This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Molecular Phylogenetics and Evolution 54 (2010) 607–616 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev Evidence for a vicariant origin of Macaronesian–Eritreo/Arabian disjunctions in Campylanthus Roth (Plantaginaceae) Mike Thiv a,*, Mats Thulin b, Mats Hjertson c, Matthias Kropf d, Hans Peter Linder e a Botany Department, Staatliches Museum für Naturkunde, Rosenstein 1, D-70191 Stuttgart, Germany Department of Evolution, Genomics and Systematics, University of Uppsala, Norbyvägen 18D, 75236 Uppsala, Sweden c Museum of Evolution, University of Uppsala, Norbyvägen 16, 75236 Uppsala, Sweden d Institut für Botanik, Universität für Bodenkultur, Gregor Mendel-Str. 33, A-1180 Vienna, Austria e Institute of Systematic Botany, University of Zurich, Zollikerstrasse 107, CH 8008 Zurich, Switzerland b a r t i c l e i n f o Article history: Received 26 May 2009 Revised 5 October 2009 Accepted 6 October 2009 Available online 13 October 2009 Keywords: Campylanthus Plantaginaceae Biogeography Phylogeny Vicariance Trans-Saharan disjunction Macaronesia a b s t r a c t The numerous disjunct plant distributions between Macaronesia and eastern Africa–Arabia suggest that these could be the relicts of a once continuous vegetation belt along the southern Tethys, which has been fragmented by Upper Miocene–Pliocene aridification. We tested this vicariance hypothesis with a phylogenetic analysis of Campylanthus (Plantaginaceae), based on nuclear and plastid DNA sequence data. Our results indicate a basal split within Campylanthus giving rise to Macaronesian and Eritreo–Arabian lineages in the Pliocene/Upper Miocene. This is consistent with the vicariance hypothesis, thus obviating the need to postulate trans-Saharan long-distance dispersal. The biogeography of Campylanthus may parallel patterns in other plant groups and the implications for our understanding of the biogeography of northern and eastern Africa, and Arabia are discussed. Crown Copyright Ó 2009 Published by Elsevier Inc. All rights reserved. 1. Introduction Disjunct distribution patterns are common in many organism groups, and their patterns and causes have early gained the interest of natural scientists (e.g., Darwin, 1859). Two main mechanisms resulting in a fragmented distribution can be distinguished. Vicariance results from an emerging barrier splitting a continuous range into two or more separate parts, while in case of (long-distance) dispersal disjunct distribution areas are the result of dispersal over pre-existing barriers (Nelson, 1978; de Queiroz, 2005). The relative importance of these processes has been intensively debated in the last decades. Until a decade ago vicariance explanations were preferred, as they result in more general hypotheses or explanations for similar disjunctions across different organismic groups. Moreover, vicariance hypotheses were falsifiable, for example if the phylogeny of a taxon was incongruent with the hypothesised vicariance history. Vicariance resulting from earth history events, such as plate tectonics, could affect whole biotas, while dispersal was considered to be mainly taxon-specific. * Corresponding author. Fax: +49 (0) 711 8936100. E-mail addresses: thiv.smns@naturkundemuseum-bw.de (M. Thiv), Mats.Thulin@ebc.uu.se (M. Thulin), Mats.Hjertson@evolmuseum.uu.se (M. Hjertson), matthias.kropf@boku.ac.at (M. Kropf), peter.linder@systbot.uzh.ch (H.P. Linder). Consequently, dispersal was only accepted a priori for cases without a vicariance explanation, e.g., for the colonisation of volcanic oceanic islands. In recent decades long-distance dispersal has emerged again as the most popular process for shaping current disjunct distributions (e.g., Renner, 2004; de Queiroz, 2005; Keppel et al., 2009). This has largely been the result of molecular dating techniques, which often falsified the vicariance prediction that the divergence of the variant taxa should be at least as old as the geological event that caused the disjunction (e.g., plate movements, mountain range formations; de Queiroz, 2005). A typical example of a disjunction is between eastern Africa– Arabia and the Macaronesian Islands (in a traditional sense comprising the Canary Islands, Madeira, Cape Verde Islands, Azores and Salvage Islands, of which, however, the Cape Verde Islands were shown to have closer links to tropical Africa, Vanderpoorten et al., 2007, see discussion; Bramwell, 1976; Quézel, 1978; Deil and Müller-Hohenstein, 1984; Mies, 1995). Examples include Campylanthus Roth (Plantaginaceae), Dracaena L. (Dracaenaceae/ Ruscaceae), Hemicrambe Webb (Brassicaceae), Parolinea Webb (Brassicaceae), Aeonium Webb & Berthel. (Crassulaceae), Camptoloma Benth. (Scrophulariaceae), and Pulicaria L./Vieraea Sch. Bip. (Asteraceae; Andrus et al., 2004). These disjunctions have been interpreted as relicts of a late Miocene continuous, subtropical flora in northern Africa (Hooker, 1878; Engler, 1879; Meusel, 1055-7903/$ - see front matter Crown Copyright Ó 2009 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2009.10.009 Author's personal copy 608 M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 1965; Axelrod, 1975; Sunding, 1979; Bramwell, 1985; Thiede, 1994; Thulin, 1994; Marrero et al., 1998). This vegetation belt was fragmented by aridification in the Upper Miocene/Pliocene/ Quaternary that resulted in the formation of the Sahara and the consequent extinction of many plant groups in this area. Some species of a few members of this flora survived in the Canary Islands and Arabia/eastern Africa, leading to the present day disjunct patterns (Sunding, 1979; Bramwell, 1985). This is a typical vicariance scenario and is therefore referred to as the vicariance hypothesis. Such biogeographic implications have mostly been inferred from taxonomy-based distribution patterns. Only a few molecular phylogenetic studies addressing this question have been conducted, and these confirmed a sister group relationship between the Macaronesian and eastern African/Arabian species for Camptoloma, Aeonium and Plocama (Kornhall et al., 2001; Mort et al., 2002; Backlund et al., 2007). Most taxa, however, do not fit the vicariance hypothesis (e.g., Pulicaria/Vieraea, Andrus et al., 2004), as the closest relatives of the Macaronesian species are not found in East Africa, but in the Mediterranean basin. Here, we test the vicariance hypothesis using a prominent example of Macaronesian–Eritreo/Arabian disjunctions, the palaeotropical genus Campylanthus (Plantaginaceae, formerly Scrophulariaceae). Campylanthus includes 18 species of shrubs and subshrubs, two being restricted to Macaronesia (Canary and Cape Verde Islands) and 16 occurring from east Africa, Arabia, to Pakistan (Fig. 1, Miller, 1980, 1982, 1988; Hjertson, 1997, 2003; Hjertson and Miller, 2000; Kilian et al., 2002; Hjertson and Thulin, 2006; Hjertson et al., 2008). We used a molecular phylogenetic analysis and a molecular dating approach employing a relaxed molecular clock, to address the following questions: (1) Do the Macaronesian and Eritreo–Arabian taxa constitute sister clades as suggested by cladistic analyses based on morphological data (Hjertson, 1997, 2003), thus consistent with a vicariance scenario? (2) If so, does the age of the divergence between the Eritreo–Arabian and Macaronesian clades coincide with the late Miocene onset of Saharan aridification, as predicted by the vicariance hypothesis? (3) Finally, we aim at reconstructing the biogeographic history of Campylanthus. To this end, we analysed sequence data of the plastid (atpB-rbcL intergenic spacer) and the nuclear genomes (nrITS, Baldwin et al., 1995; Alvarez and Wendel, 2003) from nearly all currently recognised Campylanthus taxa and used a Bayesian relaxed clock approach to date relevant nodes. 