Skip to main content
NCBI home page
Proceedings of the Royal Society B: Biological Sciences logoLink to Proceedings of the Royal Society B: Biological Sciences
. 2005 Jul 11;272(1572):1577–1586. doi: 10.1098/rspb.2005.3124

Synergistic effects of combining morphological and molecular data in resolving the phylogeny of butterflies and skippers

Niklas Wahlberg 1,*, Michael F Braby 2,5, Andrew VZ Brower 3, Rienk de Jong 4, Ming-Min Lee 3, Sören Nylin 1, Naomi E Pierce 5, Felix AH Sperling 6, Roger Vila 5, Andrew D Warren 3, Evgueni Zakharov 6
PMCID: PMC1560179  PMID: 16048773

Abstract

Phylogenetic relationships among major clades of butterflies and skippers have long been controversial, with no general consensus even today. Such lack of resolution is a substantial impediment to using the otherwise well studied butterflies as a model group in biology. Here we report the results of a combined analysis of DNA sequences from three genes and a morphological data matrix for 57 taxa (3258 characters, 1290 parsimony informative) representing all major lineages from the three putative butterfly super-families (Hedyloidea, Hesperioidea and Papilionoidea), plus out-groups representing other ditrysian Lepidoptera families. Recently, the utility of morphological data as a source of phylogenetic evidence has been debated. We present the first well supported phylogenetic hypothesis for the butterflies and skippers based on a total-evidence analysis of both traditional morphological characters and new molecular characters from three gene regions (COI, EF-1α and wingless). All four data partitions show substantial hidden support for the deeper nodes, which emerges only in a combined analysis in which the addition of morphological data plays a crucial role. With the exception of Nymphalidae, the traditionally recognized families are found to be strongly supported monophyletic clades with the following relationships: (Hesperiidae+(Papilionidae+(Pieridae+(Nymphalidae+(Lycaenidae+Riodinidae))))). Nymphalidae is recovered as a monophyletic clade but this clade does not have strong support. Lycaenidae and Riodinidae are sister groups with strong support and we suggest that the latter be given family rank. The position of Pieridae as the sister taxon to nymphalids, lycaenids and riodinids is supported by morphology and the EF-1α data but conflicted by the COI and wingless data. Hedylidae are more likely to be related to butterflies and skippers than geometrid moths and appear to be the sister group to Papilionoidea+Hesperioidea.

Keywords: molecular systematics, total evidence, Insecta

1. Introduction

Butterflies are arguably the best loved group of invertebrates and have been a source of inspiration for generations of natural historians and scientists. As a result, their generic- and specific-level classification is reasonably stable and the majority of taxa have been named (Ackery et al. 1999). Also unique for a large group of invertebrates is the immense biological knowledge amassed for many species, allowing butterflies to be used as a model group of organisms for wide ranging studies in ecology, evolution, population genetics and developmental biology (Boggs et al. 2003). Despite this wealth of information, the phylogenetic relationships and higher classification of the major groups of butterflies have remained contentious and competing hypotheses lack strong empirical support (Ehrlich 1958; Scott 1985; de Jong et al. 1996; Vane-Wright 2003).

The classification of butterflies and skippers has been based almost exclusively on the morphology of adult specimens for close to 250 years (Ackery et al. 1999), despite the rampant homoplasy in these morphologically variable insects. The usefulness of characters from immature stages has long been acknowledged (Müller 1886) but they have only recently been automatically incorporated in a phylogenetic context (Harvey 1991; Penz & Peggie 2003; Freitas & Brown 2004). Study of these features remains severely hampered by the lack of detailed descriptions or preserved specimens of larvae and pupae from which characters may be discerned.

Traditionally, the butterflies and skippers have been placed in two super-families and five families: Hesperiidae (skippers) are usually placed in their own super-family Hesperioidea while all other butterflies (Papilionidae, Pieridae, Lycaenidae and Nymphalidae) are placed in Papilionoidea. There is little agreement on the rank and monophyly of various groups and relationships among and within the families are largely unresolved (Vane-Wright 2003). For example, Riodinidae are sometimes separated from Lycaenidae, and Nymphalidae have been divided into as many as nine families. Papilionoidea and Hesperioidea have traditionally been considered sister taxa but recently the moth-like family Hedylidae (represented by the single genus Macrosoma) has been suggested to be more closely related to Papilionoidea than Hesperioidea (Scoble 1986, 1992), though this placement has been questioned (Weintraub & Miller 1987). Within Papilionoidea, Pieridae are either the sister group to the Papilionidae (Ehrlich 1958; Scott 1985) or to the (Nymphalidae+(Lycaenidae+Riodinidae)) (Kristensen 1976; de Jong et al. 1996; Weller et al. 1996; Ackery et al. 1999). Lycaenidae+Riodinidae are usually, but not always, considered to be the sister group of Nymphalidae (Scott 1985; Robbins 1988; de Jong et al. 1996).

DNA sequences have rarely been used to explicitly assess family level relationships of butterflies and skippers for exemplars of all families (Martin & Pashley 1992; Weller et al. 1996). These studies were based on short sequences of the nuclear 28S ribosomal DNA and the partial sequences of the mitochondrial gene ND1. The resulting trees were rooted with skippers, thus the monophyly of Papilionoidea was not tested. Molecular data have been utilized in a number of studies within lower taxa of butterflies, such as genera, tribes and subfamilies (Sperling 2003); however, there has been little coordination of effort among the studies cited in Sperling (2003). Thus, the many sequences available on public databases such as GenBank cannot be simply collated and analysed together because they do not represent homologous gene regions (Caterino et al. 2000). To alleviate this difficulty, we have agreed to sequence the same three genes for all studied taxa in our respective laboratories. The benefits of such cooperation are self-evident. The three genes we have chosen have been employed with great success in a variety of prior butterfly studies (Brower & Egan 1997; Brower 2000; Campbell et al. 2000; Caterino et al. 2001; Monteiro & Pierce 2001; Wahlberg et al. 2003; Megens et al. 2004), partly due to the availability of polymerase chain reaction primers that work well with all butterflies studied to date.

