HHIHHWHIHIIWIIIMmll

llHILIHHHWHHH

3%

THESiS l\\\\\\\\\ Mll\\\\\\\\\\\\\\\\\\ \\l \\\\\\\\\\1\\\2\\\\\\\\l 3 1293

This is to certify that the

dissertation entitled

PHYLOGENY AND EVOLUTION OF FORAGING SPECIALIZATION IN THE TYRANT FLYCATCHERS

presented by

Jeffrey S. Birdsley

has been accepted towards fulfillment of the requirements for

Ph. D . degree in Zoology

953227 ' Major professor W

Dang/Q06 [798/

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

LIBRARY Michigan State University

PLACE IN RETURN Box to remove this checkout from your record. To AVOID FINE return on or before date due. MAY BE RECALLED with earlier due date if requested.

DATE DUE DATE DUE DATE DUE

1/” Wu PHYLOGENY AND EVOLUTION OF FORAGING SPECIALIZATION IN THE TYRANT FLYCATCHERS By Jeffrey S. Birdsley

A DISSERTATION

Submitted to Michigan State University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

W. K. Kellogg Biological Station and Department of Zoology

1998 ABSTRACT

PHYLOGEN Y AND EVOLUTION OF FORAGING SPECIALIZATION IN THE TYRANT FLYCATCHERS By Jeffrey S. Birdsley

Little is known about the evolutionary implications of ecological specialization. Specialized species have long been considered to be evolutionary "dead-ends", having lower potential to evolve into different fonns than do their more generalized relatives. I tested this idea by performing a phylogenetic reconstruction of the evolution of specialization and generalization in foraging tactics in the tyrant flycatchers (Aves: Tyrannidae). I first performed a phylogenetic analysis of nearly all of the approximately 100 tyrannid genera using all available morphological and behavioral data in the literature. I then mapped foraging behavior onto the resulting most-parsimonious trees. In the flycatchers, specialized lineages have not been dead-ends but have given rise to different forms as often as have generalized lineages. A possible explanation is that, although specialized foraging tactics may be accompanied by specialization of some morphological structures which are inefficient for many other tactics, some of the specialist’s remaining morphology may not be so specialized. These structures may often allow expansion of the specialist’s foraging repertoire, particularly to new tactics which don’t require use of the specialized structures. Results of the phylogenetic analysis support three of the five previously proposed tyrannid assemblages, as well as several lower level relationships. No higher level relationships are well supported but there is some support for a basal placement of the flatbills and tody-tyrants. Nasal septum characters that past workers have considered conservative and phylogenetically informative are supported as important synapomorphies of the kingbird assemblage and a restricted Empidonax assemblage. A nasal capsule character provides equivocal support for a Myiarchus assemblage without Attila. Several foraging tactics are supported as homologous within large groups of taxa, demonstrating a strong historical component to foraging ecology. ACKNOWLEDGEMENTS

I wish to thank the members of my guidance committee Torn Getty, Don Beaver, Fred Dyer, Don Straney, and Scott Winterstein. I thank my advisor Tom Getty for pushing me to find the important questions to ask. The quality of this dissertation was greatly improved as a result. I thank Tom and the rest of the faculty at Kellogg Biological Station for providing a truly incredible amount of financial and equipment support. This work would not have been possible without years of unrestricted access to KBS’s excellent computer facilities. Thanks to John Gorentz for years of help and some last-minute assistance to improve the quality of my presentation. Carolyn Hammarskjold did much invaluable library legwork for me, especially during the long months I spent writing this dissertation. You can stop worrying Carolyn, I finally took that vacation. I wish to thank a host of KBS and MSU colleagues and friends for inspiration, ideas, support and companionship. In particular, Andy Turner, Chris Rogers, Casey Huckins, Jeff White, Jeff Dudycha, and Rich Leschen were always there to talk and were very helpful in the development of this research and of my approach to biology. Rob Olendorf, Jill Fisher, Sandy Halstead, Kellie Ellis, Doug Jakubiak, Jackie Smith, Elizabeth Smiley, Beth Capaldi, John Wallace, Steve Fradkin, Mike Rondinelli, Lisa Horth, Puja Batra, Emily Lyons, and Steven Mom'ssey all helped make this journey possible and enjoyable. Finally and most importantly, I must thank Becky Fuller for the encouragement and ‘push’ I needed to see this through. You’re awesome Becky, keep it up! TABLE OF CONTENTS

LIST OF TABLES ...... v

LIST OF FIGURES ...... vi

CHAPTER 1 INTRODUCTION ...... l

CHAPTER2 PHYLOGENY OF THE TYRANT FLYCATCHERS BASED ON MORPHOLOGY AND BEHAVIOR ...... 3 INTRODUCTION ...... 3 METHODS ...... 5 RESULTS AND DISCUSSION ...... 8 Flatbill and tody-tyrant assemblage ...... 15 Kingbird assemblage ...... 16 Empidonax assemblage ...... l7 Myiarchus assemblage ...... l9 Elaenia assemblage ...... l9 Problematic genera ...... 20 Homology of nasal septum characters ...... 22 Summary ...... 24

CHAPTER3 THE EVOLUTIONARY IMPLICATIONS OF ECOLOGICAL SPECIALIZATION 26 INTRODUCTION ...... 26 METHODS ...... 27 Reconstructing the evolution of foraging behavior ...... 27 Hypothesis testing ...... 29 RESULTS ...... 29 DISCUSSION ...... 33 Specialization is not a cul-de-sac ...... 33 Evolution of tyrannid foraging behaviors ...... 38 Conclusion ...... 39

APPENDIX ...... 42

LIST OF REFERENCES ...... 61

iv LIST OF TABLES

Table l - The average number of unambiguous changes from each foraging behavior state into any of the others as reconstructed on trees estimated with foraging behavior included. Note that cells on the diagonal from upper left to lower right contain the frequency of stasis for each state ...... 34

Table 2 - Relative rates of stasis and change for each foraging behavior state ...... 35

Table 3 - Relative rates of stasis and change for pooled generalist states and for pooled specialist states ...... 36

Table 4 - Relative rates of stasis and change for pooled generalist states and for pooled specialist states as reconstructed on trees estimated with foraging behavior excluded ...... 36 LIST OF FIGURES

Figure 1 - Strict consensus tree derived from all most-parsimonious trees. Synapomorphies providing unambiguous support in all most-parsimonious trees are numbered. Numbers in circles indicate support for each node in the form of a decay index giving the number of extra steps required to show a node as unsupported. Higher index values indicate more robust clades. Descriptions of characters in Appendix ...... 10

Figure 2 - Adams consensus tree derived from all most-parsimonious trees. Asterisks denote clades which are not present in any of the most—parsimonious trees 13

Figure 3 - Evolution of tyrannid foraging behavior as reconstructed on one of the most parsimonious trees chosen randomly from the set of trees estimated with foraging behavior included. In the legend, generalist foraging behavior states are listed in lower-case letters, specialist states in all capitals ...... 31

vi Chapter 1

INTRODUCTION

Little is known about the evolutionary implications of ecological specialization. Specialized species have long been considered to be evolutionary "dead—ends", having lower potential to evolve into different forms than do their more generalized relatives. Simpson (1953) and Rensch (1959) promoted the ideas that evolution tends to proceed from generalized to specialized forms and that specialized forms have a greater probability of extinction. These ideas continue to prevail in ecology and evolutionary biology despite little evidence to support them. Rausher (1993) presented theoretical evidence that specialization will tend to be an evolutionary cul-de-sac, but empirical tests of this idea have been difficult to conduct since they require examination of long-term macroevolutionary trends. The field of phylogenetic systematics holds much promise for providing the tools necessary to tackle this and other macroevolutionary problems. A phylogenetic hypothesis provides a framework upon which to reconstruct the evolution of behavioral and ecological characters and uncover historical patterns of transition between generalized and specialized states. Recent phylogenetic studies of the evolution of host-use breadth in insects have begun to provide evidence of transitions in both directions - generalist to specialist and specialist to generalist - as well as transitions between different types of specialists and between different types of generalists (Thompson 1994, Futuyma et a]. 1995, Miiller 1996, Pellmyr et al. 1996). However, most of these studies are constrained by comparative data sets and phylogenies too small to detect more than a handful of transitions in host-use. Much more evidence from larger phylogenies and comparative data sets is needed to test the idea that some types of transitions are less likely than others. Recent advances in phylogenetic comparative methods (Brooks and McLennan 1991, Harvey and Page] 1991) and computer software (Maddison and Maddison 1992, Swofford 1993) are making such

1 large-scale studies increasingly tractable. The tyrant flycatchers (Aves: Tyrannidae) are an ideal group with which to address this problem. They are a large family of about 370 species in nearly 100 genera for which exhaustive comparative morphological and behavioral data have been published. These data provide the basis for inferring phylogenetic relationships within the family. One of the behavioral data sets (Fitzpatrick 1980) documents a range foraging techniques and interspecific patterns of specialization and generalization within this range. In this dissertation, I describe a cladistic analysis of all available comparative data for the flycatchers and develop hypotheses of evolutionary relationships of nearly all genera. This provides a phylogenetic framework upon which to reconstruct the evolution of foraging behavior and identify transitions between generalized and specialized states. This reconstruction will allow a test of the hypothesis that specialization constrains foraging behavior evolution. 1 also compare the results of my cladistic analysis with previously published hypotheses of flycatcher relationships. Chapter 2

PHYLOGENY OF THE TYRANT FLYCATCHERS BASED ON MORPHOLOGY AND BEHAVIOR

INTRODUCTION The tyrant flycatchers (T yrannidae) are a primarily Neotropical family of suboscine passerine birds in the superfamily Tyrannoidea which also includes the cotingas (Cotingidae) and manakins (Pipridae). The relationships among the 90 to 100 genera of tyrant flycatchers were historically based on characters of external morphology such as the shape of the bill, wing and tail, and color and pattern of plumage (Sclater 1888, Berlepsch, 1907, Hellmayr 1927). Warter’s (1965) survey of cranial osteological variation in the Tyrannoidea revealed variation in the morphology of the nasal capsule and septum, palatine and palatomaxillary bones, interorbital septum, and general shape of the cranium. Warter felt that the nasal region held the most promise for providing useful taxonomic characters and recognized several basic states of the configuration of the nasal septum. Building upon Warter’s (1965) work, W. E. Lanyon (1984, 1985, 1986, 1988a, b, c) completed a nearly exhaustive survey of tyrannoid cranial variation and used five different states of the nasal septum to establish the monophyly of five separate tyrannid assemblages which collectively included nearly all traditional tyrannid genera Ames (1971) documented great variation in suboscine syringeal musculature and support elements. In contrast, the oscine passerines have relatively uniform syringes. The variation in suboscines is thought to exist because, unlike oscines which learn their songs, suboscines apparently sing innate songs (Kroodsma 1996). Species differences in songs presumably arise through evolutionary changes in syringeal morphology in suboscines. This has resulted in a complex of characters with great potential for recovering phylogenetic relationships. Ames ( 1971) had a relatively limited sample of taxa available for study and did not use stains on the syringeal support elements and so could not readily distinguish

3 4 between ossified and cartilaginous elements. Lanyon (1984, 1985, 1986, 1988a, b, 0) studied a nearly exhaustive sample of tyrannoid syringes which he double-stained with alcian blue for cartilage and alizarin red for ossified bone to distinguish the ossified (primarily tracheal) series of elements from the cartilaginous (primarin bronchial) series. This allowed recognition of homologous elements for comparison among taxa and the recognition of elements which are partly ossified and partly cartilaginous. Lanyon used syringeal characters to group taxa within his five assemblages and used characters of the cranium, nesting behavior, plumage, and egg coloration wherever necessary to establish relationships within syringeal groups and occasionally at higher levels. This procedure is the equivalent of weighting these syringeal characters more heavily than others, effectively disallowing convergence in them. Nasal septum characters were effectively weighted even more heavily than syringeal characters since Lanyon assumed them to be synapomorphies of his five assemblages and disallowed convergence in them. Lanyon developed hypotheses of generic relationships based on shared similarities, but because he did not develop explicit character polarization arguments based on his outgroup comparisons, some of his groups were supported by plesiomorphies. For example, in the Elaenia group Lanyon (1988a) used the poorly ossified condition of the anterior segment of the nasal septum to support a clade comprised of Capsiempis, Phaeomyias, and Nesotriccus. He then used the fully ossified condition to support a clade comprised of Serpophaga and Anairetes. Only one of these character states is the derived state and the other is the plesiomorphic, or ancestral, state in this group. Logically, only the derived state may argue for close relationship of the taxa which possess it. I performed a reanalysis of Lanyon’s data plus any other data I could find in the literature. Unlike Lanyon, I used rigorous cladistic methods to identify derived character states and used modern computer software to generate a phylogeny based on the distribution of these derived states. I also weight characters equally and analyze them together in one character matrix containing nearly all tyrannid genera, allowing the possibility of convergence in any character. This provides tests of homology of each of the five states of the nasal septum and a test of monophyly of each of the five tyrannid assemblages Lanyon proposed. In addition, I include five traditional tyrannid genera (Colonia, Machetornis, Muscigralla, Phyllomyias, and Tachuris) which Lanyon could not 5

place in any of his assemblages, and one piprid genus (Neopipo) which Mobley and Prum

(1995) have argued is a tyrannid. Monophyly of the Tyrannidae has been questioned on the basis of DNA-DNA hybridization data (Sibley and Ahlquist 1985). However, a number of syringeal characters appear to support the monophyly of the Tyrannidae. Lanyon (1984, 1986, 1988a) hypothesized the Tyrannidae to be monophyletic based on the presence of internal syringeal cartilages. Prum (1990) questioned the homology of all internal syringeal cartilages but hypothesized monophyly of the Tyrannidae based on the presence of Mm. obliqui ventrales, a pair of intrinsic syringeal muscles. Here I follow the latter authors in assuming monophyly of the Tyrannidae and I add an additional, heretofore unrecognized, syringeal synapomorphy: in nearly all tyrannids, the B1 and B2 syringeal support elements are connected at their ventral tips. This connection is absent in cotingids and piprids (Prum 1992, pers. comm.).

