Phylogenetic studies of the Melitaeini (Lepidoptera: Nymphalidae: Nymphalinae) and a revision of the genus Chlosyne Butler

MISSING IMAGE

Material Information

Title:
Phylogenetic studies of the Melitaeini (Lepidoptera: Nymphalidae: Nymphalinae) and a revision of the genus Chlosyne Butler
Physical Description:
x, 798 leaves : ill. ; 29 cm.
Language:
English
Creator:
Kons, Hugo L., 1974-
Publication Date:

Subjects

Subjects / Keywords:
Nymphalidae -- Phylogeny   ( lcsh )
Cladistic analysis   ( lcsh )
Entomology and Nematology thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Entomology and Nematology -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2000.
Bibliography:
Includes bibliographical references (leaves 790-796).
Additional Physical Form:
Also available online.
Statement of Responsibility:
by Hugo L. Kons, Jr.
General Note:
Printout.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 025854725
oclc - 46727538
System ID:
AA00025758:00001

Full Text










PHYLOGENETIC STUDIES OF THE MELITAEINI (LEPIDOPTERA: NYMPHALIDAE: NYMPHALINAE) AND A REVISION OF THE GENUS
CHLOSYNE BUTLER













By

HUGO L. KONS, JR.














A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA 2000









Dedicated to the collectors of Lepidoptera:



An extensive amount of work crucial to the success of this project was completed before it even began, in the form of the extensive collections of specimens available for study. These specimens were collected, curated, and labeled from all over the world, and represent the cumulative efforts of many amateur and professional Lepidopterists collecting in many states and nations over a period exceeding 100 years. Many of the Melitaeini are strikingly beautiful butterflies long popular among collectors, and their beauty combined with diurnal habits have contributed to their representation in collections. Nevertheless, some taxa remain poorly represented, and more material would have been highly desirable. Many other groups of Lepidoptera, let alone other insects, are far more poorly represented in collections today.

In the face of the expanding human population and ever increasing destruction of natural habitats, and consequent crisis for understanding and preserving Earth's biodiversity, the need for collecting and systematic studies of insects has never been greater. Yet, tragically, amateur and professional Lepidopterists in many lands presently have their efforts stifled by a gauntlet of misguided collecting restrictions based on ignorance, territorialism, and/or financial greed, including: restrictions preventing the acquisition of material for study, restrictions preventing collectors from retaining the specimens they collect, restrictions on transporting dead specimens across political lines, an overburden of unnecessary red tape and/or timely delays associated with obtaining permits, restrictions which discriminate against amateur Lepidopterists and/or people outside particular political boundaries, and/or exorbitant fees required to purchase permits even for noncommercial scientific collecting. It is a sad tragedy at a time when collecting is needed more then ever, it is nonetheless discouraged more than ever, thus severely limiting the prospects of success for any state, national, or international effort to understand, document, and preserve biodiversity.

In recognition of the indispensable role of collectors to the study and preservation of

Lepidoptera biodiversity, I dedicate this work to the collectors of Lepidoptera, past, present, and future, amateur and professional, and institutional and private, in the hope that the gauntlet of misguided regulations so plaguing the efforts of Lepidopterists around the world will undergo repeal or sweeping reform in a future time of far greater enlightenment and cooperation.















ACKNOWLEDGMENTS

I wish to acknowledge a number of people who provided valuable assistance with this project. For serving on my doctoral committee, providing advice and assistance throughout this project, and for review of this manuscript, I thank Thomas Emmel, James Lloyd, Jonathen Reiskind, Frank Slansky, and John Heppner. For providing access to or loan of specimens critical to this study I thank John Heppner (Florida State Collection of Arthropods), Robert Robbins and Donald Harvey (National Museum of Natural History), Lee and Jackie Miller (Allyn Museum), Keith Willmott, and Jason Hall. I am grateful to Gerardo Lamas for sending me scanned images of several Chiosyne types in the British Museum. For providing work space in the Florida State Collection of Arthropods I thank John Heppner. Thomas Walker provided access to an exceptional camera lucida microscope set up for doing the genitalia illustrations. I thank Steve Lasley and Nick Hostettler for computer assistance. I also thank Thomas Emmel and the University of Florida's Department of Entomology and Nemnatology (especially the graduate coordinator, Grover Smart) for providing research and teaching assistantship funding, respectively, for some semesters. Donald Harvey and Robert Robbins provided considerable advice and assistance during a visit to the National Museum of Natural History, and Robert Robbins assisted with securing a short term visitors grant for me to visit that institution. Debbie Hall was most helpful for assistance dealing with the University of Florida's bureaucracy. For helpful discussions and/or debates on species problems, cladistics, and/or other issues in taxonomy I thank James Lloyd, Donald



ill









Harvey, Robert Robbins, Jonathen Reiskind, Keith Willmott, Jason Hall, Byron Adams, Vitor Becker, Clay Scherer, J. Akers Pence, and Paul Choate. Finally, I thank my parents, Hugo and Sharon Kons, Sr., for encouragement throughout my time as a graduate student.











































iv
















TABLE OF CONTENTS

pane
ACKNOWLEDGMENTS.............................................................................11i

ABSTRACT ......................................................................................... ix

CHAPTERS

IINTRODUCTION...............................................................................1.

Overview .........................................................................................1.
Natural Taxa, Testable Hypotheses, and a Universal Taxon Characteristic ................... 3
Species and Subspecies Concepts Used in this Work: Theoretical Concepts of the
Species/subspecies Category and Practical Applications for Delimitating
Species/subspecies Taxa ......................................................................... 7
The Species Problem ...................................................................... 7
The Theoretical Species Concept Used in this Work .................................. 9
Applying the Evolutionary Species Concept in Practice to Delimitating Species
and Subspecies of Butterflies .....................................................11..I
Subspecies Concepts Rejected in this Work........................................... 13

2 PHYLOGENETIC ANALYSIS OF THE MELITAEIM (LEPIDOPTERA: NYMPHALIDAE: NYMPHALINAE) AND A REVISION OF THE HIGHER CLASSIFICATION OF THE MELITAE1NI......................................................... 16

Introduction.......................................................................................... 16
Materials and Methods
Out Groups................................................................................ 19
Type of Characters Coded for Analysis................................................ 23
In Group Taxa Examined .............................. :................................. 24
Preparation of Melitaeini Genitalia for Character Analysis........................ 26
Production of Melitaeini Genitalia Figures to Illustrate Character States ......... 27 Phylogenetic Analysis ................................................................... 27
Results............................................................................................. 28
Characters and Character States for a Phylogenetic Analysis of the Melitaeini
(Lepidoptera: Nymphalidae)......................................................... 28
Phylogenetic Analysis ................................................................... 82
Discussion ........................................................................................... 86
Revised Higher Classification of the Melitaeini ................................................. 95
Melitaeini Tutt ........................................................................... 96


v









Euphydryiti H iggins ........................................................................................... 98
M elitaeiti T utt ....................................................................................................... 105
G nathotrichiti Subtribe N ..................................................................................... 107
Phycioditi H iggins ................................................................................................. 114
Poladryiti Subtribe N ........................................................................................... 116
C hlosyniti Subtribe N ........................................................................................... 12 1
Key to the Subtribes of the Melitaeini Based on Genitalic Characters .................... 123

3 PHYLOGENETIC STUDIES OF THE CHLOSNYITI AND POLADRYITI (LEPIDOPTERA: NYMPHALIDAE: NYMPHALINAE: MELITAEINI) .............................. 213

In troduction .......................................................................................................................... 2 13
M aterials and M ethods ......................................................................................................... 215
M aterial E xam ined ................................................................................................ 2 15
C haracter C oding .................................................................................................. 2 18
Phylogenetic Analyses Including the Entire Data Set ........................................... 219
Investigation of the Impact of Different Numbers of Taxa in the Analysis on Tree
and C haracter Statistics .................................................................................... 222
Test for Congruence Between Genitalic and Pattern Characters .......................... 223
Test of the Impact of Different Numbers of Taxa on Boot Strap Scores ........... 224
R esu lts .................................................................................................................................. 2 24
Character Coding Overview .................................................................................. 224
Characters and Character States for a Phylogenetic Analysis of the Chlosyniti and
Poladryiti Based on Morphological Data ......................................................... 226
PhylogeneticAnalyses Based on the Entire Data Set ............................................ 332
Variation in Tree Statistics with Different Numbers of Taxa in the Analysis ...... 338
Variation in Character Statistics with Different Numbers of Taxa in the
A n aly sis ............................................................................................................ 339
Variation in the Proportion of Homoplastic Characters with Different Numbers of
T axa in the A nalysis ......................................................................................... 340
Congruency Between Tree Topologies Derived from Genitalic and Pattern
C h aracters ......................................................................................................... 34 1
Variation in Boot Strap Scores for Various Clades as a Result of the Number of
Taxa Included in the Analysis .......................................................................... 343
D iscu ssio n ............................................................................................................................ 34 4
Proposed Hypothesis for the Phylogeny of Chlosyniti and Poladryiti .................. 344
Systematic Check List of the Chlosyniti ............................................................... 344
Phylogenetically Invalid or Unsupported Generic Concepts ................................ 346
Phylogenetically Invalid or Unsupported Species Concepts ................................ 350
Different Models for Character State Polymorphisms .......................................... 357
Variation in Tree and Character Statistics with Different Numbers of Taxa ....... 365
Variation in Boot Strap Scores with Different Numbers of Taxa in the
A n aly sis ............................................................................................................ 36 7
Equal Character Weighting, Successive Character Weighting, and Successive
Character State Weighting: An Alternative Method Proposed for Generating
O ptim al T rees ................................................................................................... 372


vi









Congruency Between Tree Topologies Derived From Genitalic and Pattern
Characters ......................................................................................................... 382

4 PHYLOGENETIC REVISION AND MORPHOLOGICAL CHARACTERIZATION OF THE CHLOSYNITI (LEPIDOPTERA: NYMPHALIDAE: NYMPHALINAE: M ELITAEIN I) ........................................................................................................................... 439

Introduction .......................................................................................................................... 439
M aterials and M ethods ......................................................................................................... 440
Key to the Genera, Species, and Subspecies of the Chlosyniti ................................ 444
Antillea Higgins ................................................................................................................... 459
Antillea proclea (Doubleday and Hewitson) ........................................................ 464
Antillea pelops (Drury) ......................................................................................... 466
The M icrotia Bates and Chlosyne Butler Clade .................................................................. 468
M icrotia Bates ..................................................................................................................... 469
M icrolia eleda (Hew itson) .................................................................................... 472
M icrotia elva Bates ............................................................................................... 478
M icrotia dymas (Edwards) .................................................................................... 483
M icrotia coracara (Dyar) ..................................................................................... 486
M icrotia anomalus (Godm an & Salvin) ............................................................... 490
Chlosyne Butler .................................................................................................................... 491
Chlosyne harrisii (Scudder) .................................................................................. 495
Chlosyne kendalloruin Opler ................................................................................ 507
Chlosyne nycteis (Doubleday) .............................................................................. 510
Chlosyne gorgone (Hubner) .................................................................................. 516
Chlosyne hoffmanni .............................................................................................. 521
Chlosynepalla (Boisduval) ................................................................................... 529
Chlosyne whitneyi (Skinner) ................................................................................. 536
Chlosyne gabbii (Behr) ......................................................................................... 543
Chlosyne acastus (W H Edwards) ...................................................................... 549
Chlosyne definita (Aaron) ..................................................................................... 556
Chlosyne ezra (Hew itson) ..................................................................................... 564
Chlosyne chinaliensis (Tinkham ) .......................................................................... 570
Chlosyneperlula (M enetries) ............................................................................... 578
Chlosyne theona (M enetries) ................................................................................ 586
Chlosyne leanira (Felder & Felder) ...................................................................... 598
Chlosyne endeis (Godm an & Salvin) .................................................................... 632
Chlosyne marina (Geyer) ...................................................................................... 636
Chlosyne melitaeoides (Felder & Felder) ............................................................. 644
Chlosyne erodyle (Bates) ...................................................................................... 647
Chlosyne inelanarge (Bates) ................................................................................. 653
Chlosyne eumeda (Godm an & Salvin) .................................................................. 656
Chlosyne hylaeus (Godm an & Salvin) .................................................................. 659
Chlosyne californica (W right) .............................................................................. 661
Chlosyne lacinia (Geyer) ...................................................................................... 667
Chlosyne ehrenbergi (Geyer) ................................................................................ 685


vii









Chlosyne hippodrome (Geyer) .............................................................................. 689
Chlosyne narva (Fabricius) ................................................................................... 694
Chlosyne gaudealis (Bates) ................................................................................... 699
Chlosynejanais (Drury) ........................................................................................ 705
Chlosyne rosita Hall .............................................................................................. 716

5 SUM M ARY AND CONCLUDING REM ARKS ............................................................... 757

APPENDIX PLATES OF ADULT CHLOSYNITI & POLADRYITI .................................. 763

REFERENCES .......................................................................................................................... 790

BIOGRAPHICAL SKETCH ..................................................................................................... 797









































viii















Abstract of Dissertation Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

PHYLOGENETIC STUDIES OF THE MELITAEINI (LEPIDOPTERA:
NYMPHALIDAE: NYMPHALINAE) AND A REVISION OF THE GENUS CHLOSYNE BUTLER

By

Hugo L. Kons, Jr.

December 2000

Chair: Thomas C. Emmel
Major Department: Entomology and Nematology

An evolutionary hypothesis is presented for the phylogenetic relationships within the Melitaeini (Lepidoptera: Nymphalidae: Nymphalinae), based on cladistic analyses of morphological characters. Melitaeini is revised at the subtribal level, and the concepts of Euphydryiti and Phycioditi but not Melitaeiti are upheld as natural groupings. To create a natural classification scheme, Melitaeiti is restricted to a clade of Eurasian taxa, and three new subtribes are proposed for taxa formerly placed in Melitaeiti: Chlosyniti, Poladryiti, and Gnathotrichiti. Species level phylogenetic analyses are presented for all subtribes except the Melitaeiti and Phycioditi. Euphydryiti is revised at the generic level, and includes the genera Euphydryas Scudder, Hypodryas Higgins, and Eurodryas Higgins, with Occidryas Higgins synonymized with Euphydryas. Gnathotrichiti is revised at the generic level and includes a single monophyletic genus, Gnathotriche Felder and Felder, with Gnathotrusia Higgins placed in synonymy. Poladryiti is revised




ix









at the generic level, and includes three monophyletic genera: Atlantea Higgins, Higginsius Hemming, and Poladryas Bauer. Chlosyniti is revised in detail at the generic and species levels, and includes three monophyletic genera: Antillea Higgins, Microtia Bates, and Chlosyne Butler, with the genera Dymasia Higgins, Texola Higgins, Charidryas Scudder, Thessalia Scudder, and Anemaca placed into synonymy. Thirtyseven species taxa and seventeen subspecies taxa are recognized within the Chlosyniti. Detailed camera lucida drawings illustrate all genitalic characters and character states. The phylogenetic analysis of Poladryiti and Chlosyniti is used as a case study to investigate some issues of broader implication to systematic biology, including equally versus successively weighting characters, tree statistics, homoplasy, boot strap scores, and polymorphisms. Separate analyses of genitalic and pattern characters were conducted for the Cholsyniti/Poladryiti data matrix to investigate the proportion of groupings in conflict between independent data sets for different methods of analysis. The percentage of conflicting groupings was 0% for equal weighting of characters with parsimony, 36.8% for successive weighting and parsimony, 77% for a phenetic algorithm (UPGMA), and 100% between pairs of random trees. The absence of incongruence between independent data sets for the former analysis is argued as evidence supporting the effectiveness and theoretical validity of the phylogenetic methods used in this study.














x














CHAPTER 1
INTRODUCTION


Overview

The Melitaeini (Lepidoptera: Nymphalidae: Nymphalinae) include a group of over 265 butterfly species with representatives in Nearctic, Palearctic, and Neotropical regions with the greatest diversification in the last (Higgins 1960, Higgins 1981). The New World representatives of this group have been the subject of several major revisions, including Higgins (1960), Higgins (1978), and Higgins (1981). The classifications resulting from these revisions were based primarily on similarities and differences in adult morphology. However, previous to the present work, the Melitaeini had never been investigated with the tools of cladistic analyses to construct an evolutionary hypothesis of the natural relationships within the group.

My initial interest in the Melitaeini was based on an interest in conducting a

phylogenetic study of the genus Chlosyne Butler and its close relatives. What attracted me to this genus included a great complexity in morphological variation associated with pattern characters (an impression that was immediately evident examining specimens in museum drawers), and as a strikingly patterned group of butterflies long popular among collectors, the genus was particularly well represented in collections. The combination of complex morphological variation and extensive material available for study led me to believe (correctly) that the Chlosyne were a group well suited for pursuing my interests in phylogenetic studies of Lepidoptera. Early in the project, I decided that to achieve my



I






2

goals with respect to a phylogenetic study of the genus Chlosyne, a study of the higher relationships within the Melitaeini as a whole was desirable, and I expanded my interest to include a phylogenetic study of the tribe. While I studied only a sample of the taxa which Higgins placed in the Phycioditi and of the Palearctic fauna which Higgins (1981) placed in the Melitaeti (designed only to test if these groupings were monophyletic and if so to determine their placement on the phylogenetic tree), I was able to include almost all other melitaeinine taxa in the investigations covered in this work.