Recently, the utility of traditional morphological characters for inferring phylogenetic relationships has been heavily criticized and a purely molecular approach has been advocated (Hebert et al. 2003; Scotland et al. 2003). Such a stand is not supported by the fundamental characteristics of the two kinds of data (Miller et al. 1997; Baker & Gatesy 2002) nor by the actual use of morphological data in combination with molecular data (Jenner 2004). Here we investigate the relationships of higher taxa of butterflies and skippers using both traditional morphological data and new molecular data. We also demonstrate the synergistic impact of morphological data on the results of the simultaneous analysis of the combined dataset.

2. Material and Methods

Selection of taxa for sequencing was based on a published morphological study (de Jong et al. 1996) and covered all the major lineages in each butterfly and skipper family (see appendix A). In a few cases the genus coded for morphological characters was not available for molecular work. In such cases, a closely related genus was sequenced instead (see appendix A). Sequences of Cytochrome Oxidase subunit I (COI, 1531 bp), Elongation Factor-1α (EF-1α, 1225 bp) and wingless (403 bp) were generated in the laboratories of Brower, Pierce, Sperling and Wahlberg according to local protocols (see Brower 2000; Caterino et al. 2001; Monteiro & Pierce 2001; Wahlberg et al. 2003). Species identifications, voucher codes and deposition sites and GenBank accession numbers are given in appendix A. The morphological data of de Jong et al. (1996) were revised, with several characters being recoded, to yield a matrix of 99 characters (see electronic appendix for details). These were mainly scored from adult butterflies, comprising 39 wing venation, 19 leg, 14 head, 21 thoracic and two abdominal characters. In addition, four characters were taken from immature stages.

Heuristic parsimony searches were performed with NONA 2.0 (Goloboff 1998) via Winclada (Nixon 2002; 1000 random additions of taxa with 10 trees kept during each replication) for each dataset separately, for the combined molecular dataset and for the entire combined dataset. All characters were given equal weight. Out-groups were not specified a priori in any of the analyses and trees were rooted with a pyralid.

Bayesian phylogenetic analyses (Huelsenbeck et al. 2001) were performed using the programme MrBayes 3.1 (Ronquist & Huelsenbeck 2003). The GTR+G+I model of substitution (chosen using the programme MrModeltest; Nylander 2002) was fitted to each molecular partition and a rate variable model was fitted to the morphological data. Parameters were estimated for each of the genes and the morphology simultaneously (four partitions). Six chains (one cold and five heated) were run for 10 000 000 generations with trees sampled every 1000 generations. To check when stationarity was reached, likelihood values were graphically inspected and the first 1000 sampled trees were discarded as ‘burn-in’. Similar analyses were run without the morphological data.

Nodal support for the cladogram was assessed using Bremer support (BS) (Bremer 1988, 1994) values and bootstrap analysis. BS values and related indices (see below) were calculated with the aid of TreeRot (Sorensen 1999) and PAUP* (Swofford 2001). Bootstrap values were calculated with NONA 2.0 using 1000 pseudo-replicates with 10 random addition replicates per pseudo-replicate.

There are several indices related to BS that describe the interactions of datasets in a combined analysis framework at a node-by-node and character-by-character basis (Baker & DeSalle 1997; Gatesy et al. 1999). Each data partition contributes additively to the total BS in the combined analysis framework, giving values known as partitioned Bremer support (PBS; Baker & DeSalle 1997; Gatesy et al. 1999). A positive PBS value indicates that a given data partition supports a given node while a negative value indicates conflict from a given data partition.

Patterns of homoplasy differ in different datasets (Barrett et al. 1991; Chippindale & Wiens 1994; Olmstead & Sweere 1994) and analysing the different datasets together can bring forth the underlying phylogenetic signal common to all datasets (Baker & DeSalle 1997; Baker et al. 1998). BS values can allow us to compare the effects of combined and separate analyses through an index known as hidden Bremer support (HBS; Gatesy et al. 1999). HBS is the difference between the BS of a given node in the combined analysis and the sum of the BS of each partition for the same node in the most parsimonious tree for that partition. This index allows us to evaluate whether there is increased character support for a clade in combined analysis of multiple datasets compared with the sum of support for that clade in the separate analyses of the different partitions. Positive values indicate increased character support in the combined analysis whereas negative values indicate increased character conflict in the combined analysis.

3. Results

Our analysis is based on sequence data from two nuclear gene regions (1225 bp of EF-1α and 403 bp of Wingless) and one mitochondrial gene region (1531 bp of COI) for a total of 3159 base pairs and 99 morphological characters revised from a published study (de Jong et al. 1996) comprising a matrix of 57 taxa and 3258 characters (of which 1290 are parsimony informative; table 1). Analyses of the data partitions on their own results in many unconventional relationships (see electronic appendix). Our results for the morphology partition differ from those of de Jong et al. (1996), mainly due to our sparse out-group sampling (de Jong et al. sampled 15 out-group taxa).

Table 1.

Summary of character variation in the data partitions used in this study.

data partition characters parsimony informative shortest tree(s) CI RI
morphology 99 88 506 0.32 0.68
COI 1531 566 4972 0.19 0.25
EF-1α 1225 403 4179 0.17 0.32
wingless 403 233 2508 0.18 0.42
total 3258 1290 12459 0.21 0.34
Open in a new tab

Analyses based on the combined molecular data also fail to recover some of the traditionally recognized higher taxa as monophyletic groups (figure 1). The parsimony analysis finds many unconventional relationships (figure 1a). The low support for most nodes is probably due to large amounts of homoplasy and sparse sampling (there are about 20 000 recognized species of butterflies; Robbins 1982). Attempting to take into account these problems by using a parameter rich model (each gene region having unique parameters; see electronic appendix) and Bayesian methodology to estimate the phylogenetic relationships results in a more conventional phylogeny, although there are still several highly unconventional clades (figure 1b). For instance, the Bayesian analysis places Papilionidae as sister to the rest of the butterflies, skippers and hedylids with a high posterior probability (100%) and also places the nymphalid satyrine clade (sensu Wahlberg et al. 2003) as sister to the Riodinidae+Lycaenidae clade with a posterior probability of 100%. Such relationships have never been suggested previously in the literature.

