<%BANNER%>

Evolutionary History of Simaroubaceae (Sapindales)

Permanent Link: http://ufdc.ufl.edu/UFE0017000/00001

Material Information

Title: Evolutionary History of Simaroubaceae (Sapindales) Systematics, Biogeography and Diversification
Physical Description: 1 online resource (176 p.)
Language: english
Creator: Clayton, Joshua
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: ancestral, area, biogeography, calibration, cladogenesis, dispersal, diversification, extinction, fossil, historical, molecular, morphology, phylogeny, rates, reconstruction, sapindales, simaroubaceae, systematics, vicariance
Botany -- Dissertations, Academic -- UF
Genre: Botany thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Simaroubaceae (Sapindales) are a geographically widespread and ecologically diverse clade of pantropical and temperate trees and shrubs, consisting of 22 genera and ca. 100. Phylogenetic relationships within the family have been poorly understood, and detailed studies of the evolutionary history of pantropical clades are still relatively few. Some of the underlying causes of shifts in diversification rates are investigated in Simaroubaceae, both historical movements and morphological innovation. Molecular and morphological data are used to reconstruct the phylogeny of Simaroubaceae and clarify generic limits, employing Maximum Parsimony (MP) and Bayesian approaches. Both MP and Bayesian analyses of combined data from four gene regions produce well-supported phylogenies, and these data support the monophyly of most traditionally circumscribed genera. Morphological data also support most genera, but provide little backbone resolution compared to molecular data. Bayesian uncorrelated rates analyses and robust fossil calibrations are used in divergence date estimation, and a recently developed likelihood method for reconstructing ancestral areas (the Dispersal-Extinction-Cladogenesis (DEC) model) is employed. Simaroubaceae exhibit an early history of range expansion between major continental areas in the Northern Hemisphere, but reconstruction of ancestral areas for lineages diverging in the early Tertiary are sensitive to the parameters of the model used. A North American origin is suggested for the family, with migration via Beringia by ancestral taxa. In contrast to traditional views, long-distance dispersal events are inferred to be common, particularly in the late Oligocene and later. Morphological data reconstruct clades of genera, but show a high degree of homoplasy. Character-state reconstructions reveal a number of putative morphological synapomorphies for major clades in Simaroubaceae, and a complex history of character-state transitions. Shifts in diversification rate were examined using a number of statistical techniques, both temporal and topological. Four nodes in the phylogeny represent a point of diversification rate increase, and all can be attributed to a biogeographical shift, namely the arrival of the ancestor of Simaba, Simarouba and Pierreodendron in the New World, the movement of Castela from North to South America, the expansion of the ancestor of Brucea, Soulamea and Amaroria into Africa, and the arrival of Soulamea on New Caledonia. The topological imbalance demonstrated at the root node of Leitneria may be the result of extinction, as the fossil history of Leitneria indicates this clade was more diverse in the past. No morphological innovations of putative evolutionary significance were associated with increases in diversification rate, but analytical techniques in this area need further development.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Joshua Clayton.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Soltis, Douglas E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-02-28

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0017000:00001

Permanent Link: http://ufdc.ufl.edu/UFE0017000/00001

Material Information

Title: Evolutionary History of Simaroubaceae (Sapindales) Systematics, Biogeography and Diversification
Physical Description: 1 online resource (176 p.)
Language: english
Creator: Clayton, Joshua
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: ancestral, area, biogeography, calibration, cladogenesis, dispersal, diversification, extinction, fossil, historical, molecular, morphology, phylogeny, rates, reconstruction, sapindales, simaroubaceae, systematics, vicariance
Botany -- Dissertations, Academic -- UF
Genre: Botany thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Simaroubaceae (Sapindales) are a geographically widespread and ecologically diverse clade of pantropical and temperate trees and shrubs, consisting of 22 genera and ca. 100. Phylogenetic relationships within the family have been poorly understood, and detailed studies of the evolutionary history of pantropical clades are still relatively few. Some of the underlying causes of shifts in diversification rates are investigated in Simaroubaceae, both historical movements and morphological innovation. Molecular and morphological data are used to reconstruct the phylogeny of Simaroubaceae and clarify generic limits, employing Maximum Parsimony (MP) and Bayesian approaches. Both MP and Bayesian analyses of combined data from four gene regions produce well-supported phylogenies, and these data support the monophyly of most traditionally circumscribed genera. Morphological data also support most genera, but provide little backbone resolution compared to molecular data. Bayesian uncorrelated rates analyses and robust fossil calibrations are used in divergence date estimation, and a recently developed likelihood method for reconstructing ancestral areas (the Dispersal-Extinction-Cladogenesis (DEC) model) is employed. Simaroubaceae exhibit an early history of range expansion between major continental areas in the Northern Hemisphere, but reconstruction of ancestral areas for lineages diverging in the early Tertiary are sensitive to the parameters of the model used. A North American origin is suggested for the family, with migration via Beringia by ancestral taxa. In contrast to traditional views, long-distance dispersal events are inferred to be common, particularly in the late Oligocene and later. Morphological data reconstruct clades of genera, but show a high degree of homoplasy. Character-state reconstructions reveal a number of putative morphological synapomorphies for major clades in Simaroubaceae, and a complex history of character-state transitions. Shifts in diversification rate were examined using a number of statistical techniques, both temporal and topological. Four nodes in the phylogeny represent a point of diversification rate increase, and all can be attributed to a biogeographical shift, namely the arrival of the ancestor of Simaba, Simarouba and Pierreodendron in the New World, the movement of Castela from North to South America, the expansion of the ancestor of Brucea, Soulamea and Amaroria into Africa, and the arrival of Soulamea on New Caledonia. The topological imbalance demonstrated at the root node of Leitneria may be the result of extinction, as the fossil history of Leitneria indicates this clade was more diverse in the past. No morphological innovations of putative evolutionary significance were associated with increases in diversification rate, but analytical techniques in this area need further development.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Joshua Clayton.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Soltis, Douglas E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-02-28

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0017000:00001


This item has the following downloads:


Full Text

PAGE 1

1 EVOLUTIONARY HISTORY OF SIMAROUBACEAE (SAPINDALES): SYSTEMATICS, BIOGEOGRAPHY AND DIVERSIFICATION By JOSHUA WILLIAM CLAYTON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Joshua William Clayton

PAGE 3

3 To WilsonPhillips-X, for all the good times

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my comm ittee, Doug Soltis, Pam Soltis, Walter Judd, Steve Manchester, and Gustav Paulay. I am very grateful to everyone who provided material for this study: Christopher Quinn at the Sydney Botanic Ga rdens, Frieda Billiet at the National Botanic Garden of Belgium, Libby MacMillan at the Roya l Botanic Gardens, Edinburgh, Andrew Ford at CSIRO, Bruce Wannan of EPA, Queensland, Richard Abbott at FLMNH, Wang Hong at the Shishuangbanna Tropical Botanical Garden, China, Serena Lee at Singapore Botanic Gardens, Kew DNA Bank, MOBOT DNA Bank, and curators of MO, E, AAU, WAG, NY and CAY. Thanks also to Rick Ree (Fie ld Museum, Chicago, USA) and Stephen Smith (Yale University, USA) for help with Lagrange, Alexei Drum mond for comments rega rding BEAST, Aleksej Hvalj (Komorov Institute, St. Petersburg, Russia) for providing information on Leitneria fossils, and Wayt Thomas, Michael Moore, Scot Ke lchner, Susanne Renner and three anonymous reviewers for their helpful comments. Funding was provided by the National Science Foundation (angiosperm AtoL EF-0431266 to DES, PSS et al.; DDIG DEB-0710202 to JWC, DES), Botanical Society of America, and American Society of Plant Taxonomists. Finally, I would like to thank all my friends in the Soltis Lab, Cr ysta, Andrew, and my family, for their support.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................8LIST OF FIGURES .........................................................................................................................9ABSTRACT ...................................................................................................................... .............12 CHAPTER 1 INTRODUCTION .................................................................................................................. 142 OVERVIEW OF SIMAROUBACEAE (SAPINDALES) ..................................................... 16Family Description ............................................................................................................ .....16Vegetative Morphology ......................................................................................................... .17Vegetative Anatomy ............................................................................................................ ...18Inflorescence Structure ...........................................................................................................18Flower Structure .....................................................................................................................19Embryology .................................................................................................................... ........20Pollen Morphology .................................................................................................................20Karyology ..................................................................................................................... ..........21Reproductive Biology .............................................................................................................21Fruits and Seeds ......................................................................................................................21Dispersal ..................................................................................................................... ............23Phytochemistry ................................................................................................................ .......23Distribution and Habitats ..................................................................................................... ...24Fossil History ..........................................................................................................................25Affinities .................................................................................................................... .............25Relationships within the Family ............................................................................................. 26Uses and Economic Importance ............................................................................................. 27Key to the New World Genera ............................................................................................... 27Key to the Old World Genera ................................................................................................. 28Descriptions of Genera ...........................................................................................................301.Picrasma Blume ................................................................................................... 302.Castela Turpin ......................................................................................................313.Holacantha A.Gray ............................................................................................... 314.Ailanthus Desf. .....................................................................................................325.Leitneria Chapm. ..................................................................................................326.Soulamea Lam. .....................................................................................................337.Amaroria A.Gray ..................................................................................................338.Brucea J.F.Mill .....................................................................................................349.Laumoniera Noot. .................................................................................................3410.Nothospondias Engl. .............................................................................................35

PAGE 6

6 11.Picrolemma Hook.f. ..............................................................................................3512.Quassia L. .............................................................................................................3613.Samadera Gaertn. ................................................................................................. 3614.Eurycoma Jack ......................................................................................................3715.Gymnostemon Aubrv. & Pellegr. ........................................................................3716.Perriera Courchet ................................................................................................. 3817.Hannoa Planch. .....................................................................................................3818.Odyendea (Pierre) Engl. ....................................................................................... 3919.Iridosma Aubrv. & Pellegr. ................................................................................3920.Pierreodendron A.Chev. ......................................................................................3921.Simarouba Aubl. ...................................................................................................4022.Simaba Aubl. ........................................................................................................413 MOLECULAR PHYLOGENY OF SIMAROUBACEAE BASED ON CHLOROPLAST AND NUCLEAR MARKERS ..................................................................42Introduction .................................................................................................................. ...........42Methods ..................................................................................................................................45Taxon Sampling ...............................................................................................................45DNA Extraction, Amplificat ion and Sequencing ............................................................ 46Alignment and Indel Coding ........................................................................................... 47Maximum Parsimony Analyses and Data Congruence ................................................... 48Bayesian Analyses and Pa rtitioning Strategies ............................................................... 49Results .....................................................................................................................................51Data Congruence .............................................................................................................51Phylogenetic Analyses and Tree Topology .....................................................................52Partitioning Strategies ..................................................................................................... 53Discussion .................................................................................................................... ...........54Data Congruence and Partitioning Strategies .................................................................. 54Systematics of the Simaroubaceae .................................................................................. 574 LIKELIHOOD ANALYSIS OF GEOGR APHIC RANGE EVOLUTION IN SIMAROUBACEAE .............................................................................................................. 72Introduction .................................................................................................................. ...........72Methods ..................................................................................................................................74Taxon Sampling, DNA Sequencing and Phylogenetic Analyses .................................... 74Divergence Date Estimation ............................................................................................75Fossil calibration ......................................................................................................75Molecular rates analyses .......................................................................................... 78Biogeographic Analyses Using a Likelihood Approach .................................................80Results .....................................................................................................................................83Divergence Date Estimation ............................................................................................83Biogeographic Analyses ..................................................................................................84Discussion .................................................................................................................... ...........86Divergence Date Estimation ............................................................................................86Biogeographic Analyses ..................................................................................................87

PAGE 7

7 Likelihood models .................................................................................................... 87Origin and early history of Simaroubaceae .............................................................. 91Long-distance dispersal ............................................................................................ 925 DIVERSIFICATION AND MORP HOLOGICAL EVOLUTION IN SIMAROUBACEAE ............................................................................................................ 105Introduction .................................................................................................................. .........105Methods ................................................................................................................................108Phylogeny Estimation and Character Evolution ............................................................ 108Shifts in Diversification Rates ....................................................................................... 112Correlates of Shifts in Diversification Rate ................................................................... 114Results ...................................................................................................................................116Phylogeny Estimation and Character Evolution ............................................................ 116Diversification Analyses ...............................................................................................116Discussion .................................................................................................................... .........1186 CONCLUSIONS .................................................................................................................. 138 APPENDIX A SPECIMEN DATA FOR MOLECULAR ANALYSES ...................................................... 142B SOURCES OF MORPHOLOGICAL DATA ...................................................................... 145C SPECIMEN DATA FOR MORPHOLOGICAL ANALYSES ............................................ 147D MORPHOLOGICAL CHARACTERS USED IN PHYLOGENETIC ANALYSES ..........150LIST OF REFERENCES .............................................................................................................158BIOGRAPHICAL SKETCH .......................................................................................................176

PAGE 8

8 LIST OF TABLES Table page 3-1 List of the 22 genera of the Si maroubaceae grouped by Englers (1931) tribal classification. ............................................................................................................... ......663-2 Primers used for PCR amplification and sequencing. ....................................................... 673-3 Wilcoxon sum of rank test results show ing pairwise comparisons between data partitions. ................................................................................................................... ........683-4 Results of maximum parsimony analyses for individual data partitions and combined analyses of Simaroubaceae. ...............................................................................................693-5 Summary of results from data partitioning analyses. .........................................................694-1 Fossil calibrations used in BEAST analyses ......................................................................974-2 Divergence dates (in Ma) resulting from molecular rates analyses for eight different calibration schemes. .......................................................................................................... .974-3 Dispersal network model implemented in Lagrange, ........................................................ 984-4 Results for biogeographic models tested. .......................................................................... 994-5 Ancestral range permutations for the chosen model (M3),. ............................................. 1005-1 Results of maximum parsimony analyses for morphological data and combined morphological and molecular data of Simaroubaceae. .................................................... 1285-2 Synapomorphies for major clades of Simaroubaceae ...................................................... 1285-3 Character evolution on the Simar oubaceae phylogeny for 26 characters under maximum parsimony reconstructions (MPR). ................................................................. 1295-4 Characters tested for correlations with diversifica tion rate shift. .................................... 130

PAGE 9

9 LIST OF FIGURES Figure page 3-1 Strict consensus of eight most pars imonious trees recovered from a combined analysis. ..................................................................................................................... .........703-2 Phylogram of the majority rule consensus of trees for Simaroubaceae ............................. 714-1 Map showing extant geographic distribu tion and approximate fossil localities for Simaroubaceae. ................................................................................................................ 1014-2 Phylogram of the majority rule consensus of trees for Simaroubaceae,. ......................... 1024-3 Chronogram with 95% HPD bars, base d on BEAST analyses using two fossil calibrations .................................................................................................................. .....1034-4 Ancestral area reconstruction for Simaroubaceae using model M3, with ancestral ranges NS, NF, NA, FM, FA, FE, AU, AE and MAU, and stratified dispersal probabilities between areas (D1). ..................................................................................... 1045-1 Phylogram randomly selected from 100,000 most-parsimonious trees recovered from 71 morphological characters for Simaroubaceae.. ........................................................... 1315-2 Phylogeny randomly selected of 100,000 mo st parsimonious trees recovered from a combined analysis. ...........................................................................................................1325-3 Unambiguous character-state changes based on maximum parsimony reconstruction for reproductive traits.. ..................................................................................................... 1335-4 Unambiguous character-state changes based on maximum parsimony reconstruction for vegetative traits and general ecology. ........................................................................ 1345-5 Crown group analysis of clades of Simaroubaceae using methods of Magalln and Sanderson (2001).. ...........................................................................................................1355-6 Stem group analysis of clades of Si maroubaceae using methods of Magalln and Sanderson (2001). ............................................................................................................1365-7 Chronogram for Simaroubaceae (reproduced from Figure 4-3) showing shifts in diversification rates. .........................................................................................................137

PAGE 10

10 LIST OF ABBREVIATIONS A Mainland Asia and southeast Asia (in AAR) AAR Ancestral area reconstruction AIC Akaike Information Criterion BS Bootstrap support D1 or D2 Dispersal scenario 1 or 2 DIVA Dispersal-vicariance Analysis E Europe (in AAR) F Tropical Africa (in AAR) HPD Highest posterior density LRT Likelihood ratio test N North and Central America and Caribbean Islands (in AAR) M Madagascar (in AAR) M1, M2 etc Biogeographic likelihood model 1, 2 etc Ma Million years ago MCMC Markov chain Monte Carlo MPR Most parsimonious reconstruction Myr Million years NALB North Atlantic land bridge NW New World OW Old World PP Posterior probability S South America (in AAR) SE Asia Southeast Asia sp. Species (singular)

PAGE 11

11 spp. Species (plural) U Australia, New Guinea, Papua New Guinea, New Caledonia and the Pacific Islands (in AAR) UCLN Uncorrelated lognormal

PAGE 12

12 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EVOLUTIONARY HISTORY OF SIMAROUBACEAE (SAPINDALES): SYSTEMATICS, BIOGEOGRAPHY AND DIVERSIFICATION By Joshua William Clayton August, 2008 Chair: Douglas E. Soltis Major: Botany Simaroubaceae (Sapindales) are a geographica lly widespread and ecologically diverse clade of pantropical and temperate trees and shrubs, consisting of 22 genera and ca. 100. Phylogenetic relationships within the family ha ve been poorly understood, and detailed studies of the evolutionary history of pantropical clades are still rela tively few. Some of the underlying causes of shifts in diversifi cation rates are investigated in Simaroubaceae, both historical movements and morphological i nnovation. Molecular and morphol ogical data are used to reconstruct the phylogeny of Simaroubaceae an d clarify generic limits, employing Maximum Parsimony (MP) and Bayesian approaches. Both MP and Bayesian analyses of combined data from four gene regions produce well-supporte d phylogenies, and th ese data support the monophyly of most traditionally circumscribed genera. Morphologi cal data also support most genera, but provide little backbone resolution compared to molecu lar data. Bayesian uncorrelated rates analyses and robust fossil ca librations are used in divergence date estimation, and a recently developed likelihood method for reconstructing ancestral areas (the Dispersal-ExtinctionCladogenesis (DEC) model) is employed. Sima roubaceae exhibit an early history of range expansion between major continental areas in the Northern Hemisphere, but reconstruction of ancestral areas for lineages diverging in the early Tertiary are sensitive to the parameters of the

PAGE 13

13 model used. A North American orig in is suggested for the family, with migration via Beringia by ancestral taxa. In contrast to traditional views, long-distance dispersal events are inferred to be common, particularly in the late Oligocene and la ter. Morphological data reconstruct clades of genera, but show a high degree of homoplasy. Character-state recons tructions reveal a number of putative morphological synapomorphies for majo r clades in Simaroubaceae, and a complex history of character-stat e transitions. Shifts in diversificati on rate were examined using a number of statistical techniqu es, both temporal and topological. Fo ur nodes in the phylogeny represent a point of diversification ra te increase, and all can be attributed to a biogeographical shift, namely the arrival of the ancestor of Simaba, Simarouba and Pierreodendron in the New World, the movement of Castela from North to South America, the expansion of the ancestor of Brucea Soulamea and Amaroria into Africa, and the arrival of Soulamea on New Caledonia. The topological imbalance demonstrat ed at the root node of Leitneria may be the result of extinction, as the fossil history of Leitneria indicates this clade was more diverse in the past. No morphological innovations of putativ e evolutionary significance were associated with increases in diversification rate, but analytical techni ques in this area need further development.

PAGE 14

14 CHAPTER 1 INTRODUCTION Opportunities to undertake in-depth studies of the evolutionary history of groups of organisms are ever-increasing, with advances in technology for collecting molecular data, and constant improveme nts and refinements to analytical techniques. The search for general patterns in biome evolution and biodiversit y gradients is reliant on studies of individual clades. One such clade with potential for a detailed treatment is the pantropical angiosperm family Simaroubaceae. Simaroubaceae (in the Sapindales) are a geograph ically widespread and ecologically diverse clade of trees and shrubs, cons isting of 22 genera and ca. 100 species. Previous phylogenetic analyses of the family were limited to a singl e gene and seven genera (Fernando et al., 1995), and relationships among the genera, until this study, remain poorly understood. The family has a history of taxonomic uncertainty, and has only recently been recircumscribed in a phylogenetic framework (Fernando and Quinn, 1995). Molecular data are explored using a variety of methods for phylogenetic reconstruction, a nd special attention will be given to model selection in partitioning the data. A molecular phylogeny has al so provided the opportunity to revisit and consolidate data on the na tural history of the group, and produ ce the first general synopsis of the family since its recircumscription. Several pantro pical clades have been examined in terms of their extant and ancestral geogr aphic distributions (e.g. Lavin et al., 2000, 2004; Renner et al., 2001; Davis et al., 2002; Richards on et al., 2004; Zerega et al., 2005), including other members of the Sapindales (Weeks et al., 2005; Muellner et al., 2006), but historic al biogeography in the Simaroubaceae had not been addressed, until the i nvestigation presented herein. The familys current and fossil geographic distribution may be the culmination of a number of historical dispersal events, both long-distan ce and overland migrations via putative land bridges, as in studies of other pantropical clades. With a robust phylogeny and divergence date estimates,

PAGE 15

15 hypotheses of past movements can be looked at through ancestral area reconstruction. A number of angiosperm clades have also been investigated in terms of variation in species diversity, and possible correlates to this diversity (e.g. Eriksson and Bremer, 1991; Hodges and Arnold, 1995; Richardson et al., 2001a, 2001b; von Hagen and Kadereit, 2003; Becerra, 2005; Kay et al., 2005; Hughes and Eastwood, 2006; Good-Avila et al., 2006; Moore and Donoghue, 2007). Sufficient variation in diversification rates has been found in Simaroubaceae, and so a similar approach to some of these studies was undertaken. This study allowed the examination of factors both intrinsic to the organisms, that is, key morphological innovations, and the external environment, ecology and historical dispersal events, that might influence speciation rates, and thus covary with diversification rate shifts in the family. In this study I apply the latest techniques in data analysis to tackle the following questions: What are the phylogenetic relationships within Simaroubaceae? Over what timescale has the family and its constituent subclades evolved? How have historical migration events influenced the extant and fossil geographic distributions of the family? What accounts for the variation in species diversity of clades within the family?

PAGE 16

16 CHAPTER 2 OVERVIEW OF SIMAROUBACEAE (SAPINDALES) Simaroubaceae A.P. de Candolle, Prodromus I. 733 (1824), n om. cons. Ailanthaceae (Arnott) J. Agar dh (1858). Castelaceae J. Agardh (1858). Holacanthaceae Jadin (1901). Leitneriaceae Benth. & Hook.f. (1880). Quassiaceae Bertolini (1827). Simabaceae Horaninow (1847). Soulameaceae Endlicher (1874). Family Description Trees and shrubs, occasionally with thorns. Pi th conspicuous, with tr iterpenoid compounds of the quassinoid type throughout vege tative tissues. Leaves alternate, spirally arranged, exstipulate (stipules found in Picrasma ), pinnately compound or unifoliolate (rarely trifoliolate); leaflets entire, coarsely toothed, serrate or basally lobed, sometimes with conspicuous pitted or flattened glands beneath or above, venation pinnate, brochidodromus or occasionally reticulate. Hairs mostly simple, unicellular or multicellular, sometim es glandular-capitate. Inflorescences terminal or axillary, determinate thyrses, sometimes app earing raceme-like, pseudo-umbellate, catkin-like or flowers clustered in leaf axils. Flowers bi sexual, polygamous or unisexual, actinomorphic, bracteate (bracts large an d surrounding flowers in Leitneria ), pedicels bracteolate, occasionally jointed; sepals 4 (0 in Leitneria ), fused below, calyx sometimes splitting unevenly, occasionally bearing glands; petals 4() (0 in Leitneria ), free; stamens 4(), free, filaments often with hairy appendage, anthers do rsifixed, basifixed or versatile, opening by 2 longitudinal slits, introrse (occa sionally extrorse to latrorse); ovary superior, (1)2 carpels, free or fused basally, occasionally fused axially and deeply lobed, axile placentation, one ovule per locule, anatropous; stylodia free to connate, occas ionally absent, stigmas divergent, stellately spreading, or a single slightly lobed or cap itate stigma; fruit 1 samaroid or drupaceous

PAGE 17

17 mericarps, exocarp thin, fleshy, occasionally dry, nut-like, often carinate, endocarp reticulate; testa membranaceous, cotyledons pla noconvex, endosperm mostly lacking. A family of 22 genera and about 109 species, mainly tropical and subtropical but some temperate species. Vegetative Morphology The family is woody, composed of large trees up to 50 m, shr ubs, subshrubs, and occasionally suffructescent plants with all the leaves basal in Simaba. The wood is pithy or fistulous (Cronquist, 1944 d), making it lightweight, and the bark and twigs are often striated. The family is typified by a bitter taste to the ba rk and twigs, on account of quassinoid compounds in scattered secretory cells throughout the vegeta tive structures (Cronqui st, 1981). Thorns are present in Castela and in Holacantha where they occur at the tip s of all branches (Cronquist, 1944d). Leaves are predominantly once-pinnately co mpound, arranged spirally around cylindrical stems. Unifoliolate leaves have evolved multiple times (based on studies of character evolution, Clayton, unpublished data), a nd are characteristic of Castela Leitneria, Amaroria and Samadera and are found in six species of Soulamea (Jaffre and Fambart, 2002) and two species of Simaba ( S. monophylla and S. obovata). The leaves of Holacantha are reduced to scales or absent entirely, except in the seedlings (Cronquist, 1944 d). Leaflets are alte rnate, subopposite or opposite, but always opposite in Quassia, which has a distinctive winged and jointed rachis. Leaflet shape is diverse, but st rongly asymmetrical leaf bases are common in compound leaves. Leaf margins are predominantly enti re, but are serrate or coarsely toothed in temperate species of Ailanthus Picrasma and Brucea Stipules are found in Picrasma (Nooteboom, 1962 b), in which they are triangular, ovate or orbicular, and early caducous.

PAGE 18

18 Vegetative Anatomy Wood anatomy is described in detail by Webber (1936) and Record and Hess (1943; New World genera). Growth rings are present but indistinct, and diff use porous or ring-porous ( Ailanthus, Leitneria ). Wood is dominated by fi ber-tracheids, except in Holacantha and Castela in which wood fibers are libriform (Webber, 1936). Vessels have spiral thickenings in Castela, Holacantha and Leitneria but these are rare or absent in the rest of the family. Normal wood parenchyma cells are sparse to m oderately abundant, th e cells often septate and crystalliferous (Webber, 1936; Record and Hess, 1943). Verti cal secretory canals are common in stems (Spiekerkoetter, 1924), and in Leitneria resin ducts are described as present in the margin of the pith (Record and Hess, 1943). Nodes are trior multilacunar, and calcium oxalate crystals are present in parenchymatous tissues (Cronquist, 19 81). Flattish or concave glands are common on leaf surfaces, typically towards the margin, a nd often associated with teeth if present. Multicellular secretory glands are found on the abaxial surface of the sepals of Samadera (Nair and Joseph, 1957). Primarily unicellular, but also multicellular and glan dular hairs are common on the inflorescence axes and floral organs (Nai r and Joseph, 1957; Nair and Joshi, 1958; Nair and Sukumaran, 1960; Nooteboom, 1962b). Inflorescence Structure Inflorescences can be axillary or terminal, a nd are determinate thyr ses, with the dichasia often appearing fasciculate or reduced to a si ngle flower, giving the appearance of a panicle ( sensu Weberling, 1989). Thyrses vary between open and spreading (e.g. Ailanthus, Eurycoma, Picrolemma ), and narrow, elongate a nd sparsely branched (e.g. Brucea, Soulamea, Amaroria ). In Simarouba and Picrolemma the staminate thyrses are larger and have more flowers than the carpellate thyrse s (Cronquist, 1944b ). Picrasma has a short, broad, round ed thyrse with a long peduncle (often described as a cyme), and in Samadera the inflorescence axis is condensed to

PAGE 19

19 form a pseudo-umbel (Nair and Joseph, 1957). Quassia amara has a distinctive long raceme-like thyrse, occasionally branch ed at the base, and in Castela the flowers are occasionally solitary or clustered in the leaf axils (Cronquist, 1944d), as in some Samadera. In Leitneria the inflorescence is an erect or occasionally pe ndulous catkin-like thyrse: in the staminate inflorescence the flowers cluster in cymules of three in the axils of large, spirally-arranged bracts; in the carpellate inflores cence the flowers are solitary in the bract axils (Abbe and Earle, 1940). Flower Structure Flowers in Sim aroubaceae are small, actinomorphic, open and 4or 5-merous (trimerous in Soulamea ), with an intrastaminal nectary disc. Petals are usually red, pink, yellow, pale green or white. Unlike the majority of the family, Quassia has flowers with elongate, glabrous petals (sometimes with hairs at the ba se) that are coherent forming a tube, and the stamens and style are exserted. Leitneria is unusual in having asepalous and apetalous flowers, although Abbe and Earle (1940) observed vestig ial perianth structures. Leitneria also has a unicarpellate gynoecium, with vascular bundles suggesti ng reduction from a bicarpella te gynoecium (Abbe and Earle, 1940). The androecium in the family is most commonly obdiplostemonous, although it is reduced to haplostemony in Picrasma, Brucea, Picrolemma and Eurycoma In the latter two genera the stamens alternate with stam inodes in the staminate flowers. In Pierreodendron the outer whorl of stamens is doubled. Adaxial s cale-like appendages on the filaments occur in eleven genera, and are taxonomica lly significant at the species level, varying in shape, length, pubescence and bifurcation. In unisexual flowers, vestigial staminodes and pistillodes are common. Filaments are inserted at the base of the nectary disc, which can vary between strongly lobed, cushion-like, tall and cylindrical, conical or inconspi cuous. The disc usually enlarges in

PAGE 20

20 fruit. The gynoecium of Soulamea is reduced to two or three fuse d carpels, and is a single carpel in Amaroria Embryology Embryology for the fami ly was reviewed by Mauritzon (1935). Detailed studies of embryo anatomy are available for Ailanthus (Narayana, 1957), Samadera (Nair and Joseph, 1957), Brucea (Nair and Sukumaran, 1960) and Leitneria (Pfeiffer, 1912), and the following characteristics should be considered typical fo r the family: the anther wall consists of an epidermis, a fibrous endothecium, two to thr ee middle layers and a multinucleate secretory tapetum (binucleate in Ailanthus excelsa ); microsporogenesis is simultaneous; pollen tetrads are tetrahedral and decussate, shed at the two-celled stage; ovu les are anatrapous or hemianatropous, crassinucellate and bitegmic, the i nner integument forming the zig-zag micropyle; the nucellus is multinucleate, and the nucellar ep idermis divides to form a cap; the archesporium can be multicellular or unicellular ( Ailanthus ), only one archesporial cell developing further; megaspores are arranged linearly (a solit ary T-shaped tetrad is reported for Ailanthus integrifolia ); the chalazal megaspore develops into a Polygonum -type embryo sac; fertilization may be chalazogamous, mesogamous or por ogamous (Wiger, 1935), but only porogamy is confirmed in Samadera and Brucea ; endosperm development precedes embryo development, and is of the Nuclear type. Pollen Morphology Basak (1963, 1967) and Moncada and Machado (1987) used light microscopy to survey pollen m orphology in Quassia, Samadera Simarouba, Simaba, Eurycoma, Soulamea Ailanthus, Brucea, Castela and Picrasma and Zavada and Dilcher (1986) examined Leitneria with SEM and TEM. Pollen grains are 3zonocolporate, typically 20 m long by 13 m wide, prolate in equatorial view (sometimes subspheroidal in Castela and spheroidal in Samadera and

PAGE 21

21 Leitneria ) and planaperturate, with distinctly lalongate endoape rtures; however, Quassia pollen grains are suboblate and angulaper turate, with a square type of endoaperture. Exine is 2 m thick, and the surface pattern finely to coarsely reticulate, sometim es verrucate, in most genera. The exine is striate in Soulamea and Brucea and striato-reticulate in Quassia Karyology Simaroubaceae have a ba se chromosome number of 8 (Stevens, 2006). Bennett and Leitch (2005) record 2 n = 64 in Ailanthus integrifolia which would suggest the plant is octoploid. Raven (1975) reports x = 16 for Leitneria and Castela coccinea has 2n =26 (Bernardello et al., 1990). Reproductive Biology Simaroubaceae can be he rmaphroditic, monoecious or dioecious. The extent of selfcompatability is unknown; however, flowers of Quassia amara have been shown to self-fertilize (Roubik et al., 1985). Insect-pollination predominates in the family, the flowers typically being small, actinomorphic, open, fragrant and borne in thyrses, attracting generalist small insects such as small bees and moths (e.g. Aubrville, 1962; Hardesty, 2005). Quassia amara is hummingbird-pollinated, as suggest ed by the raceme-like inflorescences bearing deep pink or red tubular flowers. Roubik et al. (1985) observed the role of nectar robbers in re productive fitness of Q. amara revealing that flowers were visited by nectar robbing bees ( Trigona) and hummingbirds, as well as the primary hummingbird pollinator. Leitneria shows strong morphological divergence towards wind-pollination in that the flowers lack a perianth and nectary disk, and are borne in catkin-like inflor escences that develop before the leaves emerge. Fruits and Seeds Fruits in the fam ily are predominantly schizocarpous with drupaceous mericarps, and typically only 1 carpels reach maturity. The drup es have a thin pericarp, in which the exocarp

PAGE 22

22 can be fleshy (e.g. Hannoa, Quassia, Simaba), woody and fibrous ( Samadera ) or thin and dry ( Eurycoma Leitneria some Brucea ). The fleshy fruits can be pale yellow to red to deep purpleblack, with a bitter taste, globose, obovoid, ovoid or ellipsoid, and between 0.3 cm and 10 cm long. The drupes are often carinate or bicarinate and flattened, and in Samadera indica are strongly laterally dorso-ventrally compressed with a narrow unilateral thinner edge in the apical half. In Ailanthus each carpel develops into a samaroid me ricarp, elliptic in shape and tapering at each end. Variation in samara morphology is discussed in some detail by Nooteboom (1962b) and Corbett and Manchester (2004). In Soulamea the carpels remain fused in fruit, forming a dry, narrowlyto broadly-winged, obcor date fruit. Fernando and Qui nn (1992) discuss variation in pericarp anatomy in the family in detail. The ex ocarp varies in thickness and lignification, and in Ailanthus is lacking except for the epidermal laye r. Fernando and Quinn (1992) describe the endocarp as consisting of a br oad homogeneous zone of irre gularly arranged isodiametric sclereids with a strongly lignified inner epidermis. Castela and Picrasma lack the typical lightly lignified mesocarp and parenchymatous outer mesocarp. Nothospondias has an unusual Spondias -type endocarp, similar to that found in the Anacardiaceae (Fernando and Quinn, 1992). The embryo is straight or curved, and consis ts of two large pla noconvex cotyledons and a short plumule. Most Simaroubaceae have little or no endosperm, except for Brucea (Nair and Sukumaran, 1960) and some Soulamea (Nooteboom, 1962b). Fatty oil and aleuron bodies are the most common seed storage products in the family, but starch is also reported from seeds of Simaba and Perriera (Linsbauer, 1926) and Leitneria (Pfeiffer, 1912), and reserve celluloses also occur (Czaja, 1978; Stevens, 2006). The seed coat is thin and hard, undistinguished or with scattered lignified cells (Stevens, 2006), and is described as membranaceous in some genera.

PAGE 23

23 Dispersal Fleshy drupaceous fruits of Sim aroubaceae are dispersed by fruit-eating birds and mammals, often primates (e.g. Hardesty et al., 2005). The samaroid mericarps of Ailanthus disperse over small distances by wind. Fruits of Samadera indica a species that frequents alluvial and swamp forest, and Soulamea amara, a littoral species, are dispersed by water (Nooteboom, 1962b), which may account for their broad geographical distributions. Leitneria is also suspected to be water dispersed, typically growing in freshwater and brackish swamps. In all cases buoyancy is provided by an air cavity between seed and endocarp. Phytochemistry Simaroubaceae are characterized by their quass inoid chemistry. Quassinioids are triterpenoid derivatives, biosynthetically related to the limonoids of Rutaceae and Meliaceae (Da Silva and Gottlieb, 1987), and are considered by some (Dreyer, 1983; Waterman, 1983) to be further steps down the oxidative pathway of limonoids. Quassinoid structural and chemical characteristics are summarized in Waterman and Grundon (1983) and Da Silva and Gottlieb (1987), who report 35 different structural types in Picrasma alone. Pentacyclic triterpenes are also common (Hegnauer, 1983). Alkaloids have been reported in nine Simaroubaceae genera (Mester, 1983), most commonly tryptophan derived, but also a quinolone alka loid is reported in Ailanthus Only a single simple coumarin has been detected in the family, in Picrasma and Ailanthus (Gray, 1983). Of the flavonoi d groups, flavonol glycosid es and glycoflavone are reported in Ailanthus (Harborne, 1983) and fl avonols and flavones in Leitneria (Giannasi, 1986). Essential oils contained within secretory cells and resin canals cont ain a low proportion of volatile compounds compared to Rutaceae a nd Meliaceae, and are in smaller amounts (Hegnauer, 1983). Tannin content is low to considerable, and with re latively high levels of gallic and ellagic acid (Hegnauer, 1983), although Leitneria lacks ellagic acid (Giannisi, 1986).

PAGE 24

24 Distribution and Habitats Simaroubaceae have a pr imarily pantropical distribution; however, some species of Brucea, Castela Holacantha, Ailanthus and Picrasma are subtropical, and Ailanthus altissima, Picrasma quassioides and Leitneria floridana grow in temperate climates. Generic diversity is split evenly among the New World, Africa, and Asia and Australasia; however, half of the species in the family occur in the New World. Picrasma is disjunct among Asia, SE Asia and Central and South America, Brucea is disjunct between Africa and SE Asia, and Soulamea has one species in the Seychelles, one widespread in Malesia and Polynesia, and the remainder endemic to New Caledonia. Samadera is primarily Australian and SE Asian, but S. indica occurs as far west as India and Madagascar. Several genera in the Simaroubaceae consist of one or two species with restricted geographic ranges, the majority of these gene ra being in Africa. Simaba (25 spp.) is the most species-rich genus and is restricted to Centra l and South America. Simaroubaceae are found in moist lo wland tropical forest (although Brucea mollis is recorded as a high as 1800m in the Philippines, and Odyendea gabonensis at 2500m in Gabon), dry deciduous forest, and open sandy or savannah-type vegetation. Soulamea amara is a littoral species, Castela and Holacantha are found in desert and dry scrub environments, and Leitneria Samadera indica and occasionally Pierreodendron inhabit swamp forest. Eurycoma is classified as silicicolous, showing a preference for acidic, leached sandy soils (Nooteboom, 1962 b). Dating and biogeographic analyses (see Chapter 4) suggest the family originated in North America in the early Tertiary. However, ancient vi cariant and dispersal patt erns in the family are obscured by a multitude of more recent migratio n events, within and between the continents, post-Oligocene.

