Using fossils and molecular data to reveal the origins of the Cape proteas (subfamily Proteoideae)
13
Using fossils and molecular data to reveal the origins of the Cape proteas (subfamily Proteoideae) Hervé Sauquet a,b,c, * , Peter H. Weston b , Nigel P. Barker d , Cajsa Lisa Anderson e , David J. Cantrill c,f , Vincent Savolainen a,g a Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond TW9 3DS, UK b National Herbarium of New South Wales, Botanic Gardens Trust, Mrs Macquaries Road, Sydney, NSW 2000, Australia c Department of Palaeobotany, Swedish Museum of Natural History, P.O. Box 50007, 104 05 Stockholm, Sweden d Department of Botany, Rhodes University, Grahamstown 6140, South Africa e Department of Systematic Biology, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, SE-752 36 Uppsala, Sweden f National Herbarium of Victoria, Royal Botanic Gardens, Melbourne, Private Bag 2000, South Yarra, Vic. 3141, Australia g Imperial College London, Silwood Park Campus, Buckhurst Road, Ascot, Berkshire SL5 7PY, UK article info Article history: Received 4 January 2008 Revised 10 December 2008 Accepted 11 December 2008 Available online 24 December 2008 Keywords: Molecular dating Bayesian dating Penalized likelihood Fossil calibration Cape Floristic Region Cape clades Proteaceae abstract The angiosperm family Proteaceae is a distinct component of the Cape Floristic Region biodiversity hot- spot with 330 endemic species. Phylogenetic analyses of subfamily Proteoideae using sequence data from one nuclear and six plastid loci show that most of this diversity is contained in two distinct Cape floral clades. Molecular dating analyses, using Bayesian and penalized likelihood methods and four phylogenet- ically supported fossil age constraints, reveal contrasting histories for these two clades. The genus Protea belongs to a lineage that may have been in Africa since the Late Cretaceous but began to diversify in the Cape only 5–18 Myr ago. In contrast, the Leucadendrinae clade presumably arrived in the region no earlier than 46 Myr ago by long-distance dispersal from an Australian ancestor and the extant members of this clade began to diversify in the Cape 22–39 Myr ago. These results join a growing number of case studies that challenge the commonly accepted view that most of the Cape flora radiated synchronously in the Late Miocene and Early Pliocene when a Mediterranean climate settled in the region. Ó 2008 Elsevier Inc. All rights reserved. 1. Introduction With more than 6000 endemic plant species, the Cape Floristic Region (CFR) of South Africa is both one of the most species-rich re- gions in the World and one under serious threat from human impact, which has led to its recognition as one of 34 biodiversity hotspots (Myers et al., 2000; Mittermeier et al., 2004). Although its extraordinary diversification is thought to be linked with the on- set of a Mediterranean climate in the Late Miocene (Linder et al., 1992; Goldblatt and Manning, 2000), the age and origins of its floral components remain poorly known overall. For instance, it is still un- clear whether all the largest clades in the Cape have diversified at the same time, when their ancestors first appeared in the region, and where they came from. Central to these questions is how to ob- tain reliable time estimates for important events in the phyloge- netic history of the Cape flora. Excellent progress has been made recently in developing molecular dating methods that do not as- sume a strict molecular clock (e.g., Thorne and Kishino, 2002; San- derson, 2002) and these have been applied to a number of plant clades in the Cape (Richardson et al., 2001; Goldblatt et al., 2002; Linder and Hardy, 2004; Edwards and Hawkins, 2007; Forest et al., 2007). However, partly due to the lack of a suitable fossil record for these groups, these studies have often relied on a single, distant calibration point, sometimes a secondary age estimate derived from a previous, higher-level dating study (e.g., Goldblatt et al., 2002). Since the use of multiple calibration points is now widely consid- ered to be critical for divergence time estimation when molecular rates vary (Near and Sanderson, 2004; Benton and Donoghue, 2007; Rutschmann et al., 2007), it is desirable to apply these meth- ods to at least one plant group with a good fossil record. Proteoideae, a subfamily of Proteaceae, appear to be an excel- lent candidate for several reasons. First, they are abundant in the Cape, comprising 13 genera and 331 species, 99% of which are en- demic (Table 1; Cowling and Lamont, 1998). Second, the group has left a rich and widespread track in the fossil record throughout Southern Gondwana. This has the potential to provide many cali- bration points across the family. Last, Proteoideae are flagship members of the Cape flora. They are diagnostic of the ‘fynbos’, oc- 1055-7903/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2008.12.013 * Corresponding author. Address: Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond TW9 3DS, UK. Fax: +44 2083325310. E-mail address: [email protected] (H. Sauquet). Molecular Phylogenetics and Evolution 51 (2009) 31–43 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev
Transcript of Using fossils and molecular data to reveal the origins of the Cape proteas (subfamily Proteoideae)
Molecular Phylogenetics and Evolution 51 (2009) 31–43
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier .com/locate /ympev
Using fossils and molecular data to reveal the origins of the Cape proteas(subfamily Proteoideae)
Hervé Sauquet a,b,c,*, Peter H. Weston b, Nigel P. Barker d, Cajsa Lisa Anderson e, David J. Cantrill c,f,Vincent Savolainen a,g
a Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond TW9 3DS, UKb National Herbarium of New South Wales, Botanic Gardens Trust, Mrs Macquaries Road, Sydney, NSW 2000, Australiac Department of Palaeobotany, Swedish Museum of Natural History, P.O. Box 50007, 104 05 Stockholm, Swedend Department of Botany, Rhodes University, Grahamstown 6140, South Africae Department of Systematic Biology, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, SE-752 36 Uppsala, Swedenf National Herbarium of Victoria, Royal Botanic Gardens, Melbourne, Private Bag 2000, South Yarra, Vic. 3141, Australiag Imperial College London, Silwood Park Campus, Buckhurst Road, Ascot, Berkshire SL5 7PY, UK
a r t i c l e i n f o
Article history:Received 4 January 2008Revised 10 December 2008Accepted 11 December 2008Available online 24 December 2008
Keywords:Molecular datingBayesian datingPenalized likelihoodFossil calibrationCape Floristic RegionCape cladesProteaceae
1055-7903/$ - see front matter � 2008 Elsevier Inc. Adoi:10.1016/j.ympev.2008.12.013
* Corresponding author. Address: Jodrell LaboratoryRichmond TW9 3DS, UK. Fax: +44 2083325310.
E-mail address: [email protected] (H. Sau
a b s t r a c t
The angiosperm family Proteaceae is a distinct component of the Cape Floristic Region biodiversity hot-spot with 330 endemic species. Phylogenetic analyses of subfamily Proteoideae using sequence data fromone nuclear and six plastid loci show that most of this diversity is contained in two distinct Cape floralclades. Molecular dating analyses, using Bayesian and penalized likelihood methods and four phylogenet-ically supported fossil age constraints, reveal contrasting histories for these two clades. The genus Proteabelongs to a lineage that may have been in Africa since the Late Cretaceous but began to diversify in theCape only 5–18 Myr ago. In contrast, the Leucadendrinae clade presumably arrived in the region noearlier than 46 Myr ago by long-distance dispersal from an Australian ancestor and the extant membersof this clade began to diversify in the Cape 22–39 Myr ago. These results join a growing number of casestudies that challenge the commonly accepted view that most of the Cape flora radiated synchronously inthe Late Miocene and Early Pliocene when a Mediterranean climate settled in the region.
� 2008 Elsevier Inc. All rights reserved.
1. Introduction
With more than 6000 endemic plant species, the Cape FloristicRegion (CFR) of South Africa is both one of the most species-rich re-gions in the World and one under serious threat from humanimpact, which has led to its recognition as one of 34 biodiversityhotspots (Myers et al., 2000; Mittermeier et al., 2004). Althoughits extraordinary diversification is thought to be linked with the on-set of a Mediterranean climate in the Late Miocene (Linder et al.,1992; Goldblatt and Manning, 2000), the age and origins of its floralcomponents remain poorly known overall. For instance, it is still un-clear whether all the largest clades in the Cape have diversified atthe same time, when their ancestors first appeared in the region,and where they came from. Central to these questions is how to ob-tain reliable time estimates for important events in the phyloge-netic history of the Cape flora. Excellent progress has been maderecently in developing molecular dating methods that do not as-
ll rights reserved.
, Royal Botanic Gardens, Kew,
quet).
sume a strict molecular clock (e.g., Thorne and Kishino, 2002; San-derson, 2002) and these have been applied to a number of plantclades in the Cape (Richardson et al., 2001; Goldblatt et al., 2002;Linder and Hardy, 2004; Edwards and Hawkins, 2007; Forest et al.,2007). However, partly due to the lack of a suitable fossil recordfor these groups, these studies have often relied on a single, distantcalibration point, sometimes a secondary age estimate derived froma previous, higher-level dating study (e.g., Goldblatt et al., 2002).Since the use of multiple calibration points is now widely consid-ered to be critical for divergence time estimation when molecularrates vary (Near and Sanderson, 2004; Benton and Donoghue,2007; Rutschmann et al., 2007), it is desirable to apply these meth-ods to at least one plant group with a good fossil record.
Proteoideae, a subfamily of Proteaceae, appear to be an excel-lent candidate for several reasons. First, they are abundant in theCape, comprising 13 genera and 331 species, 99% of which are en-demic (Table 1; Cowling and Lamont, 1998). Second, the group hasleft a rich and widespread track in the fossil record throughoutSouthern Gondwana. This has the potential to provide many cali-bration points across the family. Last, Proteoideae are flagshipmembers of the Cape flora. They are diagnostic of the ‘fynbos’, oc-
Table 1Proteaceae genera in South Africa and the Cape Floristic Region (classificationaccording to Weston and Barker, 2006; distribution and species numbers afterMabberley, 1997; Rourke, 1998, and Rebelo, 2001).
Genus Distribution Totalspp.
CFRspp.
