Molecular and Vol. 20, No. 3, September, pp. 460–473, 2001 doi:10.1006/mpev.2001.0984, available online at http://www.idealibrary.com on

Examining in a Large Radiation of Madagascan Butterflies (: : Mycalesina) Based on Mitochondrial DNA Data E. Torres,* D. C. Lees,† R. I. Vane-Wright,† C. Kremen,‡ J. A. Leonard,§ and R. K. Wayne§ *Department of Biology and Microbiology, California State University, 5151 State University Drive, Los Angeles, California 90032-8201; †Biogeography and Conservation Laboratory, Department of Entomology, The Natural History Museum, Cromwell Road, London SW7 5BD, United Kingdom; ‡Department of Ecology and Evolutionary Biology, Princeton University, Eno Hall, Princeton, New Jersey 08544; and §Organismic Biology, Ecology and Evolution Department, 621 Charles E. Young Drive South, Los Angeles, California 90095-1606

Received September 26, 2000; revised March 14, 2001

is despite a long history of interest in the biogeography The satyrine butterfly subtribe Mycalesina has of this long-isolated and biotically unique island. undergone one of the more spectacular evolutionary In the past two decades, a much deeper understand- radiations of butterflies in the Old World tropics. Per- ing of the geological history of Madagascar has been haps the most phenotypically pronounced diversifica- gained. It is now thought that its rift from the east tion of the group has occurred in the Malagasy region, African part of Gondwanaland commenced around 165 where 68 currently recognized are divided Ma and slowly reached its approximate present posi- among five genera. Here, we report the results of phy- tion, at most 400 km distant from Mozambique, in the logenetic analyses of sequence data from the cyto- mid Cretaceous around 125 Ma (Rabinowitz et al., chrome c oxidase II and cytochrome b mitochondrial 1983). Recent phylogenetic evidence and discoveries of genes, for a total of 54 mycalesine taxa, mostly from predating the K/T boundary (65 Ma) suggests Madagascar. These molecular data complement an ex- isting data set based on male morphological charac- that the modern day vertebrate fauna had origins en- ters. The molecular results support the suggestion tirely independent of the island’s Cretaceous fauna from morphology that three of the five Malagasy (Krause et al., 1999). There is mounting circumstantial genera are paraphyletic and support the monophyly of evidence for rafting or other nontectonic dispersal, as at least three major morphological clades. Novel hy- originally suggested by Simpson (1940) for vertebrates potheses of terminal taxon pairs are generated by the (see, e.g., Caccone et al., 1999). Periods of lowered sea molecular data. Dense taxon sampling appears to be level have raised the chances of stepping stone faunal crucial for elucidating phylogenetic relationships exchange to and from Madagascar via small archipel- within this large radiation. A potentially complex sce- agos such as the Comoros, the Chagos, the Seychelles nario for the origin of Malagasy mycalesines is pro- bank, and the Mascarene Plateau. The last major ma- posed. © 2001 Academic Press rine regression occurred just 18,000 years ago (Haq et al., 1987; Colonna and Casanova, 1996). Due to the great depth of the Mozambique channel and counter- INTRODUCTION colonizing trade winds, dispersal from Africa has prob- ably long been difficult. One suggestion to the contrary Island radiations, notably those in the Pacific Ocean is based on uplift of the Davie Ridge (McCall, 1997; but archipelagos such as the Galapagos (Darwin, 1859) see Krause et al., 1999). and Hawaiian (Dobzhansky, 1951; Mayr, 1963; Car- Members of the subtribe Mycalesina Reuter, 1896 son, 1968; Otte, 1989) Islands, have played a major role [ϭ Mycalesini of Miller, 1968] (Lepidoptera: Nympha- in our present day understanding of evolutionary and lidae: Satyrinae: ), which have weak to mod- ecological patterns and processes. Examination of the erate dispersal powers, have undergone one of the evolutionary history of radiations on larger islands and more spectacular evolutionary radiations of butterflies their associated archipelagos, however, has progressed in the Old World tropics. This subtribe is distributed more slowly. For political reasons, it was not until the from northeast Australia to Asia as far west as the 1980s and early 1990s that equivalent levels of inves- Himalayan foothills, and in subsaharan Africa north to tigation were initiated on the biota of Indian Ocean Sudan and Ethiopia, with roughly equivalent numbers islands, in particular of Madagascar (Kull, 1996). This of species in the Indo–Australian, Afrotropical, and

460 1055-7903/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. MADAGASCAN PHYLOGENETICS 461

Malagasy regions. The sister relationship of the sub- Lees has greatly clarified the alphataxonomy of the Mycalesina is presently unresolved (A. V. Z. group, recognizing about 35% synonymy, some of it due Brower, pers. comm.). Like most other members of the to unrecognized sexual and seasonal dimorphism, in- putative satyrine/morphine/brassoline clade of the cluding the eight species placed in Houlbertia, for (Brower, 2000), the subtribe feeds in the which no females were recognized (d’Abrera, 1980; al- larval stage on , which include though see d’Abrera, 1997; Lees, 1997). This “phantom (grasses and ), Cyperaceae (sedges), and Zin- extinction” was exactly balanced by the recognition of giberaceae (gingers)/Marantaceae (prayer plants) 24 new species-level taxa (Lees, 1997). Lees performed (Ackery, 1988; Lees, 1997). Pollen attributable to phylogenetic analyses on 429 morphological characters Poaceae first appears in the Paleocene and becomes for 84 spp. of mycalesines (including some naked-eyed prominent in the Oligocene record (Muller, 1981), from representatives and all hairy-eyed species) from Mada- which time a number of butterflies attributed to gascar, Africa, and the Comoros. He concluded that the the Elymniini (subtribe Lethina) are known from genera Westwood, Houlbertia, and the large France (Nel et al., 1993) and Brazil (Martins-Neto et Henotesia are paraphyletic or polyphyletic, but al., 1993); no fossil Mycalesina are known. Larval My- that Masoura Hemming is monophyletic and that Ad- calesina in Madagascar feed on Poaceae (Ackery, 1988) miratio is monobasic. A number of fairly robust clades and Cyperaceae (D. C. Lees, C. Kremen, and R. van based on male morphological characters emerged from Buskirk, unpublished). Adults feed mostly on fallen this work (summarized in Table 1), although the rela- fruit and sap (Parsons, 1998) but nectar-feeding is tionships among these clades proved highly unstable. known (Lees, 1997). The adults are fairly long lived, Phylogenetic studies have been slow to commence on sometimes for several months. The community niche biotic radiations in Madagascar, but an increasing partitioning of mycalesine butterflies along gradients number have used molecular data (e.g., Yoder, 1994; in Madagascar has been examined (Kremen, 1992, Emberton, 1995; Dubois et al., 1996; Baum et al., 1998; 1994) and suggests that environmental responses are Caccone et al., 1999; Kerdelhue et al., 1999; Cibois et fine-grained enough to be informative for bioinventory al., 1999; Jansa et al., 1999; Johnson et al., 2000). work and conservation planning. The subtribe My- Recent molecular work (Rumpler et al., 1994; Del Pero calesina has also played a role in the understanding of et al., 1995; Yoder et al., 1996; Yoder and Irwin, 1999) evolutionary and developmental genetics in , suggested that lemurs arrived in Madagascar around largely due to extensive work carried out on the Afri- 55–60 Ma as the result of a single founder event. can anynana, in particular on the genetics of However, recent works on the radiations of tenrecs wing eyespots (Holloway et al., 1993; Holloway and (Asher, 1999) and murid rodents (Jansa et al., 1999) Brakefield, 1995; Monteiro et al., 1997) and develop- have not supported strict monophyly. Lees (1997) was mental plasticity (Carroll et al., 1994; Brakefield, 1998; unable to demonstrate strict monophyly for Madagas- Brakefield et al., 1996, 1998). can mycalesine butterflies (ca. 68 spp.). Lees also sug- The Mycalesina contains about 230 described species gested that the origination of the Malagasy mycalesine grouped into 11 genera (Condamin, 1973; Ackery et al., butterfly radiation may have been from India rather 1995; Aoki et al., 1982; Lees, 1997). Circumscription of than Africa, an hypothesis radically different from that genera has been based on few morphological characters of Miller (1968). Lees’ results suggested the existence and has tended to be parochial in nature (Vane-Wright, of two clades of African hairy-eyed mycalesines (cur- 1971). Traditionally, mycalesine species have been as- rently placed in Henotesia; Kielland, 1994). He pro- signed to genera based on gestalt (e.g., Oberthu¨r, posed that, following a dispersal event from India (aid- 1916), details of forewing venation (Eliot, 1992), and ed by low sea level or trade winds during the Tertiary), presence or absence of setae on the compound eyes there were at least two subsequent colonizations of (Vane-Wright, 1971; Condamin, 1973). The 68 species Africa by founders from Madagascar. Elucidation of of mycalesines in the Malagasy region (35% of which the origins of taxa on Madagascar would enrich our are undescribed) are currently placed in 5 genera. Most understanding of the history of diversification of radi- Afrotropical hairy-eyed mycalesines, including 11 spe- ations on the island. Our work here provides an inde- cies on the African mainland (Kielland, 1994), have pendent test of one of the most extensive morphological been placed within Henotesia Butler. However, some analyses for any large Malagasy radiation (Lees, 1997). Malagasy hairy-eyed taxa have been placed in other Here we report the results of phylogenetic analyses genera based on distinctive wing patterns and shapes of sequences from two mitochondrial genes, cyto- (e.g., Heteropsis drepana Westwood, species of Houl- chrome oxidase II (COII) and cytochrome b (cyt b). We bertia Oberthu¨ r, and Admiratio paradoxa Mabille; see present a larger analysis of COII for a total of 54 d’Abrera 1980, 1997). mycalesine terminal taxa, mostly from Madagascar. No prior explicit phylogenetic work had been carried We use the sequence data from the cyt b region in a out on the Mycalesina in Madagascar until that on simultaneous analysis with COII for a subset of 26 male adult morphology by Lees (1997, unpublished). taxa. The molecular analyses are used first as an in- 462 TORRES ET AL.

