This post is adapted from a thesis chapter by Patricio Salazar.
Arriving at a stable taxonomy of the diversity of Heliconius species and races, including the H. melpomene / H. erato complex, was a long and arduous endeavour that lasted nearly two hundred years (Turner 1967). The task was obviously complicated by mimicry, as well as polymorphism and geographical variation. However, even after the taxonomy became relatively stable, the nature of the numerous colour-pattern forms—labelled until then as variations, forms, types, stages or aberrations, found in some localized regions such the border between Surinam and French Guiana—remained an unsolved mystery (Beebe 1955). It was necessary to establish a field site in the tropics that would allow long-term studies of the butterflies’ life history as well as breeding experiments. That opportunity was opened by the establishment New York Zoological Society field station, at Simla (Trinidad), in 1950 (Beebe 1952).
Beebe (1955) appears to provide the first direct evidence that many of the uncommon and narrowly distributed colour pattern forms, found in the catalogues of butterfly collectors and systematists until then, were in fact the product of hybridization between colour pattern forms found in monomorphic populations distributed over wider areas. He proved this by rearing broods of several wild caught females from the Surinam/French Guiana hybrid zone (although he did not use the term ‘hybrid zone’). Most importantly, he provided the first published account of an inter-racial cross performed in captivity. It was a cross between a dennis-rayed/yellow-banded male of H. erato collected in Surinam with a red-banded female from Trinidad (Figure 1).
Figure 1. Plate II from Beebe (1955) showing the parents and brood of the first ever published interracial cross of Heliconius carried out in captivity. Top left: male parent, H. erato amazona from Surinam (now H. e. erato). Top right: female parent, H. e. hydara from Trinidad. All 13 offspring inherited most phenotypic features of the father, except for the colour the ‘broken’ forewing band, which is reddish instead of bright yellow as in the father (original image is black and white).
Beebe (1955) did not use terms like ‘forewing band’, ‘dennis’ or ‘rays’ (although he described them as ‘radii’, in Latin) that are now the standard morphological nomenclature to describe the major colour-pattern elements of H. melpomene, H. erato and several other Heliconius species. No reference was made either to the possibility that colour pattern elements could be inherited in a Mendelian manner. The major contributions of his paper were therefore to report that a parent with a given colour pattern can deliver polymorphic offspring and—most importantly—these offspring can be produced by hybridization of other colour pattern morphs. In spite of these discoveries, (Beebe 1955) did not describe the population in Surinam where the parental butterflies were collected as a hybrid zone.
The first full description of the Mendelian inheritance of colour pattern loci in Heliconius was that of Turner and Crane (1962) in 37 broods of H. melpomene. In this landmark paper they reported on the inheritance of the three major pattern elements aforementioned: the forewing (FW) band, the ‘dennis’, and the hindwing rays (Figure 2). This seems to be the first publication in which these names were used for the morphological description of Heliconius colour patterns, and is the one that established the now-standard notation B, D and R for the loci controlling these pattern elements. Turner and Crane (1962) did not only describe the phenotypic effects of B, D, R, but also inferred their physical linkage and provided an estimate of the recombination rate between D and R. They did not report crosses for H. erato, but inferred the inheritance of B, D and R and their possible linkage in this species by analysing data from Beebe (1955). Despite using the same notation for H. erato loci, Turner and Crane (1962) did not make any explicit suggestion of homology between H. melpomene and H. erato. Finally, it is worth noticing that their original description of the B locus differs from our current interpretation of its phenotypic effects. In the original description B determines the ‘width’ of the FW band, with character states ‘broad’ and ‘narrow’, both red in colour, whereas in the current interpretation B determines the ‘colour’ of the FW band, specifically the presence/absence of red (Sheppard et al. 1985; Mallet 1989; Naisbit et al. 2003). The establishment of this nomenclature for describing the morphology of Heliconius wing patterns for genetic analysis was very useful for the long-term research program of description of colour pattern loci that followed Turner and Crane’s (1962) seminal work.
