Phylogenetics, Evolution, and Systematics of Holodonata with Special Focus on Wing Structure Evolution

Material Information

Phylogenetics, Evolution, and Systematics of Holodonata with Special Focus on Wing Structure Evolution Morphological, Molecular and Fossil Evidence
Bybee, Seth
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (186 p.)

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Entomology and Nematology
Committee Chair:
Branham, Marc A.
Committee Members:
Frank, J. Howard
Soltis, Pamela S.
Whiting, Michael
Graduation Date:


Subjects / Keywords:
Aircraft wings ( jstor )
Edge effects ( jstor )
Fossils ( jstor )
Insect morphology ( jstor )
Insects ( jstor )
Phylogenetics ( jstor )
Phylogeny ( jstor )
Plant morphology ( jstor )
Taxa ( jstor )
Topology ( jstor )
Entomology and Nematology -- Dissertations, Academic -- UF
evolution, fossil, holodonata, homoplasy, morphology, odonata, paleoentomology, paleontology, phylogenetics, systematics, wing
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Entomology and Nematology thesis, Ph.D.


This research has broad implications to several fields of science: entomology, phylogenetic systematics, paleontology and biomechanics. Findings include the discovery of wing structures that evolve together over time and have shaped the major forms of flight seen among dragonflies today. A better understanding of the origins of the structures that support insect flight and their evolution through specific adaptations is also presented. This research provides a model system for researchers working on flight, particularly those focused on micro air vehicles. This research also has the ability to present basic information on evolution and phylogenetics to the general public via a charismatic insect group that are both attractive and well-known. Findings are relevant to the high school and university classroom and provide a clear avenue to present principle of evolution to students. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis (Ph.D.)--University of Florida, 2008.
Adviser: Branham, Marc A.
Electronic Access:
Statement of Responsibility:
by Seth Bybee.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Bybee, Seth. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
430115396 ( OCLC )
LD1780 2008 ( lcc )


This item has the following downloads:

Full Text




2008 Seth M. Bybee 2


To my wife for her endless supply of support, encouragement and attention and my children for their never ending curiosity. To my parents for all their love and guidanc e and my siblings for their examples. To all those that gave me a chance. 3


ACKNOWLEDGMENTS I am indebted to the long hours of dedicated he lp and support that I re ceived from all those who served on my graduate committee. This proj ect would not have been possible without their endless generosity, advice, and council. I am sincerely grateful to Dr. Alexandr Rasn itsyn and the paleoentomological staff at the Paleoentomological Institute, Dr. Andrew Ross at The Natural History Museum, and Dr. Andr Nel at le Musum National d'Histoire Naturelle, who helped to make this research possible and who were superb hosts. Many thanks to Dr. Davi d Grimaldi for his support, encouragement and insight into this research. Thanks to Dr. David Wahl for his help with optimizing the photographic equipment. This research was supported by grants from the Society of Systematic Biologists, the Explorers Club, the Systematic Association, the Entomological Society of America, and the Florida Entomological Society. The DNA component of this research was funded by NSF grants DEB 0120718, NSF DEB 0206505, NSF Research Experience for Underg raduates (REU) program DBI-0139501, Brigham Young University's Office of Original Research and Creative Activities, and the Entomological Foundation. This research represen ts a large collaborative effo rt from the odonate community and I am grateful for their ge nerosity with specimen donatio ns and identifications. I am particularly grateful to Je ff Skevington, Dennis Paulson, Akihiko Sasamoto, Karen Gaines, Rasmus Hovmller, Jurg DeMarmels, L.-J. Wang, Adrian Trapero, Robert Larsen, Ken Tennessen, Bill Mauffray and the International O donate Research Institute, Gainesville, FL, USA. I give special thanks to Andrew Rehn for providing specimens, specimen identification, discussions and character coding of his character systems, and his though tful critique of my research. I am also grateful to the personnel bo th current and past from the Branham and Whiting 4


laboratories for their collection of specimens, useful comments on this manuscript, and encouragement. 5


TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES .........................................................................................................................10 LIST OF FIGURES .......................................................................................................................11 ABSTRACT ...................................................................................................................................12 CHAPTER 1 PHYLOGENETICS............................................................................................................... .14 Introduction .............................................................................................................................14 What Types of Data Are Used in Insect Phylogenetics?........................................................15 Morphological Data .........................................................................................................16 Molecular Data ................................................................................................................17 Combined Analysis or Total Evidence ............................................................................18 What Are the Major Modern Metho dologies Used in Phylogenetics? ...................................18 Maximum Parsimony ......................................................................................................19 Statistical Approaches .....................................................................................................20 Maximum likelihood (ML) ......................................................................................20 Bayesian ...................................................................................................................20 What Is Going on in Insect Phylogenetic Studies? .................................................................21 2 SCANNING ROCKS FOR DATA: FOSSIL IMAGING AN D DATABASING..................25 Introduction .............................................................................................................................25 Materials and Methods ...........................................................................................................26 Collections Studied ..........................................................................................................26 Scanning ..........................................................................................................................26 Macrophotography ..........................................................................................................29 Results .....................................................................................................................................29 Discussion ...............................................................................................................................30 Conclusions .............................................................................................................................32 3 ODONATA PHYLOGENY...................................................................................................36 Introduction .............................................................................................................................36 Materials and Methods ...........................................................................................................39 Taxon sampling ...............................................................................................................39 Laboratory Methods ........................................................................................................40 Data Analysis ...................................................................................................................41 Results .....................................................................................................................................43 DO Analysis....................................................................................................................43 6


Bayesian Analysis ...........................................................................................................44 Parsimony Analyses ........................................................................................................45 Additional Analyses (Fossils Excluded) .........................................................................45 Data Completeness, Support, Congruence and Gene Contribution................................46 Data completeness: ...................................................................................................46 Support: ....................................................................................................................47 Congruence: .............................................................................................................47 Overall Gene Contribution .......................................................................................48 Discussion ...............................................................................................................................49 Odonata ............................................................................................................................49 Epiprocta ..........................................................................................................................50 Anisoptera ........................................................................................................................50 Superfamily and Family Monophyly ...............................................................................52 Zygoptera .........................................................................................................................54 Superfamily and Family Monophyly ...............................................................................55 Contribution of Fossil Taxa .............................................................................................59 Recommendations for Classifi cation Based on the Phylogenies ....................................60 Conclusions .............................................................................................................................60 4 ODONATA WING EVOLUTION.........................................................................................70 Introduction .............................................................................................................................70 Materials and Methods ...........................................................................................................71 Character Optimization (Smart) ......................................................................................71 Character Optimization (Wing Form) .............................................................................72 Results and Discussion ...........................................................................................................72 Pterostigmanodal Brace Complex .................................................................................72 Costal Wing Base & CostalScP Junction Complex ......................................................73 Conclusions .............................................................................................................................74 5 HOLODONATA PHYLOGENY...........................................................................................77 Introduction .............................................................................................................................77 Protodonata ......................................................................................................................77 Protanisoptera ..................................................................................................................78 Protozygoptera .................................................................................................................79 Triadophlebioptera..........................................................................................................80 Tarsophlebioptera ............................................................................................................81 Epiprocta ..........................................................................................................................81 Zygoptera .........................................................................................................................82 Historical Classification ..................................................................................................84 Materials and Methods ...........................................................................................................85 Results .....................................................................................................................................87 Combined Topology ........................................................................................................87 Morphological Topology .................................................................................................87 Additional topologies ......................................................................................................88 Discussion ...............................................................................................................................88 7


Combining morphological and molecular data ...............................................................88 Missing data .....................................................................................................................89 Classification ...................................................................................................................90 Holodonata ......................................................................................................................91 Protodonata ......................................................................................................................91 Protanisoptera ..................................................................................................................92 Triadophlebioptera..........................................................................................................92 Protozygoptera .................................................................................................................93 Zygoptera .........................................................................................................................94 Tarsophlebioptera ............................................................................................................95 Epiprocta ..........................................................................................................................96 Trends in Wing Evolution...............................................................................................97 Challenges .......................................................................................................................97 Conclusions .............................................................................................................................98 6 CHARACTER HOMOPLASY OF THE HOLODONATE WING.....................................105 Introduction ...........................................................................................................................105 Material and Methods ...........................................................................................................106 Results ...................................................................................................................................109 Discussion .............................................................................................................................111 Conclusion ............................................................................................................................113 APPENDIX A LIST OF MORPHOLOGICAL CHARACTERS (CHAPTERS 4&5)................................117 List of All Morphological Characters and Character States Used in this Analysis ..............117 Head Characters (Rehn, 2003) ......................................................................................117 Wing Articulation Characters (Rehn, 2003) ..................................................................118 Wing Venation Characters (Rehn, 2003)......................................................................118 Miscellaneous characters (Rehn, 2003).........................................................................122 Additional Characters Added to Rehn (2003) ...............................................................124 Additional Characters (Wheeler et al., 2001) ................................................................125 New characters ..............................................................................................................126 B MORPHOLOGICAL CHARACTE R MATRIX (CHAPTERS 3&4).................................127 C SUPPORT VALUES FIGURE 4-2 THROUGH FIGURE 4-6............................................134 D MORPHOLOGICAL CHARAC TERS (CHAPTERS 5&6)................................................137 E TOPOLOGIES OF ONLY EXTANT TAXA (CHAPTER 5).............................................160 F MATRIX OF MORPHOLOGI CAL CHARACTER CODINGS.........................................161 LIST OF REFERENCES .............................................................................................................177 8


BIOGRAPHICAL SKETCH .......................................................................................................186 9


LIST OF TABLES Table page 3-1. List of all taxa and GenBank accession numbers..................................................................68 5-1. Monophyly of the suborders of Holodonata........................................................................100 5-2. Tree statistics for the topol ogies from Figs. 5-1 and 5-2.....................................................101 6-1. Average consistency index (CI) and retention index (RI)...................................................115 10


LIST OF FIGURES Figure page 1-1. Monophyly, paraphyly and polyphyly...................................................................................23 1-2. Approximated number of possible phylogenetic trees..........................................................24 2-1. Direct vs. oblique light................................................................................................. .........33 2-2. Demonstration of the dept h of field flat bed scanner............................................................34 2-3. Comparison of depth of fields...............................................................................................35 3-1. Previous odonate hypotheses.............................................................................................. ...62 3-2. Parsimony topology (fossils included)..................................................................................63 3-3. Bayesian topology (fossils included)..................................................................................... 64 3-4. Parsimony topology (fossils excluded)..................................................................................65 3-5. Bayesian topology (fossils excluded)....................................................................................6 6 3-6. Two major resulting hypotheses........................................................................................... .67 4-1. Optimization of structural wing characters ( Pterostigmanodal brace complex ).................75 4-2. Optimization of structural wing characters ( costal wing base & costalScP junction complex).............................................................................................................................76 5-1. Major curent hypotheses of Holodonata...............................................................................102 5-2. Morphological topology.......................................................................................................103 5-3. Combined topology......................................................................................................... .....104 6-1. Major wing regions and th e longitudinal vein system.........................................................116 11


Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PHYLOGENETICS, EVOLUTION AND SY STEMATICS OF HOLODONATA WITH SPECIAL FOCUS ON WING STRUCTURE EVOLUTION: MORPHOLOGICAL, MOLECULAR AND FOSSIL EVIDENCE By Seth M. Bybee December 2008 Chair: Marc A. Branham Major: Entomology and Nematology A brief review of phylogenetic methods and th eory with a focus on insect phylogenetics is presented. A morphological study of fossil dragon flies was a heavy focus of this work, and gathering information from fossil insects re quired some novel methods. Methods for fossil insect imaging and databasing are presented. The ma jor focus of this research consists of two major phylogenetic questions focused on dragonfly-l ike insects. The first is a comprehensive morphological and molecular phylogenetic analys is of dragonfly phylogeny, focused primarily on extant lineages, although fossil lineages were included and analyzed in a simultaneous analysis. The legitimacy of hi gher-level family groups and th e phylogenetic relationships among families were tested. Thirteen families were supported as monophyletic and eight as nonmonophyletic, although two were recovered as m onophyletic under Bayesian analyses. Epiprocta and Zygoptera were recovered as monophyletic. Epiophlebiidae and the le stid-like damselflies are sister to the Epiprocta and Z ygoptera, respectively. Characters associated with wing structure were optimized revealing two wing character complexes: the pterostigmanodal brace complex and the costal wing base & costalScP junction complex. In turn, these two complexes appear to be associated; the pterostigmanodal brace comple x allowing for further modification of the 12


wing characters comprised within the costal wing base & costalScP junction complex leading the modern odonate wing. The second project also included extant dra gonflies but focused heavily on their fossil record (26 extant and 60 fossil Holodonata ta xa) and included 352 morphological characters and DNA (~6kb). The orders of Holodonata were fo und to be monophyletic and all major suborders of Odonata were also found to be monophylet ic. The topologies where similar for both morphological data and combined morphological and molecular data: (Protodonata (Protanisoptera (Triadophlebioptera (Proto zygoptera (Zygoptera + Tarsophlebioptera + Epiprocta))))). Zygoptera and Epiprocta were found to be nonmonophylet ic and the subordinal status of Tarsophlebioptera is placed in question. Wing forms appear to move from a strongly petiolate wing to a less petiolate or non-petiolat e wing among the taxa from each suborder. Characters of the wing venation are found to be extremely homoplasious, but much less so when analyzed together with DNA data. 13


CHAPTER 1 PHYLOGENETICS Introduction Phylogenetics is a relatively young and rapidly growing subdiscipline of the much older and larger discipline known as systematics. There are three complementary subdisciplines within systematics: taxonomy, classification, and phylogenetics. While taxonomy is used to describe the Earths biodiversity through identifying kinds and pl acing like kinds into definable groups, based on variation (usually termed species), and classification is used to arrange these species into a hierarchical sche me of relationships (class, order, family, etc.), phylogenetics is often used to unite taxonomy and classification by organizing the accumulated data from each species into groups representing a broader biological c ontext. This contex t is usually one of evolutionary significance. One of the main purposes of phyloge netics is to unite related organisms into a natural or monophyletic group. A monophyletic group is an assemblage of organisms that includes a common ancestor and all of its descendants. Occasionally, taxa are assembled into groups that represent artificial groupings of organisms, i.e., those which are not natural. These groupings which are called paraor polyphyletic (Fig. 11) have fallen into disfavor as they exclude closely related taxa A paraphyletic group contains a common ancestor but does not include all the descendants of that ancestor. A polyphyletic group does not contain the common ancestor nor all clos e relatives of a group of taxa. Determining whether a group is mono-, para-, or polyphyletic is done by including species of a hypothesized group in a phylogenetic analysis, which group s species based on inferences of their shared ancestry. Homology is the similarity of biol ogical structures due to shared ancestry; thus statements of homology are hypotheses about the or igin and similarity of biologi cal features among organisms. There are many forms of biological data upon which homology statements can be based, such as 14


DNA, morphological, or behavioral data. Phylogenetics offers a rigorous framework by which multiple homology statements can be tested simu ltaneously, resulting in phylogenetic trees that represent patterns of re latedness, like a very large family tree. Phylogenetics is the process by which hypotheses of evolutionary relationship (p hylogeny) can be formulated and tested. A revised classification is a desired ou tcome of phylogenetics, among others. Since phylogenetics is dependent on our abi lity to make homology statements based on data gathered from both extant and extinct orga nisms, a knowledge of Earths biodiversity is essential. Building phylogenetic trees that refl ect monophyletic groups is e ssential to scientists researching the processes of evol ution. Many insects have interest ing behaviors such as flight, communication and sociality, to name only a few, that are central to understanding many of the biological processes observed among insects and that have served to shape their evolutionary history for millions of years. The phylogenetics of insects is made more interesting given that insects are the most diverse group of organisms on the planet, and as such, they have the most diverse range of morphologies, behaviors, ecological adaptations and natural histories. Organizing the diversity of insects into monophyle tic groups within a phylogenetic framework is a major task and will be the chief focus for th e field of phylogenetics for many years to come. What Types of Data Are Used in Insect Phylogenetics? Phylogenetics is dependent on data gathered from heritable variation that can be directly compared via homology statements. Scientists have to be sure that each character within the data set being compared is homologous across all species or kinds in th e data set. For example, it would not do to compare the elytra of a beetle with the wings of a bird b ecause beetle elytra and bird wings do not share a common origin, have sp ecial similarity or share similarity in their position, even though they are both part of the flight mechanisms for their respective organisms. Structures such as bird wings a nd beetle elytra share no evoluti onary or developmental past and 15


are called analogous characters Only homologous characters s hould be used in a phylogenetic analysis. Traditionally, the most common type of da ta used in insect phylogenetic analyses are morphological data, although molecular data are fast becoming the norm among most of todays phylogenetic research. Other t ypes of information, such as behavior, ecology, physiology and developmental characters, can also be used to reconstruct phylogenetic relationships. Morphological Data Insects have a great deal of morphological variation due to the number of ecological niches they inhabit and their diversity. Insect s make use of most of the organic materials on Earth, from dung to flesh, from fungi to trees, and from mud to silk. Insects eat, produce, construct, and even culture these materials. The morphological dive rsity and innovation found among insects is unmatched by any other group of organisms. This astounding morphological variation is made even more im pressive by the insects ability to metamorphose. While many insects, such as grasshoppers, feed on the same diet and inhabit a sim ilar ecological niche their whole lives, most holometabolous insects exhibit morphological chan ges reflective of a shift in ecological niche during the immature and adult life stages. For example, female mosquito larvae have opposed mandibles and feed mostly on detritus or micro-organisms. However, as adults their mouthparts have become modified into a se rrated syringe-like struct ure suitable for piercing skin and extracting blood. There are also extreme shifts in the structure of some hemimetabolous insects due to changes in ecological niches dur ing the immature and adult life stages. For example, mayflies, dragonflies and stoneflies are aquatic during their immature stages and undergo large morphological changes in gill, eye, and skeletal structure to become successful terrestrial adults. Many insect s undergo modifications of their morphology between instars (each developmental stage between moult s ( ecdysis ) until the adult stage is reached) such as praying16


mantids, true bugs, thrips, and strepsipterans. Th ese shifts in structure have evolved in response to factors such as food shortages, the need for protection, or life stage requirements. Morphological data are also crucial to phylogene tic studies that seek to include insects from the fossil record since fossils (with th e rare exception of some encased in amber) by definition lack DNA. Fossils also play a major ro le in allowing scientists to better understand the current morphological adaptations seen among related present-day taxa because they often exhibit the more primitive form s of morphological features. Molecular Data Today phylogenetic analyses are increasin gly including molecular data for several reasons. Over the past decade, the amplification, comparison a nd analysis of DNA sequence data have become more straightforward, reliable, fast er, cheaper, and more routine than ever. This trend will likely continue. Further, both mitoc hondrial and nuclear genome s are being isolated from insects to form phylogenetic hypotheses. There are, however, stil l challenges to using molecular data within a phylogenetic context. Th e software used to align sequence data (i.e., provide homology statements between nucleo tides of a DNA sequence) and perform phylogenetic analyses is still in adequate for large data sets consisting of more than ~600 specimens for only a few genes. Nonetheless, molecular methods have proven to be very powerful and insightful for resolving many questions concerning in sect phylogenetics. Molecular data have also proven to be help ful when resolving relationships among species or groups of insects that have evolved unique morphological or be havioral adaptations that are difficult to homologize with existing data. (see What is going on in insect phylogenetic studies? below). 17


Combined Analysis or Total Evidence It is widely accepted that a powerful appro ach for generating phylogenetic hypotheses is to combine large amounts of morphological and molecu lar data within a single analysis. This is referred to as a combined or total evidence an alysis. This approach is generally preferred because it allows molecular, morphological, and any other type of data, to influence the resulting phylogenetic hypothesis in the same analysis si multaneously. This technique often produces more robust estimates of phylogenetic relationshi ps and evolutionary hypotheses than when using only one type of data. Simply put, a total evidence analysis produces a phylogenetic hypothesis generated from the largest amount of data at hand and in which a ll data contribute to the construction of the hypothesis. What Are the Major Modern Method ologies Used in Phylogenetics? Phylogenetic analyses result in tree diagrams called phylogenies, or simply, trees Tree reconstruction within phylogenetics is no simple task. The number of possible trees increases exponentially as the number of taxa (individual specimens or gene sequences) included in the analysis increases (Fig.1-2). The total number of possible trees for a given number of taxa (n) can be determined with the following formula: (2n 3)! 2 n-2 (n-2) As there are so many potential trees when ev en only a few taxa are included within a phylogenetic analysis (e.g., ~ 14 million potential tr ees for 12 taxa), there are multiple methods for tree reconstruction currently employed toda y. Modern analyses are performed almost exclusively using three major methods: Maximum Parsimony (called simply MP or parsimony) and two model-based statistical approaches : Maximum Likelihood (ML) and Bayesian. 18


Maximum Parsimony Maximum Parsimony (MP) is simplistic in its approach to phylogenetic analysis. MP finds its origins in Occams Razor, lex parsimoniae, which states that th e explanation of a phenomenon that makes the fewest assumptions is the preferred explanation. Trees are scored such that the most parsimonious, or simplest explanation for relationships among the taxa included in the analysis given the data is most defensible. Thus, the tree that is most parsimonious has the fewest evolutionary steps and is the favored hypothesis. This approach does not mandate that evolution always occurs in a parsimonious fashion, but rather uses the principle of parsimony to choose among compe ting phylogenetic hypotheses (i.e., phylogenetic trees). MP is commonly used among entomologists, esp ecially those using morphological data. It is also useful for molecular sequence data a nd generally gives very similar results to those generated by both maximum likelihood and Baye sian approaches. The most often cited weakness of MP is that of long branch attraction when using molecular data. Because there are only four possible nucleotides at any given site along a DNA sequence, the result is an increased probability of two or more taxa with rapidl y evolving DNA (i.e., high s ubstitution rates) having the same nucleotide at the same position. Long br anch attraction takes pl ace when at least two taxa representing rapidly evolving lineages are inferred to be closely related, due to similarities among their rapidly evolving DNA, when in fact th ey are not. This problem is often overcome in MP analyses by including a broader range of taxa and thus more diversity in molecular sequence data that serves to break up phylogenetic relationships that coul d be potentially recovered as a result of long branch attraction. 19


Statistical Approaches Maximum likelihood (ML): ML is used with molecular sequence data and is a more complex approach to tree reconstruction than MP. ML differs from MP in that an explicit model of sequence evolution is used to provide an approximation of the datas origin. In addition, because this approach is model-based, it is vi ewed as a parametric approach to estimating phylogenetic hypotheses. ML is seen by many systematists as a more attractive method because of its statisti cal foundation. Although becoming more rapid, ML is still a sl ow process for large data sets. Until computer computational power catches up with the demands of imposing a complex model over large molecular data sets, MP will continue to be used. However, ML is probably used more often than parsimony for moderately sized data sets. In general, both ML and MP usually converge on the same phylogenetic relationshi ps. Although ML outperforms MP in some respects (e.g., it is less susceptible to long branch attraction), its utility is directly linked to the ability of the model used during tree reconstruction to represent the data accurately. A model that does not accurately represent the data will bias the resulting tree. Bayesian: This approach to tree reconstruction is a recent development but is becoming more and more prevalent. Bayesian phylogenetics finds its origins in Bayesian statistics, and uses a priori evidence (prior probabiliti es, although rarely used in phylogenetics) to infer a revised probability (posterior probabilities) that a phylogenetic hypothesis may be true. Bayesian analyses use the likelihood function and the same evolutionary models for DNA sequence data as ML. Just as with ML, an estimate of Bayesian phylogenetic relationships is only as good as the model used to estimate them. Morphological models, that use the likelihood function generated from the data, can be employed in a Bayesian framework. These mode ls have been shown to perform well with 20


morphological data, demonstra ting that simultaneous analys es of both morphological and molecular data can be performed usi ng Bayesian as well as MP methods. Bayesian analyses employ a Markov chain Mo nte Carlo process to provide a sample of trees based on their likelihood and thus can produce a credible sa mple of trees from which a probability of relationship among groups and taxa can be directly estimated. Many phylogeneticists view this ability, coupled with the speed of Bayesian analyses (which is generally much quicker than ML with bootstrapping), as a great benefit. What Is Going on in Insect Phylogenetic Studies? Over the last decade there has been an influx of large-scale phylogenetic studies undertaken by the entomological community. These efforts are producing significant phylogenetic estimates of relations hip between the orders of inse cts and within some of the largest orders of insects (e.g., Hymenoptera, Lepidopter a, Diptera and Coleoptera). From these phylogenetic reconstructions, entomologists can directly test hypotheses of evolution and diversification across higher taxa. Current research has led to novel hypotheses concerning some of the more problematic relationships among the orders of Insecta. For example, 1.) The relationships among the lineages of the group Dictyoptera (orders Mantodea, Isopte ra and Blattodea) has long been in question and there are now more data s upporting Isoptera as a derived group within Blattodea. Placing Isoptera within the Blattodea allows scientists to examine and answer questions concerning the evolution of gut endosymbionts that allow for th e digestion of cellulose for both termites and some cockroaches. 2.) Strepsiptera has also been a problematic order as many systematists have long thought it was most closely related to Coleop tera. Currently, molecular data combined with morphological data support Strepsip tera as being most closely re lated to Diptera, forming the group Halteria, in reference to th e halteres that both groups po ssess in place of well developed 21


wings. This research has lead to the hypothesis that by manipulating only one gene, halteres can occur on either the mesoor metathorax. 3.) An other major dilemma within the classification of Insecta is the Paleoptera problem. Paleoptera are composed of Ephemeroptera (mayflies) and Odonata (dragonflies and damselflies) and both ar e hypothesized to represent the most primitive forms of extant winged insects. The relations hip between these two primitive groups of living insects as well as to the other insect orders ha s long been debated as these relationships appear directly linked to the origin and evolution of inse ct flight. The most convincing evidence to date is based on both morphological and molecular da ta which indicate the Ephemeroptera as the most primitive form of extant winged insects. [This chapter was originally written a nd submitted as a stand alone topic for The Encyclopedia of Entomology, 2 nd edition The chapterss format follows that of the encyclopedias intstructions for authors.] 22


Figure 1-1. An illustration of the three types of groupings: monophyletic, paraphyletic and polyphyletic. Phylogenetics seeks to identify, cat egorize and name only monophyletic groups. Note that there is no common ancestor for a group that is polyphyletic. 23


Figure 1-2. The approximated number of possibl e unrooted phylogenetic trees for 3-20 taxa (after Schuh, 2000) 24


CHAPTER 2 SCANNING ROCKS FOR DATA: FOSSIL IMAGING AND DATABASING Introduction Insects represent one of the most diverse groups of organisms currently living on our planet. What is often overlooked is their ma gnificent fossil record dating to at least the Devonian (>400 MYA; Grimaldi & Engel, 2005). Fossil insects represen t a unique source of morphological insight that is of ten passed over due to difficulty in locating certain fossils, the expense of shipping fossils as loan material due to weight and fragility, as well as the expense of traveling to each museum with hol dings of a particular group. In addition, methods of observing and documenting the morphological features of foss il insects can be subtle and are not widely discussed. To this end, I propose two complementary methods, one traditional (macrophotography) and the other mo re novel (flatbed scanning), to digitally document insect compression fossils. Flatbed scanners have been used for imaging living organisms, most notably for plants and plant pests (McConnell, 2006; Skaloudova et al., 2006), but also for extant insects, particularly dragonflies (Mitchell and Lasswell, 2000). Because my research is focused on the essentially two-dimensional compression fossils of Holodonata (the group containing both extant and fossil dragonflies and damselflies) scanning also lends itself nicely to this task. One challenge regarding imaging fossils, especially holodonates, is their large size ( Meganeuropsis permiana had a wingspan of ~710mm). Because many flatbe d scanners are capable of making scans of 320mm x 230mm, they are capable of imaging n early all large fossil insects. Since these specimens have not been pinned and crammed into a unit tray for storage, many have multiple and often large labels associated with them. It is not uncommon for some specimens to have folded letters included with them from promin ent scientists containing important historical 25


observations; these represent precious artifacts in their own right. Some co llections curate small fossils together in the same box or unit tray, according to a given taxon or geological formation and many compression fossils have two halves, each possessing important morphological details. The flatbed scanner is a wonderful tool that can usually capture all of this information (i.e., multiple labels, letters, both halves of the specimen, etc.) in a single scan, producing an instant image that can be used to document the entire contents of each unit tray or box in high resolution. Digital macrophotography is a compleme ntary method for documenting compression fossils. Because the rock matrix surrounding foss il specimens may be reflective or similar in color to the fossil, the ab ility to manipulate lighting often makes macrophotography indispensable. Furthermore, the rock matrix surrounding a fossil may not allow for the fossil to lie flat against that scanner, producing a chal lenge for a flatbed scanners depth of field. Materials and Methods Collections Studied To date I have digitized the holodonate type holdings from three major museums, the Paleoentomological Institute (Moscow, Ru ssia), The Natural History Museum (London, England), and le Musum Nationa l d'Histoire Naturelle (Paris, France). Each museum has outstanding fossil holdings that are exceptional for fossil insects. Taken together, these collections have unmatched holodonate holdings da ting back well into the Carboniferous and up to the Oligocene, a period of over 300 MY. Scanning Scanning fossil material must be done carefully, not only because the fossils may be fragile and irreplaceable but also because the scanner will need to be replaced if one is not careful. Typically all the contents of a unit tray or box an d all associated text data (i.e., from labels, 26


identification, etc.) can be placed on the scanner. If specimens are small, multiple fossils may be placed on the scanner together. Once in place, two or three scans of each fossil were taken. Step 1: an initial scan at 180-300 dpi serves to document the entire holdings of a single curated box or unit tray. Step 2: a second scan of each fossil was made at 1200-2400 dpi, depending on the size of the fossil (this step was repeated for the comp lementary half of the compression fossil, if one existed). Step 3: the specimen was scanned at more than 2400 dpi. This step was usually reserved for small specimens or structures that are important to my research (e.g., the nodus or wing articulation). All scans were made at 48 bit color under standa rd settings set by the software. I have tested four scanners, the Cano Scan 9950F, 8800F, and LiDe 600F, and the Epson 3200. All of these machines scan at a resolu tion of up to 4800 x 9800 dpi, except for the Epson 3200, which scans 1200 x 3200, and are meant to pr oduce high-quality scans of photonegatives and photographs. Each scanner operated equally we ll for my purposes with the exception of the CanoScan LiDe 600F. The LiDe 600F is a small, lightweight scanner that draws its power directly from the computer and is meant to be car ried within a briefcase. For these reasons it was attractive to our research. Duri ng testing this scanner was only capable of scanning documents, as its depth of field appeared to be much less than 2mm. An advantage of using high end scanners is that scanning times are very shor t. Although scanning time is dependent on the size and resolution of the scan, scans taken at 180300 dpi take much less than 15 seconds. The higher resolution scans (1200 dpi and above) can ta ke several minutes. Multiple scans may be set up at once by creating multiple scanning boxes using the cursor and mouse and setting the dpi for each scanning box separately. At this point, on e simply initiates the sc an and can walk away for some 10 minutes. Additionally, by scanni ng directly into Photoshop, images can be 27


immediately altered and saved in se veral different file formats. For my purposes, files are stored as .tifs. Specific materials are needed to have a successf ul scanning trip. Besides checking that all electronic equipment is capable a nd adequately equipped (e.g., adap ters) to operate properly in a foreign country, there are some materials that will make the process easier, faster and safer. Most fossils are not heavy enough to break or crack th e glass of a scanner but they could scratch it, thereby compromising future scans. This is easily overcome by placing a thin, clear sheet of plastic, such as a standard overhead projector sheet, on top of the scanners glass bed and under the fossil. It also helps to have a chammy r eady to wipe away any dust build-up on the plastic sheet or the flatbed scanner. Another indispensable tool is a broad rubber band (~5mm wide). By positioning the rubber band on its side, one can prop fossils up so that they will lie in the same plane as the glass bed. The rubber band al so keeps a good grip on the overhead sheet, whereas paper wedged beneath a fossil tends to sli p. A thin plastic or paper ruler, placed on the glass bed, provides an accurate source of scale fo r the image. Some phot o manipulation software packages have the ability to provide measurements of size directly from images as long as they are scanned to 100% actual size. However, as a quick reference and as it was my intention to disseminate these images over the web in different sizes and file formats, a ruler to provide scale scanned with each image served my needs and proved to be efficient. A black cloth was laid over the fossil(s) on the scanner bed to cut down on interference from outside light. The cloth also serves as a paper weight for the labels a nd other associated papers keeping them in close contact with the glass bed. A laptop computer is needed to both dr ive the scanner and run photo manipulation software. Along with the laptop, I used severa l small-profile, USB-pow ered, travel-friendly 28


external hard drives to backup all images. A ll files were backedup each day on each hard drive and the files generated from that day were erased from the lapt op. I generated an average of 20GB of images from each trip and leaving this amount of data on my desktop was simply not possible. Macrophotography Digital photographs for each fossil were ta ken with a Nikon D70 digital SLR (6.1 megapixel), fitted with an AF Micro Nikkor le ns (105mm/2.8) and a MacroLume TTL Promaster ring flash. To provide greater contrast between the fossil and rock matrix, a single external halogen light source was used to provide oblique illumination to provide greater contrast across the surface texture of the fossil. Each fossil was photographed at least twice. The first image was taken to document the fossil and the rock in which it was contained and as a backup to the scanned material. The second image was as a greatly magnified shot used to document fossilized st ructure(s). As some fossils were very large, images of the wing were taken in sections. Generally, I took 5-6 images for each wing >75mm. Images were taken in .RAW format and later transformed to .tif files using Photoshop. A grey or black cloth was us ed both as background and for positioning of the fossil so that the specimen was as horizontal as possible during photography. A cloth rather than grey card stock was best as it could be fit around the fossil allowing for better manipulation of light. Results I found that imaging insect fossils with a fl atbed scanner provided several advantages. Scanning provides the opportunity for high-resolution images of th e entire specimen, even when large. The entire contents of a curated box or unit tray (e.g., multiple foss ils and labels) can all be scanned at once. A scanner can be programme d to scan automatically, thereby providing the 29


researcher time to prepare additional fossils for scanning. This method of imaging uses direct light to evenly illuminate the subject to pr eserve details that otherwise might be lost. The disadvantages of scanning include tr aveling with heavy, bulky equipment (e.g., a scanner and laptop computer); a shallow depth of field (~5mm) ; a relatively long scan time (when compared to a digital camera); and the production of large file s (3-200 MB depending on scan size and resolution). A reflective rock matrix is also a challenging material to image using the direct lighting of a scanner. My digital images produced using both s canning and digital phot ographic techniques are or will be available to the scientific community for study on the morphological image database MorphBank ( They will also be hosted at, the worlds largest website devoted to Odonata for use by the general pu blic and educators. Discussion The use of flatbed scanners and macrophotography in imaging insect fossils is both complementary and effective. The two methods differ most in three main aspects: lighting (direction of illumination), depth of field, and ability to minimize light reflected from the rock matrix surrounding fossils. The two methods of illumination used to image the insect compression fossils were direct and oblique lighting. Each can be used to comp lement the other as they often can accentuate different features of a subject. Direct ligh ting can reduce or eliminate artifacts caused by reflections or shadows, thereby producing an exac t representation of the s ubject. In the direct lighting subject, light is directed onto the subject from the same direction as an eye viewing it, thereby evenly illuminating the subject. The diffic ulty of using direct li ghting is that it can be too harsh to clearly show subtle differences in the texture of a material. Oblique lighting often produces more drama in an image by creating shadows that demark changes in the surface 30


texture of a fossil, thereby adding more contrast. With this type of illumination, light is directed toward the subject at an oblique angle (e.g., fr om the side) to create slight shadows that accentuate subtle features of the s ubjects surface. This is perhap s most dramatically illustrated via the corrugations of the odonate wing (compare Figs. 2-1a & b). The difficulty of using oblique illumination when imaging fossil insects is that the generated shadows can partially or completely obscure other features that are present. Scanners project only direct lig ht on the scanned object, resulti ng in images that capture all structures of the wing, particularly venati on, in a very clear image (Fig. 2-1a). Macrophotography provides the flexib ility to adjust the direction of the light source (e.g., direct and/or oblique illumination). The use of oblique lighting accentuates the topography of the wing by creating shadows (Fig 2-1b). Both images, one that comprehensively documents details of the wing (e.g., wing venation) and the other th at accentuates surface topography, are ideal and necessary complements from which to study morphological characters. Depth of field is the biggest difference be tween imaging fossils with a scanner or by photography. The scanners I used had a depth of field of approximately 5mm, but began to quickly deteriorate at ~3.5mm (Fi g. 2-2). Depth of field is only an issue when a fossil is sunken within the rock matrix (i.e., it is not in the same plane as or resting close to the scanners glass bed). The issue that arises due to the scanners de pth of field is that portions of the wing that are set within the rock matrix at more than 3.5mm be gin to lose focus (see white boxes in Fig. 2-3a). The combination of a fossil embedded within the ro ck matrix and the direct light produces scans that are flat (compare the scan in Fig. 2-3a with the photograph in Fig. 2-3b). Additionally, keeping the entire fossil in fo cus when manually photographing fossils, especially when large 31