TABLE 1 Well-Supported Morphological Species Groups among Afrotropical Mycalesines

Morphological % Spp. Species sampled for DNA phylogeny cladeNo. spp. known sampled (additional morphological clades in parenthetic notation) Reference

narova20 70 Houlbertia narova, Ho. wardii, Ho. passandava, Henotesia strato, Hen. sp. 10 Lees, 1997 (ϭdifficilis?), Hen. sp. 18 (ϭdifficilis?), Hen. sp. 12, Hen. sp. 20, Hen. sp. 22, Hen. sp. 32, Hen. sp. 37, Hen. sp. 74, Hen. sp. 75, Hen. sp. 76, Hen. sp. 77

strigula15 60 Henotesia strigula, maeva group (Hen. maeva ϩ Hen. sp. 16 ϩ Hen. laetifica), Lees, 1997 angulifascia group (Hen. angulifascia ϩ Hen. sabas ϩ Hen. turbans ϩ Hen. sp. 73), Hen. sp. 49

subsimilis7 100 Henotesia subsimilis, Hen. pauper, Hen. comorana, Hen. sp. 7, Hen. sp. 14, Hen. Lees, 1997 sp. 14A, Hen. sp. 23, Hen. sp. 25, Hen. sp. 30

bicristata4 100 Henotesia bicristata ϩ (Hen. parvidens ϩ (Hen. uniformis ϩ Hen. avelona)) Lees 1997

Masoura6 17 Masoura antahala Lees, 1997

ankova3 67 Henotesia ankova, Hen. turbata Lees, 1997

exocellata3 33 Henotesia exocellata Lees, 1997

simonsii8 13 Henotesia perspicua Kielland, 1994

Heteropsis1 100 Heteropsis drepana (clade with Masoura) Lees, 1997

elisi2 0 Kielland, 1994

iboinia2 50 Henotesia parva Lees, 1997

Hallelesis2 50 halyma Condamin, 1993

Bicyclus76 4 hewitsonii group (Bicyclus zinebi); safitza group (Bicyclus cottrelli, B. anynana) Condamin, 1993

Admiratio1 100 Admiratio paradoxa Lees, 1997

ankaratra1 100 Henotesia ankaratra (clade with Admiratio) Lees, 1997

fraterna1 100 Henotesia fraterna Lees, 1997

vola1 100 Henotesia vola Lees, 1997

fuliginosa1 100 Henotesia fuliginosa Lees, 1997

undulans1 0 Lees, 1997

Hen. sp. 541 0 Lees, 1997

peitho1 100 Henotesia peitho Kielland, 1994

Note. Name of morphological clade includes approximate number of species in the clade as a subscript. Clades with a subscript of 1 represent rogue taxa which could not be assigned to a larger clade. dependent test of the support for classical generic tax- DNA from both sexes for each taxon (Table 2). About onomy based on morphology and the relationships of one-third of mycalesine butterfly species are known to major morphological clades and species groups. The span at least half the latitudinal range in Madagascar, identity of morphospecies and certain sexually dimor- whereas most have narrower ranges (Lees et al., 1999); phic taxa are also addressed with molecular data. Fi- it is difficult to be sure from morphological evidence nally, these molecular data are used as an independent whether allopatric exemplars are conspecific or sister test of the monophyly of the Madagascan radiation to species. For such taxa, we sequenced representatives of examine whether there was a single founding event each morphospecies when possible. Here we report on and regional radiation (“strict monophyly”) versus the analysis of 432 bp of sequence from the COII gene multiple events and subsequent dispersal or reinva- for 54 terminal taxa (including representatives of He- sions. notesia from Africa and the Comoros) (Table 2). Addi- tionally, we sequenced 380 bp of the cyt b sequence MATERIALS AND METHODS data for a subset of 26 terminal taxa. The two data sets were analyzed separately and in a simultaneous anal- Sample Collection and Preservation ysis. We sequenced three outgroup genera: Bicyclus Butterflies were collected over a period from 1992 to Hewitson (B. cottrelli and B. zinebi from Africa; B. 1995 (Fig. 1) by methods previously described (Kre- anynana from the Comoros), Hallelesis (H. halyma men, 1994). Most specimens were preserved by dehy- from Africa), and (M. francisca from SE dration in small envelopes or glass vials containing Asia). We were unable to obtain fresh material from silica beads, with the exception of one taxon preserved the Western Ghats of India for outgroups that were in 100% ethanol. Either method of preservation proved included in Lees (1997) morphological analysis. Se- adequate for DNA isolation and amplification. quence data have been deposited in GenBank (Acces- sion Nos. AY040122–AY040202). Taxon Sampling Use of traditional generic assignment in this paper To confirm the assignment of males and females to does not imply that generic names represent natural species, where possible, we obtained and sequenced groups. For convenience, major morphological clades MADAGASCAN BUTTERFLY PHYLOGENETICS 463

FIG. 1. Collection localities in Madagascar with existing rainforest indicated by dark shading.