2. Material and methods 2.1. Taxon sampling The genus Campylanthus consists of 18 species, 14 of which were included in our analyses (Table 1). The remaining four – the Somalian C. anisotrichus and C. parviflorus, the Omanian C. hajarensis and the Pakistani C. ramosissimus – are only rarely collected due to their very restricted distribution and, therefore, no material was available for our study. The position of Campylanthus within the former Scrophulariaceae was unresolved (Reichenbach, 1828; Bentham and Hooker, 1886; Wettstein, 1891; Hjertson, 1997, 2003; Fischer, 2004) until recent molecular phylogenetic analyses (Albach et al., 2005) placed it as sister to the Globularia L./Poskea Vatke Table 1 Taxon sampling, vouchers, defined areas of endemism and Genbank numbers: MA, Macaronesian Region; SA, South Arabian Province; SE, Somalo-Ethiopian Province; SO, Socotran Province. Sequences marked with an asterisk were used for Bayesian dating. Taxon Voucher Aragoa abietina H.B. & K. Aragoa cupressina H.B. & K. Campylanthus antonii Thulin Campylanthus chascaniflorus A.G. Mill. Campylanthus glaber Benth. Campylanthus hubaishanii N. Kilian & P. Hein Campylanthus incanus A.G. Mill. Campylanthus junceus Edgew. Campylanthus mirandae A.G. Mill. Campylanthus pungens O. Schwartz Campylanthus salsoloides (L.f.) Roth See reference See reference Thulin et al. 9534 Miller 7731 Kilian 3140 Kilian YP1169 Campylanthus Campylanthus Campylanthus Thulin Campylanthus Campylanthus sedoides A.G. Mill. somaliensis A.G. Mill. reconditus Hjertson & spinosus Balf. (f.) spinosus Balf. (f.) Campylanthus yemenensis A.G. Mill. Campylanthus yemenensis A.G. Mill. Digitalis lutea L. Digitalis lutea L. Globularia punctata Lapeyr. Globularia repens Lam. Globularia salicina Lam. Plantago major L. Plantago major L. Plantago raoulii Decne. Plantago stauntonii Reichardt Plantago debilis R. Br. Plantago spathulata Hook. f. Plantago atrata Hoppe Plantago coronopus L. Plantago uniflora L. Veronica filiformis Sm. Herbarium Origin Defined area of distribution nrITS atpB-rbcL spacer UPS E B B Yemen Oman Cape Verde Islands Yemen SA SA MA SA AJ459404* AJ459402* FM207412* FM207413* FM207414* FM207416* Thulin et al. 10720 Kilian 6036 Morris 561 Kilian 5138 Wieringa & Janzen 3485 Miller & Nyberg 9451 Thulin & Warfa 5560 Thulin et al. 10652 UPS B E B WAG SE SA, SE SA SA MA FM207415* FM207417* FM207418* FM207419* FM207421* FM207436 FM207437 FM207438 FM207439 FM207441 BM UPS UPS Somalia Yemen Oman Yemen Spain (Canary Islands) Oman Somalia Somalia SA SE SE FM207422* FM207425* FM207420* FM207447 FM207442 FM207440 Kilian 2374 Lavranos & Carter 23477 Miller 3076 Wood 75/225 Brune s.n. See reference Joßberger & Brune s.n. See reference See reference See reference Joßberger & Brune s.n. See reference See reference See reference See reference See reference See reference See reference Joßberger s.n. B WAG Yemen (Socotra) Somalia SO SE FM207426* FM207427* FM207443 FM207444 E E STU Yemen Yemen Germany SA SA FM207423* FM207424* FM207445 FM207446 FM207428 STU Germany STU Germany STU Germany AY591266* AY492105* AF313039* AY101861* AY101867* AY101870* AY101868* AY101869* AY101895* AY101882* AY101885* GU143559 FM207432 FM207433 FM207434 FM207435 FM207431 FM207430 FM207429 Author's personal copy M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 clade. The clade Campylanthus, Globularia, Poskea, in turn, belongs to a larger lineage including, among others, Plantago L., Veronica L., and Digitalis L. Accordingly, we chose members of Digitalis, Veronica, Plantago and Globularia as outgroups for the combined data set. Due to lack of material, we merged plastid and nuclear DNA sequences of G. punctata and G. repens to represent a single chimeric member of Globularia. For molecular dating, a modified nrITS data set was used by adding more species of Plantago and Aragoa Kunth (Bello et al., 2002) to the taxon sample, which were relevant for the calibration of the tree. 2.2. Laboratory techniques DNA was extracted from herbarium material or silica dried samples using the DNeasy plant extraction kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Amplifications were performed using 1.5 mM buffer, 0.625 mM MgCl2, 0.2 mM dNTPs, 0.05 U/ll Taq DNA polymerase (Amersham Biosciences), 0.325 lM primer (ITS-A and ITS-B [Blattner, 1999] for nuclear ITS and atpB-F2 and atpB-R5 [Manen et al., 1994] for the plastid atpB-rbcL intergenic spacer) and 5 ng/ll DNA template. PCR profiles consisted of 33 cycles of 94 °C for 1 min, 50–55 °C for 1 min, and 72 °C for 2–3 min. PCR products were cleaned using the PCR purification kit (Qiagen) and sequenced in both directions with the PCR primers using BigDye 3.1 Terminator chemistry (Applied Bioystems, Foster City, California). The resulting products were separated on an ABI PRISM 3100 automated sequencing systems (PE Biosystems). 2.3. Data analysis Sequences were initially aligned using Clustal X (vers. 1.81; Thompson et al., 1997) and then adjusted manually. The sequence data from the two genomes were analysed separately to test for congruence. The congruence assumption was rejected if trees from the two data sets contained incongruent groupings supported by >70% bootstrap (Mason-Gamer and Kellogg, 1996). The 70% is arbitrary, but has been widely used (e.g., Moline et al., 2007). As no incongruent nodes were retrieved, the nuclear and chloroplast data sets were combined in a total evidence approach (Johnson and Soltis, 1998; Wiens, 1998). The alignment of the combined data set is available as supplement. Maximum likelihood analyses for the combined data set were performed using PAUP4.0 (Swofford, 2002). A GTR+C+I model was found to be the best-fit substitution model determined by using AIC as implemented in MODELTEST 3.6 (Posada and Crandall, 1998). Using the model parameters suggested by MODELTEST, a 609 heuristic search with 100 random-addition-sequence replicates, TBR branch swapping and steepest descent option in effect was conducted. The same options were used for ML bootstrap analyses with 100 bootstrap replicates. We tested the hypothesis that Macaronesian and Eritreo–Arabian taxa are not sister to each other, by comparing the maximum likelihood topology with a tree in which these lineages were constrained not to be sister to each other. A one-tailed SH test (Shimodaira and Hasegawa, 1999) with a test distribution generated by using 1000 bootstrap replicates with full optimisation was conducted as implemented in PAUP4.0. The parametric bootstrap procedure (Goldman et al., 2000; Stefanovic and Olmstead, 2004) included the determination of ML parameters for the described constrained topologies. Based on these parameters, 99 simulated data sets were created using Seq-Gen (Rambaut and Grassly, 1997). The simulated data sets were analysed using maximum parsimony with closest sequence addition, Multrees on and TBR branch swapping, testing for significant differences in lengths between the constrained tree as null hypothesis and the optimal tree. 2.4. Molecular dating To date the divergence events in Campylanthus a relaxed molecular clock was used. The dating analyses were conducted using the nrITS data set, because this allowed the inclusion of sequences of Aragoa, the sister group of Plantago, for which the chloroplast marker was not available. To test for strict clock-like evolution of the nrITS