Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/224630738

Pollination,biogeographyandphylogenyofoceanicIslandbellflowers(Campanulaceae)

ARTICLEinPERSPECTIVESINPLANTECOLOGYEVOLUTIONANDSYSTEMATICS·JANUARY2012

ImpactFactor:3.32·DOI:10.1016/j.ppees.2012.01.003

CITATIONS

11

DOWNLOADS

26

VIEWS

244

5AUTHORS,INCLUDING:

JensMOlesen

AarhusUniversity

128PUBLICATIONS5,452CITATIONS

SEEPROFILE

MarisaAlarcón

SpanishNationalResearchCouncil

28PUBLICATIONS155CITATIONS

SEEPROFILE

JuanJoséAldasoro

69PUBLICATIONS495CITATIONS

SEEPROFILE

CristinaRoquet

FrenchNationalCentreforScientificResea…

28PUBLICATIONS341CITATIONS

SEEPROFILE

Availablefrom:JensMOlesen

Retrievedon:20June2015

R

P(

Ja

b

c

d

e

a

ARR1A

KABILPMMM

I

ctyhioefiem2(

1d

Perspectives in Plant Ecology, Evolution and Systematics 14 (2012) 169– 182

Contents lists available at SciVerse ScienceDirect

Perspectives in Plant Ecology, Evolution and Systematics

j o ur nal homepage: www.elsev ier .de /ppees

esearch article

ollination, biogeography and phylogeny of oceanic island bellflowersCampanulaceae)

ens M. Olesena,∗, Marisa Alarcónd, Bodil K. Ehlersc, Juan José Aldasorob, Cristina Roquetd,e

Department of Bioscience, Aarhus University, Ny Munkegade 114, DK-8000 Aarhus, DenmarkReal Jardín Botánico de Madrid, Consejo Superior de Investigaciones Científicas (CSIC), Plaza de Murillo, 2, E-28014 Madrid, SpainInstitute of Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense, DenmarkInstitut Botànic de Barcelona (CSIC-ICUB), Passeig del Migdia s. n., Parc de Montjuïc, E-08038 Barcelona, SpainUniversité Grenoble 1, CNRS, Laboratoire d’Écologie Alpine, UMR 5553, BP 53 F-38041 Grenoble Cedex 9, France

r t i c l e i n f o

rticle history:eceived 18 June 2011eceived in revised form6 November 2011ccepted 16 January 2012

eywords:zoresird

sland biogeographyizard

a b s t r a c t

We studied the pollination biology of nine island Campanulaceae species: Azorina vidalii, Musschia aurea,M. wollastonii, Canarina canariensis, Campanula jacobaea, Nesocodon mauritianus, and three species ofHeterochaenia. In addition, we compared C. canariensis to its two African mainland relatives C. eminii andC. abyssinica. We asked to what extent related species converge in their floral biology and pollinationin related habitats, i.e. oceanic islands. Study islands were the Azores, Madeira, Canary Islands, CapeVerde, Mauritius, and Réunion. Information about phylogenetic relationships of these species and theirrelatives were gathered from atpB, matK, rbcL and trnL-F regions, building the most complete phylogenyof Campanulaceae to date. Six of the island bellflower species were bird-pollinated and two (A. vidaliiand M. aurea) were lizard-pollinated. Insects also visited some of the species, and at least C. jacobaea hadboth insect- and self-pollination. Several morphological traits were interpreted as adaptations to bird and

ollinationacaronesiaascarenesolecular dating

lizard pollination, e.g. all had a robust flower morphology and, in addition, bird-pollinated species werescentless, whereas lizard-pollinated species had a weak scent. These examples of vertebrate pollinationevolved independently on each island or archipelago. We discuss if these pollination systems have anisland or mainland origin and when they may have evolved, and finally, we attempt to reconstruct thepollinator-interaction history of each species.

ftcCRp(

oeoOe

ntroduction

Oceanic islands offer unique opportunities for studying the pro-esses of species dispersal, establishment, and speciation, becauseheir ecosystems often are biologically simple and geologicallyoung and consequently, they may be relatively easy to compre-end (Kim et al., 2008). Long-distance dispersal events to oceanic

slands usually only encompass a single species and not fragmentsf food webs (however, see Holt, 2010). Consequently, a speciesstablishing on an oceanic island, experiences ecological releaserom its antagonists in its source continent (Janzen, 1985). On thesland, it meets a community of native species to which it will bevolutionarily naïve and consequently, its potential interactions

ay be constrained (e.g. Aizen et al., 2008; Emerson and Gillespie,

008). This different continental past or “ghost of linkage past”inspired by Connell, 1980) has to be taken into account in order

∗ Corresponding author. Tel.: +45 30126090.E-mail address: jens.olesen@biology.au.dk (J.M. Olesen).

adif2et

433-8319/$ – see front matter © 2012 Elsevier GmbH. All rights reserved.oi:10.1016/j.ppees.2012.01.003

© 2012 Elsevier GmbH. All rights reserved.

ully to understand the biology and, particularly, the biotic interac-ion context of an island species. However in spite of this differentontinental past, island lineages often converge in their evolution.onsequently, island selection regimes have to be very similar (e.g.icklefs and Schluter, 1993; Losos and Ricklefs, 2009). An exam-le of island convergence is flightlessness in different bird lineagesJohnson and Stattersfield, 2008).

The well-studied bellflower family Campanulaceae has turnedut to offer excellent opportunities in the study of this ecological-volutionary compromise between the biotic past and presencef island species and their mainland ancestors (e.g. Olesen, 1985;lesen and Valido, 2003a; Valido et al., 2004). We examined thextent of convergent or parallel evolution in pollination biologymong related island bellflower species, all being limited in theiristribution to tropical-subtropical oceanic islands. Recent stud-

es have clarified the main phylogenetic relationships within the

amily (Eddie et al., 2003; Roquet et al., 2008, 2009; Haberle et al.,009), and thus also its overall biogeographical history (Cellineset al., 2009; Roquet et al., 2009), but these studies have also inves-igated the importance of pollinators to the evolution of the flower

1 gy, Ev

mspeeblptb

M

S

Et(esmW(irnrsyoesteasw

bcfep(ppRTt

wcbaeclRDarlr2

S

toInisdTi(leMnaA

F

icseamspladvo(scAI

P

r(oatwa

dV

PT“1b

70 J.M. Olesen et al. / Perspectives in Plant Ecolo

orphology within the family (Roquet et al., 2008). Here, wetudied the pollination biology of nine island endemics of Cam-anulaceae and placed our results within a phylogenetic contextncompassing the entire family. Specifically, we asked (1) to whatxtent related island bellflower species converge in their floraliology and pollination in related habitat, i.e. if specific island pol-

ination morphs exist? And if so, (2) how can information abouthylogeny and pollinator interactions, including both the past andhe present pollinator fauna, explain the evolution of these islandellflowers?

aterials and methods

tudy family

The Campanulaceae sensu stricto is a cosmopolitan (but mainlyurasian) family with 600–950 species in 35–55 genera, sorted intohree tribes: Campanuleae, Wahlenbergieae, and PlatycodoneaeEddie et al., 2003; Cosner et al., 2004; Roquet et al., 2008; Haberlet al., 2009). Platycodoneae and Wahlenbergieae s. str. clades areuccessive sisters to core Campanuleae (Fig. 1). Campanuleae s. str.ainly includes genera from the Northern Hemisphere, whereasahlenbergieae s. str. contains only Southern Hemisphere taxa

Cosner et al., 2004). The evolutionary history of Campanulaceaencludes long-distance dispersal between continents and bursts ofapid diversification. The ancestor of the family may have origi-ated in Asia, and later colonised Africa and the Mediterraneanegion (Haberle et al., 2009; Roquet et al., 2009). The initial diver-ification within the Platycodoneae is dated to 15–25 mya (millionears ago), that of Wahlenbergieae s. str. to 15–20 mya, and thatf the two main clades of Campanuleae to 10–15 mya (Roquett al., 2009). Cooler and drier climate in Late Neogene and expan-ion of open steppe habitats in SW Asia and E Africa may haveriggered these diversifications (Axelrod and Raven, 1972; Rohlingt al., 1998; Fernandes et al., 2006). Tertiary climate change mustlso have affected the pollinators of the family because pollinatorseem to have exerted a strong selective pressure on the floral traits,hich are very diverse in the family (Roquet et al., 2008).

Within Campanuleae, open, rotate flowers are visited mainlyy a diverse fauna of Syrphidae, Muscidae, small bees and Xylo-opa, whereas more campanulate flowers are visited by a restrictedauna of Bombus species and other large bees (Skov, 1999; Roquett al., 2008). However, the Campanulaceae also includes speciesollinated by vertebrates, such as birds (Olesen, 1985) and lizardsElvers, 1978; Hansen et al., 2006). Some taxa are selfers (e.g. Cam-anula uniflora, Ægisdóttir and Thórhallsdóttir, 2006; J.M. Olesen,ers. obs.; C. dichotoma, Nyman, 1992; C. propinqua and C. fastigiata,oquet et al., 2008; C. americana, Galloway et al., 2003; the genusriodanis, Trent, 1940; McVaugh, 1948; and island populations ofhe generally sexual C. microdonta, e.g. Inoue et al., 1996).

Insect-pollinated flowers are mainly blue-violet (can be alsohite or yellow), protandrous, scented, with small amounts of con-

entrated nectar, and have a “delicate” morphology. Among them,ee-pollinated flowers usually have a campanulate corolla, bee-fly-nd butterfly-pollinated flowers are tubular, while flowers of self-rs often are small and less conspicuous (e.g. Proctor et al., 1996). Inontrast, bird-pollinated flowers are red, orange or yellow, scent-ess, with plenty of dilute nectar, and a robust morphology (e.g.odríguez-Gironés and Santamaria, 2004; Cronk and Ojeda, 2008;alsgaard et al., 2008; Rodríguez-Rodríguez and Valido, 2008). In

few cases, bird-pollination in Campanulaceae is expected to be

elict. During glaciations and dry periods of the Neogene, main-and species may have partially disappeared, but survived in mesicefuges such as oceanic islands or mountaintops (Valido et al.,004).

Twc2

olution and Systematics 14 (2012) 169– 182

tudy species and sites

Eleven Campanulaceae species were studied (Table 1). In addi-ion, the floral biology of several greenhouse specimens of variousrigins was studied. Another bellflower described from Cape Verdeslands is C. bravensis, but the reported corolla colour and shape areot enough to discriminate it from C. jacobaea, which is included

n our sample (Roquet et al., unpubl.). Recently, a third Musschiapecies, M. isambertoi M. Seq., R. Jardim, M. Silva & L. Carvalho, wasescribed from the Desertas (Menezes de Sequeira et al., 2007).he differential characters of this species are: monocarpy (whichs not the case for M. aurea), an unbranched inflorescence 1.5 m tall0.4 m in M. aurea, 1–2 m in M. wollastonii), and anthers 2–3 timesarger than those found in the other two Musschia species. How-ver, sequences of trnL-F show no nucleotide difference between. isambertoi and M. aurea (Roquet et al., unpubl.). Only one

on-flowering individual/juvenile of H. borbonica was found atn altitude of 1730–1740 m on the trail from Hell-Bourg to Capnglais, Réunion. The plant was c. 0.4 m tall.

lower morphology and pollinator observations

Flower measurements were made with a caliper. In early morn-ng hours, volume of nectar per flower was measured as standingrop using micropipettes, and nectar sugar concentration was mea-ured with a pocket refractometer and expressed as w/w% sucrosequivalents. Presence or absence of floral scent was scored from

sample of flowers stored for 24 h in a plastic bag. This crudeethod only allowed us to detect volatiles by the human olfactory

ense. The site of secondary pollen presentation (typical to Cam-anulaceae; Erbar and Leins, 1989; Yeo, 1993) was identified and

evel of dichogamy, i.e. temporal order of maturation of anthersnd stigma was scored by examining buds and flowers of differentevelopmental stages. Compiled on field trips over 7 years, obser-ations of pollinators were made in all populations. The amountf time spent per population varied from 12 (C. eminii) to 40 hN. mauritianus and C. canariensis, population on the Anaga penin-ula, Tenerife). Flower visitors are here called pollinators becauseontact with pollen brush and stigma in most cases was observed.ll observations of pollinators are original unless otherwise stated.

nsects were collected and identified later.

hylogenetic inference and dating analyses

We retrieved sequences of four cpDNA regions (atpB, matK,bcL, trnL-F) for all the Campanulaceae taxa available in Genbankelectronic supplement material ESM 1 – Tables 1 and 2). Sequencesf Abrophyllum ornans (F. Muell.) Hook. f., Cuttsia viburnea F. Muell.nd Lobelia cardinalis L. were added as outgroups. We sequencedhe trnL-F region for 14 taxa in order to complete the samplingith the oceanic bellflowers studied here. The new sequences were

dded to Genbank (ESM 1 – Table 2).Total DNA was extracted from herbarium material or silica gel-

ried plant tissue using the “DNeasy Plant Mini Kit” (Qiagen Inc.,alencia, CA), according to the manufacturer’s instructions.

PCR amplifications were performed with the thermocyclerTC-100TM Programmable Thermal Controller (MJ Research Inc.).he trnL-F region was amplified using the external primersc” and “f” and internal primers “d” and “e” (Taberlet et al.,991), amplifying the trnL (UAA) intron and the intergenic spaceretween the trnL (UAA) 3′ exon and the trnF (GAA) 5′ exon.

he PCR profile consisted of 1 min 35 s at 95 ◦C; 5 min at 80 ◦C,hile DNA-polymerase (Ecotaq, Ecogen S. R. L.) was added; 34

ycles of 1 min denaturing at 93 ◦C, 1 min annealing at 50 ◦C, min extension at 72 ◦C; and a final extension of 10 min at

J.M. Olesen et al. / Perspectives in Plant Ecology, Evolution and Systematics 14 (2012) 169– 182 171

Table 1Study species and sites.

Species Range Habitat No. study pops Flower biology studysites

1 Azorina vidalii (Watts.)Feer

All Azores Isls exceptGraciosa Isl.