PAGE 25

25 Fossil History Fossils of the distinctive samaroid f ruits of Ailanthus are found across the entire Northern Hemisphere, dating from the early Eocene up to th e Pleistocene (Corbett and Manchester, 2004). Three extinct species have been recognise d, with the earliest o ccurrence a samara of A. confucii from the Green River Formation, Wyoming. Leaf fossils are also known with reasonable certainty from the Oligocene of Germany and Mi ocene of China, exhibiting distinctive basal teeth with enlarged glands on the leaflets, characteristic of extant A. altissima (Corbett and Manchester, 2004). Leitneria has fossil fruits in western Siberia from the Oligocene and in Europe from the Miocene to the Pliocene (Dor ofeev, 1994; Nikitin, 2006a), but no fossil record from North America. Fossils fruits in transverse section have identical endocarp anatomy to extant Leitneria floridana (Dorofeev, 1994). Less well unders tood are fossil fruits of Chaneya an extinct genus from the Tertiary of North America, Europe and Ea stern Asia (Wang and Manchester, 2000; Teodoridis and Kvacek, 2005). Teodoridis and Kvacek (2005) suggest an affiliation with the extant genus Picrasma based on gynoecial morphology and persistent winglike petals; however, the fossil ha s distinctive oil cells typical of Rutaceae. Fossil leaves formerly reported as Leitneria from the Eocene of Tennessee (Berry, 1916) were subsequently reassigned to Rubiaceae, based on stipule c onfiguration, epidermal anatomy a nd leaf architecture (Roth and Dilcher, 1979). Fossil pollen of Simaroubaceae has been reported for Ailanthus (Song et al., 2004) and Leitneria (Machen, 1971), but given the lack of distinctive morphological characteristics in extant Simaroubaceae pol len (Basak, 1963, 1967; Moncada and Machado, 1987), these are considered unreliable. Affinities In the traditional circumscription, Simaroubaceae s.l. com prised six subfamilies (Engler, 1931). However, molecular work by Fernando et al (1995; also see Gadek et al., 1996) showed

PAGE 26

26 the family to be polyphyletic, with subfamilies originating in several places within eurosids I and II (sensu APG II, 2003). Members of subfamily Simarouboideae, however, form a wellsupported monophyletic group (excluding Harrisonia) within Sapindales. Leitneria a genus traditionally segregated into the monotypic family Leitneriaceae on account of its windpollinated flowers (Cronquist, 1981; Takhtajan, 1996), was also found to be part of the Simarouboideae clade. Hence, the subfamily was recircumscribed as Simaroubaceae s.s., a clade of 20 genera and approximately 95 species (Fernando and Quinn 1995). Nothospondias a monotypic genus sometimes placed in Anacardiaceae (Engler, 1905), is a member of the family (Van der Veken, 1960; Chapter 3). Also included is Laumoniera, a monotypic genus from Sumatra (Nooteboom, 1987) that was omitted from the family recircumscription of Fernando and Quinn (1995). Simaroubaceae are well supported as a memb er of a Simaroubaceae+Rutaceae+Meliaceae clade in Sapindales (Gadek et al., 1996; Kllersj et al., 1998; Savolainen et al., 2000 b; Soltis et al., 2000), but the familys sister group is st ill undetermined, with data supporting three alternative topologies Rutaceae sister to Si maroubaceae (Gadek et al., 1996); Meliaceae sister to Simaroubaceae (Chase et al., 1999; Muelln er et al., 2006); Rutaceae sister to Meliaceae (Fernando et al., 1995; Stevens, 2006). Trad itional morphological and phytochemical classifications typically suggest an affilia tion with Rutaceae and Meliaceae (e.g. Cronquist, 1981; Takhtajan, 1996). Relationships within the Family Englers (1931) classi fication of Simaroubaceae s.l. divided subfam ily Simarouboideae (Simaroubaceae s.s.) into three tribes: Simaroubeae, Picr asmeae and Soulameae. Tribes were delimited by presence or absence of filament appendages and the degree of fusion of carpels. Of Englers tribes, molecular data (Chapter 3) show only Soulameae, composed of Soulamea and

PAGE 27

27 Amaroria to be monophyletic. Relationships based on DNA sequence data from the chloroplast genome (rbcL atpB matK ) and nuclear genome (1kb of phyC ) produced a well-resolved and well-supported phylogeny (excluding Laumoniera and Iridosma ; Chapter 3), with all genera except Soulamea being monophyletic. For details of phyl ogenetic relationships see Chapter 3. Putative synapomorphies for genera are italicized in genus descriptions. Uses and Economic Importance A range of biological properties has been demonstrated by the quassinoids of Simaroubaceae, includin g antimalarial, antileukem ic, antiviral, insecticidal and amoebicidal properties (Polonsky, 1983; Klocke et al., 1985), and correspondi ngly, many genera are used locally as medicinal plants. Quassia amara and Picrasma quassioides have been used to aid digestion, and treat chroni c dyspepsia. Fruits of Brucea javanica were imported into Europe as a drug (Nooteboom, 1962 b), and the plant is used locally in Malaysia to treat malaria and dysentry. Eurycoma is used to treat malaria, diabetes, hypertension and stomach ache, typically by boiling the roots for drinking. Ailanthus is known in traditional Chines e and Korean medicine as a treatment for digestive complaints, haemo rrhoids and mastitis. Simaroubaceae are not commercially harvested for timber but are used loca lly in building in some areas of both the Old and New World. Leitneria (corkwood) is one of the lightes t known woods, and has been used traditionally by fisherman for net floats. Ailanthus (Tree of Heaven), Simarouba (paradise tree) and Quassia amara are cultivated and planted as ornamentals. Key to the New World Genera 1. Petals and sepals 0 or vestigial; flowers surrounded by large, hirsute bracts 5. Leitneria 1. Petals +, sepals +; bracts not large, not surr ounding flowers 2 2. Stamens with appendaged filaments 3

PAGE 28

28 2. Filaments lacking appendage 5 3. Leaf rachis distinctly winged and jointed 12. Quassia 3. Leaf rachis not winged, not jointed 4 4. Flowers unisexual; stigmas as long as style or longer, stellately spreading; leaflets alternate 21. Simarouba 4. Flowers bisexual; stigmas capitate or lobed; leaflets typically opposite or subopposite 22. Simaba 5. Leaves unifoliolate or absent; plant often armed with thorns; stamens twice as many as petals 6 5. Leaves pinnately compound; plant without thorns; stamens equal in number to petals 7 6. Plant with leaves; petals 4; stamens 8 2. Castela 6. Plant leafless, or leaves reduced to scales; petals 6; stamens 12 3. Holacantha 7. Staminodes present in staminate flowers; inflorescence elongate, narrowing above; fruit ellipsoi d, elongate, 200 mm long 11. Picrolemma 7. Staminodes absent or in carpellate fl owers only; inflores cence short, broad and rounded; fruit globose less than 15mm long 1. Picrasma Key to the Old World Genera 1. Stamens with appendaged filame nts 2 1. Filaments lacking appendage 8 2. Leaves unifoliolate; inflor escence a pseudo-umbel 13. Samadera 2. Leaves pinnately compound; inflorescence not umbellate 3 3. Leaf rachis jointed and often narrowly winged 12. Quassia

PAGE 29

29 3. Leaf rachis not jointed, not winged 4 4. Stamens alternating with outer whorl of staminodes or staminodal scales in staminate flowers; induplicateva lvate aestivati on; Indomalesia 14. Eurycoma 4. Staminodes absent in staminate flower s; contorted, imbricate, occasionally valvate aestivation; tropical Africa 5 5. Petals 7, valvate in bud; stamens 12 19. Iridosma 5. Petals 4, imbricate or contorted in bud; stamens 8() 6 6. Stamens (10)15(); leaves with 11 leaflets, up to 1 m long; leaflets apex with hard, pointed gland 20. Pierreodendron 6. Stamens 8; leaves with 3 leafle ts, less than 60 cm long; leaflets without hard pointed gland at apex 7 7. Calyx in bud irregularly rupturing in to 2 lobes; 5 petals; 10 stamens; 5 carpels; fruits 15 mm long 17. Hannoa 7. Calyx fused with 4() very shor t obtuse lobes; 4() petals; 8() stamens; 4 carpels; fruits 50 mm long 18. Odyendea 8. Gynoecium 1 or 2() fused carpels 9 8. Gynoecium (2)3 carpels (i f 2 then carpels free) 10 9. Gynoecium a single carpel; fruit ovoid, not winged; flowers 4or 5-merous 7. Amaroria 9. Gynoecium two or th ree carpels; fruit obcordate, winged; flowers predominantly 3-merous 6. Soulamea 10. Fruit samaroid 4. Ailanthus 10. Fruit drupaceous, fleshy or dry and nut-like 11 11. Stamens equal in number to petals 12

PAGE 30

30 11. Stamens twice as many as petals 14 12. Inflorescence short, broad and rounded; se pals and petals persistent in fruit, accrescent; fruit globose 1. Picrasma 12. Inflorescence mostly unbranched, elonga te; petals caducous in fruit; fruit ovoid or ellipsoid, or nut-like with 2 ribs when mature 13 13. Leaves imparipinnate; stigma s free, recurving; fruit 7 mm long 8. Brucea 13. Leaves paripinnate; stigmas connate, discoid; fruit 45 mm long 9. Laumoniera 14. Carpels 2; inflorescence axillary; Madagascar 16. Perriera 14. Carpels 4; inflorescence typical ly terminal; African tropics 15 15. Leaves with 19 leaflets; flowers 4-merous; fruits up to 45 mm in length; tropical west Africa 10. Nothospondias 15. Leaves with 13 leaflets; flowers 5-merous; fruits about 100 mm in length; Cote DIvoire endemic 15. Gymnostemon Descriptions of Genera 1. Picrasma Blume Picrasma Blume, Bijdr. Fl. Ned. Ind. 247 ( 1825); Cronquist, Brittonia 5:1 28 (1944). Small trees, sometimes to 20 m, or shrubs; monoecious or dioecious. Leaves imparipinnate, stipules present early caducous; leaflets opposite to subopposite, petiolula te, entire or serratecrenate, glabrous or nearly so, w ithout glands. Flowers in axillary, short and broad, rounded determinate thyrses with puberulent axes; sepals 4(), free or basally fused; petals 4(), valvate, mostly glabrous; stamens 4(), filaments lacking appendage anthers dorsifixed, staminodes absent in staminate flowers; disc fleshy, sometimes conical, glabrous or hairy; carpels (2)4(), free, stylodia fused above, so metimes free, stigmatic branches filiform,

PAGE 31

31 recurved. Fruit 1() drupaceous mericarps, globose not carinate, 5 mm long, exocarp red to blue-black at maturity, pericarp fleshy. Eight species, two in Asia and SE Asia, six in Mexico to Argentina, and Caribbean islands. 2. Castela Turpin Castela Turpin, Ann. Mus. Natl. Hist. Nat. 7:78 (1806); Cronquist, J. Arnold Arbor. 25:122 (1944). Shrubs, erect or trailing, or small trees to 5 m; dioecious; arme d with (occasionally branching) thorns. Leaves unifoliolate, petiol ate, entire, glabrous to tome ntose-pubescent, without glands. Flowers solitary, clustered in leaf axils, or in axillary, sparsely-flowered determinate thyrses, with typically a single, sparsely to densely hairy axis; sepals 4( ), basally fused; petals 4(), imbricate, glabrous to occasionally pubescent; stamens 8 or 10, filame nts lacking appendage, anthers dorsifixed, staminodes absent in stamin ate flowers; disc fleshy, ring-like, glabrous; carpels 4(), weakly united or free, stylodia fused at base, stig matic branches linear, divergent or recurved, occasionally circinately rolled. Fruit 1() drupaceous mericarps, lenticular, bicarinate, 6 mm long, exocarp red at maturity, pericarp fleshy. Twelve species from southern United States to Argentina, the Caribbean islands and the Galpagos. 3. Holacantha A.Gray Holacantha A.Gray, Pl. Nov. Thurb. 310 (1854) ; Cronquist, Brittonia 5:128 (1944). Depressed, ascending or erect shr ubs or sm all trees to 5 m; dioecious; armed with thorns at branch tips. Essentially leafless or leaves scale-like. Flowers in ax illary, short, densely-flowered determinate thyrses with one or two strongly hirsute axes, or app earing fasciculate in leaf axils; sepals 5, basally fused; petals 6, im bricate, strigose on abaxial surface; stamens 12 filaments lacking appendage, anthers dorsifixed, staminodes absent in st aminate flowers; disc narrow and ring-like, densely hairy to glabrous; carpels 6 weakly united, style short and

PAGE 32

32 broad, stigmatic branches stellately spreading. Fruit 1 drupaceous mericarps, ovoid and slightly compressed, sometimes carinate on abaxia l side, 5 mm long, e xocarp red or greenish at maturity, pericarp fleshy. Two species from so uthern California, southern and western Arizona to Mexico. 4. Ailanthus Desf. Ailanthus Desf., Mm. Acad. Roy. Sci. (Paris) I, 8:265 (1786); N ooteboom Fl. Males., Ser. 1, Spermat. 6:193 (1962 b). Large trees to 60 m; dioecious or monoecious. Le aves imparipinnate or paripinnate; leaflets opposite, subopposite or alternate below, petiolulate, entire to co arsely toothed, glabrous to densely pubescent, with sometimes large abaxial glands, occasionally domatia present as hair tufts at leaf base. Flowers in axillary or termin al determinate thyrses with glabrous to sparsely hairy axes; sepals 5(), fused basally or calyx cupular with very shor t lobes; petals 5(), induplicate-valvate glabrous to pubescent; stamens 10, f ilaments lacking appendage, anthers ventrifixed, staminodes absent in staminate flowers; disc fles hy, glabrous; carpels 2, stylodia free to fused, stigmatic branches peltate, stellately spreading, sometimes recurved. Fruit 1 samaroid mericarps with elongate, memb ranous wings tapering towards the ends 25 mm long, exocarp brown at maturity, pericarp dry. Five species from Turkestan, India, China, SE Asia, and northern Australia. 5. Leitneria Chapm. Leitneria Chapm., Fl. South. U.S., 428 (1860). Small tree to 6 m ; typically dioecious. Leaves unifoliolate petiolate, entire, villous, without glands. Flowers in axillary, elongate, highly reduced thyrses appearing catkin-like with cymules of 1 flowers in staminate inflorescence, flowers solitary in carpellate inflorescence, surrounded by densely hirsute bracts and arranged on an single glabrous axis; perianth 0 in

PAGE 33

33 staminate flowers, vestigial in carpellate flowers ; stamens (1)4 per flower in bract axil, filaments lacking appendage, anthers basifixed to dorsifixed, staminodes absent in staminate flowers; disc absent or rudimentary ; carpel 1 stigmatic branch dist ally expanded, recurved. Fruit a drupe, narrowly ellipsoid, conspicuously flattened, bicarinate, 12 mm long, exocarp brown at maturity, pericarp dry to occasionally fleshy. One species, L. floridana in SE United States. 6. Soulamea Lam. Soulamea Lam., Encycl. 1:449 (1783); Jaffr & Fam bart, Adansonia 24:159 (2002). Shrubs or small trees to 5() m; dioecious or bisexual (S. amara). Leaves unifoliolate or imparipinnate; leaflets opposite, petiolulate, leav es petiolate, entire and often revolute, densely pubescent or glabrous on adaxial surface, sometim es with glands. Flower s in axillary, elongate determinate thyrses, typically with a single, often ferruginous-tomentose major axis; sepals 3( 5), basally fused; petals 3(), glabrous to pubescent towards the base; stamens 6(), filaments lacking appendage, anthers basifixe d to dorsifixed, staminodes absent in staminate flowers; disc fleshy, glabrous; carpels 2(), c onnate, stylodia free, flattened, hor izontally appressed to carpel, stigma fleshy, rarely reniform. Fruit samaroid, 2-celled, obcordate, flattened, with a distinct wing, 10 mm long, exocarp brown at maturity pericarp dry. Thirteen species, one widespread in SE Asia and Polynesia ( S. amara), one endemic to the Seychelles ( S. terminalioides ), and eleven species endemic to New Caledonia. 7. Amaroria A.Gray Amaroria A.Gray, Bot. U. St. Expl. Exped. 1:356. I. 40 (1854); Smith, Fl. Vit. Nova Vol. 3:479 487 (1985). Small tree to 15() m; dioecious. Leaves uni foliolate, petiolate, entire, glands unknown. Flowers in axillary. elongate determinate thyrse s with a single major axis; sepals 4, basally

PAGE 34

34 fused; petals 4, glabrous or sometimes short strigillose along adaxial midline; stamens 8 or 10, filaments lacking appendage, anthers dorsifixed, staminodes absent in st aminate flowers; disc fleshy, globose; carpel 1 stigma sessile. Fruit a drupe, ovoid to subglobose, slightly flattened, sometimes inconspicuously carinate, 17 mm long, exocarp greenish yellow, becoming white at maturity, pericarp fleshy. One species, A. soulameoides endemic to Fiji. 8. Brucea J.F.Mill Brucea J.F.Mill, Icon. Anim. Pl. t. 25 (1780). Shrubs or sm all trees to 12 m; dioecious or polygamous. Leaves imparipinnate; leaflets opposite, petiolulate to subsessile, entire or crenat e-serrate, ferruginous-pubescent to glabrous, with dotted glands associated with periphe ral secondary venation underneath Flowers in axillary, elongate determinate thyrses, typically with a single glabrous to dens ely pubescent major axis; sepals (3)4(), basally fused; petals (3)4(), imbric ate, glabrous to densely pubescent; stamens (3)4(), protruding between disc lobes, fila ments lacking appendage, anthers basifixed to dorsifixed, staminodes absent in st aminate flowers; disc fleshy, gl abrous; carpels (3)4(), free or united at the base, stylodia fu sed at base, stigmatic branches linear, recurved or bending inwards. Fruit 1() drupaceous mericarps, ovoid, bicarinate, 4 mm long, exocarp red to black at maturity, pericarp dry to thinly fleshy. Six to seven species, trop ical Africa to tropical and subtropical Asia and northern Australia. 9. Laumoniera Noot. Laumoniera Noot., Blumea 32:383 (1987). Small tree to 16 m; dioecious. Leaves paripinnate ; leaflets petiolulate, entire, glands unknown. Flowers in axillary determinate thyrses, typically with a single pubescent axis; sepals 4, basally fused; petals 4, sparsely pubescent; stamens 4, filaments lacking appendage, staminodes absent in staminate flowers; disc fleshy, slightly ha iry; carpels 4, free, stigmas sessile, connate, discoid,

PAGE 35

35 covering top of ovaries Fruit 1 drupaceous mericarps, ellipsoid, 450 mm long, exocarp yellow at maturity, pericarp fleshy. One species, L. bruceadelpha, Indonesia. 10. Nothospondias Engl. Nothospondias Engl., Bot. Jahrb. Syst. 36:216 (1905); Van der Veken, Bull. Jardin Bot. tat Bruxelles 30:105 (1960). Tree to 25 m; dioecious. Leaves im paripinnate; leaf lets opposite to alternate, petiolulate, entire, glabrous, without glands. Flowers in axillary or terminal determinate thyrses, with multiple densely pubescent axes; sepals 4, basally fused; petals 4, slightly imbricate, glabrous to puberulent; stamens 8, filaments lacking appenda ge, anthers basifixed, staminodes absent in staminate flowers; disc fleshy, glabrous; carpels 4, free, style fused below. Fruit 1 drupaceous mericarps, ovoid-ellipsoid, 20 mm long, exocarp yellow to orange at maturity, pericarp fleshy. One species, N. staudtii in tropical west Africa. No apparent morphological synapomorphies, but a Spondias -type endocarp is unique to Nothospondias. 11. Picrolemma Hook.f. Picrolemma Hook.f., Gen. Pl. (Bentham & Hooker f. ) i. 312 (1862); Cronquist, Brittonia 5:128 147 (1944). Small shrubs, up to 6 m ; dioecious. Leaves imparipinnate; leaflets opposite to sometimes alternate below, petiolulate, en tire, glabrous, punctate glands associated with secondary venation underneath Flowers in terminal determinate thyrses, with multiple glabrous axes; sepals (4)5, basally fused; petals (4)5, imbricate, glabrous; stamens 5, filaments lacking appendage, anthers dorsifixed, staminodes alternating with pe tals in staminate flowers ; disc fleshy, glabrous; carpels (4)5, free, stylodia free but cohering, stigmatic branches fleshy, club-like Fruit 1 drupaceous mericarps, ellipsoid and slightly elongate, not carinate 20 mm long, exocarp brown to red at maturity, pericarp fleshy. Two species from Peru and Brazil.

PAGE 36

36 12. Quassia L. Quassia L., Sp. Pl. ed. 2, 1:553 (1762); Engler, Na t. Pflanzenfam. (Engler & Prantl) 19a:377 379 (1931). Shrub or small tree to 8 m; bi sexual. Leaves im paripinnate, rachis and petiole conspicuously winged in Q. amara, narrowly winged or wingless in Q. africana, articulated ; leaflets opposite, sessile, entire, glabrous, with punctate glands towards leaf apex ad axially. Flowers in axillary or terminal determinate thyrses, appearing raceme-like in Q. amara, with puberulent axes; sepals 5, free, overlapping at base; petals 5, contorted glabrous or basally pubes cent inside, cohering into a tube in Q. amara ; stamens 10, filaments with basal appe ndage, anthers dorsifixed, staminodes absent in staminate flowers; disc fleshy, narrowing towards ba se, glabrous; carpels 5, free, stylodia fused, stigma capitate or slightly lobed. Fruit 1 drupaceous mericarps, obovoid to ellipsoid, bicarinate, 10mm l ong, exocarp dark red at maturity, pericarp fleshy. Two species, 1 neotropical, 1 in tropical west Africa. 13. Samadera Gaertn. Samadera Gaertn., Fruct. Sem. Pl. 2:352 (1791). Sm all tree, occasionally up to 20 m; bisexual. Leaves unifoliolate entire, glabrous, with scattered punctate glands. Flowers in axillary or terminal pseudo-umbels axes glabrous to puberulent, or clustered in leaf axils; sepals (3 )4(), free or mostly fused with short lobes, occasionally with a concave gland; petals (3)4(), imbricate or contorted, glabrous to pubescent abaxially; stamens 8 or 10, filaments with basal appendage, anther s dorsifixed, staminodes absent in staminate flowers; disc fleshy, conical or cylindrical, glabrous; carpels 4, stylodia fused, stigma capitate or slightly lobed. Fruit 1() drupaceous mericarps, ovoid, ellipsoid or semicircular and flattened, slightly to strongl y carinate, 5 mm long, exocarp or ange to red or brown at

PAGE 37

37 maturity, pericarp fleshy or dry and woody. Five to six specie s from Madagascar, Indo-China, SE Asia and Australia. 14. Eurycoma Jack Eurycoma Jack, Malay. Misc. ii. 7:44 (1822); Noot eboom, Fl. Males., Ser. 1, Sperma t. 6:193 226 (1962b). Small trees to 10 m, or rarely shrubs; monoecious or dioecious. Leaves imparipinnate; leaflets opposite to subopposite, sessile or nearly so, some times appearing articulated, entire, glabrous, without glands. Flowers in ax illary determinate thyrses, multiple axes with thick, capitateglandular hairs ; sepals 5(), basally fused, with capitate-glandular hairs ; petals 5(6), induplicate-valvate pubescent, with capitate-glandular hairs ; stamens 5(), filaments with very small appendage near base, anthers dorsifixed, 5() staminodes alternating with stamens in staminate flowers ; disc inconspicuous; carpe ls 5(), free, stylodia connate or cohering, stigma lobed, peltate. Fruits 1 nut-like mericarp s, ovoid, bicarinate, 10 mm long, exocarp brown at maturity, pericarp dry. Three species, tropical SE Asia, Sumatr a, Malay peninsula, Borneo, S. Philippines. 15. Gymnostemon Aubrv. & Pellegr. Gymnostemon Aubrv. & Pellegr., Bull. Soc. Bot. France, 84:181 (1937). Large tree; bisexual or polygamous. Leaves im paripinnate; leaflets opposite to subopposite, subsessile, entire, glab rous, with punctate glands regularly spaced towards apex adaxially. Flowers in axillary or terminal determinate t hyrses, with multiple densely short-hairy axes; sepals 5, fused with short lobes; petals 5, s lightly imbricate, vill ous; stamens 10, filaments lacking appendage, anthers dorsifi xed, staminodes absent in st aminate flowers; disc fleshy, pubescent; carpels 5, free, stylodia fused, stigma si mple or slightly lobed. Fruit typically a single

PAGE 38

38 drupaceous mericarp, ovoid, up to 100 mm long, pericarp fleshy, fibrous. One species, G. zaizou, endemic to Cote DIvoire. 16. Perriera Courchet Perriera Courchet, Bull. Soc. Bot. France 52:284 ( 1905); Perrier de la Bathie, Fl. Madagasc. 105:1 (1950). Tree to 30 m; typically bisexual. Leaves im paripinnate; leaflets opposite to subopposite, subsessile, entire, pubescent when young, becoming glabrous, with punctate glands regularly spaced towards apex adaxially. Flowers in axillary determinate thyrses, with multiple pubescent axes; sepals 5, basally fused; petals 5, induplicate-valvate slightly villous; stamens 10, filaments lacking appendage, anthers dorsifi xed, staminodes absent in stam inate flowers; disc fleshy; carpels 2 slightly united at base, stylodia fused, stig matic branches divergen t. Fruit typically a single drupaceous mericarp, ovoid, up to 50 mm long, exocarp pale yellow at maturity, pericarp fleshy. One or two species endemic to Madagascar. 17. Hannoa Planch. Hannoa Planch., London J. Bot. 5:566 (1846). Trees to 50 m or shrubs, sometim es suffrutescen t; typically bisexual. Leaves imparipinnate; leaflets opposite to alternate, s ubsessile to petiolulate, entire, glabrous, with punctate glands on upper surface, more so towards margins. Flowers in terminal or occasionally axillary determinate thyrses, with multiple glabrous to sparsely pubescent axes; sepals 5, or often calyx rupturing into 2 irregular lobes; petals 5, imbricate, puberulent to densely tomentose; stamens 10, filaments with appendage, anthers dorsifixed, staminodes absent in staminate flowers; disc fleshy, sometimes with gynoecium sunken within, glabrous ; carpels 5, free, styl odia fused, stigmatic branches short, spindly lobes. Fruit 1 drupaceous mericarps, ellipsoid or ovoid, slightly

PAGE 39

39 bicarinate, 15mm long, exocarp red to purplish brown at matur ity, pericarp fleshy. Five to seven species in tropical Africa. 18. Odyendea (Pierre) Engl. Odyendea (Pierre) Engler, Nat. Pflanzenfam. (E ngler & Prantl) III 4:21 5 (1896); Aubrv. & Pellegr. Fl. Gabon 3:33 (1962). Tree to 30m; bisexual. Leaves imparipinnate; le aflets opposite to subopposite petiolulate, entire, glabrous, with punctate glands on upper surface, mo re so towards margins. Flowers in terminal or axillary determinate thyrses, with multiple glabrous axes; sepals 4 (), calyx cupular with short or absent lobes; petals 4(), imbricate, puberulent adax ially; stamens 8(), filaments with densely hairy appendage, anthers dorsifixed, staminodes absent in st aminate flowers; disc fleshy, subcylindrical, with gynoecium slightly immersed within, glabrous; carpels 4 free or united at base, stylodia fused, stigmatic branches very short, divergent. Fruit a single drupaceous mericarp, obovoid to ellipsoid, strongly carinate, up to 70 mm long exocarp red at maturity, pericarp fleshy. One species, O. gabonensis endemic to Gabon and Cameroon. 19. Iridosma Aubrv. & Pellegr. Iridosma Aubrv. & Pellegr., Fl. Gabon 3:47 (1962). Tree; bisexual. Leaves imparipinnate; leaflets opposite to subopposite, subsessile, entire, glabrous, g lands unknown. Flowers in determinate thyrses, with multiple pubescent axes; calyx cupular, irregularly undulating; petals (7)8, valvate villous; stamens 12 filaments with appendage, anthers dorsifixed, staminodes absent in staminate flowers; disc fleshy, pubescent; carpels 4 free, stylodia spirally twisted to form single column, stigma peltate, stellate. Fruit unknown. One species, I. letestui endemic to Gabon and Cameroon. 20. Pierreodendron A.Chev. Pierreodendron A.Chev., Vg. Utiles Afrique Trop. Fran. 9:257 (1917).

PAGE 40

40 Tree to 15 m; bisexual. Leaves imparipinnate; l eaflets subopposite to alternate, petiolulate, entire, glabrous to sparsely pubescent below, wi thout glands. Flowers in axillary or terminal determinate thyrses, with one or two major axes; sepals 5, calyx c upular with short lobes; petals 5, imbricate or contorted; stamens (10)15() filament appendage short with small free tip, anthers basifixed, staminodes absent in stam inate flowers; disc fleshy, sometimes with gynoecium sunken within, glabrous ; carpels 5, free, stylodia fuse d, stigmatic branches short, divergent, or stigma discoid. Fruits 1 drupaceous mericarps, obl ong-ellipsoid, laterally compressed, 70 mm long, exocarp yellow at matur ity, pericarp fleshy, fibrous. Two species from tropical Africa. 21. Simarouba Aubl. Simarouba Aubl., Hist. Pl. Guiane 2:859 (1775); Cronquist, Bull. Torrey Bot. Club 71:226 (1944). Shrubs and trees to 35 m; dioecious. Leaves paripinnate or imparipinnate; lea flets alternate to occasionally subopposite, petiolulate, entire, glabrous or densely tomentose below, with punctate glands scattered on upper surface, more so to wards apex. Flowers in terminal determinate thyrses, with multiple glabrous axes; sepals 5, basally fused; petals 5, imbricate or contorted, glabrous; stamens 10, filaments with glabrous to pubescent appendage, anthers dorsifixed, staminodes absent in staminate fl owers; disc fleshy, short, glabrous to pubescent; carpels 5, free or weakly united, stylodia fused below, stigmatic branches stellately sp reading, recurved. Fruit 1 drupaceous mericarps, ovoid or ellipsoid, slightly flattened, bicarinate, 10 mm long, exocarp orange-red to black at maturity, peri carp fleshy. Six species in Central and South America, the Caribbean is lands and south Florida.

PAGE 41

41 22. Simaba Aubl. Simaba Aubl., Hist. Pl. Guiane. 1:409 (1775) ; Cronquist, Ll oydia 7:81 (1944); Thomas, Brittonia 36:244 (1984). Trees to 30m, shrubs, rarely suffructescent with all leaves basal; bisexual. Leaves paripinnate or imparipinnate, trifoliolate or rarely unifoliolate; leaflets usually opposite, petiolulate to sessile, entire, glabrous to occasionally pubescent, with punctate glands on upper surface, and occasionally with conspicuous apical gland. Flowers in terminal or axillary determinate thyrses, with multiple glabrous to densely pubescent axes or occasionally reduced to axillary clusters; sepals (4)5, basally fused; petals (4)5, imbricate, puberulent to densely pubescent; stamens (8 )10, filaments with appendage, degree of fusion between filament and appendage variable, anthers dorsifixed, staminodes absent in staminat e flowers; disc fleshy, cylindrical, glabrous to densely pubescent; carpels (4)5, free or weakly united, stylodia fused, stigma capitate or slightly lobed. Fruit 1() drupaceous mericarps, ellipsoid to obovoid, lentic ular, slighty carinate or occasionally strongly winged, 10() mm l ong, exocarp orange, red, brown, black or yellow at maturity, pericarp fleshy. 25 species in tropical South America with one species S. cedron extending into Central America.

PAGE 42

42 CHAPTER 3 MOLECULAR PHYLOGENY OF SIMAROUBACEAE BASED ON CHLOROPLAST AND NUCLEAR MARKERS1 Introduction Simaroubaceae s.s. are a small yet morphologically dive rse angiosperm family of tropical and temperate trees and shrubs in the Sapi ndales. In the traditional circumscription, Simaroubaceae s.l. comprised six subfamilies (Engler, 1931) and encompassed considerable diversity in secondary chemistry and macrom orphology (Cronquist, 1981). However, molecular work by Fernando et al. (1995; also see Gadek et al., 1996) showed the family to be polyphyletic, with subfamilies orig inating in several places within eurosids I and II ( sensu APG II, 2003). Members of subfamily Simarouboideae, re presented in Fernando et al. (1995) by eight taxa, form a well-supported monophyletic group (excluding Harrisonia ) within the Sapindales. Leitneria a genus traditionally segregated into the monotypic family Leitneriaceae on account of its wind-pollinated flowers (Cronquist, 1981; Takhta jan, 1996), was also found to be part of the Simarouboideae clade. Hence, the subfamily was recircumscribed as Simaroubaceae s.s., a clade of 20 genera and approximately 95 species (F ernando and Quinn, 1995). Two additional genera, Laumoniera and Nothospondias may also need to be included in the family (Fernando and Quinn, 1992; Nooteboom, 1987). Tribal, generic and species relationships within Simaroubaceae s.s. (henceforth referred to as Simar oubaceae), however, remain unresolved. Synapomorphies for the Simaroubaceae are trit erpenoid compounds of the quassinoid type, five carpels united only by their styles and separating in fruit and one ovule per locule (Judd et al., 2002). Floral diversity ranges from wind-pollinated catkin-lik e inflorescences in Leitneria (Cronquist, 1981) to hummingbird-pollinated flowers in Quassia amara (Roubik et al., 1985) to 1 Reproduced with permission from Clayton et al., 2007. International Journal of Plant Sciences 168:1325-1339. 2007 by The University of Chicago. All rights reserved.

PAGE 43

43 insect pollination in the remain ing genera. Habit ranges from large rainforest and temperate forest trees, to forest understory, coastal and desert shrubs. Simaroubaceae comprise three tribes (in Simarouboideae sensu Engler, 1931): Simaroubeae, Picrasmeae and Soulameae (see Table 3-1), delimited by the presence or absence of filament appendages and the degree of fusion of the carpels. The three tribes were divided into seven subtribes based primarily on the nature of the androecium, particularly variation in the number of stamens and presence of stami nodes (Engler, 1931). The only recognized subgeneric classification in the family is the division of the largest genus Simaba into three sections, Tenuiflorae, Floribundae and Grandiflorae, based primarily on fl ower size (Engler, 1874; Cronquist, 1944c ). A recircumscription of Quassia (Nooteboom, 1962 a) included all members of Simaroubeae except Eurycoma but many authors have not a ccepted this expanded view of Quassia (e.g. Feuillett, 1983; Fernando and Quinn, 1992). Laumoniera, a monotypic genus from Sumatra, was described as a close relative of Brucea (Nooteboom, 1987); however, the status of Laumoniera as a new genus in the family has not been investigated further, and thus was omitted from the more recent family circumscri ption (Fernando and Quinn, 1995). Based on the morphological description (Nooteboom, 1987), we consider Laumoniera a distinct member of the family. Nothospondias is a monotypic genus originally described in the Anacardiaceae (Engler, 1905) and has a Spondias-type endocarp (Fernando and Qu inn, 1992). However, it was referred to Simaroubaceae by Van der Veken (1960) based on morphological characters, and its phylogenetic placement is yet to be confirmed. Simaroubaceae are well supported as a memb er of a Simaroubaceae+Rutaceae+Meliaceae clade in Sapindales (Gadek et al., 1996; Kllersj et al., 1998; Savolainen et al., 2000 b; Soltis et al., 2000), but the familys sister group is stil l undetermined, with da ta supporting all three

PAGE 44

44 alternative topologies Rutaceae sister to Si maroubaceae (Gadek et al., 1996; Cronquist, 1981); Meliaceae sister to Simaroubaceae (Chase et al., 1999; Muellner et al., 2006); Rutaceae sister to Meliaceae (Stevens, 2001). Simaroubaceae have a primarily pantropical distribution (Table 3-1); however, Leitneria, Castela Holacantha, Ailanthus, Picrasma and Brucea have subtropical and temperate members. Ailanthus has an extensive fossil record dating from the early Eocene, across the entire Northern Hemisphere (Corbett and Manchester, 2004), Leitneria has fossil fruits dating from the Oligocene of western Siberia (D orofeev, 1994; Nikitin, 2006a) and Chaneya an extinct genus from the tertiary of North America, Europe and Eastern Asia (Wa ng and Manchester, 2000; Teodoridis and Kvacek, 2005), is suggested to have an affiliation with fruits of the extant genus Picrasma The goal of this study is to reconstruc t phylogenetic relationships within the Simaroubaceae, using broad taxon sampling and mu ltiple genetic markers. Sequences of the chloroplast genes rbcL, atpB and matK have repeatedly been of utility in familyand genus-level phylogeny reconstruction (e.g. Hoot et al., 1995; Chase et al., 1999; Muellner et al., 2003; Kathriarachchi et al., 2005; Wilson, 2005). Mo re recently, the low-copy nuclear gene phyC has also proven useful in phylogenetic reconstruction at this level (M athews et al., 1995 ; Davis et al. 2002; Kathriarachchi et al., 2005; Saarela et al., 2007), provi ded homologous copies are analyzed. Previous molecular phylogenetic studies of the family (F ernando et al., 1995; Gadek et al., 1996) were based only on rbcL sequence data, which failed to resolve relationships among genera, except to show that Ailanthus was sister to the rest of the family. Furthermore, sampling was restricted to just seven of the 22 genera, each represented by a single species. We therefore conducted a phylogenetic study employing nearly co mplete generic sampling and sequence data

PAGE 45

45 from four genes, to elucidate relationships in the family. The resulting phylogeny is used to examine generic limits in the family, with particular reference to Nootebooms (1962 a) controversial broad r ecircumscription of Quassia. The position of Nothospondias, a genus that has been affiliated with both Simaroubaceae a nd Anacardiaceae, is also examined using a broader sample of Sapindales. In addition, the effects of five differe nt data partitioning strategies in Bayesian analyses focused on Simaroubaceae are explored using Bayes factors, likelihood scores and resulting topo logies and clade support. Methods Taxon Sampling We conducted a broad analysis of Sapindales to ascertain the familial placement of the monotypic genus Nothospondias. Sequences of rbcL and atpB for 64 Sapindales and three Malvales outgroups were assembled from Genbank as well as from this study, including 36 Simaroubaceae and three Anacardiaceae taxa. For the focused study of Simaroubaceae, 19 of the 20 genera sensu Fernando and Quinn (1995) were sampled, plus Nothospondias for a total of 67 ingroup accessions encompassing 58 species (Appendix 1). Material wa s taken from herbarium specimens obtained from E, MO, NY, AAU and CAY, from silica-dried leaf material from the Royal Botanic Gardens, Edinburgh, Singapore Botanic Gardens, National Botanic Garden of Belgium, MOBOT DNA Bank and from wild plants in Florida, northern Australi a and China. Extracted DNA was obtained from the Kew DNA Bank and previous studies of Simaroubaceae (Fernando et al., 1995). We were unable to obtain material of Laumoniera (omitted from Fernando and Quinns (1995) recircumscription, but considered a member of the Simaroubaceae by Nooteboom (1987)) and Iridosma (a monotypic genus from Gabon, included in Fernan do and Quinn (1995)). Three species each of Meliaceae and Rutaceae (Appendix 1) were used as outgroups based on Fernando et al. (1995)

PAGE 46

46 and Gadek et al. (1996). In addition, a more distant outgroup from Sapindaceae ( Acer ) was included. Outgroup sequences were determined for this study or obtained from GenBank. DNA Extraction, Amplific ation and Sequencing Total genomic DNA was extracted from silica-dr ied and herbarium leaf material using CTAB extraction (Doyle and Doyle, 1990) or DNeas y Plant Mini Kits (Qiagen Inc., Valencia, CA), with the addition of Proteinase K for probl ematic accessions. Extractions that were difficult to amplify were cleaned using Promega Wizard Cleanup kits. Three plastid loci, rbcL, atpB and matK with ~500 additional bp of trnK intron on either side of the matK exon, and one nuclear locus, ~1000 bp of phyC starting ~550 bp downstream in exon 1, were amplified and sequenced. Plastid loci were amplified using the primer s shown in Table 3-2. Amplification of DNA was conducted with a Biometra T3 Thermocycler (Biometra, Gttingen, Germany) or an Eppendorf Mastercycler (Brinkmann Inc., Westbury, NY) using PCR reactions in 25 l volumes, containing 2.5 units of Taq polymerase, 0.5 M of each primer, 0.1 mM of each dNTP in an equimolar ratio, 10X buffer containing 1.5 mM MgCl2, 10 ng of genomic DNA, 1M Betaine and dH2O. PCR cycling conditions for chloroplast loci were as follows: 1) an initial heating step at 95C for 5 min, 2) 94C for 1 min, 3) the initial annealing temperature was 58 C for one minute and 4) elongation at 72 C for 2.5 min. Steps 2 were repeated for 6 cycles, except for dropping the annealing temperature by 1 C in each of the si x cycles until 52 C, and then the annealing temperature was maintained at 52 C and steps 2 were repe ated for a total of 34 cycles. PCR products were cleaned using ExoS ap or Promega Wizard Cleanup kits. Weak PCR products were combined and concentrated using the Promega Wizard Cleanup kit. DNA sequencing was performed on a Beckman-Coulter CEQ 8000 Automated Sequencer with DTCS chemistry (Beckman Coulter, Fullerton, CA) or an ABI 3730xl DNA Analyzer

PAGE 47

47 (Applied Biosystems, Foster City, CA) with th e T7 primer and BigDye Terminator Cycle Sequencing chemistry, using amplification primers plus additional sequencing primers (Table 32). Sequencherv4.2 (Gene Codes Corp., Ann Arbor, MI) was used to assemble complementary sequences, and sequences were depo sited in GenBank (see Appendix 1). Partial phyC sequence was amplified using primers designed from GenBank sequences (see Table 3-2). The reaction mixture was as for chloroplast loci, using the following PCR conditions: 1) an initial heating step at 95 C for 5 min, 2) 94C for 1 min, 3) annealing at 53C for 1 min and 4) elongation at 72 C for 2.5 min. Steps 2 were repeated for 5 cycles, except for dropping the annealing temperature by 1 C in each of the five cycles until 48 C, and then the annealing temperature was maintained at 48 C and steps 2 were repe ated for a total of 35 cycles. PCR products were sequenced as above. Sequences obtained for three Ailanthus accessions and one Soulamea accession showed overlapping peaks (we were unable to sequence phyC for two Ailanthus species). These PCR products were therefore cloned using TOPO TA Cloning Kits (Invitrogen Corp., Carlsbad, CA, USA) PCR products of species from seven other genera were also cloned to confirm their status as single copies Sixteen clones per sample were chosen (to ensure multiple phyC copies were sequenced if present) from colonies grown on agar plates containing kanamycin, and amplified an d sequenced using M13F and M13R bacterial primers on an ABI 3730xl DNA Analyzer. Alignment and Indel Coding Nucleotide alignments for each locus were assembled using C lustal X v1.83 (Thompson et al., 1997). Alignment of coding regions in the analysis focused on Simaroubaceae was straightforward, with few manual adjustments; however, the trnK intron had areas of ambiguity caused by large insertions in individual sp ecies. A total of 332 ambiguous positions was

PAGE 48

48 excluded from phylogenetic analyses, 273 bp of which were large insertions in up to 12 species. Difficulties were encountered in assessing homol ogy of the majority of indels within the trnK intron and thus coding them, due to ambiguity in the alignment. Indel coding was therefore not used in further analyses; however, easily code d indel synapomorphies are noted in the Results. Alignment of rbcL and atpB in the broader Sapindales analys is was also straightforward, and no indels were present. Maximum Parsimony Analyses and Data Congruence Gene partitions for the focused study of Sima roubaceae were analyzed separately to identify areas of incongruence between genes and between plastid and nucl ear data (Huelsenbeck et al., 1996; Johnson and Soltis, 1998; Barker an d Lutzoni, 2002; Hipp et al., 2004). Maximum parsimony (MP) analyses were conducted us ing PAUP* 4.0b10 (Swofford, 2002). MP analyses for each partition used heuristic searches of 1,000 random addition replicates, with TBR branch swapping, MulTrees in effect, and saving all tr ees. Bootstrap support (Felsenstein, 1985) was estimated from 1000 bootstrap replicates using 10 random additions per replicate, with TBR branch swapping, MulTrees in effect, and saving all trees. A parsimony analysis of phyC clones revealed two copy types in Ailanthus; all other species from eight genera (including Soulamea ) that were cloned had a single copy. The placement of one Ailanthus copy was congruent with the plas tid phylogenies, and therefore a single clone from this copy wa s selected at random for each Ailanthus species, to be included in further phylogenetic analyses. The second Ailanthus copy was found to be sister to a clade of Leitneria+Soulamea+Amaroria+Brucea (i.e., the equivalent of clade III in figures 1 and 2) in the phylogeny of clones. Its inc ongruent position suggests it is a paralog resulting from a duplication event in phyC and is under different evolutionary constraints than the first copy. Thus it was removed from further analyses.