The major objectives of this work can be broadly defined as follows: 1) To

determine whether the Melitaeini comprise a natural group (Chapter 2); 2) To derive an evolutionary hypothesis of the major clades within the Melitaeini and of their relationships to each other (Chapter 2); 3) To determine which taxa are most closely related to the genus Chlosyne Butler, and then to derive an evolutionary hypothesis for the relationships among these taxa (Chapter 3); 4) To use the genus Chlosyne and its relatives as a case study to investigate some issues of broader interest to the field of phylogenetic systematics (Chapter 3); 5) To modify existing classification schemes of the Melitaeini, if necessary, based on the criteria of monophyly (=creating a natural classification scheme) and stability (Chapters 2 to 3), and to characterize the natural higher groupings of taxa morphologically; and 6) To revise and morphologically characterize Chlosyne and its relatives, recognizing species and subspecies taxa based on phylogenetic, morphological, distributional, and if available, biological evidence (Chapter 4).

I divide this work into three major chapters. The first (Chapter 2) is a study of the higher relationships within the Melitaeini, and of the relationships within the species






3


comprising several of the smaller clades, based on morphological characters coded from genitalic structures. The second (Chapter 3) is a detailed phylogenetic study of Chiosyne and its relatives, and uses this group as a case study to investigate a number of issues of interest to phylogenetic systematics. The final major chapter (Chapter 4) utilizes the knowledge obtained from conducting the studies in Chapter 3, combined with other information, to revise and morphologically characterize Chiosyne and its close relatives.

An overriding theme that I hope prevails throughout this work is systematics as a scientific discipline based upon testable hypotheses of true patterns that exist in nature. It has been my goal to study and analyze the morphology of the Melitaeini with the best scientific tools available to the modern systematist, and to part ways to the greatest extent possible with arbitrary decision-making which has been prevalent in systematics in the past. Particulars of this goal are presented in all of the three major chapters, but in the introduction to the entire work, I find it appropriate to address two issues of general significance: 1) What constitutes a valid taxon in my vision of biological systematics; and 2) The complexities of the species problem, and how the modern tools of systematics can be used in actual practice to delimit natural species and subspecies taxa within Lepidoptera (in particular), based on testable hypotheses.


Natural Taxa, Testable Hypotheses, and a Universal Taxon Characteristic

In a review of the previous literature regarding the classification of the Melitacini, I find quotes such as the following to be plentiful: "these differences are so marked that I have not found it possible to write a generic synopsis to include both..." (Higgins 1960), "The specialised characters [of CharidryasJ do not appear to be sufficiently marked to justify generic separation" (Higgins 1960), "the species associated with Charidrv as ... are






4


sufficiently distinct in maculation to merit recognition" (Ferris 1989), "Higgins included three Mexican species in the genus, but the other two species are about as close to Microtia and Dymasia as they are to the type species, Texo1a eleda" (Bauer 1975), and "The dark populations in southern Mexico have been referred to as the nominotypical subspecies but are sufficiently different to recognize" (Austin and Smith 1998b). An overriding theme that I find in common with these and many similar statements regarding taxa delimitation in the literature is arbitrary decisions being based on a worker's perception of similarity and difference. These statements and many others like them likewise illustrate the conflict and instability that results from classifications based on such criteria, as different workers often have differing views on what is most similar, dissimilar, or sufficiently different to warrant recognition. What I find absent in the above quotes, is a testable hypothesis upon which taxon delimitation is based, and by which future workers objectively may analyze evidence to determine if the taxa delimited are valid. Furthermore, of what relevance to the goals of biological systematics are my personal perceptions (or those of any other worker) regarding what entities are most similar or different?. While accepted under the International Code of Zoological Nomenclature, I personally reject the criteria of sufficient similarity/sufficient difference as a basis for delimitating taxa of any rank in the scientific discipline of biological systematics because 1) Such criteria represent arbitrary opinions rather than testable hypotheses (regardless of a workers level of experience); and (2) Such criteria often result in the delimitation of artificial taxa.

I define a natural taxon as a group of organisms representing a true product of the natural process of evolution with a real existence in nature, as opposed to an artificial






5

taxon, which represents a product of the human imagination with no real existence in nature. In other words, natural taxa are monophyletic evolutionary lineages, including by definition all the descendents of a common ancestor. Artificial taxa are any groupings of organisms which a human being artificially creates, and do not represent an evolutionary lineage. I make no distinction between paraphyletic and polyphyletic taxa. A paraphyletic group has been defined (originally by W. Hennig) as a group not including all of the descendents of a common ancestor (Scotland 1992a), while a polyphyletic group has been defined as "a group in which the most recent common ancestor is assigned to some other group and not to the group itself' (Scotland 1992a, quoted from E. Wiley via J. Farris). Because all life is believed (based on current evidence) to have a common origin, any conceivable grouping of extant organisms that is not monophyletic is by definition paraphyletic. Consequently, for the purposes of classifying extant organisms in a natural classification scheme, the only true distinction is between monophyletic (natural) and nonmonophyletic (artificial) groups.

Artificial classification schemes can be very useful for certain purposes, such as deciding what to include in a book, what to include in a university course, what to include in a pest control manual, etc. Some examples of paraphyletic groups sharing some type of similarity used for these purposes include reptiles, invertebrates, moths (excluding butterflies), the Lepidoptera of Florida, Florida pests of citrus, medically important insects, urban pests, beautiful moths, etc. Although I recognize the utility of such groupings for these purposes, I argue that such groupings (or any nonmonophyletic grouping) have no place being formally named or ranked in the scientific discipline of biological systematics for several reasons: 1) They are artificial, do not exist in nature,






6


and lack evolutionary significance; 2) The scientific method cannot be used to test their validity, as they have no characteristic corresponding to a testable hypothesis in common with other groups; 3) They are a poor fit with a hierarchical classification of sets within sets and nonoverlapping sets used in systematic classification; and 4) They would result in highly unstable classifications, because there are many different ways people can view organisms as being similar or different.

As the study of real patterns in nature and the use of testable hypotheses are an obvious part of any scientific discipline, if systematics is to be viewed as a scientific discipline the importance of the former two reasons is evident. With regard to the third reason, systematic classification does not allow, for example, one species to be a member of two different genera. If one were to ask the question, is Taxon A more similar to Taxon B or to Taxon C, different workers could easily arrive at different answers because there are many ways that taxa can be viewed as similar or different, yet Taxon A cannot be placed in two different genera even if it shares some type of similarities with both. However, there is only one way that Taxa A, B, and C can be related in terms of evolutionary relationships.

Although the hierarchical classification of systematics was developed before the theory of evolution, I argue that this hierarchy nonetheless lends itself well to a natural classification and relatively poorly to an artificial one. The decision that Taxon A is more similar to B than C is ultimately an arbitrary opinion, and any other worker could just as validly arrive at the opposite opinion. The decision that Taxon A is more closely related to B than C is a testable hypothesis. In order to change Taxon A's placement with Taxon B in a natural classification, one would need to provide additional evidence






7


showing that the initial hypothesis is no longer favored, as opposed to arbitrarily deciding that Taxon A is more similar to Taxon C. Even in a natural classification, disputes can still arise over the validity of particular taxa as two workers may reach conflicting hypotheses from the same evidence. However, the scientific method is self-correcting over time, as new evidence and/or an improved theoretical framework favors one hypothesis over its alternatives.

In all aspects of this work, I choose the criterion of monophyly as a universal requirement for the validity of a taxon of any rank. While monophyly provides a scientific basis for testing the validity of taxonomic concepts, it does not address another issue in the systematic classification of organisms: how many natural taxa does one formally name and how are they ranked. Above the species level, I see no way such a question can be answered with the scientific method, and the issue is a matter of bookkeeping rather than science. However, I argue that making such bookkeeping decisions objectively is preferable to arbitrarily doing so. This goal can be achieved by applying a secondary criterion of nomenclatural stability, as discussed in Chapters 2 and 3 where phylogenetic evidence is used to convert the existing classification schemes into a natural classification scheme.


Species and Subspecies Concepts Used in this Work: Theoretical Concepts of the
Species/subspecies Category and Practical Applications for Delimitating Species/subspecies Taxa

The Species Problem:

Ereshefsky (1992) makes an important distinction regarding the use of the term it species." One usage is the species category, relating to what characteristics make an entity a species, and the other usage is species taxa, that which is formally named as a






8


species by taxonomists. Biologists have widely differing views on how the species category is defined (Ereshefsky 1992), and consequently species taxa are delimited in a variety of different ways. In my view, a clear-cut universally applicable species concept is unachievable, both in theory and practice. If species are required to be natural (=monophyletic) lineages, the speciation process in its simplest special case involves a dichotomous branching event where an original population of one species diverges over time into two species. Bell (1997) points out that the unit event of the evolutionary process is an episode of variation (=reproduction in sexually reproducing organisms) followed by an episode of selection. It seems untenable that between two such episodes there would occur a clear transformation of an entity changing from one species to a different species, or from a subspecies to a species, or from a population to a subspecies. Whatever characteristics are assigned to the species category or a practical method of delimitating species taxa, because evolution is an ongoing process, as an ancestral lineage begins to diverge into two lineages (by a series of episodes of reproduction followed by episodes of selection) that may eventually become clear-cut separate species, there will be a period in-between when the two divergent lineages are in the process of acquiring the characteristics that make them species. The status of such lineages will always represent a gray area in delimitation of species taxa. Also, while it is clear that organic diversity falls into clusters (separate lineages) as opposed to a continuum of genetic variation (Ereshefsky 1992), Mishler and Donoghue (1992) point out that units corresponding to these clusters may not be directly comparable, and that discontinuities between clusters (such as morphological versus reproductive) may not correspond. I would infer from my studies of the Chlosyniti that no two clusters of organic diversity






9


that I recognize as basal taxa are equivalent units, because divergence between clusters varies in kind and degree. In my view, it is essential that taxonomists first objectively describe the patterns of variation that they find in nature, and then translate this information into species/subspecies decisions, as opposed to basing descriptions of the patterns of variation in nature on an attempt to conform to a particular species/subspecies concept.


The Theoretical Species Concept Used in this Work

Although I find the goal of a universally applicable clear-cut species concept

untenable, I do conclude that one of the species concepts proposed has the greatest utility to the field of biological systematics. In my view, a species concept useful to biological systematics should have two important characteristics: 1) All species units are natural (=monophyletic) taxa; and 2) Species units have an additional biologically and evolutionarily significant characteristic that distinguishes them from higher and lower monophyletic groups, such as genera and populations, respectively. However, with any species concept one must just accept that there are some cases where lineages will be in a state of transition in the speciation process, and in these cases species taxa delimitations will have to be made more arbitrarily.

The species concept that in my view best achieves the above goals is the

evolutionary species concept described by Wiley (1992), where species are defined by: "A species is a single lineage of ancestral descendent populations of organisms which maintains its identity from other such lineages and which has its own evolutionary tendencies and historical fate" (Wiley 1992). 1 regard a more concise definition as: a species is a monophyletic lineage biologically incapable of reticulating with a different






10


evolutionary lineage. The point where an evolutionary lineage loses its ability to merge with another lineage is in my view the theoretically and biologically significant event that separates a species lineage from a lower lineage. Although Wiley (1992) argued that a major advantage of his species concept was its applicability to all types of organisms, with respect to Lepidoptera (sexually reproducing organisms) a comparison can be made with the Biological Species Concept of Mayr (1963), where species were defined as "Groups of actually or potentially interbreeding populations, which are reproductively isolated from other such groups." For sexually reproducing organisms, the criteria of 11not actually or potentially interbreeding" and "maintains its identity from other such lineages" are very similar, with the exception of "gray area" cases where some amount of interbreeding occurs between two lineages yet they remain somewhat distinct. While reproductively isolated lineages may remain very similar morphologically over long periods of time, as exemplified by morphologically very similar butterflies and moths isolated on different mountain ranges, for two separate lineages to reticulate into one, reproduction must occur. In fact, with respect to sexually reproducing organisms, in my view the evolutionary species concept is a consequence of putting the biological species concept into an evolutionary context, achieved by adding a requirement that species units be natural products of the evolutionary process. A very serious flaw of the biological species concept is that it may recognize artificial (=nonmonophyletic) taxa not useful for phylogenetic studies when applied to allopatric populations (Mishler and Donoghue 1992). Regarding allopatric populations, O'Hara (1994) points out a flaw common to both the biological and evolutionary species concepts. In some situations, the decision of whether two lineages are separate species is based on knowing what will happen in the






11

future (will circumstances arise in the future where two currently allopatric lineages will come together and be capable of loosing their distinct identities or interbreeding?). Applying the Evolutionary Species Concept in Practice to Delimitating Species and Subspecies of Butterflies:

As a place to start, I search for gaps in the range of morphological variation of one or more characters. Assemblages of individuals that I find have a consistent morphological discontinuity become basal taxonomic units to be entered into the phylogenetic analysis. The presence of a consistent morphological discontinuity represents a testable hypothesis by which to initially delimit terminal taxa. Of course, this gap criterion is not a perfect one, but it does provide a place to start. It is possible for two distinct lineages to exist within a group of individuals that I cannot distinguish morphologically. Hypothetically, those lineages may not even be sister lineages. It is also possible that gaps between groups of taxa could be the result of polymorphism in the same lineage (C. hylaeus versus C. eumeda? See Chapter 4), or a result of insufficient collecting to show a true geographic continuum that exists in nature (C. rosita mazarum versus C. rosita riobalensis See Chapter 4). An advantage of selecting a relatively well collected group to study, such as Chlosyne, is that the risk associated with these two types gap errors is minimized. The monophyly of terminal Operational Taxonomic Units (OTUs) identified by morphological discontinuity need not be hypothesized based on the gap alone. Additional evidence to support the monophyly of terminal OTUs can be achieved by entering them into the phylogenetic analysis twice and seeing if they come out as sister taxa (supporting monophyly) or in a polytomy (providing evidence neither for nor against monophyly).






12


The testable hypotheses on which species delimitation are based include the

following: 1) Is the species taxon monophyletic?; and 2) is the species taxon biologically capable of reticulating with another evolutionary lineage (present or future)?. The first hypothesis is tested by cladistic analysis. Evidence that I find that substantially supports (although not necessarily guarantees) the second hypothesis includes any one or a combination of the following: 1) Sympatric occurrence of two distinct taxa; 2) Discontinuity in the range of morphological variation of a sclerotized structure, especially in the structure of the genitalia; 3) A lineage has no sister species, and is sympatric with at least one of the taxa in its sister dlade; or 4) Evidence of hybrid inviability with closely related lineages. In these cases, the evolutionary species concept is applied easily in actual practice, because the evidence supporting the testable hypotheses of an evolutionary species is fairly clear.

However, now consider a monophyletic group of two allopatric taxa with identical genitalia, no known information regarding hybrid inviability, but with morphological discontinuities in wing pattern or other pattern characters. Based on this evidence, these taxa may be hypothesized as distinct evolutionary lineages, but it is impossible to know whether or not they are evolutionary species. This is where I feel a subspecies concept can be useful. Theoretically, I define the subspecies category as follows: A subspecies is a single lineage of ancestral descendent populations of organisms that currently maintains its identity from other such lineages but that has not lost its biological ability to reticulate with another lineage except by geographic isolation. The future fate of a subspecies could be a reticulation event with another subspecies or further divergence into a separate species. In practical application, I delimit subspecies






13


taxa based on the presence of evidence to suggest they are distinct lineages but the absence of evidence to suggest they are evolutionary species. In other words, my definition of a lepidopteran subspecies in practical application is: a monophyletic lineage, allopatric with its closest relatives, with which it has identical genitalia (and other scierotized structures) but discontinuities in wing pattern. A subspecies could be increased in rank to a species if evidence is acquired that favors the hypothesis tested to delimit an evolutionary species over an alternative hypothesis.


Subspecies Concepts Rejected in this Work:

The vast majority of Chlosyne subspecies included by various authors do not meet either my criteria for a valid taxon or a valid subspecies. A common method of delimiting subspecies within the Chlosyniti has been to name populations, groups of populations, or groups of individuals comprising parts of different populations, representing points along a continuum of geographic variation. Geographic gradients are often non-uniform, with phenotypes remaining more constant over some areas and changing more rapidly over others. Subspecific status is assigned to populations occupying various geographic areas based on the criterion (extremely arbitrary in my view) that the worker(s) naming the subspecies feel it is sufficiently distinct to warrant recognition (see quote by Austin and Smith 1 998a above). Populations that represent intermediates between the named populations are not assigned to any subspecies taxon, and are referred to as subspecies A/B blends or intermediate populations (see Austin and Smith 1 998a, Austin and Smith 1 998b) or as unassigned or undetermined populations (see Smith and Brock 1988). Because some populations or individuals within the species are omitted from the subspecies concepts, the subspecies concepts are by definition






14


artificial. I reject all such subspecies as artificial taxa delimitated by arbitrary decisions rather than natural taxa delimitated by testable hypotheses.