Coastal cliff 6 3 islands: Santa Maria,Terceira, Flores

2 Campanula jacobaea C.E. Sm. ex Hook.

5 of the 6 largest CapeVerde Isls

Humid slope 2 Tarrafal, San Nicolauand Cova, Santo Antao(Cape Verde)

3 Musschia wollastoniiLowe

Madeira Isls Madeira Isl.: laurelforest 700–1000 m;Desertas Isls: rocky cliff

1 Ribeiro Frio (MadeiraIsl.)

4 Musschia aurea (L. f.)Dumort.

Madeira Isls Madeira Isl.: S and Ncoasts rocky cliff<350 m, Curral dasFreiras 800–900 m;Desertas Isls: rocky cliff

2 S coast at Sao Gonc alo(Madeira Isl.), 2populations each ofseveral hundredindividuals

5 Canarina canariensis(L.) Vatke

W Canary Isls Laurel forest, forestedge, Erica shrub,shady ravine300–1100 m

4 Tenerife (Anaga, Bco delas Moradas), La Palma(Los Tilos)

6 Canarina eminiiAschers. ex Schweinf.

Ethiopia, Sudan,Uganda, W Kenya,Tanzania, Zaire,Rwanda, Burundi

Forest 1500–3200 m 1 Mt. Elgon, E Uganda

7 Canarina abyssinicaEngl.

Ethiopia, Sudan, EUganda, W Kenya, NTanzania

Grassland, rocks, forestgap, riverine1350–2500 m

1 No obs.

8 Heterochaenia ensifolia(Lam.) DC

Réunion Isl. Forest 2000–2250 m 1 N Cirque de Cilaos

9 Heterochaenia rivalsiiBadré & Cadet

Réunion Isl. Shrubland, montaneplateau c. 2400 m

1 Plaine des Cafres

10 Heterochaeniaborbonica Badré &Cadet

Réunion Isl. Forest c. 1700 m 1 juvenile No obs.

11 Nesocodon mauritianus(Richardson) Thulin

Mauritius Isl. Humid cliff c. 400 m 1 Cascade 500 Pieds

Study period Life form Reproductiveepisodes

No. flowers/plant Flower colour andorientation

Flowerlength × width(mm)

Floral scent

1 July–August Shrub Polycarpic 1–50 infl. à 5–40flowers, plant mean is250 flowers

White-purple, pendant 23–25 × 10–12 Faint

2 April–May Terrestrial perennialherb

Polycarpic 1–5 infl. à 1–5 flowers.Plant mean is 15flowers

Blue or white,horizontal

20–31 × 17–21 None

3 August–September Shrub Monocarpic 1 infl. with 100–1000flowers

Yellow or purple,upright

45–55 × 65–90 None

4 August Shrub Polycarpic 1 infl. with <100flowers

Yellow, upright 25 × 28 Sweet

5 November–May Terrestrial perennialherb

Polycarpic Solitary flower Red-orange, pendant 40–70 × 40–70 None

6 August Epiphytic herb Polycarpic Solitary flower Red-orange, pendant 50–83 × 25 None7 No obs. Terrestrial perennial

herbPolycarpic Solitary flower Red-orange, pendant 40–75 × 30–50 None

8 January–February Shrub Polycarpic Several infl. with atotal of <500 flowers

Yellow-purple,pendant

15 × 15 None

9 January–February Perennial herb Monocarpic? 1 infl. with 35–400flowers

Yellow or purple,horizontal

20 × 30 None

10 No obs. Shrub No obs. 1 infl. à c. 20 flowers Blue, pendant No obs. No obs.11 October–November Shrub Polycarpic 1–10 solitary flowers Blue, pendant 50 × 30 None

Nectar: �m and %w/w sucrose Secondary pollen presentation site(dichogamy: s: strong, w: weak)

Pollinator Reference

1 9 �m 22% Style (s) Lizards, insects, wind This study2 No obs. Style (s) Selfing and insects This study3 95 �l 13% Below stigma (w) Birds, insects This study, Olesen and Valido (2003a)4 24 �l 12% Below stigma (w) Lizards, insects This study, Elvers (1978), Olesen and Valido (2003b)5 10 �l 20% Style (s) Birds This study, Ceballos and Ortuno (1951), Hedberg et al.

(1961), Olesen (1985), Hedberg (1988)6 No obs. Style (s) Birds, insects This study, Rodríguez-Rodríguez and Valido (2011)7 No obs. Style (s) Birds This study8 42 �l 8% Style & part below stigma (w) Birds This study, Badré (1976)9 9 �l 9% Style & part below stigma (w) Birds This study, Badré (1976)

10 No obs. no obs. No obs. Friedmann (1998)11 25 �l 15% Style (s) Birds This study, Richardson (1979), Hansen et al. (2006)

Isl. and Isls, island and islands.

172 J.M. Olesen et al. / Perspectives in Plant Ecology, Evolution and Systematics 14 (2012) 169– 182

Fig. 1. Phylogenetic tree of the Campanulaceae using four markers: atpB, matK, rbcL and trnL-F. The tree was constructed using maximum likelihood, bootstrap valuesare above the branches and posterior probabilities below. Trait state reconstruction of the pollination type of the bellflowers was carried out using likelihood ancestralr stated entireg nds is

7BswLm

sa

tdvcw

econstructions (Mesquite). The proportional likelihoods of the different characteriagram (grey for bird/lizard pollination and black for insect pollination). If a pie isrey spot at the branch tip indicates vertebrate pollination. Presence on oceanic isla

2 ◦C. The PCR-Beads kit (“puRetaq Ready-To-Go”, Amershamiosciences Inc., Piscataway, NJ) was used for the poorly pre-erved DNA of herbarium specimens. These amplified productsere purified using PCR Clean-up kit spin filter columns (MoBio

aboratories, Carlsbad, CA) following the manufacturer’s recom-

endations.PCR products were cleaned using the “QIAQuick® DNA cleanup

ystem” (Qiagen Inc.) according to manufacturer’s instructionsnd sequenced with the trnL-F c and trnL-F f primers for the

aveA

s in the ancestral reconstructions are indicated by the area grey/black in each piely black for a group, younger pies are not shown in order to simplify the figure. A

given by an asterisk after the species name.

rnL-F region. DNA sequencing of PCR-purified templates wasone using reactions based on chemistry of “Big Dye® Terminator

3.1” (PE Biosystems, Foster City, CA) following the proto-ol recommended by the manufacturer. The products obtainedere analysed on an ABI Prism® 3730 PE Biosystems/Hitachi

utomated sequencer in the Serveis Cientificotècnics de la Uni-ersitat de Barcelona, and the resulting chromatograms weredited with Chromas 2.0 (Technelysium Pty Ltd., Tewantin,ustralia).

gy, Ev

erawpttemBere1dwwm

itTRvousb(sfre1mCcC1toafac

T

ltMbRdtp

R

P

o

CeopC(ui2iCtagwlfit

D

F

J.M. Olesen et al. / Perspectives in Plant Ecolo

For each region, sequences were aligned with MAFFT (Katoht al., 2005) and checked by eye with Bioedit (Hall, 1999). The fouregions were concatenated in a supermatrix with FASconCAT (Kucknd Meusemann, 2010). Phylogenetic analyses were carried outith Maximum-Likelihood (ML) and Bayesian Inference (BI). Therogram MrModeltest 2.2 (Nylander, 2004) was used to determinehe best-fitting model of sequence evolution for each data set usinghe Akaike Information Criteria (AIC). The best-fitting sequencevolution model for each region was the General Time Reversibleodel (GTR + I + gamma) (Rodríguez et al., 1990). We conducted

I analyses with MrBayes 3.1, unlinking the model parameters forach region (Ronquist and Huelsenbeck, 2003). Two independentuns were performed with four chains each during 10 million gen-rations. One tree was sampled for each of 100 generations, yielding00,000 trees. Twenty-five percent of the trees were eliminateduring the burn-in phase. Convergence of both runs was checkedith AWTY (Nylander et al., 2008). Maximum Likelihood analysesere conducted with RaxML (Stamatakis, 2008) under the GTRCATodel and defining the partitions.Bayesian dating was done with the programs baseml (included

n the PAML package; Yang, 1997), and estbranches and mul-idivtime, both part of the package Multidivtime (v. 9/25/03;horne et al., 1998; Kishino et al., 2001; Thorne and Kishino, 2002;utschmann, 2005). Multidivtime provides direct confidence inter-als for all dated nodes. The 50% majority rule consensus treesbtained from the BI analyses of each marker were used. Wesed two different calibration points to place minimal age con-traints on internal nodes in the phylogeny: (1) A fossil seed foundy Lancucka-Srodoniowa (1979) from the Early-Middle Miocene16 mya), described by the author as Campanula sp., because itstructure resembles seeds of Campanula or related genera. Thisossil was fixed and used to constrain the node of the mostecent common ancestor to all the above-cited genera. Cellineset al. (2009) used this fossil as an internal calibration point at6.5–17.5 my, supposing that this seed is characteristic of C. pyra-idalis. However, in our opinion, this fossil should be used for all

ampanula species, because its morphology (reticulate with largeells on the testa) also is observed in other species, i.e. C. carpatica,. glomerata, and C. scouleri (Oganesian, 1985; Shetler and Morin,986); and (2) an upper age constraint of 59 mya was deduced fromhe data of Bell et al. (2010), who calculated the age of many cladesf the Angiosperm tree by calibrating it with fossil constraints. Theges obtained by these authors ranged between 64 and 41 myaor the separation of Campanulaceae and Lobeliaceae. Geologicalges of islands were not used as calibration points in order to avoidircular reasoning.

rait state evolution

Based on the phylogeny obtained here for the Campanu-aceae, we reconstructed ancestral character states of pollinatorypes using the maximum likelihood approach implemented in

esquite (Maddison and Maddison, 2009), after pruning theranches leading to taxa for which no pollination data are available.econstructions were made on supported nodes over a posterioristribution of trees from the Bayesian analysis (1000 trees withhe highest posterior probability). We used the Markov k-state 1arameter model for these reconstructions.

esults

hylogenetic relationships

ML and BI analyses yielded similar topologies, confirming previ-us known relationships within Campanulaceae (Eddie et al., 2003;

olution and Systematics 14 (2012) 169– 182 173

osner et al., 2004; Park et al., 2006; Roquet et al., 2008; Cellineset al., 2009; Haberle et al., 2009). The 50%-majority rule treebtained with ML is shown in Fig. 1, with the main groups of Cam-anulaceae labelled. According to the phylogenetic relationships,ampanulaceae could be divided into three groups: Platycodoneaebasal to the two other groups), Wahlenbergieae and Campan-leae. Within Campanuleae, the species were distributed mainly

n two clades: Rapunculus and Campanula s. str. clades (Eddie et al.,003; Roquet et al., 2008). The oceanic bellflowers species stud-

ed here were distributed into four clades: the three species ofanarina form a monophyletic group included in Platycodoneae;he Heterochaenia species, together with Nesocodon mauritianusnd Berenice arguta (a Mascarene endemic), form a monophyleticroup within Wahlenbergieae; Musschia wollastonii and M. aureaere sister species, and together with other Campanula and Gadel-

ia species they formed a basal clade within the Campanuleae;nally, Azorina vidalii and Campanula jacobaea were embedded inhe Campanuleae, in a basal clade within the Campanula s. str. clade.

istribution, habitat, and reproductive biology

Floral biology of all species studied is summarised in Table 1 andigs. 2 and 3, and below are additional details.

(1) Azorina vidalii (Figs. 2a–c, 3a): At all six populations, insectsvisited the flowers. However, only at the three eastern popu-lations (Sao Lorenc o and Ponto Castello on Santa Maria, andPorto Martens on Terceira), the lizard Lacerta dugesii (Milne-Edwards) (Lacertidae) was observed to drink nectar and eatpollen and stylar hairs from the pollen brush. This is thefirst record of lizard pollination in the Azores archipelago. AtPonto Castello, lizards foraged actively on the flowers duringthe entire day (Fig. 2c), whereas at Sao Lorenc o, their activ-ity already ceased at 0930. However, here the visitation ratewas the highest with up to ten individuals simultaneouslyon a single plant. When visiting lizards or wind were shak-ing the flowers, clouds of pollen were released. Thus somewind pollination might take place as well. At both popula-tions on Santa Maria, Macroglossa stellatarum (L.) (Sphingidae)was recorded as pollinator too. On several occasions, it wasobserved to be the first pollinator in the morning, e.g. on the30th of August at Sao Lorenc o two M. stellatarum made 50 andthree visits between 0630 and 0700, well before any lizardsbegan to forage. At Ponto Castello, Vespidae species were alsoseen as pollinators. At Sao Lorenc o, several insects (Vespi-dae species and Apis mellifera (L.) (Hymenoptera), Muscidaeand Calliphoridae species (Diptera), and the Monarch butter-fly Danaus plexippus (L.)) were observed as pollinators. Theyconsumed pollen (except D. plexippus) from the pollen brushand drank nectar. At the three western populations (one onTerceira and two on the most western island Flores), no lizardswere observed and the only pollinators were honeybees andSepsis flies, both being very numerous. Together with a fewother small dipterans, many Sepsis were caught at the corollamargin by its sticky, oozing latex. Thus the species was polli-nated by lizards, insects, and wind on the eastern islands, andby only insects and wind on the western islands. The nativeCanary bird Serinus canaria (L.) (Fringillidae) is a potential pol-linator candidate and the Azores Bullfinch Pyrrhula murina

(Godman) is a well-known flower- and bud-eater (Bannermanand Bannerman, 1966: plate 28). In addition, the bellflowersTrachelium caeruleum L. and the widespread Campanula erinusL. grow on the islands. They are naturalised insect-pollinated

174 J.M. Olesen et al. / Perspectives in Plant Ecology, Evolution and Systematics 14 (2012) 169– 182

Fig. 2. Island Campanulaceae. (a–c) Azorina vidalii, Azores, Portugal, c. 5 m a.s.l., July 2000: (a) volcanic coast, Porto Martens, Terceira, (b) pollen- and nectar-harvesting Apismellifera, beach, Faya Grande, Flores, (c) nectar-drinking Lacerta dugesii, coastal slope, Sao Lourenc o, Santa Maria, (d) Musschia aurea and Lacerta dugesii, which gleans pollenfrom the underside of one of the stigmatic lobes, coastal slope, Sao Gonc alo, Madeira, Portugal, c. 20 m, August 2000, (e–f) Musschia wollastonii, laurel forest, Ribeiro Frio,Madeira, Portugal, c. 880 m, August 2000, (g) Musschia aurea, coastal slope, Sao Gonc alo, Madeira, Portugal, c. 20 m, August 2000, (h) Canarina canariensis, laurel forest, Anaga,Tenerife, Spain, c. 500 m, March 2001, (i–j) Canarina eminii, wet montane forest, Kapkwai, Mount Elgon, Uganda, c. 2200 m, August 2002, (k–l) Canarina abyssinica, near AddisAbeba, Ethiopia, October 2002 (photo courtesy P. Binggeli), (m) Nesocodon mauritianus, being visited by a Merle (Hypsipetes olivaceus) piercing a flower, Cascade 500 Pieds,Mauritius, c. 400 m, October 1997 (photo courtesy J. Madsen), (n) Nesocodon mauritianus, flower with red nectar oozing out, greenhouse specimen, Aarhus Univ., Denmark;(o–p) Heterochaenia ensifolia, montane forest, Cirque de Cilaos, Réunion, c. 2100 m, January 1999, and (q–r) Heterochaenia rivalsii, Philippia shrub, Plaine des Cafres, Réunion,c. 2400 m, January 1999.