PAGE 49

49 Congruence between all data partitions, and between combined plastid and nuclear partitions, was assessed with the ILD test (Fa rris et al., 1994), implemented in PAUP as the partition homogeneity test, using 100 replicates and search stra tegies as for MP analyses. However, ILD results must be interpreted with cauti on as the test is sensitive, particularly in data sets with rate heterogeneity be tween partitions (Darlu and Lecointre, 2002). Therefore, Wilcoxon sum of rank tests, which take into account relative levels of clade support when assessing congruence, were also implemented following Ze rega et al. (2005). Comparisons were made between all pairs of individual gene partitions and between combined plastid and nuclear ( phyC ) data, for bootstrap cons traint trees at 70%, 80% and 90% thresholds. For the focused study of Simaroubaceae, combined analysis of all data was conducted as described above for the separate analyses, except with bootstrap s upport assessed using 10,000 bootstrap replicates. For the broader an alysis of Sapindales, the combined rbcL and atpB data were analyzed as above, using 1000 boot strap replicates to assess support. Bayesian Analyses and Partitioning Strategies A Bayesian analysis of the combined data set focused on Sima roubaceae was performed, with the data partitioned by gene (with the trnK intron, excluding the matK exon, a single separate partition) and by codon position for the c oding regions, totalling 13 separate partitions. Models of nucleotide substitution for each pa rtition were determined using Modeltest v3.6 (Posada and Crandall, 1998). The Akaike informa tion criterion (AIC) was used to selected an appropriate model, based on the relative informational distance between the ranked models in the Modeltest output. In all cases there were no to pological differences am ong the five highest ranked models, therefore the highest ranked model was chosen for each partition, as follows: rbcL codon position 1(p1) GTR+I+G, p2 F81+I, p3 GTR+G; atpB p1 TrN+G, p2 TrN+I+G, p3 K81uf+G; matK exon p1 TVM+G, p2 GTR+G, p3 TVM+G; phyC p1

PAGE 50

50 HKY+G, p2 TVM+G, p3 TVM+G; trnK intron TIM+I+G. Analyses were implemented in MrBayes v3.1.1 (Huelsenbeck and Ronquist, 2001; Ronquist and Hu elsenbeck, 2003). Two independent analyses each ran for 5 million gene rations using four Markov chains and with all other parameters at default valu es; trees were sampled every 1000th generation with a burn-in of 100,000 generations; model parameters were unlinked across partitions. Stationarity of the MCMC was determined by the average standard deviation of split frequencies between runs and by examination of the distribution of the poste rior in Tracer v1.3 (Rambaut and Drummond, 2003). Majority rule consensus trees were created in PAUP using the resulting posterior distribution of trees. For the broader analysis of the Sapindales data were partitio ned by gene and codon position, with the following substitution models determined using Modeltest: rbcL codon position 1(p1) GTR+I+G, p2 F81+I+G, p3 GTR+G; atpB p1 GTR+G, p2 K81uf+I+G, p3 TVM+G. Bayesian analyses were conducted fo llowing the methods desc ribed above, with a burn-in of 100,000 generations. For the focused study of Simaroubaceae, concerns about model specification and overparameterization of the data (Rannala, 2002; Le mmon and Moriarty, 2004; Nylander et al., 2004; Sullivan and Joyce, 2005; Kelchner and Thomas, 2007) were addressed by running five separate partitioning strategies: 1) no partitioning of the data (P1); 2) one chloroplast partition and one nuclear partition (P2); 3) partitioning by codon position only plus a separate intron partition (P4); 4) partitioning by gene and intron only (P5); 5) full partitioning by gene and codon position plus intron (P13). Number of data partitions for each partitioning strategy (P ) is indicated by a numerical subscript. Par titions were decided upon a priori under the assumption that different genes and the three codon positions may evolve at different rates, both within and between

PAGE 51

51 plastid and nuclear genomes. Competing partitioning strategies were assessed with Bayes factors, comparison of the distri bution of likelihood sc ores and resulting posterior distributions of trees, following the methods described in Brandley et al. (2005). Harmonic means of likelihoods were estimated using the sump command in MrBayes. Br andley et al. (2005) concluded that this method was adequate when compared to manual calculation of harmonic means from the posterior distri bution of likelihoods. Likelihood scores for their partitioning strategies, and the differences between them, are co mparable in magnitude to likelihood scores in this study. Bayes factors were calculated as the ratio of the harmonic means of the two competing hypotheses (partitioning strategies). The traditional cutoff for the support of a competing hypothesis over the null h ypothesis is 2ln Bayes factor > 10 (Kass and Raftery, 1995), indicating strong support for the alternative hypothes is, so this cutoff was used here. Arithmetic means and 95% credibility intervals of likeli hood scores were calculated and compared among partitioning strategies, and posterior probabilitie s of clades were assessed across the five topologies generated by the diffe rent partitioning strategies. Results Data Congruence Initial observation of separate and combined MP tree topologies and Bayesian topologies for the focused study of Sima roubaceae indicat ed all incongruence was between poorly supported clades, due to short branches or conf licting signal, and no ha rd incongruence existed among partitions. ILD tests (Fa rris et al., 1994) showed signifi cant incongruence among all four loci (p = 0.01), and between plastid and nuclear data (p = 0.01), but not among the three plastid loci (p=0.12). Results of th e Wilcoxon sum of rank tests (Table 3-3) showed no significant incongruence between any pairs of data partitions, except when the phyC tree search was constrained by the 70% atpB bootstrap consensus. However, significant incongruence was

PAGE 52

52 observed between combined plastid and phyC data with all constraints, except when the phyC tree search was constrained by the 90% plastid bootstrap consensus. Inspection of the plastid and phyC strict consensus trees suggests the majo r sources of conflict are the positions of Nothospondias and Quassia in the phylogeny. In the phyC tree, Nothospondias appears in a polytomy with Picrolemma and a clade of Samadera + Quassia +clade V, whereas in the plastid tree Nothospondias is found in the same position as in the combined analysis (see figure 1). Quassia is sister to clade V in the phyC tree, whereas in the plastid tree Quassia is sister to a clade of Samadera and clade V, as in the Bayesian topology. Phylogenetic Analyses and Tree Topology Results of MP analyses of i ndividual partitions and combin ed data for th e family are provided (Table 3-4). MP analysis of the combined data set yielded eight most-parsimonious trees (figure 1). Results of the fully partitioned Bayesian analysis yielde d a tree topology that was nearly identical to the MP strict consensu s (figure 2), with the exception of the position of the Quassia clade, which diverges before Samadera in the Bayesian analysis and after Samadera in the MP strict consensus. Neither position re ceived high posterior probability (PP = 0.91) or bootstrap support (BS = 62%). Most relationships within Simaroubaceae have high PP and BS values in the combined MP analysis and fully partitioned Bayesian analysis. An early-diverging clade of Picrasma+Holacantha+Castela (clade I, figure 1 and 2) has low PP (0.74) and BS support (78%). A clade of Leitneria+Soulamea+Amaroria+Brucea (clade III) has high PP (1.0) and BS support (100%). Soulamea is rendered paraphyletic by the inclusion of Amaroria which is sister to S. amara+S. terminalioides (PP = 0.99; BS = 84%). The position of Nothospondias staudtii as diverging after clade III has low BS support (<50%) in the MP analysis, but high PP (0.98) in the fully partitioned Bayesian analysis. Relationshi ps among the four major lineages of clade V

PAGE 53

53 (figures 1 and 2) have low PP (< 0.50) and BS (< 50%) values. Within clade V, there is high PP (1.0) for a clade of Hannoa, Gymnostemon, Perriera, Odyendea zimmermannii and Brucea tenuifolia but low BS support (58%). A clade of Simaba, Simarouba and Pierreodendron has high PP (1.0) but BS support of only 71%. The monophyly of Simarouba has high PP (1.0) and BS (100%) values, but low support values for Pierreodendron as sister (PP = 0.69; BS = 69%). The monophyly of Simaba has low a PP (0.75) and BS (53 %) value, but consists of two subclades with high support values ( PP = 1.0, BS = 99% and PP = 1.0, BS = 100%, respectively). Indels within the matK coding region were all specific to closely related species: indels support the monophyly of Picrasma Samadera (excluding S. indica ), Samadera bidwillii+S. spB, and a clade of Simaba cedron+S. ferruginia+S. insignis respectively. Unambiguous trnK intron indels support a clade of Picrasma+Holacantha+Castela, a clade of Simaba cedron+S. ferruginia+S. insignis and the monophyly of Eurycoma and Picrasma respectively. For the other easily coded trnK indels, there was an obvious tendency towards homoplasy. In both MP and Bayesian analyses of the broader Sapindales data set, a clade of Simaroubaceae has high BS (100%) and PP (1.0) values; there is high PP (0.98), but low BS support (65%) for Meliaceae as sister to Simaroubaceae. Nothospondias was found to be nested within the Simaroubaceae clade in the same position as in the focused Bayesian analysis (trees not shown). Partitioning Strategies Absolute differences in 2l n Bayes factors between partitioning strategies (Table 3-5) showed that the most complex model (P13) was a much better explanation of the data than other partitioning strategies. This was also evident in the mean ln L of P13, which was over 400 likelihood units better than the next best partitioning strategy, P5. Partitioning by gene (P5) and

PAGE 54

54 by codon position (P4) showed very similar mean likelihood scores, but partitioning by gene was determined to be a significantly better model than codon partitioning (2 ln Bayes factor = 14.60). No partitioning of the data re sulted in a lower mean ln L than the next most complex partitioning strategy (P2) and was decisively rejected as the best method of data partitioning according to all Bayes factors. Topologies generated from the di fferent partitioning strategies were identical. Differences in posterior probabilities between competing topologies were only apparent in weakly supported nodes (see Table 3-5 for examples). A general trend of increasing clade support with more complex partitioning was only observed in the positions of Nothospondias (node B) and Quassia (node C). Support for the sister relationship of Pierreodendron and Simarouba (node D) showed an inverse relationship of partitioning complexity to clade support, as did support for Simaba (node E). Discussion Data Congruence and Partitioning Strategies In the focused study of Simaroubaceae, soft incongruence b etween the four loci was relatively rare and was typically due to a paucity of informative characters in rbcL and atpB both of which made it difficult to resolve finer-scale relationships at the species level and therefore yielded high numbers of shortest trees for these two genes (Table 3-4). This problem was particularly evident in clade V, with several branches having a single s ubstitution or a length of zero. Sequences of matK including a portion of the trnK intron, provided much better resolution than either rbcL or atpB due to the high number of parsimony-informative characters (Table 3-4). PhyC also proved highly informative, particularly in resolving relationships within clade V.

PAGE 55

55 There were no instances of hard incongruence observed between the trees of each partition, so despite the results of the ILD test and Wilc oxon sum of rank tests for nuclear versus plastid data, a combined approach was favored, due to the synergistic effects of combining data (Olmstead and Sweere, 1994; Huelsenbeck et al., 1996; Sullivan, 1996; Soltis et al., 1998). This approach is particularly effec tive with the broad level of taxon sampling employed in this study (Lecointre et al., 1993; Graybeal, 1998; Hedtke et al., 2006) and th e range of evolutionary rates across the data sets being comb ined (Olmstead and Sweere, 1994; Sullivan, 1996). In particular, combining data resolved and greatly improved support for several finer-scale groupings that were not observed or poorly supported in the trees from separate data partitions (e.g. the Simaba+Simarouba+Pierreodendron clade and the position of Perriera + Gymnostemon as sister to Hannoa). The incongruent position of Quassia between parsimony and Bayesian analyses is likely due to the long-branch to Quassia in the parsimony analysis (see Bergsten, 2005). Analyses of different partit ioning strategies tested the relative importance of model specification and concerns about correct model se lection in a complex multigene data set. Overparameterization of the data, by using a fully pa rtitioned, complex substitution model, was not evident, as the most complex part itioning strategy proved to be the best choice for this data set, as in other studies (Nylander et al., 2004; Brandley et al., 2005) This result, however, should be viewed with caution, given that little is known about how tree topology and clade support are affected by the random error associated with in creasing number and decreasing size of partitions (Rannala, 2002; Cummings et al., 2003; Lemmon and Moriar ty, 2004; Nylander et al., 2004; Brandley et al., 2005; Lewis et al., 2005; Sullivan and Joyce, 2005; Kelchner and Thomas, 2007). The increases in clade support with decreasing model complexity seen in the position of Pierreodendron (node D) and support for the monophyly of Simaba (node E; Table 3-5)

PAGE 56

56 exemplify the difficulties in assessing model perf ormance, and the problems of nonidentifiability and random error in measures of clade suppor t for weakly supported nodes (Rannala, 2002; Sullivan and Joyce, 2005). This is especially true for short internodes (Alfaro et al., 2003; Erixon et al., 2003; Brandley et al., 2005; Lewis et al., 2005) such as those observed at nodes D and E (figure 2), because simpler models can produce inflated estimates of clade support (Suzuki et al., 2002; Lemmon and Moriarty, 2004; Nylander et al., 2004; Brandley et al., 2005). Nylander et al. (2004) noted the strong influenc e of correctly modelli ng withinas well as between-partition rate vari ation, a factor that was not explored in this st udy, other than assuming that the Akaike information criterion (Akaike, 1974) determined the substitution model best suited to each data partition (Posada and Buckley, 2004). Nyla nder et al. (2004) and Lemmon and Moriarty (2004) found that failure to model among-site rate heteroge neity correctly had the greatest influence on likelihood scores, and caused the largest decrease in likelihood values when rate heterogeneity was not accounted for. This probl em is unlikely to have affected our results, as all partition models in this study (except for rbcL codon position 2) estimated a gamma shape parameter and therefore incorpor ated among-site rate heterogeneity. However, base frequencies and transition/transversion rates may have been more complex than necessary, and thus the within-partition models could have used ex cessive parameters (Nylander et al., 2004). Despite the concerns reviewed above, topologi es generated from the different partitioning strategies were identical with just a few differences in s upport values for weakly supported clades, suggesting that the unde rlying phylogenetic signal is r obust and resistant to model misspecification (Kelchner and Thomas, 2007). The superiority of parameterizing by gene over codon position (Table 3-5) may be due to the rang e of rates of evolution across the four genes used; for example, the PHY gene family is reported to ev olve about 10 times faster than rbcL

PAGE 57

57 (Mathews et al., 1995). This variation in evolutio nary rates is therefore more accurately modelled by gene partitioning, rather than by combining first, second and third codon positions across genes that evolve differently. The differences in posterior probabilities of weakly supported nodes among all five topologies are most likely due to a lack of informative characters, borne out by the short branches of these nodes (figure 2). Resolution of and support for these nodes will likely improve, not with a better-fitting model, but with the addition of more sequence data, and in some instances more taxa. Systematics of the Simaroubaceae Molecular data have clarified evolutionary relationships among the genera and many species of th e Simaroubaceae. Our discussion of relationships will focus on results of the combined, fully partitioned Bayesian analysis of the focused study (figure 2). The inclusion of Picrasma Castela and Holacantha has revealed that Ailanthus is not sister to the rest of the family, as previously suggested in earlier st udies (Fernando et al., 1995; Gadek et al., 1996). Instead, Ailanthus diverges after the Picrasma+Castela+Holacantha clade. Englers (1931) tribes Picrasmeae and Simaroubeae are not monophyletic; however, Soulameae (consisting of Soulamea and Amaroria ) has high PP (1.0) and BS (100%) support. Although the clade of Picrasma+Holacantha+Castela has low PP and BS values in the combined analysis, it does appear in the MP strict consensus of most partitions and the combined analysis. A single 6-bp gap in the 5 trnK intron also supports th is grouping. Cronquist (1944 a) considered leafless Holacantha a specialized offshoot of Castela an observation supported here by the early divergence of this ditypic genus from Castela Cronquist (1944a,d) also considered Picrasma and Castela to be closest relatives in th e Simaroubaceae, based on leaflet morphology and having similar drup aceous fruits. Biogeographically, Picrasma Castela and

PAGE 58

58 Holacantha are primarily New World, with just P. quassioides and P. javanica found in mainland and southeast Asia. Fernando et al. (1995) confirmed Leitneria as a member of the Simaroubaceae, but its relationship to other genera was unresol ved. The present stud y clearly places Leitneria within clade III (figures 1 and 2) as sister to a clade of Brucea Soulamea and the monotypic Amaroria Morphological synapomorphies for clade III may be difficult to detect due to the dramatic floral divergence of Leitneria compared to other clade members: Leitneria is anemophilous whereas other clade members are insect-pollinated. Leitneria shares simple leaves with a long petiole with several species of Soulamea, and Leitneria has a single carpel, susp ected to be reduced from a bicarpellate gynoecium (Trelease, 1895), the condition found in Soulamea In contrast, Brucea has a four-carpellate gynoecium and pinnately compound leaves However, the androecium of Leitneria is reduced to four stamen s (Judd et al., 2002) as in Brucea Hence, a reduction in numbers of reproductive parts characterizes clade III. Further inves tigation is needed to identify morphological synapomorphies for the clade, which may be in the form of cryptic characters. The placement of the monotypic Amaroria within a paraphyletic Soulamea is in agreement with Nootebooms (1962b) suggestion that Amaroria should not be consid ered separate from Soulamea The genera differ chiefly in carpel number (one in Amaroria ; two in Soulamea ). However, only rbcL and atpB sequence data were available for Amaroria and only three of the 14 species of Soulamea were sampled here, so additional data are needed to confirm this placement of Amaroria Nothospondias is found to belong within the Simaroubaceae based on a broad analysis of Sapindales. In the analyses focused on the Simaroubaceae, Nothospondias diverges after clade III in the Bayesian analysis, but is in an unr esolved position in the parsimony analysis. The

PAGE 59

59 similarity of the endocarp anatomy in fruits of Nothospondias staudtii to the distinctive Spondias -type endocarp found in the Anacardiaceae (Fernando and Quinn, 1992) may therefore be considered similarity due to homoplasy. Quassia, the first genus name published for the Simaroubaceae, by Linnaeus in 1762 (Cronquist, 1944 d), has been ascribed to various member s of the tribe Simaroubeae, notably as a synonym of Samadera Simaba, Hannoa and Odyendea (Pierre, 1896; Nooteboom, 1962 a, b; Hewson, 1985; Mabberley, 1997). Nooteboom (1962 a) considered Quassia Samadera Simarouba, Simaba, Hannoa, Odyendea and Pierreodendron to constitute a broadly defined Quassia s.l. prompted by examination of a new species, Q. borneensis, which had traits in common with both New World Simaba and African genera Hannoa and Odyendea. In his recircumscription of Quassia s.l. Nooteboom (1962a) recognized four sections: sect. Quassia, sect. Samadera, sect. Simarouba and sect. Simaba. Within this sectional classification of Quassia s.l. (Nooteboom, 1962a), sect. Quassia consisted only of Q. amara a New World species with a winged, jointed leaf rachis, a racemelike inflorescence, pedicels articulated in the middl e and erect, glabrous petals (with hairs at the base). However, Q. africana was also observed to have very narrow wings on a jointed leaf rachis (Nooteboom, 1962 a) and glabrous petals (with hairs at the base), although it typically has a mulit-branched inflorescence and spreading petals. Furthermore, the short pedicels of Q. africana can make it difficult to distinguish th e position of articulation (Cronquist, 1944 d). For these reasons Nooteboom (1962 a) and Cronquist (1944 c ) referred Q. africana to sect. Simaba in Quassia s.l. and the genus Simaba, respectively. However, Engl er (1931) considered both Q. amara and Q. africana to constitute the genus Quassia s.s Our data show a well-supported clade of Q. amara and Q. africana (PP = 1.0; BS = 100%). Pubescence of the petals and wings on the

PAGE 60

60 rachis, considered unimportant in generic delimitation by Cronquist (1944 c ) and Nooteboom (1962a), may be synapomorphies for a narrowly defined Quassia. Quassia amara has unique features including a raceme-like inflorescence, pedicel articulati on and erect corolla forming a tube-like flower, which are likely newly acquired traits asso ciated with hummingbird pollination (Roubik et al., 1985). Furthermore, Ba sak (1967) noted pollen grains of Q. amara to be an entirely different type to that of Samadera Simaba and Simarouba. Following Nooteboom (1962 a), sect. Samadera consisted of just two species, Q. indica and Q. harmandiana, ranging from Madagascar to southeas t Asia. The two species are diagnosed as having large simple leaves and pseudo-umbe llate or racemose inflorescences. However, Nooteboom placed two other species, Q. baileyana and Q. bidwillii in sect. Simaba, despite both having simple leaves and Q. baileyana having a stalked pseudo-umbel (Q. bidwillii is described as having flowers in clusters in the axils of the le aves). Nooteboom (1962a) also acknowledged that Q. baileyana although placed in sect. Simaba, forms the connection with sect Samadera as regards the structure of inflorescence and flowers. Quassia baileyana and Q. bidwillii occur in Australia, suggesting a closer association to sect. Samadera (southeast Asia and Madagascar) than to the ot her New World members of sect. Simaba. Molecular data revealed that Q. indica, Q. baileyana and Q. bidwillii ( sensu Nooteboom, 1962a) form a clade with high PP (1.0) and BS (100%) values. This supports their recognition as the separate genus, Samadera (Bennett, 1872; Engler, 1931; Backer a nd Van den Brink, 1965), along with two other Australian species, Samadera sp.B and Samadera sp.C, that are yet to be formally described. In the circumscription of the genus Samadera sensu Bennett (1872), Engler (1931) and Backer and Van den Brink (1965), large simp le leaves and umbellate/pseudoumbellate inflorescences may be synapomorphies. Although Q. harmandiana was not sampled, based on its morphological

PAGE 61

61 similarity to Q. indica (Nooteboom, 1962 a), it is expected to be part of the Samadera clade. Quassia borneensis (for which molecular data were unobt ainable), although found in Malaysia, is described as having similarities to the African members of sect. Simaba, notably Q. gabonensis ( sensu Nooteboom 1962a ). However, a single specimen of Q. borneensis observed for this study has simple leaves, so the position of this species remains unclear. In clade V, no resolution was found among the four subclades. This lack of resolution can be attributed to very short branches, consisting of just one or two nucleotide substitutions supporting the basal branching pattern (figure 2), despite the use of 6 kb of sequence data. One possible explanation for these short branches is that clade V underwent a rapid radiation. More sequence data or more rapidly evolving genes are clearly needed to resolve the branching pattern among the subclades of clade V. The resolution of four subclades in clade V refutes Nootebooms (1962 a) broad circumscription of Quassia, with three well-supported and ge ographically distinct groups, along with a single segregate species, Odyendea gabonensis Eurycoma identified by Nooteboom (1962a) as the only genus in this clad e deserving of recognition outside Quassia s.l. has high PP (1.0) and BS (100%) values and is diagnosed by its induplicate-valvate aestivation and outer whorl of staminodes (Nooteboom 1962 a). Hannoa, Gymnostemon and Perriera also form a clade with a high PP (1.0) in Bayesian analyses, with the inclusion of Odyendea zimmermannii and Brucea tenuifolia The clade does not receive a high BS va lue (58%) in the MP analysis and is subtended by a short internode (figure 2); theref ore, Bayesian analyses may be over-estimating support for this clade (Suzuki et al., 2002; Alfaro et al., 2003; Erixon et al., 2003; Brandley et al., 2005; Lewis et al., 2005). However, the clade is composed only of species found in Africa and Madagascar. Brucea tenuifolia appears as sister to O. zimmermannii and from the limited

PAGE 62

62 herbarium material available for these two sp ecies, leaf morphology between the two is very similar, so a misidentification of B. tenuifolia is suspected. Regardless of this possible misidentification, O. zimmermannii may need to be reassigned to Hannoa. In Nootebooms (1962a) recircumscription of Quassia, H. undulata, H. chlorantha H. klaineana and O. zimmermannii were synonymous under Q. undulata in sect. Simaba, as the characters distinguishing the different species, such as number and shape of leaf lets, length of lateral petiolules, and flower size, were regarded as highly variable and thus not useful. Based on results of our phylogenetic analyses of Simaroubaceae, Nootebooms observations may be corroborated by the very short branches within Hannoa and low PP and BS values within the clade (Figs. 1 and 2). A re-examination of mo rphological characters in the Hannoa clade will be beneficial to clarify species limits. The well-supported clade of Perriera (a genus of two species endemic to Madagascar) and Gymnostemon (a monotypic endemic of Cte dIvoire) was not surprising, as Aubrville and Pellegrin (1937) suggested the tw o genera were closely related. Perriera was traditionally placed in tribe Picrasmeae due to the lack of a filament appendage (Engler, 1931; Perrier de la Bathie, 1950). Gymnostemon, which was not included in Englers tribal classification, also lacks a filament appendage (Aubrville and Pellegrin, 19 37; Hutchinson and Dalziel, 1954). Given that these are the only two genera in clade V without a filament appendage, loss of the appendage is likely a synapomorphy for Perriera and Gymnostemon. Perriera and Gymnostemon were not considered by Nooteboom in his broad circumscription of Quassia. The placement of Odyendea gabonensis outside the African clade in the basal polytomy of clade V (figure 2) is surprising, given its morphological sim ilarity to members of Hannoa. More sequence data and a reexamination of morphology may clarify the position of this species in clade V.

PAGE 63

63 The largest subclade in clade V, which consists of Simaba, Simarouba and Pierreodendron has a high PP (1.0) but relatively low BS support (71%). A close relationship of Pierreodendron to Simarouba and Simaba in particular has never previously been suggested based on morphology. The monophyly of Simarouba is supported by high PP (1.0) and BS (100%) values, and its status as a genus distinct from Simaba is also suggested. However, the monophyly of Simaba is not well supported (PP = 0.75; BS = <50%). Simaba is distinct from Simarouba based on its perfect flowers with capitate or lobed stigmas, and typically opposite leaflets, as compared to the unisexual flowers, long divergent stigmas and offset leaflets of Simarouba (Cronquist, 1944b,c,d). Simaba consists of two subclades, one of which corresponds to Englers (1874) section Tenuiflorae (containing the S. guianensis complex (Thomas, 1985; Franceschinelli and Thomas, 2000)), which is ch aracterized by small flowers with puberulent petals and is primarily found in the moist Amazon basin (Cronquist, 1944 c ). The second subclade consists of sections Floribundae and Grandiflorae; these are larger-flowered species with villous-tomentose petals, primarily found in the drier regions of southern and eastern Brazil and Paraguay (Cronquist, 1944 c ). Section Floribundae is paraphyletic with respect to Grandiflorae. However, only S. cedron was sampled from section Grandiflorae so more taxa are needed to assess the monophyly of the two sec tions. The variable positions of accessions of S. guianensis and low PP and BS values within section Tenuiflorae reflect the difficulties of delimiting species and subspecies in this clad e (Thomas, 1985, pers. comm.). More taxa and a more rapidly evolving gene region will help to clarify relationshi ps in this group. Overall the generic limits in the Sima roubaceae are supported by molecular data. Amaroria is nested within Soulamea as suggested by Nooteboom (1962b), and Nothospondias, sometimes placed in Anacardiaceae, is supported as include d in Simaroubaceae. Molecular data reveal

PAGE 64

64 several clades, with high PP and BS values, that correspond to traditional generic limits in tribe Simaroubeae, and thus we consid er a broad circumscription of Quassia (Nooteboom, 1962 a) unnecessarily conservative, in agreement with pr evious authors (Porter, 1973; Cavalcante, 1983; Feuillett, 1983; Fernando and Quinn, 1992). Quassia should be limited to just two species, Q. amara and Q. africana, and simple-leaved Old World Quassia species should be recognized as Samadera Simaba, Simarouba, Pierreodendron, Hannoa, Gymnostemon, Perriera, Eurycoma and Odyendea gabonensis are recognized as distinct entiti es based on our molecular data, in agreement with quassinoid chemistry (Da Silv a and Gottlieb, 1987), pericarp anatomy (Fernando and Quinn, 1992) and diagnostic morphological characters (Engler 1931; Cronquist, 1944 b,c,d). The relative importance that Engler (1931) placed on characters such as appendaged filaments and the nature of the androecium in delimiting tribes and subtribes is not borne out by the molecular phylogeny. Engler placed Perriera with genera of tribe Picrasmeae (e.g. Brucea, Ailanthus, Picrasma ), because they all lack appendaged filaments. However, molecular data place Perriera within tribe Simaroubeae (e.g. Simaba, Quassia, Hannoa, Eurycoma ) in the phylogeny. Although variation in th e androecium characters wa s commonly used by Engler (1931) to delimit subtribes, the phylogeny reveals la bility in the nature of the androecium, with typically diplostemonous flowers in the family, but haplostemonous flowers in Picrasma Brucea, Picrolemma and Eurycoma pleiostemonous flowers in Pierreodendron and staminodes occurring in the staminate flowers of Picrolemma and Eurycoma. Thus, haplostemony and staminodes evolved multiple times in the family. This study demonstrates the effectiveness of combining data from two different genomes and genes with a variety of rates of evolution (Olmstead and Sweere, 1994; Sullivan, 1996; Huelsenbeck et al., 1996). A total evidence approach, coupled with broad taxonomic sampling,

PAGE 65

65 has produced a well-resolved and well-supported phyloge ny at all levels from the species to the family. Comparisons among different strategies for partitioning the data in combined Bayesian analyses have revealed that the most complex model provides the best f it to the data using the Bayes factor criterion and examination of lik elihood scores. Partitioni ng analysis has also highlighted the importance of exploring the effects of modelling on resulting posterior probability distributions and tree topologies (Cummings et al., 2003). Results should be interpreted cautiously, due to the lack of data concerning the effects of random error on tree topology and clade support (Lewis et al., 2005), and further work is needed to explore withinpartition modelling (Lemmon and Moriarty, 2004; Nylander et al., 2004) and signs of nonidentifiability among model parameters (Rannala, 2002; Sullivan and Joyce, 2005). Despite this caution, the identical t opologies across partit ioning strategies suggest an underlying phylogenetic signal that is strong enough to allay some of the concerns about incorrect model choice and over-parameterization, particularly wh en the goal is to produce a robust topology, and place less importance on accurately estimating branch lengths (Lemmon and Moriarty, 2004; Kelchner and Thomas, 2007).

PAGE 66

66 Table 3-1. List of the 22 genera of the Si maroubaceae grouped by Englers (1931) tribal classification, with number of species per ge nus, number of species sampled in this study and geographic distribution. Taxa la beled unknown are those not included in Englers circumscription of the subfamily Simarouboideae. TRIBE Genus Clade number (this study) Number of species Number of species sampled Geographic Distribution SIMAROUBEAE Eurycoma V 3 2 Southeast Asia Hannoa V 4 3 Tropical Africa Odyendea V 2 2 Tropical west Africa Pierreodendron V 2 1 Tropical Africa Quassia 2 2 New World, tropical Africa Samadera IV 7 5 Old World tropics Simaba V 25 8 Central and South America Simarouba V 6 4 Central and South America PICRASMEAE Ailanthus II 5 4 Asia to Australia Brucea III 7 8 6 Old World tropics Castela I 12 5 SW U.S., Mexico, Central and South America Holacantha I 2 1 SW U.S., Mexico Perriera V 1 2 1 Madagascar Picrasma I 8 5 New World, Asia, southeast Asia Picrolemma 2 1 South America SOULAMEAE Amaroria III 1 1 Fiji Soulamea III 13 3 Seychelles, southeast Asia, New Caledonia and Pacific Islands UNKNOWN TRIBE Gymnostemon V 1 1 Cte dIvoire Iridosma V 1 0 Tropical west Africa Laumoniera 1 0 Sumatra Leitneria III 1 1 SE U.S. Nothospondias 1 1 Tropical west Africa Total ~109a57 a Note total number of species is estimated to be higher than Fernando and Quinn (1995).

PAGE 67

67Table 3-2. Primers used for PCR amplification and sequencing. Locus Primer name Amp/Seq Sequence Reference rbcL Z1 A+S ATGTCACCACAAACAGAAACTAAAGCAAGT Zurawski et al. 1984 3 A+S CTCGGAGCTCCTTTTAGTAAAAGATTGGGCCGA Zurawski et al. 1984 346F S ATGTTTACTTCCATTGTGGGTAATGTATTT Zurawski et al. 1984 895R S ACCATGATTCTTCTGTCTATCAATAACTGC Zurawski et al. 1984 atpB S2 A+S TATGAGAATCAATCCTACTACTTCT Hoot et al. 1995 S1494R A+S TCAGTACACAAAGATTTAAGGTCAT Hoot et al 1995 S335 S ACGTGCTTGGGGAGCCTGTTGATAA Hoot et al. 1995 S1186R S TGTCCTGAAGTTCTTTGTAACGTTG Hoot et al 1995 matK trnK-3932F A+S CCACGACTGATCCTGAAAGG This study trnK-2R* A+S ATCCCCGTGTCAACCAATAG This study 501F A+S GTTCAACCCCTTCGCTACTG This study 935R A+S TTTCCATTTAGTCATCAGAAGAGG This study 147F S TCATTGGAAAATGGGGGTTA This study 1388R S TTTACGAACCAAACTTTTAACACA This study phyC PHYC-smbF1 A+S GGCAYTGAARTCATAYAARCTTGC This study PHYC-smbR1 A+S CCRCCCCACTTGATCTCYTT This study PHYC-304F S CGCAAGCTTCCAGATTTCTT This study PHYC-699R S AGCATGTCACAAAGCACAGTTT This study

PAGE 68

68 Table 3-3. Wilcoxon sum of rank test results sh owing pairwise comparisons between data partitions for three bootstrap constraint t opologies, ranging from mo st stringent (70%) to least stringent (90%). Comparisons with significant incongru ence (p < 0.05) are indicated with an asterisk. Partition 70% Bootstrap constraint topology rbcL atpB matK phyC plastid rbcL 0.279 0.070 0.056 atpB 0.488 0.315 0.185 matK 0.352 0.383 0.072 phyC 0.415 *0.013 0.204 *0.005 plastid *0.014 80% Bootstrap constraint topology rbcL atpB matK phyC plastid rbcL 0.446 0.094 0.056 atpB 0.421 0.342 0.258 matK 0.347 0.394 0.074 phyC 0.608 0.553 0.204 *0.007 plastid *0.014 90% Bootstrap constraint topology rbcL atpB matK phyC plastid rbcL 0.458 0.111 0.075 atpB 0.438 0.341 0.223 matK 0.347 0.329 0.128 phyC 0.546 0.686 0.396 0.284 plastid *0.031

PAGE 69

69Table 3-4. Results of maximum parsimony analyses for individual data partitions and combined analyses of Simaroubaceae. a indicates MaxTrees limit was reached. Number of ingroup taxa Aligned length % missing data Variable characters Informative characters Number of shortest trees Length of best tree Consistency index Retention Index rbcL 67 1394 1.12 260 153 100,000a524 0.608 0.831 atpB 65 1438 4.92 276 160 524 446 0.722 0.854 matK + partial trnK intron 66 2184 3.54 874 526 8 1569 0.705 0.854 p hy C 62 954 5.63 452 285 360 969 0.635 0.813 Plastid combined 67 5016 4.76 1410 839 8 2557 0.684 0.846 All Combined 67 5970 5.93 1859 1132 8 3544 0.667 0.835 Table 3-5. Summary of results from data partitioning analyses. Fr om left: 2ln Bayes factors comp aring the five different partit ioning strategies; arithmetic mean ln L and credibility intervals for each partitio n strategy; comparisons between posterior probability clade support for five exemplar nodes that are weakly supported in the phylogeny (Figure 2). Partitioning Strategy 2ln Bayes Factors for alternative hypotheses Likelihood scores Posterior probabilities of nodes A E (Figure 2) P13 P5 P4 P2 Mean ln L Upper 95% CI Lower 95% CI A B C D E P1 1391.34 550.00 535.40 58.88 31031.49 31013.72 31051.20 0.74 0.91 0.69 0.94 0.87 P2 1332.46 491.12 476.52 30999.67 30981.38 31020.84 0.83 0.92 0.78 0.94 0.88 P4 855.94 14.60 30756.37 30736.61 30777.39 0.56 0.90 0.71 0.88 0.83 P5 841.34 30754.10 30733.42 30775.83 0.80 0.93 0.93 0.76 0.81 P13 30324.82 30301.60 30348.62 0.74 0.98 0.91 0.69 0.75

PAGE 70

70 Figure 3-1. Strict consensus of eight most parsimonious trees recovered from a combined analysis of plastid genes rbcL, atpB and matK (including partial trnK intron) and 1kb of nuclear gene phyC for Simaroubaceae. Bootstrap support (BS) greater than 50% is shown above branches, decay indices are shown below. Major clades (I V) of Simaroubaceae are indicated on the right hand side. Simaba orinocensis Simaba orinocensis Simaba guianensis Simaba guianensis Simaba guianensis Simaba polyphylla Simaba guianensis Simaba glabra Simaba glabra Simaba cuneata Simaba ferruginia Simaba insignis Simaba cedron Simarouba glauca cultivated Simarouba glauca wild Simarouba tulae Simarouba amara Simarouba versicolor Pierreodendron africanum Odyendea gabonensis Eurycoma apiculata Eurycoma longifolia Hannoa undulata Hannoa undulata Hannoa chlorantha Hannoa sp.A Hannoa sp.B Hannoa klaineana Gymnostemon zaizou Perriera madagascariensis Quassia africana Quassia amara Samadera bidwillii Samadera sp.B Samadera sp.C Samadera baileyana Samadera indica Picrolemma sprucei Picrolemma sprucei Brucea antidysenterica Brucea ferruginia Brucea guineensis Brucea javanica Brucea mollis Soulamea amara Soulamea terminalioides Amaroria soulameoides Soulamea morattii Leitneria floridana Nothospondias staudtii Ailanthus altissima Ailanthus altissima var tanakai Ailanthus integrifolia Ailanthus triphysa Ailanthus fordii Castela erecta Castela erecta Castela tortuosa Castela retusa Castela tweedii Castela coccinea Holacantha emoryi Picrasma antillana Picrasma excelsa Picrasma crenata Picrasma quassioides Picrasma javanica Melia azedarach Unknown Meliaceae Swietenia mahogani Casimiroa edulis Zanthoxylum sp. Cneorum tricoccon Acer saccharum 100 >10 100 >10 100 >10 100 >10 100 >10 93 2 100 >10 97 3 84 2 100 >10 100 >10 100 >10 100 >10 88 3 100 >10 72 1 100 >10 100 >10 78 3 100 >10 80 3 100 >10 62 1 100 >10 58 1 100 >10 100 >10 76 2 100 >10 100 10 100 >10 92 3 100 >10 98 9 100 >10 100 >10 97 9 100 >10 100 >10 98 >10 87 4 100 >10 71 2 <50 1 <50 1 69 2 100 >10 <50 1 76 2 89 2 54 1 94 6 100 >10 75 259 1 99 4 100 >10 95 4 100 >10 97 7 99 9 100 >10 53 1 64 1 100 >10 95 4 100 >10 V III III IVMeliaceaeRutaceaeSapindaceae

PAGE 71

71 Figure 3-2. Phylogram of the majority rule consensus of trees for Simaroubaceae, sampled from the posterior distribution in the Bayesian analysis partitioned by gene and codon position (5,000,000 generations, sampled every 1000 generations; burnin = 100,000 generations). Posterior probabilities (PP) are shown on branches. Branch lengths are proportional to the mean number of nucleotide substitutions per site, calculated from the posterior distribution density. Five nodes compared between different partitioning strategies are highlighted (A E). Major clades (I V) of Simaroubaceae are indicated on the right hand side. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Simaba polyphylla Swietenia mahogani Quassia amara Simaba cedron Simaba cuneata Odyendea gabonensis Soulamea terminalioides Hannoa undulata Simaba orinocensis Hannoa chlorantha Picrasma excelsa Quassia africana Brucea ferruginia Picrolemma sprucei Simarouba glauca wild Castela tweedii Pierreodendron africanum Hannoa sp.A Castela coccinea Brucea mollis Ailanthus altissima var tanakai Simarouba glauca cultivated Castela tortuosa Eurycoma apiculata Hannoa klaineana Simarouba amara Soulamea amara Simaba orinocensis Samadera indica Brucea guineensis Simaba guianensis Soulamea morattii Gymnostemon zaizou Nothospondias staudtii Picrasma quassioides Ailanthus altissima Ailanthus triphysa Picrasma javanica Simaba ferruginia Perriera madagascariensis Simaba glabra Ailanthus fordii Acer saccharum Simarouba tulae Picrasma antillana Picrolemma sprucei Cneorum tricoccon Castela erecta Unknown Meliaceae Simaba guianensis Ailanthus integrifolia Holacantha emoryi Simaba guianensis Samadera sp.B Samadera bidwillii Leitneria floridana Melia azedarach Hannoa undulata Brucea antidysenterica Castela erecta Castela retusa Hannoa sp.B Picrasma crenata Zanthoxylum sp. Simaba insignis Samadera sp.C Amaroria soulameoides Simaba guianensis Simaba glabra Simarouba versicolor Samadera baileyana Casimiroa edulis Brucea javanica Eurycoma longifolia 0.01 substitutions per siteVIII I II IV Meliaceae Rutaceae A E D C B 1 1 10.991 1 1 1 1 1 1 1 1 10.741 1 1 1 1 1 1 1 1 1 1 10.290.3710.69 0.91 0.98 0.750.99 0.99 0.99 0.57 0.75 0.97 0.89 0.84 0.99 0.55