(Table 1) are designated by subscript notation (i.e., the remainder of the individual for morphologic investiga-

“narova20” clade). The subscript represents the approx- tion. Fragments of the cyt b and COII genes were imate total number of species known in a particular amplified by the polymerase chain reaction (PCR). For clade. COII amplification, primers S2792 [5Ј-ATACCTCG ACGTTATTCAGA-3Ј] and A3772 [5Ј-GAGACCAT- DNA Methods TACTTGCTTTCAGTCATCT-3Ј] (Brower, 1994) were Total genomic DNA was isolated from individual used to amplify an approximately 800-bp fragment and butterflies by homogenization and overnight incuba- sequences were subsequently obtained with the inter- tion in lysis buffer (1ϫ TNE, 10% SDS, and proteinase nal primer S3126 [5Ј-ATTA(C/T)AAAATAGA(A/G)CTT K) at 55°C, followed by standard phenol/chloroform CICC-3Ј] (Brower, 1994) and with A3772. For cyt b, extraction and ethanol precipitation. The resulting primers CB1 [5Ј-TATGTACTACCATGAGGACAAAT- DNA was resuspended in sterile water and maintained ATC-3Ј] and CB2 [5Ј-ATTACACCTCCTAATTTATT- frozen until further use. In most cases, only a portion of AGGAAT-3Ј] were used for both PCR and sequencing. the thorax was used for DNA extraction to preserve the Each PCR mixture contained template DNA (amounts 464 TORRES ET AL.

TABLE 2 Taxa Sampled for DNA Analysis

Taxon Collection locality COII CytB Habitat Range

Admiratio paradoxa (Mabille) Madagascar: Maharira, Ranomafana Nat. Park 1M R W (Butler) Comoros: Grande Comoro, above M’Vouni 1F R W Bicyclus cottrelli (van Son) Africa 1F 1M S W Bicyclus zinebi (Butler) Boabeng, Nigeria 1M R N Hallelesis halyma (Fabricius) Nigeria: Ankasa 1M 1M R N Henotesia angulifascia (Butler) Madagascar: Ranomafana National Park 3M, 1F R W Henotesia ankaratra (Ward) Madagascar: 3 km S. Antalaha, 10 m Masoala 1M 1M S W Peninsula Henotesia ankaratra (Ward) Madagascar: Montagne d’Ambre 1M 1M S W Henotesia ankova (Ward) Madagascar: Ambohitantely 2M 2M R W Henotesia avelona (Ward) Madagascar: Anjanaharibe Sud 1160 m 1M R W Henotesia bicristata (Mabille) Madagascar: Masoala, Antsamanarana R. 540 m 1M 1M R N Henotesia comorana (Oberthu¨r) Comoros: Grande Comoro above M’Vouni 1M R N Henotesia exocellata (Mabille) [ϭHeteropsis Madagascar: Ankazomivady, 32 km S. of Ambositra 1M, 1F 1M, 1F R W antsianakana Oberthu¨r] Henotesia fraterna (Butler) [ϭnarcissus Fabr.] Madagascar: Masoala Peninsula 2M 2M S W Henotesia fuliginosa (Mabille) Madagascar: Ranomafana National Park 1M, 1F R N Henotesia laetifica (Oberthu¨r) Madagascar: Analalava 195 m, N.E. Madagascar 1M R N Henotesia maeva (Mabille) Madagascar: Masoala Peninsula 3M, 1F 2M, 1F R W Henotesia parva (Butler) Madagascar: Ranomafana National Park 1M, 1F 1M, 1F R W Henotesia parvidens (Mabille) Madagascar: Vohiparara 1200 m, Ranomafana 1F R N National Park Henotesia pauper (Oberthu¨r) Madagascar: Ranomafana National Park 1M, 1F 1M, 1F R W Henotesia peitho (Plotz) Nigeria: Oban Hills 1M R N Henotesia perspicua (Trimen) Tanzania: Tarangire National Park 1000 m 1M S W Henotesia sabas (Oberthu¨r) Madagascar: Anjanaharibe 1250 m 1F R N Henotesia strato (Mabille) Madagascar: Marojezy R. Lokoho 1M R N Henotesia strigula (Mabille) Madagascar: Masoala Peninsula 1M 1M R W Henotesia subsimilis (Butler) Madagascar: Ranomafana National Park 1M 2M R W Henotesia turbans (Oberthu¨r) Madagascar: Ranomafana National Park 2M, 1F 2M R W Henotesia turbata (Butler) Madagascar: Ranomafana National Park 1M, 1F 1M, 1F R N Henotesia uniformis (Oberthu¨r) Madagascar: Masoala Peninsula 1M, 2F 1M, 2F R W Henotesia vola (Ward) Madagascar: Ranomafana National Park 1F 1F R W Henotesia 7 (ϭpauper?) Madagascar: Masoala Peninsula 1F 2M R W Henotesia 10 (ϭdifficilis?) Madagascar: Ranomafana National Park 1M, 1F 2M R W Henotesia 12 [undescribed] Madagascar: Ranomafana National Park 1F R W Henotesia 14 [ϭsubsimilis?] Madagascar: Maromizaha, 28 km E Moramanga 1M 1M R N Henotesia 14A [ϭsubsimilis?] Madagascar: Masoala Peninsula 1F R N Henotesia 16 [undescribed] Madagascar: Ivontaka Sud 1F R N Henotesia 18 (ϭdifficilis?) Madagascar: Masoala Peninsula 1F 1F R N Henotesia 20 [undescribedϭsp. 77] Madagascar: Masoala Peninsula 1F R W Henotesia 22 [undescribed] Madagascar: Anjanaharibe Sud, 1240 m 2M 2M R N Henotesia 23 [undescribed] Madagascar: 32 km S. of Ambositra 2M 2M R N Henotesia 25 [ϭsubsimilis?] Madagascar: Ankazomivady, 32 km S. Ambositra 1F R N Henotesia 30 [undescribed] Madagascar: 32 km S. of Ambositra 1F R N Henotesia 32 [undescribed] Madagascar: Montagne d’Ambre 1M R N Henotesia 37 [undescribed] Madagascar: Andohahela 1M R N Henotesia 49 [undescribed] Madagascar: Anjanaharibe Sud 1F R N Henotesia 73 [undescribed] Madagascar: Anjozorobe 1400 m 1M R N Henotesia 74 [undescribed] Madagascar: Anjanaharibe Sud, 1350 m 1M R N Henotesia 75 [undescribed] Madagascar: Anjanaharibe Anjavidibe summit, 1530 m 2M R N Henotesia 76 [undescribed] Madagascar: Andavakoera 410 m 2M R N Henotesia 77 [undescribedϭsp.20] Madagascar: Anjanaharibe Sud 1240 m 1M R W Heteropsis drepana Westwood Madagascar: Anjanaharibe Sud 1320 m 1M R N Heteropsis drepana Westwood Madagascar: Andohahela 1M 1M R W Houlbertia narova (Mabille) Madagascar: Ranomafana National Park 1M 1M R W Houlbertia passandava (Ward) Madagascar: Ranomafana National Park 2M 2M R N Houlbertia wardii (Mabille) Madagascar: Masoala Peninsula 1M 2M R N Masoura antahala (Ward) Madagascar: Analalava 100 m 1M 1M R N [Femaleϭbenacus Mabille] Madagascar: Marojezy R. Lokolo Ambodimifotitra 1F 1F R W Japan 2M 1M