Figure 2. Schematic representation of the major colour pattern elements in H. melpomene and H. erato. Variation in colour, shape or full presence/absence of these pattern elements makes up for most of the colour pattern diversity in the melpomene / erato mimicry system, as well as in various other species across the Heliconinii (From: Nijhout et al. 1990).
Between the mid 1960s and the mid 1980s an intense research program to characterize colour pattern loci in H. melpomene and H. erato continued, at the NYZS station at Simla in Trinidad, at the University of Campinas in Brazil, and at the University of Liverpool in the UK. During this period the bulk of all major Mendelian loci in H. melpomene and H. erato were characterized. This work is summarized in the comprehensive works of Turner (1972) and Sheppard et al. (1985). The latter was written largely by John Turner posthumously after Phillip Sheppard passed away Sheppard et al. (1985) described 12 loci for melpomene (9 of them assigned to 6 linkage groups) and 16 for erato (13 of them in 8 linkage groups). Even though further crosses have been reported after this period (e.g. Mallet 1989; Jiggins and McMillan 1997; Naisbit et al. 2003), including other species besides H. melpomene and H. erato, no other major colour pattern loci have been identified.
In the late 1990s, the interest of the new generations of Heliconius geneticists shifted towards the identification of the actual genes that control colour pattern at the molecular level. However, no major progress occurred until 2005, when the position of the major colour pattern loci D, Ac/Sd and Yb/Cr was located in linkage maps that recovered all chromosomes of several Heliconius species (Jiggins et al. 2005; Tobler et al. 2005; Joron et al. 2006a; Kapan et al. 2006; Kronforst et al. 2006c). The identification of molecular markers genetically linked to wing pattern loci permitted several new insights. Primarily, these collaborative studies answered a longstanding question in Heliconius genetics: are the convergent similarities between mimetic species determined by the same set of homologous loci, or by a different set of independently-evolved genes? The answer was more surprising than expected: not only convergent phenotypes (such as those observed in the melpomene / erato mimetic radiation) are determined by the homologous loci, but also these same loci were largely responsible for divergent colour patterns in other Heliconius species, most notably H. cydno and H. numata (Joron et al. 2006a; Kronforst et al. 2006c).
In addition, the development of linked molecular markers permitted a second insight, which is perhaps less widely recognised. Since markers could now be scored in multiple families segregating for different phenotypic elements within each species, it was possible to establish that many of the distinct ‘loci’ described in previous work were in fact allelic variants of the same small handful of patterning loci. Thus, the majority of the 12 loci described in H. melpomene and 16 in H. erato are in fact allelic effects of just three linkage groups. Within some of these linkage groups, several tightly linked elements are found that can be separated by occasional recombinants. A example is the B/D locus – these elements were originally thought to be separated by ~27cM (Sheppard et al. 1985), but molecular mapping and association studies indicate that they are in fact two very tightly linked elements around 10kb apart (Baxter et al. 2008; S. Baxter Pers. Comm.).
The obvious questions after the linkage-mapping research yielded its most significant results were then: what are the actual genes that control colour patterns and how do these genes work to produce the diversity Heliconius colour patterns? Answering these questions has taken a few more years, but the recent discovery that differential expression of the optix and WntA genes determine the divergent and convergent phenotypes produced by the D and Ac/Sd loci respectively represent a major advance (Reed et al. 2011; Martin et al. 2012). The evidence for the involvement of these genes is primarily fine-scale mapping based on large mapping families and expression studies showing spatial patterns of expression correlating with phenotype. In the case of the WntA gene, further functional evidence was also obtained through the use of a heparin injections that are known to disrupt Wnt signalling, which disrupted FW band phenotypes. The discovery of the gene(s) that determine the Yb/Cr phenotypes has been elusive so far, but in spite of this the emerging challenge is to figure out the genetic/developmental pathways that produce the alternative phenotypic states switched by already characterized loci (Joron et al. 2006b).