METHODS

I included all 95 genera which served as terminal taxa in Lanyon's (1984, 1985, 1986, 1988a, b, c) five assemblages and my nomenclature follows his. Lanyon provided character support for monophyly of 64 of these genera. Lanyon (1986, 1988a) split the genera Mecocerculus and Myiophobus into three groups each and placed them in different parts of his phylogenies yet retaining their generic names until more comparative data are available. Here I split these genera as Lanyon did. Lanyon (1988b) argued for merging “Terenotriccus” etythrurus into Myiobius and I follow his recommendation but keep Myiobius erythrurus as a terminal taxon separate from the remaining Myiobius. Several traditional tyrannid taxa have a nasal septum morphology that is either unknown or is sufficiently unique that Lanyon (1986, 1988a) refrained from placing them in any of his assemblages, maintaining them incertae sedis: Phyllomyiasfasciatus, P. griseiceps, P. griseocapilla, and the monotypic genera Colonia, Culicivora, Machetomis, Muscigralla, and Tachuris. I included all of these taxa in my analysis except for Phyllomyias griseocapilla and Culicivora. For these two taxa, I had neither cranial nor syringeal data. The specimens apparently do not exist except for one syrinx of Culicivora which Lanyon (1988a) examined but does not describe and which I have not examined. 1 6 also included the genus Neopipo, traditionally placed in the Pipridae, because Mobley and Prum (1995) hypothesized it to belong in the Tyrannidae based on syringeal and plumage data. I extracted character data primarily from Lanyon (1984, 1985, 1986, 1988a, b, c). These papers include photographs of representative syringes for all terminal taxa and photographs of crania for most. Since Lanyon worked on one assemblage at a time, he often did not describe, for a given character, the distribution of states across all tyrannid genera and in outgroups. In these cases I made family-wide assessments and outgroup comparisons using Lanyon’s photographs and other literature sources describing tyrannoid crania (W arter 1965) and syringes (Ames 1971; McKitrick 1985; Prum and Lanyon 1989; Prum 1990, 1992; Mobley and Prum 1995). Lanyon did not make extensive use of syringeal musculature characters so I extracted these from Ames (1971). Since Ames (1971) did not use stains on syringeal support elements I gave precedence to all more recent information on these elements. I consulted Hilty and Brown (1986), Stiles and Skutch (1989), Ridgely and Tudor (1994) and similar guides for plumage characters. Foraging behavior characters are from Fitzpatrick (1980). For nest and egg characters I conducted an exhaustive literature search, drawing information from many sources and adding two which Lanyon did not use: presence of a visor over the entrance of enclosed nests

(character 74) and nests used as a dormitory (character 75). In addition I examined crania, skins, and Lanyon’s cleared-and-stained syringeal specimens at the American Museum of Natural History for any character data which was incomplete or inconclusive in the literature. I polarized character variation in the ingroup by comparison to outgroups (see Appendix for character analyses). A character state which is unique to some part of the ingroup I considered to be derived from the character state found in the remainder of the ingroup and in the outgroups. In cases where all ingroup character states were also found among outgroups, I attempted to argue character polarities using the method of Maddison et a1. (1984) assuming the following outgroup structure: (oscines (Old World suboscines (Fumarioidea ((Cotingidae, Pipridae)Tyrannidae)))). I was unable to confidently polarize one major character, the cranial interorbital septum, the alternative states of which Lanyon (1988a, b, c) used to define several groups. Therefore, I did not use this character. 7

I coded morphological and behavioral variation as 62 binary characters and 14 unordered multistate characters. For some character complexes I hypothesized transition series and coded these as a series of binary characters. I proposed transition series if similar derived states shared significant detail but some states had additional details which appeared to be further derived. Character complexes which appeared to consist of alternative variations of a character I coded as an unordered multistate character and refer to the alternative states by .1, .2, etc. after the character number. I coded ancestral states as 0 and derived states as 1, 2, etc. I coded unknown character states as “?”. I analyzed the data cladistically using PAUP 3.1.1 (Swofford 1993). The matrix was too large to use exact algorithms for finding the globally most-parsimonious (M-P) trees, so I used the heuristic search option which finds locally M-P trees. To increase the probability of finding the globally M-P trees, I repeated this search using all possible combinations of branch-swapping algorithms and stepwise addition options. For each search I set a limit of 5,000 trees to be held in memory. When a taxon was coded as having multiple states for a given character, PAUP interpreted this as uncertainty, not polymorphism. I weighted characters equally. I felt this was justified in the absence of objective criteria for weighting some characters more heavily than others. Equal weighting schemes do, however, carry the assumption of low rates of change in all characters

(Felsenstein 1982). I used consensus trees to summarize the resulting set of M-P trees. I computed a strict consensus tree which identifies groups found in all M-P trees. I also computed an Adams consensus tree which provides greater resolution than the strict consensus by identifying “problem” taxa and placing them apart from groups which consistently appear. For example, if a large group exists in all M-P trees but in some of them is “invaded” by one taxon or a small group of taxa, the Adams consensus leaves the large group intact and places the problem taxa outside of it in a polytomous position whereas the strict consensus simply collapses the large group. The Adams consensus tree provides much information on patterns and trends among alternative trees, but it should be interpreted with care as it may not correspond to any of the M-P trees. To evaluate the evidential support for relationships depicted in the strict consensus tree I used Sorenson’s (1996) TreeRot program and PAUP to compute a decay index (Bremer 1988, Killersjo et al. 1992) for 8 each node present in the strict consensus. Iran ten replicates with random addition sequence and a limit of 800 trees held in memory. The decay index gives the number of extra steps required to show a node as unsupported. For example, if a node which occurs in all M-P trees is not present in trees which are one step longer, that node gets a decay index of 1. Higher index values indicate more robust clades.

RESULTS AND DISCUSSION The cladistic analysis of 76 characters (a total of 100 apomorphies) and 104 taxa resulted in more than 5,000 M-P trees of length 372 and a consistency index (CI) of 0.40, a result not unusual for data sets of this size (e. g. Griffiths 1994). Sanderson and Donoghue (1989) showed that the CI is negatively correlated with the number of taxa in a tree. According to their model, a tree with 60 taxa is expected to have a CI of 0.35. Their model could not be extended beyond 60 taxa, but it is clear that a CI of 0.40 may be as high as could be expected from a tree containing 104 taxa. I found several ‘islands’ of M-P trees using different combinations of PAUP’s search options and found additional islands in the course of computing decay indices. The 9th and 10th decay index replicates found no smaller index values. The strict consensus of all M-P trees (Figure 1) indicates support for the monophyly of many small groups of genera and of three assemblages which correspond more or less to Lanyon’s (1984, 1985, 1986) kingbird assemblage (Phelpsia through Griseotyrannus in Figure 1), Myiarchus assemblage (Rhytiptema through Myiarchus) and Empidonax assemblage (Silvicultrix through Cnemarchus). The monophyly of neither the Elaenia assemblage (Lanyon 1988a) nor the flatbill and tody-tyrant assemblage (Lanyon 19880) is supported. No higher-level relationships are well supported with the present data. The Adams consensus tree (Figure 2) is better resolved but should not be treated as a phylogenetic hypothesis as it is not consistent with any single M-P tree and a few of the clades depicted are not present in any of the M-P trees. However, for the purposes of the following discussion it illustrates many results which are not apparent in the strict consensus tree. Here I discuss my results as compared to each of Lanyon's (1984, 1985, 1986, 1988a, b, c) proposed assemblages. Character numbers refer to characters in the present analysis. Figure 1 - Strict consensus tree derived from all most-parsimonious trees. Synapomorphies providing unambiguous support in all most-parsimonious trees are numbered. Numbers in circles indicate support for each node in the form of a decay index giving the number of extra steps required to show a node as unsupported. Higher index values indicate more robust clades. Descriptions of characters in Appendix A.

lO

outgroup

Platyn'nchus

Todirostrum

Poecilotriccus

Attila

Euscarthmus

Suin’n‘

Camptostoma

Omithion

Serpophaga

Anairetes

Uromyias

Inezia

Myiophobus

Myiophobus I,o

Myiophobus p,r

Myiotriccus

Mecocerculus

Mecocerculus h,p,s Mecocerculus c,m Tyranniscus Colonia Sublegatus Rhynchocyclus Tolmomyias Deltarhynchus 26.1,'69.6 Hamphotrigon I 18.71 (D Phylloscartes l Mionectes 6, 7. 492.34, 76 (4) Leptapogon Machetomis Muscigralla Pseudotn'ccus Corythopis Pseudocolopteryx Polystictus

Neopipo Pyrrhomyias Himndinea |_.__ Elaenia Myiopagis Tyrannulus

Capsiempis

Phaeomyias Nesotriccus

Hemitriccus Lophotn’ccus Oncostoma Cnipodectes

Onychorhynchus

Myiobius a,b, v Myiobius erythrurus Figure 1

11

Rhytiptema Casiomis Sirystes and) Myiarchus Stigmatura 5.7.a® Psaudelaenia 3426) Tachuris Phyllomyias fasciatus 42.29 ® «69.5® Phyllomyias gn'saiceps Phelpsia Pitangus .13, 14. 15 Q Philohydar Legatus Myiczetetes Myicdynastes Canopies Megarynchus Tyrannopsis Tyrannus Empidonomus Gn‘seatyrannus Silvicultrix Ochthoeca Colorhamphus Alectrurus Arundinicola Fluvicola Satrapa Ochthomis Muscisaxicola Agriomis Neoxolmis Hetemxolmis Xolmis Lessonia Hymenaps Knipolegus Pyrocephalus Sayomis Cnemotn'ccus Xenotn’ccus Mitrephanes Contopus Empidonax Aphanotriccus La thrctriccus Gubemetes Muscipipra Myiotheretes Polioxolmis Cnemarchus Figure 1 (cont’d) 12

Figure 2 - Adams consensus tree derived from all most-parsimonious trees. Asterisks denote clades which are not present in any of the most-parsimonious trees. 13 outgroup thnchocyclus Talmomyias Cnipodectes Onycharhynchus Myiobius a,b,v Myiobius erythruws Todlrostrum Poecilotn‘ccus Hamitriccus Lophotriccus Oncostoma Platyrinchus Legatus Myiozetetes Phelpsia Pitangus Philohydar Myiodynastes Canopies Megarynchus Tyrannopsis Tyrannus Ernpidonomus Griseotyrannus Myiophobus I,a Myiophobus Neopipo Pyrrhomyias Hirundlnea

Atfila Deltadvynchus Hamphotfigon Rhytiptema Casiomis

Slrystes Myiarchus Myiophobus p,r Mionectes Leptopogon meefius Phylloscartas Colonia Sublegatus Myiom'ccus Machetomis Muscigralla

Figure 2

l4

Pseudotriccus Corythopis Eleenia Myiopegis Tyrannulus Euscarthrnus Pseudocoloptatyx Polystictus Stigmatura Pseudelaenie Techuris Phyllomyies fescietus Phyllomyias griseiceps inezie Mecocerculus Ummyies Semaphege Aneiretes Capsiempis Pheeomyies Nesotn’ccus Suiriri Omimian Camptostoma Mecocerculus h,p,s Mecocerculus c,m Tyrenniscus Silvicultrix Ochthoece Colorhemphus Hymenops Xenotn‘ccus Cnemotriccus Empidonex Aphenotriccus Lethrotriccus Arundinicole Fluvicole Alectrurus Contapus Mitraphenes Pyrocephelus Gubemetes Muscipipra Se yomis Knipolegus Ochthomis Setrape Muscisaxicole Lessonia Agriornis Xolmis Heteroxolmis Naoxolmis Myiotheretes Polioxolmis Figure 2 (cont’d) Cnemarchus 15

F latbill and tody-tyrant assemblage Lanyon (1988c) defined a flatbill and tody-tyrant assemblage based on the possession of: poorly ossified nasal septa (character 1); enclosed, pendant nests (character 72); and a state of the cranial interorbital septum in which the supraorbital fenestra is reduced. 1 did not use the interorbital septum character because the three alternative states present in the Tyrannidae (1--supraorbital fenestra reduced, 2--both fenestrae open and 3- both ossified) are all present in the first outgroup, cotingas and manakins, and in the second outgroup, the fumarioids. States 1 and 3 are present in Old World suboscines and 2 and 3, at least, are present in oscines. State 1 may be derived at some point within the suboscines, but given this distribution it is not possible to polarize any of these states within the Tyrannidae using the method of Maddison et al. (1984). All M-P trees contain a flatbill and tody-tyrant assemblage (Cnipodectes through Oncostoma in Figure 2) supported by characters 1 and/or 72.2. Some trees include Rhynchocyclus and Tolmomyias within the tody-tyrant clade comprised of Todirostrum, Poecilotriccus (including Taeniotriccus), Hemitriccus (including Myiornis), Lophotriccus (including Atalotriccus), and Oncostoma and therefore the monophyly of neither the tody- tyrants nor the assemblage is supported in the strict consensus (Figure 1). However, Lophotriccus and Oncostoma are sister groups and their sister group is Hemitriccus in all M-P trees. Lanyon’s (1988c) flatbill clade comprised of Rhynchocyclus, Tolmomyias, Onychorhynchus, and Platyrinchus is not supported. A sister group relationship between Rhynchocyclus and Tolmomyias is supported, but they and Platyrinchus may belong elsewhere in the family. In most M-P trees Platyrinchus is supported as sister group to the kingbird assemblage by its dorsally inserting Mm.obliqui ventrales (character 50.2) and its concealed crown patch (character 53)(Figure 2). Platyrinchus was the only member of Lanyon’s (1988c) flatbill and tody-tyrant assemblage which does not build a pendant, enclosed nest with a side entrance (character 72). Platyrinchus builds an open cup nest as do many genera in the kingbird assemblage, but this character may be plesiomorphic in the Tyrannidae and thus unable to support this relationship. Lanyon (1988c) could not determine the relationships of Cnipodectes within the assemblage, maintaining it in a polytomy with his flatbill clade and tody-tyrant clade. My results suggest that Cnipodectes is the sister group to a clade comprised of l6

Onychorhynchus and Myiobius (including erythrurus). The presence of long rictal bristles (character 65) and rufous tails (character 60; lost within Myiobius) supports this relationship. I discuss the hypothesized sister relationship between Onychorhynchus and Myiobius below with the Empidonax assemblage. In all M-P trees, the flatbills and tody-tyrants are basal, either alone or with Platyrinchus and the kingbird assemblage, to a large clade containing all other tyrannids (Figure 2). This large clade is supported by the presence of Mm. obliqui ventrales (character 50) and double, complete syringeal A elements (characters 12 and 13) which are absent in at least some flatbills and tody-tyrants. The monophyly of this large clade is not supported in all trees because the kingbird assemblage and the clade containing Neopipo, Pyrrhomyias, and Hinmdinea are sometimes included in it, sometimes not.

Kingbird assemblage The kingbird assemblage is supported containing exactly the genera which Lanyon (1984) placed in it (Phelpsia through Griseotyrannus in Figure 1). However, my hypothesis of relationships within this assemblage is less well resolved than Lanyon’s (1984). This is due to the fact that Lanyon (1984) used alternative derived states of the ventral connection of the B1 and B2 syringeal support elements to support two major clades within the assemblage: Phelpsia, Pitangus, Philohydar, Legatus, and Myiozetetes have flattened connections (character 26.1) while the remaining seven genera in the assemblage have rounded connections (character 26.2). Since flattened connections also occur in nearly all tyrannids, my family-wide analysis indicates that flattened connections are ancestral in the kingbird assemblage and thus cannot argue for relationships within it. However, the derived presence of rounded connections (character 26.2) supports the same clade (Myiodynastes, Canopias, Megarynchus, Tyrannopsis, Tyrannus, Empidonomus, and Griseotyrannus) which Lanyon (1984) hypothesized. My results are also congruent with Lanyon’s (1984) hypothesis of relationships within this clade of seven genera. In addition to the flattened connections of the B1 and B2 syringeal support elements (character 26.1), Lanyon (1984) grouped Phelpsia, Pitangus, Philohydar, Legatus, and Myiozetetes together on the basis of the A3 element providing greater support for each bronchus. My coding scheme differs in that I coded my hypothesized transition series 17

(characters 12-15) in terms of the number of double, complete A elements (having 1 or having 2 or more) not their identity (A2, A3, or A4) to simplify coding. The result is that two of Lanyon’s (1984) nodes -- one supporting this clade of five genera, another supporting a clade comprised of Phelpsia, Pitangus, and Philohydar -- are not supported in some M-P trees.