J.M. Olesen et al. / Perspectives in Plant Ecology, Evolution and Systematics 14 (2012) 169– 182 175

Fig. 3. Island Campanulaceae. (a) Bud and open flower of Azorina vidalii, Azores; (b) Musschia wollastonii, Madeira; (c) Musschia aurea, Madeira; (d) open flower and budo abyssR .

f Canarina canariensis, the Canary Islands; (e) Canarina eminii, Uganda; (f) Canarinaéunion; (i) Heterochaenia rivalsii, Réunion; and (j) Heterochaenia ensifolia, Réunion

species, which have arrived much later to the islands than theancestor of A. vidalii (Fig. 1; Roquet et al., 2008).

(2) Campanula jacobaea: It was mainly visited by small solitaryHalictidae bees, but in some populations visitation rate was solow that self-pollination seemed more likely (Table 1). The lat-ter is supported by a greenhouse experiment. Self-pollination(18 flowers) produced a seed set of 50% compared to outcross-ing hand-pollination (24 flowers) (M. Alarcón and J.J. Aldasoro,unpubl.).

(3) Musschia wollastonii (Figs. 2ef, 3b): Various insects and thebird Blackcap (Sylvia atricapilla (L.)) (Sylviidae or Timali-

idae, see Alström et al., 2006) were pollinators. Latex oozedout from the edges of the stigmatic lobes and entangledsome of the flower-visiting insects. Among nectar-drinkinginsects were the abundant bumblebee Bombus maderensis

inica, Ethiopia; (g) Nesocodon mauritianus, Mauritius; (h) Heterochaenia borbonica,

(Erlandsson), Vespidae species and Artogeia rapae (L.) (Lep-idoptera: Pieridae), and the more rare Lycaena phlaeas (L.)(Lepidoptera: Lycaenidae). They were probably only nectarthieves and not pollinators, because the distance between nec-tar and stigma/stigmatic brush was larger than the body size ofthese animals. However, pollen-collecting Syrphidae species(Diptera) and the abundant Muscidae species might functionas pollinators. Sylvia atricapilla, on the other hand, touchedboth stigma and pollen area while drinking nectar (Olesen andValido, 2003b).

(4) Musschia aurea (Figs. 2dg, 3c): The main pollinators were Lac-

erta dugesii and Bombus maderensis. This lizard is small andcan easily move around on the inflorescences (snout-venterlength 60–80 mm; n = 8). Birds have never been reported aspollinators of this species, but Sylvia atricapilla and Serinus

1 gy, Evolution and Systematics 14 (2012) 169– 182

(

(

S

bssiA

Fig. 4. Regression of nectar concentration on corolla length for the studied Campan-ulaceae and two additional Wahlenbergia species. Colour indicates pollinator type:blue = insects, green = lizards, and red = birds. Data for Campanula rotundifolia L. arefrom Cresswell and Robertson (1994) and those for Wahlenbergia berteroi and W.fernandeziana Skottsb. are from Bernardello et al. (2000). (For interpretation of thert

tecttbbVplZv(

rlstTspedab9chstN

76 J.M. Olesen et al. / Perspectives in Plant Ecolo

canaria may be potential pollinators, which besides L. duge-sii (Elvers, 1978), are the only known, potential vertebratepollinators in the archipelago.

(5) Canarina canariensis (Figs. 2h, 3d): Its pendant flower excludesinsects. On Tenerife, pollinators were the five birds Specta-cled Warbler (Sylvia conspicillata (Temminck)), Black-headedWarbler (S. melanocephala (Gmelin)) (Sylviidae or Timali-idae), the Blue Tit (Parus teneriffae (Lesson)) (Paridae), Canarybird (Serinus canaria) and Chiff-chaff (Phylloscopus canariensis(Vieillot)) (Phylloscopidae), and some invertebrates, e.g. Com-mon Pill-woodlouse Armadillidium vulgare Latreille (OrderIsopoda) and the bee Lasioglossum viride Brullé (Halictidae),but on La Palma, only P. canariensis was observed as pollinator(Olesen, 1985; Ollerton et al., 2009; Rodríguez-Rodríguez andValido, 2011). Even semi-slugs (Plutonia spp., Gastropoda) andBlack Rat (Rattus rattus (L.)) may be pollinators (Rodríguez-Rodríguez and Valido, 2011). The Canary Islands have otherbellflowers, such as Wahlenbergia lobelioides (L.f.) DC, Trache-lium caeruleum, Campanula erinus, C. dichotoma L., and Legousiaspecies. Their floral morphology suggests insect pollination (T.caeruleum), mixed insect pollination-selfing (W. lobelioides, C.dichotoma), and selfing (C. erinus and Legousia species) and allare probably naturalised on the islands.

(6) Nesocodon mauritianus: (Figs. 2mn, 3g). The nectar is red,which is an unusual trait (Olesen et al., 1998; Hansen et al.,2007a). According to experiments, red colouration may attractlizards (Hansen et al., 2006). Two bird species were seenvisiting the flowers: the common Red-whiskered Bulbul Pyc-nonotus jocosus (L.) (Pycnonotidae) and, on one occasion, theMauritian Merle Hypsipetes olivaceus (Jardine & Selby) (Pyc-nonotidae). The former robbed flowers by piercing holes attheir base or inserted its head legitimately into the flower toreach the nectar and might thus act as a pollinator (Olesenet al., 1998).

(7) Heterochaenia ensifolia: Flowering plants were 75–300 cm tall(Figs. 2op, 3j) and flowers were visited by the Mascarene GreyWhite-eye Zosterops borbonicus (Boddaert) (Zosteropidae orTimaliidae, see Alström et al., 2006).

(8) Heterochaenia rivalsii: Flowering plants were(35–)100–200 cm tall (Figs. 2qr, 3i). The flowers werealso visited by Z. borbonicus.

(9) Heterochaenia borbonica: Flowering plants are 1–2 m tall(Badré, 1976). Number of inflorescences is unknown. Thespecies may be polycarpic and flowers are probably pollinatedby white-eyes (Zosterops spp.). The Mascarenes only have onemore native Campanulaceae species, Berenice arguta Tul., butthe flowers of this species are only a few mm in size and mightbe selfed or visited by small insects only.

10) Canarina eminii: Flowering plants were 2–4 m tall(Figs. 2ij, 3e). The species had polycarpy. The MariquaSunbird (Nectarinia mariquensis (Smith), Nectariniidae) wasobserved as pollinator 2–3 times per hour.

11) Canarina abyssinica: Flowering plants were up to 2–3 m tall(Figs. 2kl, 3f). The species had polycarpy. It was pollinated bysunbirds (P. Binggeli, pers. com.; J.J. Aldasoro and M. Alarcón,pers. obs.).

ummary of reproductive biology

Reconstruction of ancestral pollination states suggested thatird/lizard pollination had appeared four times in Campanulaceae

. str. (Pie charts representing relative proportion(s) of recon-tructed states are shown on the nodes in Fig. 1), i.e. bird pollinationn the Indian Ocean species and in Canarina, lizard pollination inzorina and birds/lizard in Musschia. This analysis indicated also

FtpD

eferences to color in this figure legend, the reader is referred to the web version ofhis article.)

hat bee-pollination was ancestral in the family and might havevolved towards bird/lizard pollination at least in three of theseases after colonisation of oceanic islands. However, there werewo exceptions: Canarina canariensis was probably pre-adaptedo bird-pollination because of its origin in the African continent,efore it or its ancestor colonised the Canaries. Several islandellflowers, however, remained insect-pollinated, e.g. the Capeerdean Campanula jacobaea, which was bee pollinated or possiblyartially selfing. Other island species not studied here, are also pol-

inated by insects, such as species of Wahlenbergia growing in Newealand (Lord, 2008) or on Robinson Crusoe Island where they areisited by bees and butterflies and possibly they also self-pollinateAnderson et al., 2000, 2001; Bernardello et al., 2000, 2001).

The Canarian/Madeiran vertebrate-pollinated bellflowers hadeddish-yellowish flowers with a strong, rigid corolla, producingarge quantities of nectar, whereas the Azorean and Mascarenepecies were blue or white, but they have similar corolla tex-ure and nectar characteristics as the Canarian/Madeiran ones.he insect-pollinated bellflowers were blue or whitish, producingmaller quantities of more concentrated nectar. Flowers of lizard-ollinated species were smaller than that of bird-pollinated ones,xcept for Heterochaenia. Flower orientation and shape were pen-ant and bell-shaped, respectively, except for Musschia, which hadn upright flower. Lizard-pollinated flowers had scent, whereasird-pollinated ones were scentless. Nectar volume varied from

to 92 �l and the concentration was weak, 8–22%, whereas theoncentration of nectar of insect-pollinated species usually wasigher than 30%. However, there was a wide variation in flowerize among bird- and lizard-visited species (Figs. 2 and 3) and nec-ar concentration and flower size were negatively correlated (Fig. 4;ectar conc. = −32.7 log10(Corolla tube length) + 66.9, R2

adj. = 0.35,

= 5.82, P < 0.04). The site of secondary pollen presentation washe style in Azorina, Nesocodon, and Canarina, on both style andart of the stigma in Heterochaenia, and on the stigma in Musschia.ichogamy was weak in Musschia and Heterochaenia, and stronger

gy, Ev

iibctt

D

naew

sf(CroHaNeCMMsttp7tc

D

E

ut1aOl2eqce

ptbflvof“t

T

iabni(SeBim1ec

(PnlmcMT

ewo(Cztd

ffS2udI

osoattvwVMLeiplar

J.M. Olesen et al. / Perspectives in Plant Ecolo

n Azorina, Nesocodon, and Canarina. In general, studied coastalsland-bellflowers were pollinated by lizards and montane speciesy birds (Table 1). Features such as flower size, shape, orientation,olour, nectar volume and concentration, secondary pollen presen-ation, and dichogamy followed more phylogeny than pollinatorype.

ating island colonisation

Mean age estimates and their credibility intervals for relevantodes are given in Table 2. The age estimate with the highest prob-bility obtained by BRC with multiple calibrations is indicated forach node in the Bayesian consensus phylogenetic trees obtainedith the marker data (Fig. 1).

In this paragraph ‘might have’ is tacitly assumed in everyentence. The tribe Platycodoneae diverged from the rest of theamily 36–40 mya (Fig. 1). Although its ancestral area was AsiaRoquet et al., 2009), it also included the African-Macaronesiananarina, which split from the rest of the tribe 25.5 mya. Cana-ina canariensis split from its African ancestors 8.3 mya. Separationf Wahlenbergieae and Campanuleae occurred 25.9 mya, but theeterochaenia–Berenice–Nesocodon group split from Wahlenbergiand the rest of the species of Wahlenbergieae s. str. 20.3 mya.esocodon and Berenice split from Heterochaenia rivalsii and H.nsifolia 6.7 mya. The ancestor of the clade of Musschia and theampanula peregrina group appeared 13.4 mya, possibly in theediterranean basin (Roquet et al., 2009), while the ancestor ofusschia appeared 8.8 mya. The separation of the two Musschia

pecies was 2.0 mya. Two other dispersal events brought the ances-or of Azorina to the Azores 8.3 mya and Campanula jacobaea tohe Cape Verde Islands 1.2 mya. Thus our data revealed a tem-oral sequence of four island colonisations, the oldest took place.8–8.4 mya for the Canarina and Heterochaenia-Nesocodon clades,he intermediate 4.1–4.2 mya for the ancestors of Azorina and Muss-hia and the most recent took place 1.4 mya for C. jacobaea.

iscussion

volution and pollination of island bellflowers

The pollination landscape of oceanic islands is said to have somenique properties: 1, a scarcity of flower-visiting insects (both inotal abundance and in species number) (MacArthur and Wilson,967; Janzen, 1973; Andrews, 1979; Anderson et al., 2001; Olesennd Jordano, 2002; Olesen, 2003; Olesen and Valido, 2003a,b, 2004;lesen et al., 2011); and 2, a package of generalised vertebrate pol-

inators, viz. bats, birds and lizards (Olesen, 1985; Olesen et al.,002a,b; Olesen and Valido, 2003a,b; Dupont et al., 2003; Validot al., 2004). Both properties, however, are difficult to verify anduantify, mainly because it is hard to find an appropriate “adja-ent mainland” comparison (however, see Inoue, 1988, 1990; Inouet al., 1996).

Reconstruction of the ancestral pollination biology of the Cam-anulaceae using maximum likelihood suggests that pollinators ofhe ancestors of our study bellflowers were insects, and in generalees (Fig. 1). Bellflowers colonising islands might either experienceower selection from birds and lizards or be adapted to pollinatingertebrates prior to their colonisation, or both. Given our estimates

f ancestral states, vertebrate pollination evolved independentlyour times in the Campanulaceae, i.e. our species experienced fourout-of-Africa” events, and the shift in pollinators was probably inhree of the cases associated with island colonisation.

twst

olution and Systematics 14 (2012) 169– 182 177

he Atlantic Ocean colonisations

How the present Macaronesian islands actually were coloniseds uncertain, because several seamounts closer to Africa witnessbout a presence of former islands, that for many taxa might haveeen stepping stones to the extant islands. The Azores consists ofine islands divided into three groups based on geographical prox-

mity: the eastern includes Santa Maria (8.1 my) and Sao Miguel4.0 my), the central includes Terceira (3.5 my), Graciosa (2.5 my),ao Jorge (0.6 my), Pico (0.3 my), and Faial (0.7 my), and the west-rn Flores (2.2 my) and Corvo (0.7 my) (Abdel-Monem et al., 1975;orges and Hortal, 2009). The maximum distance between any two

slands is 600 km, viz. between Santa Maria and Corvo. The mini-um distance to Madeira is 860 km and to the Iberian Peninsula

584 km. Sao Miguel and Santa Maria were the first to be discov-red around 1427. Thus the 8.3-my Azorina vidalii might first haveolonised Santa Maria.

The Madeiran Archipelago consists of the island of Madeira4.6 my), the Porto Santo islands (14 my), and the Desertas (3.6 my).ublished age estimates vary much. Madeira is 700 km from theearest mainland, and 300 km south of the Madeiran Archipelago

ies another archipelago, the Salvagens (12 my). Thus the 8.8-y Musschia ancestor probably colonised Porto Santo first, but

ould also have reached Madeira. Until 18,000 ya, the Desertas andadeira probably were one island (Feraud et al., 1981; Mitchell-

homé, 1985; Portugal-Ferreira et al., 1988).The Canarian archipelago consists of seven larger islands: The

astern Lanzarote (16 my) and Fuerteventura (21 my) and theestern Gran Canaria (15 my), Tenerife (7.5 my, some parts are

lder), La Gomera (14 my), La Palma (3 my), and El Hierro (1.5 my)Anguita et al., 2002; Carracedo et al., 2002). Thus the 8.3-myanarina canariensis may have moved through a wetter Lan-arote/Fuerteventura or directly to Gran Canaria or Tenerife, andhen later to the other western islands when these emerged andeveloped suitable habitats.