sequences, a likelihood ratio test was performed by comparing the scores of ML trees with and without a molecular clock enforced (Felsenstein, 1981; Sanderson, 1998; Nei and Kumar, 2000). The Bayesian dating method (Thorne et al., 1998; Thorne and Kishino, 2002) uses a probabilistic model to describe the change in evolutionary rate over time and uses the Markov chain Monte Carlo (MCMC) procedure to derive the posterior distribution of rates and time. It allows multiple calibration points and provides direct credibility intervals for estimated divergence times and substitution rates. To compare results yielded by different methods, we used two programs, BEAST 1.4.8 (Drummond and Rambaut, 2007) and MULTIDIVTIME (Thorne et al., 1998; Kishino et al., 2001; Thorne and Kishino, 2002). For BEAST analysis model parameters (GTR+C+I, A/C: 0.9395, A/G: 1.68171, A/T: 1.8252, C/G: 0.2043, C/T: 3.5625, C shape parameter 1.7796 and proportion of invariable sites 0.3089) for the nrITS data set as selected by AIC were used as initial values for Jeffreys priors. A relaxed clock model with an uncorrelated lognormal rate change was chosen. After tuning the operators using BEAST’s auto-optimisation option, analyses used random starting trees under the coalescent process and a spe- Fig. 1. Map of northern Africa and Arabia showing the total distribution of Campylanthus (modified after Hjertson, 2003). Author's personal copy 610 M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 ciation model following a birth–death process as tree prior, with two runs of 5  107 generations each, sampling every 103 generations. Resulting posterior distributions for parameter estimates were checked in Tracer 1.4.1 (Drummond and Rambaut, 2007) and maximum credibility trees, representing the maximum a posteriori topology, were calculated after removing burn-in with Tree Annotator (version 1.4.7). The xml file is available as supplement. We used MULTIDIVTIME following Thiv et al. (2006) and a stepby-step manual by Rutschmann (2004). This method requires a fully resolved topology. Therefore, polytomies due to almost zero-branch lengths were resolved in accordance with a previous phylogenetic analysis based on morphological data (Figs. 6 and 4 in Hjertson (1997, 2003), respectively). Model parameters for the F84+C model (Kishino and Hasegawa, 1989) were estimated using the module BASEML in PAML (Yang, 1997). Based on those maximum likelihood branch lengths of the rooted tree together with a variance–covariance matrix of the branch length estimates were determined with ESTBRANCHES (Thorne et al., 1998). We used MULTIDIVTIME to approximate the posterior distributions of substitution rates and divergence times by using a multivariate normal distribution of estimated branch lengths (provided here by ESTBRANCHES) and running a MCMC procedure with the following settings for the prior distributions: 1.50 for both rttm and rttmsd, 0.07 for both rtrate and rtratesd, 0.4 for both brownmean and brownsd, and 42 million years ago (mya) for bigtime, which represents the age of the stem node of the clade including Callitriche, Plantago, and Digitalis (Wikström et al., 2001). We ran the Markov chain for at least 103 cycles and collected one sample every 100 cycles, after an unsampled burn-in of 104 cycles. We repeated the analyses in BEAST and MULTIDIVTIME twice using different initial conditions to assure convergence of the Markov chain and combined the results. Geological calibration dates and fossil data were used for molecular dating estimates (e.g., Richardson et al., 2001; Forest et al., 2005). For BEAST and MULTIDIVTIME analyses two calibration points were used. Plantago stauntonii is a species endemic to the Pacific island of New Amsterdam (Tongatapu Islands), of which the geological age is known to be 0.5–0.7 mya (Rønsted et al., 2002). If we assume that speciation of the island taxon from its mainland ancestor occurred following dispersal to the island, then the geological age of islands represents the maximum age of endemic taxa. There is no evidence that New Amsterdam is part of a hot spot system (Rønsted et al., 2002), in which an endemic species could have established earlier on an now-drowned island, before dispersing to the current island, thus distorting the real age of the species (Heads, 2005). Therefore, we used the older age of 0.7 mya as the upper bound for the crown group age of the clade of P. stauntonii, P. debilis and P. spathulata. A minimum age for the stem node age of Plantago was derived from pollen fossils attributed to Plantago. Records extend to the Middle/Upper Miocene (Nagy, 1963; Gray, 1964; Muller, 1981). Dates of Plantago fossils were published by Naud and Suc (1975), (6.4 mya, France) and Mohr (1984), (5.3–7.2 mya, Germany). These dates are also supported by molecular dating using P. stauntonii, yielding an estimate of ca. 5.5 mya (Rønsted et al., 2002). Still, to our knowledge the oldest fossil record attributed to Plantago dates to the Sarmatian (Upper Middle-Miocene; 12.8–11.6 mya, Harzhauser and Piller, 2004). This polyporate pollen was described as Plantaginacearumpollis and resembles the one of the extant Plantago lanceolata L. (Nagy, 1963). Although Aragoa and Plantago share similar exine structures (Bello et al., 2002) their pollen can be distinguished by number and shape of the apertures. Plantago is characterised by 2–14-porate pollen (Saad, 1982) while Aragoa has tricolpate pollen (Nilsson and Hong, 1993). Despite the vast fossil record of Plantago we cannot rule out the existence of older fossils. Accordingly, we used 11.6 mya as the lower bound for the stem node of Plantago. 2.5. Biogeographical analyses A crucial initial step in cladistic biogeography is the definition of the organisms’ area of distribution (Linder, 2001). We recognised Macaronesia (= MA) including the Canary and Cape Verde Islands, and the three provinces of the Eritreo–Arabian subregion according to Takhtajan (1986): Somalo-Ethiopia, SE; South Arabia, SA and Socotra, SO. The distribution ranges of the species were based on Hjertson (1997, 2003) and are indicated in Fig 3. Except for C. junceus and C. spinosus, all species of Campylanthus are restricted to a single region. Campylanthus anisotrichus (SE), C. parviflorus (SE), the Omanian C. hajarensis and C. ramosissimus from Pakistan (for which no molecular sequence data were available), likely do not effect the reconstruction of ancestral areas (see Section 4). Recent biogeographical analytical programs take into account the connectivity between areas. We used Lagrange 2.0.1 (Ree and Smith, 2008) for the reconstruction of biogeographic areas based on an ultrametric subtree of the BEAST consensus topology including all Campylanthus accessions and Globulariaceae. The two Globularia species were treated as a single terminal taxon because the exact patterns between the unsampled Poskea and the remaining Globularia species were not considered. The distribution of Globulariaceae (SA, SE, SO and MA) reflects occurrences of Poskea in SA, SE and SO and Globularia in SE and MA, however, not taking into account the distribution of Globularia far outside that of Campylanthus. Still, similar results for Campylanthus were yielded when all taxa from the BEAST analysis were coded (not shown). The basal split was dated to 13 mya as suggested by BEAST analysis. All combinations of areas were allowed in the adjacency matrix and baseline rates of dispersal and local extinction were estimated. Two models were considered. The first model did not constrain links between any areas throughout time. The second model included two constraints. The aridification of the Sahara (about 7 mya, Schuster et al., 2006) was taken into account by reducing the symmetrical dispersal rate between MA and the remaining areas to 0.1 in a time frame of 0–7.0 mya. Furthermore, the split between Socotran and east African accessions of C. spinosus was regarded as result of recent long-distance dispersal because these areas have not been linked with each other in the Pliocene. Instead, the Socotran archipelago was formerly connected to Arabia (15–18 mya; Fleitmann et al., 2004, Van Damme, 2009). Therefore, we regarded triplets with SO as unlikely ancestral areas, especially for basal nodes, and excluded SO–SA–SE, SO–SA–MA and SO–SE–MA, not SO–SE–SA–MA because this was coded for Globularia, and limited the total number of areas to three. Due to low temporal resolution the repeated closure of the Red Sea could not been modelled (see Section 4). 