GrevilleoideaeMacadamieae:Macadamiinae
Brabejum L. CFR 1 1
ProteoideaeLeucadendreae:Leucadendrinae
Diastella Salisb. exKnight
CFR 7 7
Leucadendron R.Br. CFR (and E South Africa) 85 84Leucospermum R.Br. CFR (and E South Africa to
Zimbabwe)48 45
Mimetes Salisb. CFR 13 13Orothamnus Pappeex Hook.
CFR 1 1
Paranomus Salisb. CFR 19 19Serruria Salisb. CFR 54 54Sorocephalus R.Br. CFR 11 11Spatalla Salisb. CFR 20 20Vexatorella Rourke CFR 4 4
PetrophileaeAulax Bergius CFR 3 3
ProteeaeFaurea Harv. South Africa to tropical Africa (and
Madagascar)15 1
Protea L. South Africa to tropical Africa 112 69
Total Proteaceae 393 332
32 H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43
cupy a wide range of CFR habitats, and may thus be regarded as atypical component of the Cape. Tribe Proteeae (sensu Johnson andBriggs, 1975) was listed by Linder (2003) as one of 33 Cape floralclades to foster more research on the timing and patterns of plantradiation in the region. Phylogenetic analyses have shown that thisgroup is polyphyletic and composed of three different cladesnested in subfamily Proteoideae, each with a sister group outsideAfrica (Hoot and Douglas, 1998; Barker et al., 2002, 2007; Westonand Barker, 2006; Table 1). An additional member of Proteaceaeoccurring in the Cape is Brabejum, a monotypic genus in subfamilyGrevilleoideae.
This paper aims to: (1) reconstruct a phylogeny of subfamilyProteoideae with all of the genera sampled and sequence data fromseven loci; (2) identify the biogeographic origins of the three Pro-teoideae components of the Cape flora; (3) date the origin of thesethree clades using multiple fossil age constraints and methods thatdo not assume a strict molecular clock.
2. Materials and methods
2.1. Taxon sampling
Taxon sampling for this study was designed so as to maximizecharacter sampling over the seven loci selected, while includingat least one representative of each genus of Proteoideae. This ledus to combine sequence data from different species of the samegenus where necessary (Table 2). While species-level phylogeniesare scarce in the family, we avoided, to the best of our knowledge,combining sequence data from species of genera whose mono-phyly has been questioned (e.g., Barker et al., 2002), sampling inone specific subclade of such genera as much as possible. Proteais the only genus for which enough data were available to be rep-resented by more than one terminal taxon. The lumping of Protearepens and Protea roupelliae was supported by the species-levelphylogeny of Barraclough and Reeves (2005), which showed that
these two species belong to the same subclade of Protea, to theexclusion of the other two Protea representatives in our analysis(Protea cynaroides and Protea neriifolia). Specific attention was alsopaid not to use the original atpB and atpB–rbcL sequences of Proteafrom Hoot and Douglas (1998), which have turned out to be pro-duced from a misidentified specimen of Leucadendron (see Barkeret al., 2007).
In addition to representatives of each genus of Proteoideae, wehave included both genera of Symphionematoideae, the putativesister group of Proteoideae (Weston and Barker, 2006), as well assix representatives of Grevilleoideae, one representative of Persoo-nioideae, and Bellendena. Furthermore, the two closest outgroupsof Proteaceae (Platanus and Nelumbo) as well as the putative sistergroup of Proteales (Sabiaceae) and a more distant outgroup (Buxa-ceae) were also represented by one terminal each in the analysis toensure proper rooting and calibration of the molecular dating pro-cedure. This is based on congruent evidence from numerousmolecular studies of angiosperm and eudicot phylogeny (Hootet al., 1999; Qiu et al., 2000, 2005; Soltis et al., 2000; Hilu et al.,2003; Kim et al., 2004; Worberg et al., 2007).
2.2. DNA sequencing
New ITS sequences were generated by N.P.B. and P.H.W. for thisstudy following the protocols outlined in Barker et al. (2002). NewrbcL sequences were generated by N.P.B. as in Barker et al. (2007).New matK data were produced by C.L.A. as follows. DNA was ex-tracted from herbarium or silica gel-dried leaf material usingDNeasy plant mini kit (Qiagen), following the manufacturer’sinstructions. The complete matK gene and the 30-end of the trnK in-tron were amplified by PCR using a combination of primers specif-ically designed for Proteaceae (Table 3), using either Accu PowerPCR PreMix (Bioneer) or puRETaq Ready-To-Go PCR Beads kit(Amersham biosciences) and conditions as follows: 3 min at95 �C; 35 cycles of 30 s at 95 �C, 1 min at 50 �C, and 2 min at72 �C; and 10 min at 72 �C. The PCR products were purified withthe QIAquick PCR Purification (Qiagen) or the Milllipore purifica-tion system and sequenced by Macrogen Inc., Korea (protocols atwww.macrogen.com). New rbcL and trnL intron/trnL–trnF spacersequences by H.S. were produced mostly from samples of theKew DNA Bank or silica gel-dried leaf material extracted followinga modified version of the CTAB protocol (Doyle and Doyle, 1987).rbcL was amplified using a combination of two universal and twoProteaceae-specific primers (Table 3) and PCR as follows: 2 minat 94 �C; 30 cycles of 1 min at 94 �C, 1 min at 48 �C, and 1.5 minat 72 �C; and 4 min at 72 �C. The trnL intron and trnL–trnF spacerwere amplified using the universal primers of Taberlet et al.(1991) and PCR as follows: 3 min at 94 �C; 35 cycles of 1 min at94 �C, 1 min at 52 �C, and 2 min at 72 �C; and 10 min at 72 �C.PCR products were purified using QIAquick columns and sequenc-ing reactions were conducted using 26 cycles of 10 s at 96 �C, 5 s at50 �C, and 4 min at 60 �C. Individual sequences were read by anautomated ABI 3100 capillary sequencer from Applied BiosystemsInc and assembled using Sequencher version 4.5 (Gene CodesCorporation, Ann Arbor, MI, USA).
2.3. Alignment and data set assembly
New sequences as well as selected sequences imported fromGenBank were aligned for each locus using BioEdit version 7.0.9(Hall, 1999). A global alignment was first produced using the Clu-stalW algorithm as implemented in this software and then checkedand edited by eye where necessary. Regions of noncoding DNA thatcould not be aligned unambiguously due to many overlapping inser-tion–deletions of different length were excluded from phylogeneticanalyses (mainly atpB–rbcL spacer and ITS). In addition, the particu-
Table 2Species, voucher details, and GenBank Accession Nos. for the seven loci combined in this study. Voucher information is only provided for new sequences, which are marked withthe initials of their author in brackets. Except for samples obtained directly from the Kew DNA Bank or plants cultivated in Kirstenbosch, the herbarium where the voucher isdeposited is indicated (GRA, Albany Museum, Grahamstown; K, Kew; MEL, Melbourne; NSW, New South Wales; P, Paris; S, Stockholm; UPS, Uppsala).
Taxon atpB atpB–rbcL ITS matK rbcL trnL intron trnL–trnF
Buxus microphyllaSiebold & Zucc.(Buxaceae)
EF380351 EF380351 EF380351 EF380351 EF380351 EF380351
Nelumbo mix (Nelumbonaceae) N. lutea (Willd.) Pers.AF528853
N. lutea (Willd.)Pers. EF377291
N. nucifera Gaertn.AM396514
N. lutea(Willd.) Pers.DQ182337
N. lutea (Willd.)Pers. AY145359
N. lutea (Willd.)Pers. AY145359
Platanus occidentalisL. (Platanaceae)
DQ923116 DQ923116 DQ923116 DQ923116 DQ923116 DQ923116
Sabia mix(Sabiaceae)
S. swinhoeiHemsl. exForb. &Hemsl.AF093395
S. campanulataWall. AM183436
S. japonica Maxim.AM396512
S. campanulata Wall.AM183414
S. japonica Maxim.AM397158
S. japonica Maxim.AM397158
Adenanthos mix A. obovatusLabill.AF060404
A. obovatusLabill. AF060739
A. barbiger Lindl.AF508822
A. sericeus Labill.EU169606 [CLA] KewDNA Bank 10151
A. sericeus Labill.DQ875835
Agastachys odorataR.Br.
AF060393 AF060717 EU169607 [CLA] Kew6488 (K)
DQ875824
Aulax mix A. umbellata(Thunb.) R.Br.AF060732
A. cancellata (L.)Druce EU676063[NPB]Kirstenbosch
A. umbellata (Thunb.)R.Br. EU169610 [CLA]D. Andersson s.n. (UPS)
A. cancellata (L.)Druce DQ875863
A. umbellata(Thunb.) R.Br.EU676043 [HS]Kew DNA Bank7930
A. umbellata(Thunb.) R.Br.EU676047 [HS]Kew DNA Bank7930
Austromuelleratrinervia C.T.White
AY837825 AF060720 EU676064 [PHW]NSW 418608
AY823202 DQ875865 AY823214 AY823219
Beauprea mix B. montana(Brongn. &Gris) VirotAF060402
B. montana(Brongn. & Gris)Virot AF060749
B. montana(Brongn. & Gris)Virot EU676065[NPB] NSW368725
B. spathulaefoliaBrongn. & GrisEU169612 [CLA] O.H.Selling 162a (S)
B. montana (Brongn.& Gris) VirotDQ875833
Beaupreopsispaniculata(Brongn. & Gris)Virot
EU642641 [HS] J.-C.Pintaud P387 (P)
EU676044 [HS] J.-C. Pintaud P387 (P)
EU676048 [HS] J.-C. Pintaud P387 (P)
Bellendena montanaR.Br.
AF060390 AF060715 EU676066 [PHW]NSW 418648
EU169613 [CLA] O.H.Selling 105 (S)
DQ875821
Buckinghamia mix B. celsissimaF.Muell.AF060433
B. celsissimaF.Muell.AF060742
B. celsissimaF.Muell.DQ499133
B. ferruginifloraForeman & B.HylandEU169614 [CLA] C.Wiley 2005:10 (UPS)
B. ferruginifloraForeman & B.HylandDQ875861
B. celsissimaF.Muell. AF482145
B. celsissimaF.Muell. AF482190
CarnarvoniaaraliifoliaF.Muell.
