Butterflies and moths are a large group of insects called Lepidoptera. Lepidopteran insects are characterized by wings covered with scales and bristles. These scales are variously colored, and a single scale serves as an image unit (or “pixel”). These scales form diverse mosaic color patterns on wings. One group of butterflies that shows highly diverse color patterns is the family Nymphalidae, from which a common overall color pattern was derived as the nymphalid groundplan (Nijhout 1978, 1991, 2001; Otaki 2009, 2012a; Taira et al. 2015). The nymphalid groundplan is composed of three major symmetry systems (the border, central, and basal symmetry systems) and two peripheral systems (wing root and marginal systems), and all five systems are thought to be produced based on the same mechanism (Otaki 2012a; Taira et al. 2015). A unit of a symmetry system in a single wing compartment is composed of a single core element and a pair of paracore elements located at the distal and proximal sides of the core element (Otaki 2012a).
Among the symmetry systems, the border symmetry system is probably the most conspicuous in many nymphalid butterflies. It is composed of a border ocellus (an eyespot) as a core element and a pair of parafocal elements (distal and proximal parafocal elements) as paracore elements (Nijhout 1991, 2001; Dhungel and Otaki 2009; Otaki 2009, 2012a). Moving from the center to the peripheral area, a typical eyespot is composed of a white focal spot at the center (often called a “focus”), an inner black disk, a light-colored ring, and an outer black ring. A typical eyespot can be found in the African satyrine butterfly, Bicyclus anynana, one of the most popular species in butterfly biology (Beldade and Brakefield 2002; Carroll et al. 2004). Physical damage at the prospective eyespot focus in Junonia coenia (Nijhout 1980a, 1991), B. anynana (French and Brakefield 1992), Ypthima argus (Otaki et al. 2005a), Junonia orithya (Otaki et al. 2005a), and Junonia almana (Otaki 2011a), together with transplantation experiments (Nijhout 1980a, 1991; French and Brakefield 1995; Brakefield et al. 1996; Beldade et al. 2008), demonstrated that the center of the prospective eyespot behaves as an organizing center for the eyespot during the pupal stage. However, actual eyespots are highly diverse, and various deformations from the typical eyespot pattern occur (Nijhout 1990, 1991; Otaki 2011b). For example, the white focal spot is often missing, and the various rings are often distorted differently in a single eyespot.
Since the last decade of the twentieth century, many candidate genes that could specify eyespots have been identified based on their expression patterns (Carroll et al. 1994; Brakefield et al. 1996; Keys et al. 1999; Brunetti et al. 2001; Reed and Serfas 2004; Monteiro et al. 2006; Saenko et al. 2011; Tong et al. 2012). These genes are expressed during the late larval to the early pupal stages in the wing tissues, which is when the color pattern determination takes place (Nijhout 1980a). Among them, the most notable gene is probably Distal-less (Dll). It has been shown that Dll expression recapitulates the locations of organizing centers that were predicted by a reaction–diffusion model (Carroll et al. 1994; Nijhout 1990, 1991, 1994, 1996), which has often been interpreted as meaning that Dll expression defines an organizing center and that Dll is a master gene for eyespot determination. In addition to the eyespot focal determination, it has also been suggested that Dll determines eyespot size (Brakefield et al. 1996; Beldade et al. 2002).
However, functional tests for Dll were not performed until recently. One study using transgenic B. anynana butterflies showed that Dll plays a role in eyespot size regulation as well as in black spot induction (Monteiro et al. 2013). One study using the blue pansy butterfly, J. orithya, together with a novel surgical technique, showed a weak correlation of the individual Dll expression level with the individual eyespot size (Adhikari and Otaki 2016). However, sexually dimorphic eyespot size in this species (i.e., female eyespots are larger than male ones) cannot be explained by the Dll expression levels; female forewings have lower Dll levels than male ones (Adhikari and Otaki 2016). Subsequently, using J. orithya with a baculovirus gene transfer method (Dhungel et al. 2013), it has been shown that Dll can induce fragmentary patterns of an eyespot but not an entire eyespot (Dhungel et al. 2016). More elegantly, Dll deletion using genome editing has produced a deformation of eyespot, an increase of eyespot number and size, and an emergence of dark patches in Vanessa cardui and J. coenia, suggesting a role of Dll in eyespot repression (Zhang and Reed 2016). Taken together, although Dll is unlikely to be sufficient for the entire eyespot pattern formation, it plays an important role in eyespot development.
