Evolution of reproductive traits and selfing syndrome in the sub-endemic Mediterranean genus Centaurium Hill (Gentianaceae)

© 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 216 Botanical Journal of the Linnean Society, 2019, 191, 216–235. With 4 figures. Evolution of reproductive traits and selfing syndrome in the sub-endemic Mediterranean genus Centaurium Hill (Gentianaceae) VANIA JIMÉNEZ-LOBATO1,*, MARCIAL ESCUDERO1,, ZOILA DÍAZ LIFANTE1, CRISTINA ANDRÉS CAMACHO1, ALEJANDRA DE CASTRO1, GUILHEM MANSION2, LOUIS ZELTNER3 and JUAN ARROYO1 1Departamento de Biología Vegetal y Ecología, Facultad de Biología, Universidad de Sevilla, Spain 2Freelance writer, Zurich, Switzerland 3Laboratoire de Botanique Evolutive, Université de Neuchâtel, Neuchâtel, Switzerland Received 18 June 2018; revised 22 May 2019; accepted for publication 18 June 2019 In flowering plants, the shift from outcrossing to selfing is associated with a set of correlated changes in morphological and reproductive features known as the ‘selfing syndrome’. Species of the sub-endemic Mediterranean genus Centaurium Hill (Gentianaceae) exhibit a wide array of flower traits related to pollination biology and different ploidy levels. We explored if the evolutionary transitions of seven flower traits and life cycle, typically associated with the selfing syndrome, are related to polyploidy, diversification patterns, divergence times and the geological and climatic history of the Mediterranean Basin. Using 26 species of Centaurium we reconstructed a phylogenetic tree, inferred the ancestral states of the selected traits, estimated their phylogenetic signal and tested the correlative evolution among them. We found a significant increase in diversification rates during the Quaternary. Anther length, flower size, herkogamy and polyploidy undergo rapid state transitions without phylogenetic signal that could be the result of adaptation for selfing. Changes in character states do not show evidence of correlative evolution among them, as would be predicted during the evolution of selfing syndrome. The evolution of reproductive traits in Centaurium has probably relied on a more diverse array of drivers than just reproductive assurance or polyploidy events. KEYWORDS: comparative method – flower evolution – herkogamy – Iberian Peninsula – mating system – Mediterranean Basin – phylogenetic tree – polyploidy – reproductive assurance – selfing. INTRODUCTION Flowering plants have moulded their history in close association with pollinators. The world’s astonishing diversity of flowers is regarded as a logical outcome in plants pollinated by animals (Baker & Stebbins, 1965). On the other hand, during many periods of time and in many regions angiosperms have had to confront a lack or scarcity of adequate pollinators. In the past, the ability to self-fertilize has played a role in the persistence of certain lineages under environmental conditions characterized by a shortage or unsuitability of pollinators (Baker, 1955; Pannell, 2015). The shift from outcrossing to partial or complete selfing is one of the most frequent evolutionary transitions in flowering plants (Goodwillie, Kalisz & Eckert, 2005; Eckert et al., 2010). The critical stage in the transition towards selfing is the loss of a selfincompatibility system (SI) (Baker, 1955; Goodwillie, 1999). Once SI is lost, self-compatible (SC) species can outcross, self or develop a mixed mating system. Although a mating system can be measured directly by outcrossing rates (t), the flower phenotype and reproductive changes associated with transitions towards selfing are frequently used as a proxy given the difficulties inherent in estimating outcrossing rates in a large number of populations and species over time (Goodwillie et al., 2010; Sicard & Lenhard, 2011). The selfing syndrome consists of a set of correlated reproductive features usually associated with self-pollination, be it autonomous or mediated by pollinators. For instance, a reduction in the distance between anthers and stigmas within the same Keywords=Keywords=Keywords_First=Keywords HeadA=HeadB=HeadA=HeadB/HeadA HeadB=HeadC=HeadB=HeadC/HeadB HeadC=HeadD=HeadC=HeadD/HeadC Extract3=HeadA=Extract1=HeadA REV_HeadA=REV_HeadB=REV_HeadA=REV_HeadB/HeadA REV_HeadB=REV_HeadC=REV_HeadB=REV_HeadC/HeadB REV_HeadC=REV_HeadD=REV_HeadC=REV_HeadD/HeadC REV_Extract3=REV_HeadA=REV_Extract1=REV_HeadA BOR_HeadA=BOR_HeadB=BOR_HeadA=BOR_HeadB/HeadA BOR_HeadB=BOR_HeadC=BOR_HeadB=BOR_HeadC/HeadB BOR_HeadC=BOR_HeadD=BOR_HeadC=BOR_HeadD/HeadC BOR_Extract3=BOR_HeadA=BOR_Extract1=BOR_HeadA EDI_HeadA=EDI_HeadB=EDI_HeadA=EDI_HeadB/HeadA EDI_HeadB=EDI_HeadC=EDI_HeadB=EDI_HeadC/HeadB EDI_HeadC=EDI_HeadD=EDI_HeadC=EDI_HeadD/HeadC EDI_Extract3=EDI_HeadA=EDI_Extract1=EDI_HeadA CORI_HeadA=CORI_HeadB=CORI_HeadA=CORI_HeadB/HeadA CORI_HeadB=CORI_HeadC=CORI_HeadB=CORI_HeadC/HeadB CORI_HeadC=CORI_HeadD=CORI_HeadC=CORI_HeadD/HeadC CORI_Extract3=CORI_HeadA=CORI_Extract1=CORI_HeadA ERR_HeadA=ERR_HeadB=ERR_HeadA=ERR_HeadB/HeadA ERR_HeadB=ERR_HeadC=ERR_HeadB=ERR_HeadC/HeadB ERR_HeadC=ERR_HeadD=ERR_HeadC=ERR_HeadD/HeadC ERR_Extract3=ERR_HeadA=ERR_Extract1=ERR_HeadA INRE_HeadA=INRE_HeadB=INRE_HeadA=INRE_HeadB/HeadA INRE_HeadB=INRE_HeadC=INRE_HeadB=INRE_HeadC/HeadB INRE_HeadC=INRE_HeadD=INRE_HeadC=INRE_HeadD/HeadC INRE_Extract3=INRE_HeadA=INRE_Extract1=INRE_HeadA App_Head=App_HeadA=App_Head=App_HeadA/App_Head BList1=SubBList1=BList1=SubBList BList1=SubBList3=BList1=SubBList2 SubBList1=SubSubBList3=SubBList1=SubSubBList2 SubSubBList3=SubBList=SubSubBList=SubBList SubSubBList2=SubBList=SubSubBList=SubBList SubBList2=BList=SubBList=BList applyparastyle “fig//caption/p[1]” parastyle “FigCapt” *Corresponding author: E-mail: [email protected] Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 217 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 flower (herkogamy) (Herlihy & Eckert, 2007) or a diminution in the size of perianth traits that attract pollinators (Goodwillie et al., 2010) are commonly reported. However, selfing increases homozygosity in populations, which itself may generate inbreeding depression (Charlesworth & Willis, 2009), a lack of potential for novel allelic combinations (Glemin, Bazin & Charlesworth, 2006; Wright et al., 2008), or drive a drastic increase in extinction rates (Goldberg et al., 2010; Igic & Busch, 2013). For this reason, selfing has been described as a hypothetical evolutionary ‘deadend’ (Stebbins, 1957). The advantage of selfing over outcrossing can be explained by several factors. The two main theories are (1) the ‘automatic advantage of selfing’ (Fisher, 1941), based on the 3:2 potential gene-level advantage of selfing genes over genes causing outcrossing, and (2) selfing as a mechanism of reproductive assurance (Baker, 1955; Stebbins, 1957). The latter factor refers to the fact that, under conditions of pollen limitation, an increase in the self-fertilization rate may in fact be beneficial for populations and ensure sexual reproduction (Morgan & Wilson, 2005; Eckert, Samis & Dart, 2006). The evolution of selfing is promoted in colonizer, pioneer and annual/short-lived species in recently available and disturbed habitas, probably because these species frequently have to withstand a scarcity or dearth of pollinators (Barrett, 2010; Pannell, 2015). The reproductive assurance hypothesis has been tested both at macro- and micro-evolutionary levels (e.g. Goodwillie, 1999; Kalisz, Vogler & Hanley, 2004; Moeller & Geber, 2005). Moreover, modern genomic tools are now being applied in an attempt to unravel the underlying genetic mechanisms in, for example, selfing Capsella Medik. species (Foxe et al., 2009; Sas et al., 2016; Sicard et al., 2016). A third but less well understood theory explaining the evolution of selfing is polyploidy. Polyploid plants frequently have higher selfing rates than their diploid relatives (Stebbins, 1950; Barringer, 2007; Glick et al., 2016) due potentially to a reduction in inbreeding depression. This could be provoked by the presence of multiple gene copies hindering the formation of homozygotes, given the ability of polyploids to break down the SI and the minority-cytotype disadvantage (Levin, 1975), which can occur in