PAGE 72

72 CHAPTER 4 LIKELIHOOD ANALYSIS OF GEOGR APHIC RANGE EVOLUTION IN SI MAROUBACEAE Introduction The number of phylogeny-based biogeographi cal treatments for tropical groups has grown considerably in recent years (Renner et al., 2001; Richardson et al., 2004; Davis et al., 2002; Lavin et al., 2000, 2004; W eek s et al., 2005; Zerega et al ., 2005; Muellner et al., 2006), with improved phylogeny reconstruction and dating techniques, and a diversity of methods for ancestral area reconstruction (AAR). However, resolved a nd well-supported phylogenies of pantropical clades, that is, clades with taxa distributed in each of the neotropics, tropical Africa and tropical Asia and Australia, are still relativ ely few (e.g. Renner et al ., 2001; Davis et al., 2002; Lavin et al., 2004; Weeks et al., 2005; Zerega et al., 2005; Muellner et al., 2006). Simaroubaceae are an ideal pantropical candidate for exploring hypotheses of intercontinental migration, given their well-supported relations hips, taxon sampling and broad geographical distribution (Chapter 3). Although the family is much smaller in species diversity (~109 species) than many well-studied pantropical clades, a number of intriguing disjuncti ons warrant a detailed investigation. All three major tropical disjunction s, between Asia and the New World, between Africa and the New World, and between Africa a nd Southeast Asia, occur in the family, and occur multiple times. Furthermore, Simaroub aceae have tropical, subtropical and temperate elements, including both mesophy tic and dry-adapted taxa. Thes e factors add an ecological dimension to the interpretation of biogeographi cal patterns (Riddle, 1996; Wiens and Donoghue, 2004) and may provide evidence fo r migration and speciation as a result of climatic change. Recent studies have shown the biogeographical history of a number of tropical angiosperm clades (e.g. Renner et al., 2001; Davis et al., 2002; Weeks et al., 2005; Zerega et al., 2005; Muellner et al., 2006) to be the result of interactions between c limate change, migration via land-

PAGE 73

73 bridges and long-distance dispersa l, and the synthesis of data such as these has provided an analytical framework for the study of the origins and maintenance of tropical biomes as a whole (Pennington et al., 2004a, 2006; Davis et al., 20 05). Additionally, the timing of radiations of strictly tropical taxa from s ubtropical and temperate ancestors, may reveal how adaptive shifts contribute to the modern tropical community (Lav in et al., 2004), and Simaroubaceae provide an opportunity for such a study. Although the family does not constitute a significant component of biome diversity in terms of species numbers as in, for example, Fabaceae (Lavin et al., 2004) or Melastomataceae (Renner et al., 2001), it may provide data on how smaller families enhance the phylogenetic diversity (PD; Fa ith, 1992a; Barker, 2002) of species-rich environments. Simaroubaceae have a variety of fruit types a nd morphology; fruits can be small and nutlike (e.g. Brucea, Eurycoma ), small and fleshy (e.g. Picrasma Castela ), large and fleshy (e.g. Gymnostemon, Pierreodendron ), dry and winged (samaras; e.g. Ailanthus and Soulamea ) and woody and floating ( Samadera indica ). Reconstructing dispersal ev ents in the history of the family may provide some insight into how disp ersal capabilities are manifest in fruit type. The goals of this study are to: 1) reconstruc t divergence dates within Simaroubaceae using recently developed molecular rates analyses and fossil calibration; 2) propose a center of origin for the Simaroubaceae based on molecular rates an alyses and AARs; 3) use suitable interand intracontinental dispersal routes their maintenance and disappea rance, to test hypotheses of vicariant cladogenesis in the hi story of the family; and 4) identify potential long-distance dispersal events when vicariant hypotheses of speciation seem doubtful, and relate these to fruit morphology. Consideration will be given to the eff ects of past episodes of climate change, the ecological requirements of exta nt species, and how these might be reflected in ancestral distributions. Maximum-likelihood analysis of ge ographic range evolutio n (Ree et al., 2005; Ree

PAGE 74

74 and Smith, 2008) has yet to be applied to a pantropical group such as Simaroubaceae. The complex history of intera nd intracontinental migration th at is suggested by phylogenetic relationships between the extant species, and the areas they inhabit, should be well suited to a model-based method, rather than parsimony or event-based methods (dispersal-vicariance analysis (DIVA); Ronquist, 1997) th at are currently used. A DIVA an alysis is included in the study for comparative purposes. Methods Taxon Sampling, DNA Sequencing and Phylogenetic Analyses Taxa used in this study were the same as t hose of Clayton et al. (2007), with the addition of four species of Simaba (plus a second accession of S. cuneata ) and one species of Simarouba. Sequences for these new taxa were determined using the same methods as Clayton et al. (2007) for four gene regions, rbcL atpB matK plus partial trnK intron, and phyC (Genbank accession numbers EU546227 EU546249). These taxa were then added to the 67 accessions used in the previous study, totalling 73 ingroup accessions overall, and 62 out of ca. 109 species. Phylogenetic analyses consisted of Bayesian infe rence. Sequence data were partitioned by gene region and codon position for coding regions. This partitioning stra tegy was determined to be the best-fitting model using Bayes F actors (Chapter 3). Substitution models for each partition were determined with the Akaike information criter ion (AIC) in Modeltest, version 3.6 (Posada and Crandall, 1998). Two independent analyses were run for 10,000,000 generations each, using four Markov chains and default parameters; trees we re sampled every thousandth generation, with a burn-in of 150,000 generations, a nd stationarity of the Markov chain Monte Carlo (MCMC) was determined using the average standard deviation of split frequencies between runs. The data matrix and resulting tree were submitted to TreeBASE (matrix acce ssion number M3859; study accession number S2060). Compared to analyses in Clayton et al. (2007), no changes to the

PAGE 75

75 topology, its resolution, or clade support were found with the additional species. A likelihoodratio test (LRT) was performed on the combin ed data in PAUP* 4.0b10 (Swofford, 2002), to determine if the regions sequenced are evol ving in a clock-like fashion using maximum likelihood. The substitution model for the combined data set (GTR+I+G) was determined using AIC in Modeltest. Results of this test (2 = 245 and d.f. = 79, p < 0.0001) rejected the molecular clock hypothesis, and a relaxe d clock method was used. Divergence Date Estimation Divergence date estimation was performed using the uncorrelated lognorm al (UCLN) model of molecular rates estimation (Drumm ond et al., 2006), implemented in BEAST (v1.4.7, Drummond and Rambaut, 2007), a Bayesian approach to molecular rates analysis. This method uses a global sampling strategy (Markov Chain Monte Carlo) to maximize both phylogenetic inference and molecular rate estimation on br anches simultaneously (Drummond et al., 2006), thus eliminating the problems associated with selecting a tree topology and branch lengths a priori. The method also allows for calibration date priors to have a proba bility distribution, to reflect uncertainties in fossil dates. Fossil calibration The phylogeny was calibrated with four fossils: samaras of two different fossil species of Ailanthus (Corbett and Manchester, 2004), Leitneria fruit endocarps (Dorofeev, 1994; Nikitin, 2006a) and Chaneya fruits (W ang and Manchester, 2000; Teodoridis and Kvacek, 2005). Ailanthus samaras are common in de posits throughout the Tertia ry across the Northern Hemisphere (Fig. 1), with earliest occurrences in the Middle to Early Eocene. The earliest example is from the Fossil Butte fish quarries of the Green River formation, Wyoming, dated at 52.2 52.7 Ma. Fossil Ailanthus samaras share all the fruit syna pomorphies of the extant genus that are discernible in the preserved remains. However, attributing the fossil fruits specifically to

PAGE 76

76 one or another extant species is difficult, due to the limited nu mber of characters available (Corbett and Manchester, 2004). Three fossil species are recognized: A. confucii the most commonly occurring species in the genus fossil history, and closest morphologically to extant A. altissima ; A. tardensis known from a single loca lity in Hungary in th e Early Oligocene, and closest in form to extant A. triphysa ; A. gigas known from a single locali ty in Croatia in the Early Oligocene, and described as bei ng intriguingly similar to extant A. integrifolia (Corbett and Manchester, 2004). Although the three fossil species have affinities to extant species, there is variation in characteristics such as fruit size and position of the ventral vein. Ailanthus confucii specimens are the earliest example of the genus; however, A. fordii is the sister to all other extant species (although A. excelsa has not been sampled; Chapter 3). For these reasons three different Ailanthus calibration points were tested in the molecular rates analyses. Firstly, the earliest fossil occurrence ( A. confucii 52.7 Ma) was used to date the node at which Ailanthus diverges from the rest of the Simaroubaceae (the Ailanthus stem lineage). Secondly, the earli est fossil occurrence ( A. confucii 52.7 Ma) was used to date the node at which A. altissima diverges from its sister group (A. integrifolia + A. triphysa ). Thirdly, the Ailanthus gigas fossil (Early Oligocene) was used to date the node at which A. integrifolia diverges from its sister species ( A. triphysa ). Given the numerous occurrences of Ailanthus fossils in deposits across the Northern Hemisphere, and their consiste nt appearance throughout the Tertiary, up until the Pleistocene of Japan (Corbett and Manchester, 2004), the date for A. confucii was considered a robust calibration, as in previous dating analyses (Muellner et al., 2006, 2007). This robustness is considered in terms of how closely the earliest known appearance of the fossil approximates to the first appearance of the genus: Ailanthus samaras are readily preserved, straightforwar d to identify and relatively common, and the sudden appearance

PAGE 77

77 of samaras in the fossil record in the Early Eocene may indicate that the genus did not exist before that time. Proximity of the assigned fossil date to the initial cladogenesis event in geological time is critical to accu rate divergence date estimation. Leitneria has fossilized endocarps from the Early Oligocene to the Pl iocene of Western Siberia and the Miocene and Pliocene of Eur ope (Fig. 1). Several species are described (Dorofeev, 1994; Nikitin, 2006a) and show a stri king similarity to the endocarp anatomy of extant L. floridana Nikitin (2006a) recorded a putative Leitneria sp. from the Late Eocene (Priabonian), but its precise affinities are que stioned by the author, therefore Early Oligocene was used in calibrations. As extant Leitneria is a monotypic genus, the only logical position for the calibration point is the node at which L. floridana diverges from its sister clade. A third fossil is the extinct genus Chaneya (Wang and Manchester, 2000; Teodoridis and Kvacek, 2005), which has certain affinities with extant Picrasma Although the fossil appears to have oil cells in the perianth, which are lack ing in Simaroubaceae, an accrescent pentamerous calyx surrounding one, two or three globose meri carps when mature, is consistent with Picrasma (Wang and Manchester, 2000). A Chaneya calibration point at the Picrasma stem group was therefore tested for compatibility with the othe r fossils, using its earliest occurrences in the Middle Eocene of western North America and Asia. However, the po tential utility of Chaneya is limited by its uncertain relationship. Fossil pollen of Simaroubaceae was not consid ered for this study, given the lack of distinctive morphological characteristics in extant Simaroubaceae po llen (Basak, 1963, 1967; Moncada and Machado, 1987). Reports of Ailanthus fossil pollen (Song et al., 2004) and Leitneria fossil pollen (Machen, 1971) were considered unreliable.

PAGE 78

78 In BEAST v1.4.7 (Drummond and Rambaut, 2007), calibration points take the form of priors with a probability distribution. Each prior tested (Table 4-1) took a lognormal distribution, with the means and standard deviations designated to reflect our confiden ce in the fossils used. The Ailanthus confucii fossil had a narrow lognormal distri bution, given its well-characterized fossils and more precise stratigraphic age. Leitneria was given a broader probability distribution than A. confucii to account for the questionable Priabonian species. Both A. gigas and Chaneya calibrations had larger standard deviations because of the uncertain affiliations of these fossils. The hard lower bounds represent the youngest dates (D onoghue and Benton, 2007) for stratigraphic ages of epoch boundaries (Gradstein and Ogg, 2004), and in the case of Ailanthus confucii the more precise dating of the Fossil Butte fish quarries at Green River, Wyoming. In total, eight different combinations of fossil plac ements (here referred to as calibration schemes; Table 4-2) were tested, to determine how well fossils and divergence dates correlated with each other. Calibration at the family level (e.g. Wi kstrm et al., 2001, 2004; Muellner et al., 2007) was not used, to avoid compounding the error inhere nt in secondary calibra tion dates from other molecular rates analyses (Graur and Martin, 2004). However, results for the age of the Simaroubaceae are discussed with reference to di vergence dates from Muellner et al. (2007). Molecular rates analyses Analyses for each calibration scheme (T able 2) were performed in BEAST v1.4.7 (Drummond and Rambaut, 2007) using the various calibration points desc ribed above. Each analysis was run using the four-gene, 62-species data set plus outgroups. A combined data set of the four coding markers, with the exclusion of the non-coding trnK intron, used GTR+I+G as the model of nucleotide substitution, determined w ith the AIC in Modeltest v3.6 (Posada and Crandall, 1998). The non-coding intron was excluded because it was more difficult to align than the coding region (Chapter 3) a nd contained large indels in so me species; misalignment could

PAGE 79

79 cause discrepancies in branch leng ths that in turn would affect estimates of molecular rates along those branches. The coding regions were partitioned into three codon positions, based on results of partitioning strategies examined in Clayton et al. (2007; partitioning by gene region is not currently implemented in BEAUti v1.4.7 (Drummond and Rambaut, 2007)). Both the substitution model and rate heterogeneity were unlinked across codon positions. Priors other than the calibration points were uniform and set at defa ult values, and the tree root prior was set as a uniform distribution, 0 132 Myr (the estimated age of the angiosperms (Hughes, 1994)). In calibration scheme 2 this prior produced a tree with zero likelihood and so was removed. Each BEAST analysis used a starting tree with initial branch lengths satisfying the upper and lower bounds for the calibration priors being used, cons tructed using pathd8 (Britton et al., 2007). An initial analysis was run using the fossil placem ent in calibration scheme 1 and no topological constraints, producing a tree with an identical topology to the orig inal phylogenetic analysis (Fig. 2), except that Quassia was sister to Clade V (PP = 0.86; also seen in parsimony analyses of Clayton et al. (2007)), and all outgroups formed a clade (PP = 0.41) sister to Simaroubaceae. We therefore constrained the topology of all further BEAST analyses to have Samadera as sister to Clade V, and Rutaceae sister to Simaroubaceae, as per previous Bayesian analyses (Fig. 1; Clayton at al., 2007). Although there is some ev idence to suggest application of a relaxed molecular clock leads to improved phylogeny es timation (Drummond et al., 2006), we preferred to follow the more thorough phylogenetic analyses fr om the previous chapte r, particularly with the single outgroup clade seen in the unconstr ained analysis. For each calibration scheme, two independent analyses ran for 10,000,000 genera tions each, sampling every 1000 generations. Stationarity of the MCMC was determined by comparing standard deviations of node times between the two independent runs, and observing burn-in plots for the posterior and all model

PAGE 80

80 parameter values in Tracer v1.4 (Ramba ut and Drummond, 2003). A burn-in of 1,000,000 generations was removed from each run. For all es timated parameters, an effective sample size (ESS) of > 150 was obtained to ensure the pa rameter space had been sufficiently sampled (except in the case of calibration scheme 2, wher e two further runs were needed for an ESS > 150 for all parameters). Once stationarity had be en determined, the two independent runs for each scheme were combined using LogCombiner v1.4.7 (Drummond and Rambaut, 2007). The calibration scheme ultimately selected for biogeogr aphic analyses, out of the eight combinations of fossils tested (see Results), was generated a second time with four separate 10-milliongeneration runs and with scale operators adjusted to recommendations in the BEAST output from the original run for that calibration scheme (BEAST input .xml in Supplementary Information). The four runs were combined using LogCombiner v1.4.7 (Drummond and Rambaut, 2007), ensuring ESS > 600 for all parameters. Biogeographic Analyses Using a Likelihood Approach Biogeographic data were compiled from speci es distributions in the literature (e.g. Cronquist, 1944a,b,c,d; Nooteboom 1962a,b), and GIS mapping of specimen localities in ArcGIS v9.0 (http://www.esri.com /software/arcgis/). The major areas in which Simaroubaceae are distributed were categorized broadly into seven areas for AAR (Fig. 1). Areas were delimited by continental divisions based on present and pa st separation of major landmasses, with the exception of Asia and Europe. Interpretation of the results, however, incorporates more specific geographic information, especially for island ta xa. Figure 4 shows geogr aphical distributions assigned to terminals for AAR, base d on the following categories: 1) N North and Central America (north of the Panamanian Isthmus), and Caribbean Islands; 2) S South America; 3) F Tropical Africa; 4) M Madagascar; 5) A Mainland Asia, including but not restricted to India, China, Bhutan, Myanmar, Japan and Russi a (fossil localities), plus insular SE Asia,

PAGE 81

81 including but not restricted to Malaysia, Indonesia and the Philippines; 6) U Australia, New Guinea, Papua New Guinea, New Caledonia and the Pacific Islands; 8) E Europe (fossil localities). Widespread species were assigned to more than one area. The program Lagrange (Ree and Smit h, 2008) was used for AAR. Unlike DIVA (Ronquist, 1997), it incorporates an explicit model of dispersa l routes available at historical intervals, and estimates vagility and extinction parameters, correlating stochastic events with lineage persistence, as part of the DEC model (Dispersal-Ex tinction-Cladogenesis; Ree and Smith, 2008). Five different combinations of para meters (here referred to as models M1-M5) were tested to assess how model complexity and a lteration of certain parameters would affect the resulting AARs. The model of dispersal route availability was developed based on geologic history and the presence and dissolution of la nd bridges and island chains, and climatic data. In particular, geological events such as the presence of the North Atlantic Land Bri dge, the Southern track between Australia and South America, island chains between North and South America, the collision of Australian and Asian plates, and the closure of th e Panamanian Isthmus (Tiffney, 1985a; Morley, 2003; Scotese, 2001) are accounted fo r in dispersal probabilities. Two different dispersal scenarios were tested. The first dispersal scenario (D1) uses probabilities ranging from 0.1 for well-separated areas to 1.0 for contiguous landmasses (Table 4-3). The second scenario (D2) uses the same connections but imposes much smaller dispersal probabilities (0.01) for areas not directly connected (Table 4-3). The five models are as follows: 1. M1 Taxa are free to disperse between any area at any time, but ar e restricted to the following ancestral ranges (based on potential ly plausible and extant ranges): NS, NF, NE, NA, SU, FM, FA, FE, AU, AE, MAU;

PAGE 82

82 2. M2 Ancestral ranges are unrestricted and taxa are subject to disp ersal probabilities of scenario D1; 3. M3 Ancestral ranges are restricted as in M1 and subject to disper sal probabilities of scenario D1; 4. M4 Ancestral ranges are restricted as in M1 and subject to disper sal probabilities of scenario D2; 5. M5 Ancestral ranges are restricted as in M1 (excluded E, NE, FE, AE) and subject to dispersal probabilities of scenario D1; fossil taxa are removed from the analysis. An additional analysis was performed with no constraints on the model, in which taxa are free to disperse at any time and ancestral ranges are unrestricted. However, this analysis proved too computationally expensive, and no results we re obtained. Fossil areas were integrated by manually adding a terminal for Ailanthus confucii (coded as present in E, N and A) as the sister to extant Ailanthus, and a Leitneria terminal (coded as present in E and A) as sister to extant Leitneria Chaneya was not included in the analysis. The non-overlapping ra nges of the fossil and extant lineages of Leitneria led us to place the fossil lineage as diverging from the extant lineage at 28.4 Ma, the latest stra tigraphic date for the Early Oligocene, when the fossils first appeared. Ailanthus fossils were more difficult to integrate, given the presence of fossils in Asia, where extant species also live, and the possibil ity of these fossils representing part of the Ailanthus stem lineage. We tested three different points of divergence for the fossil lineage: diverging at 52.2 Myr, i.e. 2.7 Myr from the base of the stem branch, halfway along the stem branch, and 2 Myr from the tip of the stem branch. All three produced near identical likelihood scores and AARs (data not shown) Therefore, the divergence at 52.2 Ma was chosen, as this had the highest likelihood score of th e three positions. Input files for Lagrange are published in Supplementary Information. Model 3 (M3) was chosen for further anal ysis and discussion of the biogeographical implications of AAR in Simaroubaceae. This choi ce was not made on statistical grounds, as M3

PAGE 83

83 has a much lower likelihood than M1, M2 and M5 (T able 4-4), with only a slight increase in the number of parameters. A less constrained model may fit the data well, but at the expense of realism. This is the case here, where freedom to disperse anywhere at any time, ancestral ranges that encompass non-contiguous and sometimes well-s eparated areas, and exclusion of fossils, fail to adequately represent the biology of the orga nisms. We optimized the chosen model (M3) by examining the effects of different ancestral ranges between the Ol d World and New World (namely NA, NE, NF and SU) on the likelihood score of the model. These potential migration routes are important to pantropical clades, a nd in Simaroubaceae, disjunctions between the Old World and New World are common throughout the family. Each ancestral range was tested alone, along with all possible comb inations (Table 4-5). It is not clear how to compare likelihood scores between models with cha nges in the specified ancestral ranges, because the models are not nested, and the additional para meters involved in having fewer an cestral ranges are fixed at a boundary value of zero (R. Ree, pers. comm.). Th e resulting likelihood scores were therefore compared directly. A DIVA analysis (Ronquist, 1997) was also performed on the data, to compare AARs between this event-based approach and the explic it model-based approach of Lagrange (Ree and Smith, 2008). Tree topology and area assignments fr om Lagrange were also used in DIVA, which ran with default program parameters. Two analyses were conducted, with maxareas unlimited and constrained to two. Results Divergence Date Estimation Tracer output statis tics for BEAST analyses show the data are not evolving in a clocklike manner (coefficient of variation for calibration scheme 7 = 0.31 [0.22, 0.40]), and substitution rates between ancestor-descendant lineages are uncorrelated (covariance for

PAGE 84

84 calibration scheme 7 = 0.0071 [-0.16, 0.15]). Therefor e, the UCLN model of rate variation was the most appropriate method for estimating diverg ence dates, as opposed to methods such as r8s (Sanderson, 1997, 2002) and multidivtime (Thorne a nd Kishino, 2002), which use an a priori assumption of autocorrelation between ancestor-descendent lineages. However, the covariance statistic appears to bias re sults against finding autocorrelation (A. Drummond, pers. comm.). Results of the different calibration schemes are s hown with dates for fossil nodes as well as other example nodes from the tree (Table 2), and lowe r and upper 95% highest posterior densities are given in square brackets. Fossils of Leitneria and Ailanthus confucii (calibrating the Ailanthus stem group) corresponded reasonably well with each other (calibration schemes 1 and 5); therefore, these two fossils were used together (calibration scheme 7) to produce the chronogram for biogeographic analyses (Fig. 3). In calibra tion scheme 7, as well as in scheme 1, the divergence date of Leitneria was estimated to be older th an the fossil dates. Using A. confucii at the divergence of A. altissima (scheme 2) produced very old dates thr oughout the tree, with the genus diverging from its sister clade at 139.3 [95.0,190.0] Ma. Ailanthus gigas gave higher age estimates on its own (family crown group 87.3 [67.9,105.3] Ma), and in the final chronogram (calibration scheme 7) the divergence of A. integrifolia was considerably younger than the A. gigas fossil, hence its exclusion from the final analysis. Chaneya alone produced younger dates than the other calibrations, estimating the divergence of Ailanthus at 35.2 [27.8,42.9] Ma. Biogeographic Analyses Likelihood scores and estimates of dispersal and extinction rates fo r the five models tested are shown in Table 4-4, along with AARs for selected node s in the phylogeny, and results of DIVA analyses. Models M1 and M2 showed num erous alternative re constructions for nodes throughout the tree (within the confidence window of a 2 log likelihood unit difference (Edwards, 1992; Ree and Smith, 2008)). For example, the root node in M2 had 33 alternative

PAGE 85

85 reconstructions. Dispersal and extinction rate esti mates were lower for M1 and M2 than the more constrained models. In M3, M4 and M5, alternative reconstructi ons were common, but typically involved the same areas, and only differed in how the ranges evolved at the node. DIVA results showed unconstrained analyses ha d numerous equally optimal rec onstructions at deeper nodes in the phylogeny (Table 4-4). With ancestral ranges rest ricted to two areas, results were similar to Lagrange, particularly the less co nstrained models. Optimal recons truction required 30 dispersal events for the unconstrained an alysis, and 32 dispersals when maxareas was limited to two. The effects of different ancestral range co mbinations in M3 are shown in Table 4-5, ordered by decreasing likelihood. An alyses including SU and NE typically had lower likelihoods and higher dispersal and extincti on rates than other co mbinations. Although the likelihood scores could not be compared using an objective criterio n such as the AIC (Akaike, 1974), the top four ancestral range permutations had scores over 8 log likelihood units better than the next best model, and all produced near-identical AARs (Table 4-5), except for the root node. The combination with the highest likelihood was NA + NF, and this is shown in Figure 4-4. For clarity, the chronogram has been pruned of selected terminals that have a sister taxon in the same area. For nodes where likelihood scores were not significantly different between multiple reconstructions, the relative proba bility of the global likelihood is shown. Figure 4 shows three circumstances of range inheritance. Firstly, ther e are instances of dispersal resulting in range expansion, which are common throughout the tree, especially in the Miocene. Secondly, there are local extinction events (thr ee are hypothesized), which are in ferred when a daughter lineage inherits a different range from its parent (a ra nge expansion prior to extinction is inferred). Thirdly, there is vicariance by cladogenesis, wh ere the ancestral range encompassing two or more areas subdivides between daughter lineages. Also seen in Figure 4-4 are numerous

PAGE 86

86 occurrences of incipient speciation, whereby an ancestral range of two areas subdivides, with one daughter inheriting the widespread ancestral range and one daught er inheriting just one of the two ancestral areas. An example is th e root node of clade III (Fig. 4-4). Discussion Divergence Date Estimation The different placements of the three fossil c onstraints strongly aff ected the divergence dates estim ated, especially date s prior to the divergence of Ailanthus, that is, the divergence of Picrasma Castela and Holacantha, and the age of the family. Prior to the divergence of Ailanthus however, only minor differences were seen between different fossil placements in Clades III, IV and V. There are many sources of error in divergence time estimation (Sanderson, 2002; Graur and Martin, 2004; N ear and Sanderson, 2004). These in clude, but are not limited to: accuracy of dating fossil strata ; correctly identifying fossils (e.g. the ambiguity around the Chaneya assignment); assigning fossil dates to the most appropriate nodes in the phylogeny; poorly supported nodes in the phy logeny; justifying the nature of the prior probability distribution of dates around the fossils; errors in estimating rates of molecular evolution, such as the problems inherent in decoupling rates and time, correctly estimating branch lengths on the topology, and assuming certain characteristics of molecular evolution (e .g. the uncorrelated lognormal distribution prior for substitution rates). However, these potential sources of error do not necessarily inhibit our abi lity to make general statements about migration patterns and dispersal events, in a system which encompasses large-scale and relatively slow processes, such as continental drift and land bridge forma tion and dissolution, over long periods of time. Cross-calibration of fossils showed some compatibility between Ailanthus confucii and Leitneria fossils, but only when A. confucii was associated with the Ailanthus stem group. Calibration scheme 7 was chosen as the best chronogram for biogeographic analyses, as it

PAGE 87

87 included the two fossil placements that had the be st fossil records and showed some overlap in the probability distributions of node ages. In calibration schemes with multiple calibrations, A. confucii at the Ailanthus stem lineage was the most influential calibration point. The older ages for the Leitneria divergence produced in calibration scheme s 1, 4, 7 and 8 could be accounted for by the Leitneria sp. described by Nikitin (2006a) as questi onably Priabonian, that is, older than the conservative lower bound at the end of the Early Oligocene. Although the Chaneya calibration (scheme 6) produced dates younger than those derived from the more extensive fossil records of Ailanthus and Leitneria the mean dates produced for the Picrasma stem and crown in calibration scheme 7 were 67.4 [57.8,77.8] Ma and 14.2 [9.2,20.0] Ma, respectively, and the Chaneya fossils fall between these dates. Therefore, an association between Chaneya and Picrasma remains a possibility. The final chronogram (Fig. 3) illustrates when the major clades of Simaroubaceae arose. Muellner et al. (2007) estimated the age of the Simaroubaceae crown group at 52 Myr, as this was the most conservative placement for the Ailanthus fossil, given their limited taxon sampling. With the addition of Picrasma Castela and Holacantha in this study, the ag e of the crown group is estimated to be Maastrichtian. The mean age of the stem group (88.4 [75.9,101.4] Ma) is difficult to compare to dates provided by Muellner et al. (2007), given that the sister relationship of the family remains poorly resolved. Biogeographic Analyses Likelihood models Reconstructions have a tendency to become more ambiguous deep er in the phylogeny (the so-called widespread ancestor problem; Ree et al., 2005), whic h is seen in the number of ambiguous nodes in the AARs. This is an inevitable result give n the amount of time that has passed since the Simaroubaceae first arose, and the extrapolations made from geographic

PAGE 88

88 distributions of the extant tips. The highly ambi guous reconstructions for M2 throughout the tree, especially at the deeper nodes, s uggested the need for a biogeogra phic model that better explains the data. The impact of the fossils was primarily the inclusion of Europe as the seventh area in analyses, but they only influenced AARs in anal yses which did not include NA (Table 4-5). In reality the connection between Euro pe and North America was an in tegral part of the NALB, but with only two terminals assigned to Europe in the tree, the NE an cestral range did not feature in the chosen analysis. The restriction of ancestral ranges to one or two (and in one case three) areas helps to counterbalance the widespread an cestor problem (Ree et al., 2005), and is an assumption based on the low likelihood of an ancestral species having a range across three or more of the geographic areas described above. Given the extant species distributions, only Samadera indica has a presence in three areas. Specification of an cestral ranges was an infl uential factor in how ancestral areas were reconstructe d, particularly the inclusion or exclusion of NA, NF or NE. These differences indicate the critical nature of the path between the Old World and the New World, and which is more plausible for Si maroubaceae. Comparison of likelihood scores between NA-, NE-, NFor SU-constrained models (Table 4-5) shows that NA has the highest likelihood for a single ancestral range, and in combination NA and NF have the highest likelihood (Fig. 4). This suggests tr ans-Beringial migration to be an important determining factor in Simaroubaceae biogeography, more so than the NALB. Exclusively trans-Atlantic disjunctions (excluding Ailanthus fossil species which could have migrated via Beringia or the NALB) were only hypothesized for three nodes, one of which arose after the NALB was available (the root node of Simaba, Simarouba and Pierreodendron), and one of which the timing is unknown (the dispersal of Nothospondias to Africa). Models that included NF and NE,

PAGE 89

89 but excluded NA, increased probability of migr ation via the NALB as opposed to Beringia a priori, and may be more biologi cally realistic for tropical gr oups. However, the early-diverging clades of Simaroubaceae, including genera such as Picrasma Ailanthus and Leitneria continue to maintain a temperate element, supporting the Beringial track. The Southern track (SU) produced significantly lower likelihoods than the other three ranges, due to how disconnected South America and Australia were through much of the Tertiary. North America was typically the optimal reconstruction for the root node in th e ancestral range permuta tions with the highest likelihoods. Ree and Smith (2008) note that in their analysis of Psychotria widespread ranges are rapidly reduced to single areas by cla dogenesis or local extinction. However, in Simaroubaceae, many nodes retain widespread dist ributions, especially deeper in the phylogeny. Most of these widespread distributions break down through vicariance, with few lineages going extinct locally ( Picrolemma and Nothospondias ). Whether these ancestral ranges are plausible over a significant period of time is debatable, but the assumpti on of trans-Atlantic and transBeringial disjunctions in the early history of th e family, has been suggested for other tropical angiosperm families. The likelihood model could be streamlined further by introducing stratified ancestral ranges, such that ancestral distributions can vary with different periods in the history of the group. The differences in AARs between dispersal scenarios D1 and D2 were small, for example, a possible European-North American ancestor for Ailanthus and clade III in D2 (Table 4-4). However, AAR of almost all nodes was the sa me in the optimal re construction between D1 and D2. As expected, the inclusion of more stringent dispersal probabilities in D2 produced a lower overall likelihood score and much higher estimated rates of dispersal and extinction. The impact of more stringent dispersal was s een in very few clades, but nota bly for dispersals between North

PAGE 90

90 and South America in Holacantha and Castela due to the reduced pr obability of movement between these two areas in D2 (data not shown). The fact th at both scenarios produced nearidentical AARs leads us to believe that only extr eme changes in dispersal probabilities, such as a reduction to zero between areas, wi ll result in AARs that affect our interpretation of the data. Because of the similarities between D1 and D2, results for the more relaxed dispersal probabilities in D1 (which had a much higher likelihood scor e) are illustrated in Figure 4-4 and discussed in the context of historical biogeography. Ultimately we must assess whether the modelbased approach is a valid one, given the ambiguity in the results and the number of model parameters that have been introduced to fit the data. In any likelihood analysis th ere is a trade-off between a model too simplistic to describe the data accurately, and one that is overly complex, leading to overconfidence in one particular result. Ambiguity is seen in AARs for all m odels, and there are differences among the five models (Table 4-4) and ancestral range permutations (Table 4-5), especially for deeper nodes. This ambiguity is likely a reflection of a simplis tic model attempting to describe the multitude of historical events that have shaped the familys history. Despite this lim itation, there is an overall biogeographical signal concurrent among all of th e different models tested. Optimal AARs are congruent among models for lineages diverging in th e Late Oligocene and later, and the signal is one of multiple recent range shifts. These recent shifts appear to overshadow reconstruction of events deeper in the familys history, and as a result, Paleocene, Eocene and Early Oligocene reconstructions are much more sensitive to th e specification of the model. However, when alternative scenarios within each model have very similar likelihoods, the areas involved are typically the same (Table 4-4). It is how thes e areas are inherited, whether through vicariance, incipient speciation or dispersal followed by exti nction, that distinguishe s the alternative AARs.

PAGE 91

91 The likelihood method encourages a more fluid approach to geographic range evolution and historical biogeography, a discip line that has traditionally been dominated by the vicariancedispersal dichotomy. DIVA produced results similar to Lagrange when ancestral ranges were confined to two areas, but South America was more commonly involved in DIVA AARs than in likelihood analyses (Table 4-4). Without the maxareas c onstraint, most deeper nodes in the phylogeny had implausibly widespread distributions. Some improbable ancestral ranges were observed, such as SA and FU, demonstrating the limitation of this method in a complex system. Also there is no objective method to favor one reconstruction over another at any particular node. Freedom from temporal constraints (Donoghue and Moore, 20 03) and limited ancestral ranges removes extinction as a factor in shaping distributions, as alternative vicariant events will always be cheaper in the DIVA cost matrix. Origin and early history of Simaroubaceae Given the small size of Simaroubaceae (ca. 109 s pp.), which is more comparable to a large genus than previously studied tropical families (Melastomataceae: 3000 spp.; Malpighiaceae:1200 spp., Meliaceae: 550 spp.; Bu rseraceae: 500 spp.; Moraceae: 1500 spp. including temperate elements; Annonaceae: 2300 spp.; Dalbergioid legumes: 1100 spp. (species numbers from Judd et al., 2007; Lavin et al., 2000)), the complex reconstruction of biogeographical events proposed, and the resulting extant geographical heterogeneity, is perhaps more surprising. This is especially true given th e origin of the family in the Late Cretaceous, a similar timeframe to many of the families me ntioned above. Additionally, although attempts were made to include extinct species in AAR, fu rther extinctions in the family are hypothesized, especially for depauperate and isolated lineages, such as Nothospondias (arising 46.7 [41.8,51.5]

PAGE 92

92 Ma, a single species in Africa) and Picrolemma (diverging 39.1 [33.7,44.4] Ma, two species in South America). Based on the results of models including foss il data, the family originated in North America, with expansion to Asia via Beringia early on in the Tertiary. This is hypothesized if we assume ancestral species were adapted to temperate climates, and several extant species from early-diverging clades retain this characteristic. However, if we assume ancestral species were tropical, we could rule out NA and SU as ancestr al ranges, and the resu lts for NF + NE (the highest likelihood ancestral range permutation without NA or SU) support movement via the NALB, which has been proposed for other tropical groups (Lavin et al., 2000; Renner et al., 2001; Davis et al., 2002; Richards on et al., 2004; Weeks et al., 2005; Zerega et al., 2005; Muellner et al., 2006). The origin of clade I is North American only; however, the relationship of Picrasma to Castela and Holacantha is not strongly supported in phylogenetic analyses (Chapter 3), which could impact AAR for the earliestdiverging nodes in the phylogeny. Clade III is hypothesized to have originated in Asia and North America, with subsequent dispersals to Africa and SE Asia. The presence of Nothospondias in Africa is the result of a migration via the NALB followed by local extinction in North Ameri ca, although a later epis ode of long-distance dispersal cannot be ruled out. Similarly, Picrolemma dispersed to South America with subsequent local extinction, possibl y via island chains, or potentially much la ter after the closure of the Panamanian Isthmus. Long-distance dispersal The break-up of Gondwanaland, traditionally th e most parsimonious explanation for tropical disjunctions (Raven and Axelrod, 1974), has now been replaced by hypotheses incorporating knowledge of phylogenetic relation ships, more realistic divergence dates, availability of dispersal routes (Morley, 2003), and a general acceptance of long-distance

PAGE 93

93 dispersal as a major driving for ce in extant plant distributions (Givnish and Renner, 2004; Lavin et al., 2004; Renner, 2004a; De Queiroz, 2005) The African-Mesoamerican/South American disjunctions in Simaroubaceae mostly occurred af ter the NALB was viable for tropical groups (Tiffney, 1985a). The ancestor of Quassia amara and Q. africana supposedly maintained a transAtlantic disjunction for 22 Myr, which underwent vicariance, an unlikely scenario without the NALB for much of that period. NF and NE ancesto rs are unlikely in the Miocene when tropical vegetation was restricted to lo wer latitudes; therefore, long-di stance dispersal is the best explanation for these disjunctions, and disper sals between Africa and South America are hypothesized for a number of well-studied clades (Lavin et al., 2004; Renner, 2004a; De Queiroz, 2005). Leitneria may be another candidate for tran s-Atlantic or trans-Pacific longdistance dispersal, but this is dependent on how the fossils are integrated, and if they are considered part of the extant lineage. In our in terpretation, migration via Beringia is most likely, but the presence of fossils in Europe and Russia, with no definitive fossils in North America, might suggest a late arrival to North America via long-di stance dispersal. Disjunctions between mainland Asia and Nort h America occur in a number of temperate genera (Gray, 1859; Li, 1952; Donoghue et al., 2001; Xiang a nd Soltis, 2001; Donoghue and Smith, 2004), with phylogenetic relationships be tween the two areas sh owing a variety of patterns. The disj unction seen in Picrasma between two Asian/SE Asian species and six Central and South American and Caribbean species, is less common (e.g. Magnolia Azuma et al., 2001; Ehretia Gottschling et al., 2004; Hedyosmum Zhang and Renner, 2003). Reconstructions suggest Picrasma arrived in Asia from North America, but the timing of this migration is unknown, as ancestral Picrasma could have maintained a North American-Asian ancestral range beginning at any time along the stem branch of the genus. This ancestral distribution may be

PAGE 94

94 reflected in fossils of Chaneya which are widespread across the Northern Hemisphere (Fig. 1). Trans-Beringial migration is plausible for c ool-adapted species (Sanmartin et al., 2001; Donoghue et al., 2001), and Picrasma quassioides is a temperate species that occurs as far north as northern Japan. Range expansion into the As ian tropics may have prompted the incipient speciation of P. javanica with the ancestral NA range eventually undergoing a vicariant split ~9.9 [6.1,14.0] Ma, and the New World species moving south. Alternatively, a long-distance dispersal event by the ancestor of P. quassioides could be inferred, but this would be more plausible if it were adapted to a climate sim ilar to that inhabited by the New World species. Dispersal between Africa and Asia has been pr oposed for clades such as Crypteroniaceae (Conti et al., 2002) under the hypothesis of ra fting on the Indian sub-continent, and Exacum (Yuan et al., 2005) and Melastomataceae (Renne r, 2004b) through long-distance dispersal. Eurycoma nested within the African species, and sister clades Brucea and Soulamea fit with the idea of long-distance dispersal, given the significant ocean barrier betw een Africa and mainland and SE Asia during the Oligocen e (Scotese, 2001; Morley, 2003), coupled with climatic cooling. Long-distance dispersal to Madaga scar has been suggested for many taxa, both animals (Poux et al., 2005) and plants (e.g. Adansonia (Baum et al., 1998), Exacum (Yuan et al., 2005), Nepenthes (Meimberg et al., 2001), Melastomataceae (Re nner, 2004b)). Simaroubaceae are represented on Madagascar by two species, both of which arri ved there by long-distan ce dispersal in the Miocene. With the presence of numerous is land chains among mainland Asia, the SE Asian islands and Australia during la ter epochs (Meimberg et al ., 2001; Cannon and Manos, 2003; Morley, 2003), island-hopping dispersal explains the widespread distri butions of species of Ailanthus Picrasma Brucea, and Samadera. Soulamea like a number of other genera e.g. New Caledonian Sapotaceae (Bartish et al., 2005), Cyrtandra (Cronk et al., 2005), and Weinmannia

PAGE 95

95 (Bradford, 2002), has readily dispersed between New Caledonia and surrounding islands, notably Fiji and Tuvalu, reaching as far as the Seychelles ( S. terminalioides ). In the New World, Castela and Holacantha show evidence of multiple dispersals between North and Central and South America, w ith speciation on several Caribbean islands and the Galpagos. This pattern reflects the amphitr opical disjunctions of ot her dry-adapted clades such as Tiquilia subg. Tiquilia (Moore et al., 2006) and Hoffmannseggia (Simpson et al., 2005). New World disjunctions are also seen in Simarouba and Picrasma and both genera have at least three species distributed on Caribbean islands, plus Simarouba glauca in south Florida, implying one or more overwater dispersal events within the last ten million year s (Lavin et al., 2003; Santiago-Valentin and Olmst ead, 2004; Morris et al., 2007). There is no obvious link between dispersabi lity and diaspore size in Simaroubaceae, especially in the tropical clad es, which tend to have drupaceous, birdor mammal-dispersed fruits (e.g. Hardesty et al., 2005). The fruits of Castela Picrasma and Simarouba are small, birddispersed drupes, and so north-south dispersal ma y be facilitated by the migratory patterns of fruit-eating birds in the New World. Simila rly, dispersal of the large fleshy drupes of Gymnostemon and Perriera between mainland Africa and Mada gascar, may involve migrating birds (Renner, 2004b). Wind-dispersed Ailanthus was widespread across the Northern Hemisphere based on fossil evidence (Cor bett and Manchester, 2004), and extant A. altissima is a weedy species well known for its dispersability (Benvenuti, 2007). The fruits of Soulamea and Samadera indica are dry, with an air cav ity allowing them to float (Nooteboom, 1962b), which potentially explains the widespr ead distribution of these taxa around the Indian Ocean basin, in conjunction with ocean currents, such as monsoon circulation seen today (Schott and McCreary,

PAGE 96

96 2001). Leitneria which inhabits swamp and coastal forest s, also has floating fruits, which could have facilitated an oceanic migration to the New World from Europe or Asia.