(see box, Fig 2-1b), is difficult. As long as the wing is flat with the surface of the rock face, as is often the case, the scanner will keep the entire wing surface in focus (Fig. 2-1a). Reflective rock matrices can complicate the imaging of insect compression fossils. Certain types of rock matrices that contain fossils can be reflective due to thei r light color or mineral composition. An increased amount of reflective li ght obscures subtle feat ures, especially those that are light in color. Macrophotography has the advantage in that the amount and direction of illumination can be adjusted or the fossil itself can be moved to an angle at which less light is reflected back toward the lens, resulting in less bleached images. The ability to manipulate both the method of illumination and the orientation of th e fossils allows for capt uring greater contrast between the fossil and its surrounding matrix. Multiple images complement each other as the manipulation usually results in the ability to photograph only a portion of a fossil. Conclusions I discuss using the flatbed scanner and macrophotography as two methods for imaging insect compression fossils. Both methods represent viable cost-effective imaging techniques that are complementary. Scanning represents an extremely low-cost, high-throughput solution to digitizing the insect fossil holdings (at least type specimens), and I encourage its use. Fossils can contribute a great deal to the st udy of evolution, and image database s are a valuable tool for both storing and disseminating image data to the scient ific community. My hop e is that the use of these simple techniques for imaging fossils in conjunction with the use of online databases for the Holodonata will catalyze research on this and other groups of insects. 32


Figure 2-1. Comparison of direct vs oblique light: a) the image ca ptured from a scan with only direct illumination of the fossil; b) an im age taken with a digital camera while under oblique light. The white box de notes an area of the image where sharp focus was lost using macrophotography due to the large size of the wing. 33


Figure 2-2. Demonstration of the depth of field when using a CanoScan 9950F flatbed scanner. Each bar denotes 2.5mm and the yellow bar ma rks the deterioration of the depth of field. 34


Figure 2-3. Comparison of depth of fields: a) scan of a fossil that has portions of the wing embedded in the rock matrix (i.e., wing re gions are not in the same plane as the scanner bed). The white boxes highlight area s of the wing where sharp focus is lost due to a shallow depth of field; b) A di gital photograph of the same fossil showing improved depth of field resulting in sh arper focus throughout the entire wing. 35


CHAPTER 3 ODONATA PHYLOGENY Introduction The order Odonata, commonly known as dr agonflies and damselflies, includes approximately 5500 species divided into two suborders ( Bridges, 1994 ; Rehn, 2003 ): the morphological diverse suborder Zygo ptera and the slightly more sp ecies rich suborder Epiprocta (defined as Anisoptera + Epi ophlebiidae). Dragonflies are among the more popular insects with the public because they are large, often colorf ul, highly visible insects, and for this reason odonates are the focus of many conservation efforts. As they are wholly insectivorous, feeding on small insects such as mosquitoes, dragonflies al so show some promise as biological control agents in augmentative releases of the im mature stages to confined habitats ( Sebastian et al., 1990 ). Nevertheless, odonates are much more than the "mosquito hawks" and "charismatic mega fauna" of the insect world. Odonates are among th e most primitive of winged insects and there has been much controversy surrounding their phylogenetic position ( Kristensen, 1991 ; Whiting et al., 1997 ; Wheeler et al., 2001 ; Ogden and Whiting, 2003 ; Kjer, 2004 ). They exhibit some of the most distinctive behaviors among insects, pa rticularly their flight and mating behavior. Owing to the uniqueness of their behavior and phylogenetic position among insects and both life stages being tied to a diversity of aquatic habitats they have serv ed as the subject of study in many evolutionary, biomechanical, be havioral and ecological studies ( Corbet, 1999 ). In depth cladistic studies at th e familial, generic and species level have been performed for odonates ( Brown et al., 2000 ; Misof et al., 2000 ; Artiss et al., 2001 ; Carle and Kjer, 2002 ; Pilgrim et al., 2002 ; Turgeon and McPeek, 2002 ; von Ellenrieder, 2002 ; O'Grady and May, 2003 ; Dumont et al., 2005 ; Kiyoshi and Sota, 2006 ; Stoks and McPeek, 2006 ), but an overall estimate of dragonfly phylogeny that combines bot h molecular and morphological data for extant 36


and fossil groups has yet to be done. Estimat es of odonate phylogeny based on explicit, quantitative analyses have only re cently appeared, and higher-level relationships within the order are greatly debated ( Bechly, 1996 ; Lohmann, 1996 ; Trueman, 1996 ; Misof et al., 2001 ; Rehn, 2003 ; Saux et al., 2003 ; Hasegawa and Kasuya, 2006 ). Monophyly of the currently recognized extant suborders of Odonata are still debate d, as are the phylogenetic relationships of the constituent families. The majority of phylogenetic hypotheses for Odonata as a group are based largely on morphological characters asso ciated with the wing ( Needham, 1903 ; Munz, 1919 ; Fraser, 1954, 1957 ; Tillyard and Fraser, 1938 ; Carle, 1982 ; Pfau, 1991 ; Bechly, 1996 ; Trueman, 1996 ). Several of the post-Hennigian phylogenies based on wing venation suffer from poor taxon sampling, insufficient morphological characters, and in some cases are not based on modern cladistic analysis but on "intuitive" arrangeme nts of taxa and lack any formal analyses. Hypotheses based on molecular sequence data are confined primarily to the suborder Epiprocta and limited chiefly to mitochondrial markers ( Artiss et al., 2001 ; Misof, 2002 ; Saux et al., 2003 ). Despite Rehn's (2003 ) well-documented morphological matrix and a long history of morphology research ( Tillyard, 1917 ; Asahina, 1954 ) all molecular hypotheses fo r Odonata to date have ignored morphology for inclusion in a total evidence analysis. Several contradicting hypotheses have recently been proposed for higher-level phylogenetic relationships of odonates (for an in -depth review of precladistic hypotheses of Odonata, see Rehn, 2003 ; Westfall and May, 2006 ). Bechly (1996 ) performed an analysis of wing structure, but because he used intuitive rather than quantitative methods of analysis, resulting in subjectively weighted morphological characters as more important than others, his results are difficult to examine objectively. Based on this analysis, Bechly (1996 ) hypothesized a 37


monophyletic Zygoptera with the superfamilies ( sensu Bechly, 1996 ) Calopterygoidea + Amphipterygoidea and Lestoidea + Coenagrionoidea as sister ( Fig. 3-1a ). Trueman (1996 ), in an effort to combine both extant and fossil odonates, performed cladistic analysis limited to wing vein characteristics. Trueman (1996 ) proposed a paraphyletic Zygopt era with Perilestidae, a member of the Lestinoidea, as sister rela tive to the remaining families of Zygoptera ( Fig. 3-1b ). Rehn (2003 ) produced a matrix based on morphological characters including but not limited to the wing, broadly sampled the suborder Zygoptera, and analyzed these data in a standard cladistic analysis. Rehn (2003 ) also recovered a monophyletic Zygoptera with Philoganga + Diphlebia (considered here as members of the Calopterygoidea) as sister to the remaining Zygoptera ( Fig. 3-1c ). Recently, Saux et al. (2003 ), based on 12S, and Hasegawa and Kasuya (2006 ), based on 16S and 28S, proposed similar hypot heses for the group. Hasegawa and Kasuya preferred independent analyses of each gene rather than a tota l evidence approach, based on nine of the 32 families of Odonata, and found Zygoptera to be strongly supported as paraphyletic and Lestidae, a family traditionally thought to be within Zygoptera, as most closely related to Epiprocta ( Fig. 3-1d ). There is no consensus as to the m onophyly nor the sister lineage of the suborder Zygoptera among the cu rrent higher-level hypotheses ( Fig. 3-1 ). There is consensus, however, among all authors as to the monophyly of Anisoptera and the family Petaluridae as sister to the remaining familie s of Anisoptera. Moreover, all four recent hypotheses place Epiophlebia superstes (one of two species from the subor der formally known as Anisozygoptera) as sister to the remaining Epiprocta ( Fig. 3-1ad ). A broad sampling of DNA sequence data from a wide selection of taxa across both Epiprocta and Zygoptera, coupled with morphological data for ex tant and fossil groups, would assist in illuminating and strengthening our unde rstanding of the relati onships between major 38


odonate lineages and their evolution. Present herein is the first comprehensive analysis of higherlevel phylogenetic relationships in Odonata, based on combined morphological and molecular data, with an emphasis on testing the legitimacy of current higher-level classification schemes (order, suborder and family), evaluating the phylogenetic pa tterns between families, and evaluating the contribution of fossils in order to address evolutionary questions focused on features of the odonate wing (see Chapter 4). Materials and Methods Taxon sampling A recent world list of Odonata ( Schorr et al., 2007 ) was followed as the source of familial names and Bridges (1994 ) was used for taxonomic names included in this analysis above the family rank. There are currently 34 families and 600 genera recognized within the extant Odonata ( Bridges, 1994 ; Schorr et al., 2007 ). I obtained multiple ex emplars representing 30 of the 34 families and 109 genera ( Table 3-1 ). Only the families Thaumatoneuridae (one extant species), Neopetaliidae (one extant species), Hemiphlebiidae (one extant species) and Dicteriabidae (two extant spec ies) are not represented in th is analysis. The taxon sampling broadly represents all major morphological varia tion within the group for both extant and fossil taxa. Because Geropteridae and Protodonata are es tablished as the sister taxa to Odonata ( Riek and Kukalov-Peck, 1984 ; Brauckmann and Zessin, 1989 ; Zessin, 1991 ; Rehn, 2003 ), I used a eugeropterid (Geropteron ) and two protodonates (Tupus and Oligotypus ) as outgroups to polarize the morphological character s. Other outgroup taxa have be en selected to polarize the molecular data from the basal, wingle ss insects (Archaeognatha and Zygentoma; Whiting et al. 1997 ; Wheeler et al., 2001 ) and from the most basal-winged insect order (Ephemeroptera; Ogden and Whiting, 2003, 2005 ), along with several neopteran in sects selected from Polyneoptera (Phasmida, Blattodea, Embiopter a, Plecoptera and Orthoptera) ( Table 3-1 ). 39


Laboratory Methods Genomic DNA was extracted from specimens pr eserved in 95% ethanol using the Qiagen DNeasy protocol for animal tissue (Valencia, CA USA). Muscle tissue was dissected from the leg and/or thorax region. Geno mic DNA and specimen vouchers were stored in 95% ETOH at 80 C and have been deposited in the Insect Genomics Collection (IGC), M.L. Bean Museum, Brigham Young University (Provo, UT, USA). E ach DNA voucher specimen is labeled, listed and cross-referenced with its genomic DNA within the IGC's electronic database. The molecular data set is composed of six genes: 12S ribosomal (12S rDNA, 0.4 kb) and 16S ribosomal (16S rDNA, 0.6 kb) and the protei n coding genes cytochrome oxidase subunit II (COII, 0.6 kb) from the mitochondrion, and Hist one 3 (H3, 0.5 kb), 18S ribosomal (18S rDNA, 2.0 kb) and 28S ribosomal (28S rDNA, 1.3 kb) from the nucleus. Primers for these loci are given in Whiting (2001 ), Bybee et al. (2004 ), Ogden and Whiting (2005 ), and Terry and Whiting (2005 ). The 18S rDNA, 28S rDNA, COII and H3 gene s were each amplified using a three-step polymerase chain reaction (PCR) at 40 cycles with an annealing temperature of 54 C for 28S rDNA, 50 C for 18S rDNA and COII. The 12S rDNA and 16S rDNA genes were amplified using a touchdown method with the annealing temp erature starting at 62 C and decreasing 1 C every other cycle to 42 C over 40 cycles of a st andard three-step PCR. All PCR products were visualized via agarose gel elec trophoresis to assure proper amp lification and detect possible contamination using negative controls. Products were purified using Montage PCR Cleanup Kit (Millipore) and cycle-sequenced using BigD ye Terminator chemistry (version 3, ABI). Sequences were generated using an ABI 3100 capillary sequencer at the DNA Sequencing Center, Brigham Young University. Complementar y strands were sequenced with sufficient fragment overlap to reduce sequencing errors All sequences were further examined for contamination on GenBank via the nucleotide nucleotide BLAST (blastn) search function. 40


In addition, 119 morphological characters from Rehn (2003 ) were coded for each taxon in the analysis. Rehn's (2003 ) analyses were based on a detaile d study of the skeletal morphology (13 from the head; nine from the wing articu lation; 24 miscellaneous) and wing venation (63 characters), complemented with larval charact ers (10 characters). This data set has been expanded to 153 total characters to provi de more resolution within Epiprocta ( see Appendix A for an explanation of characters and morphological data matrix). Data Analysis All topologies presented in this paper are the result of combining all data partitions in a simultaneous analysis (SA). SA of partitione d data have demonstrated the importance of discovering hidden support to furthe r corroborate and refute clades recovered when partitions are analyzed separately. This is true for both parsimony ( Chippendale and Wiens, 1994 ; Brower and Egan, 1997 ; Gatesy et al., 1999, 2003 ; Cognato and Vogler, 2001 ; Wheeler et al., 2001 ; O'Grady and Kidwell, 2002 ; Ogden and Whiting, 2003 ) and maximum likelihood methods ( Gatesy and Baker, 2005 ). SA also provides the opportunity for both molecular and morphological data partitions to inform each other such that hi dden underlying signal among all data sets is discovered ( Wheeler et al., 2001 ; Ogden and Whiting, 2003 ). In a group such as Odonata that has not only a long history of mor phological study but a rich fossil record as well, the most appropriate way to create a phyl ogeny that would allow for addr essing the objectives is by SA. The alignments for COII and H3 were generated in Sequencher 4.1 (GeneCodes, 2002, Sequencher v4.5, Ann Arbor, MI) based on conser vation of codon reading frame. The taxa Cyclogomphus heterostylus and Lestoidea conjuncta contained a single codon insertion at nucleotide positions 11 and 300 within the ali gned COII data partition, respectively; these autapomorphic characters were excluded from the analysis. Sequences for the ribosomal genes were initially aligned manually in Sequencher 4. 1 to identify conserved and variable regions. 41


These regions were then subdivided into partitions in order to assist the search strategy in finding more optimal solutions during direct optimization ( Giribet, 2001 ). All partitions were analyzed via Direct Optimization (DO) in POY 3.0. ( Wheeler et al., 1996 ). DO searching was implemented using the following basic parameters: "fitchtrees noleading norandomizeoutgroup impliedalignment sprmaxtrees 1 tbrmaxtrees 1 maxtrees 2 holdmaxtrees 2 slop 2 checkslop 2 buildspr bu ildmaxtrees 1 random 4 treefuse fuselimit 2 fusemingroup 2 fusemaxtrees 2 numdriftchanges 5 driftspr numdriftspr 2 drifttbr numdrifttbr 2 slop 2 checkslop 2 seed 1" on an IBM SP 2 superc omputer at the Brigham Young University Fulton Supercomputing Center. The COII, H3 and morphology data sets were treated as prealigned data in the POY analysis. ClustalX was also used to generate an a lignment for each partition from the ribosomal genes using the default parameters, and these a lignments were concatenated into a single multiple alignment. COII, H3 and morphology data sets were treated as prealigned and simply concatenated on to the end of the ClustalX ribosomal alignment. Pars imony analyses of the ClustalX alignment were performed in PAUP*.4.0b10 ( Swofford, 2002 ) using 100 random additions with TBR swapping with gaps treated as both missing and 5th state characters. This same alignment was then test ed via Modeltest version 3.06 ( Posada and Crandall, 1998 ) to select the most justified substitution model for analyses run in MrBayes 3.1 ( Ronquist and Huelsenbeck, 2003 ). The matrix was analyzed on a Dell desktop computer for 6 500 000 generations using four chains with a sample frequency once pe r 1000 generations. All analyses as described above were also performed with fossils both included and excluded. This was done to understand the contribution of fossil taxa to th e overall analysis, and to examine the effects of missing morphological data. 42


Partitioned Bremer support values we re computed on the DO topologies via PAUP*.4.0b10 using a command file generated in TreeRot 2Vc ( Sorenson, 1999 ). Partitioned Bremer supports were computed three times to ensure consistency am ong the values recovered and thus repeatability. Partitioned Bremer support values were normalized by dividing the total number of parsimony informative characters per partition by total Bremer score per partition in order to examine the overall contribution of each partition. Bootstrap values were also generated from the DO implied alignments via PAUP *.4.0b10 with 100 random replicates. Posterior probabilities (PPs) were recovered by performing a majority rule consensus on all trees from the 6 500 000 generations minus the "burn-in". In all analyses, trees were rooted to Archaeognatha sp., the taxon representing the insect order Ar chaeognatha, given that there is significant evidence supporting this group as being the most basal insect order ( Whiting et al., 1997 ; Wheeler et al., 2001 ; Giribet et al., 2004 ). Results When data sets were analyzed individua lly (morphology and each molecular locus) or together (all molecular loci and morphology) no major incongruence was observed using the monophyly of families as the criterion to diagnos e "congruence." Relationships "higher" than familial but lower than subordinal rank were more variable between phylogenetic criteria (parsimony versus Bayesian), and this is repres ented by nodal maps located at the base of each node (Figs 3-2 through 3 ). Here, convincing evidence is pres ented that the data sets have worked together in a synergistic way to produce a large and robust estimate of odonate phylogeny. DO Analysis When the combined data matrix including fossils was subjected to DO in POY 3.0 ( Wheeler et al., 1996 ) with transversions, transitions and gaps treated as equal (1 : 1 : 1), 43


one most parsimonious tree was found of length 29 460 (CI = 0.345 and RI = 0.614; Fig. 3-2 ). Odonata was supported as monophyletic being placed as sister to Protodonata. Two major clades were recovered among the ingroup representing Epiprocta + Tarsophlebia (node 17, Fig. 3-2 ) and Zygoptera (node 61, Fig. 3-2 ). Epiophlebia superstes was recovered as the sister lineage to the remaining Epiprocta. All anisopteran families with multiple representatives were found to be monophyletic. Anisoptera subdivided into two major clades. One consists of the families Austropetaliidae, Aeshnidae, Petaluridae, Co rdulegastridae, Chlorogomphidae and Gomphidae (node 22) and the second of the families Corduliidae and Libellulidae (node 41, Libelluloidea). Nearly half of the zygopteran families represented by multiple taxa were recovered as nonmonophyletic, and the families (Perilestidae + Sy nlestidae + Chorismagrionidae + Lestidea) formed a monophyletic clade sister to the remaining Zygoptera. Bayesian Analysis All data partitions including fossils were combined in a SA using the Mk model for morphology ( Ronquist et al., 2005 ). Modeltest selected GTR + I + G as the best-justified model from the ClustalX alignment. The first 1000 tree s were discarded as "burn-in" as indicated by graphing the generations in Mi crosoft Office Excel 2003. The majo rity rule topology recovered from Bayesian analyses with PPs > 90 is presented in Fig. 3-3 (exact PPs for each node in Fig. 33 are found in Appendix C). Monophyletic groupings of taxa corresponding broadly to the "family level" were largely c ongruent with those found in the DO topologies, with the exceptions of a non-monophyletic Corduliidae and relations hips among the megapodagrionid lineages. Relationships above the family level were la rgely variable when compared with the DO topology, although the monophyletic clad e of (Perilestidae + Synlestidae + Chorismagrionidae + Lestidae) was still supported as sister to all Zyg optera as was the relations hip of Platystictidae as the next most sister lineage to the remaining Zygoptera. Only under Bayesian analyses ( Figs 3-3 44


and 3-5 ) were the Aeshnidae + Austropetaliidae recove red as sister to all remaining Anisoptera. Relationships among taxa below the family level were also somewhat variable when comparing the DO and Bayesian topologies. Fo ssil ingroup and outgroup taxa tr aditionally placed as sister lineages to the Epiprocta + Zygoptera clade were combined in a polytomy of all outgroup taxa under Bayesian analysis (see below, Contribution of fossil taxa ). Parsimony Analyses MP analyses of the ClustalX alignment with gaps treated as both missing and 5th state characters are represented by shad ed (i.e., present in the MP anal yses) or non-shaded boxes (i.e., not present in the MP analyses) at each node (Figs 3-2 through 3 ). When gaps were treated as missing, 16 480 most parsimonious trees at leng th 28 753 were recovered (CI = 0.241, RI = 0.523). When gaps were treated as 5th states, 895 trees were recovered at length 33 798 (CI = 0.255, RI = 0.547). Resolution among relationships above the family level in the strict consensus topology were collapsed a great deal across th e topology when gaps were treated as missing, particularly among Zygoptera, which was represented as a co mplete polytomy sister to Epiprocta. These results tended to agree with a recent analysis ( Ogden and Rosenberg, 2007 ), which showed that treating gaps as a 5th stat e character was generally more accurate than treating them as missing data. When gaps from th e ClustalX alignment were treated as 5th state characters, the resolution within the strict consensus tree increas ed markedly, particularly among outgroup taxa, although Zygoptera still remained la rgely unresolved (these results are not presented here but are represented on Fig. 3-2 in the node legend for nodes 61, 70, 72, 75, 76 and 82). Additional Analyses (Fossils Excluded) In order to evaluate how the fossil taxa aff ected the overall topologies, the above analyses were "rerun" with all fossil taxa excluded. All topologies for ingr oup taxa were largely congruent 45


according to optimality criterion regardless of the in clusion or exclusion of all fossils (see below, Contribution of fossil taxa ). The DO search produced 16 most parsimoni ous trees at length 29 415 (CI = 0.345, RI = 0.614, Fig.3-4 ). When gaps were treated as missing for the ClustalX alignment, MP recovered four most parsimonious trees at length 28 673 (CI = 0.241, RI = 0.521). When gaps were treated as fifth states, 31 trees were recovered at length 33 707 (CI = 0.255, RI = 0.546). The Bayesian topology minus fossils ( Fig.3-5 ) was analyzed as described above and produced a topology nearly completely congruent with Fig. 3-3 Data Completeness, Support, Congruence and Gene Contribution Data completeness: Compression fossils were used to code the eight fossil taxa included in the analyses presented in Missing data among fossil taxa were extensive, not only due to the obvious lack of DNA data but for missing morphological data as well. Geropteron, Tupus/Megatypus and Oligotypus represented outgroup fossil taxa and had missing morphological data in the amounts of 55%, 53.5% and 59.5%, respectively. Ingroup fossil taxa included Ditaxineura, Kennedya, Tarsophlebia, Hete rophlebia and Liassoph lebia, and each had missing data in the amounts of 54%, 54%, 40% 45% and 45%, resp ectively. Generally, morphological data for fossil taxa are more comple te the more recent the fossil, but overall most fossils lacked morphological data for characters of the head and body as the wings are often well preserved. Extant taxa from the i ngroup with associated molecula r data represented 105 genera (). Of these 105 ingroup taxa, 100% had comple te or partial 18S and 28S data, 92% had sequence data for COII, 97% had sequence data for 12S and 16S, and 99% had H3 sequence data. Figs 3-2 and 3-3 Table 3-1 46


Support: As has been seen in other ordinal level insect studies ( ; ; ), overall results revealed highe st support was concentrated more at the apical nodes than at the basal nodes, for both optimality criteria used here. Topologies generated for each analysis recove red congruent relationships in monophyletic groupings when support values were also high (i .e., PPs were > 90, bootstraps (BS) were > 70 and summed partitioned Bremer support (BrS) we re > 10) regardless of mode of analysis (C). Wheeler et al., 2001 Robertson et al., 2004 Terry and Whiting, 2005 Appendix Congruence: There were very few differences betw een the topologies that included and excluded fossil taxa across a single analyti cal methodology. There was, however, significant difference in both the placement of fossil taxa and relationshi ps among family groups between analytical methodologies. Odonata (including the fossil taxa Ditaxineura Kennedya Tarsophlebia Heterophlebia and Liassophlebia ) was found to be monophyletic only in parsimony analyses (i.e., DO and MP) where gaps were treated as 5th state characters. Familial relationships were variable between methodol ogies, particularly among the Epiprocta, the megopadrionids and calopterygoids, where DO recovered many relationships that were completely unexpected and di fficult to synthesize (e.g., Cordulegaster + Petaluridae, the complete polyphyly of Calopterygoidea and Megapodagr ionidae, the paraphyl y of Perilestidae). All analyses performed with only the extant lineages supported a monophyletic Odonata regardless of optimality criterion. The suborder Epiprocta was recovered as monophyletic in all analyses. Zygoptera was also always recovere d as monophyletic except when fossils were included and gaps were treated as missing in the MP analysis. All families that could be tested for monophyly within Epiprocta were recove red as monophyletic with the exception of Corduliidae. Of the 15 zygopteran families that had multiple representatives, only seven were 47


recovered as monophyletic across a ll analyses. In general, Bayesi an analyses produced estimates of extant odonate relationships that are mo re in line with current hypotheses based on morphology. However, only MP and DO reconstructe d relationships among fossil taxa that are harmonious with current hypotheses. The inclusion or exclusion of one or more extant outgroup taxa was never performed among the finalized data set, thus no discussion as to the sensitivity of ingroup relationships to extant outgroup selection is included here. Howe ver, in preliminary DO analyses, where there were no morphological data included for outgroup taxa, there appeared to be no affect on ingroup relationships when extant outgroup taxa were included or excluded. There was no explicit test of sensitivit y of ingroup relationships to outgrou p selection performed during either preliminary or final analyses. Overall Gene Contribution Summing BrS values partitions across the DO topology and dividing these by the total number of parsimony in formative characters indicates the relative signal contribution of each gene across the topology ( ; ). The contributions of each partition were as follows: 12S = 11.4%; 16S = 16%; 18S = 5.2%; 28S = 45.5%; H3 = 5.7%; COII = 8.2%; and morphology = 5.8% ( C). All genes examined had appropriate variability in length, nucleotides and dispersed support and provided relatively equal contributions to the topology, except 28S. The ribosomal marker 28S contributed a large signal that was relatively well distributed across all nodes within the phylogeny. The 28S gene is a large and commonl y used gene within insect phylogenetics because its rate heterogeneity contributes info rmation at both higher and lower levels within phylogenies ( ; ; ). At Baker and DeSalle, 1997 Ogden and Whiting, 2005 Appendix Wheeler et al., 2001 Ogden and Whiting, 2003, 2005 Terry and Whiting, 2005 5% of the signal, 18S contribute s the least of all data partitions due to its conserved nature 48


within odonates, a result in stark contrast to Ephemeroptera, the other "primitive" winged insect group, where a molecular phylogeny of similar taxon sampling size and breadth and gene selection yielded a contribution of 23.7% of the overall signal to the topology ( ). Ogden and Whiting, 2005 Discussion Analyses supported several clades that have been generally rec ognized as monophyletic within Odonata ( Fraser, 1957 ; Carle, 1982 ; Bechly, 1996 ; Trueman, 1996 ; Rehn, 2003 ; Hasegawa and Kasuya, 2006 ). However, many other traditi onal groups were not supported. Examining only the work of Rehn (2003 ), Bechly (1996 ) and Trueman (1996 ) reveals a consensus between my results and theirs on th e monophyly of Odonata and Epiprocta, and the results of Rehn and Bechly for a monophyletic Zygoptera. There is also a strong consensus of the extant families within Epiprocta as monophyletic within the lite rature as a whole ( Carle and Kjer, 2002 ; Misof, 2002 ; von Ellenrieder, 2002 ). The paraphyly among zygopteran lineages is more extensive than what was observed by Rehn (2003 ), but several families were recovered as non-monophyletic only in the DO and MP analyses. Odonata Odonata, including the fossil taxa Kennedya Ditaxineura and Tarsophlebia were supported (node 14, Fig. 3-2 ; BS = 86) and recovered only under DO and MP when gaps were treated as 5th state characters. All extant odonate taxa that were sampled were strongly supported as a monophyletic group (node 12, Figs 3-4 and 3-5 ; BS = 100; BrS = 229; PP = 81); this was recovered under all criteria. Ther e is little debate in the literat ure that Odonata are monophyletic and that Kennedya and Ditaxineura make up the sister lineages of Odonata. The placement of the fossil taxon Tarsophlebia as sister to the Epiprocta + Zygoptera ( Bechly, 1996 ; Trueman, 1996 ; Rehn, 2003 ; Fleck et al., 2004 ) or as sister only to the Epiprocta 49