3. Results The aligned sequence lengths were 646 bp (ITS1, 5.8 SrDNA, ITS2) and 523 bp (atpB-rbcL spacer), resulting in a total of 1169 bp for the combined data set (including 1.9% of cells with missing data), of which 332 were variable and 180 were potentially parsimony-informative. The selected optimal model of sequence evolution for this combined data set was the general time reversible (GTR+C+I) model (Rodriguez et al., 1990): unequal base frequencies (A = 0.2656, C = 0.2252, G = 0.2431, T = 0.2661), six substitution types (A/C: 1.3514, A/G: 1.7381, A/T: 1.1212, C/G: 0.3735, C/T: 4.2855), gamma distribution of rates among sites with alpha shape parameter 0.8682 and proportion of invariable sites 0.4097. The analysis using these parameters yielded a ML tree with a log-likelihood score of lnL = 4232.01 (Fig. 2). Author's personal copy M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 The combined data set strongly supports the monophyly of Campylanthus with a bootstrap (= BS) of 100%. Within the genus, C. glaber from the Cape Verde Islands and the Canarian C. salsoloides constitute the sister group (BS 97%) to the eastern African–Arabian species (BS 85%). Within the latter clade, C. mirandae, C. pungens, and C. junceus/C. yemenensis (= C. pungens group; BS 89%) form a clade sister to a group including C. chascanifolius, C. sedoides, C. antonii/C. hubaishanii (= C. sedoides group, BS 98%) and C. somaliensis. The two clades in turn are sister to a group including C. incanus/ C. reconditus and C. spinosus (BS 82%). The alternative hypothesis of the Macaronesian and all Eritreo– Arabian taxa not being sister groups resulted in a difference of the log-likelihood of 4.47 in the SH test and in a difference of tree length of 2 steps in the parametric bootstrap analysis, neither test thus rejecting the alternative hypothesis (P = 0.173 and P = 0.41, respectively). 3.1. Molecular dating and biogeographical analyses of Enforcing the molecular clock resulted in a log-likelihood score lnL = 3754.77 for the nrITS data set (Table 1, Fig. 3). The com- 611 parison between clock and non-clock trees ( lnL = 3717.86) by applying the likelihood ratio (LR) test significantly rejected clocklike evolution for the dataset (LR = 73.82; df = 27, P < 0.001). The results of the two runs using BEAST were very similar, and were therefore combined. The same applied to the two runs using MULTIDIVTIME. The estimated mean ages and 95% highest posterior density intervals (HPD) are shown in Table 2. Although the ages inferred by BEAST are generally younger than those by MULTIDIVTIME, both methods date the basal split within Campylanthus to the Upper Miocene/Pliocene, with estimates between 2.00–8.08 and 3.02–9.84 mya (node a, Table 2). A chronogram of one of the two runs made by BEAST is shown in Fig. 3. The results of the biogeographical analyses are given in Table 2. Most relevant to the questions addressed in this paper are the results for the divergence between the Macaronesian and the Eritreo–African clades (node a in Fig. 3). Depending on the underlying model, the geographical division with the highest probability was either MA and SA/SE/SO (model 1) or MA and SA/SE (model 2). Among Arabian and east African species, the results are quite similar, however, larger incongruence between the models was found in following cases. The highest probability for node c Digitalis lutea Veronica filiformis Plantago major Globularia punctata/repens punctata repens 70 C. glaber 3140 97 C. salsoloides 3485 98 C. incanus 10720 84 C. reconditus 10652 100 82 C. spinosus 2374 100 C. spinosus 23477 C. somaliensis 5560 85 C. chascaniflorus 7731 98 C. sedoides 9451 81 C. antonii 9534 99 90 C. hubaishanii 116 1169 9 C. mirandae 561 C. pungens 5138 89 C. junceus 6036 82 0.01 substitutions/site C. yemenensis 3076 100 C. yemenensis 75/225 Fig. 2. ML tree of Campylanthus based on combined nrITS and atpB-rbcL intergene sequences. ML-Bootstrap values (>50%) are on branches. Numbers following Campylanthus (= C.) species names refer to collection numbers shown in Table 1. Author's personal copy 612 M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 Digitalis D i g i t a l i s lutea lutea Aragoa A r a g o a abietina abietina Aragoa A r a g o a cupressina cupressina Plantago P l a n t a g o coronopus coronopus Plantago P l a n t a g o uniflora uniflora Plantago Pla n ta g o atrata a tra ta Plantago P l a n t a g o major major min. 11.6 mya Plantago P l a n t a g o raoulii raoulii Plantago P l a n t a g o stauntonii stauntonii Plantago P l a n t a g o debilis debilis max. 0.7 mya Plantago P l a n t a g o spathulata spathulata Globularia G l o b u l a r i a repens repens Globularia G l o b u l a r i a salicina salicina e d c g h a f i b 20 15 10 5 0 C.. spinosus C spinosus SE C.. spinosus C spinosus SO C.. reconditus C re co n d itu s SE C.. incanus C in ca n u s SE C.. mirandae C mira n d a e SA C.. pungens C pungens SA C.. junceus C ju n ce u s SA, SE C.. yemenensis C ye me n e n sis SA C.. yemenensis C ye me n e n sis SA C.. somaliensis C so ma lie n sis SE C.. chascaniflorus C ch a sca n iflo ru s SA C.. sedoides C se d o id e s SA C.. hubaishanii C hubaishanii SA C.. antonii C a n to n ii SA C. glaber MA C.. salsoloides C salsoloides MA mya Fig. 3. Chronogram of Campylanthus of the Bayesian dating analysis using BEAST based on nrITS sequences. Lower case letters refer to nodes as in Table 2. Letters behind Campylanthus (= C.) species indicate defined areas of distribution (MA, Macaronesia; SE; Somalo-Ethiopia; SA, South Arabia; SO, Socotra). Arrows define calibration nodes with given ages. Gray bars indicate 95% HPD of age estimates. in Fig. 3 was SE|SA, SE using model 2, whereas the node was optimised SE, SO|SA under model 1. The models also differed for node f favouring a split between Arabia (SA|SA) or SA|SA, SE. 4. Discussion 4.1. Phylogeny Our phylogenetic analysis using combined chloroplast and nuclear data corroborate the monophyly of Campylanthus and the basal bipartition between the Macaronesian clade of C. glaber and C. salsoloides and the lineage of the remaining species from eastern Africa, Arabia, and likely southern Asia. This is in accordance with the morphological analysis by Hjertson (1997, 2003). We tested alternative scenarios using SH test and parametric bootstrap. The SH test is a nonparametric test which is a recommended way of assessing support when the number of candidate trees is small (Shimodaira, 2002) and parametric bootstrap is shown to be a statistically sound method of evaluating different alternative topological hypotheses (Huelsenbeck et al., 1996; Goldman et al., 