I also do not view such subspecies concepts as useful for describing geographic variation. Consider the character of forewing length which varies geographically in C. theona (Austin and Smith 1998b). Austin and Smith (1998b) report the mean measurements and range (low-high) for a variety of artificial subspecies taxa, but do not report such information for populations designated as "blend populations". If the goal is to describe geographic variation, the blend populations arejust as significant as any population within the artificial subspecies concept, and may be of particular interest if they represent areas of higher variability or a steeper phenotypic gradient. For the purposes of quantifying geographic variation, I argue the best approach would have been to measure individual populations and to present this data, as opposed to presenting measurements for an artificial assemblage of populations.

The second type of subspecies concept which I reject is naming populations within a continuum of variation but with reported different averages in the range of variation for a character compared with adjacent populations. I distinguish between population differences and subspecies differences. Because selective pressures are unlikely ever to be completely identical between two separate populations, invariably every local population has a difference in the average range of variation for some character compared with adjacent populations. If such differences provide a basis for naming subspecies, every local population, no matter how temporary, would have to be named as a subspecies. If one requires that the average difference be sufficiently distinct






15


to warrant subspecies recognition, once again taxa are being delimitated by arbitrary opinions as opposed to testable hypotheses.















CHAPTER 2
PHYLOGENETIC ANALYSIS OF THE MELITAEINI (LEPIDOPTERA:
NYMPHALIDAE: NYMPHALINAE) AND A REVISION OF THE HIGHER CLASSIFICATION OF THE MELITAEINI


Introduction

The higher classification of Nymphalidae that I adopted as a starting point is the one proposed by Harvey (1991). Harvey's (1991) concept of Nymphalidae includes approximately 6,452 species worldwide (Shields 1989), and thirteen subfamilies including Heliconiinae, Nymphalinae, Limenitidinae, Charaxinae, Apaturinae, Morphinae, Brassolinae, Satyrinae, Calinaginae, Danainae, Tellervinae, Ithominae, and Libytheinae. Previous classifications, including Ehrlich (1958), Miller and Brown (1981), and Ackery (1988), ranked Melitaeinae as a subfamily of a narrower concept of Nymphalidae, with some of Harvey's (1991) subfamilies ranked as separate families. Higgins (1981) included three tribes within Melitaeinae, the Melitaeini, Phyciodini, and Euphydryini. A consequence of Harvey's (1991) classification is that Melitaeinae is given a tribal rank and the three tribes of Higgins (1981) are now ranked as subtribes Melitaeiti, Phycioditi, and Euphydryiti.

Harvey (1991) recognized three tribes within the Nymphalinae, including

Nymphalini, Kallimini, and Melitaeini. The synapomorphy that Harvey (1991) used to define this group was the larval character of a scolus on abdominal segment 9 (A9) ventral and posterior to the filiform setae. Harvey (1991) noted that Heliconiinae also




16






17


have a scolus on A9, but because it occurs in a different position (dorsal to the filiform setae), he presumed that it was not homologous to that of the Nymphalinae.

Harvey (1991) reported that systematic relationships within the Nymphalinae were poorly understood, particularly at the generic and tribal levels. The relationships beamong Nymphalini, Kallimini, and Melitaeini are unknown. Nymphalini and Melitaeini share a feature unique to the Nymphalidae of having filiform setae present on A 1,2, while Kallimini and Melitaeini share a feature unique to the Melitaeini of having the filiform setae on A9 occurring on the sclerotized base of the scolus (Harvey 1991). Consequently, at least one of these characters must have been subject either to reversal, or potentially less likely, to independent acquisition.

As noted by Harvey (1991), Higgins (1981) never provided a synapomorphy for the Melitaeini. Harvey (1991) reported the notched saccus in the male genitalia as a unique character for Melitaeini among the Nymphalidae, but actually several Phycioditi lack a notched saccus, so this character does not represent a universal synapomorphy. Harvey (1991) also noted that the distribution of filiform setae on Melitaeinine larvae is diagnostic within the Nymphalidae, based on the combination of filiform setae on A1,2 and on A9 at the sclerotized base of the scolus. However, although the combination of these features is a good identification character, it provides no evidence of monophyly for the Melitaeini, because each is present in some out group taxa.

The following is a summary of Higgins' (1981) higher classification of the

Melitaeini, which includes three subtribes (ranked as tribes by Higgins (1981)) and 31 genera. The species that Higgins (1981) included in each genus appear on pages 165-171 of his 1981 review of the classification of the Melitaeini.






18




Higgins' (1981) higher classification of the Melitaeini (with tribes down-ranked to subtribes):

Euphydryiti Higgins: Euphydryas Scudder, Hypodryas Higgins, Occidryas Higgins, and Eurodryas Higgins

Melitaeiti Tutt: Mellicta Billberg, Melitaea Fabricius, Poladryas Bauer, Didymaeformia Verity, Cinclidia Hubner, Chlosyne Butler, Thessalia Scudder, Texola Higgins, Dymasia Higgins, Microtia Bates, Gnathotriche Felder & Felder, Gnathotrusia Higgins, Higginsius Hemming, Antillea Higgins

Phycioditi Higgins: Phyciodes Hubner, Phystis Higgins, Anthanassa Scudder, Dagon Higgins, Telenassa Higgins, Ortilia Higgins, Tisona Higgins, Eresia Boisduval, Castilia Higgins, Janatella Higgins, Mazia Higgins Unplaced Genus: Atlantea Higgins


The support of a final phylogenetic tree can be no stronger than the support of any assumption made along the way, and because phylogenetic trees generated in parsimony analysis with morphological data are rooted and the characters are then polarized based on the out group (Kitching 1992a), the out group decision is critical. Although the best evidence currently available suggests that Nymphalinae forms a monophyletic group including Nymphalini, Kallimini, and Melitaeini (Harvey 1991), the relationships among these three tribes are unclear (Harvey 1991). Consequently the closest relatives of the Melitaeini are unknown. A similar problem of out group uncertainty was addressed with respect to a phylogenetic study of Rekoa Kaye (Lycaenidae: Theclinae) by Robbins (1991). Robbins (1991) presented evidence that Rekoa was related to Arawacus Kaye and Thereus Johnson, but was unsure which represented the sister group and unsure of the of relationships within these two genera. Robbins (1991) solved this problem by including taxa in these two genera as the out group taxon, and by coding them






19


collectively as one out group taxon. This out group taxon was coded for a state not present in the in group (=9) for characters not useful for tree rooting because one of the following was the case: 1) Out group character information was missing; 2) all in group character states occurred in the out group; or 3) None of the in group character states occurred in the out group. For the remaining characters, Robbins (1991) coded the out group taxon as the state or combination of states occurring in the in group that also occurred in the out group.

The objectives of this study were to code morphological characters for cladistic analysis to 1) Test the hypothesis that the Melitaeini form a monophyletic group; (2) To reconstruct a phylogenetic hypothesis of relationships within the Melitaeini; and (3) To improve the characterization of the Melitaeini based on adult morphology. Objective 2 included testing the validity of the subtribes and genera recognized by Higgins (1981 ) (with the exception of those within the Phycioditi and Eurasian Melitaeiti, for which the aim of the study was only to determine whether those groups were monophyletic and the placement of their clades on the Melitaini phylogeny), and to revise the higher classification of the Melitaeini such that natural evolutionary relationships are reflected by the higher taxa recognized. Finally, a major aim of this study was to determine which genera are related to Chlosyne, and to determine an out group for an additional detailed phylogenetic study of Chlosyne.


Materials and Methods


Out Groups






20


1 examined a number of taxa within Nymphalini and Kallimini to try to avoid making unwarranted assumptions on character polarization based on missing data. I adopt a similar out group approach to that of Robbins (199 1) for combining Nymphalini and Kallimini as the out group for a phylogenetic study of the Melitaeini, with my coding guidelines outlined below. Some of my coding decisions are different from those of Robbins (1991) in order to provide more specific information in the matrix, such as distinguishing among the three scenarios where Robbins coded the out group "9". However, their affect on the phylogenetic analysis is unchanged. For the purposes of discussion, I refer to this method as the "Cumulative Out Group Method."

I designate "c" as the symbol for a state occurring in out group taxa that never

occurs within the in group, and "n" and "p" as different character states that occur in both the in group and out group (in the actual data matrix "n" or "p" would be designated as a numerical state, except for multistate characters with more than 10 states where state assignments follow the sequence "0,1,2,3,4,5,6,7,8,9 ,a,b,d" (letters are used because MacClade 3.07 interprets "10,' as "0 & 1"). The following is a summary of how I coded characters for the cumulative out group: 1) All out group taxa exhibit a state (n) which also occurs in the in group: the out group is coded "n". 2) Some out group taxa have state n, while others have state p, and both states n and p occur within the in group: the out group is coded "n,p" (=n&p in MacClade) (if no states occur within the in group other than n and p, coding the out group as "T' or "c" would be equivalent in terms of affecting the computer analysis, but less informative regarding the distribution of character states among out group taxa). The same coding methodology would apply to characters with three or more states occurring in both the out group and in group. 3) No






21


out group taxon exhibits a state that occurs in the in group: the out group is coded "c" (the number of states occurring in out group taxa, or how they would be designated is irrelevant to cumulative out group coding). 4) Some out group taxa exhibit a state that does not occur in the in group, while others exhibit one or more states that do occur in the in group (n and p): the out group is coded "c,n,p." (5). Some out group taxa are coded for a particular character, but state information is unknown for one or more taxa included in the cumulative out group: the out group is coded "?". 6) A character clearly falls into discrete states (n and p) within the in group, but not among out group taxa: the out group is coded "?", while in group taxa are individually coded "n" or "p" as appropriate.

In an attempt to include a sample representative of the character state variation within Nymphalini and Kallimini, I examined the males of the type species of each Nymphalini and Kallimini genus listed in Harvey's (1991) Table B.2 except for a few genera for which I did not have males of the type species available (Bassaris, Symbrenthia, and Mynes). Information on the specific types for these genera was obtained from Hemming (1967). I also included a representative of Liminetidini, Colobura dirce, in the out group. The taxa comprising the cumulative out group are listed in Table 1 under "Out Group Taxa". If the cumulative out group method is applied to the phylogenetic analysis of a genus where several genera are potentially sister taxa, for scoring the cumulative out group it would be best to include all species taxa in the genera which are potential sister taxa. However, with respect to the phylogenetic analysis of a tribe such as Melitaeini with two subtribes as potential sister taxa, an approach of examining all species taxa in Kallimini and Nymphalini would be prohibitively impractical.






22


With respect to the cumulative out group method, I have polarized characters within the Melitaeini based on two assumptions. The first is that the sister group to Melitaeini includes and is exclusively composed of some combination of taxa currently classified within the Nymphalini and Kallimini. The second assumption is that the sample of generic types is representative of the character states that occur both within the true Melitaeinine out group and the in group (Melitaeini). I have made no assumption that particular taxa or groupings of taxa within the Nymphalini and Kallimini are most closely related to Melitaeini.

In addition to the use of the cumulative out group method, I developed a second method for polarizing characters within the Melitaeini (in a separate analysis) which I term the "Representative Out Group Method." Use of the representative out group method requires that characters are first scored for the cumulative out group method. From the resulting data matrix, I then selected representative out group taxa, such that the group of representative taxa collectively includes all character states occurring within the in group. These taxa were then entered into the data matrix individually rather than collectively, and specific states were assigned for each taxon in place of the "c" designation (invalid for this analysis because "c" can refer to more than one character state). I scored some additional characters that are invariant within Melitaeini but for which there is variation in states among the representative out group taxa. The reason for doing this was so that all the representative out group taxa would not, evidence permitting, come out as a unresolved basal polytomy where the number of equally parsimonious out group permutations would be multiplied by the number of equally






23

parsimonious trees within the in group, thereby greatly increasing the time needed to do an analysis in PAUP.


Type of Characters Coded for Analysis

The data matrix obtained for the Melitaeini is based exclusively on genitalic

characters. I searched for other external characters of sclerotized structures, but did not find variation that could be coded into discrete states within the in group. Scale pattern characters were not used in this analysis, because I did not have series of material from throughout the geographic range for a portion of the taxa examined (wing pattern can be highly geographically variable within the Melitaeini, but with a few nonproblematic exceptions I have found almost no geographic variation in genitalia). Another reason for not using pattern characters in this analysis is that many taxa in the Phycioditi and Eurasian Melitaeiti were not included in the analysis, and while the samples examined may provide a very good indication of the range of variation in relevant genitalic characters (=characters informative with respect to the monophyly of Phycioditi and Eurasian Melitaeiti and the placement of their clades on the phylogeny, as opposed to only informative with respect to relationships within these groups), they may not with respect to wing pattern.

Homologous characters existing in more than two forms among the taxa

examined were coded as multistate characters as opposed to multiple binary characters. Coding of certain complex multistate characters is discussed on a character by character basis in the "Results: Characters and character states for a phylogenetic analysis of the Melitaeini based on genitalic characters" section below. The general guidelines I strived to follow when making character coding decisions include: (1) no a priori assumptions






24

are made regarding character state order, (2) character coding schemes which weight the same change more than once are avoided, (3) coding schemes which may produce unwarranted bias regarding character state polarity are avoided, (4) a continuum of variation is not arbitrarily divided up into a series of discrete states, (5) characters which seem to be completely dependent (such as two separate wing patches always being the same color-more relevant to the following chapter) are coded only once, (6) for characters involving further division of a character state into additional states which do not apply to taxa lacking the state to be divided, "?" coding is avoided if but only if this can be accomplished without weighting an identical step twice, and (7) characters which can be coded into discrete states are not deleted because a priori I feel they may be homoplastic.


In Group Taxa Examined

Taxa examined and coded for characters are listed in Table 1. All described Melitaeini taxa other than Phycioditi and Eurasian Melitaeiti were included with the following exceptions for which I did not have material available for study: Eurodryas alexandria (Staudinger), Eurodryas orientalis (Herrich-Schafter), Gnathotrusia steinii (Dewitz), and Atlantea cryptadia. Also, I did not dissect the few specimens I examined of Atlantea perezi (Herrich-Schaffer); however, judging from figures 11-13 (pg. 391) in Higgins (1960), I suspect there would be very few if any differences in character coding between this taxon and the other Atlantea examined. Several taxa had to be coded for males only, since I had no females available for study. These taxa are Texola anomalus (Godman & Salvin), Higginsiusfasciatus (Hopffer), Gnathotriche sodialis Staudinger, and Gnathotrusia mundina (Druce).






25

Since the Phycioditi and Eurasian Melitaeiti are large groups, with 137 and 67

species, respectively, listed in Higgins (1981), and since the aim of this study was only to determine if these groups are monophyletic and where their clade appears on the Melitaeinine phylogeny, I limited my coverage of these groups to a sample. I coded characters for 18 Phycioditi, including the type species of each genus. I also examined Higgins (1981) Figures 183-477, which include genitalia drawings of males of most of the Phycioditi, and while these figures are limited in their detail, they do suggest the sample of taxa examined provides good representation of the range of genitalic variation within Phycioditi. I suspect most or all characters and character states not included in the sample are particular to certain Phycioditi, and consequently not important to an analysis aiming only to determine if the Phycioditi are monophyletic and the position of their clade on the Melitaeinine tree. I coded characters for 13 species of Eurasian Melitaeiti, including the types of each genus plus several additional representatives.

The number of male individuals which I dissected varied among the in group taxa. For taxa examined in the genera Chlosyne, Thessalia, Charidryas, Anemaca, Texola, Dymasia, Microtia, Antillea, and Poladryas, I dissected a minimum of three individuals from different parts of a taxon's range, and in nearly all cases more than three individuals were examined. The sole exception is Texola anomalus, for which I was only able to examine one specimen. For Phycioditi and Eurasian Melitaeini, two or more individuals were examined for the type species of each genus, and one to two individuals for the remaining taxa. For Euphydryiti, three or more individuals were dissected for Nearctic taxa and the types of each genus, while one to two individuals were dissected for remaining taxa. I dissected only one individual of Gnathotriche sodialis and






26


Gnathotrusia mundina, two individuals of Atlantea tulita, and three or more individuals of the remaining taxa. The vesica was everted, or partially everted, for at least one male specimen of every taxon in the data matrix except for Janatella leucodesma and Ortilia liriope.