The Cape Verde islands are also much older to the east (26 myor Sal and Boavista, 7–20 my for Maio) than to the west (5.5 myor Brava, 6.5 my for S. Nicolau, 7 my for S Antao, and 10 my forantiago; Mitchell-Thomé, 1985; Plesner et al., 2003; Duprat et al.,007), i.e. ages already being high, when the ancestor of Campan-la jacobaea colonised 1.4 mya. The sister species to C. jacobaea, C.ichotoma (Fig. 1), is an annual, which also grows in the Canaryslands, N Africa, S Spain and S Italy.

The five Macaronesian archipelagos have 17 lizard species, allf them are potential nectar consumers and pollinators (J.M. Ole-en, unpubl.). Three extinct lizards are known. However, only fourf the lizards have actually been observed as pollinators (A. Validond J.M. Olesen, unpubl.). The lizard Lacerta dugesii is endemic tohe Madeiran Archipelago and the Salvagens, and it is introducedo the Azores and the Iberian Peninsula. It is omnivorous, achievesery high densities and ranges from sea level to 1800 m. It is aell-known pollinator, being observed visiting 13 plant species (A.alido and J.M. Olesen, unpubl.). Its closest relative is probably theoroccan L. perspicillata (Duméril & Bebron) (Harris et al., 1998).

acerta dugesii colonised the Madeiran Archipelago 2.8 mya (Brehmt al., 2003). This is a late colonisation compared to the age of theslands and may indicate that lacertids, in general, are poorer dis-ersers than geckos (Brehm et al., 2003). Today, all other Madeiran

izards are introduced. The living sisters of Musschia and Azorinare “Campanula” species such as C. primulifolia Brot. and C. pereg-ina Hoffm. & Link, and Gadellia lactiflora (Bieb.), distributed along

he Mediterranean shores. All have rotate, white or bluish flowers,hich attract many generalist insects (M. Alarcón and J.J. Alda-

oro, unpubl.). Thus the Musschia ancestor must for an extendedime-span have relied on other pollinators, e.g. migrant birds or

178 J.M. Olesen et al. / Perspectives in Plant Ecology, Evolution and Systematics 14 (2012) 169– 182

Table 2Estimated age and standard deviation (SD) of nodes in the phylogeny discussed in the paper, using Bayesian Relaxed Molecular Clock (Multidivtime) with multiple calibrations.Node numbers correspond to the ones given in the chronogram (see Fig. 1).

Nodes Multidivtime age (SD)million years

Confidence interval (lowerlimit–upper limit)

Separation of Platycodoneae and the ancestors of Wahlenbergieae and Campanuleae 38.5 (1.9) 34.71–42.48Separation of the ancestor of Platycodon, Codonopsis, Leptocodon, and Cyananthus, and the

ancestor of Canarina25.5 (2.5) 20.59–30.35

Separation of Canarina canariensis and the ancestor of C. eminii and C. abyssinica 8.3 (3.0) 3.55–15.01Separation of Wahlenbergieae and Campanuleae 25.9 (2.4) 21.26–30.50Separation of the ancestor of Wahlenbergia, Prismatocarpus, Roella, Merciera, Microcodon,

Craterocapsa, and Theilera, and the ancestor of all Heterochaenia, Berenice, and Nesocodon20.3 (2.4) 15.75–25.05

Separation of the ancestor of the ancestor of Berenice arguta, Nesocodon mauritianus,Heterocodon rariflorum, and Heterochaenia ensifolia, and the ancestor of Heterochaeniaborbonica

8.9 (2.4) 4.92–14.37

Separation of the ancestor of the ancestor of Berenice arguta and Nesocodon mauritianus, andthe ancestor of Heterochaenia rivalsii and Heterochaenia ensifolia

6.7 (1.6) 3.85–10.28

Beginning of diversification of Campanuleae (separation of Musschia clade) 24.7 (2.3) 19.99–29.29Beginning of diversification of Musschia clade (split of Gadellia) 13.4 (2.3) 9.33–18.22Separation of Campanula primulifolia and the ancestor of Musschia 8.8 (2.5) 4.08–13.78Separation of Musschia aurea and M. wollastonii 2.0 (1.8) 0.06–6.77Divergence of Rapunculus clade and Campanula clade 21.3 (2.3) 16.83–25.76Beginning of diversification of Rapunculus clade 17.4 (2.2) 13.34–21.95Beginning of diversification of Campanula clade 19.9 (2.2) 15.52–24.27Separation of the ancestor of Campanula sect. Roucela (C. pinatzii, C. creutzburgii, C. drabifolia,

and C. erinus) and the ancestor of Campanula group mollis including Azorina12.0 (2.0) 8.49–16.11

efb2ttma

b((lb23ia(pPticOsWtbta

tMraCac

tt

T

Mr2a2

sdlMa(dsiflmcvavitt

wbpA

Separation of Azorina and its sister clade in the C. mollis group

Separation of C. jacobaea and its sister C. dichotoma

xtinct lizards. Seven endemic lacertid Gallotia species are knownrom the Canaries. Gallotia atlantica, G. caesaris, and G. galloti haveeen observed as pollinators to eleven plant species (Valido et al.,002, 2004; A. Valido and J.M. Olesen, unpubl.). Since the colonisa-ion history of the Gallotias began shortly after the appearance ofhe eastern islands, i.e. 17–20 mya (Cox et al., 2010), these species

ight have been and might be important mutualists (pollinatorsnd seed dispersers) to many Canarian plants.

Macaronesia has nine species of actual nectar birds. Threereeding species of Sylvia, viz. (1) S. melanocephala on the Canaries,2a) S. atricapilla gularis on the Cape Verde Islands and the Azores,2b) S. a. heineken on Madeira and the Canaries, and (3) S. conspicil-ata on Cape Verde Islands, Canaries and Madeira. (1) and (2) areoth young species, but their actual age is unknown (Alström et al.,006). The first colonisation of the two former species might go

my back in time (Dietzen et al., 2008). (1) visits ≥5 native andntroduced species, (2b) visits ≥4 species, and (3) ≥2 species. Inddition, (4) the endemic Phylloscopus canariensis visits ≥9 species,5) the native Canary Serinus canaria and (6) the introduced Euro-ean Serin S. serinus (L.) visit together ≥4 species, (7) the nativearus teneriffae (Lesson) (Paridae) visits ≥6 species, and finally, (8)he possibly native Passer hispaniolensis (Temminck) and (9) thentroduced P. domesticus (L.) (Ploceidae) have only been seen on aouple of introduced species (Olesen, 1985; Valido et al., 2004; J.M.lesen, pers. obs.). The first colonisation by Phylloscopus canarien-

is could have occurred as early as 2.5 mya (Helbig et al., 1996).ithout proper phylogenies for these bird groups it is not possible

o reconstruct the ancient pollinator fauna for the Atlantic Oceanellflower colonisations. Our suggestion is that the ancestors ofheir present-day bird pollinators arrived later than the bellflowersnd the first pollinators might thus have been Gallotia lizards.

In general, E African Canarinas are pollinated by sunbirds, buthe relatives of Nectarinia mariquensis observed as pollinator at

t. Elgon only diversified 3–5 mya (Warren et al., 2003). Cana-ina, however, separated from the other Platycodoneae 25.5 mya,

nd its colonisation in the Canary Islands took place 8.3 mya. Thus,anarina must have been pollinated by other bird groups or by thencestors of modern sunbirds. The fleshy Canarina fruit, which isonsumed by birds (J.M. Olesen, pers. obs.), makes its dispersal to

nods

8.3 (1.7) 5.38–11.881.4 (1.2) 0.05–4.65

he Canaries easy to accept. Canarina canariensis disappeared fromhe continent when the humid forests of NC Africa dried out.

he Indian Ocean colonisations

The Mascarenes are an archipelago of three large islands,auritius, Réunion, and Rodrigues, and several small islets sur-

ounding Mauritius. The age of Mauritius and Réunion is 7.8 and.1 my, respectively. The estimate for Rodrigues varies between 1.5nd 15 my (Fisk et al., 1989; Duncan and Storey, 1992; Warren et al.,003).

The Mascarene flora includes c. 885 native flowering plantpecies and an additional 775 are naturalised (Strahm, 1993). Aiverse assemblage of flower-visiting insects and vertebrates pol-

inate these plants. Compared to their size and oceanic isolation, theascarenes are relatively rich in insects, and 2000 species of native

nd introduced insects are known from the island of MauritiusMotala et al., 2007). The islands have 22 native vertebrates, candi-ating as pollinators. Flower visitation has not been observed for allpecies, but based on information from the Mascarenes and otherslands we expect seven bird species, 13 lizard species and twoying foxes to service as pollinators and potential pollinators ofany native and introduced plants (J.M. Olesen, unpubl.). Histori-

ally, we know nine extinct vertebrate species, which also may haveisited flowers. Thus when the Portuguese made landfall on therchipelago in 1528 (Cheke and Hume, 2008), the flower-visitingertebrate fauna may have summed up to 31 species. Additionally,n historical time, two birds and ten lizards have been introducedo the islands. The phylogeny and colonisation history of most ofhe flower-visiting vertebrates are known.

The extant native Mascarene flower-visiting birds are threehite-eyes (Zosterops), two fodies (Foudia, Ploceidae), and two

ulbuls (Hypsipetes). White-eyes came to the SW Indian Ocean,robably from Asia, using a string of seamounts as stepping-stones.

founding white-eye arrived 1.8 mya to either Mauritius or Réu-

ion, maybe from Grand Comore, and 1.2 my later colonised thether island. Rodrigues has no white-eye. In Mauritius and Réunion,ifferent selection regimes produced an ancestral Olive White-eyepecies on one island and a Grey White-eye on the other, finally

gy, Ev

tgpTeoc2

oibM(Rww

ipsab2Tcat(j1H

pfIpiPasofU2paeRliMe2

ihefseLAho

h2Rfih“gaaSoV

oCcp1msi1sd

4MtaittBRbmhaHat

mdbpmmsahtrHpsbwmo

J.M. Olesen et al. / Perspectives in Plant Ecolo

he olive colonised the other island 0.64 mya, and, vice versa, therey the other island 0.43 mya, ending up with the present sym-atric grey-olive species pair on each island (Warren et al., 2006).he olives are regarded as two species: Mauritius Olive White-ye (Z. chloronothos (Vieillot)) and Réunion Olive-White-eye (Z.livaceus (L.)), whereas the greys are only subspecies of the Mas-arene Grey White-eye (Z. borbonicus (Boddaert)) (Gill, 1971; Balen,008).

The phylogeny of fodies (Foudia) is unknown. It is a genusf seven extant and one extinct species, being endemic to theslands in the W Indian Ocean. They originated from Africa, proba-ly colonised the Seychelles first, then the Mascarenes, and finallyadagascar and Comores. The Mascarene species are F. rubra

Gmelin) from Mauritius and F. flavicans (Newton) from Rodrigues.éunion once had a fody (F. delloni (Cheke and Hume, 2008)), whichent extinct around 1675. The Malagasy F. madagascariensis (L.)as introduced to the Mascarenes after 1750.

Five extant species of bulbuls (Hypsipetes) are native to theslands of the W Indian Ocean and they constitute a mono-hyletic group (Warren et al., 2005). The Mascarenes have twopecies: the extant H. olivaceus (Jardine & Selby) from Mauritiusnd the extinct H. borbonicus (Forster) from Réunion (subfossileones of a Rodrigues Hypsipetes sp. are known; Cheke and Hume,008). Bulbuls colonised the region from India and the Seychelles.he highland of Grand Comoro and Réunion (H. borbonicus) wasolonised 2.1 mya and later came Aldabra, Madagascar, Mauritiusnd the lowland of the Comores. The Mauritian H. olivaceus andhe widespread H. madagascariensis (Müller) diverged 0.4–1.8 myaWarren et al., 2005). The alien Red-whiskered Bulbul (Pycnonotusocosus (L.)) was introduced to Mauritius in 1892 and Réunion in965 (Mandon-Dalger et al., 2004; Amiot et al., 2007; Cheke andume, 2008; Linnebjerg et al., 2009).

Flying foxes (Pteropodidae) are important fruit dispersers andollinators. On oceanic islands, they are reported to visit >25 plantamilies and >50 species for floral resources (J.M. Olesen, unpubl.).n reviews, e.g. Dobat and Peikert-Holle (1985), 15 species of islandteropodids are listed as nectar drinkers. On the W Indian Ocean

slands, we find seven extant and one extinct species of the genusteropus, three of them being Mascarene. They do not constitute

monophyletic group but were assembled through three coloni-ations (O’Brian et al., 2009). The Mascarene species were partf the second and third colonisation. Pteropus rodricensis Dobsonrom the Island of Rodrigues constituted the second colonisation.ntil 1846–1906, it was also found on Mauritius (Cheke and Hume,008). The phylogeny for the species in the third colonisation isoorly resolved (O’Brian et al., 2009). It is a clade of four species,nd one of them is the Mauritius Flying Fox P. niger (Kerr) (Nyhagent al., 2005). Until 1772–1801, this species was also found onéunion and bone material is known from Rodrigues. The phy-

ogenetic and biogeographical position of the extinct P. subnigers unknown. It disappeared from Réunion 1772–1860 and from

auritius 1864–1873 (Cheke and Hume, 2008). It was the small-st of all Pteropus and definitely a flower visitor (Cheke and Hume,008; J.M. Olesen, unpubl.).

The islands in the W Indian Ocean once had one of the richestnsular faunas of lizards in the world, and is still today, in spite ofuman-caused extinctions, fairly rich. They arrived to the regionither from Africa and Madagascar or with the Equatorial currentrom SE Asia and Australia. The day geckos (Phelsuma) came to themaller Indian Ocean islands from Madagascar. The two gecko gen-ra Nactus and Lepidodactylus and the skinks Cryptoblepharus, and

eiolopisma made an incredible journey of at least 5600 km fromsia or Australia (Austin and Arnold, 2006). Hemidactylus geckos,owever, are probably the most exclusive long-distance dispersersf all (Carranza and Arnold, 2006).

fil

A

olution and Systematics 14 (2012) 169– 182 179

When humans arrived to the Mascarenes, this archipelago stillad c. 20 species of lizards. Today, 13 are left (Cheke and Hume,008). Mauritius (incl. its islets) has seven species out of 17 left,odrigues has one out of three left, and Réunion has three out ofve left. Ten new lizard introductions from outside the Mascarenesave occurred. These are mainly members of a cosmopolitan club ofhouse-geckos”, in addition to some “new” Phelsumas from Mada-ascar and the Seychelles. Many of these lizards serve or may serves pollinators and seed dispersers (Nyhagen et al., 2001; Olesennd Valido, 2003a; Hansen et al., 2007b; Hansen and Müller, 2009).even of the 13 native species and five of the introduced ones arebserved as pollinators, and all except three being Phelsumas (A.alido and J.M. Olesen, unpubl.).