AF060407 AF060726 EU642774 EU169616 [CLA] C.Wiley 2005:7 (UPS)
DQ875838
Cenarrhenes nitidaLabill.
AF060396 AF060746 EU676067 [NPB]NSW 397495
DQ875827
Conospermum mix C. mitcheliiMeisn.AF060398
C. mitcheliiMeisn.AF060728
C. mitchelii Meisn.AF508814
C. taxifolium C.F.Gaertn.EU169617 [CLA] C.Wiley 2005:2 (UPS)
C. mitchelii Meisn.DQ875829
Diastella mix D. divaricata(Bergius) RourkeAF508836
D. divaricata (Bergius)Rourke EU169619[CLA] D. Andersson s.n.(UPS)
D. divaricata (Bergius)Rourke EU676071[NPB] KirstenboschHTC 286
D. parilis Salisb. exKnight EU676045[HS] Kew DNABank 9345
Dilobeia thouarsiiRoemer &Schultes
EU169620 [CLA]Larsson et al. L093(UPS)
EU642642 [HS]Larsson et al. L093(UPS)
EidotheazoexylocaryaA.W.Douglas &B.Hyland
AF060395 AF060714 DQ875826
Faurea mix F. macnaughtoniiE.PhillipsAJ699262
F. macnaughtoniiE.PhillipsAF508817
F. forficuliflora BakerEU169625 [CLA]Larsson et al. L094(UPS)
F. saligna Harv.EU676072 [NPB]Kirstenbosch 559/76
F. macnaughtoniiE.PhillipsAJ698320
F. macnaughtoniiE.PhillipsAJ698228
Floydia praealta(F.Muell.)L.A.S.Johnson &B.G.Briggs
AF060416 AF060713 EU676068 [PHW]NSW 368734
EU169626 [CLA] C.Wiley 2005:8 (UPS)
DQ875845 AF482147 AF482192
Franklandia fucifoliaR.Br.
AF060400 AF060721 EU169627 [CLA] A.S.George 4403 (S)
DQ875831
(continued on next page)
H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43 33
Table 2 (continued)
Taxon atpB atpB–rbcL ITS matK rbcL trnL intron trnL–trnF
Isopogon mix I. buxifoliusAF060403
I. buxifoliusAF060734
I. buxifoliusAF508818
I. anethifolius (Salisb.)Knight EU169637 [CLA]D.H. Vitt 27315 (UPS)
I. sphaerocephalusLindl. DQ875834
Leucadendron mix L. salignumBergiusAF060405
L. salignumBergiusAF060723
L. salignumBergiusAY692172
L. uniflorum PhillipsEU169641 [CLA] C.L.Anderson 5 (UPS)
L. chamelaea (Lam.)I.Williams EU642643[HS] Kew DNA Bank9346
L. chamelaea(Lam.) I.WilliamsEU676046 [HS]Kew DNA Bank9346
L. chamelaea(Lam.) I.WilliamsEU676049 [HS]Kew DNA Bank9346
Leucospermum mix L. bolusii Gand.AF508847
L. pedunculatumKlotzsch EU169642[CLA] S.L. Williams 849(S)
L. bolusii Gand.AM235083
Macadamia mix M. integrifoliaMaiden &BetcheAY837827
M. janseniiC.L.Gross &P.H.WestonAF060750
M. integrifoliaMaiden & BetcheEU642798
M. integrifolia Maiden &Betche AY823204
M. ternifolia F.Muell.U79172
M. integrifoliaMaiden & BetcheAF482140
M. integrifoliaMaiden & BetcheAF482185
Mimetes mix M. hottentoticusE.Phillips &Hutch. AF508838
M. hirtus (L.) Salisb. exKnight EU169647 [CLA]J.P. Rourke 1643 (S)
M. hottentoticusE.Phillips & Hutch.EU676073 [NPB]Kirstenbosch 61/78
Orothamnus zeyheriPappe ex Hook.
AF508810 U37132
Paranomus mix P. reflexus(E.Phillips &Hutch.) Fourc.AF508851
P. bracteolaris Salisb. exKnight EU169653 [CLA]P. Goldblatt 3075 (UPS)
P. reflexus (E.Phillips& Hutch.) Fourc.EU676074 [NPB]Kirstenbosch 1116/83
Persoonia mix P. kateraeP.H.Weston &L.A.S.JohnsonAF197615
P. lanceolata Andr.EU676069 [NPB]
P. linearis Andr.EU169654 [CLA] J.H.Ross 3460 (S)
P. falcata R.Br.EU676075 [NPB]NSW 397506
Petrophile mix P. circinataKippist exMeisn.AF060401
P. circinataKippist exMeisn.AF060735
P. canescens A.Cunn. exR.Br. EU169655 [CLA]C. Wiley 2005:3 (UPS)
P. biloba R.Br.DQ875832
Protea cynaroides(L.) L.
DQ875866 AJ699186 EU169658 [CLA] C.L.Anderson 4 (UPS)
DQ875837 AJ698243 AJ698150
Protea mix P. roupelliaeMeisn. AJ699224
P. roupelliaeMeisn. AF508816
P. repens (L.) L.U79182
P. roupelliae Meisn.AJ698282
P. roupelliae Meisn.AJ698189
Protea neriifolia R.Br. AJ699238 EU169659 [CLA] C.L.Anderson 10 (UPS)
AJ698296 AJ698203
Serruria mix S. trilophaAF508824
S. trilopha EU169663[CLA] J.P. Rourke 1590(S)
S. aemula Salisb. exKnight EU676076[NPB] Kirstenbosch816/88
Sorocephaluspinifolius (Salisb.ex Knight)Rourke
AF508829 EU676077 [NPB] J.Rourke
Spatalla incurva(Thunb.) R.Br.
AF508833 EU676078 [NPB]Kirstenbosch 999/89
Sphalmiumracemosum (C.T.White)B.G.Briggs,B.Hyland & L.A.S.Johnson
AF060408 AF060719 EU676070 [PHW]MEL 2027222
DQ875839
Stirlingia latifolia(R.Br.) Steud.
AF060397 AF060738 AF508835 EU169666 [CLA] B.H.Smith 1530 (S)
DQ875828
Symphionemamontanum R.Br.
AF060394 AF060733 AF508813 EU169667 [CLA] B.Nordenström & A.Anderberg 728 (S)
DQ875825
Synaphea mix S. mediaA.S.GeorgeAF060399
S. mediaA.S.GeorgeAF060729
S. spinulosa(Burm.f.) Merr.AF508812
S. spinulosa (Burm.f.)Merr. EU169669 [CLA]B. Nordenström & A.Anderberg 461 (S)
S. spinulosa (Burm.f.)Merr. DQ875830
Vexatorella alpina(Salisb. exKnight) Rourke
AF508853 EU169672 [CLA] J.P.Rourke 1800 (S)
EU676079 [NPB]Kirstenbosch 536/90
34 H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43
larly difficult alignment of the ITS region led us to exclude any out-groups of Proteaceae for this locus and adopt in a few fragments ofthe region a local alignment strategy, whereby blocks of similarsequences were aligned and gaps created for all remainingsequences.
The seven edited data sets were combined following a maxi-mum taxon sampling strategy (as outlined in Table 2) with missingdata coded as question marks. The total combined matrix createdhas 41 taxa and 8841 characters and is available from the authorson request.
Table 3matK and rbcL primers used in this study (primers marked with an asterisk weredesigned specifically for Proteaceae).
Primer 50–30 Sequence
matK-570F* CCTTCYTTGCATTTATTGCGAmatK-1059F* GGGTTATCTTTCGAGTGTACGACTmatK-1270F* ATCGATTTGGGCGGATATGCAAmatK-1840R* TCGGATGGAGTAGAWAATTTYCTTmatK-1765R* TGCACACGGCTTTCYCTATGTATmatK-1340R* GCACAYGAAAGTCGAAGTATATACmatK-770R** GAAATGTGTTCGYTCAAGAAGGGCTCCrbcL-1F ATGTCACCACAAACAGAAACrbcL-717R* ACCTGCYGTAGCATTCAAGrbcL-636F GCGTTGGAGAGATCGTTTCTrbcL-1369R* TCCATACCTCACAAGCAGC
H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43 35
2.4. Phylogenetic analyses
The supermatrix was analyzed using both maximum parsimonyand Bayesian statistical inference. In addition, preliminary parsi-mony analyses were conducted on each individual data set to de-tect and exclude anomalous sequences of GenBank producedfrom misidentified specimen samples. Parsimony analyses wereconducted in PAUP� version 4.0b10 (Swofford, 2002) using heuris-tic searches of 1000 random taxon addition replicates and TBR(tree bisection and reconnection) branch swapping. Branch sup-port was evaluated using 1000 bootstrap replicates each with arandom taxon addition sequence and a maximum of 100 trees.Bayesian analyses were conducted in MrBayes version 3.1.2 (Huel-senbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003)with two parallel searches of three heated and one cold chainrun for 1,000,000 generations and sampled every 100th iteration.Stationarity of the runs was checked graphically by plotting log-likelihood values against generations and numerically by examin-ing the final average standard deviation of split frequencies. Thefirst 25% trees were discarded as the ‘burn-in’ phase before calcu-lating a majority-rule consensus tree with posterior probabilities.Bayesian analyses were conducted both with a uniform GTR + C + Imodel over the entire combined data set and with a partitionedmixed model with optimal models for each data set selected withMrModeltest version 2.2 (Nylander, 2004).
2.5. Fossil calibration
The age constraints used in molecular dating analyses in thispaper were derived from phylogenetic analyses of fossil taxa usingthe data matrix and methods of Sauquet et al. (2009). Two mainapproaches have been proposed to reconstruct the phylogeneticposition of fossil taxa using both morphological characters andmolecular data (available only for extant taxa). The first is to com-bine both sources of data in a ‘‘total evidence approach” (Eernisseand Kluge, 1993; Jordan and Hill, 1999; Hermsen et al., 2006;Manos et al., 2007). The second approach, applied here, is to usethe molecular phylogeny of extant taxa as a backbone constraintin parsimony analyses of extant and fossil taxa based on morphol-ogy (method also known as ‘‘molecular scaffold approach”; Spring-er et al., 2001; Magallón, 2004; Manos et al., 2007). A summary ofthis approach and the particular fossil taxa used to constrain thepresent analyses are outlined below.