Figure 1.

Figure 1

Open in a new tab

Phylogenetic analyses of the three molecular datasets combined. (a) Strict consensus of three equally maximum parsimonious trees (length 11 908, consistency index 0.18, retention index 0.32), numbers above branches are Bremer support, those to the right of the node are Bootstrap values; (b) tree resulting from Bayesian analysis (average likelihood=−49 408.8), numbers above or below branches are posterior probabilities for the node to the right of each number. Colour codes represent families as follows: pink, Hedylidae; red, Hesperiidae; green, Papilionidae; yellow, Pieridae; purple, Riodinidae; blue, Lycaenidae; orange, Nymphalidae.

In contrast, the analyses of the combined molecular and morphology dataset provide strong support for the monophyly of all traditionally recognized higher taxa, except Nymphalidae which has moderate BS (6), high posterior probability (100%) but no bootstrap support (less than 50%; figure 2). Both analyses place Riodinidae as sister to Lycaenidae with a monophyletic Nymphalidae sister to the former two. Pieridae is placed as sister to (Nymphalidae+(Lycaenidae+Riodinidae)). Hedylidae is placed as sister to Hesperioidea+Papilionoidea. Support values for most of the nodes describing higher taxa are strong (BS ≥ 9, bootstrap ≥ 80%, posterior probability ≥ 95%; figure 2, table 2). The exceptions are the nodes describing Hesperioidea+Papilionoidea, Pieridae as sister to (Nymphalidae+(Lycaenidae+Riodinidae)) and monophyly of Nymphalidae which have BS values of 6 or less, although the same nodes have posterior probabilities higher than 95%.

Figure 2.

Figure 2

Open in a new tab

Phylogenetic analyses of the combined molecular and morphological datasets. (a) Single most parsimonious tree (length 12 459, consistency index 0.21, retention index 0.34). Numbers above branches are Bremer support/Bootstrap values for the node to the right of the numbers and italicized numbers below branches are node numbers referred to in table 1; (b) tree resulting from Bayesian analysis using mixed models (average likelihood=−51 999.2). Numbers below branches are posterior probabilities for the node to the right of each number. Taxa in parentheses are related substitutes from which sequence data were obtained. Colour codes as in figure 1.

Table 2.

Bremer support (BS), hidden Bremer support (HBS), partitioned Bremer support (PBS) and partitioned hidden Bremer support (PHBS) for nodes in the total evidence tree.