Note. Collection localities indicated on Fig. 1. Numbers of each sex sampled listed below gene region. M, male; F, female; R, rainforest; S, savannah; W, wide; N, narrow. MADAGASCAN BUTTERFLY PHYLOGENETICS 465 and dilutions varied), 5 ␮l10ϫ Mg-free reaction buffer ever, entire sequences containing indels were not ex-

(Promega), 2.5 mM MgCl2, 25 pmol each primer, 0.2 ␮l cluded because for the taxa listed above, the only se- Taq polymerase DNA (Promega) in a total volume of 50 quences that were available contained indels. In the ␮l. Amplifications were carried out in a Perkin–Elmer case of H. fraterna, we chose the exemplar that lacked Cetus DNA thermal cycler (Model 480), with conditions any indels for the analyses. The COII region included typically as follows: 28–35 cycles of denaturation at 174 parsimony-informative characters, whereas the 94°C for 1 min, 45–55°C annealing for 2 min, and 72°C combined data set of COII and cyt b contained 237. extension for 1.5 min; followed by an 8-min extension Mean pairwise sequence divergence for COII (uncor- at 72°C. PCR products were either gel purified by Gene rected for multiple hits but corrected for missing char- Clean (Bio 101) or purified with shrimp alkaline phos- acters) varied widely within each of the major - phatase and exonuclease. Direct sequencing was car- logical clades. Within the narova20 clade, mean ried out manually (Sequenase Kit, US Biochemical) sequence divergence ranged from 5.0 to 12.4% among followed by gel separation and autoradiography or by species, however, with a lower range of 0.95–2.7% dem- cycle sequencing (Prism kit; Applied Biosystems Inc.) onstrated for sister species pairs (which may in some followed by analysis with an ABI 373 automated se- cases represent morphospecies). Among the taxa be- quencer (Applied Biosystems Inc.). longing to the strigula15 clade (which was unresolved with COII data), the overall mean pairwise sequence Phylogenetic Analysis divergence ranged from 4.7 to 10.4%. Sister species Sequences were aligned manually (as the alignment pairs that were resolved by the COII data within was straightforward and unambiguous) and analyzed strigula15 showed a range of 1.7–1.9%. The bicristata4 assuming the parsimony optimality criterion and sub- clade demonstrated a range of sequence divergence sequently by maximum-likelihood for comparison. All from 7.4 to 11.4%. Within the subsimilis7 clade, a range characters were assigned equal weight (gapped posi- of 2.8–8.8% was demonstrated, with very low sequence tions excluded) and heuristic searches (stepwise addi- divergence among the morphospecies assigned to He- tion, random addition of taxa, 100 replicates, one tree notesia pauper (0–0.76%). held at each replicate) were performed in PAUP* 4.0 Phylogenetic Analyses beta version (Swofford, 1998). Results of alternative weighting schemes (i.e., transversions weighted, third Cytochrome oxidase II. Equally weighted parsi- positions weighted) and maximum-likelihood (HKY85 mony analysis of the COII region with three outgroup model [to reflect unequal base frequences and transi- genera included (Bicyclus, Mycalesis, and Hallelesis tion/transversion biases]) are presented in tabular [Hallelesis and Mycalesis designated as outgroups]), form (phylogenetic trees are not presented here). An resulted in three equally parsimonious trees (Fig. 2A). estimate of the support at each node on the parsimony Mycalesis falls within a large clade of mostly Malagasy trees was assessed by 100 bootstrap replications species. The Bicyclus species fall into two separate (Felsenstein, 1985) and Bremer support indices (Bre- positions internal to the Malagasy radiation, with B. mer, 1994) with AutoDecay v 4.0.1 [heuristic search cottrelli and B. anynana forming a clade. option, random addition of taxa, number of repli- The large genus Henotesia is clearly not monophy- catesϭ20] (Eriksson, 1998). letic, as suggested by morphology (Lees, 1997). The Malagasy genera Houlbertia, Heteropsis, and Admira- tio are nested within a large clade consisting mostly of RESULTS both African and Malagasy representatives of Henote- sia. The genus Houlbertia contains eight species Data Matrix Description (D’Abrera, 1980; Ackery et al., 1995) or nine species Base frequencies for the COII and cyt b regions were with the addition of Henotesia peitho (D’Abrera, 1997). comparable (A:33%, C:15%, G:13%, and T:39% for COII However, H. peitho clearly falls external to the well- and A:31%, C:17%, G:9%, and T:43% for cyt b). Tran- supported narova20 clade that contains the four repre- sition/transversion ratios were 2.35:1 for the COII re- sentatives of Houlbertia included in this study. Houl- gion and 2.48:1 for cyt b. There were no indels in the bertia passandava falls here as sister to an cyt b region and two in the coding region of COII. The undescribed species pair and not to the other species of first indel was three bases in length and was present in Houlbertia. Heteropsis is currently treated as dibasic Henotesia fuliginosa and in H. fraterna. The second (D’Abrera, 1980; Ackery et al., 1995); however, one of indel varied in length and was either three (Bicyclus the species (Heteropsis antsianakana) is a synonym of zinebi and Houlbertia narova), six (Henotesia fuligi- Henotesia exocellata (Lees, 1997). The COII results nosa), or nine (Henotesia vola) bases in length. Gapped suggest that H. exocellata has a different sister taxon characters corresponding to these indels were excluded than Heteropsis drepana, a result also suggested by from the analyses because in H. fraterna they were morphology. Admiratio is monobasic. The genus Ma- found in a single exemplar of two for the taxon. How- soura currently contains five species (Ackery et al., 466 TORRES ET AL.

FIG. 2. (A) Strict consensus of three equally parsimonious trees based on analysis of COII sequence data. All characters assigned equal weights and gapped positions were excluded from the analysis. Numbers above branches represent bootstrap values Ͼ50%; numbers below indicate decay indices. Taxa from localities other than Madagascar indicated by *. Tree Length ϭ 955, CI ϭ 0.2588, RI ϭ 0.5189, RC ϭ 0.1532. (B) Phylogram of one of the most equally parsimonious trees used to construct the strict consensus tree.