Since I found very little intraspecific variation in the samples dissected other than for the exact shape of the saccus for some taxa, I find it very unlikely that dissecting additional individuals of the taxa examined would result in changing character coding schemes for the characters coded. Fewer female specimens were dissected relative to males, and for taxa other than Chlosyne, Thessalia, and Charidryas, in most cases I dissected only one or two females per taxon.

For the genera Chlosyne, Thessalia, Charidryas, and Anemaca, all 41 Operational Taxonomic Units (OTUs) I recognize have the same character state for each character in the Melitaeini data matrix, with very few exceptions noted under the appropriate characters. Consequently, I entered these taxa into the data matrix collectively as "Chlosyne A" and "Chlosyne B". This is preferable to entering these taxa into the matrix collectively as one taxon, because a single collective OTU would not permit the analysis to show if there is evidence for the monophyly of these taxa (if two or more taxa are coded with identical states for all characters, that by itself does not imply monophyly).


Preparation of Melitaeini Genitalia for Character Analysis

I studied genitalic characters from genitalia with their natural shape intact (as

opposed to putting genitalia on slides as was done by Higgins (1960) and Higgins (1981), which involves destroying the three dimensional structure of Melitaeinine genitalia). Genitalia were dissected using standard techniques of removing a specimen's abdomen






27


and dissolving away proteinaceous material with a 10% KOH solution. The maximum magnification at which I examined genitalia was 75X, and light sources included fiber optics and/or lighting from below the microscope platform. Some preparations were stained with chlorozol back dye to highlight membranous areas.


Production of Meiitaeini Genitalia Figures to Illustrate Character States

All the genitalia illustrations (Figs. 1-292) are from camera lucida drawings. I drew these figures at 50X magnification, except for Figure 292 (drawn at 25X magnification). All the drawings were done in pencil, with differing degrees of shading used to show the difference between darker and lighter areas. I scanned each drawing, retaining its original size, with a Hewlett Packard Scan Jet ADF at 200 DPI with the output type set at grayscale. The drawings were then scanned to Adobe Photoshop 5.0.2, where I selected "Image: Adjust: Brightness/Contrast" and increased the contrast to 40%. The drawings were then saved as TIFF files, and imported into Microsoft PowerPoint, again with the original size of the drawing retained. Subsequent alterations in size (if any) as indicated on the figures, were done in PowerPoint.


Phylogenetic Analysis

Eight separate analyses were conducted with PAUP version 4.Ob4a for Macintosh, four for the cumulative out group method and four for the representative out group method. The four types of analyses included a heuristic search with equally weighted characters, a series of heuristic searches with successively weighted characters (weighting was based on character resealed consistency indices) until the tree length no longer changed, a boot strap analysis with equally weighted characters, and a boot strap






28


analysis with successively weighted characters. Strict consensus trees were calculated from heuristic searches, and 50% consensus trees were calculated from the boot strap analyses. Tree scores (consistency index, retention index, and rescaled consistency index) were calculated in PAUP based on parsimony informative characters only.

All heuristic searches were conducted with characters unordered (in my view ordering characters almost always introduces additional unwarranted assumptions into the analysis). Also, all heuristic searches were run to completion and the strict consensus trees were calculated from all of the equally parsimonious trees. For rooting options, for the cumulative out group method the "make in group monophyletic" option was selected. For the representative out group method, the option "root tree at internal node with basal polytomy" was selected.

Boot strap analyses were run with the fast stepwise addition algorithm and based on 10,000 replications. The option "retain groups with frequency >50% was used for each boot strap analysis. For analyses based on successively weighted characters, the option "sample characters with equal probability but apply weights" was selected.


Results


Characters and Character States for a Phylogenetic Analysis of the Melitaeini (Lepidoptera: Nymphalidae)

Almost all of the below character states occurring within the Melitaeini are illustrated in the camera lucida genitalia drawings presented in Figures 1-292, except for a few states particular to Phycioditi and European Melitaeiti which may be valuable to future investigations of relationships within these clades, but are irrelevant to establishing if these groups are monophyletic or how they are related to other clades. While at least one






29


figure is referenced for each illustrated character state, the below references to figures to illustrate the character states are not comprehensive; a state is often illustrated in more figures than are specifically referenced for that state. Since genitalia drawings are presented in phylogenetic order (with some exceptions to allow for more efficient use of space), and each drawing depicts multiple character states, the figure references for character states do not occur in sequence.


Characters

# =A character state which does not occur within the Melitaeiti but occurs in Nymphalini or Kallimini out group taxa.

C=For the collective out-group coding, a character state which is different from any that occurs in the Melitaeini.

Male genitalia

Characters of the valvae:

1. Hollow projection on each valve off of the inner wall, projecting off of a distinct plate on the inner posterior side of the valve (well posterior of the anterior edge of the tegumen):

#1 =Absent.

2=Present (Fig. 1).

3=Vestigial, a hollow projection is absent but a reduced plate is present in the

same position (Fig. 64).

The Kallimini generic type of Doleschalia (bisaltide) has a projection extending from the inner wall of the valvae, but it differs in position and structure from that of the Melitaeiti. I code this projection as state 3. Out group taxa Colobura dirce and Salamis augustina






30

and have hollow inner valve projections in roughly the same position on the valve as those of Melitaeini, however the shape of these projections and form of the valvae for these taxa is dissimilar to that found in Melitaeini. Some other out group taxa have dissimilar inner valve projections in different positions relative to the Melitaeinae, and many out group taxa have no inner valve projections. The collective coding for the outgroup is "1 & c", since no out-group taxa have inner valve projections that could reasonably be assumed to be homologous to those of the Melitaeini (and in any case certainly not coded as a Melitaeini state), yet for some taxa, homology can not definitively be assumed a priori to be lacking either.

2. For those taxa with state 2 of the preceding character, the general orientation of the inner valve process:

l=Posterior lateral (Fig. 1).

2=Anterior lateral (Figs. 59 & 62).

The differences between state 1 and 2 can be thought of as a hinge at the base of the valve projection, with the hinge up in state 1 and down in state 2 (contrast next character).

In Gnathotrusia mundina the inner valve process is vestigial and can not be clearly assigned to either state. I code this taxon as "?". This is the only Melitaeini examined which was found to have a vestigial inner valve process (G. epione, which I have not examined, may have this characteristic as well). Out group taxa lacking a inner valve process are coded 0, while those coded c for the preceding character are coded c for this character as well.






31

3. For those taxa with character I state 2, the curve of the posterior side of the inner valve process in ventral view:

l=Concave (Fig. 37).

2=Convex (Fig. 108).

3=Both sides fairly straight, neither distinctly concave or convex(Figs. 89 & 87). The exact angle from which the male genitalia are viewed affects the apparent shape of the inner valve process. In ventral aspect, as the anterior end of the genitalia are pushed down (increasing the posterior aspect of the view) the posterior edge begins to look concave for both state I and state 2 taxa. The state differences appear to be due to differences in how the base of the inner valve process has been rotated. Within the genus Chlosyne both states 1 and 2 occur, in addition to an intermediate state which occurs in four taxa, which does not affect this analysis. One taxon (C. leanira and its subspecies) exhibits integrades between state 2 and the intermediate state, supporting the rotation hypothesis. However, Eurodryas desfontaini appears to have independently acquired a state 2 process due to a change in shape rather than rotation, as this taxa has a forked inner valve process with the dorsal fork is in the same position as other Euphydryiti with state 1.

Since Gnathotriche sodialis and G. exclamationis have the inner valve process hinged down, the frame of reference for these taxa is different. However, if I am correct that this variation is due to the base of the process being hinged down, these taxa would be state 1. To avoid making this additional assumption, I code these taxa as "?" along with G. mundina which has a vestigial inner valve process. The out group is coded the same as for the preceding character.






32


4. For those taxa with the projection of character 1 state 2, the shape of this projection:

1 =Curved with a posterior fork (the fork in E. desfontaini is reduced to a slight

inward extending point whereas it is strong in other taxa examined with this

state) (Figs. 3 & 6-14).

2=Curved and pointed (bluntly or sharply) with a smooth simple wall (Fig. 37).

3=Dorso-ventrally flattened with five short pointed projections on the anterior

side (Fig. 43).

4=Saw-like and dorso-ventrally flattened, wedge-shaped with outer side smooth

and the inner side with approximately 8-12 prominent teeth along much of the

surface. The teeth are asymmetrical between valves and vary from three

dimensional to flattened, and single or double in the same individual (Fig. 34).

5=Approximately proximal half arched dorsally inward and roughly perpendicular

to its lateral plane, distal half abruptly changes direction, arched dorsally and projecting posteriorly inward where it tapers to a point; a short, blunt, broad

projection on the anterior side at the corner where the projection abruptly changes

direction (Fig. 81).

6=Curved and pointed and dorso ventrally flattened, edge entire except several

small but conspicuous teeth on the dorsal /outer edge.

7=B3ase broad, then narrowing, mid section serrate on sides and of similar width,

tip abruptly narrowing and then tapering to a smooth sided point.

#8=Y-shaped and covered with many prominent teeth and setae. Out-group coding is identical to the preceding character.






33


Variation of a forked inner valve projection provides a rich source of additional characters. Considerable independent evidence from other characters indicates the inner valve projection being forked is a terminal derived state. Consequently, taxa lacking a forked inner valve projection are coded "0" for the next series of characters.

5. For those taxa with state I of the preceding character, extent of the fork of the inner valve projection:

I=Strongly forked (Figs. 7-14).

2=Slightly forked (Figs. 3 & 6).

6. For those taxa with character 4 state 1, characteristic shape of the distal edge of the dorsal fork of the inner valve projection:

I1=Narrowly rounded off (Figs. 7-11 & 14).

2=Sharply pointed (Figs. 12-13).

3=Very broadly rounded off (Figs. 3 & 6).

7. For those taxa with character 4 state 1, the occurrence of prominent teeth on one side of the distal end of the dorsal fork of the inner valve process:

P=Present (Figs. 7-1 1 & 14).

2=Absent (Figs. 3,6, 12-13).

8. For those taxa with state 1 of the preceding character, the orientation of the dentate side of the dorsal fork of the inner valve process:

I Posterior (Figs. 9-1 1 & 14).

2=Anterior (Figs. 7-8).

There is no independent evidence from other characters to suggest that the presence of teeth on the dorsal fork is a terminal derived state; however, this feature is limited to taxa






34

with a forked inner valve process for which there is much independent evidence of monophyly. Consequently, Euphydryiti lacking teeth on the dorsal fork of the inner valve process are coded "?", and taxa lacking a forked inner valve process are coded "0".

9. For those taxa with character 4 state 1, the orientation of the distal part of the dorsal fork:

= Inward toward the midline (Figs. 3,6,9,11 & 14).

2=Strongly ventral, weakly inward (Figs. 7 & 8).

3=Strongly inward, weakly ventral (Figs. 12 & 13).

4=Dorsally inward (Fig. 10).

10. For those taxa with character 4 state 1, the characteristic of the base of the dorsal fork:

l=Narrow (of similar width to the tip), elongate and not or very slightly tapering

(Figs. 7-8).

2=Broad (distinctly wider than the tip), short, and tapering (Figs. 3,6,10-12 & 14).

3=Narrow, tapering, and elongate (Fig. 13).

The orientation of the tip of the ventral fork falls into discrete states for those taxa with a forked inner valve projection. However, some of the taxa with a forked projection such as E. desfontaini and E. aurinia suggest the dorsal fork could be thought of as a process off of the ventral fork, and that the ventral fork is likely homologous to the inner valve process of other Melitaeini while the dorsal fork is derived. Consequently, it would be invalid to assign derived states of the ventral fork to the Euphydryiti group and code remaining taxa 0. Orientation of the tip of the inner valve process does not fall nicely into discrete states throughout the Melitaeini, however, this is no reason to discard this






35


information within the Euphydryiti. For the following character, I code taxa lacking a forked inner valve process as "?". 11. For those taxa with a forked inner valve process, the orientation of the tip of the ventral fork:

l=Curved ventrally outward (Figs. 7-8).

2=Curved ventrally inward (Figs. 3 & 9-14).

3=Strongly inward and slightly ventrally (Fig. 6).

The variation within taxa with a saw-like toothed inner valve process (character 4 state 4) can be further subdivided into discrete states. Independent character evidence suggests that this form of the inner valve process is a terminal derived state. 12. For those taxa with a saw-like inner valve process, the distal extent of the teeth on the process:

1=Distal half free of teeth.

2=Teeth continue distinctly distal of the distal half (Fig. 34). Taxa lacking a saw-like inner valve process are coded 0. 13. For those taxa with a inner valve process (character 1 state 2), the shape of distal end of this process:

I =Narrow and sharply or bluntly pointed (Fig. 1).

2=Flared out and serrate (Figs. 10-14).

Most state I taxa examined have a sharply pointed process; however, it is bluntly pointed in some derived Chlosyne. This variation is addressed in the data matrix for the Chlosyniti.





36

14. A distinct hollow ventral valve process at the posterior end of each valve, in part anterior to the part of the valve with extensive long hair-like setae:

0=Absent (Figs. 15,39,45,60,64.1-64.2 & 76).

1=Present (Figs. 1,36,42?,82,85,91,96,101,114,116,118 & 120).

Coding this character for some European taxa is problematic. The Cinclidia examined have a broad posterior part of the valve extended as a posterior process. At the base of this process, there is a ventral hollow projection which appears to be slightly posterior to the position of the other state 1 taxa. The most objective coding in my judgement is to code this taxon as state 1. However, for characters pertaining to the form of the ventral valve process, I code the taxa in Cinclidia as "?" to avoid weighting a potentially independently acquired character state twice. While the Mellicta examined all have a ventral process in the same position as the other state I taxa, this problem does not apply as the form of the process in Mellicta is different from all other state I taxa. The process at the posterior ventral comer of the valve in D. didyma and D. trivia is formed from the part of the valve with hair like setae, as opposed to a process anterior to the part of the valve with hair like setae, and is therefor coded as a state of character 18.

Most of the out group taxa lack any ventral valve process. However, a few taxa have a ventral valve process which is quite dissimilar to anything found in the Melitaeini in terms of structure and position. I code the out-group taxa collectively as "O&C." 15. For those taxa with a ventral valve process, the curve of the ventral valve process:

I=A posteriorly curved pointed process with a convex anterior side and concave

posterior side (Figs. 1,82,85,91,96,101,114,116,118 & 120).






37


2=A process of variable shape that's distal most point curves anteriorly

(sometimes only slightly) with the anterior wall of the distal most point concave

and the posterior wall convex (Fig. 36). Taxa lacking a ventral valve process are coded 0. 16. For those taxa with a ventral valve projection of character 14, the variation in shape of the process among state I and state 2 taxa for the preceding character can further be divided into discrete states:

I =Curved and pointed with the walls entire (or with one tooth-see the following

character) (Figs. 1,85,91,96,101,114,116,118 & 120).

2=Anterior edge smooth with 1-2 small teeth distally, posterior edge serrate with

several (approximately 6) distinct teeth (Fig. 82).

3=Veryjagged, anterior portion of process with 2-3 prominent teeth in a row

(some additional smaller teeth may be present in-between) with a similar distal

extent, posterior portion of process extends much farther distally as a curved,

pointed projection with 0-1 smaller teeth on its anterior edge and 1-2 smaller teeth

on its posterior edge (posterior to the longest prong there may be 3 or 0 short

teeth) (Fig. 36).

4=A deeply forked projection, with the longest (anterior) fork curving anteriorly

and the shorter (posterior) fork curving posteriorly.

5=A single broad process with a serrate anterior side and an entire posterior side. States I and 2 are for taxa which have state I for the preceding character, while states 3-6 are for taxa which have state 2 for the preceding character. The above coding avoids biasing the analysis for or against the homology between these two states of character 14,






38


and avoids the unnecessary question mark coding that would result from splitting this into two separate characters. Taxa with no ventral valve process are coded 0. Within the group of taxa with character state I for the preceding character, the variation can be further subdivided into discrete states. 17. For those taxa with character 6 state 1, the presence of a small additional distal tooth at the distal end of the process:

I =None.

2=A small tooth is present just proximal to the distal end on the anterior side

(Fig. 107).

3=A small tooth is present just proximal to the distal end on the posterior side

(Fig. 94).

Taxa with a ventral process but lacking state I for the preceding character are coded "?". Taxa lacking a ventral valve process are coded 0. 18. Posterior part of valve (area with many hair-like setae) extended to form a distinct pointed valve projection, located posterior to the ventral projection of character 5 (if present).

I =None (See also note for character 3 1

2=Posterior ventral side extends as an elongate, narrow, hollow process (Figs.

1,85,115,117,119 & 121).

3=A flattened non-hollow projection is present (Fig. 93).

4=Posterior dorsal side extended as an elongate, narrow, hollow process (Fig. 39).






39


5=Posterior side of valve with multiple hollow processes, with the most

prominent one at the posterior ventral corner, and the next most prominent at the

posterior dorsal corner (Fig. 45).