node clade BS total HBS total PBS morph PBS COI PBS EF1α PBS Wgl PHBS morph PHBS COI PHBS EF1α PHBS Wgl BS morph BS COI BS EF1α BS Wgl
1 9 26 −6.0 −4.5 12.0 7.5 −3.0 3.5 18.0 7.5 −3 −8 −6 0
2 9 −13 −6.0 −4.5 12.0 7.5 −1.0 −2.5 13.0 −22.5 −5 −2 −1 30
3 Hedy+Hesp+Papi 10 22 4.0 3.5 4.5 −2.0 5.0 −0.5 13.5 4.0 −1 4 −9 −6
4 Hesp+Papi 4 31 −0.8 −2.4 2.8 4.4 0.2 4.6 14.8 11.4 −1 −7 −12 −7
5 Hesperiidae 13 9 12.0 −0.8 2.5 −0.8 0.0 3.3 4.5 1.3 12 −4 −2 −2
6 7 5 3.0 −2.0 1.0 5.0 1.0 −2.0 1.0 5.0 2 0 0 0
7 9 12 −1.0 2.0 3.0 5.0 3.0 −2.0 7.0 4.0 −4 4 −4 1
8 3 12 0.0 −1.0 0.0 4.0 2.0 4.0 2.0 4.0 −2 −5 −2 0
9 16 7 0.0 0.0 4.0 12.0 2.0 −2.0 4.0 3.0 −2 2 0 9
10 Papilionoidea 14 40 9.0 4.5 −0.5 1.0 4.0 10.5 13.5 12.0 5 −6 −14 −11
11 Papilionidae 21 16 6.0 0.5 8.5 6.0 1.0 5.5 10.5 −1.0 5 −5 −2 7
12 12 26 −1.0 6.0 8.0 −1.0 3.0 16.0 7.0 0.0 −4 −10 1 −1
13 4 13 2.0 6.0 −2.0 −2.0 1.0 11.0 −1.0 2.0 1 −5 −1 −4
14 10 14 1.3 6.0 1.7 1.0 1.3 9.0 1.7 2.0 0 −3 0 −1
15 Pie+Nym+Lyc+Rio 1 35 2.0 −3.5 3.5 −1.0 4.0 −3.5 29.5 5.0 −2 0 −26 −6
16 Pieridae 19 32 2.5 −0.5 13.5 3.5 4.5 0.5 12.5 14.5 −2 −1 1 −11
17 6 7 1.0 −1.0 4.0 2.0 −1.0 5.0 −2.0 5.0 2 −6 6 −3
18 7 22 −1.3 −2.7 7.7 3.3 −0.3 0.3 8.7 13.3 −1 −3 −1 −10
19 32 15 2.3 5.6 22.4 1.8 1.3 7.6 2.4 3.8 1 −2 20 −2
20 17 26 3.0 7.0 11.0 −4.0 4.0 10.0 10.0 2.0 −1 −3 1 −6
21 Nym+Lyc+Rio 9 43 5.5 6.0 3.5 −6.0 3.5 12.0 22.5 5.0 2 −6 −19 −11
22 Lyc+Rio 12 41 8.0 17.0 −6.0 −7.0 9.0 17.0 14.0 1.0 −1 0 −20 −8
23 Riodinidae 26 1 4.0 6.0 10.0 6.0 3.0 2.0 1.0 −5.0 1 4 9 11
24 26 1 3.5 17.5 −7.0 12.0 3.5 12.5 −12.0 −3.0 0 5 5 15
25 2 8 −1.0 2.0 6.0 −5.0 0.0 3.0 6.0 −1.0 −1 −1 0 −4
26 16 3 5.0 −12.0 20.0 3.0 4.0 −8.0 8.0 −1.0 1 −4 12 4
27 Lycaenidae 24 23 5.0 4.0 6.0 9.0 8.0 4.0 20.0 −9.0 −3 0 −14 18
28 9 45 3.5 10.0 1.0 −5.5 8.5 14.0 21.0 1.5 −5 −4 −20 −7
29 7 1 2.6 −1.4 3.4 2.4 6.6 −2.4 −0.6 −2.6 −4 1 4 5
30 10 6 3.0 5.0 1.0 1.0 7.0 1.0 1.0 −3.0 −4 4 0 4
31 29 9 3.0 5.0 8.0 13.0 5.0 2.0 2.0 0.0 −2 3 6 13
32 8 53 1.0 5.0 0.0 2.0 4.0 9.0 29.0 11.0 −3 −4 −29 −9
33 5 47 1.5 0.3 −3.0 6.3 2.5 5.3 26.0 13.3 −1 −5 −29 −7
34 7 38 0.0 0.0 6.0 1.0 0.0 3.0 27.0 8.0 0 −3 −21 −7
35 3 15 2.3 −1.3 4.3 −2.3 5.3 0.7 8.3 0.7 −3 −2 −4 −3
36 19 4 1.3 0.7 18.3 −1.3 3.3 1.7 4.3 −5.3 −2 −1 14 4
37 Nymphalidae 6 22 3.0 4.0 3.8 −4.8 −1.0 6.0 14.8 2.2 4 −2 −11 −7
38 2 36 1.0 −6.0 10.0 −3.0 9.0 3.0 19.0 5.0 −8 −9 −9 −8
39 3 29 0.0 3.5 −1.0 0.5 4.0 9.5 6.0 9.5 −4 −6 −7 −9
40 9 16 1.5 −3.0 10.0 0.5 3.5 5.0 8.0 −0.5 −2 −8 2 1
41 1 9 1.0 −7.0 7.5 −0.5 2.0 3.0 5.5 −1.5 −1 −10 2 1
42 3 14 0.0 3.5 −1.0 0.5 1.0 6.5 7.0 −0.5 −1 −3 −8 1
43 6 39 0.0 14.5 −10.0 1.5 5.0 21.5 5.0 7.5 −5 −7 −15 −6
44 11 30 1.0 6.5 0.0 3.5 1.0 17.5 5.0 6.5 0 −11 −5 −3
45 2 29 0.0 7.5 −2.0 −3.5 2.0 14.5 9.0 3.5 −2 −7 −11 −7
46 2 20 0.0 7.5 −2.0 −3.5 3.0 7.5 9.0 0.5 −3 0 −11 −4
47 2 19 0.0 7.5 −2.0 −3.5 1.0 5.5 7.0 5.5 −1 2 −9 −9
48 10 10 2.0 2.5 7.0 −1.5 2.0 4.5 4.0 −0.5 0 −2 3 −1
49 13 23 0.0 14.8 −1.6 −0.2 0.0 14.8 5.4 2.8 0 0 −7 −3
50 11 15 −2.0 18.0 −6.0 1.0 −2.0 17.0 −3.0 3.0 0 1 −3 −2
51 9 13 3.0 −1.5 12.5 −5.0 2.0 −5.5 13.5 3.0 1 4 −1 −8
52 1 9 −2.0 1.0 −1.0 3.0 0.0 −1.0 8.0 2.0 −2 2 −9 1
53 7 14 −3.0 3.0 4.0 3.0 −2.0 4.0 3.0 9.0 −1 −1 1 −6
54 10 24 −2.0 9.0 2.0 1.0 −2.0 5.0 14.0 7.0 0 4 −12 −6
Total 543 1066 83.7 167.3 221.3 70.7 133.7 293.3 488.3 150.7 −50 −126 −267 −80
Open in a new tab

PBS values show that in the combined analysis all the data partitions contribute positively to the support of the relationships in figure 2 (table 2). Since the magnitude of BS is contingent on the number of parsimony informative characters, dividing the total PBS for a partition by the number of parsimony informative characters of that partition gives us an index of the relative informativeness of each partition. The index clearly shows that the morphological (0.95) and EF-1α (0.56) partitions contribute most of the support for the relationships in figure 2a. These two partitions support almost all the major nodes and there is weak conflict derived from them at three nodes (major nodes shown in bold in table 2). The COI (0.29) and wingless (0.30) partitions show a higher level of conflict at many of the nodes yet both show strong support for the major nodes of interest.

Despite the strong support for most clades, we find that each data partition shows a great deal of homoplasy when mapped on to the tree found in combined parsimony analysis (table 1). However, combining the different data partitions reveals strong hidden support for almost all of the major nodes (table 2), indicating that combined analysis enhances the inherent phylogenetic signal of the data compared with analysing the partitions separately. Partitioning the HBS values indicates that all four data partitions show substantial hidden signals that are revealed when analysed in combination with the other datasets.

4. Discussion

In the discussion of their analysis of butterfly relationships based on morphology (the dataset employed here), de Jong et al. (1996) expressed a degree of dissatisfaction with the power of morphological features to resolve many clades. They questioned whether additional morphological data would improve the results and called for the pursuit of alternative sources of evidence. However, they qualified these controversial statements by arguing that the most productive avenue would be to combine those alternative data with morphological characters, as we have done here. Many current phylogenetic studies do not take morphological data into account for a variety of reasons (see Baker & Gatesy 2002; Scotland et al. 2003; Jenner 2004). As we have shown, ignoring the morphological data in this case resulted in several spurious clades, regardless of the method of analysis used (figure 1), including a paraphyletic Nymphalidae with regard to Lycaenidae+Riodinidae, a sister relationship between Hedylidae and Papilionidae and a paraphyletic Papilionoidea with regard to Hesperioidea and Hedyloidea.