Empidonax assemblage A restricted Empidonax assemblage is supported containing the same genera which Lanyon (1986, 1988b) placed in it (Silvicultrix through Cnemarchus in Figure 1) except for the four genera which comprise his Myiophobus group: Myiophobus, Pyrrhomyias, Hirundinea, and Myiobius. In some, but not all, M-P trees the monophyly of this restricted Empidonax assemblage is supported by the presence of a nasal septum with basal trabecular plate and anterior notch (character 4.1) and wide posterior forking (character 9.1). In the remainder of trees these characters support this assemblage with Myiophobus included basally. Within the Empidonax assemblage, these results support Lanyon’s (1986) hypothesis of a basal position of the chat-tyrants Ochthoeca, Silvicultrix and Colorhamphus. These results do not support the monophyly of Lanyon’s (1986) Ochthoeca group comprised of these three genera plus Amndinicola, F luvicola, and Alectrurus because the latter three genera share several plumage, habitat and nasal capsule characters with other Empidonax assemblage genera such as Heteroxolmis and Hymenops. Most M-P trees support the monophyly of an Amndinicola, F luvicola, and Alectrurus clade (Figure 2). The monophyly of a large clade comprised of all Empidonax assemblage genera except Ochthoeca, Silvicultrix and Colorhamphus is supported in all M-P trees by one to three unambiguous synapomorphies including syringeal characters 14, 19, -29, and 48 and plumage character -56. 1. However none of these characters provides support in all trees. Lanyon (1986) hypothesized monophyly of a bush-tyrant clade comprised of Myiotheretes, Polioxolmis and Cnemarchus but was unable to resolve relationships among them. These results support this clade and support Polioxolmis and Cnemarchus as sister groups based on the presence of well developed Mm. obliqui ventrales (character 51). Lanyon (1988a) used this character in the Elaenia assemblage but apparently did not 18 recognize it in the Empidonax assemblage. The double, complete, ossified syringeal A elements (characters 13 and 15) and concealed crown patch (character 53) present in Lanyon’s (1986, 1988b) Myiophobus group are not present in the rest of the Empidonax assemblage and argue for the placement of these genera elsewhere. In the set of M-P trees, Myiophobus, Pyrrhomyias, and Hirundinea are allied with several different groups outside the Empidonax assemblage and this is reflected in their polytomous placement in the Adams consensus (Figure 2). A sister group relationship between Myiobius and Onychorhynchus is fairly well supported by the presence of at least two double, complete, ossifled syringeal A elements (characters 12, 14, 15). Myiobius also has the long rictal bristles (character 65) and erythrurus has the rufous tail (character 60) found in this “flatbill” clade. A close association between Myiobius and Onychorhynchus has been hypothesized by a number of past workers (Ames 1971, Traylor 1977, Traylor and Fitzpatrick 1982) based on these syringeal and plumage characters. Myiobius builds a pendant, enclosed nest as do Onychorhynchus and Cnipodectes, but in the present analysis this character does not support the relationship of Myiobius to these genera since I coded the globular nests (character 72.1) of Myiobius and elongate nests (character 72.2) of Onychorhynchus and Cnipodectes as alternative unordered derived states. These results suggest that the globular nests of Myiobius are derived from elongate nests. Myiobius (formerly Terenotriccus) erythrurus is the sister group to the other Myiobius species, congruent with Lanyon’s (1988b) merger of erythrurus into Myiobius. A sister group relationship between Pyrrhomyias and Hirundinea is supported, in agreement with Lanyon (1986, 1988b). Assuming it really is a tyrannid and not a piprid, Neopipo is supported as their sister group, in agreement with one of Mobley and Prum’s (1995) alternative hypotheses. However, cranial data for Neopipo are still needed in order to fully evaluate this hypothesis. The nasal septa of “Myiophobus” lintoni, ochraeeiventris, phoenicomitra, and roraimae have trabecular plates which are elevated above the base of the septum (character 4.2) which caused Lanyon (1986, 1988a) to remove them from Myiophobus and place them in the Elaenia assemblage. He refrained from creating new generic names until more comparative data are available. Myiophobus, now including only the species cryptoxanthus, fasciatus, flavicans, inornatus and pulcher, is supported as the sister group 19 to “Myiophobus” lintoni and ochraceiventris in some M-P trees (Figure 2), but in no trees are “Myiophobus” phoenicomitra and roraimae sister to other Myiophobus. These results support Lanyon’s (1986, 1988a) conclusion that Myiophobus as traditionally delimited (T raylor 1979, Sibley and Monroe 1990) is not monophyletic, with at least phoenicomitra and roraimae representing a distinct genus.

Myiarchus assemblage The monophyly of the Myiarchus assemblage (Lanyon 1985)(Attila through Myiarchus in Figure 2) is supported in some, but not all, M-P trees. This clade is supported by the presence of cavity nesting (character 70). A restricted assemblage comprised only of Rhytiptema, Casiornis, Sirystes, and Myiarchus is supported in all trees, as is a Deltarhynchus - Ramphotrigon clade. Lanyon (1985) described extreme similarity among the crania and syringes of Myiarchus, Sirystes, and Casiomis. Unable to develop a hypothesis of their relationships, he grouped them together as a single terminal taxon in his phylogeny. My results indicate that Sirystes and Myiarchus are sister groups. This relationship is supported by the presence of dorsally divergent B1 and B2 syringeal support elements (character 28), a character which Lanyon (1984) used in the kingbird assemblage but apparently did not recognize in the myiarchines.

Elaenia assemblage The monophyly of the Elaenia assemblage (Lanyon 1988a) is not supported (Figure 1). However, in all M-P trees a large clade containing most or all of the genera of the Elaenia assemblage plus the restricted Empidonax assemblage is supported by the presence of Mm. tracheolaterales which cover the ventral surface of the trachea and diverge at insertion (character 49.1). This clade always includes the genera from Pseudotriccus through Cnemarchus in Figure 2 and in some trees also includes Myiophobus, “Myiophobus” lintoni, ochraceiventris, phoenicomitra, roraimae, and Mionectes through Muscigralla in Figure 2. All M—P trees also contain a ‘tyrannulet’ clade corresponding more or less to Lanyon’s (1988a) Elaenia group in the Elaenia assemblage (Euscarthmus through Tyranniscus in Figure 2). This group is characterized in all trees by the apparently homologous presence of superciliary eye stripes (character 62), specialized perch-gleaning 20 foraging behavior (character 69.5), and unmarked eggs (character 76). The clade comprised of Elaenia, Myiopagis, and Tyrannulus is, in some M-P trees, included in this group and Euscarthmus, Pseudocolopteryx, and Polystictus sometimes come out of it. Therefore its monophyly is not supported in the strict consensus (Figure 1). These results support the monophyly of several of Lanyon’s (1988a) smaller clades within the Elaenia assemblage. A clade comprised of Elaenia, Myiopagis, and Tyrannulus and a clade comprised of Capsiempis, Phaeomyias, and Nesotriccus are both supported. Sister group relationships between Stigmatura and Pseudelaenia, between Pseudotriccus and Corythopis, between Zimmerius and Phylloscartes and between Mionectes and Leptopogon are supported. Tyranniscus is the sister group to “Mecocerculus” calopterus and minor, supporting Lanyon’s (1988a) grouping of these taxa and his conclusion that the genus Mecocerculus as traditionally delimited (T raylor 1979, Sibley and Monroe 1990) is polyphyletic. In some of the M-P trees, “Mecocerculus” hellmayri, poecilocercus, and stictopterus are the sister group to the clade comprised of Tyranniscus and “Mecocerculus” calopterus, minor based on the presence of an additional, posterior trabecular plate in the nasal septum (character 6), but in no trees is Mecocerculus (leucophrys) closely related to these groups. The genera Pseudocolopteryx and Polystictus, which Lanyon (1988a) placed in a large polytomy in his Elaenia group, appear to be sister taxa. This relationship is supported in all M-P trees by two or three unambiguous synapomorphies including habitation of open grassland or marsh (character 68.1), syringeal characters 21.2, —43, and 50.1, and a trabecular plate in the nasal septum which is slenderly forked posteriorly (character 9.2). Lanyon (1988a) did not consider the posterior forks to be homologous in these two genera, apparently because the fork of Pseudocolopteryx is much deeper than that of Polystictus. My hypothesis of their homology is not refuted by the present data set since they are homologous in all M-P reconstructions.

Problematic genera This analysis resulted in hypotheses of relationships for most of the problematic genera which Lanyon (1986, 1988a) maintained incertae sedis. A clade comprised of the problematic Tachuris, Phyllomyiasfasciatus and P. griseiceps is supported by the loss of two derived characters: the elevated trabecular plate in the nasal septum (character 4.2; 21 cranial data for P. griseiceps lacking) and a Myiarchus-like configuration of the syringeal B1 and B2 elements (character 29). The absence of these derived features should not be interpreted as evidence that these taxa do not belong in the Tyrannidae. They all posses the acute ventral connections of the syringeal B1 and B2 elements (character 26.1) which appears to be a synapomorphy of the Tyrannidae, and they posses the derived Mm. obliqui ventrales (character 50) found in most tyrannids. They also share a derived syringeal character with Stigmatura and Pseudelaenia: a cartilaginous bronchial plate with robust internal cartilages (character 34.2) which supports a clade comprised of these five taxa. Lanyon (1988a) thought the syringes of Phyllomyiasfasciatus and P. griseiceps to be very different from each other and speculated that Phyllomyias (comprised of fasciatus, griseiceps, and griseocapilla) is polyphyletic. I examined Lanyon’s cleared-and-stained syringeal specimens (three of fasciatus and one of griseiceps) and found them to be quite similar. The syrinx of griseiceps gives the impression of being a more ossified version of the fasciatus syringes. Their cartilaginous bronchial plates are very similar in detail and the internal cartilages of both species are attached to narrow caudal extensions of the bronchial plates. Besides the degree of ossification, the only conspicuous difference between the syringes of the two species is the shorter internal cartilages of griseiceps which lack the large, amorphous ventral and caudal extension present in fasciatus. These differences are within the range of intrageneric variation in the Tyrannidae. These results indicate that Phyllomyiasfasciatus and P. griseiceps are closely related, possibly sister species (pending examination of crania of P. griseiceps and P. griseocapilla and syringes of P. griseocapilla). The problematic Machetornis and Muscigralla are supported as sister groups by three characters: the loss of a Myiarchus—like configuration of the syringeal B1 and B2 elements (character 29), habitation of open grassland or marsh (character 68.1), and ground specialist foraging behavior (character 69.9). The problematic Colonia is supported as sister to Sublegatus in all M-P trees by two unambiguous synapomorphies: the presence of a superciliary eye stripe (character 62) and either triangular internal cartilages (character 37) or loss of double, complete, ossified syringeal A elements (character 13). 22

Homology of nasal septum characters My results provide a test of the homology of each of the states of the nasal septum which Lanyon (1984, 1985, 1986, 1988a, c) used to define his assemblages. The homology of the poorly ossified septum (character 1), one of the three characters which Lanyon (19880) used to define his flatbill and tody-tyrant assemblage, is supported in all M-P trees for Cnipodectes, Onychorhynchus, Todirostrum, Poecilotriccus, Hemitriccus,

Lophotriccus, Oncostoma, and in some trees Rhynchocyclus, Tolmomyias, and Platyrinchus as well. The poorly ossified septum in Tachuris is convergent in all trees. Despite the fact that he used this character to support this assemblage, Lanyon (1984, 1985, 1988c) hypothesized poorly ossified or unossified nasal septa to be ancestral in the Tyrannidae because they are widely found among suboscines and among birds in general. More recently, Prum and Lanyon (1989) hypothesized ossified nasal septa such as those in the Schijfomis group (Cotingidae) to be ancestral within the tyrannoids and this was the basis for my polarization argument for character 1. The polarity of this character may need closer examination. In my broad survey of passerine crania, I found that most oscines have unossified septa; Old World suboscines have both ossified and unossified septa and Prum’s (1993) phylogeny suggests ossified septa are ancestral in this group; fumarioids have unossified septa except for in the highly ossified crania of Dendrocolaptes; cotingid and piprid septa range from the more or less ossified septa of piprids, Schtflornis, Pachyramphus, and others to the fully ossified septa of Rupicola, Ampelion, Gymnoderus and others. Using the algorithm of Maddison et a1. (1984) and the outgroup structure previously described, if we assume the unossified state is ancestral in fumarioids and the ossified state is ancestral in piprids and cotingids, then the ancestral state of the Tyrannoidea is unossified (contra Prum and Lanyon [1989]) and the ancestral state of the Tyrannidae is equivocal. Greater resolution of higher level fumarioid and tyrannoid relationships may help assess the polarity of these character states within the Tyrannidae. But of perhaps greater value than the degree of ossification is the information potentially existent in the morphology of the cartilaginous parts of those septa which are poorly ossified. At least some of the tody-tyrants probably have cartilaginous trabecular plates within their cartilaginous septa as evidenced by the ossification of these structures in occasional specimens (Lanyon 1988c). I agree with Warter (1965) that a study of these 23 cartilaginous parts will be necessary to make full use of nasal septum characters in tyrannoid birds. Lanyon (1984) defined his kingbird assemblage partly by the presence of an ossified nasal septum which lacks both a trabecular plate and an internal support rod. I coded both the ossified state and the lack of a trabecular plate as ancestral in tyrannids, but the lack of an internal support rod as derived (character 3). In all M-P trees this character is homologous in all kingbird assemblage taxa and is convergent in Attila, in Colonia, and in Machetomis. The polarization of trabecular plate presence/absence (character 4) may need closer examination since trabecular plates are present in some cotingids and Old World suboscines (W arter 1965, pers. obs.). If flatbills, tody-tyrants, other tyrannoids, or even fumarioids with poorly ossified nasal septa turn out to have cartilaginous trabecular plates, the absence of a plate may be derived in the kingbird and Myiarchus assemblages. Lanyon (1985) defined his Myiarchus assemblage (without Attila) partly by the presence of an ossified nasal septum with an internal support rod but lacking a trabecular plate. I coded all of these character states as ancestral in the Tyrannidae, so there are no nasal septum characters supporting this group, with or without Attila, in my analysis. Lanyon (1985) recognized that these genera (except Attila) have ossified nasal capsules (character 10) which surround the septum, but he did not explicitly include this in his list of characters. In some M-P trees in the present analysis, character 10 supports the Myiarchus assemblage (Deltarhynchus, Ramphotrigon, Rhytiptema, Sirystes, Casiomis, and Myiarchus) without Attila. I examined the same four crania of Attila spadiceus and two of A. cinnamomeus which Lanyon ( 1985) did and, contrary to his report, found one (A. cinnamomeus, AMNH 11616) with a completely ossified nasal capsule. A larger sample of Attila crania may show a significant tendency for nasal capsules to ossify and thus may give additional support for the currently tenuous inclusion of Attila in the Myiarchus assemblage. Lanyon (1986) defined his Empidonax assemblage solely by the presence of an ossified nasal septum with basal trabecular plate and anterior notch (character 4.1) and posterior forking (character 9.1). In all M-P trees, character 4.1 is homologous in all genera in the restricted Empidonax assemblage (Silvicultrix through Cnemarchus in Figure 1) and in some trees Myiophobus, Pyrrhomyias, and Hirundinea as well. In all trees, 24 character 4.1 is convergent in Myiobius. Lanyon (1988a) described the surprising appearance of this type of nasal septum in six species of Elaenia, providing further evidence that this seemingly unique character has arisen convergently. In these Elaenia species and in Pyrrhomyias, Hirundinea, and Myiobius the trabecular plate is not conspicuously forked posteriorly (character 9.1) as it is in the restricted Empidonax assemblage. In all M-P trees, character 9.1 is homologous in all genera (except lost in Cnemarchus) in the restricted Empidonax assemblage and in some trees Myiophobus as well. Nasal septa with basal trabecular plates and anterior notches (character 4.1) and posterior forking (character 9.1) appear to be synapomorphies of a restricted Empidonax assemblage which may include Myiophobus basally. Lanyon (1988a) defined his Elaenia assemblage solely by the presence of an ossified nasal septum with a trabecular plate elevated above the ventral edge of the septum (character 4.2). In all M-P trees, character 4.2 is homologous in all genera of ‘tyrannulets’ and allies (Pseudotriccus through Tyranniscus in Figure 2). It is also homologous in a smaller group (“Myiophobus” phoenicomitra and roraimae through Muscigralla in Figure 2). In some trees, character 4.2 is homologous in all of these taxa and is a synapomorphy of a clade comprised of all Elaenia and Empidonax assemblage genera. There is no consistent support for any one hypothesis of transition between trabecular plate states 4.1 and 4.2, but the short, broad plate of Colonia (state 4.3) is derived from elevated plates (state 4.2) in all M-P trees. This makes sense because the plate in Colonia, although uniquely shaped, is (barely) elevated above the ventral edge of the septum and has no anterior notch. Lanyon (1988a) examined one cranium each of Phyllomyiasfasciatus and Tachuris and both lacked trabecular plates (although the septum of Tachuris was simply unossified). In all M-P trees, these are reconstructed as a single derived loss of ossified, elevated trabecular plates (state 4.2).

