Cryptic variation in butterfly eyespot development: the importance of sample size in gene expression studies Robert D. Reed, Ã Po-Hao Chen, and H. Frederik Nijhout Department of Biology, Duke University, Durham, NC 27708, USA Ã Author for correspondence (email: rreed@uci.edu) SUMMARY Previous studies have shown that development can be robust to variation in parameters such as the timing or level of gene expression. This leads to the prediction that natural populations should be able to host developmental variation that has little phenotypic effect. Cryptic variation is of particular interest because it can result in selectable phenotypes when ‘‘released’’ by environmental or genetic factors. Currently, however, we have little idea of how variation is distributed between genes or over time in pattern formation processes. Here we survey expression of Notch (N), Spalt (Sal), and Engrailed (En) during butterfly eyespot determination to better understand how pattern formation may vary within a population. We observed substantial heterochronic variance in the progress of spatial expression patterns for all three proteins, suggesting some degree of developmental buffering in eyespot development. Peak variance for different proteins was found at both early and late stages of development, contrasting with previous models suggesting that the distribution of variance should be more temporally focused during pattern formation. We speculate that our observations are representative of a standing reservoir of cryptic variation that may contribute to phenotypic evolution under certain circumstances. Our results also provide a strong cautionary message that gene expression studies with limited sample sizes can be positively misleading in terms of inferring expression pattern time series, as well as for making cross-species phylogenetic comparisons. INTRODUCTION Populations can harbor significant cryptic genetic variation that has little or no phenotypic effect under normal conditions (Gibson and Dworkin 2004; Dworkin 2005). Experimental work has shown that cryptic variation can produce selectable phenotypic variation under circumstances such as environ- mental change (Waddington 1953; Suzuki and Nijhout 2006), hybridization (Lauter and Doebley 2002), genetic perturb- ation (Rendel 1959; Polaczyk et al. 1998), and the comprom- ise of developmental buffering mechanisms (Rutherford and Lindquist 1998; Queltsch et al. 2002; Cowan and Lindquist 2005). In addition, cryptic variation in gene regulation can be used to make inferences about the mech- anisms underlying the robustness of developmental circuits (Houchmandzadeh et al. 2002; Horikawa et al. 2006). Despite its evolutionary and developmental significance, very little is known about the nature of cryptic variation. A certain por- tion of cryptic variation probably exists only as latent genetic information, having no effect on biological processes under normal conditions. Some cryptic variation, however, might manifest developmentally or physiologically, but simply not have an effect on phenotypes that is obvious or accessible to investigators. Recent theoretical and experimental insights into the ro- bustness of development to variation in gene expression and/ or protein distribution (von Dassow et al. 2000; Ho- uchmandzadeh et al. 2002; Meir et al. 2002; Lucchetta et al. 2005; Horikawa et al. 2006; Veitia and Nijhout 2006) lend support to the idea that some cryptic variation could be ex- pressed developmentally. To date, however, this hypothesis remains poorly tested because few direct measurements of intraspecific developmental variation have been made. To address this issue, and to better understand how a develop- mental process can vary in a population, we conducted a survey of spatiotemporal gene expression patterns associated with eyespot color pattern determination in a population of Junonia (Precis) coenia butterflies. Butterfly eyespot determination occurs in final-instar imag- inal wing disks when several signaling molecules and tran- scription factors are expressed in a group of cells that will become the eyespot center, or ‘‘focus’’ (Carroll et al. 1994; Brakefield et al. 1996; Keys et al. 1999; Reed and Serfas 2004). Shortly after pupation, the focal cells send out a diffusible signal that induces surrounding cells to express various tran- scription factors and take on the identity of color rings in the adult wing pattern (Nijhout 1980; Brunetti et al. 2001). The eyespot-associated proteins so far identified belong to devel- EVOLUTION & DEVELOPMENT 9:1, 2–9 (2007) & 2007 The Author(s) Journal compilation & 2007 Blackwell Publishing Ltd. 2