ufdcimages.uflib.ufl.eduufdcimages.uflib.ufl.edu/uf/e0/05/05/72/00001/crowl_a.pdf ·...
TRANSCRIPT
EVOLUTION AND BIOGEOGRAPHY OF CAMPANULACEAE: FROM GLOBAL PATTERNS TO SHALLOW, SPECIES-LEVEL PROCESSES IN THE
MEDITERRANEAN BASIN
By
ANDREW ALAN CROWL
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2016
© 2016 Andrew Alan Crowl
To SED, for giving me the confidence to complete this while constantly reminding me to not take it too seriously.
“It was the tension between these two poles - a restless idealism on one hand and a
sense of impending doom on the other - that kept me going.” HS Thompson
4
ACKNOWLEDGMENTS
First, my deepest gratitude to my friend and supervisor, Dr. Nico Cellinese, who
taught me everything I know about phylogenetics, biogeography, and “species” – and to
always use quotes when talking about them. Her knowledge and supervision made this
work possible...and enjoyable. Also many thanks to my committee: Dr. Pam Soltis, Dr.
David Reed, and Dr. Matthew Smith. Our many insightful discussions were integral to
my successful completion of this dissertation.
Thanks to my mom for her green thumb and a childhood introduction to botany,
my dad for passing on his science genes, Gregg (DOM) Stull for all the ‘late night
discussions’, Heather-Rose Kates and Rebecca Stubbs for the soft discussions –
scientific and otherwise, Volta and The Bull for providing a big, wooden table and cold
beverages while writing, Benjamin Crowl, Michael Dudley, and Thomas Wondergem for
help with the grueling fieldwork, Michael Chester and Clayton Visger for help with
chromosome counts and niche modeling, Dr. Walter Judd for his help with
morphological work, Dr. Matthew Gitzendanner for always having an answer to the
many, many random questions I had throughout the years, my many collaborators and
co-authors around the world for their helpful manuscript edits and discussions, and
finally, my amazing wife, Susan Dalton, for putting up with me through all of this,
listening to me ramble about DNA and bellflowers, following me through the
Mediterranean, and believing in me.
Financial support for this research was provided by a Doctoral Dissertation
Improvement Grant from the National Science Foundation (DEB-1501676), with
additional funding from the Botanical Society of America, Society of Systematic
Biologists, and American Society of Plant Taxonomists.
5
TABLE OF CONTENTS page
ACKNOWLEDGMENTS .................................................................................................. 4
LIST OF TABLES ............................................................................................................ 9
LIST OF FIGURES ........................................................................................................ 10
ABSTRACT ................................................................................................................... 11
CHAPTER
1 OVERVIEW OF CAMPANULACEAE AND INTRODUCTORY REMARKS ............. 13
2 A GLOBAL PERSPECTIVE ON CAMPANULACEAE: BIOGEOGRAPHIC, GENOMIC, AND FLORAL EVOLUTION ................................................................. 15
Introduction ............................................................................................................. 15 Phylogenetics ................................................................................................... 16 Biogeography ................................................................................................... 17 Genome Evolution ............................................................................................ 18 Secondary Pollen Presentation ........................................................................ 18 Summary .......................................................................................................... 20
Methods .................................................................................................................. 20 Sampling .......................................................................................................... 20 Phylogenetic Analyses ..................................................................................... 21 Character Reconstruction ................................................................................. 22 Divergence Dating ............................................................................................ 23 Biogeographic Analyses ................................................................................... 24 Genome Evolution ............................................................................................ 25
Results .................................................................................................................... 26 Phylogenetics ................................................................................................... 26 Divergence Dating ............................................................................................ 27 Character Reconstruction ................................................................................. 28 Biogeographic Analyses ................................................................................... 29 Genome Evolution ............................................................................................ 29
Discussion .............................................................................................................. 30 Phylogenetics ................................................................................................... 30 Floral Evolution ................................................................................................. 31 Biogeography ................................................................................................... 32 Genome Evolution ............................................................................................ 36 Conclusion ........................................................................................................ 39
3 PHYLOGENY OF CAMPANULOIDEAE (CAMPANULACEAE) WITH EMPHASIS ON THE UTILITY OF NUCLEAR PENTATRICOPEPTIDE REPEAT (PPR) GENES ........................................................................................................ 46
6
Introduction ............................................................................................................. 46 Methods .................................................................................................................. 50
Taxon Sampling, Amplification, & Sequencing ................................................. 50 Phylogenetic Analyses ..................................................................................... 52 Dating Analyses ............................................................................................... 53
Results And Discussion .......................................................................................... 55 Phylogenetic Resolution ................................................................................... 56
Plastid loci .................................................................................................. 56 Pentatricopeptide repeat (PPR) loci ........................................................... 60 Combined plastid and PPR loci .................................................................. 62
Plastid–Nuclear Incongruence .......................................................................... 65 Final Conclusions ................................................................................................... 68
4 EVOLUTION AND BIOGEOGRAPHY OF THE ENDEMIC ROUCELA COMPLEX (CAMPANULACEAE: CAMPANULA) IN THE EASTERN MEDITERRANEAN ................................................................................................. 76
Introduction ............................................................................................................. 76 Oceanic and Continental Islands ...................................................................... 77 Geologic and Climatic History of the Eastern Mediterranean Basin ................. 77 Roucela Complex ............................................................................................. 79 Summary .......................................................................................................... 80
Materials and Methods............................................................................................ 81 Taxon Sampling and DNA Amplification ........................................................... 81 Phylogenetic Analysis ...................................................................................... 81 Species Tree .................................................................................................... 82 Molecular Dating .............................................................................................. 83 Biogeographic Analysis .................................................................................... 84 Ecological Niche Modeling ............................................................................... 85 Diversification ................................................................................................... 86
Results .................................................................................................................... 88 Phylogenetic Analyses ..................................................................................... 88
Plastid ........................................................................................................ 88 Nuclear ...................................................................................................... 89 Species tree and combined dataset ........................................................... 90
Divergence Time Estimates and Ancestral Range Estimation .......................... 91 Niche Modeling ................................................................................................. 92 Diversification ................................................................................................... 92
Discussion .............................................................................................................. 93 Evolution and Biogeography of the Roucela Clade .......................................... 93
Break-up of Aegean landmass ................................................................... 94 Cyprus disjunctions .................................................................................... 96 Campanula erinus ...................................................................................... 97 Niche modeling .......................................................................................... 97 Diversification ............................................................................................. 98
Concluding Remarks ........................................................................................ 99
7
5 GENE TREE DISCORDANCE PROVIDES EVIDENCE FOR CRYPTIC DIVERSITY AND INSIGHTS INTO THE EVOLUTION OF A POLYPLOID COMPLEX IN A MEDITERRANEAN CAMPANULA (CAMPANULACEAE) CLADE .................................................................................................................. 105
Introduction ........................................................................................................... 105 Methods ................................................................................................................ 107
Sampling ........................................................................................................ 107 Molecular Data ............................................................................................... 107 Data Processing ............................................................................................. 108 Phylogenetic Analysis .................................................................................... 109 Coalescent Species-Tree Analyses ................................................................ 110 Bayesian Concordance Analysis .................................................................... 111 Network Analysis ............................................................................................ 111 Ploidy Estimation ............................................................................................ 112 Morphology ..................................................................................................... 112
Results .................................................................................................................. 113 Phylogenetic Analyses ................................................................................... 113 Species-Tree and Network Analyses ............................................................. 114 Ploidy.............................................................................................................. 115 Morphology ..................................................................................................... 116
Discussion ............................................................................................................ 116
6 PHYLOGENETIC DEFINITION OF THE ROUCELA CLADE ............................... 124
Introduction: Roucela ............................................................................................ 124 Phylogenetic Definition: Roucela .................................................................... 124 Etymology: Roucela ....................................................................................... 124 Reference Phylogeny: Roucela ...................................................................... 125 Composition: Roucela .................................................................................... 125 Diagnostic Apomorphies: Roucela ................................................................. 125 Synonyms: Roucela ....................................................................................... 125 Comments: Roucela ....................................................................................... 125
Introduction: Holoerinus ........................................................................................ 127 Phylogenetic Definition: Holoerinus ................................................................ 127 Etymology: Holoerinus ................................................................................... 128 Reference Phylogeny: Holoerinus .................................................................. 128
Introduction: Tetraerinus ....................................................................................... 128 Phylogenetic Definition: Tetraerinus ............................................................... 128 Etymology: Tetraerinus................................................................................... 128 Reference Phylogeny: Tetraerinus ................................................................. 128 Composition: Tetraerinus ............................................................................... 129 Diagnostic Apomorphies: Tetraerinus ............................................................ 129 Synonyms: Tetraerinus................................................................................... 129 Comments: Tetraerinus .................................................................................. 129
Introduction: Octoerinus ........................................................................................ 129 Phylogenetic Definition: Octoerinus ................................................................ 129
8
Etymology: Octoerinus ................................................................................... 129 Reference Phylogeny: Octoerinus .................................................................. 130 Composition: Octoerinus ................................................................................ 130 Diagnostic Apomorphies: Octoerinus ............................................................. 130 Synonyms: Octoerinus ................................................................................... 130 Comments: Octoerinus ................................................................................... 130
LIST OF REFERENCES ............................................................................................. 131
BIOGRAPHICAL SKETCH .......................................................................................... 149
9
LIST OF TABLES
Table page 2-1 Comparison of the three data sets included in this study. .................................. 40
2-2 Summary of the two divergence-time estimate methods. ................................... 40
2-3 Likelihood statistics from ancestral area estimation models implemented in BioGeoBEARS. .................................................................................................. 41
3-1 Chloroplast and nuclear markers used in this study. .......................................... 70
10
LIST OF FIGURES
Figure page 2-1 Comparison of phylogenetic results from the coding-genes-only data set. ......... 41
2-2 Chronogram of the Campanulaceae clade showing ancestral range estimations and hypothesized polyploidy events. ............................................... 42
2-3 Ancestral estimation of floral symmetry in the Campanulaceae. ........................ 43
2-4 Estimation of ancestral character states for characters related to secondary pollen presentation in the Campanulaceae. ....................................................... 44
2-5 Ks plots showing the distribution of synonymous substitutions among gene duplicates. .......................................................................................................... 45
3-1 Plastid phylogeny of the Campanuloideae clade. ............................................... 71
3-2 PPR phylogeny of the Campanuloideae clade. .................................................. 72
3-3 Combined plastid and PPR phylogeny of the Campanuloideae clade. ............... 73
3-4 Support for plastid-only and combined plastid-PPR trees. ................................. 74
3-5 Divergence time estimates for combined plastid and PPR tree. ......................... 75
4-1 Occurrence map for the Roucela complex. ...................................................... 100
4-2 Results from concatenated and species tree analyses. .................................... 101
4-3 Chronogram of the Roucela clade showing ancestral range estimation. .......... 102
4-4 Niche modeling results for selected taxa in the Roucela complex. ................... 103
4-5 Tempo and pattern of diversification of the Roucela clade. .............................. 104
5-1 Sample localities, species distributions, and comparison of concatenated analyses. .......................................................................................................... 120
5-2 Comparison of species-tree analyses. .............................................................. 121
5-3 Hypothesized hybridization scenario. ............................................................... 122
5-4 Analyses of morphology data. .......................................................................... 123
11
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
EVOLUTION AND BIOGEOGRAPHY OF CAMPANULACEAE: FROM GLOBAL
PATTERNS TO SHALLOW, SPECIES-LEVEL PROCESSES IN THE MEDITERRANEAN BASIN
By
Andrew Alan Crowl
December 2016
Chair: Nicoletta Cellinese Major: Botany
The Campanulaceae are a diverse clade of flowering plants, encompassing more
than 2300 species, inhabiting myriad habitats from tropical rainforests to arctic tundra.
Using a phylogenetic framework, I inferred the placement and timing of major
biogeographic, genomic, and morphological changes in the history of the group. This
study highlights the diversity and complexity of historical processes driving evolution
within the Campanulaceae from broad, global patterns to shallow, species-level
processes in the Mediterranean Basin.
First, a robust, multi-gene phylogeny, including all major lineages, is presented to
provide a broad, evolutionary perspective of this clade across six continents. Ancestral
range estimation supports an out-of-Africa diversification following the KPg extinction
event. Chromosomal modeling provides evidence for numerous genome-wide
duplication events prior to movements into new, and often harsh, habitats.
Morphological reconstructions support the hypothesis that switches in floral symmetry
and anther dehiscence were important in the evolution of secondary pollen presentation
mechanisms.
12
The Mediterranean Basin is among the most biologically diverse areas in the
world, harboring enumerable poorly understood, species-rich groups. Here, I focus on
the Roucela complex (Campanula subgenus Roucela), 12 annual species found
primarily in the eastern Mediterranean Basin. Plastid and low-copy nuclear markers
were employed to reconstruct evolutionary relationships and provide insights into
patterns of endemism and diversification through time. Diversification of the Roucela
clade appears to have been primarily the result of vicariance driven by the break-up of
an ancient landmass. Contrary to past studies, my findings suggest the onset of the
Mediterranean climate has not promoted diversification in the Roucela complex and, in
fact, may be negatively affecting these species.
Because cryptic diversity was detected within the currently recognized,
widespread species, Campanula erinus, I used a Hyb-Seq approach to obtain two
genomic datasets (nuclear and plastome) across 105 C. erinus individuals, representing
27 populations spanning the Mediterranean Basin. Two lineages were recovered: a
western Mediterranean tetraploid lineage and an eastern Mediterranean octoploid
lineage. Nuclear gene tree topologies and network analyses indicate a hybrid origin for
the octoploid. The Cretan endemic C. creutzburgii (tetraploid) and the western
Mediterranean C. erinus (tetraploid) are implicated as parental lineages.
13
CHAPTER 1 OVERVIEW OF CAMPANULACEAE AND INTRODUCTORY REMARKS
With over 2300 species found across six continents (Lammers, 2007b), the
angiosperm clade, Campanulaceae, offers an opportunity to illuminate the historical
mechanisms responsible for the distribution of widespread, pan-tropical and pan-
holarctic biota, two questions that have long been of interest to botanists (Croizat, 1958;
Raven and Axelrod 1974; Sanmartin and Ronquist, 2004; Beaulieu et al., 2013).
Campanulaceae include five major lineages, variously treated as separate
families or subfamilies (e.g., Lammers, 1992, 2007a, 2007b). Little is known about the
early evolution of this group, primarily due to a poor fossil record and unresolved
relationships among the major lineages. Decades of systematic and phylogenetic
investigations into this diverse group have failed to satisfactorily resolve deep
phylogenetic relationships (see Ch. 2 for further discussion and citations). A robust,
comprehensive phylogeny is necessary to generate hypotheses regarding when, where,
and how this clade came to inhabit such a diverse array of habitats and exhibit
significant morphological variation.
Because many relationships at low phylogenetic scales remained unclear, two
low-copy nuclear loci from the pentatricopeptide repeat (PPR) gene family were
developed for use within the Campanuloideae clade (Ch. 3). This study represents the
first inclusion of low-copy nuclear genes for phylogenetic reconstruction in
Campanuloideae.
The Mediterranean Basin is among the most biologically diverse areas in the
world, harboring enumerable poorly understood, species-rich groups. Of the
approximately 25,000 plant species found in the Mediterranean, ca. 50-60% are found
14
nowhere else in the world (Cowling et al., 1996; Thompson et al., 2005). Understanding
evolutionary processes and species diversity is of special interest in this region given
the exceptionally high degree of endemism and high proportion of rare and endangered
taxa. Here I focus on the Roucela complex (Campanula subgenus Roucela). The
current distribution and high level of endemism in this clade appears to be the result of
allopatric speciation driven by the break-up of an ancient landmass (Ch. 4) with more
recent climatic fluctuations negatively affecting these sub-tropically adapted taxa. Gene-
tree discordance and ploidy estimates suggest allopolyploidy has created cryptic
diversity within the currently recognized, widespread species, Campanula erinus. (Ch.
5).
15
CHAPTER 2 A GLOBAL PERSPECTIVE ON CAMPANULACEAE: BIOGEOGRAPHIC, GENOMIC,
AND FLORAL EVOLUTION
Introduction
Historical mechanisms responsible for the distribution of widespread, pan-tropical
or pan-holarctic biota have long been of interest to botanists (Croizat, 1958; Raven and
Axelrod 1974; Sanmartin and Ronquist, 2004; Beaulieu et al., 2013). The
Campanulaceae are a clade that has diversified across both areas almost equally well.
The Campanulaceae are a large angiosperm group encompassing over 2300
species found on six continents (Lammers, 2007b). Little is known about its early
evolution, primarily due to a poor fossil record and unresolved relationships among the
major lineages. A robust, comprehensive phylogeny is necessary to generate
hypotheses regarding when, where, and how this clade came to inhabit such a diverse
array of habitats and exhibit significant morphological variation.
The Campanulaceae include five major lineages, variously treated as separate
families or subfamilies (e.g., Lammers, 1992, 2007a, 2007b): (i) Campanuloideae, a
group encompassing ca. 1000 species distributed worldwide with a center of diversity in
the holarctic, include primarily small perennials and represents the only lineage in the
Campanulaceae with radial floral symmetry; (ii) Lobelioideae, encompassing ca. 1200
species worldwide and a center of diversity in the New World tropics, include taxa with a
diverse array of habits from small herbs to tree-like ‘giant lobelias’; (iii) Cyphioideae
include 65 perennial herbs restricted to Africa; (iv) Nemacladoideae are a group of 19 Reprinted with permission from American Journal of Botany, Inc. Original publication: Crowl A.A., Miles N.W., Visger C.J., Hansen K., Ayers T., Haberle R., & Cellinese N. (2016) A global perspective on Campanulaceae: Biogeographic, genomic, and floral evolution. American journal of botany, 103, 233–245. Online access: http://www.amjbot.org/content/103/2/233
16
species known as ‘thread-plants’ and distributed in the southwestern United States and
northern Mexico; (v) Cyphocarpoideae include three poorly known species of small
annuals endemic to the Atacama Desert of Chile.
Phylogenetics
Early attempts to reconstruct phylogenetic relationships within the
Campanulaceae relied on morphology and single-gene datasets (Cosner et al., 1994;
Gustafsson and Bremer, 1995; Gustafsson et al., 1996; Haberle, 1998). These studies
included all five major lineages and confirmed the monophyly of the Campanulaceae
but recovered differing relationships within the family.
Phylogenetic analysis of morphological data resulted in a single clade including
taxa with bilateral floral symmetry sister to the radially-symmetric Campanuloideae
(Gustafsson and Bremer, 1995). Early molecular studies using the rbcL plastid gene
suggested a different topology: Cyphioideae was inferred as the earliest diverging clade
with a successive sister relationship of Cyphocarpoideae, Lobelioideae,
Nemacladoideae, and Campanuloideae (Cosner et al., 1994; Gustafsson et al., 1996).
Haberle (1998) and Ayers and Haberle (1999), using the nrITS region, found
Nemacladoideae sister to Campanuloideae and Cyphioideae sister to Lobelioideae plus
Cyphocarpoideae. To our knowledge, a study including all five Campanulaceae
lineages has not been conducted since.
The combined molecular and morphological analysis of Lundberg and Bremer
(2003) suggested Nemacladoideae sister to Lobelioideae and Campanuloideae sister to
Cyphioideae. These relationships have also been suggested by more recent, multi-gene
studies (Tank and Donoghue, 2010; Knox, 2014), though with low statistical support.
17
Recent molecular dating analyses of the Campanulaceae have shown the
divergence of these major lineages occurred within a relatively short time period
(Beaulieu et al., 2013; Knox, 2014), providing a potential explanation for the difficulty
previous studies have encountered in accurately resolving relationships within the
clade.
Biogeography
Accurate estimates for the age of the Campanulaceae and its individual lineages
within the clade have important implications for uncovering the biogeographic history of
this group and understanding the historical processes responsible for the current,
cosmopolitan distribution. Past estimates suggest an origin for the Campanulaceae in
the Cretaceous to the Paleogene (67-40 million years ago [Ma]; Wikstrom et al., 2001;
Bell et al. 2010; and Knox, 2014). Diversification of the Campanuloideae has been
estimated to begin at 56-23 Ma (Cellinese et al., 2009; Roquet et al., 2009; Mansion et
al., 2012; Crowl et al., 2014; Crowl et al., 2015) and the Lobelioideae at 88-50 Ma
(Antonelli, 2009).
With improved species sampling we are able to estimate divergence dates for all
five major lineages and infer putative dispersal events, which have led to the
cosmopolitan distribution of this clade. A Southern Hemisphere origin has been
hypothesized for the Asterales (Bremer and Gustafsson, 1997; Beaulieu et al. 2013),
agreeing with the out-of-Africa scenario seen in past analyses of Campanulaceae (Knox
et al., 2006; Antonelli, 2009). However, these studies focused mainly on the tropical
clades, Lobelioideae and Cyphioideae. Our sampling allows for more accurate
estimations regarding the ancestral area of this group and biogeographic patterns within
the clade.
18
Genome Evolution
Whole-genome duplication (WGD; polyploidy) events have been found to
facilitate movement into new environments and changes in morphology (see Stebbins,
1985; Leitch and Leitch, 2008; Flagel and Wendel, 2009; Parisod, 2012; te Beest et al.
2012). A comprehensive understanding of the relationships within Campanulaceae will
provide the basis to elucidate what role genomic evolution has played in the
evolutionary history this group. A large variation in chromosome number exists across
the clade, with every base number between 6 and 30 (and as high as 40) reported
(Lammers, 2007a). A long-standing debate exists regarding polyploidy in the group, with
little resolution about how to interpret chromosome counts in relation to genomic history
(Lammers, 1993; Stace and James, 1996). To date, the question of whether the high
chromosome numbers found in Campanulaceae are the result of an ancestral
polyploidy event, followed by descending disploidy or numerous, recent polyploidy
events, has not been answered.
Secondary Pollen Presentation
Secondary pollen presentation involves the relocation of pollen from the anthers
to disparate floral organs (Howell et al., 1993; Yeo, 1993). This feature is found across
many clades of angiosperms and throughout the Campanulaceae, with a variety of
mechanisms present (Erbar and Leins, 1989, 1995; Leins and Erbar, 1990, 2005; Yeo,
1993). However, in all cases, pollen presentation occurs via introrse anthers. In the
Campanuloideae and Nemacladoideae the anthers are connivent, forming an ‘anther
ring’ (or a ‘pollen box’ in the case of Cyphioideae), while they are connate in the
Lobelioideae, forming an ‘anther tube’ (Leins and Erbar, 1990; 2003).
19
Multiple, specialized mechanisms have evolved within the group. The
Campanuloideae exhibit a simple “deposition mechanism” where pollen is released onto
stylar hairs within the anther ring prior to anthesis (Leins and Erbar, 1990). The stamens
then quickly wither while the style elongates, exposing the pollen to the outside
environment. This is followed by invagination of the hairs, causing pollen to be released
(Leins and Erbar, 1990) and the style to become characteristically pitted. Within the
Campanuloideae, Phyteuma species have evolved a more complex ‘brushing
mechanism.’ This mechanism allows for deposition of pollen onto stylar hairs with the
aid of the corolla, which remains fused at the apical tip, and elongation of the style. The
growing style eventually ruptures through the corolla hood, presenting pollen to
pollinators (Leins and Erbar, 2006). The simple deposition mechanism of the
Cyphioideae is characterized by a pollen box in which the five connivent anthers form a
wall and the stylar tip forms the base of the box (Leins and Erbar 2003). Pollen is shed
into the box and onto the stylar tip prior to anthesis. Unlike in the Campanuloideae, the
simple deposition mechanism of Cyphioideae does not include style elongation or
invagination of the stylar hairs (Leins and Erbar, 2003, 2005). The pump mechanism of
the Lobelioideae is similar to that found in the Asteraceae (often referred to as a plunger
mechanism): following release of the pollen into the anther tube, stylar growth forces
pollen out of the top of the anther tube (Erbar and Leins, 1989; Leins and Erbar, 1990).
Very little is known about secondary pollen presentation in the Nemacladoideae, though
a preliminary study (C. Erbar and P. Leins, University of Heidelberg, personal
communication) suggests this mechanism is similar to that found in the
20
Campanuloideae with pollen deposition onto stylar hairs and retraction of these hairs at
anthesis. The mechanism of pollen presentation within Cyphocarpoideae is unknown.
Leins and Erbar (2006) presented an evolutionary interpretation of secondary
pollen presentation within the Campanulaceae, suggesting the pump mechanism of
Lobelioideae as the ancestral state for the clade. However, the precise phylogenetic
position of all lineages was unknown at the time. A robust, complete phylogeny of the
Campanulaceae has important implications for understanding how secondary pollen
presentation mechanisms may have evolved within this group. Furthermore,
understanding the genomic history and floral character evolution will allow us to uncover
potential floral development and genomic changes associated with secondary pollen
presentation.
Summary
Previous studies of the Campanulaceae have relied on either a small number of
gene regions or limited taxon sampling, resulting in incomplete hypotheses. We have
constructed the largest dataset to date, including 16 gene regions (plastid and nuclear)
and representatives (over 900 species) from all five major lineages to better understand
the evolutionary history of this diverse, cosmopolitan clade of flowering plants.
Methods
Sampling
Custom python scripts were used to mine GenBank
(http://www.ncbi.nlm.nih.gov/genbank) for all available Campanulaceae sequence data.
Sixteen loci were obtained: 15 plastid and one nuclear locus (atpB-rbcL Spacer, atpB,
atpF, atpF-atpH Spacer, atpH, ITS, matK, ndhF, pbsA-trnH Spacer, pbsA-trnK Spacer,
petD, rbcL, rpoC1, trnL-trnF Spacer, trnT-trnL Spacer, trnV-trnK Spacer). Additionally,
21
we extracted the same 15 plastid loci from 52 currently available plastomes (Knox,
2014). Targeted PCR (see methods of Crowl et al., 2014) was used to further fill gaps
by sequencing these individual plastid regions for Cyphocarpus rigescens Miers,
Porterella carnosula Torr, Nemacladus glanduliferus Jeps., and Pseudonemacladus
oppositifolius (B.L.Rob) McVaugh. We included 67 of the approximately 84 accepted
genera in the Campanulaceae.
The inclusion of a sufficient number of outgroup taxa may be necessary to
recover the early branching order of Campanulaceae lineages (Knox, 2014), therefore,
we included a diverse sampling of species in the Asterales, including taxa from the
Argophyllaceae, Goodeniaceae, Pentaphragmataceae, Phellinaceae, Rousseaceae,
and Stylidiaceae. Sequence data for outgroup taxa–Abrophyllum ornans (F.Muell.)
Benth., Carpodetus serratus J.R.Forst. & G.Forst., Corokia cotoneaster Raoul, Cuttsia
viburnea F. Muell., Pentaphragma ellipticum Poulsen, Phelline lucida Vieill. Ex Baill.,
Roussea simplex Sm., Scaevola sp., and Stylidium adnatum R.Br—were obtained from
GenBank and the 1KP project database (Matasci et al., 2014; http://www.onekp.com).
Phylogenetic Analyses
We constructed three datasets: (i) all-genes dataset - a total evidence alignment
(Kluge, 1989) containing 16 loci for 921 taxa; (ii) plastid-only dataset - alignment
containing only the 15 plastid loci; and (iii) coding-genes-only dataset - a reduced
alignment containing only the seven protein-coding regions.
Alignments were initially constructed using MUSCLE 3.7 (Edgar, 2004) using
default parameter values with a maximum of eight iterations. The alignments were then
manually adjusted (based on compositional similarity; Simmons, 2004) in Geneious
version 6.1.3 (http://www.geneious.com, Kearse et al., 2012) to correct problematic
22
regions. These alignments contain a substantial amount of missing data (Table 2-1).