Summary Cladistic analysis of all currently available data supports three of the five previously proposed tyrannid assemblages, as well as several lower level relationships. No higher level relationships are well supported but there is some support for a basal placement of the flatbills and tody-tyrants. It may be possible to code some characters differently than I 25 have and the effect this might have on results is unknown. For example, my particular coding of the ordered transition series of double, complete syringeal A elements (characters 12-17) results in three of these characters (12, 14, and 15) giving fairly strong support to an Onychorhynchus - Myiobius clade. How this result might change had I coded these characters in terms of the identity of double, complete A elements (A2, A3, A4) rather than their number (having 1 or having 22), is unknown. . Nasal septum characters that past workers have considered conservative and phylogenetically informative are supported as important synapomorphies of the kingbird assemblage and a restricted Empidonax assemblage. A nasal capsule character provides equivocal support for a Myiarchus assemblage without Attila. Many nodes have minimal support and so much more data are needed to resolve relationships in the Tyrannidae, especially at higher levels. Future investigations of cartilaginous septum morphology may provide informative characters in those tyrannids in which the nasal septum does not normally ossify. CHAPTER 3

THE EVOLUTIONARY IMPLICATIONS OF ECOLOGICAL SPECIALIZATION

INTRODUCTION The idea that specialization is a general cause of phylogenetic constraint follows from the assumption that modification of characters during the evolution of a specialist results in loss of some functions and this in turn results in loss of evolutionary potential (Futuyma and Moreno 1988). Generalists are assumed to have a wider range of features and functions from which to draw when faced with environmental change. Rausher (1993) presented theoretical evidence that specialization will tend to be an evolutionary cul-de-sac because simultaneous changes in both behavior and physiology/morphology are required in order for a specialist to I) accept, and 2) perform efficiently on, new resources. Transitions from generalist to specialist occur more readily because only behavior need change initially, and these changes need only involve a loss of some behaviors. The generalist’s physiology and morphology are already able to perform to some degree on the contracted range of resources. Rausher (1993) used this model to explain what is thought to be a general trend in herbivorous insects from host generalists to specialists (Berenbaum 1981, 1983; Price 1983). The phylogeny of the tyrant flycatchers presented in Chapter 2 of this dissertation and a complete comparative data set on tyrannid foraging behavior (Fitzpatrick 1980) provide an unprecedented opportunity to examine large-scale patterns of evolution of specialization and generalization and perform statistical tests of the relative frequencies of different types of transitions. Fitzpatrick (1980) collected data on the foraging behavior of nearly all of the 90-100 flycatcher genera and recognized a number of quantitatively-defined specialist and generalist foraging mode categories. Fitzpatrick (1985) hypothesized that taxa specializing on a particular foraging mode are less evolutionarily flexible than those that generalize on a variety of modes. He showed that these behavioral specialists often have specialized

26 27 wings, bills, or legs that are at the extremes of morphological continua in the Tyrannidae. He reasoned that these extreme morphologies produce trade-offs causing specialists to experience decreased performance when attempting a new foraging mode and that this places them at a competitive disadvantage with species already using that mode. If true, this would make it difficult to depart from their specialized behavior. Based on this line of reasoning, Fitzpatrick (1985) developed evolutionary scenarios in which generalized foragers gave rise to the various specialized genera within each of three tyrannid subfamilies. In this chapter I describe a phylogenetic analysis of the evolution of specialized and generalized foraging behaviors in the approximately 100 genera of tyrant flycatchers and a test of the hypothesis that specialization constrains foraging behavior evolution. If specialized lineages are found to give rise to generalized lineages as often as the reverse transition then specialization is not supported as a general cause of phylogenetic constraint. I also test Fitzpatrick’s (1985) scenarios of behavioral evolution in flycatchers and discuss specific patterns of behavioral diversification suggested by my analysis.

METHODS Reconstructing the evolution of foraging behavior I extracted foraging behavior data from Fitzpatrick (1980) who characterized the foraging behavior of nearly every tyrannid genus and recognized ten ‘foraging mode’ categories. Five categories represent specialist modes which be defined as the use of one particular capture method for greater than 50% of prey captures. Five categories represent generalist modes which he defined as using no single capture method for greater than 50% of prey captures. Since I did not have Fitzpatrick’s raw data, I could not examine the effect of different definitions of specialization (e.g. use of one particular capture method for greater than 80% of prey captures). I coded each of these ten foraging mode categories as alternative unordered derived states of a single character, with the exception of one: I hypothesize the ‘fruit/upward hover-glean generalist’ mode to be the primitive state because that is the mode which seems to most closely correspond to the foraging behavior of the majority of piprids and cotingids (Snow 1982, Hilty and Brown 1986, Marini 1992) which comprise the immediate outgroup. 28 Many workers believe that when studying character evolution, the character of interest should be excluded from the estimation of the underlying phylogeny in order to avoid circular reasoning when using the phylogeny to reconstruct the evolution of that character (Coddington 1988, Sytsma 1990, Brooks and McLennan 1991, Sillén-Tullberg and Moller 1993, Block and Finnerty 1994, Meyer et al. 1994, Thompson 1994). More recently, de Queiroz (1996) argued that the problem is better characterized as one of bias. If the character of interest is one of the characters included in a parsimony analysis to estimate the underlying phylogeny, its reconstruction on that phylogeny may be biased toward underestimating the number of steps or transitions that character has undergone. This potential bias is inherent in parsimony methods (Felsenstein 1985, Saitou 1989) but does not necessarily make it inappropriate to include the character of interest in estimation of the underlying phylogeny. Whether it is inappropriate or not depends on the particular hypotheses being tested (Maddison 1990, de Queiroz and Wimberger 1993, de Queiroz 1996). If I reconstruct the evolution of foraging behavior on phylogenies estimated using that character (and many others), any underestimate of the actual number of transitions in foraging behavior is not expected to differ for generalists versus specialists. I should be able to perform an unbiased test of the hypothesis that transitions out of specialist states are as common as transitions out of generalist states. However, underestimating the number of transitions will reduce my total sample size of transitions, thus reducing the power of statistical tests. For this reason I estimated phylogenies both including and excluding foraging behavior and reconstructed the evolution of foraging behavior on both types of trees. I used MacClade software (Maddison and Maddison 1992) to reconstruct the evolution of foraging behavior on each of 900 phylogenetic trees representing samples of 100 trees from each of nine ‘islands’ of most-parsimonious (M-P) trees found in the cladistic analysis described in Chapter 2 of this dissertation. Foraging behavior was included in the estimation of these trees. I also reconstructed the evolution of foraging behavior on each of a similar number of trees sampled from two ‘islands’ of M-P trees estimated with foraging behavior excluded. 29

Hypothesis testing I used MacClade’s chart function to count the number of unambiguous instances of state changes and stasis in foraging behavior on each tree. MacClade does this by visiting each non-terminal branch and determining its starting state and the states of the two descendant branches above it. If these three states are all different from each other, the reconstruction of foraging behavior on that branch is ambiguous and the changes to the two new states are not charted. It is considered ambiguous because MacClade’s algorithms do not allow ancestral branches to be polymorphic and so two new states are not allowed to arise at a single node. If at least two of the three states are the same, that branch is unambiguously reconstructed as having that state. The state of at least one descendant branch will then be the same as the ancestral branch (= stasis) and the state of the other descendant branch will either be the same or different (= change) from the ancestral branch. Ambiguous reconstructions on single branches will reduce the number of changes charted, but ambiguous reconstructions on several successive branches can also reduce the number of instances of stasis if the ancestral state of a branch appears again higher in the tree as one of several alternative states, such that a reconstruction of stasis on several intervening branches is one of several possible reconstructions. To the extent that there are ambiguous reconstructions on a tree, the total number of changes and stasis may be reduced, but this is not expected to differ for generalists versus specialists. Because the number of branches on which a particular state is reconstructed affects the number of opportunities it has to change, I compared relative rates of stasis and change for each foraging behavior state. I charted, across all trees, the average number of changes and stasis in each of the ten foraging behavior states. I used a Mantel-Haenszel test of general association on the resulting 10x2 table to test for differences between generalist and specialist states in their relative rates of stasis and change. In an alternative analysis I pooled the five generalist states and pooled the five specialist states and performed a chi- square analysis on the resulting 2X2 table.

RESULTS Figure 3 shows the evolution of tyrannid foraging behavior reconstructed on one of the M-P trees chosen randomly from the set of trees estimated with foraging behavior 30

Figure 3 - Evolution of tyrannid foraging behavior as reconstructed on one of the most parsimonious trees chosen randomly from the set of trees estimated with foraging behavior included. In the legend, generalist foraging behavior states are listed in lower-case letters, specialist states in all capitals.

31

1U outgroup . Cnipodectes 3‘3? .E Onychorhynchus - gfi Myiobius a b v

- E Todirostrum

;- I Flhynchocyclus :- E Tolmomyias - E} Legatus ' ' - E] Myiozetetes husxxxxmm I Phelpsia 1% I Pitangus :Q‘Wfi I Philohydar E._- Myiodynastes m E Canopies

1: I Megerynchus m E] Tyrennopsis ;: I Tyrannus s. I Empidonomus 2. I Griseotyrannus Deltarhynchus . _ E Fiamphotrigon g2:32:22a:32::assszzzsassszszzsssss IE Myiophobus ”snag agar:::::::::::::::=:::assaaazsB Myiophobus I o :: E: aa;aaaaaaasaaaaaaaaaaaaasaaWrochIus p,r gagggg Eg WmfiMyiotn'ccus gggggg 35.222222222225: IMachetomis §§ EE 3: I Muscigrella a.- E55355 2- Neopipo :23: -_ 3%: I Pynhomyias :: 3% I Himndinoa :4 22: Colonia if: m Sublegatus 2;}? Zimman'us f: $5.5 a Phylloscartes :- E] Mionectes 333' E] Leptopagon

Figure 3

Figure Imxlnbo a (uni) -

3 mourned -

(cont’d) MMVH wruav - axlurs ouvIvun fl l l I. I .- ‘1 I _I “II lI ‘I III -'I ‘Iu'Ip'IiI‘I .I‘In‘w l I - I I

NVEI'iQHOHad- 1|.IEIII1I, II; 'I‘ Ex. ‘I. ll. ‘n punorfi Ieeu m IEI;I'.I, hymn; milieu uoiad pasopua ;ggggg uaafli-renou pIemno m a mu llml m sass

uaeIB-Ia/lou piemdn 11111] [:1 f” W ‘be.‘h‘s‘i‘bbflbbi‘t\‘flfi‘x\K‘k‘tk‘k‘t‘t\K‘gfl‘h\‘bfi\\h&\‘\:\fi:§

IoIAeuaq BugfieIOI ,2 ms

gray/Hm; xxemxxxxxxp \EKHRXH‘fiVRRRXRXK‘EEL‘flK\\\\‘L\"hfin\\‘h\\\:‘§§ s‘msms

W .5" 143’ W W‘s .rr' \ss 32 sweets *sIs smmm ssxsxmxxxxxmmxmslxm

\. Emma‘s

' sssssfis‘s‘s

3:1 II" 1:: $252922?“ wanna- we-.. ,u. ,3}. Innuuunnnnl'uunn maul-I mun

i...... " I: I,

““335535 n: nnnnunuunu

aunanuuu

— jIEE

._

I r 11...... W5 I... 3533;; ' .

33..., nqss‘ «- a; ____;I "m " MAE] """ ;5;; _E 7“ 'v

f1

__E I :5

_- -.G :: :- .; .; g. -: [j [j E [j I E EEusceIthmus E] I I I a I a I I E I I I I I I Arundinicola I I I I I Ochthooce ,- . Mococamulus MonpegIs Capsiampis Pheeomyias Pseudatnccus Alectrurus Polystictus Pseudacoloptaiyx Phy/Iamyias Aphanotn'ccus Inez/a Camptostome Carythapis Knipo/egus SI'IVI'cu/trix Uromyies Phy/Iomyias Elaenia Xenotriccus Sayom/s Fluwcala C Lathrotriccus Mitrephanes Ochthomis Tyrannuius Muscipipra Cantopus Empidonax Pyrocephalus Xolmis Lessonia Agnomis Musasaxico/a Setrapa Hymenops Neoxolmis My/otherstes Heteroxalmis Polioxolmis Cnemarchus O nemotriccus ubemetes l O rh

amp h

us griseiceps fescietus c.m 33 included. This reconstruction illustrates many features common to reconstructions on all M-P trees. The ‘upward strike specialist’ state characteristic of fiatbills and tody-tyrants is always reconstructed on several of the deepest branches in the tree, suggesting that it is the ancestral tyrannid foraging behavior. This is because the flatbills and tody-tyrants are placed in a basal position in all M-P trees (see Chapter 2). This result is obtained even when foraging behavior is excluded from the estimation of the underlying phylogeny. Table 1 shows the average number of unambiguous changes in each foraging behavior state as reconstructed on trees estimated with foraging behavior included. Note that cells on the diagonal contain the frequency of stasis for each state. When, for each state, the values in the non-diagonal cells are summed to give a single value for change, the resulting 10X2 table (Table 2) shows no significant differences among the states in their relative rates of stasis and change (P = 0.97). However, several of these cells have few or zero observations. A table which pools the data for the five generalist modes and pools the data for five specialist modes (Table 3) also shows no differences between generalists and specialists in relative rates of stasis and change (P = 0.46). The same conclusion is reached when foraging behavior is excluded from estimation of the underlying trees (Table 4)(P = 0.25).

D I S C U S S I O N Specialization is not a cul-de-sac Foraging behavior specialization does not appear to have constrained diversification in the tyrant flycatchers. Specialist lineages have given rise repeatedly to generalists and to other types of specialists and show no greater tendency to be derived than do generalist lineages. These conclusions are reached whether foraging behavior is included or excluded from estimation of the phylogenetic hypothesis. Miiller (1996) reached a similar conclusion in a phylogenetic analysis of the evolution of host-plant specialization and generalization in anthidiine bees. He found that there had been at least four transitions from specialist to generalist, at least eight of specialists between different plant taxa, but none from generalist to specialist. The present study is, as far as I am aware, the first to examine the evolution of specialization and generalization on a phylogenetic framework large enough to make statistical inferences about relative rates of different classes of transitions. Table l - The average number of unambiguous changes from each foraging behavior state into any of the others as reconstructed on trees estimated with foraging behavior included. Note that cells on the diagonal from upper left to lower right contain the frequency of stasis for each state.