Fossil palynomorph species were selected for phylogeneticanalysis based on the availability of suitable observations (prefer-ably a combination of light, scanning electron, and transmissionelectron microscopy). Each fossil was scored for 22 pollen charac-ters documented for all extant genera of Proteaceae. Each fossilwas mapped onto the phylogeny using parsimony analyses con-ducted in PAUP� with heuristic searches of 100 replicates of ran-
dom taxon addition sequence, with the phylogeny of extantgenera (Sauquet et al., 2009) enforced as a backbone constraint. Be-cause multiple most parsimonious positions were obtained formost fossils, a simultaneous analysis of all fossils yielded too manymost parsimonious combinations to disentangle the position ofeach fossil. In addition, on the reasonable assumption that lesshomoplastic characters should be given more importance in find-ing fossil affinities (an assumption implicitly made by most pale-obotanists when considering fossil relationships without aphylogenetic analysis), a second set of parsimony analyses wasconducted after characters were reweighted by maximum valueof their rescaled consistency indices based on the extant phylog-eny. This is similar to the logic behind successive weighting (Farris,1969) except that the fossil is not allowed to participate in thereweighting scheme and the extant phylogenetic backbone is fixed.
Unweighted and weighted parsimony analyses gave similar andconsistent results, although weighted analyses occasionally re-sulted in fewer most parsimonious trees. Fossils were consideredsuitable for calibration if they were sufficiently known in thepalynostratigraphic literature (i.e., reported more than once),matched less than 10 most parsimonious positions in both un-weighted and weighted analyses, and these positions were concen-trated in a local portion of the tree. Fossils were considered toprovide a minimum age for the most recent common ancestor ofall of the most parsimonious positions. When two or more fossilswere found to provide a minimum age for the same node, the old-est fossil was chosen as a minimum age constraint. Following thestrict guidelines above, we obtained four fossil age constraintsfor the present study.
Beaupreaidites elegansiformis Cookson emend. Milne is a wellknown fossil taxon occurring in Australia, New Zealand, CampbellIsland, and the Antarctic Peninsula, with a total range of Campa-nian to Pleistocene (Pocknall and Crosbie, 1988; Askin and Baldoni,1998; Milne, 1998). It was studied by a combination of LM, SEM,and TEM by Milne (1998) and proposed to be a relative of the ex-tant New Caledonian genus Beauprea, based on its colpoid aper-tures, which are unique in Proteaceae. In our analyses, this fossilbranches consistently in the clade of Beauprea, Faurea, and Proteaor on its stem branch. It therefore allows us to give a minimumage of Campanian to the stem node of this clade.
Cranwellipollis palisadus (Couper) Martin & Harris has been re-ported from Australia, New Zealand, and the Antarctic Peninsula,with a total range of Campanian to Late Eocene (Couper, 1960;Dettmann and Jarzen, 1996; Askin and Baldoni, 1998). It was stud-ied with LM and SEM by Martin (1995) and was suggested to be apossible relative of the extant Southwest Australian genus Frank-landia, based on its unique clavate sexine. Our analyses confirmthis relationship and the fossil is therefore used as a minimumage constraint for the stem node of Franklandia.
Propylipollis annularis (Cookson) Martin & Harris is a wellknown palynomorph from Australia, New Zealand, and the Antarc-tic Peninsula, with a total range of Campanian to Late Miocene(Milne, 1994; Askin and Baldoni, 1998). It was studied in detailwith LM, SEM, and TEM by Milne (1994) and was proposed to bea close relative of either Lambertia or Xylomelum (Dettmann andJarzen, 1998), based on distinct apertures lined with abundantgranular endexine, which are synapomorphic of subfamily Grevil-leoideae (Sauquet et al., unpublished data). Our analyses do notagree with the specific affinities to the extant genera previouslysuggested but do confirm its position well nested among Grevil-leoideae (not on their stem lineage). We therefore use this fossilas a minimum age constraint for crown Grevilleoideae.
Triorites africaensis Jardiné & Magloire has been reported inIvory Coast, Senegal, and Peru, with a total range of Cenomanianto Turonian (Jardiné and Magloire, 1965; Brenner, 1968). Wardand Doyle (1994) studied this species with LM, SEM, and TEM
36 H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43
and argued that it may be the earliest known representative of Pro-teaceae. Our analyses are consistent with this and place the fossilunequivocally on the stem lineage of Proteaceae. We thereforeuse this fossil as a minimum age constraint for the stem node ofthe family.
Finally, because at least one node must be fixed or given amaximum age in most molecular dating algorithms, we use thewell documented origin of tricolpate pollen in the fossil recordnear the Late Barremian/Early Aptian boundary (Hughes andMcDougall, 1990; Doyle and Hotton, 1991; Doyle, 1992) as amaximum age constraint for the root in the present study. Thisis justified by the facts that tricolpate pollen is completely absentfrom extensively scrutinized earlier sediments and is a clear syn-apomorphy of eudicots as a whole (Doyle and Endress, 2000; Fur-ness and Rudall, 2004; Doyle, 2005), and that the most recentcommon ancestor of Proteales and Sabiaceae (the root in our tree,after Buxaceae are pruned out; see below) is probably not muchyounger than the origin of eudicots (e.g., Anderson et al., 2005,found stem-group Proteales to be only 1 Myr younger thancrown-group eudicots).
To convert stratigraphic ages into absolute ages, we used thegeological time scale of Gradstein et al. (2004). Following a conser-vative approach, we used the absolute age of the upper (youngest)bound of the oldest stage in which a fossil has been confirmed as aminimum age constraint on its assigned node.
2.6. Dating analyses
Estimation of divergence times was performed using two meth-ods that do not assume a strict molecular clock. The same set ofminimum and maximum age constraints described above wasused in all analyses.
The Bayesian autocorrelation molecular dating approach(Thorne et al., 1998; Kishino et al., 2001; Thorne and Kishino,2002) was conducted using a combination of the PAML version3.15 (Yang, 1997) and the multidistribute (Thorne and Kishino,2002) software packages following Rutschmann (2005). This meth-od allowed estimation of branch lengths for each partition of thedata set using the estbranches program and integration of the se-ven resulting sets of branch lengths all at once in multidivtime,while allowing missing taxa in individual partitions. The topologyobtained from the Bayesian partitioned analysis was used as theinput tree for all steps of the dating analysis. However, missingtaxa had to be pruned for the estimation of model parameters inthe first step (using the baseml procedure of the PAML package).The multidivtime analysis was run for 1,000,000 generations, sam-pling every 100th generation and discarding the first 100,000 gen-erations as the ‘burn-in’ phase. The prior for the age of the rootnode (rttm) was set to 125 Myr with a standard deviation (rttmsd)of 25 Myr, and bigtime was set to 150 Myr. This is justified usingthe same argument as above, that the age of the root is approxi-mated by the age of the eudicots. However, the maximum age con-straint on the root was also used in addition to these parameters,providing a more rigorous approach, given observations that rttmhas little influence on the analysis and that preliminary searcheswere effectively bound by bigtime. The prior for the rate of molec-ular evolution at the root node (rtrate) was set to 0.008 (calculatedas the average amount of evolution from the root to the tips of thetree, averaged over the seven data sets, divided by rttm) with astandard deviation (rtratesd) of 0.008.
Penalized likelihood dating (Sanderson, 2002; hereafter referredto as PL) was implemented using r8s version 1.71 (Sanderson,2003). The phylogram obtained from the Bayesian partitionedanalysis was used as the input tree, after pruning out the earli-est-diverging taxon (Buxus) for consistency with the Bayesian dat-ing analysis. A cross-validation procedure was performed to test 17
values of the smoothing parameter (s) scaling between 1 and10,000,000. A value of 10 had the lowest v2-score and was there-fore selected as the optimal smoothing parameter for the actualanalysis.
In addition to the main set of dating analyses described above,we ran various sets of complementary sensitivity analyses (usingthe same methods and parameters) to evaluate the influence ofthe fossil age constraints on our divergence time estimates. Theseanalyses are detailed in the next section.
3. Results
3.1. Phylogenetic analyses
Separate parsimony analyses of the seven individual data setsproduced mostly congruent consensus trees, some of which werepoorly resolved due to low levels of variation (e.g., rbcL; resultsnot shown). We found no significant conflict among these analysesand therefore combined all the data sets. Parsimony and Bayesiananalyses of the supermatrix produced congruent, well resolvedconsensus trees (Fig. 1). MrModeltest selected an optimal GTR +C + I model for the ITS, rbcL, and atpB partitions, and an optimalGTR + C model for the remaining four partitions. Both the unparti-tioned and the partitioned Bayesian analyses appeared to reachconvergence within 100,000 generations (lnL = �35,079.35 and�34,059.73, respectively; average standard deviation of split fre-quencies = 0.005047 and 0.016044, respectively, after 1,000,000generations) and the parsimony analysis yielded only two mostparsimonious trees (L = 4217 steps). Bootstrap support (bs) andposterior probabilities (pp) are congruent overall, although a fewbranches well supported in the Bayesian analysis received onlyweak bootstrap support. This may partly be explained by missingtaxa for individual partitions in our analysis, which increases thechance that bootstrap resampling will not sample enough informa-tive characters to correctly reconstruct the phylogenetic position ofthese taxa. The main topological differences between the parsi-mony and Bayesian analyses concern relationships within Grevil-leoideae. However, this clade of 47 genera was only representedby six terminal taxa in this study and the conflicting topologiesare very poorly supported. The unpartitioned Bayesian analysisproduced exactly the same topology as the partitioned analysis,with identical or only slightly varying posterior probabilities(results not shown).