PAGE 97

97Table 4-1. Fossil calibrations used in BE AST analyses, for testing compatibility be tween fossils, and constructing the final chronogram for biogeographic analyses. Calibration point Fossil Prior distribution Mean / SD of lognormal distribution Hard lower bound / Mean / Soft upper bound (95%) (Ma) A Ailanthus confucii lognormal 1.0 / 0.5 52.2 / 55.3 / 58.4 B Ailanthus gigas lognormal 1.0 / 1.0 28.4 / 32.9 / 42.5 C Leitneria spp. lognormal 1.5 / 0.75 28.4 / 34.3 / 43.8 D Chaneya spp. lognormal 1.0 / 1.0 40.4 / 44.9 / 54.5 Table 4-2. Divergence dates (in Ma) resulting from molecular rates analyses for eight different calibration schemes. Numbers in bold are dates for calibrated nodes, using pr iors described in Table 4-1. Lower and upper 95% highest posterior densities are shown in square brackets. The latter four nodes are exam ples taken from the phylogeny for comparative purposes. Ailanthus stem group (A. confucii fossil) A. altissima divergence ( A. confucii fossil) A. integrifolia divergence ( A. gigas ) Leitneria divergence Picrasma stem group (Chaneya fossil) Picrasma crown group Clade V crown Family crown Quassia crown 1 55.1[52.9,58.1] 22.3[14.5,30.1] 17.5[10.2,25.1] 42.8[36.8,48.0] 68.6[59.3,79.3] 14.3[9.4,19.9] 20.9[17.0,25.4] 71.0[62.7,79.8] 13.6[7.9,19.9] 2 139.3[95.0,190.0] 55.2[52.9,58.3] 43.3[30.2,54.0] 107.6[71.1,148.4] 173.6[112. 8,236.8] 36.2[20.2,54.6] 53.0[35.6,73.7] 179.5[118.1,243.1] 34.1[16.3,52.2] 3 68.9[53.7,84.3] 35.0[29.6,41.9] 30.4[28.5,33.5] 53.2[40.5,66.3] 84.3[65.0,104.0] 18.5[11.1,26.8] 26.3[19.4,33.4] 87.3[67.9,105.3] 17.0[8.4,25.6] 4 55.8[53.0,59.7] 33.2[29.2,37.7] 30.0[28.5,32.3] 43.3[37.3,49.2] 70.2[58.4,81.6] 15.0[9.9,21.0] 21.8[17.9,26.2] 72.7[63.5,83.4] 14.1[7.9,20.4] 5 43.1[35.0,53.9] 17.3[10. 6,25.0] 13.5[7.1,20.4] 33.3[28.8,40.4] 53.7[41.6,69.0] 11.1[6.9,16.0] 16.4[12.4, 21.5] 55.5[43.5,70.3] 10.6[5.8,16.2] 6 35.2[27.8,42.9] 14.1[8.8,19.9] 11.0[6.2,16.3] 27.2[20.9,33.8] 44.0[40.5,51.0] 9.2[5.6,13.1] 13.5[10.2,17.3] 45.4[40.5,53.6] 8.6[4.8,12.9] 7 54.9[52.8,57.4] 21.9[14.3,29.8] 17.1[10.3,25.1] 40.7[34.3,46.4] 67.4[57.8,77.8] 14.2[9.2,20.0] 20.6[16.8, 24.6] 69.8[62.1,78.6] 13.4[7.7,19.4] 8 55.3[53.0,58.4] 33.1[29.1,37.6] 30.0[28.5,32.3] 41.2[34.5,47.4] 69.1[58.7,81.5] 15.0[9.6,20.6] 21.6[17.7, 26.2] 71.5[63.0,82.2] 13.9[7.8,20.4]

PAGE 98

98Table 4-3. Dispersal network model implemented in Lagrange, showing probabilities of dispersal betw een areas for two different scenarios (see text), based on data from Tiffney (1985a), Morley (2003) and Scotese ( 2001). Probabilities are for movement in both directions. Area connections Time period (Ma) D1D2 All overwater dispersals without island chains 70 0 0.1 0.01 Areas not adjacent between 70Ma and present 70 0 0.1 0.01 North America South America 70 45 0.25 0.01 45 5 0.75 0.01 5 0 1.0 1.0 North America Europe 70 30 0.75 0.1 30 0 0.1 0.01 North America Mainland and SE Asia 70 30 0.5 0.1 30 0 0.1 0.01 North America Africa 70 30 0.5 0.1 30 0 0.1 0.01 South America Australia 70 45 0.5 0.1 45 0 0.1 0.01 Africa Madagascar 70 0 0.5 0.1 Africa Mainland and SE Asia 70 45 0.75 0.1 45 30 0.75 0.01 30 0 0.1 0.01 Africa Europe 70 30 0.75 0.1 30 5 0.5 0.01 5 0 0.25 0.01 Mainland and SE Asia Australia 70 45 0.1 0.01 45 30 0.5 0.01 30 0 0.75 1.0 Mainland Asia Europe 70 0 1.0 1.0

PAGE 99

99Table 4-4. Results for biogeographic models tested, including lik elihood scores (-lnL) and estima tes of dispersal (D) and extin ction (E) rates (events per Myr). In AARs for example nodes, areas shown indicate the range inhe rited by the daughter lineages. In cases where two ranges are separated by a bar, the first area is inherited by th e upper branch on Figure 4, the second area is inherited by the lower branch. Relative probability of each AAR is shown in parentheses for multiple reconstructions. For DIVA results, all equally optimal recons tructions are shown, separated by commas. Model Constraints -lnL D E Examples of reconstruc tions for ancestral taxon of the following clades: Whole family Clade I Sister clade to clade I Ailanthus Clade III Nothospondias Clade V M1 Ancestral areas restricted (see text) 137.3 0.011 0.005 N (0.24) NA | N (0.23) A | NA (0.15) A | N (0.13)a N (0.59) N | NA (0.17) NA | A (0.29) A (0.29) A | NA (0.09) N (0.08)a A (0.64) A | NA (0.13) N (0.09) A (0.49) A | NA (0.31) A | N (0.09) N (0.46) A (0.24) N | F (0.47) N | A (0.35) NF | F (0.07) M2 Ancestral areas unrestricted; dispersal probabilities stratified by age using D1 133.2 0.023 0.000 NSFAU | N (0.06) SFAU | N (0.06) NSAU | N (0.03) SFAU | S (0.03)a N (0.58) S (0.24) F | NAE(0.10) F | NA (0.05) N | NA (0.03) NSFU | A (0.03)a A | NA (0.45) A (0.18) A | NE (0.17) A | NAE (0.15) A | NA (0.45) A | NE (0.17) A | NAE (0.15) F | NE (0.10)a F (0.45) SFU | F(0.09) NSFU | F (0.07) NF | F (0.07) NSF | F(0.35) F | FA (0.32) SF | F (0.16) S | FA (0.06) M3 Full constraint (M1+M2); dispersal probabilities set to D1 143.9 0.045 0.005 NA | N (0.31) N (0.28) A | NA (0.11) N | NA (0.07)a N (0.65) N | NA (0.23) NA | A (0.34) N | NA (0.21) N | A (0.10) A (0.09)a A (0.51) A | NA (0.28) A | N (0.13) A | NA N (0.53) NF | F (0.18) F (0.10) F | FA (0.63) N | A (0.14) NF | F (0.12) M4 Full constraint (M1+M2); dispersal probabilities set to D2 203.6 0.087 0.011 NA | N (0.34) N | NS (0.25) A | NA (0.13) N | NA (0.07)a N (0.46) N | NA (0.24) NS | S (0.19) NA | A (0.28) N | NA (0.23) N (0.13) A | NA (0.09)a A | NA (0.44) A (0.36) E | NE (0.08) A | NA (0.70) FA | A (0.18) E | NE (0.12) NF | F (0.53) N (0.25) F (0.18) N | A (0.44) F | FA (0.29) NF | F (0.22) M5 Full constraint (M1+M2) with fossils excluded; dispersal probabilities set to D1 120.5 0.065 0.006 A | NA (0.20) A | N (0.18) N (0.17) NA | N (0.15)a N (0.56) N | NA (0.26) A (0.22) NA | A (0.22) F | A (0.22) N | A (0.18)a A A | N (0.21) F (0.15) A (0.15) F | N (0.11)a F (0.40) N (0.27) NF | F (0.11) F | FA (0.64) NF | F (0.13) N | A (0.12) DIVA 1 Unconstrained NSFAUE N, S, NA, SA, NSA FAUE, NFAUE, SFAUE, NSFAUE A UE, NUE, FUE, NFUE, AUE, NAUE, FAUE, NFAUE F, SF F, SF DIVA 2 Maxareas = 2 A, NA, SA N, S, NA, SA A A A F, SF F, SF a Further alternative reconstruc tions (not shown) are within 2 log likelihood units of the optimal reconstruction.

PAGE 100

100Table 4-5. Ancestral range permutations for the chosen model (M3), ordered by decreasing likelihood. Likelihood scores (-lnL), and estimates of dispersal (D) and extinction (E) rates (eve nts per Myr) are shown. For example nodes, AARs follow the format of Table 4-4; only the AAR with th e highest relative probability is shown. Ancestral ranges -lnL D E Examples of reconstructio ns for ancestral taxon of the following clades: Whole family Clade I Sister clade to clade I Ailanthus Clade III Nothospondi as Clade V NA + NF 142.8 0.047 0.005 N (0.37) N (0.74) NA | A (0.36) A (0.55) A | NA N (0.63) F | FA (0.61) NA + NF + SU 142.9 0.047 0.005 N (0.37) N (0.74) NA | A (0.36) A (0.55) A | NA N (0.63) F | FA (0.61) NA + NE + NF 143.8 0.045 0.005 NA | N (0.31) N (0.66) NA | A (0. 34) A (0.51) A | NA N (0.53) F | FA (0.63) All possible a 143.9 0.045 0.005 NA | N (0.31) N (0.65) NA | A (0. 34) A (0.51) A | NA N (0.53) F | FA (0.63) NE + NF + SU 152.0 0.072 0.011 N | S (0.44) S (0.48) NE | E (0.25) E (0.31) E | NE (0.66) N (0.53) F | FA (0.48) NE + NF 154.4 0.074 0.013 N (0.37) N (0.58) NE | E (0.22) E (0.34) E | NE (0.62) N (0.47) F | FA (0.44) NA + NE 156.1 0.063 0.010 N (0.31) N (0.63) NA | A (0. 23) A (0.40) A | NA N (0.20) N | A (0.25) NA + NE + SU 156.2 0.063 0.010 N (0.31) N (0.62) NA | A (0.23) A (0.40) A | NA (0.70) N (0.20) N | A (0.25) NA 158.8 0.056 0.008 N (0.37) N (0.74) NA | A (0.41) A (0.59) A | NA (0.83) A (0.53) NA | A (0.46) NA + SU 158.9 0.055 0.008 N (0.37) N (0.73) NA | A (0.41) A (0.59) A | NA (0.83) A (0.53) NA | A (0.45) NE + SU 161.4 0.092 0.016 N | S (0.39) S (0.44) NE | E (0 .29) E (0.33) E | NE E (0.34) NE | E (0.49) NE 163.3 0.095 0.018 N (0.44) N (0.60) NE | E (0.2 7) E (0.35) E | NE E (0.34) NE | E (0.50) NF + SU 170.6 0.072 0.011 A | U (0.31) S (0.28) NF | F (0.38) A | FA (0.29) F | NF (0.53) F (0.47) F | FA (0.48) NF 177.5 0.069 0.012 F (0.29) F (0.35) F | FA (0.26) A (0.34) F | NF (0.38) F (0.63) F | FA (0.49) SU 206.0 0.121 0.022 S (0.24) S (0.44) U (0.48) U (0.76) U | SU (0.50) U (0.51) U | A a Equivalent to M3 in Table 4-4.

PAGE 101

101 Figure 4-1. Map showing extant geographic distribution and approximate fossil localities for Simaroubaceae. Colors and letters refer to the seven areas assigned to terminals in biogeographic analyses and shown in Figure 4-4. N

PAGE 102

102 Figure 4-2. Phylogram of the majority rule consensus of trees for Simaroubaceae, sampled from the posterior distribution in the Bayesian analysis partitioned by gene and codon position (combination of two runs of 10,000,000 generations, sampled every 1000 generations; burn-in = 150,000 generations). Posterior probabilities (PP) are shown on branches. Branch lengths are proportional to the mean number of nucleotide substitutions per site, calculated from the posterior distribution density. Simaba glabra Simaba glabra Simaba suffruticosa Simaba cuneata Simaba cuneata Simaba moretti Simaba paraensis Simaba cedron Simaba trichilioides Simaba ferruginia Simaba insignis Simaba orinocensis Simaba orinocensis Simaba guianensis Simaba guianensis Simaba guianensis Simaba polyphylla Simaba guianensis Simarouba glauca Simarouba glauca Simarouba tulae Simarouba berteroana Simarouba amara Simarouba versicolor Pierreodendron africanum Hannoa undulata Hannoa undulata Hannoa chlorantha Hannoa sp. A Hannoa sp. B Hannoa klaineana Gymnostemon zaizou Perriera madagascariensis Eurycoma apiculata Eurycoma longifolia Odyendea gabonensis Samadera bidwillii Samadera sp. B Samadera sp. C Samadera baileyana Samadera indica Quassia africana Quassia amara Picrolemma sprucei Picrolemma sprucei Nothospondias staudtii Brucea antidysenterica Brucea ferruginia Brucea guineensis Brucea javanica Brucea mollis Soulamea amara Soulamea terminalioides Amaroria soulameoides Soulamea morattii Leitneria floridana Ailanthus altissima Ailanthus altissima Ailanthus integrifolia Ailanthus triphysa Ailanthus fordii Castela erecta Castela tortuosa Castela erecta Castela retusa Castela tweedii Castela coccinea Holacantha emoryi Picrasma antillana Picrasma excelsa Picrasma crenata Picrasma quassioides Picrasma javanica Casimiroa edulis Zanthoxylum sp. Cneorum tricoccon Melia azedarach Unknown Meliaceae Swietenia macrophylla Acer saccharum0.005 substitutions/site0.991 1 1 1 1 1 1 1 1 1 1 1 10.89 0.570.70 0.730.99 0.99 0.99 0.97 0.42 0.320.91 0.970.51 0.840.680.310.740.99 0.88 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

PAGE 103

103 Figure 4-3. Chronogram with 95% HPD bars, based on BEAST analyses using two fossil calibrations, Ailanthus confucii and Leitneria sp., at nodes indicated with an open bar. Nodes labelled A, B, C and D refer to the positions of the four fossils used to assess different calibration schemes (Table 4-1). Geological time scale (Gradstein and Ogg, 2004) is shown at the bottom, and major clades of Simaroubaceae are indicated on the right hand side. I V IV III II OutgroupsSimarouba versicolor Hannoa chlorantha Perriera madagascariensis Simaba glabra Samadera bidwillii Simabaorinocensis Hannoa undulata Simaba paraensis Simaba guianensis Pierreodendron africanum Samadera sp.B Simarouba berteroana Simaba insignis Samadera sp.C Simaba cedron Simaba cuneata Simaba glabra Simaba moretti Simaba trichilioides Eurycoma longifolia Simaba ferruginia Eurycoma apiculata Samadera indica Simarouba tulae Gymnostemon zaizou Simaba cuneata Simarouba amara Simabaorinocensis Simaba guianensis Hannoa undulata Simabapolyphylla Samadera baileyana Simaba guianensis Hannoa sp.B Quassia africana Odyendea gabonensis Simabaguianensis Simarouba glauca Simarouba glauca Simaba suffruticosa Hannoa klaineana Hannoa sp.A Castela erecta Brucea ferruginia Acer saccharum Picrasma quassioides Leitneria floridana Zanthoxylum sp. Unknown Meliaceae Nothospondias staudtii Brucea guineensis Picrasma antillana Casimiroa edulis Holacantha emoryi Castela tweedii Brucea mollis Ailanthus integrifolia Ailanthus altissima Ailanthus fordii Soulamea morattii Picrasma javanica Brucea antidysenterica Ailanthus altissima Castela retusa Soulamea amara Picrolemma sprucei Picrolemma sprucei Ailanthus triphysa Picrasma crenata Brucea javanica Melia azedarach Castela coccinea Swietenia macrophylla Quassia amara Amaroria soulameoides Castela erecta Picrasma excelsa Castela tortuosa Soulamea terminalioides Cneorum tricoccon OligoMiocene Late CretaceousPaleoEocene Pli Early CretaceousA B C D A

PAGE 104

104 Figure 4-4. Ancestral area reconstruction for Simaroubaceae using model M3, with ancestral ranges NS, NF, NA, FM, FA, FE, AU, AE and MAU, and stratified dispersal probabilities between areas (D1). AARs with highest likelihood are shown as colored boxes at each node. Single area boxes indicate an ancestor confined to a single geographic area; combined boxes indicate an ancestor with a distribution encompassing two or more areas; two boxes separated by a space indicate the ancestral ranges inherited by each of the daughter lineages arising from the node. For nodes with alternative reconstructions (within 2 log likelihood units of the maximum), the relative probability of the global likelihood for the optimal reconstruction is given. Three modes of range inheritance (range expansion, local extinction and vicariance by cladogenesis) are indicated as symbols on branches in the phylogeny (see Results). For clarity, selected sister terminals from the same single area have been pruned from the chronogram. Fossil lineages are shown with a dotted line. Geological time scale (Gradstein and Ogg, 2004) is shown at the bottom, and major clades of Simaroubaceae are indicated on the right hand side. Quassia amara Picrasma crenata Picrasma javanica Leitneria fossil Picrolemma sprucei Simarouba glauca Simaba cedron Gymnostemon Castela retusa Hannoa klaineana Simaba ferruginia Simaba paraensis Perriera Soulamea amara Samadera baileyana Simarouba tulae Picrasma quassio. Odyendea Simaba orinocensis Brucea mollis Castela coccinea Holacantha emoryi Samadera bidwillii Simarouba amara Eurycoma apiculata Ailanthus fossil Simaba cuneata Nothospondias Brucea guineensis Soulamea termina. Simaba guianensis Eurycoma longifolia Samadera indica Ailanthus integ. Ailanthus altissima Ailanthus fordii Quassia africana Castela erecta Picrasma antillana Simaba glabra Simarouba versicolor Pierreodendron Ailanthus triphysa Castela tweedii Brucea javanica Leitneria floridana Soulamea morattii Picrasma excelsa Amaroria soulame. S N F M A U ENorth America, Central America, Caribbean islands South America Europe Australia, New Caledonia and Pacific islands Mainland Asia and insular SE Asia Madagascar Tropical Africa S S S F F A A A A S S S S S S S S S M F N S N S A A S F F F A F M A A S N N N N S A N N N U U A A M U A U A U U U A E A U N A E N S A U 0.37 0.36 0.49 A A U U0.47 U A U M 0.77 A U M0.75 A N N0.63 0.44 F N F A0.61 0.61 F A F0.76 F F M0.67 A F0.67 A N F N F N N N0.74 0.67 S0.32 A U U0.54 F N N N0.69 S N S N N S S N 0.70 0.83 Dispersal resulting in range expansion Local extinction Vicariance by cladogenesis U XX X X PlioceneMiocene Oligocene Eocene Paleocene F I V I V III II0.74 N A N A A N S N A A A0.55 N A0.74 F A F A 0.82 S N S N S N S S N N A A F A F A0.64 0.71 0.81 0.51 S N 0.53

PAGE 105

105 CHAPTER 5 DIVERSIFICATION AND MORPHOLOGICAL EVOLUTION IN SIMAROUBACEAE Introduction W iens and Donoghue (2004) suggest there is a large void between ecology and historical biogeography in the interpretation of biodive rsity gradients. Although studies encompassing m ajor biomes (Cracraft, 1985; Dick and Wr ight, 2005; Pennington a nd Dick, 2004; Pennington et al. 2004a; 2004b; 2006; Mittlebach et al., 200 7) provide an overview, the origins of biodiversity cannot be explored without detailed studies of clades with well-established phylogenetic relationships. The underlying causes of biodiversity are often a complex interaction of factors (de Queiroz, 2002; Lavin et al ., 2004; Wiens and Donoghue, 2004), and so pinpointing specific determinants may be best suited to a data exploration approach. In the context of a clade such as Simaroubaceae, this can be done by detecti ng shifts in species diversification rates within a phylogenetic framework, and examining biological correlates to these sh ifts. Diversification rate shifts, that is, net changes in rates of speciation or extinction between clades in a phylogenetic tree, can be considered both in a temporal (Nee et al., 1992; Pybus and Harvey, 2000; Magalln and Sanderson, 2001) and a topol ogical (Chan and Moore, 2002) context. The correlates of species diversity can be historical, ecological an d intrinsic to the organism (Cracraft, 1985; Herrera, 1989; Er iksson and Bremer, 1992; Dodd et al., 1999; Lavin et al., 2004; Pennington and Dick, 2004; Penni ngton et al. 2004a; 2004b; 2006; Dick and Wright, 2005; Wiens and Donoghue, 2004), and for the present st udy this translates to two identifiable variables: dispersal into new environments (and subsequent adaptive radiation across environmental gradients), a nd key morphological innovation. Reconstruction of the biogeographic history of Simaroubaceae (C hapter 4) has allowed the identification of long-distance, island-hopping and overland dispersa l events ancestrally within

PAGE 106

106 specific clades. Dispersal events are significant because they drive cladogenesis, and the isolation of one of the daughter lineages in the newly colonized environment may or may not promote diversification. However, ecologica l adaptation is manifest in morphological innovation, and even subtle changes in local en vironmental factors, rather than a major geographical isolation event, can cause niche sp ecialization and promote speciation (Funk et al., 2006), for example, by diversifying floral form to take advantage of pollinator diversity, resulting in reproductive isol ation (Hodges and Arnold, 1995; Wase r, 1998; Kay et al., 2005; von Hagen and Kadereit, 2003). Thus we might expect to find links to divers ification rate shifts between either morphological char acters considered important to niche specialization, such as reproductive morphology, or a histori cal dispersal event, or both, for a given lineage in which a diversification rate shift is det ected. Historical dispersal events and environmental variation have been linked to patterns of diversification in studies of specific clades (Richardson et al., 2001a, 2001b; von Hagen and Kadereit, 2003; Becerra, 2005; Moore and Donoghue, 2007; Alfaro et al., 2007; Wiens et al., 2007), and even experimentally in a controlled environment (Fukami et al., 2007). Similarly, morphological correlates to specie s diversity have been examined for a number of angiosperm groups (Herrera, 1989; Eriksson and Bremer, 1991; Eriksson and Bremer, 1992; Dodd et al., 1999; Good-Avila et al., 2006), a commonly cited exam ple being the evolution of nectar spurs (Hodges and Arnold, 1995; Ree, 2005). Testing hypotheses of key innovations has only recently been undertaken with a more cr itical approach in a phylogenetic framework (de Queiroz, 2002; Ree, 2005; Leschen and Bu ckley, 2007; Moore and Donoghue, 2007), to attempt to overcome the problems associated with simp le measures of clade diversity through species numbers. When parsing out the relative effects of the two major factors hypothesized to be involved in diversification rate in creases, there is a subsequent i ssue of contingency (de Queiroz,

PAGE 107

107 2002; Davies et al., 2004), for example, the contribution of innovative dispersal mechanisms to future chances of dispersal (Moore and Donoghue, 2007), or rapid morphological diversification of the daughter population from the parent population in order to take advantage of new ecological opportunities (von Hagen and Kadereit, 2003; Moore and Donoghue, 2007). This study will take a data exploration approach to address the following questions: Does the phylogenetic history of Simaroubaceae show significant shifts in diversification rates between lineages? Do any significant diversification rate shifts correlate to morphological character-state changes, historical biogeographic movements, both of these factors, or neither? As a necessary precursor to answering these questions, an analysis of morphological variation and evolution in Simaroubaceae was undertaken. Morphological data extracted from the literature and herbarium material was firstly used to determine if any phylogenetic signal is present for the selected characters. Particular consideration was given to coding of the characters. These were compared to the molecular phylogeny for the family (Chapter 3) to see which clades are supported by morphological data. A second analysis was performed on a combined matrix of molecular and morphological data. The purpose of this was to see if the morphological data have any impact on the molecular phylogeny, in changing topology and support for clades (Pennington, 1996; Nylander et al., 2004). It also serves to determine the phylogenetic placement of numerous taxa for which no molecular data have been obtained, particularly the unsampled monotypic genera Laumoniera and Iridosma. Potential morphological synapomorphies for major clades of Simaroubaceae were also identified. Selected morphological traits are used to reconstruct character evolution in Simaroubaceae, to examine the lability of various traits deemed of putative evolutionary significance, and their pattern of inheritance. A number of different techniques for diversification rate analysis have been applied to the previous molecular phylogeny, and to phylogenetic data obtained in this study. Ultimately, any diversification rate

PAGE 108

108 shifts are examined in the context of character-state transitions and historical biogeographical events (Chapter 4). Methods Phylogeny Estimation and Character Evolution Morphological and anatomical characters were recorded from species descriptions in the literature (Trelease, 1895; Engler, 1931; Aubrville and Pellegrin, 1937; Abbe and Earle, 1940; Cronquist, 1944a,b,c,d; Perrier de la Bathie, 1950; Hutchinson and Dalziel, 1954; Gilbert, 1958; Aubrville, 1962; Nooteboom, 1962a, 1962b, 1987; Basak, 1963, 1967; Wild and Phipps, 1963; Backer and Van den Brink, 1965; Porter, 1971, 1973; Li, 1977; Cavalcante, 1983; Feuillett, 1983; Thomas, 1984; Hewson, 1985; Smith, 1985; Thomas, 1985; Moncada and Machado, 1987; Fernando and Quinn, 1992; Pirani, 1987, 1997; Franceschinelli and Yamamoto, 1999; Franceschinelli et al., 1999; Franceschinelli and Thomas, 2000; Stannard, 2000; Hahn and Thomas, 2001; Jaffre and Fambart, 2002; Thomas, 2002) and 120 herbarium sheets were examined (see Appendix B and C) macroscopically and with a binocular microscope. The tabulated characters were then converted into a data matrix of coded characters in MacClade v.4.08 (Maddison and Maddison, 2004). The following criteria were used for coding characters in discrete states for maximum parsimony and Bayesian analysis: Species were considered the operational taxonomic units (OTUs) in the matrix. Character data were sythesized from multiple species in the literature and herbarium specimens. For continuous characters the total variation was used (maximum and minimum values), and variation in meristic and discrete characters was coded as ambiguous. See Appendix D for details. If more than 11 of the 22 genera had missing data for a character, it was not used. Continuous characters were only included if two or more non-overlapping character states could be assigned. Non-overlapping was defined as any gap between the highest and lowest measurements of the character. See Appendix D for details. All characters were coded as unordered.

PAGE 109

109 All characters had equal weight. Character states were coded as missing, rather than absent, if they are dependent on a second character in the matrix which is also absent, to avoid non-independence e.g. petal characters in Leitneria, which has no petals. Morphological characters were not coded for outgroups, because of the difficulty involved in selecting suitable outgroup taxa. Both Rutaceae and Meliaceae are families with a large amount of morphological variation, even more so than Simaroubaceae, and the position of the root of Simaroubaceae would be biased towards whichever taxa were selected as outgroups. Reconstruction of an ancestral Rutaceae or Meliaceae taxon, that encompassed the ancestral traits of these families, would be the most suitable outgroup in this case, but these data are unavailable. Therefore, for morphological analyses, Picrasma and Castela+Holacantha were together designated as a functional outgroup, based on their position as sister to the rest of the family in previous molecular analyses (Chapter 3), and each genus was also tested individually as an outgroup. A maximum parsimony (MP) analysis was conducted on the morphological data alone for 105 taxa. The analysis was performed in PAUP* 4.0b10 (Swofford, 2002), using a heuristic search of 100 random addition replicates with TBR branch swapping, MulTrees in effect, and saving all trees, with a maxtrees limit set to 1000 trees for each replicate. A second heuristic search was made using the shortest trees resulting from the first search, with a maxtrees of 100,000. ACCTRAN character optimization was used. A strict consensus of most-parsimonious trees was made, and support for clades was assessed by bootstrapping (Felsenstein, 1985), using 1000 bootstrap replicates with one random addition per replicate, maxtrees set to 1000 for each replicate, TBR branch swapping, MulTrees in effect. A Bayesian analysis of morphological data was also performed under the Mk1 model of character evolution (Lewis, 2001), that is, one symmetrical transition rate between character states, with a gamma rate distribution. The analysis

PAGE 110

110 was implemented in MrBayes v3.1.1 (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003). Two independent analyses each ran for 10 million generations using four Markov chains and with all other parameters at default values; trees were sampled every 1000th generation with a burn-in of 100,000 generations. Coding bias wa s set to variable for the morphological data, to account for the fact that no constant characters were included in the morphological matrix. Stationarity of the MCMC was determined by the average standard deviation of split frequencies between runs and by examination of the distribution of the posterior in Tracer v1.4 (Rambaut and Drummo nd, 2003). A majority-rule consensus tree was created using the posterior di stribution of trees. The resul ting estimations of phylogeny for morphology alone were compared to the molecular phylogeny for the family (Chapter 3) to see which clades, if any, were recovered by the morphological data. A second MP analysis was performed on a combined matrix of molecular and morphological data, using the same heuristic s earch criteria as the morphology alone, except with DELTRAN character optimization. Bootstrap support was assessed as above. A combined Bayesian analysis was also performed with th e molecular data partitioned by codon position and gene region using models of nucleotide evolut ion determined using the AIC in Modeltest (Posada and Crandall, 1998; see Chapter 3). Mo rphological data were included as a separate partition, modeled under the Mk1 model of charact er evolution (Lewis, 2001). Analyses were performed in MrBayes v3.1.1 (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003). Two independent analyses each ran for 40 million generations using four Markov chains and with all other parameters at default values; trees were sa mpled every 1000th generation with a burn-in of 10 million generations; model parame ters were unlinked across partitions, and the prior for the rate parameter was allowed to vary between partitions. Coding bias was fixed for

PAGE 111

111 molecular data and set to variable for the mo rphological data. Stationarity of the MCMC was determined as above. Character evolution was examined usi ng parsimony (Swofford and Maddison, 1987). Morphological characters were not investigated in these analyses if they were unique to fewer than four taxa, or were potent ially subject to phenotypi c variation. Also excluded were characters with significant amounts of missing data, and those in which coding was deemed too subjective, especially continuous quantitative traits. Character s of particular interest were those that might be considered evolutionarily significant (e .g. Herrera, 1989; Eriksson and Bremer, 1991, 1992; Waser, 1998; Dodd et al., 1999; Moore and Donoghue 2007), and characters that have evolved multiple times across the phylogeny. We therefore focused on reproductive/floral traits, dispersal biology and habit. A total of 22 ch aracters were examined, plus four additional characters not used in phylogeny reconstruction: pollination synd rome (insect vs. bird vs. wind), mode of dispersal (biotic vs. abiotic) and general ecol ogy, which was broken up into two characters temperate vs. subtropical/tropical and arid vs. mesophytic vs. swamp/riverine/coastal. Parsimony mapping of characters was performed in MacClade v.4.08 (Maddison and Maddison, 2004). Characters were mapped on a phylogeny that was ra ndomly chosen from most-parsimonious tree topologies of combined molecula r and morphological data that were compatible with a 70% majority-rule consensus and clades havi ng PP > 0.95 and BS > 70%. The outgroups were removed from this tree, along with four taxa that appeared sister to the rest of their respective genera due to a lack of morphological data: Picrasma selleana Castela jacquinifolia Simarouba laevis and Soulamea fraxinifolia Simaba cavalcantei was also removed as it appeared sister to Samadera likely due to a lack of informative characters. Although some have suggested circularity in mapping characters on a phylogeny which has been constructed using these

PAGE 112

112 characters (Coddington, 1988; Armbruster, 1992), the large amount of molecular data should limit the effects of non-independence (Luckow a nd Bruneau, 1997). Furthermore, the full data set provides the opportunity to use over 100 species. Characte rs were mapped using the MPR function in MacClade, and minimum number of unambiguous state changes, and number of most-parsimonious reconstr uctions were calculated. Shifts in Diversification Rates Temporal methods for detecting sh ifts in diversification rate ( r ) are numerous, and two of the most commonly used statistics were chosen. The first is the relative cladogenesis statistic (Nee et al., 1992), which is calcul ated as the likelihood that at any given time slice across the phylogeny, each lineage at that time slice will give rise to a number of species, summing to the total number of species, such that each lineage ha s diversified at the same stochastic rate. Lineages deviating from this nul l expectation are therefore consid ered significantly more diverse than expected. The second statistic follows methods employed by Magalln and Sanderson (2001) to calculate r which are less sensitive to taxon sampling, because calculations of diversification rate use only cl ade diversity and time since orig ination of the stem or crown lineage. We calculated r for major clades of Simaroubaceae using both crown and stem lineages. Confidence limits are set as a function of the bac kground diversification rate of the family as a whole, and clades that fall out side the 95% confidence interval are considered hyperdiverse or significantly depauperate in species. The method also requires the user to specify an extinction rate, and therefore a range of values from 0 to 0.9 was tested (Alf aro et al., 2007). Taxon sampling is important for this method with respect to measuring rates for stem and crown lineages, because using the crow n group assumes the sampling is sufficient to encompass the most inclusive node for the clade. The crown group method was used for most clades in Simaroubaceae because most unsampled taxa are limited to the three large genera, Soulamea

PAGE 113

113 Castela and Simaba. In Simaba the deepest node has probably been sampled because the genus consists of two sister clades, which are mor phologically distinct, and members of both clades have been sampled in the phylogeny. In the case of Castela and Soulamea it may be unreasonable to assume the deepest node has been sampled, and in this case we also calculated diversification rates with the stem group me thod, along with a rate for the New Caledonian Soulamea (for which only a single species was sample d). The stem group method also had to be applied to monotypic genera Leitneria, Nothospondias and Odyendea, which have no crown group, and Picrolemma and Holacantha, which have two species bu t only one was sampled. All temporal methods were performed in R using the laser package (Rabos ky, 2006) and the geiger package (Harmon et al., 2008). For both the rela tive cladogenesis statistic and the method of Magalln and Sanderson, the most parsimonious interpretation is a shift in diversification rate implied at the node ancestral to the root of the significantly mo re diverse clade (Purvis et al., 1995). A topological method was also used for detectin g shifts in diversification rate, as this would allow the use of the combined molecular and morphological data set, and does not suffer the trickle down effect seen in the temporal methods (Moore et al., 2004 ). Trees resulting from the combined data had much better taxon sampling than molecular data only, but could not be included in temporal methods described above because divergence dates are estimated using nucleotide substitutions only. The topological met hod relies on tree shape, and detects nodes that are significantly imbalanced using a shift statistic ( Chan and Moore, 2002; Moore et al., 2004; Moore and Donoghue, 2007). The shift statistic ( ) is calculated for all nodes, using the difference between likelihood ratios at each node. Nodes are treated as three-taxon trees, and a likelihood ratio is calculated for the ingroup-outgroup diversity partition, and the diversity

PAGE 114

114 partition between the two ingroup clades. The lik elihood ratio is a ratio between a one-rate model, where both clades are expected to have the same branching rate, and a two-rate model, where both clades are expected to have different branching ra tes (see Chan and Moore, 2002; Moore et al., 2004; Moore and Donoghue, 2007). The program SymmeTREE (Chan and Moore, 2005) implements the topological method (Chan a nd Moore, 2002), and returns shift statistics and their signifcance level for w hole tree imbalance and for indivi dual nodes. The software uses a tree as the only input, and so the tree used in parsimony character mapping was used. Correlates of Shifts in Diversification Rate Both biogeography and morphological evoluti on were considered for correlation to diversification rate shifts. Bioge ographical shifts are taken direc tly from analyses performed in Chapter 4. All nodes after which a range expansi on or vicariant event occurred were used as points of biogeographic shift (see Chapter 4). Most of these range shifts were categorized as long-distance dispersal events, with few c onsidered overland migrations. These nodes are identified in Figure 5-7. The bioge ographic shifts were compared visually to diversification rate shifts determined in analyses described earlier. Shifts along terminal lineages were excluded (e.g. range expansion in Samadera indica or Brucea javanica ), as no shift in diversification rate can be inferred. Morphological traits used to examin e character evolution we re also tested as potential key innovations. The method of Ree (2005) was used to test wh ether the morphological traits used for character mappi ng covary with diversificatio n rate. This method involves calculating a test statistic by mapping charac ters on the Simaroubaceae chronogram using stochastic mapping under a conti nuous-time Markov model (Huelse nbeck et al., 2003). Bayesian stochastic mapping allows for multiple state cha nges along branches, thus better dealing with long branches than parsimony methods (Cunning ham et al., 1998; Huelsenbeck et al., 2003). A diversification rate is ca lculated simply as the number of hi storical branching events with one

PAGE 115

115 state of the character divided by the time spent in that state (based on the stochastic mapping of the character). The difference between the rates fo r one state (the putative key innovation) versus the second state (the ancestral stat e) is then calculated. This pro cess is repeated for a number of mappings and the mean difference in diversificati on rates between the two states is used as the observed test statistic. For the nu ll distribution, Monte Carlo simula tion is used to generate any number of trees with the same topology as the Simaroubaceae chronogram but with branch lengths generated under a pure birth process. For each of these trees, the test statistic is generated as it was for the Simaroubaceae chronogram, and the resulting values constitute the null distribution, against which the obser ved test statistic can be compar ed to determine its posterior predictive p value (Ree et al., 2005; Moore and Donoghue, 2007). The process is implemented in keyinnotest ( http://www.phylodiversit y.net/rree/software.html ). The characters chosen were recoded as binary traits if th ey were not already, with multista te characters being broken down into presence/absence of a part icular state. Currently keyinnotest does not accommodate missing data, so missing states were recoded based on mo st-parsimonious reconstructions determined in MacClade. Given the temporal co mponent of the analysis, only taxa present in the molecular rates data set could be included (C hapter 4). To generate the observ ed test statistic for each of the 36 binary character traits, 100 stochastic mappings were performed on the Simaroubaceae chronogram. For each character the tree length was scaled to the minimum number of characterstate changes needed to observe the distribution of states at the terminals (Ree, 2005). The null distribution was generated with 1000 pure birth trees, and 10 stochastic mappings on each simulated tree.