So we have an interesting problem: morphological data on their own provide weak support for, or fail to resolve, many nodes and molecular data on their own yield trees that contain implausible clades, some of which are strongly supported. A common claim from molecular systematists is that a large number of independent characters are needed to be able to estimate phylogenetic relationships robustly and reliably (Rokas et al. 2003). Yet for many groups of organisms, it is still not feasible to generate large amounts of DNA sequence data from a variety of gene regions. In Lepidoptera, for example, it has been a challenge to discover primers that are universal enough to amplify gene regions that are of broad phylogenetic utility in resolving all hierarchic taxonomic levels, from species to super-families (Friedlander et al. 1994). We feel that the propensity of limited molecular datasets to imply wrong, yet well supported, topologies is a problem that is not fully appreciated. Researchers now generally try to avoid the mistakes of early molecular systematics when dramatic rearrangements of phylogeny were often proposed on the basis of small and sparsely sampled datasets. However, there is still a tendency to reach very general conclusions from very specific and limited data. In particular, we feel that it should be more appreciated that adding more data is superior to adding ever more thorough analyses of existing data when the phylogeny is uncertain.

Our solution to this problem has been to combine our sequences with available morphological data. Morphological features are often uniquely derived complex structures that unequivocally unite major taxa, such as the osmeterial glands of papilionid larvae and the tricarinate antennae of nymphalid adults, and we see them as an intrinsic component of a robust treatment of this important group. Even though all characters are given equal weight and the morphological characters are vastly outnumbered by the molecular data, the low intrinsic homoplasy of the morphological features allows them to establish a structural framework in the combined analysis to which the morphological data contributes synergistically. Although the morphological data on their own were unable to give unequivocal answers in our analyses, the combined analyses (regardless of method used) have given us a robust phylogenetic hypothesis for the butterflies and skippers and the whole is greater than the sum of its parts, as is shown by the positive HBS values.

Our analyses of the combined dataset provide evidence to settle a long-standing question on the position and rank of Riodinidae. Our data give strong support for a Lycaenidae+Riodinidae sister group relationship, rather than embedding Riodinidae within Lycaenidae (de Jong et al. 1996) or placing them as sister to Nymphalidae (Robbins 1989). This result is concordant with an earlier molecular study that included a larger sample of taxa for the three groups but only used evidence from wingless (Campbell et al. 2000). The high BS and full congruence among the different datasets (table 2) indicates that only a great deal of contradictory new evidence could overturn this sister relationship. We believe that the current analysis provides a long-awaited empirical rationale for ranking Lycaenidae and Riodinidae as separate families.

The monophyly of Hesperiidae, Papilionidae and Pieridae is also strongly supported (figure 2) and there is little or no conflict between the data partitions at these nodes (table 2). These three families are generally considered to form monophyletic groups, though occasionally some small groups have been split off into their own families (e.g. Megathymidae from Hesperiidae and Baroniidae from Papilionidae). The case of Nymphalidae, however, is more complicated. Morphology, COI and EF-1α sequence data support the monophyly of the family while wingless conflicts moderately. The hidden BS of the partitions are positive, with the molecular partitions exhibiting larger values than the morphological partition. This suggests that there is a large amount of homoplasy (noise) in the molecular partitions for Nymphalidae that is overcome by the combined analysis of all data. The basal branch leading to Nymphalidae appears to be much shorter than the branches leading to the other families (figure 2b). One interpretation of such a pattern might be that the major lineages in Nymphalidae diverged very rapidly from one another. This may explain the difficulty in determining homologous states of morphological characters among nymphalids (Freitas & Brown 2004) and the difficulty in finding uniquely derived features to define the family (de Jong et al. 1996).

Relationships of taxa within families generally agree with recent family level studies (Brower 2000; Campbell et al. 2000; Caterino et al. 2001; Hall & Harvey 2002; Wahlberg et al. 2003; Freitas & Brown 2004), though some incongruent relationships can be noted. These incongruities are probably the result of insufficient taxon sampling in this study. The studies cited above included many more taxa per group and were generally able to resolve the relationships of major lineages within the families with higher support.

This is the first study to provide strongly supported answers to many of the questions about butterfly relationships so eloquently raised by Vane-Wright and co-workers (de Jong et al. 1996; Vane-Wright 2003). Our data show that the six currently recognized skipper and butterfly families (Hesperiidae, Papilionidae, Pieridae, Lycaenidae, Riodinidae and Nymphalidae) are monophyletic entities, that Lycaenidae and Riodinidae are sister taxa, that Nymphalidae are sister to Lycaenidae+Riodinidae, that Pieridae are sister to (Nymphalidae+(Lycaenidae+Riodinidae)) and lend support to the hypothesis that Hedylidae are more closely related to the butterflies and skippers than the geometrids within which they were once classified. Many of these taxa have been intuitively grouped based on striking morphological novelties since the time of Linnaeus but this study is the first to provide robust empirical support for relationships based on rigorous analysis of characters, both morphological and molecular. These results once again emphasize the importance of combining evidence from different sources as a means to dilute the bias of homoplasy within any individual data partition (Farris 1983; Miller et al. 1997).

Acknowledgments

This work has been supported in part by the Swedish Research Council (to S.N. and N.W.), USA NSF grants (to N.P. and A.B.), a Canadian NSERC grant (to F.S.), a Fulbright Foundation grant (to R.V.) and an Australian Research Council Fellowship and Fulbright Postdoctoral Fellow Award (to M.F.B.). We are very grateful to K. Brown Jr., A. Chichvarkhin, S. S. Collins, P. DeVries, A. Freitas, J. Hall, D. Janzen, D. Lohman, C. Stefanescu and K. Willmott for providing specimens used in this study. We thank Elisabet Weingartner and Carlos Peña for help in the laboratory.

Appendix

See table 3

Table 3.