2000). Curiously, our parametric bootstrap and SH tests indicated that alternative topologies cannot be rejected. Such trees placed the two Macaronesian species nested inside the Eritreo–Arabian taxa as sister to C. incanus, C. reconditus and C. spinosus. This could be due to low phylogenetic signal, however, as the basal split between Macaronesian and Eritreo–Arabian Campylanthus gains high bootstrap and morphological (Hjertson, 1997, 2003) support, we favour this topology as our working hypothesis. Most of the Eritreo–Arabian clades show uniform geographic patterns. Well-supported sister group to all Arabian species plus C. somaliensis is the eastern African and partly Socotran group of C. incanus/C. reconditus/C. spinosus. Based on the common occurrence of adpressed hairs on the vegetative parts, this clade should also include the unsampled Somalian C. anisotrichus and C. parviflorus (Hjertson, 1997, 2003). Other morphological characters support the incorporation of C. ramosissimus from Pakistan and C. hajarensis from Oman into the mainly Arabian C. pungens clade (Hjertson, 1997, 2003; Hjertson et al., 2008). The monophyly of the C. sedoides group including C. somaliensis is corroborated by particular style shapes (Hjertson, 1997, 2003). Overall, the relationships among Eritreo–Arabian species of Campylanthus inferred from molecular data differ from those based on morphological characters (Hjertson, 1997, 2003), and none of these clades exactly match those in our study. Nonetheless, several taxonomic affinities as discussed by Hjertson (1997, 2003) are corroborated by our data, e.g., the close relationships between C. incanus and C. spinosus, and between C. pungens, C. junceus, C. yemenensis, and C. mirandae. 4.2. Vicariance hypothesis The vicariance hypothesis (e.g., Axelrod, 1975; Bramwell, 1985) predicts that (i) Macaronesian and Eritreo–Arabian taxa should be sister groups and (ii) that the age of their split should fall in the Upper Miocene to Pliocene, the period in which the aridification of the Sahara took place, which, according to palaeoecological evidence, started at least 7 mya (Schuster et al., 2006). Both criteria Author's personal copy 613 M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 Table 2 Results of the Bayesian dating using BEAST and MULTIDIVTIME showing combined mean ages, 95% HPD (all in mya) of two runs, and the biogeographic reconstructions using Lagrange based on models 1 and 2 (see Section 2). Abbreviations follow Table 1, nodes refer to Fig. 3. BEAST MULTIDIVTIME Lagrange model 1 Split [SA, SE, SO|MA] [SA, SE|MA] [MA|MA] [SE, SO|SA] [SE|SA, SE] [SO|SA, SE] [SE|SA] [SE, SO|SE] [SE|SE] [SE|SO] [SA|SA] [SA|SA, SE] Node a Mean 4.68 95% HPD 2.00–8.07 Mean 6.15 95% HPD 3.02–9.84 b c 1.16 3.34 0.16–2.61 1.40–5.65 3.12 4.90 0.59–6.76 2.10–8.47 d 1.45 0.35–2.90 3.33 0.95–6.70 e f 0.99 2.34 0.00–0.32 1.00–4.02 0.83 3.73 0.02–3.12 1.40–7.06 g 1.38 0.49–2.45 2.64 0.82–5.58 h 1.03 0.17–2.06 1.48 0.14–3.99 i 0.81 0.23–1.57 1.80 0.38–4.40 [SA|SA] [SA|SA, SE] [SA|SA] [SA|SA, SE] [SA|SE] are met in Campylanthus. First, the split between the Macaronesian clade and the Eritreo–Arabian one is corroborated, and, second, the divergence time of these groups is estimated to be 4.68 (2.00–8.07) and 6.15 (3.02–9.84) mya using BEAST and MULTIDIVTIME, respectively. Whereas the results of MULTIDIVTIME coincide with the proposed inception of climate change 7 mya, age estimates by BEAST are somewhat younger. All present day Campylanthus species inhabit dry regions and show several adaptations to these environments (Hjertson, 1997, 2003). Possibly, ancestral, northern African populations were already drought adapted and were able to colonise drier niches within the sclerophyllous, evergreen woodland belt before they became subsequently victim of progressive, extreme aridification. This could explain a possible delayed extinction, especially of sclerophyllous taxa in the Saharan belt. Additionally, also in agreement with the vicariance hypothesis, our biogeographic analyses indicated a basal split between western and eastern groups of Campylanthus encompassing Macaronesia and east Africa/Arabia and possibly Socotra (Table 2). These findings make the postulation of trans-Saharan long-distance dispersal redundant. The climatic shift across northern Africa since the Upper Miocene should have led to similarly vicariant trans-Saharan distribution patterns in other plant groups. Such taxa were recently reviewed by Andrus et al. (2004). The evidence from phylogenetic analyses of such taxa is ambiguous. A vicariance scenario is refuted, for instance, for Tolpis Adanson (Asteraceae, Park et al., 2001; Moore et al., 2002) and Ceropegia L. (Apocynaceae, Bruyns, 1985). Biogeographic relationships between Macaronesia and east Africa/Arabia do exist for Aeonium. Here, the east African A. leucoblepharum Webb ex A. Rich. is deeply nested within the Macaronesian clade (Mort et al., 2002) and has probably dispersed from west to east no later than in the Pliocene or Pleistocene (Kim et al., 2008; Thiv et al., in press), possibly facilitated by the presence of numerous small seeds. For other taxa, available data are consistent with the vicariance hypothesis, for instance for Nanorrhinum Betsche (Plantaginaceae, Ghebrehiwet, 2000), Kleinia Mill. (Asteraceae, Pelser et al., 2007; M. Thiv, unpubl. data) and Dracaena (Marrero et al., 1998). The same applies to Camptoloma (Kornhall et al., 2001), in which sister taxa occur across the Sahara, and have links to southern Africa, as also shown for Plocama Ait. (Backlund et al., 2007) illustrating a pattern of the Rand flora (Sanchez-Meseguer et al., 2009). Even the present day distribution of Globularia in Macaronesia/Mediterranean area/northern Europe and Poskea in Eritreo– Lagrange model 2 lnL 18.17 19.13 17.72 18.41 19.45 19.52 20.36 17.97 19.39 17.72 18.29 18.86 17.87 19.72 17.98 19.25 17.72 Rel. prob. 0.6402 0.2452 1 0.5026 0.1788 0.1665 0.0718 0.7797 0.189 1 0.5693 0.3221 0.8605 0.1365 0.7749 0.2175 1 Split [SA, SE|MA] [SE|MA] [MA|MA] [SE|SA, SE] [SE|SA] [SE|SE] [SE|SE] [SE, SO|SE] [SE|SO] [SA|SA, SE] [SA|SA] [SE|SE] [SA|SE] [SA|SA] [SA|SA, SE] [SA|SA] [SA|SA, SE] [SA|SE] lnL 18.82 19.68 18.36 19.14 19.62 19.97 19.04 19.14 18.36 19.26 19.59 20.41 21.17 18.74 19.56 18.86 19.35 18.36 Rel. prob. 0.6279 0.265 1 0.4581 0.2814 0.2002 0.5036 0.4583 1 0.4073 0.2911 0.1286 0.0602 0.6848 0.3004 0.6042 0.3687 1 Arabia may reflect such a vicariance origin (Wagenitz, 2004; Albach et al., 2005). 