The genus Phelsuma is monophyletic and encompasses a totalf 38–40 species, including two extinct ones (Austin et al., 2004;arretero et al., 2005; Rocha et al., 2007). In general, Phelsumas arelosely associated with flowering plants, being attracted by nectar,ollen, fruit and flower-visiting insects on which they prey (Vinson,975). They and their nectar and fruit plants may share a com-on evolutionary history on the W Indian Ocean islands. Phelsumas

eem to have reached the Mascarenes in one colonisation event,.e. Rodrigues 5.1 mya, Mauritius 4.6 mya and much later Réunion.2 mya (Austin et al., 2004; Rocha et al., 2007). The first Mauritianpecies were wet forest species and first later did they invade therier lowland.

The four Mascarene bellflowers are found in the mountains from00 m (N. mauritianus) to 1700–2400 m (Heterochaenia). All theascarene nectar birds, both native and introduced, are seen in

he upland of the islands, at least part of the year (Cheke, 1987),nd are thus, at least with respect to habitat, potential pollinators,.e. six and five bird species to Nesocodon and Heterochaenia, respec-ively. Most lizards are lowland species, but a few can be found inhe Mauritian highland forests, especially Hemiphyllodactylus typusleeker, Phelsuma cepediana Merrem, and P. rosagularis Mertens. Inéunion, no lizard reaches the altitudes of Heterochaenia. Phelsumaorbonica Mertens may, however, be seen up to 1000 m. Nesocodonauritianus is known from three sites. At Cascade 500 Pieds, theabitat is very humid and often windy and only birds, but no insectsnd lizards, can be expected as pollinators (Olesen et al., 1998).owever, the other two sites are within the range of P. ornata Graynd their visitation is likely (Hansen et al., 2006). Heterochaenia, onhe other hand, has to be bird-pollinated.

The phylogenetic analyses tell (1) that the white-eyes arriveduch later to the Mascarenes than the bellflowers; (2) that we

o not know if fodies could have been the first pollinators of theellflowers; (3) that Hypsipetes borbonicus might be among the firstollinators of Heterochaenia rivalsii and H. ensifolia and their com-on ancestor; (4) that the two extant Mascarene flying foxes areuch younger than the bellflowers, and (5) that although Phel-

umas appeared in the Mascarenes later than the bellflowers, theyre the most likely pollinator candidates living in the Mascareneighland. Our conclusion is that Nesocodon mauritianus might, inhe first place, have been pollinated by the Phelsuma cepediana/P.osagularis ancestor and later maybe also by Zosterops spp. andypsipetes olivaceus, and finally, in this century, these evolutionaryartners were partly replaced by the introduced Pycnonotus joco-us. Heterochaenia was probably initially pollinated by Hypsipetesorbonicus and later by Zosterops species. However, we do not knowhat pollination role the extinct Pteropus subniger and the fodiesight have played. In a pollen analysis of seven specimens of skins

f P. subniger deposited in the Natural History Museum, London,

ve pollen types were found, but none belonged to the Campanu-

aceae (J.M. Olesen, unpubl.).Heterochaenia and Nesocodon may originate in the wet E

frican or Malagasy forests 20.3 mya. In E Africa, large forests

1 gy, Ev

w(flobi

C

r(iHradbhndltoofdnabfraanypbpm

haQatabdga

Wforihbs(1aec

oApKefsdbiHpNnWasi

optcipdaf

A

NmiCd(tsaCeU

A

i

R

A

A

A

A

A

80 J.M. Olesen et al. / Perspectives in Plant Ecolo

ere widespread at this time, while savannas developed 9.4 myaCerling, 1992). If the ancestor of Heterochaenia/Nesocodon had blueowers they may have been pollinated by large bees in Madagascarr E Africa, e.g. by Xylocopa and Pachymeles. Today, only one largeee, X. fenestrata Fabricius is known from the Mascarenes and it is,

n general, confined to the dry lowlands.

oncluding remarks

Oceanic islands have a poor insect fauna or at least the mostemote ones, e.g. the Juan Fernández Islands. Bernardello et al.2001) write about the Juan Fernández Islands: . . .total insect activ-ty was very, very low. Native insect pollinators are virtually absent.owever, insect richness and abundance on islands need more

esearch. It is known that low species density on islands and thus reduced interspecific competition may lead to very high abun-ance (i.e. density compensation, MacArthur et al., 1972), e.g. inirds and lizards (Olesen et al., 2011). Thus island bellflowers mayave converged in their floral biology towards vertebrate polli-ators due to a shortage of insects. We are, however, not able toiscern between convergent and parallel evolution because of a

ack of information about the variation in the ancestral charac-er states for our island clades. Except for Canarina, they probablyriginated from insect-pollinated ancestors. Thus the floral biol-gy of our Campanulaceae clades has undergone parallel evolutionrom insect to vertebrate pollination. Thus the island bellflowersescribed here represent a natural experiment with four phyloge-etic independent repeats (Canarina, Musschia, Azorina/Campanula,nd Heterochaenia/Nesocodon). They are all pollinated by verte-rates, but became so in slightly different ways. Most of them have aew floral traits associated with bird pollination, e.g. robust corollas,ed-yellow corolla, plenty of diluted nectar, and no odour (Faegrind Pijl, 1971). These characters are not evenly present in all speciesnd there are several variants: the shape of corolla: campanulate toearly rotate, the colour of the corolla: blue, purple, red, orange, orellow, and different flower orientations, form of secondary pollenresentation, and level of dichogamy. The recent establishment ofird and reptile species on islands, variation in the regional insularool of vertebrate pollinators, and various phylogenetic constraintsay explain the lack of a greater degree of convergence.Except for cyclones in the W Indian Ocean, our study islands

ave a stable climate, which may allow an accumulation of speciesnd convert them into important refuges of biodiversity. However,uaternary changes in atmospheric circulation, sea-level changesnd anthropogenic disturbance mediated many invasions, extinc-ions and changes in flora and fauna at the end of the Tertiarynd Quaternary (Cronk, 1997). Consequently, the relationshipsetween vertebrates and plants may have changed many timesuring this period. This may be the evolutionary-ecological back-round for the impressive diversity and great lability in plant–birdnd plant–lizard relationships, observed among island bellflowers.

Our sample of island bellflowers is certainly not complete.ahlenbergia berteroi Hook., for example, an endemic bellflower

rom the Robinson Crusoe Island, has floral traits in common withur study species, such as a large (2 cm long), strong corolla of aeddish colour with white stripes, no scent and a more or less hor-zontal orientation (J.M. Olesen, pers. obs.). Flies and maybe ants,owever, are the only observed pollinators (Anderson et al., 2001),ut the local hummingbirds, the endemic Sephanoides fernanden-is (King) and the native S. sephaniodes (Lesson), are candidatesColwell, 1989; Roy et al., 1998). These birds pollinate at least

4 native plant species and many introduced in the archipelagond are thus super generalists (Bernardello et al., 2001; Olesent al., 2002a). Wahlenbergia berteroi grows in low numbers onoastal cliffs near the town San Juan Bautista and may today be

A

olution and Systematics 14 (2012) 169– 182

utcompeted by simultaneously flowering garden species, such asbutilon, Eucalyptus, and Nasturtium (J.M. Olesen, pers. obs.). Insectollination is rarely observed on the island (Skottsberg, 1928;uschel, 1952; Wilson, 1973; Anderson et al., 2001; Bernardellot al., 2001), and this may allow the hummingbirds to expand theireeding niche to other food items. Sephanoides fernandensis is evenaid to sip fermented sap from old fig fruits on the ground, and too so to the extent that it intoxicated falls off its perch, risking toe predated upon by cats (I. Leiva, pers. com.). Another group of

sland bellflowers are the Wahlenbergias from New Zealand and St.elena; they have a floral morphology suggesting insect- and self-ollination (Petterson, 1997; Lord, 2008), e.g. the much-studiedew Zealand W. albomarginata Hook., which has a broad polli-ator fauna of insects (Primack, 1983). St. Helena once had fourahlenbergia species, two of which are now extinct (Cronk, 1987),

nd, in addition, Trimeris scaevolifolia (Roxb.) Mabb. All three extantpecies have white, 4–12 mm large flowers, probably pollinated bynsects.

In the introduction, we postulated that island Campanulaceaeffered excellent opportunities in the study of convergence andarallel evolution in similar functional traits, viz. traits relatedo their floral biology. All our study species (except Canarinaanariensis) evolved towards vertebrate pollination in insect-poorsland environments and they probably all originated from insect-ollinated continental ancestors. Due to phylogenetic constraints,ifferent traits in different species functioned as adaptations tottract local vertebrate pollinators. We anticipate that other plantamilies may show similar patterns.

cknowledgements

We thank J. Christiansen, Y.L. Dupont, F. Pedersen, P. Damgårdielsen, J. Rasmussen, A. Valido, P. Binggeli, J.A. Carvalho, F. Fried-ann, C. Thébaud, W.M.M. Eddie, and E. Dias for assistance,

nformation and critique of manuscript, and we are grateful to A.K.hristensen and C.L. Løjtnant for access to their unpublished stu-ent report (2004. Phylogenetic analysis of Nesocodon mauritianusCampanulaceae) applied by nrITS sequences – a biogeographical con-emplation, Univ. Copenhagen, 43 pp.). Our study received financialupport from the Danish Natural Science Research Council (JMO)nd the Carlsberg Foundation (JMO), and also from the Spanishentral Administration (Ministerio de Educación y Ciencia, refer-nce: REN2003-04397, JJA, CR), and the Agència de Gestió d’Ajutsniversitaris i de Recerca, Generalitat de Catalunya (JJA, CR).

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at doi:10.1016/j.ppees.2012.01.003.

eferences

bdel-Monem, A.A., Fernández, L.A., Boone, G.M., 1975. K–Ar ages from the east-ern Azores group (Santa Maria, São Miguel and the Formigas Islands). Lithos 8,247–254.

izen, M.A., Morales, C.L., Morales, J.M., 2008. Invasive mutualists erode native pol-lination webs. PLoS Biol. 6, 396–403.

lström, P., Ericson, P.G.P., Olsson, U., Sundberg, P., 2006. Phylogeny and classifica-tion of the avian superfamily Sylvioidea. Mol. Phylogenet. Evol. 38, 381–397.

miot, C., Lorvelec, O., Mandon-Dalger, I., Sardella, A., Lequilliec, P., Clergeau, P.,2007. Rapid morphological divergence of introduced Red-whiskered BulbulsPycnonotus jocosus in contrasting environments. Ibis 149, 482–489.

nderson, G.J., Bernardello, G., López, P.S., Crawford, D.J., Stuessy, T.F., 2000. Repro-

ductive biology of Wahlenbergia (Campanulaceae) endemic to Robinson CrusoeIsland (Chile). Plant Syst. Evol. 233, 109–123.

nderson, G.J., Bernardello, G., Stuessy, T.F., Crawford, D.J., 2001. Breeding systemand pollination of selected plants endemic to Juan Fernández Islands. Am. J. Bot.88, 220–233.

gy, Ev

A

A

A

A

A

Æ

B

B

B

B

B

B

B

B

C

C

C

C

C

C

C

CC

C

C

C

C

C

C

C

D

D

D

D

D

D

E

E

E

E

F

F

F

F

FG

G

H

H

H

H

H

H

H

H

H

H

H

I

I

I

J

JJ

K

K

J.M. Olesen et al. / Perspectives in Plant Ecolo

ndrews, R.M., 1979. Evolution of life histories: a comparison of Anolis lizards frommatched island and mainland habitats. Breviora 454, 1–51.

nguita, F., Márquez, A., Castineiras, P., Hernán, F., 2002. Los volcanes de Canarias.Rueda, Madrid.

ustin, J.J., Arnold, E.N., 2006. Using ancient and recent DNA to explore relation-ships of extinct and endangered Leiolopisma skinks (Reptilia, Scincidae) in theMascarene Islands. Mol. Phylogenet. Evol. 39, 503–511.

ustin, J.J., Arnold, E.N., Jones, C.G., 2004. Reconstructing an island radiation usingancient and recent DNA, the extinct and living day geckos (Phelsuma) of theMascarene Islands. Mol. Phylogenet. Evol. 31, 109–122.

xelrod, D.I., Raven, P.H., 1972. Evolutionary biogeography viewed from plate tec-tonic theory. In: Behnke, J.A. (Ed.), Future Directions in the Life Sciences. Am.Inst. Biol. Sci., Washington DC, pp. 218–236.

gisdóttir, H., Thórhallsdóttir, T., 2006. Breeding system evolution in the Arctic,a comparative study of Campanula uniflora in Greenland and Iceland. ArcticAntarctic Alpine Res. 38, 305–312.

adré, F., 1976. Campanulacées. Flore des Mascareignes 111. Sugar Res. Inst,Mauritius.

alen, S.V., 2008. Family Zosteropidae (white-eyes). In: del Hoyo, J., Elliott, A., Sar-gatal, J. (Eds.), Handbook of Birds of the World, vol. 13. Lynx, Barcelona, pp.402–485.

ell, C.D., Soltis, D.E., Soltis, P.S., 2010. The age and diversification of the angiospermsre-visited. Am. J. Bot. 97, 1296–1303.

annerman, D.A., Bannerman, W.M., 1966. Birds of the Atlantic Islands, v. 3. A historyof the birds of the Azores. Oliver and Boyd, Edinburgh.

ernardello, G., Anderson, G.J., Stuessy, T.F., Crawford, D.J., 2001. A survey of flo-ral traits, breeding systems, floral visitors, and pollination systems of theangiosperms of the Juan Fernández Islands (Chile). Bot. Rev. 67, 255–308.

ernardello, G., Galetto, L., Anderson, G.J., 2000. Floral nectary structure and nectarchemical composition of some species from Robinson Crusoe Island (Chile). Can.J. Bot. 78, 862–872.

orges, P.A.V., Hortal, J., 2009. Time, area and isolation, factors driving the diversifi-cation of the Azorean arthropods. J. Biogeog. 36, 178–191.