Past studies, however, suggest a supermatrix approach, such as presented here, can
be successful, provided a sufficient proportion of phylogenetically-informative sites are
present in the dataset (Driskell et al., 2004; Burleigh et al., 2009). All alignments are
available through the Dryad repository (http://dx.doi.org/10.5061/dryad.322vn ).
Maximum likelihood analyses were run in RAxML v. 8.0.24 (Stamatakis, 2014)
using the GTRCAT model with separate model partitioning for each gene, spacer
region, and codon position in the coding regions. All analyses were run with 10
independent runs on distinct starting trees and 1000 thorough bootstrap replicates.
Character Reconstruction
We estimated ancestral states for a number of characters related to pollen
presentation in order to provide insights into the evolution of these mechanisms in
Campanulaceae. Using the all-genes Maximum Likelihood phylogeny, we reconstructed
ancestral states for corolla symmetry, anther dehiscence, stylar hair retraction, stylar
growth after anthesis, and the general pollen presentation mechanisms using Maximum
Likelihood approach as implemented in the APE package (Paradis et al., 2004) in R
v.3.1.0 (R Development Core Team, 2015). An Akaike Information Criterion (AIC;
Akaike, 1974) test was used to compare three transition models including a one-
parameter equal rates model, a symmetric model with equal forward and reverse
transitions between states, and an all-rates-different matrix. For the reconstruction of
secondary pollen presentation, we assigned taxa to one of three broad mechanisms:
simple deposition, brushing, or a pump mechanism.
23
Divergence Dating
We estimated divergence times using the penalized likelihood method
implemented in r8s 1.7 (Sanderson, 2002; 2003) as well as a relaxed molecular clock
method using the BEAST 2.1.2 package (Bouckaert, et al., 2014).
Optimal smoothing values for the r8s analyses were calculated for each dataset
by a cross-validation procedure. The best tree from each RAxML run was used as the
input topology and three min/max age constraints (see below) were used to estimate
divergence times for each tree.
BEAST analysis was run under an uncorrelated lognormal model for 108
generations, logging parameters every 1000 generations, and assuming a Yule
process. Due to computational constraints, the coding-genes-only dataset was used for
this analysis. Tracer v.1.6 (Rambaut et al., 2014) was used to assess effective sample
sizes (ESS values) for estimated parameters and to estimate burn-in. Twenty-five
percent of trees were removed as burn-in. We then used TreeAnnotator v.1.7.4
(Drummond et al., 2012) to summarize the statistics of our BEAST run from the
remaining trees and produce a summary tree.
The fossil record of the Campanulaceae is very poor. However, reliable fossil
seeds do exist for Campanula. These fossils are identified as Campanula sp. and
Campanula paleopyramidalis and date from the Miocene (approximately 17-16 Ma;
Lancucka-Srodoniowa, 1977, 1979). We applied a lognormal prior distribution constraint
to the node representing the last common ancestor of C. pyramidalis and C. carpatica
with a mean of 5.0, stdev=1.0, and offset =16. This, effectively, provides a 16 Ma
minimum age constraint for the fossil node. See Cellinese et al. (2009), Mansion et al.,
24
(2012), Olesen et al. (2012), and Crowl et al. (2014) for further discussion on the use of
this fossil calibration.
Two additional, previously published, calibration points were used to constrain
deeper nodes in the phylogeny. Bell et al. (2010) used numerous fossil calibrations to
date major angiosperm clades. We used date ranges from the 95% highest posterior
densities from this study to constrain the root of the Asterales (76-94 Ma) and the
Campanulaceae (41-67 Ma). We placed normal distribution priors on both of these
nodes, using the mean from each range reported in Bell et al. (2010) as the mean for
the prior distribution and a standard deviation of 5.0.
Biogeographic Analyses
Taxon occurrence data were taken from the Global Biodiversity Facility (GBIF;
http://www.gbif.org), inspected, and coded as presence or absence in seven large-
scale/global geographic areas defined by World Wildlife Federation Terrestrial
Ecoregions (Olson et al., 2001): (1) Afrotropics, (2) Australasia, (3) Indo-Malay, (4)
Nearctic, (5) Neotropics, (6) Oceania, and (7) Palearctic. Here, the Afrotropics region is
defined as it is by the World Wildlife Federation to include not only tropical regions of
Africa but also sub-Saharan Africa and the southern Arabian Peninsula.
We estimated ancestral biogeographic ranges using the BioGeoBEARS 0.2.1
package (Matzke, 2013) using the all-genes chronogram from the r8s analysis. All
models, including DEC, BAYAREA, and DIVALIKE, with and without the founder event
speciation parameter (+J), were tested and compared based on AIC scores.
BioGeoBEARS allows users to constrain the maximum number of ancestral areas to be
reconstructed at a given node; we tested each integer between two and seven.
25
Genome Evolution
Chromosome counts for 308 of the 964 ingroup taxa were available from the
digital version of the Index of Plant Chromosome Numbers (IPCN; Goldblatt and
Johnson, 1979). These counts, in combination with our all-genes-dataset phylogeny,
were used to infer ancestral chromosome numbers as well as the location and type of
chromosome number transitions recovered with the program chromEvol v2.0 (Glick and
Mayrose, 2014). Both the phylogram from RAxML and the Penalized Likelihood
ultrametric tree generated from r8s were analyzed in separate reconstructions. The
ultrametric tree from the BEAST analysis was not considered as it was only run using
the coding-genes-only dataset (due to computational restrictions) and, therefore, did not
include the full set of taxa (Table 2-1). AIC scores were used to compare eight different
likelihood models that estimate the rates of possible transitions, including whole
genome duplications, demi-duplications, and individual chromosome losses or gains,
along the phylogeny. Ten thousand simulations were performed with the best-fit model
to compare the inferred ploidal transition events with the simulated events along each
branch.
The DupPipes pipeline (Barker et al., 2008; available on the EvoPipes.net online
server; Barker et al., 2010) was used to assess evidence of WGD events by generating
a distribution of synonymous substitution rates between paralog pairs within the
assembled transcriptomes of Platycodon grandiflorus, Lobelia siphilitica, and Phelline
lucida. The pipeline uses a discontinguous MegaBLAST of a single transcriptome
against itself (all vs. all), identifying reciprocal best hits with greater than 40% sequence
similarity and a minimum length of 300 bp as probable paralogues. The reading frame
of the gene pairs is inferred through an amino acid alignment against available plant
26
genomes. The synonymous substitutions per synonymous sites (Ks) values of each
duplicate pair was calculated using PAML 4 (Yang, 1997). We then plotted a histogram
of all Ks values between zero and two using R (R Development core team, 2015). A
mixture model was implemented using the R package Mixtools 1.0.3 (Benaglia et al.,
2009) to identify underlying Gaussian distributions within the histogram. The inferred
peaks represent signatures of past duplication events.
Results
Phylogenetics
The various combinations of datasets, including the all-gene dataset, plastid-only
dataset, and coding-genes-only dataset showed a consistent backbone topology. The
coding-genes-only dataset provided the highest support for relationships between the
five major clades (Table 2-1). This is not a surprising result as this was the dataset with
the least amount of missing data and included, presumably, the slowest evolving
markers on average.
All analyses indicate that the Campanulaceae consist of five lineages
corresponding to the five traditionally-recognized sub-families or families (Figure 2-1).
Early in the evolutionary history of the group, the Campanulaceae appear to have
diverged into two clades: one including Cyphioideae sister to Campanuloideae, and the
other consisting of Nemacladoideae, Cyphocarpoideae, and Lobelioideae (Figure 2-1).
We recovered consistently high support for the placement of these lineages with the
exception of Cyphocarpoideae. Maximum Likelihood analysis of the coding-genes-only
dataset recovered this taxon as sister to the Lobelioideae with 63% bootstrap support
(Figure 2-1a). Alternatively, our Bayesian (BEAST) analysis of the same coding-genes-
27
only dataset inferred Cyphocarpoideae as sister to Nemacladoideae with a 0.98
posterior probability (Figure 2-1b).
Relationships within the largest clades, Lobelioideae and Campanuloideae, are
in agreement with previously delineated subclades, including Lobelioideae clades C1
through C8 in Antonelli (2008) and Campanuloideae clades Cam01 through Cam17 in
Mansion et al. (2012; Figure 2-2). When appropriate, we follow the clade naming
convention of these previous studies.
Within the Campanuloideae and Lobelioideae clades, we found many genera –
including Campanula and Lobelia – to be non-monophyletic, as suggested by past
analyses (e.g. Antonelli, 2008; Cellinese et al., 2009; Mansion et al., 2012; Crowl et al.,
2014). Our analyses recovered 50% of the genera for which more than one
representative was included to be paraphyletic or polyphyletic.
Interestingly, two of our datasets (the all-genes and plastid-only datasets)
recovered Wahlenbergia hederacea in the Wahlenbergieae clade, rather than sister to
Jasione – a relationship recovered in previous analyses (Cellinese et al., 2009; Haberle
et al., 2009; Prebble et al., 2011) and our coding-only dataset. In fact, the placement of
W. hederacea outside of the Wahlenbergieae led Eddie and Cupido (2014) to propose a
new generic name for this lineage. Our results, however, suggest more in-depth studies
are needed.
Divergence Dating
The divergence times recovered by r8s and BEAST provided similar results.
Estimates from r8s mostly fell within the 95% posterior credibility interval estimated by
BEAST. When they did differ, r8s ages were generally older (Table 2-2). The r8s
chronogram presented here was generated from the all-genes dataset (Figure 2-2),
28
while the BEAST chronogram was generated using the coding-genes-only dataset due
to computational constraints. We estimated the date for the split between
Campanulaceae and Rousseaceae to be approximately 76 Ma (86-67 Ma), a result
congruent with that obtained by a recent study aimed at estimating divergence dates
across the angiosperm tree of life (Magallón et al., 2015). Crown age estimates for the
major lineages within the family are presented in Table 2-2.
Character Reconstruction
The ‘equal-rates’ model was selected as the best-fit model for reconstructing the
evolution of corolla symmetry. Radial symmetry was inferred as the ancestral state for
the most recent common ancestor of Rousseaceae and Campanulaceae with a
transition to bilateral symmetry at the crown of the Campanulaceae clade. A reversal
back to radial symmetry was inferred for the Campanuloideae (Figure 2-3).
We inferred the ‘all-rates-different’ model as the best fit for all other characters.
Extrorse anther dehiscence was reconstructed as the ancestral state for the most recent
common ancestor of Rousseaceae and Campanulaceae with a transition to introrse
dehiscence and connivent anthers at the base of Campanulaceae (Figure 2-4). We
inferred the absence of both stylar hair retraction and stylar growth after anthesis as the
ancestral state of the Campanulaceae. These characters appear to have evolved twice
independently within the clade (Figure 2-4). A lack of secondary pollen presentation was
reconstructed as the ancestral state for the most recent ancestor of Rousseaceae and
Campanulaceae with a transition to simple deposition at the base of Campanulaceae
(Figure 2-4).
29
Biogeographic Analyses
The BioGeoBEARS analyses identified the BAYAREA+J model as the best fit
model and AIC scores showed this to be significantly better than BAYAREA without the
founder event parameter, “J” (p-value<<0.05). Allowing a maximum of two ancestral
areas in the reconstruction produced the lowest likelihood (Table 2-3).
The range expansion (d), range extinction (e), and founder-events (j) parameters
were found to be: d=0.0004, e=1.0e-7, and j=0.01. Our results indicate that the model
relied on founder-events far more than range expansion or extinction. This may be
explained by the easy dispersal of the minute seeds, characteristic of the family
(Antonelli, 2009).
The Afrotropics was inferred as the ancestral range for Campanulaceae as well
as for Cyphioideae and Lobelioideae (Figure 2-2). Multiple dispersal events into the
New World (Nearctic and Neotropics) appear to have occurred early in the evolutionary
history of the family, including the early diverging lineages of Nemacladoideae and
Cyphocarpoideae. The Afrotropics was recovered as the ancestral area of the
Cyphioideae/Campanuloideae clade. The Palearctic was recovered as the ancestral
range of the Campanuloideae, though the Afrotropics was inferred as the ancestral
range of the non-Cyanantheae taxa in the Campanuloideae clade (Figure 2-2),
indicating two dispersals into the Palearctic in early Campanuloideae history.
Genome Evolution
Chromosome modeling using the ultrametic (r8s-generated) and non-ultrametric
(RAxML generated) trees differed in the base number inferred for the family as well as
some of the polyploidy events inferred. For both trees the model of chromosome
30
evolution was equal, or very nearly equal, in AIC score. We restrict our discussion of
polyploidy events to only those shared by at least three taxa in our all-genes phylogeny.
Our DupPipes analyses found 3058 gene pairs with Ks<2 in the assembled
transcriptome dataset of Platycodon grandiflorus, 3087 in Lobelia siphilitica, and 8510 in
Phelline lucida. Two peaks are evident in the histograms of Platycodon grandiflorus with
medians of 0.5 and 1.55 Ks. Two similar peaks were found in Lobelia siphilitica Ks plots
with medians of 0.6 and 1.6 Ks (Figure 2-5). The analysis of Phelline lucida
(Phellinaceae) showed three peaks with medians of 0.05, 0.25, and 1.25 Ks (Figure 2-
5).
Discussion
Phylogenetics
Our phylogenetic results are generally in agreement with recent studies involving
the family (Lundberg and Bremer, 2003; Tank and Donoghue, 2010; Knox 2014). Our
analyses resolve the relationships of the five major lineages with high support except for
the placement of Cyphocarpoideae. We consistently recovered this lineage in the clade
containing Lobelioideae and Nemacladoideae but inferred conflicting sister relationships
with these lineages using maximum likelihood and Bayesian methods.
Cyphocarpoideae was represented in our dataset by only one species (of three extant
species) with seven plastid regions. Improved sampling of taxa and gene regions may
help resolve the position of this clade with respect to the Nemacladoideae and
Lobelioideae (Hansen et al., unpublished data). This may still prove difficult, however,
due to the apparent ancient, rapid diversification of these lineages (ca. 55-65 Ma;
Figures 2-1, 2-2).
31
Floral Evolution
Ancestral character state estimation supports the hypothesis that the most recent
common ancestor of extant Campanulaceae taxa had zygomorphic floral symmetry and
exhibited a simple deposition mechanism of secondary pollen presentation, with no
retraction of stylar hairs, no stylar growth after anthesis, and introrse anther dehiscence.
Our results support the hypothesis that the evolution of secondary pollen
presentation in Campanulaceae may have been functionally related to the evolution of
bilateral symmetry and anther dehiscence within the family. Our reconstruction inferred
a single transition to secondary pollen presentation, coinciding with the transition to
bilateral floral symmetry and introrse anther dehiscence at the base of the
Campanulaceae (Figures 2-3, 2-4). The evolution of radial floral symmetry with free
petals and extrorse anthers, as in Rousseaceae, to sympetalous, bilateral flowers in
Campanulaceae may have been associated with an adjusted anther orientation to better
allow for pollen deposition and uptake by pollinators. The introrse orientation of the
anthers was subsequently further modified to form an anther ring, tube, or box. We
recovered a simple deposition mechanism as the ancestral state for the group. From
this ancestor, multiple, specialized secondary pollen presentation strategies then
evolved within the group, including the pump mechanism of Lobelioideae and the
brushing mechanism of Phyteuma (Campanuloideae).
Our results indicate that Campanulaceae acquired bilateral flowers with simple
deposition early in the evolutionary history of the clade. The reversal back to radial floral
symmetry in Campanuloideae appears to have also been accompanied by a change in
the pollen presentation mechanism, namely, the retraction of stylar hairs. Because
pollinators do not enter the flower from a consistent angle, as is the case with bilateral
32
flowers, the deposition method with retractable hairs around the circumference of the
style is presumably more successful at releasing pollen to pollinators, regardless of the
angle at which the pollinator approaches the flower.
Bilateral flowers are known to attract butterfly, bat, and bird pollinators, all of
which are common in the tropics – the current centers of diversity for Cyphioideae and
Lobelioideae. The evolution of radial flower symmetry in the Campanuloideae has been
hypothesized to have evolved due to selection by bee and fly pollination, common
pollinators in northern temperate regions of the holartic (Yeo, 1993) – the current center
of diversity for the Campanuloideae. This hypothesis finds some support from our
biogeographic and character reconstruction analyses, which show that Campanuloideae
evolved the radial flower morphology at approximately the same time the ancestor of
this clade dispersed from the Afrotropics into the northern temperate Palearctic region.
We caution, however, that this must be viewed simply as a hypothesis, as these
changes occurred approximately 61-46 Ma (Figure 2-2), when the holartic was
experiencing a much different climate than presently (Zachos et al., 2001) and thus,
possibly different pollinator ecology.
Biogeography
The break up of the Pangaea landmass has long been invoked as a historical
factor responsible for distribution patterns of taxa with cosmopolitan distributions
(Huggett, 2004). Our age estimates, however, post-date this event and vicariance alone
can, therefore, not explain the current distribution of this group. Rather, more recent
dispersal must be invoked to explain the worldwide distribution of the Campanulaceae
clade. This claim is further supported by our biogeographic analyses. Our dataset
strongly favors the BAYAREA+J model, indicating that vicariance is not a significantly
33
important process compared to dispersal and founder events (Matzke, 2013). Results
from our biogeographic reconstruction are provided in Figure 2.
A Southern Hemisphere origin of Campanulaceae agrees with the hypothesized
origin of the Asterales and the rest of the Campanulidae (Beaulieu et al., 2013). The
many early-diverging African lineages within Cyphioideae, Campanuloideae, and
Lobelioideae suggest an African origin for the entire Campanulaceae. This is supported
by our ancestral range estimation and a trend pointed out by Linder (2014) that the
majority of African plants have diversified from within Africa rather than by migration to
this region from other parts of the world (Linder, 2014).
The presence of a long branch leading to the Campanulaceae clade with many
short internodes at the base of the group, dated to approximately 52-68 Ma, may be a
signature of mass extinction followed by rapid diversification at the KT Boundary. Our
results suggest that the Campanulaceae may have survived the KT extinction in Africa.
This is similar to the findings of Bockenak-Khelladi et al. (2010) and Bardon et al.
(2013), who suggested this scenario for the Poaceae and Chrysobalanaceae,
respectively. Sub-Saharan Africa, especially highland or montane areas, may have
acted as possible refugia during this event based on the global distribution of iridium
anomalies that mark the impact debris from the Chicxulub crater (Claeys et al., 2002)
and the hypothesized distribution of wildfires ignited by ejecta from the crater impact
(Kring and Durda, 2002), both of which are absent from large parts of Africa.
The Cyphioideae clade is found primarily in the Cape/Austro-temperate floral
region of Africa, where it originated and diversified in situ, with no apparent movement
34
out of Africa. Cyphioideae shows a similar crown age (ca. 20 Ma) to that of other clades
that have diversified in this region (Linder, 2014).
The current distributions of Nemacladoideae and Cyphocarpoideae both appear
to be the result of independent dispersal events (into the Nearctic and the Neotropics,
respectively) from the Afrotropics, early in the evolutionary history of Campanulaceae.
Similar to the Cyphioideae, both clades then diversified in situ, with no subsequent
movement out of these regions.
While the Lobelioideae clade originated in the Afrotropics (48-64 Ma), the
greatest diversity is found in the Neotropics. New World lobelioids are the results of
numerous, independent dispersal events from the Afrotropics into the Neotropics, and
subsequent exchanges between the Neotropics and the Nearctic (nine such events
were inferred in our dataset). All Nearctic lobelioid taxa included here are the result of
migrations from the Neotropics. The timing of these movements (40-50 Ma)
corresponds to the time when the Caribbean Plate created islands between North and
South America, providing a potential migration route (Morley, 2003). Neotropical
lineages, on the other hand, appear to be the result of a combination of movements
from the Afrotropics and the Nearctic.
The diversification of the Campanuloideae began approximately 46-61 Ma, when
land bridges connected parts of the Asian and European landmasses (e.g., McKenna,
1983). This also corresponds to the Paleocene-Eocene Thermal Maximum when the
Holarctic region experienced an increase in temperature and humidity, greatly
facilitating biological interchange across the Northern Hemisphere (Zachos et al., 2001).
35
The timing of the major radiation of Campanuloideae into the Palearctic
corresponds with the collision of the Arabian Peninsula with Eurasia, creating a land
bridge between Africa and the Palearctic in the late Eocene (~35 Ma). Multiple,
additional interchanges between the Afrotropic region and the Palearctic were
recovered in the Campanuloideae (Campanula balfourii, Wahlenbergia lobelioides, and
W. hederacea) and the Lobelioideae (Lobelia sessilifolia and L. urens). Canarina
represents another apparent instance of movement between these regions (Mairal et
al., 2015). Sequence data for African Canarina taxa, however, were not available at the
time we constructed this dataset.
Dispersal in the direction of northern temperate regions to the tropics is rare in
the Campanuloideae. The directionality of the above-mentioned movements was
inferred from the Afrotropics to the Palearctic for all but one taxon. We recovered one
dispersal event into the Afrotropics from the Palearctic – Campanula balfourii, endemic
to Socotra. Within the Lobelioideae clade, however, this dispersal pattern occurred
more frequently. As discussed above, numerous Nearctic-Neotropical exchanges were
recovered. Five of these events were inferred to have occurred from the Nearctic into
the Neotropics.
Our biogeographic analyses inferred multiple, independent movements from the
Palearctic to the Nearctic, all occurring between approximately 15-40 Ma. In fact, all
Nearctic campanuloids included in this study appear to be the result of dispersals from
the Palearctic. These events likely occurred through Beringia or a North Atlantic land
bridge, which may have persisted until more recently than previously thought (Denk et
36
al., 2011). Further investigation into the historical processes driving the European-North
American disjunctions are in progress (Crowl et al., unpublished data).
Similarity in floras of the Southern Hemisphere tropics is well known and
movement between them appears to be more common than between northern
temperate regions (Sanmartin and Ronquist, 2004). We inferred at least four
interchanges between these regions within the Lobelioideae, which diversified quickly
and frequently across both Neotropical and Afrotropical regions, and one recent
dispersal event within the Campanuloideae. Our biogeographic estimation suggests all
instances of movement were in the direction of the Afrotropics to the Neotropics.
We found migration from the Afrotropics to the Nearctic as a very rare
occurrence. We infer only one such movement, at the base of the Nemacladoideae.
However, if Cyphocarpoideae is sister to Nemacladoideae (as recovered in the BEAST
analysis), this would support a scenario of migration first into the Neotropics and then,
secondarily, into the Nearctic.
Genome Evolution
Two hypotheses have been put forth regarding chromosome evolution in the
Campanulaceae. Lammers (1993) postulated that the family evolved from a low diploid
base number, while Stace and James (1996) suggested an ancestral polyploidy event,
followed by descending disploidy leading to reduced numbers in the clade. Our results
from chromosome modeling support the former hypothesis and indicate an ancestral
base number of nine for the group. This appears to be primarily informed by the counts
of n=9 in the early diverging clades of Cyphioideae and Nemacladoideae, and is in
agreement with the prediction of Bremer et al. (2001), who postulated the same base
number for the Asterales.
37
Our Ks plots of two Campanulaceae and one Phellinaceae species further
support our prediction of the diploid base number. Platycodon grandiflorus, with
chromosome counts of n=9, and Lobelia siphilitica, with a count of n=7, show two
peaks, indicating two separate duplication events (Figure 2-5). Similar peaks were
recovered in the outgroup Phelline lucida, suggesting these events pre-date the
diversification of the Campanulaceae clade. The peaks we observe are likely the
duplication event shared by all eudicots, approximately 117 Ma (Jiao et al., 2012), and
the duplication shared by all angiosperms, approximately 192 Ma (Jiao et al., 2011).
Phelline lucida has a chromosome count of n=17 and shows evidence of a third, more
recent genome duplication that was not evident in the Campanulaceae taxa.
A previous study using gene-to-chromosome hybridizations of South American
lobelioids with high chromosome numbers (n=14) found evidence for a recent
duplication event shared among these tetraploid taxa (Vanzela et al., 1999). Results
from our chromosome modeling analyses are in agreement with this finding, again
bolstering the claim of a diploid ancestor for Campanulaceae and multiple, recent
duplication events within the clade.
Though past studies have suggested polyploidy likely played an important role in
the survival of major angiosperm lineages during the KT mass extinction event (Fawcett
et al., 2009; Soltis and Burleigh, 2009), we failed to find evidence for a polyploidy event
at the base of the Campanulaceae clade. Similarly, though previous studies have found
duplications associated with floral symmetry changes in related angiosperm clades,
such as the Dipsacales and Asteraceae (Howarth and Donoghue, 2005; Chapman et
al., 2008; Howarth and Donoghue, 2009), we found no evidence for polyploidy events
38
associated with such changes. This suggests that there was no duplication of the floral
symmetry gene regulatory pathway and subsequent subfunctionalization. However, as
described by Schranz et al., (2012) and confirmed by Tank et al., (2015),
polyploidization may facilitate key innovations following a lag-time – therefore, it is
difficult to fully disentangle the role of polyploidy from floral evolution in this clade.
In Campanulaceae, polyploidy appears to have played a role in the evolution of
island endemics, montane species, and insular woodiness. Interestingly, our analyses
also recovered polyploidy events at the base of the clades containing vernal pool
specialists, such as Downingia, and the serpentine soil specialists, Campanula exigua
and C. griffinii.
Hawaiian lobelioids are known polyploids (Lammers, 1988, 1993). Our
chromosome modeling predicts these taxa share a polyploidy event with African,
Neotropical, and Indo-Malayan species (clade C4; Figure 2-2). The giant lobelioids of
Hawaii and Africa appear to have evolved their woody habit from a single, common
ancestor (Antonelli, 2009). Our results confirm this assertion and indicate that this was
likely associated with a genome duplication event.
The evolution of woodiness in the Neotropical Centropogon-Burmeistera-
Siphocampylus clade (C8; Figure 2-2; referred to as the “CBS” clade by Batterman and
Lammers, 2004; Antonelli, 2009) appears to be associated with polyploidization. We
hypothesize a single polyploidy event at the base of this clade.
The woody Nesocodon-Heterochaenia-Berenice clade (Wahlenbergieae; Figure
2-2) from the Mascarene Islands is predicted to share a recent polyploidy event. We
infer, for the first time, the sister species to this clade as the East African, herbaceous,
39
diploid Wahlenbergia krebsii (n=7). We recovered three additional examples of recent
polyploidy events associated with woody island endemics: Azorina vidalii of the Azores,
Musschia wollastonii from Madeira, and Lobelia physaloides from New Zealand.
Ploidal differences within woody genera are known to be rare (Stebbins, 1971)
and most of the examples in our dataset follow a trend of the polyploidy event predating
the evolution of woodiness. An exception is Lobelia boninensis, a woody Hawaiian
lobelioid nested in a woody clade and predicted to have a recent, secondary polyploidy
event. Woody species are also thought to have higher chromosome numbers and
derived from tropical herbaceous species (Stebbins, 1971). Multiple Campanulaceae
taxa support this trend. In our data, the woody clades lacking a predicted polyploidy
event may simply be due to a lack of chromosome counts for those taxa (for example,
Theilera, Rhigiophyllum, Siphocodon, Lobelia rotundifolia, and some Wahlenbergia
taxa).