Proteoideae appear to be monophyletic (pp = 0.95, bs = 53%)and sister to Symphionematoideae (pp = 1.00, bs < 50%). All gen-era of Proteoideae except Eidothea group into two main clades,which we refer to in this study as Proteoideae I and ProteoideaeII. The position of Eidothea at the base of Proteoideae is equivocalin the Bayesian analysis. It is sister to Proteoideae I in the strictconsensus of the parsimony analysis but with bootstrap supportless than 50%. Proteoideae I consist of Conospermeae plus threesmall genera (Beaupreopsis, Cenarrhenes, and Dilobeia). The mono-phyly of Conospermeae is strongly supported and the tribe ap-pears to be sister to a weakly supported clade of these threegenera. Proteoideae II appear to be formed of three main clades:Beauprea plus Proteeae, Franklandia plus Petrophileae, and Leuca-dendreae. While there is strong support for the monophyly ofLeucadendreae, the relationship of Beauprea and Proteeae on theone hand, and that of Franklandia and Petrophileae on the other,appear to be more unstable. However, both Proteeae and Petro-phileae are very well supported as monophyletic tribes. The rela-tionships at the base of Leucadendreae are very stronglysupported, with Adenanthos as the sister group of Leucadendrinae,and Leucadendron diverging first in Leucadendrinae. The relation-ships among the remaining genera of Leucadendrinae are onlyweakly supported.
Fig. 1. Phylogenetic relationships obtained in the partitioned Bayesian analysis of molecular sequence data from seven loci (8841 bp). Dotted branches collapse in the strictconsensus of two most parsimonious trees. Thin, vertical dotted lines represent additional or alternative relationships that appear in the strict consensus of two most parsimonioustrees. Posterior probabilities from the Bayesian analysis are indicated above branches, parsimony bootstrap values below (‘—‘ denotes branches with less than 50% bootstrapsupport). Classification according to Weston and Barker (2006). Abbreviations: GREVIL, Grevilleoideae; SYM, Symphionematoideae; Conosp, Conospermeae; Petr, Petrophileae.
H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43 37
Table 4Age estimates (in Myr) for selected nodes in the phylogeny (Fig. 2).
Node Constraint PL Multidivtime
Mean Confidence interval
A Root (crown eudicots) <125.0 125.0 123.2 118.9–124.9B Crown Proteales 116.7 119.8 113.0–124.1C Stem Proteaceae >93.5 107.2 112.8 104.7–120.1D Crown Proteaceae 79.6 85.1 79.9–91.1E Crown Proteoideae 74.8 78.7 74.3–83.7F Stem [Beauprea + Proteeae] >70.6 72.5 75.1 71.9–79.6G Stem Proteeae 68.9 72.7 67.6–77.9H Crown Proteeae 28.4 28.2 20.2–37.1I Stem Franklandia >70.6 70.6 71.9 70.6–75.0J Crown Petrophileae 55.8 55.9 47.3–64.6K Stem Leucadendrinae 34.5 37.3 29.3–46.2L Crown Leucadendrinae 28.5 30.3 22.3–39.3M Crown Grevilleoideae >70.6 70.6 73.4 70.7–78.9
38 H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43
3.2. Dating analyses
Bayesian (multidivtime) and penalized likelihood (PL) datinganalyses produced very similar divergence time estimates (Table4). In general, PL ages were slightly younger than multidivtimeaverage estimates and closer to the lower (younger) bound ofmultidivtime 95% confidence intervals. The only exception is theage inferred for the root, which is exactly at 125 Myr for PL, butyounger on average with multidivtime (123.2 Myr) with a confi-dence interval of 118.9–124.9 Myr. This indicates that both analy-ses were effectively bound by the maximum age constraint for theroot set at 125 Myr. Crown Proteales are dated at 116.7 (PL) or113.0–124.1 (multidivtime) Myr. The divergence of Platanus andProteaceae is estimated at 107.2 (PL) or 104.7–120.1 (multidiv-time) Myr. This is older than the oldest confirmed fossil record ofProteaceae (Triorites africaensis at 93.5 Myr, used as a minimumage constraint for this node). This result suggests either that theearly fossil record of Proteaceae is incomplete or that the main syn-apomorphies of Proteaceae pollen may not have evolved immedi-ately after they separated from the branch leading to extantPlatanus. Crown Proteaceae are younger at 79.6 (PL) or 79.9–91.1(multidivtime) Myr. Among the four internal minimum age con-straints used in the analyses, two effectively appear to havestretched ages down. These are the divergence of Franklandia withits sister group Petrophileae, dated at 70.6 (PL) or 70.6–75.0 (mul-tidivtime) Myr and the age of crown Grevilleoideae, dated at 70.6(PL) or 70.7–78.9 (multidivtime) Myr.
The divergence of Beauprea (New Caledonia) and Proteeae(Africa) is estimated at 68.9 (PL) or 67.6–77.9 (multidivtime) Myrand the divergence of Faurea and Protea at 28.4 (PL) or 20.2–37.1(multidivtime) Myr. The split between Aulax (South Africa) andPetrophile (Australia) is estimated at 55.8 (PL) or 47.3–64.6 (multi-divtime) Myr. Finally, the divergence of Adenanthos (Australia) andLeucadendrinae (South Africa) is estimated at 34.5 (PL) or 29.3–46.2 (multidivtime) Myr and the age of crown Leucadendrinae at28.5 (PL) or 22.3–39.3 (multidivtime) Myr.
The first set of complementary sensitivity analyses tested theinfluence of the maximum age given to the root (125 Myr). Analy-ses were repeated with this constraint set to 115 or 135 Myr, withall other parameters unchanged. As expected, the age estimates forthe root and deeper nodes in the tree shifted in the same directionas the root calibration. However, age estimates for all nodes withinProteaceae (including their crown node) varied only by a maxi-mum of 2 Myr and this difference became insignificant in moreshallow nodes of the phylogeny (Fig. 3). For instance, the age esti-mate for the divergence of Adenanthos and Leucadendrinae variedonly between 34.4 and 34.6 Myr with penalized likelihood and29.2–45.9 and 29.4–46.2 Myr with multidivtime. These resultssuggest that the age estimates for the main nodes of interest in this
study are largely unaffected by the exact value of the maximumage given to the root.
In another experiment, we removed the maximum age con-straint on the root to test the effects of the rttm (prior age of theroot node) and bigtime parameters in multidivtime. This resultedin an estimate of 138.4–149.9 Myr for the age of the root, suggest-ing that rttm (set at 125 Myr) has little influence and the analysis iseffectively bound by bigtime (set at 150 Myr). Deep nodes in thephylogeny were also affected and significantly older than in themain analysis, but again the age estimates for the shallower nodeswere hardly affected.
Finally, we tested the influence of internal minimum age con-straints by removing all of them and using a maximum age of125 Myr for the root as the only constraint in the analyses. Thisresulted in dramatically younger age estimates for all nodes abovethe root with both penalized likelihood and multidivtime. Forinstance, the divergence of Adenanthos and Leucadendrinae wasestimated at 20.8 (PL) or 10.2–22.5 (multidivtime) Myr.
4. Discussion
4.1. Phylogeny
Assuming Buxus is the most external outgroup in our analysis,the relationships among Sabia, Nelumbo, Platanus, and Proteaceaeare entirely consistent with higher-level studies of angiosperm oreudicot phylogeny (Hoot et al., 1999; Qiu et al., 2000, 2005; Soltiset al., 2000; Hilu et al., 2003; Kim et al., 2004; Worberg et al.,2007). The general structure of the Proteaceae phylogeny obtainedhere is consistent with the trees of Hoot and Douglas (1998), Wes-ton and Barker (2006), and Barker et al. (2007), with minor excep-tions. Hoot and Douglas (1998) found Bellendena to be sister to allremaining Proteaceae, but this relationship was poorly supported(bs = 61%) and our analysis strongly favors its position as sister tosubfamily Persoonioideae (here represented by Persoonia; pp =1.00, bs = 84%). The sister group relationship of Symphionematoi-deae and Proteoideae found by Hoot and Douglas (1998) and Wes-ton and Barker (2006) was challenged by Barker et al. (2007), whereGrevilleoideae instead appeared as the sister group of Proteoideae,with no significant support. Our Bayesian analysis shows strongsupport for the sister group relationship of Symphionematoideaeand Proteoideae (pp = 1.00), but parsimony remains equivocal(bs < 50%). Because this study was not designed to address relation-ships within Grevilleoideae, these will not be discussed here.
The relationships obtained within Proteoideae are also broadlyconsistent with previous analyses (Hoot and Douglas, 1998; Wes-ton and Barker, 2006; Barker et al., 2007) and are identical to thoseof Sauquet et al. (2009), reconstructed from a broader data setencompassing other subfamilies of Proteaceae. All tribes and sub-
H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43 39
tribes of Proteoideae named in the phylogenetic system of Westonand Barker (2006) are confirmed as monophyletic groups with verystrong support (pp = 1.00, bs > 98%; Fig. 1). To our knowledge, thisis the first time that the small genera Beaupreopsis (New Caledonia)and Dilobeia (Madagascar) have been sampled in any molecularphylogenetic analysis. Despite the lack of parsimony support,Bayesian posterior probability values uphold the placement ofthese two taxa in a clade with the Tasmanian endemic Cenarrhenesin Proteoideae I. The main difference with previous studies lies inthe relationships among Beauprea, Franklandia, Proteeae, Petrophi-leae, and Leucadendreae in the Proteoideae II clade, for whichvarious alternatives have been proposed but never with good sup-port or sampling of all of the genera. Though branch support in thispart of the tree remains weak, our study favors a sister relationshipof Beauprea with Proteeae and of Franklandia with Petrophileae.Interestingly, although both of these relationships are new anddo not appear obvious on macro-morphological grounds, they areconsistent with pollen characters. Beauprea and Proteeae sharecolpoid to ellipsoid elongated apertures (rather than the strictlyporate apertures characteristic of Proteaceae), while Franklandiaand Petrophileae both have distinctly convex to globose pollenwith foot layer ending abruptly at the pore margin and changinglittle in the apertural region (Sauquet and Cantrill, 2007).