( Carle, 1982 ; Nel et al., 1993 ; Bechly, 1995 ) is debated. Tarsophlebia was supported as sister to Epiprocta and Zygoptera by Ba yesian methods (node 16, Fig. 3-3 ; PP = 98) and MP when gaps were treated as 5th state characters, as sister to Epiprocta under DO (node 18, Fig. 3-2 ) and its relationship was unresolved in MP when ga ps were treated as missing. Owing to the unconventional arrangement among some of the ex tant higher level rela tionships from the DO tree and the apparent difficulty of Bayesian me thods to deal with "basal" fossil ingroup and outgroup taxa, a definitive placement of Tarsophlebia is still elusive. Epiprocta The monophyly of Epiprocta was recovered under all modes of analysis investigated when fossils were included (node 18, Fig. 3-2 ; BS = 80; BrS = 16; PP = 77) and when fossils were excluded (node 13, Fig. 3-4 ; BS = 100; BrS = 229; PP = 100). Epiophlebia was sister to the remainder of Epiprocta in all analyses. The fossil group Heterophlebia + Liassophlebia (node 20, Fig. 3-2 ; PP = 99; BS = 77; BrS = 1) was placed as sister to the Anisoptera (node 19, Fig. 3-2 ; BS = 70; BrS = l; PP = 71) under all analyses with the exceptions of MP when gaps were treated as missing. Thus, these results corroborate Ep iprocta (Anisoptera + Epiophlebiidae) as designated by Lohmann (1996 ) and recognized by Rehn (2003 ). Anisoptera All families within Anisoptera represented by multiple taxa, except Corduliidae, were recovered and strongly supporte d as monophyletic acro ss all topologies. Ou tside the general trend of the libelluloids being more closely rela ted to the cordulegasterids and chlorogomphids than other anisopteran families and Petaluridae being the sister group to all other Anisoptera, there is no clear pattern of relationship between the familial lineages that emerges from the literature (i.e., between Petaluri dae, Aeshnidae and Gomphidae; Fraser, 1954, 1957 ; Hennig, 1981 ; Carle, 1982 ; Pfau, 1991 ; Bechly, 1996 ; Lohmann, 1996 ; Misof et al., 2001 ; Rehn, 2003 ). 50


There were two hypotheses of familial relationship for Anisoptera that reflected the analytical methodology used (DO versus Bayesian). Under DO there was a bifurcation of two major groups (node 21, Fig. 3-2 ), one composed of the families Gomphidae, Austropetaliidae, Aeshnidae, Chlorogomphidae, Cordulegastridae and Petaluridae, and the other composed of the libelluloid families Corduliidae, Macromiidae, Libe llulidae and Synthemistidae (node 21, Fig. 3-2 ; node 14, Fig. 3-4 ), a result that is in line with Hasegawa and Kasuya (2006 ). Topologies for Anisoptera recovered from Bayesian (node 21, Fig. 3-3 ; node 14, Fig. 3-5 ) differed by placing all families in a pectinate assemblage with Aeshnidae + Austropeta liidae as sister to all other families and the libelluloids (node 41, Fig. 3-3 ; node 34, Fig. 3-5 ) as the most derived lineage of Anisoptera. The relationship of Aeshnidae as sist er to a pectinate assemblage of anisopteran lineages is a novel hypothesis for the group. Neither of the backbones for either topology (nodes 22, 23, 28, Fig. 32 ; 127, 128, Fig. 3-3 ) had high support, making it difficult to accept one hypothesis over another. However, the Bayesian topologies corres ponded better to anisopt eran "morphological transitions" involving characters of wing venation and the naiad labium. For example, the primitive condition of the labium is having the labium retracted directly under the head with two stout lateral lobes used to seize and subdue prey when the labium is protracted. The more derived condition is having the lateral lobes of the labium forming a "mask" that fits tightly over the face and just under the eyes. The Bayesian topologies ( Figs 3-3 and 3-5 ) suggested a gradient of intermediates between the most derived naiad labial mask and the most primitive retracted labium. The Bayesian topology also appears to be more in line with previous hypotheses of Anisopteran phylogeny ( Fraser, 1954, 1957 ; Bechly, 1996 ; Misof et al., 2001 ; Rehn, 2003 ), especially in reference to m odern hypotheses involving relations hips between Libelluloidea, Chlorogomphidae and Cordulegastridae ( Carle, 1982 ; Rehn, 2003 ). Unfortunately, the 51


evolutionary position among the families of Anisopte ra was not clearly resolved here and may be difficult as a result of a deep pectinate evolutionary history ( Ogden and Rosenberg, 2006 ). Nonetheless, two hypotheses were established that can be further tested in future research ( Fig. 3-6 ). Superfamily and Family Monophyly The superfamily Aeshnoidea has been proposed to include the families Aeshnidae, Austropetaliidae, Gomphidae and Petaluridae or only Aeshnidae, Austropetaliidae and Petaluridae. Data from this study do not support a monophyletic Aeshnoidea. Anisopteran superfamilies that have been proposed and were found to be monophyletic include Cordulegastroidea, Gomphoidea and Libelluloidea. Cordulegastroidea were represented by two taxa in this analysis and was placed as sister to the libelluloids (nodes 132 and 116, Figs 3-3 and 3-5 respectively) or as sister to the Petaluridae (nodes 28 and 21, Figs 3-2 and 3-4 respectively). Gomphoidea consisted solely of the family Gomphidae and was cl early monophyletic. Though its taxonomic elevation as a superfamily and phylogenetic position have been contentious ( Fraser, 1954, 1957 ; Bechly, 1996 ; Misof et al., 2001 ; Hasegawa and Kasuya, 2006 ), the groups is well characterized by a club tail (present to some extent in all taxa) and possess several primitive characters (e.g., eyes separated). Go mphoidea (i.e., Gomphidae) was placed as the sister taxa to Libelluloidea + Cordulegastroidea (nodes 128 and 112, Figs 3-3 and 3-5 respectively) or as sister to the clade containing the members of "Aeshnoidea" and Cordulegastroidea (nodes 22 and 15, Figs 3-2 and 3-4 respectively). The superfamily Libelluloidea was str ongly supported as monophyletic (node 41, Fig. 3-2 ; BS = 100; BrS = 80; PP = 100). The major morphological character defining the Libelluloidea is the anal loop that provides support to the anal re gion of the hind wing. It is constructed of veins forming a "boot" shape that is present to different degrees in the majority of taxa within the 52


superfamily. It is hypothesized that the more primitive lineages within the group have an incomplete boot (i.e., missing the "toe" or "foot region of the boot altogether) and the more derived lineages have a fully complete boot shap e (i.e., having a well developed "heel and toe"; Carle and Kjer, 2002 ). The "boot" was a consistently recovered character for the superfamily and when comparing the differing t opologies (with the exception of Fig. 3-2 ) it appeared an incomplete boot is the more primitive condition with a more complete boot (i.e., having a well developed "heel and toe") being the more de rived condition and common among the libellulids. A major point of contention among authors concerning Epiprocta is the monophyly and relationships between the libelluloid families (e.g., the monophyly of Corduliidae and the familial rank of Synthemistidae and Macromiidae). Libellulidae was strongly supported as monophyletic by all analyses (node 42, Fig. 3-2 ; BS = 100; BrS = 77; PP = 100) as were the synthemistids (node 56, Fig. 3-2 ; BS = 100; BrS = 42; PP = 100). The elevation of the synthemistids to familial status, at least as it is currently defined, was not supported as it is consistently placed within or sister to severa l "corduliid" lineages. Macromiidae's placement was variable between topologies and is also difficult to define. Co rduliidae (if Synthemistidae is considered a subfamily) was monoph yletic only under the DO analysis when fossils are included (node 50, Fig. 3-2 ; BrS = 10). Research on the phylogeny of the libelluloids is already under way and includes a more extensive taxon sampling th an presented here and should provide more resolution and evolutionary insight (J. Ware, personal communication). Eleven of 12 extant families of Epiprocta were represented in this analysis. Aeshnidae, Petaluridae, Gomphidae, Synthemistidae and Li bellulidae were all represented by two or more taxa and were recovered as m onophyletic lineages with strong support. Corduliidae was also represented by multiple taxa, but was recovered as non-monophyletic among all but one analysis 53


(see Fig. 3-2 ) and was never supported by high br anch support. Austropetaliidae, Chlorogomphidae, Cordulegastridae, Macromiidae and the monoge neric Epiophlebiidae were all represented by a single taxon, thus the monophyly of these groups could not be tested here. Zygoptera Rehn (2003 ) found the monophyly of the Zygoptera to be supported by nine morphological synapomorphies. Previous molecular phylogenetic analyses focused on Odonata have failed to recover the monophyly of Zygoptera ( Saux et al., 2003 ; Hasegawa and Kasuya, 2006 ). However, the SA analysis of molecular + morphological data and the SA analysis of just the molecular data (tree not shown) support the monophyly of Zygopter a. Our results suggest that a paraphyletic Zygoptera is an artifact of limited taxon sampli ng within the suborder, the use of a single molecular marker, and failing to perform combined molecular analyses. In the analyses presented here, the monophyly of Zygoptera was recovere d under both phylogenetic criteria (node 61, Figs 3-2 and 3 ; BS = 78; BrS = 4; PP = 94; node 54, Figs 3-4 and 5 ; BS = 100; BrS = 10; PP = 100). While Anisoptera appears to have two welldefined hypotheses for the relationships among families (Figs 3-2 through 3-5 ), relationships between zygopteran families were more variable when comparing all topologies. A zygopteran relationship that wa s constant among all analyses was the clade composed of the lestid-lik e families (Perilestidae + Synlestidae + Chorismagrionidae + Lestidea) as the sister group to all other Zygoptera. While, recent molecular analyses have placed the Les tidae as sister to the Epiprocta ( Saux et al., 2003 ; Hasegawa and Kasuya, 2006 ), lestid-like damself lies have long been thought to be ancient. Fraser (1957 ) proposed a similar superfamily containing all lestid-lik e damselflies and Lestinoidea, and placed it sister to Calopterygoidea + Epiprocta based on characteristics of the RP midfork that he perceived as primitive ( Rehn, 2003 ). 54


Superfamily and Family Monophyly All zygopteran superfamilies, with the exception of Calopt erygoidea as described by Bridges (1994 ) but including Lestoidea (nodes 157 and 147, Figs 3-3 and 3-5 respectively), were found to be non-monophyletic. Calopterygoid ea is supported as monophyletic only under Bayesian analyses when fossils are excluded (node 147, Fig. 3-5 ) with high support (PP = 100). Calopterygoidea was recovered as completely polyphyletic unde r DO and MP methods. Dumont et al. (2005 ) also reported recovering the monophyl y of Calopterygoidea only when using a Bayesian or Likelihood approach. Rehn (2003 ) found Calopterygoidea (minus Amphipterygidae) to be monophyletic, defined by two synapomorph ies, (1) RP3,4 Straight with RP1,2 branching anteriorly, and (2) crossveins present in the RP -MA space between the arculus and the distal end of the quadrangle. Amphipterygidae's taxonomi c placement within the Calopterygoidea and the family's definition has been contentious. In th e past Amphipterygidae has included the genera Philoganga, Diphlebia Amphipteryx Pentaphlebia Devadatta and Rimanella More recently the family has been defined as only Amphipteryx Pentaphlebia Devadatta and Rimanella based on the presence of gill tufts in the larval stage ( Novelo-Gutierrez, 1995 ). The other two genera, Philogonga and Diphlebia either have been left within the Amphipterygidae as a separate subfamily (Philogonginae and Diphlebiinae; Fraser, 1957 ; Bridges, 1994 ) or placed in a unique family (Diphlebiidae) along with Lestoidea ( Novelo-Gutierrez, 1995 ). Several of these genera were included in this analysis ( Philoganga Diphlebia Devadatta Rimanella and Lestoidea). Philoganga Devadatta and Rimanella had variable placements between topologies, but in general they tended to pair toge ther (though never all in the same clade) and nest with either Megapodagriondae or Calopterygoidea. The relati onship between these genera may gain support with the addition of molecular data for Rimanella a taxon represented only by morphological data However, Diphlebia and Lestoidea form a well supported clade and are strongly supported 55


as sister to the Euphaeidae acr oss all topologies (nodes 84, 78 on Figs 3-2, 3-3 and 3-3, 34 respectively) A notable, though not well supporte d, clade was the placement of Rimanella + Devadatta with the Diphlebia, Lestoidea + Euphaeidae clade (node 148, Fig. 3-5 ). Polythoridae is placed within Megapodagrionidae under DO ( Figs 3-2 and 3-4 ). However, Bayesian analyses ( Figs 3-3 and 3-5 ) resulted in the placement of Polythor idae within the Cal opterygoidea, leaving a largely monophyletic Megapodagrionidae. The Lestinoidea have traditionally included the lestid-like fam ilies (Perilestidae + Synlestidae + Chorismagrionidae + Lestidae), Megapodagrionidae, and at times Lestoideidae (defined as Lestoidea Philoganga and Diphlebia ). Rehn (2003 ) recovered Megapodagrionidae as a pectinate assemblage leading to Lestinoi dea + Coenagrionoidea. No analyses performed here recovered Megapodagrionidae as sister to the lestid-like damsel flies, but rather as a sister group to the calopterygoids (Bayesian analyses, Figs 3-3 and 3-5 ) or at least a sister to a lineage of the polyphyletic Calo pterygoidea (DO analyses, Figs 3-2 and 3-4 ). Megapodagrionidae was never recovered as monophyletic in either Bayesian or DO topologies, due to the monospecific family Pseudolestidae (i.e., Pseudolestes) or several amphipterygids, Polythoridae and Pseudolestidae nesting within, respectively. Fraser (1957 ) removed Pseudolestes from the Megapodagrionidae along with seve ral other genera because they represented "strictly isolated forms" that made it difficult to give a strict definition to Megapodagrionidae. However, if Pseudolestes is considered a megapodagrion, Me gapodagrionidae was monophyletic among the Bayesian topologies and had a PP of 100 when fossils were excluded (node 139, Fig. 3-5 ; node 154, Fig. 3-3 ). Perilestidae and Synlestidae, members of the le stid-like damselflies were recovered as nonmonophyletic. The paraphyly of Perilestidae is doubtful, as there are only two extremely similar 56


genera in this family. The paraphyly of the perilestids is most likely due to a lack of data overlap in the genes best suited to address questions of close relationship (C OII, H3, 12S, 16S; see Table 3-1 ). Chorismagrion risi represents the monospecific family Chorismagrionidae and was placed within Synlestidae with very high support. Chorismagrion risi as a member of the Synlestidae has also been suggested based on similarities of th e agnostic behavior of the final-instar larvae ( Rowe, 2004 ). Chorismagrionidae has been grouped with Hemiphlebia mirabilis representing the monospecific family Hemiphlebiidae in the superfamily Hemiphleboidea ( Bridges, 1994 ). Neither Rehn (2003 ) nor our results support C. risi as a member of the Hemiphleboidea. Although Hemiphlebia mirabilis is not included here, ot her morphological evidence ( Rehn, 2003 ) suggests it may simply be a member of the primitive lestid -like damselflies. Coenagrionoidea was never supported as monophylet ic in these analyses due to the family Platystictidae being placed as th e closest sister lineage to the re st of Zygoptera af ter the lestidlike damselflies. Platystictidae is unique among petiolate "coenagr ionoid" lineages and as a result was removed from the Protoneuridae and elevated to familial status by Tillyard and Fraser, 1938 ). To our knowledge, the placement and support of Platystictidae as sist er to the rest of Zygoptera has never before been hypothesize d. Other coenagrionoid lineages supported as monophyletic include Isostictidae, Pseudostigmatid ae and Platycnemididae (in the Bayesian topologies, Figs 3-3 and 3-5 ). Members of Pseudostigmatidae form a spectacular group of large forest canopy dwelling zygopterans from Sout h America. Although clearly a monophyletic group, the rank of Pseudostigmatidae as a family ma y be in question as it consistently places within or next to lineages traditionally assigne d to Coenagrionidae. Recently a SA analysis of molecular and morphological data r ecovered the East African species Coryphagrion grandis as sister to the Pseudostigmatidae ( Groeneveld et al., 2007 ). This species occupies an exceptionally 57


similar habitat and morphological features with pseudostigmatids, as well as the similar breeding and feeding behaviors. Although, Groeneveld et al. (2007 ) did not make an effort to place the group within a superfamily, Rehn (2003 ) also recovered C. grandis + Pseudostigmatidae and placed the group within the Coenagrionoidea. Isostictidae was supported as a monophyletic lineage sister to the coen agrionoid lineages (minus Platystic tidae) across all topologies (except Bayesian analyses where fossils were excluded) and was placed as the sister lineage to all remaining coenagrionoid lineages. Rehn (2003 ) also found Isostictidae to be monophyletic as supported by a single synapomorphy from a la rval character (prementum with many long raptorial setae). Isostictidae was raised to familial status without any precise evidence ( Lieftinck, 1975 ) and the placement of Isostictidae here is completely novel. Platycnemididae's classification has been problematic in the litera ture being defined as bot h a constituent of the subfamily Coenagrioninae ( Munz, 1919 ) and a family ( Selys, 1863 ). The paraphyly of the family recovered herein under DO was also recovered by Rehn (2003 ); however, the monophyly of Platycnemididae is recovered under Bayesian an alyses and with high branch support (PP = 92) when fossils are included ( Fig. 3-3 ). One relationship involving Platycnemididae that is highly supported across all topologies is th e sister group relationship with the Phylloneura and Nososticta a lineage of the paraphyletic "Protoneurid ae". Protoneuridae is not recovered as monophyletic in any analyses and is separated into two lineages, one from the Old World that is sister to the Platycnemididae a nd the other from the New World whose placement is difficult to resolve. In all, 19 of the 22 extant families of Zygoptera are represented in this analysis. Lestidae, Platystictidae, Chlorocyphidae, Calopterygidae, Euphaeidae, Po lythoridae, Isostictidae and Pseudostigmatidae were all represented by two or more taxa and were recovered and strongly 58


supported as monophyletic lineages. Two other families represented by two or more taxa, Perilestidae and Platycnemididae (discussed ab ove), were recovered as monophyletic only under a Bayesian framework. Chorismagrionidae and Pseudolestidae are monospecific families and Diphlebiidae and Lestoideidae are monogeneric fa milies. Each family was represented by only one taxon in this analysis; thus, the monophyly of these families coul d not be specifically tested. However, results suggest that some of these families are simply apomorphic genera within existing families (e.g., Pseudolestes as a member of the Megapodagrionidae). Contribution of Fossil Taxa On the whole, fossils did not have a larg e effect on the overall topology or on ingroup relationships. Moreover, due to la rge amounts of missing data that broke up several relationships among the outgroup taxa (e.g., the monophyly of Plecoptera, node 125, Fig. 3-3 ), fossil taxa appeared only to interfere w ith the Bayesian analysis ( Fig. 3-3 ). MrBayes failed to recover several nodes where fossil taxa had only partia l, more plesiomorphic morphological data (e.g., Meganeura and Kennedya). This "interference" of fossil taxa is likely due to the problem of missing data inherent to most very old comp ression fossils and methods that because of a statistical nature require th ere to be only nominal amounts of missing data. The ingroup relationships do not appear to be affected by fossil taxa (e.g., Liassophlebia ) nor by other extant taxa that are represented exclusively by more complete morphological data (e.g., Coenagrion). Thus, although there was not a specific test performed here, analyses based on a parsimony criterion (DO and MP) appeared to not be that affected by la rge amounts of missing data among more unique taxa. The parsimony analyses appeared to be less sensitive to missing data than the Bayesian analyses in that the parsimony tr ees had a much more robust placement of both primitive ingroup and outgroup fossil taxa. These observations make it reasonable to proceed to study the group's classification and evolution of morphological featur es by further inclusion of 59


many more fossils under a parsimony framework. Ou r results suggest that questions addressing the evolution of odonate morphologi cal or behavioral features (e .g., wing veins or flight) must include a broad representation of fossils from the group's exceptional fossil history. Recommendations for Classification Based on the Phylogenies Based on the results of all analyses taken t ogether the following taxonomic refinements are recommended (indicated as on Figs 3-2, 3-3 and 3-6 ). Anisoptera: the superfamily Gomphoidea should be formally re cognized as it is a distinct and well supported lineage among all analyses performed; Synthemistidae is always strongly supported as sist er to several lineages of Corduliidae and should be considered a subfam ily (i.e., Synthemistinae) until further research, including a broader taxon sampli ng, can elucidate the higher-level classification of this group; Macromiidae should be a subfamily of the corduliids for the same reasons. Taxonomic refinements for Zygoptera at this point include the formal synonymizing of the monospecific families Pseudolestidae Kirby (1900 ) and Chorismagrionidae Tillyard and Fraser (1938 ) within the families Megapodagrionidae Tillyard (1917 ) and Synlestidae Tillyard (1917 ), respectively; the superfamily Lestinoidea being strictly defined as the lestid-like families Perilestidae + Synlestidae (including Chorismagrion ) + Lestidae and may likely include Hemiphlebiidae; Pseudostigmatidae should forma lly be assigned to the Coenagrionoidea. Conclusions The research represents the first overall comprehensive morphological and molecular phylogenetic analysis of odonate phylogeny. The analysis includes 30 of the 34 extant families, representing 20% of the genera. Thirteen families, re presented by multiple taxa, were supported as monophyletic and eight as non-monophyletic, a lthough two of the eight were recovered as monophyletic under a Bayesian criterion. Seven fa milies were represented by only one species and monophyly was not testable. Epiprocta and Zygoptera were supported as monophyletic 60


under all criteria (except Zygoptera under MP when gaps were treated as missing). Ditaxineura is supported as the sister lineage to the rest of Odonata under DO. Epiophlebiidae and Lestinoidea (the lestid-like damselflies) are sister lineag es of the remaining Epiprocta and Zygoptera, respectively, across all analyses. In general, relationships higher than familial but below subordinal rank did not have hi gh support. The inclusion of one or more genes capable of recovering and/or providing addi tional support to deeper relations hips will be important for future analyses especially within the Zygoptera. Fossil taxa did not appear to provide signa l imperative to recove ring a robust phylogeny, however, they were crucial fo r understanding the evolution of key morphological wing features and are likely just as crucial fo r understanding the evol ution of other morphol ogical features (see Chapter 4). As a basal-winged insect group, Holodonata (both fossil and extant odonates) holds the key to many interesting and important questions concerning insect classification and insect flight, its origins and its evolut ion. A broader taxon sampling of ex tant and fossil taxa as well as additional data, both morphological and molecular, will be necessary and is already under way by the authors in an attempt to understand the co mplex evolutionary history of Odonata as well as the evolution of the odonate wing throughout their more than 300 million year history. 61


Figure 3-1. Representations of previous hypothese s for the relationships between subordinal groups (Zygoptera and Epiprocta), basal superfamily groups within Zygoptera and the basal family group within the Epiprocta a nd Anisoptera. All three hypotheses recover Epiophlebiidae as sister to a monophyletic Anisoptera (creating Epiprocta) and the basal anisopteran as Petalu ridae (shown in gray). indicates non-monophyly of groups. (a) Represents the classification of Bechly (1996) showing a monophyletic Zygoptera with a perfect bifu rcation of the major lineag es (note that Bechlys classification scheme for zygopteran s ubfamilial groupings varies from current schemes, especially for Amphipterygoi dea). (b) Hypothesis of Trueman (1996) placing Perilestidae (i.e., Perilestes ) as sister to a non-m onophyletic Zygoptera and Amphipterygidae (i.e., Amphipteryx ) as sister to Epiproct a. (c) Hypothesis of Rehn (2003) recovering a monophyletic Zygoptera and non-monophyletic Calopterygoidea (i.e., Diphlebia + Philogenia ) as the most inclusive lineage of the group. (d) Molecular results of Hasegawa and Ka suya (2006) str ongly supporting a nonmonophyletic Zygoptera, the Coenagrionoidea as sister to the remaining Zygoptera and the Lestidae as sister to Epiprocta. 62


Figure 3-2. Single most parsimoni ous tree recovered from a simultaneous analysis with fossil taxa included of all data partitions and in POY (Direct Optimization) under an equal weighting scheme and 250 replicates (L: 29460; CI: 0.345; RI: 0.614). Exact nodal supports are listed in Appendix C. Non-monophyletic groups are indicated by suggested updates to the classification ar e indicated by (see Recommendations for classification based on the phylogenies ), and indicates a fossil taxon. 63


Figure 3-3. Bayesian analysis with fossil taxa included resulting from a simultaneous analysis of all data partitions. Majority rule consensus of topology ge nerated via MrBayes with 6 500 000 generations using the model GTR + I + G for molecular data sets and the Mk Model for Morphology. Nodes found only under Bayesian analysis (i.e., not recovered in Fig. 3-2) are outlined in gra y. Exact supports for all nodes are listed in Appendix C. Non-monophyletic groups are indicated by suggested updates to the classification are indicated by (see R ecommendations for classification based on the phylogenies), and indicates a fossil taxon. 64


Figure 3-4. Strict consensus tree of 16 most parsimonious trees fr om a simultaneous analysis of all data partitions when fossil taxa were excluded in POY (Direct Optimization) under an equal weighting scheme and 250 re plicates (L 29415, CI 0.348, RI 0.617). Exact nodal supports are listed in Appendix C. 65


Figure 3-5. Bayesian analysis resu lting from a simultaneous analysis of all data partitions with fossil taxa excluded. Majority rule consen sus of topology generated via MrBayes with 6 500 000 generations using the model GTR + I + G for molecular data sets and the Mk Model for Morphology. Nodes found only under Bayesian analysis (i.e., not recovered in Fig. 3-4) are outlined in gra y. Exact supports for all nodes are listed in Appendix C. 66


Figure 3-6 The two major hypotheses resulting from simultaneous analyses under both parsimony, Hypothesis I (MP and DO) and likelihood, Hypothesis II (Bayesian) optimality criteria. Ditaxineura and Kennedya were left out Hypothesis II as fossil outgroup and basal fossil ingroup taxa were difficult to resolve due to missing data. Non-monophyletic groups are indicate d by suggested updates to the classification are indicated by *(see Recomme ndations for classi fication based on the phylogenies), and indicates a fossil taxon. 67


Table 3-1. List of all taxa used in this an alysis with GenBank accession numbers 68


Table 3-1 Continued 69


CHAPTER 4 ODONATA WING EVOLUTION Introduction Odonates are well known for their ability as ae rial acrobats. Unlike most other flying insects, odonates possess a direct mechanism of flight in which flight musculature connects directly to the base of the wi ngs. The contraction of specific muscles attaching directly to the wing results in the wing being raised and lowered. This mechanism of flight is cont rasted against the indirect flight mechanism in which the wi ngs are raised and lowe red by a combination of muscle contractions and the flex produced in the cuticle of the insect thorax ( Chapman, 1982 ). Because of the relative simplicity of the direct flight mechanism, and because they are among the most primitive insect lineages with flight, odona tes have long been the focus of biomechanical studies of insect flight. Odonatoids have been aerial predators since at least the Carbonifer ous and today exhibit one of the most advanced forms of flight of any organism ( Wootton et al., 1998 ). Nearly all insects have muscles that attach directly to the wing base (direct flight musculature) that plays a part in their flight ability. However, odonates are one of only tw o insect orders, the other being Blattodea, that make use of direct flight musculature to drive the wing's down stroke. The aptitude of odonates for fast forward-thrust flight powered by direct flight musculature, coupled with the ability to hover in mid-air and their relatively large size make them a favorite study organism for biomechanical studies oriented around insect flight ( Wootton, 1991 ; Sudo et al., 1999 ; Kesel, 2001 ; Wootton, 2002, 2003 ). All odonates owe their flight ability to "smart" mechanisms, such as the nodus, that provide rigid support while maximizing aerodynamic properties of the wing ( Wootton et al., 1998 ). The structure and function of odonate wing elements, such as the nodus and pterostigma, has been reviewed in the literature ( Wootton, 1991, 70


1992 ) and is fairly well understood in the biomechan ic sense. However, the evolution of these wing elements has not been investigated from a phylogenetic perspectiv e, nor are there any testable hypotheses to date that have explored the possible linkage of multiple wing elements. Herein, is a simple, yet novel approach to test for potential evolutiona ry linkages between wing elements (e.g., the nodus, pterostigma and elements of the wing articulation) by optimizing these characters across a topology. To my knowledge th is represents the first attempt to produce testable hypotheses for wing struct ures that are linked throughout an insects group's evolutionary history. Wootton (1992, 2002, 2003 ) and Wootton et al. (1998 ) have appealed for just such an investigation to be undertaken in order to better understand th e interaction between the wing elements. Wootton (1991 ) and Wootton et al. (1998 ) identified several odonate wing features that drastically improve the aerodynamic properties of the wing through increased structural support; they referred to these as "smart mechanisms". These include the nodus, which serves as a major structural support along the leading edge of the wing and the pterostigma, which reduces vibration at the wing tip dur ing flight. An investigati on of the congruence between morphological elements of the wing led to the identification of character complexes that appeared to have played a major role in the evolution of the odonate wing. Materials and Methods Character Optimization (Smart) I have coded two of these smart mechanisms (represented by characters 32, 33, 35 in Appendix A and described below) and additional mo rphological characters a ssociated with flight (characters 16, 30, 31 Appendix A) for all 130 taxa from the analysis carried out in Chapter 3 and mapped them on the DO topology ( Fig. 3-2 ), as this topology br oadly represents the subordinal relationships found amon g all analyses and is more resolved than the Bayesian 71


topology when fossil taxa are included. The ch aracter mapping for the character to be investigated was unambiguous and is illustrate d on a simplified version of the DO topology ( Figs 4-1 and 42) Character Optimization (Wing Form) Characters associated with overall wing form an d mechanistic characters associated with the uniqueness of Odonata flight (e.g., nodus and wi ng articulation) were included within the SA analysis and optimized on the DO topology using th e default parameter settings in MacClade 4.06 OS X. Only if MacClade recovers an ambiguous character mapping is an alternative optimization (ACCTRAN or DELTRAN) required. Characters associated with wing form and mechanics that were optimized were: costal axalar e (character 16); costal triangle (character 30); flexion line between distal edge of BxC and costal margin (character 32); primary and secondary braces of nodus (nodal crossvein and subnodus) (c haracter 33); Pterostigma (character 35). Results and Discussion Pterostigmanodal Brace Complex Optimizing the three states coded for the primary and secondary braces of the nodus (character 33) as well as the thr ee states of the pterostigma (cha racter 35) on the topology reveals a tight evolutionary congruence (i.e., optimizing together on the same nodes of the topology) between these two distantly located wing structures ( Fig. 41). The nodus is the main structural element along the leading edge of the wing and serves to brace the wing while twisting during flight. The two crossveins (nodal and subnodal) coded for in character 33 serve as major structural supports to the nodus. The nodal support s are absent in the out group taxa (character state 0), while character state (1) developed in ScP-RA and RA-RP1 spaces but not aligned, is present in Ditaxineura Character state (2) well developed and aligned, is present in the clade Kennedya + Tarsophlebia + Epiprocta + Zygoptera, which co mprises the best fliers of the 72


modern odonates. This progression from the absence of supports for the nodus to the presence of supports show an increasing trend for more structural rigidity within the wing ( Fig. 4-1 ). The pterostigma is usually identifiable as a colored cell located just behind th e leading edge in the distal portion of the wing. The pterostigma serves to reduce the self-excited vibrations in the distal portion of the wing during wing flapping and has been show n to raise critical speed by 10 25% in one species of dragonfly ( Norberg, 1972 ). The pterostigma is absent in the outgroup taxa and present in two forms in the ingroup: (state 1) present in C-RA and RA-RP spaces in the Ditaxineura and (state 2) present in only the C-RA space in the clade Kennedya + Tarsophlebia + Epiprocta + Zygoptera ( Fig. 4-2). Both of the above characters were coded as unordered multistate characters. Costal Wing Base & CostalScP Junction Complex Also closely linked to the evolution of the wing are other charac ters located at or near the wing articulation and the nodus. Ther e are four unreversed synapomor phies from wing structures that support the monophyly of th e modern Odonata (defined here as all modern groups plus Tarsophlebia ; Fig. 3-2 node 16): costal axalare is fully fused, costal triangle is fully formed with the anterior subcosta (ScA) completely fused to co stal margin, presence of a flexion line between the distal edge of BxC and the costal margi n, and the complete formation of the nodus by the posterior subcostal vein turni ng sharply forward to meet the costa at nearly a right angle (characters 16, 30, 31, 32, respectively) All of these character states are derived and appear to contribute to an increased wing pivoting ability a nd structural support within regions of the wing ( Fig. 42). These characters arise in unison among the modern Odonata ( Fig.4-2 ), possibly in response to a more well developed nodal bracin g structure (character 33) and pterostigma (character 35) that allowed fo r higher performance flight ( Fig. 4-1 ). In addition, all of these wing characters (16, 30, 35) appear linked throughout the evolutionary history of Odonata, 73


attaining their most derived st ates in modern odonates. Take n together, the pterostigmanodal brace complex and the costal wing base and costalScP junction complex account for a large part of the flight performance ga ins seen in modern Odonata. The wing characters associated with the nodus and pterostigma appear to arise in unison and show a directional progression toward wing m odifications that produce wing rigidity capable of supporting higher performance flight than was theoretically possible with preexisting wings. The suite of wing characters asso ciated with the costal wing base and the costalScP junction complex arise basal to this clade and also appear to arise in unison, thus obscuring a picture of sequential addition of wing character specializati on. This is likely an artifact of limited taxon sampling of the older lineages (f ossils) within Holodonata (= Odona ta and their extinct relatives). A large number of wing vein characters used in the analysis plotted well inside the ingroup, defining most of the extant odonate lineages. Th ese characters appear to represent additional refinements of the modern odonate wing follo wing the apparent basal origins of wing modifications that provided both additional strength and rigi dity to the wing. Conclusions This research produced several preliminar y testable hypotheses concerning the odonate wing; however, there are more complex hypotheses of wing and flight evolution to be addressed. For example, based on the structural elements of the odonate wing, how many modes of flight are there among modern taxa and wh at insight does the fossil record provide as to their origins and evolution? Which morphological features of the wing came first and allowed for further refinements of other structural wing features? At what rate do wing elements appear to evolve? Many of these questions can be addressed by a more refined coding of wing elements and wing forms (e.g., What defines a broad-based and petio late wing? How many origins are there for both wing types?). 74