Conclusion
This study is the largest to date on the cosmopolitan and diverse
Campanulaceae, providing a broad phylogenetic perspective on biogeography,
chromosomal, and morphological evolution. Much work still needs to be done within the
Campanulaceae and it is our hope that this study will act as a framework for future
studies in diverse areas of research within this clade. For example, our hypotheses of
polyploidy will not only provide a framework for more in-depth investigations into
genomic evolution, but also help interpret phylogenetic results inferred from nuclear
markers within putative polyploid clades. Additionally, the scenario of floral
morphological evolution put forth in this study will be useful for interpreting floral
40
developmental genetics and guide future efforts in the investigation of secondary pollen
presentation and pollinator interactions.
Supplemental figures have not been included here. All supplementary material
pertaining to this study can be found in the online version of the manuscript:
http://www.amjbot.org/content/103/2/233.full.
Table 2-1. Comparison of the three data sets included in this study.
Dataset
Taxa
sampled
Characters in dataset
Proportion
of gaps
Bootstrap support for sister relationships
(Cyphioid.,Camp.) (Nemacl.(Cypho.,Lobel.)) Coding-only 468 7,284 58.03% 95 80(63)
Plastid-only 922 17,445 81.25% 75 85(37)
All-genes 974 18,606 82.82% 83 84(47)
Table 2-2. Summary of the two divergence-time estimate methods. Crown ages are
given for the major Campanulaceae clades in millions of years before present. Parenthetical values represent 95% highest posterior densities of ages for that node.
Crown clade r8s BEAST MRCA of Rousseaceae and Campanulaceae Campanulaceae
92 67
76 (67-86) 64 (56-72)
Campanuloideae 63
53 (46-61)
Cyphioideae 23
22 (14-33)
Lobelioideae 63
57 (49-65)
Nemacladoideae 60
44 (27-59)
41
Table 2-3. Likelihood statistics from ancestral area estimation models implemented in
BioGeoBEARS. Model
lnL
AIC
Likelihood Ratio Test P-value
BAYAREA -636.4641 1277 4.5e-82
BAYAREA+J -452.3424 910.7
DEC -515.1424 1034 3.6e-25
DEC + J -461.4439 928.9
DIVALIKE -550.9410 1106 1.6e-33
DIVALIKE+J -478.1609 962.3
Figure 2-1. Comparison of phylogenetic results from the coding-genes-only data set. A)
Maximum likelihood analysis. B) Bayesian analysis. Each of the five clades has been collapsed to a single branch for clarity and an illustration of a representative floral morphology is provided for each. Note the inconsistent placement of Cyphocarpoideae. Numbers on nodes are bootstrap support values (A) and posterior probability values (B). Values on the tips indicate support for the monophyly of the major lineages. Branch lengths are drawn proportional to time.
42
Figure 2-2. Chronogram of the Campanulaceae clade showing ancestral range
estimations and hypothesized polyploidy events. Branches are colored relative to geographic area indicated on the map. Gold circles on branches represent placement of hypothesized polyploidy events. Histogram at the tips of the tree represent chromosome count data used in this study. Lobelioideae clade names based on Antonelli (2008), Campanuloideae clade names refer to those of Mansion et al. (2012).
43
Figure 2-3. Ancestral estimation of floral symmetry in the Campanulaceae. Radial
symmetry is shown in black and bilateral symmetry is shown in red. Pie charts indicate the marginal likelihoods of the character reconstruction at that node. Node highlighted in green indicates ancestral state for the Campanulaceae.
44
Figure 2-4. Estimation of ancestral character states for characters related to secondary
pollen presentation in the Campanulaceae. Character states are given for each reconstruction. Pie charts indicate the marginal likelihoods of the character reconstruction at that node. Node highlighted in red indicates ancestral state for the Campanulaceae. Branching order and clade positions are identical to those in Figure 2-3.
45
Figure 2-5. Ks plots showing the distribution of synonymous substitutions among gene
duplicates. A) Platycodon grandiflorus. B) Lobelia siphilitica. C) Phelline lucida. Black lines indicate inferred peaks, representing signatures of past duplication events.
46
CHAPTER 3 PHYLOGENY OF CAMPANULOIDEAE (CAMPANULACEAE) WITH EMPHASIS ON
THE UTILITY OF NUCLEAR PENTATRICOPEPTIDE REPEAT (PPR) GENES
Introduction
Campanulaceae Jussieu are a nearly cosmopolitan group of flowering plants
comprising five subfamilies (Campanuloideae, Lobelioideae, Nemacladoideae,
Cyphioideae, and Cyphocarpoideae), approximately 84 traditionally circumscribed
genera, and more than 2300 species (Lammers, 2007). Historically, there has been
much disagreement as to intrafamilial classification (de Candolle, 1839; Fedorov, 1957;
Takhtajan, 1987), primarily due to the polyphyly of the largest genera, Campanula and
Wahlenbergia (Cellinese et al., 2009; Haberle et al., 2009; Prebble et al., 2011; Mansion
et al, 2012; Prebble et al., 2012).
The heterogeneous Campanuloideae Burnett (approximately 1054 species) are
found primarily in the Northern Hemisphere and are most abundant in temperate areas
of the Old World (Lammers, 2007). They are found from temperate to sub-tropical
areas and occupy a wide variety of habitats, from steppes to high elevation
mountainous regions. Some species have wide distribution ranges, spanning entire
continents, while others are narrow endemics, e.g., restricted to single islands.
Previous studies of the Campanulaceae and Campanuloideae have typically
focused on a few chloroplast genes (Cellinese et al., 2009; Haberle et al., 2009;
Antonelli, 2008; Roquet et al., 2008), one chloroplast marker with expanded taxon
Reprinted with permission from Public Library of Science. Original publication: Crowl A.A., Mavrodiev E., Mansion G., Haberle R., Pistarino A., Kamari G., Phitos D., Borsch T., & Cellinese N. (2014) Phylogeny of Campanuloideae (Campanulaceae) with Emphasis on the Utility of Nuclear Pentatricopeptide Repeat (PPR) Genes. PLoS ONE, 9, e94199. Online access: http://dx.doi.org/10.1371/journal.pone.0094199
47
sampling (Mansion et al, 2012; Borsch et al., 2009), or gene order in the highly
rearranged chloroplast genome (Cosner et al., 1991; Cosner et al., 1993; Cosner et al.,
2004). These studies have made significant progress toward a robust phylogenetic
hypothesis of the group and have highlighted the high level of paraphyly and polyphyly
of many traditionally circumscribed genera, especially Campanula and Wahlenbergia.
However, species-level relationships are yet to be understood and the most widely used
plastid markers within the family (i.e., atpB, matK, and rbcL) have not been able to
provide a significant level of resolution. Furthermore, focusing solely on maternally
inherited markers may obscure the role that hybridization may have played in the
evolutionary history of this group.
Additional studies have attempted to use nuclear data by including the internal
transcribed spacers (ITS) sequences of nuclear ribosomal DNA (Prebble et al., 2011;
Prebble et al., 2012; Roquet et al., 2008; Eddie et al., 2003; Park et al., 2006; Antonelli,
2009; Wendling et al., 2011). Although potentially informative at the species level, this
region is considerably difficult to align with confidence in positional homology across
wide phylogenetic distances in the Campanuloideae and is further complicated by
potential concerted evolution and high levels of homoplasy (for further discussion and
concerns see (Alvarez and Wendel, 2003), but see (Feliner and Rossello, 2007)).
Ultimately, past studies including ITS have shown its significant limitation in resolving
species level relationships and providing accurate information on the placement of
several genera (e.g., Jasione and Musschia).
Inferring robust phylogenies for species-rich clades is of great importance for
understanding processes of speciation, hybridization, and patterns in historical
48
biogeography while posing a major challenge to systematists (Mansion et al, 2012;
Borsch et al., 2009; Avis, 1994). Resolving relationships at low taxonomic levels can be
difficult for taxa that are closely related and/or recently diverged. Furthermore,
relationships at the interspecific level can be complicated by hybridization and
introgression. Thus, multiple rapidly evolving, independent nuclear markers may be
useful, and even necessary, to accurately reconstruct species-level phylogenies (Sang,
2002).
Current molecular and phylogenetic methods allow researchers to obtain large,
multi-gene datasets for phylogenetic studies. However, because of highly conserved
genome organization, gene order, and gene content of the chloroplast genome across
much of angiosperm diversity (but see Cosner et al., 1991; Cosner et al., 1993; Cosner
et al., 2004; Haberle, 2008 for exceptions in the Campanuloideae) and the relative ease
of developing universal primers for both chloroplast and nuclear ribosomal DNA, these
have been the most widely used sources of molecular data for plant phylogenetics
(Small et al., 2004). Although universal markers are more labor-intensive to develop
due to gene duplications and deletions (Sang, 2002), under-utilized low-copy nuclear
genes can be of great value to molecular phylogenetic studies.
Low-copy nuclear genes have a number of advantages over plastid regions: they
are unlinked (Steele et al., 2008), possess increased sequence variation (Gaut, 1998),
and are bi-parentally inherited (Sang, 2002; Small and Wendel, 2004; Sang, 2000).
Unlinked nuclear genes allow for multiple, independent datasets and, therefore,
independent estimates of phylogenetic relationships. This affords researchers with the
ability to utilize coalescent-based species tree approaches, the results of which may be
49
more accurate in the presence of incomplete lineage sorting. In contrast, the
chloroplast and mitochondrial genomes provide single markers due to gene linkage
(Steele et al., 2008; Moore, 1995). Furthermore, the often higher rate of sequence
evolution of low-copy nuclear genes (Sang, 2002; Small and Wendel, 2004) may allow
for greater phylogenetic resolution in clades containing slowly evolving or recently
diverged taxa.
However, working with nuclear markers has its limitations. Successfully
designing primers and amplifying target sequences can be quite difficult and labor-
intensive steps such as cloning are often necessary. In order to confidently reconstruct
species relationships it is of great importance to compare orthologous loci rather than
paralogous copies (Sang, 2002). Because most nuclear genes belong to multi-gene
families with different lineages containing losses or duplications the search for orthology
is a crucial limitation of working with nuclear genes (Sang, 2002; Doyle et al., 2003) and
great care must be taken. Focusing on single- or low-copy nuclear genes, however,
can alleviate this limitation.
In this study we compare results based on five chloroplast markers and 2 single-
copy nuclear loci from the pentatricopeptide repeat (PPR) gene family. The
phylogenetic utility of these nuclear loci for plant phylogenetic reconstruction has been
previously demonstrated by Yuan et al. (Yuan et al., 2009). The PPR loci were found to
have a single orthologue in both Oryza sativa and Arabidopsis thaliana and a rapid rate
of evolution, useful at the intergeneric and interspecific levels (Yuan et al., 2009).
Following empirical studies on Verbenaceae (Yuan et al., 2010), recent studies have
50
demonstrated the utility of these genes in plant phylogenetics (Drew and Sytsma, 2013;
Lu-Irving and Olmstead, 2013).
One of the strengths of using PPR loci is that orthology has been previously
assessed (Yuan et al., 2009). Additionally, because they are intronless, issues with
highly polymorphic introns are avoided (Yuan et al., 2009; Yuan et al., 2010).
In this study we evaluate the utility of two PPR loci to resolve evolutionary
relationships within the Campanuloideae. Our results suggest that these markers, when
considered separately and in combination with plastid data, can be informative tools for
phylogenetic reconstruction and for the detection of putative hybridization events.
Methods
Taxon Sampling, Amplification, & Sequencing
Taxa spanning the Campanuloideae clade were included in this study to test the
utility of two single-copy nuclear genes for reconstructing relationships across this large,
taxonomically diverse group as well as resolving relationships between closely related
species. Cyphia elata (Cyphioidae) and Solenopsis minuta (Lobelioideae) were used as
outgroup taxa based on previous studies (Haberle et al., 2009; Gustafsson and Bremer,
1995; Lundberg and Bremer, 2003).
A number of chloroplast (atpB, matK, petD, rbcL, and trnL-F) and ITS sequences
were taken from previously published works available from Genbank; additional taxa,
including all PPR sequences, were amplified as described below. Total genomic DNA
was extracted from silica dried leaf tissue and herbarium specimens following a
modified cetyltrimethyl ammonium bromide (CTAB) extraction protocol (Doyle and
Doyle, 1987).
51
Nuclear (PPR) primers were designed after Yuan et al. (Yuan et al., 2010). We
screened the primer pairs discussed in this study and found AT1G09680 and
AT3G09060 to give clean results when PCR products were directly amplified, with very
few polymorphic sites, suggesting a single copy of these loci within all tested
individuals. Following Lu-Irving & Olmstead (2013), these will hereafter be referred to
as PPR11 (AT1G09680) and PPR70 (AT3G09060), from the order in which Yuan et al.
(Yuan et al., 2009) list them. We found that for the PPR11 locus, 320F and 1590R
primers were the most successful within the Campanuloideae (Table 3-1). For the
PPR70 locus, we used the 930F and 2080R primers. Both of these primer pairs were
used for PCR amplification and sequencing. Chloroplast primer sequences used in this
study are also shown in Table 3-1.
In order to further verify orthology, we screened eight taxa across the
Campanuloideae for multiple copies of both PPR loci. Cloning followed the StratClone
PCR Cloning Kit protocol (Stratagene) following the manufacturer’s instructions.
Between two and eight colonies were picked, amplified, and sequenced using T7 and
T3 primers. An initial phylogeny included directly sequenced PCR products as well as
cloned sequences. Only a single sequence type of PPR70 was found in all individuals.
However, two distinct fragments were amplified using the PPR11 primers for
Campanula pelviformis, C. glomerata, and C. tubulosa. These two distinct sequence
types differed greatly in nucleotide composition and size (approximately 300 bp and 850
bp). We were able to easily distinguish between the ‘large-copy’ and ‘small-copy’ by
simple gel electrophoresis, suggesting cloning was unnecessary and paralogy was not
an issue in this dataset if all amplified fragments were of appropriate length. Given a
52
single band was visualized using gel electrophoresis for all subsequent taxa, we
proceeded with direct sequence following amplification.
All new sequences were amplified in 50 μl PCR reactions containing: 1 μl DNA,
10 μl 5X buffer, 5 μl of 25 mM MgCl2, 10 μl Betain, 4 μl of 0.1 μM dNTPs, 5 μl of 5 μM
primers, 1.25 units Taq polymerase (produced in the lab from E. coli), and water was
added to bring to volume. Amplification reactions for nuclear loci were run on an
automated thermal cycler under the following conditions: (1) initial denaturation was
carried out at 95°C for 2 min; (2) five cycles of 95°C for 1 min, 53°C for 1 min, and 72°C
for 2 min; (3) 32 cycles of 95°C for 1 min, 48°C for 1 min, and 72°C for 2 min; (4) a final
elongation step at 72°C for 12 min. Plastid regions were amplified following Haberle et
al. (Haberle et al., 2009) and Borsch et al. (Borsch et al., 2009).
Sequencing was carried out on an ABI Prism 3700 automated sequencer
(Applied Biosystems). Sequences were inspected, assembled, and edited using
Sequencher 4.9 (Gene Codes Corp., Ann Arbor, MI, USA). Initial alignments were
carried out using Muscle (Edgar, 2004) and manually adjusted in Se-Al v2.0 (Rambaut,
2002). Polymorphic sites in heterozygotes were coded using standard IUPAC
ambiguity codes. All sequences have been deposited in GenBank.
Phylogenetic Analyses
JModelTest (Posada, 2008) was used to determine appropriate models of
molecular evolution for all datasets using the Akaike Information Criterion (AIC) and
comparing –ln likelihood scores. The best-fitting models for each dataset are given in
Table 3-1.
All individual gene datasets were analyzed independently (atpB, matK, rbcL,
trnL-F, PPR11, and PPR70) before we analyzed the concatenated chloroplast and PPR
53
datasets. Because the individual datasets recovered largely congruent results, we
combined the PPR and chloroplast loci, using the plastid dataset as a ‘guide’ and
including only PPR accessions for which plastid data was also available. The combined
plastid-PPR matrix included 124 ingroup taxa and 7727 characters. All datasets were
analyzed using maximum likelihood and Bayesian Inference.
Maximum likelihood analyses were run in RAxML (version 7.0.4; Stamatakis,
2006) using the most appropriate model for each dataset. One thousand bootstrap
replicates were generated to measure clade support. Bayesian analyses were
conducted with MrBayes (version 3.1.2; Huelsenbeck and Ronquist, 2001; Ronquist and
Huelsenbeck, 2003) with the following settings. The maximum likelihood model
employed 6 substitution types (nst=6), with rate variation across sites modeled using a
gamma distribution, as well as a proportion of sites being invariant (rates=invgamma).
The Markov Chain Monte Carlo search was run with 4 chains for 5000000 generations,
with trees sampled every 1000 generations. We visually assessed convergence using
AWTY (Nylander et al., 2008).
Although a multi-species coalescent approach is likely to give more accurate
results for multiple unlinked partitions when compared to analyses of concatenated
datasets (e.g., Maddison and Knowles, 2006), our sampling of a single to a few
individuals per species and only three independent loci is likely insufficient to accurately
infer the species tree for Campanuloideae (Knowles and Kubatko, 2010).
Dating Analyses
Dating analyses were carried out with BEAST v1.7.4 (Drummond et al., 2012)
under an uncorrelated lognormal model. Twenty million generations were run logging
parameters every 1000 generations. Tracer v.1.5 (Drummond et al., 2012) was used to
54
visualize log files, assess success of runs, and calculate “burn-in” for each analysis.
Post burn-in trees were summarized with TreeAnnotator v.1.7.4 (Drummond et al.,
2012).
Although fossils for calibrating the Campanulaceae tree are limited,
Campanuloideae fossil seeds are available. These fossils, identified as Campanula sp.
and Campanula paleopyramidalis, date to the Miocene of the Nowy Sącz Basin in
Poland (Lancucka-Srodoniowa, 1977; Lancucka-Srodoniowa, 1979). Geological and
palynological studies have dated freshwater deposits of this formation to the Karpatian,
approximately 17-16 MYA (Oszczypko and Stucklik, 1972; Oszast and Stuchlik, 1977;
Nemcok et al., 1998).
Following Cellinese et al., 2009, we used the age of the well-determined C.
paleopyramidalis fossil as a constraint for the most recent common ancestor of C.
pyramidalis and C. carpatica. A lognormal prior distribution was applied to the fossil
constraint with a mean of 5.0, stdev of 1.0, and offset of 16. This gave a minimum age
constraint of 16 MYA for the node where the fossil was assigned, placing most of the
prior probability on this younger age, but still allowing older ages for this constrained
node. Placing this constraint on the most recent common ancestor of all Campanula
species gave marginally younger ages, as expected (Crowl, unpublished data), without
significantly changing our conclusions. Therefore, we restrict our discussion to the
former analysis because we agree with the identification provided by Lancucka-
Srodoniowa (1977) and Lancucka-Srodoniowa (1979).
We used two additional calibrations for the root of the tree based on a recent
study (Bell et al., 2010), which estimated dates for a number of major angiosperm
55
clades. Date ranges from the 95% highest posterior densities from this study were used
to constrain the split between the Campanuloideae and the Lobelioideae (41-67 MYA)
and the root of the Campanuloideae (28-56 MYA). Normal distribution priors were
placed on each of these nodes using the mean from each range reported in Bell et al.
(2010) as the mean for the prior distribution: 54 MYA for the Campanuloideae-
Lobelioideae split and 42 MYA for the Campanuloideae root and a stdev of 5.0 for both.
Results And Discussion
We generated 137 PPR11 sequences and 203 PPR70 sequences. The ITS
matrix included 209 taxa. The final chloroplast matrices consisted of 119 atpB, 120
matK, 183 petD, 125 rbcL, and 185 trnL-F sequences (Table 3-1). Results from the
plastid (Figure 3-1), PPR (Figure 3-2), and combined plastid-PPR (Figure 3-3) datasets
are discussed below. Dating analyses of these three datasets gave similar results and
we restrict our discussion to the combined plastid-PPR dated phylogeny (Figure 3-5).
Previous studies have attempted to utilize nuclear data to resolve relationships
within the Campanuloideae by using the ribosomal internal transcribed spacer (ITS)
region. As a way to further test the utility of PPR loci and directly compare results from
the nuclear genome, we inferred relationships using ITS sequences obtained from
GenBank (http://www.ncbi.nlm.nih.gov/genbank/). The ITS tree, despite a much larger
taxon sampling than the PPR dataset, fails to provide significant resolution within the
Campanuloideae clade. Although we recovered congruence between these datasets,
the sampling between them differs dramatically and we refrain from discussing these
results in detail.
56
Phylogenetic Resolution
Plastid loci
Chloroplast loci have been the markers of choice for the majority of past
phylogenetic studies on Campanulaceae. This is primarily due to the fact that universal
primers are readily available and these regions can be rapidly amplified with relative
ease. Furthermore, the high phylogenetic signal obtained at deep levels has made
them attractive to studies aimed at gaining a general understanding of the
Campanulaceae and resolving higher level relationships. Here we synthesize results
from the five chloroplast-marker datasets and identify clades of interest for this study.
Rather than discuss in detail each individual clade (which has been done in a recent
study, Mansion et al., 2012), we simply highlight those that are relevant to our
discussion on the utility and complications of the PPR markers. When appropriate,
reference is made to the ‘Cam’ clade number of Mansion et al., 2012 for each clade
discussed in this study.
All individual markers gave largely congruent results, which are generally in
agreement with recent hypotheses (Cellinese et al., 2009; Haberle et al., 2009; Mansion
et al, 2012; Roquet et al., 2008). We, therefore, combined all chloroplast loci into a
dataset that includes 124 ingroup taxa (Table 2-1). Figure 3-1 shows results from the
Maximum Likelihood analysis of the combined plastid dataset. Bayesian analyses
generated congruent results.
The resolution and support along the backbone of the chloroplast phylogeny is
largely consistent with past studies (Cellinese et al., 2009; Haberle et al., 2009; Mansion
et al, 2012). We found moderate support for the core Campanuloideae (node A: sensu
Borsch et al., 2009), which includes all Campanula species and close relatives. Node B
57
is only weakly supported while we recovered strong support for node C and the two
subclades, C1 and C2. Clade C contains the majority of Campanuloideae diversity and
represents the split between two species-rich clades that include Campanula species
and several segregate genera. The C1 and C2 clades, respectively, roughly correspond
to the Campanula s. str. and Rapunculus groups defined by traditional taxonomic
studies (Fedorov, 1957; Eddie et al., 2003; Boissier, 1875).
At the base of the Campanuloideae we found two clades that correspond to the
Platycodonoideae, including Platycodon and Canarina, and the Wahlenbergioideae,
including a polyphyletic Wahlenbergia. These results are also consistent with recent
studies (Haberle et al., 2009; Mansion et al, 2012; Eddie et al., 2003; Roquet et al.,
2009).
Clade D. The Southern Hemisphere Clade D is strongly supported and
represents the group that has been previously referred to as the Wahlenbergioideae
(Cellinese et al., 2009; Haberle et al., 2009; Eddie et al., 2003). The placement of this
clade as sister to the core Campanuloideae is maximally supported and includes
Heterochaenia ensifolia as sister to the polyphyletic Wahlenbergia, with the South
African Merciera tenuifolia nested within the Southern Hemisphere Wahlenbergia
species. Wahlenbergia hederacea falls outside of this clade and seems more closely
related to Jasione (Clade E). This taxon is morphologically distinct from others in the
genus and occurs in the northernmost range of Wahlenbergia (Haberle et al., 2009).
The non-monophyly of Wahlenbergia has been consistently recovered in past studies
(Cellinese et al., 2009; Haberle et al., 2009; Prebble et al., 2011; Mansion et al, 2012;
Prebble et al., 2012; Roquet et al., 2009; Olesen et al., 2012).
58
Clade E. Jasione plus Wahlenbergia hederacea are found to be sister to the rest
of the core Campanuloideae (Clade A). The placement of these taxa has been
problematic in past studies. Analyses based on ITS by Eddie et al. (2003) failed to
resolve their relationship to other members of the Campanuloideae and indicated them
as “transitional” taxa (transitional between “wahlenbergioids” and “campanuloids”).
Although our combined chloroplast dataset recovers the core Campanuloideae with
moderate support (node A) and strongly excludes W. hederacea from the
Wahlenbergioideae clade, we fail to obtain support for the exact placement of W.
hederacea or for this clade with respect to Clade F.
Clade F (Cam01). The next diverging group contains Gadellia, Musschia, and
two Eastern Mediterranean Campanula species, C. peregrina, and C. lactiflora. This
clade is strongly supported but its placement within Clade A is unclear (node B).
Similarly to Clade E, the exact placement of these “transitional” taxa (Eddie et al., 2003)
has been, so far, uncertain. Examining individual chloroplast gene trees reveals weak
support and inconsistencies in the position of clades E and F, leading to overall weak
support at node B. Dating analyses (see combined plastid and PPR dataset discussion
below) suggest rapid divergence of these two clades. Consequently, this dataset may
contain too few synapomorphies to confidently place these taxa.
Clade G (Cam14). This complex, within the Campanula s. str. clade (C1),
contains three morphologically similar species of annual Campanulas representing the
C. drabifolia complex (Carlström, 1986), sometimes referred to as the Roucela species
complex (Crowl et al., 2015). This maximally supported clade is composed of taxa
restricted to the Mediterranean Basin - most narrowly endemic within the Aegean
59
Archipelago. C. drabifolia (endemic to the mainland of Greece) is sister to a clade that
includes the widespread C. erinus and the Cretan endemic C. creutzburgii. Our results
are consistent with past studies (Cellinese et al., 2009; Haberle et al., 2009; Mansion et
al., 2012; Roquet et al., 2009) that found the C. drabifolia complex as monophyletic and
sister to a clade composed of western Mediterranean and North African taxa.
Clade H. This Cretan clade is recovered with high support. Campanula
pelviformis, C. carpatha, C. laciniata, and C. tubulosa are all endemic to Crete and
Karpathos islands except C. laciniata, which is also found in the Cyclades islands. This
clade is likely the result of a single introduction into the Cretan area and one of the few
examples of in situ diversification in the Cretan Campanuloideae (Cellinese et al.,
2009).
Clade I (Cam04). This highly supported clade contains the paraphyletic
Legousia (distributed primarily in southern Europe), and a North American clade
containing C. reverchonii and Triodanis species. Our plastid results confirm the non-
monophyly of Legousia with L. falcata sister to the North American clade (Cellinese et
al., 2009; Haberle et al., 2009; Mansion et al., 2012), suggesting a single introduction
into North American taxa and possible hybridization (see PPR results and discussion).
This North American-Mediterranean disjunction begs further study.
Clade J (Cam12). This poorly supported group is primarily composed of taxa
distributed in central and southern Europe (Alps and Apennines) and includes the
circumpolar C. rotundifolia. With the exception of Clade K and a couple species pairs
(C. cochleariifolia-cespitosa, and C. macrorhiza-forsythii), very little resolution is
obtained within Clade J. Clade K is composed primarily of taxa distributed in the Alps
60
and adjacent areas with the exceptions C. forsythii, endemic to Sardinia, and the
widespread and taxonomically problematic C. rotundifolia, distributed throughout
northern Europe and North America. Populations of C. rotundifolia show high
phenotypic plasticity and the cytological studies of Kovanda (1977) indicate at least two
ploidy levels, diploid (2n = 34) and tetraploid (2n = 68). Although Clade J includes the
majority of European alpine Campanula species, others are found within the C1 clade
(e.g., C. alpestris), suggesting multiple introductions into the Alps.
Pentatricopeptide repeat (PPR) loci
Both PPR loci (PPR11 and PPR70) recovered relationships that are largely
congruent with each other and with the chloroplast dataset. For example, the major
split within the core Campanuloideae (Clade C: C1 and C2), the placement of the
Wahlenbergioideae (Clade D), and other significant clades are recovered with both
PPR11 and PPR70 datasets and are consistent with results based on chloroplast data.