To: From:

OOOOOOO C." ©6100 fruit/upward hover-glean 7.57 0.11 fiooooo

fruit/hawk 0000

NMV’WOI‘QO‘

generalists outward hover-glean 0.55 0.55 0.55 34

enclosed perch hawk 0.11 12.52 0.11 0.33

near ground 0.17 18.24 1.64

PERCH-GLEAN 3.33 O 0.87 0.38 37.09

UPWARD STRIKE 0.11 1.54 1.48 0.82 0.81 35.56 1

SPECIALISTS AERIAL HAWK 0.11 0.11 0.52 0.55 0 14.94

PERCH-TO-GROUND 8.12

GROUND 0.11 0 0.22 0.33

35

Table 2 - Relative rates of stasis and change for each foraging behavior state.

stasis change

fruit/upward hover-glean 7.57 0.88

fruit/hawk 2 0

generalists outward hover-glean 9.62 1.65

enclosed perch hawk 12.52 1.48

near ground 18.24 3.27

PERCH-GLEAN 37.09 4.58

UPWARD STRIKE 35.56 5.76

SPECIALISTS AERIAL HAWK 14.94 3 .4

PERCH-TO-GROUND 8. 12 0. 12

GROUND 6.22 0.66

36

Table 3 - Relative rates of stasis and change for pooled generalist states and for pooled specialist states.

stasis change

generalists 50 7

specialists 102 15

Table 4 - Relative rates of stasis and change for pooled generalist states and for pooled specialist states as reconstructed on trees estimated with foraging behavior excluded.

stasis change

generalists 36 3 specialists 76 10

37

Arguments for specialization as an evolutionary cul-de-sac are couched in terms of specialization on one or few resources and they assume the presence of performance trade- offs associated with trying novel resources. In this study I examined patterns of behavioral evolution and so this work constitutes a test of the idea that specialization is a cul—de-sac only if: 1) behavioral specialization is correlated with resource specialization, and 2) there are performance trade-offs associated with feeding on novel resources. Below I discuss the evidence for both in flycatchers: Are behavioral specialists also resource specialists ?—If behavioral repertoire breadth is poorly correlated with diet breadth, behavioral specialists might actually be resource generalists and be no more sensitive to environmental change than are behavioral generalists. Resources have not been quantified for most tyrannid species, but there is evidence of some correlation between behavioral repertoire breadth and diet breadth. Extreme diet specialists include the upward strike specialist Myiobius (‘Terenotriccus’) erythrurus which feeds almost exclusively on planthoppers (Homoptera) which jump great distances to evade attack (Sherry 1984). The bird’s short, explosive flight to scoop an insect off the underside of a leaf appears be a “surprise” tactic allowing them to specialize on planthoppers. Other flycatchers specializing on the upward strike tactic eat a greater variety of insects, but they tend to be cryptic, immobile insects found on the underside of large understory leaves and so are relatively inaccessible to other birds which glean insects they can reach on foot (Sherry 1984). Aerial hawk specialists Colonia and some Contopus often specialize on flying, social hymenoptera which tend to be aggregated near hives and nests (Bent 1942, Beaver and Baldwin 1975, Sherry 1984). These birds sally and return repeatedly to a favored perch located near an insect aggregation. Other aerial hawk specialists such as Tyrannus and some Contopus have broader diets. The behavioral generalists Myiarchus, Sayomis, and Empidonax all have quite generalized diets (Bent 1942). We need much more data, but it appears likely that there is a positive correlation between behavioral repertoire breadth and resource breadth in flycatchers. Are behavioral specialists also morphological and physiological specialists ?-If behavioral specialists do not experience decreased performance when attempting novel foraging techniques, they may be no more sensitive to environmental change than are behavioral generalists. Fitzpatrick (1985) suggested that such performance trade-offs do 38 exist in flycatchers and described several correlations between specialized foraging behaviors and putative morphological adaptations which could cause trade-offs. However, these correlations used species as data points and thus their interpretation is problematical. Since some species are more closely related to each other than they are to other species, they do not represent independent data points (Felsenstein 1985). Evidence of adaptation must be sought in the number of historically independent times particular behaviors and morphologies have evolved together and thus, Fitzpatrick’s (1985) adaptation hypotheses await historically-based phylogenetic tests. The results of such tests will suggest appropriate functional studies aimed at detecting performance trade-offs. There are no such functional studies in sallying insectivorous passerine birds, but their behavior and ecology is highly convergent with that in insectivorous bats. The relationship between wing shape and foraging behavior in bats has been extensively studied using both theoretical and functional approaches (Jones et al. 1993, Norberg 1994). This work suggests that wing shape is tightly linked to the performance of specific foraging behaviors and that trade-offs between performance of different behaviors are certain to exist. It is likely that behaviorally specialized flycatchers are also morphological and physiological specialists. If behaviorally specialized flycatchers are also resource specialists and morphological/physiological specialists, then why do they appear to be no more evolutionarily constrained than generalists? An examination of some specific patterns of behavioral evolution in the tyrant flycatchers may suggest an explanation.

Evolution of tyrannid foraging behaviors The reconstruction of behavioral evolution (Figure 3) constitutes an historically- based test of Fitzpatrick’s (1985) hypotheses of foraging behavior evolution in the tyrant flycatchers. My results provide some support for the hypothesis that the specialized terrestrial foraging habits of the fluvicoline ground-tyrants (Muscisaxicola through Cnemarchus in Figure 3) are derived ultimately from near ground generalist foragers such as Pyrocephalus and Ochthoeca. However, this result is not obtained on all M-P trees. The perch-to-ground sallying bush-tyrants, Shrike-tyrants, and monjitas are, on all M-P trees, derived from ground foragers such as Muscisaxicola and Lessonia which search for prey while walking and standing on the ground. 39 Others of Fitzpatrick’s (1985) hypotheses, which were based on the assumption that specialized foragers are derived from generalized foragers, are not supported. In particular, my results do not support the hypothesis that upward strike specialization is highly derived. Fitzpatrick (1980, 1985) suggested that lineages specializing on this mode represent the best example of loss of evolutionary flexibility or potential. This is because upward striking is the most stereotyped tyrannid foraging technique and these species posses several morphological specializations such as extremely wide bills, short and rounded wings, long legs, and short tails. He hypothesized that these extreme morphologies produce performance trade-offs which make it difficult to depart evolutionarily from their specialized behavior. On all M-P trees in the present study, upward strike specialization is the ancestral tyrannid foraging behavior and is reconstructed on several of the deep branches of the tree. While this specialization does appear to be evolutionarily conserved, surviving in several present day lineages, it also appears to have had the flexibility to give rise to many other sorts of generalist and specialist foragers. Another specialized foraging behavior which appears to have a long, conserved history is perch-gleaning. All M-P trees contain a large clade of about 23 genera of tyrannulets and allies which forage in the very active manner of warblers and vireos, capturing most of their prey while perched, without sallying. This behavior appears to homologous among all species possessing it, having arisen once in their common ancestor. Yet, as evolutionarily conserved as it appears to be, it has evidently given rise numerous times to fruit/upward hover-glean generalists. Hover-gleaning and frugivory are minor components of the repertoire of perch-glean specialists (Fitzpatrick 1980), and it is apparently easy for them to evolve greater reliance on those techniques. I hypothesize that this is because, unlike hover-gleaning, perch-gleaning does not involve sallying to prey and therefore wing morphology would not affect its performance. There would then be no performance trade-off involved in the use the hover-glean and perch-glean techniques (unless bill morphology introduced one) and a bird could use both techniques efficiently.

Conclusion The above hypothesis may provide a general explanation for the results of this study and others which find that transitions out of specialized states occur as easily as 40

transitions out of generalist states (e.g. Thompson 1994, Futuyma et al. 1995, Mfiller 1996, Pellmyr et al. 1996). The existence of performance trade-offs is central to arguments that specialization is an evolutionary cul-de-sac, and they undoubtedly exist for particular morphological structures. A forager with wings highly specialized for use of just one resource may be extremely inefficient on other resources acquired on the wing, but may often be able to escape this ‘sallying cul-de-sac’ by switching to resources which do not require sallying. This specialist may have leg, foot, or bill morphologies with less extreme performance trade-offs. In terms of an adaptive landscape, what was assumed to be an adaptive peak really turns out to be an adaptive ridge which provides a route away from the cul-de-sac. If we consider the organism as a whole, it is probably not as constrained in its resource use as it would seem from consideration of single morphological structures which may be highly specialized. Functional studies of a number of morphological traits and a number of activities will be necessary to accurately characterize particular species as especially sensitive to environmental change or as ‘evolutionarily constrained’. This explanation depends upon the existence of an adequate degree of plasticity in foraging behavior. Fitzpatrick (1980) documented some plasticity in showing that many specialists minor in alternative foraging techniques. One can also observe them to resort to alternative techniques under occasional, adverse environmental conditions. For example, Contopus aerial hawk specialists often resort to much hover-gleaning on unusually cold mornings when insects are not flying (pers. obs.). These birds can use a variety of sallying techniques but most of the time choose not to, probably because their morphology is such that they are energetically inefficient at techniques other than aerial hawking. Despite such behavioral plasticity, these results indicate that on some level there can be a strong, conserved phylogenetic component to foraging behavior and ecology in tyrant flycatchers. The perch-glean specialists and upward strike specialists, in particular, have inherited their foraging behaviors from distant ancestors. Predominant paradigms which explain foraging behavior in terms of the distribution of resources and individual decision- making (Krebs and Davies 1991) need to account for the possibility that related species share homologous behaviors, sometimes in very different ecological settings. Explanations of these behaviors are not complete without an historical perspective which incorporates comparative studies of related species and phylogenetic analysis of the origin of the 41 behavior and the selective environment in which it is likely to have evolved. APPENDIX APPENDIX

Descriptions of characters used in the phylogenetic analysis

Cranial 1. Nasal septum poorly ossified.--In the flatbill and tody-tyrant assemblage (Lanyon 1988c) and apparently in Tachuris (Lanyon 1988a), the nasal septum usually does not ossify, leaving only a shallow dorsal remnant of a septum in cleaned museum specimens. In other tyrannids and in cotingids and piprids, the septum usually ossifies to varying degrees. Prum and Lanyon (1989) hypothesized that the ossified nasal septum is primitive within tyrannoids. I hypothesize the poorly ossified condition to be derived in Tachuris, the flatbills, and the tody-tyrants. 2. Anterior segment of nasal septum poorly ossified.--In Nesotriccus, Capsiempis, and Phaeomyias, the region of the nasal septum anterior to the internal support rod is poorly ossified (Lanyon 1988a). I hypothesize this condition to be derived in these three taxa. 3. Internal support rod reduced.--In the kingbird assemblage (Lanyon 1984) and in the genera Attila, Colonia, and Machetomis (pers. obs.), there is no conspicuous internal support rod within the ossified nasal septum. There is only a pattern or grain running anterodorsally to posteroventrally on the surface of the septum in the position where the rod is located in the septa of other tyrannoids. I hypothesize the reduced state to be derived in the above tyrannids. 4. Nasal septum with transverse trabecular plate.--In many tyrannids, the ossified nasal septum has a transverse trabecular plate anterior to the internal support rod. In the Empidonax assemblage (Lanyon 1986, 1988b) this plate is at the base (ventral edge) of the septum and there is a distinctive notch in the septum anterior to the trabecular plate. In the Elaenia assemblage (Lanyon 1988a) this plate is elevated above the ventral edge of the septum, so that there appears to be a mid-sagittal ridge on the ventral side of the trabecular plate. There is no anterior notch associated with this type of trabecular plate. The nasal septum of Colonia has a trabecular plate which is unique among tyrannids (Lanyon 1986). It is barely elevated above the ventral edge of the septum and occupies only a very short 42 43 segment of the anterior part of the septum. It is very broad, truncated posteriorly, and has no anterior notch. Trabecular plates, both basal and elevated, are present in some cotingids and Old World suboscines (W arter 1965, pers. obs.). I hypothesize the trabecular plates of tyrannids to be derived; the basal plate with anterior notch (4.1), elevated plate (4.2), and short, broad plate of Colonia (4.3) are alternative unordered derived states of a single character. 5. Trabecular plate reduced, anterior.--In Pseudotriccus and Corythopis, the elevated trabecular plate is reduced to a small anterior section of the nasal septum (Lanyon 1988a). I hypothesize this condition to be derived in these two taxa. 6-8. Additional, posterior trabecular plate.--In several genera in the Elaenia assemblage (Lanyon 1988a) there is an additional, small trabecular plate posterior to the internal support rod. In Elaenia, Tyranniscus, and Mecocerculus (calopterus, stictopterus, and poecilocercus) this posterior plate is flat like anterior plates. In Mionectes and Leptopogon, the posterior plate is bulbous. In Stigmatura and Pseudelaenia, both anterior and posterior plates are bulbous. Posterior plates are found elsewhere only in occasional specimens of Omithion and Camptostoma (Lanyon 1988a) and bulbous anterior plates are found elsewhere only in some cotingids (e.g. Rupicola)(Warter 1965, pers. obs.). I hypothesize posterior trabecular plates (6), bulbous posterior plates (7), and bulbous anterior plates (8) to be derived in an ordered transition series, and code them as a set of additive binary characters. 9. Trabecular plate forked posteriorly.-In all genera in the Empidonax assemblage

(Lanyon 1986, 1988b) except Cnemarchus, Pyrrhomyias, Hirundinea, and Myiobius, the trabecular plate is forked posteriorly. In seven genera of the Elaenia assemblage (Pseudocolopteryx, Polystictus, Capsiempis, Phaeomyias, Nesotriccus, Semaphaga, and Anairetes), the trabecular plate is also forked posteriorly, but the fork is narrower than those in the Empidonax assemblage, appearing as a slender tuning fork (Lanyon 1988a). These morphologies are unique among tyrannoids and I hypothesize them to be derived; relatively wide forks (9.1) and slender forks (9.2) are alternative unordered derived states of a single character. 10. Nasal capsule ossified.--In the Myiarchus assemblage (except for Attila), and in

Arundinicola, F luvicola, Alectrurus, Lathrotriccus, Heteroxolmis, Gubernetes, Colonia, 44

Zimmerius, Mionectes, Sublegatus, Myiotriccus, Myiopagis, Tyrannulus, Suiriri, and some Phylloscartes, the nasal capsule is fully ossified including alinasal walls and turbinals. In other tyrannids and in most suboscines, the alinasal walls and turbinals do not remain in cleaned museum specimens. Ossified nasal capsules occur in a few cotingids and fumarioids (Warter 1965, pers. obs.), and in some Old World suboscines where the unossified condition appears to be primitive (pers. obs.). I follow Lanyon (1985, 1986, 1988a) and hypothesize ossified nasal capsules to be derived in the above tyrannids. 11. Medial ridge in frontal region of cranium.--In the kingbird assemblage and in Machetomis (Lanyon 1984), there is a medial ridge extending from the nasal-frontal hinge posteriorly to near the top of the cranium. This ridge is found elsewhere only in some Old World suboscines (pers. obs.) and I hypothesize it to be derived in the kingbird assemblage and Machetomis.

Syringeal 12-17. Double, complete A elements.--In about two-thirds of the tyrannid genera, some of the caudal-most of the (primarily tracheal) ‘A’ series of syringeal support rings are double (encircling each bronchus) and complete (completely encircling each bronchus)(Ames 1971). This morphology also occurs in some cotingids and piprids (Prum 1992). Based on outgroup comparison to fumarioids, Old World suboscines, and oscines, Prum (1992) hypothesized that double, complete A elements are derived in tyrannoids and, further, are derived in piprids independently of those in cotingids. I hypothesize that double, complete A elements are derived in tyrannids independently of those in other tyrannoids. The caudal-most A element, A1, is never complete, but any of elements A2 through A5 may be. The number of these that are complete varies among taxa as does their degree of ossification. Prum (1992) hypothesized a transition series from the primitive state in which these A elements are medially cartilaginous to the derived state in which they are fully ossified. Lanyon (1984) and Griffiths ( 1994) hypothesized transition series from few to greater numbers of complete bronchial A elements. I incorporate both of these hypotheses into the following ordered transition series coded as a set of six additive binary characters: (12) At least one A element forms a complete ring around each bronchus; derived in the 45

flatbills (Lanyon 1988c), the kingbird assemblage (Lanyon 1984), the Phylloscartes group of the Elaenia assemblage (Lanyon 1988a), the Empidonax assemblage except for the Ochthoeca group (Lanyon 1986, 1988b), the Myiarchus assemblage (Lanyon 1985), and the genera Machetomis (Lanyon 1984), Neopipo (Mobley and Prum 1995), Colonia, and Muscigralla (pers. obs.). (13) One A element forms a complete, fully ossified ring around each bronchus; derived in the flatbills Rhynchocyclus, Tolmomyias, and Platyrinchus, the kingbird assemblage (except for Pitangus and Philohydar), Zimmerius, Mionectes, Leptopogon, Myiophobus (ochraceiventris, phoenicomitra) and Myiotriccus in the Phylloscartes group of the Elaenia assemblage (Lanyon 1988a, pers. obs.), Myiophobus, Pyrrhomyias, and Hirundinea of the Empidonax assemblage, Deltarhynchus, Ramphotrigon, and some Attila, Rhytipterna, and Casiomis in the Myiarchus assemblage, and in the genera Machetomis, Neopipo, and Muscigralla. (14) At least two A elements form a complete ring around each bronchus; derived in Onychorhynchus, Pitangus, Philohydar, the Empidonax and Muscisaxicola groups (except for Gubemetes and Muscipipra), and Hymenops, Knipolegus, Ochthomis, Satrapa, and Myiobius (including erythrurus) of the Empidonax assemblage. (15) At least two A elements form complete, fully ossified rings around each bronchus; derived in Onychorhynchus, Pitangus, Philohydar, and Myiobius (including erythrurus). (16) A3 elements contribute to support of each bronchus through fusion to the complete A2 elements; derived in Legatus and Myiozetetes (Lanyon 1984) and Colonia (pers. obs.). (17) A2 elements contribute to support of each bronchus through fusion to the complete A3 elements; derived in Pyrrhomyias, Hirundinea, and Neopipo (Lanyon 1986, Mobley and Prum 1995). 18. Dorsal segments of AI elements are cartilaginous.--In Zimmerius, Phylloscartes, Elaenia, Myiopagis, Tyrannulus, Suiriri, Uromyias, Capsiempis, Nesotriccus (Lanyon 1988a), and in Phyllomyias griseiceps (pers. obs.), large dorsal segments of the A1 elements are cartilaginous. In most other tyrannoids, the A1 elements are more fully ossified, sometimes with small dorsal tips cartilaginous. Large cartilaginous dorsal segments are found in some piprids (Prum 1992) where there is also a tendency towards fully cartilaginous Als. I hypothesize the large cartilaginous dorsal segments of these 46 tyrannids to be derived independently of those in piprids. 19-20. Medial segments of A2 's modified for attachment of internal cartilages.-4n Cnemotriccus, Aphanotriccus, Lathrotriccus, Xenotriccus, Sayomis, Contopus, Mitrephanes, and Empidonax, the cartilaginous medial segments of the A2 elements are enlarged and bulbous caudally, presumably for the attachment of the internal cartilages (Lanyon 1986). In Sayomis, Contopus, Mitrephanes, and Empidonax, this medial cartilaginous segment is usually disconnected from the remaining (ossified) portion of the A2 element. These morphologies are unique among tyrannoids. I hypothesize modified medial segments of A2s (19) and disconnected modified segments (20) to be derived in an ordered transition series, and code them as a set of additive binary characters. 21. Single (tracheal) A elements fused to form a 'drum'.--In the Elaenia group of the Elaenia assemblage and Ochthoeca and Colorhamphus of the Empidonax assemblage, two or more single A elements are fused together and to the pessulus, forming a drum (Lanyon

