DISCUSSION AND CONCLUSIONS

In this chapter I have described a hypothesis of wing pattern genetics and mimetic evolution in Heliconius based on empirical observations of hybridization within the melpomene/cydno/silvaniform clade of Heliconius. These species apparently share a "tool box" for generating wing pattern which I intentionally described with minimal reference to other schemes for describing the genetics and wing pattern development in Heliconius. To relate, criticize, and harmonize this hypothesis with prior hypotheses and systems of gene nomenclature, while important, would require a much longer and potentially more confusing text than the present chapter. I placed my concept of pattern determination in Heliconius in terms of developmental genetic hypotheses based on Drosophila with two metaphorical aids. First, I compared major pattern determining zones of Heliconius wings to computer program windows to explain a possible way that regulatory gene hierarchies might operate independently in different wing compartments or lineage restriction zones. Second, I described them in terms of actual windows on walls and shutters on windows as an aid in visualizing such simplicity and elegance. Both of these analogies highlight the ease with which extensive qualitative pattern variety can arise through introgression.

Specific ways that the tool box/ introgression hypothesis assists in interpreting wing patterns of known species, races, and forms of the MCS clade were discussed in reference to particular wing surfaces. Therefore it is possible finally to return to the more general evolutionary questions, posed earlier, about Heliconius wing pattern.

1. Why are so many wing patterns available to Heliconius?

As I pointed out in the introduction, the focal question has been: "Given a given amount of available genetic variation, why does so much variety evolve?" While this is still a central problem which could occupy us for many more decades, in this chapter I shifted focus to "Why do Heliconius seem to have such a great supply of variation in the first place?" My own perspective began to shift in the early 1980's. Working with synthetic hybrid zones, I found that the sum total of patterns which exist over the entire geographical range of the MCS clade are but a fraction of the myriad of patterns potentially available to that clade even within parts of its range (e.g., Figures 4, 7, and 8). In this clade, novel patterns are generated through introgression at rates unimaginable by mutation. Thus, this chapter has been focused on how, in Heliconius , regulatory genes that establish the identity of autonomous scale cell lineages early in development could account for the remarkable capacity of a few genes to generate so much qualitative variety in pattern. It is in the relative allocation to this particular mode of gene governance over scale pattern that may set Heliconius, and species like Papilio dardanus, apart from other butterflies.

However, just as genotype variation in a population depends on allelic variation at genetic loci, the mechanisms of developmental genetics and introgression that I describe for producing novel patterns depend on hybridization between already present, distinctly patterned, species or races. Furthermore, just as standard genetic recombination can produce greater variety if frequency dependent selection or heterosis maintains genetic polymorphism in a population, so the creative potential of introgression in this case depends at some level on the diversity of wing patterns maintained in local Heliconius communities. But, as discussed earlier, wing pattern diversity in a community depends upon habitat heterogeneity and external sources of pattern variation, which brings us full circle.

Novel developmental genetic mechanisms on the one hand, and rules for the diversification and storage of novel pattern programs in local communities on the other, are adequate to explain why Heliconius has available a greater source of pattern variation to recombine through introgression than do many other nymphaline taxa such as Precis (e.g., Hafernik 1982). However, since novel wing pattern genotypes arise through introgression between already diverse races or species as we have seen, how does a new variant emerge from the dizzying milieu of patterns in a hybrid zone to achieve recognition as a wing pattern race? Ultimately it does appear as Mallet, Turner, and others have suggested, that processes like the shifting balance are needed at some stage to preserve novel genotypes in nature by increasing their frequencies to the level of recognized taxa. These taxa may later hybridize to fuel further diversification through introgressive recombination.

Once again such lines of logic bring us full circle in terms of answering the question of why Heliconius seems to have more wing pattern variation available. This is because the mechanisms and processes discussed promote each another, and positive feedback system or cycles tend to be circular. While many other butterfly genera, including members of the heliconiine tribe, use the same pigments to color wing scales; share some common mechanisms for genetic control of some aspects of scale pattern; have diverse, mimetic species and hybrid zones (e.g., Limenitis, Platt 1983), and are exposed to the evolutionary forces of drift and selection, they certainly cannot match the punctuated nature of pattern evolution seen on Heliconius wings (Turner 1983 ) nor match the rate of diversification of that genus.

What traits of Heliconius might account for its exceptional capacity to generate biodiversity as illustrated in Figure 8? I argued in the introduction the unique trait of pupal mating might promote local diversity, one of the key ingredients of the feedback system envisioned. Another unique trait of Heliconius, adult pollen feeding, by promoting individual longevity and residence time, allows local populations to persist at extremely low densities (Gilbert 1972, 1991). Such resilience of local demes in the context of metapopulations in a hybrid zone, should improve the opportunity for shifting balance to work effectively. However, while features of adult life history probably play a role, they represent quantitative rather than qualitative departures from what we know about related butterflies. I must conclude therefore, that the catalyst for setting off the complex feedback cycles which sequentially depend upon, then generate, the pattern diversity found in Heliconius, is to be found in the tool box of developmental genetics as revealed by studying the MCS clade.