Therefore, we limit our discussion to the combined PPR (PPR11 plus PPR70) dataset,
which includes 111 ingroup taxa (Table 3-1).
The PPR dataset provides a well-resolved and highly supported backbone for the
Campanuloideae phylogeny. Early diverging clades D-F are resolved with much higher
support compared to the plastid tree (Figure 3-2), though the placement of these clades
is not always consistent (see below).
The Southern Hemisphere Wahlenbergioideae (Clade D) is maximally supported
as sister to the core Campanuloideae (Clade A) and includes Campanula pelviformis.
The Musschia-Gadellia clade (Clade F) is recovered as sister to the rest of the core
Campanuloideae with moderate support (node A). This group predates the divergence
of the Jasione clade (Clade E), which is strongly supported as sister to Clade C.
61
Relationships along the backbone of the Campanuloideae have been problematic and
this is the highest support obtained for the placement of these clades to date.
The placement of the maximally supported C. drabifolia complex (Clade G) sister
to a clade containing western Mediterranean and North African taxa is consistent with
plastid analyses.
The PPR dataset recovered a monophyletic Legousia sister to a clade containing
a Cretan endemic (C. cretica) and a North American species (C. reverchonii) within
Clade I. This is the first molecular study to suggest Legousia as monophyletic and,
specifically, L. falcata as sister to the rest of the Legousia clade (Clade I; Figure 3-2).
The phenomenon of incomplete lineage sorting is more prominent in recently
diverged taxa (Maddison and Knowles, 2006). Because of its diploid nature and
biparental inheritance, the nuclear genome may have an effective population size four
times larger than that of the chloroplast genome. The expected time to coalescence is
therefore four times longer (Sang, 2002; Moore, 1995), thereby increasing the
probability of finding ancestral polymorphisms in taxa of recent origin when using
nuclear loci (Olsen and Shaal, 1999). This poses a problem in the Campanuloideae as
many taxa are the result of recent diversification (Figure 3-5; Mansion et al., 2012;
Roquet et al., 2009). As a result, we conclude that lineage sorting is likely causing the
lack of species monophyly in the PPR tree (Figure 3-2). Given our sampling, however,
the amount of non-monophyly inferred (only a single species with high statistical
support) using these markers is minimal.
Similar to the plastid tree, the PPR markers alone provided very little resolution
within the C2 clade.
62
Combined plastid and PPR loci
Because the results of the individual analyses were largely congruent, we chose
to combine datasets in order to explore relationships of all included taxa and further
explore the utility of the PPR loci. The combined plastid-PPR dataset included the
same 121 taxa present in the chloroplast dataset.
This phylogeny is largely congruent with the results based on chloroplast data,
likely because of a strong signal being contributed by the more numerous plastid
markers. However, we found the addition of PPR loci to increase support at many
nodes within the Campanuloideae phylogeny. Interestingly, the increase in support
values spans from the backbone of the phylogeny (e.g., node A, node B, and node C) to
the terminal lineages. In Figure 3-3 we highlight the 23 nodes for which BS support
values increased compared to the plastid dataset. Figure 3-4 shows relative support for
these two trees with darker branches indicating increasing support.
The combined dataset corroborates the placement of the Jasione and Musschia
clades in the PPR tree (which contradicts the plastid results), though with only moderate
support. We found the divergence of the Musschia clade (Clade F) to pre-date the
divergence of the Jasione clade (Clade E). Although suggested before (Mansion et al.,
2012; Olesen et al., 2012), this is the strongest support to date for this relationship
(Figure 3-3). Dating analyses suggest a rapid divergence of these two clades in the
Late Eocene (Figure 3-5), providing a possible explanation for their uncertain
placement.
As in the plastid dataset, Wahlenbergia is again found to be polyphyletic.
However, the placement of W. hederacea is unsupported. A recent expanded
phylogeny of Wahlenbergia supports the exclusion of W. hederacea from the
63
Wahlenbergioideae clade (Prebble et al., 2011; Prebble et al., 2012). Although we were
unable to test the monophyly of Wahlenbergia with both PPR loci, the individual PPR11
analysis corroborates this relationship.
The combined plastid-PPR dataset places Campanula pelviformis in the Cretan
Clade H rather than Clade D as in the PPR tree (Figure 3-2), consistent with the
chloroplast results. Again, this is likely the result of five chloroplast markers contributing
more phylogenetic signal than the two PPR loci. Dating analyses support the
hypothesis that Clade H may have been the result of an in situ radiation in the Cretan
area (Cellinese et al., 2009) with the stem of this clade dating to approximately 9 million
years ago and diversification of the crown clade estimated at approximately 6 million
years ago (Figure 3-5).
Analysis of the combined plastid-PPR dataset inferred Legousia to be
monophyletic with strong support (Figure 3-3, Clade I). Three North American species
(Triodanis and C. reverchonii) form a clade sister to the Legousia clade. We are
currently investigating the relationship of Legousia with North American taxa at a finer
scale using complete taxon sampling and both plastid and nuclear data (Crowl et al., in
prep).
The combined plastid-PPR dataset recovered increased support Clade J, with
Campanula cenisia and C. elatinoides in a sister relationship and sister to this clade. A
clade containing C. excisa, C. fragilis, C. cochleariifolia, and C. caespitosa is found to
be sister to the rest of this clade with C. herminii and C. arvatica diverging next and
strongly supported as sisters. The overall resolution and support within Clade I is
increased when PPR loci are combined with the plastid matrix (Figure 3-3). However,
64
many relationships within Clade K could not be resolved with statistical confidence even
when nuclear loci were included (Figure 3-4). This is likely due to the recent origin of
this clade (Figure 3-5) and possibly a rapid radiation into the Alps.
Our results infer non-monophyly of Alpine and Caucasian taxa, indicating
multiple introductions into these areas, likely during the Pleistocene glaciation. Dating
analyses indicate all of these taxa diversified prior to the major glaciation of northern
and central Europe during the Pleistocene and, therefore, may have originated in
different areas. Both of these areas acted as refugia for many taxa during times of
unfavorable climatic conditions (Schönswetter et al., 2002; Tribsch et al., 2003; Dubey
et al., 2006) and it is likely that many Campanuloideae taxa followed this or a similar
pattern.
Of the six species that were found to be non-monophyletic in the PPR tree, three
of these (Campanula martinii, C. marchesettii, and C. bertolae) were found to be
monophyletic in the combined analyses (Clade K; Figure 3-3). We were unable to
assess the monophyly of the remaining three taxa because sequence data for multiple
accessions was not available.
Dating analyses indicate the Campanuloideae to be approximately 50 million
years old and the diversification of the core Campanuloideae (Clade A) beginning
approximately 38 million years ago. Much of the diversification of this group, however,
occurred within the last 5-10 million years (Figure 3-5), potentially complicating
phylogenetic reconstruction.
65
Plastid–Nuclear Incongruence
As discussed above, results from individual datasets gave largely congruent
results. However, we did discover discordant patterns between plastid and nuclear loci.
Topological contradictions between the two datasets are discussed below.
Lack of allelic monophyly within nuclear gene datasets and incongruence
between nuclear and plastid datasets may be caused by lineage sorting, gene flow, or
gene duplication leading to paralogous copies (Doyle, 1992; Maddison, 1997; Wendel,
1998; Small and Wendel, 2000). Because paralogy issues were accounted for within
the PPR dataset, the incongruences found here are likely due to incomplete lineage
sorting or hybridization.
Six species are found to be non-monophyletic in the PPR tree, although support
is low. These include C. bononiensis and C. rapunculoides in Clade C1, and C.
elatinoides, the recently described C. martinii (Fenaroli et al., 2013), C. bertolae, and C.
marchesetii in Clade J. C. elatinoides is the only species for which the non-monophyly
is well supported. Much of Campanuloideae diversity has occurred very recently (within
the last 10 MYA; Figure 3-5; Mansion et al., 2012) and, consequently, rapid divergence
events may have hindered lineage sorting. Therefore, the lack of allelic monophyly for
these six taxa could be the result of incomplete lineage sorting. Alternatively, recent
hybridization events may also explain these results. Many of the taxa found to be non-
monophyletic have overlapping or adjacent distributions in Italy and across the Alps,
suggesting past or recent hybridization events may be likely. However, further
investigation into phenology and other potential sources of pre- or post-zygotic
reproductive isolation between these sympatric species is required to fully understand
the patterns we have found here.
66
The placement of Clade E and Clade F are inconsistent between datasets with
the divergence of Clade E pre-dating the divergence of Clade F in the plastid tree
(Figure 3-1). This relationship is reversed in the nuclear tree, and with higher support.
This pattern is again recovered in analyses of the combined plastid plus nuclear
dataset. Dating analyses indicate these two clades diverged within approximately one
million years of each other (Figure 3-5). The disagreement in the placement of these
clades between plastid and nuclear markers – and even between studies using plastid
data – is likely to be the result of an ancient, rapid radiation (Figure 3-5).
Comparing PPR and chloroplast trees reveals a possible ancient hybridization
event. While the chloroplast dataset places Campanula pelviformis in a clade
containing other Cretan taxa (Clade H in Figure 3-1), the PPR dataset places this taxon
near the base of the Campanuloideae in Clade D, with Prismatocarpus and
Wahlenbergia, two Southern Hemisphere taxa. Multiple accessions of C. pelviformis
were included in our analyses, to verify the unusual placement of this taxon. This result
is quite surprising, however, and we cannot discount the possibility that this pattern is
due to retention of ancient paralogues in PPR11 or some other misleading phenomenon
that begs in-depth investigation into this taxon.
Further hybridization is possible for Legousia. Our results based on the
chloroplast dataset (Figure 3-1) are in agreement with other studies that have
consistently found Legousia to be non-monophyletic. Analyses based on the PPR
dataset, however, infer this group as monophyletic with strong support (Clade I; Figure
3-2), a result not previously recovered. While the plastid tree indicates Campanula
reverchonii closely related to L. falcata (Figure 3-1), the nuclear tree recovered this
67
taxon in a clade sister to a monophyletic Legousia (Clade I; Figure 3-2). The fact that
the non-monophyly occurs in the plastid tree while monophyly is recovered with the
nuclear dataset suggests that incomplete lineage sorting is unlikely (see discussion
below). Two possible scenarios could explain the paraphyletic Legousia inferred with
plastid data. As stated above, this could be the result of hybridization. Alternatively,
this may be the result of a long distance dispersal event from within the Legousia clade
to North America. In this case, the paraphyly inferred is simply due to insufficient time
for plastid loci to reach reciprocal monophyly. We are currently investigating this issue
in more detail (Crowl et al., unpublished).
While we can suggest general patterns of hybridization in Campanuloideae, it is
difficult to infer specific hybridization events because of incomplete taxon sampling (the
Campanuloideae include over 1,000 extant species, of which we have sampled
approximately 11% here). However, the increased phylogenetic resolution afforded by
these nuclear loci will make them useful in future studies aimed at disentangling
relationships within clades of closely related taxa and/or species complexes where
complete or near-complete sampling is possible (Crowl et al., 2015). Furthermore,
although we present likely scenarios regarding hybridization events, it is difficult to
distinguish between the processes of incomplete lineage sorting and hybridization in
causing discordance between gene trees. This is an active area of research and recent
studies have suggested methods in which species trees are employed to distinguish
between these two processes (eg. Kubatko et al., 2009; Larget et al., 2010; Joly, 2012).
We leave this for future studies, where complete taxon sampling of specific clades is
possible.
68
Final Conclusions
This study represents the first inclusion of low-copy nuclear markers for
phylogenetic reconstruction in Campanuloideae. The PPR loci included here present a
powerful tool for Campanulaceae phylogenetics as they provide independent
estimations of relationships, allowing researchers to uncover hybridization events and,
when used in combination with plastid data, yield increased resolution at both deep and
shallow phylogenetic levels. These loci were easy to sequence, required no cloning,
and the sequence alignments were straightforward across a large taxonomic breadth.
Our analyses recovered known relationships - often with increased statistical
support - and suggested relationships not previously recovered. For example, we
resolved the placement of two early diverging groups, the Jasione clade and Musschia-
Gadellia clade, with increased confidence. Because of the putative ancient, rapid
diversification that seems to have occurred (Figure 3-5), rapidly evolving markers such
as the PPR loci are necessary to capture this event and resolve the placement of such
clades.
Consistent with past studies, we find further evidence for the non-monophyly of
Wahlenbergia. Although we failed to find support for the precise placement of W.
hederacea, the non-monophyly of this group has now been corroborated by studies
employing plastid, nrITS, and low-copy nuclear data.
PPR loci analyzed alone and in combination with plastid data also recovered
relationships not previously suggested. Our results indicate that Legousia is, in fact, a
monophyletic group, as expected when considering morphology, a result missed when
only considering data from the chloroplast genome. The paraphyly inferred by
chloroplast analyses may be due to past hybridization, or the result of a single, long-
69
distance dispersal event from within the Legousia clade to North America. In the case
of the latter, the non-monophyly consistently recovered with plastid data is the result of
an insufficient amount of time passing for these loci to reach reciprocal monophyly.
This result highlights the importance of using multiple, independent loci when inferring
phylogenetic relationships and assessing taxon monophyly.
In order to obtain a comprehensive understanding of the evolutionary history of
the Campanulaceae, it is apparent that numerous independent, rapidly evolving loci will
be needed. Although many relationships still remain unresolved, likely due to the recent
origin and rapid diversification of many Campanula species (Figure 3-5), the inclusion of
the PPR loci presented here bring us one step closer to inferring a species level
phylogeny of this diverse clade of angiosperms. These markers may be of great use
especially in studies aimed at clades in which complete or near complete sampling is
possible, allowing for the discovery of past hybridization events, an aspect of a taxon’s
evolutionary history not captured when only organellar markers are considered.
Supplemental figures have not been included here. All supplementary material
pertaining to this study can be found in the online version of the manuscript:
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0094199.
70
Table 3-1. Chloroplast and nuclear markers used in this study.
Locus
Primer sequences (5’-3’)
Num. taxa
Total characters included
Variable Sites
Parsimony inform. sites Model (AIC)
atpB F: TATGAGAATCAATCCTACTACTTCT 119 1230 319 183 TPM1+G R: TCAGTACACAAACATTTAAGGTCAT matK F: TTTCAGGARTATATTATGCACTTGCT 120 1315 679 438 GTR+I+G R: GCGAAATAGARGAAGCTCTGG petD F: GCCGRMTTTATGTTAATGC 183 1228 596 420 SYM+G R: AATTTAGCYCTTAATACAGG rbcL F: ATGTCACCACAAACAGARACTAAAGC 125 1179 328 198 TPM2 R: GCAGTTATTGATAGACAGAAAAATCATGGT trnL-F F: CGAAATCGGTAGACGCTACG 185 1024 428 245 GTR+I R: ATTTGAACTGGTGACACGAG PPR11 F: TTTGTTATGTTGATKTGGGTTTT 137 826 510 399 TPM3+I+G R: GCCAGAAATAATAGCCGTGTAAG PPR70 F: AGTGCTYTGATTCATGGGTTGTG 203 981 693 492 TVM+I+G R: ACAGCTCKRACAAGTATRTTCCA Concatenated Plastid 121 5973 2039 1252 GTR+I+G
Concatenated PPR 116 1807 1012 713 TIM3+I+G
Concatenated Plastid+PPR 121 7727 3047 1892 GTR+I+G
71
Figure 3-1. Plastid phylogeny of the Campanuloideae clade. Best tree from maximum
likelihood analysis of combined plastid dataset: atpB, matK, petD, rbcL, and trnL-F. Numbers above branches are bootstrap values >50%. Numbers below branches indicate posterior probabilities from Bayesian analysis. Letters A-K refer to nodes and clades discussed in the text.
72
Figure 3-2. PPR phylogeny of the Campanuloideae clade. Best tree from maximum
likelihood analysis of combined PPR dataset: PPR11 and PPR70. Numbers above branches are bootstrap values >50%. Numbers below branches indicate posterior probabilities from Bayesian analysis. Letters refer to nodes and clades discussed in the text.
73
Figure 3-3. Combined plastid and PPR phylogeny of the Campanuloideae clade. Best
tree from maximum likelihood analysis of combined plastid-PPR dataset: atpB, matK, petD, rbcL, trnL-F, PPR11, and PPR70. Numbers above branches are bootstrap values >50%. Numbers below branches indicate posterior probabilities from Bayesian analysis. Letters refer to nodes and clades discussed in the text. Nodes for which bootstrap values are increased compared to the plastid-only analysis are highlighted in blue.
74
Figure 3-4. Support for plastid-only and combined plastid-PPR trees. Maximum-
likelihood trees from the plastid dataset (left) and the combined plastid-PPR dataset (right) with taxon names removed. Branches are shaded relative to BS support with darker branches indicating higher support. Letters correspond to clade/node names in text and in Figure 3-1 and Figure 3-3. Support for many clades is increased with the inclusion of PPR loci while other areas of the tree remain poorly supported.
75
Figure 3-5. Divergence time estimates for combined plastid and PPR tree.
Chronogram from BEAST analysis of the combined plastid-PPR dataset. Scale bar is in millions of years before present. Green star indicates placement of fossil constraint. Green diamonds indicate age constraints obtained from Bell et al. (2010). Numbers above branches indicate mean age estimates for clades (in millions of years). Error bars around nodes correspond to 95% highest posterior distributions of divergence times for clades discussed in the text.
76
CHAPTER 4 EVOLUTION AND BIOGEOGRAPHY OF THE ENDEMIC ROUCELA COMPLEX
(CAMPANULACEAE: CAMPANULA) IN THE EASTERN MEDITERRANEAN
Introduction
Spatial patterns of biological diversity are shaped by numerous factors, including
biotic interactions, habitat heterogeneity, area, climatic constraints, isolation, and
anthropogenic events (Huston, 1994). Uncovering the relative contributions of these
factors and evolutionary dynamics responsible for driving endemism is essential to
understanding plant diversity and may have important implications for conservation.
Endemic species are non-randomly distributed across terrestrial habitats and
appear to be concentrated in specific regions, or ‘hotspots’ of biodiversity (de Candolle,
1875; Kruckeberg and Rabinowitz, 1985; Myers et al., 2000), such as the
Mediterranean Basin (e.g., Médail & Quézel, 1997; Thompson, 2005). The complex, but
well understood, climatic and geologic history of this region provides an ideal setting for
studying endemism, evolution, and biogeography.
While the western Mediterranean Basin has been relatively well studied (e.g.,
Mansion et al., 2008; Mansion et al., 2009), the eastern basin remains poorly
understood. With a high degree of endemism and both oceanic and continental islands
present, this region affords a unique opportunity to better understand the processes
leading to endemism on these distinctly different classes of islands within the same
geographic area.
Reprinted with permission from John Wiley & Sons, Inc. Original publication: Crowl A.A., Visger C.J., Mansion G., Hand R., Wu H.-H., Kamari G., Phitos D., & Cellinese N. (2015) Evolution and biogeography of the endemic Roucela complex (Campanulaceae: Campanula) in the Eastern Mediterranean. Ecology and Evolution, 5, 5329–5343. Online access: http://onlinelibrary.wiley.com/doi/10.1002/ece3.1791/full
77
Oceanic and Continental Islands
Historically, islands were viewed as fragments of continents until Charles Darwin
and Alfred Russell Wallace made a distinction between continental islands, which have
had a past connection with the mainland, and oceanic islands - those that have arisen
from the ocean and have no history of continental connection (Darwin, 1859; Wallace,
1902).
These two types of islands are fundamentally different, both geologically and
biologically. Oceanic islands are formed by volcanic activity or tectonic events and arise
from the ocean, never having been in contact with an organismal source. They,
therefore, have initially empty ecological niche space. Continental islands, in contrast,
are formed by tectonic events or rising sea levels causing the break-up or isolation of a
fragment from the continent and contain a balanced flora and fauna at the time of
isolation. Crete, Kasos, Karpathos, Rhodes, and the numerous small islands off the
west coast of Turkey represent continental systems included in this study while Cyprus
is of oceanic origin.
Geologic and Climatic History of the Eastern Mediterranean Basin
The geologic and climatic history of the eastern Mediterranean Basin since the
Miocene is a complex combination of tectonic events, sea level changes, volcanism,
and a trend toward summer drought and increased seasonality. All of these events have
had a profound effect on the flora and fauna of the area (Thompson, 2005). Below we
lay out those that had the largest impact on biogeographic patterns in the eastern
Mediterranean and are, thus, potential drivers of diversification and current distribution
patterns in this region.
78
A continuous landmass (termed Ägäis) stretched from present day Turkey to
present day Greece, until approximately 12 Ma, when rising sea levels and tectonic
activity caused it to break up (Creutzburg, 1963; Dermitzakis, 1990; Triantis & Mylonas,
2009). This began the formation of the Aegean Archipelago and formed many of the
continental islands in the eastern Mediterranean. During this time (12-9 Ma), the Mid-
Aegean Trench (MAT) formed, causing a tectonic split between Crete and Karpathos,
stretching northward (Creutzburg, 1963; Dermitzakis, 1990).
With the closure of the Mediterranean’s connection with the Atlantic Ocean
approximately 5.96 Ma, a major desiccation of the Mediterranean Basin occurred (Hsu
et al., 1973; Krijgsman et al., 1999). The Messinian Salinity Crisis (MSC) led to the re-
connection of many islands to each other and to the mainland, potentially facilitating
dispersal between previously isolated areas. Approximately 5.33 Ma, this barrier was
broken and a rapid re-flooding of the basin occurred, leaving many islands once again
isolated (Krijgsman et al., 1999). This event led to extreme aridity and likely caused
significant extinction in sub-tropical lineages and diversification within arid-adapted
groups (Fiz-Palacios et al., 2010; Jimenez-Moreno et al., 2010).
Cyprus has an incredibly complex paleogeographic history and is one of the most
isolated islands in this region (Moores et al., 1984). The current configuration of the
island is, in fact, the result of a connection between two separate oceanic islands – the
Troodos Massif to the southwest and the Kyrenia Range to the north. The Troodos
Massif was likely an island by the Late Miocene, at which time the Kyrenia Range
began to rise (Hadjisterkotis et al., 2000). The collision of these two mountain ranges
was followed by the uplift of the central and coastal areas, resulting in the present island
79
formation during the Pliocene-Pleistocene transition (Hadjisterkotis et al., 2000; Yerkes,
2012). The Troodos Massif is dominated by diabase and serpentine soil while the
Kyrenia range is composed primarily of limestone.
More recent, and perhaps less dramatic, are the eustatic sea level changes
during the Pleistocene. Glacial and interglacial periods saw many islands in close
proximity to each other and the mainland going through periods of isolation and
reconnection. For example, land bridges likely connected many of the eastern Aegean
islands with each other and with mainland Turkey during the Last Glacial Maximum,
when the sea level was approximately 120 m below present levels (Shackleton, 1987).
Climatic fluctuations have been quite dramatic and significant in this area.
Subtropical conditions persisted through the early Miocene (23-16 Ma) with high
summer rainfall and little seasonal temperature changes. A gradual decrease in
summer rainfall and a trend toward increased aridification and seasonality began in the
middle Miocene (9-8 Ma) and continued into the Pliocene, leading to the establishment
of the current Mediterranean climate (3.4-2.8 Ma; Suc, 1984).
Roucela Complex
In this study we focus on the Roucela complex - referred to as the drabifolia
species complex by Carlström (1986) - which includes small, herbaceous, annual
Campanula species restricted to the Mediterranean Basin, characterized by the
presence of unappendaged calyx lobes (Carlström, 1986; Lammers, 2007). In the last
available revision, Carlström (1986) disentangled the group, formally recognizing 12
morphological species. One additional species, C. lycica, was later described and
added to this complex by Tan and Sorger (1987).
80
This taxonomically difficult group has historically been considered at various
ranks, including its own genus distinct from Campanula (Roucela; Dumortier, 1822), and
a subgenus of Campanula (Damboldt, 1976; Lammers, 2007). Here we repurpose the
old genus name, Roucela, in reference to the clade that includes the taxa subject of our
study.
The current distribution and high level of endemism in the Roucela complex
makes it an ideal model for understanding historical drivers of speciation and endemism
in the Mediterranean Basin. Species in the group are primarily found in the eastern
Mediterranean, with very restricted distributions, many endemic to a single or a few
islands in the Aegean Archipelago, western Turkey, or Cyprus (Figure 4-1). An
exception is the widespread C. erinus, found from the Azores, southern Europe and
northern Africa, to the Arabian Peninsula – an area broadly corresponding to the
Mediterranean climate zone. Interestingly, this is the only known self-compatible taxon.
Summary
Here we present a phylogeny of the Roucela complex, inclusive of all 13 taxa
traditionally recognized in this group. We assess the monophyly of the complex and
infer potential gene flow between species by utilizing five plastid and two nuclear loci
recently found to be informative in the Campanuloideae (Crowl et al., 2014). We use
molecular dating, biogeographic reconstruction, and diversification analyses to establish
timing of major splitting events and generate hypotheses regarding potential drivers
(and inhibitors) of diversification in this clade. Finally, we infer potential climatic niche
space of Roucela species by employing ecological niche modeling techniques to test
the hypothesis that climatic constraints are responsible for the narrow species
distributions observed.
81
Materials and Methods
Taxon Sampling and DNA Amplification
Taxon sampling for this study included multiple accessions of all 13
representatives of the Roucela complex as defined by Carlström (1986) and Tan and
Sorger (1987). In order to estimate divergence times, infer the placement of these taxa
with other Campanula species, and test the monophyly of the group, we analyzed
Roucela accessions within the context of a larger Campanuloideae dataset. Sampling
for these analyses followed Crowl et al. (2014).
Total genomic DNA was extracted from field collected, silica dried leaf material
and herbarium specimens following a modified cetyltrimethyl ammonium bromide
(CTAB) extraction protocol (Doyle and Doyle, 1987). We amplified and sequenced five
plastid regions (atpB, matK, petD, rbcL, and the atpB-rbcL intergenic spacer region) and
two low-copy nuclear loci (PPR11 and PPR70) from the pentatricopeptide repeat (PPR)
gene family.
All sequences were amplified following Crowl et al. (2014). Because orthology of
nuclear loci has been assessed by this previous study, PCR products of appropriate
size (approximately 800-1000 bp) were sequenced directly. All sequences have been
deposited in GenBank.
Phylogenetic Analysis
We used jModelTest 2 (Darriba et al., 2012) to determine appropriate models of
molecular evolution for individual loci. Maximum likelihood analyses were run in RAxML
(version 7.0.4; Stamatakis, 2006) with 1000 bootstrap replicates. Given that individual
gene trees provided congruent results we combined the five plastid and the two nuclear
markers into independent datasets in order to compare histories from these different
82
genomes. We then constructed a combined (concatenated) dataset including all seven
markers. The combined dataset included only accessions for which more than one
marker was available (Figure 4-2). Cyphia elata was used as the outgroup for plastid
and combined analyses while Solenopsis minuta served as the outgroup for nuclear
analyses due to availability of sequences. Datasets were partitioned by gene. The
combined alignment is available through the Dryad repository
(doi:10.5061/dryad.v6p3h).
Because phylogenetic analysis placed Campanula scutellata outside the Roucela
clade (see Results), this taxon was excluded from all subsequent analyses.