Some Nymphalini and Kallimini with valvae very uncharacteristic of Melitaeini have posterior valvae projections which appear highly dissimilar to any of those found in Melitaeini. Such taxa are coded as C for this character. 19. A broad area of the posterior end of the valvae, including the area with dense hairlike setae and the barer area ventral to it, expanded into a broad posterior valve projection:

0=None.

I=A broad dorsally curved projection with a convex ventral surface and a concave

dorsal surface (much shorter than the dorsal surface), bearing two prominent

ventral projections (Fig. 42).

20. Inner wall of ventral part of valvae turned roughly 90' and lobed, forming a pocket shaped roughly like a half circle with many spine-like setae on the posterior surface.

I =Absent.

2=Present (Figs. 2,3,5 & 15).

The shape and lobing of this structure appears somewhat different in Eurodryas desfontaini from the other taxa with this structure; however, a structure with identical spine-like setae is present on the same part of the valve.

I considered coding this variation as two characters, one for the spine-like setae and another for the lobe of the valvae; however, since the spines occur on the lobe and all






40


taxa with the lobe have spines, there would seem a good chance that these characters are not independent.

2 1. For those taxa with state 2 of the preceding character, the spine like setae are:

I =Numerous, dozens are present (Figs. 2,3 & 15).

2=Only five, especially large ones are present (Fig. 6).

Independent evidence from other characters suggests state 2 of the preceding character is a terminal derived state; therefore taxa lacking this state for the preceding character are coded 0.

22. The shape of the area containing dense setae on the inner posterior side of the valve:

Il=A roughly bean-shaped plateau with a flat surface (except for setae sockets).

2=The area is curved and not forming a plateau.

3=The area forms a plateau with a broad ventral side, widening slightly and then

tapering to almost a point on the dorsal side.

4=The area forms a plateau with a narrow, rounded ventral side and a broadly

rounded dorsal side.

All Melitaeini have an opening between where the plates of a valve fold around and overlap, always visible in ventral view. In most taxa, this opening ranges from not visible to barely visible along the ventral edge in lateral view. However, the position of the opening is notably different in Euphydryiti. 23. Position of the above ventral valve opening in lateral viewv:

I =Completely and clearly visible in lateral view dorsal to the ventral edge of the

valve (Fig. 15).






41


2=Partly visible to not visible in lateral view at the ventral edge of the valve (Figs.

42,45,114,116,120).

24. The form of the above ventral valve opening:

l=A distinct opening (Figs. 1,2,34,37,40,43,58,61,63,75,80,94,108,110,112 &

122).

2=A slit (Figs. 87,89 & 106).

25. The anterior extent of the above opening with respect to the ventral posterior edge of the vinculum:

I1=Terminates well posterior to the vinculumn (Fig. I & all other figures of ventral

male genitalia capsules).

2=Terminates anterior of the vinculumn (Fig. 80).

There is considerable variation in the anterior extent and length of the above valve opening and in the posterior extent of the vinculumn within the Melitaeini, most of which does not fall into discrete states. However, only taxa in the genus Atlantea have the vinculumn overlapping the anterior ends of the valve, including the ventral valve opening.

All Melitaeini have a patch of setae on the inner side of the valve (around the

vicinity of the anterior to posterior midpoint). The extent of folding of the valvae in this area determines whether the sockets for these setae are visible in ventral view. 26. The visibility of the sockets for the above setae in ventral view.

1 =Visible (Fig. 1 & all other figures of ventral male genitalia capsules).

2=Concealed (Figs. 99 & 106).

Note that some Phycioditi have the setae sockets just visible at the inner edges of the valve in ventral view. The out group is coded "?" for this character, because there is a






42


great diversity of valve structure and setae arrangements within the Nymphalini and Kallamini not found within the Melitaeini. 27. Posterior ventral inner wall of each valve extended inward in a flat, triangular projection with the anterior side approximately perpendicular to the midline of genitalia and the posterior side slanted inward.

1 =Absent.

2=Present (Figs 58 & 62-63).

28. Proximity of the valvae to each other in ventral and dorsal view:

0=Valvae distinctly separated (Fig. 1 and all other figures of male genitalia

capsules in ventral or dorsal aspect).

I =Valvae touching each other or nearly so (Figs. 34-35).

The only taxon that was somewhat ambiguous in scoring this character is Mellicta aurelia. In this taxon the valvae are still distinctly separated, but not by as much as other state 0 taxa.

29. The presence of a patch of short setae on the dorsal half of the posterior lateral sides of the valvae (distinct from the long setae at the posterior edge of the valvae):

0=Present (Figs. 15,36,39,42,45,64.2,76,82,96,101,114, & 176-192 (minus

178,179,183 7 187).

l=Absent (Figs. 60,64.1,116,118 & 120).

This character can be difficult to detect with a camera lucida setup, and it was inadvertently omitted from some of the first figures I produced of lateral genitalia capsules, including Figs. 91,178,179,183 & 187. These taxa actually do have the patch of






43


short setae characteristic of state 0, although these setae are not illustrated by these figures.

30. Posterior tapering of the entire valvae (This does not refer to a distinct posterior projection off of an otherwise non-tapering valve):

I1=Valvae do not extend and taper to a point posteriorly (All Figs. of lateral

genitalia capsules minus those indicated below)..

2=Valvae extend posteriorly and taper to a point, with the area of hair-like setae

located along this extension (Figs. 60,64.1,64.2 & 76).

#3=Valvae distinctly tapering but nowhere near to a point as in state 1, ends of

valvae broadly rounded.

3 1. For those taxa with state 2 of the preceding character, the characteristic of the posterior terminus of the valvae:

I1=Posterior end of valve with a short, curved, pointed tooth (Fig. 75).

2=Posterior end of valve with a long, narrow, tubular extension with a flared,

flatter end (extension arches dorsally, tip points ventrally) (Higgins 198 1, Fig. 275

pg. 207).

3=Long, tapering tubular extensions ending in a sharp, inward curved point (Figs.

58-64.2).

Taxa lacking state 2 of the preceding character do not have a distinct posterior valve terminus. Evidence suggests state 2 of the preceding character is a terminal derived state; however, taxa lacking this state are coded "'?". This is because some taxa without a valve that tapers to a point have a distinct posterior projection off of a non tapering valve (character 18). An alternative coding scheme could be to code the above three states as






44


three new states for character 18. An analysis with this alternative coding scheme did not affect the resulting consensus tree generated by PAUP. 32. In lateral view, the orientation of the valvae:

0=Much more posterior than ventral (Fig. 76).

1 =Much more ventral than posterior (All other figures of male genitalia capsules

in lateral aspect).

32.1. Outer lateral side of each valve with a groove, originating near the anterior-dorsal comer of the valve and slanting posterior ventrally across the valve for much of its length.

0=Absent.

I =Present (Figs. 60 & 64.1). Characters of the juxta:

A number of Melitaeini have various ridges or plateaus on the ventral surface of the juxta, while others have a predominantly smooth featureless juxta. The nymphalid out groups have the juxta features quite different from the Melitaeini. 33. Ventral surface of the juxta:

I =With a mid ventral ridge extending to the posterior edge of the juxta and

originating a variable distance anteriorly (Fig. 5).

2=Fairly smooth and lacking distinct ridges or plateaus (Figs.

4,34,40,43,61,63,75,80,87,89 & 94).

3=Narrow ridge tapering anterior to posterior along the midline, flared anteriorly

into a broad triangular plateau (Fig. 37).






45


4=A prominent raised plateau composed of a diamond-shaped posterior section,

and widening anteriorly into a broad triangular section with slightly concave

sides. The lateral sides of the plateau are steep and distinct while the anterior and

posterior sides more gradually slope dorsally to the level of the remainder of the

juxta (Figs. 1, 108 & 134-155) .

5=A distinct posteriorly directed triangular-shaped plateau, including the entire

anterior part of the juxta as the base of the triangle; the sides and tip are bordered

by steep dorsal slopes (Figs. 1 10 & 122).

6=A distinct raised plateau with the anterior side narrow and transversing the

width of the juxta, the middle greatly constricted with deeply concave sides

adjacent to steep dorsal slopes, and the posterior section broad and gently sloping

dorsally (Figs. 112-113).

7=Roof-shaped, the ventro-anterior face is triangular (pointed ventrally) with a distinct boundary, and each side of the posterior part is rectangular and slanted

ventrally inward, with the two sides meeting along the ventral midline (Fig. 99).

8=Smoothly curved anteriorly, roof-shaped with a sharp ventral keel in

approximately the posterior 4 (Fig. 94).

One of the taxa with state 5 has the apex of the plateau extended as a posteriorly directed spine. This variation is not coded, as only one taxon has this feature. The spine is dissimilar to that found in most Euphydryiti, which possess a posterior juxta projection composed of two parts close together.

34. For those taxa with state 1 of the preceding character, the anterior extent of the mid ventral juxta ridge:






46

l=Terminating distinctly before the anterior edge of the juxta (Fig. 5).

2=Extending to the anterior edge of the juxta.

All these taxa have the ridge keeled posteriorly, but those with the ridge extending anteriorly to the anterior margin have this part of the ridge smooth. This variation coincides exactly with the above character, thus it is not coded since it does not appear to constitute independent variation. Independent evidence from other characters suggests that state 1 of the preceding character is a terminal derived state, so taxa lacking this state are coded "0" for the above character. 35. The presence of dense, minute, granulose structures on the ventral midline and other portions of the ventral surface of the j uxta:

0=Absent.

l=Present (Fig. 5).

36. Ventral posterior midline ofjuxta terminating in two distinct, ventral-posteriorly projecting points (an anterior and posterior component). The two projections are very close together and appear to be a single projection unless viewed from the lateral side, where the separation can be detected.

I =Absent.

2=Present (Figs. 2 & 5).

37. For those taxa with state 2 of the preceding character, the relative sclerotization of the posterior projections of the juxta:

l=Far more heavily sclerotized than the remainder of the juxta, appearing almost

black.






47

2=Somewhat more heavily sclerotized than the remainder of the juxta, however

not nearly to the degree as state I and clearly not appearing almost black (Figs. 2

&5).

Since this character is scored by the sclerotization of the juxta relative to that of the projection in the same specimen, differences in how long different specimens were left in KOH should not prevent the ability to accurately score this character. However, the specimens scored for this character were placed in KOH from the same stock solution at the same time, and left in KOH at the same temperature (room temperature) and then dissected in succession (except for Hypodryas iduna). However, I found other dissected specimens left in KOH for differing amounts of time could still be coded the same way without ambiguity.

Taxa lacking character 18 state 2 are coded 0, with the exception of Eurodryas desfontaini. Many other independent characters support the monophyly of the Euphydryiti clade, and E. desfontaini is the only Euphydryiti examined which lacks the derived state of the preceding character. I avoid coding this taxon as 0 for the above character, because the absence of paired juxta projections is the only character suggesting E. desfontaini may be more primitive than the other Euphydryiti examined. Coding E. desfontaini as 0 rather than ? would produce an unwarranted bias. 38. For those taxa with paired posterior juxta projections, the lateral compression of these projections:

l=Distinctly laterally compressed (Fig. 5).

2=Not laterally compressed (Fig. 2).






48

Taxa lacking paired posterior juxta projections are coded the same as for the preceding character.

Characters of the saccus:

39. Forking of the saccus due to invaginated projections:

O=Unforked.

l=Forked (All figures of male genitalia capsules in dorsal and ventral aspects). 40. Further characterization of anterior-dorsal projections of the saccus formed from invaginations of the saccus:

1 =Two bilaterally symmetrical prominent projections, one on each side (Figs.

1,2,5,34,37,40,44,59,61,63,74,81,94,99,106,108,111,134-173).

2=Saccus tapering anteriorly and triangular, then forming a narrow extension

forked with two small extensions at the anterior most end (Figs. 86 & 89).

3=A single prominent projection to the right of the midline in ventral view

(Higgins 1981, Fig. 302 pg. 211).

4=Saccus extends in a centered rectangular projection, slightly forked at the

anterior end (Higgins 1981, Fig. 272 pg. 207).

5=A single prominent centered projection (Higgins 1981, Fig. 416 pg. 227).

6=Similar to 5, except the central projection is slightly forked at the anterior end

(Higgins 1981, Fig. 475 pg. 236).

7=Saccus roughly triangular with an abrupt constriction at about one-half its

length, slightly forked at anterior most end (Fig. 112).

#8=Entire saccus projects anteriorly in an elongate, hollow, extension, as opposed

to any distinct invaginated projections from the saccus as in states 1-7.






49


#9=Entire saccus projects anteriorly in a broad, triangular extension.

#A=Saccus broad and emarginate anteriorly with no extensions.

These character states are subdivisions of state 0 and state I of the preceding character. The above approach as opposed to splitting this variation up into two characters (one for state 0 taxa and one for state 1 taxa of the preceding character) avoids unnecessary "?" coding.

There is great variation among the taxa with paired saccular projections (state 1), but this variation was not found to fall into discrete states. In addition, exact shape of the saccus was found to exhibit considerable intraspecific variation within the Chlosyniti (Figs. 167&169, & 225-260). The forms of character states 2 and 7 were found to be consistent, although there is some interspecific variation within state 2. States 3-6 are particular to Phycioditi, and saccus shapes of Phycioditi are illustrated in Higgins (1981). Also, note some Phyciodes have a single centered saccular projection as do some Nymphalid out group genera. While all are coded state 5, the shape and length of these projections between the Phycioditi state 5 taxa and out group state 5 taxa varies greatly. 41. For taxa with a pair of saccular projections, the presence of a broad, flat plate of lightly sclerotized tissue between the projections:

1 =Absent.

2=Present (Figs. 58-59,61-62 & 63-64.1).

Overwhelming evidence from other characters indicates a forked saccus is a derived state of a binary character for the Melitaeini and those Melitaeini with a non-forked saccus underwent a reversal. Since the above character can not be coded for taxa lacking a






50

forked saccus, I code the Melitaeini without a forked saccus as "?" and the out group as

0.

42. Given state 2 of the preceding character, the points of attachment of the connecting plate of tissue between the saccular fork:

1 =Anterior margin connects across the distal ends of the fork (Figs 64 & 64.1).

2=Anterior margin connects across the vicinity of the midpoint of the fork (Figs.

59 & 61).

Independent evidence from other characters suggests state 2 of the preceding character is a terminal derived state. Taxa lacking this state are coded "0" for the above character. 43. Extent of the saccus at the middle of the ventral anterior side of the genitalia:

I =Prominent/well developed.

2=Reduced to a very narrow bridge (Fig. 2).

44. Exposure of openings to the hollow projection(s) of the saccus:

1 =Openings concealed by vinculum (Figs. 1,5,34,37,44,43,58,61,63,80,

87,89,94,99,108,110,134-155).

2=Vinculum anterior of openings leaving them exposed (Figs. 75,112 & 122).

#3=Vinculum slightly posterior or even with saccular projections posterior opening, however the vinculum is orientated so far ventral that the saccular

projection's openings appear to be exposed. Characters of the tegumen/uncus:

45. Development of the uncus:

O=Tegumen expanded into a well developed uncus (Figs. 35,59,62,74,86,90,95,

99,106,109,111,113,123,156-173).






51


1=Tegumen reduced to a thin, simple, sclerotized bridge between the valvae

(Figs. 1,4,16-23,25,38,41.

Within the state 0 taxa (nymphalid out groups and many Melitaeini) there is considerable variation in the shape and size of the well developed uncus, which I was not able to reliably code into discrete states. There is relatively little variation in the simple tegumen of state I taxa, other than for its exact width. While Euphydryiti have prominent posterior projections on the tegumen, it is otherwise a simple bridge characteristic of state

1.

There are some taxa which were somewhat ambiguous in scoring this character, in particular Gnathotriche mundina and Atlantea. I code these taxa as "?". Higgins (1981) illustrates a well developed conical tegumen for G. mundina, however, the sole specimen of this taxon I was able to examine has the posterior border of the tegumen straight across (Fig. 64) with the juxta shape more reminiscent of Higgins (1981) illustration of the tegumen. Since I have found tegumen shape to exhibit very little intraspecific variation within the Melitaeini (including in G. exclamationis), and since Higgins greatly compressed the genitalia he examined by placing them on slides, I hypthesize that the tegumen illustration for G. mundina in Higgins (1981) is in error. 46. Each outer side of tegumen with a very heavily sclerotized plate containing a raised, pointed lateral ridge:

1 =Absent.

2=Present (Figs. 100-102).

47. Clusters of small, heavily sclerotized, pointed teeth covering each posterior corner of the uncus:






52


1 =Absent.

2=Present (Higgins 1981, Fig. 475 pg. 236).

48. A heavily sclerotized convexly curved claw on each posterior side of the uncus:

1 =None.

2=An entire claw is present (Figs. 74 & 76).

3=A dentate claw is present (Higgins 1981, Fig. 473 pg. 235).