The phylogeny obtained for Leucadendreae is in general consis-tent with that of Barker et al. (2002), based on ITS sequence data.The Australian genera Isopogon and Adenanthos are very stronglysupported as the two earliest-diverging taxa in the group, withAdenanthos sister to the African clade Leucadendrinae. Within Leu-cadendrinae, the large genus Leucadendron is clearly sister to therest of the genera (pp = 1.00, bs = 100%). While the ‘MOLD’ cladeof Barker et al. (2002; consisting of Mimetes, Orothamnus, Leuco-spermum, and Diastella) is well supported (pp = 1.00, bs = 91%),their ‘SSPV’ clade (Sorocephalus, Spatalla, Paranomus, and Vexatorel-la) only appears in the strict consensus of our parsimony analysisand collapses in our Bayesian analysis. However, this clade wasonly weakly supported (bs = 67%) in Barker et al. (2002). Finally,Serruria appears here to be sister to the ‘MOLD’ clade, while itwas sister to the ‘MOLD’ + ‘SSPV’ clades in Barker et al. (2002),but neither of these relationships is well supported.
4.2. Biogeographic origin of South African Proteaceae
This study implies radically different biogeographic histories forthe two main clades of Proteaceae in South Africa, Proteeae andLeucadendrinae (Fig. 2). The lineage leading to the extant Africangenera Faurea and Protea appears to have split from its New Cale-donian sister group Beauprea ca. 67–78 Myr ago (node G). Whilethis age is too young to be compatible with a scenario based onlyon vicariance between Africa and New Caledonia, it suggests thepossibility that Proteeae ancestors may have been around on theAfrican mainland for quite a long time, possibly since the Late Cre-taceous. Among the few fossils that we were able to assign confi-dently to a narrow clade of the phylogeny, Beaupreaiditeselegansiformis is of particular interest. Although this taxon hasnot been reported in South Africa, it is well known from other frag-ments of Southern Gondwana, including the Antarctic Peninsula(Campanian to Eocene; Askin and Baldoni, 1998), Australia (Eoceneto Miocene; Pocknall and Crosbie, 1988), and New Zealand (Paleo-cene to Pleistocene; Pocknall and Crosbie, 1988). Our phylogeneticanalyses have shown that this taxon belongs to the clade of Beaup-rea, Faurea, and Protea or diverged from its stem branch (hence itsuse as a minimum age constraint for the stem node of this clade;Fig. 2, node F). The distribution of this fossil suggests that this tax-on occurred over a large ancestral area of Southern Gondwana, giv-ing rise to the extant members of this group in New Caledonia(Beauprea) and Africa (Faurea and Protea) but becoming extinct in
Antarctica, Australia, and New Zealand. This is consistent withthe old crown age inferred for this clade (67–78 Myr). However,as noted earlier, this postdates the separation of Africa from therest of Southern Gondwana and therefore an old long-distance dis-persal event, perhaps in the Late Cretaceous, must be invoked toexplain the presence of Proteeae in Africa.
In contrast, the split between the stem lineage of the SouthAfrican clade Leucadendrinae and their Australian sister groupAdenanthos appears to have occurred much more recently, ca.29–46 Myr ago (node K). This age is consistent with the estimateof 39.7 ± 9.3 Myr found by Barker et al. (2007) and is robust to var-iation in method or maximum age constraint on the root (Fig. 3). Asnoted by Barker et al. (2007), this is much too young to agree withany scenario of vicariance between Africa and Australia. Simplemapping of geographic distributions onto the phylogeny usingFitch parsimony clearly shows the South African Leucadendrinaeto be nested among Australian taxa (Sauquet et al., 2009). The mostlikely explanation is therefore that the stem lineage of Leucaden-drinae arrived in South Africa after 46 Myr ago by long-distancedispersal from an Australian ancestor. At least from a morpholog-ical point of view, this is not an aberrant scenario, given that mostmembers of this group have small seeds that seem fit for long-dis-tance dispersal by wind (Rourke, 1998; Barker et al., 2007). Finally,the divergence between the small South African genus Aulax andits Australian sister group Petrophile is older, at about 47–65 Myr(node J), a younger estimate on average than that of Barker et al.(2007), who found an age of 64.9 ± 11.9 Myr for this node. Alsonested among Australian taxa, the presence of Aulax in South Africais most likely also explained by a long-distance dispersal eventfrom an Australian ancestor. This scenario is again facilitated bythe wind dispersal capabilities of seeds of extant members of bothAulax and Petrophile (Foreman, 1995; Rourke, 1998).
4.3. Diversification of Proteoideae in the Cape
This study dates the divergence of Faurea and Protea at about20–37 Myr ago and the crown group of Protea around 5–18 Myr.In an analysis based on multiple plastid loci and AFLP data, Barrac-lough and Reeves (2005) obtained a phylogeny of 86 species of Pro-tea. Their results clearly indicated that the subtropical and tropicalspecies of the genus formed a clade nested among CFR species andthey concluded that Protea originated in the Cape. Their analysisrecovered a different topology for the three selected representa-tives of the genus in this study (Fig. 1). However, according tothe tree of Barraclough and Reeves (2005), the most recent com-mon ancestor of these three representatives is near the basal nodeof Protea. We may therefore approximate that the beginning of thediversification of the genus in terms of extant species is dated bythe crown-group age of Protea in the present study. Our resultsplace this diversification at the time of final establishment of awinter-rainfall regime in the region in the Late Miocene. Barrac-lough and Reeves (2005) found a comparable age estimate of ca.15 Myr.
In contrast, the age inferred for the crown group of Leucaden-drinae, which marks the beginning of the diversification of thesecond significant clade of Cape Proteaceae, is much older at22–39 Myr (node L; Fig. 2). However, if we exclude the genusLeucadendron, the diversification of its sister clade of nine generadid not start until at least 10 Myr later, around 12–27 Myr ago. Thisis still older than the origin of the modern Mediterranean climatein the region but implies a long lapse period of at least 17 Myr be-tween the presumed arrival of Leucadendrinae in South Africa andthe diversification of this clade of nine genera. Using ITS sequencedata, Barker et al. (2004) produced a near-complete species-levelphylogeny of Leucadendron. Due to low molecular divergence, bothresolution and branch support were weak overall, yet their results
Fig. 2. Chronogram obtained from the multidivtime dating analysis using molecular sequence data from seven loci (with the topology shown in Fig. 1) and five ageconstraints based on fossil evidence (see text). Black circles denote the nodes on which age constraints were applied. White circles denote nodes of interest discussed in thetext (see also Table 4). Areas shaded in gray highlight the three clades of Proteoideae present in the Cape Floristic Region (CFR). The vertical line marks the final onset of aMediterranean climate in the CFR (Linder, 2003). Absolute time scale according to Gradstein et al. (2004). Abbreviations: Afr, Africa; Aus, Australia; Mad, Madagascar; NC, NewCaledonia; Qld, Queensland; Tas, Tasmania; WA, Western Australia.
40 H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43
suggested an origin of the genus in the Karoo Mountain and South-eastern centers of endemism of the CFR (Linder, 2003). Their study
did not include a dating analysis, but the pattern of branch lengthscompared with our study is suggestive of an old crown age (i.e.,
Fig. 3. Variation in absolute age estimates for selected nodes according to method(multidivtime or penalized likelihood) and maximum age constraint on the root.Letters refer to nodes marked in Fig. 2 and listed in Table 4.
H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43 41
Leucadendron may have begun to diversify shortly after its splitfrom the rest of Leucadendrinae). Further analyses that will includemany species sampled from each genus of Leucadendrinae and se-quence data from multiple loci will clarify the timing and tempo ofdiversification of this large clade in the Cape. At the moment, with-out dated phylogenies more completely sampled at the species le-vel, it is difficult to correlate the diversification of Proteaceae in theCape with the climatic changes that led to the final onset of a Med-iterranean climate in the region. In addition, recent diversificationof extant species does not preclude older speciation bursts fol-lowed by intensive extinction. However, the present study clearlyindicates a complex pattern, challenging the view of synchronousdiversification in all Cape clades.
Additional, complementary evidence on the history of diversifi-cation in the Cape might come from the fossil record. Unfortu-nately, the lack of a robust palynostratigraphic framework andthe lack of suitable descriptions of pollen attributed to Proteaceaemake it difficult to draw any decisive conclusions from these re-cords. As noted by Linder (2003), fossil evidence on Tertiary florasin the region is extremely limited. This is due in part to the geolog-ical history that has seen Miocene uplift and erosion of Late Creta-ceous and early Tertiary sediments, leaving only scattered depositsthat are difficult to correlate and date (Partridge and Maud, 1987).Sedimentation is also limited by the relatively high elevation ofsouthern Africa and a narrow strip between the modified rift scarpand the coast where material could accumulate. Few deposits arewell documented, and those that contain fossil plant material havenot been fully examined.
To date, the best documented evidence is the Arnot Pipe pollenflora from the uppermost Cretaceous to Paleocene of Namaqualand(Scholtz, 1985), which includes several palynomorph species attrib-uted to Proteaceae. Unfortunately, the quality of the data (mostlylight microscopy) is not sufficient to conduct thorough phyloge-netic analyses of these taxa, but the evidence is sufficient to showthe presence of Proteaceae in South Africa in the latest Cretaceousto Paleocene. Proteaceae palynomorph taxa were also describedfrom more recent deposits in Knysna (Thiergart et al., 1963) andthe west coast of Namaqualand near Hondeklipbaai (de Villiersand Cadman, 1997, 2001). Although originally thought to be datedfrom the Plio-Pleistocene, the Knysna record is most likely from theMiocene (Thwaites and Jacobs, 1987). Linder (2003) regarded thedating of the Namaqualand record as problematic, but recent re-search from offshore sites suggests an Eocene occurrence of taxatraditionally regarded as Miocene in age (Scott et al., 2006).