Morphological studies also advanced. Butterfly wings exhibit coordinated scale size distributions in addition to coordinated scale color distributions (Kusaba and Otaki 2009; Dhungel and Otaki 2013; Iwata and Otaki 2016). The largest scales in an eyespot are often at the central area in J. orithya (Kusaba and Otaki 2009) and J. almana (Iwata and Otaki 2016). This finding, together with the observation that scale size is proportional to the size of scale-building cells (Henke 1946; Sondhi 1963), led us to propose the ploidy hypothesis that morphogen signals for color patterns are identical to ploidy signals (Iwata and Otaki 2016).
Additionally, the pupal surface has cuticle focal spots that correspond to adult eyespots in various butterfly species (Nijhout 1980a; Otaki et al. 2005a). Two Junonia species that have large eyespots in adult wings, J. orithya and J. almana, indeed have large and distinct pupal cuticle focal spots, whereas a Junonia species that has small eyespots in adult wings, J. hedonia, has small ones (Taira and Otaki 2016). Interestingly, the size of the cuticle spot is correlated with the size of the corresponding eyespots in J. orithya and Y. argus (Otaki et al. 2005a). Similar correlations were also obtained among serial eyespots on a single wing in J. orithya (Taira and Otaki 2016). The three-dimensional structures of pupal cuticle focal spots as well as adult wings were revealed recently (Taira and Otaki 2016).
Moreover, physiologically induced changes of color patterns, which are typically considered positional and morphological changes of elements, have been investigated in detail (Nijhout 1984; Otaki 1998, 2007, 2008a, b; Otaki and Yamamoto 2004a, b; Serfas and Carroll 2005; Otaki et al. 2005b, 2010; Mahdi et al. 2010, 2011; Hiyama et al. 2012). Meanwhile, an invention of a real-time in vivo observation system made it possible to record how wing tissues develop inside the pupal case (Iwata et al. 2014). Developing epithelial cells are elongated vertically as well as horizontally (Ohno and Otaki 2015a), confirming a century-old histological study (Mayer 1896). Long-range slow calcium waves have been discovered in pupal wing tissues, which may function as signals to coordinate development throughout a wing (Ohno and Otaki 2015b).
This information should collectively evaluate the feasibility of mechanistic models for color pattern determination. Historically, morphogen gradient models have been proposed and used to explain various experimental results (Nijhout 1978, 1980a, 1981, 1990, 1991; French and Brakefield 1992, 1995; Brakefield and French 1995; Monteiro et al. 2001; Serfas and Carroll 2005; Otaki 2008a). Nijhout (1990) examined the diverse eyespot patterns of nymphalid butterflies and identified 36 pattern categories, which were used to construct a gradient-based model. These models are based on the simple diffusion of a putative morphogen that forms a gradient, together with differentiation thresholds inherently programmed into immature scale cells. Abrupt changes of the cellular interpretation of a smooth gradient were attained mathematically by a sigmoidal curve, resulting in two thresholds and three colors (Nijhout 1991).