Information on specimens used for molecular studies.

family subfamily species specimen ID GenBank ID COI GenBank ID EF-1α GenBank ID wingless
Pyralidae Ostrinia nubilalis FS.b-5 AF170853 AF173392 —
Noctuidae Feltia jaculifera FS.b-150 U60990 AF173390 AY569039
Geometriidae Archiearis parthenias NW107-1 DQ018928 DQ018899 DQ018869
Uraniidae Urania leilia NW96-7 DQ018927 DQ018898 DQ018868
Hedylidae Macrosoma sp. FS.b-983 AF170854 AF173393 AY569042
Hesperiidae Coeliadinae Hasora khoda AW97 DQ018930 DQ018901 DQ018871
Hesperiidae Hesperiinae Ochlodes sylvanoides AW50 DQ018931 DQ018902 DQ018872
Hesperiidae Trapezitinae Trapezites symmomus AW89 DQ018932 DQ018903 DQ018873
Hesperiidae Pyrginae Urbanus dorantes AW280 DQ018929 DQ018900 DQ018870
Hesperiidae Pyrginae Pyrgus communis FS.b-901 AF170857 AF173396 AY569043
Hesperiidae Pyrrhopyginae Mysoria ambigua AW138 DQ018933 DQ018904 DQ018874
Papilionidae Baroniinae Baronia brevicornis FS.a-167 AF170866 AF173406 AY569044
Papilionidae Parnassiinae Parnassius phoebus FS.a-8 AF170872 AF173412 AY569045
Papilionidae Papilioninae Graphium agamemnon FS.b-900 AF170874 AF173414 AY569046
Papilionidae Papilioninae Papilion machaon FS.a-27 AF044006 AF044819 AY569124
Papilionidae Papilioninae Troides helena FS.b-974 AF170878 AF173418 AY569047
Pieridae Coliadinae Eurema hecabe MFB-00-P036 DQ018935 AY870587 DQ018876
Pieridae Coliadinae Colias eurytheme FS.b-543 AF044024 AF173400 AY569040
Pieridae Pierinae Delias belladonna DL-01-N104 DQ082773 AY870510 DQ082808
Pieridae Pierinae Pieris napi FS.b-943 AF170861 AF173401 AY569041
Pieridae Dismorphiinae Dismorphia zathoe MFB-00-P231 AY954566 AY870578 AY954596
Pieridae Pseudopontiinae Pseudopontia paradoxa SC-01-T380 AY954564 AY870580 AY954594
Lycaenidae Curetinae Curetis bulis MWT-93-A028 DQ018942 DQ018909 AF233549
Lycaenidae Lipteninae Baliochila minima SP-93-P006 DQ018938 DQ018905 DQ018879
Lycaenidae Miletinae Liphyra brassolis KD-94-T063 DQ018940 DQ018907 AF233551
Lycaenidae Miletinae Miletus ancon KF-94-P022 DQ018941 DQ018908 AF233550
Lycaenidae Poritiinae Poritia erycinoides MWT-93-B007 DQ018939 DQ018906 DQ018880
Lycaenidae Theclinae Callipsyche behrii AS-92-Z034 DQ018943 DQ018910 DQ018881
Lycaenidae Theclinae Thecla coelicolor DY-93-G038 DQ018945 DQ018912 DQ018883
Lycaenidae Theclinae Lucia limbaria KD-94-Q002 DQ018944 DQ018911 DQ018882
Lycaenidae Lycaeninae Lycaena helloides NP-99-W131 DQ018948 DQ018915 DQ018886
Lycaenidae Polyommatinae Echinargus isola AS-92-Z185 DQ018947 DQ018914 DQ018885
Lycaenidae Polyommatinae Anthene emolus MWT-93-A051 DQ018946 DQ018913 —
Lycaenidae Polyommatinae Neurellipes staudingeri RD-98-U112 — — DQ018884
Riodinidae Riodininae Riodina lysippus PDV-94-A050 DQ018952 DQ018919 AF233540
Riodinidae Riodininae Emesis nr mandana PDV-94-T022 DQ018950 DQ018917 DQ018888
Riodinidae Euselasiinae Euselasia nr orfita PDV-94-A022 DQ018951 DQ018918 DQ018889
Riodinidae Styginae Styx infernalis GL-02-N259 DQ018949 DQ018916 DQ018887
Riodinidae Hamearinae Hamearis lucina NW84-13 DQ018953 DQ018920 DQ018890
Nymphalidae Libytheinae Libythea celtis NW71-1 AY090198 AY090164 AY090131
Nymphalidae Danainae Danaus plexippus NW108-22 DQ018954 DQ018921 DQ018891
Nymphalidae Calinaginae Calinaga buddha NW64-3 AY090208 AY090174 AY090141
Nymphalidae Charaxinae Charaxes castor NW78-3 AY090219 AY090185 AY090152
Nymphalidae Satyrinae Oeneis jutta EW4-1 DQ018958 DQ018925 DQ018896
Nymphalidae Satyrinae Haetera piera CP01-84 DQ018959 DQ018926 DQ018897
Nymphalidae Morphinae Amathusia phidippus NW114-17 DQ018956 DQ018923 DQ018894
Nymphalidae Morphinae Morpho peleides NW66-5 AY090210 AY090176 AY090143
Nymphalidae Morphinae Opsiphanes quiteria NW109-10 DQ018957 DQ018924 DQ018895
Nymphalidae Limenitidinae Limenitis reducta NW67-2 AY090217 AY090183 AY090150
Nymphalidae Heliconiinae Heliconius hecale NW70-6 AY090202 AY090168 AY090135
Nymphalidae Heliconiinae Actinote stratonice NW90-14 AY218233 AY218252 DQ018892
Nymphalidae Heliconiinae Argynnis paphia NW76-12 AY090200 AY090166 AY090133
Nymphalidae Biblidinae Biblis hyperia NW106-3 DQ018955 DQ018922 DQ018893
Nymphalidae Cyrestinae Marpesia orsilochus AB-RB250 AY788604 AY788706 AF246532
Nymphalidae Apaturinae Apatura iris NW69-6 AY090199 AY090165 AY090132
Nymphalidae Nymphalinae Melitaea cinxia NW73-14 AY788656 AY788776 AY788536
Nymphalidae Nymphalinae Nymphalis antiopa NW70-2 AY218246 AY218266 AY218284
Nymphalidae Nymphalinae Colobura dirce NW68-11 AY090228 AY090196 AY090162
Open in a new tab