Figure 4-1. The optimization of structural wing characters representing the Pterostigmanodal brace complex across a reduced DO phylogeny. The most ancestral character state (0) is shown in orange for all characters, st ate 1 in blue and state 2 in green. The optimization for both the pterostigma and the primary and secondary brace veins of the nodus corresponding to characters 33 and 35 is shown. See Appendix A for characters referenced here.* this wing is from Meganeuropsis americana which was not included in this analysis but repres ents a more complete wing upon which to illustrate the character states. All wings we re modified from Carpenter (1992) and Rehn (2003). 75


Figure 4-2. The optimization of st ructural wing characters repres enting the costal wing base & costalScP junction complex across a simplif ication of Fig. 3-2. The most ancestral character state (0) is shown in orange and st ate 1 in blue for all characters. This figure illustrates the optimization of AxC, costal tria ngle, flexion line between distal edge of BxC and costal margin, and the junction of costa and ScP corresponding to characters 16, 30, 31, 32, respectively. See Appendix A for characters referenced here.a The articulation of Tarsophlebia is unknown.* This wing is from Meganeuropsis americana, which was not included in this analys is, but represents a more complete wing on which to demonstrate the character states. All wings and wing articulations were modified from Carpenter (1992) and Rehn (2003). 76


CHAPTER 5 HOLODANTA PHYLOGENY Introduction Members of superorder Holodonata represent a spectacular diversity of ancient dragonflylike insects. The diversity of structure in wing and body shape is spectacular as are the behaviors exhibited among modern holodonates. They are most well known for their secondary genitalia and sperm competition, though their unmatched ability to fly and prowess as areal predators has not gone unnoticed. Holodonata has the most diverse and exquisite ly preserved fossil record of any insect group, with thousands of known compression foss ils stemming from the Carboniferous (>300 MYA) to the Miocene (<30 MYA), and more th an 70 known from amber. The large holodonate fossil record has a broad representation of fossils that fall bot h within and basal to extant groups, permitting an extensive survey of wing vein and wing structure evolution. Holodonata is composed of two orders: Prot odonata and Odonata. Protodonata is known only from the Paleozoic and is famous for its gi gantism. Odonata is composed of six suborders reaching well into the Paleozoic. Four of thes e suborders are composed entirely of fossils (Protanisoptera, Protozygoptera, Triadophlebiopter a, Tarsophlebioptera) and are sister to the extant lineages of dragonflies (Epiprocta) a nd damselflies (Zygoptera), which compose the modern Odonata. Order Protodonata The largest known insect to have ever lived, Meganeura permiana with a wing span of 71cm (28 inches), was a protodonate. Protodonata originated in the Late Carboniferous ~320 million years ago (MYA) and stretched into the Late Permian ~250 MYA. The fossil record for the group is thought to be relativ ely complete, however, new specimens are still being discovered 77


(Nel et al. submitted). Among the most signific ant and more recent protodonate discoveries is that of small protodonates, with a wingspan of ~50mm (2 inches), comparable to small to average sized modern Epiprocta, from the Lodve of France (Nel et al., submitted). Protodonata are represented in the fossil record mostly by wings and wing fragments. Their venation is simple with longi tudinal veins stretching out in a straight or arched path from the wing articulation. Longitudinal veins are also less fused basally than modern Odonata. In comparison to modern Odonata the wing structur es that supported protodonate flight are less rigid, rudimentary and in some cases non-existe nt (e.g. arculus, nodus and pterostigma). One feature of these wings that are consistent with modern Anisopte ra is the broadening of the hind wing in comparison with the forewing, a trait that is most certainly convergent. Rarely remnants of the head, thorax and abdo men are found. From these uncom mon and usually incomplete skeletal fossils it can be seen that the head is large with large eyes and strikingly large mandibles, legs are lined with stou t spines and attached to a large t horax while the abdomen is long and slender. Suborder Protanisoptera This group derives its name from the percei ved similarity to Anisoptera among classical paleoentomologists due to the similarity of overall wing shape between these two groups. Protanisoptera show many of th e structural precursors that become standard features of the modern odonate wing (e.g., rudimentary nodus and pterostigma). Though wings and wing fragments are well known for the group, skelet al impressions remain largely unknown. As a result, the only serious modern attempt at deciphering their phylogeny is centered entirely on wing morphology (Huguet et al., 2002) From the very few skeletal impressions that are known to science, it is clear that the thorax is rela tively large and likely housed a complex musculature and a broad abdomen with the first abdominal segment being short similar to modern Odonata. 78


Additionally, at least abdominal segment two appear s to possess remnants of secondary genitalia (Rasnitsyn and Pritykina, 2002), which in modern Odonata are found on both segments two and three. Protanisoptera are known only from th e Permian, 295-250 MYA, and became the dominant odonatoid at ~284 MYA, about the tim e the Protodonata appear to have died out (Rasnitsyn and Pritykina, 2002). Pr otodonata fossil specimens are recorded from Russia, the USA, Brazil and Australia and likely had an ex tensive distribution during their peek in the Permian (Huguet et al., 2002). Suborder Protozygoptera This group was so named for the similarity of wing shape they share with modern Zygoptera (see descriptio n below).The group represents an excellent example of extreme petiolation of the holodonate wing (a basal restriction of the wi ng). The classification of the groups that make up Protozygoptera is also the most contentious of the fossil suborders. There has been much speculation that Protozygoptera is non-monophyletic (Bechly 1996) and several classifications have been proposed for the gr oup (see Carpenter, 1992 ; Bechly 1996; Rasnitsyn and Pritykina, 2002). Presently, a phylogenetic analysis using m odern methods and discrete morphological characters addressi ng the higher-level relationships for the group as a whole is lacking. Thus, the classification of this group is likely to change significantly over the next several years, as researchers carefu lly approach the study of this group. Protozygoptera make their appearance in the foss il record along side Protanisoptera in the Permian but their fossil record continues well into the Late Jurassic ~161 MYA (Grimaldi and Engel, 2005) or Early Cretaceous ~ 146 (Rasnitsyn and Pritykina, 2002). The groups wing morphology is rudimentary in comparison to mo dern Odonata, with the pterostigma lacking bracing crossveins and the nodus and subnodus though more modern in appearance compared to 79


Protodonata and Protanisopter a are not as rigidly or ganized as in extant taxa. Also, the basal portion of the medial vein is st ill present at the wing base as a rudimentary vein, a feature found in all ancient odonates and lost in modern odonates. Suborder Triadophlebioptera Triadophlebioptera represent the most puzzli ng of all odonatoid groups. Members of the group appear to have a completely formed nodus, along with alignment of the nodal crossveins, though the formation of the nodus is difficult to d ecipher for some specimens due to the extreme reduction of the nodal space. Further, the group lacks a pterostigma and the primary antenodal crossveins (Ax1 and Ax2; major structural elemen ts supporting the pre-nodal leading edge of the wing) appear to be reduced, lost altogether or lacking only A x2. To complicate its placement among the other odonatoid lineag es Triadophlebioptera exhib its the fusion of several longitudinal wing veins that can be extreme (e.g., the fusion of CuA, CuP and AA at their base). Triadophlebioptera are known only from the Pe rmo-Triassic period and little is known about their skeletal morphology. The vast majority of fossils are wing frag ments, most of which are incomplete. Further, many tr iadophlebiopteran fossils are pres erved in a very fine grained rock matrix that is easily worn from handling or degraded over time while in storage (personal observation). This condition makes it difficult to examine type specimens to verify the venation and structures described in the or iginal descriptions (personal obser vation). Relatively little has been published on Triadophlebioptera beyond species descriptions and much of what has been published has been in Russian journals during the cold war era making this literature difficult to attain. One exception is a phylogenetic analysis of 17 characters and nine genera that resulted in a phylogeny with modest resolution due to the difficult morphology and incompleteness of fossils being studied (Nel et al., 2001). 80


Suborder Tarsophlebioptera This group holds the distinction as being th e likely sister group taxon to all modern Odonata (Bechly, 1996; Rehn, 2003; Bybee et al., 2008) or as the sister group to Epiprocta (Bechly, 1995; Fleck et al., 2004, Bybee et al., 2008). Tarsophlebioptera have a completely formed nodus and the beginnings of a comple te anisopteran triangl e in the hind wing. Tarsophlebioptera is thought to have existed as far back as the Permian, due to its hypothesized position as sister to all modern Odonata, but is known only from the Jurassic, as a contemporary with several other lineages that ar e basal to modern Anisoptera. Tarsophlebia also appears in the fossil record ~30MYA before the appearance of modern Anisoptera, though both groups coexisted during the Malm peri od of the Jurassic ~145-135MYA. The fossil record for Tarsophlebioptera, if it s origins do indeed date to the Permian, is incomplete. However, there are several nearly complete and well preserved fossil specimens. One in particular, Tarsophlebia examina housed at le Musum Nationa l dHistoire Naturelle in Paris is nearly a complete specimen consisti ng of fore and hind wings, a head, thorax and abdomen along with secondary genitalia and termin al anal appendages. This specimen provides an excellent resource for morphological study. Suborder Epiprocta The suborder Epiprocta, commonly referred to as dragonflies, was recently erected by Lohmann (1996) and combines the former subor ders Anisozygoptera and Anisoptera on the basis of the presence of an epiproct, among ot her features (Anisozygoptera do not form a monophyletic group but instead forms a pectinate a ssemblage of taxa leading to a monophyletic Anisoptera (Nel et al., 1993). The paraphyletic grade is referre d to as the anisozygopteran grade throughout this paper). Th e epiproct is a large plate-like appendage found at the terminal end of the abdomen that is ventra l to and directly betw een the cerci. The epiproct is used along 81


with the cerci to grasp females over the head during copulation. The fossil record for this group is excellent and especially rich with specimens representing modern families that arose in the late Jurassic. Within the Epiprocta there are several fossil groups that show spectacular variations on the epiproctan wing plan, such as the Aeschnidiid ae with their heavily structured wings and tight venation. This group had wings that were greatl y expanded in the anal region, much more than what is observed among the extant species of t oday, and a pseudo-subcostal vein, present in the antenodal and postnodal spaces between the subcostal and costal veins. Both the fore and hind wings of all Epiprocta are heavil y structured with stoutly fused veins, especially in the basal portion, to support a more rigorous forward-thrust type of fl ight. A good example of heavy structuring is the discoidal cell, found as a sing le cell usually without cr ossveins in Zygoptera, which has been modified into thr ee separate triangles (hypertriangl e, triangle and subtriangle) in the modern epiproctan wing. The anal region of the hind wing is also greatly expanded, modified, and structured in comparison to ot her Holodonata (many Zygo ptera have no anal region within the wing). The head is globular and all exhibit a relatively robust body form in comparison to Zygoptera. Though skeletal morphol ogy appears to be relatively conserved across the group, some fossil specimens exhibit intric ate auricles (expansions of the abdomen originating around the secondary gen italia) that are much more impr essive than what is observed among modern Anisoptera. Overall, the ep iproctan ground plan for wing shape and wing structural elements (e.g., nodus and arculus) a ppear to have undergone relatively little modification since the late Jurassic ~154 MYA. Suborder Zygoptera The damsels of the insect world seem aptly na med for their striking color, ability to hover and their very gracile, slight build in comp arison to dragonflies. Another feature that distinguishes this suborder from other odonatoids is the width of the head being elongate (nearly 82


always with a space between the ey es that it larger than the widt h of a single eye). Zygoptera, though slightly less species rich than the Epipro cta, are the most morphologically diverse group of all the known suborders of Holodonata. Though nearly all members of the group have petiolate wings (some calopterygoids being the exception) the amount of va riation in skeletal morphology and diversity of wing form is impressi ve. This group also has managed to diversify into ecological niches unoccupied by any mode rn dragonfly (e.g., phytotelmata such as bamboo and bromeliads and waterfalls). Zygoptera ha ve a world-wide distribution, with diversity increasing closer to the equator. Major epicen ters for the currently known diversity are centered in Central and South America and Southeast Asia, while the continent of Africa is by comparison much less diverse (Corbet, 1999). Defining the age of Zygoptera is more difficult than for the other suborders. The fossil record for the group is sparse and seems much mo re incomplete. Epiprocta arises in the late Triassic and due to the position of Zygoptera to Epiprocta, Zygoptera sh ould also approximate this age. Zygoptera are very ra re in the fossil record (excep t Steleopteridae) until the early Cretaceous (Fleck et al., 2001). Th eir rarity is a puzzle because in sects much more delicate then Zygoptera have been preserved in the fossil reco rd. Because other holodonates of similar size and structure have been discovered throughout the Jurassic and Triassic (e.g., Isophlebiidae), it is thought that Zygoptera may have been a minorit y group or living in a different type of environment during this time period (Nel, Personal Communication). The oldest known damselflies appear in the late Jurassic (Fleck et al., 2001) follow ed by a gap of more than 80 MY until the first coenagrionid damselfly appears in the early Cretaceous ~ 170MYA (Grimaldi and Engel, 2005). However, Saxonagrion minutus, a holodonate very similar to ancient Zygoptera, is known from the Upper Permian and is thought to be the very first known damselfly (Nel, 83


personal communication; Nel et al. 1999). The major groups of m odern day damselflies either do not have a fossil record or a ppear in the fossil record during the Cenozoic at the end of the Eocene (Grimaldi and Engel, 2005). Historical Classification The monophyly, definition and general nomen clature used for each of the holodonate suborders over the past half century is mind be nding (see Table 5-1. For an in depth review of holodonate monophyly, classification and nomencl ature see Bechly, 1996 and 2007). The monophyly of each suborder has been questioned and tested, sometimes with modern methods. The suborders of Protanisoptera, Zygoptera a nd Epiprocta have been tested under modern phylogenetic methodologies and are well supported as monophyletic (Huguet et al., 2001; Rehn, 2003; Bybee et al., 2008; Carle et al., 2008). Anisoptera, once a suborder containing only the dragonflies before being placed within the s uborder Epiprocta by Lohmann (1996), has been accepted as a natural group since before Fras er (1954, 1957). Until recent phylogenetic analyses recovered strong evidence for zygopteran monophyl y (Rehn, 2003; Bybee et al., 2008), Carle et al., 2008) the group was hypothesized by several other researchers to be nonmonophyletic (Trueman, 1996; Pfau, 1991; Hennig, 1981). The monophyly of Protodonata, Protozygoptera and Tarsophlebioptera remains contentious and untested within a large-sc ale cladistic analysis including all the major suborders and a large tax on sampling. The work of G. Bechly and A. Nel along with several others (Fleck and Jarzembowsk i) has provided a classification scheme for Holodonata. This scheme has been continuously updated on the Internet (Bechly, 2007) and provides an extensive framework within which ta xonomists, especially those that study fossils, can work. It is this classification scheme that will be tested herein. The relationship between the holodonate suborde rs has also been in flux, though there is general agreement among thes e hypotheses (Fig. 5-1): the sister group to Odonata is 84


Protodonata, the basal most lineage of Odonata is Protanisoptera and Zygoptera and Epiprocta are sister. There is much more variation among existing hypotheses regarding the classification for the remaining suborders as some groups we re excluded or not yet described when some hypotheses were articulated (see Table 5-1), but generally (Triadophlebioptera (Protozygoptera (Tarsophlebioptera + Zygoptera + Epiprocta))) is the trend among current hypotheses. Due to the publication of several data sets that included molecular data the current classification scheme for Holodona ta that is built primarily on wing venation has come into question (see list papers from Bechly, 2007). We propose the first comprehensive phylogenetic analysis, which includes fossil and extant taxa sampled for a wide range of morphological and molecular characters as a robus t estimate of holodonate phylogeny. Materials and Methods Bechly (1996, 2007) provides an updated classi fication scheme based on hand and brain analyses and a compilation of other hypothese s based on morphological cladistic studies, primarily from Europe. The clas sification scheme represents all odonatoids above the rank of genus, though recently it has been shown that his classification scheme for modern Odonata is at odds with schemes generated from molecula r and morphological data analyzed using computational phylogenetic methods (Bechly 2007). Nonetheless, there is no other classification scheme as complete or up to date for fossil taxa This classification is the scheme that will be tested herein though with slight ly modified group names as presented in the introduction. A review of the Holodonata l iterature and papers that included explicit morphological characters since 1993 (i.e., defined characters and character states used within a modern phylogenetic analysis) was performed. All morphol ogical characters from these papers were data based according to characters system (e.g., wing, head, thorax, abdominal, etc) using Excel. 85


This effort resulted in ~880 characters. These ch aracters were then refi ned in order to remove redundancy among characters and to combine and addr ess character states th at were applicable to a lower level analysis but needed to be rele vant within an analysis addressing the subordinal relationships. The resulting mo rphological data matrix consists of 352 morphological characters from both skeletal (69 in total) and wi ng morphologies (283 in total; Appendix D). A total of 86 taxa (26 extant and 60 fossils) we re included in this anal ysis. Taken together these taxa represent all the major suborders of Holodonata and broadly reflect the morphological diversity in the group. Extant taxa along with associated molecular data from six molecular markers from Bybee et al. (2008 and chapter 4) were included in th is analysis (Table 5-1). All taxa were coded for morphological data where the characters were applicable and fossil specimens were complete. Analyses were run using TNT 1.1 (Goloboff, 2003) on the combined morphological and molecular data matrix and the morphological matr ix alone using the defaults. A subset of analyses was run to observe how these data resolved relationship among only extant taxa (Appendix E). All analyses were run with gaps treated as missing, under a new technologies search using default parameters. Statistical approaches to morphology, particularly those implemented in a Bayesian framework, appear to provide robust phylogenetic results. Bybee et al. (2008) found that statistical approaches to morphology did not provide adequate resolution when fossil taxa were included in analyses due to missing data. Since missing data make up a significant amount of the data matrix no Bayesian approach to phylogenetic inference was conducted herein. In all analyses of both fossil and extant taxa together, trees were rooted to Eugeropteron given that there is significant evidence supporting this group as being the most 86


basal order of Odonatoptera (i.e., all odonatoid insects; Bechly, 1996; Rehn, 2003; Bybee et al., 2008). Bremer support values were computed via PAUP*.4.0b10 using a command file generated in TreeRot 2Vc ( Sorenson, 1999 ). Bremer supports were computed two times to ensure consistency among the values recovered and thus repeatability. Bootstrap values were also generated via PAUP*.4.0b10 with 100 random replicates. Results Combined Topology Two most parsimonious trees were recovered when both morphological and molecular data were combined (Strict consensus Fig. 5-2 see Table 5-2 for topological st atistics). The overall resolution for the topology was excellent, though support values were generally low for both bootstraps and Bremer values. The general trend of low support values is likely due mostly to large amount of missing data among fossil taxa as support values were higher among extant taxa. All fossil suborders were found to be monophyle tic, except Zygoptera and Epiprocta (Anisoptera is monophyletic) Morphological Topology Eight most parsimonious trees resulted from an analysis of the morphological data set (strict consensus Fig. 5-2; see Table 5-2 for topological st atistics). The resolu tion of this data set is less than that of the combined analysis, ye t all fossil subord ers are monophyletic, Zygoptera is again recovered as nonmonophyletic as is Epip rocta (Anisoptera forms a monophyletic group with poor resolution). Overall the morphological and combined t opologies are congruent concerning the monophyly of, and the relationships between th e suborders. Major differences between topologies centered around the Epiprocta, specif ically the groups trad itionally defined as 87


Tarsophlebioptera and Anisozygopt era and the lack of resolu tion among the Anisoptera clade from the morphological topology. Additional topologies Additional topologies were generated using only the extant taxa (Appendix E). These analyses included: morphological characters alone, molecula r data alone, and the total data set. The motivation in running these analyses was to explore the data sets individua lly and as a whole. Despite having a morphological matr ix that is 200 characters larg er than that of Bybee et al. (2008) these data sets provide results that ar e relatively congruent with Bybee et al., and demonstrate the importance of a large taxon sampli ng to gain an appropriate picture of odonate phylogenetics and classification. Discussion Combining morphological and molecular data When comparing the morphological and combin ed topologies the most obvious difference is the resolution obtained in the combined topo logy (Fig. 5-2) versus the morphological topology (Fig. 5-3). The inclusion of 27 ex tant taxa and their associated mo lecular data resulted in fewer trees due to resolution of fossil clades that were not resolved by mo rphological data alone, particularly for Zygoptera and Epiprocta. This is evidenced best by comparing the clade the Epiprocta + Tarsophlebioptera clade. Upon th e inclusion of molecu lar data there is a noticeable change in topology and as a result relationships among the included taxa of Epiprocta, particularly the basal members of this clade, be come nearly completely resolved. Superficially, it appears that the inclusion of molecular data, though only 31% of the taxon sampling had associated molecular data, served to settle the extant classification which in turn settles the relationships among the fossils sister and basal to modern Epiprocta. 88


Another interesting result is the effect that a relatively small number of taxa with associate molecular data have on the polarity of characters from a relatively large morphological data set. When looking at the clades from the morphological tree (Fig. 5-2) and comparing them to the combined tree (Fig. 5-3) the most derived taxa of the clades from the morphological tree become more primitive members of the clades in the co mbined tree. The effects of DNA data are far reaching within the tree. For example, Synthemis an extant libelluloid, changes position from one of the more derived lineages within Libelluloidea to a more primitive position and Batkenia a derived member of the Protozygoptera clade is placed as sister to all Protozygoptera. Missing data Because even the most complete fossil insect specimen is usually only two dimensional, the exception being those preserved in amber, all fossil taxa lack some morphological data and of course molecular data. For taxa included in th is work none appears to have been negatively affected by missing data, (negatively affected being defined herein as becoming a wildcard taxon within the analysis or being placed inside a clade that is contra the evidence from the fossil record.). Ditaxineurella is a protanisoptran fossil with remnants of only the apical wing portion beginning just prior to the nodus and Megatypus consists of the basal portion of the wing with radial veins present but has no portion of the nodus included. Each taxon was coded for 112 and 95 morphological characters respectively, re sulting in only 7% an d 6% of all possible parsimony informative characters (PINs) included in the combined analysis. Despite the lack of data both taxa were placed in posi tions that are consistent with the fossil record and the current subordinal classification. The relationships between and among taxa from the entirely fossil suborders seems to be little effected by the inclusion of molecular data (though there is affect on Protozygoptera as discussed above). However, molecular data have a much more visible influence on the 89


relationship of the taxa included in the modern clade of Epiprocta. Epiprocta has a fairly complete fossil record with more and more speci es being described all the time (the zygopteran fossil record is incomplete and fossil damselflies are rare, resulti ng in a more limited fossil taxon sampling presented here). The completeness of the fossil record and quality of specimens themselves, especially concerning wings, has resulted in large phyloge netic hypotheses based almost entirely on wing venation. However, wing venation has been shown to be homoplasious within the modern Anisoptera (Pilgrim et al, 2008; Ware et al., 2007), making it difficult to examine or accept current hypotheses of relationsh ip between the major groups of Epiprocta. This problem with topologies based on homoplas ious characters of wing venation is best demonstrated by Epiprocta and the morphologi cal topology (Fig. 5-2). Though it may seem counter intuitive, the complete ness of the epiproctan fossil re cord may have preserved more homoplasious wing characters than have been preserved among other fossil groups, resulting in a largely unresolved epiproctan clade. Regardless of the explanation for the lack of resolution among the primitive Epiprocta, adding molecular data not only improved the extant classification but had broad effects on the topo logy among fossil taxa as well. Classification Due to the overall similarity between both t opologies discussion of the classification will be focused primarily on the topology resulting from the combined data set (Fig. 5-3). This topology is more resolved, invol ves more parsimony informative characters (PIN : 1542 v. 317 in the morphological matrix) and incorporates morphol ogical data analyzed with a large amount of molecular data, thus providing an opportunity for bot h data sets to work together to give a robust estimate of Holodonate classification. 90


Holodonata Only one outgroup taxon for all of Holodonata wa s included in this analysis, thus little can be said regarding the monophyly of Holodonata. Howeve r, there is little deba te that this group is monophyletic, at least from what is currently known from the fossil record (Carpenter, 1992). Other fossil groups closely rela ted to Protodonata include the Eugeropteridae (the outgroup taxon here) that are sister to Hol odonata and together form the gr oup Odonatoptera, Ephemeroptera and Paleodictyoptera. Holodonata occupy an isolat ed phylogenetic positi on, making selection of an appropriate morphological outgroup problem atic. Few phylogenetically informative characters are shared between or can be coded for Holodonata and the other two extant pterygote lineages (Ephemeroptera and Neoptera). Extinct pterygote lineages (e .g. Paleodictyoptera) are equally distant and are known only from fossils, thus yielding mostly inapp licable and/or missing data. It is fortunate that odona toids have a rich fossil history, with Geroptera (the suborder containing Eugeropteron) being sufficiently known to provide the morphological root for Holodonata. Protodonata Protodonata is recovered as monophy letic. Several ancient fossil taxa have been included and excluded from the Protodonata since its form ation. Most recently, Bechly, (1996) excluded the ancient Erasipteroides from Protodonata on the basis of a uniquely retained archaedictyon, and the long median stem as well as the lack of known fossils that include a body. Bechly also states that the pectinate br anching of the anterior anal 2 vein could also serve as a synapomorphy for Eriasipteroides with the rest of the Protodonata. Erasipteroides is included within the Protodonata and as sister to Namurotypus Three genera form a polytomy sister to the re maining Protodonata when molecular data are included (Fig. 5-3) making it difficult to identif y a sister group within Protodonata. However, 91


when only morphological data are included Oligotypus is recovered as sister to all Protodonata. The fossil record for such an ancient group is re latively complete but still lacking and current knowledge of the group is still growing (Nel et al., submitted). Protanisoptera The monophyly of Protanisoptera is again recovered and reaffirmed as monophyletic here (Fig. 5-2; Table 5-2). Bechly li sts several synapomorphies that we re not directly coded for this analysis but do appear to provi de further evidence for the groups monophyly (Bechly, 2007). The basal most lineage of the group is unclear with both Polytaxineuridae and Permaeschnidae forming a polytomy sister to the remaining genera. Triadophlebioptera Triadophlebioptera forms a monophyletic and pe ctinate assemblage of genera with Triadophlebia as sister to all other Triadophlebioptera and Zygophlebia + Neritophlebia representing the most derived lineages. Twelve synapomorphies support the monophyly of the group, (see Appendix D, characters 82, 88, 111, 112, 113, 206, 239, 274, 277, 278, 331, 332 and 335). The placement of Zygophlebia as a derived member of th e Triadophlebioptera is very interesting when focusing on wing evolution because it has the broadest petiole of all members of the group and its wing structures (e.g., nodus ) are much more modern than the other members of the group. Triadophlebia has a very narrow wing petio le with the medial, cubital and anal longitudinal vein stems seemingly fuse d and the nodal crossvei ns are not aligned and not perpendicular to the radial po sterior vein. Toward the middle of the pectinate assemblage of genera, Mitophlebia s petiole is less than that of Zygophlebia with the anterior anal vein not fused to the cubital and medial vein stems, th e nodal crossveins aligned and more perpendicular to the radial posterior vein, though not entirely. Zygophlebia exhibits the least amount of petiolation and there is relatively extensive sepa ration of the anal vein from the medial and 92


cubital vein bases. There is also more separati on and structural support in the distal portion of the petiole with the fusion of CuA & CuP & AA2 becoming unfused and only CuA and CuP remaining fused for a distance. Of most interest in this sequence of evoluti on is the ability of the longitudinal veins to fuse and unfuse (though not en tirely clear if there was actually fusion), as well as the wing progressing from a very petiolate to much le ss petiolate, structured and presumably a more rigid wing base This structuring and lack of fusion of the longitudinal veins of CuA and CuP and their relationship to AA is very much reminiscent to modern damselflies particularly members of the Coenagrionidae. The formation of a nodus is a synapomorphy for all member of Odonata including Triadophlebiopter a, but what is significant in the primitive triadophlebiopteran taxa is the formation of the nodal crossveins from being unaligned, a very primitive state, to being aligned and perpendicular to the radial posterior vein (among the more derived taxa) a state that is observed in all extant Odonata. Protozygoptera Protozygoptera is also rec overed as monophyletic with Batkenia as the sister group to all other Protozygoptera. Bechly (2007) also hypothesized the monophyly of the group, though the combined topology differs greatly from Bechlys proposed classification. The morphological data alone also recover a monophyl etic Protozygoptera, but the t opology within the clade differs from that of the combined analysis and Bechlys classification. It is cl ear that an expanded taxon sampling and focused study of the Protozygoptera would be greatly benefi cial to understanding the phylogenetic relationships of the group. Like the basal member of Triadophlebioptera Batkenia also exhibits an extensive petiole and fusion of the wing veins with more derived members of the group exhibiting a comparatively expanded wing petiole. Unlike Triadophlebioptera it is more difficult to identify any clear path of nodal crossvein modification. Again a more extensive taxon sampling and focused 93


examination of the most complete fossil speci mens for the group would surely result in substantial findings regarding th e evolution of wing structures within this enigmatic group. Zygoptera Zygoptera has been supported as monophylet ic using both morphological data alone (Rehn, 2003) and molecular data alone (Carle et al., 2008) and molecular and morphological data together (Bybee et al., 2008). Zygoptera was not recovered as monophyletic but as a gradient of clades leading to Epiprocta (i .e., the anisozygopteran grade). Steleopteron + Auliella members of the family Steleopteridae, represent the most primitive lineage of the grade with Phenacolestes as sister to the Epiprocta. A paraphylet ic Zygoptera has been proposed in the past when only morphological characters were considered (Table 5-1). Further, other authors using a single ribosomal gene have also proposed a similar zygopteran grade leading to a monophyletic Epiprocta (Saux et al., 2003; Hasegawa and Kasuya, 2006). R ecovering Zygoptera as nonmonophyletic is perhaps the most surprising finding of this research because current opinion among odonate systematists is that a lack of data was to blame fo r recovering a paraphyletic Zygoptera in the past (Rehn et al, 2003; Bybee et al., 2008; Carle et al., 2008). This is certainly not the case here as this resear ch represents the most extensiv e morphological and molecular data matrix to date, which was analyzed with the mo st modern methods of parsimony analysis (i.e., ratchet, tree drift, etc). Further, Rehns morphological data were included within the 352 characters analyzed here (Appendix D). There are several areas that will need to be addressed in order to establish Zygoptera as truly nonmonophyletic. First, the taxon samp ling must be expanded. The taxon sampling included here represents 14 extant and 7 fossil da mselflies. Bybee et al., (2008) included an additional 53 extant damselflies that have associ ated molecular data that were not included here due to the difficulty of coding each specimen fo r 352 morphological characters. The 14 extant 94


taxa included in this analysis were chosen fo r their morphological diversity and because they represented distinct lineages on the topology of Bybee et al. (Fig. 3-2). By being selective in taxon sampling and choosing taxa that represented the broad range of morphological data the damselfly missing links, those that provided topological stability in Bybee et al, may have been inadvertently excluded. Additionally, the effects of hom oplasious wing characters on the topologies needs to be better understood (e .g., which wing regions produce morphological characters that are less homoplasious? And wh at are the effects of homoplasy on the overall topology?). The clade represen ting Zygoptera in Bybee et al. (2008) and Carle et al. (2008) differed significantly from that of Rehns (2003) (see these papers for a complete review) and when focused on the evolution of wings and wing st ructures there was no clear answer as to how the Zygopteran wing evolved. There appears to be a large amount of convergence between wing form and structure that is perhaps more tied to habitat niche than e volutionary origins. Elucidating the monophyly of Zygoptera wi ll be a difficult task for the future. Tarsophlebioptera There are three genera that make up Tarsophlebioptera (Tarsophlebia Tarsophlebiopsis and Turanophlebia ) two of which where included in this analysis, Tarsophlebia and Tarsophlebiopsis and form a monophyletic group. The monophyly of Tarsophl ebioptera is well established (Fleck et al., 2004). What is much less well established and contentious is the placement of this group in relation to Zygoptera and Epiprocta (Fig. 5-1). Arguments for the placement of Tarsophlebioptera as sister to the modern Odonata is based largely on the number of tarsomeres (Protodonata has 5; Pr otozygoptera is hypothesi zed to have 4; Tarsophlebia has what looks like 4 and Turanophlebia has what looks like 3 (Fleck et al., 2004)). Fleck et al. performed a phylogenetic analysis of hypothesi zed ground plan morphological conditions and recovered Tarsophlebioptera as sister to th e Epiprocta. This research recovered 95