neopolyploids co-occurring with their diploid progenitors (Barringer, 2007). Polyploidization is regarded as a unidirectional and irreversible process (Ramsey & Schemske, 2002; Meyers & Levin, 2006; the reversion to the original ploidy level, polyhaploidy, albeit possible, is extremely rare). Polyploids have sometimes been considered ‘dead-ends’ that are present more often at the tips of phylogenies and tend to be nested within larger diploid groups (Mayrose et al., 2011, 2015). However, it is known that many plant radiations have been preceded by polyploidy events (Soltis et al., 2014). Evolutionary patterns of mating systems along lineages support the hypothesis of outcrossers as ancestors of selfing taxa (Takebayashi & Morrell, 2001; Wright et al., 2008). The SI to SC transition is frequent and follows Dollo’s law of irreversibility (Goodwillie, 1999; Igic, Bohs & Kohn, 2004). Similarly, the switch from outcrossing to selfing is most frequently encountered in clades in which SC has already evolved but which are not necessarily selfers (Takebayashi & Morrell, 2001). However, controversy still reigns regarding the ancestral and derived states of selfing traits, which in many cases are idiosyncratic of particular plant groups. For instance, as predicted, in some taxa such as Clarkia xantiana A.Gray (Moeller & Geber, 2005), Gilia achilleifolia Benth. (Takebayashi, Wolf & Delph, 2006), Eichhornia paniculata Solms (Vallejo-Marín & Barrett, 2009) and Mimulus guttatus DC. (Ritland & Ritland, 1989) the evolution of the selfing syndrome from outcrossing relatives has been accompanied by an increase in selfing rates. However, in other cases, floral traits related to outcrossing such as large corollas and approach herkogamy (the stigma protruding beyond the anthers; Webb & Lloyd, 1986) seem to be derived from states associated with selfing (small corolla and non-herkogamy) (Olmstead, 1989; Armbruster, 1993; de Vos, Wüest & Conti, 2013). It is evident that the directional transition from outcrossing towards selfing has occurred repeatedly in angiosperms (Eckert et al., 2010). Conversely, there is as yet no consensus as to (1) whether the selfing syndrome always evolves as a consequence of the evolution of selfing or is a by-product of the relatedness between species with no current adaptive value; (2) whether or not there is a directional trajectory of character states along lineages and (3) whether or not there are consistent patterns of phenotypic and genetic correlations between ‘selfing traits’ or between these traits and polyploidy. Studies using a phylogenetic framework to examine the evolution of the reproductive traits putatively associated with the respective mating strategies remain scarce. The family Gentianaceae Juss. and, in particular, the genus Centaurium Hill exhibit a wide array of flower forms and architectures (see Fig. 1) in, above all, structures related to pollination and reproductive biology (Struwe et al., 2002; Pringle, 2014). Available data in the literature suggest that the SC system is widespread in this family and in Centaurium species (Pringle, 2014; and Supporting Information, Table S1), which will help any analysis of the value of flower variation in selfing, regardless of a virtually constant SC. Hence, Centaurium is a good model for studying how a mating system such as the selfing syndrome Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 218 V. JIMÉNEZ-LOBATO ET AL. © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 219 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 may evolve mediated by morphological flower traits, independently of incompatibility systems (Randle, Slyder & Kalisz, 2009). Centaurium consists of 27–30 species/subspecies, some of which are polyploids (Mansion, Zeltner & Bretagnolle, 2005; Díaz Lifante, 2012). This genus is sub-endemic to the Mediterranean Basin, and just a few Centaurium species are found in temperate Europe, western and central Asia, North Africa and North America. Unfortunately, data on the predominant mating systems in Centaurium are very scarce and reports suggest that they are mainly facultative or obligate selfers (Pringle, 2014) (Supporting Information, Table S1). In three species – C. erythraea Rafn subsp. erythraea, C. pulchellum (Sw.) Druce and C. littorale subsp. littorale (Turner) Gilmour – selfing is associated with the evolution of the selfing syndrome as a reproductive assurance mechanism dependent on pollinator behaviour (Brys & Jacquemyn, 2011). However, it is not clear whether or not the selfing syndrome occurs throughout the whole genus or whether it is only found in one or just a few related clades. Diversification patterns in Centaurium are expected to be related to the complex geological and climatic history of the Mediterranean Basin, as has already been disentangled for a number of lineages (e.g. Cistaceae Juss., Guzmán & Vargas, 2009). In particular, transitions towards selfing could be related to harsher environmental conditions, as in the case of C. erythraea subsp. erythraea at micro-evolutionary scale (Brys & Jacquemyn, 2012). In addition, the high percentage of polyploids suggests that genome duplication may also have played an important role in the evolution of selfing in the genus (Stebbins, 1950; Mable, 2004). In this study, we explored the diversification patterns and divergence times, as well as the evolutionary transitions and trajectories, of the flower traits that are typically related to the mating system, life history and polyploidy in the genus Centaurium. As a result of the close relationship between the selfing strategy and certain floral traits reported in this genus (Brys & Jacquemyn, 2011, 2012), we used the traits putatively related to the selfing syndrome as a proxy for the mating system. Specifically, our goals were (1) to clarify the phylogenetic relationships within the genus, (2) to infer divergence times, (3) to explore diversification patterns, (4) to map the transitions and trajectories of polyploidy, and reproductive and life-history traits on the phylogenetic tree, (5) to study the phylogenetic signal in polyploidy and reproductive and life-history traits and, finally, (6) to test whether or not there is correlative evolution between herkogamy, flower size, flower display and polyploidy, expected under a transition towards the selfing syndrome. In attempting to answer these evolutionary questions, we hope to throw more light on how different plants confront environmental changes, in particular those associated with disturbances and pollinator decline, two processes that are predicted to intensify as a result of global change. MATERIAL AND METHODS Study plantS Centaurium, mainly found in the Mediterranean Basin, Europe and western Asia, is the most speciesrich genus in subtribe Chironiinae of Gentianaceae (Mansion et al., 2005). Its monophyly has already been demonstrated (Mansion et al., 2005). The most recent systematic studies of the genus have identified 27–30 species and subspecies, with c. 40% diploid and 60% polyploid taxa (probably c. 30% autopolyploids and 30% allopolyploids; Zeltner, 1970; Melderis, 1972a, b; Mansion et al., 2005; Díaz Lifante, 2012). Previous phylogenetic analyses based on sequences from plastid DNA (trnL intron and trnL-F spacer) and nrDNA (ITS) have inferred several major clades (Mansion et al., 2005). Mansion et al. (2005) report that allopolyploidy has played an important role in the evolutionary history of this genus. However, the allopolyploidy/ autopolyploidy origin of these polyploids has not yet been proven. The exception is C. discolor (Gand.) Ronniger, which has been shown to be an allopolyploid species (Guggisberg, Bretagnolle & Mansion, 2006). In the current study, we expand on the work by Mansion et al. (2005) and increase the number of taxa and studied DNA regions. A total of 69 samples belonging to 29 taxa were included in our study: 26 taxa (18 species and 8 subspecies) of Centaurium, Schenkia spicata (L.) G. Mans. and Exaculum pusillum Caruel as sister lineages, and Blackstonia perfoliata (L.) Huds. as an outgroup (Mansion & Struwe, 2004; Mansion et al., 2005). The taxonomic treatment proposed by Mansion & Struwe (2004), with modifications by Díaz Lifante Figure 1. Reproductive traits analysed in Centaurium species. A, Medium-to-low flower display in C. pulchellum (left) and C. tenuiflorum (right); B, High flower display in C. quadrifolium subsp. barrelieri. Variation in herkogamy level: C, High lateral herkogamy in C. grandiflorum subsp. boissieri; D, Medium herkogamy in C. grandiflorum subsp. majus; E, Nonherkogamy in C. pulchellum; F, Outline of a C. discolor flower; G, outline of a C. suffruticossum flower and H, Outline of a C. scilloides flower. Photographs A, D, E: Z. Díaz Lifante; B: V. Jiménez-Lobato; C: R. Santos-Gally and F–H were obtained from Díaz Lifante (2012). Fs, flower size; Al, anther length; Anz, zygomorphic androecium symmetry; Anr, radial androecium symmetry; Stc, curved style position; Ste, erect style position and Hk, herkogamy. Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 220 V. JIMÉNEZ-LOBATO ET AL. © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 (2012) for the species present in the Iberian Peninsula, was followed (Supporting Information, Table S2). The collection of the material was performed both in the field and in herbaria [Universidad de Sevilla (SEV), Real Jardín Botánico de Madrid (MA), Universidad de Santiago de Compostela (SANT), Universidad de Valencia (VAL), Universidad de Málaga (MGC), Neuchatel (NEU)]. Thirty-six accessions from ITS and the trnL-F spacer-trnL intron complex were downloaded from GenBank (Supporting Information, Table S3). trait characterization Nine traits (seven floral traits, polyploidy and life history) were selected to analyse evolutionary patterns. The selection of traits was based on their putative relationship with selfing and their variability across species samples. Trait values were obtained by direct observations in the field or greenhouses, from measurements of herbarium material and from diverse taxonomic and floristic sources (Supporting Information, Table S4). The ploidy (Pl) level was taken from Mansion et al. (2005) and the values for the rest of the traits from Mansion & Struwe (2004), Mansion et al. (2005) and Díaz Lifante (2012). Traits were coded as binaries as follows: ploidy level (Pl), 0: diploid, 1: polyploid; flower display (Fd, as total number of flowers in anthesis phase simultaneously), 0: low [pauciflora (< 30 flowers per plant)], 1: high [multiflora (> 30 flowers per plant)]; flower size (length) (Fs), 0: small- to-medium (< 15 mm), 1: large (> 15.1 mm); anther length (Al), 0: medium-to-large (> 2.1 mm), 1: short (< 2 mm); androecium symmetry (An), 0: radial, 1: zygomorphic; style position (S), 0: erect, 1: curved; herkogamy (Hk), 0: low (contact between stigmas and anthers), 1: high (no contact between stigmas and anthers); and stigma length (Sl), 0: medium-to-large (> 0.8 mm), 1: short (< 0.7 mm). The life-history traits were classified as follows: life cycle (Lc), 0: annual or biennial; 1: perennial. A binary categorization of certain continuous data (Fd, Fs, Al, Hk, Sl) was employed for the sake of simplicity in further phylogenetic correlation analyses with natural binary traits. Categories were based on the knowledge of the diagnostic value of the distribution of measurements published by Díaz Lifante (2012) for the Iberian plants, completed with measurements of herbaria specimens from other Euro-Mediterranean regions based in the same criteria (Díaz Lifante, 2012). The low sample number (N = 26) precluded a proper analysis of the frequency distribution of these continuous traits, which would have enabled us to select cut-off points. dna extraction, molecular markerS and Sequencing We extracted and isolated the total genomic DNA from 30 mg of leaf tissue using Invisorb Spin Plant Mini Kit (Stratec molecular, Berlin, Germany). For each sample, six plastid DNA regions (matK, psbA-trnH intergenic spacer, trnL intron + trnL-F intergenic spacer, rpoB, rpoC and atpF-atpH intergenic spacer) and three nuclear DNA regions [internal transcribed spacer (ITS), external transcribed spacer (ETS) and NADPH- cytochrome P450 reductase (CPR1)] were amplified and sequenced. PCR reactions were performed using a Veriti 96 Well Thermal Cycler (Thermo Fisher Scientific Inc., MA, USA) at the University of Seville, and sequencing reactions were performed in a Bio- Rad DNA Engine Tetrad 2 Peltier Thermal Cycler (Bio-Rad, CA, USA) using the ABI BigDye Terminator v.3.1 Cycle Sequencing Kit (Thermo Fisher Scientific Inc., MA, USA) in the Macrogene Europe Laboratory (Amsterdam, the Netherlands). The PCR mixture consisted of 5 μL 5× reaction buffer, 1 μL dimethyl sulphoxide solution, 1 μL 10 mg mL−1 bovine serum albumin solution, 1 μL 10 μM for each primer, 0.1 u per μL polymerase (My Taq Red), 1–2 μL template DNA and 11–12.5 μL distilled water, resulting in a total volume of 25 μL per reaction. We used the primers and PCR parameters employed previously in published studies on Centaurium species: Mansion et al. (2005) for ITS and trnL-F; Schwarz et al. (2009) for CPR1; Schaefer et al. (2011) for matK; Fazekas et al. (2008) for atpF-atpH, rpoB and rpoC and Kress & Erickson (2007) for psbA-trnH. The ETS DNA region had not been previously amplified or sequenced for any Centaurium species. We followed the conditions specified by Heenan et al. (2010) for Gentianella Moench. Due to the low success rates of amplification and sequencing, we designed a reverse primer for ETS (ETS-Gent_Rw: 5´-CAT GCA TGG CTT AAT CTT TGA -3´). The sequences obtained were edited with Sequencher v.4.1.4 (Gene Codes Corporation) and aligned using Muscle (Robert, 2004). We obtained 13 DNA matrices, one for each DNA region (nine), a combined plastid DNA matrix, an ITS + ETS combined matrix (including only those samples for which we were able to amplify and sequence ITS and ETS DNA regions), a nuclear DNA combined matrix (with only those samples for which we were able to amplify and sequence ITS and ETS DNA regions) and a nuclearplastid DNA combined matrix (including only those samples for which we were able to amplify and sequence ITS and ETS DNA regions). The accession numbers and localities sampled are described in the Supporting Information, Table S5. phylogenetic analySeS Bayesian inference analyses were performed on the 13 DNA matrices (see above) using a Markov chain Monte Carlo (MCMC) as implemented in MrBayes v.3.2 (Huelsenbeck et al., 2001; Ronquist & Huelsenbeck, Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 221 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 2003; Ronquist et al., 2012). The simplest model of DNA evolution for each DNA region was selected using the AIC criterion from the analysis of each matrix in jModeltest v.2.1.3 (Guindon & Gascuel, 2003; Darriba et al., 2012). Two independent analyses of four Metropolis-coupled Markov chains were run for 5 million generations for single DNA region matrices and 10 million generations for the combined matrices, with convergence and stationarity determined using Tracer v.1.6 (Rambaut et al., 2014). After a burn-in of 25%, the remaining trees were used to construct a majority-rule consensus tree using posterior probability (PP) values as a measure of clade support (Alfaro, Zoller & Lutzoni, 2003). diverSification timeS A chronogram with absolute times was first estimated using combined nuclear (ITS1 and ITS2) and plastid (trnL intron and trnL-F spacer) matrices in BEAST2 (Drummond et al., 2012; Bouckaert et al., 2014) including one randomly selected sample per Centaurium species and one sample per main lineage from subtribe Chironiinae (Mansion & Struwe, 2004), downloaded from GenBank (Supporting Information, Table S3). Clock and tree models were linked for the nuclear and plastid partitions. The site model was unlinked (two different nucleotide mutations schemes were used based on jModeltest results). Two tree models (Yule model and Birth-Death model) in two alternative analyses were performed and the tree model with the best fit using the Bayes factor was selected. Trees were estimated using an uncorrelated lognormal relaxed clock model. Three calibration points (two secondary calibration points and one calibration point based on a geological event) were used: (1) a