Supplementary Material

rspb20053124s02.pdf (204.6KB, pdf)

References

  1. Ackery P.R, de Jong R, Vane-Wright R.I. The butterflies: Hedyloidea, Hesperoidea and Papilionoidea. In: Kristensen N.P, editor. Lepidoptera, moths and butterflies. 1. Evolution, systematics and biogeography. Handbook of zoology 4 (35), Lepidoptera. de Gruyter; Berlin: 1999. pp. 263–300. [Google Scholar]
  2. Baker R.H, DeSalle R. Multiple sources of character information and the phylogeny of Hawaiian drosophilids. Syst. Biol. 1997;46:654–673. doi: 10.1093/sysbio/46.4.654. [DOI] [PubMed] [Google Scholar]
  3. Baker R, Gatesy J. Is morphology still relevant? In: DeSalle R, Wheeler W, Giribet G, editors. Molecular systematics and evolution: theory and practice. Birkhauser Verlag; Basel: 2002. pp. 163–174. [Google Scholar]
  4. Baker R.H, Yu X, DeSalle R. Assessing the relative contribution of molecular and morphological characters in simultaneous analysis trees. Mol. Phyl. Evol. 1998;9:427–436. doi: 10.1006/mpev.1998.0519. [DOI] [PubMed] [Google Scholar]
  5. Barrett M, Donoghue M, Sober E. Against consensus. Syst. Zool. 1991;40:486–493. [Google Scholar]
  6. Boggs C.L, Watt W.B, Ehrlich P.R, editors. Butterflies: evolution and ecology taking flight. Rocky mountain biological lab symposium series. University of Chicago Press; 2003. [Google Scholar]
  7. Bremer K. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution. 1988;42:795–803. doi: 10.1111/j.1558-5646.1988.tb02497.x. [DOI] [PubMed] [Google Scholar]
  8. Bremer K. Branch support and tree stability. Cladistics. 1994;10:295–304. [Google Scholar]
  9. Brower A.V.Z. Phylogenetic relationships among the Nymphalidae (Lepidoptera), inferred from partial sequences of the wingless gene. Proc. R. Soc. B. 2000;267:1201–1211. doi: 10.1098/rspb.2000.1129. 10.1098/rspb.2000.1129 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Brower A.V.Z, Egan M.G. Cladistic analysis of Heliconius butterflies and relatives (Nymphalidae: Heliconiiti): a revised phylogenetic position for Eueides based on sequences from mtDNA and a nuclear gene. Proc. R. Soc. B. 1997;264:969–977. 10.1098/rspb.1997.0134 [Google Scholar]
  11. Campbell D.L, Brower A.V.Z, Pierce N.E. Molecular evolution of the wingless gene and its implications for the phylogenetic placement of the butterfly family Riodinidae (Lepidoptera: Papilionoidea) Mol. Biol. Evol. 2000;17:684–696. doi: 10.1093/oxfordjournals.molbev.a026347. [DOI] [PubMed] [Google Scholar]
  12. Caterino M.S, Cho S, Sperling F.A.H. The current state of insect molecular systematics: a thriving tower of babel. Annu. Rev. Entomol. 2000;45:1–54. doi: 10.1146/annurev.ento.45.1.1. [DOI] [PubMed] [Google Scholar]
  13. Caterino M.S, Reed R.D, Kuo M.M, Sperling F.A.H. A partitioned likelihood analysis of swallowtail butterfly phylogeny (Lepidoptera: Papilionidae) Syst. Biol. 2001;50:106–127. [PubMed] [Google Scholar]
  14. Chippindale P, Wiens J. Weighting, partitioning, and combining characters in phylogenetic analysis. Syst. Biol. 1994;43:278–287. [Google Scholar]
  15. de Jong R, Vane-Wright R.I, Ackery P.R. The higher classification of butterflies (Lepidoptera): problems and prospects. Entomol. Scand. 1996;27:65–101. [Google Scholar]
  16. Ehrlich P.R. The comparative morphology, phylogeny and higher classification of the butterflies (Lepidoptera: Papilionoidea) The University of Kansas Science Bulletin. 1958;39:305–370. [Google Scholar]
  17. Farris J.S. The logical basis of phylogenetic analysis. In: Platnick N.I, Funk V.A, editors. Advances in cladistics. vol. 2. Columbia University Press; New York: 1983. pp. 7–36. [Google Scholar]
  18. Freitas A.V.L, Brown K.S.J. Phylogeny of the Nymphalidae (Lepidoptera: Papilionoidea) Syst. Biol. 2004;53:363–383. doi: 10.1080/10635150490445670. [DOI] [PubMed] [Google Scholar]
  19. Friedlander T.P, Regier J.C, Mitter C. Phylogenetic information content of five nuclear gene sequences in animals: initial assessment of character sets from concordance and divergence studies. Syst. Biol. 1994;43:511–525. [Google Scholar]
  20. Gatesy J, O'Grady P, Baker R.H. Corroboration among data sets in simultaneous analysis: hidden support for phylogenetic relationships among higher level artiodactyl taxa. Cladistics. 1999;15:271–313. doi: 10.1111/j.1096-0031.1999.tb00268.x. [DOI] [PubMed] [Google Scholar]
  21. Goloboff P.A. published by author; New York: 1998. NONA. [Google Scholar]
  22. Hall J.P.W, Harvey D.J. A survey of androconial organs in the Riodinidae (Lepidoptera) Zool. J. Linn. Soc. 2002;136:171–197. [Google Scholar]
  23. Harvey D.J. Higher classification of the Nymphalidae, appendix B. In: Nijhout H.F, editor. Higher classification of the Nymphalidae. Smithsonian Institution Press; Washington DC: 1991. pp. 255–273. [Google Scholar]
  24. Hebert P.D.N, Cywinska A, Ball S.L, deWaard J.R. Biological identifications through DNA barcodes. Proc. R. Soc. B. 2003;270:313–322. doi: 10.1098/rspb.2002.2218. 10.1098/rspb.2002.2218 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Huelsenbeck J.P, Ronquist F, Nielsen R, Bollback J.P. Bayesian inference of phylogeny and its impact on evolutionary biology. Science. 2001;294:2310–2314. doi: 10.1126/science.1065889. [DOI] [PubMed] [Google Scholar]