An alternative coding scheme would be to code for presence or absence of a claw, and then have a separate character for whether the claw is smooth or dentate. However, the structure of the dentate and entire claws and their exact position is quite different, so the above coding seemed most appropriate. It also may be appropriate to combine the two preceding characters as a single multistate character. These character states are particular to the Phycioditi, and not informative other than for relationships within the Phycioditi, and consequently extraneous to the goals of this analysis. Examination of more Phycioditi taxa would give more insight into the range of variation occurring within Phycioditi and the best way to code it.

49. Hollow, paired, pointed posterior projections formed from invaginations of the inner wall of the tegumen (some of the out-group states refer to single projections-only paired projections occur in the Melitaeini):

O=None.

1 =Projections divide the tegumen into a deep fork; they are connected by only a

very thin bridge of tegumen between them and anteriorly (Figs. 4,16-23 & 25).






53


2=Projections do not divide the tegumen, but extend from its sides and are connected by a longer and much wider (relative to state 1) area of tegumen on their anterior side (Figs 35,80,81,86,90 & 95). 3=Ventral lateral inner side of tegumen extended as long, narrow, hollow invaginations covered extensively with many small teeth on the inner surface (Fig. 64).

4=Ventro-lateral outer sides of tegumen extended as minute triangular pointed projections (Fig. 44).

#5=A large, deeply forked posterior projection arising from the center of the tegumen.

#6=Paired prominent projections extend from the middle of the tegumen (versus the sides in state 2), one on each side of the dorsal midline with very little space between them, a wide area of tegumen occurs anterior to these projections. #7=Entire tegumen extends posteriorly in an elongate pointed posterior projection.

#8=Entire posterior tegumen deeply forked, occurring as two elongate tapering pointed projections.

#9=Entire posterior tegumen extends as an elongate (concave in posterior view) spatulate projection.

#A=A single elongate tapering posterior pointed projection at the midline. #B=Entire tegumen extends posteriorly as an elongate tapering projection but with a broadly rounded terminus.






54


#D=A single elongate centered projection widened posteriorly with two lateral

pointed extensions.

Due to the notably different structure/characteristics and position between the invaginations of the above in group states, I did not code for the presence or absence of paired invaginated projections of the tegumen. 50. For those taxa with character 49 state 1, the shape of the tegumen projections:

l=Elongate, narrow, and curving ventrally to a point distally (Figs. 23-24).

2=Short, broad, and triangular with the outer side slanted and the inner side nearly

vertical (Fig. 21).

3=Prominent and triangular with a broad base and a narrow straight tip (Fig. 16).

4=Base broadly triangular but the distal portion narrower and slanting inward

toward the midline; the distal portion is blunt and more finger-like than triangular

(Fig. 19).

5=Long and pointed with the outer side concave and the inner side curving

outward distally while being more vertical/concave basally (Figs. 18,20,22 & 25).

6=Prominent, proximal approximately 2 broad and triangular, distal 2 narrower

and triangular curving sharply outward at the distal extremity (Fig. 4).

7=Short and broad, basal 2 somewhat rectangular, distal 2 projects from lateral

side of base, triangular, and with the inner side concave and outer side fairly

straight (Fig. 17).

Independent evidence from other characters suggests state 1 of the preceding character is a terminal derived state. Consequently, taxa lacking this state are coded "0".






55


5 1. For those taxa with character 49 state 1, the presence of an additional pointed projection below the apex on the inner sides of the tegumen projections:

I =Ab sent.

2=Present (Figs. 19 & 2 1).

52. For those taxa with character 49 state 2, further characterization of the projections off of the tegumen:

I=About as long as length of tegumen anterior to the projections or slightly

shorter, for most of length at most slightly tapering; however, apically projections

curve inward a short distance and taper abruptly to a point (Fig. 35).

2=About as long as length of tegumen anterior to the projections or slightly

shorter, outer side convex and inner side straighter and slightly concave, apex

pointed (Fig. 2).

3=Small, triangular, and much shorter than the length of the tegumen anterior to

the projections, apex pointed (Figs. 86 & 90).

4=Weakly triangular and much shorter than the length of the tegumen anterior to

the projections, outer side somewhat convex, inner side slightly convex

proximally and straight distally, apex blunt (Fig. 80).

Independent evidence from other characters suggests state 2 of the third preceding character independently arose twice; however, there is no overlap of the above states between the two clades where this state arose. Consequently, taxa lacking state 2 of the third preceding character can be coded 0 without risk of doubly weighting an independent acquisition.






56


52.1. For those taxa with character 49 state 2, the presence of setae on the tegumen projections:

I =Present.

2=Absent.

As noted, independent evidence from other characters suggests character 49 state 2 independently arose as a terminal derived state in two lineages, and taxa in each lineage have the state of setae absent. To code them both as such has the danger of weighting a likely independent acquisition twice. However, this character is nonetheless informative within the lineage that includes taxa with and without setae on the projections (Poladryas, Atlantea and Higginsius), and the character should not be discarded. I code these taxa for the above character, those Mellicta with character 49 state 2 and taxa with other tegumen projections as "?", and other taxa as "0". All Euphydryiti have minute granulose patches on their tegumen projections. These are the only taxa with these granulose patches on the tegumen, and there seems no reason to assume the presence of these patches is dependent on having a particular type of tegumen projection. In some taxa these patches are just detectable at 50X magnification with good lighting, while in others they are conspicuous, with some apical granulose patches enlarged into teeth-like patches.

53. The presence of minute granulose patches on the tegumen:

0=Absent.

I =Present.

54. For those taxa with state I for the preceding character, the apparency of these patches at 50X magnification with good lighting:






57

l=Very conspicuous (Figs. 19, 23 & 24).

2=Obscure (generally not detectable with the camera lucida on).

Independent evidence from other characters suggests state 1 of the preceding character is a terminal derived state, and taxa lacking this state are coded "40". Characters of the Phallus:

55. Laterally flattened keel on the anterior and anterior-ventral end of the phallus:

1=Present (Figs. 28,48-50,57,66-67,72,78,84,88,92,97-98,103-105,126-133,213253).

#2=Absent.

#3=Absent, but a flared wedge-shaped hollow structure is present.

#4=Absent, but a hollow, open, ventral scoop-shaped structure is present.

56. Orientation of the posterior opening of the phallus relative to the anterior opening:

1 =Posterior opening is ventral and approximately 180 degrees around from dorsal

anterior opening (Figs.26-29 and all other phallus figures).

2=Anterior opening is lateral and approximately 90 degrees around to the left (in

ventral view) with respect to the ventral dorsal opening (Figs. 30-33).

#3=Anterior opening is neither distinctly dorsal or ventral, posterior opening is

ventral.

#4=Posterior and anterior openings are both dorsal.

All Melitaeini have a pattern formed by areas of lightly and more heavily sclerotized tissue on the ventral posterior surface of the phallus, posterior to the supersensory membrane. Variation can be coded with respect to the pattern of the lightly sclerotized area, and whether it extends anteriorly beyond the supersensory membrane.






58

57. Sclerotization pattern of the ventral side of the phallus posterior to the supersensory membrane:

I =Elaborate spear-shaped design (Figs. 1 & 213-233).

2=Posteriorly lightly sclerotized all the way across, anterior to this area the darkly

sclerotized area begins along the sides, widening until extending all the way

across forming a concave border (Figs. 26,30,49-51,53 & 104).

3=A narrow band of dark sclerotization on each side, and lightly sclerotized in the

middle (Figs. 65,68,73,79,88,92 & 98-99).

4=Lightly sclerotized area ovoid, with darkly sclerotized area reduced to an

extremely narrow band on each side at one point, and expanding anterior and

posterior to this point (Figs. 127,129,131,133).

5=Similar to state 2 except for a narrow, posteriorly tapering triangular section of

more heavily sclerotized tissue originating at the supersensory membrane, and

the border of between the lightly and darkly sclerotized areas is convex (Fig. 28).

6=Like state 3 except narrow inward anterior slanting extensions of sclerotized tissue from the sides (just anterior to the plate attached to the vesica) (Fig. 97).

Several taxa, Euphydryas phaeton, E. editha, and the E. anicia complex have the phallus twisted 90 degrees such that the lateral side is equivalent to the ventral side for other taxa. This information is coded as a separate character, and assigning a new derived state for the ventral sclerotization pattern would weight the same feature twice. The lateral side of these taxa is typical of state 2, and they are coded as such. The same applies to character 60 below.






59

58. For those taxa with state 2 for the preceding character, the point of the posterior extent of the more heavily sclerotized area:

I =Much closer to the posterior opening than the supersensory membrane (Figs.

49-51 & 53).

2=Closer to the supersensory membrane than the posterior opening (Figs. 26,30 &

104).

Taxa lacking state 2 for the preceding character are coded "?" 59. The anterior extent of the lightly sclerotized area on the ventral phallus surface:

0=Reaching (and may surpass) the supersensory membrane (Figs. Figs. 26,30,4951,53,65,68,73,79,88,92,97-98,104,127,213-214,216-224).

I =Terminating distinctly posterior to the supersensory membrane, but closer to the supersensory membrane than the posterior phallus opening (Figs. 129,131,

133, 225-227,230,232-233).

2=Terminating distinctly posterior to the supersensory membrane, but closer to

the posterior phallus opening than the supersensory membrane (Fig. 53).

This character varies within the genus Chlosyne. I code Chlosyne as "?" in this analysis but code Chlosyne taxa individually for this character in the Chlosyniti data matrix. 60. Sclerotization pattern of the dorsal surface of the phallus:

l=The longitudinal middle is more heavily sclerotized relative to the sides, and

this sclerotized area tapers anterior to posterior (within this area the midline may

appear lighter than the sides) (Figs.29 & 31).

2=Uniformly sclerotized (the sides may appear darker in dorsal view due to their

proximity to sclerotized area of the phallus on the ventral side) (Figs. 69 & 77).






60


61. A hollow tube open on both ends on the distal posterior end of the phallus.

I =Absent.

2=Present (Figs. 83,88,92,97 & 98).

62. Posterior dorsal end of phallus distal to the posterior ventral side: I =Entire scierotized area on posterior dorsal side of phallus in the form of an elongate

triangular tapering area extending distinctly distal to the ventral side of the phallus

opening (Figs. 29-30).

2=Not extending distinctly distal to the posterior ventral end (Figs. 49-5 7 & 77-79).

3=With an elongate tapering extension with the sides slightly concave (Figs 88,92 & 9798).

4=A triangular extension with a squared off tip (Figs. 65 & 7 1).

5=A broad extension with a convex base (Figs. 104,127,129,131,133 & 213-233).

63. For those taxa with state 5 of the preceding character, the presence of a triangular projection at the distal posterior edge of the phallus extension:

I1=Present (Fig. 104).

2=Absent (Figs. 127,129,131,133 & 213-233).

Independent evidence from other characters suggests state 5 of the preceding character is a terminal derived state with one independent acquisition in G. exciarnationis, therefor taxa lacking this state are coded "O"and G. exclamnationis is coded "?". 64. For those taxa with an extension of the dorsal posterior side of the phallus distal to the ventral posterior side, the sclerotization pattern on the ventral side of this extension:

I1=Sides and middle darkest (the sides may be as dark or darker than the middle) with the sides and middle separated by an area of lightly sclerotized tissue of variable width (best

illustrated in Figs. 1 & 228, difficult to see in some figures of state I taxa because the






61

lightly sclerotized area between the sides and middle is narrow and requires high

magnification to illustrate).

2=A triangle within a triangle pattern, with the arrow head pattern in-between lightly

sclerotized (Fig. 104).

3=Uniformly sclerotized except may appear darker on the sides where they are folded

over (Figs. 26,28,30,65,69,71,88,92 & 97-98).

Taxa with no extension on the dorsal posterior side of the phallus are coded "?". The following characteristics associated with the vesica were examined by everting the vesica for specimens of all included species with the exception of some Phycioditi. Within the Phycioditi, which are notoriously difficult to evert the vesica, I found the character states of these characters to be invariant, and for those taxa where I did not evert the vesica I still was able to score the state for every character except for the presence of teeth on the ventral surface of the everted vesica. I was also unable to evert the vesica on some of the out-group taxa. Failure to evert some of the out group taxa was problematic for some of the following characters but not others, as for some characters it was clear out group taxa do not contain the pertinent structures (such as plates on the posterior end of the phallus attached to the vesica) and consequently lack any forms of these structures. For the analysis where Anartia, Junonia, and Catacroptera were used as out groups, the above complications do not apply because the vesicas of all these taxa were everted.

65. The presence of minute teeth on the ventral surface of the everted vesica:

I=Present (Figs. 52 & 54-57).

2=Absent.






62


State 1 is not to be confused with a sculpturing of small bumps (as opposed to distinct pointed teeth) which occur in the Euphydryiti. None of the out group taxa examined have teeth on the vesica; however, since some outgroup taxa were not everted the out group is coded "?" for this character.

The presence of the above minute teeth appears to be a unique synapomorphy for European Melitaeiti; however, the teeth may be absent in D. trivia. I was unable to confidently assign a state to D. trivia even under maximum power-this taxon may have extremely minute teeth present but I could not be sure. If D. trivia does have these teeth, they are greatly reduced relative to any of the other state I taxa examined. At present, I code D. trivia as "?". Other taxa where this character is coded "?" are taxa for which I did not evert the vesica, but in all likelihood would have lacked teeth on the ventral everted vesica.

66. Proximal part of the everted vesica encircled by dense, minute, granulose patches (conspicuous at 50X magnification with bright lighting).

O=Absent.

l=Present (Figs. 27-28 & 33).

The out group is coded "?" for the same reasons as for the preceding character, although none of the vesicas examined for out group taxa were found to possess this feature. 67. Given state I for the preceding character, the length of the band of granulose patches encircling the vesica relative to its width:

1 =Distinctly wider than long (Fig. 28).

2=Distinctly longer than wide (Figs. 27 & 33).






63


Taxa lacking state I for the preceding character are coded "?", since the failure to evert the vesica on some out group taxa precludes reliably concluding that the presence of the granulose patches is a terminal derived state. 68. The presence of a pair of disjunct sclerotized patches containing minute teeth on the ventral surface of the everted vesica:

O=Absent.

1=Present (Fig. 28).

For the two taxa examined which possess state 1, in E. desfontaini the sclerotized patches are shorter and more distal, while in E. aurinia they are longer and more proximal. No out group taxa were found to possess state 1; however, since some out group taxa did not have the vesica everted the out group is coded "?". 69. Sclerotization of ventral part of vesica, viewed when everted:

1=None, this area is clear and membranous, staining blue if chlorozol black is

applied.

2=Present, this area does not stain blue.

#3=Many thin, longitudinal, lightly sclerotized bands proximally on vesica for a

length exceeding /2 the length of the phallus.

For those taxa designated as not having been examined with the vesica everted (also not stained), this character was examined at high power by placing the phallus in a glass petri dish with a bright light source underneath. With this technique, I found I was able to score state 1 and state 2 specimens prior to everting the vesica. 70. A sclerotized plate on the vesica (best seen when vesica everted) attached to each ventral lateral side of the phallus.






64


1=Absent (Figs. 27-28 & 33).

2=Present (Figs. 48,54-56,66,69,78,83-84,105,112 & 211-212).

71. For those taxa with state 2 of the preceding character, a second plate, bearing teeth, attached to the vesica and connected to the distal end or distal-ventral comer of the plate attached to the phallus :

l=Absent (Figs. 27-28,33,54 & 66).

2=Present (Figs. 48,52,55-56,69,78,83-84,105,112 & 211-212).

The structure of the plates attached to the vesica in Gnathotriche sodialis is very unusual. A single teeth-bearing plate is attached to the phallus over a broad area, including the entire posterior sclerotized end of the phallus from the point where the phallus begins to constrict. There is no clear separation of two plates in contrast to the other Melitaeinine taxa with teeth-bearing plates. The unique plate of G. sodialis is broad and triangular with the posterior side perpendicular to the plane of the phallus and the ventral side appearing almost as a continuation of the phallus. Four teeth are present, with three running approximately perpendicular to the posterior edge of the phallus and a small fourth tooth (on the right side only, based on the one specimen examined) ventroposterior to the second posterior-most tooth and near the posterior edge of the plate. I do not know if the single teeth-bearing plate attached to the vesica is homologous to the distal or proximal plates of other Melitaeini, or if the plates are so fused together that they are no longer individually detectable. If the teeth bearing plate is homologous to the distal plate of other Melitaeini, G. sodialis actually underwent a loss of the basal plate and subsequent expansion of the distal plate to reconnect with the posterior end of the phallus. This would represent an unique reversal within the Melitaeini, and an






65


autapomorphic state for the form of the distal plate. If the plate in G. sodialis is homologous to the proximal plate of other Melitaeini, then it would share a reversal with G. mundina and have an autapomorphic state for the form of the distal plate. If the apparent single plate in G. sodialis is actually the extreme fusion of both plates, then G. sodialis has an autapomorphic state for the distal plate, and does not share a reversal with G. mundina. Since these three possible scenarios have considerably different implications for coding, and I have no objective basis for favoring one of them, I code G. sodialis as "?" for the next three characters. 72. For taxa with state 2 of the preceding character, the relative size of the teeth on the distal plate.

l=Minute, all of a similar size range (Figs. 52 & 55-56).