1995); however, we have sequenced only a single rep- peitho, H. perspicua, and nine other species) was sug- resentative and therefore cannot address its mono- gested to be paraphyletic by the morphological analy- phyly. ses of Lees (1997). Sequencing of an additional African The two African species of Henotesia that we sam- hairy-eyed clade (e.g., the dibasic Henotesia elisi group; pled (H. peitho and H. perspicua) fell in two different Kielland, 1994) may be required to further clarify the places in the equally weighted COII strict consensus position of African hairy-eyed mycalesines. tree (Fig. 2A) and were internal to the Malagasy radi- The equally weighted COII analysis (Fig. 2A) unam- ation. With transversions at all three codon positions biguously supported two major morphological clades: weighted 2:1 (trees not shown), H. peitho and H. per- the subsimilis7 group and the narova20 group (with the spicua appear as sister taxa and to most of the exception of the taxon Henotesia vola, which is often Malagasy taxa. However, this result was not well sup- enigmatic in the morphological analyses). A fully re- ported with either bootstrap or decay indices. Relation- solved, although less-well-supported, bicristata4 group ships of the African hairy-eyed species of Henotesia (H. was also recovered. These three morphologically de- MADAGASCAN BUTTERFLY PHYLOGENETICS 467

FIG. 2—Continued

fined clades were recovered under all weighting plus Henotesia ankaratra and Heteropsis plus Ma- schemes assuming parsimony and by maximum-likeli- soura. Relationships among the major morphological hood (Table 3). The strigula15 group is rendered clades and outgroup genera are not supported. paraphyletic by the intrusion of several taxa under A phylogram of one of the equally parsimonious trees equal weighting and fewer taxa under differential (Fig. 2B) reveals relatively longer branches for Halle- weighting schemes and maximum-likelihood. A mono- lesis halyma, Bicyclus zinebi, B. cottrelli, B. anynana, phyletic strigula15 group has not been recovered under Henotesia peitho, Henotesia turbata, Henotesia per- any weighting scheme or method used (Table 3). The spicua, Henotesia fuliginosa, Heteropsis drepana, He- angulifascia5 group (nested within the strigula15 notesia exocellata, Masoura antahala, Admiratio para- group) is also paraphyletic except under maximum- doxa, Henotesia fraterna, Henotesia parvidens, and likelihood and under parsimony with transversions Henotesia parva. The phylogenetic placement of sev- weighted 3:1 (Table 3). The other major morphological eral of these “rogue” taxa listed above were also poorly clades that are not supported with COII are Admiratio resolved in the morphological analyses. The inclusion 468 TORRES ET AL.

TABLE 3 which is phenotypically distinct from H. pauper (Kre- men, 1992, 1994; Lees, 1997). Support for the Monophyly of Morphological Species Groups Simultaneous analysis of cytochrome b and cyto- chrome oxidase II for 26 taxa. Simultaneous analysis Morphological Equal 3rd Position of cyt b and COII for 26 taxa (with missing data for the species groups weights Tv 2:1 Tv 3:1 Tv 2:1 ML COII region for H. sp. 14) resulted in seven equally narova 63/4 76/7 76/10 64/4 M parsimonious trees (Fig. 3). The strigula15, narova20, strigula NM NM NM NM NM and subsimilis7 groups were supported, but the bicris- maeva NM NM NM NM NM tata group was not recovered. There was no support angulifascia NM NM Ͻ50/1 NM M 4 subsimilis 97/8 100/13 100/16 97/10 M for relationships among the major morphological bicristata Ͻ50/1 Ͻ50/3 52/6 Ͻ50/1 M clades in any of the trees for the subset of 26 taxa. ankova NM NM NM NM NM Bicyclus NM NM NM NM NM DISCUSSION Note. Tv, transversions; ML, maximum-likelihood; M, monophy- letic; NM, nonmonophyletic. Bootstrap value/decay index indicated Traditional Generic Groupings for each analysis. Mitochondrial DNA sequences contradict monophyly of the large hairy-eyed Afrotropical mycalesine genus Henotesia. This result is not surprising considering the of Henotesia fuliginosa and B. zinebi had a strong and size of the genus (56 species listed in Ackery et al., destabilizing effect on the topologies of morphological 1995) and the taxonomic assignment of species to this trees (Lees, 1997). genus based principally on wing shapes and patterns Several terminal sister pairs in the COII tree (Fig. and the single less-labile character of hairy eyes. The 2A) are supported by morphology: the allopatric spe- presence or absence of setae on the compound eye is a cies pairs (Henotesia angulifascia ϩ H. sp. 73), (Heno- character that has been used for and system- tesia sp. 22 ϩ H. sp. 76), (Henotesia sp. 77 ϩ H. sp. 12), atics in other groups. However, it has been and (Henotesia sp. 75 ϩ H. strato). Several taxon pairs shown that in Drosophila multiple types of genetic that contradict the previous morphological analysis are mutations can code for an identical phenotype with suggested by the COII tree: (Henotesia sabas ϩ H. respect to the presence or absence of setation on the turbans), (Henotesia ankova ϩ H. sp. 49), (Henotesia compound eye (DeSalle and Grimaldi, 1993). Moreover, sp. 32 ϩ H. sp. 74), and (Admiratio paradoxa ϩ Heno- within the Mycalesina, this character appears useful tesia peitho). Some sister pairs within the subsimilis7 for systematics only as a loss character (loss of eye clade are also in conflict with morphology, but the setation is known throughout Bicyclus and in the di- newly suggested sister pairs are often supported by basic Hallelesis). However, even here caution is re- certain morphological synapomorphies (Lees, 1997). quired: stereoscan investigations of various naked- However, not all taxa of the Madagascan radiation eyed nymphalids reveals that, in reality, many have have been obtained for sequencing and are, therefore, very short inter-ommatidial hairs, suggestive that this not represented in this study. feature may readily be reversible (R. I. Vane-Wright, Identities of allopatric morphospecies suggested by unpublished). Kremen (1992, 1994) and Lees (1997) are also sup- We attribute the of Bicyclus in the COII ported by the COII data. Two morphospecies (Henote- trees in this study partially to taxon sampling prob- sia sp. 10 and H. sp. 18) attributed by Kremen (1992, lems within the large genus Bicyclus which contains 1994) and Lees (1997) to H. difficilis form a highly about 80 species (D’Abrera, 1997), but more likely to a supported clade. Henotesia difficilis exhibits strong lack of resolution of any basal clades with this partic- sexual dimorphism, such that the male and female ular region of COII. We consider that the relationships were originally placed into three additional different of the Malagasy taxa to African taxa is best addressed species: Henotesia undulosa (Oberthu¨ r) (male), H. in a large simultaneous analysis with dense sampling macropthalma (Oberthu¨ r) (female), and H. iboina of both geographic radiations (see Omland et al., 1999) (Ward) f. fitensis (female: Oberthu¨ r, 1916; Lees, 1997; and/or the addition of genes which will potentially re- see also d’Abrera, 1997). Within the subsimilis7 species solve the more basal relationships within the subtribe group, the molecular data suggest that H. pauper is Mycalesina. rather indistinct from H. sp.7[ϭH. pauper (Lees Future Higher Systematic Revision 1997)] and morphospecies H. sp. 23 and H. sp. 30. Morphospecies H. sp. 23 and H. sp. 30 are sympatri- The results presented here suggest a promising basis cally endemic to one of the very few remaining isolated for a fresh higher-taxic classification of Afrotropical plateau forests in Madagascar. They represent male mycalesines that should commence with those clades and female, respectively, of the same dimorphic taxon that prove robust within existing molecular and mor- MADAGASCAN BUTTERFLY PHYLOGENETICS 469

FIG. 3. Strict consensus of seven trees based on equally weighted simultaneous parsimony analysis of cyt b and COII for 26 taxa. Tree Length ϭ 1026; CI ϭ 0.3640; RI ϭ 0.4244; RC ϭ 0.1787. phological analyses. Three morphology-based clades Henotesia bicristata was never suggested to be closely

(the subsimilis7, narova20, and bicristata4 clades) are related to the avelona3 group prior to the cladistic resolved here, whereas monophyly of the strigula15 and morphological study of Lees (1997), and this position is turbata3 clades have not yet been justified. The compo- also corroborated by molecular data. sition of these groups may well require readjustment. Male genitalia have provided many important mor- Molecular results may be valuable to reinterpret the phological synapomorphies, supporting a number of value of different morphological data sets (i.e., wing vs clades also suggested by molecular data. For example, genitalic characters). the angulifascia5 group with its extremely distinctive