4.4. Comparison with other angiosperm Cape clades
The ages and origins of Cape clades have been investigated in anumber of other angiosperm taxa. These studies (reviewed inLinder and Hardy, 2004, and Edwards and Hawkins, 2007) haveproposed ages for Cape floral clades ranging from about 2 to46 Myr. Unfortunately, for most of these groups, no fossil age con-straint phylogenetically close to the Cape clade analyzed was avail-able, which led authors to use more external fossil age constraints(e.g., Forest et al., 2007), secondary calibration points (e.g., Goldbl-att et al., 2002), or geological calibration points (e.g., Richardsonet al., 2001). Given the growing consensus that the integration ofmultiple fossil age constraints is critical for the accuracy of molec-ular dating analyses (Near and Sanderson, 2004; Benton andDonoghue, 2007; Rutschmann et al., 2007), a direct comparisonof these ages across studies is not necessarily meaningful.
As noted by Edwards and Hawkins (2007), the variation in abso-lute ages obtained for different clades may also be due in part tovarying approaches to the delimitation of Cape clades, taxon sam-pling, and molecular dating methods. In order to correct for thesefactors of variation, they assembled and analyzed a large data setto estimate simultaneously the ages of two Cape clades in the Faba-ceae. Their results indicate a contemporaneous origin of these twoclades (Crotalarieae p.p. and Podylarieae), around 44–46 Myr ago,which is much older than the origin of the modern Mediterraneanclimate in the CFR. With lower taxon sampling density but moremolecular data to estimate branch lengths and more fossil age con-straints, our study has provided another simultaneous test of the ageof two Cape clades, namely, Protea and the Leucadendrinae. In thiscase, we found significantly different age estimates (5–18 and22–39 Myr, respectively). Although species-level studies will berequired to test the patterns of diversification in these Cape clades,our results add to growing evidence that the Cape floral diversitycannot be simply explained by synchronous radiation in many plantgroups in response to a common trigger (Linder and Hardy, 2004).
Acknowledgments
We are grateful to Félix Forest for advice on Bayesian phyloge-netic and dating analyses; Katherine Downs for assistance inextracting and sequencing ITS nrDNA from several species; NahidHeidari for matK primer design and sequencing; and three anony-mous reviewers for helpful comments on this paper. Part of thiswork was carried out using the resources of the ComputationalBiology Service Unit from Cornell University and the Edit serverof the Royal Botanic Gardens, Kew. This research was supportedby grants to N.P.B. and P.H.W. from the Royal Botanic Gardensand Domain Trust, Sydney, grants to D.J.C. and Kåre Bremer fromthe Swedish Research Council, and a Marie Curie Outgoing Interna-tional Fellowship (021943-HOTMED) from the European Commis-sion to H.S., V.S., and P.H.W.
References
Anderson, C.L., Bremer, K., Friis, E.M., 2005. Dating phylogenetically basal eudicotsusing rbcL sequences and multiple fossil reference points. Am. J. Bot. 92, 1737–1748.
Askin, R.A., Baldoni, A.M., 1998. The Santonian through Paleogene record ofProteaceae in the southern South America–Antarctic Peninsula region. Aust.Syst. Bot. 11, 373–390.
Barker, N.P., Weston, P.H., Rourke, J.P., Reeves, G., 2002. The relationships of thesouthern African Proteaceae as elucidated by internal transcribed spacer (ITS)DNA sequence data. Kew Bull. 57, 867–883.
Barker, N.P., Vanderpoorten, A., Morton, C.M., Rourke, J.P., 2004. Phylogeny,biogeography, and the evolution of life-history traits in Leucadendron(Proteaceae). Mol. Phylogenet. Evol. 33, 845–860.
Barker, N.P., Weston, P.H., Rutschmann, F., Sauquet, H., 2007. Molecular dating ofthe ‘Gondwanan’ plant family Proteaceae is only partially congruent with thetiming of the break-up of Gondwana. J. Biogeogr. 34, 2012–2027.
42 H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43
Barraclough, T.G., Reeves, G., 2005. The causes of speciation in flowering plantlineages: species-level DNA trees in the African genus Protea. In: Bakker, F.T.,Chatrou, L.W., Gravendeel, B., Pelser, P.B. (Eds.), Plant Species-LevelSystematics: New Perspectives on Pattern and Process. Regnum Vegetabile142. Koeltz, Königstein, pp. 31–46.
Benton, M.J., Donoghue, P.C.J., 2007. Paleontological evidence to date the tree of life.Mol. Biol. Evol. 24, 26–53.
Brenner, G.J., 1968. Middle Cretaceous spores and pollen from northeastern Peru.Pollen et Spores 10, 341–383.
Couper, R.A., 1960. New Zealand Mesozoic and Cainozoic plant microfossils. NewZealand Geological Survey, Paleontological Bulletin 32, 1–87.
Cowling, R.M., Lamont, B.B., 1998. On the nature of Gondwanan species flocks:diversity of Proteaceae in Mediterranean south-western Australia and SouthAfrica. Aust. J. Bot. 46, 335–355.
de Villiers, S., Cadman, A., 1997. The palynology of Tertiary sediments from apalaeochannel in Namaqualand, South Africa. Palaeontologia Africana 34, 69–99.
de Villiers, S., Cadman, A., 2001. An analysis of the palynomorphs obtained fromTertiary sediments at Koingnaas, Namaqualand, South Africa. J. Afr. Earth Sci.33, 17–47.
Dettmann, M.E., Jarzen, D.M., 1996. Pollen of proteaceous-type from latestCretaceous sediments, southeastern Australia. Alcheringa 20, 103–160.
Dettmann, M.E., Jarzen, D.M., 1998. The early history of the Proteaceae in Australia:the pollen record. Aust. Syst. Bot. 11, 401–438.
Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities offresh leaf tissue. Phytochemical Bulletin 19, 11–15.
Doyle, J.A., Endress, P.K., 2000. Morphological phylogenetic analysis of basalangiosperms: comparison and combination with molecular data. Int. J. PlantSci. 161, S121–S153.
Doyle, J.A., Hotton, C.L., 1991. Diversification of early angiosperm pollen in acladistic context. Pollen et Spores 44, 169–195.
Doyle, J.A., 1992. Revised palynological correlations of the lower Potomac Group(USA) and the Cocobeach sequence of Gabon (Barremian–Aptian). Cretac. Res.13, 337–349.
Doyle, J.A., 2005. Early evolution of angiosperm pollen as inferred from molecularand morphological phylogenetic analyses. Grana 44, 227–251.
Edwards, D., Hawkins, J.A., 2007. Are Cape floral clades the same age?Contemporaneous origins of two lineages in the genistoids s.l. (Fabaceae).Mol. Phylogenet. Evol. 45, 952–970.
Eernisse, D.J., Kluge, A.G., 1993. Taxonomic congruence versus total evidence, andamniote phylogeny inferred from fossils, molecules, and morphology. Mol. Biol.Evol. 10, 1170–1195.
Farris, J.S., 1969. A successive approximations approach to character weighting.Syst. Zool. 18, 374–385.
Foreman, D., 1995. Petrophile. In: McCarthy, P. (Ed.), Flora of Australia. CSIROAustralia, Melbourne, pp. 149–193.
Forest, F., Nänni, N., Chase, M., Crane, P., Hawkins, J.A., 2007. Diversification of alarge genus in a continental biodiversity hotspot: temporal and spatial origin ofMuraltia (Polygalaceae) in the Cape of South Africa. Mol. Phylogenet. Evol. 43,60–74.
Furness, C.A., Rudall, P.J., 2004. Pollen aperture evolution—a crucial factor foreudicot success? Trends Plant Sci. 9, 154–158.
Goldblatt, P., Manning, J., 2000. Cape Plants. A Conspectus of the Cape Flora of SouthAfrica. National Botanical Institute, Pretoria.
Goldblatt, P., Savolainen, V., Porteous, O., Sostaric, I., Powell, M., Reeves, G.,Manning, J.C., Barraclough, T.G., Chase, M.W., 2002. Radiation in the Cape floraand the phylogeny of peacock irises Moraea (Iridaceae) based on four plastidDNA regions. Mol. Phylogenet. Evol. 25, 341–360.
Gradstein, F.M., Ogg, J.G., Smith, A.G., Agterberg, F.P., Bleeker, W., Cooper, R.A.,Davydov, V., Gibbard, P., Hinnov, L., House, M.R., Lourens, L., Luterbacher, H.-P.,McArthur, J., Melchin, M.J., Robb, L.J., Shergold, J., Villeneuve, M., Wardlaw, B.R.,Ali, J., Brinkhuis, H., Hilgen, F.J., Hooker, J., Howarth, R.J., Knoll, A.H., Laskar, J.,Monechi, S., Powell, J., Plumb, K.A., Raffi, I., Röhl, U., Sanfilippo, A., Schmitz, B.,Shackleton, N.J., Shields, G.A., Strauss, H., Dam, J.V., Veizer, J., Kolfschoten, T.v.,Wilson, D., 2004. A Geologic Time Scale 2004. Cambridge University Press.
Hall, 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.
Hermsen, E., Nixon, K., Crepet, W., 2006. The impact of extinct taxa on understandingthe early evolution of angiosperm clades: an example incorporating fossilreproductive structures of Saxifragales. Plant Syst. Evol. 260, 141–169.
Hilu, K.W., Borsch, T., Muller, K., Soltis, D.E., Soltis, P.S., Savolainen, V., Chase, M.W.,Powell, M.P., Alice, L.A., Evans, R., Sauquet, H., Neinhuis, C., Slotta, T.A.B.,Rohwer, J.G., Campbell, C.S., Chatrou, L.W., 2003. Angiosperm phylogeny basedon matK sequence information. Am. J. Bot. 90, 1758–1776.
Hoot, S.B., Douglas, A.W., 1998. Phylogeny of the Proteaceae based on atpB andatpB–rbcL intergenic spacer region sequences. Aust. Syst. Bot. 11, 301–320.
Hoot, S.B., Magallón, S., Crane, P.R., 1999. Phylogeny of basal eudicots based on threemolecular data sets: atpB, rbcL, and 18S nuclear ribosomal DNA sequences. Ann.MO Bot. Gard. 86, 1–32.