2=Prominent, some teeth distinctly larger than others (relative to state 1) (Figs.

70,78,83,84,97,105,125 & 211-212).

While the size difference between state I and 2 is quantitative, the difference is sufficiently obvious that quantification is not necessary to assign two distinct states (see the above referenced figures). However, note that within state 2 taxa, larger species tend to have larger teeth than smaller ones, but none were found to be ambiguous in coding as distinct from state 1.

Didymaeformia didyma lacks a second plate on the vesica but has minute teeth

(identical to other state I taxa) on the vesica in the area where the second plate would be. This taxon is coded as state 1. Independent evidence from other characters suggests the second teeth bearing plate may have underwent reversal. All taxa with the distal teeth bearing plate have a proximal plate to which it is attached (although this could not be






66


interpreted in one taxon, G. sodialis). Taxa with a proximal plate but lacking the second teeth bearing plate are coded "?", while taxa lacking either plate are coded "0". 73. For those taxa with state 2 of the preceding character, the arrangement of teeth on the teeth-bearing plate attached to the vesica:

l=Teeth in one row along an arch (Figs. 105,125 & 211-212).

2=Teeth are not confined to one row (Figs. 52,55-56,70,78,83-84 & 97).

74. For those taxa with a plate bearing teeth attached to the vesica, characteristics of this plate (these descriptions are based on a view with the vesica everted):

I=A narrow, heavily sclerotized crescent curving posteriorly and bearing teeth

(usually 6-10) projecting posteriorly along the crescent (ridges of the anterior base

of the teeth occur on the outer side of the crescent) (Fig. 211).

2=As in state I except dorsally the plate extends as an anteriorly curving arc (this state occurs in only 3 taxa of the genus Chlosyne; all other Chlosyne have state 1)

(Fig. 212).

3=Base rectangular, on posterior end a ventrally projecting, anteriorly curved, and

heavily sclerotized crescent bearing teeth as in state 1 (Fig. 105).

4=Elongate and more lightly sclerotized relative to the basal plate with an

outward arch at the very base, covered with many minute teeth on surface and

edges (Figs. 52 & 55-56).

5=Narrow crescent with prominent teeth on its surface, six teeth long anterior to

posterior and no more than three teeth wide (Fig. 70).






67


6=Arched plate with prominent sclerotized teeth spread over outer surface (much

less elongate and more heavily sclerotized than state 4, with many fewer but

larger teeth) (Figs. 78,83-84 & 97).

Taxa lacking a pair of plates attached to the vesica and posterior phallus are coded "0". Taxa with such plates but lacking a second teeth bearing plate are coded "?". 75. A laterally flattened, sclerotized plate attached to the very posterior edge of the dorsal midline of the phallus, and attached to the vesica for the remainder of its ventral edge:

I =Absent.

2=Present (Fig. 78).

This appears to be what Higgins (1981) referred to as an "ostium keel." 76. For those taxa with state 2 of the preceding character, basic shape of this plate (viewed when vesica everted):

I =Triangular dorsal projection on proximal side (not illustrated).

2=Proximal side entire (Fig. 78).

State 2 of the preceding character occurs only in the Phycioditi, but there is no evidence to conclude it is a terminal derived state. Consequently, Phycioditi lacking state 2 of the preceding character are coded "?". However, as independent characters provide clear evidence the Phycioditi are a monophyletic group, other Melitaeini taxa can be coded

14011.

77. Posterior dorsal side of phallus "hinged" and changes its orientation from posterior to dorsal posterior when the vesica is everted:

0=This feature absent.






68


l=This feature present (Fig. 105).

78. The dorsal side of the everted vesica with a prominent, three dimensional sclerotized structure attached to the dorsal posterior edge of the phallus. The structure is entire with an anteriorly directed lobe proximally and elongate and tapering distally, extending well past the teeth-bearing plates of the everted vesica:

0=Absent.

1 =Present (Figs. 47-48 & 55).


Female genitalia

Characters of the Corpus Bursae:

79. The presence of a sclerotized plate on the ventral posterior side of the corpus bursae.

l=Present (Figs. 261-265,267-272,274,276-284 & 286-292).

#2=Absent.

Note that in Mellicta asteria the plate is very lightly sclerotized relative to other state I taxa, and may require higher magnification and more light to see. Since I only examined one specimen of this taxon, this feature may not be consistent. 80. For those taxa with character 79 state 1, extensions of plate on ventral posterior side of corpus bursae:

I =Plate extends anteriorly in two, narrow extensions, one on each side of the

plate (Figs. 261-265,278-284 & 286-292).

2=Plate lacks a pair of anterior extensions (Figs. 267,269-272,274 & 276).

The extensions are atypically short in Mellicta alatauica; however, the posterior arch of the plate is very characteristic of other Mellicta species examined.






69


Independent evidence suggests the presence of a plate on the ventral posterior

corpus bursae is a terminal derived state for all Melitaeini, and the out group is coded "0" for this and the next set of characters. 81. For those taxa with character 79 state 1, the posterior curve of the corpus bursae plate at the plates anterior end:

l=Anterior edge not curved posteriorly.

2=At least part of anterior edge distinctly curved posteriorly (Figs. 269-272).

82. For those taxa with character 79 state 1, the separation of plate on ventral posterior side of corpus bursae:

1 =Plate distinctly separated by membranous tissue along ventral midline (Fig.

261).

2=Plate with continuous sclerotization along ventral midline (Figs. 262-265,267272,274,276,278,279-284 & 286-292).

83. For those taxa with character 79 state 1, a raised sclerotized ridge (highest in middle) along midline of dorsal (inner) surface of the sclerotized plate on the ventral posterior side of the corpus bursae.

I =Absent.

2=Present (Fig. 269).

84. For those taxa with character 79 state 1, the presence of many small sclerotized teeth on the inner surface of the plate on the ventral corpus bursae:

1 =Present.

2=Absent.






70


The size of these teeth are too small to effectively illustrate with camera lucida drawings. although in some of the larger specimens I have illustrated a jagged edge along parts of the plate folded such that the teeth are visible along the edge (Fig. 261, for example). Taxa lacking character 42 state 1 are coded 0. 85. For those taxa with character 80 state 2, the shape of the plate on the ventral corpus bursae:

l=Very long and narrow, well over 5 times as long as the width of the anterior

side, posterior side widest and slightly wider than anterior side (Fig. 267).

2=Rectangular and wider than long, posterior side about as wide as anterior side

(Fig. 276).

3=Stout and bulged, at most 3 times as long as wide (usually less) (Figs. 269272).

4=Wider than long, anterior side about 1 .5-2X the length of the posterior side,

posterior edge irregular, sides widen posterior to anterior in a concave curve,

anterior edge extends along the midline (Fig. 274).

Independent evidence does not suggest state 2 of the fourth preceding character is a terminal derived state, so taxa lacking this state are coded "?" Out group taxa (with no plate on the ventral corpus bursae) are coded "0".

There is considerable variation within state 3, a state particular to the Phycioditi, which may potentially provide useful information for a detailed phylogenetic study of the Phycioditi.

86. For those taxa with character 80 state 1, the extent of the sclerotized plate on the ventral posterior side of corpus bursae before it diverges posteriorly:






71


1 =The unforked posterior portion of the plate extends for notably less than /2 the

length of the plate (Figs. 261,265,278-284 & 286-290)).

2=The unforked posterior portion of the plate extends for nearly 2 the length of

the plate or more (Figs. 262-264).

Note that Didymaeformia trivia has the plate a single nondivergent piece for nearly its entire length, before forking only slightly at the anterior most end.

Outgroup taxa are coded "0" and in group taxa lacking character 80 state I are coded "?".

The next four characters refer to aspects of variation in the signa. Melitaeinine signa are composed of aggregations of small teeth on the inner side of the corpus bursae. Some out group taxa have signa while others do not, and none of those that do have them in the same patterns found in the Melitaeini. The presence of signa teeth is obviously not a terminal derived state, so taxa lacking signa must be coded as "?" for characters pertaining to signa variation.

87. Aggregations of small teeth on inner surface of corpus bursae:

I =Absent.

2=Present.

88. For those taxa with character 88 state 2, the magnification required for the teeth of the signa to be readily apparent with a fiber optic light source:

l=Readily apparent in a spot plate at 25X1.5 magnification.

2=Extremely minute and barely if at all detectable in a spot plate at 25X1.5

magnification.






72

The signa teeth are easier to detect, and detectable at a lower magnification, when the genitalia are placed in a glass petri dish with water and a very bright light source from underneath. There were some cases where scoring this character within the Phycioditi was somewhat ambiguous, and examination of additional Phycioditi may warrant the conclusion that this character should not be used in a phylogenetic study of that group. 89. For those taxa with character 88 state 2, the arrangement of the aggregations of small teeth on the inner surface of the corpus bursae:

l=Continous band (of similar width) of teeth encircles corpus bursae (Fig. 261).

2=Distinct signa patches of teeth present on lateral sides, but connected by a narrower bridge of teeth on the ventral side (in Melitaea cinxia and Mellicta

britomartis the bridge of teeth is broken at the ventral midline) (Figs. 262-265).

3=Teeth confined to two distinct patches on the lateral sides of the corpus bursae

(Figs. 269-270,278,280,282,284,286,287,290 & 291).

4=A continuous band of teeth which is widest on the lateral sides, narrow across

the dorsal side, and relatively wider (than dorsally) across the ventral side (Fig.

267).

5=Nearly continuous band of irregular width encircles the corpus bursae, except for a narrow strip along the ventral midline which is nearly devoid of teeth (Figs.

274-276).

#6=The teeth densely cover the entire corpus bursae anterior of the ductus bursae. 90. For those taxa with character 88 state 2, the relative sclerotization of the teeth on the corpus bursae:






73


0=Appearing non-sclerotized to very lightly sclerotized (Figs. 261-265,267270,274,276,278,280 & 282).

1 =Prominently sclerotized (Figs. 284,286 & 289-291).

The amount of time the preparation is left in KOH has little effect on the ability to score this character.

91. Corpus bursae with a distinct protruding bulge on each anterior lateral side (signa patches of teeth cover the surface of the bulge in taxa which possess this character):

I =Absent.

2=Present (Fig. 270).

92. Ventral corpus bursae anterior to the ventral plate relative to the dorsal side.

O=Bulged out no more than 3X the distance on the ventral side relative to the

dorsal side (Figs. 268 & 291-292).

1 =Bulged out over I OX the distance on the ventral side relative to the dorsal side

(Fig. 265).

Note that this character must be scored in lateral view. Ductus Bursae:

93. A distinct sclerotized plate (originating at the posterior end of ductus bursae at the junction with the ventral genital opening) on the ventral ductus bursae separate from and posterior to the plate on the ventral corpus bursae (the plate on the corpus bursae would not have to be present to score this character due to the difference in position of the two plates).

l=Present (Fig. 261).

2=Absent.






74


Ostium Bursae:

94. A distinct sclerotized tube (ostium bursae) extending distal of the plates of lamella postvaginallis and lamella antevaginallis.

1 =Absent.

2=Present (Figs. 269-270 & 272).

All Phycioditi have state I and no other Melitaeini do. There is considerable variation in the shape of the ostium bursae within the Phycioditi, which may provide useful phylogenetic information for a phylogenetic study of that group. Lamellae and associated structures:

95. A membranous (stains blue in chlorozol black) pouch with many longitudinal wrinkle lines on its ventral surface, connected from the posterior lateral sides of the lamella postvaginallis around the anterior end of the lamella antevaginallis, and extending posteriorly over the genital opening and ostium bursae.

1 =Absent.

2=Present (Figs. 269-270 & 272).

96. Ventral lateral sides of lamella antevaginallis distinctly more sclerotized than a membranous-lightly sclerotized area in-between, giving the appearance extensions.

1 =This feature absent, sclerotization of the edge of the of lamella antevaginallis

fairly uniform.

2=This feature present (Figs. 284-291).

97. A raised posteriorly orientated hollow and compressed ridge on the lamella antevaginallis forming roughly a half circle around the ventral side of the corpus bursae opening:






75


0=Absent.

1=Present (Figs. 279-282).

98. For those taxa with state 1 of the preceding character, the form of the ridge on the ventral side of the corpus bursae opening:

1=Of similar width throughout, with the sides at most slightly produced over the

ventral midpoint (Figs. 280 & 282).

2=The sides are flared out as very prominent posterior-lateral extensions, and the

ventral midpoint of the ridge is barely produced over the lamella antevaginallis

(Figs. 279 & 281).

Independent evidence indicates state 1 of the preceding character is a terminal derived state, therefor taxa lacking this state are coded "0". 99. Development of the lamella antevaginallis relative to the lamella postvaginallis:

l=Well developed, as large as the lamella postvaginallis or only slightly smaller

(Figs. 261-263,265,274-275,277-292).

2=Very reduced, much smaller than lamella postvaginallis (Figs. 267-272). Higginsius miriam has the lamella antevaginallis much smaller than the lamella postvaginallis, yet the lamella antevaginallis is not very reduced but rather the lamella postvaginallis is unusually expanded. I code this taxon as state 1. In some out group taxa there is no distinction between lamella antevaginallis and lamella postvaginallis, rather there seems to be a single plate around the genital opening. These taxa are coded "?". 100. The formation of a pouch by the lamella postvaginallis and lamella antevaginallis around the opening to the ductus bursae:

0=Not forming a pouch, projecting in opposite or widely divergent directions.






76


1 =Forming a partial open pouch, with the separation wide enough that the

opening to the ductus bursae can be seen in ventral posterior view without prying

the plates apart with a forceps, projecting at an acute angle (Figs.

283,286,285,288,290,291 & 292-this pouch is not apparent for all taxa which

possess it at the angles from which they are illustrated).

2=Forming a closed pouch, with the opening to the ductus bursae difficult to see unless the plates are pried apart with a forceps, projecting at a small acute angle

(Figs. 262-265).

While taxa with state 0 could be easily scored, there were some cases where distinguishing between state 1 and state 2 became somewhat subjective, mainly for Antillea. There can be some variation in how much the pouch is closed within the same taxon, as was noted for some Chlosyne and Mellicta athalia. Texola tend to have the partial wider than Chlosyne. However, most individuals could be readily assigned to one of the above states. Note that Phycioditi lack any pouch formed from the lamella, however they have an analogous membranous pouch in a similar position.

Upon casual inspection, Atlantea may appear to have a closed pouch of state 2 similar to Mellicta. However, the pouch in Atlantea is formed by a posteriorly curved broad emarginate process extending off of the anterior edge of the lamella antevaginallis and fused with the lamella postvaginallis at its base. The lamella antevaginallis itself projects opposite the lamella postvaginallis typical of state 0. In state 2 taxa where the lamella antevaginallis plate is angled acutely with respect to the lamella postvaginallis, the abdominal stemite is fused to the distal end of the pouch. In Atlantea, where the






77

pouch is formed by an extra process, the abdominal sternite is attached to the base of the lamella antevaginallis plate, and the process is free of the stemite throughout its length.

The out group taxon Catacroptera cloanthe has a pouch formed by a extension of the abdominal tergite over the lamella antevaginallis. The posterior edge of the pouch formed by the sternite is at the anterior edge of the ductus bursae opening. The lamella antevaginallis itself is characteristic of state 0. 101. Given state 2 of the preceding character, the edge of the lamella postvaginallis with respect to the lamella antevaginallis at the distal edge of the pouch:

I =Extended nearly even with the lamella antevaginallis (Fig. 264).

2=Extending well posterior of the lamella antevaginallis (Figs. 262-263 & 265266).

While independent evidence suggests state 2 of the preceding character is a terminal derived state, taxa lacking state 2 of the preceding character are coded "?" because this character is not necessarily dependent on having state 2 for the preceding character. However, taxa with the lamella pouch can be readily assigned to discrete states for this character, which is not the case for taxa with the lamellae divergent. 102. An extension of the outer edges of the lamellae antevaginallis beyond the plane of the rest of the lamella antevaginallis on the side opposite the ductus bursae opening, forming concave depressions on each side:

0=This area is fairly smooth, lacking such extensions.

1 =The above extensions are present (Fig. 283).

103. Folding of the distal edge (farthest from the genital opening) of the lamella antevaginallis:






78


0=Unfolded.

I =Folded over on the inner side such that the distal edge of the lamella

antevaginallis is a three-dimensional structure (Fig. 264).

104. Sides of lamella antevaginallis near the base with a sclerotized structure with a convex outer side folded around the sides of the lamella antevaginallis and attached to the abdominal sterna ventrally and the lamella postvaginallis dorsally:

0=This structure absent.