4.3. Biogeographic history The aridification of the Sahara may have divided a formerly widespread ancestor of Campylanthus into a western and eastern group. The western lineage is today represented by C. salsoloides on the Canary Islands and C. glaber on the Cape Verde Islands. The divergence time between these species in the Pleistocene–Pliocene (node b, Table 2) is much younger than the geological ages of the oldest Canarian island (Fuerteventura, 20.6 mya, Carracedo et al., 2002) and of Cape Verde island (Miocene, Mitchell-Thomé, 1972; Rothe, 1982; Boekschoten and Manuputty, 1993). A vicariance scenario giving rise to a differentiation between the two Macaronesian species can be ruled out because there is no evidence that these two volcanic archipelagos were ever connected with each other (Carracedo et al., 2002). Accordingly, Campylanthus probably colonised these islands via long-distance dispersal. With the data available it is not possible to determine whether the Macaronesian clade originated in the Canary Islands and dispersed to the Cape Verde Islands, like Sonchus and Aeonium (Kim et al., 2008), or vice versa. Whatever the route among the island, the ancestor must have come from the mainland. Close biogeographic relationships between north-western Africa/Mediterranean on the one hand and the Canary islands on the other hand have been suggested for numerous plants. Phylogenetic analyses, e.g., of the Sonchus alliance (Asteraceae), dragon trees, Ixanthus Griseb. (Gentianaceae) or Macaronesian Crassulaceae–Sempervivoideae (Kim et al., 1996, 2008; Marrero et al., 1998; Thiv et al., 1999, in press; Mort et al., 2002) corroborated the hypothesis that the mainland served as source area for the island colonisation (Médail and Quézel, 1999; Sanmartín et al., 2008). A third explanation is that C. salsoloides and C. glaber could have reached the islands independently from (north-)western Africa as shown for bryophytes and pteridophytes (Vanderpoorten et al., 2007), where this lineage later became extinct. The eastern clade of African–Arabian species of Campylanthus is geographically structured, and these biogeographic patterns could reflect another case of vicariance. The mostly African group of C. incanus, C. reconditus, and C. spinosus is sister to the remaining primarily Arabian taxa. Most of these species grow along the Gulf of Aden, the Somalia block and parts of the eastern Red Sea (Fig. 1). Author's personal copy 614 M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 Their basal split probably falls into the Lower Pliocene/Upper Miocene (node c: Table 2). During this period repeated land connections between Africa and Arabia enabled several African mammal species to colonise Central Europe. This migration route was closed when oceanic rifting opened the south-central Red Sea and the propagation of the Sheba Ridge widened the Gulf of Aden (Steininger et al., 1985; Bosworth et al., 2005). These events correspond in time to the Messinian salinity crisis of the Mediterranean basin 6 mya and might have separated the primarily African and Arabian groups of Campylanthus. This pattern is indicated by splits between SA and SE in Lagrange analyses (Table 2: node c), but show rather low likelihoods under both models. Although Lagrange indicates a division between SE and SA for S. somaliensis and its sister (node i), vicariance is unlikely because the estimated age is much younger than the geological separation. Alternatively, all Eritreo/Arabian disjunctions could also be explained with dispersal events. In the dispersal scenario either eastern Africa or Arabia was the area of origin. If eastern Africa was the area of origin (Table 2: node c, model 2), then the southern Arabian C. sedoides and C. pungens groups, irrespective of C. junceus, are the result of either two independent (Table 2: nodes g and i) or a single dispersal event from eastern Africa (Table 2: node f, model 1). In this scenario, the east African distribution of C. somaliensis is more likely explained by secondary back-dispersal from Arabia because no evidence for vicariance is coeval with the age determination. If the entire group is postulated to have originated in Arabia, then Africa was colonised twice (Fig. 3: nodes d and i). Under a parsimony criterion, however, these dispersalist scenarios for the African–Arabian taxa are less likely since they require more steps than a vicariance explanation. Even if the vicariance hypothesis is accepted, long-distance dispersals must have been involved where no geographic links can be assumed. This is the case for C. junceus which occurs in Arabia and Africa, for the Sindian C. ramossisimus from Arabia to Pakistan and for C. spinosus from east Africa to Socotra. The estimated age of the Socotran accession of up to 3.12 mya clearly falls within the geological age of the island group of at least 15 mya (Fleitmann et al., 2004) and is younger than that of other Socotran plants (Aerva Forsk., Thiv et al., 2006), which seems plausible regarding the conspecifity of African and Socotran C. spinosus. In conclusion, our data support the hypothesis for a vicariant origin of the disjunct distribution of Campylanthus between Macaronesia and east Africa–Arabia. This may be the result of the aridification of the Sahara in the Upper Miocene and Pliocene. Vicariance resulting from climatic changes, as suggested for Campylanthus, may be a much more common process then hitherto assumed. Acknowledgments The authors thank Gerald M. Schneeweiss (Vienna, Austria), Merijn Bos (Stuttgart, Germany) and Arno Wörz (Stuttgart, Germany) for valuable comments on the paper, Frank Rutschmann (Bern, Switzerland), Daniele Silvestro (Frankfurt, Germany) and Andreas Franzke (Heidelberg, Germany) for helpful advice for using MULTIDIVTIME, Lagrange, and BEAST, respectively, Veronika Wähnert (Freiburg, Germany), Johanna Eder (Stuttgart, Germany) and Barbara Mohr (Berlin, Germany) for the evaluation of the fossil record, Norbert Kilian (Berlin, Germany) for plant material and literature, Mohamed Ali Hubaishan (AREA Research Station Mukalla), Ahmed Said Sulaiman (EPA Socotra), Said Masood Awad Al-Gareiri (Dept. Agriculture Socotra), and Mohamed El-Mashjary (EPA Sanaa; all Yemen) for support of the field work on Socotra. The field work was conducted as part of the BIOTA Yemen Project funded by the German Ministry for Research and Education (BMBF). This study was supported by Grants of the German Research Founda- tion (DFG, Th830/1-1) and the Claraz-Schenkung (Switzerland) to the first author. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ympev.2009.10.009. References Albach, D.C., Meudt, H.M., Oxelman, B., 2005. Piecing together the ‘‘new’’ Plantaginaceae. Am. J. Bot. 92 (2), 297–315. Alvarez, I., Wendel, J.F., 2003. Ribosomal ITS sequences and plant phylogenetic inference. Mol. Phylogenet. Evol. 29, 417–434. Andrus, N., Trusty, J., Santos-Guerra, A., Jansen, R.K., Francisco-Ortega, J., 2004. Using molecular phylogenies to test phytogeographical links between East/South Africa–Southern Arabia and the Macaronesian islands – a review, and the case of Vierea and Pulicaria section Vieraeopsis (Asteraceae). Taxon 53, 333–346. Axelrod, D.I., 1975. Evolution and biogeography of Madrean–Tethyan sclerophyll vegetation. Ann. Missouri Bot. Gard. 62, 280–334. Backlund, M., Bremer, B., Thulin, M., 2007. Paraphyly of Paederieae, recognition of Putorieae and expansion of Plocama (Rubiaceae–Rubioideae). Taxon 56, 315– 328. Baldwin, B.G., Sanderson, M.J., Porter, J.M., Wojciechowski, M.F., Campbell, C.S., Donoghue, M.J., 1995. The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Ann. Missouri Bot. Gard. 82, 247– 277. Bello, M.A., Chase, M.W., Olmstead, R.G., Rønsted, N., Albach, D.C., 2002. The Páramo endemic Aragoa is the sister genus of Plantago (Plantaginaceae; Lamiales): evidence from plastid rbcL and nuclear ribosomal ITS sequence data. Kew Bull. 57, 585–597. Bentham, G., Hooker, J.D., 1886. Genera plantarum, vol. 2. Part 2. Reeve & Co., London. Blattner, F.R., 1999. Direct amplification of the entire ITS region from poorly preserved plant material using recombinant PCR. BioTechniques 27, 1180– 1186. Boekschoten, G.L., Manuputty, J.A., 1993. The age of the Cape Verde Islands. Courier Forschungsinstitut Senckenberg 159, 3–5. Bosworth, W., Huchon, H., McClay, K., 2005. The Red Sea and Gulf of Aden Basins. J. Afr. Earth Sci. 43, 334–378. Bramwell, D., 1976. The endemic flora of the Canary Islands. In: Kunkel, G. (Ed.), Biogeography and Ecology in the Canary Islands. Junk, The Hague, pp. 207–240. Bramwell, D., 1985. Contribución a la biogeografia de las Islas Canarias. Bot. Macaronés 14, 3–34. Bruyns, P., 1985. The genus Ceropegia on the Canary Islands (Asclepiadaceae– Ceropegieae): a morphological and taxonomic account. Beitr. Biol. Pflanzen 60, 427–458. Carracedo, J.C., Pérez, F.J., Ancochea, E.J.M., Hernán, F., C.R., Cubas, Casillas, R., Rodriguez, E., Ahijado, A., 2002. Cenozoic volcanism II: the Canary Islands. In: Gibbons, W., Moreno, T. (Eds.), The Geology of Spain. The Geological Society of London, Bath, pp. 439–472. Darwin, C., 1859. On the Origin of Species. Murray, London. Deil, U., Müller-Hohenstein, K., 1984. Fragmenta Phytosociologica Arabiae-Felicis I – Eine Euphorbia-balsamifera-Gesellschaft aus dem jemenitischen Hochland und ihre Beziehungen zu makaronesischen Pflanzengesellschaften. Flora 175, 407– 426. Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7, 214. doi:10.1186/1471-2148-7-214. Engler, A., 1879. Versuch einer Entwicklungsgeschichte der Pflanzenwelt, insbesondere der Florengebiete seit der Tertiärperiode. I. Die extratropischen Gebiete der nördlichen Hemisphäre. Engelmann, Leipzig. 202. Felsenstein, J., 1981. Evolutionary trees from DNA sequences: a maximum likelihood approach. J. Mol. Evol. 17, 368–376. Fischer, E., 2004. Scrophulariaceae. In: Kadereit, J.W. (Ed.), The Families and Genera of Vascular Plants. Springer, Berlin, Heidelberg, pp. 333–432. Fleitmann, D., Matter, A., Burns, S.J., Al-Subbary, A., Al-Aowah, M. A., 2004. Geology and Quaternary climate history of Socotra. Fauna of Arabia 20, 27–43. Forest, F., Savolainen, V., Chase, M.W., Lupia, R., Bruneau, A., Crane, P.R., 2005. Teasing apart molecular-versus fossil-based error estimates when dating phylogenetic trees: a case study in the birch family (Betulaceae). Syst. Bot. 30 (1), 118–133. Ghebrehiwet, M., 2000. Taxonomy, phylogeny and biogeography of Kickxia and Nanorrhinum (Scrophulariaceae). Nordic J. Bot. 20, 655–690. Goldman, N., Anderson, J.P., Rodrigo, A.G., 2000. Likelihood-based tests of topologies in phylogenetics. Syst. Biol. 49, 652–670. Gray, J., 1964. Northwest American Tertiary palynology. The emerging picture. In: Cranwell, L.M. (Ed.), Ancient Pacific Floras. University of Hawaii Press, Honolulu, pp. 21–30. Harzhauser, M., Piller, W.E., 2004. Integrated stratigraphy of the Sarmatian (Upper Middle Miocene) in the western Central Paratethys. Stratigraphy 1, 65–86. Heads, M., 2005. Dating nodes on molecular phylogenies: a critique of molecular biogeography. Cladistics 21, 62–78. Author's personal copy M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 Hjertson, M., 1997. Systematics of Lindenbergia and Campylanthus (Scrophulariaceae). Acta Univ. Upsal. 331, 1–46. Hjertson, M., 2003. Revision of the disjunct genus Campylanthus (Scrophulariaceae). Edinb. J. Bot. 60 (2), 131–174. Hjertson, M., Miller, A.J., 2000. A new species and a new combination in Campylanthus (Scrophulariaceae). Edinb. J. Bot. 57, 221–226. Hjertson, M., Thulin, M., 2006. A new species of Campylanthus (Scrophulariaceae) from Somalia. Nord. J. Bot. 23 (6), 707–709. Hjertson, M., Henrot, J., Thulin, M., 2008. Campylanthus hajarensis sp. nov. and a new record of Campylanthus (Scrophulariaceae) from Oman. Nord. J. Bot. 26, 35–37. Hooker, J.D., 1878. On the Canarian flora as compared with the Maroccan. In: Hooker, J.D., Ball, J. (Eds.), Journal of a Tour in Marocco and the Great Atlas. Macmillan, London, pp. 404–421. Huelsenbeck, J.P., Hillis, D.M., Jones, R., 1996. Parametric bootstrapping in molecular phylogenies: application and performance. In: Ferris, J.D., Palumbi, S.R. (Eds.), Molecular Zoology: Advances, Strategies, and Protocols. Wiley, New York, pp. 19–45. Johnson, L.A., Soltis, D.E., 1998. Assessing congruence: empirical examples from molecular data. In: Soltis, D.E., Soltis, P.S., Doyle, J.J. (Eds.), Molecular Systematics of Plants II: DNA Sequencing. Kluwer Academic Publishers, New York, pp. 297–348. Keppel, G., Lowe, A.J., Possingham, H.P., 2009. Changing perspectives on the biogeography of the tropical South Pacific: influences of dispersal, vicariance and extinction. J. Biogeogr. 36, 1035–1054. Kilian, N., Hein, P., Bahah, S.O., 2002. A new species of Campylanthus (Scrophulariaceae) from Ras Fartak, Al-Mahra, and notes on other species of the genus in Yemen. Willdenowia 32, 271–279. Kim, S.C., Crawford, D.J., Francisco-Ortega, J., Santos-Guerra, A., 1996. A common origin for woody Sonchus and five related genera in the Macaronesian islands: molecular evidence for extensive radiation. Proc. Natl. Acad. Sci. USA 93, 7743– 7748. Kim, 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 Macaronesia. PLoS One 3 (5), e2139. doi:10.1371/journal.pone.0002139. Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29, 170–179. Kishino, H., Thorne, J.L., Bruno, W.J., 2001. Performance of a divergence time estimation method under a probabilistic model of rate evolution. Mol. Biol. Evol. 18, 352–361. Kornhall, P., Heidari, N., Bremer, B., 2001. Selagineae and Manuleeae, two tribes or one? Phylogenetic studies in the Scrophulariaceae. Pl. Syst. Evol. 228, 199–218. Linder, H.P., 2001. On areas of endemism, with an example from the African Restionaceae. Syst. Biol. 50, 892–912. Manen, J.-F., Natali, A., Ehrendorfer, F., 1994. Phylogeny of Rubiaceae–Rubieae inferred from the sequence of a cpDNA intergene region. Pl. Syst. Evol. 190, 195–211. Mason-Gamer, R.J., Kellogg, E.A., 1996. Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae (Gramineae). Syst. Biol. 45, 524–545. Marrero, A., Almeida, R.S., González-Martín, M., 1998. A new species of the wild dragon tree, Dracaena (Dracaenaceae) from Gran Canaria and its biogeographic implications. Bot. J. Linn. Soc. 128, 291–314. Médail, F., Queézel, P., 1999. The phytogeographic significance of S.W. Morocco compared to the Canary Islands. Pl. Ecol. 140, 221–244. Meusel, H., 1965. Die Reliktvegetation der Kanarischen Inseln in ihren Beziehungen zur süd-und mitteleuropäischen Flora. In: Gersch, M. (Ed.), Gesammelte Vorträge über moderne Probleme der Abstammungslehre. Friedrich-Schiller University, Jena, pp. 117–136. Mies, B.A., 1995. On the comparison of the flora and vegetation of the island groups Socotra and Macaronesia. Bol. Mus. Munic. Funchal, Supl. 4, 455–471. Miller, A.G., 1980. A revision of Campylanthus. Notes Roy. Bot. Gard. Edinb. 38, 373– 385. Miller, A.G., 1982. Further notes on Campylanthus. Notes Roy. Bot. Gard. Edinb. 40, 331–332. Miller, A.G., 1988. Two new species of Campylanthus. Notes Roy. Bot. Gard. Edinb. 45, 73–76. Mitchell-Thomé, R., 1972. Outline of the geology of the Cape Verde Archipelago. Geol. Rundsch. 61, 1087–1109. Mohr, B.A.R., 1984. Die Mikroflora der obermiozänen bis unterpliozänen Deckschichten der Rheinischen Braunkohle. Paleontogr. Abt. B 191, 29–133. Moline, P., Thiv, M., Ameka, G.K., Ghogue, J.-P., Pfeiffer, E., Rutishauser, R., 2007. Comparative morphology and molecular systematics of African Podostemaceae–Podostemoideae, with emphasis on Dicraeanthus and Ledermanniella from Cameroon. Int. J. Plant Sci. 168, 159–180. Moore, M.J., Francisco-Ortega, J., Santos-Guerra, A., Jansen, R.K., 2002. Chloroplast DNA evidence for the roles of island colonization and extinction in Tolpis (Asteraceae: Lactuceae). Am. J. Bot. 89, 518–526. Mort, M.E., Soltis, D.E., Soltis, P.S., Francisco-Ortega, J., Santos-Guerra, A., 2002. Phylogenetics and evolution of the Macaronesian clade of Crassulaceae inferred from nuclear and chloroplast sequence data. Syst. Bot. 27, 271–288. Muller, J., 1981. Fossil pollen records of extant angiosperms. Bot. Rev. 47, 1–142. Nagy, E., 1963. Some new spore and pollen species from the Neogene of the Mecsek Mountain. Acta Bot. Hung. 9, 387–404. Naud, G., Suc, J.