PAGE 116

116 Results Phylogeny Estimation and Character Evolution The 71 morphological characters used for phylogeny estima tion are provided in Appendix D. Summary statistics for both morphology only a nd combined MP analyses are shown in Table 5-1. A strict consensus for mor phological data only is shown in Figure 5-1, with clade support shown on branches at cutoffs of 70% BS and 0.95 PP. Several genera and some major clades of Simaroubaceae were resolved by morphological data, but backbone structure was poorly supported. For combined morphological and molecula r data, a strict consensus is shown (Fig. 52), with bootstrap support and Bayesian posteri or probabilities. The t opology was similar to the previous molecular phylogeny (Chapter 3); how ever, support within genera and for some backbone nodes was greatly reduce d. The previously unplaced genus Laumoniera was sister to Nothospondias in the analysis of morphology only, but its position was unresolved in the combined analysis. Iridosma was found in a clade with Gymnostemon and Perriera in the combined analyses, but was unresolved within Clade V for morphology only. Table 5-2 shows morphological synapomorphies for the major clad es in Simaroubaceae, including taxa in which the characters have reversed or evolved in para llel elsewhere. Table 5-3 shows details for MPR for the 26 characters used for character mapping, including the number of unambiguous character-state changes required to explain the data, the number of most-parsimonious reconstructions for each character, and ancestr al states for major clades of Simaroubaceae. Diversification Analyses The relative cladogenesis statistic (Nee et al., 1992) identified si x nodes in the phylogeny with significant shifts in divers ification rate. All but one of these are attrib uted to the trick ledown effect, where a particularly diverse clade causes the successively deeper clades in which it is nested to appear more diverse than expected (Moore et al., 2004). In th is case, the shallowest

PAGE 117

117 clade is assumed to be the clade with signifi cantly elevated levels of diversity, which in Simaroubaceae is the New World subclade of Clade V (plus Pierreodendron ). The diversification rates cal culated using methods of Ma galln and Sanderson (2001) are illustrated in Figures 5-5 and 5-6, with 95% conf idence interval boundary lines for three different extinction rates (E). Fo r crown groups (Fig. 5-5), no clades fall outside the 95% bounds when E = 0.9. Eight crown clades are signi ficantly diverse for E = 0 and three clades for E = 0.5. Of the eight clades, six are equi valent to the shifts detected by the relative cladogenesis statistic, and thus five of these are eliminated under the assump tion of the trickle-down effect (Moore et al., 2004), leaving Simaba as the least inclusive clade. The other two crown clades above the 95% bound for E = 0 were Soulamea and Castela although these assume the true crown node was sampled for each, which may not be the case. Wh en these two clades were sampled as stem groups, only Castela fell outside the 95% CI fo r E = 0. At the lower 95% bound, Quassia, Perriera + Gymnostemon, and Asian Brucea had significantly fewer species than expected for E = 0. For the stem groups (Figure 5-6) only one clade, New Caledonian Soulamea, was significantly more diverse than expected, for E = 0 and E = 0.5. Monotypic genera Leitneria, Nothospondias and Odyendea fall below the lower 95% bound for all three extinction rates, but it is difficult to evaluate diversification in clades of a single species (Wiens et al., 2007). Results from SymmeTREE analyses (Chan and Moore, 2005) showed only one node to have a significant shift in diversification rates, the root node of Clade III, that is, the divergence of Leitneria from Brucea and Soulamea One other node was considered near-significant (p=0.11), the root node of the family, that is, an imbalance between Clade I and the rest of the Simaroubaceae. SymmeTREE also detected significant variation in divers ification rates across the whole tree (p < 0.01). Nodes at which a diversification rate shift was detected, using either

PAGE 118

118 temporal or topological methods de scribed above, are indicated in Figure 5-7. Also shown in Fig. 5-7 are nodes at which biogeographical shifts are proposed to have occurred. Results for analysis of key innovations are shown in Table 5-4. Of the 36 binary char acters tested, four showed significant correlation, with p < 0.05, but none at p < 0.01. Discussion Morphological data exhibited high levels of homoplasy, but also showed a number of characters were phylogenetically informative, more so at the generic level th an deeper in the tree. In both MP and Bayesian analyses, clade suppor t was generally low; however, the MP strict consensus wa s resolved for several genera, and bootstrap support (> 70%) and posterior probabilities (> 0.95) supported most of this generic resolution (Fig. 5-1). Genera of Simaroubaceae are well-supported because each has unique characteristics, for example, the winged and jointed rachis of Quassia, the trimerous flowers and dry, obcordate fruits of Soulamea or the cymose inflorescence and globose fruits of Picrasma Morphology also provided some resolution for larger clades found in molecular analyses (Chapter 3), such as a clade of Picrolemma plus its sister group. The sister relationship of Picrasma and Castela + Holacantha to the rest of the family could be an artifact of the lack of outgroup taxa (see Methods). However, regardle ss of which of the two genera are selected to be the outgroup (based on results of molecular data ), the other is always the next -diverging clade in the family, supporting their early divergence in the family. W ithout outgroup taxa it is impossible to say if their forming a clade is supported by mo rphology, or if morphology supports the Picrasma lineage diverging before the Castela lineage or vice versa. Ther e are no synapomorphies for a clade of Picrasma Castela and Holacantha that could not alternatively be interpreted as pleisomorphies in the absence of outgroup taxa, e.g. the absence of leaf glands; flowers with four carpels. Furthermore, both Picrasma and Castela + Holacantha have synapomorphies unique to

PAGE 119

119 the family, that could be shared with outgroup taxa, e.g. cymes in Picrasma ; spines in Castela and Holacantha. The combined analyses of molecular and morphological data pr oduced a phylogeny (Fig. 5-2) similar to the previous molecular phylogeny (Chapter 3). The introdu ction of taxa without molecular data reduced support within genera su bstantially, but only significantly impacted the topology of Clade V. The backbone remains poor ly supported, and based on analyses of morphology alone, there is a cons iderable amount of homoplasy in the data (CI = 0.29). Of particular note in both combined a nd morphology-only analyses is the genera Iridosma (Aubrville, 1962) and Laumoniera (Nooteboom, 1987). Iridosma appears closely related to its African counterparts in Clade V; however, herbarium material was unavailable for examination, so a phylogenetic position with improved support could be found with more data. Laumoniera was initially described as closely related to Brucea (Nooteboom, 1987), as they shared tetramerous flowers, a narrow thyrsoid infloresce nce and lack of a filament appendage, and both are found in SE Asia. However, Laumoniera being sister to Nothospondias is hypothesized under critical analysis of morphological data. Nothospondias, although found only in Africa, also shares tetramerous flowers, a narrow thyrse and an absence of a filament appendage with Laumoniera, and additionally has large, fleshy, yellow, ellipsoid fruits, similar to those of Laumoniera, and quite unlike the small, ovo id, scarcely fleshy drupes of Brucea It is also curious that in his description of Laumoniera, Nooteboom did not mention any prominent submarginal glands on the undersurface of the le aflets, glands that ar e characteristic of Brucea and which Nothospondias lacks. Again, no herbarium material of Laumoniera was available for examination, but based on these data, Laumoniera is more closely related to Nothospondias than Brucea More data will determine whether differences between the two are sufficient to maintain

PAGE 120

120 generic status, or whether Laumoniera should be placed in Nothospondias under the new combination Nothospondias bruceadelpha (Noot.) J.W.Clayton. Analyses of character evolution reveal a complex history of tra it evolution, particularly for a family with only ca. 100 species, although charact er evolution data from related families, for comparison, are scarce. A high rate of transi tion, and high numbers of most-parsimonious reconstructions, are seen between states for quan titative characters such as size and color of flowers and fruits, and degree of corolla pubescence. This is partly due to the number of states assigned to these characters, and the subjectivit y of coding such characters, with some taxa showing conditions close to the boun daries of the different states It may also be due to the influence of the environment, and local adaptation to shifts in pollinators, dispersers and the immediate surroundings that support a particular growth form. Ce rtain reproductive characters also show lability, such as the types of inflorescence, the number of stamens and the number of carpels. For example, the ancestor of the family is reconstructed to have had four carpels, but in Clade III there are examples of reduction to one, two and three car pels. Characters which show a low amount of change and few most-parsimonious reconstructions are strong synapomorphies for defining some of the major clades in the phylogeny. The presence of a filament appendage was traditionally used to de limit tribes (Engler, 1931), and it is still upheld as a good synapomorphy for all taxa diverging after Picrolemma with a single reversion in Perriera and Gymnostemon. Punctate glands have received little attent ion in the literature, except in individual species descriptions, and when Aubrville and Pe llegrin (1937) suggested a relationship between Perriera and Gymnostemon on the basis of punctate glands. Th ey are, however, only present in all taxa diverging after Picrolemma except for two apparent losses. Clade III can be defined as having reduced flowers (if we also code Leitneria s absent perianth as reduced), and this clade

PAGE 121

121 also shows a tendency towards having spiciform inflorescences. Ancestral states for the family correspond to ancestral states for the Sapi ndales (e.g. pinnately compound leaves), but interestingly, Sapindales are hypot hesized to have pentamerous flowers ancestrally (Ronse de Craene, 2008), depending on the rosid topology, whereas Simaroubaceae are ancestrally tetramerous. Diversification rate shifts in the family ar e few and dependent on the methods used, but the family as a whole is suggested to have significant nodal imbalance based on tree shape alone. The diversification ra te has increased in Castela a clade of Brucea Soulamea and Amaroria New Caledonian Soulamea and either the New World members of Clade V or Simaba (depending on the method). Although the strengt h of the Magalln and Sanderson (2001) method is its applicability to clades with unsampled taxa, the reliance on species diversity may fail to detect rate shifts in clades that show a recen t upturn in diversificati on rate, which has yet to manifest itself in increased clade diversity (Ree, 2005), or in clades wh ere a rapid increase in diversification rate is followed by a period of stasis. Methods ba sed on clade diversity, measured in species numbers, are phylogenetically non-i ndependent, unlike methods based on tree shape (Chan and Moore, 2002; Moore et al., 2004), which look for rate shifts relative to parent and daughter lineages. This may not be a problem when species-rich clades appear recently, such as those detected here, but it is the cause of the trickle-down effect (Moore et al., 2004) seen in the elevated diversity of more inclusive clades containing Simaba. The current study found four signifi cant correlations between dive rsification rate shifts and putative key innovations, that is, presence of thorns, evolution of haplostemony from diplostemony, having a short style, and movement into a temper ate habitat. There is always potential for other morphological characters to impact diversif ication, particularly cryptic

PAGE 122

122 characters for which the ecological impacts ar e poorly understood. The significance and nearsignificance of some of the characters is, however, questionable base d on visual inspection of the tree, and may be an artifact of insufficient stochastic mappings, insufficient sampling to generate the null distribution, or imbalance in character-state frequencies. In all significant cases, simulated trees create a very narrow null distribu tion, and the significance of thorns may be an artifact of the long branch leading to Castela which would result in a wide variance around the mean time spent having thorns, leading to an anomalous result. Regardless, none of the characters were significant at p < 0.01, whereas in previous examples where this method has shown significance (nectar spurs in Aquilegia : Ree, 2005; stamen number in Valerianaceae: Moore and Donoghue, 2007), the predictive post erior p value was < 0.001. A conservative approach is taken in this study, given uncertainty around the sampling, and no significant key innovations are proposed. As stat ed by Moore and Donoghue (2007), th e statistical properties of this test need further exploration. Normalizing the null distribution for different states depending on their frequency of occurrence may be necessa ry (Ree, 2005). Furthermore, these types of studies have an important caveat. In order for character-state transiti ons to covary with rate shifts with true statistical significance, a larger sample size would be needed, with repeated examples of covariance across a number of more distantly related clades (Ree, 2005). In most cases, a single character is shown to correlate to a si ngle rate shift (Ree, 2005; Moore et al., 2006; Leschen and Buckley, 2007; Moore and Donoghue, 2007; Wiens et al., 2007), resulting in a single data point, although it could be argued th at highly homoplasious characters may represent more data points (Ree, 2005). Without a certain level of statistical confidence, the conclusions drawn from such a study must account for the po ssibility of random chance producing the results seen, and the myriad of interact ions of key innovations with ot her organismal traits and the

PAGE 123

123 environment (de Queiroz, 2002). Very few studies such as these have been undertaken (Moore et al., 2006; Leschen and Buckley, 2007; Moor e and Donoghue, 2007), and Moore and Donoghue found biogeographical movements to be most commonly correlated with diversification rate increases rather than putative key innovations Looking at morphological characters across a broader phylogenetic spectrum, for example, with in the Sapindales, would reveal more about diversification in the Simaroubaceae, and stochastic effects. Biogeographical movements appear connected to increased species diversity in Simaroubaceae, but with the number of dispersal events occurring in the recent history of the family, it is difficult to parse out significant events if they are surr ounded by events showing no apparent connection to diversification. In Simaroubaceae diversification is linked in two instances to dispersal to and movement within the New World, one inst ance to arrival on New Caledonia, and one instance to movement into Africa from Asia. There are, however, clades which also show movement between North a nd South America and no accompanying significant increase in diversification ( Picrasma, Picrolemma and Simarouba). Because exploring historical biogeography on a global scale limits the state tr ansitions that are examined, these patterns can be compared to other studies in which increased diversification was also associated, especially with movement within the New World (Moore and Donoghue, 2007; Hughes and Eastwood, 2006; Wiens et al., 2007), but this is typically coupled with movement across an environmental gradient. This is contrary to statistical tests for key innovations, of which few have been undertaken, and which have poten tially hundreds of morphological st ate transitions to examine. In the case of Castela the increased diversity of the clade correlates with a movement of the ancestor into South America. However, repeated dispersal between arid environments of North and South America (and to the Galpagos Islands) is more likely to be the driver of this

PAGE 124

124 increased speciation rate (Lavin et al., 2004). If the initial adaptation to arid environments of North America in the Oligocene a nd Miocene were the cause of an adaptive radiation, we would expect to see significantly incr eased speciation rates including Holacantha as well as a significant test statistic for the aridity character (Table 5-4, character 36), which is not the case. Moore and Donoghue (2007) describe dispersal into Latin America as a promoter of diversification rate increase, and it is altitudinal environmental fluctation within the Andes that has caused allopatric speciati on, leading to high diversificat ion rates (also see Hughes and Eastwood, 2006; Wiens et al., 2007). Long-distance dispersal events are the initial promoter insofar as they expose the founder to a variety of new niches in which smaller-scale processes can operate towards speciation (von Hagen and Kadereit, 2003; Kay et al., 2005; Funk et al., 2006). Lavin et al. (2004) argue that metacommunity processe s are the driving force behind current legume distributions, with historical dispersal events frequent enough to suggest that ecological preadaptation determines extant geographical patterns, which in turn promotes speciation due to isolation by distance. Alt hough Simaroubaceae are far less diverse in total species numbers compared to legumes, their deep history and multiple post-Oligocene dispersal events might be expected to generate a similar pattern of speciation thro ugh dispersal. Recent speciation has occurred in clades with species adapted to seasonally dry or arid environments, such as previously mentioned Holacantha and Castela African Brucea species, and a number of species in Clade V, e.g. Hannoa in African savannah habitat, and Simaba sect. Floribundae in the drier areas of Brazil and Paraguay (Cronquist, 1944c). In Simaba, the dry-adapted Floribundae clade, the much more species-rich clade of Simaba, is sister to the Amazonian species, and character mapping sugges ts primary rainforest is the ancestral habitat in Clade V. A

PAGE 125

125 more fine-scale approach to categorizing ecologi cal traits, such as seasonal rainfall averages, might provide some insight into the dynamics of dry forest speciation (Pennington et al., 2004b). New Caledonian Soulamea were not resolved as a clad e outside the other species of Soulamea and Amaroria in morphological analyses, and none of the characters examined in ancestral-state reconstruction mapped strictly to the New Caledonian species. It is more likely that Soulamea experienced an adaptive radiation associ ated with new arrivals to an island environment (Whittaker, 1998). Conversely, Soulamea amara has a very widespread distribution, and is the on ly bisexual species in Soulamea All other species are confined to a single island or group of islands, a nd are dioecious. The ab ility to self is crit ical to establishment in a new environment distant from the parent population (Whittaker, 1998) whereas in the other species dioecy may confer an advantage to maintaining heterozygosity in limited space and resources. Shifts in diversification rate are always considered in a relative context, compared either to background levels of diversification in the family or to more and less inclusive nodes. Temporal methods are usually interpreted from the point of increased, rather th an decreased, rates of diversification, as these are of interest when looking at biodiversity gradients, adaptive radiations, and typical cause and effect relationships (A lfaro et al., 2007; Moore and Donoghue, 2007; Wiens et al., 2007). Clades with re duced levels of diversity, such as Leitneria Nothospondias, Picrolemma and Quassia, are not as easily interprete d, because low diversity can be attributed just as easily to extinction as to a slow-down in diversific ation rate. The correlating factors of biogeography and mo rphological change could be implicated in an increased extinction rate, but more typically this is attributed to factors extrinsic to the organism (Cracraft, 1985), such as climate change. Although Leitneria registers as a significantly depauperate

PAGE 126

126 lineage, based on the methods of Magall n and Sanderson (2001) and implied by node imbalance (Chan and Moore, 2006), its fossil histor y shows a greater divers ity of species in the past (Dorofeev, 1994; Nikitin, 2006a). It is interesti ng to note that this lineage has such a striking shift in floral and pollen morphology from insect to wind-pollination, yet a single extant species remains. Wind pollination has evolved many times independently in the angiosperms (Soltis et al., 2005), and a number of studies (e.g. Eriksson and Bremer, 1992; Dodd et al., 1999) suggest wind-pollinated groups have lower diversification rates. Shifts to wind pollination are typically coupled with temperate environments (Culley et al., 2002), which rarely support the significant levels of species diversity seen in the tropi cs (Wiens and Donoghue, 2004), so a slow-down in diversification might be expected. Pollinator shif ts have been hypothesized as drivers of tropical forest diversity (Gentry, 1982) due to reproductive isolation. Quassia amara demonstrates a shift from insect to hummingbird pollination, yet this lineage consists of a single New World species, diverging from its African sister species approximately 13 Ma. Costus subg. Costus was demonstrated to show rapid speciation under th e influence of shifts between hummingbird and bee pollination (Kay et al., 2005), which was co upled with local environmental fluctuation. However, empirical examples in a phylogenetic context such as Costus are few, and no conclusions can be drawn regarding Quassia. It should also be noted that extinct species are more likely to be recovered as fossils in temp erate lineages, given pres ervational and collecting biases, and tropical lineages such as Nothospondias and Picrolemma could be the remnants of more diverse clades. The aridification of Afri ca has been hypothesized as a major cause of extinction in the Miocen e (Davis et al., 2002b; Morley et al., 2003; Plana, 2004) so we might expect reduced species diversity in ancient African lineages (of which Nothospondias is the

PAGE 127

127 oldest in Simaroubaceae). Conversely, Picrolemma is much less diverse than expected compared to the other incursions of Simaroubaceae into South America ( Simarouba+ Simaba and Castela ). Whether we are examining diversification as a function of biog eography, morphological evolution or stochastic processes, we are faced with the problem of phylogeny as a whole. For every clade studied, there is a un ique evolutionary history that cannot be replicated in an experimental design, and inferences made must ta ke this into account. Only when results can be synthesized across a number of pl ant clades and patterns are repeat ed, can we begin to draw firm conclusions about the origins of extant biodivers ity in the Simaroubaceae. Despite this, our study has suggested recent dispersal events, rath er than morphological innovation, may have contributed significantly to increased diversification rates in the family.

PAGE 128

128Table 5-1. Results of maximum parsimony analyses for morphological data and combined morphological and molecular data of Simaroubaceae. a indicates MaxTrees limit was reached. Number of ingroup taxa Number of characters Variable characters Informative characters Number of shortest trees Length of best tree Consistency index Retention Index Morphological 105 71 71 67 100,000a 404 0.290 0.776 Combined 115 6035 1924 1208 100,000a 4036 0.614 0.822 Table 5-2. Synapomorphies for major clades of Simaroubaceae, includ ing taxa inside the clade show ing a reversal in the characte r, and parallel acquisitions or lo sses outside the clade. Only characters ha ve a CI greater than 0.1 and unambiguous reconstruction on the tree are included. Clade Synapomorphies Clade I (Picrasma Castela and Holacantha ) No unambiguous synapomorphies Castela and Holacantha Thorns present Clade III (Leitneria Brucea Soulamea and Amaroria ) Narrow, elongate thyrse with a single major axis Corolla length less than 3 mm (parallel evolution in Ailanthus fordii, Castela erecta, C. retusa, C. macrophylla, Picrasma crenata, Samadera bidwillii ) Reticulately wrinkled endocarp Striate pollen exine (except for Leitneria in which it is minutely verrucate) Clade V (Simaba to Odyendea inclusively) No unambiguous synapomorphies Clade V to Quassia inclusively Punctate glands on lamina Bisexual plants (reversal in Eurycoma and Simarouba ; parallel evolution in Soulamea amara) Filament appendage present Stigma inconspicuous or capitate (reversal in Simarouba) Clade V to Picrolemma inclusively Five carpels (reversals in Iridosma Perriera Odyendea and Samadera indica ; parallel in Ailanthus ) Crystalliferous cells present; parallel evolution in Castela coccinea ; data is limited for this character) Fused portion of style much long er than free stylodia branches Ailanthus to Clade V inclusively No unambiguous synapomorphies

PAGE 129

129Table 5-3. Character evolution on the Simaroubaceae phylogeny for 26 characters under maximum parsimony reconstructions (MPR). Ancestral states for selected nodes are s hown for ACCTRAN optimization. Asterisk i ndicates an alternative reconstruction under DELTRAN optimization. No. Character No. of states No. of most parsim onious recons. No. of changes under MPR Ancestral state for family Ancestral state for Clade I Ancestral state for Clade III Ancestral state for Clade V 1 Height (life form) 2 2 9 Shrub or small tree < 12 m Shrub or small tree < 12 m Shrub or small tree < 12 m Shrub or small tree < 12 m 2 Thorns 2 1 1 Absent Absent Absent Absent 3 Leaf type 3 6 9 Pinnate Pinnate Unifoliolate* Pinnate 4 Marginal teeth 2 4 6 Entire Entire Entire Entire 5 Leaf glands 2 12 5 Equivocal Absent Present Present 6 Leaf gland type 2 1 6 Absent Prominent, dark Punctate, light 7 Inflorescence type 4 15 5 Equivocal Eq uivocal Typical thyrse Typical thyrse 8 Thyrsoid structure 2 15 6 Equivocal Narrow, elongate Open, multiple axes 9 Sex 2 1 4 Unisexual Unisexual Unisexual Bisexual 10 Merosity 4 5 8 Tetramerous Tetramerous Tetramerous Pentamerous 11 Corolla hairs 4 340 18 Glabrous Glabrous Glabrous Pubescent on both sides 12 Corolla length 4 90 15 2.1 8 mm 2.1 8 mm 0 2 mm 2.1 8 mm 13 Corolla color 2 6 9 White/yellow/green White/y ellow/green White/yellow/green White/yellow/green 14 Shape of nectar disc 3 37 8 Fleshy, broad Fleshy, broad Fleshy, broad Columnar* 15 Nature of androecium 3 4 7 Equivocal Equivocal Haplostemonous* Diplostemonous 16 Filament appendage 2 1 2 Absent Absent Absent Present 17 Number of carpels 6 6 10 Equivocal Four One* Five 18 Style length 2 1 3 Free portion longer Free portion longer Free portion longer Free portion shorter 19 Stigma shape 4 2 7 Linear, recurved Linear, recurved Linear, recurved Inconspicuous 20 Fruit type 6 1 5 Ellipsoid drupe Ellipso id drupe Ellipsoid drupe Ellipsoid drupe 21 Drupe size 3 36 12 Equivocal 13 mm 13.1 25 mm 13.1 25 mm 22 Drupe color when ripe 2 12 4 Orange/red/black Orange/red/black Orange/red /black Orange/red/black 23 Pollinator 3 1 2 Insect Insect Insect Insect 24 Mode of dispersal 2 1 4 Biotic Biotic Biotic Biotic 25 Temperate vs. tropical habitat 2 2 4 Tropical Tropical Tropical Tropical 26 Moisture in habitat 3 4 8 Mesophytic Mesophytic Mesophytic Mesophytic

PAGE 130

130Table 5-4. Characters tested for correlati ons with diversification rate shift. Binary character st ates are shown, along with po sterior predictive p value (values in bold are significant (p < 0.05)). Character State 0 State 1 P 1 Maximum height (life form) Shrub < 12 m Tree > 15 m 0.42 2 Thorns Absent Present 0.03 3 Leaf type Pinnately compound Unifoliolate or leafless 0.89 4 Margin teeth Entire Crenulate, serrate or shallowly undulating 0.06 5 Punctate glands Absent Present 1.00 6 Prominent, dark glands Absent Present 0.12 7 Typical thyrse inflorescence Not typically thyrsoid Typical thyrse 0.88 8 Fascicle-like inflorescence Not fasciculate Fasciculate 0.07 9 Umbellate inflorescence Not umbellate Umbellate 0.88 10 Cymose inflorescence Not cymose Cymose 0.06 11 Thyrse structure (narrow) Not spiciform Spiciform 0.67 12 Thyrse structure (open) Absent or narrow Broad, compound, open, 0.85 13 Sex Unisexual (or androdioecious) Bisexual 1.00 14 Trimerous No Yes 0.12 15 Tetramerous No Yes 0.06 16 Pentamerous No Yes 0.92 17 Petal hair Glabrous, or o ccasional hairs Pubescent 0.99 18 Small flowers Flowers > 2 mm Flowers 2 mm 0.09 19 Medium flowers 2 mm or > 8 mm 2.1 8 mm 0.55 20 Large flowers < 9 mm 9 mm 1.00 21 Petal color (abaxial) White/yellow/green Pink/red/violet 0.18 22 Disc shape Ring-like, fleshy, lobed or absent Columnar 0.98 23 Androecium Diplostemonous (or pleiostemonous) Haplostemonous 0.02 24 Filament appendage Absent Present 0.97 25 Number of carpels More than one One 0.45 26 Short style Style long, stigma inconspicuous Style short, stigma long, or discoid 0.04 27 Long style Style short, stigma long or discoid Long style, inconspicuous stigma 0.99 28 Fruit type Fleshy, drupaceous Dry, samaroid 0.14 29 Drupe size (excl. wing, keel) 25 mm 29 mm or more 1.00 30 Red drupes Green, yellow or n/a Orange to red to blue-black 0.42 31 Yellow or green drupes Orange to red to blue-black or n/a Yellow or green 0.95 32 Pollinator Insect Bird or wind 0.14 33 Dispersal Biotic Abiotic 0.21 34 Habitat1 Tropical Temperate 0.03 35 Habitat2 Not swamp Swamp, riverine, coastal 0.77 36 Habitat3 Not arid Arid 0.11

PAGE 131

131 Figure 5-1. Phylogram randomly selected from 100,000 most-parsimonious trees recovered from 71 morphological characters for Simaroubaceae. Thick branches appear in the strict consensus. Asterisks indicate >70% bootstrap support and > 0.95 posterior probability, respectively. Samadera indica Pierreodendron africanum Samadera baileyana Samadera bidwillii Iridosma letestui Perriera madagascariensis Gymnostemon zaizou Simaba cedron Simaba docensis Simaba polyphylla Simaba paraensis Eurycoma apiculata Eurycoma longifolia Eurycoma harmandiana Simaba guianensis Simaba guianensis Simaba guianensis Simaba guianensis Simaba insignis Simaba praecox Simaba glabra Simaba warmingiana Simaba ferruginia Simaba cuneata Simaba cuneata Simaba moretti Simaba cavalcantei Simaba trichilioides Quassia africana Quassia amara Simaba glabra Simaba orinocensis Simaba obovata Simaba orinocensis Hannoa undulata Hannoa undulata Simaba suffruticosa Hannoa chlorantha Hannoa ferruginia Odyendea gabonensis Hannoa klaineana Hannoa sp.B Hannoa sp.A Simarouba amara Simarouba versicolor Simarouba berteroana Simarouba glauca Simarouba glauca Simarouba laevis Simarouba tulae Picrolemma sprucei Picrolemma sprucei Picrolemma huberi Ailanthus excelsa Ailanthus triphysa Ailanthus fordii Ailanthus altissima Ailanthus altissima Ailanthus integrifolia Soulamea dagostinii Soulamea morattii Soulamea terminalioides Soulamea pelletieri Soulamea trifoliata Soulamea rigaultii Soulamea tomentosa Soulamea cycloptera Soulamea amara Soulamea pancheri Soulamea muelleri Soulamea cardioptera Soulamea fraxinifolia Amaroria soulameoides Laumoniera bruceadelpha Nothospondias staudtii Brucea javanica Brucea mollis Brucea antidysenterica Brucea tenuifolia Brucea guineensis Brucea macrocarpa Leitneria floridana Holacantha emoryi Holacantha stewartii Castela coccinea Castela spinosa Castela calcicola Castela peninsularis Castela depressa Castela retusa Castela erecta Castela erecta Castela tortuosa Castela galapegia Castela macrophylla Castela jacquinifolia Castela tweedii Picrasma antillana Picrasma excelsa Picrasma javanica Picrasma quassioides Picrasma crenata Picrasma mexicana Picrasma cubensis Picrasma selleana 1 change **/**/**/**/**/*/ -*/ -*/ -*/ -*/ -*/ -*/ -*/ -*/

PAGE 132

132 Figure 5-2. Phylogeny randomly selected of 100,000 most parsimonious trees recovered from a combined analysis of rbcL, atpB, matK (including partial trnK intron) and phyC, and 71 morphological characters for Simaroubaceae. Grey branches collapse in the strict consensus. Asterisks indicate >70% bootstrap support and > 0.95 posterior probability, respectively. Simaba glabra Simaba warmingiana Simaba glabra Simaba cuneata Simaba cuneata Simaba suffruticosa Simaba moretti Simaba paraensis Simaba docensis Simaba insignis Simaba praecox Simaba ferruginia Simaba cedron Simaba trichilioides Simaba orinocensis Simaba orinocensis Simaba obovata Simaba guianensis Simaba guianensis Simaba guianensis Simaba polyphylla Simaba guianensis Simarouba berteroana Simarouba tulae Simarouba glauca Simarouba glauca Simarouba amara Simarouba versicolor Simarouba laevis Pierreodendron africanum Hannoa undulata Hannoa undulata Hannoa chlorantha Hannoa ferruginia Hannoa sp.A Hannoa sp.B Hannoa klaineana Odyendea gabonensis Eurycoma apiculata Eurycoma longifolia Eurycoma harmandiana Perriera madagascariensis Gymnostemon zaizou Iridosma letestui Quassia africana Quassia amara Samadera bidwillii Samadera sp.B Samadera sp.C Samadera baileyana Samadera indica Simaba cavalcantei Picrolemma huberi Picrolemma sprucei Picrolemma sprucei Laumoniera bruceadelpha Nothospondias staudtii Soulamea cardioptera Soulamea cycloptera Soulamea tomentosa Soulamea muelleri Soulamea rigaultii Soulamea pancheri Soulamea amara Soulamea dagostinii Soulamea trifoliata Soulamea terminalioides Soulamea pelletieri Soulamea morattii Soulamea fraxinifolia Amaroria soulameoides Brucea antidysenterica Brucea ferruginia Brucea tenuifolia Brucea guineensis Brucea macrocarpa Brucea javanica Brucea mollis Leitneria floridana Ailanthus altissima Ailanthus altissima Ailanthus integrifolia Ailanthus triphysa Ailanthus fordii Ailanthus excelsa Castela erecta Castela erecta Castela galapegia Castela tortuosa Castela retusa Castela macrophylla Castela tweedii Castela calcicola Castela spinosa Castela peninsularis Castela coccinea Castela depressa Holacantha emoryi Holacantha stewartii Castela jacquinifolia Picrasma antillana Picrasma excelsa Picrasma crenata Picrasma quassioides Picrasma mexicana Picrasma cubensis Picrasma javanica Picrasma selleana Casimiroa edulis Zanthoxylum sp. Cneorum tricoccon Melia azedarach Unknown Meliaceae Swietenia macrophylla Acer saccharum **/*/ -**/**/**/**/*/ -*/ -**/**/**/**/**/**/**/**/**/*/ -*/ -**/**/**/**/**/**/**/**/*/ -**/*/ -**/**/**/*/ -**/**/**/*/ -*/ -**/*/

PAGE 133

133 Figure 5-3. Unambiguous character-state changes based on maximum parsimony reconstruction for reproductive traits. Numbers by character-state transitions refer to character numbers in Table 5-3. P icra s ma cuben sis Pic ra s ma ja v ani ca Pic ra s ma mexi c an a Picrasma qua s sioide s Pic ra s ma c rena ta P icra s ma antillan a Pic ra s ma ex cels a Holacan t ha emory i Hola c an t ha st ewar tii Ca st ela depre ss a Castela c oc c ine a Ca st ela peninsulari s Ca st ela calci c ol a Cas t ela s pino sa Ca st ela t weedi i Cas t ela ma c rophyll a Cas t ela re t us a Ca st ela t or t uo sa Ca s tela galapegi a Cas t ela erec ta A ilanthu s ex c els a Ailanthu s fo r di i Ailanthus altissi ma A ilan t hu s al tis sima A ilanthus integrifoli a A ilan t hu s t riph ysa Lei t neria f loridan a Bruc ea m olli s B ru c ea javanic a B ru c ea ma c ro c arp a B rucea guineen sis B rucea t enui f oli a B ru c ea an t idysen t eric a B ru c ea f errugini a A maroria s oulameoide s S oulamea mora tt ii S oulamea pelle t ier i S oulamea terminalioide s S oulamea dagos t ini i Soula m ea t r ifoliat a S oulamea amar a S oulamea pan c her i S oulamea rigaul tii Soula m ea m uelle ri S oulamea t omen t os a S oulamea cardiop t er a S oulamea cy c lop t er a Laumoniera bruceadelph a Notho s pondia s s taudti i Pic rolemma spru cei P icrolemma huber i S amadera indi ca S amadera baile y an a S amadera sp C S amadera bidwilli i S amadera s pB Q uas s ia af ri c an a Qua ss ia amar a Ir idos m a letestu i P erriera madaga s carien sis G ymnos t emon z aizo u E ur y coma harmandian a E ur y coma api c ula ta Eu ryc om a longifoli a O dyendea gabonensi s Hannoa klainean a Hannoa sp A Hannoa sp B Hannoa f errugini a Hannoa c hloran tha Hannoa undula ta Hannoa undula ta P ierreodendron africanu m S imarouba amar a S imarouba versicolo r S imarouba ber t eroan a S imarouba t ula e S imarouba glauca S imarouba glauca S imaba guianensis S imaba guianen sis S imaba pol y phyll a S imaba guianensis S imaba guianen sis S imaba obova ta S imaba orinocen sis S imaba orino c en sis S imaba c edro n Si m aba t ric hilioide s S imaba f errugini a S imaba prae cox S imaba docen sis S imaba in s igni s S imaba paraen sis S imaba more tt i S imaba glabr a S imaba glabra S imaba warmingian a Si m aba suff r uti cosa S imaba cunea ta S imaba cunea ta 18 19 9 16 19 12 9 18 19 12 13 21 12 13 15 21 10 11 17 21 9 15 11 16 17 18 19 10 17 12 13 21 7 21 20 21 22 11 12 13 12 19 10 11 13 20 13 11 9 11 17 20 21 11 21 11 14 10 20 11 11 12 13 12 19 12 7 10 14 17 19 20 12 Fruit F loral Inf lores cenceCharacters that change unambiguous ly on b ranch

PAGE 134

134 Figure 5-4. Unambiguous character-state changes based on maximum parsimony reconstruction for vegetative traits and general ecology. Numbers by character-state transitions refer to character numbers in Table 5-3. P icra s ma cuben sis Pic ra s ma ja v ani ca Pic ra s ma mexi c an a Picrasma qua s sioide s Pic ra s ma c rena ta P icrasma antillan a Pic ra s ma ex cels a Holacan t ha emory i Hola c an t ha st ewar tii Ca st ela depre ss a Castela c oc c ine a Ca st ela peninsulari s Ca st ela calci c ol a Cas t ela s pino sa Ca st ela t weedi i Cas t ela ma c rophyll a Cas t ela re t us a Ca st ela t or t uo sa Ca s tela galapegi a Cas t ela erec ta Ca st ela ere ct a V en A ilanthu s ex c els a Ailanthu s fo r di i Ailanthus altissi ma A ilan t hus al tis sima A ilanthus integrifoli a A ilan t hus t riph ysa Lei t neria f loridan a Bruc ea m olli s B ru c ea javanic a B ru c ea ma c ro c arp a B rucea guineen sis B rucea t enui f oli a B ru c ea an t idysen t eric a B ru c ea f errugini a A maroria s oulameoide s S oulamea mora tt ii S oulamea pelle t ier i S oulamea terminalioide s S oulamea dagos t ini i Soula m ea t r ifoliat a S oulamea amar a S oulamea pan c her i S oulamea rigaul tii Soula m ea m uelle ri S oulamea t omen t os a S oulamea cardiop t er a S oulamea cy c lop t er a Laumoniera bruceadelph a Notho s pondia s s taudti i P icrolemma huber i Pic rolemma spru cei S amadera indi ca S amadera baile y an a S amadera sp C S amadera bidwilli i S amadera s pB Q uas s ia af ri c an a Qua ss ia amar a Ir idos m a letestu i P erriera madaga s carien sis G ymnos t emon z aizo u E ur y coma harmandian a E ur y coma api c ula ta Eu ryc om a longifoli a O dyendea gabonensi s Hannoa klainean a Hannoa sp A Hannoa sp B Hannoa f errugini a Hannoa c hloran tha Hannoa undula ta Hannoa undula ta P ierreodendron africanu m S imarouba amar a S imarouba versicolo r S imarouba ber t eroan a S imarouba t ula e S imarouba glauca S imarouba glauca S imaba guianensis S imaba guianen sis S imaba pol y phyll a S imaba guianensis S imaba guianen sis S imaba obova ta S imaba orinocen sis S imaba orino c en sis S imaba c edro n Si m aba t ric hilioide s S imaba f errugini a S imaba prae cox S imaba docen sis S imaba in s igni s S imaba paraen sis S imaba more tt i S imaba glabr a S imaba glabra S imaba warmingian a Si m aba suff r uti cosa S imaba cunea ta S imaba cunea ta 6 1 1 1 26 3 27 26 27 1 5 26 1 1 27 23 27 3 24 26 1 24 3 3 3 27 1 4 27 5 23 24 25 26 1 24 4 25 27 4 2 26 27Charactersthat change unambiguous ly on b ranch V egetative General ecology

PAGE 135

135 Figure 5-5. Crown group analysis of clades of Simaroubaceae using methods of Magalln and Sanderson (2001). Confidence intervals for three extinction rates are shown. 1 10 100 1000 0 10 20 30 40 50 60 70 80 Number of species Age of crown group (Myr) E = 0.0 E = 0.9 E = 0.5Clade V Family Asian Brucea Quassia Gym+Per Soulamea Castela Simaba NW Clade V 1 SamaderaClade V 2 Quassia -Clade V 3 Picrolemma -Clade V3 2 1

PAGE 136

136 Figure 5-6. Stem group analysis of clades of Simaroubaceae using methods of Magalln and Sanderson (2001). Confidence intervals for three extinction rates are shown. 1 10 100 1000 0 10 20 30 40 50 60 70 80 90 100 Number of species Age of stem group (Myr) Soulamea Leitneria Nothospondias Holacantha Picrolemma Odyendea New Caledonian Soulamea Castela FamilyE = 0.0 E = 0.9 E = 0.5

PAGE 137

137 Figure 5-7. Chronogram for Simaroubaceae (reproduced from Figure 4-3) showing shifts in diversification rates based on the methods of Magalln and Sanderson (2001), the relative cladogenesis statistic (Nee et al., 1992), and SymmeTREE (Chan and Moore, 2005). Also shown are nodes at which biogeographic shifts between two areas are hypothesized, based on biogeographic analyses in Chapter 4. I V IV III IISimarouba versicolor Hannoa chlorantha Perriera madagascariensis Simaba glabra Samadera bidwillii Simabaorinocensis Hannoa undulata Simaba paraensis Simaba guianensis Pierreodendron africanum Samadera sp.B Simarouba berteroana Simaba insignis Samadera sp.C Simaba cedron Simaba cuneata Simaba glabra Simaba moretti Simaba trichilioides Eurycoma longifolia Simaba ferruginia Eurycoma apiculata Samadera indica Simarouba tulae Gymnostemon zaizou Simaba cuneata Simarouba amara Simabaorinocensis Simaba guianensis Hannoa undulata Simabapolyphylla Samadera baileyana Simaba guianensis Hannoa sp.B Quassia afr icana Odyendea gabonensis Simabaguianensis Simarouba glauca Simarouba glauca Simaba suffruticosa Hannoa klaineana Hannoa sp.A Castela erecta Brucea ferruginia Picrasma quassioides Leitneria floridana Nothospondias staudtii Brucea guineensis Picrasma antillana Holacantha emoryi Castela tweedii Brucea mollis Ailanthus integrifolia Ailanthus altissima Ailanthus fordii Soulamea morattii Picrasma javanica Brucea antidysenterica Ailanthus altissima Castela retusa Soulamea amara Picrolemma sprucei Picrolemma sprucei Ailanthus triphysa Picrasma crenata Brucea javanica Castela coccinea Quassia a mara Amaroria soulameoides Castela erecta Picrasma excelsa Castela tortuosa Soulamea terminalioides OligoMioceneLate CretaceousPaleoEocenePli Clade is signi !cantly more diverse than expected; E = 0 (Magallon and Sanderson method) Rate shift implied by relative cladogenesis statistic (excluding trickle-down e ect) Rate shift detected by SymmeTREE Clade is signi !cantly less diverse than expected; E = 0 (Magallon and Sanderson method) Geographic range shift Rate shift implied by Magallon and Sanderson method

PAGE 138

138 CHAPTER 6 CONCLUSIONS Application of the latest tec hniques in systematics has been critical to im proving our understanding of the evolutionary history of Simaroubaceae, particularly as the family has received little attention in recent years. P hylogenetic reconstruction has been aided by the familys small size, and successful procurement of plant material, which has been obtained for almost all genera and the entire geographic ra nge of the family. Sequencing a combination of both well-established and less often used gene regions, and applying the latest approaches to modeling sequence evolution in a Bayesian framework, has produced a robust phylogeny. Combining this phylogeny with fossils of Ailanthus and Leitneria and a recently developed model of uncorrelated Bayesian molecular rates analysis (Dru mmond et al., 2006), has generated divergence date estimations for th e family. In turn these dates, along with geographical data for extant and extinct species, have been used to test hypotheses of histor ical migration and longdistance dispersal events, under a very recently developed likelihood model of geographic range evolution (Ree and Smith, 2008). Finally, the phylogeny has been used to examine character evolution in the family, and search for correlates of diversification rate shifts. Not since Englers (1931) classification ha s the entire family been dealt with in a comprehensive overview. Since then, five of Englers six subfamilies, and Harrisonia from Simarouboideae, have been found to be more cl osely related to various taxa throughout the rosids (Fernando et al., 1995) than to subfam ily Simarouboideae. Cronquist produced synopses of Castela (1944a), Simarouba (1944b) and Simaba (1944c), and an overview of the American genera (1944d), and Nooteboom (1962a) gave an ex cellent description of the genera and species found in Malaysia, with reference to the surround ing areas. He also recircumscribed several genera as the single genus Quassia, a classification not favored by other systematists of

PAGE 139

139 Simaroubaceae. The African genera and species in Simaroubaceae have received little attention except in Floras of individual regions (Aubrville, 1962; Stannard, 2000). Chapter 2 is the first time the family and its constituent genera have been described as a whole, since their phylogenetic recircumscription (Fernando and Qu inn, 1995). The chapter provides a key to the genera, plus morphological de scriptions, approximate speci es numbers and geographical distribution. The natural hist ory of the group, including anatomy, phytochemistry, reproductive biology, dispersal mechanisms, and important economic uses, is also summarized. Chapter 3 focuses on reconstructing th e molecular phylogeny of Simaroubaceae. Approximately six thousand base pa irs of sequence data have prove d sufficient in resolving most nodes in the phylogeny, and these data were tested thoroughly for th e effects of how the genetic data are partitioned. Model selection is an increasingly important aspect of phylogeny reconstruction, as molecular data from multiple gene regions across all genomes become more accessible, and analytical techniques have to adapt to larger data sets. This study showed that the partitioning of the four gene regions did little to influence th e underlying phylogenetic signal. By selecting commonly used genes ( rbcL, atpB, matK ), the study has allowed for potential integration of Simaroubaceae into larger scale Tree of Life and bar-coding projects. Additionally, with the use of phyC a less well-studied gene, the phyloge ny has improved our understanding of the utility of this nuclear marker. Historical biogeography is a fi eld undergoing considerable change, so the application of the latest techniques used in Chapter 4 is para mount to maximizing the potential of geographic and phylogenetic data. Pantropical clades are ofte n logistically more difficult to sample, given their widespread nature, and for Simaroubaceae this is particularly true with the relatively low number of species, leading to isolated lineages and little sympatry. However, this was not a

PAGE 140

140 limiting factor, with successful DNA extraction fr om herbarium material and contributions of silica-dried leaves from botanists worldwide. Robust fossil calibration with a combination of Ailanthus and Leitneria fossil fruits has produced a chronogram that proposes a Late Cretaceous origin for the family, a time scale similar to other pantropical families. Despite the many uncertainties associated with biogeographical analyses for geographically widespread and ancient groups, such as errors in molecular rate estimation and the relative simplicity of ancestral area reconstruction, many clear patterns in th e biogeographic history of Simaroubaceae can be inferred from phylogenetic and fossil data. The fa mily shows patterns of overland migration via putative land bridges, coupled with a number of inferred long-distance dispersal events. The break-up of Gondwanaland, traditionally the mo st parsimonious explanation for tropical disjunctions (Raven and Axelr od, 1974), has now been replaced by hypotheses incorporating knowledge of phylogenetic relationships, more realistic divergence dates, availability of dispersal routes (Morley, 2003), and a general a cceptance of long-distance dispersal as a major driving force in extant plant distributions (Gi vnish and Renner, 2004; Lavin et al., 2004; Renner, 2004a; De Queiroz, 2005). The North Atlantic Land Bridge is often implicated in pantropical disjunctions, but testing a variet y of potential ancestral ranges between the Old World and the New World has illustrated the importance of a trans-Beringial connection for Simaroubaceae. The biogeographic model requires further refine ment to accommodate stratified ancestral ranges and more realistic fossil integrat ion, but is an excellent first st ep in moving analyses towards a model-based approach, which can include testing hypotheses of a vari ety of biogeographic scenarios. Chapter 5 takes a data exploration approach towards patterns of diversification in the family, similar to the recent study by M oore and Donoghue (2007). Morphological data do

PAGE 141

141 contain a phylogenetic signal which recovers some of the deep-level relationships found with molecular data, and also shows many of the gene ra to be well-characterized morphologically. Up to four significant diversificat ion rate increases are found, depending on the methods used. These shifts are typically associated with biogeographic dispersal events but a number of dispersals have not resulted in diversific ation rate increases; therefore, a variety of factors, both environmental and intrinsic to the organism, are probably involved (de Queiroz, 2002). The technique used to examine putative key innova tions is only recently developed, and requires further simulation-based studies to determine its statistical propertie s and behavior in a variety of phylogenetic systems. However, it has allowed th e exploration of a nu mber of morphological characters and how state changes affect the frequency of cladogenesis. By a conservative estimation, none of the 36 characters examined had a significant association, but this part of the study would benefit from improved sampling for st ochastic mapping. Extendi ng this approach to the Sapindales would hopefully yield a wealth of interesting data. The order would provide a broad phylogenetic spectrum across which more recent character-state changes could be considered independent, whilst th e types of morphologica l characters examined are comparable, for example, floral structure, fruit morphology, etc. In conclusion, only with these types of detailed studies of in dividual clades can we begin to create an overall picture of the evolutionary history of tropical and temperate biodiversity. For such a small and relatively understudied family, Simaroubaceae has produced a wealth of insights into the modeli ng of molecular data, pantropical historical biogeography, morphological evolution and diversification.