One of the most distinctive morphological synapo- antler-like gnathos and the subsimilis7 group with its morphies within the radiation is the dorso-ventrally symmetrically skittle-shaped valve base (Lees, 1997)

flattened aedeagus that characterizes the narova20 are supported by our molecular results. In fact, male clade. Although the genus Houlbertia is likely to be genitalia may provide a more informative morphologi- paraphyletic, all members fall within the narova20 cal character system than either wing ocellus homo- clade and possess a dorso-ventrally flattened aedeagus, logues or venation, two character sets examined by a placement entirely consistent with the molecular re- Lees (1997). In this context it is interesting to note that sults. Houlbertia is recognized phenetically by the dark de Jong et al. (1996, Table 10) found butterfly wing blue/violet metallic coloration of the male, but this venation characters to be the most homoplasious character also occurs in traditional Henotesia (e.g., H. among five partitions of adult morphological features. obscura; D’Abrera, 1980) and adults of Houlbertia ap- The remarkably differentiated inflated vein system pear otherwise poor in apomorphic characters. In con- of the male hindwing within the Madagascar radiation, trast, the bicristata4 clade is strongly autapomorphic. while lacking in the narova20 clade, is very strongly 470 TORRES ET AL.

developed in the bicristata4, strigula15 and subsimilis7 higher percentage of taxa sampled (Table 1). Separate clades and provides many potential morphological sy- analyses of cyt b and COII for a subset of 26 taxa napomorphies (Lees, 1997). Only the inflated vein resulted in poorly resolved and conflicting topologies characters of the bicristata4 and the subsimilis7 clades for each gene alone (data not shown). The combination and a loss of the system within narova20 are congruent of the two gene regions for the subset of 26 taxa did not with our molecular data. Because only one exemplar substantially improve the resolution, although in the was sampled, we were unable to assess here the mono- combined analysis the strigula15 clade was resolved. phyly of the morphologically distinctive genus Ma- The exemplar approach (choosing representative taxa soura. However, the monophyly of Heteropsis plus Ma- from each clade in one cladogram) to taxon sampling soura, which has been suggested by a unique may be inadequate for this large and closely related configuration of inflated veins, alar scent brushes, and radiation. We plan to sequence an additional 16ϩ taxa similar uncus shapes, is not supported by the molecu- within the island radiation to further clarify phyloge- lar data. This adds support to the likelihood of devel- netic relationships within the group. opmental plasticity and convergence in the inflated Biogeography vein system, which appears to lack strong phylogenetic signal. Likewise, Ackery and Vane-Wright (1984) The explosive radiation of mycalesine species in the found the inflated hindwing veins in many Parantica Afrotropical region provides a potentially informative species (Danainae) to be homoplasious, even though scenario with which to test hypotheses of biogeography suggestive of various possible affinities. and dispersal in the Malagasy region. Miller’s (1968) The sister taxon relationships among the major mor- hypothesis that both naked-eye and hairy-eyed my- phological clades and with a number of “rogue” taxa calesines spread from Asia to Africa, with a subsequent including Masoura still remains obscure. Undoubtedly, dispersal of hairy-eyed species to Madagascar, appears a simultaneous analysis of mitochondrial DNA with to be contradicted by the appearance in this analysis of nuclear DNA (i.e., wingless) would generate a more Henotesia peitho and H. perspicua nested within the robust hypothesis based on genetic data; such analyses Madagascan radiation. Interestingly, this also oc- have considerable potential to tease out hidden phylo- curred in the morphological analyses of Lees (1997) genetic signal from independent data sets. This has when all 11 African Henotesia species were included been shown, for example, in the case of a comparable but when no African Henotesia was specified as an radiation of nymphalid butterflies and at a more basal outgroup. H. peitho falls in both studies in a different level (e.g., Brower and Egan, 1997; Brower and position within the tree. Also, the sister taxon relation- DeSalle, 1994, 1998). Ultimately, these additional data ships are different for these two taxa in molecular and will be used in combination with morphology once com- morphological trees. In no case, however, did H. peitho pletely compatible data sets are obtained. group with taxa within the paraphyletic (or perhaps polyphyletic) genus Houlbertia, in which it was com- Alphataxonomy bined by d’Abrera (1997). Whereas none of these basal The utility of both the cyt b and the COII genes for relationships are strongly supported, at least an in- alphataxonomy is confirmed and, at least in the case of triguing new biogeographic hypothesis is implied. Henotesia pauper, suggests that morphospecies H. sp. There may have been a minimum of two invasions into 23 and H. sp. 30 may each be allopatric populations Africa from Madagascar, one putatively within a rain- within H. pauper. Analysis of the COII and cyt b se- forest-adapted lineage (that included an ancestor of H. quence data, consisting of multiple exemplars for each peitho). A westward direction of dispersal across the taxon, was consistent with the assignments of species Mozambique channel is far more probable climatically, recognized by Kremen (1992, 1994) and Lees (1997). especially in the case of weakly flighted butterflies Although in this study we did not explicitly sample (Lees, 1997). Colonization of Africa from Madagascar taxa to test Lees’s hypotheses of conspecificity of sexual has also been inferred to occur on several occasions and seasonal forms in this highly confused butterfly based on a cladistic analysis of the considerably more group, no radical conflicts are apparent at this stage of dispersive nymphalid butterfly genus Acraea (Pierre, sampling. In all cases, sequences for both sexes of 1994). Moreover, even the most widespread and eury- morphospecies were identical. The potential of mtDNA topic species of Bicyclus have not penetrated the Mal- to clarify or confirm synonymic hypotheses based on agasy region from the Ethiopian Region further east geographic variation and seasonal or sexual polymor- than the Comoros, and this may have happened just phism in these butterflies thus seems considerable. once, with subsequent island hopping within the Co- moros as demonstrated by Bicyclus anynana (see maps Effects of Taxon Sampling in Condamin, 1973). The paucity of Lycaenidae and Taxon sampling density may have affected the lack Danainae in Madagascar, which are well represented of resolution within certain morphological clades, with in Africa, also suggests that colonization from west to the more consistently recovered clades containing a east has been difficult. Carcasson (1964) suggested MADAGASCAN BUTTERFLY PHYLOGENETICS 471 that a close relationship between two kite swallowtails ilionoidea and Hesperoidea of the Afrotropical Region” (P. R. Ack- (Graphium evombar in Madagascar and G. junodi in ery, C. R. Smith, and R. I. Vane-Wright, Eds.), pp. 287–303. Mozambique) offered evidence of the reverse flow, but CSIRO, Melbourne. this supposed close relationship is now considered to be Aoki, T., Yamaguchi, S., and Uemura, Y. (1982). “Butterflies of the South East Asian Islands. III. Satyridae, Libytheidae,” Plapac, false (Smith and Vane-Wright, 2000). These remarks Tokyo. beg the question of the ultimate origin(s) of the hairy- Asher, R. J. 1999. A morphological basis for assessing the phylogeny eyed mycalesine radiation in Madagascar. of the “Tenrecoidea” (Mammalia, Lipotyphla). 