I =This structure present (Figs. 262-266).

This and the preceding character were examined by prying the lamellae apart with a forceps as well as examining them from ventral aspect. 105. The presence of a lightly sclerotized area on the lamella postvaginallis posterior to the ductus bursae opening:

1=Absent (Fig. 261).

2=Present (Figs. 262-264,270,273-274,277-291 (several illustrations depict taxa

with this state where it can not be seen due to the angle from which the illustration

was produced).

Note that taxa found to have state 1 are also taxa with a sclerotized ductus bursae. These taxa have the distal posterior side of the ductus bursae lightly sclerotized. It may be possible that the sclerotized end of the ductus bursae in these taxa was formed from an invagination of the lamella antevaginallis and lamella postvaginallis sclerites, and that the presence of a sclerotized ductus bursae and the above character may not truly be independent.






79

106. Given state 2 of the preceding character, the extent of the lightly sclerotized area at the base of the lamella postvaginallis:

1=The entire base of the lamella postvaginallis is lightly sclerotized (Figs. 262263).

2=The more heavily sclerotized area extends to the base of the lamella

postvaginallis on the sides (Figs. 264,270,274,277-287). Taxa lacking state 2 of the preceding character are coded "T'?". 107. Given state 2 of the second preceding character, the continuity of the lightly sclerotized area at the lamella postvaginallis base with the posterior edge of the lamella postvaginallis:

1 =Narrowly contiguous (Fig. 283).

2=Not contiguous (Figs. 262-275 & 277-292).

3=Very broadly contiguous (Fig. 282).

108. Vicinity of the horizontal midline of the lamella postvaginallis:

0=Smooth, lacking a ridge (Figs. 261,266,267,269-270,273,275,277(there is a

concave depression not a ridge),278-288 & 290-291).

1=Extends ventro anteriorly as a hollow ridge (Figs. 262-264).

109. Anterior edge of lamella antevaginallis with a strongly posteriorly curved broad emarginate process overlapping the lamella postvaginallis for most of its length and enclosing the ductus bursae opening in a pouch. Dorsally the process has posterior extensions fused with the lamella postvaginallis on each side of the base which form "walls" around the genital opening.

0=Lacking a process.






80


l=With the process described above (Figs. 274-275). Ovipositor Lobes:

110. A distinct sclerotized pouch anterior to the ovipositor lobes formed from an extension of the anterior base of the ovipositor lobes on the ventral side.

#0=Absent.

1 =Present (Figs. 268,288-290 & 291-292).


Out Group Characters

The following characters provide no information pertinent to the relationships within the Melitaeini, as all Melitaeini have the same states for these characters. These characters were coded for an analysis using a smaller number of Kallimini and Nymphalini representatives as the out group (Junonia coenia, Catacroptera cloanthe, Anartia chrysopelea, A. jatrophe, A. fatima, and A. amathea). These characters may be helpful for future work on the relationships within or between these genera, but are by no means an exhaustive list. They are irrelevant to and not used in the analysis using collective out group coding.

111. Each anterior side of the vinculum projecting anteriorly in a prominent, broad and smooth-sided, heavily sclerotized anterior-ventral projection:

0=Absent.

1 =Present.

112. When viewed at a ventral posterior angle, a flattened U-shaped plate between the valves and anterior of the uncus, densely covered with numerous long, thick hairs projecting dorso posteriorly over the plate:

O=This feature absent.






81


1 =This feature present.

113. Outer posterior side of valve curved over 180' forming a short extension over the inner posterior lateral side bearing 5-8 (n=2) teeth on the ventral side and three teeth along the vicinity of midline on the surface:

O=This feature absent.

l=This feature present.

114. Outer dorsal posterior side of valvae curved over the ventral posterior inner side in a curved, flat extension tapering to a triangular point projecting inward with an anterior slant. The projection bears five (n=2) prominent teeth on its posterior edge:

O=This feature absent.

I =This feature present.

115. Posterior ventral comer of valvae extended inward in a triangular extension:

O=Absent.

1 =Present.

116. A sclerotized lateral ridge of pointed teeth on each posterior lateral side of the everted vesica, separate from the phallus:

1 =Present.

2=Absent.

117. Ventral anterior sides of phallus with a flat, flared, lateral extension slanting dorso anteriorly:

I =Present.

2=Absent.






82


118. Anterior end of phallus with a scoop-shaped membranous pouch, with the convex border of the scoop on the dorsal side and the ventral side slightly concave.

0=Absent.

1 = Present.

119. Ventral anterior side of phallus with a laterally expanded plate attached to the ventral and lateral sides of the phallus (a suture line is clearly visible). The plate is arched ventrally and U-shaped when viewed in anterior aspect:

0=Absent.

1 =Present.

120. Form of the ductus bursae:

O=Very short or vestigial.

1 =Long and tubular.

121. Form of the posterior end of the corpus bursae::

0=Corpus bursae tapers posteriorly.

I1=Tube-like and of similar width, distinctly narrower than the remainder of the

corpus bursae


Phylogenetic Analysis

A heuristic search with equally weighted characters with the representative out group method yielded 5 10 equally parsimonious trees with a consistency index of 0.85 8, a rescaled consistency index of 0.827, and a retention index of 0.965. The strict consensus tree of this analysis is presented in Figure 293. The consensus tree (Figure 293) is almost completely resolved except for relationships within the Phycioditi dlade. In fact, almost all the equally parsimonious trees are a result of permutations within the






83


Phycioditi clade. If all but two Phycioditi are deleted (or the P. tharos three species clade plus any other Phycioditi taxon), only two most parsimonious trees are obtained (before successive weighting CI=0.892, RI=0.970, RC=0.865; after successive weighting CI=0.940, RI=0.983, RC=0.924) with no change in any of the other clades. Of 115 parsimony informative characters, 88 characters require only the minimum number of steps on the most parsimonious trees, and 27 characters are homoplastic. Status of individual characters is presented in Table 2.

The strict consensus tree supports the monophyly of Higgins' (1981) Euphydryiti and Phycioditi, but indicates Melitaeiti is paraphyletic. Higgins (1981) concepts of Antillea, Higginsius, Atlantea, Eurodryas, and Hypodryas come out as monophyletic groups. However Higgins (1981) concepts of the following genera come out as nonmonophyletic (*) or make a different genus paraphyletic(#): Texola*, Dymasia#, Microtia#, Gnathotriche*, Gnathotrusia#, Occidryas*, and Euphydryas#. As noted, insufficient taxa were examined to test Higgins (1981) concepts of Phycioditi genera or Melitaea, Mellicta, Cinclidia, and Didymaeformia, however the taxa in the latter four genera have grouped together with the limited sample examined. The Chlosyne group including Chlosyne, Thessalia, Charidryas, and Anemaca, comes out as a monophyletic group.

The shortest possible trees obtainable from successive weighting with the

representative out group method were obtained from the second successive weighting analysis. Twelve equally parsimonious trees were obtained, with a consistency index of

0.926, retention index of 0.982, and rescaled consistency index of 0.910. The topology of the strict consensus tree of successively weighted characters (Figure 294) is identical to






84


that obtained from equally weighted characters, except there is greater resolution within the Phycioditi clade.

Boot strap 50% consensus trees from equally (Figure 295) and successively

weighted characters (Figure 296) with the representative out group method support the same clades with one notable exception. The clade with Chlosyne comes out as sister clade to the clade with Poladryas with a score of 51% for successively weighted characters, but this clade only has a score of 40.5% for equally weighted characters. Most clades from the two analyses have scores within 5% of each other, and only one other clade has a score with a difference greater than 9%; the clade uniting the two representatives of Cinclidia and Melitaea has a score 13% lower for equally weighted characters. None of the clades supported by the boot strap analyses are in conflict with the strict consensus trees for equally or successively weighted characters, and all of the clades appearing on the strict consensus tree of equally weighted characters appear on both 50% boot strap consensus trees with three exceptions. One exception is the missing union of the Chlosyne group and Poladryas group on the boot strap tree from equally weighted characters as noted. In addition, the relationships between the three clades coming out as the sister clade to Euphydryiti are unresolved on both boot strap trees. Finally, Mellicta varia, M. asteria, and M. aurelia came out as a resolved clade on the strict consensus trees but this clade was collapsed on both of the boot strap consensus trees.

With the cumulative out group method, the heuristic search of equally weighted characters yielded 2,546 equally parsimonious trees with a consistency index of 0.863, retention index of 0.963, and rescaled consistency index of 0.831. The strict consensus






85


tree (Figure 297) is identical to the one obtained for the representative out group method, except the three species clade of M. varia, M. aurelia, and M. asteria is collapsed. Once again, most of the most parsimonious trees are due to permutations within the Phycioditi clade, and if Phycioditi are removed except for the P. tharos three species clade and one other representative, only 10 most parsimonious trees are obtained (CI=0.9, RI=0.968, RC=0.871 with equal character weighting; CI=0.946, RI=0.983, RC=0.929 with successive character weighting). Of 108 parsimony informative characters, 83 only require the minimum number of steps on the most parsimonious trees while 25 characters are homoplastic. The status of individual characters is presented in Table 3.

The shortest possible trees derived from successive weighting were obtained after two repetitions of successive weighting. Forty most parsimonious trees were obtained, with a consistency index of 0.933, retention index of 0.983, and rescaled consistency index of 0.917. The strict consensus tree (Figure 298) is identical to the one derived from the representative out group method except for the same three collapsed Mellicta taxa noted above.

Boot strap 50% consensus trees for equally and successively weighted characters obtained with the cumulative out group method differ little with respect to their complementary trees derived from the representative out group method. The resolution obtained is identical, and the scores differ by less than 10% in every case, and the scores for nearly all clades differ by no more than 5%.

It is not possible to report exact number of character state changes on each branch length because a number of characters can be drawn on a tree in more than one equally parsimonious way. Figure 301 presents the minimum and maximum number of possible






86


state changes for each branch length, as calculated with MacClade 3.07. Figure 302 gives character numbers for those characters which undergo an unambiguous state change on each branch. The tree used to obtain this information includes the representative out group (the number of unambiguous state changes is less with the cumulative out group, because the cumulative out group is treated as one taxon) with the Junonia and Anartia clades drawn as paraphyletic. If these clades are drawn as monophyletic, there are no unambiguous state changes at the Melitaeini ancestral node, but the other values are unaffected. Switching the order of these clades in the paraphyletic out group arrangement has no affect, however. Also, since MacClade 3.07 does not calculate unambiguous changes below polytomies, even if all taxa above the polytomy node have the same state, I used the arrow under "Tools" to remove polytomies for Figs. 301-302 in an arrangement compatible with a most parsimonious tree.


Discussion

The different types of analyses including heuristic searches with equally weighted characters, heuristic searches with successively weighted characters, boot strap analyses with equally and successively weighted characters, and their application to the representative and cumulative out group methods, all support congruent tree topologies, The results of these analyses provide a well resolved and well supported phylogenetic hypothesis for a number of Melitaeinine clades, and provide a framework for detailed studies of relationships within the three clades not investigated in detail in this analysis: the genus Chlosyne, the Phycioditi, and the Eurasian taxa Higgins (1981) placed in Melitaeini. A detailed phylogenetic study of Chlosyne is presented in the following chapter.






87


One significance of the results is that they allow for the construction of a well supported phylogenetic classification of Melitaeinine subtribes based on the criteria of monophyly and stability. I find that Robbins and Henson's (1986) guidelines for applying these criteria to constructing a classification work well for Melitaeinine subtribes: "[1] If a genus [or other higher taxon] is monophyletic, do not change the name. [2] If a genus is not monophyletic, choose the combination of monophyletic generic groupings that will create the fewest name changes. [3] If another option causes more name changes now but will be more stable in the future because of better evidence of monophyly, then present the reasons and evidence for that choice" (content in brackets [] inserted by the author]. However, while I apply these three guidelines to my subtribal classification scheme, I do not adhere to a strict interpretation of the first guideline at the generic level because there are cases where synonymizing a monophyletic genus to make a paraphyletic genus monophyletic achieves the goal of monophyly with fewer name changes. Furthermore, I always choose to synonymize monotypic genera that make another genus paraphyletic. Monotypic genera are without value to reflecting evolutionary relationships, plus retaining such genera often would involve more name changes than synonymizing them and never require fewer. Consequently, I amend Robbins and Henson's (1986) guidelines 1 and 2 as follows: "If a genus is not monophyletic, or if a genus is monophyletic but makes another genus paraphyletic, choose the combination of new monophyletic generic groupings and/or generic synonymizations that will create the fewest name changes. If a genus is monotypic and comes out within a different genus, synonymize the monotypic genus (or the genus representing the clade to which it belongs, if the monotypic genus has the older name). If






88

two or more alternative natural classifications require equivalent numbers of name changes, choose the option which requires the fewest changes to valid taxonomic categories." I accept Robbins and Henson's (1986) guideline number 3 verbatim.

Applying the above guidelines leads to the retention of Euphydryiti Higgins and Phycioditi Higgins, which represent long standing taxa that come out as well supported monophyletic groups in all of the analyses. Application of the second guideline involves splitting Melitaeiti into three subtribes, with Melitaeiti now restricted to the Eurasian taxa Higgins (1981) included in his concept of Melitaeiti, Gnathotrichiti as a new subtribe forming the sister clade to Phycioditi, and a third subtribe including the Poladryas and Chlosvne clades. However, application of Robbins and Henson's (1986) third guideline favors naming the Chlosyne and Poladryas clades separately (as Chlosyniti and Poladryiti), since the evidence of the individual monophyly of these two clades is much stronger than the evidence of their combined monophyly. While placing Gnathotrichiti within the Phycioditi would require the same number of name changes and be as phylogenetically valid as giving it Gnathotrichiti subtribal status, I choose the former action because it does not require changing the boundaries of a valid taxon (Phycioditi).

Several factors can be considered when evaluating how well supported a

particular grouping of taxa is, including congruence between independent data sets, character quality, and the number of extra steps required by less parsimonious arrangements. Congruence between independent data sets is used in the following chapter, where the phylogenetic analysis includes a character set from genitalic characters plus a character set from scale pattern characters, but it cannot be used to evaluate the results presented in this chapter since all characters are based on genitalia. Character






89


quality includes simple versus complex characters, and the degree of character state uniqueness among related taxa and out groups (for example, is a character state supporting a clade a unique synapomorphy or a state that must have independently evolved more than once to account for its distribution on the most parsimonious tree(s)?). Unique character states which require only one step to account for their distribution on the most parsimonious tree(s) arguably provide greater support than individual character states which appear to have independently evolved more than once. Also, character states based on complex sclerotized structures are arguably less likely to evolve more than once (see the following chapter for empirical evidence supporting this hypothesis). I argue in the next chapter that boot strap scores may be a very poor method of evaluating support for various clades, and that clades appearing only in an analysis based on successive character weighting are of very dubious validity. In general it is logical that the greater the number of extra steps required for an alternative topology, the better supported a particular clade is. However, I hypothesize that clades supported by relatively few steps are not necessarily poorly supported if they are supported by several complex character states but with some conflicting evidence from simple homoplastic characters.

With respect to the subtribal classification which I propose for Euphydryiti, Melitaeiti, Gnathotrichiti, Phycioditi, Poladryiti, and Chlosyniti, these clades are supported by multiple complex character states (see subtribe descriptions below), and a minimum of three to twelve additional steps would be required to break apart any subtribal clade. Also, there is consistency between equally and successively weighted character analyses. A discussion of all the universal synapomorphies from binary






90

characters and terminal derived states of multistate characters is included in the subtribe descriptions below. Of course, since not all taxa in Phycioditi and Melitaeiti were examined, there is the possibility of reversals or additional character states that I am unaware of. However, given the number of characters supporting these clades, plus in the case of some characters of Phycioditi, insight provided by Higgins (1981) camera lucida drawings, I find the possibility that examination of additional taxa would significantly undermine the evidence of monophyly for these groups to be very remote.

The value of an analysis from the representative out group method in addition to the cumulative out group method is that the representative method yields fewer ambiguous state designations at the ancestral node than when the out group is collectively treated as one taxon If there is reason to suspect the taxa included in the cumulative out group and in the representative out group do not form a monophyletic sister group to the in group (this is the case for Nymphalini and Kallimini (Harvey 1991)), making the representative out group paraphyletic in MacClade is useful for investigating the number of additional steps required for alternative tree topologies.

In order to investigate the number of extra steps required for alternative

arrangements between the subtribe clades, I constructed a matrix (Table 4) including each possible pair of subtribe clades, and each possible pair of subtribe clades grouped with another subtribe clade or pair of subtribes. Values entered into the matrix represent the minimum number of extra steps required for a tree with a particular pair of clades grouped together, as determined by checking each arrangement in MacClade 3.07 from the tree obtained from equally weighted characters and with the two clades of the representative out group drawn as paraphyletic (in either order). Since the cumulative