rehm, A., Jesús, J., Spínola, H., Alves, C., Vicente, L., Harris, D.J., 2003. Phylogeographyof the Madeiran endemic lizard Lacerta dugesii inferred from mtDNA sequences.Mol. Phylogenet. Evol. 26, 222–230.

arracedo, J.C., Pérez, F.J., Ancochea, E., Meco, J., Hernán, F., Cubas, C.R., Casillas,R., Rodríguez, E., Ahijado, A., 2002. Cenozoic volcanism II: the Canary Islands.In: Gibbons, W., Moreno, T. (Eds.), The Geology of Spain. Geolog. Soc. London,London, pp. 439–472.

arranza, S., Arnold, E.N., 2006. Systematics, biogeography, and evolution of Hemi-dactylus geckos (Reptilia, Gekkonidae) elucidated using mitochondrial DNAsequences. Mol. Phylogenet. Evol. 38, 531–545.

arretero, M.A., Harris, D.J., Rocha, S., 2005. Recent observations of reptiles in theComoro Islands (West Indian Ocean). Herpetol. Bull. 91, 19–28.

eballos, L.F.d.C., Ortuno, F.M., 1951. Estudio sobre la vegetacion y la flora forestalde las Canarias occidentales. Ministerio Agricultura, Madrid.

ellinese, N., Smith, S.A., Edwards, E.J., Kim, S.-T., Haberle, R.C., Avramakis, M.,Donoghue, M.J., 2009. Historical biogeography of the endemic Campanulaceaeof Crete. J. Biogeog. 36, 1253–1269.

erling, T.E., 1992. Development of grasslands and savannas in East Africa duringthe Neogene. Palaeogeog. Palaeoclimatol. Palaeoecol. 97, 241–247.

heke, A.S., 1987. The ecology of smaller landbirds of Mauritius. In: Diamond, A.W.(Ed.), Studies of Mascarene Island birds. Cambridge Univ. Press, Cambridge, pp.151–207.

heke, A.S., Hume, J., 2008. Lost Land of the Dodo. T & A.D. Poyser, London.olwell, R.K., 1989. Hummingbirds of the Juan Fernández Islands: natural history,

evolution and population status. Ibis 131, 548–566.onnell, J.H., 1980. Diversity and the coevolution of competitors, or the ghost of

competition past. Oikos 35, 131–138.osner, M.E., Raubeson, L.A., Jansen, R.K., 2004. Chloroplast DNA rearrangements in

Campanulaceae: phylogenetic utility of highly rearranged genomes. BMC Evol.Biol. 4, 1471–2148.

ox, S.C., Carranza, S., Brown, R.P., 2010. Divergence times and colonization of theCanary Islands by Gallotia lizards. Mol. Phylogenet. Evol. 56, 747–757.

resswell, J.E., Robertson, A.W., 1994. Discrimination by pollen-collecting bumble-bees among differentially rewarding flowers of an alpine wildflower Campanularotundifolia L. (Campanulaceae). Oikos 69, 304–308.

ronk, Q.C.B., 1987. The history of endemic flora of St Helena: a relictual series. NewPhytol. 105, 509–520.

ronk, Q.C.B., 1997. Islands, stability, diversity, conservation. Biodiv. Conserv. 6,477–493.

ronk, Q.C.B., Ojeda, I., 2008. Bird-pollinated flowers in an evolutionary and molec-ular context. J. Exp. Bot. 5, 715–727.

alsgaard, B., Martín, A.M.G., Olesen, J.M., Timmermann, A., Andersen, L.H., Ollerton,J., 2008. Pollination networks and functional specialization, a test using LesserAntillean plant–hummingbird assemblages. Oikos 117, 789–793.

ietzen, C.H., García del Rey, E., Castro, G.D., 2008. Phylogenetic differentiation ofSylvia species (Aves, Passeriformes) of the Atlantic islands (Macaronesia) basedon mitochondrial DNA sequence data and morphometrics. Biol. J. Linnean Soc.

95, 157–174.

obat, K., Peikert-Holle, T., 1985. Blüten und Fledermäuse. Waldemar Kramer,Frankfurt am Main.

uncan, R.A., Storey, M., 1992. The life cycle of Indian Ocean hotspots. Am. Geophys.Union Monogr. 70, 91–103.

K

olution and Systematics 14 (2012) 169– 182 181

upont, Y.L., Hansen, D.M., Olesen, J.M., 2003. Structure of a plant–pollinatornetwork in the high altitude sub-alpine desert of Tenerife, Canary Islands. Ecog-raphy 26, 301–310.

uprat, H.I., Friis, J., Holm, P.M., Grandvuinet, T., Sørensen, R.V., 2007. The volcanicand geochemical development of São Nicolau, Cape Verde Islands: constraintsfrom field and 40Ar/39Ar evidence. J. Volcanol. Geothermal Res. 162, 1–19.

ddie, W.M.M., Shulkina, T., Gaskin, J., Haberle, R.C., Jansen, R.K., 2003. Phylogenyof Campanulaceae s. str. inferred from its sequences of nuclear ribosomal DNA.Ann. Missouri Bot. Gard. 90, 554–575.

lvers, I., 1978. The Madeiran lizard–flower connection observed in a natural habitat.Botaniska Notiser 131, 159–160.

merson, B.C., Gillespie, R.G., 2008. Phylogenetic analysis of community assemblyand structure over space and time. Trends Ecol. Evol. 23, 619–630.

rbar, C., Leins, P., 1989. On the early floral development and the mechanisms ofsecondary pollen presentation in Campanula, Jasione and Lobelia. Bot. Jahrb. Syst.111, 29–55.

aegri, K., Pijl, L.v.d., 1971. The Principles of Pollination Ecology. Pergamon Press,London.

eraud, G., Schimncke, H.U., Lietz, J., Gastaud, J., Pritchard, G., Bliel, U., 1981. NewK/Ar ages, chemical analyses and magnetic data of rocks from the islands ofSanta Maria (Azores), Porto Santo and Madeira (Madeira Archipelago) and GranCanaria (Canary Islands). Bull. Volcanol. 44, 359–375.

ernandes, C.A., Rohling, E.J., Siddall, M., 2006. Absence of post-Miocene Red Sealand bridges: biogeographic implications. J. Biogeog. 33, 961–966.

isk, M.R., Duncan, R.A., Baxter, A.N., Greenough, J.D., Hargraves, R.B., Tatsumi, Y.,1989. Réunion hotspot magma chemistry over the past 65 m.y.: results from Leg115 of the Ocean Drilling Program. Geology 17, 934–937.

riedmann, F., 1998. Fleurs rares des Iles Mascareignes. L’Ile aux Images, Mauritius.alloway, L.F., Etterson, J.R., Hamrick, J.L., 2003. Outcrossing rate and inbreeding

depression in the herbaceous autotetraploid Campanula americana. Heredity90, 308–315.

ill, F.B., 1971. Ecology and evolution of the sympatric Mascarene white-eyes, Zos-terops borbonica and Zosterops olivacea. Auk 88, 35–60.

aberle, R.C., Dang, A., Lee, T., Penaflor, C., Cortes-Burn, H., Oestreich, A., Rauben-son, L., Cellinese, N., Edwards, E.J., Kim, S.-T., Eddie, W.M.M., Jansen, R.K., 2009.Taxonomic and biogeographic implications of a phylogenetic analysis of theCampanulaceae based on three chloroplast genes. Taxon 58, 715–734.

all, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor andanalysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 95–98.

ansen, D.M., Beer, K., Müller, C.B., 2006. Mauritian coloured nectar no longer amystery, a visual signal for lizard pollinators. Biol. Lett. 2, 165–168.

ansen, D.M., Kiesbüy, H.C., Jones, C.G., Müller, C.B., 2007b. Positive indirect inter-actions between neighboring plant species via a lizard pollinator. Am. Nat. 169,534–542.

ansen, D.M., Müller, C.B., 2009. Pollination of the enigmatic Mauritian endemicRoussea simplex (Rousseaceae): birds or geckos? Int. J. Plant Sc. 170, 42–52.

ansen, D.M., Olesen, J.M., Mione, T., Johnson, S.D., Müller, C.B., 2007a. Colourednectar: distribution, ecology, and evolution of an enigmatic floral trait. Biol. Rev.82, 83–111.

arris, D.J., Arnold, E.N., Thomas, R.H., 1998. Relationships of the lacertid lizards(Reptilia: Lacertidae) estimated from mitochondrial DNA sequences and mor-phology. Proc. R. Soc. Lond. B 265, 1939–1948.

edberg, O., 1988. Classical taxonomy versus modern systematic studies in Africanbotany. Monogr. Syst. Bot. Missouri Bot. Garden 25, 567–573.

edberg, O., Hedberg, I., Jonsell, B., Junell, L., Lundqvist, N., Olsson, O., 1961.Monograph of the genus Canarina L. (Campanulaceae). Svensk Bot. Tidskr. 55,17–62.

elbig, A.J., Martens, J., Seibold, I., Henning, F., Schottler, B., Wink, M., 1996. Phy-logeny and species limits in the Palearctic Chiffchaff Phylloscopus collybitacomplex: mitochondrial genetic differentiation and bioacoustic evidence. Ibis138, 650–666.

olt, R.D., 2010. Towards a trophic island biogeography: reflections on the interfaceof island biogeography and food web ecology. In: Losos, J.B., Ricklefs, R.E. (Eds.),The Theory of Island Biogeography Revisited. Princeton Univ. Press, Princeton,pp. 143–185.

noue, K., 1988. Pattern of breeding-system change in the Izu Islands in Campanulapunctata: bumblebee-absence hypothesis. Plant Sp. Biol. 3, 125–128.

noue, K., 1990. Evolution of mating systems in island populations of Campanulamicrodonta: pollinator availability hypothesis. Plant Sp. Biol. 5, 57–64.

noue, K., Maki, M., Masuda, M., 1996. Evolution of Campanula flowers in relation toinsect pollinators on islands. In: Lloyd, D.G., Barrett, S.C.H. (Eds.), Floral Biology.Chapman & Hall, New York, pp. 377–400.

anzen, D.H., 1973. Sweep samples of tropical foliage insects, effects of seasons,vegetation types, elevation, time of day, and insularity. Ecology 54, 687–701.

anzen, D.H., 1985. On ecological fitting. Oikos 45, 308–310.ohnson, T.H., Stattersfield, A.J., 2008. A global review of island endemic birds. Ibis

132, 167–180.atoh, K., Kuma, K., Toh, H., Miyata, T., 2005. MAFFT version 5: improvement in

accuracy of multiple sequence alignment. Nucleic Acids Res. 33, 511–518.im, S.C., McGowen, M.R., Lubinsky, P., Barber, J.C., Mort, M.E., Santos-Guerra, A.,

2008. Timing and tempo of early and successive adaptive radiations in Mac-aronesia. PLoS One 3 (5), e2139.

ishino, H., Thorne, J.L., Bruno, W.J., 2001. Performance of a divergence time estima-tion method under a probabilistic model of rate evolution. Mol. Biol. Evol. 18,352–361.

1 gy, Ev

K

K

L

L

L

L

M

M

M

M

MM

M

M

N

N

N

N

N

O

O

O

O

O

O

O

O

OO

O

O

O

P

P

P

P

P

P

R

R

R

R

R

R

R

R

R

R

R

R

R

S

S

S

S

S

T

T

T

T

V

V

V

W

W

W

W285–287.

82 J.M. Olesen et al. / Perspectives in Plant Ecolo

uck, P., Meusemann, K., 2010. FASconCAT, version 1.0. ZoologischesForschungsmuseum Alexander Koenig, Bonn.

uschel, G., 1952. Los insectos de las islas Juan Fernández. Introducción. Rev. Chil.Entomol. 2, 3–6.

ancucka-Srodoniowa, M., 1979. Macroscopic plant remains from the freshwaterMiocene of the Nowy Sacz Basin (West Carpathians, Poland). Acta Palaeobot.20, 74–75.

innebjerg, J.F., Hansen, D.M., Olesen, J.M., 2009. Gut passage effect of the introducedRed-whiskered Bulbul (Pycnonotus jocosus) on germination of invasive plantspecies in Mauritius. Aust. Ecol. 34, 272–277.

ord, J.M., 2008. A test for phylogenetic conservatism in plant–pollinator relation-ships in Australian and New Zealand alpine floras. N. Z. J. Bot. 46, 367–372.

osos, J.B., Ricklefs, R.E., 2009. Adaptation and diversification on islands. Nature 457,830–836.

acArthur, R.H., Diamond, J.M., Karr, J.R., 1972. Density compensation in islandfaunas. Ecology 53, 330–342.

acArthur, R.H., Wilson, E.O., 1967. The Theory of Island Biogeography. PrincetonUniv. Press, Princeton.

addison, W.P., Maddison, D.R., 2009. Mesquite: A Modular System for EvolutionaryAnalysis. v.2.72., http://mesquiteproject.org.

andon-Dalger, I., Clergeau, P., Tassin, J., Rivière, J.N., Gatti, S., 2004. Relationshipsbetween alien plants and an alien bird species on Réunion Island. J. Trop. Ecol.20, 635–642.

cVaugh, R., 1948. Generic status of Triodanis and Specularia. Rhodora 50, 38–49.enezes de Sequeira, M., Jardim, R., Silva, M., Carvalho, L., 2007. Musschia isambertoi

M. Seq., R. Jardim, M. Silva and L. Carvalho (Campanulaceae): a new species fromthe Madeira Archipelago (Portugal). Ann. Jardín Bot. Madrid 64, 135–146.

itchell-Thomé, R.C., 1985. Radiometric studies in Macaronesia. Boletim MuseuMunicipal Funchal 167, 52–85.

otala, S.M., Krell, F.T., Mungroo, Y., Donovan, S.E., 2007. The terrestrial arthro-pods of Mauritius, a neglected conservation target. Biodivers. Conserv. 16,2867–2881.

yhagen, D.F., Kragelund, C., Olesen, J.M., Jones, C.G., 2001. Insular interactionsbetween lizards and flowers: flower visitation by an endemic Mauritian gecko.J. Trop. Ecol. 17, 755–761.

yhagen, D.F., Turnbull, S.D., Olesen, J.M., Jones, C.G., 2005. An investigation intothe role of the Mauritian flying fox, Pteropus niger, in forest regeneration. Biol.Conserv. 122, 491–497.

ylander, J.A.A., 2004. MrModeltest v2. Evol. Biol. Centre, UppsalaUniv, Uppsala, program distributed by author, http://www.ebc.uu.se/systzoo/staff/nylander.html.

ylander, J.A.A., Wilgensbusch, J.C., Warren, D.L., Swofford, D.L., 2008. AWTY (arewe there yet?): a system for graphical exploration of MCMC convergence inBayesian phylogenetics. Bioinformatics 24, 581–583.

yman, Y., 1992. Pollination mechanisms in six Campanula species (Campanu-laceae). Plant Syst. Evol. 181, 97–108.