-P., 1975. Contribution á l’étude paléofloristique des Coirons (Ardèche): premières analyses pollinique dans les alluvions sous-basaltiques et 615 interbasaltiques de Mirabel (Miocène supérieur). Bull. Soc. Géol. France 17, 820–827. Nei, M., Kumar, S., 2000. Molecular Evolution and Phylogenetics. Oxford University Press, Oxford. Nelson, G., 1978. From Candolle to Croizat: comments on the history of biogeography. J. Hist. Biol. 11, 269–305. Nilsson, S., Hong, D.y., 1993. The taxonomic significance of Aragoa-pollen (Scrophulariaceae). Opera Bot. 121, 275–278. Park, S.-J., Korompai, E.J., Francisco-Ortega, J., Santos-Guerra, A., Jansen, R.K., 2001. Phylogenetic relationships of Tolpis (Asteraceae: Lactuceae) based on ndhF sequence data. Pl. Syst. Evol. 226, 23–33. Pelser, P.B., Nordenstam, B., Kadereit, J.W.L.W., 2007. An ITS phylogeny of tribe Senecioneae (Asteraceae) and a new delimitation of Senecio L. Taxon 56 (4), 1077–1104. Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817–818. de Queiroz, A., 2005. The resurrection of oceanic dispersal in historical biogeography. Trends Ecol. Evol. 20, 68–73. Quézel, P., 1978. Analysis of the flora of Mediterranean and Saharan Africa. Ann. Missouri Bot. Gard. 65, 479–534. Rambaut, A., Grassly, N.C., 1997. Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Comput. Appl. Biosci. 13, 235–238. Ree, R.H., Smith, S.A., 2008. Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 4–14. Reichenbach, H.G.L., 1828. Conspectus Regni Vegetabilis per Gradus Naturales Evoluti. Carolum Cnobloch, Leipzig. Renner, S., 2004. Plant dispersal across the tropical Atlantic by wind and sea currents. Int. J. Pl. Sci. 165, S23–S33. Richardson, J.E., Weitz, M.F., Fay, M.F., Cronk, Q.C.B., Linder, H.P., Reeves, G., Chase, M.W., 2001. Rapid and ancient origin of species richness in the Cape Flora of South Africa. Nature 412, 181–183. Rodriguez, F., Oliver, J.L., Marin, A., Medina, J.R., 1990. The general stochastic model of nucleotide substitution. J. Theor. Biol. 142, 485–501. Rønsted, N., Chase, M.W., Albach, D.C., Bello, M.A., 2002. Phylogenetic relationships within Plantago (Plantaginaceae): evidence from nuclear ribosomal ITS and plastid trn L-F sequence data. Bot. J. Linn. Soc. 139, 323–338. Rothe, P., 1982. Zur Geologie der Kapverdischen Inseln. Courier Forschungsinstitut Senckenberg 52, 1–9. Rutschmann, F., 2004. Bayesian Molecular Dating using PAML/Multidivtime. A Stepby-Step Manual. University of Zurich, Switzerland. Saad, S.I., 1982. Palynological studies in the genus Plantago L. (Plantaginaceae). Pollen et Spores 28, 43–60. Sanchez-Meseguer, A., Aldasoro, J.J., Anderson, C.L., Sanmartín, I., 2009. Origin of the Afro-Mediterranean Rand flora: integrating phylogenetic and model-based biogeographic methods in the study of the genus Hypericum L. (Guttiferae). Abstract Systematics 2009. Available from: <http://www.boerhaavextern.nl/ biosyst/Site/downloads/Systematics2009abstracts.pdf>. Sanderson, M.J., 1998. Estimating rate and time in molecular phylogenies: beyond the molecular clock? In: Soltis, D.E., Soltis, P.S., Doyle, J.J. (Eds.), Molecular Systematics of Plants. Kluwer, Boston, Dordrecht, London, pp. 242–264. Sanmartín, I., van der Mark, P., Ronquist, F., 2008. Inferring dispersal: a Bayesian approach to phylogeny-based island biogeography, with special reference to the Canary Islands. J. Biogeo. 35, 428–449. Schuster, M., Duringer, P., Ghienne, J.F., Vignaud, P., Mackaye, H.T., Likius, A., Brunet, M., 2006. The age of the Sahara desert. Science 311, 1138–1139. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of loglikelihoods with applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116. Shimodaira, H., 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51, 492–508. Stefanovic, S., Olmstead, R.G., 2004. Testing the phylogenetic position of a parasitic plant (Cuscuta, Convolvulaceae, Asteridae): Bayesian Inference and the parametric bootstrap on data drawn from three genomes. Syst. Biol. 53, 384– 399. Steininger, F.F., Rabeder, G., Rögl, F., 1985. Land mammal distribution in the Mediterranean Neogene: a consequence of geokinematic and climatic events. In: Stanley, D.J., Wezel, F.-C. (Eds.), Geological evolution of the Mediterranean Basin. Springer Verlag, New York, Berlin, pp. 559–571. Sunding, P., 1979. Origins of the Macaronesian flora. In: Bramwell, D. (Ed.), Plants and Islands. Academic Press, London, pp. 13–40. Swofford, D.L., 2002. PAUP* 4.0 – Phylogenetic Analysis Using Parsimony (* and Other Methods). Sinauer Association, Sunderland, MA. Takhtajan, A., 1986. Floristic regions of the world. University of California Press, Berkeley, Los Angeles, London. Thiede, J., 1994. New aspects of the phytogeographic relations of Macaronesian Crassulaceae. In: Seyani, J.H., Chikuni, A.C. (Eds.), Proceedings of the XIIIth Plenar Meeting AETFAT. National Herbarium and Botanic Gardens of Malawi, Zomba, pp. 1121–1143. Thiv, M., Struwe, L., Kadereit, J.W., 1999. The phylogenetic relationships and evolution of the Canarian laurel forest endemic Ixanthus viscosus (Ait.) Griseb. (Gentianaceae): evidence from matK and ITS sequence variation, and floral morphology and anatomy. Pl. Syst. Evol. 218, 299–317. Thiv, M., Thulin, M., Kilian, N., Linder, H.P., 2006. Eritreo–Arabian affinities of the Socotran Flora as revealed from the molecular phylogeny of Aerva (Amaranthaceae). Syst. Bot. 31, 560–570. Author's personal copy 616 M. Thiv et al. / Molecular Phylogenetics and Evolution 54 (2010) 607–616 Thiv, M., Esfeld, K., Koch, M., in press. Studying adaptive radiation at the molecular level: a case study in the Macaronesian Crassulaceae–Sempervivoideae. In: Glaubrecht, M., Schneider, H. (Eds.), Evolution in Action – Adaptive Radiations and the Origins of Biodiversity. Springer Verlag, Hamburg. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The ClustalX windows interface. flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876– 4882. Thorne, J.L., Kishino, H., Painter, I.S., 1998. Estimating the rate of evolution of the rate of molecular evolution. Mol. Biol. Evol. 15, 1647–1657. Thorne, J.L., Kishino, H., 2002. Divergence time and evolutionary rate estimation with multilocus data. Syst. Biol. 51, 689–702. Thulin, M., 1994. Aspects of disjunct distributions and endemism in the arid parts of the Horn of Africa, particularly Somalia. In: Seyani, J.H., Chikuni, A.C. (Eds.), Proceedings of the XIIIth Plenar Meeting AETFAT. National Herbarium and Botanic Gardens of Malawi, Zomba, pp. 1105–1119. Van Damme, K., 2009. Socotra archipelago. In: Gillespie, R. G., Clague, D.A. (Eds.), Encyclopedia of Islands. University of California Press, London, pp. 846–851. Vanderpoorten, A., Rumsey, F.J., Carine, M.A., 2007. Does Macaronesia exist? Conflicting signal in the bryophyte and pteridophyte floras. Am. J. Bot. 94, 625– 639. Wagenitz, G., 2004. Globulariaceae. In: Kadereit, J.W. (Ed.), The Families and Genera of Vascular Plants. Springer Verlag, Berlin, Heidelberg, pp. 159–162. Wettstein, R.v., 1891. Scrophulariaceae. In: Engler, A. (Ed.), Die natürlichen Pflanzenfamilien nebst ihren Gattungen und wichtigeren Arten, insbesondere den Nutzpflanzen. Engelmann, Leipzig, pp. 39–107. Wiens, J.J., 1998. Combining data sets with different phylogenetic histories. Syst. Biol. 47, 568–581. Wikström, N., Savolainen, V., Chase, M.W., 2001. Evolution of angiosperms: calibrating the family tree. Proc. R. Soc. Lond. B Biol. Sci. 268, 2211–2220. Yang, Z., 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. CABIOS 13, 555–556.