved adap- tations to a varying environment (1–3). The adaptive importance of polyphenisms has been demonstrated in many cases, and many studies have shown that the threshold for the switch between alternative phenotypes can evolve in response to external selective pressures (4–9). Although much work has been done on the evolutionary maintenance of polyphenisms and the evolutionary shifts of polyphenic thresholds (10, 11), little is known about the evolutionary and developmental mechanism behind the ori- gin of these threshold traits. We tested the hypothesis that a polyphenism can evolve through genetic stabilization of a stress-induced phenotype, a process known as genetic assimilation (2). Because related species are likely to share genetic and devel- opmental backgrounds, we reasoned that ex- posing hidden genetic variation by stress (12) may allow us to evolve a polyphenic regula- tory mechanism in a monophenic species that shares a recent common ancestor with a poly- phenic species. We studied this possibility by evolving a larval color polyphenism in the to- bacco hornworm, Manduca sexta, a monophenic species with green larvae (13); a related species, M. quinquemaculata, exhibits a larval color polyphenism, developing a black phenotype at 20-C and a green phenotype at 28-C(14). Because thermal stress is commonly encoun- tered in the wild (15), we chose to use tem- perature stress to obtain phenocopies (16). Wild-type larval coloration was robust to thermal stress, with the fifth instar larva re- maining green after heat shock during the mid and late fourth larval instar. We also examined the effect of thermal stress in the black mutant line of M. sexta. The black mutation is a sex- linked recessive allele that reduces juvenile hormone (JH) secretion (17), which results in an increased melanization of the larval epi- dermis. The black mutant phenotype can be rescued by treatment with JH (17), yielding a normal green-colored larva. Larvae of the black mutant are black at physiologically tolerable temperatures ranging from 20-C to 28-C (fig. S2). Heat shocks during the sen- sitive period of the fourth larval instar gen- erated fifth instar larvae with colors that ranged from normal black to nearly normal green, with the majority showing a slight color change (Fig. 1 and fig. S2). The black strain was most sensitive to a 6-hour heat shock applied less than 8 hours before apolysis (the detachment of the epidermis from the cuticle, which is the first step in the molting process), at the molt from the fourth to the fifth larval instar (fig. S1). The diversity of heat shock–induced pheno- types provided us with a range of phenotypic variants upon which we could artificially select. We established two lines: one selected for in- creased greenness upon heat treatment (poly- phenic line), the other for decreased color change upon heat treatment (monophenic line). About 300 larvae were reared and heat-shocked every generation, and approximately 60 with the most desirable phenotypic response were selected to establish the subsequent generation. An unselected control line was heat-shocked every generation to monitor any change that was not a direct result of selection. The re- sponse to selection (Fig. 2A) shows that the induced color change is heritable. The varia- tion in the phenotype is continuous rather than discrete, which indicates that the induced color change is under polygenic control. The mono- phenic line lost its response to temperature shock after about the seventh generation of selection and remained black thereafter, with little phenotypic response to heat shock. The reaction norms of the three lines in the 13th generation are shown in Fig. 2B. The unselected control line has a narrow threshold between 30-C and 33-C, with the inflection point at 32.7-C. As a result of selection, two major evolutionary changes have taken place in the polyphenic line: (i) completely green coloration at lower temperatures of 28-C, not seen in the control line (fig. S2), and (ii) a threshold shift to a lower temperature with the inflection point at 28.5-C. The monophenic line remained black at all temperatures. Thus, selection resulted in the evolution of different phenotypes at different constant environmental temperatures (fig. S2) and changed the shape of the reaction norm (Fig. 2B) so that the response to a small temperature change in the transition region became more discrete, or switchlike. The time of the sensitive period for heat shock corresponds to the time of the JH- sensitive period for epidermal color determi- nation (17, 18). Topical application of JH to unselected black mutant during this sensitive period reverses the black phenotype to the green wild-type color. Dopa decarboxylase (DDC), the enzyme that converts dopa to dopa- mine in the melanin synthesis pathway, is first synthesized about 16 hours after this sensitive period (19), which indicates that heat shock Department of Biology, Duke University, Durham, NC 27708, USA. *To whom correspondence should be addressed. E-mail: ys16@duke.edu Fig. 1. The range of lar- val coloration observed in the heat-shocked lar- vae of the black mutant. The numbers below rep- resent the scoring sys- tem used to quantify the color change: 0 is completely black, and 4 is completely green. Non–heat-shocked black mutant and non–heat- shocked wild-type lar- vae of M. sexta have the phenotypic scores of 0 and 4, respectively. REPORTS 3 FEBRUARY 2006 VOL 311 SCIENCE www.sciencemag.org 650