1986, 1988a). In several genera, this drum does not include A2, but A3 and higher elements are fused. In others, the drum includes A2. In Serpophaga and Anairetes, the drum includes A2 but is, on average, not as robustly fused as in other genera. Fused tracheal A elements are found in some piprids and cotingids where Prum (1992) has hypothesized them to be independently derived. I hypothesize tracheal drums to be derived in tyrannids and have coded the variation in drum morphology as alternative unordered derived states of a single character. The drum is robust and starts at A3 (21.1) in Elaenia, Myiopagis, Tyrannulus, Tyranniscus, and Mecocerculus (calopterus and minor). The drum is robust and starts at A2 (21.2) in Ornithion, Camptostoma, Mecocerculus (stictopterus, poecilocercus, and leucophrys), Suiriri, Pseudocolopteryx, Polystictus, Uromyias, Capsiempis, Phaeomyias, Nesotriccus, Ochthoeca, and Colorhamphus. The drum is poorly fused and starts at A2 (21.3) in Serpophaga and Anairetes. 22. Nodule on lateral surface of AI and A2.--In Aphanotriccus and Lathrotriccus, there is an ossified nodule on the lateral surface of each Al and A2 element, with a cartilaginous connection between them. This morphology is unique among tyrannoids and I hypothesize it to be derived in these two taxa. 23. Pessulus extends anteriorly to divide tracheal A elements dorsally.-—In Hemitriccus, Lophotriccus, and Oncostoma, the ossified pessulus extends anteriorly on the dorsal side 47 of the trachea to divide four or more A elements. This morphology is unique among tyrannoids and I hypothesize it to be derived in these three taxa. 24. Pessulus is an ossified extension of the first single complete A element.--In the Phylloscartes group of the Elaenia assemblage (except for Myiophobus ochraceiventris), and in Pseudelaenia (Lanyon 1988a), Pyrrhomyias, Neopipo (Mobley and Prum 1995),

Cnipodectes, Todirostrum, some Tolmomyias (Lanyon 1988c), Machetomis, Muscigralla, and Colonia (pers. obs.), the pessulus is an ossified caudal extension of the dorsal side of the first single, dorsally complete A element. This morphology is present in some cotingids and piprids but is absent in fumarioids. It has been hypothesized to be independently derived in at least some tyrannids (Mobley and Prum 1995). I hypothesize it to be derived in all of the above tyrannids. 25. Pessulus large, concave.--In Pseudotriccus and Corythopis, the ossified pessulus is relatively large and concave dorsally. Its greatest breadth is at the point where the internal cartilages attach (Lanyon 1988a). This morphology is unique among tyrannoids with the possible exception of a few cotingids (Prum and Lanyon 1989). I hypothesize it to be derived in Pseudotriccus and Corythopis. ' 26. BI -2 connected ventrally.--In all tyrannids except Myiotriccus, F luvicola, Pyrocephalus, Hirundinea, Myiobius, and Myiophobusflavicans, the bronchial, cartilaginous B1 and B2 elements are connected at their ventral tips. In Myiodynastes, Conopias, Megarynchus, Tyrannopsis, Tyrannus, Empidonomus, and Griseotyrannus, this connection is symmetrically rounded (Lanyon 1984). In all other tyrannids in which it is present, this connection is more flattened or acute. Ventral connections between B1 and B2 elements are absent in cotingids (Prum, pers. com.) and in piprids where B1 and B2 are connected dorsally (Prum 1992). I hypothesize ventrally connected B1 and B2 elements to be derived in tyrannids and I code acute connections (26. 1) and rounded connections (26.2) as alternative unordered derived states of a single character. 27. B] -2 connections are close medially.--In Megarynchus, Tyrannopsis, Tyrannus, Empidonomus, and Griseotyrannus, the bronchi are close together in the region of B1 and B2, so that the left and right ventral connections of those elements nearly touch one another medially (Lanyon 1984). This morphology is unique among tyrannoids and I hypothesize it to be derived in these five taxa. 48

28. 81-2 diverge dorsally.--In Myiozetetes, Pseudotriccus, Sirystes, and Myiarchus, the B1 and B2 elements diverge dorsally so that their dorsal tips lie relatively far apart. This morphology is unique among tyrannoids and I hypothesize it to be derived in these four taxa. 29. 81-2 shaped as in Myiarchus.--In the Myiarchus assemblage, the Elaenia assemblage (except for Mionectes, Suiriri, Capsiempis, and Nesotriccus), and in Phelpsia, Philohydar, Legatus, Cnemotriccus, Aphanotriccus, Lathrotriccus, Xenotriccus, Contopus, Empidonax, Lessonia, Satrapa, Muscisaxicola, Agriomis, Xolmis, Heteroxolmis, Myiotheretes, Myiophobus, Neopipo, Colorhamphus, and some Silvicultrix and Ochthoeca, the B1 and B2 elements have a unique configuration. B1 is narrow except for a very broad dorsal end. B2 is narrow except for a broad, triangular ventral end which is barely attached to the ventral end of B1. This morphology is unique among tyrannoids and I hypothesize it to be derived in these taxa. 30. BI elements broad laterally.--In Onychorhynchus and Platyrinchus, the BI elements are very broad in their central (lateral) regions and less broad toward the ventral and dorsal tips. This morphology is unique among tyrannoids and I hypothesize it to be derived in these two taxa. 31. BI elements with bulbous dorsal ends.--In Gubemetes and Muscipipra, the dorsal ends of the B1 elements are bulbous or swollen. This morphology is unique among tyrannoids and I hypothesize it to be derived in these two taxa. 32. 82 elements with Y-shaped ventral ends.--In Platyrinchus, Sublegatus, Myiophobus (roraimae and phoenicomitra), Empidonax, Gubemetes, and Muscipipra, the ventral end of the B2 elements are forked or shaped like the letter 'Y'. This morphology is unique among tyrannoids and I hypothesize it to be derived in these taxa. 33. Syrinx and trachea laterally compressed.-—In Lophotriccus and Oncostoma, the entire trachea and tracheobronchial junction are laterally compressed so that they are unusually narrow when viewed dorsally or ventrally, and broad when viewed laterally. Similar morphology occurs elsewhere only in Machetomis where the tracheobronchial junction, but not the trachea, is laterally compressed (per. obs.). I hypothesize the compressed trachea and tracheobronchial junction to be derived in Lophotriccus and

Oncostoma. 49 34. Cartilaginous bronchial plate.--In the tody-tyrant group (Lanyon 1988c), in Stigmatura and Pseudelaenia (Lanyon 1988a), and in Tachuris (Ames 1971, pers. obs.) and Phyllomyias (fasciatus and griseiceps, at least; pers. obs.), there is a horseshoe-shaped cartilaginous plate which connects the dorsal tips of the incomplete bronchial A elements and the pessulus. It forms the dorsal or craniodorsal margins of the medial tympaniforrn membranes. The internal cartilages are located near the caudal ends of the horseshoe, sometimes attached to it. In the tody-tyrant group, the internal cartilages are very small, delicate rods, while in the other taxa possessing the cartilaginous plate, the internal cartilages are as robust as those in most tyrannids. Similar 'medial bronchial cartilage bars' occur without internal cartilages in some piprids where they have been hypothesized to be independently derived (Prum 1992). I hypothesize the bronchial plates of tyrannids to be derived. I code plates with small, delicate, rod-like internal cartilages (34.1) and plates with robust internal cartilages (34.2) as alternative unordered derived states of a single character. 35. Cartilaginous tracheal plate.--In Stigmatura (Lanyon 1988a) and Phyllomyias fasciatus (pers. obs.), the cartilaginous pessulus extends cranially so that the lower tracheal A elements up to A6 or A8 are cartilaginous dorsally. This morphology is found elsewhere only in some piprids where it has been hypothesized to be independently derived (Prum 1992). I hypothesize cartilaginous tracheal plates to be derived in Stigmatura and Phyllomyiasfasciatus.

INTERNAL CARTILAGES: The internal tympaniforrn membranes in the syringes of all tyrannids and some cotingids contain cartilaginous structures which are often attached to A elements. Because there is extreme variation in the shape, position, and attachment of internal cartilages, Prum and Lanyon (1989) hypothesized them to be derived in cotingids independently of those in tyrannids. Prum (1990, pers. comm.) even questions their homology within all tyrannids, and recommends using unique, detailed morphologies of internal cartilages as characters, while refraining from the hypothesis of homology of internal cartilages as broadly defined. Here I follow this recommendation and the precedent of Lanyon (1984, 1985, 1986, 1988a, b, c) in coding shared, detailed morphologies of tyrannid internal cartilages as independent, derived characters. 50

36. Internal cartilages narrow, linear.--In Rhynchocyclus and Tolmomyias (Lanyon 1988c), and in Phylloscartes, Leptopogon, and Pseudotriccus (pers. obs.), the internal cartilages are narrow and linear. 37. Internal cartilages triangular.--In Sublegatus, Myiophobus (roraimae and phoenicomitra), and Myiotriccus (Lanyon 1988a), in Alectrurus, Ochthoeca, Colorhamphus, Hymenops, Knipolegus, and Ochthomis (Lanyon 1986), in Neopipo (Mobley and Prum 1995), and in Machetomis, Muscigralla, and Colonia (pers. obs.), the internal cartilages are triangular, with the base (the cranial end) much broader than the rest of the cartilage. 38. Internal cartilages spatulate.--In Onychorhynchus and Platyrinchus (Lanyon 1988c), the internal cartilages are broad and have a thickened dorsal edge. 39. Internal cartilages J or L shaped.-In Philohydar, Myiodynastes, Conopias, Tyrannus, Empidonomus, and Griseotyrannus, the internal cartilages are thin and bent in a ‘J’ or ‘L’ shape. In Legatus, Megarynchus, Tyrannopsis, Neoxolmis, Gubemetes, Rhytiptema, Sirystes, Casiomis, and Myiarchus, the internal cartilages are also ‘1’ or ‘L’ shaped but are significantly more robust. I code thin ‘J’ or ‘L’ shaped cartilages (39.1) and robust ‘J ’ or ‘L’ shaped cartilages (39.2) as alternative unordered derived states of a single character. 40. Internal cartilagesforked.--In Cnipodectes, Pseudotriccus, Corythopis, and some Leptopogon, the internal cartilages are forked caudally. 41. Internal cartilages with ventral extension.-In Myiopagis, Tyrannulus, Omithion, and Camptostoma (Lanyon 1988a), and in Phyllomyiasfasciatus (pets. obs.), the internal cartilages have poorly staining, amorphous ventral extensions from the caudal half of the cartilage. In Mecocerculus stictopterus and poecilocercus, there is a well-formed, squarish ventral extension (Lanyon 1988a). I code amorphous (41.1) and well-formed (41.2) ventral extensions of the internal cartilages as alternative unordered derived states of a single character. 42. Internal cartilages with broad, amorphous caudal segments.-In Tyranniscus and Mecocerculus calopterus and minor, the internal cartilages are narrow cranially but broaden and become amorphous caudally. 43. Internal cartilages narrow, curved.--In Mecocerculus leucophrys, Uromyias, 51

Capsiempis, Phaeomyias, Nesotriccus, Serpophaga, and Anairetes, the internal cartilages are relatively narrow and curved (medially concave). 44. Internal cartilages straight, expanded caudally.--In Deltarhynchus and Ramphotrigon, the internal cartilages are relatively long, straight, and expanded caudally. They are somewhat twisted, with the cranial half flattened at right angles to the internal tympaniform membrane and the caudal half flattened within the plane of the membrane. 45. Internal cartilages attached to tracheal drum and incomplete As.--In Myiopagis, Tyrannulus, Capsiempis, Phaeomyias, and Nesotriccus, the internal cartilages are attached to the dorsal end of an incomplete A element (either A1 or A2) as well as to the tracheal drum. In all other taxa which have a drum, the internal cartilages are attached only to the drum. 46. Internal cartilages attached to ventral side of tracheobronchialjunction.—-In Rhytiptema, Sirystes, Casiornis, and Myiarchus, the internal cartilages are attached to ventral or medic-ventral segments of A2 or A3. In all other tyrannids, the internal cartilages attach to the dorsal side of various structures in the tracheobronchial junction. 47. Narrow strand of cartilage between ventral ends of 32 and B3.--In Pyrrhomyias and Hirundinea, there is, in addition to the large internal cartilages, a narrow strand of cartilage in the internal tympaniform membrane uniquely placed between the ventral ends of B2 and B3. This morphology is unique among tyrannoids and I hypothesize it to be derived in these two taxa. 48. Cartilaginous plug of tissue between A2s.--In most of the Empidonax assemblage (in the Empidonax group, the Muscisaxicola group, the Knipolegus group, Ochthomis, and Satrapa), there is a plug of tissue just caudal to the tracheobronchial junction and between the cartilaginous medial segments of the A2s. This plug stains weakly for cartilage. This morphology is unique among tyrannoids and I hypothesize it to be derived in these taxa. 49. Mm. tracheolaterales cover ventral surface of trachea-4n the Elaenia assemblage (except for Suiriri, Phaeomyias, Nesotriccus, Anairetes, Tyrannulus, and some Elaenia and Myiopagis), the Empidonax assemblage (except for Aphanotriccus and the Myiophobus group), and in Megarynchus, Tyrannopsis, Tyrannus, Griseotyrannus, Machetomis, Muscigralla, Colonia, Tachuris, Platyrinchus, Casiornis, Phyllomyiasfasciatus and 52 griseiceps, and some Myiarchus and Lophotriccus, the Mm. tracheolaterales, which are a pair of extrinsic syringeal muscles, cover the ventral surface of the trachea (Ames 1971, pers. obs.). Typically, the right and left Mm. tracheolaterales widen ventrally and meet above A20, sometimes lower, and completely cover the ventral and ventrolateral surface of the trachea from there posteriad, diverging just before their insertion in the region of the syrinx. In three of the above taxa (Mionectes, Leptopogon, and Pyrocephalus), the left and right Mm. tracheolaterales do not diverge at their insertion, but converge to a point and insert midventrally on the lower A elements. In all other tyrannids, the Mm. tracheolaterales are thin muscles covering only the lateral portions of the trachea. Among other tyrannoids, all three states occur (Prum and Lanyon 1989, Prum 1990) but since the latter state in which the Mm. tracheolaterales cover only the lateral portions of the trachea is the state found in the next two outgroups, fumarioids and Old World suboscines (Prum and Lanyon 1989, Prum 1993), I hypothesize this to be the primitive state in the tyrannoids and in the tyrannids. I hypothesize Mm. tracheolaterales covering the ventral surface of the trachea to be derived in the above tyrannids and I code those which diverge to insert (49.1) and those which converge to insert (49.2) as alternative unordered derived states of a single character. 50. Mm. obliqui ventrales.--All tyrannids except Machetomis, Onychorhynchus, Pyrrhomyias, Hirundinea, Neopipo, Myiobius, and at least some Todirostrum (cinereum),

