DISCUSSION AND CONCLUSIONS (continued)

6. Does one clade typically drive the evolution of novel patterns?

Müllerian pairs from the pupal mating ESS and the MCS clade are known to coevolve locally with respect to polygenic adjustments in pattern elements such as width of HW window in melpomene and erato (Gilbert 1983). However, the pattern variety among races of erato (ESS) and melpomene (MCS) and the precise mimetic tracking of one by the other clearly shows that novel steps occur and that mimetic tracking ensues. The question is whether novel steps are taken by an ESS species in some places while by its counterpart MCS species in others.

At least two ESS species do not appear to be comodels driving pattern evolution among other Heliconius. First, H. charitonius, the most palatable Heliconius species tested by Chai (1990) is remarkably uniform over its range from Texas and Florida to Panama and does not participate with other Heliconius in mimicry. (In Peru a sibling species of H. charitonius converges on the ithomiine Elzunia.) Likewise, the pupal mating species, H. hecalesia, along with H. hecale, converges on the pattern of the much more unpalatable ithomiines (Chai 1990).

These species aside, there are several reasons to implicate the ESS clade as the principal generator of novelty. First, to the degree that females of pupal mating species lack the ability to choose mates, sexual selection is minimized as a conservative force maintaining wing pattern (in contrast to MCS females which mate multiply, ESS females are typically monogamous, the only mating usually occurring at eclosion). Second, pupal mating species, in comparison to the MCS clade, form larger, and more gregarious nocturnal roosting groups (personal obs.). Dense pre-roost congregations might increase the apparent local density of a novel trait to local predators and promote its establishment (Mallet 1986, Mallet and Gilbert 1995). Third, pupal mating species often occur in marginal habitats on the edge of Heliconius distribution where they can evolve novelty by mutation without the swamping effects of gene flow.

Fourth, the group of ESS species which include sapho, hewitsoni, and eleuchia utilize the woody Astrophea group Passiflora and all deposit large clusters of eggs on new shoots which appear in pulses and grow very rapidly. Consequently, in these species, large cohorts of new adults often pulse into local concentrations of their older kin, which themselves had previously emerged in a particular host patch (personal obs., Deinert this volume, Kapan 1998). Such population structure provides unusual opportunity for local inbreeding which in turn, would promote the rapid karyotype evolution in this group (see Brown et al. 1992) as well as fixation of rare combinations of mutant alleles. The result might be rapid production of genetically isolated species with novel karyotypes and wing patterns (e.g., H. eleuchia) Moreover, species of this group are unique among Heliconius in "hyper-active" sequestration of host cyanogens (Engler 1998), and may be less dependent on adult pollen feeding for unpalatibility as is assumed for most Heliconius (Gilbert 1991). Occasional large local congregations, unpalatibility, and high potential for inbreeding are possible ingredients for the evolution of pattern novelty without invoking introgression.

The working hypothesis emerging from these considerations is that populations of species in the ESS clade are more likely to autonomously establish new adaptive peaks for warning pattern than are those of the MCS group. Species of the pupal mating group tend to be better isolated from each other, often by chromosome differences, and thus form, as sapho and Hewitsoni illustrate in Costa Rica, stable templates for their Müllerian partners.

Whether MCS Heliconius possess unique tools in their pattern tool box is a question deserving of intensive study. Because of frequent failure of interspecies crosses within the ESS clade, it is more difficult to make progress in defining the nature of shared pattern genes in this group than is the case for the MCS clade I have discussed here. It is worth noting that Turner (1981, 1983), in applying Haldane's sieve to polarize pattern genes as traits for constructing parallel phylogenies of erato and melpomene, predicted that both species descended from ancestors possessing yellow bands (=windows) on both FW and HW. This, along with the fact that it is possible to obtain this predicted phenotype at the bottom of an epistatic pecking order of pattern genes in crosses of cydno and pachinus, (e.g., Figure 4 upper right g), suggests that similar underlying mechanisms are at work in ESS and MCS clades. Yet, from the crosses I have conducted (not shown here) or reviewed in the literature involving erato races, and those involving different species at the species/race boundary (e.g., erato and himera, Jiggins and McMillan 1997), it is apparent that some of the shutter tools in the "pattern tool box" of the MCS clade represent developmental innovations unique to the cydno/melpomene part of clade. I suggest that these innovations may have allowed this group to diversify rapidly against the template of ESS-generated adaptive peaks typically characterized by warning patterns unique to Heliconius. To understand the level at which coevolution might operate in these parallel radiations, it will be helpful to know to what degree the ESS side has taken the evolutionary initiative and what aspects of the parallel wing pattern evolution of these clades have been based on similar developmental genetic systems throughout their joint histories.