Species Tree
Multi-species coalescent approaches are likely to give more accurate results for
multiple unlinked partitions when compared to analyses of concatenated datasets
(Maddison and Knowles 2006). We conducted species tree analyses using a Bayesian
multispecies coalescent approach implemented in *BEAST (BEAST v.1.8.0; Heled and
Drummond 2010). Our dataset consisted of three independent loci: one plastid (all
plastid markers were treated as a single locus) and two nuclear loci. Simulation studies
(Maddison and Knowles 2006; McCormack et al. 2009; Heled and Drummond 2010)
have found that three independent loci across multiple individuals will accurately
recover the species tree for groups with divergence times much younger than we expect
in this study. We used a Yule prior for the species tree and applied the best-fit model
identified by jModelTest 2 (Darriba et al., 2012) for each locus. This analysis was run
with a chain length of 107 and a 10% cutoff was used for the burn-in.
83
Molecular Dating
Molecular dating analyses were performed under a relaxed molecular clock to
estimate divergence times for the Roucela complex. In order to utilize a fossil constraint,
we included these taxa within the context of a larger Campanuloideae dataset.
BEAST (v.1.8.0; Heled and Drummond 2010) analyses were run under an
uncorrelated lognormal model – which assumes no autocorrelation of rates – for 108
generations, logging parameters every 1000 generations, and assuming a Yule
process. We allowed BEAST to simultaneously infer tree topology and divergence dates
with no constraints on topology except for the outgroup. We used our three independent
loci (plastid, PPR11, PPR70) as input; jModelTest 2 (Darriba et al., 2012) was used to
find the best-fit model for each locus (GTR+Γ+I for all partitions). Tracer v.5.0
(Drummond et al., 2012) was used to assess effective sample sizes (ESS values) for
estimated parameters and to assess burn-in. Ten percent of trees were removed as
burn-in and summary statistics were calculated from the remaining trees using
TreeAnnotator v.1.7.4 (Drummond et al., 2012) to provide a summary tree (Figure 4-3).
The fossil record of the Campanulaceae is especially scarce (Lammers, 2007).
However, reliable fossil seeds do exist for Campanula. These fossils are identified as
Campanula sp. and Campanula paleopyramidalis from the Miocene (17-16 Ma) of the
Nowy Sacz Basin in Poland (Lancucka-Srodoniowa, 1977; 1979). See Crowl et al.
(2014) for further discussion.
A lognormal prior distribution was applied to the most recent common ancestor of
C. pyramidalis and C. carpatica with a mean of 5.0, stdev of 1.0, and an offset of 16,
giving a minimum age constraint for the fossil node. Furthermore, two additional
calibration points were used to constrain deeper nodes. We used previously obtained
84
dates from a recent study (Bell et al., 2010), which used numerous fossil calibrations to
date major clades within angiosperms. Date ranges from the 95% highest posterior
densities were used to constrain the node representing the split between the
Campanuloideae and the rest of the family (41-67 Ma) as well as the crown of the
Campanuloideae (28-56 Ma). Normal distribution priors were placed on both of these
nodes, using the mean from each range reported in Bell et al. (2010) and a stdev of 5.0.
Additionally, we estimated divergence times using a multispecies coalescent
(species tree) approach in *BEAST. Though species tree methods are often preferable
to concatenation approaches, they may not be appropriate for estimating divergence
times if gene flow is present between species, as is likely in the Roucela complex (see
Results). Estimation errors in divergence times can be greatly impacted by the migration
of even a single individual (Leaché et al., 2014). We confirmed this assertion and found
divergence dates an order of magnitude younger than the BEAST analysis and previous
estimates (Crowl et al., 2014; Mansion et al., 2012).
Biogeographic Analysis
We estimated ancestral ranges using the BioGeoBEARS package (Matzke 2013)
in R. All models, including DEC, BAYESAREALIKE, and DIVALIKE were tested.
Additionally, this program implements a founder-event speciation parameter (+J), which
may be important in island systems (Matzke, 2013). We used likelihood-ratio tests and
AIC values to compare the fit of these models to the data. Each taxon was coded for
presence/absence in six geographic areas: mainland Greece, Crete, Kasos/Karpathos,
Rhodes, Cyprus, and Turkey. Turkey was coded to include the islands immediately off
the west coast, which have been connected to the mainland in recent geologic history
relative to the diversification of the Rouclea taxa. For this analysis, we used the BEAST
85
maximum clade credibility tree with outgroups removed and species collapsed to a
single representative lineage. The maximum number of ancestral areas to be
reconstructed at each node was set to six.
We conducted three analyses in BioGeoBEARS: i) A non-time-stratified analysis
allowing unconstrained dispersal between all areas through time; ii) A time-stratified
analysis, with variable dispersal rates across five time intervals by subdividing the
phylogeny at 0.5 Ma, 5.33 Ma, 5.96 Ma, 12 Ma, and 25 Ma, corresponding to major
changes in the connections and formation of landmasses in the eastern Mediterranean
Basin; iii) Due to the uncertainty in the placement and monophyly of C. erinus, we
omitted this taxon from the chronogram and conducted a stratified analysis identical to
analysis (ii), above.
Ecological Niche Modeling
We used ecological niche modeling techniques to estimate potential climatic
niches for six species within the Roucela clade in order to better understand the nature
of endemism in the group and test the hypothesis that climatic constraints may be
responsible for the narrow distributions observed. Occurrence data were gathered from
both field observations and museum collections aggregated in GBIF (www.gbif.org).
Duplicate localities and points that were collected well outside of the expected range of
a taxon (likely misidentified specimens) were removed. Climatic datasets used had a
resolution of 1 km2. We, therefore, restricted occurrence points to a single locality per 1
km2 in order to avoid spatial autocorrelation and sampling bias using ENMtools (Warren
et al., 2010). Number of individuals varied from 21 (C. creutzburgii) to 124 (C. drabifolia)
after the above quality control. Due to insufficient sampling (fewer than 10 occurrences)
we were unable to confidently include the rare, narrow endemics C. kastellorizana, C.
86
lycica, C. podocarpa, C. raveyi, and C. veneris. Campanula erinus was excluded due to
its widespread distribution, occurring beyond the eastern Mediterranean (and, thus,
beyond the scope of this study) and insufficient sampling across its range.
Twenty global climate and elevation layers with a spatial resolution of 1 km2 were
obtained from the WorldClim database (Hijmans et al., 2005). Layers were clipped to
the eastern Mediterranean Basin using QGIS (Open Source Geospatial Foundation
Project, 2014) and 10,000 randomly distributed points were sampled across each layer.
The 10,000 values from the 20 layers were tested for correlation using JMP Pro v.11.
Climate layers found to be highly correlated (>0.70 Pearson’s correlation coefficient)
were excluded from subsequent analyses. When two layers were found to be highly
correlated, one was removed. This approach was implemented for all pairwise
comparisons until all remaining layers were below the 0.70 correlation threshold.
We used Maxent (v.3.3.3k; Phillips et al., 2006) to infer potential climatic niches.
This method requires occurrence-only data and has been found to perform well with
sample sizes as low as 10, a useful feature when studying narrow endemics
(Hernandez et al., 2006; Pearson et al., 2007). Default settings were used, with the
following exceptions: 10 subsampled replicates, test percentage of 15%, and 5000
maximum iterations. Statistical evaluation of niche and distribution model predictions
was done using the area under the curve (AUC) of the receiver operating characteristic
statistic, which provides a way to assess the ability of the model to correctly predict
distributions of the training points.
Diversification
We utilized a number of diversification methods implemented in R (v.3.1.0; R
Development Core Team, 2008) to better understand patterns of the timing and tempo
87
of lineage diversification within the Roucela clade and test the hypothesis that allopatric
speciation caused by the break-up of the Aegean landmass is responsible for much of
the diversification within the group. The chronogram from the BEAST analysis was
trimmed to include only a single accession for all Roucela taxa. Campanula erinus was
also reduced to a single accession, giving a topology that approximated the species
tree. Although this currently recognized species may represent multiple cryptic taxa,
until further studies can disentangle this complex with confidence, we chose to follow
current taxonomy and represent C. erinus as a single lineage for these analyses.
Lineage-through-time (LTT) plots were constructed for the post-burn-in posterior
distribution of trees using the APE package (v.3.1-1; Paradis et al., 2004). We
calculated the gamma statistic (Pybus & Harvey, 2000) as implemented in GEIGER
(Harmon et al., 2008) to test if diversification rates have been constant through time for
this clade.
We then further explored diversification rates through time following the approach
of Simpson et al. (2011). This method uses the LASER package (Rabosky, 2006) and
calculates diversification rates across the tree by estimating the number of nodes and
their corresponding branch lengths within a sliding window. We used a window width of
five million years.
A variety of diversification models were then fit to our data. Five models were
tested within a maximum-likelihood framework using the LASER package. Two of these
models – pure birth (PB) and birth-death (BD)– assume constant rates, while the
remaining models allow for temporal rate variability: linear diversity dependent (DDL),
exponential diversity-dependent (DDX), and two-parameter Yule (y2r). We then
88
computed and compared Akaike Information Criterion (AIC) scores to assess model fit.
Our sampling included all extant species within the Roucela complex and, therefore,
incomplete taxon sampling should not be an issue in these analyses.
Results
Phylogenetic Analyses
All analyses recovered the Roucela complex as monophyletic with the exclusion
of the Balkan endemic, Campanula scutellata. This taxon falls well outside the Roucela
clade, a result not surprising when considering its distinct morphology and chromosome
number. The broader corolla and larger overall size of this species led Carlström (1986)
to question its placement within Roucela. Our analyses corroborate Mansion et al.
(2012) and place C. scutellata with other annuals in the Megalocalyx clade.
Both concatenation and species tree analyses were performed in order to
compare results from these different methods. Though results were very similar, we
chose to present both trees as the phylogeny resulting from the concatenated dataset
provides relationships within species (between populations), allowing us to make
inferences regarding biogeographic history of taxa and potential gene flow between
species. Results from these analyses are discussed below.
Plastid
Our plastid dataset recovered a strongly supported Roucela clade sister to a
clade containing the North African taxa Feeria angustifolia and C. edulis, the Azorian
endemic Azorina vidalii, and C. mollis, distributed in the western Mediterranean. This
placement within the Campanuloideae is consistent with previous studies (Cellinese et
al., 2009; Haberle et al., 2009; Mansion et al., 2012; Crowl et al., 2014).
89
Within the Roucela complex three clades can be distinguished, corresponding to
an ‘eastern grade’ and a ‘western clade’. The eastern grade is composed of two clades
found only east of the Mid-Aegean Trench (MAT). The earliest diverging clade includes
two species, C. pinatzii (found on Karpathos and Kasos) and the very narrow endemic
C. kastellorizana (restricted to the island of Kastellorizo). The second eastern clade
contains C. raveyi (western Turkey), C. rhodensis (endemic to Rhodes), C. lycica
(western Turkey and Kastellorizo), C. delicatula (SW Turkey, SE Aegean, and Cyprus),
C. veneris (narrowly endemic in the Troodos Mountains of Cyprus), and C. podocarpa
(SW Turkey, eastern Aegean Islands, and Cyprus).
The ‘western clade’ includes two species found exclusively west of the MAT, C.
creutzburgii (endemic to Crete) and C. drabifolia (mainland Greece), as well as the
widespread C. erinus and one eastern species, C. simulans (eastern Aegean Islands
and western Turkey).
Campanula erinus was inferred to be polyphyletic. Interestingly, plastid markers
found individuals of this taxon to fall into three clades, roughly corresponding to
geographic regions. Support for relationships among C. erinus populations and other
taxa in this clade, however, are not sufficient to draw meaningful conclusions.
Nuclear
Nuclear and plastid loci gave largely similar results. The placement of the
Roucela complex within the Campanuloideae is consistent with both datasets. The
nuclear dataset also recovered three clades within a monophyletic Roucela. Although
the relationships among these clades are poorly supported, the content of the clades is
largely consistent.
90
Paralogy was previously tested for the nuclear genes within this clade (Crowl et
al., 2014), and therefore, incongruences in this study are likely the result of incomplete
lineage sorting and/or hybridization. We recovered two likely instance of hybridization.
Two accessions of C. lycica are found to be sister to C. kastellorizana in the nuclear
phylogeny. These taxa are sympatric on the island of Kastellorizo, where these
individuals were collected, suggesting the inconsistent placement of C. lycica between
nuclear and plastid markers may be the result of interspecific gene flow. A second
instance is possible within C. erinus, though a more in-depth study needs to be carried
out to verify this assertion.
Species tree and combined dataset
Results from the independent plastid and nuclear loci did not show significant
(highly supported) incongruences; therefore, we combined all loci into a single dataset.
The concatenated dataset generated similar results to the plastid-only analyses, but
with increased support for many relationships. As an alternative, results from the
concatenated dataset were compared to a species tree. Our concatenated and species
tree analyses recovered equivalent species relationships (with one exception) within a
strongly supported Roucela clade (Figure 4-2).
Similar to the plastid-only results, we found three well-supported clades that
correspond to an eastern grade and a western clade. Relationships within these clades
are nearly identical to those obtained with the plastid-only dataset. By concatenating
plastid and nuclear loci, however, we were able to increase resolution toward the
terminals of the phylogeny, uncovering intra-specific (population-level) structure.
This dataset recovered two clades within C. podocarpa: one contains only
accessions from the Turkish mainland while the other includes only individuals from
91
Cyprus. A sister relationship is inferred for C. podocarpa and the narrow Cyprus
endemic, C. veneris. Similarly, within the C. delicatula clade, we recovered a grade of
Turkish accessions sister to a clade (although poorly supported) of Aegean populations.
Resolution within the western clade was not significantly improved in the
combined dataset. Although we recovered increased support for the monophyly of C.
simulans, this dataset did not provide definitive results regarding relationships within the
clade or monophyly of C. creutzburgii, C. drabifolia, or C. erinus. The species tree
analysis, however, recovered well-supported relationships for all four taxa. This analysis
found the widespread C. erinus to be sister to C. creutzburgii, a species endemic to the
island of Crete. However, all other analyses (non-species-tree methods) recovered C.
erinus as polyphyletic with respect to C. creutzburgii or C. drabifolia.
Divergence Time Estimates and Ancestral Range Estimation
Divergence estimates obtained from a concatenated dataset of the broader
Campanuloideae clade generated ages older than expected given the low level of
morphological divergence between Roucela taxa. This is congruent with the results of
Mansion et al. (2012). The Roucela clade is inferred to have originated in the early
Miocene (Figure 4-3). The early-diverging eastern clade dates to approximately 22 Ma.
The split between the western clade and eastern clades occurred approximately 19 Ma.
Ancestral range estimation using BioGeoBEARS recovered the DEC model as
the best-fit model and inclusion of the founder event parameter (+J) significantly
(p<0.05) improved the log likelihood of this model for all analyses. The time-stratified
analysis excluding C. erinus was inferred to be significantly better than other analyses..
The origin of the Roucela clade was inferred as Karpathos + Rodos + Turkey +
the eastern Aegean (Figure 4-3). During the early Miocene, all of these areas were
92
connected as a single landmass, suggesting the eastern Aegean landmass as the
ancestral range. Multiple vicariance and dispersal events were inferred during cycles of
island connection and isolation as this landmass fragmented (Figure 4-3).
Niche Modeling
Of the climatic layers included in this study, elevation, mean temperature of
coldest quarter, precipitation of wettest quarter, precipitation of driest quarter, mean
diurnal range, temperature seasonality, minimum temperature of coldest month, and
mean temperature of wettest quarter were found to be least correlated.
Using the above climatic layers and current distribution data, niche modeling
results recovered an average AUC score of 0.926 to 0.950, with standard deviations
from 0.030 to 0.065, suggesting low rates of inaccurate predictions.
Geographically, fundamental niches appear to be broader than realized niches in
all species tested (Figure 4-4). In other words, realized niches appear to be a subset of
the fundamental niche space and there exists unoccupied, potentially suitable habitat
beyond the current distribution of all taxa.
Diversification
The LTT plot shows a rapid accumulation of lineages through the mid-Miocene,
then quickly tapers-off beginning approximately 8-7 Ma (Figure 4-5). We obtained a
gamma value of -2.2976 and a p-value of 0.01, indicating we can reject the null
hypothesis of ‘rates have not decreased over time’ for this clade. The large negative
value for the gamma statistic can be interpreted as evidence for a decrease in
speciation rate through time. The LTT plot and sliding window analysis verify this.
Results from the sliding window method indicate a burst of speciation between
93
approximately 12 Ma and 8 Ma (Figure 4-5). This is followed by a decline in
diversification, persisting to the present.
We obtained AIC values for the diversification models listed above ranging
between 33.864 and 43.073. Our results, based on the AIC score for each model,
suggest that the DDL (logistic diversity-dependent; AIC=33.864) model and the y2r
(two-rate Yule; AIC=35.605; st=8.9) are the best fit to the data. Because these models
are within 2 AIC values of each other, we consider both equally likely (Burnham &
Anderson, 2002).
Discussion
Evolution and Biogeography of the Roucela Clade
Both dispersal and vicariance appear to be historically important processes in
driving the biogeographic patterns we observe in the Roucela clade. The group likely
originated from a Eurasian ancestor (see also Mansion et al., 2012) during the early
Miocene when the area was experiencing a sub-tropical climate (Figure 4-3). Contrary
to past studies of Mediterranean biota (e.g. Yesson and Culham, 2006; Postigo et al.,
2009; Fiz-Palacios & Valcárcel, 2013), our results indicate the onset of the
Mediterranean climate has not promoted diversification within the clade. In fact, the shift
to increased seasonality and decreased rainfall appears to have greatly slowed the rate
of diversification (Figure 4-5). Speciation was, instead, likely the result of ancient
geologic and tectonic events, which led to numerous cycles of island connection and
isolation (Figure 4-3).
Dating and diversification analyses indicate the timing of diversification within the
Roucela clade coincides with the break-up of an ancient landmass (Ägäis) in the
Miocene, suggesting vicariance as a likely driver of diversification in the clade (Figure 4-
94
3; Figure 4-5). Additionally, we found geologic events such as the formation of the Mid-
Aegean Trench and the Messinian Salinity Crisis to be historically important in the
evolutionary history of this group.
Break-up of Aegean landmass
The complex geologic history of the Aegean has played a significant role in
shaping biogeographic patterns in the area. One such event is the fragmentation of an
ancient landmass during the middle Miocene. Tectonic movements and changing sea
levels, beginning approximately 12 Ma, caused the Aegean landmass to break apart –
eventually forming the continental island system of the Aegean Archipelago
(Dermitzakis and Papanikolaou, 1981).
Molecular dating and ancestral range estimation suggest the Roucela clade
originated while this landmass was still intact and we find evidence that much of the
diversification within the group is the result of vicariance caused by its break-up (Figure
4-3; Figure 4-5). Specifically, Campanula drabifolia, C. rhodensis, C. pinatzii, and C.
kastellorizana were found to be the result of vicariance driven by rising sea levels and
continental fragmentation (Figure 4-3).
The separation of the eastern and western Aegean, approximately 12-9 Ma, led
to the formation of the Mid-Aegean Trench (Creutzburg, 1963; Dermitzakis, 1990),
which has long acted as a barrier to dispersal and has been hypothesized to be a major
factor in the evolutionary history of many taxa, including reptiles (Poulakakis et al.,
2005, 2008, 2013; Lymberakis and Poulakakis, 2010), scorpions (Parmakelis et al.,
2006), land snails (Kornilios et al., 2009), and plants (Strid, 1996; Bittkau and Comes,
2005). This formation appears to be important in the current species distribution within
the Roucela clade.
95
Roucela taxa show a clear geographic pattern and can be divided into eastern
and western species relative to the Mid-Aegean Trench, with the exception of C. erinus,
which is widespread throughout the Mediterranean and C. delicatula, which occurs
primarily east of the trench but is also distributed on islands just west of it (Carlström,
1986).
Dating analysis indicates that C. delicatula diverged prior to the formation of the
Mid-Aegean Trench (Figure 4-3), giving a possible explanation for the distribution of this
taxon. However, this species seems to have dispersed to Cyprus (see below),
suggesting that dispersal over this barrier cannot be discounted as a possibility.
Unfortunately, individuals from west of the trench were not available to test these
hypotheses.
All analyses confirm the presence of three distinct clades within the group. Two
clades composed exclusively of eastern taxa form a grade sister to a clade containing
two western species (C. creutzburgii and C. drabifolia), the widespread C. erinus, and
one eastern species, C. simulans – endemic to western Turkey and adjacent islands.
The placement of C. simulans within the western clade is peculiar as the
divergence of this species occurred after the formation of the Mid-Aegean Trench,
suggesting it is the result of a single dispersal event to the eastern Aegean, across this
barrier (Figure 4-2; Figure 4-3). However, our results suggest this dispersal occurred
during the Messinian Salinity Crisis, when many landmasses in the eastern Aegean
were again connected due to a drastic sea-level drop (Figure 4-3). The Mid-Aegean
Trench may have represented a less significant barrier to dispersal during this time.
96
Cyprus disjunctions
Cyprus has a complex paleogeographic history. Past researchers (e.g., Schmidt,
1960) considered it a continental island, a landmass once connected to the mainland.
However, more recent studies have shown this island to be of oceanic origins, formed at
the junction of the Eurasian and African plates (Gass, 1968; Moores and Vine, 1971)
and one of the most isolated islands (both geologically and biologically) in the
Mediterranean Basin (Moores et al., 1984).
In addition to the widespread C. erinus, three Roucela taxa are found on Cyprus:
C. veneris, C. delicatula, and C. podocarpa (Figure 4-1). C. delicatula and C. podocarpa
show narrow, disjunct distributions between western Turkey and Cyprus. Interestingly,
these taxa are sympatric in western Turkey but have non-overlapping distributions on
Cyprus. Campanula veneris is a narrow Cypriot endemic, found only in the Troodos
Mountains, where it is parapatric with C. podocarpa.
The current, non-overlapping distributions of C. delicatula and C. podocarpa on
Cyprus are likely the result of edaphic differences and the separate geologic histories of
the two landmasses comprising the present day island. The apparent low dispersal
ability of these taxa suggest geodispersal events during a time of low sea-level.
Results from dating analyses estimate the split between Turkish accessions of C.
podocarpa and individuals from Cyprus at approximately 5 Ma, suggesting dispersal (or
range expansion) of this taxon during the MSC, when ‘stepping-stone’ islands may have
connected the two islands to the mainland (Hadjisterkotis et al., 2000). Divergence of
the Cypriot population of C. delicatula from the mainland accessions was estimated to
occur less than one million years ago, suggesting dispersal of this taxon occurred
during Pleistocene glacial cycles.
97
Campanula erinus
The evolutionary history of C. erinus remains enigmatic and phylogenetic
relationships continue to be largely unresolved. Species tree analyses indicate C. erinus
is sister to the Cretan endemic, C. creutzburgii, while analyses of concatenated and
individual gene datasets suggest it is polyphyletic and a close relationship with both C.
creutzburgii and C. drabifolia is inferred (Figure 4-2). Dating analyses indicate relatively
young divergence dates for all the C. erinus individuals included in this study (with the
exception of one Spanish accession), suggesting that the non-monophyly may be the
result of incomplete lineage sorting. However, phylogenetic placement of populations
using plastid markers appears to follow a loose geographic pattern – indicating that it is
possible this taxon represents a cryptic species complex containing multiple lineages of
taxa exhibiting low levels of morphological divergence. Two chromosome numbers have
been reported for C. erinus (2n=28 and 56; Carlström, 1986), suggesting polyploid
populations exist.
Whether the putative polyploids are the result of hybridization or autopolyploidy,
the result of a single or multiple events, and if C. erinus represents multiple,
independent evolutionary lineages that have simply not been recognized by taxonomists
is currently being investigated (Crowl et al., in prep.). Disentangling the historical
processes that have led to the apparent non-monophyly of C. erinus in this study
promises to provide even further insights into the evolutionary processes in the
Mediterranean Basin.
Niche modeling
Our results indicate that, for all taxa tested, the realized distributions of Roucela
species are much narrower than potential niche space (Figure 4-4). The narrow
98
endemism of these taxa in the Eastern Mediterranean, therefore, does not appear to be
the result of climatic constraints.
No discernible adaptations to dispersal have been observed in Roucela species.
Field observations suggest seed dispersal by gravity is likely. This dispersal mechanism
may also be partly responsible for the narrow distribution of these taxa. In other words,
these endemic species have no way to effectively fill their fundamental niche.
Diversification
We reject the hypothesis of a constant rate of species diversification through time
for the Roucela complex. The sliding window analysis indicates an elevated
diversification rate between approximately 8-12 Ma (Figure 4-5). This corresponds to
the break-up of the Aegean landmass, further supporting our hypothesis that allopatric
speciation caused by this event is likely responsible for much of the diversity in the
Roucela clade. The diversification models found to be most significant for this dataset –
the DDL and y2r model – provide two potential explanations regarding the apparent
slow-down in diversification rate that follows.
The DDL model predicts speciation rates to be density dependent and, therefore,
decline (linearly) through time as the number of lineages increase, and thus, ecological
niche space becomes saturated (Rabosky and Lovette, 2008).
The y2r model, on the other hand, suggests that the clade has diversified under a
single rate until, at some point in time, a switch occurs and the clade begins diversifying
at a new rate. The timing of this rate shift was estimated at approximately 8 Ma for the
Roucela clade (Figure 4-5), corresponding to a trend toward increased aridification and
away from a sub-tropical climate in the Mediterranean Basin, providing a likely
explanation for decreased speciation in a clade adapted to sub-tropical conditions. It is
99
important to note, however, that recent extinction cannot be discounted as a possible
cause for the pattern we observe.
Concluding Remarks
Our results provide detailed insights into the evolutionary history of Campanula
species in the eastern Mediterranean. This area has had a complex, but well
understood, geologic and climatic history. By reconstructing evolutionary relationships
and estimating divergence times we are able to test the relative importance of specific
historical events that have contributed to the evolutionary history of the Roucela clade.
We have shown that the evolutionary history and current distributional patterns of these
taxa are the result of both dispersal and vicariance events and have been shaped by
numerous events through the Miocene and onward.
Specifically, our results indicate the break-up of an ancient Aegean landmass to
be responsible for driving diversification of these species. Contrastingly, climatic shifts
beginning approximately 8 Ma, and eventually leading to the current Mediterranean
climate, appear to be responsible for a decrease in diversification rate in the Roucela
clade. Continental island endemics likely originated via vicariance, whereas the oceanic
island endemic on Cyprus appears to be the product of a single dispersal event from the
mainland, followed by in situ diversification. The narrow endemism of Roucela taxa
does not appear to be due to climatic constraints, but is likely linked to the apparent low-
dispersal ability of this group.
By studying the Roucela complex in an evolutionary and biogeographic context,
we highlight the diversity and complexity of historical processes driving plant evolution
in the Mediterranean Basin.
100
Supplemental figures have not been included here. All supplementary material
pertaining to this study can be found in the online version of the manuscript:
http://onlinelibrary.wiley.com/doi/10.1002/ece3.1791/full.
Figure 4-1. Occurrence map for the Roucela complex. Occurrence maps for taxa in the
Roucela complex based on field observations and herbarium collections. Light blue: Campanula drabifolia. Orange: C. creutzburgii. Purple: C. delicatula. Black: C. simulans. Green: C. rhodensis. Pink: C. pinatzii. Dark blue: C. raveyi. Yellow: C. podocarpa. Red: C. kastellorizana. White: C. lycica. ‘X’: C. veneris. Occurrences of the widespread C. erinus not shown. Dashed line indicates approximate location of the Mid-Aegean Trench.