15: 231– New phylogenetic signal from additional genes, com- 252. bined with more exhaustive taxon sampling of African Baum, D. A., Small, R. L., and Wendel, J. F. (1998). Biogeography and Asian taxa, will help to resolve the history of and floral evolution of Baobabs (Adansonia, Bombacaceae) as in- origination within this group. Molecular data have tre- ferred from multiple data sets. Sys. Biol. 47: 181–207. mendous potential to illuminate evolutionary history, Brakefield, P. M. (1998). The evolution–development interface and for example to examine the tempo and modes of origi- advances with the eyespot patterns of Bicyclus butterflies. Hered- ity 80: 265–272. nation that pertain to this large radiation, and this Brakefield, P. M., Gates, J., Keys, D., Kesbeke, F., Wijngaarden, could help shed light on basic questions already posed P. J., Monteiro, A., French, V., and Carroll, S. B. (1996). Develop- about Lepidoptera diversification in Africa and the In- ment, plasticity and evolution of butterfly eyespot patterns. Na- dian Ocean (Vane-Wright, 1997). Hypotheses of gradu- ture 384: 236–242. alistic or punctuated rates of diversification can be Brakefield, P. M., Kesbeke, F., and Koch, P. B. (1998). The regulation tested within the context of adaptive radiations, pro- of phenotypic plasticity of eyespots in the butterfly Bicyclus vided that sister taxon relationships offering controls anynana. Am. Nat. 152: 853–860. for age can be firmly established. Patterns of diversity, Bremer, K. (1994). Branch support and tree stability. Cladistics 10: especially species richness, strongly reflect gradients of 295–304. diversity for other taxa on Madagascar (Lees et al., Brower, A. V. Z. (1994). Phylogeny of Heliconius butterflies inferred 1999). Knowledge of phylogenetic and geographic di- from mitochondrial DNA sequences (Lepidoptera: Nymphalidae). Mol. Phylogenet. Evol. 3: 159–174. versity patterns of narrow endemism in these butter- Brower, A. V. Z. (2000). Phylogenetic relationships among the flies has the potential to better inform the conservation Nymphalidae (Lepidoptera), inferred from partial sequences of the planning process: comparisons with other cladistic wingless gene. Proc. R. Soc. Lond. B 267: 1201–1211. analyses will allow much more rigorous tests of levels Brower, A. V. Z., and Egan, M. G. (1997). Cladistic analysis of of representativeness in geographic patterns of genetic Heliconius butterflies and relatives (Nymphalidae: Heliconiiti): A diversity. revised phylogenetic position for Eueides based on sequences from mtDNA and a nuclear gene. Proc. R. Soc. Lond. B 264: 969–977. Brower, A. V. Z., and DeSalle, R. (1994). Practical and theoretical ACKNOWLEDGMENTS problems for choice of a DNA sequence region in insect molecular systematics, with a short review of published studies using nuclear We thank the Xerces Society, the Wildlife Conservation Society, gene regions. Ann. Entomol. Soc. Am. 87: 702–716. and the National Geographic Society for support for field collections. Brower, A. V. Z., and DeSalle, R. (1998). Patterns of mitochondrial This was helped in particular by an accord with the University of versus nuclear DNA sequence divergence among nymphalid but- Antananarivo, and we express our thanks to the late Vincent terflies: The utility of wingless as a source of characters for phy- Razafimahatratra for helping with permits, to all the students and logenetic inference. Insect Mol. Biol. 7: 73–82. other Malagasy who helped in many ways with fieldwork, and to the government of the Comoros and to Patricia Moehlman for arranging Caccone, A., Amato, G., Gratry, O. C., Behler, J., and Powell, J. R. permits to supply us with additional Afrotropical material. We are (1999). A molecular phylogeny of four endangered Madagascar most grateful to Dr. Takahishi of Japan for supplying us with DNA tortoises based on mtDNA sequences. Mol. Phylogenet. Evol. 12: of Mycalesis francisca. Ilik Saccheri carried out many of the DNA 1–9. extractions. We thank in particular Nurit Ziv and Veronica Villa at Carcasson, R. H. (1964). A preliminary survey of the zoogeography of UCLA for laboratory assistance. Antonia Monteiro and Andy Brower African butterflies. East Afr. Wildl. J. 2: 122–157. assisted greatly with various queries. The morphological work by Carroll, S. B., Keys, D. N., Paddock, S. W., Panganiban, G. E. F., David Lees was funded by a BBSRC grant (No. 93307999) to Robert Selegue, J. E., and Williams, J. A. (1994). Pattern formation and Wayne. Sequencing was funded by an NSF grant (No. 9225059) to eyespot determination in butterfly wings. Science 265: 109–114. Claire Kremen and by a Leverhulme Foundation grant (F/696/F) to Dick Vane-Wright. A travel grant was awarded to David Lees from the Carson, H. L. (1968). The population flush and its genetic conse- CEA-CREST program at California State University, Los Angeles. quences. In “Population Biology and Evolution” (R. C. Lewontin, Ed.), pp. 123–137. Syracuse Univ. Press, Syracuse, NY. Cibois, A., Pasquet, E., and Schulenberg, T. S. (1999). Molecular REFERENCES systematics of the Malagasy babblers (Passeriformes: Timaliidae) and warblers (Passeriformes: Sylviidae), based on cytochrome b Ackery, P. R. (1988). Hostplants and classification: A review of and 16S rRNA sequences Mol. Phylogenet. Evol. 13: 581–595. nymphalid butterflies. Biol. J. Linn. Soc. 33: 95–203. Colonna, M., and Casanova, J. (1996). Sea-level changes and ␦18O Ackery, P. R., and Vane-Wright, R. I. (1984). “Milkweed Butterflies,” record for the past 34,000 yr from Mayotte reef, Indian Ocean. Cornell Univ. Press, New York. Quatern. Res. N. Y. 46: 335–339. Ackery, P. R., Smith, C. R., and Vane-Wright, R. I. (1995). In “Car- Condamin, M. (1973). Monographie du Genre Bicyclus (Lepidoptera: casson’s African Butterflies: An Annotated Catalogue of the Pap- Satyridae). Mem. Inst. Fondam. Afrique Noire 88: 1–324. 472 TORRES ET AL. d’Abrera B. L. (1980). “Butterflies of the Afrotropical Region,” 1st ed. Kremen, C. (1994). Biological inventory using target taxa: A Lansdowne Editions, Melbourne. case study of the butterflies of Madagascar. Ecol. Appl. 4: 407– d’Abrera B. L. (1997). “Butterflies of the Afrotropical Region, Part 1. 