normal distribution prior with a mean of 27.7 Mya and SD of 4.0 Myr was applied to the crown node of subtribe Chironiinae (Mathews, Ruigrok & Mansion, 2015); (2) a normal distribution prior with a mean of 9.0 Mya and SD of 1.7 Myr was applied to the crown node of the clade Gyrandra Griseb. – Sabatia Adans. (Mathews et al., 2015) and (3) a uniform prior with an upper bound of 15.1 Myr, lower bound of 0.0 Myr and offset of 1.0 Myr was applied to the crown node of the clade Blackstonia Huds. – Ixanthus Griseb. (Mansion & Struwe, 2004). This last calibration was implemented using a flat prior since the dispersal-speciation of Ixanthus on the island of Tenerife could have happened any time since its origin to the present. This strategy was adopted given the absence of reliable fossils for subtribe Chironiinae and, specifically, for the genus Centaurium (Mansion & Struwe, 2004). The settings of the calibration points were determined using the recommendations by Ho & Phillips (2009). Analyses were conducted using three independent MCMC runs of 10 million generations each. Run convergences and burn-ins were assessed in Tracer v.1.6 (Lemey et al., 2009). Maximum clade credibility trees were calculated with TreeAnnotator v.2.3.2 (Drummond & Rambaut, 2007) using a PP limit of 0.95, a maximum clade credibility tree and the mean heights options. To obtain an ultrametric tree (that is, a chronogram with branch length proportional to the absolute time in million years) with the highest resolution for performing the comparative phylogenetic analyses, we conducted a second analysis with BEAST2. The first BEAST2 analysis inferred respective ages of 10.3 Mya (95% highest posterior densities [HPD]: 8.3–12.3) for the crown node of the genus Centaurium and 14.9 Mya (95% HPD: 12.6–17.2) for crown node of the clade Schenkia Griseb. – Exaculum Caruel – Centaurium. Those estimates were then used as secondary calibration points in a 29-taxa dataset analysis (26 Centaurium samples, one per taxon randomly selected, plus one Schenkia, one Exaculum and one Blackstonia sample) using ITS, ETS and CPR1 DNA regions. The rest of the parameters were set as in the preliminary analysis. Despite sequencing and combining six plastid DNA regions in the MrBayes analyses, the combined plastid tree had a low resolution (mostly polytomies) and retrieved exceptionally only a few poorly supported clades, which were incongruent with the inferred nuclear tree and with previous taxonomic treatments (see Results). Accordingly, plastid regions were not included in this analysis because of the lack of resolution and probable partial incongruence (although this was difficult to evaluate due to the lack of resolution). SpecieS tree and diverSification timeS A species tree with absolute times was estimated from ITS, ETS, CPR1 and plastid gene trees with StarBEAST2 (Drummond et al., 2012; Bouckaert et al., 2014). We used 50 samples (1–2 samples per taxa) for which we were able to amplify and sequence ITS and ETS (plastid sequences were available for these 50 samples and for most CPR1). Clock model, tree model and site model were unlinked for ITS, ETS, CPR1 and plastid partitions. A calibrated Yule model was selected as the tree model for gene trees, which were estimated using an uncorrelated lognormal relaxed clock model. As indicated above, for subtribe Chironiinae the BEAST2 analysis inferred ages of 10.3 Mya (95% HPD: 8.3–12.3) for the crown node of the genus Centaurium and 14.9 Mya (95% HPD: 12.6–17.2) for the crown node of the clade Schenkia – Exaculum – Centaurium. These estimates were used again as secondary calibration points. Analyses were conducted using twelve independent MCMC runs of 500 million generations each. Run convergences and burn-ins were assessed in Tracer v.1.6 (Lemey Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 222 V. JIMÉNEZ-LOBATO ET AL. © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 et al., 2009). A maximum clade credibility tree was calculated with TreeAnnotator v.2.3.2 (Drummond & Rambaut, 2007) using a PP limit of 0.90, a maximum clade credibility tree and the mean heights options. The tool DensiTree (Drummond & Rambaut, 2007) was also used to represent inferred trees (we plotted consensus trees representing the HPD of 95% of the species tree searches). diverSification patternS The 29 taxa consensus tree and 1000 randomly selected post-burn-in trees obtained from the BEAST2 analysis were pruned using the drop.tip function implemented in R package ape (Paradis, Claude & Strimmer, 2004; R Development Core Team, 2015) to remove samples from Schenkia, Exaculum and Blackstonia. The pruned trees were used to study diversification patterns and trait evolution. Lineages-through-time (LTT) analyses were performed using the pruned BEAST2 consensus tree and 1000 post-burn-in trees with the functions mltt.plot and ltt.lines implemented in R package ape (Paradis et al., 2004). The gamma statistic (Pybus & Harvey, 2000) was also calculated for the 1000 post-burn-in trees as implemented in the ltt R function in phytools (Revell, 2012). The diversification pattern was studied by fitting different models of diversification (Pure- Birth, Birth-Death, DDL, DDX, yule2rate, yule3rate and yule4rate) on the consensus tree, as well as on the 100 post-burn-in trees, using the function fitdAICrc implemented in the R package laser (Rabosky, 2007). anceStral State reconStruction We reconstructed the evolution of traits as binary characters using the one-state symmetrical Markov model (Mk1) and the two-state asymmetrical Markov model (Mk2), implemented in the R package diversitree (Fitzjohn, 2012). We assumed that we had a full sample of the genus Centaurium (although the species limits are not totally clear in this genus, we believe we sampled most of its evolutionary significant units). In the case of polyploidy, we constrained the model to be unidirectional, thereby assuming an irreversible transition from polyploidy to diploidy. phylogenetic Signal The phylogenetic signal was estimated with the D measure used by Fritz & Purvis (2010), appropriate for binary traits, and implemented in the function phylo.d in R package caper (Orme, 2013). D is based on the sum of the differences between the two pairs of tips of each branch in the phylogenetic tree, scaled by the sum resulting from the differences obtained in a Brownian and a random phylogenetic pattern of evolution (Fritz & Purvis, 2010). A D statistic value of 1 (D = 1) means that the observed binary trait has a phylogenetically random distribution, while a value of 0 (D = 0) supports a clumped or Brownian pattern. The D value can fall outside this range with D < 0 in extremely clumped cases and D > 1 in case of overdispersion between character states. In addition, we tested the significance of the departure of D from 0 and 1 with P values obtained from 1000 permutations of the parameters used by D. trait correlation analySeS We used Pagel’s model (Pagel, 1994) for the correlated evolution of two binary traits as implemented in the function fitPagel of the R package phytools (Revell, 2012). We ran the analyses twice assuming both equal and different rates of transitions between character states. The analysed traits were: flower size, flower display, herkogamy and polyploidy; all of which are typically related to the selfing syndrome. RESULTS phylogenetic inference The combined nuclear matrix contained 1535 aligned characters (635 characters from ITS, 388 from ETS and 512 from CPR1). The topology recovered by MrBayes (Supporting Information, Fig. S1) showed three wellseparated clades, similar to those reported by Mansion et al. (2005): Clade A consisting of C. pulchellum, C. centaurioides (Roxb.) R.S.Rao & Hemadri, C. capense C.R.Broome, C. mairei Zeltner, C. discolor, C. maritimum (L.) Fritsch and C. malzacianum Maire; clade B consisting of C. tenuiflorum (Hoffmanns. & Link) Fritsch and C. serpentinicola Carlström; and clade C, whose internal lineage relationships are not very well resolved, consisting of three main subclades clustering as Clade C1: C. chloodes (Brot.) Samp., C. somedanum M.Laínz, C. scilloides (L.f.) Samp. and C. littorale subsp. littorale; Clade C2: C. grandiflorum subsp. boissieri (Willk.) Z.Díaz, C. quadrifolium subsp. barrelieri (L.M.Dufour) G.López, C. quadrifolium subsp. quadrifolium (L.) G.López & C.E.Jarvis and C. quadrifolium subsp. parviflorum (Willk.) Pedrol (this latter taxon is diploid and seems to be polyphyletic as one of its samples is sister to clade C2) and, finally, Clade C3: C. suffruticosum (Griseb.) Ronniger, C. grandiflorum (Pers.) Ronniger subsp. grandiflorum, C. grandiflorum subsp. majus (Hoffmanns. & Link) Z.Díaz, C. turcicum (Velen.) Ronniger, C. erythrea subsp. rumelicum (Velen.) Melderis, C. littorale subsp. uliginosum (Waldst. & Kit.) Melderis, C. erythraea subsp. erythraea and C. erythraea subsp. rhodense (Boiss. & Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 223 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 Reut.) Melderis. The relationships between species and subspecies in this latter clade were poorly resolved and not well supported. Finally, C. quadrifolium subsp. linariifolium (Lam.) G.López forms a polytomy with the remaining species in clade C. Trees recovered from Bayesian inference from nuclear data regions, ITS, ETS, CPR1 and ITS-ETS (Supporting Information, Figs S2–S5, respectively), showed congruence for the main clades and for the within-clades relationships (PP > 0.90) and also were congruent with the nuclear combined ITS-ETS-CPR1 tree (Fig. S1). The combined plastid matrix contained 3772 aligned characters. The matrices of the six plastid DNA regions each revealed a limited number of informative characters (inferred trees displayed extremely low resolution; results not shown). The plastid tree inferred from the concatenated plastid sequences (also with a small number of informative characters) was also poorly resolved at within-clades relationships (Supporting Information, Fig. S6). The few inferred clades with high-to-moderate support were mostly incongruent with the taxonomical species delimitation, the subgeneric classification and the trees inferred from nuclear DNA regions (that mainly concern clades A and B). The tree inferred from combined nuclear and plastid DNA regions (Fig. S7) displayed similar results to the tree inferred from combined nuclear regions (Supporting Information, Fig. S1). diverSification time analySeS In the analysis of Chironiinae, the Yule model tree was better supported than the one based on the Birth-Death model (Bayes factor = 3.046). The divergence time in the Yule model tree for the crown node of Centaurium (PP = 1) inferred from ITS and trnL intron-trnLF of Chironiinae using BEAST2 (Fig. 2A) was 10.25 (Mya) (95% HPD: 6.37–15.17). The crown node of clade B–C of Centaurium was dated to 8.22 Mya (95% HPD: 4.55–12.5; PP = 0.99) and that of clade A to 7.29 Mya (95% HPD: 4.19–10.94; PP = 1.00). The crown node for clade B was dated to 0.67 Mya (95% HPD: 0.09–1.47; PP = 1.00) and that of clade C to about 2.73 Mya (95% HPD: 1.41–4.23; PP = 1.00). The topology of the maximum clade credibility (MCC) tree of Centaurium analysis (Fig. 2B) inferred from ITS, ETS and CPR1 using BEAST2 was congruent with the Centaurium clade in tree of the analysis of Chironiinae and also topologically congruent with the tree inferred from ITS, ETS and CPR1 nuclear regions using MrBayes (Supporting Information, Fig. S1). The crown node of Centaurium (PP = 1) was 10.32 Mya (95% HPD: 7.46–13.18). The crown node of clade B–C (PP = 1) was 8.13 Mya (95% HPD: 5.39–10.93); for clade A (PP = 1) was 5.8 Mya (95% HPD: 3.55–8.09); for clade B (PP = 1) was 0.35 Mya (95% HPD: 0.02– 0.82; PP = 1.00) and that of clade C to about 2.71 Mya (95% HPD: 1.57–4.08; PP = 1.00) (Fig. 2B). SpecieS tree and timeS of diverSification Species tree inferred using StarBEAST2 supported the monophyly of Centaurium and resolved two main clades (A–B and C) within (all with a PP = 1). In addition within the clade only two clades were supported with a PP of 0.91–0.94 (Fig. 3A). The remaining clades in the inferred species tree had a support lower than 0.90 PP. This pattern is also show in Fig. 3B in which just clades C and A–B are clearly separated. The lack of resolution at within-clade level in the species trees was explained by the fact that the single partitions (ITS, ETS, CPR1 and plastid) were not very explanative. As result, the gene trees were poorly resolved (mostly polytomies and poorly supported clades) and the species trees resulting from the combination of the gene trees was also poorly resolved (Fig. 3A, B). The diversification dates between BEAST2 and StarBeast2, taking into account the confidence intervals, are congruent. Nevertheless, the dates inferred in StarBeast2 are in general younger that the ones inferred in BEAST2, especially in the shallower nodes in StarBeast2 that are inferred as very recent. The MCC tree from the BEAST2 analysis of Centaurium was much better resolved as ITS, ETS and CPR1 were linked for the tree model in the MCMC iterations; the resulting MCC tree was better resolved as it combined the information from all the tree DNA regions (Fig. 2B). This lack of phylogenetic resolution from the species trees from StarBEAST2 precluded its use in further analyses. We used instead the MCC tree from the BEAST2 analysis of the genus Centaurium (Fig. 2B). diverSification patternS The gamma value for the MCC tree from Centaurium analysis using BEAST2 tree was 1.98 (P = 0.047), while the mean gamma value for the 1000 postburn-in trees was 2.05 (SD = 0.54). This indicates that nodes are concentrated towards the tips, which in turn suggests a recent and accelerated net diversification rate towards the present. The best model for explaining the diversification pattern in the MCC tree was yule3rate (AIC = −32.50); the second and third best models were yule4rate (AIC = −30.90) and yule2rate (AIC = −27.69), respectively. Yule3rate predicts an initial background diversification rate of 0.19 lineages per million years (lin. MY-1), with a significant increase in the diversification rate to 1.26 lin. MY-1 by 1.32 Mya, and with significant decrease in the diversification rate to 0.22 lin. MY-1 by 0.53 Mya. This model is also the best model for explaining the diversification patterns Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 224 V. JIMÉNEZ-LOBATO ET AL. © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 in 595 of the 1000 post-burn-in trees. There was a mean initial background diversification rate of 0.17 lin. MY-1 (SD = 0.08), with a significant increase in the diversification rate up to 1.41 lin. MY-1 (SD = 1.02) by 1.94 Ma (SD = 1.10 Mya) and a significant decrease in the diversification rate to 0.31 lin. MY-1 (SD = 0.17) by 0.73 Mya (SD = 0.72). For the rest of trees, the best models for explaining diversification patterns were yule2rate (280 trees) and yule4rate (120 trees). In yule2rate, the initial increase in diversification Figure 2. Phylogenetic trees inferred using BEAST2 for subtribe Chironiinae based on A, ITS, trnL-F spacer and trnL intron, and B, for Centaurium based on ITS, ETS and CPR1. Letters A–C refer to the clades reported by Mansion et al. (2005). Italic numbers associated with nodes refer to the mean estimated ages. The bars in grey show 95% highest posterior densities (HPD) of the estimated ages. Numbers above branches are posterior probability (PP) values (only shown > 0.90). Time scale in millions of years is below the trees and the Messinian Salinity Crisis (MSC) and the onset of the Mediterranean Climate (OMC) are highlighted in blue. Each code after taxon names (CENT) is associated with an accession number in GenBank. Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 231 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 we thank Michael Lockwood for the revision of the English. The authors declare no conflict of interest. REFERENCES Ainouche ML, Fortune PM, Salmon A, Parisod C, Grandbastien MA, Fukunaga K, Ricou M, Misset MT. 2009. Hybridization, polyploidy and invasion: lessons from Spartina (Poaceae). Biological Invasions 11: 1159–1173. Alfaro ME, Zoller S, Lutzoni F. 2003. Bayes or bootstrap? A simulation study comparing the performance of Bayesian Markov chain Monte Carlo sampling and bootstrapping in assessing phylogenetic confidence. Molecular Biology and Evolution 20: 255–266. Andreasen K, Baldwin BG. 2001. Unequal evolutionary rates between annual and perennial lineages of checker mallows (Sidalcea, Malvaceae): evidence from 18S-26S rDNA internal and external transcribed spacers. Molecular Biology and Evolution 18: 936–944. Armbruster WS. 1993. Evolution of plant pollination systems: hypotheses and tests with the Neotropical vine Dalechampia. Evolution 47: 1480–1505. Aronson J, Kigel J, Shmida A, Klein J. 1992. Adaptive phenology of desert and Mediterranean populations of annual plants grown with and without water stress. Oecologia 89: 17–26. Baker HG. 1955. Self-compatibility and establishment after ‘long-distance’ dispersal. Evolution 9: 347–349. Baker HG, Stebbins GL. 1965. The genetics of colonizing species. New York: Academic Press. Barrett SCH. 2002. Evolution of sex: the evolution of plant sexual diversity. Nature Reviews Genetics 3: 274–284. Barrett SCH. 2010. Understanding plant reproductive diversity. Philosophical Transactions of the Royal Society B – Biological Sciences 365: 99–109. Barrett SCH, Harder LD, Worley AC. 1996. The comparative biology of pollination and mating in flowering plants. Philosophical Transactions of the Royal Society of London B 351: 1271–1280. Barringer BC. 2007. Polyploidy and self-fertilization in flowering plants. American Journal of Botany 94: 1527–1533. Bouckaert R, Heled J, Kühnert D, Vaughan T, Wu CH, Xie D, Suchard MA, Rambaut A, Drummond AJ. 2014. BEAST 2: a software platform for Bayesian evolutionary analysis. PLoS Computational Biology 10: 1–6. Brys R, Jacquemyn H. 2011. Variation in the functioning of autonomous self-pollination, pollinator services and floral traits in three Centaurium species. Annals of Botany 107: 917–925. Brys R, Jacquemyn H. 2012. Effects of human-mediated pollinator impoverishment on floral traits and mating patterns in a short-lived herb: an experimental approach. Functional Ecology 26: 189–197. Brys R, Van Cauwenberghe J, Jacquemyn H. 2016. The importance of autonomous selfing in preventing hybridization in three closely related plant species. Journal of Ecology 104: 601–610. Brys R, Vanden BA, Mergeay J, Jacquemyn H. 2014. The contribution of mating system variation to reproductive isolation in two closely related Centaurium species (Gentianaceae) with a generalized flower morphology. Evolution 68: 1281–1293. Carrizo GC. 2001. Androecium ontogeny in Normania triphylla (Lowe) Lowe (Solaneae Solanaceae). Anales Jardín Botánico de Madrid 59: 67–74. Charlesworth D, Willis JH. 2009. The genetics of inbreeding depression. Nature Reviews. Genetics 10: 783–796. Cohen JI. 2016. Floral evolution in Lithospermum (Boraginaceae): independent origins of similar flower types. Botanical Journal of the Linnean Society 2: 213–228. Coleman M, Liston A, Kadereit JW, Abbott RJ. 2003. Repeat intercontinental dispersal and Pleistocene speciation in disjunct Mediterranean and desert Senecio (Asteraceae). American Journal of Botany 90: 1446–1454. Cronk Q, Moller M. 1997. Genetic of floral symmetry revealed. Trends in Ecology and Evolution 12: 85–86. Cubas P. 2004. Floral zygomorphy, the recurring evolution of a successful trait. BioEssays 26: 1175–1184. Cusimano N, Renner SS. 2010. Slowdowns in diversification rates from real phylogenies may not be real. Systematic Biology 59: 458–464. Darriba D, Taboada GL, Doallo R, Posada D. 2012. jModelTest 2: more models, new heuristics and parallel computing. Nature Methods 9: 772. Datson PM, Murray BG, Steiner KE. 2008. Climate and the evolution of annual/perennial life-histories in Nemesia (Scrophulariaceae). Plant Systematics and Evolution 270: 39–57. Díaz Lifante Z. 2012. Centaurium. In: C. Romero Z, Qintanar A, eds. Flora Iberica 11. Madrid: Real Jardín Botánico, CSIC, 49–81. Donoghue MJ, Ree RH, Baum DA. 1998. Phylogeny and the evolution of flower symmetry in the Asteridae. Trends in Plant Science 3: 311–317. Drummond AJ, Rambaut A. 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology 7: 214. Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29: 1969–1973. Duggen S, Hoernle K, van den Bogaard P, Rupke L, Phipps MJ. 2003. Deep roots of the Messinian salinity crisis. Nature 422: 602–606. Eaton DAR, Ree RH. 2013. Inferring phylogeny and introgression using genomic RADseq data: an example from flowering plants (Pedicularis: Orobanchaceae). Systematic Biology, 62: 689–706. Eckert CG, Kalisz S, Geber MA, Sargent R, Elle E, Cheptou PO, Goodwillie C, Johnston MO, Kelly JK, Moeller DA, Porcher E, Ree RH, Vallejo-Marín M, Winn AA. 2010. Plant mating systems in a changing world. Trends in Ecology & Evolution 25: 35–43. Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 232 V. JIMÉNEZ-LOBATO ET AL. © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 Eckert CG, Samis K, Dart S. 2006. Reproductive assurance and the evolution of uniparental reproduction in flowering plants. In: Harder LD, Barrett SCH, eds. The ecology and evolution of flowers. Oxford: Oxford University Press, 183–203. Elle E, Hare JD. 2002. Environmentally induced variation in floral traits affects the mating system in Datura wrightii. Functional Ecology 16: 79–88. Escudero M, Balao F, Martín-Bravo S, Valente L, Valcárcel V. 2018. Is the diversification of Mediterranean Basin plant lineages coupled with karyotypic changes? Plant Biology 20(Suppl. 1):166–175. Etienne RS, Rosindell J. 2012. Prolonging the past counteracts the pull of the present: protracted speciation can explain observed slowdowns in diversification. Systematic Biology 61: 204–213. Fazekas AJ, Burgess KS, Kesanakurti PR, Graham SW, Newmaster SG, Husband BC, Percy DM, Hajibabaei M, Barrett SC. 2008. Multiple multilocus DNA barcodes from the plastid genome discriminate plant species equally well. PLoS One 3: e2802. Fenster CB, Armbruster WS, Wilson P, Dudash MR, Thomson JD. 2004. Pollination syndromes and floral specialization. Annual Review of Ecology, Evolution, and Systematics 35: 375–403. Fisher R. 1941. Average excess and average effect of a gene substitution. Annuals of Eugenics 11: 53–63. Fitzjohn RG. 2012. Diversitree: comparative phylogenetic analyses of diversification in R. Methods in Ecology and Evolution 3: 1084–1092. Fiz-Palacios O, Valcárcel V. 2013. From Messinian crisis to Mediterranean climate: a temporal gap of diversification recovered from multiple plant phylogenies. Perspectives in Plant Ecology, Evolution and Systematics 15: 130–137. Foxe JP, Slotte T, Stahl EA, Neuffer B, Hurka H, Wright AI. 2009. Recent speciation associated with the evolution of selfing in Capsella. Proceedings of the National Academy of Sciences of the United States of America 106: 5241–5245. Fritz SA, Purvis A. 2010. Selectivity in mammalian extinction risk and threat types: a new measure of phylogenetic signal strength in binary traits. Conservation Biology 24: 1042–1051. Gene Codes Corporation. Sequencher version 4.1.4 DNA Sequence Analysis software, Gene Code Corporation, Ann Arbor, MI, USA. http://genecodes.com. Glemin S, Bazin E, Charlesworth D. 2006. Impact of mating systems on patterns of sequence polymorphism in flowering plants. Proceedings of the Royal Society B 273: 3011–3019. Glick L, Sabath N, Ashman T-L, Goldberg E, Mayrose I. 2016. Polyploidy and sexual system in angiosperms: is there an association? American Journal of Botany 103: 1–13. Goldberg EE, Kohn JR, Lande R, Robertson KA, Smith SA, Igic B. 2010. Species selection maintains selfincompatibility. Science 330: 493–495. Goodwillie C. 1999. Multiple origins of self-compatibility in Linanthus section Leptosiphon (Polemoniaceae): Phylogenetic evidence from internal-transcribed-spacer sequence data. Evolution 53: 1387–1395. Goodwillie C, Kalisz S, Eckert CG. 2005. The evolutionary enigma of mixed mating systems in plants: occurrence, theoretical explanations, and empirical evidence. Annual Review of Ecology, Evolution, and Systematics 36: 47–79. Goodwillie C, Sargent RD, Eckert CG, Elle E, Geber MA, Johnston MO, Kalisz S, Moeller DA, Ree RH, Vallejo- Marin M, Winn AA. 2010. Correlated evolution of mating system and floral display traits in flowering plants and its implications for the distribution of mating system variation. The New Phytologist 185: 311–321. Guggisberg A, Bretagnolle F, Mansion G. 2006. Allopolyploid origin of the Mediterranean endemic, Centaurium bianoris (Gentianaceae), inferred by molecular markers. Systematic Biology 31: 368–379. Guindon S, Gascuel O. 2003. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology 52: 696–704. Guzmán B, Vargas P. 2009. Historical biogeography and character evolution of Cistaceae (Malvales) based on analysis of plastid rbcL and trnL-trnF sequences. Organisms Diversity and Evolution 9: 83–99. Heenan PB, Mitchell AD, de Lange PJ, Keeling J, Paterson AM. 2010. Late-Cenozoic origin and diversification of Chatham Islands endemic plant species revealed by analyses of DNA sequence data. New Zealand Journal of Botany 48: 83–136. Herlihy CR, Eckert CG. 2007. Evolutionary analysis of a key floral trait in Aquilegia canadensis (Ranunculaceae): genetic variation in herkogamy and its effect on the mating system. Evolution 61: 1661–1674. Hijmans RJ, Gavrilenko T, Stephenson S, Bamberg J, Salas A, Spooner DM. 2007. Geographical and environmental range expansion through polyploidy in wild potato (Solanum section Petota). Global Ecology and Biogeography 16: 485–495. Ho SY, Phillips MJ. 2009. Accounting for calibration uncertainty in phylogenetic estimation of evolutionary divergence times. Systematic Biology 58: 367–380. Howe HF. 2016. Making dispersal syndromes and networks useful in tropical conservation and restoration. Global Ecology and Conservation 16: 152–178. Huber KT, Moulton V. 2006. Phylogenetic networks from multi-labelled trees. Journal of Mathematical Biology 52: 613–632. Huelsenbeck JP, Ronquist F, Nielsen R, Bollback JP. 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294: 2310–2314. Igic B, Bohs L, Kohn JR. 2004. Historical inferences from the self-incompatibility locus. New Phytologist 161: 97–105. Igic B, Busch JW. 2013. Is self-fertilization an evolutionary dead end? The New Phytologist 198: 386–397. Jesson LK, Barrett SCH. 2002. The genetic of mirror-image flowers. Proceedings of the Royal Society B 269: 1835–1839. Johnson SD, Steiner KE. 2000. Generalization versus specialization in plant pollination systems. Trends in Ecology & Evolution 15: 140–143. Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 233 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 Kalisz S, Vogler DW, Hanley KM. 2004. Context-dependent autonomous self-fertilization yields reproductive assurance and mixed mating. Nature 430: 884–887. Kim SC, McGowen MR, Lubinsky P, Barber JC, Mort ME, Santos-Guerra A. 2008. Timing and tempo of early and successive adaptive radiations in Macaronesia. PLoS One 3: e2139. Kress JW, Erickson LD. 2007. Coding rbcL gene complements the non-coding trnH-psbA spacer region. PLoS ONE 6: e508. Lemey P, Rambaut A, Drummond AJ, Suchard MA. 2009. Bayesian phylogeography finds its roots. PLoS Computational Biology 5: 1–16. Levin, DA. 1975. Minority cytotype exclusion in local plant populations. Taxon 24: 35–43. Liu L, Yu L, Kubatko L, Pearl DK, Edwards SV. 2009. Coalescent methods for estimating phylogenetic trees. Molecular Phylogenetics and Evolution 53: 320–328. Mable BK. 2004. Polyploidy and self-compatibility: is there an association? New Phytologist 162: 803–811. Magri D, Palombo MR. 2013. Early to middle Pleistocene dynamics of plant and mammal communities in south west Europe. Quaternary International 288: 63–72. Mansion G, Struwe L. 2004. Generic delimitation and phylogenetic relationships within the subtribe Chironiinae (Chironieae: Gentianaceae), with special reference to Centaurium: evidence from nrDNA and cpDNA sequences. Molecular Phylogenetics and Evolution 32: 951–977. Mansion G, Zeltner L, Bretagnolle F. 2005. Phylogenetic patterns and polyploid evolution within the Mediterranean genus Centaurium (Gentianaceae - Chironieae). Taxon 54: 931–950. Mathews KG, Ruigrok MS, Mansion G. 2015. Phylogeny and biogeography of the eastern North American rose gentians (Sabatia, Gentianaceae). Systematic Botany 40: 811–825. Mayrose I, Zhan SH, Rothfels CJ, Arrigo N, Barker MS, Rieseberg LH, Otto SP. 2015. Methods for studying polyploid diversification and the dead end hypothesis: a reply to Soltis et al. (2014). The New Phytologist 206: 27–35. Mayrose I, Zhan SH, Rothfels CJ, Magnuson-Ford K, Barker MS, Rieseberg LH, Otto SP. 2011. Recently formed polyploid plants diversify at lower rates. Science 333: 1257. Melderis A. 1972a. Taxonomic studies on the European species of the genus Centaurium Hill. Botanical Journal of the Linnean Society 65: 224–250. Melderis A. 1972b. Centaurium. In: Tutin TG, Heywood VH, Burges NA, Valentine DH, Walters SM, Webb DA, eds. Flora Europaea. Cambridge: Cambridge University Press, 56–59. Meyers LA, Levin DA. 2006. On the abundance of polyploids in flowering plants. Evolution 60: 1198–1206. Moeller D, Geber M. 2005. Ecological context of the evolution of self-pollination in Clarkia xantiana: population size, plant communities, and reproductive assurance. International Journal of Organic Evolution 59: 786–799. Moen D, Morlon H. 2014. Why does diversification slow down? Trends in Ecology & Evolution 29: 190–197. Morgan MT, Wilson WG. 2005. Self-fertilization and scape from pollen limitation in variable pollination environments. Evolution 59: 1143–1148. Oberprieler C. 2005. Temporal and spatial diversification of circum-Mediterranean Compositae-Anthemideae. Taxon 54: 951–966. Olmstead R. 1989. Phylogeny, phenotypic evolution, and biogeography of the Scutellaria angustifolia complex (Lamiaceae): inference from morphological and molecular data. Systematic Botany 14: 320–338. Orme D. 2013. The caper package: comparative analysis of phylogenetics and evolution in R. Pagel M. 1994. Detecting correlated evolution on phylogenies: a general method for the comparative analysis of discrete characters. Proceedings:Biological Sciences 255: 37–45. Pannell JR. 2015. Evolution of the mating system in colonizing plants. Molecular Ecology 24: 2018–2037. Paradis E, Claude J, Strimmer K. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20: 289–290. Parisod C, Besnard G. 2007. Glacial in situ survival in the western Alps and polytopic autopolyploidy in Biscutella laevigata L. (Brassicaceae). Molecular Ecology 16: 2755–2767. Parisod C, Holderegger R, Brochmann C. 2010. Evolutionary consequences of autopolyploidy. The New Phytologist 186: 5–17. Petanidou T, Kallimanis AS, Tzanopoulos J, Sgardelis SP, Pantis JD. 2008. Long-term observation of a pollination network: fluctuation in species and interactions, relative invariance of network structure and implications for estimates of specialization. Ecology Letters 11: 564–575. Pringle J. 2014. Morphological characteristics of the family Gentianaceae. In: Rybczyński JJ, Davey MR, Mikuła A, eds. The Gentianaceae - Volume 1: characterization and ecology. Berlin, Heidelberg: Springer-Verlag, 1–329. Pybus OG, Harvey PH. 2000. Testing macro-evolutionary models using incomplete molecular phylogenies. Proceedings of the Royal Society B 267: 2267–2272. Rabosky DL. 2007. LASER: a maximum likelihood toolkit for detecting temporal shifts in diversification rates from molecular phylogenies. Evolutionary Bioinformatics Online 2: 273–276. Rambaut A, Suchard MA, Xie D, Drummond AJ. 2014. Tracer v.1.6. Ramsey J, Schemske DW. 1998. Pathways, mechanisms, and rates of polyploid formation in flowering plants. Annual Review of Ecology and Systematics 29: 467–501. Ramsey J, Schemske DW. 2002. Neopolyploidy in flowering plants. Annual Review of Ecology and Systematics. 33: 589–639. Randle AM, Slyder JB, Kalisz S. 2009. Can differences in autonomous selfing ability explain differences in range size among sister-taxa pairs of Collinsia (Plantaginaceae)? An extension of Baker’s Law. The New Phytologist 183: 618–629. R Development Core Team. 2015. R: a language and environment for statistical computing. Vienna: R Foundation for statistical computing. ISBN 3-900051-07-0, URL http:// www.R-project.org. Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 234 V. JIMÉNEZ-LOBATO ET AL. © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 Revell LJ. 2012. phytools: an R package for phylogenetic comparative biology (and other things). Methods in Ecology and Evolution 3: 217–223. Ritland C, Ritland K. 1989. Variation of sex allocation among 8 taxa of the Mimulus guttatus species complex (Scrophulariaceae). American Journal of Botany 76: 1731–1739. Robert CE. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research 32: 1792–1797. Ronquist F, Huelsenbeck JP. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, Larget B, Liu L, Suchard MA, Huelsenbeck JP. 2012. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Systematic Biology 61: 539–542. Rudall PJ, Bateman RM. 2004. Evolution of zygomorphy in monocot flowers: iterative patterns and developmental constraints. New Phytologist 162: 25–44. Sargent RD. 2004. Floral symmetry affects speciation rates in angiosperms. Proceedings of the Royal Society B 271: 603–8. Sas C, Müller F, Kappel C, Kent TV, Wright SI, Hilker M, Lenhard M. 2016. Repeated inactivation of the first committed enzyme underlies the loss of benzaldehyde emission after the selfing transition in Capsella. Current Biology 26: 3313–3319. Schaefer H, Hardy OJ, Silva L, Barraclough TG, Savolainen V. 2011. Testing Darwin’s naturalization hypothesis in the Azores. Ecology Letters 14: 389–396. Schwarz H, Liu B, Peters S, Barillas W, Beerhues L. 2009. Purification, cDNA cloning and functional expression of NADPH-cytochrome P450 reductase from Centaurium erythraea cell cultures. Plant Biology 11: 300–6. Sicard A, Kappel C, Lee YW, Woźniak NJ, Marona C, Stinchcombe JR, Wright SI, Lenhard M. 2016. Standing genetic variation in a tissue-specific enhancer underlies selfing-syndrome evolution in Capsella. Proceedings of the National Academy of Sciences 113: 13911–13916. Sicard A, Lenhard M. 2011. The selfing syndrome: a model for studying the genetic and evolutionary basis of morphological adaptation in plants. Annals of Botany 107: 1433–1443. Soltis DE, Segovia-Salcedo MC, Jordon-Thaden I, Majure L, Miles NM, Mavrodiev EV, Mei W, Cortez MB, Soltis PS, Gitzendanner MA. 2014. Are polyploids really evolutionary dead-ends (again)? A critical reappraisal of Mayrose et al. (2011). New Phytologist 202: 1105–1117. Stebbins GL. 1950. Variation and evolution in plants. New York: Columbia University Press. Stebbins GL. 1957. Self fertilization and population variability in the higher plants. The American Naturalist 91: 337–354. Stebbins L. 1985. Polyploidy, hybridization, and the invasion of new habitats. Annals of the Missouri Botanical Garden 72: 824–832. Struwe L, Kadereit JW, Klackemberg J, Nilsson S, Thiv M, von Hagen KB, Albert VA. 2002. Systematics, character evolution, and biogeography of Gentianaceae, including a new tribal and subtribal classificaton. In: Struwe L, Albert V, eds. Gentianaceae: systematic and natural history. Cambridge: Cambridge University Press, 21–309. Suc J. 1984. Origin and evolution of the Mediterranean vegetation and climate in Europe. Nature 307: 429–432. Takebayashi N, Morrell PL. 2001. Is self-fertilization an evolutionary dead end? Revisiting an old hypothesis with genetic theories and a macroevolutionary approach. American Journal of Botany 88: 1143–1150. Takebayashi N, Wolf DE, Delph LF. 2006. Effect of variation in herkogamy on outcrossing within a population of Gilia achilleifolia. Heredity 96: 159–165. Tank DC, Olmstead RG. 2008. From annuals to perennials: phylogeny of subtribe Castillejinae (Orobanchaceae). American Journal of Botany 95: 608–625. Valente LM, Savolainen V, Vargas P. 2010. Unparalleled rates of species diversification in Europe. Proceedings of the Royal Society of London B 277: 1489–1496. Vallejo-Marín M, Barrett SC. 2009. Modification of flower architecture during early stages in the evolution of selffertilization. Annals of Botany 103: 951–962. Vamosi JC, Vamosi SM. 2010. Key innovations within a geographical context in flowering plants: towards resolving Darwin’s abominable mystery. Ecology Letters 13: 1270–1279. Vilatersana R, Susanna A, Garcia-Jacas N, Garnatje T. 2000. Karyology, generic delineation and dysploidy in the genera Carduncellus, Carthamus and Phonus (Asteraceae). Botanical Journal of the Linnean Society 134: 425–438. de Vos JM, Wüest RO, Conti E. 2013. Small and ugly? Phylogenetic analyses of the ‘selfing syndrome’ reveal complex evolutionary fates of monomorphic primrose flowers. Evolution 68: 1042–1057. Webb CJ, Lloyd DG. 1986. The avoidance of interference between the presentation of pollen and stigmas in angiosperms II. Herkogamy. New Zealand Journal of Botany 24: 163–178. Wright SI, Ness RW, Foxe JP, Barrett SCH. 2008. Genomic consequences of outcrossing and selfing in plants. International Journal of Plant Sciences 169: 105–118. Zeltner L. 1970. Recherches de biosystématique sur les genres Blackstonia Huds. et Centaurium Hill (Gentianaceae). Bulletin de la Société Neuchâteloise des Sciences Naturelles 93: 1–164. Živković S, Dević M, Filipović B, Giba Z, Grubišić D. 2007. Effect of NaCl on seed germination in some Centaurium Hill. species (Gentianaceae). Archives of Biological Science Belgrade 59: 227–231. Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019 THE EVOLUTION OF SELFING SYNDROME IN CENTAURIUM 235 © 2019 The Linnean Society of London, Botanical Journal of the Linnean Society, 2019, 191, 216–235 SUPPORTING INFORMATION Figure S1. Bayesian consensus tree of Centaurium obtained from the concatenated nuclear DNA matrix using MrBayes. Letters A–C refer to the clades and subclades reported by Mansion et al. (2005) and referred to in the text. Each code after taxon names (CENT) is associated with an accession number in GenBank. Figure S2. Bayesian consensus tree of Centaurium obtained from the ITS matrix using MrBayes. Letters A–C refer approximately to the clades and subclades reported by Mansion et al. (2005) and referred to in the text. Each code after taxon names (CENT) is associated with an accession number in GenBank. Figure S3. Bayesian consensus tree of Centaurium obtained from the ETS matrix using MrBayes. Letters A–C refer approximately to the clades reported by Mansion et al. (2005) and referred to in the text. Subclades C1–C3 are not clearly delimited and only clade C in shown. Each code after taxon names (CENT) is associated with an accession number in GenBank. Figure S4. Bayesian consensus tree of Centaurium obtained from CPR1 matrix using MrBayes. Letters A–C refer approximately to the clades reported by Mansion et al. (2005) and referred to in the text. Subclades C1–C3 are not clearly delimited and only clade C in shown. Each code after taxon names (CENT) is associated with an accession number in GenBank. Figure S5. Bayesian consensus tree of Centaurium obtained from the concatenated ITS-ETS DNA matrix using MrBayes. Letters A–C refer approximately to the clades and subclades reported by Mansion et al. (2005) and referred to in the text, with the exception of C. mairei (CENT58) associated to clade B and C. quadrifolium parviflorum (CENT60) associated to subclade C1. Each code after taxon names (CENT) is associated with an accession number in GenBank. Figure S6. Bayesian consensus tree of Centaurium obtained from the concatenated plastid DNA matrix using MrBayes. Letters A–C refer approximately to the clades reported by Mansion et al. (2005) and referred to in the text. Subclades C1–C3 are not clearly delimited and only clade C in shown. Each code after taxa names (CENT) is associated with an accession number in GenBank. Figure S7. Bayesian consensus tree of Centaurium obtained from the concatenated nuclear and plastid DNA matrix using MrBayes. Letters A–C refer approximately to the clades reported by Mansion et al. (2005) and referred to in the text. Subclades C1–C3 are not clearly delimited and only clade C in shown. Each code after taxon names (CENT) is associated with an accession number in GenBank. Figure S8. Reconstruction of the ancestral states for Centaurium species obtained from Mk1 model, displayed on Nuclear DNAtree. A, Androecium symmetry (An); B, Style position (St) and C, Stigma length (Sl). Asterix (*) refers to significant supported nodes (PP > 0.94). Downloaded from https://academic.oup.com/botlinnean/article-abstract/191/2/216/5545651 by University of New England user on 25 September 2019