101
Figure 4-2. Results from concatenated and species tree analyses. Maximum
Likelihood tree (left) and species tree from *BEAST analysis (right) for the Roucela clade. Bootstrap support (>50%) and posterior probability values (>0.50) given above branches. Photo of C. podocarpa by Charalambos Christodoulou. Remaining photos by AA Crowl.
102
Figure 4-3. Chronogram of the Roucela clade showing ancestral range estimation.
Summary chronogram from Bayesian dating analysis (BEAST). Outgroups not shown. Monophyletic species have been reduced to a single lineage, C. erinus excluded. Current distribution of each taxon is indicated on the terminals of the tree. Cr: Crete, TuEA: Turkey and East Aegean islands, Gr: mainland Greece, Cy: Cyprus, Ro: Rodos, Ka: Karpathos and Kasos, E Aeg: east Aegean landmass. Internal colored squares indicate most likely ancestral area recovered by BioGeoBEARS under the DEC+J model. Corners represent ranges immediately following a speciation event. Circles with an arrow denote dispersal events while circles with a line denote vicariance. Horizontal grey bars represent 95% HPD confidence intervals. Maps are paleogeographic reconstructions of the Aegean area through time re-drawn from Kasapidis et al., 2005 and Parmakelis et al., 2006 with water depicted as white and land shaded. Areas are colored as coded in the biogeographic analysis, showing connections and isolation through time.
103
Figure 4-4. Niche modeling results for selected taxa in the Roucela complex. Results
from climatic niche modeling analyses for four representative Roucela taxa: Campanula creutzburgii, C. delicatula, C. pinatzii, and C. simulans. Colors represent inferred fundamental climatic niche space for each species. Black dots indicate representative occurrence points to indicate approximate, current distributions. Results suggest realized distributions represent a subset of fundamental climatic niche space for all taxa tested.
104
Figure 4-5. Tempo and pattern of diversification of the Roucela clade. Results from
diversification analyses. A) Log lineage-through-time plot (LTT). LTT plots for the posterior distribution of trees from BEAST analysis (post burn-in) shown in grey. Dotted line indicates hypothetical constant diversification. B) Chronogram for the Roucela clade from BEAST analysis. C. erinus reduced to a single accession to approximate the species tree topology. C) Diversification rate through time using the sliding window approach of Simpson et al. (2011). Diversification rate is calculated as the number of nodes over the sum of all branch lengths within a window of given length.
105
CHAPTER 5 GENE TREE DISCORDANCE PROVIDES EVIDENCE FOR CRYPTIC DIVERSITY
AND INSIGHTS INTO THE EVOLUTION OF A POLYPLOID COMPLEX IN A MEDITERRANEAN CAMPANULA (CAMPANULACEAE) CLADE
Introduction
Cryptic species are those for which a lack of morphological differentiation has
hindered recognition of genetically distinct lineages. Accurate species delimitation is a
difficult but critical aspect of systematics because, although the subject of much debate,
species are regarded as fundamental units of biogeography, ecology, and conservation.
As threats to biodiversity mount, accurate assessments are increasingly important as
inaccurate delimitation of species may hinder biological inferences and conservation
efforts (Wiens, 2007). In this study we focus on a group of endemic Mediterranean
plants, in which putative cryptic diversity has been postulated (Crowl et al., 2015).
The Mediterranean Basin is among the most biologically diverse areas in the
world, harboring innumerable poorly understood, species-rich groups (Myers, 1990;
Médail & Quézel, 1997). Understanding evolutionary processes and species diversity is
of special interest in this region given the exceptionally high degree of endemism and
number of rare and threatened taxa (Greuter, 1991; Médail & Quézel, 1997).
The Roucela clade (Campanulaceae) comprises 12 currently recognized, mostly
narrowly endemic species of bellflowers found primarily in the eastern Mediterranean
Basin (Carlström, 1986; Crowl et al., 2015). These taxa exhibit a high degree of
endemism within this region, often confined to a single or a few islands, with the
exception of Campanula erinus. As currently recognized, this taxon is distributed across
the Mediterranean Basin. Baker’s Law, which states that self-compatible individuals are
more likely to be successful colonizers following a long-distance dispersal event than
106
self-incompatible individuals (Baker 1955), may provide insights into this pattern. The
ability to self-pollinate (Carlström, 1986) likely explains the abnormally broad distribution
pattern of C. erinus within an otherwise highly endemic clade of plants.
The recent phylogenetic analyses of Crowl et al. (2015) recovered strong support
for evolutionary relationships of these taxa with the exception of one clade. These
analyses, based on five plastid and two nuclear loci, failed to disentangle the group
containing the Cretan endemic, C. creutzburgii, C. drabifolia, endemic to the mainland
of Greece, C. simulans from southwestern Turkey, and the most widespread taxon, C.
erinus. Of these, only C. simulans was strongly supported as monophyletic. To increase
phylogenetic resolution within this clade and test hypotheses regarding hybridization
and cryptic speciation, we increased population sampling of each taxon and constructed
a genomic dataset comprising 130 nuclear loci and near-complete plastomes for 105
individuals.
We used these datasets to test hypotheses concerning the nature of previously
unrecognized diversity within the Roucela clade in the Mediterranean Basin, as
suggested by a previous study (Crowl et al., 2015). Specifically, results from numerous
phylogenetic analyses, including concatenation and species-tree approaches, suggest
two cryptic lineages within the currently recognized, widespread species, C. erinus.
These lineages are consistent with both geography and ploidy, with a tetraploid clade
consisting of populations found in the western Mediterranean Basin and an octoploid
lineage restricted to the eastern portion of the basin (Figure 5-1). Network analyses,
corroborated by nuclear gene-tree topologies, indicate a hybrid origin for the octoploid.
107
The narrow Cretan endemic, C. creutzburgii (a tetraploid) and the western
Mediterranean C. erinus (also a tetraploid) are implicated as parental lineages.
The target enrichment approach taken here provided a powerful genomic dataset
to uncover previously overlooked diversity in the eastern Mediterranean Basin. Accurate
species assessments in this hotspot of biodiversity are especially critical in the face of
future climate change, which is projected to increase aridity in the Mediterranean Basin
(e.g., Gao and Giorgi, 2015) and, thus, further impact subtropically adapted taxa such
as the Roucela clade (Crowl et al., 2015).
Methods
Sampling
Taxon sampling included 105 representatives from six species in the Roucela
clade. We sampled 2-5 individuals from 27 populations spanning the distribution of
Campanula erinus from the Azores to Cyprus (Figure 5-1). Four individuals of C.
creutzburgii, four individuals of C. drabifolia, and two individuals of C. simulans were
also included. On the basis of recent phylogenetic results (Crowl et al. (2015), two
additional Roucela species were used as outgroups: C. rhodensis and C. lycica. DNA
was extracted from silica-dried and herbarium material following a modified CTAB
extraction protocol (Doyle and Doyle 1987).
Molecular Data
Previous analyses using plastid data and a small number of nuclear loci indicated
difficulty in resolving phylogenetic relationships of Campanula erinus, C. creutzburgii, C.
drabifolia, and C. simulans (Crowl et al., 2015). We, therefore, obtained a large, multi-
locus nuclear dataset and plastome dataset to disentangle species relationships. This
was achieved using a sequence capture approach (Cronn et al., 2012; Mandel et al.,
108
2014) for the purpose of target enrichment prior to sequencing the reduced-
representation libraries.
MarkerMiner (Chamala et al., 2015) was used to discover and develop probes for
single-copy nuclear loci. We used four assembled Campanulaceae transcriptomes
(Lobelia siphilitica, Platycodon grandiflorus, Campanula delicatula, and Campanula
erinus; available through the 1KP project; onekp.com). This method uses reciprocal
BLAST (Altschul et al., 1997) searches, a database of single-copy nuclear genes (De
Smet et al., 2013), and clustering steps to identify putative orthologous and single-copy
loci across input sequences. Arabidopsis thaliana was used as a reference to estimate
intron/exon boundaries.
Probes were designed for 246 nuclear loci, ranging from 120-3,680 bp in length,
conserved across the Roucela clade. In-solution biotinylated probes were synthesized
using a custom MYbaits target enrichment kit (MYcroarray, Ann Arbor, MI;
http://www.microarray.com). 120mer probes (10,000 total baits) were used with 2x tiling
density. Additionally, we isolated 90 putatively single-copy nuclear genes conserved
across the Campanulaceae clade, useful for future studies in the family. Library building
and capture reactions were carried out by RAPiD Genomics (Gainesville, FL;
http://www.rapid-genomics.com). Samples were sequenced using the Illumina HiSeq
3000 platform (2x100 reads).
Data Processing
Quality filtering of Illumina reads was carried out using cutadapt (Martin, 2011)
and sickle (Joshi, 2011) to remove adapter sequences and trim low-quality nucleotides.
Default parameters were used. Custom python scripts were then used in combination
with those available from the HybPiper pipeline (Johnson et al., 2016). This pipeline
109
uses BWA (Li and Durbin, 2009) to align reads to target sequences and SPAdes
(Bankevich et al., 2012) to assemble these reads into contigs. If multiple contigs that
contained sequences representing at least 75% of the original bait length were found,
these were flagged as potential paralogs and removed from all downstream analyses. A
further filtering step was conducting by manual inspection of gene trees to remove
paralogous loci.
Consensus contigs were aligned to the original probe sequences. The resulting
loci were not trimmed to the original probe length, however, allowing the sequences to
extend into putative intronic regions. After quality filtering and removal of potential
paralogous loci, 130 orthologous loci contained sequences for all 109 sampled taxa (no
missing data).
Plastomes were assembled in a similar way, using Trachelium caeruleum
(Haberle et al., 2008) as a reference. Aligned contigs were trimmed to the plastome
reference length.
Phylogenetic Analysis
Individual gene and plastome alignments were constructed using MAFFT
(v.7.245; Katoh et al., 2002; 2013). Plastomes were considered as a single marker for
all subsequent analyses. We estimated individual nuclear gene trees as well as a
concatenated phylogeny using maximum likelihood (ML) with the program RAxML
(v.7.3.2; Stamatakis, 2006). The ML search was run using 10 distinct starting trees and
1000 bootstrap replicates to measure support. PartitionFinder (v.2.0.0; Lanfear et al.,
2012) was used to infer the optimal partitioning schemes and models of molecular
evolution for the alignments using the rcluster search option.
110
Initial results indicated rogue taxa were present in the nuclear dataset. We used
Rogue NaRok (Aberer et al., 2013) to identify such OTUs. This analysis, optimized for
support using a majority rule consensus threshold, identified four individuals of C. erinus
as rogue samples. These accessions were removed from all datasets and the ML
analyses were re-run as above.
Coalescent Species-Tree Analyses
The relatively young age of the C. erinus complex suggests lineage sorting has
the potential to confound results from concatenation approaches (Crowl et al., 2015).
We, therefore, utilized recently developed coalescent methods to estimate a species
tree for the clade in this study.
ASTRAL-II (v.4.10.0; Mirarab and Warnow, 2015), which estimates the species
tree that maximizes the number of shared quartet trees given a set of gene trees, has
been found to be consistent and accurate in simulations compared to alternative
coalescent approaches (Mirarab et al., 2014). The 130 ML gene trees (best trees)
inferred using RAxML were used as input, and local posterior probabilities were
estimated to provide support for relationships. With respect to individuals being
assigned to species, C. erinus populations were assigned to eastern-Mediterranean
(octoploid) and western-Mediterranean (tetraploid) lineages while C. drabifolia and C.
creutzburgii populations were kept separate, as suggested by the ML analyses.
Additionally, we used SVDquartets (Chifman and Kubatko, 2014) implemented in
PAUP* (v.4.0a147; Swofford, 2002) to verify results generated by ASTRAL-II. A
coalescent approach originally intended for SNP data, SVDquartets has been shown to
perform well on multilocus datasets despite violating the assumption that sites are
independent (Chifman and Kubatko, 2014). We used the concatenated nuclear data
111
matrix as input, evaluated 100,000 random quartets, and assessed support using 100
bootstrap replicates.
Bayesian Concordance Analysis
Though current species-tree methods assume no migration between populations
(Heled and Drummond 2010; Bryant et al. 2012), concordance analyses can still
recover primary phylogenetic signal in the presence of gene flow (Larget et al., 2010).
We, therefore, summarized topological concordance among loci using BUCKy (v.1.4.4;
Baum, 2007; Ane et al., 2007; Larget et al., 2010). Due to computational constraints, it
was necessary to reduce our molecular dataset to the eight major lineages recovered in
previous analyses. To test the impact of the discordance parameter (alpha),
independent analyses were run using alpha=1, alpha=10, and alpha=1000. All analyses
were run with four Markov chain Monte Carlo (MCMC) chains for 1 million generations.
Burn-in was set to 10%.
Network Analysis
We further explored the possibility of hybridization using the program SNAQ
(Solís-Lemus and Ané, 2016). This approach estimates a phylogenetic network under
the coalescent model to account for incomplete lineage sorting, while allowing for
reticulation events within a pseudo-likelihood framework. In order to infer the species
network, we obtained a table of quartet concordance factors from our previous BUCKy
analysis using the bucky.pl script provided in the BUCKy v.1.4.4 package. Tetraploid
and octoploid C. erinus populations were regarded as separate lineages as in all
previous species-tree estimations. We ran two separate analyses, allowing one
(hmax=1) and two (hmax=2) hybridization events. Both were executed with 10
112
independent runs on a starting tree (reduced topology from nuclear concatenated
analysis).
Ploidy Estimation
Chromosome counts were obtained from the literature (Carlström, 1986). To infer
ploidy of the 27 C. erinus populations included in our phylogenetic analyses, we used a
modified flow cytometry method described in Roberts et al. (2009). Approximately 4mg
of silica-dried sample material was combined with 2mg of Pisum sativum as an internal
standard in a 1.5-ml Eppendorf tube containing 2-3 zirconia beads and placed in a bead
mill for 1-3 seconds. Cold lysis buffer (500 ul) was added to the ground material and
filtered through cell culture tubes. RNaseA (1 ul) was then added. Finally, 35 ul of
Propidium Iodide staining solution was added to each tube of suspended nuclei.
Samples were run on a BD Accuri C6 flow cytometer (BD Biosciences, San Jose,
California) at the University of Florida. Two to five individuals per population were used
to confirm ploidy estimates.
Morphology
To determine if there were measurable morphological differences between the
two lineages of C. erinus identified by molecular data, we focused on bract teeth, a
morphological feature found to be taxonomically informative within the Roucela
complex. Carlström (1986) showed that the length of the bract teeth unambiguously
distinguishes C. creutzburgii from C. erinus. This feature has the added advantage of
being well preserved in herbarium specimens, regardless of specimen age or
preservation method, as opposed to floral characteristics, which do not preserve well in
this group. Using ImageJ (v.2.0.0), we measured bract length, bract area, and bract
tooth length for 22 digitized specimens of C. creutzburgii and 252 specimens of C.
113
erinus across their ranges. To account for tooth length differences due to confounding
factors such as age of the plant or bract age on individual plants, we corrected tooth
measurements by dividing these values by the length of the entire bract and the area of
the bract.
Results
Phylogenetic Analyses
Though plastid regions were not targeted in our probe design, we obtained a
significant amount of data from the plastid genome as a byproduct of the Illumina
sequencing run. We recovered between 83.1% and 99.7% (avg. = 95.7%) plastome
coverage for all samples. Maximum likelihood analysis of the plastome dataset found
strong support for two Campanula erinus clades (Figure 5-1). Tetraploid populations
were maximally supported as monophyletic and sister to C. drabifolia. Octoploid
populations were moderately supported (BS=86) as monophyletic and sister to two
individuals of C. creutzburgii from western Crete. The placement of the remaining C.
creutzburgii samples was not supported but inferred to be sister to a monophyletic C.
simulans.
ML analysis of the concatenated nuclear dataset recovered maximal support for
all relationships discussed below. A monophyletic C. creutzburgii was nested within the
octoploid C. erinus populations, rendering the octoploids paraphyletic. A tetraploid clade
was inferred to be sister to this octoploid assemblage. Campanula drabifolia was found
to be non-monophyletic. Campanula simulans was again recovered as monophyletic
and sister to the rest.
Individual nuclear gene trees showed high levels of phylogenetic discordance.
Close inspection of gene trees indicated two major topologies were being recovered.
114
Approximately 30% of nuclear loci indicated tetraploid and octoploid populations of C.
erinus were reciprocally monophyletic (consistent with the ASTRAL-II and SVDquartets
species-tree analyses; see below). The second topology, recovered with approximately
34% of genes, indicated the octoploid lineage was sister to C. creutzburgii. This was
consistent with the plastome dataset and BUCKy analyses (see below). The remainder
(36%) of the gene trees did not fall into these strictly defined categories. However, when
we accounted for low statistical support for relationships of the C. erinus lineages in
these gene trees, we found that nearly all of them approximated one of the two
previously discussed topologies. This more relaxed assignment of gene trees
suggested that nearly half of the sampled genome approximated the plastome-like
topology (close association of the octoploid lineage with C. creutzburgii; 43%), while the
other half (49%) indicated a close association between the octoploid and tetraploid C.
erinus populations.
Species-Tree and Network Analyses
Species-tree analyses of the nuclear dataset recovered relationships consistent
with the individual gene trees but differed from the ML concatenation results. Both
ASTRAL-II and SVDquartets recovered species trees in which tetraploid and octoploid
C. erinus lineages were reciprocally monophyletic when individuals were assigned to
lineages (Figure 5-2, panels A and B). Support for the C. erinus sister relationship,
however, was low (PP=0.79 in ASTRAL-II; BS=72 in SVDquartets). Bayesian
concordance analysis, as implemented in BUCKy, inferred a primary concordance tree
in which the octoploid C. erinus lineage was sister to C. creutzburgii, while the tetraploid
C. erinus lineage was sister to C. drabifolia (Figure 5-2, panel C), consistent with the
115
plastome phylogeny. These relationships had concordance factors of CF=0.342 and
CF=0.329, respectively (see also results from manual inspection of gene trees above).
Interestingly, ASTRAL-II and SVDquartets differed in the reconstruction of the
specie-tree topology when individuals were assigned to populations, rather than
assigning them to the lineages discussed above. When we used the population
assignments, ASTRAL-II estimated a topology similar to the concatenation analyses,
with C. creutzburgii populations nested within octoploid C. erinus populations (Figure 5-
2, panel E). There was, however, very low support for this relationship. The
SVDquartets analysis using population assignments strongly supported a monophyletic
octoploid C. erinus, and C. creutzburgii sister to C. simulans (Figure 5-2, panel D).
SNAQ analyses suggested a single hybridization event (Figure 5-3). The best
network (-loglik = 129.62) included a single hybrid edge between C. creutzburgii and the
tetraploid C. erinus, indicating these as the parental lineages of the octoploid C. erinus.
A paraphyletic C. drabifolia was found to be sister to the tetraploid C. erinus, and a
monophyletic C. creutzburgii sister to C. simulans. These analyses estimated a gamma
value of 0.466 for tetraploid C. erinus and gamma=0.534 for C. creutzburgii parental
lineages. Our manual inspection of gene trees yielded similar results with 49% of gene
trees indicating a close association of the octoploid lineage with the tetraploid lineage
and 43% indicating an association with C. creutzburgii.
Ploidy
Flow cytometry results provided easy-to-interpret ploidy estimates for all
individuals tested. Both tetraploid (2n=28) and octoploid (2n=56) populations were
found within C. erinus. The two ploidal levels correspond to geography, with all
identified tetraploid populations occurring west of Greece and all octoploid populations
116
occurring in the eastern Mediterranean Basin (Figure 5-1). All C. creutzburgii and C.
drabifolia populations were verified as tetraploid. Carlström (1986) cited an earlier
chromosome count for C. creutzburgii as 2n=56, but warned that this record needs
confirmation as this same count had been reported for C. erinus, which is sympatric on
the island of Crete. Our results validate this concern and we propose 2n=28 may be a
more accurate count for C. creutzburgii based on genome size estimations of five
individuals from three populations, though this should be further verified.
Morphology
Though molecular data and genome size estimates suggest multiple lineages
within C. erinus, previous morphological work failed to distinguish these two groups. Our
morphological dataset agrees with this, indicating no measurable difference between
the two lineages based on bract tooth length (Figure 5-4), suggesting these lineages
represent cryptic diversity. Measurements of the bract tooth showed a clear difference
between C. erinus (both tetraploid and octoploid populations) and C. creutzburgii
(Figure 5-4).
Discussion
Phylogenetic analyses, with corroboration from genome-size estimates and
morphologic data, suggest cryptic diversity is present within Campanula erinus, as
currently recognized. Due to the highly similar morphologies of the two lineages
recovered here, this diversity had, until now, been overlooked. Flow cytometry
estimates of genome size found evidence for tetraploid and octoploid populations.
These populations are geographically distinct, with octoploids occurring in the eastern
Mediterranean – Greece, Aegean islands, and Cyprus – while the tetraploids inhabit a
wide area in the western Mediterranean (Figure 5-1). Unfortunately, our population
117
sampling was insufficient to determine the precise geographic location (approximately
mainland Greece or the Balkans) that demarcates the eastern distribution limit of
tetraploid and the western limit of octoploid populations. More in-depth sampling would
provide this boundary and indicate whether or not the tetraploid and octoploid lineages
have overlapping distributions in this region.
Conflicting phylogenetic signal between plastid and nuclear datasets, and within
the nuclear dataset, appears to be the signature of hybridization. The sister relationship
of the octoploid C. erinus lineage with the tetraploid C. erinus and tetraploid C.
creutzburgii, suggests a hybrid origin for this taxon and implicates these tetraploid taxa
as parental lineages. Network analyses, estimated under the coalescent while allowing
for hybridization, confirm this assertion (Figure 5-3). These analyses, with corroboration
from our manual inspection of gene trees, suggest the octoploid lineage of C. erinus
was likely the result of an allopolyploid event (or events) with near-equal contributions
from the two parental lineages.
The non-monophyly of octoploid populations recovered in our phylogenetic
analyses based on the concatenated nuclear dataset (Figure 5-1, panel C) may be
evidence of multiple polyploid origins or simply the result of incomplete lineage sorting.
The results from our species-tree analyses, unfortunately, do not satisfactorily resolve
this issue. Our SVDquartets analysis, in which individuals were assigned to populations,
recovered a maximally supported octoploid C. erinus clade (Figure 5-2, panel D) –
suggesting ILS is the cause for the non-monophyly in the concatenation analyses, while
this same assignment of individuals analyzed with ASTRAL-II (Figure 5-2, panel E)
confirms the non-monophyly of octoploid populations recovered in the concatenation
118
analyses. A more in-depth investigation is necessary to confirm whether a single event
led to the octoploid lineage, or if polyploid formation has been recurring.
Dating analyses of Crowl et al. (2015), in which age estimates for the Roucela
clade were inferred within the broader Campanuloideae, suggest the timing of the C.
erinus–C. creutzburgii hybridization event may have coincided with the Messinian
Salinity Crisis. The closure of the Mediterranean Basin’s connection with the Atlantic
Ocean (5.96-5.33 Ma) led to a significant desiccation of the Mediterranean Sea (Hsu et
al., 1973; Krijgsman et al., 1999). This event led to the connection of previously isolated
islands to each other and the mainland, potentially facilitating range-expansion and
sympatry of C. erinus and C. creutzburgii. Crowl et al. (2015) found that the relatively
recent onset of the Mediterranean climate may have caused extinction in the Roucela
complex. This provides another possible explanation for the formation of a hybrid taxon
from two currently non-sympatric species, as it is conceivable that these taxa had wider
distributions in the past. A more in-depth survey of ploidal levels within populations
found in Crete and mainland Greece would provide further insights into the precise
mechanism underlying this apparent allopolyploid event, or events.
Though past researchers have argued that polyploidy may allow for broad
ecological tolerance and, thus, broad geographic ranges (see Stebbins 1950), the
narrow endemic polyploid taxa in the Roucela clade indicate it clearly is not polyploidy
per se that is responsible for the wide distribution of C. erinus. Our results suggest that
the distribution pattern observed within C. erinus is the result of two factors. First, what
has been historically considered C. erinus is, in fact, composed of two lineages. The
assertion that a single species is distributed across the Mediterranean Basin is,
119
therefore, erroneous. Second, because the octoploid C. erinus appears to have retained
the ability to self pollinate from the parental tetraploid C. erinus, this cytotype was likely
able to rapidly and widely disperse across the eastern part of the basin, in contrast to
the numerous narrowly distributed taxa in this same region – a result that corroborates
Baker’s Law (Baker 1955).
This study provides evidence for cryptic speciation in a small clade of flowering
plants in the Mediterranean Basin. Our results highlight the utility of target enrichment
approaches for obtaining multilocus, genomic datasets for thorough assessments of
species diversity and the need to carefully consider gene-tree discordance within such
datasets. Much work needs to be done in regards to species assessments across the
Tree of Life in this biodiversity hotspot. With the help of genomic data, future studies will
surely uncover further cryptic diversity, as has been done here, providing more accurate
assessments of biodiversity in this fragile region of the world.
120
Figure 5-1. Sample localities, species distributions, and comparison of concatenated
analyses. A) Occurrence map of lineages under investigation. Colors are consistent with those in panels B and C, with tetraploid and octoploid Campanula erinus lineages labeled as such on the map. Circles indicate sampled populations of C. erinus, diamonds indicate sampling localities of C. creutzburgii populations, and squares indicate sampling localities of C. drabifolia populations included in this study. B) Results from maximum likelihood analysis of plastome dataset. Numbers at nodes indicate bootstrap support for relationships and clades discussed in the text. C) Results from maximum likelihood analysis of concatenated nuclear dataset. Numbers at nodes indicate bootstrap support for relationships and clades discussed in the text. Branch colors are consistent with panel B.
121
Figure 5-2. Comparison of species-tree analyses. A) Species-tree topology estimated
with SVDquartets when individuals were assigned to the major lineages recovered with concatenation analyses (see Figure 5-1). Numbers at nodes indicate bootstrap support values of relationships. B) Species-tree topology estimated with ASTRAL-II when individuals were assigned to lineages, as in panel A. C) Primary concordance tree from Bayesian Concordance Analysis in Bucky. Numbers indicate concordance factors and are, thus, estimates of the proportion of the genome for which a relationship is true. Branch lengths are in coalescent units. D) Topology estimated with SVD quartets when individuals were assigned to separate populations, rather than the major lineages as in panel A. Numbers indicate bootstrap support values. E) Topology estimated with ASTRAL-II when individuals were assigned to separate populations. Numbers indicate local posterior probability support values. Branch lengths shown in coalescent units.
122
Figure 5-3. Hypothesized hybridization scenario. A) Phylogenetic network estimated
with SNAQ showing the tetraploid C. erinus and C. creutzburgii as putative parental lineages of the octoploid C. erinus (dashed lines). Numbers on dashed lines indicate inheritance probabilities from SNAQ analysis. These values represent the proportion of the genome estimated to have been contributed by each parental lineage. To simplify the figure, we have collapsed the C. creutzburgii lineages (found to be reciprocally monophyletic) to a single lineage. B) Depiction of the hypothesized hybridization event indicating maternal and paternal genomes involved in the allopolyploid event. Large circles represent nuclear genomes while smaller squares represent plastomes. Red and purple colors indicate parental contributions from each tetraploid.
123
Figure 5-4. Analyses of morphology data. A) Graph of bract tooth length -vs- total bract
length for C. creutzburgii (red), C. erinus 4x (purple), and C. erinus 8x (pink). Solid lines are linear regressions for each species, grey bars indicate 95% confidence intervals. B) Box-plot of bract tooth length corrected for total bract length for the three lineages. Colors consistent with those in (A). Illustrations of the different morphologies shown in upper left corner.