Tarsophlebioptera within the Epiprocta as a mo re primitive member of the anisozygopteran grade toward Anisoptera, similar to what Trueman (1996) recovered. Though an unusual placement for the group there are several comp elling arguments for why this placement is warranted. First, there is very little known about the skeletal morphology of the Protanisoptera, Triadophlebioptera and Protozygoptera. Though there is more known of the skeletal morphology of the members of the anisozygoptera n grade still relativel y little is known for most species included therein that would allow for conclusive mo rphological evidence that could exclude Tarsophlebioptera from the Epiprocta or provide solid evidence for its placement as sister to Epiprocta + Zygoptera. The Triassic and Jurassic represent a time period of major diversification among the basal Ep iprocta (represented by the dive rsity of fossils within the anisozygopteran grade leading toward Anis optera) and tarsophlebiopterans exhibit many features that unite them with th is grade (e.g., nodal structure, di scoidal cell/triangle and lacking vein CuP). Secondly, when consulting the fossil record tarsophlebiopterans are known only from the Jurassic and the fossils that are known are fairly robust in structure and well preserved making it difficult to accept that the fossil record of this group would not st retch to the Permian. Certainly, there would be some remnant of a grou p as robust as Tarsophleb ioptera at least into the Triassic if it did indeed exist into the Permia n. What is greatly lack ing in our knowledge to provide conclusive evidence for the placement of this taxon is better fossil preservation among fossils from the other holodonate suborders (Fleck et al., 2004). Epiprocta With the exception of Tarsophlebioptera be ing placed among the more primitive lineages of Epiprocta (discussed above), Epiprocta is recovered as monophyletic. Sogdopteron and Progonophlebia members of Euparazygopteridae, form th e sister group to all Epiprocta. The anisozygopteran grade proposed by Nel et al (1993) and defined by Lohmann (1996) is 96


recovered here and is extensive (Fig. 5-3). Th e Anisoptera form a monophyletic group with the Aeschnidiidae (i.e., Misofaeschnidium and Aeschnidium ) represented as the most basal member of this group. Other relations hips include Gomphidae (i.e., Gomphus ) as the most primitive member of Aeshnoidea and the recovery of Libelluloidea. Pilgrim and von Dohlen (2008) recovered Gomphidae as sister to the libelluloids as did Bybe e et al (2008) under Bayesian analyses. Trends in Wing Evolution Several clades, Triadophlebioptera, Protozygoptera, the Zygoptera + Epiprocta/Tarsophlebioptera and at least one clade within Zygopt era (e.g., the clade containing Philogenia and Hetearina ) exhibited petiolation, a restriction of the wing base, among the more primitive members that then became broadened in th e more derived representatives of the clade. This broadening consisted of a loss of simple fusion or near fusion of longitudinal veins associated with the hind wing area and ma rgin (Triadophlebioptera, Protozygoptera & Zygoptera) to the extreme and extensive expansion of the anal region of the modern anisopteran hind wing. Additionally, the clades Triadophlebioptera and Zygoptera + Epiprocta/Tarsophlebioptera show ed a gradual alignment of th e nodal and subnodal crossveins and a migration from being non-perpendicular to the posterior radial vein to being perpendicular. Challenges Several challenges for future work regard ing holodonate classifi cation and phylogenetics exist. Many of the major challe nges that exist are simply logistic al though some are egotistical, but all can be overcome. Major collections of fossil Holodonata are sp read far apart (China, Russia, South America, throughout Europe, th e US and Japan) and few countries have centralized their fossil collec tions (e.g., Germany has many excel lent fossils spread over a handful of museums and the same is true for the UK), though some have centralized collections 97


that are exceptional (Paleontological Institute, Russ ian Academy of Sciences). There is little opportunity to ship fossils in the mail like extant insect specimens. Using the mail service has proven to be too risky (two pri celess Tarsophlebioptera with intact and well preserved secondary genitalia were lost via the ma il service) and cost prohibitive (some Protodonata specimens weigh more than 10 kilograms). Further, the price of travel to each collection throughout a lifetime is plausible but for any one researcher, let alone several, to visit each major collection, produce drawings or scans of all fossil sp ecimens within a short window of ti me is currently not possible. It has proven difficult to find the funding to a ttempt such an endeavor as well (personal experience). Further, any drawings produced woul d be subject to human error and any scans or images are of limited use (i.e., they provide lim ited magnification) to the necessary fine detailed morphological work that will be required to produce a robust classification scheme from morphological data. In short, the interest a nd expertise exist for an extensive and thorough revision of Holodonata, but the f unding is limited at present. An additional challenge will be the ability to carry out the fine detailed morphological work for an expanded fossil taxon sampling. Comb ining more fossil taxa that are largely incomplete representations of the insect and re present intermediate form s in skeletal and wing structure with extant taxa will be a difficult task. However, in this analysis excellent results were gained by combining a morphological matrix with large amounts of missing data, with molecular and complete morphological data from extant taxa. It remains to be seen what will happen when a larger number of extant taxa a nd their molecular data are combin ed with more fossil taxa that are less complete than those included here. Conclusions This research represents the first combined analysis of both molecular and morphological data for an extensive taxon sampling of both fo ssil and extant taxa for an ancient group of 98


insects. The inclusion of molecular data had far reaching effects on the overall topology and appears to stabilize the relations hips of fossil taxa that were coded only for morphological data, especially within the st em groups of Epiprocta. All fossil suborders of Holodonata were r ecovered as monophyletic, what was more surprising is that the modern suborders, Zygopter a and Epiprocta, were not. Tarsophlebioptera was recovered as a basal lineage of the Epipro cta and a member of the anisozygopteran grade leading toward a monophyletic Anisoptera. A more extensive taxon sampling of both fossil and extant taxa will be e ssential to evaluate the monophyly of Zygoptera as it was recovered as polyphyletic. Missing data, which in some cases reached as much as 94% of PIN characters, did not pose a problem in recovering a resolved and well supp orted topology, as long as molecular data were included (Fig. 5-3). Molecular data appear to have helped to sort out homoplasy present among wing characters (see chapter 6) and provi de a much more rigid topology over which the morphological data c ould be worked out. 99

PAGE 100

Table 5-1. A history of monophyly of the subor ders of Holodonata including the former suborders Anisozygoptera and Anisoptera ( sensu Lohmann, 1996). atanodotorP aretposinatorP ahpromaibelhpodairT aretpogyzihcrA aretpoibelhposraT areptogyZMM PM PM PM PM PM PM PM PM??MM M M MBybee et al. (2008)M? PMM Rehn (2003) M? ??M Bechly (1995 &1996 )PM P?P Nel (1993) MMMPMM Trueman (1996)MMM M?M Brauchmann & Zessi nPMMM P Carle (1982) MM M P Hennig (1981)MM M P Fraser (1954, 1957 )PM M M M P MatcorpipE AZ A 100

PAGE 101

Table 5-2. Tree Statistics for th e topologies from Figs. 5-2 and 5-3. Consistency Index (CI) and Retention Index (RI), Parsimony In formative Characters (PIN). LengthCIRIPINTrees Morphology 21210.2690.6783178 Morphology Strict 23110.2470.639317 Combined 82900.4080.54415422 Combined Strict 82930.4080.5431542 Combined Morph*21730.2630.667317 Combined Molec*61170.460.4441225 *indicates that these partitions of the combined data set were examined independantely over the combined topology from Fig. 6combined) 101

PAGE 102

102 Fraser Protodonata Protanisoptera Protozygoptera Zygoptera Anisozygoptera Anisoptera Epiprocta a.Hennig Protodonata Protanisoptera Protozygoptera Zygoptera Anisozygoptera Anisoptera Epiprocta b.Carle Protodonata Protanisoptera Protozygoptera Zygoptera Anisozygoptera Anisoptera Epiprocta Tarsophlebiopterc.Zessin Protodonata Protanisoptera Protozygoptera Zygoptera Anisozygoptera Anisoptera Epiprocta d.atcorpipEZygopteraNel Protodonata Protanisoptera Triadophlebiamorpha Tarsophlebioptera Protozygoptera Anisozygoptera Anisoptera e.atcorpipEZygopteraBechly Protodonata Protanisoptera Triadophlebiamorpha Tarsophlebioptera Protozygoptera Anisozygoptera Anisoptera f.Trueman Protodonata* Protanisoptera Zygoptera Protozygoptera ZygopteraT a r s o p h l e b i o p t e r a Triadophlebiamorpha A n i s o z y g o p t e r a A n i s o p t e r a A n i s o z y g o p t e r a E p i p r o c t a g.Rehn Protodonata Protanisoptera Protozygoptera Tarsophlebioptera Anisozygoptera Anisoptera Epiprocta Zygopterah.Bybee et al. Protodonata Protanisoptera Protozygoptera Zygoptera Anisozygoptera Anisoptera Epiprocta Tarsophlebiopterai. Figure 5-1. Major hypotheses of Holodonata afte r Bechly (1996). The name of each major contributor is shown in gray behind each figure.

PAGE 103

Synthemis Rhinoaeshna Cordulegaster Cordulia Devadatta Gomphus Libellula Macromia Petalura Phyllopetalia Mesuropetala Cymatophlebia Bellabrunetia Parapetala Sona Hemeroscopus Euthemis Libellulium Liassogomphus Miocordulia Stenogomphus Proterogomphus Paraheterophlebia Phthitgomphus Libellulid Oxygastra Uropetala Aeschnidi Misofaeshna Diceratobasis Oligolestes Archilestes Coenagrionid Isosticta Libellago Euphaea Lais Hetaerina Litheuphaea Platycnemis Protosticta Protoneura Philogenia Lestes Microstigma Euthore Liassophlebia Heterophlebia Epiophlebia Stenophlebia Erichschmidtia Turanothemis Oreophlebia Tarsophlebiopsis Tarsophlebia Auliella Epilestes Batkenia Kennedya Phenacolestes Progonophlebia Sogdopteron Progoneura Steleopteron Permolestes Saxomyrmeleon Protomyrmeleon Triassagrion Zygophlebia Mitophlebia Neritophlebia Paurophlebia Cladophlebia Triadophlebia Ditaxineurella Callimokaltania Permaeschna Polytaxineura Ditaxineura Oligotypus Tupus Megatypus Meganeuropsis Erasipteroides Namurotypus Eugeropteron Protodonata Protanisoptera Triadophlebiamorpha Protzygoptera Zygoptera Epiprocta Epiprocta Tarophlebioptera 83 100 100 100 100 100 100 70 70 80 100 100 86 Figure 5-2. Strict consensus of eight pars imony topologies resulting from the morphological data analyzed in TNT 1.1 under default para meters in the new technologies search. For tree statistics see Table 5-2. 103

PAGE 104

Strictconsensusof2trees(0taxaexcluded) Synthemis*Rhinoaeshna*Cordulegaster*Cordulia*Devadatta*Gomphus*Libellula*Macromia*Petalura*Phyllopetalia*Mesuropetala Cymatophlebia Bellabrunetia Parapetala Sona Hemeroscopus Euthemis Libellulium Liassogomphus Miocordulia Stenogomphus Proterogomphus Paraheterophlebia Phthitgomphus Libellid*Oxygastra*Uropetala*Aeschnidi Misofaeshchn Diceratobasis Oligolestes Archilestes*Coenagrionid*Isosticta*Libellago*Euphaea*Lais Hetearina*Litheuphaea Platycnemis*Protosticta*Protoneura*Philogenia*Lestes*Microstigma*Euthore*Liassophlebia Heterophlebia Epiophlebia*Stenophlebia Erichschmidtia Turanothemis Oreophlebia Tarsophlebiopsis Tarsophlebia Auliella Epilestes Batkenia Kennedya Phenacolestes Progonophlebia Sogdopteron Progoneura Steleopteron Permolestes Saxomyrmeleon Protomyrmeleon Triassagrion Zygophlebia Mitophlebia Neritophlebia Paurophlebia Cladophlebia Triadophlebia Ditaxineurella Callimokaltania Permaeschna Polytaxineura Ditaxineura Oligotypus Tupus Megatypus Meganeuropsis Erasipteroides Namurotypus Eugeropterontotal gaps missing Protodonata Protanisoptera Triadophlebiamorpha Protzygoptera Zygoptera Epiprocta Epiprocta Tarophlebioptera Anisozygopteran grade Anisopteran7 5 1 7 6 12 7 3 18 5 1 5 1 5 23 6 51 5 7 5 10 4 3 4 10 4 5 8 3 1 3 8 2 8 1 1 1 1 2 6 1 5 4 1 4 2 3 1 3 9 2 7 1 3 1 8 Figure 5-3. Strict consensus of two parsimony topologies from TNT 1.1 when both morphological and molecular data were comb ined. For tree sta tistics see Table 5-2. 104

PAGE 105

CHAPTER 6 CHARACTER HOMOPLASY OF THE HOLODONATE WING Introduction The holodonate wing is an amazing structure and a testimony to the success of over 320 million years of evolution. The first members, the Carboniferous Protodonata, had relatively simple venation and structural wing components as compared to the modern Anisoptera of today with their rigid structural venation and expa nded hind wing. Holodonata is a group with great potential to provide insight into the evolution of wings, venation and wing structures for Insecta in general. Historically, wing venation (which includes structural adaptations as well) has served as a major source of information supporting hypothese s of insect classification (Hamilton, 1972). Perhaps in no other insect order has this been more the case than Odonata with nearly all higherlevel classifications for the group based almo st exclusively on wing morphology (Tillyard and Fraser, 1938-1940; Fraser, 1957; Carle, 1982; Bechly, 1996; Trueman, 1996; Rehn, 2003). While hypotheses based on wing venation have produ ced classifications that appear to be relatively stable for the families of Anisoptera, especially those outside of Libelluloidea (Bybee et al., 2008; Pilgrim and von Dohlen, 2008), relations hips above and below the family level for Anisoptera and the higher-level classification of Zygoptera in general do not appear to be accurately reflected by wing morphology alone (Bybee et al, 2008; Carle et al., 2008). Though several classification schemes for holod onates have been proposed using methods grounded in Hennigian phylogeneti cs, no analysis has explicitly investigated the amount of homoplasy of wing vein characters using an objective measure. Traditionally, wing structure and character homology has been discussed within the context of the fossi l record, as posthoc observations, or before any type of analysis (computational or manual) has been performed. 105

PAGE 106

Recently, Pilgrim and von Dohlen (2008) were the fi rst to examine the util ity of morphological characters from the wing venation in a modern phylogenetic context based on a combination of both morphological and molecular data for a family of Odonata (Libellulida e). Using the results from the consistency index (CI) and retention index (RI) for each venation character proposed as a synapomorphy they were able to assess the homoplasy, an d utility of each character in terms of its use as a synapomorphy within the ranked classification scheme (i .e., focused at the family and subfamily). Pilgrim and von Dohlen found that the traditional wing ch aracters defined as synapomorphies for the family Libellulidae had high CI and RI (1.00) values but that the synapomorphies supporting the subfamilial relationships recovered low indices and were thus poor synapomorphies. CI and RI values will va ry according to taxon sampling and the data set used to generate the topology, however, Pilg rim and von Dohlens taxon sampling was robust and their analyses rigorous and thei r results are unlikely to change. This research represents the first attempt to examine the amount of homoplasy, the ability to homologize and utility of the wing characte rs used to support the current holodonate classification. Material and Methods The topologies used to genera te the indices (consistency indexCI and retention index RI) presented here were generated in chapter five. See chapter fi ve for a review of the methods and materials used to reconstruct these topologies. Only a brief re view is presented here to aid the reader. A total of 352 morphological characters were used to generate the morphological topology from Chapter 5 (Fig. 5-2). These characters were then combined with molecular sequence data from 26 extant taxa (also in cluded in Fig 5-2 but only for morphological data) to produce a combined topology (Fig. 5-3). The morphological and combined topologies were used to 106

PAGE 107

examine the amount of homoplasy among wing vena tional characters, as well as the strict consensus topologies for both. Future research will improve upon this research by expanding the taxon sampling and better addressing the complexi ties of combining large scale morphological data matrices with large scale molecular matric es (i.e., missing molecular data for fossil taxa), but we find the current data and taxon sampling to be more than adequate to address the questions at hand. Using the strict consensus topologies from Chap ter 5 the CI and RI values were calculated for each morphological character individually. Ch aracters were further divided into several categories reflecting major wing region and one single character system (e.g., longitudinal venation system and anal, nodal or pterostigmal region) then plotted on a modern anisopteran wing to visualize which wing regi ons or which wing character syst ems had the highest CI and RI values. Following Pilgrim and von Dohlen (2008) I also hypothesize that the higher the CI and RI values the more robust each character and thus the more robust each overall character system or wing region is to holodonate classification. In chapter 5 eight most parsimonious trees were recovered for the morphological data set and two when the combined data set was analyze d. The CI and RI values for each character were calculated on each of the topologies (i.e., ei ght from the morphological data and two from the combined data), then averaged to get a CI an d RI value that reflected the true average of each character across all topologies. A word of caution when interp reting the results of this rese arch: it is well understood that the CI and RI values of any single character can potentially vary, sometimes greatly, between any two analyses where different topologies exis t. These values tend to vary most with the inclusion or exclusion of large amou nts of data or taxa or when different taxa or data were used 107

PAGE 108

to examine the same phylogenetic relationships (e .g. two separate researchers using two sets of data and two different taxon samplings to address the phylogeny of Odonata). These values also will change with the inclusion and exclusion of autapomorphic data, among other things (Naylor and Kraus, 1995). Though the morphological matrix supporting this research is the largest and most complete, particularly among morphological characters of the wing, of any analysis performed for Holodonata to date, the taxon samp ling is generally considered undersized to capture all morphological variation within Holodonata. The results presented here, especially the CI and RI values, are likely to remain relativ ely stable regardless of taxon sampling, however, further research may prove otherwise. Thus, a nyone citing this research as a reason to neglect wing characters or morphology from future hol odonate research should not do so. Simply because a particular wing character does not have a high CI/RI value or is not a perfect synapomorphy for a clade does not mean that it provides no useful signal within a phylogenetic analysis. For the purposes of this paper the hol odonate wing has been divided into one morphological wing system (longitudinal venation systems; Fig. 6-2a) and nine wing regions (leading edgepre-nodus, nodal, leading edgepos t nodal/pre pterostigmal, pterostigmal, wing hind margin, cubital, anal, arculus/discoidal wi ng base and Radial-medial interior; Fig. 6-2b). The longitudinal venation system stretches across the entire wing and represents much of the major morphological variati on that has served to define both hi storical and modern classification schemes and hypotheses concerning the process of evolution within the odonate wing ( Riek and Kukalov-Peck, 1984 ; Nel et al., 1993; Trueman, 2001). The wing regions (e.g., anal and leading edgepre nodus) are important to holodonate classification and represent more localized morphological variation within the wing. The syst em and regions were chosen by the author. 108

PAGE 109

Though other systems and regions could certain ly be proposed and those included here simplified or expanded the regions from Fig. 6-2 reflect major regions important to the biomechanics of holodonate flight and/or regions that have undergone major evolutionary change. The CI and RI values for all characters invo lving skeletal morphology were also generated and compared using the two trees from chapter five This was done in an effort to compare the utility of skeletal morphology, a portion of the in sect largely neglected by researches working above the genus level because most fossils do not have associated skeletal structures, with that of wing characters. When multiple trees were recovered the CI/RI values were calculated for each character from each tree and then averaged fo r the purpose of comparison between data types (e.g., wing vs. skeletal morphologi cal data) and wing regions. Results The CI and RI values for morphological characte rs of the wing were found to be relatively equal across all regions of the wing (Table 6-1). Without excep tion the inclusion of molecular data to the morphological data matrix increa sed the overall CI and RI values when taken together, often with dramatic results (e.g., th e wing hind margin regions CI improved from 0.3362 to 0.7275 and the RI improved from 0.3290.753). The wing region with the highest CI/RI values when only morphological data we re analyzed was the leading edgepost nodal region. When molecular data were included the area along the leadi ng edge again had the highest value but this time it was the pre-nodal region of the leading e dge. Generally speaking the more characters assigned to a specific wing region the lower the CI and RI values (e.g., the radial-medial interior and arculus/discoida l wing base regions had 59 and 75 characters respectively and a CI/R I value of ~0.35). 109

PAGE 110

The longitudinal vein system consisted of 69 characters and had CI/RI values of 0.3652/0.3723 when only morphology was analyzed and 0.6203/0.6283 when combined with molecular data. This system of morphological ch aracters reflects the same trends as observed across the wing regions, namely large sized data se ts have lower CI/RI values and molecular data serve to improve these indices. Skeletal morphology did not show the same trends exhibited by characters of the wing. The CI and RI values remained nearly consta nt when molecular data were included, actually decreasing slightly as a result. The subset of characters re presenting skeletal morphology was large, comprising 69 characters, but retain ed high values (CI/RI: morphology 0.7391/0.7585; molecules 0.7157/0.7478) when compared to the average CI/RI for all wing regions combined. When the average CI and RI values were ca lculated from each of the eight morphological trees and the two combined trees they were found to be very similar to the CI and RI values from the consensus topologies. The CI and RI values for only three morphological characters from the combined analysis differed from that of the c onsensus and the average of these three values across the two topologies resulte d in minimal differences between CI/RI values from the consensus. This trend was also true for the mor phological data set but to a larger extent. Eighty characters had different CI/RI values depending on the particular topology over which it was optimized. However, the average of these CI and RI values differed only minimally from the consensus CI/RI values. For this reason the di scussion that follows regarding the CI and RI values is focused on these values as calculated on the consensus topologies. However, in the future it would be best to discuss these values in terms of each individual characters average CI and RI calculated from each topology individually. 110

PAGE 111

Discussion The homoplasy values for both topologies (Figs. 5-2 and 5-3) show both CI and RI values that are generally higher than those recovered by Pilgrim and von Dohlen (2008) when focused on the Libellulidae. Pilgrim and von Dohlen do not show the average values for all morphological characters included in their research but instead only focus on those characters directly cited as synapomorphies for libellulids and its respective subfamilies. No effort was made here to examine the synapomorphies for the major groupings in this analysis thus making it difficult to compare results. Generally speaking th ey recovered values th at were less and often closer to zero than utility while values here were high in most cases and contributed to the overall topology. What is presented here shoul d be viewed as a complement to what was discovered by Pilgrim and von Dohlen, in that they overall focused heavily and thus more finely on a taxon sampling for a single family and especial ly within a single subf amily (Sympetrinae). When these results are compared with theirs is could be stated that morphological characters focused at the higher levels of Holodonata are more easily homologized. This seems to be supported by their finding that characters that were also synapomorphies for the family of Libellulidae received CI/RI values of 1.00. Generally, characters supporting hypotheses of higher level relationship for Holodonata, those of family and above, reflect major morphological changes (e.g., the formation of the complete ar culus) while characters supporting lower level classification schemes, those of subfamily and be low, reflect refinements or variation of these major morphological changes (e.g. th e position of the arcu lus within the wing). Further research looking at wing characters as synapomorphies fo r the higher levels with an increased taxon sampling will be necessary to further address these questions. The regions with the highest CI/RI values in cluded the anal region and regions along the wing margin (i.e. leading edge and wing hind margin). This result is frankly very surprising. 111

PAGE 112

These are regions of the wing that are subjec t to great amounts of morphological variation, variation this is mostly defined by a secondar ily derived wing region (a nal region) that has undergone extreme amounts of modification, by cr ossveins that fluctuate in number, position, and angle to each other and bordering longitudinal veins (leading edge), or by the expansion and contraction of fields of cells and intercalary veins between ma jor longitudinal veins (wing hind margin). An explanation for this finding coul d be that the groups most defined by characters from these wing regions such as Anisoptera ha ve a relatively well documented fossil record overall, but also a much more robust taxon sampling in this analysis than for the other included groups. The well documented fossil record of An isoptera and Epiprocta has allowed for an indepth examination of wing vein evolution with in these groups that has lead to homology statements that appear to have held up within this larger cladistic analysis. Further, the taxon sampling included in the analysis of chapter five was selected in order to examine, among other things, the anisozygopteran grade toward Anis optera. Including such a representative taxon sampling for Epiprocta may have lead to inflation of the CI/RI. Nonetheless, it is still interesting that characters from these regions have such high indices considering they were thought to be homoplastic. The nodal region is represented by only five characters and represents a major wing structure with an evolutionary origin that is relatively well understood. Despite these facts the nodal region has the lowest CI/RI values of all wing regions. The most derived states of the characters comprising the nodal region are exhibite d by the Epiprocta. Thus, it is no coincidence that the impressive improvement of the epiproctan clade due to the inclusion of DNA also lead to a dramatic improvement of the CI/RI values (Table 6-1; see also Epiprocta Figs. 5-2 and 5-3). 112

PAGE 113

The CI/RI values for skeletal structures were very high when DNA was included or excluded. This may suggest that the skeletal stru ctures that have been neglected in comparison to venational structures may be an excellent resource for future studies, particularly those centered on improving holodonate classifi cation. What is not reflected in the CI/RI values is that the vast majority of fossil taxa included here did not have skeletal structures to be coded within the morphological matrix. Additionally, several of the skeletal characters from the morphological matrix were from the immature stage and only codeable for taxa with a known naiad. As a result these values may be inflated. These results will need to be further investigated by examining a thickly sampled group of extant odona tes that have associated naiads in order to differentiate between adult and naiad character utility and to test that skeletal structures are less prone to homoplasy. Conclusion When using morphological data derived from the holodonate wing to reconstruct evolutionary relationships the in clusion of molecular data is im portant to recovering a resolved topology. This trend is likely to be true for other research conducte d on insect groups with classification schemes derived pr imarily from wing venation. Though what is presented herein would make it tempting to use only molecular data to reconstruct phylogenetic relationshi ps that is not the purpose of this paper. Any reader who follows such a path as a result of this paper has missed the point entirely. By neglecting morphological data, particularly of the wing, entomologists miss an excellent opportunity to provide classification schemes that are completely inclusive to the known biological diversity to have existed on this planet. Further, insights in to the evolution of morphological features that made insects so successful (i.e. those surr ounding flight) would be skewed towards our knowledge of modern insects or completely lost. 113

PAGE 114

This research represents only a small portion of what could be inves tigated concerning the morphology and its utility for Holodonata or any animal group. Future research could examine the homology of the hind wing versus the fore wing, major structural components of the wing individually (along with known aspects of biomec hanical significance), and characters according to rank within the classification, etc. What wa s not investigated in this topology due to the limited number of extant taxa is the effect of morphological data especially a data set of 352 characters, would have on molecula r data in a combined analysis. 114

PAGE 115

Table 6-1. Average consistency index (CI) a nd retention index (RI) for all morphological characters divided into the major wi ng regions/system and skeletal morphology. 115

PAGE 116

Figure 6-1. Major wing regions and the longitudinal vein system used to calculate the CI and RI values shown in Table 6-1. 116

PAGE 117

APPENDIX A LIST OF MORPHOLOGICAL CHARACTERS (CHAPTERS 4&5) List of All Morphological Characters and Ch aracter States Used in this Analysis Head Characters (Rehn, 2003) Shape of clypeus: (0) rectangul ar, with anteclypeus and po stclypeus forming distinct anterior and dorsal faces, respectively; (1) flattened, with ante clypeus tilted back and not distinct from dorsal facing postclypeus; (2) greatly sw ollen and rounded into prominent snout; (3) vertical, with anteclypeus and postclypeus facing anteriorly. Shape of labial palp: (0) widest at base, tapering to tip; (1) parallel-sided; (2) external edge greatly expanded; (3) square. Premental cleft: (0) well developed, at least onequarter the length of entire prementum; (1) poorly developed, no more than one-quarter the length of entire prementum; (2) absent. Shape of frons: (0) smoothly roun ded in profile; (1) angulate; (2) flattened; (3) grossly enlarged, forming most of the head anterior to the eyes. Length of pedicel and scape: (0) pedicel longer than scape; (1 ) scape longer than pedicel; (2) sce and pedice l equal in length. ap Ecdysial cleavage line: (0) well developed; (1) partially developed; (2) absent. Postfrontal suture: (0) vestigial or absent; (1) partially developed; (2) well developed. Inner dorsal margins of eyes: (0) bent at a sharp angle so that a single point marks the narrowest space between them; (1) straight, so that no narrowest point exists between them. P Movable hook of labial palp: (0) present; (1) absent. rementum: (0) bilobed; (1) not bilobed. Shape of head: (0) globular ; (1) transversely elongate. Distance between eyes: (0) less than their own width; (1) greater than their own width; (2) eyes fused at a single point; (3) eyes broadly fused along an eye seam. Shape of vertex: (0) flat, not developed into large protuberan ce; (1) conical, or developed into a large transverse ridge. 117

PAGE 118

Wing Articulation Characters (Rehn, 2003) 14 Length of costal basivenale (BxC): (0) as lo ng, from anterior to posterior margin, as the anterior platform; (1) much shorter than the anterior platform. 15 Shape of BxC: (0) triangular, widest anteri orly and well sclerotized ; (1) rectangular and partially desclerotized in its anterior half; (2) tr iangular, broadest posterior ly and well sclerotized throughout. 16 Costal axalare (AxC): (0) separated from costal fulcalare (FxC) by a sulcus or suture; (1) AxC fully fused with FxC, suture absent. 17 Large lobe on the outside edge of AxC: (0) present; (1) absent. 18 Anterior and posterior lobes of FxC: (0) s ubequal in size; (1) posterior lobe of FxC distinctly smaller than anterior lobe; (2) posterior lobe of FxC vestigial. 19 Large, proximal horn-like sclerite on posterior articular plate: (0 ) not developed; (1) fully developed and greatly enlarged; (2) well developed, but not greatly enlarged. 20 Posterior articular plate: (0 ) with a single component scleri te enlarged and distinct from the other sclerites that comprise the plate; (1) th is sclerite reduced and fully fused with the other sclerites in the posterior articular plate. 21 Shape of anterior edge semidetached plate of the scutum (SDP): (0) narrow and bluntly rounded; (1) with a U-shaped invagination; (2) straight. 22 Bulla on outer edge of SDP: (0) absent; (1) as large as edge of basalare and heavily sclerotized; (2) distinctly smaller than edge of basalare and not heavily sclerotized. Wing Venation Characters (Rehn, 2003) Antenodal crossveins (Ax): (0 ) many (at least five, but usua lly 10 or more) present in CSc space and ScR space, unaligned; (1) many present in CSc and ScR space, aligned; (2) many present in CSc space only; (3) only two in CSc space and ScR space. Width venation of hind wings vers us fore wings: (0) hind wing slightly broader and shorter than fore wing, and with similar venation; (1) hind wing and fore wing identical in size and venation; (2) hind wing broader than fo re wing and with very different venation. Primary Ax: (0) absent; (1) two present; (2) one present. Position of IR1: (0) closer to RP1 than to RP2; (1) equidistant from RP1 and RP2; (2) closer to RP2 than to RP1. 118

PAGE 119

RP midfork: (0) symmetrical; (1) RP1,2 stra ight with RP3,4 branching posteriorly; (2) RP3,4 straight with RP1,2 branching anteriorly. CuP: (0) present; (1) absent. Costal triangle: (0) incompletely forme d, with ScA basally separated from costal margin by a partially sclerotized area; (1) fully formed, with ScA completely fused to costal margi Oblique vein between RP2 and IR 2: (0) absent; (1) present. n. Flexion line between distal edge of BxC and costal marg in: (0) absent; (1) present. Junction of costa and ScP: (0 ) acute; (1) ScP turned sharply forward to meet costa at nearly a right angle. Primary and secondary braces of nodus (nodal crossvein and subnodus): (0) absent; (1) developed in ScPRA and RARP1,2 spaces, respectively, but not aligned; (2) well developed and aligned. Postnodal crossveins: (0) unaligned in the C RA and RARP spaces; (1) aligned in the CRA and RARP spaces only; (2) aligned in a transverse series to beyond IR2. Pterostigma (Pt): (0) absent; (1) present in CRA and RARP spaces; (2) present in only the CRA space; (3) secondarily lost in bot h sexes and replaced by a densely reticulate network of veins. Stigma brace vein: (0) absent; (1) present. MA RP fusion: (0) MA basally connected to RP with a strut crossvein; (1) MA directly fused to RP at an acute angle; (2) MA and RP kinked at point of fusion, superimposed as anterior arculus; (3) MA and RP appearing to arise directly from RA, with no common stem forming the anterior arculus. M stem: (0) present and complete; (1) stranded at base of wing as a vestigial veinal remnant; (2) completely reduced, or fused with Cu stem. Path of MP distal to hind angle of quadr angle: (0) continuing straight; (1) greatly arched forward; (2) absent distal of the hind angle of the quadrangle. MP origin: (0) originates basally from the M stem; (1) originates from the Cu + M stem (see Discussion in Introduction), and arches forw ard after leaving the Cu + Mstem at the Cu crossing; (2) originates from the Cu + M stem and curves posteriorly or co ntinues straight after leaving the Cu + M stem at the Cu crossing; (3 ) originates from the Cu + M stem and abruptly kinks backward distal of where it leaves the Cu + M stem to form the proximal side of the anisopteran triangle. 119