However, Otaki (2011b, c) pointed out several difficulties of the gradient models to explain actual butterfly wing color patterns. For example, an “archetypical” butterfly eyespot is likely composed of a series of repetitions of an inductive signal for black (or dark) area (Otaki 2011c). In other words, a non-black (i.e., light-colored) area between the black areas is equivalent to background (Otaki 2011c). This binary rule (stating that a series of repetitions of dark areas with light-area intervals is the basic expression of an eyespot) alone makes threshold-based diffusion models unrealistic because the black rings or disks are equivalent to each other in actual butterflies. Moreover, not just two but three or more repetitive black rings are observed in many butterflies (Otaki 2011b). Indeed, one of the “black rings” of an eyespot is a pair of discontinued elements called parafocal elements (Otaki 2009, 2011c, 2012a, b). Moreover, color pattern analysis of neighboring or serial eyespots with different structures on the same wing surface pointed out that thresholds for gradient interpretation, if exist according to the gradient models, do not vary among neighboring compartments and that these eyespots should be produced by different levels of a morphogen to reflect their morphological differences (Otaki 2011b). But it is theoretically difficult to satisfy these two points simultaneously in gradient models (Otaki 2011b). In fact, the dynamic responses of eyespots to physical damage requires flexible models that can accommodate signals from damage sites and from neighboring organizing centers (Otaki 2011a).
As an alternative model, the induction model has been proposed (Otaki 2011b, c, 2012b). The induction model is based on many case analyses of normal and experimentally induced color patterns (Otaki 2011b, c, 2012b), incorporating the principle of “short-range activation and long-range inhibition” that have been found in many biological patterns (Gierer and Meinhardt 1972; Meinhardt 1982; Meinhardt and Gierer 1974, 2000).
In either model, the status of the white focal area has not been explained well in the literature. Nijhout (1978, 1980a) proposed that a “focus” at the center of an eyespot releases a morphogen at the late larval and early pupal stages, based on which a gradient model was formulated. Since then, one tends to assume that the white focal spot directly corresponds to an organizing center for the entire eyespot. In many instances, this assumption seems to be valid; a white spot is located at the physical centers of eyespots in many nymphalid butterflies. However, this is not always the case. Nijhout (1980a) indeed pointed out that the white scales at the eyespot center do not precisely correspond to the “focus”. Likewise, there is a discrepancy between the location of the largest scales and the location of the white spots in a particular eyespot of J. almana (Iwata and Otaki 2016). Similar cases have been pointed out in Calisto herophile and other butterflies (Iwata and Otaki 2016). Moreover, the white coloration is structural rather than pigment-based (Nijhout 1980b, 1991; Iwata and Otaki 2016). In a gradient model, the area of the highest morphogen concentration above a certain threshold is supposed to become the white spot. But molecular pathways for structural color production are probably qualitatively different from those for pigment-based color production. Thus, one could think that these two production lines may be distinctly specified. In any case, the relationships between white spots and their corresponding eyespot bodies (defined as all the eyespot portions except white spots) should be clarified to understand how butterfly eyespots are constructed during development.
In this paper, we ask if white spots behave independently of eyespot bodies. We hypothesized that if uncoupling of white spots is mechanistically possible, some species of nymphalid butterflies show uncoupling color patterns naturally. More concretely, we hypothesized that it may be possible to observe white spots that are not located at the center of an eyespot in nymphalid butterfly wings and that such uncoupling behavior may be shown by morphometric analysis. Here, we focus on Calisto butterflies to test this hypothesis.
Lepidopterists in Asian (and probably in many other) countries are not familiar with the genus Calisto because they are endemic to the West Indian regions (mainly in Hispaniola, which is occupied by Haiti and Dominican Republic). Indeed, Calisto-type pear-shaped eyespot patterns were not incorporated in the pattern analysis of Nijhout (1990). However, we had an opportunity to examine specimens of Calisto butterflies. The genus Calisto is an exclusive group of satyrine butterflies in the West Indies that constitutes more than 40 species (Smith et al. 1994; Miller and Miller 2001; Askew and Stafford 2008). Among them, we here focused on eyespots of Calisto tasajera González, Schwartz & Wetherbee 1991 (González et al. 1991; Hedges and Johnson 1994) because it has unique pear-shaped eyespots that have two or more white “focal” spots. Molecular phylogenetic analysis and historical biogeography of Calisto have been reported (Sourakov and Zakharov 2011; Matos-Maraví et al. 2014). We also examined eyespots of other nymphalid butterflies to support our findings with C. tasajera. The present study argues for an uncoupling of white spots from the rest of the eyespots (i.e., eyespot bodies).