Huelsenbeck, J.P., Ronquist, F., 2001. MrBayes: Bayesian inference of phylogeny.Bioinformatics 17, 754.
Hughes, N.F., McDougall, A.B., 1990. Barremian–Aptian angiospermid pollen recordsfrom southern England. Rev. Palaeobot. Palynol. 65, 145–151.
Jardiné, S., Magloire, L., 1965. Palynologie et stratigraphie du Crétacé des bassins duSénégal et de Côte d’Ivoire. Mémoire du Bureau de Recherches GéologiquesMinières 32, 187–245.
Johnson, L.A.S., Briggs, B.G., 1975. On the Proteaceae—the evolution andclassification of a southern family. Bot. J. Linn. Soc. 70, 83–182.
Jordan, G.J., Hill, R.S., 1999. The phylogenetic affinities of Nothofagus(Nothofagaceae) leaf fossils based on combined molecular and morphologicaldata. Int. J. Plant Sci. 160, 1177–1188.
Kim, S., Soltis, D.E., Soltis, P.S., Zanis, M.J., Suh, Y., 2004. Phylogenetic relationshipsamong early-diverging eudicots based on four genes: were the eudicotsancestrally woody? Mol. Phylogenet. Evol. 31, 16–30.
Kishino, H., Thorne, J.L., Bruno, W.J., 2001. Performance of a divergence timeestimation method under a probabilistic model of rate evolution. Mol. Biol. Evol.18, 352–361.
Linder, H.P., Hardy, C.R., 2004. Evolution of the species-rich Cape flora. Philos. Trans.R. Soc. Lond. B 359, 1623–1632.
Linder, H.P., Meadows, M.E., Cowling, R.M., 1992. History of the Cape flora. In:Cowling, R.M. (Ed.), The Ecology of Fynbos: Nutrients, Fire and Diversity. OxfordUniversity Press, Cape Town, pp. 113–134.
Linder, H.P., 2003. The radiation of the Cape flora, southern Africa. Biol. Rev. 78,597–638.
Mabberley, D.J., 1997. The Plant-Book, second ed. Cambridge University Press,Cambridge.
Magallón, S.A., 2004. Dating lineages: molecular and paleontological approaches tothe temporal framework of clades. Int. J. Plant Sci. 165, S7–S21.
Manos, P.S., Soltis, P.S., Soltis, D.E., Manchester, S.R., Oh, S.-H., Bell, C.D., Dilcher, D.L.,Stone, D.E., 2007. Phylogeny of extant and fossil Juglandaceae inferred from theintegration of molecular and morphological data sets. Syst. Biol. 56, 412–430.
Martin, A.R.H., 1995. Paleogene proteaceous pollen and phylogeny. Alcheringa 19,27–40.
Milne, L.A., 1994. Relationship between Propylipollis annularis (Tertiary dispersedpollen) and extant Xylomelum (Proteaceae). In: Kurmann, M.H., Doyle, J.A. (Eds.),Ultrastructure of Fossil Spores and Pollen. Royal Botanic Gardens, Kew, pp. 193–213.
Milne, L.A., 1998. Tertiary palynology: Beaupreaidites and new Conospermeae(Proteoideae) affiliates. Aust. Syst. Bot. 11, 553–603.
Mittermeier, R.A., Robles Gil, P., Hoffmann, M., Pilgrim, J., Brooks, T., Mittermeier,C.G., Lamoreux, J., da Fonseca, G.A.B., 2004. Hotspots Revisited: Earth’sBiologically Richest and Most Endangered Terrestrial Ecoregions CEMEX,Mexico City.
Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000.Biodiversity hotspots for conservation priorities. Nature 403, 853–858.
Near, T.J., Sanderson, M.J., 2004. Assessing the quality of molecular divergence timeestimates by fossil calibrations and fossil-based model selection. Philos. Trans.R. Soc. Lond. B 359, 1477–1483.
Nylander, J.A.A., 2004. MrModeltest v2. Program distributed by the author.Evolutionary Biology Centre, Uppsala University.
Partridge, T.C., Maud, R.R., 1987. Geomorphic evolution of southern Africa since theMesozoic. S. Afr. J. Geol. 902, 179–208.
Pocknall, D.T., Crosbie, Y.M., 1988. Pollen morphology of Beauprea (Proteaceae)—modern and fossil. Rev. Palaeobot. Palynol. 53, 305–327.
Qiu, Y.L., Lee, J., Bernasconi-Quadroni, F., Soltis, D.E., Soltis, P.S., Zanis, M., Zimmer,E.A., Chen, Z., Savolainen, V., Chase, M.W., 2000. Phylogeny of basalangiosperms: analyses of five genes from three genomes. Int. J. Plant Sci. 161,S3–S27.
Qiu, Y.L., Dombrovska, O., Lee, J., Li, L., Whitlock, B.A., Bernasconi-Quadroni, F., Rest,J., Davis, C.C., Borsch, T., Hilu, K.W., Renner, S.S., Soltis, D.E., Soltis, P.S., Zanis, M.,Cannone, J.J., Gutell, R.R., Powell, M.P., Savolainen, V., Chatrou, L.W., Chase,M.W., 2005. Phylogenetic analyses of basal angiosperms based on nine plastid,mitochondrial, and nuclear genes. Int. J. Plant Sci. 166, 815–842.
Rebelo, T., 2001. SASOL Proteas: A Field Guide to the Proteas of Southern Africa,second ed. Fernwood Press, Vlaeberg.
Richardson, J.E., Weitz, F.M., Fay, M.F., Cronk, Q.C.B., Linder, H.P., Reeves, G., Chase,M.W., 2001. Rapid and recent origin of species richness in the Cape flora ofSouth Africa. Nature 412, 181–183.
Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.
Rourke, J.P., 1998. A review of the systematics and phylogeny of the AfricanProteaceae. Aust. Syst. Bot. 11, 267–285.
Rutschmann, F., Eriksson, T., Salim, K.A., Conti, E., 2007. Assessing calibrationuncertainty in molecular dating: the assignment of fossils to alternativecalibration points. Syst. Biol. 56, 591–608.
Rutschmann, F., 2005. Bayesian molecular dating using PAML/multidivtime. A step-by-step manual. Version 1.5. University of Zurich, Switzerland. Available from:<http://www.plant.ch>.
Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution anddivergence times: a penalized likelihood approach. Mol. Biol. Evol. 19, 101–109.
Sanderson, M.J., 2003. R8s: inferring absolute rates of molecular evolution anddivergence times in the absence of a molecular clock. Bioinformatics 19, 301–302.
Sauquet, H., Cantrill, D.J., 2007. Pollen diversity and evolution in Proteoideae(Proteales: Proteaceae). Syst. Bot. 32, 271–316.
Sauquet, H., Weston, P.H., Anderson, C.L., Barker, N.P., Cantrill, D.J., Mast, A.R.,Savolainen, V., 2009. Contrasted patterns of hyperdiversification inMediterranean hotspots. Proc. Natl. Acad. Sci. USA 106, 221–225.
Scholtz, A., 1985. The palynology of the upper lacustrine sediments of the ArnotPipe, Banke, Namaqualand. Annals of the South African Museum 95, 1–109.
Scott, L., Cadman, A., McMillan, I.K., 2006. Early history of Cainozoic Asteraceaealong the Southern African west coast. Rev. Palaeobot. Palynol. 142, 47–52.
H. Sauquet et al. / Molecular Phylogenetics and Evolution 51 (2009) 31–43 43
Soltis, D.E., Soltis, P.S., Chase, M.W., Mort, M.E., Albach, D.C., Zanis, M., Savolainen,V., Hahn, W.H., Hoot, S.B., Fay, M.F., Axtell, M., Swensen, S.M., Prince, L.M., Kress,W.J., Nixon, K.C., Farris, J.S., 2000. Angiosperm phylogeny inferred from 18SrDNA, rbcL, and atpB sequences. Bot. J. Linn. Soc. 133, 381–461.
Springer, M.S., Teeling, E.C., Madsen, O., Stanhope, M.J., de Jong, W.W., 2001.Integrated fossil and molecular data reconstruct bat echolocation. Proc. Natl.Acad. Sci. USA 98, 6241–6246.
Swofford, D.L., 2002. PAUP�. Phylogenetic Analysis Using Parsimony (�and OtherMethods). Version 4. Sinauer Associates, Sunderland, Massachusetts.
Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers foramplification of three non-coding regions of chloroplast DNA. Plant Mol. Biol.17, 1105–1109.
Thiergart, F., Frantz, U., Raukopf, K., 1963. Palynologische Untersuchungen vonTertiärkohlen und einer Oberflächenprobe nähe Kynsna, Südafrika. AdvancingFrontiers of Plant Science 4, 151–178.
Thorne, J.L., Kishino, H., 2002. Divergence time and evolutionary rate estimationwith multilocus data. Syst. Biol. 51, 689–702.
Thorne, J.L., Kishino, H., Painter, I.S., 1998. Estimating the rate of evolution of therate of molecular evolution. Mol. Biol. Evol. 15, 1647–1657.
Thwaites, R.N., Jacobs, E.O., 1987. The Knysna lignites: a review of their positionwithin the geomorphological development of the southern Cape Province. S.Afr. J. Geol. 90, 137–146.
Ward, J.V., Doyle, J.A., 1994. Ultrastructure and relationships of mid-Cretaceouspolyforates and triporates from Northern Gondwana. In: Kurmann, M.H., Doyle,J.A. (Eds.), Ultrastructure of Fossil Spores and Pollen. Royal Botanic Gardens,Kew, pp. 161–172.
Weston, P.H., Barker, N.P., 2006. A new suprageneric classification of the Proteaceae,with an annotated checklist of genera. Telopea 11, 314–344.
Worberg, A., Quandt, D., Barniske, A.-M., Lohne, C., Hilu, K.W., Borsch, T., 2007.Phylogeny of basal eudicots: insights from non-coding and rapidly evolvingDNA. Org. Divers. Evol. 7, 55–77.
Yang, Z., 1997. PAML: a program package for phylogenetic analysis by maximumlikelihood. CABIOS 13, 555–556.