PAGE 120

Cu crossing: (0) absent; (1) pr esent, leaving Cu stem at an oblique angle; (2) present, leaving Cu stem at a right angle. Additional crossveins immediately distal an d basal of Cu crossing: (0) absent; (1) presen t. Position of arculus: (0) distal of 2Ax, or closer to 2Ax than to 1Ax; (1) arculus between 1Ax and 2Ax, or closer to 1Ax. Posterior arculus: (0) absent; (1) developed proximal to the RP MA divergence; (2) developed at or distal of the RP MA divergence; (3) secondarily lost. Orientation of posterior arculus: (0) developed at an angle with the anterior arculus; (1) posterior arculus continuing the pa th of the anterior arculus. RP and MA divergence: (0) not strongly ar ched forward after diverging from the anterior arculus; (1) strongly arched forward after diverging from the anterior arculus. RARP space proximal to the end of ScP: (0) crossed; (1) not crossed. Crossveins in basal space: (0) present; (1) absent. Position of RP midfork: (0) located beyond 25% wing length; (1) located at less than 25% wing length. Position of RPIR2junction: (0) beyond 50% wing length; (1) at 25% wing length) at less than 25% wing length; ) secondary return to beyond 50% wing length. ; (2 (3 IR2: (0) apparently joined to RP with a crossvein;(1) fused directly to RP at an acute angle, or with a gentle forward curve. Discoidal vein: (0) absent; (1) present, form ing distal side of quadrangle between MA and M P. Subdiscoidal crossvein: (0) absent; (1) pres ent between MP and CuA and aligned with discoidal vein; (2) secondarily lost resulting from the fusion of the posterior-apical corner of the quadrangle with the hind margin of the wi ng; (3) secondarily lost in only the hind win Quadrangle (discoidal cell): (0) not di vided by a crossvein into triangle and supertriangle; (1) divided by a crossvein into triangle and supertriangl e in hind wing only; (2) divided by a crossvein into triangle and supertriangle in fore wing and hind wing. g due to the proximity of MP and CuA. Crossveins in quadrangle: (0) absent; (1) present. 120

PAGE 121

Crossveins in subquadrangle: (0) absent; (1) pres ent only in hind wing; (2) present in fore wing and hind wing; (3) present only in fore wing. Crossveins in the RPMA space between the arculus and the distal end of the quadrangle: (0) abse nt; (1) present. CuA: (0) distally forked in a simple bifu rcation; (1) not fork ed throughout its entire length; (2) with a basal bifurcat ion only; (3) pectinat e basally; (4) reduced and ending on the subdiscoidal vein, or absent; (5) reduced and e nding on the posterior side of the quadrangle. AP: (0) developed within the wing membrane ; (1) fused with the hind margin of the wing, or lost. Anal brace (AA): (0) dichotomously bran ched; (1) with no secondary branches. Quadrangle length: (0) shorter than the ba sal space; (1) longer than the basal space. Petiolation of wings: (0) absent; (1) present. Discoidal cell: (0) acute; (1) short square; (2) long square; (3) acute in fore wing, obtuse in hind wing; (4) obtuse in both pairs of wings. Position of nodus: (0) located beyond the middle of the wing; (1) located at one-third to one-half wing length; (2) located at one-quarter to one-t hird wing length; (3) located at less than one-quarter wing length. Membranule: (0) absent; (1) present. Costal nodal kink: (0) absent; (1) present. RP1IR1field: (0) expanded and filled by dic hotomous branching of RP1; (1) expanded and filled by intercalated veins; (2) narrow, with no RP1 branches or intercalated veins. RP2IR2field, intercalated sectors: (0) absent; (1) present. IR1RP2field, intercalated sectors: (0) absent; (1) present. IR2RP3field: (0) expanded and filled by dic hotomous branching of RP3; (1) expanded and filled by intercalated veins; (2) narrow, with no RP3 branches or intercalated veins. RP3MA field: (0) expanded and filled by dichotomous branching of MA; (1) expanded and filled by intercalated veins; (2) narrow, with no MA branches or intercalated veins. MAMP field: (0) expanded and filled by di chotomous branching of MA; (1) expanded and filled by intercalated veins; (2) narrow, with no MA branches or intercalated veins. 121

PAGE 122

MPCuA field: (0) expanded and filled by dichotomous branching of CuA; (1) expanded and filled by intercalated veins; (2) na rrow, with no CuA branches or intercalated veins. Width of MAMP field immediately distal of discoidal vein: (0) one cell wide immeely distal of the discoida l vein; (1) at least two cells wide distal of the discoidal vein. diat Anal loop: (0) absent; (1) pres ent, simple and sac-like; (2 ) present and elongated with a distinct midrib; (3) present w ith a well developed midrib and distinctively boot-shaped, with a well formed heeland toe. Posterior branch of ScA (ScA3,4): (0) oblique and developed within the wing membrane; (1) perpendicular to the wing axis, and not developed within the wing membrane. Basal proximity of IR2and RP3: (0) not posit ioned extremely close to one another near their origins for the length of several cells; (1) positioned extremely close to one another basally for the length of several cells. Apices of RA and RP1: (0) meeting the distal wing margin anterior to the apex of the wing itself; (1) meeting the distal wing margin posterior to the wing apex. Miscellaneous characters (Rehn, 2003) Interpleural suture: (0) complete; (1) broken in the middle with distinct upper and lower halves; (2) upper portion of sutu re absent, and only a vestigial remainder below the metathoracic spirac le. Obliquity of thorax: (0) not oblique; (1) oblique. Third segment of penis: (0) with two lateral lobes only; (1) with two apical and two lateral lobes; (2) filamentous; (3) vestigial or absent; (4) present, but with no lobes. Anterior hamules: (0) external, plate-like and quadrate; (1) external, plate-like and triangula (2) internal and folded; (3) intern al and hooked; (4) internal and vestigial. Ligula: (0) three-segmented, modified into penis; (1) one-segment ed, aids posterior hamuls in sperm transfer; (2) one-segmented, forms protective shield over modified vesicle sperm. e alis Vesicle spermalis (VS): (0) unsegmente d, unmodified storage vesicle only; (1) segmentedand modified into the intermittent organ. Posterior hamules: (0) simple, blunt and small, not projecting beyond rim of genital fossa; (1) large, clearly projecting beyond rim of genital fossa, and variously modified into 122

PAGE 123

claspers with claws, sharp tips or folds; (2) post erior hamules grossly enlarged and modified into intermnt organ. itte Epiproct: (0) very large and spatulate, not modified for grasping; (1) grasping, simple; (2) grasping lobed; (3) bifid; (4) vestigial or absent. Paraprocts: (0) simple, unmodified lobes pr ojecting from sternum of segment 10; (1) modified into inferior appendages for grasping females. Lateral abdominal gills in larva: (0) absent ; (1) present on segmen ts 2; (2) present on segments 2. Larval caudal gills: (0) abse nt; (1) present; (2) nodate. Larval gill tufts: (0) absent; (1) present. Larval rectal gills: (0) absent; (1) present. hape of larval labium: (0) flat; (1) mask-shaped and covering much of face. S Larval prementum: (0) withou t long raptorial setae; (1) w ith many long raptorial setae; (2) with only two weak setae on the median lobe. Raptorial setae on labial palps in larva: (0 ) absent; (1) several long raptorial palpal setae nt. prese Raptorial setae on moveable hook in larva: (0) absen t; (1) present. Base of larval prementum: (0) not stalked; (1) stalked. Length of second antennal segment (pedicel) in larva: (0) shorter than all other segments combined; (1) longer than a ll other antennal segments combined. Length of abdomen: (0) not greatly elonga ted for oviposition in phytotelmata (tank bromeliads, water filled tree holes, etc.); (1) a bdomen extremely elongated (total length atleast 62 mmt usually >80 mm) as a modification for ovi position in phytotelmata. bu Shape of seminal vesicle (SV): (0) rounded laterally,and anteriorly produced into two sclerotized tips connected by desclerotized me mbrane; (1) laterally produced into sharp expansions, anteriorly produced into two sclero tized tips connected by desclerotized membrane; (2) rounded laterally, anteriorly the two sclerotized tips fuse into a single tip with no membranous area. Dorsum of abdominal segment 10: (0) not developed into pyram id-like carina; (1) developed into pyramid-like carina. 123

PAGE 124

Position of wings at re st: (0) wings not held pressed together over the abdomen at rest; (1) wings held pressed together over the abdomen at rest. Relative length of abdomen a nd wings: (0) abdomen distinct ly longer than wings, or extending at least to wing tips; (1) ab domen distinctly shorter than wings. Spines projecting from ventral eye margin in larva: (0) absent; (1) present. Relative position of RP midf ork and nodus: (0) midfork not several cells distad of nodus; (1) midfork distinctly distad (by at least thr ee cells) of nodus. Posterior curve in MP distal of qua drangle: (0) absent; (1) present. Elaborate dilation and coloratio n of tibiae in second and thir d pairs of legs in males: (0) ab (1) present. sent; Large conical projections on larval caudal gills: (0) abse nt; (1) present, and gills saccoi d. Number of crossveins basal of Cu crossing: (0) severa l; (1) one; (2) none. Auricles: (0) absent; (1) present. Anal triangle in hind wing of male: (0) absent; (1) present, three-celled; (2) present, two-celled. Triangle in fore wing and hind wing: (0) of similar size, shape and proximity to arculus; (1) of different size and shape, and in hi nd wing half as far from the arculus as in fore wing; (2) of different size and shape, and in hind wing at or very close to the arculus. Internal fold of ligula (pen is): (0) not develo ped into a long fila ment; (1) developed into a long filament. Additional Characters Added to Rehn (2003) 120 Pronotal flanges: (0) absent; (1) present. 121 Highly sclerotized caudal gills : (0) absent; (1) present. 122 Ovipositor with blades: (0) spike-like; (1) absent. 123 Pterostigma: (0) short; (1) long. 124 Abdomenal lateral carinae: (0) absent; (1) present. 124

PAGE 125

125 Tibial keel: (0) abse nt; (1) present on first tibial keel; (2) present on 2nd or 3rd tibiae. 126 Posterior margin of eye: (0) not sinuate; (1) sinuate. 127 Larval antennae: (0) sixor se ven-segmented; (1) four-segmented. 128 Larval mesotarsi: (0) three-segmented; (1) twosegmented. 129 Larval moveable hook: (0) without dorsolateral spur; (1) with spur. 130 Distal margin of labial palps: (0) without deep cuts; (1 ) median cuts; (2) large cuts. 131 Distal margin of larval prementum: (0) without cleft tooth; (1) with cleft tooth. 132 Lateral spines on segment 9 in larva: (0) shorter than middorsal length of 9; (1) at least as long as 9. 133 Cerci: (0) less than 3 5 length of paraprocts; (1) greater than 3 5 length of paraprocts. 134 Abdomen: (0) not tri quetral; (1) triquetral. 135 Patella: (0) absent; (1) present. Additional Characters (Wheeler et al., 2001) 136 Superlinguae: (0) absent; (1 ) present; (2) interlocking. 137 Subimago: (0) present; (1) absent. 138 Tracheation: (0) an terior; (1) arch. 139 Direct spiracular musculatur e: (0) absent; (1) present. 140 Tentorio-lacinial muscle: (0) present; (1) absent. 141 Tentorio-mandibular muscles: (0) several bundles; (1) one. 142 Loss of some pterothoracic muscles: (0) no; (1) yes. 143 Sperm transfer: (0) indi rect; (1) copulation; (2) indirect, using claspers. 144 Male forelegs clasping: (0) absent; (1) present. 145 Male styli IX: (0) not claspers; (1) claspers. 146 Imaginal lifespan: (0) normal, f eeding; (1) Shortened, non-feeding. 125

PAGE 126

147 Larval labium: (0) not prehensile; (1) prehensile. 148 Lateral cervical sclerite in thr ee pieces: (0) absent; (1) present. 149 Pteropleura tilted backward with notum small: (0) absent; (1) present. 150 Male accessory copulatory organs: (0) absen t; (1) present. New characters CuP-kink & AA fusion: abse nt (0); present (1). CuA-CuP brace fusion: absent (0); present (1). Anal brace: not extending beyond CuP (0);extending beyond CuP (1). 126

PAGE 127


PAGE 128


PAGE 129


PAGE 130


PAGE 131


PAGE 132


PAGE 133


PAGE 134

APPENDIX C SUPPORT VALUES FIGURE 4-2 THROUGH FIGURE 4-6 Bremer supports (BrS), bootstraps (B S) and posterior probabilities (PP). 134

PAGE 135


PAGE 136


PAGE 137

APPENDIX D MORPHOLOGICAL CHARACTERS (CHAPTERS 5&6) Characters 1-69 correspond to 122; 86-122; 124-135 of Bybee et al. (2008) or Appendix A. 70 (0) nodus is formed by ScP that meets up with the costal wing margin; no discernable CP nerve; (1) nodus is formed by the CP that is inflected towards Sc P and RA, ScP bordering (or not) the distal portion of the nodus 71 Positions of Arculus: (0) between Ax1 and A x2 but closer to Ax1 th an Ax2; (1) between Ax1 and Ax2 but closer to A x2 than Ax1; (2) not between Ax1 and Ax2, distinctly basal to Ax2; (3) More or less equidistance be tween Ax1 and Ax2; (4) Not between Ax1 and Ax2 just basal to Ax2 or directly opposite. 72 Oblique cross-vein; (0) (f) sm all cross-veins reaching the distal oblique vein 'O' absent; (1) (f) small cross-veins reaching th e distal oblique vein 'O' present 73 Oblique cross-vein; (0) (h) small cross-veins reaching the distal oblique vein 'O' absent; (1) (h) small cross-veins reaching the distal oblique vein 'O' present 74 Primary Ax: (0) absent; (1) 2 pres ent; (2) 1 present. (Rehn, 2003) 75 hind v. forewing Width/venation of hind wings vs fore wings: (0) hind wing slightly broader and shorter than fore wing, and with similar venation; (1) hind wing and fore wing identical in size and venation; (2) hind wi ng broader than fore wing and with very different venation. (Rehn, 2003) 76 IR1 position @ origin: (0) closer to RP1 than RP2; (1 ) equidistance between both; (2) closer to RP2 than RP1 77 IR1 position medially (directly below anterior point of pterostigma or where might be located: (0) closer to RP1 th an RP2; (1) equidistance betw een both; (2) closer to RP2 than RP1 78 IR1 position at wing margin: (0) closer to RP1 than RP2; (1) equidistance between both; (2) closer to RP2 than RP1 79 CuP CuP: (0) present; (1) absent (Rehn, 2003). 80 RP midfork RP midfork: (0) symmetrica l; (1) RP1,2 straight with RP3,4 branching posteriorly; (2) RP3,4 straight with RP1,2 branching anteriorly. (Rehn, 2003) 81 triangle, costal Costal tria ngle: (0) incompletely formed, with ScA basally separated from costal margin by a partially sclerotized area; (1) fully formed, with ScA completely fused to costal margin. (Rehn, 2003) 137

PAGE 138

82 ScP and costa Junction of costa and ScP: (0 ) clearly acute; (1) scp turned forward to meet costa (triadophlebioptera; protomrymele ontidae); (2) ScP turned sharply forward to meet costa at nearly a right angle. (Rehn, 2003) 83 postnodal crossveins Postnodal crossveins : (0) unaligned in the C-RA and RA-RP spaces; (1) aligned in the C-RA and RA-RP spaces only; (2) ali gned in a transverse series (running across; at right angl es to the longitudinal axis) to beyond IR2. (Rehn, 2003) 84 pterostigma Pterostigma (Pt): (0) absent; (1) present in C-RA and RA-RP spaces; (2) present in only the C-RA space; (3) in bot h sexes and replaced by a densely reticulate network of veins. (Rehn, 2003) 85 MA/RP MA/RP fusion: (0) MA basally conn ected to RP with a 'strut' crossvein; (1) MA directly fused to RP at an acute a ngle; (2) MA and RP kinked at point of fusion, superimposed as anterior arcu lus; (3) MA and RP appearing to arise directly from RA, with no common stem forming the anterior arculus. (Rehn, 2003) 86 MA/RP (0)MA and and RP: superimposed as anterior arculus; (1) MA&RP no common stem forming anterioir arculus 87 MA/RP (0) Arculus kink more pronounced in hindwing than in forewing; (1) equally pronounced; (2) more pronounced in forewing than in hindwing 88 M stem M stem: (0) present and comple te; (1) stranded at base of wing as a vestigial veinal remnant; (2) completely reduced, or fused with Cu stem.(Rehn, 2003) 89 MP Path of MP distal to hind angle of qua drangle: (0) continuing straight; (1) greatly arched forward; (2) absent distal of the hind angle of the quadrangle (Rehn, 2003). 90 MP MP origin: (0) originates basally from the M stem; (1) originates from the Cu+M stem (see discussion in Introduction), and arches forward after leaving the Cu+M stem at the Cu crossing; (2) originates from the Cu+M stem and curves posteriorly or continues straight after leaving the Cu+M stem at the Cu crossing; (3) originates from the Cu+M stem and abruptly kinks backward distal of where it leaves the Cu+M stem to form the proximal side of the anisopt eran triangle (Rehn, 2003). 91 Cu crossing Cu crossing: (0) absent; (1) present, leaving Cu stem at an oblique angle; (2) present, leaving Cu stem at a right angle (Rehn, 2003). 92 arculus Posterior arculus: (0) absen t; (1) developed proximal to the RP/MA divergence; (2) developed at or distal of the RP/MA divergence; (3) secondarily lost (Rehn, 2003). 93 arculus Orientation of posterior arculu s: (0) developed at an angle with the anterior arculus; (1) posterior arculus continuing the path of the anterior arculus.(Rehn, 2003) 138

PAGE 139

94 RP ma RP and MA divergence: (0) not strong ly arched forward after diverging from the anterior arculus; (1) strongly arched forwar d after diverging from the anterior arculus. (Rehn, 2003) 95 crossveins in basal space Crossveins in basal space: (0) pres ent; (1) absent (Rehn, 2003). 96 RP midfork Position of RP midfork: (0 ) located beyond 25% wing length; (1) located at less than 25% wing length. (Rehn, 2003) 97 IR2 IR2: (0) apparently joined to RP' with a crossvein; (1) fused directly to RP' at an acute angle, or with a gentle forward curve; (2) IR2 connected to RP w/ crossvein, but apparently arising from RP3. (Rehn, 2003) 98 discoidal vein Discoidal vein : (0) absent; (1) present, form ing distal side of quadrangle between MA and MP.(Rehn, 2003) 99 subdiscoidal Subdiscoidal crossvein: (0) absent; (1) presen t between MP and CuA; (2) due to lost resulting from the fusion of the posterior-apical corner of the quadrangle with the hind margin of the wing; (3) lost in onl y the hind wing due to the proximity of MP and CuA. (Rehn, 2003) 100 subdiscoidal discoidal & subdiscoida l aligned (0) absent; (1) present 101 quadrangle Crossveins in quadrangle (disco idal cell): (0) absent; (1) present. (Rehn, 2003) 102 subquadrangle Crossveins in subquadrangle: (0) absent; (1) present only in hind wing; (2) present in fore wing and hind wing; (3) present only in fore wing (Rehn, 2003). 103 crossveins Crossveins in the RP-MA space between the arculus and the distal end of the quadrangle (discoidal vein): (0) absent; (1 ) present (Rehn, 2003). 104 AP: (0) developed within the wing membrane ; (1) fused with the hind margin of the wing, or lost. 105 petiolate Petiolation of wings: (0) absent ; (1) present. Note: In this paper, petiolation is defined as any basal fusion of the veins AA and AP that extends beyond the ScA brace 106 quadrangle Quadrangle length: (0) shorter th an the basal space; (1) longer than the basal space (Rehn, 2003). 107 discoidal Discoidal cell: (0) Acute; (1) shor t square; (2) long square; (3) acute in fore wing, obtuse in hind wing; (4) obtuse in both pairs of wings.(Nel et al., 1993). 139

PAGE 140

108 nodus Position of nodus: (0) located beyond th e middle of the wing; (1) located at onethird to one-half wing length; (2) located at one-quarter to one-third wing length; (3) located at less than one-qua rter wing length. (Rehn, 2003) 109 nodus Costal nodal kink: (0) ab sent; (1) present. (Rehn, 2003) 110 RP1-IR1 RP1-IR1field: (0) expanded and filled by dichotomous branching of RP1; (1) expanded and filled by intercalated veins; (2) narrow, with no RP1 branches or intercalated veins.(Rehn, 2003) 111 IR1-RP2 IR1-RP2 field, intercalated sectors: (0) absent; (1) present; (2) expanded by dichotomous branching of IR1 (Rehn, 2003) 112 RP2-IR2 RP2-IR2field, intercalated sect ors: (0) absent; (1) present; (2) expand by dichotomous branching of RP2. (Rehn, 2003) 113 IR2-RP3 IR2-RP3field: (0) expanded and filled by dichotomous branching of RP3; (1) expanded and filled by intercalated veins; (2) narrow, with no RP3 branches or intercalated veins (Rehn, 2003). 114 RP3-MA field: (0) expanded and filled by dichotomous branching of MA; (1) expanded and filled by intercalated veins; (2) narrow, with no MA branches or intercalated veins. (Rehn, 2003) 115 MA-MP MA-MP field: (0) expanded a nd filled by dichotomous branching of Mp; (1) expanded and filled by intercalated veins; (2) narrow, with no MA branches or intercalated veins; (3) expanded by di chotomous branching of Mp (Rehn, 2003) 116 MP-CuA MP-CuA field: (0) expanded and filled by dichotomous branching of CuA; (1) expanded and filled by intercalated veins; (2) narrow, with no CuA branches or intercalated veins; (3) ex panded and filled by dichotomous branching of MP (Rehn, 2003) 117 MP-CuA Width of MA-MP field immediatel y distal of discoi dal vein: (0) 1 cell wide immediately distal of the discoidal vein; (1) 2 cells wide distal of the discoidal vein; (2) 3-4 cells; (3) many more than 4 cells. (Rehn and 2003) 118 same but H.W. 119 ScA Posterior branch of ScA (ScA3,4): (0 ) developed within th e wing membrane; (1) developed within the wing membrane. (Rehn, 2003). 120 CuP, secondary Secondary -CuP-: (0) absent; (1) present (Rehn, 2003) 121 RA and RP1 Apices of RA and RP1: (0) mee ting the distal wing marg in anterior to the apex of the wing itself; (1) meeting the dist al wing margin posterior to the wing apex (Rehn, 2003). 140

PAGE 141

122 MP Posterior curve in MP distal of qua drangle: (0) absent; (1) present (Rehn, 2003). 123 RP midfork Relative position of RP mi dfork and nodus: (0) midfork at nodus; (1) midfork distinctly distad (often by at least 3 cells) of nodus (Rehn, 2003). 124 Cu, crossviens Number of crossveins ba sal of Cu crossing: (0) several; (1) 1; (2) none. (Rehn, 2003). 125 subdiscoidal area (0) (f) subdiscoidal area moderately transverse; (1) (f) subdiscoidal area very transverse (Fleck and Nel, 2003) 126 RP-IR2 Position of RP-IR2junction: (0) beyond 50% wing length; (1) at 25-50% wing length; (2) at less than 25% wing length; (3) secondary return to beyond 50% wing length. (Rehn, 2003) 127 CuA CuA: (0) distally forked in a simple bifurcation; (1) not forked throughout its entire length; (2) with a basal bifurcation only; (3) pectinate basa lly; (4) reduced and ending on the subdiscoidal vein, or absent; (5 ) reduced and ending on the posterior side of the quadrangle (Rehn, 2003). 128 Same as 185 but with H.W. 129 wing MA and RP3/4 (0) (f-h) MA and RP3/4 well apart near posterior wing margin; (1) (f-h) MA and RP3/4 convergent bu t not strongly near posterior wing margin; (2) (f-h) MA and RP3/4 strongly convergent near posterior wing margin (1-2 rows of cells between them); (3) (f-h) MA and RP3/ 4 parallel until wing marg in; (4) (f-h) MA and RP3/4 divergent near wing margin; (5) RP 2 branched; (6) MA branched (Fleck and Nel, 2003) 130 pterostigma brace Stigma brace vein: (0) absent; (1) present a nd well-defined; (2) present and weakly defined. (Rehn, 2003) 131 position of pterostigma brace vein: (0) meeting Pt exactly at or anterior to point/edge of pterostigma; (1) behind pt (A eschnidiidae; (2) before it 132 Same as 360.1 but H.W. 133 antenodal crossviens forewing (Ax) in C-Sc space: (0) few (3-5) present; (1) many (at least 6, but usually 10 or more) pres ent.; (2) only 2 present; (3) only 1 134 same states as 134 but H.W. 135 Forewing Antenodal crossveins (Ax) in Sc P-RA space: (0) few (3 -5) present; (1) many (at least 6, but usually 10 or more) present 2) only 2 present; (3) only 1; (4) none. (Rehn, 2003). 136 Same as 135 but H.W. 141

PAGE 142

137 single antenodal in posterior ScP+RA aligne d with single crossvein in RA-RP antenodal space: (0) absent; (1) present. 138 antenodal crossviens Antenodal crossveins (A x) alignment in C-Sc and Sc-R spaces: (0)many and all unaligned; (1) many but onl y two aligned; (2) only two present in ScP&C and ScA&RA and aligned; (3) only 2 ScA&C but not aligne d; (4) many and all aligned. 139 same but in H.W. 140 gaff (0) very long & relatively strai ght (primitive odonatoids), may be slightly undulate but not strongly sigmoidal; (1) absent (2) gaff is very short; (3) gaff is very long and sigmoidally curved. (Bechly et al., 1998, hemeroscopidae) 141 AA (0) (h) AA (+CuP) more or less parr allel with MP+ CuA: submedian space not widened in its median part; (1) (h) AA (+ CuP) not at all parallel with MP+ CuA: submedian space widened in its median part (Fleck and Nel, 2003) 142 AA:(0) Not reduced to an oblique crossvein between posterior margin and branching of Cu+AA into CuA and CuP (1) reduced to an oblique crossvein between posterior margin and branching of Cu+AA into CuA and CuP; (2) fused w/ CuP to hind wing margin (Nel et al., 2001) 143 AA:(0) (h) the principel branch of AA meets up with CuA close to the ventral angle of the discoidal space, but does not define the anal buckle/curve, in the hindwings; (1) the principel longitudinal branch of AA meets up w ith CuA, not at the ventral angle of the discoidal space, but further down, on the free part of CuA where CuA2 begins, in deliminting an anal buckle/curl in the posteriors; (2) AA meets up with CuA2 far from the origin of CuA2 (-), in the posteriors (Nel et al., 1993) 144 AA (0) AA does not make a curt, jerky at the point of contact with CuP; (1) AA does make a curt, jerky at the point of contact with CuP (Nel et al., 1993) 145 AA (0) AA is not fused with MP+Cu before the arculus; (1) AA is fused with MP+Cu for a short distance, almost before the leve l of the arculus(2)AA is fused w/ MP+CA well before the arculus (triadophe lbiomorpha) (Nel et al., 1993) 146 Anal brace (AA): (0) branched; (1) w ith no secondary branches (Rehn, 2003). 147 AA and AP (0) AA and AP are independen t; (1)Vein AA separates from AP near the extremity of the petiole; (2) AA and AP are completely fused (Henrotay et al., protpmyrmeleiontid; Nel et al, 2005; Nel and Jarzembowski, 1998) 148 AA space (0) the distal portion of AA in c ubito-anal area is well defined, appearing as a long longitudinal vein; (1) the distal po rtion of AA in cubito-anal area is not well defined, appearing as a strongl y zigzagged vein; (2) the distal portion of AA in cubitoanal area is a cross-vein; (3)absent (Huguet et al., 2002) 142

PAGE 143

149 AA; Aspl2 and 3 (0) (h) convex veins Aspl3 and Aspl2 both reaching the main stem of AA (1) (h) convex veins Aspl3 and Aspl 2 only Aspl3 reaching the main stem of AA (2) (h) convex veins Aspl3 and Aspl2 both not reaching the main stem of AA (Fleck and Nel, 2003) 150 AA1a (0) (h) an angle of AA1a below subdisc oidal space absent (1 ) (h) an angle of AA1a below subdiscoidal space present but weak; (2) (h) an angle of AA1a below subdiscoidal space present and strong (Fleck and Nel, 2003) 151 anal angle, male (0) (hindwing) male an al angle present; (1) (h indwing) male anal angle absent (Fleck and Nel, 2003; Jarzembowski and Nel, 1996) 152 anal loop (0) the hindwing anal loop is absent or posteriorly open (1) the hindwing anal loop present but rudimentar ily (less than four-celled); (2 ) the hindwing anal loop is broad (almost as wide as long) and is pent agonal or hexagonal in sh ape (more than fourcelled); (3) the anal loop is well defined by the means of (CuA + AA)b and a posterior branch of AA, but posteriorly opened; (4) the anal loop is well defined and enlongated and posteriorly closed, but wit hout a well-defined midrib (with at least 8 cells); (5) the anal loop transversely elongated, posteriorly closed and with a well-defined midrib (=Cuspl); (6) the anal loop is, posteriorly clos ed, foot-shaped with a well-defined midrib and a distinct tow; (7) (hindw ing) 'Anisopterid' anal loop sensu Fleck et al. 2003, reduced by the reduction of cubito-anal area; (8) (hindwing) 'Anisopterid' anal loop sensu Fleck et al. 2003, reduced by the development of the Aspl 1 (Nel et al., 1996; Jarzembowksi et al., 1996; Fleck and Nel, 2003) 153 anal triangle Anal triangle in hind wing of male: (0) absen t; (1) present, 3 major cells; (2) present, 2 major cells. (Rehn, 2003) 154 Antendoal crossveins (0) wing with crossv eins in the distal antesubnodal area (RARP1,2); (1) wings without crossveins in the di stal antesubnodal area (Cordulegastrid gap) (2)no CV present (zygoptera); (3) one single vein (Bech ly et al., 1998 four new) 155 antenodal area (0) (f) in the forewing the an tenodal area is not shor ter than the postnodal area; (1) (f) in the fo rewing the antenodal area is slightly shorter than the postnodal area; (2) (f) in the forewing the antenodal area is dis tincly shorter than the postnodal area (Bechly et al., petaluridae, 1998) 156 antenodal crossvein (0) the primary antenodal AX2 is situated distal of the discoidal triangle in fore wings; (1) the primary antenod al AX2 is situated basal of the discoidal triangle in fore wings (2) locate at discoidal cell or triangle (Bechly et al., 1998 four new) 157 Antenodal crossvein, secondary (0) (f) se condary antenodal cross-veins between Ax0 and Ax1 and between ScP and C absent; (1) (f) secondary an tenodal cross-veins between Ax0 and Ax1 and between ScP and C pr esent, in only one row between ScP and C; (2) (f) secondary antenodal cross-veins between Ax0 and Ax1 and between ScP and C present, in two rows or more betw een ScP and C (Fleck and Nel., 2003). 143

PAGE 144

158 Antenodal crossvein, secondary (0) absent altogether; (1) (f) secondary antenodal cross-veins between Ax1 and Ax2 present in one row; (2) (f) sec ondary antenodal crossveins between Ax1and Ax2 present in two rows ; (3) (f) secondary antenodal cross-veins between Ax1and Ax2 present in three rows ; (4) (f) secondary antenodal cross-veins between Ax1and Ax2 present in more th an three rows (Fleck and Nel, 2003). 159 Antenodal crossvein, secondary (0) (f) se condary antenodal cross-veins between Ax2 and nodus present in one row; (1) (f) s econdary antenodal cros s-veins between Ax2 and nodus present in two rows; (2) (f) sec ondary antenodal cross-veins between Ax2 and nodus present in three rows (Fleck and Nel, 2003) 160 antenodal crossveins (0) The antenodal crossveins are or are almost (witnin one) as numerous in the hindwing as in the forewing; (1) The antenodal crossveins are fewer in the hindwing than in the forewing (by two or more) (Jarzembowski, 1996) 161 antenodal crossvines (0) (f) in the fore wing the two primary antenodal crossveins between the C-ScP space have no secondary antenodals between th em; (1) (f) in the forewing the two primary antenodal crossveins in the C-ScP space are separated by only 1 secondary antenodals; (2) (f) in the forewi ng the two primary antenodal crossveins in the C-ScP space are separated by 2 or 3 seconda ry antenodals; (3) (f) in the forewing the two primary antenodal crossveins in the C-ScP space are separated by 4 or 5 secondary antenodals; (4) (f) in the forewing the two primary antenodal crossveins in the C-ScP space are separated by 6 (rarely seven) secondary antenodals; (Rehn, 2003). 162 wing antenodal supra-ScP' (0) (h) the 'antenodal supra-ScP' between C and ScP absent; (1) (h) the 'antenodal supra-ScP' be tween C and ScP present but interrupted; (2) (h) the 'antenodal supra-ScP' between C and ScP continuous, long but zigzagged; (3) (h) the 'antenodal supra-ScP' between C and ScP continuous, long and straight 163 wing antenodal supra-ScP' (0) (f) the 'ant enodal supra-ScP' between C and ScP absent; (1) (f) the 'antenodal supra-ScP' between C a nd ScP, continous, long but zigzagged; (2) (f) the 'antenodal supra-ScP' between C and ScP, long and straight 164 arculus-F.W. (0) the posterior part (arcular crossvein) not shortened; (1) the posterior part much shorter than the anterior part (RP + MA); (2) arculus posterior somewhat sigmoid (Bechly et al., 1998 four new) 165 same as 141 but H.W. 166 Arculus (0) RP&MA never fused connect ed by crossveins (eugeropteron); (1) RP&MA fused to form only anterior arculus; (2) sectors of arculu s (RP and MA) basally widely separated; (3) sectors of arculus (R P and MA) basally more or less approximate; (4) sectors of arculus (RP and MA) divergi ng from one point or even shortly fused basally (arculus stalked). (B echly et al., 1998 four new) 167 Arculus (0) (f-h) posterior pa rt of arculus distinctly str onger than other cross-veins in median space; (1) (f-h) posterior part of arculus not distinctly stronger, or not strong than other cross-veins in medi an space (Fleck and Nel, 2003) 144

