METHODS, MODELS, METAPHORICAL AIDS

Background

The ideas in this chapter on genetic control of wing pattern have emerged from 25 years of maintaining Heliconius populations in insectaries and there observing the products of semi-natural or "synthetic" hybrid zones during the era when molecular genetic mechanisms of pattern formation were first being revealed. Ideas presented later about the ecology and evolution of diversity were shaped by my own three decades of studying Heliconius populations in their community context and by my exposure over these years to the ideas and findings of other Heliconius workers who have been integrating behavioral and physiological ecology, ecological and evolutionary genetics, and phylogenetics to a degree unprecedented for any genus of animals. I have focused on one group of Heliconius, the MCS clade, because this group hybridizes most frequently in culture and nature (see Mallet et al. 1999).

During population studies of Costa Rican Heliconius communities in the early 1980's I observed that the Heliconius community shares nearly identical color-pattern themes on both sides of the montane barrier which divides that country, with one striking exception (Figure 1, middle): the wing coloration of Müllerian partners that fly in the forest understory vary from iridescent blue/black with broad white forewing patches (H. sapho, H. cydno galanthus) on the Atlantic side, to flat black with bold yellow forewing and hindwing bars (H. hewitsoni, H. pachinus) on the Pacific side. The two pupal-mating species (H. sapho and H. hewitsoni) are specialists on Passiflora pittieri and possess virtually identical life histories and larval and pupal morphology. Nevertheless, these prove to be good species: they have different chromosome numbers: 56 vs. 21 (Brown et al. 1992) and matings between them in greenhouse cultures fail to produce offspring (personal observation).

In contrast, crosses between H. cydno galanthus and H. pachinus are readily obtained. Both are hostplant or oviposition generalists using species in the genus Passiflora and have similar larval and pupal morphology. Nevertheless, it is not difficult to understand why taxonomists traditionally separated these populations into different species: no hybrid zones have been previously recognized in the (ca. 50 km wide) zone of potential contact centered on the Meseta Central of Costa Rica. However, it is now clear that a hybrid zone has been present, but not perceived, since many hybrid individuals, due of the nature of wing pattern genetics in this group, display rather novel phenotypes as accounts to follow will illustrate. Moreover, H. pachinus is identical to both H. cydno galanthus and Colombian H. c. weymeri in terms of rDNA restriction sites (Lee et. al. 1992), it groups within a clade of cydno races in terms of mtDNA (Brower 1996b), and it is fully interfertile with H. cydno galanthus or other H. cydno (personal observations). Based on these observations H. pachinus is assumed for the present to represent a well-differentiated race of H. cydno, approaching but not broaching the race/species boundary described for Heliconius by Mallet et al. (1999).

While conducting genetic crosses and maintaining cydno x pachinus offspring in insectaries with other Heliconius species, I obtained unexpected matings between H. melpomene males and H. cydno group females. F1 females of this cross are sterile in keeping with Haldane's Rule, but males are fertile and can be backcrossed to pure females of both species. By repeatedly ("iteratively") backcrossing successive generations of hybrid males showing a particular trait of one species (e.g., red FW band of melpomene) to pure females of the other species it was possible eventually to obtain hybrid males and females showing the heterospecific trait which were also fertile. I term broods from the crossing of such individuals "pseudo F2" broods since they provide a look at some of the phenotypes one might find in a true F2 cross between two species.

By this method I discovered the expression of the melpomene gene for FW red to be useful to test hypotheses developed to explain certain results of cydno X cpachinus crosses. To account for the observed interaction of pattern genes in these three phenotypically distinct taxa (Figure 4 top left: a, b, c), it was necessary to develop the concept of a shared wing pattern "tool box." This includes those pattern genes which behave essentially the same in both cydno and melpomene genomes.

Little of the genetic data underlying the "tool box" model of pattern control to be outlined in the pages to follow has been published (but see Nijhout et al. 1990). However, broods from experimental crosses have been and continue to be made accessible to interested researchers working on Heliconius genetics and butterfly wing pattern development (see Nijhout 1991). For present purposes, I refer only to robust conclusions secured by the most definitive results that I illustrate with photos of representative critical broods. Because of logistic and time constraints, I could not rear every potentially interesting brood arising from all the unique and potentially informative hybridizations that have occurred in my insectaries. Rather, I used greenhouse "hybrid zones" created with known founders to prospect for any phenotypes whose mere appearance would force a re-thinking of a current paradigm, or conversely, for rare recombinants predicted by the model. Subsequent test crosses then were focused on the most interesting phenotypes and presumptive genotypes. Finally, because obtaining large families from Heliconius females requires that they must be kept several months and fed pollen or appropriate amino acid diets (Gilbert, 1972), I often worked with the combined broods of several sisters mated to a single male or to several males of the same genotype, sacrificing details of minor individual variation for the ability to observe the behavior of major loci in larger communal families.

For several reasons, including space constraints, I have not attempted to relate my genetic hypotheses to gene classifications of prior Heliconius studies or to utilize their terminology (although ultimately that will be an informative and important exercise). A glance at the terminology of Sheppard et al's (1985) epic work on Heliconius demonstrates why only the hardcore of specialists can process and utilize the results or conclusions of the largest analysis of Heliconius genetics ever undertaken. Likewise, although subsequent works more specifically on races of the H. cydno group (e.g., Nijhout et al. 1990, Linares 1996, Kapan 1998) developed terminology in the context of knowledge of some of my own work on the MCS tool box elements and would be more compatible with the scheme I use in this chapter, I do not attempt to show the correspondence between these and the hypotheses I present here. Nijhout (1991) in his synthesis, provides the most valuable summary of Heliconius wing color and pattern genes to date. I focus on describing a few important wing pattern ground-plan elements along with those few genes whose effects and interactions are recognizable across races and species of the MCS clade as: (1) homologous and (2) significant in determining major aspects wing pattern phenotype.

Early in this work I attempted to understand underlying mechanisms for the origin of novel patterns in Heliconius in terms of general models of insect developmental genetics based on Drosophila . Although some of the early ideas from that literature are outmoded, they helped me organize hypotheses and suggested intuitive metaphors to explain the correspondence between genes and wing pattern which seemed to account for pattern variation in synthetic hybrid zones. Beyond the fact that Drosophila wing bristles and butterfly wing scales are now known to be homologous (Galant et al. 1998), I see heuristic value in retaining the Drosophila-inspired conceptual structure in thinking about how Heliconius may be different from other well-studied models of butterfly development.