’Brian, J., Mariani, C., Olson, L., Russell, A.L., Say, L., Yoder, A., Hayden, T.J., 2009.Multiple colonization of the Western Indian Ocean by Pteropus fruit bats(Megachiroptera, Pteropodidae): the furthest islands were colonized first. Mol.Phylogenet. Evol. 51, 294–303.

ganesian, M.E., 1985. Structural features of seeds. I The Campanula and Symphyan-dra (Campanulaceae). Bot. Zhurn. (Moscow & Leningrad) 70, 947–955.

lesen, J.M., 1985. The Macaronesian bird-flower element and its relation to birdand bee opportunists. Bot. J. Linnean Soc. 91, 395–414.

lesen, J.M., 2003. Island pollinators. In: Carvalho, M.A.A.P.d., Costa, G.P., Jesus,J.A., Rodrigues, D.M.M (Eds.), Island Ecosystems: Conservation and MolecularApproach. Centre Biol. Geol. Sci., Funchal, pp. 45–86.

lesen, J.M., Carstensen, D.W., Dupont, Y.L., 2011. Island pollination: ways to solvethe species poverty problem. In: Pérez-Mellado, V., Ramon, M.M. (Eds.), Islandsand Evolution, vol. 19. Institut Menorquí d’Estudis, Recerca, pp. 147–169.

lesen, J.M., Eskildsen, L.I., Venkatasamy, S., 2002a. Invasion of pollination net-works on oceanic islands: importance of invader complexes and endemic supergeneralists. Div. Distr. 8, 181–192.

lesen, J.M., Jordano, P., 2002. Geographic patterns in plant/pollinator mutualisticnetworks. Ecology 83, 2416–2424.

lesen, J.M., Rønsted, N., Tolderlund, U., Cornett, C., Mølgaard, P., Madsen, J., Jones,C.G., Olsen, C.E., 1998. Mauritian red nectar remains a mystery. Nature 393, 529.

lesen, J.M., Valido, A., 2003a. Bird pollination in Madeira Island. Ardeola 50, 65–67.lesen, J.M., Valido, A., 2003b. Lizards as pollinators and seed dispersers: an island

phenomenon. Trends Ecol. Evol. 18, 177–181.lesen, J.M., Valido, A., 2004. Lizards and birds as generalized pollinators and

seed dispersers of island plants. In: Fernández-Palacios, J.M., Morici, C. (Eds.),Island Ecosystems: Conservation and Molecular Approach. Asociación EspanolaEcología Terrestre (AEET) Excmo, La Palma, pp. 229–249.

lesen, J.M., Valido, A., Dupont, Y.L., 2002b. Polinizacíon de plantas canarias. ElIndiferente 13, 18–29.

llerton, J., Cranmer, L., Stelzer, R.J., Sullivan, S., Chittka, L., 2009. Bird pollination ofCanary Island endemic plants. Naturwissenschaften 96, 221–232.

ark, J.-M., Kovacic, S., Liber, Z., Eddie, W.M.M., Schneeweiis, G.M., 2006. Phylogenyand biogeography of isophyllous species of Campanula (Campanulaceae) in theMediterranean area. Syst. Bot. 31, 862–880.

etterson, J.A., 1997. Revision of the genus Wahlenbergia (Campanulaceae) in NewZealand. N. Z. J. Bot. 35, 9–54.

lesner, S., Holm, P.M., Wilson, J.R., 2003. 40Ar–39Ar geochronology of Santo Antãom,Cape Verde Islands. J. Volcanol. Geotherm. Res. 120, 103–121.

Y

Y

olution and Systematics 14 (2012) 169– 182

ortugal-Ferreira, M., Macedo, C., Ferreira, J., 1988. K–Ar geochronology in the Sel-vagens, Porto Santo and Madeira Islands (Eastern Central Atlantic): a 30 m.y.spectrum of submarine and subaerial volcanism. Abstr. Lunar Planet. Sci. Conf.19, 325–326.

rimack, R.B., 1983. Insect pollination in the New Zealand mountain flora. N. Z. J.Bot. 21, 317–333.

roctor, M., Yeo, P., Lack, A., 1996. The Natural History of Pollination. HarperCollins,London.

ichardson, I.B.K., 1979. A distinctive new species of Wahlenbergia (Campanulaceae)from Mauritius. Kew Bull. 33, 547–550.

icklefs, R.E., Schluter, D. (Eds.), 1993. Species Diversity in Ecological Communities:Historical and Geographical Perspectives. Univ. Chicago Press, Chicago.

ocha, S., Posada, D., Carretero, M.A., Harris, D.J., 2007. Phylogenetic affinities ofComoroan and East African day geckos (genus Phelsuma): multiple naturalcolonizations, introductions and island radiations. Mol. Phylogenet. Evol. 43,685–692.

odríguez, F., Oliver, J.L., Marín, A., Medina, J.R., 1990. The general stochastic modelof nucleotide substitution. J. Theor. Biol. 142, 485–501.

odríguez-Gironés, M.A., Santamaria, L., 2004. Why are so many bird flowers red?PLoS Biol. 2, 1515–1519, e350.

odríguez-Rodríguez, M.C., Valido, A., 2008. Opportunistic nectar-feeding birds areeffective pollinators of bird-flowers from Canary Islands: experimental evidencefrom Isoplexis canariensis (Scrophulariaceae). Am. J. Bot. 95, 1408–1415.

odríguez-Rodríguez, M.C., Valido, A., 2011. Consequences of plant–pollinator andfloral–herbivore interactions on the reproductive success of the Canary Islandsendemic Canarina canariensis (Campanulaceae). Am. J. Bot. 98, 1465–1474.

ohling, E.J., Fenton, M., Jorissen, F.J., Bertrand, P., Ganssen, G., Caulet, J.P., 1998.Magnitudes of sea-level lowstands of the past 500,000 years. Nature 394,162–165.

onquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

oquet, C., Sáez, L., Aldasoro, J.J., Susanna, A., Alarcón, M.L., García-Jacas, N., 2008.Natural delineation, molecular phylogeny and floral evolution in Campanula.Syst. Bot. 33, 203–217.

oquet, C., Sanmartín, I., García-Jacas, N., Sáez, L., Susanna, A., Wikström, N., Alda-soro, J.J., 2009. Reconstructing the history of Campanulaceae with a Bayesianapproach to molecular dating and dispersal-variance analyses. Mol. Phylogenet.Evol. 52, 575–587.

oy, M.S., Torres-Mura, J.C., Hertel, F., 1998. Evolution and history of humming-birds (Aves, Trochilidae) from the Juan Fernández Islands, Chile. Ibis 140,265–273.

utschmann, F., 2005. Bayesian Molecular Dating Using PAML/multidivtime. A Step-by-Step Manual. Univ. Zurich, Zürich, http://www.plant.ch.

hetler, S.G., Morin, N.R., 1986. Seed morphology in North American Campanulaceae.Ann. Missouri Bot. Garden 73, 653–688.

kottsberg, C., 1928. Pollinationsbiologie und Samenverbreitung auf den Juan Fer-nández Inseln. In: Skottsberg, C. (Ed.), The Natural History of the Juan Fernándezand Easter Islands, v. 2, Botany. Almqvist and Wiskell, Uppsala, pp. 763–792.

kov, C., 1999. Specialised solitary bees in Denmark – a field study. MSc thesis,Aarhus Univ., Aarhus.

trahm, W.A., 1993. The conservation and restoration of the flora of Mauritius andRodrigues. PhD thesis, Univ. Reading, Reading.

tamatakis, A., 2008. The RAxML 7.0.4 Manual, available athttp://icwww.epfl.ch/∼stamatak/.

aberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers for amplificationof three non-coding regions of chloroplast DNA. Plant Mol. Biol. 17, 1105–1109.

horne, J.L., Kishino, H., 2002. Divergence time and evolutionary rate estimationwith multilocus data. Syst. Biol. 51, 689–702.

horne, J.L., Kishino, H., Painter, I.S., 1998. Estimating the rate of evolution of therate of molecular evolution. Mol. Biol. Evol. 15, 1647–1657.

rent, J.A., 1940. Floral variations in Spercularia perfoliata (L.) A. DC. Am. MidlandNat. 23, 448–454.

alido, A., Dupont, Y.L., Hansen, D.M., 2002. Native birds and insects, and introducedhoney bees visiting Echium wildpretii (Boraginaceae) in the Canary Islands. ActaOecol. 23, 413–419.

alido, A., Dupont, Y.L., Olesen, J.M., 2004. Bird–flower interactions in the Macarone-sian Islands. J. Biogeog. 31, 1945–1953.

inson, J.M., 1975. Notes on the reptiles of Round Island. Mauritius Inst. Bull. 8,49–67.

arren, B.H., Bermingham, E., Bowie, R.C.K., Prys-Jones, R.P., Thébaud, C., 2003.Molecular phylogeography reveals island colonization history and diversi-fication of Western Indian Ocean sunbirds (Nectarinia, Nectariniidae). Mol.Phylogenet. Evol. 29, 67–85.

arren, B.H., Bermingham, E., Prys-Jones, R.P., 2006. Immigration, species radiationand extinction in a highly diverse songbird lineage: white-eyes on Indian Oceanislands. Mol. Ecol. 15, 3769–3786.

arren, B.H., Bermingham, E., Prys-Jones, R.P., Thébaud, C., 2005. Tracking islandcolonization history and phenotypic shifts in Indian Ocean bulbuls (Hypsipetes,Pycnonotidae). Biol. J. Linnean Soc. 85, 271–287.

ilson, E.O., 1973. The ants of Easter Island and Juan Fernández. Pac. Insects 15,

ang, Z., 1997. PAML: a program package for phylogenetic analysis by maximumlikelihood. Cabios Appl. Notes 13, 555–556.

eo, P.F., 1993. Secondary Pollen Presentation: Form, Function and Evolution.Springer, Wien.

ESM 1. Table 1. Sequences of Campanulaceae extracted from Genbank: marker, accession number and length. Study island bellflowers are marked in grey.

Species trnLF rbcL matK atpB

accession number

sequence length

accession number

sequence length

accession number

sequence length

accession number

sequence length

Abrophyllum ornans GQ984068 413 AF299090 1504 GQ983653 1193 AJ318962 1492 Adenophora stricta subsp. confusa - - AY655145 1375 - - - - Adenophora divaricata - - EU713430 1361 EU713323 1455 EU437656 1411 Adenophora remotiflora EF088693 868 EU643715 1344 - - - - Asyneuma anthericoides GQ254891 920 - - - - - - Asyneuma campanuloides GQ254895 940 - - - - - - Asyneuma canescens GQ254899 925 - - - - - - Asyneuma canescens EF088708 824 Asyneuma comosiforme GQ254900 914 - - - - - - Asyneuma limoniifolium GQ254905 923 EU643704 1344 - - - - Asyneuma lobelioides FJ426568 878 EU643733 669 - - - - Asyneuma virgatum - - EU713439 1375 EU713332 1530 EU437665 1467 Asyneuma pichleri GQ254918 931 - - - - - - Asyneuma trichocalycinum GQ254926 948 - - - - - - Azorina vidalii EF088696 824 EU713373 1366 EU713266 1474 EU437601 1432 Berenice arguta - - EU713446 1340 EU713339 1409 EU437672 1451 Campanula abietina EF088697 884 - - - - - - Campanula affinis EF088698 864 FJ587240 699 - - - - Campanula aizoides - - EU713436 1402 EU713329 1459 EU437662 1488 Campanula alaskana GQ244675 553 - - - - - - Campanula alliariifolia EF088700 865 EU713376 1351 EU713269 1449 EU437604 1464 Campanula alpestris EF213141 901 - - - - - - Campanula andrewsii EF088701 868 - - - - - - Campanula aparinoides EF088702 867 EU643728 1344 - - - - Campanula argaea EF088703 868 - - - - - - Campanula armena EF088704 867 FJ587242 701 - - - - Campanula arvatica - - EU713451 1376 EU713344 1533 EU437677 1466 Campanula asperuloides DQ356170 946 DQ356117 1215 - - - - Campanula aucheri - - EU713348 1363 EU713241 1530 EU437576 1476

Campanula autraniana GQ254908 877 - - - - - - Campanula balfourii EF088705 824 FJ587243 701 - - - - Campanula bellidifolia EF088706 863 FJ587244 1358 EU713240 1480 EU437575 1486 Campanula betulifolia EF213142 892 - - - - - - Campanula bononiensis - - EU713381 1354 EU713274 1445 EU437609 1441 Campanula carpatha - - EU713368 1402 EU713261 1459 EU437596 1497 Campanula carpatica - - EU713410 1331 EU713303 1533 - - Campanula cephallenica FJ426576 875 - - - - - - Campanula chamissonis EF088709 852 EU643724 1344 - - - - Campanula choruhensis GQ254909 888 - - - - - - Campanula cochleariifolia EF088710 861 FJ587247 702 - - - - Campanula collina EF213143 893 - - - - - - Campanula conferta EF088712 867 FJ587249 1351 - - - - Campanula coriacea EF088713 867 - - - - - - Campanula cretica - - EU713437 1395 EU713330 1459 EU437663 1497 Campanula creutzburgii EF088714 821 EU713397 1402 EU713290 1459 EU437625 1497 Campanula cymbalaria EF088715 902 - - - - - - Campanula dasyantha EF213144 880 - - - - - - Campanula debarensis FJ426575 875 - - - - - - Campanula decumbens EF088716 872 FJ587250 1358 - - - - Campanula dichotoma EF088717 823 FJ587251 1358 - - - - Campanula dimorphantha - - FJ587246 701 - - - - Campanula divaricata EF088718 861 EU713450 1335 EU713343 1455 EU437676 1458 Campanula drabifolia EF088719 821 FJ587252 701 - - - - Campanula edulis - - EU713374 1331 EU713267 1500 EU437602 1399 Campanula elatines FJ426577 864 EU713438 1375 AJ430387 1680 EU437664 1467 Campanula elatinoides FJ426578 868 - - - - - - Campanula erinus EF088720 821 EU713398 1402 EU713291 1459 EU437626 1497 Campanula exigua - - EU713416 1277 EU713309 1457 EU437643 1436 Campanula fastigiata EF088721 865 EU643727 1344 - - - - Campanula fenestrellata subsp. istriaca FJ426584 890 - - - - - - Campanula filicaulis EF088722 822 FJ587253 701 - - - - Campanula foliosa EF088723 868 FJ587254 701 - - - - Campanula fragilis FJ426580 867 EU713428 1363 EU713321 1533 EU437655 1387 Campanula fruticulosa EF088724 867 EU643716 1344 - - - - Campanula garganica EF213145 937 FJ587255 1358 - - - - Campanula grossheimii - - EU713393 1363 EU713286 1492 EU437621 1397 Campanula haradjanii EF088726 884 - - - - - -