124
CHAPTER 6 PHYLOGENETIC DEFINITION OF THE ROUCELA CLADE
Introduction: Roucela
Several phylogenetic studies (Cellinese et al., 2009; Haberle et al., 2009;
Mansion et al., 2012; Crowl et al., 2015) have consistently recovered a highly supported
clade that includes 12 annual Campanula species (Campanulaceae) found in the
Mediterranean Basin (Carlström, 1986). This clade has previously been referred to as
Campanula subg. Roucela (Dumort.) Damboldt or the Roucela complex, and in-depth
studies of its evolutionary history motivated the establishment of a formal phylogenetic
definition for this group.
Phylogenetic Definition: Roucela
ROUCELA (Dumort.) Damboldt 1976 [A. A. Crowl & N. Cellinese], nomen cladi
conversum.
Node-Based Definition. The least inclusive clade containing Campanula erinus
L. 1753, Campanula rhodensis A. DC. 1830, and Campanula pinatzii Greuter & Phitos
1967.
Etymology: Roucela
The name Roucela has previously been used at the rank of genus (Roucela
Dumort.) and, currently, subgenus (Campanula subg. Roucela [Dumort.] Damboldt).
Though non-monophyletic in Crowl et al. (2015; see below), the type, Campanula erinus
L. 1753 was found to fall within a clade containing other traditionally recognized taxa in
this group. Here, we repurpose Roucela to be used as a clade name approximating
Campanula subg. Roucela (Dumort.) Damboldt.
125
Reference Phylogeny: Roucela
Crowl et al. (2015; Figure 4-2).
Composition: Roucela
In the most recent taxonomic revision, Carlström (1986) recognized 12 species in
Campanula subg. Roucela (Dumort.) Damboldt: Campanula creutzburgii Greuter , C.
delicatula Boiss., C. drabifolia Sm., C. erinus L., C. kastellorizana Carlström, C. pinatzii
Greuter & Phitos, C. podocarpa Boiss., C. raveyi Boiss., C. rhodensis A. DC., C.
scutellata Griseb., C. simulans Carlström, and C. veneris Carlström. More recently, one
additional species, C. lycica Kit Tan & Sorger, was described and added to this group
by Tan and Sorger (1987). According to the phylogenetic analyses of Mansion et al.
(2012) and Crowl et al. (2015), however, C. scutellata Griseb. does not appear to
belong to this clade.
Diagnostic Apomorphies: Roucela
The Roucela clade includes dichotomously branched annual species of
Campanula with an unappendaged calyx (Carlström, 1986, Lammers, 2007).
Synonyms: Roucela
None.
Comments: Roucela
The name Roucela has previously been used at the ranks of genus and
subgenus. The group was initially recognized on the basis of morphology: small,
dichotomously branched annuals with unappendaged calyx lobes (Carlström, 1986;
Lammers, 2007). See Crowl et al. (2015; Figure 4-2) for the primary reference
phylogeny.
126
The distribution of the Roucela clade spans the Mediterranean Basin. As
traditionally recognized, the most widespread taxon, C. erinus, is found from the Azores,
southern Europe, northern Africa, and the Arabian Peninsula, an area broadly
corresponding to the Mediterranean climate zone but extending as far east as Iran.
However, see Crowl et al. (in prep.; Ch. 5) and discussion below for a revised view of C.
erinus. The remaining species occupy more restricted distributions – many, narrow
island endemics in the Aegean Archipelago, western Turkey, and Cyprus.
The phylogenetic analyses of Crowl et al. (2015) included all 13 previously
recognized species in the Roucela complex. Utilizing both plastid and nuclear markers,
this study recovered strong support for the monophyly of the group within the broader
Campanuloideae clade, with the exception of Campanula scutellata. Carlström (1986)
pointed out the morphological divergence of C. scutellata as compared to other species
in the group. Phylogenetic analyses have confirmed this observation and suggest this
species is more closely related to other annual taxa in the Megalocalyx clade (Mansion
et al., 2012; Crowl et al., 2015).
Past studies suggest the Roucela clade as being closely related to a clade
containing Northern African and Western Mediterranean taxa (Cellinese et al., 2009;
Haberle et al., 2009; Crowl et al., 2015), though increased taxon sampling indicates that
this clade may also contain Asian campanuloids (Mansion et al., 2012; Crowl et al.,
2016). A more detailed study with increased molecular and taxonomic sampling is
necessary to test these hypotheses.
Within the Roucela clade, the geologic history of the Eastern Mediterranean
appears to have played an important role in the diversification of many species, while
127
the climatic history – specifically, the shift from a subtropical climate – may have
adversely affected diversification (see Crowl et al., 2015 for an in-depth discussion).
Because Campanula erinus L. 1753 (synonym: Roucela erinus [L.] Dumort.) was
the type for the previously recognized genus and subgenus, we have included two
specimens (representing the two lineages recovered by Crowl et al. [in prep.; Ch. 5];
see below) of this taxon as internal specifiers in our Roucela clade definition.
Introduction: Holoerinus
The phylogenetic analyses of Crowl et al. (2015) recovered the taxon Campanula
erinus L. as non-monophyletic. Statistical support for this result, however, was
insufficient to draw meaningful conclusions. The more recent, phylogenomic analyses of
Crowl et al. (in prep.; Ch. 5), which increased both population and genomic sampling,
verified this assertion and suggested two cryptic clades within this currently recognized
species. Though seemingly indistinguishable on the basis of morphology, populations
belonging to these groups are recognized on the basis of geography and ploidy:
western Mediterranean tetraploids and eastern Mediterranean octoploids.
Phylogenetic Definition: Holoerinus
HOLOERINUS L. [A. A. Crowl & N. Cellinese], nomen cladi novum.
Branch-modified node-based definition. The most inclusive crown clade
containing Crowl #67 [Campanula erinus L.], 2-Jun-2012, Italy: 3 km west of Baia della
Zagare on Strada Provinciale N. 53, dirt road just before entrance to tunnel, FLAS
260389; but not Crowl #7 [Campanula creutzburgii Greuter], 8-May-2011, Western
Crete, Balos Lagoon, near Kissamos, FLAS 240137; or Cellinese #NC2000 [Campanula
drabifolia Sibth. & Sm.], 13-May-2010, Greece: Kefalonia Island, Pyloros, Village
Agonas, along the road, open phrygana.
128
Etymology: Holoerinus
To indicate the inclusion of both tetraploid and octoploid lineages, we combine
the Greek prefix ‘holo’ with the specific epithet of the currently recognized species,
Campanula erinus.
Reference Phylogeny: Holoerinus
Crowl et al. (in prep.; Ch. 5; Figure 5-3).
Introduction: Tetraerinus
Phylogenetic analyses consistently recovered strong support for a clade of
tetraploid Campanula erinus L. populations. This tetraploid clade, found throughout the
western Mediterranean Basin, occurs from the Azores to the Balkans and includes the
type specimen of Campanula erinus. The phylogenetic definition presented here
includes herbarium specimens as specifiers.
Phylogenetic Definition: Tetraerinus
TETRAERINUS L. [A. A. Crowl & N. Cellinese], nomen cladi novum.
Apomorphy-modified node-based definition. The most inclusive crown clade
exhibiting a ploidal level (tetraploid) synapomorphic with that of Crowl #67 [Campanula
erinus L.], 2-Jun-2012, Italy: 3 km west of Baia della Zagare on Strada Provinciale N.
53, dirt road just before entrance to tunnel, FLAS 260389.
Etymology: Tetraerinus
To reflect the tetraploid ploidal level of this clade, we have combined the specific
epithet of Campanula erinus with the Latin prefix, tetra.
Reference Phylogeny: Tetraerinus
Crowl et al. (in prep.; Ch. 5; Figure 5-1). See also Crowl et al. (in prep.; Ch. 5;
Figure 5-2) for results from species tree analyses.
129
Composition: Tetraerinus
The Tetraerinus clade is composed of tetraploid populations of the currently
recognized species, Campanula erinus, occurring in the western Mediterranean Basin
from the Balkans to the Azores.
Diagnostic Apomorphies: Tetraerinus
Tetraploid ploidal level, chromosome count of 2n=28.
Synonyms: Tetraerinus
None.
Comments: Tetraerinus
A genomic dataset consisting of 130 nuclear loci and near-complete plastomes
across 27 populations of Campanula erinus, spanning its distribution, provides strong
evidence for two lineages within this currently recognized taxon (Crowl et al., in prep.;
Ch. 5). Ploidy estimates suggest a strongly supported clade consisting of tetraploid
populations distributed throughout the western Mediterranean Basin.
Introduction: Octoerinus
Phylogenetic Definition: Octoerinus
OCTOERINUS A. A. Crowl & N. Cellinese, nomen cladi novum.
Apomorphy-modified node-based definition. The most inclusive crown clade
exhibiting a ploidal level (octoploid) synapomorphic with that of Crowl #2 [Campanula
erinus L.], 4-May-2011, Greece: southern Crete, small canyon 1km south of Matala,
FLAS 240140.
Etymology: Octoerinus
To reflect the octoploid ploidal level of this clade, we have combined the specific
epithet of Campanula erinus with the Latin prefix, octo.
130
Reference Phylogeny: Octoerinus
Crowl et al. (in prep.; Ch. 5; Figure 5-1). See also Crowl et al. (in prep.; Ch. 5;
Figure 5-2) for results from species-tree analyses.
Composition: Octoerinus
The Octoerinus clade is composed of octoploid individuals of the currently
recognized species, Campanula erinus, occurring in the eastern Mediterranean Basin.
Diagnostic Apomorphies: Octoerinus
Octoploid ploidal level, chromosome count of 2n=56.
Synonyms: Octoerinus
None.
Comments: Octoerinus
A phylogenetic definition for Octoerinus has been constructed to distinguish
morphologically similar octoploid and tetraploid lineages within the currently recognized
species, Campanula erinus. Though these terminal taxa are not ‘independent’ in that
the octoploid appears to be the result of an allopolyploid event in which the tetraploid
Tetraerinus (see above) and tetraploid Campanula creutzburgii are parental lineages,
the phylogenetic analyses of Crowl et al. (in prep.; Ch. 5) showed that populations of the
different ploidal levels form separate clades, consistent with geography.
Octoerinus can be distinguished from Tetraerinus based on ploidal level:
octoploid (2n=56) and tetraploid (2n=28), respectively.
131
LIST OF REFERENCES
Aberer A.J., Krompass D., & Stamatakis A. (2013) Pruning rogue taxa improves phylogenetic accuracy: An efficient algorithm and webservice. Systematic Biology, 62, 162–166.
Akaike H. (1974) A new look at the statistical model identification. IEEE Transactions on Automatic Control, 19, 716–723.
Altschul S.F., Madden T.L., Schäffer A.A., Zhang J., Zhang Z., Miller W., & Lipman D.J. (1997) Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucleic Acids Research, 25, 3389–3402.
Alvarez I.. & Wendel J.F. (2003) Ribosomal ITS sequences and plant phylogenetic inference. Molecular Phylogenetics and Evolution, 29, 417–434.
Ané C., Larget B., Baum D.A., Smith S.D., & Rokas A. (2007) Bayesian estimation of concordance among gene trees. Molecular Biology and Evolution, 24, 412–426.
Antonelli A. (2008) Higher level phylogeny and evolutionary trends in Campanulaceae subfam. Lobelioideae: Molecular signal overshadows morphology. Molecular Phylogenetics and Evolution, 46, 1–18.
Antonelli A. (2009) Have giant lobelias evolved several times independently? Life form shifts and historical biogeography of the cosmopolitan and highly diverse subfamily Lobelioideae (Campanulaceae). BMC Biology, 7, 82–103.
Baker H.G. (1955) Self-Compatibility and Establishment After “Long-Distance” Dispersal. Evolution, 9, 347–349.
Bankevich A., Nurk S., Antipov D., Gurevich A.A., Dvorkin M., Kulikov A.S., Lesin V.M., Nikolenko S.I., Pham S., Prjibelski A.D., Pyshkin A. V, Sirotkin A. V, Vyahhi N., Tesler G., Alekseyev M.A., & Pevzner P.A. (2012) SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. Journal of computational biology : a journal of computational molecular cell biology, 19, 455–77.
Bardon L., Chamagne J., Dexter K.G., Sothers C.A., Prance G.T., & Chave J. (2013) Origin and evolution of Chrysobalanaceae: insights into the evolution of plants in the Neotropics. Botanical Journal of the Linnean Society, 171, 19–37.
Barker M.S., Kane N.C., Matvienko M., Kozik A., Michelmore R.W., Knapp S.J., & Rieseberg L.H. (2008) Multiple paleopolyploidizations during the evolution of the Compositae reveal parallel patterns of duplicate gene retention after millions of years. Molecular biology and evolution, 25, 2445–55.
132
Barker M.S., Dlugosch K.M., Dinh L., Sashikiran Challa R., Kane N.C., King M.G., & Rieseberg L.H. (2010) Evopipes.net: Bioinformatic tools for ecological and evolutionary genomics. Evolutionary Bioinformatics, 2010, 143–149.
Batterman M.R.W. & Lammers T.G. (2004) Branched Foliar Trichomes of Lobelioideae (Campanulaceae) and the Infrageneric Classification of Centropogon. Systematic Botany, 29, 448–458.
Baum D. (2007) Concordance Trees, Concordance Factors, and the Exploration of Reticulate Genealogy. Taxon, 56, 417–426.
Beaulieu J.M., Tank D.C., & Donoghue M.J. (2013) A Southern Hemisphere origin for campanulid angiosperms, with traces of the break-up of Gondwana. BMC Evol Biol, 13, 1–17.
Bell C., Soltis D., & Soltis P. (2010) The Age and Diversification of the Angiosperms Re-revisited. American Journal of Botany, 97, 1296–1303.
Benaglia T., Chauveau D., Hunter D.R., & Young D.S. (2009) mixtools: An R Package for Analyzing Finite. Journal of Statistical Software, 32, 1–29.
Bittkau C. & Comes H.P. (2005) Evolutionary processes in a continental island system: molecular phylogeography of the Aegean Nigella arvensis alliance (Ranunculaceae) inferred from chloroplast DNA. Molecular Ecology, 14, 4065–4083.
Borsch T., Korotkova N., Raus T., Lobin W., & Löhne C. (2009) The petD group II intron as a species level marker: utility for tree inference and species identification in the diverse genus Campanula (Campanulaceae). Willdenowia, 39, 7–33.
Bouchenak-Khelladi Y., Verboom G.A., Savolainen V., & Hodkinson T.R. (2010) Biogeography of the grasses (Poaceae): A phylogenetic approach to reveal evolutionary history in geographical space and geological time. Botanical Journal of the Linnean Society, 162, 543–557.
Bouckaert R., Heled J., Kuhnert D., Vaughan T., Wu C.H., Xie D., Suchard M.A., Rambaut A., & Drummond A.J. (2014) BEAST 2: A Software Platform for Bayesian Evolutionary Analysis. Plos Computational Biology, 10, e1003537.
Bremer K., Backlund A., Sennblad B., Swenson U., Andreasen K., Hjertson M., Lundberg J., Backlund M., & Bremer B. (2001) A phylogenetic analysis of 100+genera and 50+families of euasterids based on morphological and molecular data with notes on possible higher level morphological synapomorphies. Plant Systematics and Evolution, 229, 137–169.
Bremer K. & Gustafsson M.H.G. (1997) East Gondwana ancestry of the sunflower alliance of families. Proceedings of the National Academy of Sciences, 94, 9188–9190.
133
Bryant D., Bouckaert R., Felsenstein J., Rosenberg N.A., & Roychoudhury A. (2012) Inferring species trees directly from biallelic genetic markers: Bypassing gene trees in a full coalescent analysis. Molecular Biology and Evolution, 29, 1917–1932.
Burleigh G.., Hilu K.., & Soltis D. (2009) Inferring phylogenies with incomplete data sets: a 5-gene, 567-taxon analysis of angiosperms. BMC Evolutionary Biology, 9, 1-11.
Burnham K.P.. & Anderson D.R. (2002) Model selection and multimodel inference : a practical information-theoretic approach. Springer, New York.
Carlström A. (1986) A revision of the Campanula drabifolia complex (Campanulaceae). Willdenowia, 15, 375–387.
Cellinese N., Smith S.A., Edwards E.J., Kim S.-T., Haberle R.C., Avramakis M., & Donoghue M.J. (2009) Historical Biogeography of the endemic Campanulaceae of Crete. Journal of Biogeography, 36, 1253–1269.
Chamala S., García N., Godden G.T., Krishnakumar V., Jordon-Thaden I.E., De Smet R., Barbazuk W.B., Soltis D.E., & Soltis P.S. (2015) MarkerMiner 1.0: A new application for phylogenetic marker development using angiosperm transcriptomes. Applications in plant sciences, 3, 1400115.
Chapman M.A., Leebens-Mack J.H., & Burke J.M. (2008) Positive selection and expression divergence following gene duplication in the sunflower CYCLOIDEA gene family. Molecular Biology and Evolution, 25, 1260–1273.
Chifman J. & Kubatko L. (2014) Quartet Inference from SNP Data Under the Coalescent Model. Bioinformatics, 30, 3317–3324.
Cosner M.E., Jansen R.K., & Crawford D.J. (1991) Structural Variation and Evolution of Chloroplast DNA in the Campanulaceae. American Journal of Botany, 78, 92.
Cosner M.E., Jansen R.K., & Lammers T.G. (1994) Phylogenetic relationships in the Campanulales based on rbcL sequences. Plant Systematics and Evolution, 190, 79–95.
Cosner M.E. & Jansen R.K. (1993) Evolution of chloroplast DNA structure in the Campanulaceae. American Journal of Botany, 80, 140.
Cosner M., Raubeson L., & Jansen R. (2004) Chloroplast DNA rearrangements in Campanulaceae: phylogenetic utility of highly rearranged genomes. BMC Evolutionary Biology, 4, 1–17.
Cowling R.M., Rundel P.W., Lamont B.B., Arroyo M.K., & Arianoutsou M. (1996) Plant diversity in Mediterranean-climate regions. Trends in Ecology & Evolution, 11, 362–366.
134
Creutzburg N. (1963) Paleogeographic evolution of Crete from Miocene till our days. Cretan Annals, 15/16, 336–342.
Cronn R., Knaus B.J., Liston A., Maughan P.J., Parks M., Syring J. V, & Udall J. (2012) Targeted enrichment strategies for next-generation plant biology. Am J Bot, 99, 291–311.
Crowl A.A., Mavrodiev E., Mansion G., Haberle R., Pistarino A., Kamari G., Phitos D., Borsch T., & Cellinese N. (2014) Phylogeny of Campanuloideae (Campanulaceae) with Emphasis on the Utility of Nuclear Pentatricopeptide Repeat (PPR) Genes. PLoS ONE, 9, e94199.
Crowl A.A., Visger C.J., Mansion G., Hand R., Wu H.-H., Kamari G., Phitos D., & Cellinese N. (2015) Evolution and biogeography of the endemic Roucela complex (Campanulaceae: Campanula) in the Eastern Mediterranean. Ecology and Evolution, 5, 5329–5343.
Damboldt J. (1976) Materials for a Flora of Turkey Part 32 Campanulaceae. Notes from the Royal Botanic Garden Edinburgh, 35, 39–52.
de Candolle A.P. (1839) Campanulaceae. Prodromus Systematis Naturalis Regni Vegetabilis pp. 784–792. Treuttel & Würtz, Paris.
De Smet R., Adams K.L., Vandepoele K., Van Montagu M.C., Maere S., & Van de Peer Y. (2013) Convergent gene loss following gene and genome duplications creates single-copy families in flowering plants. Proc Natl Acad Sci U S A, 110, 2898–2903.
Denk T., Grimsson F., Zetter R., & Símonarson L.A. (2011) Late Cainozoic Floras of Iceland: 15 Million Years of Vegetation and Climate History in the Northern North Atlantic. Springer,
Dermitzakis M.D. (1990) Paleogeography, geodynamic processes and event stratigraphy during the Late Cenozoic of the Aegean area. International Symposium on Biogeographical Aspects of Insularity pp. 263–287. Accademia Nazionale dei Lincei, Rome.
Doyle J.J. (1992) Gene Trees and Species Trees - Molecular Systematics as One-Character Taxonomy. Systematic Botany, 17, 144–163.
Doyle J.J. & Doyle J.L. (1987) A rapid DNA isolation procedure for small quantities of fresh leaf tissues. Phytochem. Bulletin, 19, 11–15.
Doyle J.J., Doyle J.L., & Harbison C. (2003) Chloroplast-expressed glutamine synthetase in Glycine and related Leguminosae: Phylogeny, gene duplication, and ancient polyploidy. Systematic Botany, 28, 567–577.
135
Drew B.T. & Sytsma K.J. (2013) The South American radiation of Lepechinia (Lamiaceae): phylogenetics, divergence times and evolution of dioecy. Botanical Journal of the Linnean Society, 171, 171–190.
Driskell A.C., Ane C., Burleigh J.G., McMahon M.M., O’Meara B.C., & Sanderson M.J. (2004) Prospects for building the tree of life from large sequence databases. Science, 306, 1172–1174.
Drummond A.J., Suchard M.A., Xie D., & Rambaut A. (2012) Bayesian Phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution, 29, 1969–1973.
Dubey S., Zaitsev M., Cosson J.F., Abdukadier A., & Vogel P. (2006) Pliocene and Pleistocene diversification and multiple refugia in a Eurasian shrew (Crocidura suaveolens group). Molecular Phylogenetics and Evolution, 38, 635–647.
Eddie W.M.M., Shulkina T., Gaskin J., Haberle R.C., & Jansen R.K. (2003) Phylogeny of Campanulaceae s. str. inferred from ITS Sequences of Nuclear Ribosomal DNA. Annals of the Missouri Botanical Garden, 90, 554–575.
Eddie W.M.M. & Cupido C.N. (2014) Hesperocodon, a new generic name for Wahlenbergia hederacea (Campanulaceae): phylogeny and capsule dehiscence. Edinburgh Journal of Botany, 71, 63–74.
Edgar R.C. (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research, 32, 1792–1797.
Erbar C. & Leins P. (1989) On the early floral development and the mechanisms of secondary pollen presentation in Campanula, Jasione, and Lobelia. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie, 111, 29–55.
Erbar C. & Leins P. (1995) Portioned pollen release and the syndromes of secondary pollen presentation in the Campanulales-Asterales complex. Flora, 190, 323–338.
Eric Schranz M., Mohammadin S., & Edger P.P. (2012) Ancient whole genome duplications, novelty and diversification: The WGD Radiation Lag-Time Model. Current Opinion in Plant Biology, 15, 147–153.
Fawcett J.A., Maere S., & Van de Peer Y. (2009) Plants with double genomes might have had a better chance to survive the Cretaceous-Tertiary extinction event. Proceedings of the National Academy of Sciences of the United States of America, 106, 5737–42.
Fedorov A.A. (1957) Campanulaceae. Flora URSS (ed. by B.K. Shishkin), pp. 459–475. Akademia Nauk, Moscow.
136
Feliner G.N. & Rossello J.A. (2007) Better the devil you know? Guidelines for insightful utilization of nrDNA ITS in species-level evolutionary studies in plants. Molecular Phylogenetics and Evolution, 44, 911–919.
Fenaroli F., Pistarino A., Peruzzi L., Cellinese N., & Cellinese N. (2013) Campanula martinii (Campanulaceae), a new species from northern Italy. Phytotaxa, 111, 27.
Fiz-Palacios O., Vargas P., Vila R., Papadopulos A.S.T., & Aldasoro J.J. (2010) The uneven phylogeny and biogeography of Erodium (Geraniaceae): radiations in the Mediterranean and recent recurrent intercontinental colonization. Annals of Botany, 106, 871–884.
Fiz-Palacios O. & Valcárcel V. (2013) From Messinian crisis to Mediterranean climate: A temporal gap of diversification recovered from multiple plant phylogenies. Perspectives in Plant Ecology, Evolution and Systematics, 15, 130–137.
Flagel L.E. & Wendel J.F. (2009) Gene duplication and evolutionary novelty in plants. New Phytologist, 183, 557–564.
Gao X. & Giorgi F. (2008) Increased aridity in the Mediterranean region under greenhouse gas forcing estimated from high resolution simulations with a regional climate model. Global and Planetary Change, 62, 195–209.
Gass I.G. (1968) Is Troodos Massif of Cyprus a Fragment of Mesozoic Ocean Floor. Nature, 220, 39–42.
Gaut B.S. (1998) Molecular clocks and nucleotide substitution rates in higher plants. Evolutionary Biology, Vol 30, 30, 93–120.
Glick L. & Mayrose I. (2014) ChromEvol: Assessing the pattern of chromosome number evolution and the inference of polyploidy along a phylogeny. Molecular Biology and Evolution, 31, 1914–1922.
Greuter W. (1991) Botanical diversity, endemism, rarity, and extinction in the Mediterranean area: an analysis based on the published volumes of the Med-Checklist. Bot. Chronika.
Gustafsson M.H.G. & Bremer K. (1995) Morphology and Phylogenetic Interrelationships of the Asteraceae, Calyceraceae, Campanulaceae, Goodeniaceae, and Related Families (Asterales). American Journal of Botany, 82, 250–265.
Gustafsson M.H.G., Backlund A., & Bremer B. (1996) Phylogeny of the Asterales sensu lato based on rbcL sequences with particular reference to the Goodeniaceae. Plant Systematics and Evolution, 199, 217–242.
Haberle R.C., Dang A., Lee T., Penaflor C., Cortes-Burn H., Oestreich A., Raubenson L., Cellinese N., Edwards E.J., Kim S.-T., Eddie W.M.M., & Jansen R.K. (2009)
137
Taxonomic and biogeographic implications of a phylogenetic analysis of the Campanulaceae based on three chloroplast genes. Taxon, 58, 715–734.
Haberle R. (1998) Phylogenetic Systematics of Pseudonemacladus and the North American Cyphioids (Campanulaceae, Sensu Lato). M.S. Thesis, Northern Arizona University, Flagstaff.,
Haberle R., Fourcade H., Boore J., & Jansen R. (2008) Extensive rearrangements in the chloroplast genome of Trachelium caeruleum are associated with repeats and tRNA genes. Journal of Molecular Evolution, 66, 350–361.
Hadjisterkotis E., Masala B., & Reese D.S. (2000) The origin and extinction of the large endemic Pleistocene mammals of Cyprus. Biogeographia, 21, 593–606.
Harmon L.J., Weir J.T., Brock C.D., Glor R.E., & Challenger W. (2008) GEIGER: investigating evolutionary radiations. Bioinformatics, 24, 129–131.
Heled J. & Drummond A.J. (2010) Bayesian inference of species trees from multilocus data. Mol Biol Evol, 27, 570–580.
Hernandez P.A., Graham C.H., Master L.L., & Albert D.L. (2006) The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography, 29, 773–785.
Hijmans R.J., Cameron S.E., Parra J.L., Jones P.G., & Jarvis A. (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25, 1965–1978.
Howarth D.G. & Donoghue M.J. (2005) DUPLICATIONS IN CYC-like Genes from Dipsacales Correlate with Floral Form. International Journal of Plant Sciences, 166, 357–370.
Howarth D.G. & Donoghue M.J. (2009) Duplications and expression of DIVARICATA-like genes in dipsacales. Molecular Biology and Evolution, 26, 1245–1258.
Howell G.J., Slater A.T., & Knox R.B. (1993) Secondary Pollen Presentation in Angiosperms and Its Biological Significance. Australian Journal of Botany, 41, 417–438.