PAGE 142

142APPENDIX A SPECIMEN DATA FOR MOLECULAR ANALYSES Sources of DNA and GenBank accession numbers for sequence data. Taxon name s in parentheses ar e original determinations. Taxon Voucher rbcL atpB matK phyC Ailanthus altissima J.W.Clayton 14 (FLAS) EU042978 EU042770 EU042840 EU042911 A. altissima var. tanakai Chase 16982 (K) EU042979 EU042771 EU042841 EU042912 A. fordii Lau 038 (MO) EU042980 EU042772 EU042842 A. integrifolia B.Hyland 15229 (QRS) EU042981 EU042773 EU042843 A. triphysa Fernando 1540 (UNSW) EU042982 EU042774 EU042844 EU042913 Amaroria soulameoides No voucher, ex SUVA, Fiji U38923 AF066856 Brucea antidysenterica Simon et al 1140 (MO) EU042983 EU042775 EU042845 EU042914 B. ferruginia Acc. 19073747 NBG Belgium EU042984 EU042776 EU042846 EU042915 B. guineensis McPherson 18015 (MO) EU042985 EU042777 EU042847 EU042916 B. javanica JTH 134 EU042986 EU042778 EU042848 EU042917 B. mollis J.R.I.Wood 6887 (E) EU042987 EU042849 EU042918 B. tenuifolia Mwangoka and Kalage 2687 (MO) EU042988 EU042779 EU042850 EU042919 Castela coccinea A.Krapovickas and C.L.Cristobal 44166 (MO) EU042989 EU042780 EU042851 EU042920 C. erecta Webster and Armbruster 23577 (MO) EU042991 EU042782 EU042853 EU042922 C. erecta subsp. texana P.Tenorio 20405 (MO) EU042990 EU042781 EU042852 EU042921 C. retusa Abisai Garcia M. 2830 (MO) EU042992 EU042783 EU042854 EU042923 C. tortuosa Juan Torres 00172 (MO) EU042993 EU042784 EU042855 EU042924 C. tweedii G.Hatschbach et al. 72435 (M O) EU042994 EU042785 EU042856 EU042925 Eurycoma apiculata No voucher ex FRIM, Malaysia EU042995 EU042786 EU042857 EU042926 E. longifolia Gwee and Samsuri GW19 (SING) EU042996 EU042787 EU042858 EU042927 Gymnostemonn zaizou L. Ake 19117 (MO) EU042997 EU042788 EU042859 EU042928 Hannoa chlorantha D.K.Harder et al. 3732 (MO) EU042998 EU042789 EU042860 EU042929 H. klaineana M.Merello et al. 1584 (MO) EU042999 EU042790 EU042861 EU042930 H. undulata Madsen 4075 (MO) EU043001 EU042792 EU042863 EU042932 H. undulata H.H.Schmidt et al. 3336 (MO) EU043000 EU042791 EU042862 EU042931 Holacantha emoryi T.K. Lowrey and C.J. Quinn,1876 (UNM) EU043002 EU042793 EU042864 EU042933 Leitneria floridana J.R.Abbott 14212 (FLAS) EU043003 EU042794 EU042865 EU042934 Nothospondias staudtii H.P.Bourobou Bourobou 474 (MO) EU043004 EU042795 EU042866 EU042935 Odyendea gabonensis G.Walters et al. 577 (MO) EU043005 EU042796 EU042867 EU042936 O. zimmermannii L.B. Mwasumbi 14191 (MO) EU043006 EU042797 EU042868 EU042937

PAGE 143

143Perriera madagascariensis Noyes et al 1079 (MO) EU043007 EU042798 EU042869 EU042938 Picrasma antillana P.Aceredo Rdgz, A.Siaca 4262 (MO) EU043009 EU042800 EU042871 P. crenata A.C.Cervi and R. Spichiger 6861 (NY) EU043010 EU042801 EU042872 EU042940 P. javanica No voucher, wild collected, China. EU043011 EU042802 EU042873 EU042941 P. quassioides Acc. 19510406 RBGE cultivated EU043008 EU042799 EU042870 EU042939 P. excelsa Hill 25443 (MO) EU043012 EU042803 EU042874 EU042942 Picrolemma sprucei ( P. pseudocoffea ) G. Bourdy GB2988 (CAY) EU043013 EU042804 EU042875 EU042943 P. sprucei C.Grandez and N.Jaramillo 2871 (MO) EU043014 EU042876 EU042944 Pierreodendron africanum Terese Butler Hart 1386 (MO) EU043015 EU042805 EU042877 EU042945 Quassia africana McPherson 16672 (MO) EU043016 EU042806 EU042878 EU042946 Q. amara Chase 18959 (K) EU043017 EU042807 EU042879 EU042947 Samadera baileyana UNSW 22894 (UNSW) EU043018 EU042808 EU042880 EU042948 S. bidwillii Craven and Walker 9339 EU043019 EU042809 EU042881 EU042949 S. indica JTH 138 EU043020 EU042810 EU042882 EU042950 S. sp. B Fernando 1538 (UNSW) EU043021 EU042811 EU042883 EU042951 S. sp. C (Quassia Barong) A. Ford 4679 EU043022 EU042812 EU042884 EU042952 Simaba cedron N.P.Taylor 690 (MO) EU043024 EU042814 EU042886 EU042953 S. cuneata R.P. Lyra-Lemos 4082 (NY) EU043025 EU042815 EU042887 EU042954 S. cuneata M.R. Barbosa s.n. EU546244 EU546227 EU546232 EU546238 S. cf ferruginia Pott and Franco 6177 (E) EU043027 EU042817 EU042889 EU042956 S. glabra ( S. cf blanchetii ) Ratter et al R8004 (E) EU043023 EU042813 EU042885 S. glabra J.A.Ratter and Valdir P. de Lima 6719 (MO) EU043028 EU042818 EU042890 EU042957 S. guianensis (S. cuspidata) Korning and Thomsen 47632 (AAU) EU043026 EU042816 EU042888 EU042955 S. guianensis Martin Timana 3683 (MO) EU043030 EU042820 EU042892 EU042959 S. guianensis L.Barrabe and M.Pechberty 185 (CAY) EU043029 EU042819 EU042891 EU042958 S.guianensis (S. orinocensis) P. Grenand 3267 (CAY) EU043034 EU042824 EU042896 EU042963 S. insignis Pirani et al. 4517 (NY) EU043031 EU042821 EU042893 EU042960 S. morettii M. Prevost and D. Sabathier 2987 EU546245 EU546228 EU546233 EU546239 S. orinocensis (S. multiflora ) Rimanchi 10347 (MO) EU043032 EU042822 EU042894 EU042961 S. orinocensis (S. multiflora ) Gillespie 2576 (MO) EU043033 EU042823 EU042895 EU042962 S. aff paraensis G.L. Farias 301 EU546246 EU546234 EU546240 S. polyphylla Molino J.-F. et al 1998 (CAY) EU043035 EU042825 EU042897 EU042964 S. suffruticosa J. Ratter et al. 3532 EU546247 EU546229 EU546235 EU546241 S. trichilioides R.C. Forzza et al. 455 EU546248 EU546230 EU546236 EU546242 Simarouba amara A.Araujo et al. 353 (MO) EU043036 EU042826 EU042898 EU042965

PAGE 144

144S. berteroana WT 14671 EU546249 EU546231 EU546237 EU546243 S. glauca (cultivated) J.R.Abbott 19605 (FLAS) EU043037 EU042827 EU042899 EU042966 S. glauca (wild collected) Acc. 19843216 NBG Belgium EU043038 EU042828 EU042900 EU042967 S. tulae Taylor 10589 (MO) EU043039 EU042829 EU042901 EU042968 S. versicolor Nee 39012 (MO) EU043040 EU042830 EU042902 EU042969 Soulamea amara Chambers 78 (MO) EU043041 EU042831 EU042903 EU042970 S. terminalioides Robertson 2529 (MO) EU 043043 EU042833 EU042905 EU042972 S. sp. J.Munziger and McPherson 589 (MO) EU043042 EU042832 EU042904 EU042971 OUTGROUPS Acer saccharum Chase 106 (NCU) L13181 AF035893 AY724265 Melia azaderach J.R.Abbott 8456 (FLAS) EU042973 EU042764 EU042834 EU042906 Swietenia mahogani/ macrophylla J.W.Clayton 12 (FLAS) AY128241 EU042765 EU042835 EU042907 Casimiroa edulis J.R.Abbott 8156 (FLAS) EU042975 EU042767 EU042837 EU042909 Zanthoxylum sp. J.W.Clayton 15 (FLAS) EU042976 EU042768 EU042838 EU042910 Cneorum tricoccon J.W.Clayton 13 (FLAS) EU042977 EU042769 EU042839 Unknown Meliaceae Acc. 19514782 NBG Belgium EU042974 EU042766 EU042836 EU042908

PAGE 145

145 APPENDIX B SOURCES OF MORPHOLOGICAL DATA Morphological data obtained from herbarium specim ens (see Appendix C), botanical literature, or both. Species Herbarium specimen Literature Ailanthus altissima X X A. excelsa X X A. fordii X X A. integrifolia X A. triphysa X X Amaroria soulameoides X Brucea antidysenterica X X B. guineensis X X B. javanica X X B. macrocarpa X B. mollis X X B. tenuifolia X Castela calcicola X C. coccinea X X C. depressa X C. erecta X X C. galapegia X X C. jacquinifolia X C. macrophylla X C. peninsularis X C. retusa X X C. spinosa X C. tortuosa X X C. tweedii X X Eurycoma apiculata X E. harmandiana X E. longifolia X X Gymnostemon zaizou X X Hannoa chlorantha X X H. ferruginia X X H. klaineana X X H. sp. A X H. sp. B X H. undulata X X Holacantha emoryi X X H. stewartii X X Iridosma letestui X Laumoniera bruceadelpha X Leitneria floridana X X Nothospondias staudtii X X Odyendea gabonensis X X Picrasma antillana X X P. crenata X X P. cubensis X P. excelsa X X P. mexicana X P. quassioides X X P. selleana X X

PAGE 146

146Perriera madagascariensis X X Picrolemma huberi X P. sprucei X X Pierreodendron africanum X X Quassia africana X X Q. amara X X Samadera baileyana X X S. bidwillii X X S. indica X X Simaba cavalcantei X S. cedron X X S. cuneata X X S. docensis X S. ferruginia X X S. glabra X X S. guianensis X X S. insignis X S. obovata X X S. orinocensis X X S. paraensis X X S. polyphylla X X S. praecox X S. suffruticosa X S. trichilioides X S. warmingiana X Simarouba amara X X S. berteroana X X S. glauca X S. laevis X S. tulae X X S. versicolor X X Soulamea amara X X S. cardioptera X S. cycloptera X S. dagostinii X S. fraxinifolia X S. morattii X S. muelleri X X S. pancheri X X S. pelletieri X S. rigaultii X S. terminalioides X S. tomentosa X S. trifoliata X

PAGE 147

147APPENDIX C SPECIMEN DATA FOR MORPHOLOGICAL ANALYSES Voucher specimens used in examin ation of morphologi cal characters Species Voucher Ailanthus altissima M.S.Franc 75 (FLAS); D.R.Windler, M.Burch & P.H.Ke nnan 3698 (FLAS); A.E. Radford 33944 (FLAS); E. Prendes (FLAS);A Cuthbert (FLAS) A. excelsa E.West (FLAS) A. fordii C.P.Lau 038 (MO) A. triphysa J.F.Maxwell 89-244 (MO) Brucea antidysenterica W. Kindeketa 309 (MO); G.Simon, N.Senti & Y. Raphael 1140 (MO); G.J.H.Amshoff 1970 (MO) B. guineensis H.H.Schmidt et al. 3594 (MO); G.McPherson 18015(MO) B. javanica P.J.Martin 36966 (MO); Paul Chai et al. 33226 (MO); Hu & But 22470 (MO) B. mollis J.R.I.Wood 6887 (MO) Castela coccinea A. Krapovickas & C.L.Cristbal 44166 (MO); T. L. Gragson 159 (MO); M.Nee 48131 (NY) C. erecta T.B. Croat & D.P. Hannon 66020 (MO); P. Tenorio L. 20405 (MO); G.L. Webster & W. S. Armbruster 23577 (MO) C. galapegia R.D. Suttkus 66-42-2 (FLAS) C. retusa Abisai Garcia M. et al. 2830 (MO) C. tortuosa Juan Torres 00172 (MO) C. tweedii G. Hatschbach, R. Goldenberg & J.M. Silva 72435 (MO) Eurycoma longifolia K. Larsen et al. 43047 (MO) Gymnostemon zaizou L. Ak 19117 (MO) Hannoa chlorantha D.K.Harder, M.Bingham, B.Luwiika & N. Zimba 3732 (MO) H. ferruginia D.W.Thomas & H.L.Mcleod 5301 (MO) H. klaineana M.Merello, H.H.Schmidt, J.Amponsah, M.Chintoh & K.Baah 1584 (MO) H. sp. A (Brucea tenuifolia) M.A.Mwangoka & A.Kalage 2687 (MO) H. sp. B (Odyendea zimmermannii) L.B.Mwasumbi 14191 (MO) H. undulata J.E.Madsen 2909 (MO); J.E.Madsen 2909 (AAU); J.E.Ma dsen 4075 (AAU); C.C.Jongkind & C.M.J.Nieuwenhuis 1887 (MO) Holacantha emoryi R.F. Thorne 45120 (FLAS) H. stewartii M.C.Johnston, T.L.Wendt & F. Chiang C. 11391 (FLAS) Leitneria floridana R.K.Godfrey 62862 (FLAS); D.Demaree 71628 (FLAS); R. K.Godfrey & A.F.Clewell 63248 (FLAS); A. Gholson (FLAS); W.Judd 3323 (FLAS); J.R.Abbott 9047 (FLAS) Nothospondias staudtii H.P.Bourobou Bourobou 474 (MO)

PAGE 148

148Odyendea gabonensis J.J. de Wilde 8470 (MO); G. McPherson 16947 (MO); G. Walter, J.Stone, G.N.Essouma, A.Mintsa & L.Ndong 577 (MO) Picrasma antillana P. Acevedo Rdgz. A. Siaca 4262 (MO) P. crenata J.C.Lindeman & J.H.de Haas 992 (NY); A.C.Cervi & R. Spichiger 6861 (NY); W.W.Thomas, J.Kallunki & J.Jardim 11919 (NY) P. excelsa M.Nee, R. Vsquez, G.Coimbra & A. Becerra 49143 (NY) G.R.Proctor 45915 (MO), J. Betancur 1338 (MO), S. R. Hill 25443 (MO), A.Grijalva & Mario S ousa 30453 (MO), G.A.Goodfriend (FLAS) P. quassioides K.Sohma 3001(MO); S.Tsugaru & T.Takahashi 18 049 (MO); S.Tsugaru & M.Sawada 18551 (MO) P. selleana W.S.Judd 4400 (FLAS) Perriera madagascariensis R.D.Noyes et al. 1079 (MO) Picrolemma sprucei C.Grndez & N. Jaramillo 2871 (MO); R. Vsquez a nd N. Jaramillo 5674(MO);G.Bourdy 2988 (CAY); M.F.Prvost, D.Sabatier & J.-F.Molino 4440 (CAY) Pierreodendron africanum T. Butler Hart 1386 (MO) Quassia africana F.J.Breteler 6597 (MO);G.McPherson 16672 (MO); X.M. van der Burgt, D.Ndoum, B.S.van Gemerden & S. Gideon 529 (WAG); J.J.F.E. de Wilde 8296 (MO) Q. amara A. Welsing, M, Merello & H. Schmidt 10 (MO); M.Nee 8638 (MO); R.Rueda & H. Cuadros 357 (MO); J.Gonzlez 511 (MO) Samadera baileyana B.Gray 1732 (MO) S. bidwillii D.Halford Z155 (MO) S. indica R.Rabevohitra 2170 (MO); W.Meijer 10134 (MO); S.Waas 378 (MO) Simaba blanchetii J.A.Ratter, S. Bridgewater & J.Batista 8004 (E) S. cedron A.Gentry & A.Perry 77869 (MO); N.P.Taylor 690 (MO) S. cuneata R.P.Lyra-Lemos 4082 (NY) S. glabra J.A.Ratter & V.P. de Lima 6719 (MO) S. guianensis M.Timana 3683 (MO); J.J.Pipoly et al. 12746 (MO); D.C.Daly et al. 8303 (NY); R.Vsquez et al. 14174 (MO); L. Barrab & M.Pechberty 185 (CAY); J.Korning & K.T homsen 47632 (AAU); J.J. de Granville 16641 (CAY) S. insignis J.R.Pirani et al. 4517 (NY) S. obovata J.J.Wurdack & L.S.Adderley 42989 (MO) S. orinocensis L.J.Gillespie 2576 (MO); M. Rimanchi 10347 (MO);R. Liesner & A.C. Gonzlez 5613 (MO); C.C.Berg & A.J.Henderson BG672 (MO) S. paraensis P.Vinha 844 (NY); S.Espinoza & C.Gualinga 827 (MO) S. polyphylla M. Alexiades 189 (NY); M.Aulestia 2227 (MO); J.-F Molino, M.-F.Prvost & D. Sabatier 1998 (CAY) S. warmingiana J.G.Jardim et al. 912 (NY) Simarouba amara P.Moreno & J.C.Sandino 6481 (MO); P.Moreno & J.C.Sandino 6167 (MO); R.Liesner & A. Gonzlez 11470 (MO); Quevedo 2399 (NY) S. berteroana T. Zanoni & F. Jimnez 45005 (MO); T.Zanoni, J. Pimentel & R. Garca 38070 (FLAS); T.Zanoni, J. Pimentel &

PAGE 149

149R. Garca 38071 (FLAS) S. tulae C.M.Taylor 10589 (MO) S. versicolor L.P. de Queiroz & N.S.Nascimento 4133 (NY); M.Nee 39012 (MO);J.G.Jardim et al. 888 (NY); A.M. de Carvalho 4062 (MO); S.Tsugaru & Y. Sano B192 (NY) Soulamea amara A.F. Chambers 78 (MO) S. cardioptera G. McPherson 3292 (MO) S. muelleri J. Munzinger 345 (MO) S. pancheri G.McPherson 2080 (MO); J. Munzinger 955 (MO) S. terminalioides S.A. Robertson 2529(MO)

PAGE 150

150APPENDIX D MORPHOLOGICAL CHARACTERS USED IN PHYLOGENETIC ANALYSES Morphological characters and character states used in phylogenetic analyses Justification is provided where characters and cha racter state delimitations mi ght be considered subjective. Character States 1 Mean height (life form) 0 Shrub or small tree < 12 m; 1 Large tree > 15m Mean height showed a gap between 12 m and 15 m, at which the boundary between character states was drawn. 2 Thorns 0 Absent; 1 Present 3 Branchlet hair 0 Glabrous; 1 Pubescent towa rds branch ends, esp ecially young shoots Glabrous shoots may have shown very occasional hairs; pubescence when present was typically dense FrequencyMean height

PAGE 151

1514 Pinnation 0 Imparipi nnate; Paripinnate Certain specimens of Ailanthus and Simarouba showed both conditions, and were coded as ambiguous 5 Leaflet phyllotaxy 0 Opposite to subopposite above; 2 Alternate above Most genera had some variation in degree of alternation on lower leaflets, but this character refers specifically to leaflets in Simarouba which are distinctly alternating al ong the entire length of the rachis 6 Mean no. lflt prs 0 One to seven; 1 Eight or more The break between the character states was drawn based on a gap in the range of m ean number of leaflet pairs 7 Leaflet shape 0 Primarily ov ate to elliptic; 1 Primarily oblong; 2 Primarily obovate Oblong and obovate leaflets were a distinct departure from the other leaflet shapes, which were more generally categorised as ovate to elliptic (see Judd et al. (2001 ) for character defin itions). Primarily refers to the dominant shape seen in the lower leaflets. 8 Leaf shape 0 Primarily ov ate to elliptic; 1 Primarily oblong; 2 Primarily obovate See character 7. 9 Leaf type 0 Leafless; 1 Simple or unifoliolate; 2 Pinnately compound FrequencyMean number of leaflets (upper bound)

PAGE 152

152 Leafless refers to Holacantha which is essentially leafless but may have scale-like structures homologous to leaves. 10 Leaflet apex 0 Rounded, truncate or emarginate ; 1 Acute to shortly acuminate; 2 Long acuminate Long acuminate refers to the apex seen in Picrolemma Quassia and one species of Picrasma in which the significantly narrowed portion of the tip has a visible length along which the sides are parallel. 11 Leaf apex 0 Rounded, truncate or emarginate; 1 Acute to shortly acuminate; 2 Long acuminate See character 10. 12 Margin teeth 0 Entire; 1 Crenulat e or serrate or shallowly undulating Shallowly undulating was seen in species of Brucea in which the margin is wavy on the same plane as the leaf blade (as oppose to undulations perpendicular to the leaf blade in character 14). 13 Margin revolute 0 Flat or slightly revolute; 1 Strongly revolute Strongly revolute is when the margin curls at least 180 i.e. completely back on itself or more. 14 Margin undulation 0 Flat; 1 Finely undulating Only Picrolemma showed the finely undulating ch aracter state, where the frequency of vertical undulation was much higher than any other taxon. 15 Mature leaf hairs abaxially 0 Glabrous to sparsely hairy; 1 Densely hairy on lamina For the very few taxa in which there was not a clear distinction between the states due to moderate and varying degrees of pubescence, the character was coded as ambiguous. 16 Mature leaf hairs adaxially 0 Glabrous to sparsely pubescent; 1 Densely pubescent See mature leaf hairs abaxially. 17 Leaf axes hairs 0 Glabrous or occasional hairs; 1 Hairy Leaf axes refers to petioles, petiolule s and the rachis. See characters 15 and 16. 18 Glands 0 Absent; 1 Present Refers to all types of glands seen. 19 Gland type 0 Punctate, light colored; 1 Prominent, usually dark colored Prominent refers to dot-like or raised glands, rather than the circular, pitted glands which usually had a lighter coloration 20 Punctate leaf gland location 0 Scattered below; 1 In vein forks below; 2 Above, marginal towards apex; 3 Above, marginal tow ards base; 4 Above, around entire margin; 5 Underneath, at base Vein forks are where secondary venation running out from the mid-rib divides in two, typically as the vein loops near the margin. This is opposed to marginal glands which were not obviously associated with the venation. At the base underneath refers to two or three cases where thr ee to four glands were situated in th e attenuate portion of the lamina base. 21 Petiolule 0 Sessile or subsessile; 1 Petiolulate 22 Rachis jointed 0 Not jointed; 1 Jointed 23 Rachis wing 0 Absent; 1 Present 24 Secondaries prominence above 0 Indistinct; 1 Sulcate or flush but still visible; 2 Prominent or clearly visible Flush refers to faint lines which appear neither sunken in th e lamina or raised, but are dark er or lighter than the lamina. 25 Secondaries prominence 0 Indistinc t; 1 Sulcate; 2 Slightly sulcate, flush or slightly raised; 3 Prominent

PAGE 153

153beneath See character 24. 26 Stipules 0 Absent; 1 Present 27 Inflorescence type 0 Typical thyrse; 1 Fascicle-like in leaf axils; 2 Umbellate; 3 Short, broad, rounded thyrse 28 Typical thyrse structure 0 Second axes below absent (flowers appearing fasciculate) or shor t and pedicel-like; 1 Second axes similar to primary axis 29 Position of inflorescence 0 Axillary; 1 Terminal 30 Inflorescence axis hairs (incl. pedicel) 0 Glabrous or very sparsely hairy; 1 Moderately to densely hairy See character 15. 31 Hair type 0 Simple; 1 Glandular Glandular refers solely to the glandular-capitate hairs of Eurycoma 32 Pedicel articulation 0 None or at base; 1 Halfway along pedicel 33 Sex 0 Bisexual; 1 Unisexual (or androdioecious) Although some unisexual plants were reported as having perfect flowers or being androdioecious in the literature, this character was coded on herbarium speci mens where possible, and unisexual was used if perfect flowers were rare. 34 Merosity 0 Trimerous; 1 Tetramerous; 2 Pentamerous; 3 Other Other refers to Castela and Iridosma. 35 Calyx shape 0 Slightly fused at base to halfway; 1 Almost entirely fused, triangular very short sepals; 2 Ripping into uneve n lobes 36 Calyx glands 0 Absent; 1 Present 37 Calyx hairs 0 Glabrous; 1 Glabrous except for marginal hair s, ciliolate; 2 Pubescent abaxially only; 3 Pubescent adaxially only; 3 Pubescent on both sides and margins Pubescence here refers to any type of hair, as taxa are reported to by villous, strigose, puberulent and tomentose as well as pubescent. 38 Petal hair 0 Glabrous, or occasional hair; 1 Pubescent adaxially; 2 Pubescent abaxially; 3 Pubescent on both sides See character 37. 39 Petal shape 0 Oblong, linear; 1 Ovate, obovate Oblong or linear is considered a special case of petal shape, where petals are stra p-like. Ovate, obovate refers to all other shapes observed. 40 Corolla length 0 2 mm'; 1 '2.1 8 mm'; 2 '9 14 mm'; 3 '17+ mm' States 2 and 3 were delimited as gaps in the distribution of corolla length, with Quassia amara flowers reaching a maximum of 50 mm in length. States 0 and 1 had no obvious break, however, on observation of the taxa making up the lower end of the distribution, 15 of the 16 taxa with mean corolla lengths equal to or less than 2 mm were members of

PAGE 154

154Clade III. Only two members of Clade III had corollas longer than 2.5 mm, and never did they exceed 3 mm. Four members of Clade III had no data on flower length. 41 Corolla aestivation 0 Valvate; 1 Indup licate-valvate; 2 Imbricate; 3 Contorted 42 Petal color (abaxial) 0 White-y ellow-green; 1 Pink-red-violet Petal color could typically vary betwee n white, creamy white, pale yellow or green between specimens of the same species and between descriptions of the same species, similarly for pink red and violet. However, only in one or two instances did color vary between states 1 and 2, and these were coded as ambiguous. 43 Disc hair 0 Glabrous; 1 Pubescent 44 Disc shape 0 Very short, ri ng-like, or absent; 1 Fleshy, l obed, broader than tall; 2 Columnar, or occ. gynoecium embedded in side 45 Androecium 0 Haplostemonous; 1 Diplostemonous; 2 Pleiostemonous, FrequencyMean corolla length (upper bound)

PAGE 155

15546 Filament hairiness 0 Glabrous; 1 Sparsely to densely hairy below, sparse or glabrous above 47 Filament appendage 0 Absent; 1 Present 48 Appendage fusion 0 Mostly adnate, just the tip free; 1 Substantial free portion Free portion free over half of the to tal length of the filament or more. 49 Dehiscence 0 Latrorse to extrorse; 1 Introrse 50 Staminodes in staminate flower 0 Absent; 1 Present 51 Number of carpels 0 Six or more; 1 One; 2 Two; 3 Three; 4 Four; 5 Five 52 Fusion of ovary 0 Free or fused at base; 1 Fused except at apex Fusion up to the apex refers to the gynoecia of Soulamea only. 53 Style length 0 Free portion shorter, fused along length; 1 Free portion equal or longer, fused at top of ovary In taxa with free portion shorter, styles were typically much longer than in state 1, and we re associated with capitate or very shortly lobed stigmatic branches. 54 Stigmatic branch shape 0 Reniform, fleshy, thick; 1 Linear, elongate, upright; 2 Linear, elongate, recurved; 3 Inconspicuous to slightly lobed 55 Fruit shape 0 Ellipsoid, obovoid or slightly ovoid; 1 Strongly ovoid; 2 Globose; 3 Samadera shape; 4 Elongate ( Ailanthus ); 5 Obcordate ( Soulamea) Strongly ovoid was restricted to the fruits of Brucea 56 Pericarp 0 Dry; 1 Fleshy 57 Fruiting gynophore 0 Not or slightly en larging; 1 Enlarging considerably, fleshy 58 Maximum drupe size 0 13 mm'; 1 '13.1 25 mm'; 2 '> 29 mm', Maximum drupe size was used rather than a mean because fruits in a number of specimens may not have reached maturity, and the measurement was preferred to be as close to matu re size as possible. The break between states 1 and 2 was according to a gap in the distribution, and se gregated the very large fruits such as Gymnostemon and Pierreodendron State delimitation between 0 and 1 was more ambiguous. However, it was useful to break up the large range (0 25 mm) into smaller ranges for observation of fruit evolution in character mapping. These states should not be regarded as robust in terms of phylogenetic analysis.

PAGE 156

156 59 Bicarination 0 None; 1 Pres ent but fruit not flattened; 2 Present, fruit flattened 60 Drupe color when ripe 0 Orange to red to blue-black; 1 Green to yellow 61 Hairs on fruit 0 Glabrous or with occasional hairs; 1 Pubescent 62 Fruit vascular bundle 0 Towards seed; 1 Intramarginal Ailanthus only. 63 Stylar scar position 0 Level with seed centre; 1 Above seed; 2 Below seed Ailanthus only. 64 Endocarp surface 0 Crustaceous, smooth; 1 Reticulately wrinkled 65 Endocarp type 0 One layer of sclereids; 1 Two layer of sclereids FrequencyMaximum fruit length (upper bound)

PAGE 157

157 Data from Fernando and Quinn (1992). 66 Crystalliferous cells 0 Absent; 1 Present Data from Fernando and Quinn (1992). 67 Mesocarp secretory canals 0 Absent; 1 Present Data from Fernando and Quinn (1992). 68 Pollen amb shape 0 Circular to subcircular; 1 Trangular to subtriangular Data from Basak (1963, 1967), Moncada and Machado (1987) and Zavada and Dilcher (1986) 69 Pollen exine 0 Reticulate; 1 Striato-reticulate; 2 Striate; 3 Smooth Data from Basak (1963, 1967), Moncada and Machado (1987) and Zavada and Dilcher (1986) 70 Pollen ora 0 Lalongate; 1 Square Data from Basak (1963, 1967), Moncada and Machado (1987) and Zavada and Dilcher (1986) 71 Pollen aperturate 0 Planaperturate; 1 Colpi in middle of flat angles; 2 Angulaperturate Data from Basak (1963, 1967), Moncada and Machado (1987) and Zavada and Dilcher (1986)

PAGE 158

158 LIST OF REFERENCES Abbe, E. C., and T. T. Earle. 1940. Inflor escence, floral anatomy and morphology of Leitneria floridana. Bull. Torrey Bot. Club 67:173. Akaike, H. 1974. A new look at stat istical model identification. IEEE Transactions on Auto matic Control 19:716. Alfaro, M. E., S. Zoller, and F. Lutzon i. 2003. Bayes or bootstrap? A simulation study comparing the performance of Bayesian Markov chain Monte Carlo sampling and bootstrapping in assessing phylogenetic confidence. Mol. Biol. Evol. 20:255. Alfaro, M., F. Santini, and C. D. Brock. 2007. Do reefs drive diversificati on in marine teleosts? Evidence from the pufferfish and their allies. Evolution 61:2104. APG II. 2003. An update of the angiosperm phylogeny group classification for the orders and families of flowering plants: APG II. Bot. J. Linn. Soc. 141:399. Armbruster, W. S. 1992. Phylogeny and the evolution of plantanimal interactions. Bioscience 42:12. Aubrville, A., and F. Pellegrin. 1937. Gymnostemon A. et P., genre nouveau de la Cte dIvoire, voisin dun endmique de Madagascar. Bul. Soc. Bot. Fr. 84:181. Aubrville, A. 1962. Simaroubaces. Pages 3352 in Flore du Gabon (A. Aubrville, N. Hall, B. Descoings, J. Koechlin, and A. Cavaco, ed s.), volume 3. Musum National d'Histoire Naturelle, Paris, France. Azuma, H., J. G. Garcia-Franco, V. Rico-Gra y, and L. B. Thien. 2001. Molecular phylogeny of the Magnoliaceae: the biogeography of tropical and temperate disjunctions. Am. J. Bot. 88:2275. Backer, C. A., and R. C. B. Van den Brink. 1965. Flora of Java Vol II Angiospermae. Families 111. N.V.P. Noordhoff, Groningen, Netherlands. Barker, G. M. 2002. Phylogenetic diversity: a qu antitative framework for measurement of priority and achievement in biodiversity conservation. Bot. J. Linn. Soc. 76:165. Barker, F. K., and F. M. Lutzoni. 2002. The util ity of the incongruence length difference test. Syst. Biol. 51:625. Bartish, I. V., U. Swenson, J. Munzinger, and A. A. Anderberg. 2005. Phylogenetic relationships among New Caledonian Sapotaceae (Ericales): Mo lecular evidence for generic polyphyly and repeated dispersal. Am. J. Bot. 92:667. Basak, R. K. 1963. Pollen morphology of Indian Simaroubaceae. Bull. Bot. Surv. India 5:381 397.

PAGE 159

159 Basak, R. K. 1967. Studies on the pollen morphology of the Simaroubaceae. Bull. Bot. Surv. India 9:63. Baum, D. A., R. L. Small, and J. F. Wendel. 1998. Biogeography and floral evolution of baobabs ( Adansonia Bombacaceae) as inferred from multip le data sets. Syst. Biol. 47:181. Becerra, J. 2005. From The Cover: Timing the origin and expansi on of the Mexican tropical dry forest. Proc. Natl. Acad. Sci. USA 102:10919 Bennett, A. W. 1872. Simarubeae. Pages 5173 in Flora of British India (J.D. Hooker, ed.) volume 1, Ranunculaceae to Sapindaceae. L. Reeve and Co, London, UK. Bennett, M. D., and I. J. Leitch. 2005. Nucl ear DNA amounts in angiosperms: progress, problems and prospects. Annals Bot. 95:45. Benvenuti, S. 2007. Weed seed movement and dispersal strategies in the agricultural environment. Weed Biol. Manag. 7:141. Bergsten, J. 2005. A review of long-br anch attraction. Cladistics 21:163. Bernadello, L. M., L. B. Stiefkens, and M. A. Piovano. 1990. Nmeros cromosmicos en dicotiledneas Argentinas. Boletn de la Sociedad Argentina de Botnica 26:149 Berry, E. W. 1916. The physical conditions and age indicated by the flora of the Alum Bluff Formation. Profess. Pap. U.S. Geol. Surv. 98E:41. Bradford, J. C. 2002. Molecular phylogenetics and morphological evolution in Cunonieae (Cunoniaceae). Ann. Mo. Bot. Gard. 89:491. Brandley, M. C., A. Schmitz, and T. W. Reed er. 2005. Partitioned Bayesian analyses, partition choice, and the phylogenetic relationships of Scincid Lizards. Syst. Biol. 54:373. Britton, T., C. L. Anderson, D. Jacquet, S. Lundqvist, and K. Bremer. 2007. Estimating divergence times in large phylogene tic trees. Syst. Biol. 56:741. Cannon, C. H., and P. S. Manos. 2003. Phylogeogr aphy of the Southeast Asian stone oaks ( Lithocarpus ). J. Biogeogr. 30:211. Cavalcante, P. 1983. Reviso taxonmica do gnero Simaba Aublet (Simaroubaceae) na Amrica do Sul. Publicaes Avulsas do Museu Goeldi 37. Belm, Par, Brasil. Chan, K., and B. R. Moore. 2002. Whole-tree met hods for detecting differe ntial diversification rates. Syst. Biol. 51:855. Chan, K., and B. R. Moore. 2005. SymmeTREE: whole-tree analysis of differential diversification rates. Bioinformatics 21:1709.