  26. Jenner R.A. Accepting partnership by submission? Morphological phylogenetics in a molecular millennium. Syst. Biol. 2004;53:333–342. doi: 10.1080/10635150490423962. [DOI] [PubMed] [Google Scholar]
  27. Kristensen N.P. Remarks on the family-level phylogeny of butterflies (Insecta Lepidoptera, Rhopalocera) Zeit. Zool. Syst. Evol. Forsch. 1976;14:25–33. [Google Scholar]
  28. Martin J.A, Pashley D.P. Molecular systematic analysis of butterfly family and some subfamily relationships (Lepidoptera: Papilionoidea) Ann. Entomol. Soc. Am. 1992;85:127–139. [Google Scholar]
  29. Megens H.-J, van Nes W.J, van Moorsel C.H.M, Pierce N.E, de Jong R. Molecular phylogeny of the oriental butterfly genus Arhopala (Lycaenidae Theclinae) inferred from mitochondrial and nuclear genes. Syst. Entomol. 2004;29:115–131. [Google Scholar]
  30. Miller J.S, Brower A.V.Z, DeSalle R. Phylogeny of the neotropical moth tribe Josiini (Notodontidae: Dioptinae): comparing and combining evidence from DNA sequences and morphology. Biol. J. Linn. Soc. 1997;60:297–316. [Google Scholar]
  31. Monteiro A, Pierce N.E. Phylogeny of Bicyclus (Lepidoptera: Nymphalidae) inferred from COI, COII, and EF-1alpha gene sequences. Mol. Phyl. Evol. 2001;18:264–281. doi: 10.1006/mpev.2000.0872. [DOI] [PubMed] [Google Scholar]
  32. Müller W. Südamerikanische Nymphalidaeraupen: Versuch eines natürlichen Systems der Nymphaliden. Zool. Jahrb. 1886;1:417–678. [Google Scholar]
  33. Nixon K.C. published by author; Ithaca, NY: 2002. Winclada. [Google Scholar]
  34. Nylander, J.A.A. 2002 MrModeltest v2.1. Department of Systematic Zoology, Uppsala University: available from author.
  35. Olmstead R, Sweere J. Combining data in phylogenetic systematics: an empirical approach using three molecular data sets in the Solanaceae. Syst. Biol. 1994;43:467–481. [Google Scholar]
  36. Penz C, Peggie D. Phylogenetic relationships among Heliconiinae genera based on morphology (Lepidoptera: Nymphalidae) Syst. Entomol. 2003;28:451–479. [Google Scholar]
  37. Robbins R.K. How many butterfly species? News of the lepidopterists' society. 1982;1982:40–41. [Google Scholar]
  38. Robbins R.K. Comparative morphology of the butterfly forleg coxa and trochanter (Lepidoptera) and its systematic implications. Proc. Entomol. Soc. Wash. 1988;90:133–154. [Google Scholar]
  39. Robbins R.K. Systematic implications of butterfly leg structures that clean antennae. Psyche. 1989;96:209–222. [Google Scholar]
  40. Rokas A, Williams B.L, King N, Carroll S.B. Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature. 2003;425:798–804. doi: 10.1038/nature02053. [DOI] [PubMed] [Google Scholar]
  41. Ronquist F, Huelsenbeck J.P. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics. 2003;19:1572–1574. doi: 10.1093/bioinformatics/btg180. [DOI] [PubMed] [Google Scholar]
  42. Scoble M.J. The structure and affinities of the Hedyloidea: a new concept of butterflies. Bull. Br. Museum Nat. Hist. (Entomol.) 1986;53:251–286. [Google Scholar]
  43. Scoble M.J. Oxford University Press; 1992. The Lepidoptera: form, function and diversity. [Google Scholar]
  44. Scotland R.W, Olmstead R, Bennett J.R. Phylogeny reconstruction: the role of morphology. Syst. Biol. 2003;52:539–548. doi: 10.1080/10635150390223613. [DOI] [PubMed] [Google Scholar]
  45. Scott J.A. The phylogeny of butterflies (Papilionoidea and Hesperoidea) J. Res. Lepid. 1985;23:241–281. [Google Scholar]
  46. Sorensen M.D. Boston University; 1999. TreeRot. [Google Scholar]
  47. Sperling F.A.H. Butterfly species and molecular phylogenies. In: Boggs C.L, Watt W.B, Ehrlich P.R, editors. Butterflies: evolution and ecology taking flight. University of Chicago Press; 2003. pp. 431–458. [Google Scholar]
  48. Swofford D.L. Sinauer Associates; Sunderland, MA: 2001. PAUP*: phylogenetic analysis using parsimony (* and other methods) (version 4.0b10) [Google Scholar]
  49. Vane-Wright R.I. Evidence and identity in butterfly systematics. In: Boggs C.L, Watt W.B, Ehrlich P.R, editors. Butterflies: evolution and ecology taking flight. University of Chicago Press; 2003. pp. 477–514. [Google Scholar]
  50. Wahlberg N, Weingartner E, Nylin S. Towards a better understanding of the higher systematics of Nymphalidae (Lepidoptera: Papilionoidea) Mol. Phyl. Evol. 2003;28:473–484. doi: 10.1016/s1055-7903(03)00052-6. [DOI] [PubMed] [Google Scholar]
  51. Weintraub J.D, Miller J.S. The structure and affinities of the Hedyloidea: a new concept of butterflies. Cladistics. 1987;3:299–304. [Google Scholar]
  52. Weller S.J, Pashley D.P, Martin J.A. Reassessment of butterfly family relationships using independent genes and morphology. Ann. Entomol. Soc. Am. 1996;89:184–192. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

rspb20053124s02.pdf (204.6KB, pdf)

Articles from Proceedings of the Royal Society B: Biological Sciences are provided here courtesy of The Royal Society

RESOURCES