422. Papilionidae, Pieridae, Nymphalidae, Satyridae,” 2nd ed., Hill Kull, C. A. (1996). The evolution of conservation efforts in Madagas- House, London. car. Int. Environ. Affairs 8: 50–86. Darwin, C. (1859). “On the Origin of Species by Means of Natural Lees, D. C. (1997). “Systematics and Biogeography of Madagascan Selection, or the Preservation of Favoured Races in the Struggle Mycalesine Butterflies (Lepidoptera: Satyrinae),” PhD thesis, Uni- for Life,” John Murray, London. versity of London. de Jong, R., Vane-Wright, R. I., and Ackery, P. R. (1996). The higher Lees, D. C., Kremen, C., and Andriamampianina, L. (1999). A null classification of butterflies (Lepidoptera): Problems and prospects. model for species richness gradients: Bounded range overlap of Entomol. Scand. 27: 65–101. butterflies and other rainforest endemics in Madagascar. Biol. J. Del Pero, M., Crovella, S., Cervella, P., Ardito, G., and Rumpler, Y. Linn. Soc. 67: 529–584. (1995). Phylogenetic relationships among Malagasy lemurs as re- Martins-Neto, R. G., Kucera-Santos, J. C., Moraes-Vieira, F. R. vealed by mitochondrial DNA sequence analysis. Primates 36: de., and Campos Fragos, L. M. de. (1993). Nova especie de 431–440. borboleta (Lepidoptera: Nymphalidae: Satyrinae) de Formacao DeSalle, R., and Grimaldi, D. (1993). Phylogenetic patterns and Tremembe, Oligoceno do Estado de Sao Paulo. Acta Geol. developmental process in Drosophila. Syst. Biol. 42: 458–475. Leopold 16: 5–16. Dobzhansky, T. (1951). “Genetics and the Origin of Species,” 3rd ed., Mayr, E. (1963). “ Species and Evolution,” Belknap Press of Colombia Univ. Press, New York. Harvard Univ. Press, Cambridge, MA. Dubois, J. Y., Rakotondravony, D., Hanni, C., Sourrouille, P., and McCall, R. A. (1997). Implications of recent geological investigations Catzeflis, F. M. (1996). Molecular evolutionary relationships of of the Mozambique Channel for the mammalian colonization of three genera of Nesomyinae, endemic rodent taxa from Madagas- Madagascar. Proc. R. Soc. Lond. B 264: 663–665. car. J. Mammal. Evol. 3: 239–260. Miller, L. D. (1968). The higher classification, phylogeny and zooge- Eliot, J. N. (1992). In “The Butterflies of the Malay Peninsula by A. ography of the Satyridae (Lepidoptera), Mem. Am. Entomol. Soc. Steven Corbet and H. M. Pendlebury” (J. N. Eliot, Ed.), 4th ed., pp. 24: 1–174. 126–127. Malayan Nature Society, Kuala Lumpur. Monteiro, A., Brakefield, P. M., and French, V. (1997). Butterfly Emberton, K. C. (1995). Phylogenetic analysis of 18 species of Mada- eyespots: The genetics and development of the color rings. Evolu- gascan acavid land snails using allozyme characters. Veliger 38, tion 51: 1207–1216. 1–7. Muller, J. (1981). Fossil pollen records of extant angiosperms. Bot. Eriksson, T. (1998). “AutoDecay ver. 4.0.” (program distributed by Rev. 47: 1–142. the author). Department of Botany, Stockholm University, Stock- Nel, A., Nel, J., and Balme, C. (1993). Un nouveau Le´pidopte`re holm. Satyrinae fossile de l’Oligoce`ne du sud-est de la France (Insecta, Felsenstein, J. (1985). Confidence limits on phylogenies: An ap- Lepidoptera, Nymphalinae). Linneana Belg. 14: 20–36. proach using the bootstrap. Evolution 39: 783–791. Oberthu¨ r, C. (1916). Observations sur une centurie d’espe`ces de Haq, B. U., Hardenbol, J., and Vail, P. R. (1987). Chronology of Le´pidopte`res Rhopaloce`res malagaches. E´ tud. Le´p. Comp. 11: fluctuating sea levels since the Triassic. Science 235: 1156–1167. 177–244. Holloway, G. J., and Brakefield, P. M. (1995). Artificial selection of Omland, K. E., Lanyon, S. M., and Fritz, S. J. (1999). A molecular reaction norms of wing pattern elements in Bicyclus anynana. phylogeny of the new world orioles (Icterus): The importance of Heredity 74: 91–99. dense taxon sampling. Mol. Phylogenet. Evol. 12: 224–239. Holloway, G. J., Brakefield, P. M., and Kofman, S. (1993). The Otte, D. (1989). Speciation in Hawaiian crickets. In “Speciation and genetics of wing pattern elements in the polyphenic butterfly, Its Consequences” (D. Otte and J. A. Endler, Eds.), pp. 482–526. Bicyclus anynana. Heredity 70: 179–186. Sinauer, Sunderland, MA. Jansa, S. A., Goodman, S. M., and Tucker, P. K. (1999). Molecular Parsons, M. J. (1998). “Butterflies of New Guinea,” Academic Press, phylogeny and biogeography of the native rodents of Madagascar Oxford. (Muridae: Nesomyinae): A test of the single-origin hypothesis. Pierre, J. (1994). Les Acraeidae des Comores et de Madagascar Cladistics 15: 253–270. (Lepidopte`res Rhopaloce`res). Entomologiste 48: 351–364. Johnson, K. P., Goodman, S. M., and Lanyon, S. M. (2000). A phylo- Rabinowitz, P. D., Coffin, M. F., and Falvey, D. (1983). The separa- genetic study of the Malagasy Couas with insights into cuckoo tion of Madagascar and Africa. Science 220: 67–69. relationships. Mol. Phylogenet. Evol. 14: 436–444. Kerdelhue, C., Le Clainche, I., and Rasplus, J. Y. (1999). Molecular Rumpler, Y., Crovella, S., and Montagnon, D. (1994). Systematic phylogeny of the Ceratosolen species pollinating Ficus of the sub- relationships among Cheirogaleidae (Primates, Strepsirhini) de- genus Sycomorus sensu stricto: Biogeogeographical history and termined from analysis of highly repeated DNA. Folia Primatol. origins of the species-specificity breakdown cases. Mol. Phylogenet. 63: 149–155. Evol. 11: 401–414. Simpson, G. G. (1940). Mammals and land bridges. J. Wash. Acad. Kielland, J. (1994). A revision of the genus Henotesia (excluding Sci. 30: 137–163. Madagascar and other Indian Ocean Islands). Lambillionea 94: Smith, C. R., and Vane-Wright, R. I. (2000). “A Revision of African 235–273. Graphium (Lepidoptera: Papilionidae).” Krause, D. W., Rogers, R. R., Forster, C. A., Hartman, J. H., Buckley, Swofford, D. L. (1998). PAUP*: Phylogenetic analysis using par- G. A., and Sampson, S. D. (1999). The Late Cretaceous vertebrate simony (and other methods), version 4.0. Sinauer, Sunderland, fauna of Madagascar: Implications for Gondwanan paleobiogeog- MA. raphy. GSA Today 9: 2–7. Vane-Wright, R. I. (1971). The systematics of Drusillopsis Oberthur Kremen, C. (1992). Assessing the indicator properties of species (Satyrinae) and the supposed Amathusiid Bigaena van Eecke assemblages for natural areas monitoring. Ecol. Appl. 2: 203– (Lepidoptera: Nymphalidae), with some observations on Batesian 217. mimicry. Trans. R. Entomol. Soc. Lond. 123: 97–123. MADAGASCAN BUTTERFLY PHYLOGENETICS 473

Vane-Wright, R. I. (1997). Mode and tempo in the evolution of mim- Yoder, A. D., Cartmills, M., Ruvolo, M., Smith, K., and Vilgalys, R. icry in African butterflies—A research programme. Metamorpho- (1996). Ancient single origin for Malagasy primates. Proc. Natl. sis (Suppl. 3): 133–138. Acad. Sci. USA 93: 5122–5126. Yoder, A. D. (1994). Relative position of the Cheirogaleidae in strep- Yoder, A. D., and Irwin, J. A. (1999). Phylogeny of the Lemuridae: sirhine phylogeny: A comparison of morphological and molecular Effects of character and taxon sampling on resolution of species methods and results. Am. J. Phys. Anthropol. 94: 25–46. relationships within Eulemur. Cladistics 15: 351–361.