PAGE 145

168 arculus (0)The arculus is in a proximal pos ition, one-4 cell from the discoidal vein, in the hindwing; (1) The arculus is above the proximal angle of the discoidal vein; (2) very far >4 cells. (Jarzembowski and Nel, 1996). 169 arculus (0)The arculus is in a proxim al position only one cell from the discoidal triangle or is above the discoidal triangle, so that the supratriangle is very short in the forewing; (1)The arculus is in a very proxima l position relative to the discoidal triangle so that the supratriangle is ve ry long (jarzembowski and Nel 1996) 170 arculus F.w. Positions of Arculus: (0) be tween Ax1 and Ax2 but closer to Ax1 than Ax2; (1) between Ax1 and Ax2 but closer to Ax2 than Ax1; (2) not between Ax1 and Ax2, distinctly basal to Ax2; (3) More or less equidistance between Ax1 and Ax2; (4) Not between Ax1 and Ax2 just basal to Ax2 or directly opposite. (R ehn, 2003; Lin et al., 2002) 171 same as 148 but H.W. 172 Aspl1 (0) (h) Aspl1 with several strong posteri or branches (1) (h) Aspl1 with 5 or more strong posterior branches (Fleck and Nel, 2003) 173 Aspl1 (0) (h) main anterior branch of As pl1 not vanishing well before posterior wing margin (1) (h) main anterior branch of Aspl1vanishing well before posterior wing margin (Fleck and Nel, 2003) 174 Aspl3 (0) (h) Aspl3 not clearl y divided into two posterior br anches; (1) (h) Aspl3 clearly divided into two posterior br anches, but Aspl4 not clearly divided into tow posterior branches; (2) (h) Aspl3 clearly divided into two posterior branches, but Aspl4 clearly divided into tow posterior bran ches (Fleck and Nel, 2003) 175 Ax0, basal brace (0) the basal brace 'Ax0' between c and ScP distincly oblique; (1) the basal brace 'Ax0' between c and Sc P not so oblique (Huguet et al., 2002) 176 Ax0, basal brace (0) the basal brace 'Ax0' is prolonged by a cross-vein between ScP and RA; (1) the basal brace 'Ax0' is not pr olonged by a cross-vein between ScP and RA (Huguet et al., 2002) 177 Ax1 (0) (f) Ax1 perpendicular to C, ScP and RA; (1) (f) Ax1 w eakly oblique; (2) (f) Ax1 strongly oblique (Fleck and Nel, 2003) 178 Ax1 and Ax2 (0) (h) Ax1 and Ax2 strongly close together; (1 ) (h) Ax1 and Ax2 moderately close together; (2) (h) Ax1 and Ax2 well apart (Fleck and Nel, 2003) 179 Ax2 (0) (f) Ax2 as strong as Ax1; (1) (f) Ax2 weaker than Ax1 (Fleck and Nel, 2003) 180 Ax2 (0) (f) Ax2 perpendicular to C, ScP and RA; (1) (f) Ax2 w eakly oblique; (2) (f) Ax2 strongly oblique (Fleck and Nel, 2003) 145

PAGE 146

181 Ax2 (0) (h) Ax2 perpendicular to C, ScP a nd RA; (1) (h) Ax2 slig htly oblique; (2) (h) Ax2 strongtly oblique (Fleck and Nel, 2003) 182 Ax2 (0) (h) Ax2 well basal of discoidal tria ngle; (1) (h) Ax2 just basal or opposite of proximal angle of proximal side of discoidal triangle; (2) (h) Ax2 well basal of MAb or opposite it; (3) (h)beyond discoida l vein (Fleck and Nel, 2003) 183 Bq corssveins, in the four wing:(0) the bridge-space (Bqs-area) between RP and IR2 basal of the subnodus is of normal width; (1) the bridge-space (Bqs-a rea) between RP and IR2 basal of the subnodus is dis tinctly narrowed; (2) the bridge-space (Bqs-area) between RP and IR2 basal of the subnodus is extremely narrow (especially in th e forwing) (Nel et al., 1998) 184 brace nervure (0) no O vein; (1) An O vein, but whose position varies, often in the distal position in respect to Sn. ; (2) O situated directly after Sn (Nel et al, 1993). 185 BxC Flexion line between distal edge of Bx C and costal margin: (0 ) absent; (1) present (Rehn, 2003). 186 C and ScA space: (0) A basal triangular scle rotized area between C and ScA, Present; (1) A basal triangular sclerotized area betw een C and ScA, absent (Huguet et al., 2002) 187 costal margin (0) the costal margin seems to branch secondarily at ScP and makes a very pronounced elbow at the level of the nodus; (1) the costal margin does not branch at ScP. It is straight (except at th e level of the eventual fissure/sulcus) (Nel et al., 1993) 188 Cr and Sn, nodal (0) (f-h) poste rior part of nodas, Cr and Sn, not strongly oblique; (1) (fh) posterior part of nodal Cr and Sn st rongly oblique (Fleck and Nel, 2003) 189 Cr nodal vein (0) the nodal vein cr between ScP and RA, not in the direct extension of ScP, more or less perpendicular to RA; (1) th e nodal vein cr between ScP and RA, in the direct extension of ScP, making a clearly acute angle with RA (Nel et al., 1993) 190 crossveins in cubital space (0) Crossveins between MP&Cu and AA basal of CuP (cubital cell sensu. Bechly 1995), absent; (1 ) Cross-veins between MP&Cu and AA basal of CuP ((cubital cell sensu. Bechly 1995), present (Huguet et al., 2002) 191 CuA (0) in the anteriors, the portion of Cu A before its division in to CuA1 and CuA2 is short; (1) in the anteriors, the portion of CuA before its division into CuA1 and CuA2 is long: the field between MP and CuA1 is at leas t as wide as that of the field between MP and MA1. (2) very long (Tupus me ga, etc) (Nel et al., 1993) 192 CuA (0) (h) free part of CuA present (1) (h) free part of Cu A absent (Fleck et al., 2003) 193 CuA F.W. (0) the free part of CuA between MP&Cu and AA is distinctly oblique; (1) the free part of CuA between MP&Cu and AA is nearly perpendicular to MP&Cu and AA (Huguet et al., 2002) 146

PAGE 147

194 same but H.W. 195 CuA (F.W.) (0) CuA does not depass nodus or terminate at nodus (1) the portion of CuA that depasses the level of the nodus is clearly longer than the proximal portion of CuA; (2) the portion of CuA that depasses the level of the nodus is a little less than the proximal portion of CuA; (3) the portion of Cu A that depasses the level of the nodus is relatively very short compar ed the proximal portion of Cu A (4) CuA begins after the nodus 196 same but H.W 197 CuA and CuP: (0) not closely parallel and with more than one row of cells between them at abase to medial section; (1) closely paralle l and with one row of cells between them at abase to medial section; (2) closely parallel & more than one row at abase to medial section (Nel et al, 2001) 198 CuA F.W. (0) Reaching the posterior wing ma ring as a solid vein; (1) apparently not reaching the posterior wing marg in (Nel et al., 2001) 199 F.W CuP. same states as 191 200 H.W. CuA same states as 191 201 H.W. CuP same states as 191 202 CuA and Mp CuA orientation to MP (0) CuA seperated from MP before general MAb between MA and MP, situated next to th e level of the arculus; (1) CuA seperated from MP at the level of MAb (2) No independent CuA; (2) CuA seperates from MP after MAb; (3) CuA apparently separating from CuP (Nel et al., voltzialestes) 203 CuA definition (0) (f) CuA is well defined in the forewing; (1) CuA is zigzagged and evanescent distally. (Nel et al, 2001; Nel and Escuille, 1992; Jarzembowski and Nel, 1996; Nel and Jarzembowski, 1998) 204 CuA definition (0) (h) CuA is well defined in the hindwing; (1) CuA is zigzagged and evanescent distally in hindwing (Nel et al, 2001; Nel and Escuille, 1992; Jarzembowski and Nel, 1996; Nel and Jarzembowski, 1998). 205 CuA definition (0) (f) CuA is straight or almost straight in forewing; (1) CuA is clearly arched in forewing (Nel et al, 2001; Nel and Es cuille, 1992; Jarzembowski and Nel, 1996; Nel and Jarzembowski, 1998) 206 CuA definition (0) (h) CuA is straight or almost straight in hindwing; (1) CuA is clearly arched in hindwing (Nel et al, 2001; Nel and Es cuille, 1992; Jarzembowski and Nel, 1996; Nel and Jarzembowski, 1998) 207 CuA+CuP+AA (0) not fused to MP; (1) fu sed to MP, but separateing basal of the base of arculus; (2) fused to MP, separati ng distal of the arculu s (Nel et al., 2001) 147

PAGE 148

208 CuA1 (0) in the anteriors, CuA1 is longue; (1 ) in the anteriors, CuA1 is very short (Nel, 1993) 209 CuA1 and MP field: (0) posteriors, the fiel d between CuA1 and MP is not particularly vaste, compared to the Cubital-anal field; (1) posteriors, the fiel d between CuA1 and MP is particularly vaste but CuA1 is not parallel to MP in an appreciable portion (at least half) of its travel/distance; (2 ) posteriors, the field between CuA1 and MP is particularly vaste but CuA1 is parallel to MP in an appreciable portion (a t least half) of its travel/distance (Nel et al., 1993). 210 CuA2 (0) CuA2, if it exists, is not parallel to the ventral margin, but is directed obliquely toward the ventral margin, in the posteriors; (1) CuA2 is, is not directed toward the ventral margin, but toward the base of the wing, parallel to the ventral margin, in the posteriors (Nel et al., 1993) 211 Cubito-anal field (0) the cells of the cubito-anal field are in the form of an almost regular five sided pentagon (or even six sided); (1) the cells of the cubito-anal field are largely 4-sided (Nel and Escuille, 1992). 212 Cubito-anal field (0) the cells of the cubito-anal field ar e not elongated; (1) the cells of the cubito-anal field are in form of an elongated pentagon. 213 CuP (0) the distal portion of CuP in the cubito-anal area is a well defined, long longitudinal vein parallel to di stal part of CuA or a crossv ein of CuA, with one row of cells between them; (1) the distal portion of CuP in the cubito-anal area is not will defined, appearing as a strongly zigzagged vein diverging from distal part of CuA; (2) the distal portion of CuP in the cubito-anal area is a crossvein (Huguet et al., 2002). 214 Cup (0) CuP near the level of the aruculus; (1) CuP in a very basal position (Nel et al., protozygotera) 215 discoidal cell (0) Discoidal cell of anterior (forewing) not transverse (broader than long); (1) Discoidal cell of ante rior (forewing) transverse (Nel et al., 1993) 216 discoidal cell (0) Discoidal cell of the poste riors (hindwing) not tr ansverse; (1) Discoidal cell of the posteriors (hindwing) transverse (Nel et al., 1993) 217 discoidal cell (0) Discoidal cell of posteri or (hindwing) open; (1) Discoidal cell of posterior (hindwing) closed by a transverse vein between MA and MP (Nel et al., 1993) 218 discoidal cell (0) Discoidal cell of anteri or (forewing) open; (1) Discoidal cell of anterior (forewing) clos ed (Nel et al., 1993). 219 discoidal cell (f) (0) in the forewing th e discoidal triangle is usually free of crossveins (unicellular); (1) in the forewing the discoidal triangle is usually divided into two cells by a crossveins; (2) in the forewing th e discoidal triangle is usually divided into three (or sometimes four) cells by two cro ssveins (or sometimes three) parallel 148

PAGE 149

crossveins; (3) in the forewing the discoidal triangle is usually at least divided into three cells by three crosvines that build a 'Ypsilon.' (Nel et al., 1998). 220 discoidal cell (f) (0) the basal side of th e forewing discoidal triangle is straight; (1) the basal side of the forewing discoidal tr iangle is sigmoidal (Nel et al., 1998) 221 discoidal cell (F) (0) (for ewing) Discoidal cell not divi ded into hypertriangle and a triangle; (1) (forewing) Di scoidal cell divided into a hypertriangle and a triangle, hypertriangle short, triangle relatively tran sverse; (2) (forewing) Discoidal cell divided into a hypertriangle and a tria ngle, hypertriangle short, triangle not transverse; (3) (forewing) Discoidal cell divi ded into a hypertriangle and a triangle, hypertriangle long, triangle not transverse; (4) (forewing) Disc oidal cell divided into a hypertriangle and a triangle, hypertriangle long, tr iangle relatively transverse; (5 ) (forewing) Discoidal cell divided into a hypertriangle a nd a triangle, hypertriangle long triangle very transverse (Fleck and Nel, 2003) 222 discoidal cell (H) (0) (hindwing) Discoida l Cell not divided into a hypertriangle and a triangle; (1) (hindwing) Discoidal cell a hypertriangle and a tria ngle, hypertriangle short, triangle relatively transverse; (2) (h indwing) Discoidal ce ll a hypertriangle and a triangle, hypertriangle short, triangle not transverse; (3) (hindwing) Discoidal cell a hypertriangle and a triangle, hypertriangle long, triangle not transver se; (4) (hindwing) Discoidal cell a hypertriangle and a triangle, hypertriangle long, triangle relatively transverse; (5) (hindwing) Discoidal cell a hypertriangle a nd a triangle, hypertriangle long, triangle very transverse Discoidal cell di vided into a hypertriangle and a triangle, hypertriangle long, triangle very transverse (Fleck and Nel, 2003) 223 discoidal hypertriangle (0) (fore) discoidal hypertriangl e free of cross-veins; (1) (fore-hindwing) discoidal hypertriangle crossed (Nel et al., 1998) petalurid, 224 Discoidal triangle (0) (h) discoidal triangl e very broad in its median part; (1) (h) discoidal triangle not very broad in its median part; (2) (h) di scoidal triangle very narrow in its median part (Fleck and Nel, 2003) 225 Discoidal triangle (0)(f) Discoi dal triangle very broad in its median part; (1) (f) Discoidal triangle not very broad in its median third; (2) (f) Discoidal triangle very narrow in its median part (Fleck and Nel, 2003) 226 Discoidal triangle (0) (h) discoi dal triangle very broad in its an terior 3rd; (1) (h) discoidal triangle moderatly broad in its anterior 3rd; (2)(h) discoida l triangle very narrow in its anterior 3rd (Fleck and Nel, 2003) 227 Discoidal triangle (0) (f) Discoidal triangle very broad in its anterior third; (1) (f) Discoidal triangle moderatley broad in its anterior third; (2) (f) Discoidal triangle very narrow in its anterior third (Fleck and Nel, 2003) 228 Discoidal triangle (0) the arrangements of cells in the foreand hindwing discoidal triangles are similar; (1) th e arrangements of cells in th e foreand hindwing discoidal triangles are different (Madsen and Nel, 1997) 149

PAGE 150

229 Discoidal triangle (0) (f ) Distal margin of discoidal triang le (MAb) straight or nearly so; (1) (f) Distal margin of discoidal tria ngle (MAb) curved (Fleck and Nel, 2003) 230 Discoidal triangle (0) (h) Distal margin of discoidal triangle (MAb) straight or nearly so; (1) (h) Distal margin of discoidal triangl e (MAb) curved; (2) (h) Distal margin of discoidal triangle (MAb) very curved (Fleck and Nel, 2003) 231 Discoidal triangle (0) (h) basal margin of discoidal triangle (MP + CuA) straight or nearly so; (1) (h) basal margin of discoidal triangle (MP + CuA) curved; (2) (h) basal margin of discoidal triangle (MP + CuA) very curved (Fleck and Nel, 2003) 232 discoidal triangle (0) The costal margin of the discoidal triangle is straight in the hindwing; (1) The costal margin of the discoi dal triangle is very rounded; (2) the costal margin of the discoidal tria ngle is slightly concave. (Jarzembowski and Nel, 1996) 233 discoidal triangle (0) The costal side of the discoidal triangle is a bout the same length as the distal side in the forewing; (1) The costal side is much shorter than the distal and proximal sides and the triangle is very narrow.(Jarzemboski and Nel, 1996) 234 Field between MA and MP (0) the field between MA and MP does not contain more than one row of cells under th e nodus; (1) the field between MA and MP contains two rows of cells under the nodus; (2) (2 ) >2 cells (Nel and Escuille, 1992) 235 Forewing discoidal triangle (0) Forewing disc oidal triangle as broad as that of the hingwing; (1) Forewing discoidal triangle mo re slender than hingwing; (2) much more slender; (3) F.W. broader then H.W. (Madsen and Nel, 1997) 236 hingwing (0) posterior margin of the hind wi ng is disctincly indented at the end of the RP3/4 and MA; (1) posterior margin of the hind wing is more or le ss straight between the apex and CuAa (Nel et al, 1998) 237 hypertriangles F.W. (0) anterior (costal) margin (MA) of hypertriangle more or less straight; (1) anterior (costal) margin (MA) of hypertriangle distinctly curved (Bechly et al, 1998; four new drgonflies) 238 Same as 245 but H.W. 239 infrasubdiscoidal (0) (f) infrasubdiscoidal spaces not numerous; (1) (f) infrasubdiscoidal spaces numerous (more than three) (Fleck and Nel, 2003) 240 infrasubdiscoidal (0) (f) the most distal infrasubdiscoidal spac es posteriorly closed; (1) (f) the most distal infrasubdiscoidal spaces the 2-3 most distal ones opened (Fleck and Nel, 2003). 241 IR2 FW (0) IR2 is a well de veloped vein; (1) IR2 is a zi gzagged vein in its distal part; (2) IR2 is a zigzagged vein beyond its distal part and nearly throughout its whole length (probably needs to be adjusted) (Bech ly et al., 1998, four new dragonflies; Nel and Jarzembowski, 1998); Henrotay, protomyrmeleontid) 150

PAGE 151

242 IR1 (0) the base of vein IR1 is basal of the middle of the pteros tigma; (1) the base of vein IR1 is below or distad of the distal si de of the pterostigma; (2) base of IR1 near midwing (Archizygoptera ).(Bechly et al., 1998) 243 F.W. (0) IR1 absent; (1) IR1 going from zig zag vein to well defined vein directly under pterostigma; (2) IR1 well defined until the pterostigma, then zigzagged 244 H.W. (0) IR1 absent; (1) IR1 going from zig zag vein to well defined vein directly under pterostigma; (2) IR1 well defined until the pterostigma, then zigzagged 245 IR2: (0) absent; (1) undul ate; (2) Strongly undulate 246 IR2: (0) not parallel with RP 2; (1) parallel with RP2 (or mo st distal branch of RP2) 247 IR1 F.W.:(0) (fore-hindwing) primary IR1 present, well developed, beginning well before the pterostigma; (1) (fore-hindwing) primary IR1rudimentary; (2) (fore-hindwing) primary IR1 absent or shorte ned, beginning at most 6 cells before the pterostigma; (3) IR1 (Lin et al., 2002; Fleck and Nel, 2003; Nel et al., 1993) 248 same but H.W. 249 IR1 (0) IR1 is basally a zigzagged vein or close to it; (1) IR1 is bassally a very clean, straight vein (Nel et al., 1998; (Nel and Jarzembowski 1998) 250 IR1 (0) vien IR1 commences on the main branch of RP1, even if it is very near to RP2 at its beginning; (1) IR1 bran ches from RP2, the original branching of IR1 on RP being reduced to an oblique crossvien; (2) The orig inal branching of IR1 on RP is indistinct (Fig. 2) (Henrotay et al., Protomyrmeleontid) 251 IR1 (0) IR1 begins distad of the free part of RP2; (1) IR1 seems to begin at or basad of the free part of RP2 but distad of the base of IR2; (2) IR1 seems to begin basad of both the free part of RP2 & IR2 (Fig 2.) (Henrotay et al., Protomyrmeleontid) 252 IR1 (0) there is no basal ps eudo-IR1; (1) there is a supe rior more or less zigzagged pseudo-IR1 befor pterostigma(Fig. 2); (2 ) pseudo IR1 originat ing under or behind pterostigma (Henrotay et al., Protomyrmeleontid) 253 IR1 and RP1 field (0) One single row of celles between IR1 and RP1 before the pterostigma; (1) two rows of cells between IR 1 and RP1 before the pterostigma; (2) More than two rows of cells between IR1 and RP1 before the pterostigma (Nel and Escuille, 1992) 254 IR1 and RP2 field (0) not more than one secondary longitudinal vein between IR1 and RP2 under the level of the pterostigma; (1) Two long secondary longitudinal veins between IR1 and RP2 under the level of the pt erostigma; (2) at least three long secondary longitudinal veins between IR1 and IR2 under the level of the pt erostigma (Nel and Escuille, 1992) 151

PAGE 152

255 IR2 (0) IR2 begins just after the beginn ing of RP3-4 (one cell at most per gap or difference-); (1) IR2 begins well after the beginning of RP3-4, but always has one cell difference or gap; (2) IR2 begins well afte r the beginning of RP3-4 (two or three cells difference, or gap); (3) IR2 & RP3 ovelappi ng or converging broadl y like an arculus at origins (Nel et al., 1993) 256 IR2 and RP2 (0) (f-h) IR2 and RP2 well apart near posterior wing margin; (1) (f-h) IR2 and RP2 not strongly convergent near posterior wing margin; (2) (f-h) IR2 and RP2 strongly convergent near posterior wing margin ; (3) (f-h) IR2 and RP 2 close together and parallel until wing margin; (4) (f-h) IR2 and RP2 divergent near wing margin only (Fleck and Nel, 2003) 257 IR2 and RP2 (0) IR2 and RP2 are not fused basally, so that IR2 does not appear as a seconday branch of RP2; (1) IR2 appears as a secondary branch of RP2, fused basally with it, (Nel and Jarzembowksi, 1998) 258 IR2 and RP2 area (0) (f-H) Area between IR2 and RP2 with four rows of cells or less in its widest part; (1) (f-H) Area between IR2 and RP2 with more than four rows of cells in its widest part (Fleck and Nel, 2003) 259 IR2 and RP2; O-oblique (0) no oblique vein; (1) Presence of two oblique veins between RP2 and IR2; (2) Presence of onl y one oblique vein (Bechly et al., 1998) 260 IR2-RP3 (0) The area between veins IR2 and RP3/4 is large in all wings; (1) The area between IR2 and RP3/4 is reduced (two or three cells across) (Jarzembowski and Nel, 1996) 261 longitudinal secondary braces (0) There are less than two longitudinal secondary branches between RP2a and RP2b; (1) Ther e are two or more longitudinal secondary branches between RP2a and RP2b (Hen rotay et al., protomyrmeleontid) 262 MA (0) (f-h) MA not straight; (1) (f-h) MA straight (Fleck and Nel, 2003) 263 MA and Mspl (F.W.) (0) (f) area between Mspl and MA with three rows of cells or less; (1) (f) area between Mspl and MA with four or five rows of cells; (2) (f) area between Mspl and MA with more than five ro ws of cells (Fleck and Nel, 2003) 264 MA and Mspl (H.W.) (0) (f) area between Mspl and MA with three rows of cells or less; (1) (f) area between Mspl and MA with four or five rows of cells; (2) (f) area between Mspl and MA with more than five rows of cells 265 (f.w.) Mspl (0) absent; (1) present 266 (f.w.) Mspl relatively distant from MA overall (by elongate cells) 267 (f.w.) MA with distal branching cutting off Mslp. 152

PAGE 153

268 (f.w.) Mslp running exactly parallel to MA for most of the length before branching: (0) absent; (1) present 269 (f.w.) Mslp with distal brances: 0) absent; (1) present 270 (same as 282.3 but H.W.) Mspl relatively dist ant from MA overall (by elongate cells) 271 (same as 282.3 but H.W.) MA with di stal branching cutting off Mslp. 272 (same as 282.3 but H.W.) Mslp running exactly parallel to MA for most of the length before branching: (0) absent; (1) present 273 (same as 282.3 but H.w.) Mslp with distal brances: 0) absent; (1) present 274 MA and Mspl (0) (f-h) convex 'MA-Mspl' ve inlets' absent or very weak between MA and Mspl; (1) (f-h) convex 'MAMspl' veinlets' present be tween MA and Mspl but not very strong and not numerous; (2) (f-h) conve x 'MA-Mspl' veinlets' present between MA and Mspl, strong and numerous(Fleck and Nel, 2003) 275 MA and RP distal portion (sp ace) (0) in the distal porti on of the area between MA and RP, there are two long rows of cells or more; (1) in the distal porti on of the area between MA and RP, there is a short double row or onl y one row of cells; (2) many cells (Madsen and Nel, 1997) 276 MA and RP3-4 field (0) importa nt vein; (1) 1important vein; (2) 2 important veins; (3) 3 or more important veins (Nel and Escuille, 1992) 277 MA, RP and base of RP3/4 (space) (0) Ar ea between MA, RP and base of RP3/4 with cross-veins; (1) Area between MA, RP and base of RP3/4 without cross-veins (Nel et al., 2005) 278 MAb or discoidal brace (DDV) (0) (f) Gene ral direction of MAb distinctly bent towards wing basis; (1) (f) Ge neral direction of MAb perpendi cular to wing basis; (2) (f) General direction of MAb di stinctly bent towards wing apex.(Fleck and nel, 2003) 279 MAb (0) (h) general directi on of MAb disctinctly bent towards wing basis; (1) (h) general direction of MAb perp endicular to wing axis; (2) (h ) general direction of MAb distinctly bent towards wing apex (Fleck and Nel, 2003) 280 Membranule (f) (0) (fore-hindwing) Membranule present; (1) (fore-hindwing) Membranule very reduced; (2 ) (fore-hindwing) Membranule absent (Fleck and Nel, 2003) 281 membranule (h); (0) the hindwing membranule is present and distin ct; (1) the hindwing membranule is present but strong ly reduced (Nel et al., 1998) 153

PAGE 154

282 MP (0) at the distal end of the MP, the posterior wing margin ha s no notch; (1) at the distal end of the MP, the posterior wing margin has a weak notch; (2) at the distal end of the MP, the posterior wing margin has a strong notch (Huguet et al., 2002) 283 MP and CU fusion (0) MP and Cu not fuse d at their extreme base; (1) MP and Cu fused basally, indestinguisable .(Nel et al., voltzialestes) 284 Mp and CuA (0) at the level of the base of the arculus, MP and CuA (&CuP&AA) remain parallel for some ways; (1) one to two cells distal to origins, MP and CuA (&CuP&AA) move aside, the distance betw een the two veins greatly increases, and approximate again distally, so that MP is strongly arched at th is point(Huguet et al., 2002) 285 MP and CuAa area (0) the hindwing area be tween MP and CuAa is narrow, with only one row of cells near the disc oidal triangle; (1) the hindw ing area between MP and CuAa is widened, with two rows of cells near the discoidal tr iangle (Bechly et al., 1998) 286 MP hindwing (0) in the hindwing MP reaches the posterior wing margin distinctly distal of the nodus; (1) in the hindwing MP reaches the posterior wing margin only slightly distal of nodus; (2) in the hindwing MP reaches the posterior wing margin on the level of the nodus, or even somewhat basally (Nel et al, 1998) 287 same as 307 but H.W. 288 MP+CuA (0) (f) Basal part (MP+CuA) Stra ight or nearly so; (1) (f) Basal margin (MP+CuA) curved; (2) (f) Basal margin (MP+ CuA) very curved (F leck and Nel, 2003) 289 MP+CuA & AA (0) the viens MP + CuA and AA do not form a brusque elbow before the arculus, in the posteriors; (1) the viens MP + CuA and AA do form a brusque elbow before the arculus, in the posteriors (Nel et al., 1993) 290 MP+CuA & AA (0) the viens MP + CuA and AA do not form a brusque elbow before the arculus, in the anteriors; (1) the viens MP + CuA and AA do form a brusque elbow before the arculus, in the anteriors (Nel et al., 1993) 291 Mspl (0) Mspl absent; (1) at least a weakly defined Msp l; (2) very well defined Mspl present in both wing pairs (Bechly et al., 1998) 292 Mspl and discoidal triangle (0) (hindwing) vein Mspl not beginning close to discoidal triangle; (1) (hindwing) vein Mspl beginning close to discoidal triangle; (2) Mspl not contiguous starts near discoidal break & starts again. (Lin et al., 2002) 293 nodus (0) The ventral portion of the nudus is situated well before the point where ScP rejoins the Costa; (1) The ve ntral portion of the nudus is situated under the point where ScP rejoins the Costa (Nel et al, Protozygoptera voltzialestes) 154

PAGE 155

294 O, basal (0) basal oblique vein is situated more than 3 cells distal of the subnode in both wing pairs; (1) Basal oblique vein is le ss than 3 cells distal of the subnode in both wing pairs. (Bechly et al., 1998b) 295 Oblique cross-vein (0) (f-h) distal oblique ve in 'O' close to first 'O'; (1) (f-h) distal oblique vein 'O' far from first 'O' (mor e than six cells) (Fleck and Nel, 2003) 296 Oblique cross-vein (0) (f-h) small cross-veins reaching the basal oblique vein 'O' absent; (1) (f-h) small cross-ve ins reaching the basal oblique vein 'O' present (Fleck and Nel, 2003) 297 Oblique cross-vein (0) (f-h) small crossveins reaching the distal oblique vein 'O' absent; (1) (f-h) small cross-ve ins reaching the distal oblique vein 'O' present (Fleck and Nel, 2003) 298 oblique postsubnodal vein (0) A distincly oblique postsubnodal cross-vein between RA and RP absent; (1) A distincly oblique postsubnodal cross-vein between RA and RP present (Huguet et al., 2002) 299 petiolate, fore wing (0) anterior wings petiolation is inferior in length to the width of the petiole; (1) wings petiolate by the fusion, at least partial, of AA with AP on a length at least equal to the width of the petiole; (2) very subequal much more than width (twice as much-) (Nel et al., 1993) 300 petiolate, hing wing (1) posterior wings petiolation is inferior in length to the width of the petiole; (2) wings petiolate by the fusion, at least partial, of AA with AP on a length at least equal to the width of the petiole. (Nel et al., 1993) 301 postdiscoidal area (0) (h) postd iscoidal area not wide with no more than two rows of cells; (1) (h) postdiscoidal area moderately wi de, with 12 rows of cells or less; (2) (h) postdiscoidal area very wide, with more than 12 rows of cells or less just distal of Mab(Fleck and Nel, 2003) 302 postdiscoidal area (0) (f) postdiscoidal area not wide with no more than two rows of cells; (1) (f) postdiscoidal area moderately wide with nine rows of cells or less; (2) (f) postdiscoidal area very wide, with more than nine rows of cells or less, just distal MAb (Fleck and Nel, 2003) 303 Petiolation tapering outward to wing (costal & anal margins not parallel) (0) absent; (1) present 304 Posterior petiolation margin composed of M+Cu+A: (0) ab sent; (1) present 305 postnodal area (0) the postn odal area of both wings is normal (not narrowed); (1) the postnodal area of both wings is very narrow, with many cells distal of the pterostigma; (2) the postnodal area of both wings is very narrow, with many cells distal of the pterostigma and a pseudo-ScP is present (Nel et al., 1998) 155

PAGE 156

306 postnodal area First postnodal crossv ein oblique: (0) ab sent; (1) present 307 Pseudo ScP: (0) absent; (1) present. 308 postnodal crossveins C-RA (0) Postnoda l cross-veins between C-RA absent; (1) Postnodal cross-veins between C-RA presen t and numerous; (2) Postnodal cross-veins between C-RA present, but not numerous (H uguet et al., 2002; Fleck and Nel, 2003) 309 postnodal crossveins RA-RP1: (0) Postnodal cross-veins between RA-RP1 absent; (1) Postnodal cross-veins between RA-RP1 pres ent and numerous present; (2) Postnodal cross-veins between RA-RP1 present, but not numerous (only a few) (Huguet et al., 2002; Fleck and Nel, 2003) 310 Postnodal supra-ScP: (0) (h) 'Postnodal supra-ScP', between C and ScP absent; (1) present (Fleck and Nel, 2003) 311 Postnodal supra-ScP (0)(h) 'Postnodal supra-ScP', between C and ScP reduced to few cells; (1) (h) 'Postnodal supra-ScP', between C and ScP long but zigzagged; (2) (h) 'Postnodal supra-ScP', between C and ScP long and straight 312 postnodal supra-ScP: (0) (f) 'postnodal supr a-ScP', between c and ScP absent; (1) (f) 'postnodal supra-ScP', between c and ScP reduc ed to few cells; (2) (f) 'postnodal supraScP', between c and ScP zigzagged; (3) (f) 'postnodal supra-ScP', between c and ScP straight; (4) (f) 'postnodal supra-ScP', betw een c and ScP stronger than ScP; (5) (f) 'postnodal supra-ScP', between c and ScP reach ing the pterostigma or nearly so (Fleck and Nel, 2003). 313 Pterostigma (0) Pterostigma pa rallel sided; (1) Pterostigma not parallel sided (Bechly, et al., 1998b) 314 Pterostigma side margins (0) parall el sided; (1) not parallel sided 315 Pterostigma (0) (Fore and hindwing) pteros tigma not basally shif ted in the basal two third of the wing; (1) (Fore and hindwing) pt erostigma basally shifted in the basal two third of the wing, midway between nodus and apex (Fleck and Nel, 2003) 316 pterostigma (0) (f-h) pterostigmal area not crossed by veinlets; (1 ) (f-h) pterostigmal area crossed by very weak veinlets; (2) (f-h) pterostigmal area by str ong veinlets; (3) (f-h) pterostigmal area crossed by veinlets and a net of cells (Fleck and Nel, 2003) 317 pterostigma (0) (f-h) basal side of pterostigma well-defined and strong; (1) (f-h) basal side of pterostigma not well-define d or absent (Fleck and Nel, 2003) 318 pterostigma (0) (f-h) distal side of pterostigma well-defined or ab sent; (1) (f-h) distal side of pterostigma not well-define d or absent (Fleck and Nel, 2003) 156