Poecilotriccus (sylvia and plumbeiceps), and Zimmerius (chrysops) (Ames 1971, McKitrick 1985, Mobley and Prum 1995), posess Mm. obliqui ventrales, a pair of ventrally-originating intrinsic syringeal muscles. In most taxa, these muscles insert on the ventral or lateral side of the syrinx, but in the kingbird assemblage and in Platyrinchus, Hemitriccus, Lophotriccus, Oncostoma, Leptopogon, Inezia, Myiopagis, Tyranniscus, Omithion, Camptostoma, Suiriri, Phaeomyias, Nesotriccus, Serpophaga, Anairetes, and some Phylloscartes, these muscles insert on the dorsal side of the syrinx (Ames 1971). Many piprids and cotingids posess intrinsic syringeal muscles, but, with the exception of those in the problematic tyrannoid Sapayoa, these are different in detail of fiber direction and form of insertion. Prum and Lanyon (1989) and Prum (1990, 1992) have hypothesized the intrinsic syringeal muscles of piprids and cotingids to be derived independently of the Mm. obliqui ventrales in tyrannids. I hypothesize Mm. obliqui 53

ventrales to be derived in tyrannids, with ventrally or laterally inserting muscles (50.1) and dorsally inserting muscles (50.2) coded as alternative unordered derived states of a single character. 51. Mm. obliqui ventrales well developed.--In the kingbird assemblage (Lanyon 1984), in Elaenia, Myiopagis, Tyrannulus, Tyranniscus, Mecocerculus (calopterus, minor, poecilocercus, and stictopterus), Omithion, Camptostoma, Suiriri, Capsiempis, and Phaeomyias (Lanyon 1988a), and in Hymenops, Polioxolmis, and Cnemarchus (pers.obs.), the Mm. obliqui ventrales are relatively well developed, appearing as round, bulging muscle masses. In other tyrannids, these muscles are less developed, appearing flatter. I hypothesize well developed Mm. obliqui ventrales to be derived in the above taxa. 52. Mm. obliqui laterales.--In the tody-tyrant group and in Tolmomyias, Legatus, Myiozetetes, Elaenia, Myiopagis, Camptostoma, Suiriri, Pseudocolopteryx, Phaeomyias, Ochthoeca, Ramphotrigon, and some Phylloscartes, there is a pair of laterally-originating and laterally-inserting intrinsic syringeal muscles called Mm. obliqui laterales (Ames 1971). In most of these taxa, these muscles exist in addition to Mm. obliqui ventrales, but apparently in some tody-tyrants, Mm. obliqui laterales are the only intrinsic syringeal muscles. These intrinsic muscles are unique among tyrannoids and Lanyon (1984) hypothesized them to be derived in tyrannids. I hypothesize Mm. obliqui laterales to be derived in the above taxa.

Plumage 53.. Crown patch.--In the kingbird assemblage (except for Phelpsia and some Conopias) and in Platyrinchus, Myiophobus, Myiotriccus, Pseudelaenia, Euscarthmus, Myiopagis, Tyrannulus, Polystictus, Serpophaga, Pyrrhomyias, Neopipo, Myiobius (except erythrurus), Machetomis, Muscigralla, Tachuris, and some Elaenia and Pseudocolopteryx, the crown plumage contains a patch of white, red, orange, or yellow which is usually concealed or semi-concealed. Several cotingids and piprids have a similar crown patch, but since it is absent in the next two outgroups, the fumarioids and Old World suboscines, it is probably derived in the tyrannoids. I hypothesize concealed or semi-concealed crown patches to be derived in the above tyrannids. 54. Sexually dichromatic plumage.--In Alectrurus, Lessonia, Pyrocephalus, 54

Hymenops, and some Knipolegus and Poecilotriccus, males and females have markedly different plumage coloration and patterns. Elsewhere, sexually dichromatic plumage is present in piprids and some cotingids and fumarioids. But since sexual monochromatism appears to be primitive in Old World suboscines, I hypothesize monochromatism to be primitive in the New World groups as well. I follow Lanyon (1986) in hypothesizing sexually dichromatic plumage to be derived in the above tyrannids. 55. Wing pattemed.—-In Lessonia, Polioxolmis, Cnemarchus, Myiotheretes, Neoxolmis, Gubemetes, Pyrrhomyias, and Hirundinea, the wing is distinctively patterned: the remiges are black or fuscous distally, and buff or cinnamon-rufous proximally. In Alectrurus, Hymenops, Heteroxolmis, and some Xolmis and Knipolegus, the remiges are similarly patterned except they are white proximally. These wing patterns are unique among tyrannoids. Similar rufous patterns are present in some fumarioids. I hypothesize these wing patterns to be derived in the above tyrannids and code buff or cinnamon-rufous patterns (55.1) and white patterns (55.2) as alternative unordered derived states of a single character. 56. Wing bars.--Many tyrannids have two wing bars. Their color can be assigned to one of three rough categories: rufous-buff, yellow, or white. Although several genera are polymorphic for presence or absence of wing bars (Poecilotriccus, Hemitriccus, Lophotriccus, Phylloscartes, Myiopagis, Pseudocolopteryx, Serpophaga, Ochthoeca, Contopus, Knipolegus, Myiotheretes, Attila, and Rhytiptema) and many are polymorphic for their color, it appears likely that this character will, in some instances, provide evidence of relationships among genera (Lanyon 1986). Rufous-buff wing bars are present in Leptopogon, Myiophobus, Euscarthmus, Mecocerculus (calopterus, minor and some leucophrys), Camptostoma, Polystictus, Nesotriccus, Silvicultrix, Colorhamphus, Cnemotriccus, Aphanotriccus, Mitrephanes, Pyrrhomyias, Deltarhynchus, Ramphotrigon, Cnipodectes, and some Poecilotriccus, Phylloscartes, Phaeomyias, Ochthoeca,

Lathrotriccus, Xenotriccus, Contopus, Empidonax, Knipolegus, Myiotheretes, and Attila. Yellow wing bars are present in Tyranniscus, Capsiempis, Rhynchocyclus, Tolmomyias,

Todirostrum, Oncostoma, and some Poecilotriccus, Hemitriccus, Lophotriccus, Phylloscartes, and Myiopagis. White wing bars are present in Muscigralla, Sublegatus, Inezia, Pseudelaenia, Elaenia, Tyrannulus, Phyllomyiasfasciatus, Mecocerculus 55

(stictopterus and poecilocercus), Suiriri, Anairetes, Satrapa, and some Mecocerculus leucophrys, Pseudocolopteryx, Phaeomyias, Serpophaga, Ochthoeca, Lathrotriccus,

Xenotriccus, Contopus, Empidonax, Knipolegus, Rhytiptema, and Myiarchus. Wing bars occur elsewhere in a few cotingid genera and some fumarioids, but since wing bars are absent in Old World suboscines, I hypothesize their absence to be primitive in the New World groups as well. I hypothesize all tyrannid wing bars to be derived and code rufous- buff (56.1), yellow (56.2), and white (56.3) wing bars as alternative unordered derived states of a single character. 57. Throat streaked.--In Uromyias, Anairetes, Agriomis, Polioxolmis, Cnemarchus, Myiotheretes, Lophotriccus, and some Hemitriccus, the throat plumage has dark streaks against a white background. This morphology is unique among tyrannoids and I hypothesize it to be derived in these eight taxa. 58. Outer web of outer rectrix pale.--In the Muscisaxicola group of the Empidonax assemblage (except for Heteroxolmis), and in Ochthoeca, Lessonia, Pyrocephalus, Satrapa, Anairetes, and some Tyrannus, Rhytiptema, and Myiarchus, the outer webs of the outer rectrices are pale (white to cinnamon-rufous). This morphology occurs elsewhere only in the cotingid Xenopsaris and I hypothesize it to be derived in the above tyrannids. 59. Tail elongated, forked.--In Alectrurus, Gubemetes, Muscipipra, and some Tyrannus, the tail is elongated and forked. This morphology occurs elsewhere only in the cotingid Phibalura and I hypothesize it to be derived in the above tyrannids. 60. Tail rufous.--In Onychorhynchus, Cnipodectes, Neopipo, Myiobius erythrurus, Attila, Casiornis, and some Myiodynastes, Pseudotriccus, Ramphotrigon, and Rhytiptema, the entire tail is rufous. Elsewhere among tyrannoids, rufous tails occur in only a couple of cotingids. Although rufous tails are common among fumarioids, they are absent in the next outgroup, the Old World suboscines. I hypothesize the absence of this trait to be primitive in the New World groups and the rufous tails of the above tyrannids to be derived. 61. Rufous underparts.--In Pyrrhomyias, Hirundinea, Neopipo, Myiobius erythrurus, Cnemarchus, and some Myiotheretes, Mionectes, Ochthoeca, and Rhytiptema, the plumage of the underparts is rufous. Elsewhere among tyrannoids, rufous underparts occur in only a couple of cotingids. Although rufous underparts are present in some fumarioids, they are 56

absent in the next outgroup, the Old World suboscines. I hypothesize the absence of this trait to be primitive in the New World groups and the rufous underparts of the above tyrannids to be derived. 62.. Superciliary.-—In the kingbird assemblage (except for Tyrannus and some Myiozetetes), the Elaenia group of the Elaenia assemblage (except for Elaenia, Tyrannulus, Pseudocolopteryx, and some Myiopagis, Serpophaga, and Anairetes), and in Sublegatus, Inezia, Stigmatura, Pseudelaenia, Phyllomyias, Alectrurus, Silvicultrix, Ochthoeca, Cnemotriccus, Ochthomis, Satrapa, Cnemarchus, Gubemetes, Colonia, Tachuris, and some Phylloscartes, Muscisaxicola, Xolmis, Myiotheretes, and Neoxolmis, there is a superciliary stripe over or through the eye which may be white, yellow, or cinnamon. These eye stripes are unique among tyrannoids. They are common among fumarioids but absent in Old World suboscines. I hypothesize superciliary eye stripes to be derived in the above tyrannids independently of those in fumarioids. 63. Plumage like Sublegatus.--Sublegatus, Inezia, Suiriri, Mecocerculus leucophrys, and Phyllomyias fasciatus have a unique combination of plumage color patterns which Lanyon (1988a) hypothesized to be derived in Sublegatus and Inezia. The plumage is olive-green above, whitish to yellowish below, with white wing bars and a white superciliary eye stripe. This combination of color patterns is unique among tyrannoids and I hypothesize it to be derived in these five taxa 64. Plumage black and white.--In Amndinicola, F luvicola, Alectrurus, Hymenops, Heteroxolmis, Corythopis, Colonia, and some Sayomis, Knipolegus, and Xolmis, the plumage is black and white or brown and white. Although such coloration is present in some cotingids, piprids, and fumarioids, it is absent in Old World suboscines. I follow Lanyon (1986) and hypothesize black and white or brown and white plumage to be derived in the above tyrannids. 65. Rictal bristles long.--In Onychorhynchus, Cnipodectes, and Myiobius, the rictal bristles are as long or nearly as long as the bill. These exceptionally long rictal bristles are unique among tyrannoids and I hypothesize them to be derived in these three taxa. 66. Pointed crest.--In Xenotriccus, Contopus, and Mitrephanes, the feathers of the crown'form a pointed, prominent crest. This morphology is unique among tyrannoids and I hypothesize it to be derived in these three taxa. 57

67. Primaries notched.—-In Tyrannus, Empidonomus, Griseotyrannus, Arundinicola, Alectrurus, Lessonia, Agriomis, Xolmis, Heteroxolmis, Polioxolmis, Neoxolmis, Muscipipra, and some Knipolegus and Myiotheretes, the outer one to three primaries are notched or attenuated. Elsewhere, notched primaries occur in a few cotingids (Snow 1982). I follow Lanyon (1986) and hypothesize notched primaries to be derived in the above tyrannids independently of those in other tyrannoids.

Behavior/Ecology

68. Inhabit open grassland or marsh.—-Machetomis, Muscigralla, Tachuris, Pseudocolopteryx, Polystictus, Arundinicola, F luvicola, Alectrurus, Lessonia, Hymenops,

Muscisaxicola, Agriomis, Heteroxolmis, Neoxolmis, Gubemetes, and some Xolmis,

Tyrannus, and Sayomis inhabit open grassland or marsh. Hirundinea inhabits Open cliffs, canyons, and roadcuts. All other tyrannids, piprids, and cotingids primarily inhabit forests and woodlands. I hypothesize the occupation of open habitats to be derived in the above tyrannids and code grassland or marsh (68.1) and cliff (68.2) habitats as alternative unordered derived states of a single character. 69. Foraging behavior.--Fitzpatrick (1980) characterized the foraging behavior of nearly every tyrannid genus and recognized ten ‘foraging mode’ categories in the family. Here I treat each of these foraging mode categories as alternative unordered derived states of a single character, with the exception of one: I hypothesize the ‘fruit/upward hover-glean generalist’ mode to be the primitive state because that is the mode which seems to most closely correspond to the foraging behavior of the majority of piprids and cotingids (Snow 1982, Hilty and Brown 1986, Marini 1992). (69.1) Fruit/hawk generalist. Legatus, Myiozetetes, and Tyrannopsis combine frugivory with aerial hawking of flying insects. They use these capture techniques in roughly equal proportions. (69.2) Outward hover-glean generalist. Con0pias, Sublegatus, Attila, Rhytiptema, Sirystes, Casiomis, and Myiarchus often glean insects from foliage surfaces via an outward or downward ‘sally’ or capture flight. The bird may slow down and hover while capturing the insect or snap it up while in direct flight, often striking the foliage and continuing to a new perch. Upward hover-gleaning and aerial hawking are additional 58

significant elements of their foraging repertoires. (69.3) Enclosed perch hawk generalist. Myiophobus, Myiotriccus, Nesotriccus, Cnemotriccus, Aphanotriccus, Lathrotriccus, Empidonax, and Myiobius (except erythrurus) have foraging repertoires composed of roughly equal proportions of aerial hawking, upward hover-gleaning, and upward striking. (69.4) Near ground generalist. Phelpsia, Pitangus, Philohydar, Arundinicola, Silvicultrix,

