Does plasticity drive speciation? Host-plant shifts and diversification in nymphaline butterflies (Lepidoptera: Nymphalidae) during the tertiary

Biological Journal of the Linnean Society, 2008, 94, 115–130. With 5 figures Does plasticity drive speciation? Host-plant shifts and diversification in nymphaline butterflies (Lepidoptera: Nymphalidae) during the tertiary SÖREN NYLIN1* and NIKLAS WAHLBERG1,2 2 Department of Zoology, Stockholm University, SE-106 91 Stockholm, Sweden Laboratory of Genetics, Department of Biology, University of Turku, FI-20014 Turku, Finland Received 15 February 2007; accepted for publication 9 July 2007 How and why the great diversity of phytophagous insects has evolved is not clear but, if the explanation is the diversity of plants as a resource, colonizations of novel plant taxa can be expected to be associated with higher net speciation rates. In the present study, we make use of recent advances in plant and butterfly systematics to trace the evolution of host-plant utilization in the butterfly subfamily Nymphalinae (tribes Nymphalini, Melitaeini, and the probably paraphyletic ‘Kallimini’). A clear historical pattern emerges, with an ancestral host-plant theme of ‘urticalean rosids’ and two major colonizations of novel distantly-related plant clades. The asterid order Lamiales was colonized by an ancestor of ‘Kallimini’ + Melitaeini and the family Asteraceae in Asterales was later colonized by Melitaeini butterflies. These colonization events appear to have been followed by increases in the rate of net butterfly diversification. Two not mutually exclusive scenarios to explain such patterns have been suggested: (1) adaptive radiation due to release from competition following host-plant shifts or (2) higher rates of net speciation during a relatively long-lasting potentially polyphagous (plastic) state. In support of the ‘plasticity scenario’, phylogenetic traces of a long-lasting stage with some potential to feed on more than one host-plant clade can still be seen, despite the ancient age of the colonizations. When angiosperm communities changed after the K/T boundary due to extinctions and subsequent diversification, herbivore taxa that could occupy several alternative niches may have had the greatest opportunity to diversify in turn. © 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 94, 115–130. ADDITIONAL KEYWORDS: feeding – generalist – niche – phenotypic plasticity – polyphagy – specialization. INTRODUCTION Phytophagous insects represent a large proportion of global biodiversity, and insect clades feeding on plants are generally more species-rich than their sister taxa (Mitter, Farrel & Wiegmann, 1988). These patterns strongly suggest that plant-feeding in insects promotes speciation and diversification, but it is not clear how and why. Evolution of the ability to feed on plants may be something of an evolutionary ‘hurdle’ for insects, a key innovation opening up new routes of evolution, as suggested by Mitter et al. (1988). The important factor may be the diversity of plants (chemical and otherwise) as a resource compared to, *Corresponding author. E-mail: for example, feeding on other insects, in particular the great diversity of angiosperms (Ehrlich & Raven, 1964; Farrell, 1998). But, if so, exactly how does plant diversity drive the origin of insect diversity? After all, most phytophagous insects are relatively specialized on their host plants (Thompson, 1994), so where is the opportunity for plant diversity to affect insect speciation? In a now classic study, Ehrlich & Raven (1964) suggested a coevolutionary scenario with adaptive radiations after shifts to competition-free host plants, using butterflies and their larval hosts for illustration. However, with the phylogenetic knowledge available at the time, the data could only be presented in the form of lists of associations between taxa, showing that related butterflies feed on related plants. In the © 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 94, 115–130 115 1 116 S. NYLIN and N. WAHLBERG MATERIAL AND METHODS BUTTERFLY PHYLOGENY The relationships among nymphalid butterflies has until recently been poorly known. This has also been true for Nymphalinae, but a series of phylogenetic studies on the family and subfamily have cleared up many questions (Brower, 2000; Nylin et al., 2001; Wahlberg & Zimmermann, 2000; Wahlberg & Nylin, 2003; Wahlberg, Weingartner & Nylin, 2003; Freitas & Brown, 2004; Wahlberg, Brower & Nylin, 2005). The current subfamily Nymphalinae consists of the tribes Nymphalini, Melitaeini, Kallimini, Victorinini, Junoniini, and possibly Coeini (for genera included in these taxa, see Table 1). The sister group to Nymphalinae is not yet clear, but the most recent molecular studies suggest that it is one of the clades Cyrestini, Pseudergolini, Biblidinae, or Apaturinae (Wahlberg et al., 2003, 2005). Within Nymphalinae, relationships have been elucidated by a morphological study (Freitas & Brown, 2004) and a molecular study with data from the mitochondrial gene cytochrome oxidase subunit I and the nuclear genes EF1-a and wingless (Wahlberg et al., 2005). Coeini (which contains the genera Historis and Baeotus only) appears to be the sister group to the rest of the species. The remaining tribes form a stable well-supported clade, with Nymphalini being sister to the rest of the species. What was once considered a poorly defined tribe Kallimini (Harvey, 1991) has now been shown to comprise several tribes that form a grade to a monophyletic Melitaeini (Wahlberg et al., 2005). Three genera with single species in each (Kallimoides, Rhinopalpa, and Vanessula) have unstable positions. This uncertainty limits the possibilities of unambiguously reconstructing host-plant shifts, and we discuss the implications of different topologies here. Figures 1, 2, 3, 4 show the relationships between genera supported by the most recent and most complete study (Wahlberg et al., 2005; a maximum parsimony analysis), along with inferences (from taxonomical statements in the literature) for genera not sampled in that study (Antillea, Atlantea, Phystis, Dagon, Ortilia, and Tisona). The alternative topology in Figure 5 is a result of a Bayesian analysis presented in Wahlberg (2006b). Note especially the different positions for Rhinopalpa and Vanessula. HOST PLANTS We have followed the most recent plant classification available, the one suggested by the Angiosperm Plant Phylogeny group (APGII, 2003), based on molecular evidence. The new order Rosales is more inclusive than the traditional one and includes the former Urticales. Still, the families Ulmaceae, Cannabaceae, Moraceae, and Urticaceae (the former Urticales) together evidently form a monophyletic clade with high support, and we refer to this clade in the following as the ‘urticalean rosids’. Regarding ‘Kallimini’ and Melitaeini, the new classification simplifies presentation of the host-plant utilization patterns observed, by bringing together important host families formerly in the orde (...truncated)