PAGE 157

319 pterostigma (0) (f) Basal side of pterostig ma without cross-vein s reaching it; (1) (f) Basal side of pterostigma with cross-vein(s) reaching it; (2) only brace basal to it. (Fleck and Nel, 2003) 320 pterostigma (0) (h) Distal side of pteros tigma without cross-veins reaching it; (1) (h) Distal side of pterostigma with crossvein(s) reaching it (Fleck and Nel, 2003) 321 pterostigma (0) (f-h) pterostigmal brace st raight; (1) (f-h) pter ostigmal brace curved and arched (Fleck and Nel, 2003) 322 pterostigma (0) (f-h) pterostigmal brace moderately oblique; (1) (f-h) pterostigmal brace very oblique; (2) (f-h) pterostigmal brace and arched (Fleck and Nel, 2003) 323 pterostigma (0) (f-h) Area near wing apex (distal of pterostigma), between RA and RP1 with one row of cells; (1) (f-h) Area near wing apex (distal of pterostigma), between RA and RP1 with two rows of cells; (2) (f-h) Area near wing apex (distal of pterostigma), between RA and RP1 with more than tw o row of cells. (Fleck and Nel, 2003) 324 Pterostigma length (0) pteros tigma as broad as long; (1) pterostigma longer than broad with length less than 8 times its width; (2 ) pterostigma longer than broad with length greater than 8 times its width (Bech ly et al, 1998; B ybee et al., 2008) 325 pterostigmal area (0) (h) Area below pteros tigma with one row of cells; (1) (h) Area below pterostigma with two rows of cells; (2) (h) Area below pterostigma with more than two rows of cells (Fleck and Nel, 2003) 326 RA (0) RA with no distal posterior branch; (1) RA with a short distal posterior branch; (2) RA with a long distal posterior branch (Huguet et al., 2002) 327 RA and RP (0) the transverse veins be tween RA and RP, before the nodus is not completely oblique; (1) The tr ansverse veins between RA and RP, before the nodus is clearly oblique (Nel et al., protozygoptera) 328 RA RP space RA-RP space proximal to the end of ScP: (0) crossed by a few crossveins; (1) not crossed; (2) crossed by many cross-veins (Rehn, 2003; Huguet et al., 2002) 329 same as 328 but H.W 330 RP and MA (0) RP and MA are well separa ted at their bases in all wings; (1) RP and MA are fused at base but diverging at or igin; (2) RP and MA are fused for a short distance distad of their bases; (3) RP & MA never fused; (3) RP & MA are never fused. (Jarzembowski, Nel, 1996) 331 RP2 (0) RP2 orignates before nodus; (1) RP 2 originates less than 4 cells after the nodus; (2) RP2 originates at between 4-8 cells after the nodus; (3)RP2 originates >8 cells after the nodus (Nel and Escuille, 1992) 157

PAGE 158

332 RP2 (0) RP2 is more or less straight; (1) RP2 is curved be neath the pterostigma; (2)RP2 is undulate prior to Pt (3)(f) at least in the forewing, RP2 is di stally indistinct and zigzagged (Bechly et al., 1998; Nel et al., 1998). 333 RP2 (0) RP2 is not divided into two bran ches; (1) RP2 is divided into two short branches; (2) RP2 is divide d into two long branches. (Nel and Jarzembowski, 389) 334 RP2 and RP1 space (0) Number of rows of cells in area between RP2 and RP1 opposite pterostigma is three or less; (1) Number of rows of cells in area between RP2 and RP1 opposite pterostigma is more th an three (Nel et al., 2005) 335 RP3 and RP4-space (0) not broad; (1) very broad, with more than 10 rows of cells along posterior wing marg in (Nel et al., 2001) 336 RP3/4 (0) RP3/4 is not divided into two well -defined branches; (1) RP3/4 is divided into two well-defined branches RP3/4a and RP3/4b, but RP3/4a is weak and more or less zigzagged; (2) RP3/4 is divi ded into two well-defined bran ches RP3/4a and RP3/4b and RP3/4a is strong and not zi gzagged (Nel et al., 2005) 337 RP3/4 (0) RP3/4 is branching normally on RP ; (1) RP3/4 is apparently branching on MA (Nel et al., 2005) 338 RP3/4-MA (0) RP3/4 and MA more or less parallel; (1) RP3/4 and MA strictly prallel with only one row of cells til l the hind margin; (2) RP3/4 and MA more or less undulate and distinctly divergent near the hind margin (Bechly et al., 1998). 339 RP4 (0) with strong secondary branches; (1) no strong sec ondary branches (Nel et al, 2001) 340 Rspl (0) Rspl absent; (1) at least a weakly defined Rspl, parallel to IR2; (2) very well defined Rspl present in both wing pairs, and pa rallel to IR2; (3) very distinct and strongly curved Rspl present in both wing pairs. (Bechly et al., 1998) 341 ScP and RA (0) (f) A secondary zigzagge d longitudinal vein between ScP and RA, basal of nodus, absent; (1) (f) A secondary zi gzagged longitudinal vein between ScP and RA, basal of nodus, present (Fleck and Nel, 2003) 342 subdiscoidal area (0) (h) subdiscoidal area divided into numerous small cells; (1) (h) subdiscoidal area divided into few large ce lls; (2) not divided (Fleck and Nel, 2003) 343 subdiscoidal area (0) (h) Basal part of subd iscoidal area, if pres ent shorter than the distal part; (1) (h) Basal part of subdiscoidal area, if present at least as long as the distal part (Fleck and Nel, 2003) 344 sub-discoidal cell (0) posteriors, the ventra l border of the subdiscoidal cell stright or concave; (1) posteriors, the ve ntral border of the subdiscoida l cell clearly convex (Nel et al., 1993) 158

PAGE 159

345 sub-discoidal cell (0) posteriors, subdiscoidal cell as long as th e discoidal cell; (1) posteriors, subdiscoidal cell longer th an discoidal cell (Nel et al., 1993) 346 sub-discoidal cell (0) posteriors, subdiscoidal cell closed ventrally; (1) posteriors, subdiscoidal cell meeting up with the ventral wing margin; (2) posteriors, AA bends to dissapears distally, subdiscoidal cell very big and open proximally. 347 sub-discoidal cell (0) (h) sub-discoidal cells at a disto-ventral angl e little or not acute, in the posteriors; (1) sub-discoidal cells in the form of a triangle a disto-ventral angle very acute, in the posteriors. (Nel et al., 1993) 348 sub-discoidal cell (0) (f) sub-discoidal cells at a disto-ventral angl e little or not acute, in the anteriors; (1) sub-discoi dal cells in the form of a tria ngle a disto-ventral angle very acute, in the anteriors (Nel et al., 1993) 349 Subdiscoidal space (0) sub-discoidal space is in the form of quadarangle; (1) subdiscoidal space is in the form of a tr iangle; (2) very small & triangulate (triadophblebiamorpha).(Nel et al., 1993) 350 supertriangle (0) The anterior margin (MA) of the supratriangle is not very rounded; (1) The anterior margin (MA) of the supratrian gle is very rounded (Jarzembowski, Nel, 1996,) 351 size, width (0) (h) the widest part of the hi ndwing is opposite Ax1 or nearly so; (1) (h) the widest part of the hindwing is distal of Ax1 (Fleck and Nel, 2003) 352 posterior margin; (0) (f-h) not ches along posterior wing margin absent or weak; (1) (f-h) notches along posterior wing marg in strong (Fleck and Nel, 2003) 159

PAGE 160


PAGE 161


PAGE 162


PAGE 163


PAGE 164


PAGE 165


PAGE 166


PAGE 167


PAGE 168


PAGE 169


PAGE 170


PAGE 171


PAGE 172


PAGE 173


PAGE 174


PAGE 175


PAGE 176


PAGE 177

LIST OF REFERENCES Artiss, T., T. R. Schultz, D.A. Polhemus, and C. Simon. 2001. Molecular Phylogenetic Analysis of the Dragonfly Genera Libellula, Ladona, and Plathemis (Odonata: Libellulidae) Based on Mitochondrial Cytochrome Oxidase I and 16S rRNA Sequence Data. Molecular Phylogenetics and Evolution 18: 348-361. Asahina, S. 1954. A Morphological Study of a Relic Dra gonfly Epiophlebia superstes Selys (Odonata, Anisozygoptera). Tokyo, Japanese So ciety for the Promotion of Science. Baker, R. H., and R. DeSalle. 1997. Multiple sources of character information and the phylogeny of Hawaiian drosphilids. Systematic Biology 46: 654-673. Bechly, G. 1995. Morphologische Untersuchungen am Flgelgeder der Rezenten Libellen und deren Stammgruppenvertreter (Inse cta; Pterygota; Odonata) unter Besonderer Bercksichtigung der Phylogentischen Systematik und des Grundplanes der Odonata. Petalura (special volume): 1341. Bechly, G. 1996. Morphologische Untersuchungen am Fl gelgeder der Rezenten Libellen und deren Stammgruppenvertreter (Inse cta; Pterygota; Odonata) unter Besonderer Bercksichtigung der Phylogentischen Systematik und des Grundplanes der Odonata. Petalura (special volume): 1402. Bechly, G., A.Nel, X. Martnez-Delcls, and G. Fleck. 1998. Four new dragonfly species from the Upper Jurassic of Germany and the Lo wer Cretaceous of M ongolia (Anisoptera: Hemeroscopidae, Sonidae, and Proterogom phidae fam. nov.). Odonatologica 27: 149-187. Bechly, G. 2007. Phylogenetic Systematics of Odonata November 2007. Online. Internet. Accessed October 2008. . Braukmann, C., and W. Zessin. 1989. Neue Meganeuridae aus dem Namurium von HagenVorhalle (BRD) und die Phylogenie der Mega nisoptera (Insecta, Odonata). Deutsche Entomologische Zeitshrift, N.F. 36: 117-215. Bridges, C. A. 1994. Catalogue of the Family-group, Genus-group and Species-group Names of the Odonata of the World. Urbana Illinois, Published privately. Brower, A., and M. Egan. 1997. Cladistic analysis of Heliconius butterflies and relatives (Nymphalidae: Heliconiiti): a re vised phylogenetic position for Eu eides based on sequences from mtDNA and a nuclear gene. Proceedings of the Royal Society of London Series B-biological sciences 264: 969-977. Brown, J. M., M. A. McPeek, and M. May. 2000. A phylogenetic perspective on habitat shifts and diversity in the North American Enallagma damselflies. Systematic Biology 49(4): 697-712. 177

PAGE 178

Bybee, S. M., S. D. Taylor, R. Nelson and M.F. Whiting. 2004. A phylogeny of robber flies (Diptera: Asilidae) at the subfamilial level: molecular evidence. Molecular Phylogenetics and Evolution 30: 789-797. Bybee, S.M., T.H. Ogden, M.A. Branham, and M.F. Whiting. 2008. Molecules, morphology and fossils: a comprehensive approach to odona te phylogeny and the evolution of the odonate wing. Cladistics 24: 477-514 Carle, F. L. 1982. The wing vein homologies and phyl ogeny of the Odonata: a continuing debate. Societas Internationalis Odona tologica Rapid Communications 4: 1-57. Carle, F. L. and K. M. Kjer. 2002. Phylogeny of Libellula Linnaeus (Odonata: Insecta). Zootaxa 87: 1-18. Carle, F. L., K.M. Kjer, and M.L. May. 2008. Evolution of Odonata, with special reference to Coenagrionidea (Zygoptera). Arthropod Systematcs & Evolution 66: 37-44. Carpenter, F.M. 1992. Treatise on Invertebrate Paleontolog y. Part R. Arthopoda. Vol 3 and 4. Superclass Hexapoda. Geological So ciety of America and University of Kansas: Boulder, CO & Lawrence, KS. Chapman, R. F. 1982. The Insects structure and function. Cambridge, Massachussetts, Harvard University Press. Chippendale, P. T., and J. J. Wiens 1994. Weighting, Partitioning, and combining characters in phylogenetic analysis. Systematic Biology 43: 278-287. Cognato, A., and A. P. Vogler. 2001. Exploring data interaction and nucleotide alignment in a multiple gene analysis of Ips (Coleoptera: Scolytinae). Systematic Biology 50: 758-780. Corbet, P. S. 1999. Dragonflies: Behavior and Ecology of Odonata. Ithaca, NY, Cornell University Press. Dumont, H. J., J. R. Vanfleteren, J. F. De Jonckheere, and P. H. H. Weekers. 2005. Phylogenetic relationships, divergence time estimation, and global biogeographic patterns of calopterygoid damselflies (Odonata, Zygoptera) inferred from ribosomal DNA sequences. Systematic Biology 54: 347-362. Fleck, G., A. Nel, G. Bechly, and X. Martnez-Delcls. 2001. Revision and Phylogenetic affinities of the Jurassic Steleopteridae Handl irsch, 1906 (Insecta: Odonata: Zygoptera). Insect Systematics & Evolution 32: 285-305. Fleck, G., and A. Nel. 2003. Revision of the Mesozoic family Aeschnidiidae (Odonata: Anisoptera). Zoologica 153: 1-180. 178

PAGE 179

Fleck, G., G. Bechly, X. Martnez-Delcl s., E. A. Jarzembowski and A. Nel. 2004. A revision of the upper Jurassic-Lower Cret aceous dragonfly family Tarsophlebiidae, with a discussion on the phylogenetic positions of the Tarsophlebiida e and Sieblosiidae (Insecta, Odonatoptera, Panodoanta). Geodiversitas 26: 33-60. Fraser, F. C. 1954. The origin and descent of the order Odonata based on the evidence of persistent archaic characters. Transactions of the Royal Ento mological Society of London (B) 23: 89-94. Fraser, F. C. 1957. A reclassification of the Order Odona ta. Royal Zoological Society of New South Wales, Sydney. Gatsey, J., G. Amato, M. Norell, R. DeSalle, and C. Hayashi. 2003. Combined support for wholesale taxic atavism in gavialine crocodylians. Systema tic Biology 52: 403-422. Gatsey, J., and R. H. Baker. 2005. Hidden Likelihood Support in Genomic Data: Can FortyFive Wrongs Make a Right? Sy stematic Biology 54: 483-492. Gatsey, J., P. M. O'Grady, and R. H. Baker. 1999. Corroboration among data sets in simultaneous analysis: Hidden support for phyl ogenetic relationships among higher level artiodactyl taxa. Cladistics 15: 271-313. Giribet, G. 2001. Exploring the behavior of POY, a program for direct optimization of molecular data. Cladistics 17: s60-s70. Goloboff, P. A., S. J. Farris, and K. Nixon. (2003). TNT: Tree analysis using New Technology. Tucumn, Argentina, Published by the author. Giribet, G., G. D. Edgecombe, J. M. Carpen ter, C. A. D'Haese, and W. C. Wheeler. 2004. Is Ellipura monophyletic? A combined analysis of basal hexapod relationships with emphasis on the origin of insects. Organisms, Diversity and Evolution 4: 319-340. Grimaldi, D., and M.S. Engel. 2005. Evolution of the Insects. Cambridge University Press, New York, New York. Groeneveld, L. F., V. Clausnitzer, and H. Hadrys. 2007. Convergent evolution of gigantism in damselflies of Africa and South America? Ev idence from nuclear and mitochondrial sequence data. Molecular Phylogene tics and Evolution 42: 339-346. Hamilton, K. G. A. 1972. The insect wing. Part II. Vein ho mology and the archetypal insect wing. Journal of the Kansas Entomological Society 45(1): 54-58. Hasegawa, E., and E. Kasuya. 2006. Phylogenetic analysis of th e insect order Odonata using 28S and 16S rDNA sequences: a comparison between da ta sets with different evolutionary rates. Entomological Science 9: 55-66. Hennig, W. 1981. Insect Phylogeny. New York, John Wiley and Sons. 179

PAGE 180

Henrotay, M., A. Nel, a nd E.A. Jarzembowski. 1997. New protomyrmeleontid damselflies from the Triassic of Australia and the Lias sic of Luxembourg, with a description of Tillyardomyrmeleon petermilleri gen.n. & sp.n. (Archizygopte ra: Protomyrmeleontidae). Odonatologica 26: 375-506. Huelsenbeck, J.P., and K.A Crandall. 1997. Phylogeny estimation and hypothesis testing using maximum likelihood. Annual Review of Ecology and Systematics 28: 437-466. Huelsenbeck, J.P., F. Ronquist, R. Nielsen, and J.P. Bollback. 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294 (5550): 2310-2314. Huguet, A., A. Nel, X. Martinez-Delcl os, G. Bechly, and R. Martins-Neto. 2002. Preliminary phylogenetic analysis of the Protanisoptera (Insecta: Odonatoptera). Geobios 35: 537-560 Jarzembowski, E.A. and A. Nel. 1996. New fossil dragonflies from the Lower Cretaceous of SE England and the phylogeny of the superfamily Libelluloidea (Insecta: Odonata). Cretaceous Research 17:67-85 Kesel, A. B. 2001. The ultralight aerofoils of insects an evolutionary masterpiece. ZoologyAnalysis of Complex Systems 103: 222-229. Kiyoshi, T., and T. Sota. 2006. Differentiation of the dragonfly genus Davidius (Odonata : Gomphidae) in Japan inferred from mitochondr ial and nuclear gene genealogies. Zoologica Science 23: 1-8. Kjer, K. M. 2004. Aligned 18S and Insect Phyloge ny. Systematic Biology 53(3): 506-514. Kristensen, N. P. 1991. Phylogeny of extant hexapods. The Insects of Australia. Melbourne, Melbourne University Press. Lieftinck, M. A. 1975. The dragonflies (Odonata) of New Caledonia and the Loyalty Islands. Part 1. Imagines. Cahiers O.R.S.T.O.M., Serie Hydrobiologie 9: 127-166. Lohmann, H. 1996. Das phylogenetische System der Anis opteren (Odonata). Entomologische Zietschrift 106: 209-252. Madsen, H.O.R, and A. Nel. 1997. Two new fossil species of Gom phaeschna Selys, 1871 in the Paleocene/Eocene of Denmark (Odonata: Aeshnoidea) Annales de la Socit Entomologique de France 33: 285-293 McConnell, J. 2006. The flat bed scanner as a close-up di gital camera. Hortscience 41(4): 1044. Misof, B. C., C. L. Anderson, and H. Hadrys. 2000. A phylogeny of the damselfly genus Calopteryx (Odonata) using mitochondrial 16S rDNA ma rkers. Molecular Phylogenetics and Evolution 15: 5-14. 180

PAGE 181

Misof, B. C., T. R. Buckley, G. Fleck, and K. P. Sauer. 2001. Phylogenetic signal and its decay in mitochondrial SSU and LSU rRNA gene fragments of Anisoptera. Molecular Biology and Evolution 18: 27-37. Misof, B. C. 2002. Diversity of Anisoptera (Odonata): Inferring speciatio n processes from patterns of morphological diversity. Zoology 105: 355-365. Mitchell, F. L., and J. L. Lasswell. 2000. Digital dragonflies. American Entomologist 46 : 110116. Munz, P. A. 1919. A venational study of the suborder Z ygoptera (Odonata) with keys for the identification of genera. Me moirs of the American Ento mological Society 3: 1-78. Naylor, G., and F.Kraus. 1995. The relationship between s and m and the retention index. Systematic Biology 44: 559-562. Needham, J. G. 1903. A genealogic study of dragon-fly wi ng venation. Proceedings of the US National Museum 26: 703-764. Nel, A., and F. Escuillie. 1992: Presence of the genus Sten olestes Scudder, 1895 in the Oligocene laminites of Reve st-des-Brousses (Luberon, Fran ce). Entomologiste 48: 337-349. Nel, A., X. Martnez-Delcls, J.C. Paicheler, and M. Henrotay. 1993. Les Anisozygoptera fossiles phylognie et classification (O donata). Bois-d'Arcy, France 311 S., S.F.O. Nel, A., F. Papier, L. Grauvogel-Stamm, and J.C. Gall. 1997. Voltzialestes triasicus gen.nov.; sp. nov., le premier Odonata Protozygoptera du Trias infrieur des Vosges (France). Palaeontologia. Lombarda, New Series 5: 25-36. Nel, A., G. Bechly, E.A. Jarzembowski, and X. Martnez-Delcls. 1998. A revision of the fossil petalurid dragonflies (Insect a: Odonata: Anisoptera: Petalu rida). Paleontologia Lombarda New Series 10: 1-68. Nel, A., E.A. and Jarzembowski. 1998. New protomyrmeleontid dragonflies from the lower Cretaceous of southern England (In secta, Odonata, Archizygoptera). Cretaceous Research 19:: 393-402 Nel, A., G. Gand., G. Fleck, O. Bethoux, and J. Lapeyrie. 1999. Saxonagrion minutus nov. gen. et sp., the oldest damsefly from the uppe r Permian of France (Odonatoptera, Panodonata, Saxonagrionidae nov. fam.). Geobios 32: 883-888. Nel, A., O. Bethoux, G. Bechly, X. Martnez-Delcls, and R. Papier. 2001. The PermoTriassic Odonatoptera of the "Protodonate" grad e (Insecta: Odonatatoptera). Annales de la Socit Entomologique de France 37: 501-525. Nel, A., J.F. Petrulevicius, and X. Martnez-Delcls. 2005. New Mesozoic Protomyrmeleontidae (Insecta: Odonatoptera: Archizygoptera) from Asia with a new phylogenetic analysis. Journal of Systematic Palaeontology 3: 187-201 181

PAGE 182

Nel, A., G. Fleck, R. Garrouste, G. Gand, J. Japeyrie, S.M. Bybee, and J. Prokop. Monograph of the Griffinflies (In secta: Odonata: Meganispoter a). Palaeontographica. Submitted Norberg, R. A. 1972. Pterostigma of insect wins an intert ial regulator of wing pitch. Journal of Compartive Physiology A: Neur oethology, Sensory, Neural, and Behavioral Physiology 81: 922. Novelo-Gutierrez, R. 1995. The larva of Amphipteryx and a reclassification of Amphipterygidae sensu lato, based upon the larva (Zygoptera). Odona tologica 24: 73-87. Ogden, T. H. and M. F. Whiting. 2003. The problem with the Paleoptera problem: sense and sensitivity. Cladistics 19: 432-442. Ogden, T. H., and M. F. Whiting. 2005. Phylogeny of Ephemeroptera (mayflies) based on molecular evidence. Molecular Phyl ogenetics and Evolution 37: 625-643. Ogden, T. H. and M. S. Rosenberg. 2006. Multiple sequence alignment accuracy and phylogenetic inference. Sy stematic Biology 55: 314-328. Ogden, T. H. and M. S. Rosenberg. 2007. How should gaps be treated in parsimony? A comparison of approaches using simulation. Mo lecular Phylogenetics and evolution 42: 817-826. O'Grady, P., and M. Kidwell. 2002. Phylogeny of the subgenus Sophophora (Diptera : Drosophilidae) based on combined analysis of nuclear and mitochondrial sequences. Molecular Phylogenetics and Evolution 22: 442-453. O'Grady, E. W., and M. L. May. 2003. A phylogenetic reassessment of the subfamilies of Coenagrionidae (Odonata: Zygoptera). J ournal of natural history 37: 2807-2834. Pfau, H. K. 1991. Contributions of functional morphology to the phylogenetic systematics of Odonata. Advances in Odonatology 5: 109-141. Pilgrim, E. M., S. A. Roush, and D.E. Krane. 2002. Combining DNA sequences and morphology in systematics: testing th e validity of the dragonfly species Cordulegaster bilineata Heredity 89: 184-190. Pilgrim, E.M., and C.D. von Dohlen. 2008. Phylogeny of the Sympetrinae (Odonata: Libellulidae): further evidence of the homoplas ious nature of wing venation. Systematic Entomlogy 33: 159-174. Pritykina, L.N. 1981. New Triassic Odoanta of middle Asia pp 5-42 in Vishniakova, V.N., G.M. Dlussky, L.N. Pritykina. (eds) Trudy Paleontologiceskogo Instituta Akademii nauk S.S.S.R. Posada, D. and K. A. Crandall. 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics 14: 817-818. 182

PAGE 183

Qi-Bin, L., A. Nel, and H. Di-Ying. 2002. Phylogenetic analysis of the Mesozoic dragonfly family Liupanshaniidae (Insecta: Aeshnoptera: Odonata) Cretaceous Research 23 4: 439-444. Rasnitsyn, A.P., and L.N. Pritykina. 2002. Superorder Libellulidea Laicharting, 1781. Order Odonata Fabricius, 1792. The dragonflies. in Rasnitsyn A.P., Quicke D.L.J. (eds). 2002. The History of Insects, Kluwer Publ., Dordrecht, Boston, London, 517 pp. Rehn, A. C. 2003. Phylogenetic analysis of higher-level relationships of Odonata. Systematic Entomology 28: 181-239. Riek, E. F., and J. Kukalov-Peck. 1984. A new interpretation of dragonfly wing venation based upon early upper carboniferous fossils from Argentina (Insecta, Odonatoidea) and basic character states in Pterygote wings. Canadian Journal of Zoology 62: 1150-1166. Robertson, J. A., J. V. McHugh, and M. F. Whiting. 2004. A molecular phlogenetic analysis of the pleasing fungus beetles (Coleoptera: Erotylidae): evolution of colour patterns, gregariousness and mycophagy. Systematic Entomology 29: 173-187. Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phy logenetic inference under mixed models. Bioinformatics 19: 1572-1574. Ronquist, F., J. P. Huelsenbeck, and P. van der Mark. 2005. MrBayes 3.1 Manual, 2005. Rowe, R. J. 2004. Agnostic behavior in final-instar lavae of Episynlestes cristatus, Synlestes tropicus and Chrismagrion risi (Odonata: Synlestidae), and relati onships within the 'Lestoidea'. australian Journal of Zoology 52: 169-181. Saux, C., C. Simon, and G. S. Spicer. 2003. Phylogeny of the dragonfly and damselfly order odonata as inferred by mitochondrial 12S ri bosomal RNA sequences. Annals of the Entomological Society of America 96: 693-699. Schorr, M., M. Lindeboom, and D. R. Paulson. 2007. World Odonata List, 2007. Schuh, R.T. 2000. Biological Systematics: pr inciples and applications. Cornell University Press, Ithaca, New York. Sebastian, A., M. M. Sein, M. M. Thu, and P. S. Corbet. 1990. Suppression of Aedes aegypti (L.) (Diptera: Culicidae) using augmentative release of dragonfly larv ae (Odonata: Libellulidae) with community participation in Yangon, Myanma r. Bulletin of Entomo logical Research 89: 223-232. Selys, E. 1863. Synopsis des Agrionines. Quatrime lgion. Platycnemis. Bulletin de l'Acadmie Royale de Belgique 2: 147. Skaloudova, B., V. Krivan, and R. Zemek. 2006. Computer-assisted estimation of leaf damage caused by spider mites. Computationa l Electronic Agri culture 53(2): 81-91. 183

PAGE 184

Sorenson, M. D. 1999. TreeRot2Vc. Boston, MA., Boston University. Stoks, R., and M. A. McPeek. 2006. A tale of two diversifications: Reciprocal habitat shifts to fill ecological space along the pond permanence grad ient. American Naturalist 168(6): S50-S72. Sudo, S., K. Tsuyuki, T. Ikohagi, F. Ohta, S. Shida, and J. Tani. 1999. A study on the wing structure and flapping beha vior of a dragonfly. JSME Interan tional Journal Series C-Mechanical systems machine elements and manufacturing 42(3): 721-729. Swofford, D. L. 2002. PAUP*; Phylogenetic Analysis Us ing Parsimony (*and other methods). Sunderland, Massachusetts, Sinauer Associates. Terry, M. D., and M. F. Whiting. 2005. Mantophasmatodea and phylogeny of the lower neopterous insects. Cladistics 21: 240-257. Tillyard, R. J. 1917. The Biology of Dragonflies. Cambridge Cambridge University Press. Tillyard, R. J., and F. C. Fraser. 1938-1940. A reclassification of the order Odonata. Australian Zoologist 9: 125-169, 195-221, 359-390. Trueman, J. W. H. 1996. A preliminary cladistic anal ysis of odonate wing venation. Odonatologica 25: 59-72. Turgeon, J., and M. A. McPeek. 2002. Phylogeographic analysis of a recent radiation of Enallgma damselflies (Odonata: Coenagrion idae). Molecular Ecology 11: 1989-2001. von Ellenrieder, N. 2002. A phylogenetic analysis of th e extant Aeshnidae (Odonata : Anisoptera). Systematic Entomology 27: 437-467. Ware, J.; M. May, and K. Kjer. 2007. Phylogeny of the higher Libe lluloidea (Anisoptera : Odonata): An exploration of the most spec iose superfamily of dragonflies. Molecular Phylogenetic and Evolution 45: 289-310 Westfall, M. J. J., and M. L. May. 2006. The Damselflies of North America. Gainesville, Florida, Scientific Publishers. Wheeler, W. C., D. Gladstein, and J. De Laet. 1996-2003. POY. Version 3.0. /pub/molecular/poy. New York. Wheeler, W. C., M. Whiting, Q. D. Wheeler, an d J. M. Carpenter. 2001. The phylogeny of the extant hexapod orders. Cladistics 17: 113-169. Whiting, M. F., J. M. Carpenter, Q. D. Wheeler, and W. C. Wheeler. 1997. The Strepsiptera problem: Phylogeny of the holometabolous insect orders inferred from 18S and 28S ribosomal DNA sequences and morphology. Systematic Biology 46: 1-68. Whiting, M. F. 2001. Mecoptera is paraphyletic: multiple genes and phylogeny of Mecoptera and Siphonaptera. Zoolog ica Scripta 31: 93-104. 184

PAGE 185

Wootton, R. J. 1991. The functional morphology of the wings of Odonata. Advances in Odonatology 5: 153-169. Wootton, R. J. 1992. Functional morphology of insect wings. Annual Review of Entomology 37: 113-140. Wootton, R. J. 2002. Design, Function and evolution in the wings of holometabolous insects. Zoologica Scripta 31: 31-40. Wootton, R. J. 2003. Reconstructing insect flight performance from fossil evidence. Acta zoologica cracoviensia 46: 88-99. Wootton, R. J., J. Kukalova-Peck, D. J. S. Newman, and J. Muzon. 1998. Smart engineering in the mid-Carboniferous: How well could Pal aeozoic dragonflies fly? Science 282: 749-751. Zessin, W. 1991. Die Phylogenie der Protomyrmele ontidae unter Einbeziehung neuer oberliassischer Funde (Archizygoptera sens. nov.). Odonatologica 20: 97-126. 185

PAGE 186

BIOGRAPHICAL SKETCH Seth Mikaya Bybee was born in Las Vegas, NV. in 1979 and moved to Provo, UT when he was four years old. Followed in the footsteps of his older brothers and parents he applied and was accepted to Brigham Young University. He bega n his freshman year at BYU in the Fall of 1996. After completing his freshman and sophomore years at BYU he served a two year mission for the Church of Jesus Christ of Latter-day Sa ints in the Southern France where he became proficient in French. Bybee re turned to BYU in the fall of 2000 to complete his degree. Bybee joined the Whiting Laboratory of Insect Ge nomics at BYU in the summer of 2002. His undergraduate research resulted in the first molecular estim ate of robber fly (Asilidae) phylogeny. He is most grateful to Michael Whiting for his friendship, encouragement, and generosity. Upon graduation from BYU with a Bachelors degree in Conservation Biology with an emphasis in wildlife ecology, Bybee joined the Branham Laboratory at the University of Florida where followed his passion for evolutio nary biology. Currentl y, Bybee has accepted a Postdoctoral position at the Univer sity of California-Irvine work ing on Insect Vision with Dr. Adriana Briscoe. Along the way Bybee met and married his wife Elizabeth Conner Bybee. They are the parents of five girls. He credits any success and scientific insight to his childrens curiosity and his wifes superhero-like support. 186