Ochthoeca, Colorhamphus, Sayomis, Pyrocephalus, and Ochthomis usually hunt from perches near the ground and use a wide variety of capture techniques including perch-to- ground and perch-to-water sallies, upward and outward hover-gleaning and striking, and aerial hawking. (69.5) Perch-glean specialists. Inezia, Stigmatura, Pseudelaenia, Euscarthmus, Myiopagis, Mecocerculus, Omithion, Camptostoma, Suiriri, Pseudocolopteryx, Uromyias, Capsiempis, Serpophaga, Anairetes, and Tachuris capture most of their prey while perched, without sallying. Hover-gleaning and frugivory are minor components of their repertoire. (69.6) Upward strike specialists. Rhynchocyclus, Tolmomyias, Onychorhynchus, Platyrinchus, Todirostrum, Poecilotriccus, Hemitriccus, Lophotriccus, Oncostoma, Myiodynastes, Phylloscartes, Pseudotriccus, Corythopis, Myiobius erythrurus, and Ramphotrigon capture most of their prey in an explosively rapid upward sally to the underside of a leaf, snatching or scooping an insect off the under-surface, usually without hovering. (69.7) Aerial hawk specialists. Megarynchus, Tyrannus, Empidonomus, Griseotyrannus, Alectrurus, Contopus, Mitrephanes, Knipolegus, Satrapa, Gubemetes, Pyrrhomyias, Hirundinea, and Colonia capture most of their prey out of the air. (69.8) Perch-to-ground specialists. Agriomis, Xolmis, Heteroxolmis, Polioxolmis, Cnemarchus, Myiotheretes, and Neoxolmis hunt from low perches, usually sallying to the ground to catch prey. Aerial hawking may be a minor component of their repertoire. (69.9) Ground specialists. F luvicola, Lessonia, Hymenops, Muscisaxicola, Machetomis, and Muscigralla stand, walk, or run on the ground and search for their prey which they pick off the ground or low vegetation. Aerial hawking from the ground may be a minor component of their repertoire. 59

70. Cavity nesting.--In the Myiarchus assemblage and in Myiodynastes, Conopias, Myiotriccus, Xolmis, Colonia, and Machetomis, nests are located in tree cavities or holes in earthen banks. This behavior occurs elsewhere in the cotingid genus Tityra and in some fumarioids, but is absent among Old World suboscines. I hypothesize cavity nesting to be derived in the above tyrannids. 71. Enclosed nest within hanging vegetation—Zimmerius, Phylloscartes, and some Camptostoma build an enclosed nest with a side entrance which is placed within a hanging mass of vegetation. Although enclosed nests are built by several other tyrannids and by some taxa in each successive outgroup, the placement of such a nest within hanging vegetation is unique and I hypothesize this behavior to be derived in these three taxa. 72. Enclosed nest pendant.--Myiobius, Leptopogon, Rhynchocyclus, Tolmomyias,

Hemitriccus, Lophotriccus, Oncostoma, and some Todirostrum suspend their globular, enclosed nests from a branch. Silvicultrix, Mionectes, Onychorhynchus, Cnipodectes, Poecilotriccus, and some Todirostrum build similar pendant nests which are elongate (1m or more). Elsewhere among New World suboscines, enclosed, pendant nests occur only in some species of the cotingid genus Pachyramphus and in a few fumariids. I hypothesize the enclosed, pendant nests of the above tyrannids to be derived independently of those in

Pachyramphus and the fumariids and code globular (72.1) and elongate (72.2) nests as alternative unordered derived states of a single character. 73. Enclosed, pendant nest retort-shaped.--The enclosed, pendant, globular nests of Rhynchocyclus and Tolmomyias are shaped like a chemists’ retort with a downward- pointing entrance spout. This nest type is unique and I hypothesize it to be derived in these two taxa.

74. Enclosed nest with visor over entrance.--The enclosed nests of Todirostrum, Poecilotriccus, Lophotriccus, Oncostoma, Mionectes, Leptopogon, Arundinicola, and Myiobius erythrurus have a short roof or visor over the side entrance. This nest feature is unique among tyrannoids and I hypothesize it to be derived in the above taxa. 75. Nest used as d0rmitory.--In Rhynchocyclus and Tolmomyias, adults sleep in their nests at all seasons (Skutch 1960). Rhynchocyclus builds separate but similar nests for reproduction and sleeping, while Tolmomyias sleeps in its reproduction nest even after reproduction is finished. This behavior is apparently unique among tyrannoids and I 60 hypothesize it to be derived in these two taxa.

Eggs

76. Eggs unmarked.--In Mionectes, Leptopogon, Inezia, Pseudelaenia, Tyrannulus, Tyranniscus, Suiriri, Pseudocolopteryx, Polystictus, Mecocerculus leucophrys, Capsiempis, Phaeomyias, Nesotriccus, Serpophaga, Anairetes, Arundinicola, Silvicultrix,

Sayomis, Colonia, Tachuris, Lophotriccus, and some Todirostrum, Euscarthmus, Empidonax, and Muscisaxicola, the eggs are plainly colored, usually some shade of white. In all other tyrannoids for which eggs have been described, the eggs are marked with dark blotches and streaks. I hypothesize unmarked eggs to be derived in the above taxa. LIST OF REFERENCES LIST OF REFERENCES

Ames, P. L. 1971. The morphology of the syrinx in passerine birds. Bulletin of the Peabody Museum of Natural History No. 37.

Beaver, D. L., and P. H. Baldwin. 1975. Ecological overlap and the problem of competition and sympatry in the western and Hammond’s flycatchers. Condor 77: 1-13.

Bent, A. C. 1942. Life Histories of North American flycatchers, larks, swallows and their allies. Bulletin of the U. S. National Museum, no. 179.

Berenbaum, M. R. 1981. Patterns of furanocoumarin distribution and insect herbivory in the Umbelliferae: plant chemistry and community structure. Ecology 62: 1254-1266.

Berenbaum, M. R. 1983. Coumarins and caterpillars: A case for coevolution. Evolution 37: 163-179.

Berlepsch, H. Graf von. 1907. Studien fiber Tyranniden. Omis 14:463-493.

Block, B. A., and J. R. Finnerty. 1994. Endotherrny in fishes: A phylogenetic analysis of constraints, predispositions, and selection pressures. Environmental Biology of Fishes 40:283-302.

Bremer, K. 1988. The limits of amino acid sequence data in angiosperrn phylogenetic reconstruction. Evolution 42:795-803.

Brooks, D. R., and D. A. McLennan. 1991. Phylogeny, ecology, and behavior. University of Chicago Press, Chicago.

Coddington, J. A. 1988. Cladistic tests of adaptational hypotheses. Cladistics 423-22. de Queiroz, A., and P. H. Wimberger. 1993. The usefulness of behavior for phylogeny estimation: Levels of homoplasy in behavioral and morphological characters. Evolution 47:46-60. de Queiroz, K. 1996. Including the characters of interest during tree reconstruction and the problems of circularity and bias in studies of character evolution. American Naturalist 148:700-708.

Felsenstein, J. 1982. Numerical methods for inferring evolutionary trees. Quarterly Review of Biology 57:379-404.

Felsenstein, J. 1985. Phylogenies and the comparative method. American Naturalist 1 25: 1- 15.

61 62

Fitzpatrick, J. W. 1980. Foraging behavior of Neotropical tyrant flycatchers. Condor 82:43-57.

Fitzpatrick, J. W. 1985. Form, foraging behavior, and adaptive radiation in the Tyrannidae. Pages 447-470 in Neotropical Ornithology (P. A. Buckley, M. S. Foster, E. S. Morton, R. S. Ridgely, and F. G. Buckley, Eds.). Ornithological Monographs 36. American Omithologists Union, Washington, D. C.

Futuyma, D. J ., M. C. Keese, and D. J. Funk. 1995. Genetic constraints on macroevolution: The evolution of host affiliation in the leaf beetle genus Ophraella. Evolution 49:797-809.

Futuyma, D. J ., and G. Moreno. 1988. The evolution of ecological specialization. Annual Review of Ecology and Systematics 19:207-233.

Griffiths, C. S. 1994. Monophyly of the Falconiformes based on syringeal morphology. Auk 111:787-805.

Harvey, P. 11, and M. D. Page]. 1991. The comparative method in evolutionary biology. Oxford University Press, New York.

Hellmayr, C. E. 1927. Catalogue of birds of the Americas and the adjacent islands. Part V. Tyrannidae. Field Museum of Natural History, Zoological Series, 13:1-517.

Hilty, S. L., and W. L. Brown. 1986. A guide to the birds of Colombia. Princeton University Press, Princeton, New Jersey.

Jones, 6., M. Morton, P. M. Hughes, and R. M. Budden. 1993. Echolocation, flight morphology and foraging strategies of some West African hipposiderid bats. Journal of Zoology 230:385-400.

Kallersjo, M., J. S. Farris, A. G. Kluge, and C. Bult. 1992. Skewness and permutation. Cladistics 8: 275-288.

Krebs, J. R., and N. B. Davies. 1991. An introduction to behavioral ecology, 3rd ed. Blackwell, Oxford.

Kroodsma, D. E. 1996. Ecology of passerine song development. Pages 3-19 in Ecology and evolution of acoustic communication in birds (D. E. Kroodsma and E. H. Miller, Eds.) Cornell University Press, Ithaca, New York.

Lanyon, W. E. 1984. A phylogeny of the kingbirds and their allies. American Museum Novitates No. 2797.

Lanyon, W. E. 1985. A phylogeny of the myiarchine flycatchers. Pages 361-380 in Neotropical omithology (P. A. Buckley, M. 8. Foster, E. S. Morton, R. S. Ridgely, and F. G. Buckley, Eds.). Ornithological Monographs No. 36. American Omithologists Union, Washington, D. C.

Lanyon, W. E. 1986. A phylogeny of the thirty-three genera in the Empidonax assemblage of tyrant flycatchers. American Museum Novitates No. 2846. 63

Lanyon, W. E. 1988a. A phylogeny of the thirty-two genera in the Elaenia assemblage of tyrant flycatchers. American Museum Novitates No. 2914.

Lanyon, W. E. 1988b. The phylogenetic affinities of the flycatcher genera Myiobius Darwin and Terenotriccus Ridgeway. American Museum Novitates No. 2915.

Lanyon, W. E. 1988c. A phylogeny of the flatbill and tody-tyrant assemblage of tyrant flycatchers. American Museum Novitates No. 2923.

Maddison, W. P. 1990. A method for testing the correlated evolution of two binary characters: Are gains or losses concentrated on certain branches of a phylogenetic tree? Evolution 44:539-557.

Maddison, W. P., M. J. Donoghue, and D. R. Maddison. 1984. Outgroup analysis and parsimony. Systematic Zoology 33:83-103.

Maddison, W. P., and D. R. Maddison. 1992. MacClade: Analysis of phylogeny and character evolution. Version 3.04. Sinauer Associates, Sunderland, Massachusetts.

Marini, M. A. 1992. Foraging behavior and diet of the helmeted manakin. Condor 94: 151-158.

McKitrick, M. C. 1985. Monophyly of the Tyrannidae (Aves): Comparison of morphology and DNA. Systematic Zoology 34:35-45.

Meyer, A., J. M. Morrissey, and M. Schartl. 1994. Recurrent origin of a sexually selected trait in Xiphophorus fishes inferred from a molecular phylogeny. Nature 368:539-542.

Mobley, J. A., and R. O. Prum. 1995. Phylogenetic relationships of the cinnamon tyrant, Neopipo cinnamomea, to the tyrant flycatchers (T yrannidae). Condor 97:650-662.

Muller, A. 1996. Host-plant specialization in western palearctic anthidiine bees (Hymenoptera: Apoidea: Megachilidae). Ecological Monographs 66:235-257.

Norberg, U. 1994. Wing design, flight performance, and habitat use in bats. Pages 205- 239 in Ecological morphology: Integrative organismal biology (P. C. Wainwright and S. M. Reilly, Eds.). University of Chicago Press, Chicago.

Pellmyr, O., J. N. Thompson, J. M. Brown, and R. G. Harrison. 1996. Evolution of pollination and mutualism in the yucca moth lineage. American Naturalist 148:827- 847.

Price, P. W. 1983. Hypotheses on organization and evolution in herbivorous insect communities. Pages 559-596 in Variable plants and herbivores in natural and managed systems (R. F. Denno and M. S. McClure, Eds.). Academic Press, New York.

Prum, R. O. 1990. A test of the monophyly of the manakins (Pipridae) and of the cotingas (Cotingidae) based on morphology. Occasional Papers of the Museum of Zoology, University of Michigan 723:1-44. 64

Prum, R. O. 1992. Syringeal morphology, phylogeny, and evolution of the Neotropical manakins (Aves: Pipridae). American Museum Novitates No. 3043.

Prum, R. O. 1993. Phylogeny, biogeography, and evolution of the broadbills (Eurylaimidae) and asities (Philepittidae) based on morphology. Auk 110:304-324.

Prum, R. O., and W. E. Lanyon. 1989. Monophyly and phylogeny of the Schijfornis group (Tyrannoidea). Condor 91:444-461.

Rausher, M. D. 1993. The evolution of habitat preference: Avoidance and adaptation. Pages 259-283 in Evolution of insect pests: Patterns of variation (Kim, K. C., and B. A. McPheron, Eds.). Wiley, New York.

Rensch, B. 1959. Evolution above the species level. Columbia University Press, New York.

Ridgely, R. S., and G. Tudor. 1994. The birds of South America, vol. 2: The suboscine passerines. University of Texas Press, Austin. Saitou, N. 1989. A theoretical study of the underestimation of branch lengths by the maximum parsimony principle. Systematic Zoology 38: 1-6.

Sanderson, M. J ., and M. J. Donoghue. 1989. Patterns of variation in levels of homoplasy. Evolution 43: 1781-1795.

Sclater, P. L. 1888. Catalogue of birds in the British Museum. Volume 14. British Museum (Natural History), London.

Sherry, T. W. 1984. Comparative dietary ecology of sympatric insectivorous neotropical flycatchers (Tyrannidae). Ecological Monographs 54:313-338. Sibley, C. G., and J. E. Ahlquist. 1985. Phylogeny and classification of New World suboscine passerine birds (Passeriforrnes: Oligomyodi: Tyrannides). Pages 396- 428 in Neotropical Ornithology (P. A. Buckley, M. S. Foster, E. S. Morton, R. S. Ridgely, and F. G. Buckley, Eds.). Ornithological Monographs 36. American Omithologists Union, Washington, D. C.

Sibley, C. G., and B. L. Monroe, Jr. 1990. Distribution and taxonomy of birds of the world. Yale University Press, New Haven, Connecticut.

Sillén-Tullberg, B., and A. P. Moller. 1993. The relationship between concealed ovulation and mating systems in anthropoid primates: A phylogenetic analysis. American Naturalist 141 : 1-25.

Simpson, G. G. 1953. The major features of evolution. Columbia University Press, New York.

Skutch, A. F. 1960. Life histories of Central American birds II. Cooper Ornithological Society, Berkeley, California.

Snow, D. W. 1982. The cotingas. Cornell University Press, Ithaca, New York.

Sorenson, M. D. 1996. TreeRot. University of Michigan, Ann Arbor. 65

Stiles, F. G., and A. F. Skutch. 1989. A guide to the birds of Costa Rica. Cornell University Press, Ithaca, New York.

Swofford, D. L. 1993. PAUP: Phylogenetic analysis using parsimony, version 3.1. Illinois Natural History Survey, Charnpaign, Illinois.

Sytsma, K. J. 1990. DNA and morphology: Inference of plant phylogeny. Trends in Ecology & Evolution 5:104-110.

Thompson, J. N. 1994. The coevolutionary process. University of Chicago Press, Chicago.

Traylor, M. A., Jr. 1977. A classification of the tyrant flycatchers (Tyrannidae). Bulletin of the Museum of Comparative Zoology 148: 129-184.

Traylor, M. A., Jr. 1979. Family Tyrannidae. Pages 1-245 in Check-list of birds of the world, vol. VIII (M. A. Traylor, Jr., Ed.). Museum of Comparative Zoology, Cambridge, Massachusetts.

Traylor, M. A., Jr., and J. W. Fitzpatrick. 1982. A survey of the tyrant flycatchers. Living Bird 1917-50.

Warter, S. L. 1965. The cranial osteology of the New World Tyrannoidea and its taxonomic implications. Ph.D. dissertation, Louisiana State University, Baton Rouge. "’llllllll"llllllllES