Abstract

The contiguous Neotropical lowland wet forests of northwest South America and Central America of are typically inhabited by 10-15 species of Passionflowers in the genus Passiflora, which are fed upon by the larvae of 10-11 species of Heliconius butterflies. In the 1970's the author investigated this interaction at the La Selva Biological Station in northeastern Costa Rica, and in addition to finding 10 species of Heliconius, discovered another, equally diverse community of herbivores: the Passiflora-feeding flea beetles (Coleoptera: Chrysomelidae: Alticini). These beetles shared Passifora species with Heliconius, and to a first approximation, divided the host plant community in the same way. Now, 40 years later, the author extended this data set, observing, collecting and rearing over 1500 flea beetles on wild Passiflora vines at La Selva. Genetic barcoding of a subsample of 110 of these specimens revealed 10 distinct species in five genera. Morphological comparison of the modern specimens with the 1970's collections revealed identical community composition over more than 40 years. Rearing in cages also revealed, for the first time, larval stages for 8 species belonging to two feeding strategies, leaf /epidermis eaters and surface rootlet/basal stem feeders.

Working on the assumption that species diversity in these two communities is controlled by characteristics of the host plant community, multiple aspects of the interaction were examined. Unlike solitary-feeding Heliconius larvae that are known to be prey to the Ectatomma ants common on Passiflora extrafloral nectaries, the adult and larval flea beetles were more common on ant-tended plants and exhibited other characteristics of being resistent to ant predation. A survey of cyanogenesis across Passiflora species revealed spatial and temporal variation within plants as well as 5 orders of magnitude variation within and between species. Three Passiflora species most used by Heliconius and flea beetles varied over a factor of 100 in cyanogenesis. Overall, flea beetles chose Passiflora with reduced or variable amounts of cyanide production, possibly coping with cyanide toxicity by eating tiny meals and growing slowly. Fast growing Heliconius larvae, with the ability to sequester cyanogenic glycosides, appeared to select host plants regardless of their cyanide production. These findings suggested the hypothesis that Heliconius and flea beetles exert opposing selection pressure on Passiflora cyanogenesis, expanding within-species variabity and acting as mutualists. Additional comparisons revealed that, unlike Heliconius that fly with their co-mimics, brightly colored flea beetle adults showed only limited tendancy toward mimicry. Also unlike Heliconius that decline greatly during extended rain and overcast weather, flea beetles remained active year round with only a small decline in the rainy season. Comparison of diet breadth revealed great similarities between butterflies and flea beetles, including a similar response to the three principal Passiflora subgenera represented at La Selva. This subgeneric response was also mirrored in a third community represented in Costa Rica by three species, the Josiine moths (Lepidoptera: Notodontidae).

A direct comparison analysis, attempting to pair each Heliconius with its ecologically equivalent flea beetle, was partially successful. One difficulty in the comparison is that five Heliconius are Passiflora species-specific, whereas only one flea beetle is monophagous. Another difficulty is that two of the butterflies and two of the flea beetle species were rarely sampled and may represent vagrants from nearby habitats. These difficulties were addressed by restricting the analysis to species well-represented in the sampling; in this case 10 species of Passiflora and eight species each of Heliconius and flea beetle. This analysis revealed the underlying community structure to consist of five distinct groups of host plants, each of which hosts four coexisting, non-competing herbivores: Heliconius "generalists" with solitary larvae, Heliconius monophages with group-feeding larvae, flea beetles with leaf-feeding larvae, and flea beetles with surface rootlet and basal stem feeding larvae. The five host plant groups exclude three species seldom used by the butterflies and flea beetles, but include the forest P. pittieri (subgenus Astrophaea), the forest P. lobata (subgenus Decaloba), the second-growth P. auriculata/P. biflora (Decaloba), the forest P. vitifolia/P. ambigua (subgenus Passiflora) and the second growth P. oerstedii (subgenus Passiflora). The 16 Heliconius and flea beetle species fit well into this niche analysis, with only one instance of two species sharing the same niche and three cases where one species occupies two niches. The remaining 13 niches are uniquely occupied by one species each. This finding suggests a tightly structured community with strong interspecific competition within each niche, but greatly relaxed competition between the four larval feeding types, even though they share the same host plant species. It also illustrates three generators of species diversity for this system: host plant differentiation at the subgeneric level, habitat specialization in host plants as well as herbivores, and finally, evolution of distinct, non-competing larval feeding-strategies. In contrast with the strong role that host plant subgenera play in generating herbivore diversity, the remarkable differentiation of Passiflora at the section and species level does not appear to contribute greatly to species diversity at La Selva.