Hsu K.J., Ryan W.B.E., & Cita M.B. (1973) Late Miocene desiccation of the Mediterranean. Nature, 242, 240–244.
Huelsenbeck J.P. & Ronquist F. (2001) MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics, 17, 754–755.
Huggett R.J. (2004) Fundamentals of Biogeography. Psychology Press,
138
Jiao Y., Leebens-Mack J., Ayyampalayam S., Bowers J.E., McKain M.R., McNeal J., Rolf M., Ruzicka D.R., Wafula E., Wickett N.J., Wu X., Zhang Y., Wang J., Zhang Y., Carpenter E.J., Deyholos M.K., Kutchan T.M., Chanderbali A.S., Soltis P.S., Stevenson D.W., McCombie R., Pires J., Wong G., Soltis D.E., & dePamphilis C.W. (2012) A genome triplication associated with early diversification of the core eudicots. Genome Biology, 13, R3.
Jiao Y., Wickett N.J., Ayyampalayam S., Chanderbali A.S., Landherr L., Ralph P.E., Tomsho L.P., Hu Y., Liang H., Soltis P.S., Soltis D.E., Clifton S.W., Schlarbaum S.E., Schuster S.C., Ma H., Leebens-Mack J., & DePamphilis C.W. (2011) Ancestral polyploidy in seed plants and angiosperms. Nature, 473, 97–100.
Johnson M.G., Gardner E.M., Liu Y., Medina R., Goffinet B., Shaw A.J., Zerega N.J.C., & Wickett N.J. (2016) HybPiper: Extracting Coding Sequence and Introns for Phylogenetics from High-Throughput Sequencing Reads Using Target Enrichment. Applications in Plant Sciences, 4, 1600016.
Joly S. (2012) JML: testing hybridization from species trees. Molecular Ecology Resources, 12, 179–184.
Joshi N. & Fass J. (2011) Sickle: A sliding-window, adaptive, quality-based trimming tool for FastQ files (Version 1.33) [Software]. Available at https://github.com/najoshi/sickle.
Katoh K., Misawa K., Kuma K., & Miyata T. (2002) MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic acids research, 30, 3059–3066.
Katoh K. & Standley D.M. (2013) MAFFT multiple sequence alignment software version 7: Improvements in performance and usability. Molecular Biology and Evolution, 30, 772–780.
Kearse M., Moir R., Wilson A., Stones-Havas S., Cheung M., Sturrock S., Buxton S., Cooper A., Markowitz S., Duran C., Thierer T., Ashton B., Meintjes P., & Drummond A. (2012) Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics, 28, 1647–1649.
Kluge A.G. (1989) A concern for evidence and a phylogenetic hypothesis of relationships among epicrates (Boidae, serpentes). Systematic Biology, 38, 7–25.
Knowles L.L. & Kubatko L.S. (2010) Estimating species trees: an introduction to concepts and models. Estimating species trees: practical and theoretical aspects (ed. by L.L. Knowles and L.S. Kubatko), Wiley-Blackwell.
139
Knox E.B. (2014) The dynamic history of plastid genomes in the Campanulaceae sensu lato is unique among angiosperms. Proceedings of the National Academy of Sciences of the United States of America, 111, 11097–102.
Kornilios P., Poulakakis N., Mylonas M., & Vardinoyannis K. (2009) The phylogeny and biogeography of the genus Zonites Montfort, 1810 (gastropoda: pulmonata): preliminary evidence from mitochondrial data. Journal of Molluscan Studies, 75, 109–117.
Kovanda M. (1977) Polyploidy and Variation in the Campanula rotundifolia Complex. Part II. (Taxonomic) 2. Revision of the Groups Vulgares and Scheuchzerianae in Czechoslovakia and Adjacent Regions.
Krijgsman W., Hilgen F.J., Raffi I., Sierro F.J., & Wilson D.S. (1999) Chronology, causes and progression of the Messinian salinity crisis. Nature, 400, 652–655.
Kring D.A. & Durda D.D. (2002) Trajectories and distribution of material ejected from the Chicxulub impact crater: Implications for postimpact wildfires. Journal of Geophysical Research: Planets, 107, 6–22.
Kruckeberg A.R. & Rabinowitz D. (1985) Biological Aspects of Endemism in Higher Plants. Annual Review of Ecology and Systematics, 16, 447–479.
Kubatko L.S., Carstens B.C., & Knowles L.L. (2009) STEM: species tree estimation using maximum likelihood for gene trees under coalescence. Bioinformatics, 25, 971–973.
Lammers T.G. (1988) Chromosome Numbers and Their Systematic Implications in Hawaiian USA Lobelioideae Campanulaceae. American Journal of Botany, 75, 1130–1134.
Lammers T.G. (2007) World Checklist and Bibliography of Campanulaceae. Royal Botanic Gardens, Kew, UK,
Lammers T.G. (2007) Campanulaceae. The families and genera of vascular plants (ed. by K. Kubitzki, J.W. Kadereit, and C. Jeffrey), pp. 26–56. Springer-Verlag, Berlin, Germany.
Lammers T.G. (1993) Chromosome Numbers of Campanulaceae. III. Review and Integration of Data for Subfamily Lobelioideae. American Journal of Botany, 80, 660–675.
Lammers T.G. (1992) Circumscription and Phylogeny of the Campanulales. Annals of the Missouri Botanical Garden, 79, 388–413.
Lancucka-Srodoniowa M. (1977) New Herbs Described from the Tertiary of Poland. Acta Palaeobatanica, 18, 37–44.
140
Lancucka-Srodoniowa M. (1979) Macroscopic plant remains from the freshwater Miocene of the Nowy Sacz Basin (West Carpathians, Poland). Acta Palaeobotanica, 20, 3–117.
Lanfear R., Calcott B., Ho S.Y.W., & Guindon S. (2012) PartitionFinder: Combined selection of partitioning schemes and substitution models for phylogenetic analyses. Molecular Biology and Evolution, 29, 1695–1701.
Larget B.R., Kotha S.K., Dewey C.N., & Ane C. (2010) BUCKy: Gene tree/species tree reconciliation with Bayesian concordance analysis. Bioinformatics, 26, 2910–2911.
Leache A.D., Harris R.B., Rannala B., & Yang Z. (2014) The influence of gene flow on species tree estimation: a simulation study. Syst Biol, 63, 17–30.
Leins P. & Erbar C. (2003) The Pollen Box in Cyphiaceae (Campanulales). International Journal of Plant Sciences, 164, S321–S328.
Leins P. & Erbar C. (1990) On the mechanisms of secondary pollen presentation in the Campanulales-Asterales-Complex. Botanica Acta, 103, 87–92.
Leins P. & Erbar C. (2006) Secondary pollen presentation syndromes of the Asterales: a phylogenetic perspective. Botanische Jahrbücher, 127, 83–103.
Leins P. & Erbar C. (2005) Floral Morphological Studies in the South African Cyphia stenopetala Diels (Cyphiaceae). International Journal of Plant Sciences, 166, 207–217.
Li H. & Durbin R. (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics (Oxford, England), 25, 1754–60.
Linder H.P. (2014) The evolution of African plant diversity. Frontiers in Ecology and Evolution, 2, 1–14.
Lu-Irving P. & Olmstead R.G. (2013) Investigating the evolution of Lantaneae (Verbenaceae) using multiple loci. Botanical Journal of the Linnean Society, 171, 103–119.
Lundberg J. & Bremer K. (2003) Phylogenetic Study of the Order Asterales Using One Morphological and Three Molecular Data Sets. International Journal of Plant Sciences, 164, 553–578.
Lymberakis P.. & Poulakakis N. (2010) Three Continents Claiming an Archipelago: The Evolution of Aegean’s Herpetofaunal Diversity. Diversity, 2, 233–255.
Maddison W.P. & Knowles L.L. (2006) Inferring phylogeny despite incomplete lineage sorting. Systematic Biology, 55, 21–30.
141
Maddison W.P. (1997) Gene Trees in Species Trees. Systematic Biology, 46, 523–536.
Magallón S., Gómez-Acevedo S., Sánchez-Reyes L.L., & Hernández-Hernández T. (2015) A metacalibrated time-tree documents the early rise of flowering plant phylogenetic diversity. The New phytologist, 207, 437–53.
Mairal M., Pokorny L., Aldasoro J.J., Alarcón M., & Sanmartín I. (2015) Ancient vicariance and climate-driven extinction continental-wide disjunctions in Africa: the case of the Rand Flora genus Canarina (Campanulaceae). Molecular ecology, 24, 1335–54.
Mandel J.R., Dikow R.B., Funk V.A., Masalia R.R., Staton S.E., Kozik A., Michelmore R.W., Rieseberg L.H., & Burke J.M. (2014) A target enrichment method for gathering phylogenetic information from hundreds of loci: An example from the Compositae. Applications in plant sciences, 2, 1300085.
Mansion G., Parolly G., Crowl A., Mavrodiev E., Cellinese N., Oganesian M., Fraunhofer K., Kamari G., Phitos D., Haberle R., Akaydin G., Ikinci N., Raus T., & Borsch T. (2012) How to handle speciose clades? Massive taxon-sampling as a strategy towards illuminating the natural history of the bell flowers (Campanula, Campanuloideae). PLoS ONE, 7, e50076.
Mansion G., Selvi F., Guggisberg A., & Conti E. (2009) Origin of Mediterranean insular endemics in the Boraginales: integrative evidence from molecular dating and ancestral area reconstruction. Journal of Biogeography, 36, 1282–1296.
Mansion G., Rosenbaum G., Schoenenberger N., Bacchetta G., Rossell√≥ J.A., & Conti E. (2008) Phylogenetic Analysis Informed by Geological History Supports Multiple, Sequential Invasions of the Mediterranean Basin by the Angiosperm Family Araceae. Systematic Biology, 57, 269–285.
Martin M. (2011) Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal, 17, 10-12.
Matasci N., Hung L.-H., Yan Z., Carpenter E.J., Wickett N.J., Mirarab S., Nguyen N., Warnow T., Ayyampalayam S., Barker M., Burleigh J.G., Gitzendanner M.A., Wafula E., Der J.P., dePamphilis C.W., Roure B., Philippe H., Ruhfel B.R., Miles N.W., Graham S.W., Mathews S., Surek B., Melkonian M., Soltis D.E., Soltis P.S., Rothfels C., Pokorny L., Shaw J.A., DeGironimo L., Stevenson D.W., Villarreal J.C., Chen T., Kutchan T.M., Rolf M., Baucom R.S., Deyholos M.K., Samudrala R., Tian Z., Wu X., Sun X., Zhang Y., Wang J., Leebens-Mack J., & Wong G.K.-S. (2014) Data access for the 1,000 Plants (1KP) project. GigaScience, 3, 17.
Matzke N.J. (2013) Probabilistic historical biogeography: new models for founder-event speciation, imperfect detection, and fossils allow improved accuracy and model-testing. Frontiers of Biogeography, 5, 242–248.
142
Mayrose I., Zhan S.H., Rothfels C.J., Magnuson-Ford K., Barker M.S., Rieseberg L.H., & Otto S.P. (2012) Recently formed polyploid plants diversify at lower rates. Science, 333, 1257.
McCormack J.E., Heled J., Delaney K.S., Peterson A.T., & Knowles L.L. (2011) Calibrating Divergence Times on Species Trees Versus Gene Trees: Implications for Speciation History of Aphelocoma Jays. Evolution, 65, 184–202.
McCormack J.E., Huang H.T., & Knowles L.L. (2009) Maximum Likelihood Estimates of Species Trees: How Accuracy of Phylogenetic Inference Depends upon the Divergence History and Sampling Design. Systematic Biology, 58, 501–508.
McKenna M.C. (1983) Holarctic Landmass Rearrangement, Cosmic Events, and Cenozoic Terrestrial Organisms. Annals of the Missouri Botanical Garden, 70, 459.
Médail F. & Quézel P. (1997) Hotspots analysis for conservation of plant biodiversity in the Mediterranean basin. Annals of the Missouri Botanical Garden, 84, 112–127.
Mirarab S., Reaz R., Bayzid M.S., Zimmermann T., Swenson M.S., & Warnow T. (2014) ASTRAL: genome-scale coalescent-based species tree estimation. Bioinformatics (Oxford, England), 30, i541-8.
Mirarab S. & Warnow T. (2015) ASTRAL-II: Coalescent-based species tree estimation with many hundreds of taxa and thousands of genes. 31, i44–i52.
Moore W. (1995) Inferring phylogenies from mtdna variation: mitochondrial-gene trees versus nuclear-gene trees. Evolution, 49, 718–726.
Moores E.M., Robinson P.T., Malpas J., & Xenophonotos C. (1984) Model for the Origin of the Troodos Massif, Cyprus, and Other Mideast Ophiolites. Geology, 12, 500–503.
Moores E.M. & Vine F.J. (1971) Troodos Massif, Cyprus and Other Ophiolites as Oceanic Crust - Evaluation and Implications. Philosophical Transactions of the Royal Society of London Series a-Mathematical and Physical Sciences, 268, 443–466.
Morley R.J. (2003) Interplate dispersal paths for megathermal angiosperms. Perspectives in Plant Ecology, Evolution and Systematics, 6, 5–20.
Myers N. (1990) The biodiversity challenge: expanded hot-spots analysis. The Environmentalist, 10, 243–56.
Myers N., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B., & Kent J. (2000) Biodiversity hotspots for conservation priorities. Nature, 403, 853–858.
143
Nemcok M., Pospisil L., Lexa J., & Donelick R.A. (1998) Tertiary subduction and slab break-off model of the Carpathian-Pannonian region. Tectonophysics, 295, 307–340.
Nylander J.A.A., Wilgenbusch J.C., Warren D.L., & Swofford D.L. (2008) AWTY (are we there yet?): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics, 24, 581–583.
Olesen J.M.., Alarcón M.., Ehlers B.K.., Aldasoro J.J.., & Roquet C. (2012) Pollination, biogeography and phylogeny of oceanic island bellflowers (Campanulaceae). Perspectives in Plant Ecology, Evolution and Systematics, 14, 169–182.
Olsen K. & Shaal B. (1999) Evidence on the origin of cassava: Phylogeography of Manihot esculenta. Proc. Natl. Acad. Sci., 96, 5586–5591.
Olson D.M., Dinerstein E., Wikramanayake E.D., Burgess N.D., Powell G.V.N., Underwood E.C., D’amico J. a., Itoua I., Strand H.E., Morrison J.C., Loucks C.J., Allnutt T.F., Ricketts T.H., Kura Y., Lamoreux J.F., Wettengel W.W., Hedao P., & Kassem K.R. (2001) Terrestrial Ecoregions of the World: A New Map of Life on Earth. BioScience, 51, 933.
Oszast J. & Stuchlik L. (1977) The Neogene vegetation of the Podhale (West Carpathians, Poland). Acta Palaeobatanica, 18, 45–86.
Oszczypko N. & Stuchlik L. (1972) Miocen słodkowodny Kotliny Sądeckiej. Wyniki badań geologicznych i palinologicznych (summary: The freshwater Miocene of the Nowy Sącz Basin. Results of geological and palynological investigations). Acta Palaeobatanica, 13, 137–156.
Paradis E., Claude J., & Strimmer K. (2004) APE: Analyses of Phylogenetics and Evolution in R language. Bioinformatics, 20, 289–290.
Parisod C. (2012) Polyploids integrate genomic changes and ecological shifts. 13th Congress of the European Society for Evolutionary Biology, Tuebingen, Germany, August 2011. The New phytologist, 193, 297–300.
Park J.-M., Kovačic S., Liber Z., Eddie W.M.M., & Schneeweiss G.M. (2006) Phylogeny and Biogeography of Isophyllous Species of Campanula (Campanulaceae) in the Mediterranean Area. Systematic Botany, 31, 862–880.
Parmakelis A., Stathi I., Spanos L., Louis C., & Mylonas M. (2006) Phylogeography of Iurus dufoureius (Brulle, 1832) (Scorpiones, Iuridae). Journal of Biogeography, 33, 251–260.
Pearson R.G., Raxworthy C.J., Nakamura M., & Peterson A.T. (2007) Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography, 34, 102–117.
144
Phillips S.J., Anderson R.P., & Schapire R.E. (2006) Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190, 231–259.
Posada D. (2008) jModelTest: Phylogenetic model averaging. Molecular Biology and Evolution, 25, 1253–1256.
Postigo Mijarra J.M., Barrón E., Gómez Manzaneque F., & Morla C. (2009) Floristic changes in the Iberian Peninsula and Balearic Islands (south-west Europe) during the Cenozoic. Journal of Biogeography, 36, 2025–2043.
Poulakakis Kapli, P., Kardamaki, A., Skourtanioti, E., Göcmen, B., Ilgaz, Ç., , N. & Yusuf Kumlutaş Avci, A., Lymberakis, P. Y. (2013) Comparative phylogeography of six herpetofauna species in Cyprus: late Miocene to Pleistocene colonization routes. Biological Journal of the Linnean Society, 108, 619–635.
Poulakakis N., Lymberakis P., Valakos E., Pafilis P., Zouros E., & Mylonas M. (2005) Phylogeography of Balkan wall lizard (Podarcis taurica) and its relatives inferred from mitochondrial DNA sequences. Molecular Ecology, 14, 2433–2443.
Poulakakis N., Pakaki V., Mylonas M., & Lymberakis P. (2008) Molecular phylogeny of the Greek legless skink Ophiomorus punctatissimus (Squamata: Scincidae): The impact of the Mid-Aegean trench in its phylogeography. Molecular Phylogenetics and Evolution, 47, 396–402.
Prebble J.M., Cupido C.N., Meudt H.M., & Garnock-Jones P.J. (2011) First phylogenetic and biogeographical study of the southern bluebells (Wahlenbergia, Campanulaceae). Molecular Phylogenetics and Evolution, 59, 636–648.
Prebble J.M., Meudt H.M., & Garnock-Jones P.J. (2012) An expanded molecular phylogeny of the southern bluebells (Wahlenbergia, Campanulaceae) from Australia and New Zealand. Australian Systematic Botany, 25, 11–30.
Pybus O.G. & Harvey P.H. (2000) Testing macro-evolutionary models using incomplete molecular phylogenies. Proceedings of the Royal Society B-Biological Sciences, 267, 2267–2272.
Rabosky D.L. (2006) LASER: A Maximum Likelihood Toolkit for Detecting Temporal Shifts in Diversification Rates From Molecular Phylogenies. Evolutionary Bioinformatics, 2, 247–250.
Rabosky D.L. & Lovette I.J. (2008) Density-dependent diversification in North American wood warblers. Proceedings. Biological sciences / The Royal Society, 275, 2363–71.
Raven P.H. & Axelrod D.I. (1974) Angiosperm Biogeography and Past Continental Movements. Annals of the Missouri Botanical Garden, 61, 539–673.
145
Roberts A. V, Gladis T., & Brumme H. (2009) DNA amounts of roses (Rosa L.) and their use in attributing ploidy levels. Plant cell reports, 28, 61–71.
Ronquist F. & Huelsenbeck J.P. (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19, 1572–1574.
Roquet C., Sáez L., Aldasoro J.J., Susanna A., Alarcón M.L., & Garcia-Jacas N. (2008) Natural Delineation, Molecular Phylogeny and Floral Evolution in Campanula. Systematic Botany, 33, 203–217.
Roquet C., SanmartÌn I., Garcia-Jacas N., S·ez L., Susanna A., Wikström N., & Aldasoro J.J. (2009) Reconstructing the history of Campanulaceae with a Bayesian approach to molecular dating and dispersal-vicariance analyses. Molecular Phylogenetics and Evolution, 52, 575–587.
Sanderson M.J. (2003) r8s: Inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics, 19, 301–302.
Sanderson M.J. (2002) Estimating Absolute Rates of Molecular Evolution and Divergence Times: A Penalized Likelihood Approach. Mol Biol Evol, 19, 101–109.
Sang T. (2002) Utility of Low-Copy Nuclear Gene Sequences in Plant Phylogenetics. Critical Reviews in Biochemistry and Molecular Biology, 37, 121–147.
Sang T. & Zhong Y. (2000) Testing hybridization hypotheses based on incongruent gene trees. Syst Biol, 49, 422–434.
Sanmartín I. & Ronquist F. (2004) Southern hemisphere biogeography inferred by event-based models: plant versus animal patterns. Systematic biology, 53, 216–43.
Schmidt W.E. (1960) Zur Struktur und Tektonik der Insel Cypern. Geologische Rundschau, 50, 375–395.
Schönswetter P., Tribsch A., Barfuss M., & Niklfeld H. (2002) Several Pleistocene refugia detected in the high alpine plant Phyteuma globulariifolium Sternb. & Hoppe (Campanulaceae) in the European Alps. Molecular Ecology, 11, 2637–2647.
Shackleton N.J. (1987) Oxygen Isotopes, Ice Volume and Sea-Level. Quaternary Science Reviews, 6, 183–190.
Simmons M.P. (2004) Independence of alignment and tree search. Molecular phylogenetics and evolution, 31, 874–9.
146
Simpson C., Kiessling W., Mewis H., Baron-Szabo R.C., & Muller J. (2011) Evolutionary Diversification of Reef Corals: A Comparison of the Molecular and Fossil Records. Evolution, 65, 3274–3284.
Small R. & Wendel J. (2000) Phylogeny, Duplication, and Intraspecific Variation of Adh Sequences in New World Diploid Cottons (Gossypium L., Malvaceae). Molecular Phylogenetics and Evolution, 16, 73–84.
Small R.., Cronn R.., & Wendel J. (2004) Use of nuclear genes for phylogeny reconstruction in plants. Australian Systematic Botany, 17, 145–170.
Smith S.A. & Donoghue M.J. (2010) Combining Historical Biogeography with Niche Modeling in the Caprifolium Clade of Lonicera (Caprifoliaceae, Dipsacales). Systematic Biology, 59, 322–341.
Solís-Lemus C. & Ané C. (2016) Inferring Phylogenetic Networks with Maximum Pseudolikelihood under Incomplete Lineage Sorting. PLOS Genetics, 12, e1005896.
Soltis D.E. & Burleigh J.G. (2009) Surviving the K-T mass extinction: new perspectives of polyploidization in angiosperms. Proceedings of the National Academy of Sciences of the United States of America, 106, 5455–5456.
Stace H.M. & James S.H. (1996) Another perspective on cytoevolution in Lobelioideae (Campanulaceae). American Journal of Botany, 83, 1356–1364.
Stamatakis A. (2006) RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics, 22, 2688–2690.
Stamatakis A. (2014) RAxML version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics, 30, 1312–1313.
Stebbins G.L. (1971) Chromosomal evolution in higher plants. Edward Arnold, London, UK.
Stebbins G.L. (1950) Variation and evolution in plants. Oxford University Press, London, UK.
Steele R.., Guisinger-Bellian M.., Linder C.R.., & Jansen R. (2008) Phylogenetic utility of 141 low-copy nuclear regions in taxa at different taxonomic levels in two distantly related families of rosids. Molecular Phylogenetics and Evolution, 48, 1013–1026.
Strid A. (1996) Phytogeographia Aegaea and the Flora Hellenica Database. Annalen des Naturhistorischen Museums in Wien, 98 B, Supplement, 279–289.
147
Swofford D.L. (2002) PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, Massachusetts.
Takhtajan A. (1987) Systema Magnoliophytorum. Editoria Nauka, Leningrad.
Tan K. & Sorger F. (1986) Even more new taxa from South and East Anatolia II*. Plant Systematics and Evolution, 155, 93–103.
Tank D. & Donoghue M.J. (2010) Phylogeny and Phylogenetic Nomenclature of the Campanulidae based on an Expanded Sample of Genes and Taxa. Systematic Botany, 35, 425–441.
Tank D.C., Eastman J.M., Pennell M.W., Soltis P.S., Soltis D.E., Hinchliff C.E., Brown J.W., Sessa E.B., & Harmon L.J. (2015) Nested radiations and the pulse of angiosperm diversification: increased diversification rates often follow whole genome duplications. The New phytologist, 207, 454–67.
te Beest M., Le Roux J.J., Richardson D.M., Brysting A.K., Suda J., Kubesová M., & Pysek P. (2012) The more the better? The role of polyploidy in facilitating plant invasions. Annals of botany, 109, 19–45.
Thompson J.D., Lavergne S., Affre L., Gaudeul M., & Debussche M. (2005) Ecological Differentiation of Mediterranean Endemic Plants. Taxon, 54, 967–976.
Tribsch A. & Schönswetter P. (2003) Patterns of Endemism and Comparative Phylogeography Confirm Palaeoenvironmental Evidence for Pleistocene Refugia in the Eastern Alps. Taxon, 52, 477–497.
Vanzela A.L.L., Cuadrado A., Vieira A.O.S., & Jouve N. (1999) Genome characterization and relationships between two species of the genus Lobelia (Campanulaceae) determined by repeated DNA sequences. Plant Systematics and Evolution, 214, 211–218.
Warren D.L., Glor R.E., & Turelli M. (2010) ENMTools: a toolbox for comparative studies of environmental niche models. Ecography, 33, 607–611.
Wendel J.F. & Doyle J.J. (1998) Phylogenetic incongruence: window into genome history and molecular evolution. Molecular systematics of plants II: DNA sequencing (ed. by D.E. Soltis, P.S. Soltis, and J.J. Doyle), pp. 265–296. Springer,
Wendling B.., Galbreath K.., & DeChaine E. (2011) Resolving the Evolutionary History of Campanula (Campanulaceae) in Western North America. PLoS ONE, 6, e23559.
Wiens J.J. (2007) Species delimitation: new approaches for discovering diversity. Syst Biol, 56, 875–878.
148
Wikström N., Savolainen V., & Chase M.W. (2003) Angiosperm divergence times: congruence and incongruence between fossils and sequence divergence estimates. Telling the evolutionary time: molecular clocks and the fossil record (ed. by P.C.J. Donoghue and M.P. Smith), pp. 142–165. Taylor & Francis, London.
Yang Z. (1997) PAML: a program package for phylogenetic analysis by maximum likelihood. Computer applications in the biosciences: CABIOS, 13, 555–6.
Yerkes R.W. (2012) Environmental Context of the Athienou Archaeological Project. Chapter 2 in “Crossroads and Boundaries: The Archaeology of Past and Present in the Malloura Valley, Cyprus,” edited by M.K. Toumazou, P.N. Kardulias, and D.B. Counts. Boston: American Schools. 13–24.
Yesson C. & Culham A. (2006) Phyloclimatic Modeling: Combining Phylogenetics and Bioclimatic Modeling. Systematic Biology, 55, 785–802.
Yuan Y.-W., Marx; H.E., & Olmstead R.G. (2010) An empirical demonstration of using pentatricopeptide repeat (PPR) genes as plant phylogenetic tools: Phylogeny of Verbenaceae and the Verbena complex. Molecular Phylogenetics and Evolution, 54, 23–35.
Yuan Y.-W. & Olmstead R.G. (2009) The pentatricopeptide repeat (PPR) gene family, a tremendous resource for plant phylogenetic studies. New Phytologist, 182, 272–283.
Zachos J., Pagani M., Sloan L., Thomas E., & Billups K. (2001) Trends, rhythms, and aberrations in global climate 65 Ma to present. Science, 292, 686–693.
149
BIOGRAPHICAL SKETCH
Andrew Crowl’s Ph.D. studies focused on the evolution and biogeography of the
Campanulaceae. He is particularly interested in understanding evolutionary patterns
and processes in the Mediterranean Basin and applying genomic data to address
questions related to biogeography, cryptic speciation, and endemism in this biodiversity
hotspot. Andrew received his doctoral degree from the Department of Biology at the
University of Florida in the fall of 2016.