PAGE 160

160 Chase, M. W., Morton, C. M., and J. A. Kall unki. 1999. Phylogenetic rela tionships of Rutaceae: A cladistic analysis of the subfamilies using evidence from rbcL and atpB sequence variation. Am. J. Bot. 86:1191. Clayton, J. W., E. S. Fernando, P. S. Soltis, and D. E. Soltis. 2007. Molecular phylogeny of the Tree-of-Heaven family (Simaroubaceae) based on chloroplast and nuclear markers. Int. J. Plant Sci. 168:1325. Coddington, J. 1988. Cladistic tests of ad aptational hypotheses. Cladistics 10:229. Conti, E., T. Eriksson, J. Schnenberger, K. J. Sytsma, and D. A. Baum. 2002. Early Tertiary out-of-India dispersal of Crypteroniaceae: evidence from phylogeny and molecular dating. Evolution 56:1931. Corbett, S. L., and S. R. Manchester. 2004. Phytogeography and fossil history of Ailanthus (Simaroubaceae). Int. J. Plant Sci. 165:671. Cracraft, J. 1985. Biological di versification and its causes Ann. Mo. Bot. Gard. 72:794. Cronk, Q. C. B., M. Kiehn, W. L. Wagne r, and J. F. Smith. 2005. Evolution of Cyrtandra (Gesneriaceae) in the Pacific Ocean: the origin of a supert ramp clade. Am. J. Bot. 92:1017. Cronquist, A. 1944a. Studies in the Simaroubaceae I: The genus Castela. J. Arnold Arboretum 25:122. Cronquist, A. 1944b. Studies in the Simaroubaceae II: The genus Simarouba. Bull. Torrey Bot. Club 71:226. Cronquist, A. 1944c. Studies in th e Simaroubaceae III: The genus Simaba. Lloydia 7:81. Cronquist, A. 1944d. Studies in the Simaroubaceae IV : Resume of American genera. Brittonia 5:128. Cronquist, A. 1981. An integrated system of classification of flowering plants. Columbia University Press, New York, USA. Cummings, M. P., S. A. Handle y, D. S. Myers, D. L. Reed, A. Rokas, and K. Winka. 2003. Comparing bootstrap and posterior probability va lues in the four-taxon case. Syst. Biol. 52:477. Cunningham, C. W., K. E. Omland, and T. H. Oa kley. 1998. Reconstructi ng ancestral character states: a critical reappraisal. Trends Ecol. Evol. 13:361. Czaja, A. T. 1978. Strke und strkespeicherung bei Gefsspflanzen. Gustav Fischer Verlag, Stuttgart, Germany.

PAGE 161

161 Da Silva, M. F. D. G. F., and O. R. Gottlie b. 1987. Evolution of quassinoids and limonoids in the Rutales. Biochem. Syst. Ecol. 15:85. Darlu, P., and G. Lecointre. 2002. When does the in congruence length differe nce test fail? Mol. Biol. Evol. 19:432. Davis, C. C., C. D. Bell, S. Mathews, and M. J. Donoghue. 2002a. Laurasian migration explains Gondwanan disjunctions: evidence from Malp ighiaceae. Proc. Natl. Acad. Sci. USA 99:6833. Davis, C. C., C. D. Bell, P. W. Fritsch, and S. Mathews. 2002b. Phylogeny of AcridocarpusBrachylophon (Malpighiaceae): Implications for Tert iary tropical floras and Afroasian biogeography. Evolution 56:2395. Davis, C. C., C. O. Webb, K. J. Wurdack, C. A. Jaramillo, and M. J. Donoghue. 2005. Explosive radiation of Malpighiales supports a mid-Cretaceous origin of m odern tropical rain forests. Am. Nat. 165:E36E65. De Queiroz, A. 2002. Contingent pr edictability in evolution: key traits and divers ification. Syst. Biol. 51:917. De Queiroz, A. 2005. The resurrection of oceanic dispersal in historical biogeography. Trends Ecol. Evol. 20:68. Dick, C., and S. Wright. 2005. Tropical mountain cr adles of dry forest diversity. Proc. Natl. Acad. Sci. USA 102:10757. Dodd, M. E., J. Silvertown, and M. W. Chase. 1 999. Phylogenetic analysis of trait evolution and species diversity variation among angiosperm families. Evolution 53:732. Donoghue, M. J., C. D. Bell, and J. Li. 2001. P hylogenetic patterns in Northern Hemisphere plant geography. Int. J. Pl. Sci. 162:S41S52. Donoghue, M. J., and B. R. Moore. 2003. Toward an integrative historical biogeography. Integr. Comp. Biol. 43:261. Donoghue, M. J., and S. A. Smith. 2004. Patterns in the assembly of temperate forests around the Northern Hemisphere. Philos T. Roy. Soc. B 359:1633. Donoghue, P. C. J., and M. J. Benton. 2007. Rocks and clocks: calibrating the Tree of Life using fossils and molecules. Tr ends Ecol. Evol. 22:424. Dorofeev, P. I. 1994. Leitneria Pages 8 in Fossil flowering plants of Russia and adjacent states (L. Budantsev, ed.), volume 3, Leitn eriaceae-Juglandaceae. Kamarov Botanical Institute, Russian Academy of Sciences, St. Petersburg, Russia. Doyle, J. J., and J. L. Doyle. 1990. Isolati on of plant DNA from fresh tissue. Focus 12:13.

PAGE 162

162 Dreyer, D. L. 1983. Limonoids of the Rutaceae. Pages 215 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Drummond, A. J., S. Y. W. Ho, M. J. Phillip s, and A. Rambaut. 2006. Relaxed phylogenetics and dating with confiden ce. PLoS Biol. 4:e88. Drummond, A., and A. Rambaut. 2007. BEAST: Baye sian evolutionary analysis sampling trees. BMC Evol. Biol. 7:214. Edwards, A. W. F. 1992. Likelihood. John Hopki ns University Press, Baltimore, USA. Engler, A. 1874. Simaroubaceae. Pages 197, tabs. 40 in Flora Brasiliensis (C. F. P. de Martius, and I. Urban, eds.), volume 12(2). Frid. Fleischer in co mm., Lipsiae, Germany. Engler, A. 1905. Anacardiaceae africanae I II. Botanische Jahrbcher Syst. 36:216. Engler, A. 1931. Simarubaceae. Pages 359 in Die natrlichen Planzenfamilien 2nd ed. (A. Engler, and K. Prantl, eds.), volum e 19a. Engelmann, Leipzig, Germany. Eriksson, O., and B. Bremer, B. 1991. Fruit characteristics, life forms, and species richness in the plant family Rubiaceae. Am. Nat. 138:751. Eriksson, O., and B. Bremer, B. 1992. Pollination systems, dispersal modes, life forms, and diversification rates in angiosperm families. Evolution 46:258. Erixon, P., B. Svennblad, T. Britton, and B. Ox elman. 2003. Reliability of Bayesian posterior probabilities and bootstrap frequencies in phylogenetics. Syst. Biol. 52:665. Faith, D. P. 1992a. Conservation evaluation and phylogenetic diversity. Biol. Conserv. 61:1. Farris, J. S., M. Kallersjo, A. G. Kluge, and C. Bult. 1994. Testing signif icance of incongruence. Cladistics 10:315. Felsenstein, J. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39:783. Fernando, E. S., and Quinn, C. J. 1992. Pericarp anatomy and systematics of the Simaroubaceae s.l. Aust. J. Bot. 40:263. Fernando, E. S., and Quinn, C. J. 1995. Picramniaceae, a new family, and recircumscription of Simaroubaceae. Taxon 44:177. Fernando, E. S., Gadek, P. A., and Quinn, C. J. 1995. Simaroubaceae, an artificial construct: Evidence from rbcL sequence variation. Am. J. Bot. 82:92. Feuillet, C. 1983. Le statut des genres Quassia L., Samadera Gaertn., Simaba Aubl. et Simarouba Aubl. (Simaroubaceae). Bull. Jardin Bot. Nat. Belg. 53:51011.

PAGE 163

163 Franceschinelli, E. V. and K. Yamamoto. 1999. Simaba docensis, a new Brazilian species of Simaroubaceae. Novon 9:345. Franceschinelli, E. V., K. Yamamoto and G. J. Shepherd. 1999. Distinctions among three Simarouba species. Syst. Bot. 23:479. Franceschinelli, E. V., and W. W. Thomas. 2000. Simaba guianensis subsp. huberi a new Venezuelan taxon of Simaroubaceae. Brittonia 52:311. Fukami, T., H. Beaumont, X. Zhang, and P. Rainey. 2007. Immigration history controls diversification in experimental adaptive radiation. Nature 446:436. Funk, D., P. Nosil, and W. Etges. 2006. Ecological divergence exhibits consistently positive associations with reproductive isolation across disparate taxa Proc. Natl. Acad. Sci. USA 103:3209. Gadek P. A., E. S. Fernando, C. J. Quinn, S. B. Hoot, T. Terrazas, M. C. Sheahan, and M. W. Chase. 1996. Sapindales: Molecular delimitati on and infraordinal groups. Am. J. Bot. 83:802. Gentry, A. H. 1982. Neotropical oristic diversity: phytogeogr aphical connections between Central and South America, Pleistocene climatic uctuations, or an accident of the Andean orogeny? Ann. Mo. Bot. Gard. 69: 557. Giannasi, D. E. 1986. Phytochemical aspects of phylogeny in Hamamelidae. Ann. Mo. Bot. Gard. 73:417. Gilbert, G. 1958. Simaroubaceae. Pages 119 in Flore du Congo Belge et du Ruanda-Urundi (Comit excutif de la Flore du Congo belge et le Jardin botanique de ltat), volume VII. I.N..A.C., Brussels, Belgium. Givnish, T., and S. Renner. 2004. Tropical intercontinental disjunc tions: Gondwana breakup, immigration from the boreotropics, and transoce anic dispersal. Int. J. Plant Sci. 165:S1S6. Good-Avila, S., V. Souza, B. Gaut, and L. Eguiarte. 2006. Timing and rate of speciation in Agave (Agavaceae). Proc. Natl. Acad. Sci. USA 103:9124. Gottschling, M., N. Diane, H. H. Hilg er, and M. Weigend. 2004. Testing hypotheses on disjunctions present in the primarily woody Boraginales: Ehretiaceae, Cordiaceae, and Heliotropiaceae, inferred from ITS1 sequence data. Int. J. Plant Sci. 165:S123S135. Gradstein, F. M., and J. G. Ogg. 2004. Geologi c Time Scale 2004 why, how, and where next! Lethaia 37:175. Graur, D., and W. Martin. 2004. Reading the entr ails of chickens: molecular timescales of evolution and the illusion of precision. Trends Genet. 20:80.

PAGE 164

164 Gray A. 1859. Diagnostic characters of phaneroga mous plants collected by Charles Wright, botanist of the U.S. North Pacific Expl oring Expedition, with observations upon the relations of the Japanese flora to that of North America, a nd other parts of the Northern Temperate Zone. Mem. Am. Acad. Arts Sci. NS 6:377. Gray, A. I. 1983. Structural di versity and distribution of c oumarins and chromones in the Rutales. Pages 97 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Graybeal, A. 1998. Is it better to add taxa or char acters to a difficult phylogenetic problem? Syst. Biol. 47:9. Hahn, W. J., and W. W. Thomas. 2001. Simaroubaceae DC. Pages 2368 in Flora de Nicaragua (W. D. Stevens, C. U. Ulloa, A. P ool, and O. M. Montiel, eds.), volume 85, part III. Missouri Botanical Garden Press, Missouri, USA. Harborne, J. B. 1983. The flavonoids of the Rutales. Pages 147 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Hardesty, B. D., C.W. Dick, A. Kremer, S. H ubbell, and E. Bermingham. 2005. Spatial genetic structure of Simarouba amara Aubl. (Simaroubaceae), a dioecious, animal-dispersed Neotropical tree, on Barro Colora do Island, Panama. Heredity 95:290. Harmon, L., J. Weir, C. Brock and R. Glor 2008. GEIGER: investigating evolutionary radiations. Bioinformatics 24:129. Hedtke, S. M., T. M. Townsend, and D. M. H illis. 2006. Resolution of p hylogenetic conflict in large data sets by increased taxon sampling. Syst. Biol. 55:522. Hegnauer, R. 1983. Chemical characters and th e classification of the Rutales. Pages 401 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Herrera, C. 1989. Seed dispersal by animals: a role in angiosperm diversification? Am. Nat. 133:309. Hewson, H. J. 1985. Simaroubaceae. Pages 188 in Flora of Australia (A. S. George, ed.), volume 25. Australian Government Pub lishing Service Canberra, Australia. Hipp, A. L., J. C. Hall, and K. J. Sytsma 2004. Congruence versus phylogenetic accuracy: Revisiting the incongruence length difference test. Syst. Biol. 53:81. Hodges, S., and M. Arnold. 1995. Spurring Plant Dive rsification: Are Floral Nectar Spurs a Key Innovation? Proc. Royal Soc. London B 262:343.

PAGE 165

165 Hoot, S.B, A. Culham, and P.R. Crane. 1995. The utility of atpB gene sequences in resolving phylogenetic relationships: Comparison with rbcL and 18S ribosomal DNA sequences in the Lardizabalaceae. Ann. Mo. Bot. Gard. 82:194. Huelsenbeck, J.P., J. J. Bull, and C.W. C unningham. 1996. Combining data in phylogenetic analysis. Trends Ecol. Evol. 11:152. Huelsenbeck, J.P., and F. Ronquist. 2001. MR BAYES: Bayesian inference of phylogeny. Bioinformatics 17:754. Huelsenbeck, J. P., R. Nielsen, and J. P. Bo llback. 2003. Stochastic Mapping of Morphological Characters. Syst. Biol. 52:131. Hughes, N. F. 1994. The enigma of angiosperm origins. Cambridge University Press, Cambridge, UK. Hughes, C., and R. Eastwood. 2006. Is land radiation on a co ntinental scale: exceptional rates of plant diversification after uplift of the Andes. Proc. Natl. Acad. Sci. USA 103:10334 10339. Hutchinson, J., and J. M. Dalziel. 1954. Flora of west tropical Africa, 2nd edition, volume 1. Crown Agents, London, UK. Jaffre, T., and J. Fambart. 2002. Quatre nouvelles espces de Soulamea (Simaroubaceae) de Nouvelle-Caldonie. Adansonia 24:159. Johnson, L.A., and D.E. Soltis. 1998. Assessing congr uence: Empirical examples from molecular data. Pages 297 in Molecular systematics of plan ts II: DNA sequencing (D.E. Soltis, P.S. Soltis, and J.J. Doyle, eds.). Kluwer, Norwell, Massachusetts, USA. Judd, W. S., C. S. Campbell, E. A. Kellogg, P. F. Stevens, and M. J. Donoghue. 2008. Plant Systematics: A Phylogenetic Approach, 3rd edition. Sinauer Associates Inc., Sunderland, Massachusetts, USA. Kllersj, M., J. S. Farris, M. W. Chase, B. Brem er, M. F. Fay, C. J. Humphries, G. Petersen, O. Seberg, and K. Bremer. 1998. Simultaneous parsimony jackknife analysis of 2538 rbcL DNA sequences reveals support for major clad es of green plants, land plants, and flowering plants. Plant. Syst. Evol. 213:259. Kass, R.E., and A.E. Raftery. 1995. Bayes factors. J. Am. Stat. Assoc. 90:773. Kathriarachchi, H., P. Hoffmann, R. Samuel, K. J. Wurdack, and M. W. Chase. 2005. Molecular phylogenetics of Phyllanthaceae inferred from five genes (plastid atpB, matK 3 ndhF rbcL and nuclear PHYC). Mol. Phylogenet. Evol. 36:112. Kay, K. M., P. A. Reeves, R. G. Olmstead, and D. W. Schemske. 2005. Rapid speciation and the evolution of hummingbird pollination in neotropical Costus subgenus Costus (Costaceae): evidence from nrDNA ITS and ETS sequences. Am. J. Bot. 92: 1899.

PAGE 166

166 Kelchner, S.A., and M. A. Thomas. 2007. Mode l use in phylogenetics: nine key questions. Trends Ecol. Evol. 22:87. Klocke, J.A., M. Arisawa, S. S. Handa, A. D. Kinghorn, G. A. Cordell, and N. R. Farnsworth. 1985. Growth inhibitory, insecticidal and an tifeedant effects of some antileukemic and cytotoxic quassinoids on two sp ecies of agricultural pests. Cell. Mol. Life. Sci. 4:379. Lavin, M., M. Thulin, J. -N. Labat, and R.T. Pennington. 2000. Africa, the odd man out: Molecular biogeography of dalbergioid legumes (Fabaceae) suggests otherwise. Syst. Bot. 25:449. Lavin, M., M. F. Wojciechowski, P. Gasson, C. H. Hughes, and E. Wheeler. 2003. Phylogeny of robinioid legumes (Fabaceae) revisited: Coursetia and Gliricidia recircumscribed, and a biogeographical appraisal of the Cari bbean endemics. Syst. Bot. 28:387. Lavin, M., B. P. Schrire, G. Lewis, R. T. Pe nnington, A. Delgado-Salinas, M. Thulin, C. E. Hughes, A. B. Matos, and M. F. Wojci echowski. 2004. Metacommunity process rather than continental tectonic history better expl ains geographically stru ctured phylogenies in legumes. Philos T. Roy. Soc. B 359:1509. Lecointre, G., H. Philippe, H. L. Van Le, and H. Guyader. 1993. Species sampling has a major impact on phylogenetic inference. Mol. Phylogenet. Evol. 2:205. Lemmon, A. R., and E. C. Moriarty. 2004. The importance of proper model assumption in Bayesian phylogenetics Syst. Biol. 53:265. Leschen, R. A. B., and T. R. Buckley. 2007. Multista te Characters and Diet Shifts: Evolution of Erotylidae (Coleopter a). Syst. Biol. 56:97. Lewis, P. O. 2001. A likelihood approach to es timating phylogeny from discrete morphological character data. Syst. Biol. 50:913. Lewis, P. O., M. T. Holder, and K. E. Holsinger. 2005. Polytomies a nd Bayesian phylogenetic inference. Syst. Biol. 54:241. Li, H-.L. 1952. Floristic relationshi ps between eastern Asia and eas tern North America. Trans. Am. Phil. Soc. 42:371. Li, H-.L. 1977. Simarubaceae. Pages 538 in Flora of Taiwan (H-.L. Li, T-. S. Liu, T-. C. Huang, T. Koyama, and C. E. DeVol, eds.), volume 3. Epoch Publishing Co., Ltd., Taipei, Taiwan, Republic of China. Luckow, M., and A. Bruneau. 1997. Circularity and Independence in Phylogenetic Tests of Ecological Hypotheses. Cladistics 13:145. Mabberley, D. 1997. The Plant Book. Cambri dge University Press, Cambridge, UK.

PAGE 167

167 Machen, J. 1971. Plant microfossils from Tertiary deposits of the Isle of Wight. New Phytol. 70:851. Maddison, D. R., and W. P. Maddison. 2004. MacCla de 4: Analysis of phylogeny and character evolution. Version 4.08. Sinauer Associates, Sunderland, MA. Magalln, S., and M. J. Sanderson. 2001. Absolute di versification rates in angiosperm clades. Evolution 55:1762. Mathews, S., M. Lavin, and R. A. Sharrock. 1995. Evolution of the phytochrome gene family and its utility for phylogenetic analyses of angiosperms. Ann. Mo. Bot. Gard. 82:296. Mauritzon, J. 1935. Kritic von J. Wigers Abhandlung Embryological studies on the families Buxaceae, Meliaceae, Simarubaceae and Burseraceae. Bot. Not. 1935:490. Meimberg, H., A. Wistuba, P. Dittrich, and G. Heubl. 2001. Molecular phylogeny of Nepenthaceae based on cladistic analysis of plastid trnK intron sequence data. Plant Biology 3:164. Mester, I. 1983. Structural dive rsity and distribution of alkalo ids in the Rutales. Pages 31 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Mittelbach, G., D. Schemske, H. Cornell, a nd A. Allen. 2007. Evolution and the latitudinal diversity gradient: speciation, extinct ion and biogeography. Ecol. Lett. 10:315. Moncada, M., and S. Machado. 1987. Los granos de polen de Simarubaceae. Acta. Bot. Cub. 45:1. Moore, B. R., K. Chan, and M. J. Donoghue. 2004. Detecting diversification rate variation in supertrees. Pages 487 in Phylogenetic supertrees: combining information to reveal the Tree of Life (O. R. P. Bininda-Edmonds, ed.) Kluwer Academic Publishers, Dordrecht, The Netherlands. Moore, B. R., Q. -Y. Xiang, J. Thorne, D. Thomas, and M. J. Donoghue. 2006. Biogeographic and biological correlates of diversification in dogwoods: Disp ersal and evolution of fruits and inflorescences in space and time Evolution Meeting (Stonybrook, NY). Moore, B. R., and M. J. Donoghue. 2007. Correlate s of diversification in the plant clade Dipsacales: geographic movement and evolut ionary innovations. Am. Nat. 170:S28S55. Moore, M. J., A. Tye, and R. K. Jansen. 2006. Patterns of long-distance dispersal in Tiquilia subg. Tiquilia (Boraginaceae): implications for the origins of amphitropical disjuncts and Galapagos Islands endemics. Am. J. Bot. 93:1163. Morley, R. J. 2003. Interplate dispersal paths for megathermal angiosperms. Perspect. Plant Ecol. 6:5.

PAGE 168

168 Morris, A. B., C. D. Bell, J. W. Clayton, W. S. Judd, D. E. Soltis, and P. S. Soltis. 2007. Phylogeny and divergence time estimation in Illicium with implications for New World biogeography. Syst. Bot. 32:236. Muellner, A. N., R. Samuel, S. A. Johnson, M. Cheek, T. D. Pennington, and M.W. Chase. 2003. Molecular phylogenetics of Meliaceae (Sap indales) based on nucle ar and plastid DNA sequences. Am. J. Bot. 90:471. Muellner, A. N., V. Savolainen, R. Samuel, and M. W. Chase. 2006. The mahogany family outof-Africa: Divergence time estimation, globa l biogeographic patterns inferred from plastid rbcL DNA sequences, extant, and fossil distribution of dive rsity. Mol. Phylogenet. Evol. 40:236. Muellner, A. N., D. D. Vassiliades, and S. Re nner. 2007. Placing Biebersteiniaceae, a herbaceous clade of Sapindales, in a temporal and ge ographic context. Plan t Syst. Evol. 266:233. Nair, N. C., and T. C. Joseph. 1957. Floral morphology and embryology of Samadera indica Bot. Gaz. 119:104. Nair, N. C., and R. K. Joshi. 1958. Floral mo rphology of some members of the Simaroubaceae. Bot. Gaz. 120:88. Nair, N. C., and N. P. Sukumaran. 1960. Floral morphology and embryology of Brucea amarissima Bot. Gaz. 121:175. Narayana, L. L. 1957. Embryology of Simaroubaceae. Curr. Sci. 26:323324. Near, T. J., and M. J. Sanderson. 2004. Assessing the quality of mol ecular divergence time estimates by fossil calibrations and fossil-ba sed model selection. Philos. T. Roy. Soc. B 359:1477. Nee, S., A. Mooers, and P. H. Harvey. 1992. Tempo and mode of evolution revealed from molecular phylogenies. Proc. Natl. Acad. Sci. USA 89:8322. Linsbauer, K. 1926. Handbuch der Pflanzenanat omie, volume 2(X). Borntraeger, Berlin, Germany. Nikitin, V. P. 2006a. Paleocarpology and stratigraphy of the Paleogene and Neogene strata in Asian Russia. Academic Publishing House "Geo", Novosibirsk, Russia. Nooteboom, H. P. 1962a. Generic delimitation in Simaroubaceae Tribus Simaroubeae and a conspectus of the genus Quassia L. Blumea 11:509. Nooteboom, H. P. 1962b. Simaroubaceae. Flora Malesiana Series I 6:193. Nooteboom, H. P. 1987. Laumoniera, a new genus of Simaroubaceae from Sumatra. Blumea 32:383.

PAGE 169

169 Nylander, J. A. A., F. Ronquist J. P. Huelsenbeck, and J. L. Nieves-Aldrey. 2004. Bayesian phylogenetic analysis of combined data. Syst. Biol. 53:47. Ollerton, J., S. D. Johnson, and A. B. Hingston. 2006. Geographical variation in diversity and specificity of pollination systems. Pages 283 308 in Plant-pollinator interactions: from specialization to generalization (N M. Waser, and J. Ollerton, eds.). University of Chicago Press, Chicago, Illinois, USA. Olmstead, R. G., and J. A. Sweere. 1994. Comb ining data in phylogenetic systematics: An empirical approach using three molecular data sets in the Solanaceae. Syst. Biol. 43:467 481. Pennington, R.T. 1996. Molecular a nd morphological data provide phylogenetic resolution at different hierarchical levels in Andira Syst. Biol. 45:496. Pennington, R. T., and C. W. Dick. 2004. The role of immigrants in the assembly of the South American rainforest tree flora. Philos. T. Roy. Soc. B 359:1611 Pennington, R. T., Q. C. B. Cronk, and J. E. Rich ardson. 2004a. Introduction and synthesis: plant phylogeny and the origin of major biomes. Philos. T. Roy. Soc. B 359:1455. Pennington, R. T., M. Lavin, D. E. Prado, C. A. Pendry, S. K. Pell, and C. A. Butterworth. 2004b. Historical climate change and speciation: neotropical seasonall y dry forest plants show patterns of both Tertiary and Quaterna ry diversification. Philos. T. Roy. Soc. B 359:515. Pennington, R. T., J. E. Richardson, and M. Lavi n. 2006. Insights into the historical construction of species-rich biomes from dated plant phylogenies, neutral ecological theory and phylogenetic community struct ure. New. Phytol. 172:605. Perrier de la Bathie, H. 1950. Simarubaces. Pages 1 in Flore de Madagascar (H. Humbert ed.), volume 105. Firmin-Didot and Co., Paris, France. Pfeiffer, W. M. 1912. The morphology of Leitneria floridana. Bot. Gaz. 53:189. Pierre, L. 1896 Plantes du Gabon. Bull Soc Linn Paris 2:1236. Pirani, J. R. 1987. Simaroubaceae. Pages 7 in Flora del Paraguay (R. Spichiger, ed.). Missouri Botanical Garden, Missouri, USA. Pirani, J. R. 1997. Simaroubceas. Pages 1 in Flora Ilustrada Catarinense (A. Reis, ed.). Itaja, Santa Catarina, Brasil. Plana, V. 2004. Mechanisms and tempo of evolut ion in the African Guineo-Congolian rainforest. Philos. T. Roy. Soc. B 359:1585.

PAGE 170

170 Polonsky, J. 1983. Chemistry and biological activity of the qua ssinoids. Pages 247 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Porter, D. M. 1971. Simaroubaceae. Pages 7546 in Flora of the Galpagos Islands (I. L. Wiggins, and D. M. Porter, eds.). Stanford Un iversity Press, Stanford, California, USA. Porter, D. M. 1973. Flora of Panama, Family 90, Simaroubaceae. Ann. Mo. Bot. Gard. 60:23. Posada, D., and K. A. Crandall. 1998. Modelt est: testing the model of DNA substitution. Bioinformatics 14:817. Posada D., and T. R. Buckley. 2004. Model se lection and model averaging in phylogenetics: Advantages of Akaike information criterion a nd Bayesian approaches over likelihood ratio tests. Syst. Biol. 53:793. Poux, C., O. Madsen, E. Marqua rd, D. R. Vieites, W. W. de Jong, and M. Vences. 2005. Asynchronous colonization of Madagascar by the f our endemic clades of primates, tenrecs, carnivores, and rodents as inferred fro m nuclear genes. Syst. Biol. 54:719. Purvis, A., S. Nee and P. H. Harvey. 1995. Macroevolutionary inferences from primate phylogeny. Proc. Royal Soc. London B 260:329. Pybus, O., and P. H. Harvey. 2000. Testing macro-evolutionary models using incomplete molecular phylogenies. Proc Royal Soc. London B 267:2267. Rabosky, D. 2006. Likelihood methods for detecting te mporal shifts in diversification rates. Evolution 60:1152. Rambaut, A., and A. J. Drummond. 2003. Tracer v1.3. Available from http://evolve.zoo.ox.ac.uk/. Rannala, B. 2002. Identifiability of parameters in MCMC Bayesian inference of phylogeny. Syst. Biol. 51:754. Raven, P. H., and D. I. Axelrod. 1974. Angi osperm biogeography and past continental movements. Ann. Mo. Bot. Gard. 61:539. Raven, P. H. 1975. The bases of angiosperm phylogeny: cytology. Ann. Mo. Bot. Gard. 68:724 764. Record, S. J., and R. W. Hess. 1943. Timbers of the New World. Yale University Press, New Haven, Conneticut, USA. Ree, R. 2005. Detecting the historical signature of key innovations using stochastic models of character evolution and cl adogenesis. Evolution 59:257.

PAGE 171

171 Ree, R. H., B. R. Moore, C. O. Webb, and M. J. Donoghue. 2005. A likelihood framework for inferring the evolution of geographic ra nge on phylogenetic trees. Evolution 59:2299 2311. Ree, R. H., and S. A. Smith. 2008. Maximum like lihood inference of geogr aphic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57:4. Renner, S. S., G. Clausing, and K. Meyer. 2001. Historical biogeogra phy of Melastomataceae: the roles of Tertiary migration and lo ng-distance dispersal. Am. J. Bot. 88:1290. Renner, S. S. 2004a. Plant dispersal across the tropical Atlantic by wind and sea currents. Int. J. Plant Sci. 165:S23S33. Renner, S. S. 2004b. Multiple Miocene Melastomataceae dispersal between Madagascar, Africa and India. Philos T. Roy. Soc. B 359:1485. Richardson, J. E., R. T. Pennington, T. Pe nnington, and P. Hollingsworth. 2001a. Rapid diversification of a species-rich genus of neotropical rain forest trees. Science 293: 2242 2245. Richardson, J. E., F. M. Weitz, M. F. Fay, Q. C. B. Cronk, H. P. Linder, G. Reeves, and M. W. Chase. 2001b. Rapid and recent origin of species richness in the Cape flora of South Africa. Nature 412:181. Richardson, J. E., L. W. Chatrou, J. B. Mols, R. H. J. Erkens, and M. D. Pirie. 2004. Historical biogeography of two cosmopolitan families of flowering plants: Annonaceae and Rhamnaceae. Philos T. Roy. Soc. B 359:1495. Riddle, B. R. 1996. The molecular phylogeographic bridge between deep and shallow history in continental biotas. Tren ds. Ecol. Evol. 11:207. Ronquist, F. 1997. Dispersal-Vicariance Analysis: A new approach to th e quantification of historical biogeogra phy. Syst. Biol. 46:195. Ronquist, F., and J. P. Huelsenbeck. 2003. MrBa yes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572. Ronse de Craene, L. P. 2008. Homology and evolution of petals in the core eudicots. Syst. Bot. 33:301. Roth, J. L. and D. L. Dilcher. 1979. Investigatio ns of angiosperms from the Eocene of North America: stipulate leaves of the Rubiaceae including a probable polyploid population. Am. J. Bot. 66:1194. Roubik, D. W., N. M. Holbrook, and G. V. Parra 1985. Roles of nectar robbers in reproduction of the tropical treelet Quassia amara (Simaroubaceae). Oecologia 66:161.

PAGE 172

172 Saarela, J. M., H. S. Rai, J. A. Doyle, P. K. En dress, S. Mathews, A. D. Marchant, B. G. Briggs, and S. W. Graham. 2007. Hydate llaceae identified as a new branch near the base of the angiosperm phylogenetic tree. Nature 446:312. Sanderson, M. J. 1997. A nonparametric approach to estimating divergence times in the absence of rate constancy. Mol. Biol. Evol. 14:1218. Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approac h. Mol. Biol. Evol. 19:101. Sanmartin, I., H. Enghoff, and F. Ronquist. 2001. Patterns of animal dispersal, vicariance and diversification in the Holarctic. Biol. J. Linn. Soc. 73:345. Santiago-Valentin, E., and R. G. Olmstead. 2004. Historical biogeography of Caribbean plants: introduction to current knowle dge and possibilities from a phylogenetic perspective. Taxon 53:299. Savolainen, V., M. W. Chase, S. B. Hoot, C. M. Morton, D. E., Soltis, C. Bayer, M. F Fay, A. Y. de Bruijn, S. Sullivan, and Y. -L. Qiu. 2000b Phylogenetics of flow ering plants based on a combined analysis of plastid atpB and rbcL gene sequences. Syst. Biol. 49:306. Schott, F. A., and J. P. McCreary. 2001. The monsoon circulation of the Indian Ocean. Prog. Oceanogr. 51:1. Scotese, C. R. 2001. Atlas of Earth Hi story. PALEOMAP Project, Arlington, TX. www.scotese.com. Simpson, B. B., J. A. Tate, and A. Weeks. 2005. The biogeography of Hoffmannseggia (Leguminosae, Caesalpinoideae, Caesalpiniea e): a tale of many travels. J. Biogeogr. 32:15. Smith, A. C. 1985. Simaroubaceae. Pages 4799 in Flora Vitiensis Nova (A. C. Smith, ed.), volume 3. Pacific Tropical Bota nical Garden, Hawaii, USA. Soltis, D. E., P. S. Soltis, M. E. Mort, M. W. Chase, V. Savolainen, S. B. Hoot, and C. M. Morton. 1998. Inferring complex phylogenies us ing parsimony: An empirical approach using three large DNA data sets for angiosperms. Syst. Biol. 47:32. Soltis, D. E., P. S. Soltis, M. W. Chase, M. E. Mort, D. C. Albach, M. Zanis, V. Savolainen, W. H. Hahn, S. B. Hoot, M. F. Fay, M. Axtell, S. M. Swensen, L. M. Prince, W. J. Kress, K. C. Nixon, and J. S. Farris. 2000. Angiospe rm phylogeny inferred from 18S rDNA, rbcL and atpB sequences. Bot. J. Linn. Soc. 133:381. Soltis, D. E., P. S. Soltis, P. K. Endress, a nd M. W. Chase. 2005. Phylogeny and evolution of angiosperms. Sinauer Associates In c., Sunderland, Massachusetts, USA. Song, Z. -C., W. -M. Wang, and F. Huang. 2004. Fossil pollen records of extant angiosperms in China. Bot. Rev. 70:425.

PAGE 173

173 Spiekerkoetter, H. 1924. Untersuchungen zur An atomie und Systematik stafrikanischer Meliaceen, Burseraceen und Simarubaceen. Bot. Arch. 7:274. Stannard, B. 2000. Simaroubaceae. Pages 1 in Flora of Tropical East Africa (H. J. Beentje, ed.). Royal Botanic Gardens, Kew, London, UK. Stevens, P. F. 2006. Angios perm Phylogeny Website v 7. http://www.mobot.org/MOBOT/research/APweb/. Sullivan, J. 1996 Combining data wi th different distributions of among-site rate variation. Syst. Biol. 45:375. Sullivan, J., and P. Joyce. 2005. Model selecti on in phylogenetics. Annu. Rev. Ecol. Evol. Syst. 36:445. Suzuki, Y., G. V. Glazko, and M. Nei. 2002. Overcredibility of molecular phylogenetics obtained by Bayesian phylogenetics. Proc. Natl. Acad. Sci. USA 99:16138. Swofford, D. L. and W.P. Maddison. 1987. Reconstructing ancestral character states under Wagner parsimony. Math. Biosci. 87:199. Swofford, D. S. 2002. Phylogenetic analysis using parsimony, version 4.0b10. Sinauer Associates Inc., Sunderland, MA. Tahktajan, A. 1996. Diversity and cl assification of flowering plants Columbia University Press, New York, USA. Teodoris, V., and Z. Kvacek. 2005. The extinct genus Chaneya Wang et Manchester in the Tertiary of Europe a revision of Porana-like fruit remains from hningen and Bohemia. Rev. Palaeobot. Palyno. 134:85. Thomas, W. W. 1984. A new species of Simaba (Simaroubaceae) from Par, Brazil, wih a key to the species north of the Am azon river. Brittonia 36:244. Thomas, W. W. 1985. The Simaba guianensis complex in northern South America. Acta Amazonica, supl. 15:71. Thomas, W. W. 2002. Simaroubaceae. Pages 685 in Guide to the vascular plants of Central French Guiana (S. A. Mori, G. Cremers, C. A. Gracie, J-. J. de Granvi lle, S. V. Head, M. Hoff, and J. D. Mitchell), part 2, Dicotyledone s. The New York Botanical Garden Press, New York, USA. Thomson, J., T. Gibson, F. Plewniak, F. Jeanmougin, and D. Higgins. 1997. The ClustalX windows interface: flexible strategies for mu ltiple sequence alignments aided by quality analysis tools. Nucleic Acids Res. 25:4876. Thorne, J. L., and H. Kishino. 2002. Divergence time and evolutionary rate estimation with multilocus data. Syst. Biol. 51:68902.

PAGE 174

174 Tiffney, B. H. 1985a. The Eocene North Atlantic land bridge: its importance in Tertiary and modern phytogeography of the northern he misphere. J. Arnold Arboretum 66:243. Trelease, W. 1895. Leitneria floridana Rept. Mo. Bot. Garden 6:1 pls. 30. Van der Veken, P. 1960. Nothospondias Engl. Simaroubaceae africaine meconnue. Bull. Jardin Botanique Etat Bruxelles 30:105. Von Hagen, K., and J. Kadereit. 2003. The diversification of Halenia (Gentianaceae): ecological opportunity versus key i nnovation. Evolution 57:2507. Wang, Y., and S. R. Manchester. 2000. Chaneya a new genus of winged fruit from the Tertiary of North America and Eastern Asia. Int. J. Plant Sci. 161:167. Wanntorp, L., and H. E. Wanntorp. 2003. The biogeography of Gunnera L.: vicariance and dispersal. J. Biogeogr. 30:979. Waser, N. 1998. Pollination, angiosperm speciation, and the nature of species boundaries. Oikos 81:198. Waterman, P. G., and M. F. Grundon. 1983. Chemistr y and chemical taxonomy of the Rutales. Academic Press Inc., London, UK. Waterman, P. G. 1983. Phylogenetic implications of the distribution of secondary metabolites within the Rutales. Pages 377 in Chemistry and chemical taxonomy of the Rutales (P. G. Waterman, and M. F. Grundon, eds.). Academic Press Inc., London, UK. Webber, I. E. 1936. Systematic anatomy of the woods of the Simarubaceae. Am. J. Bot. 23:577 587. Weberling, F. 1989. Morphology of flowers and infl orescences. Cambridge University Press, UK. Weeks, A., D. C. Daly, and B. B. Simpson. 2005. The phylogenetic history and biogeography of the frankincense and myrrh family (Burse raceae) based on nuclear and chloroplast sequence data. Mol. Phylogenet. Evol. 35:85. Whittaker, R. J. 1998. Island biogeography. Ox ford University Press, Oxford, UK. Wiens, J. J., and M. J. Donoghue. 2004. Historical biogeography, ecology and species richness. Trends Ecol. Evol. 19:639. Wiens, J. J., G. Parra-Olea, M. Garca-Par s, and D. B. Wake. 2007. Phylogenetic history underlies elevational biodiversity patterns in tropical salamanders. Proc. Royal Soc. London B 274:919. Wiger, J. 1935. Embryological studies in Bux aceae, Meliaceae, Simarubaceae and Burseraceae. Dissertation, Universi ty of Lund, Sweden.

PAGE 175

175 Wikstrm, N., V. Savolainen, and M.W. Chase. 2001. Evolution of the angiosperms: calibrating the family tree. Proc. R. Soc. B 268:2211. Wikstrm, N., V. Savolainen, and M. W. Chase. 2004. Angiosperm divergence times: congruence and incongruence between fossils and sequence divergence estimates. Pages 142 in Telling the evolutionary time: molecula r clocks and the fossil record (P.C.J. Donoghue, and M.P. Smith, eds.). Taylor and Francis, London, UK. Wild, H., and J. B. Phipps. 1963. Simaroubaceae. Pages 210 in Flora Zambesiaca (A. W. Exell, A. Fernandes, and H. Wild, eds.), volume 2, part 1. Crown Agents for Oversea Governments and Administrations, London, UK. Wilson, P. G., M. M. OBrien, M. M. Heslewood, and C. J. Quinn. 2005. Relationships within Myrtaceae sensu lato based on a mat K phylogeny. Plant Syst. Evol. 251:3. Xiang, Q. Y., and D. E. Soltis. 2001. Dispersal-vi cariance analysis of in tercontinental disjuncts: historical biogeographical implications for angi osperms in the Northern Hemisphere. Int. J. Plant Sci. 162:S29S39. Yuan, Y. M., S. Wohlhauser, M. Mller, J. Klackenberg, M. Callmander, and P. Kpfer. 2005. Phylogeny and biogeography of Exacum (Gentianaceae): a disjuncti ve distribution in the Indian Ocean basin resulted from long-dista nce dispersal and extensive radiation. Syst. Biol. 54:21. Zavada, M. S., and D. L. Dilc her. 1986. Comparative pollen morphology and its relationship to phylogeny of pollen in the Hamamelidae. Ann. Mo. Bot. Gard. 73:348. Zerega, N. J. C., W. L. Clement, S. L. Da twyler, and G. D. Weiblen. 2005. Biogeography and divergence times in the mulberry family (Moraceae). Mol. Phylogenet. Evol. 37:402416. Zhang, L. -B. and S. S. Renner. 2003. The deepes t splits in Chloranthaceae as resolved by chloroplast sequences. Int. J. Plant Sci. 164:S383S392. Zurawski G., M. T. Clegg, and A. H. D. Brown. 1984. The nature of nucleotide sequence divergence between barley and mai ze chloroplast DNA. Genetics 106:735.

PAGE 176

176 BIOGRAPHICAL SKETCH Joshua W. Clayton received his Bachelor of Science degree in plant science from the University of Edinburgh, in June, 2001. During the following year he worked as a field botanist in the upland forests of Mauritius. He began his graduate career in the fall of 2002 at the University of Edinburgh and Royal Botanic Gardens, Edinburgh, receiving his Master of Science in the Biodiversity and Taxonomy of Plants in August, 2003. His Masters thesis focused on the historical biogeography of Manilkara a genus of tropical trees. Jo sh started his doctoral work with Doug and Pam Soltis at the University of Florida in Gainesville, where he studied biogeographical patterns and evolutionary hist ory of Simaroubaceae. Upon completion of his dissertation, Josh plans to continue resear ch through a postdoctoral position in the UK.