Campanula hawkinsiana EF213146 880 EU713445 1253 EU713338 1524 EU437671 1391 Campanula herminii - - EU713447 1335 EU713340 1533 EU437673 1397 Campanula hierapetrae - - EU713395 1402 EU713288 1456 EU437623 1496 Campanula hofmannii EF088727 868 - - - - - - Campanula incurva EF088728 868 FJ587256 1358 - - - - Campanula involucrata EF088729 871 FJ587257 716 - - - - Campanula isophylla FJ426583 868 - - - - - - Campanula jacquinii - - EU713448 1421 EU713341 1459 EU437674 1497 Campanula karakuschensis EF088730 887 FJ587258 662 - - - - Campanula laciniata - - EU713351 1402 EU713244 1459 EU437579 1497 Campanula lactiflora FJ589212 875 EU643703 1344 EU713318 1530 EU437652 1399 Campanula lanata EF088731 871 EU713382 1337 EU713275 1499 EU437610 1439 Campanula lasiocarpa AB219602 913 - - - - - - Campanula latifolia DQ356169 945 EF141027 1402 EU713271 1530 EU437606 1429 Campanula ledebouriana GQ254910 883 - - - - - - Campanula lusitanica EF088733 879 EU713441 1348 EU713334 1533 EU437667 1456 Campanula lyrata EF088734 868 - - - - - - Campanula macrochlamys EF088735 877 - - - - - - Campanula macrostachya EF088736 859 - - - - - - Campanula macrostyla EF088737 867 EU643722 1344 - - - - Campanula medium EF088738 864 FJ587261 1358 EU713272 1498 EU437607 1415 Campanula mirabilis - - EU713384 1357 EU713277 1502 EU437612 1428 Campanula mollis EF088739 823 EU713375 1353 EU713268 1530 EU437603 1403 Campanula moravica EF088740 868 FJ587262 1358 - - - - Campanula olympica EF088741 883 FJ587263 1358 - - - - Campanula parryi EF213147 873 - - EU713342 1533 EU437675 1456 Campanula patula EF213148 917 - - - - - - Campanula pelviformis - - EU713350 1402 EU713243 1459 EU437578 1492 Campanula peregrina EF088742 879 EU713427 1345 EU713320 1530 EU437654 1422 Campanula persicifolia EF213149 903 FJ587264 1358 EU713324 1527 EU437657 1414 Campanula pinatzii EF088744 826 EU713396 1421 EU713289 1459 EU437624 1497 Campanula pinnatifida EF088745 864 - - - - - - Campanula polyclada EF088746 824 - - - - - - Campanula portenschlagiana FJ426587 876 - - - - - - Campanula poscharskyana EF088747 890 FJ587266 1358 - - - - Campanula prenanthoides EF088748 878 - - - - - - Campanula primulifolia EF088699 887 EU643718 1344 - - - - Campanula propinqua EF088749 867 FJ587267 1358 - - - -

Campanula ptarmicifolia var. ptarmicifolia EF088750 844 EU643710 1344 - - - - Campanula pterocaula EF088751 882 FJ587268 1358 - - - - Campanula pubicalyx EF088752 867 EU643717 1344 - - - - Campanula punctata EF088753 868 EU643725 1344 - - - - Campanula pyramidalis GQ254919 948 EU713429 1325 EU713322 1533 - - Campanula quercetorum EF088755 864 FJ587269 1358 - - - - Campanula radicosa EF213151 890 - - - - - - Campanula radula EF088756 864 FJ587270 1358 - - - - Campanula ramosa - - CMZCHLORO 1428 - - - - Campanula rapunculoides EF213152 879 FJ587271 1358 EU713285 1450 EU437620 1400 Campanula rapunculus EF088758 877 - - - - - - Campanula reatina FJ426589 815 - - - - - - Campanula reverchonii - - EU713366 1345 EU713259 1530 EU437594 1444 Campanula robinsiae - - EU713415 1343 EU713308 1479 EU437642 1428 Campanula rotundifolia EF213153 889 EU713443 1402 EU713336 1503 EU437668 1497 Campanula rumeliana EF088778 915 EU643726 1344 EU713284 1530 EU437619 1427 Campanula sarmatica GQ254921 890 EU713386 1345 EU713279 1456 EU437614 1450 Campanula savalanica EF088760 868 - - - - - - Campanula saxatilis - - EU713349 1402 EU713242 1459 EU437577 1497 Campanula saxifraga - - FJ587274 1358 - - - - Campanula saxifraga subsp. aucheri EF088761 880 - - - - - - Campanula scheuchzeri EF088762 870 FJ587275 701 - - - - Campanula sclerotricha EF088763 867 FJ587276 701 - - - - Campanula scoparia EF088764 868 FJ587277 1358 - - - - Campanula scouleri - - EU713452 1335 EU713345 1455 EU437678 1436 Campanula scutellata EF088765 863 - - - - - - Campanula secundiflora GQ254922 827 - - - - - - Campanula semisecta EF088766 862 FJ587278 624 - - - - Campanula seraglio EF213156 914 - - - - - - Campanula sibirica EF213157 927 FJ587279 1358 - - - - Campanula spatulata subsp. filicaulis - - EU713444 1402 EU713337 1459 EU437670 1497 Campanula sparsa subsp. sphaerotrix EF213159 920 - - - - - - Campanula speciosa EF088768 864 FJ587280 701 - - - - Campanula spicata EF088769 864 FJ587281 1358 - - - - Campanula stevenii - - FJ587282 1358 - - - - Campanula stevenii subsp. stevenii EF088770 794 - - - - - - Campanula stricta EF088771 868 FJ587283 1358 - - - - Campanula strigosa EF088772 869 - - - - - -

Campanula subcapitata EF088773 863 FJ587284 1358 - - - - Campanula thyrsoides - - EU643723 1344 - - - - Campanula tommasiniana GQ254923 922 - - - - - - Campanula trachelium DQ356171 943 FJ587285 1358 - - AJ235423 1482 Campanula tridentata EF213160 901 - - - - - - Campanula tubulosa - - EU713352 1402 EU713245 1459 EU437580 1497 Campanula tymphaea EF213162 900 - - - - - - Campanula uniflora FJ426574 869 - - - - - - Campanula versicolor FJ426591 875 - - - - - - Campanula waldsteiniana GQ254927 926 - - - - - - Campanula witasekiana EF213164 894 - - - - - - Campanulastrum americanum GQ254928 838 EU713419 1372 EU713312 1530 EU437646 1350 Canarina canariensis DQ356115 1402 EU713246 1530 EU437581 1444 Codonopsis dicentrifolia - - EU713357 1376 EU713250 1529 EU437585 1436 Codonopsis kawakamii - - EU713360 1315 EU713253 1484 EU437588 1439 Codonopsis lanceolata - - EU713355 1374 EU713248 1530 EU437583 1450 Codonopsis ovata - - CDZCPRBCL 1422 - - - - Codonopsis pilosula - - GQ436426 703 - - AJ236202 1497 Codonopsis viridis - - EU713356 1375 EU713249 1421 EU437584 1467 Craterocapsa tarsodes - - EU713408 1330 EU713301 1524 EU437636 1429 Cuttsia viburnea GQ984071 403 Y10676 1428 GQ983640 1229 AJ318969 1469 Cyananthus lobatus - - CNQCPRBCL 1422 EU713252 1459 EU437587 1467 Edraianthus australis EF213172 920 - - - - - - Edraianthus dalmaticus EF213188 914 - - - - - - Edraianthus dinaricus EF213189 917 - - - - - - Edraianthus glisicii EF213190 927 - - - - - - Edraianthus graminifolius EF213271 921 EU713380 1369 EU713273 1456 - - Edraianthus horvatii EF213277 918 - - - - - - Edraianthus niveus EF213280 917 - - - - - - Edraianthus owerinianus EF213283 880 - - - - - - Edraianthus parnassicus EF213284 916 - - - - - - Edraianthus pumilio EF213289 923 - - - - - - Edraianthus serbicus EF213290 877 - - - - - - Edraianthus serpyllifolius EF213304 923 - - - - - - Edraianthus siculus EF213317 917 - - - - - - Edraianthus tarae EF213321 919 - - - - - - Edraianthus tenuifolius EF213330 922 - - - - - - Edraianthus wettsteinii subsp. lovcenicus EF213337 919 - - - - - -

Feeria angustifolia EF088780 880 EU713394 1369 EU713287 1529 EU437622 1375 Githopsis diffusa - - EU713417 1277 EU713310 1457 EU437644 1440 Githopsis pulchella - - EU713420 1338 EU713313 1375 EU437647 1459 Hanabusaya asiatica - - EU713432 1364 EU713325 1533 EU437658 1418 Heterocodon rariflorum - - EU713414 1310 EU713307 1454 EU437641 1455 Jasione crispa - - EU713390 1327 EU713283 800 EU437618 1445 Jasione heldreichii EF088781 882 EU713388 1375 EU713281 1453 EU437616 1467 Jasione laevis - - EU713387 1328 EU713280 1524 EU437615 1441 Jasione montana DQ356174 955 EU713354 1330 EU713247 1524 EU437582 1407 Legousia falcata - - AY655152 1375 EU713311 1459 EU437645 1391 Legousia hybrida DQ356234 941 EU713434 1396 EU713327 1459 EU437660 1495 Legousia pentagonia - - EU713367 1428 EU713260 1459 EU437595 1494 Legousia speculum-veneris - - EU713365 1316 EU713258 1459 EU437593 1497 Leptocodon gracilis - - EU713389 1293 EU713282 1455 EU437617 1379 Lobelia cardinalis DQ356231 933 AF042659 1353 GQ248149 776 EU437598 1467 Merciera brevifolia - - AM234939 1399 - - - - Michauxia campanuloides GQ254929 912 - - - - - - Michauxia tchihatchewii EF088784 868 EU643720 1344 EU713239 1424 EU437574 1450 Microcodon glomeratum - - EU713399 1353 EU713292 1480 EU437627 1461 Musschia aurea EF088785 671 AY655154 1369 EU713304 1524 EU437638 1467 Pentaphragma ellipticum - - PNRCPRBCL 1424 - - - - Petkovia orphanidea EF213341 915 - - - - - - Petromarula pinnata FJ426585 877 EU713433 1402 EU713326 1474 EU437659 1444 Physoplexis comosa FJ426586 880 EU713362 1270 EU713255 1530 EU437590 1467 Phyteuma cordatum GQ254930 935 - - - - - - Phyteuma globulariifolium FJ426582 882 - - - - - - Phyteuma spicatum EF088787 874 EU643712 1344 EU713254 1530 EU437589 1467 Platycodon grandiflorum EF088788 889 AY655156 1358 EU713251 1459 EU437586 1434 Prismatocarpus diffusus - - AY655157 1375 EU713294 1530 EU437629 1444 Prismatocarpus fruticosus - - - - EU713299 1414 EU437634 1446 Prismatocarpus schlechteri - - AM234942 1399 EU713297 1496 EU437632 1467 Prismatocarpus sessilis - - - - EU713296 1474 EU437631 1416 Rhigiophyllum squarrosum - - EU713426 1343 - - - - Roella amplexicaulis - - AM234944 1399 - - - - Roella ciliata EF088789 879 EU713405 1375 EU713298 1459 EU437633 1467 Symphyandra armena - - EU713383 1325 EU713276 1533 EU437611 1472 Symphyandra hofmannii - - EU713377 1375 EU713270 1474 EU437605 1444 Symphyandra pendula - - EU713385 1368 EU713278 1472 EU437613 1497

Symphyandra zangezura EF213342 880 - - - - - - Theilera guthriei - - EU713409 1347 EU713302 1451 EU437637 1436 Trachelium caeruleum EF088791 851 EU713435 1428 EU713328 1545 EU437661 1408 Triodanis coloradoensis - - EU713364 1336 EU713257 1499 EU437592 1439 Triodanis perfoliata - - EU713363 1390 EU713256 1459 EU437591 Wahlenbergia angustifolia - - EU713412 1361 EU713305 1444 EU437639 Wahlenbergia berteroi - - EU713423 1342 EU713316 1300 - - Wahlenbergia capensis - - AM234947 1399 - - - - Wahlenbergia gloriosa - - EU713407 1390 EU713300 1530 - - Wahlenbergia hederacea EF088792 730 EU713400 1362 EU713293 1525 - - Wahlenbergia linifolia - - EU713424 1346 EU713317 1495 - - Wahlenbergia lobelioides EF088793 839 EU643707 1344 - - - - Table 2. New sequences of Campanulaceae: locality and accession number. Species Locality rbcL

accession number

trnLF accession number

Campanula arvatica Spain, Oviedo, Muniellos, Aldasoro & Alarcón 11190 JF747591 Campanula edulis Yemen, Alrum, Podlech 3616 JF747590 Campanula herminii Spain, Avila, Puerto del Pico, Aldasoro & Alarcón 9290 JF747588 Campanula jacobaea Cape Verde, S Antao, Cova, Aldasoro & Alarcón 9291 JF747589 Canarina abyssinica Ethiopia, Birbirsa, Gara Muleta Valley, Aldasoro & Alarcón 10424 new seq JF747584 Canarina canariensis Spain, Canarias Isls, Tenerife, Palmar, Aldasoro & Alarcón 10339b new seq JF747586 Canarina eminii Ethiopia, Haranna forest, road from Negele to Goba,

Aldasoro & Alarcón 10322 new seq JF747585

Codonopsis lanceolata Japan, Hoshu, Estebanez 1515 JF747587 Heterochaenia borbonica Réunion, Hell-Bourg to Cap Anglais, 1730–1740 m, Olesen s.n. JF747579 Heterochaenia ensifolia Réunion, Cirque de Cilaos, c. 2100 m, Olesen s.n. JF747581 Heterochaenia rivalsii Réunion, SE of Piton des Neiges, c. 2400 m, Olesen s.n. JF747580 Merciera tenuifolia South Africa, Cape, Sir Lowry Pass, Aldasoro & Alarcón 9016 JF747578 Musschia wollastonii Portugal, Madeira, Ribeiro Frio, Madeira, 880 m, Olesen s.n. JF747583 Nesocodon mauritianus Mauritius, Cascade 500 Pieds, c. 400 m, Olesen s.n. JF747582

ESM 2. Phylogenetic tree of the Campanulaceae using four markers: atpB, matK, rbcL and trnL-F (see legend to Fig. 1). Posterior probabilities are above branches and bootstrap values below.