Polyploidy and hybridization in the Mediterranean: unravelling the evolutionary history of Centaurium (Gentianaceae).

Annals of Botany 134: 247–262, 2024 https://doi.org/10.1093/aob/mcae066, available online at www.academic.oup.com/aob ORIGINAL ARTICLE Polyploidy and hybridization in the Mediterranean: unravelling theevolutionary history of Centaurium (Gentianaceae) Ana Valdés-Florido1,*, Claudia González-Toral2, Enrique Maguilla3,, Eduardo Cires2,4, Zoila Díaz-Lifante1, Cristina Andrés-Camacho1, Gonzalo Nieto Feliner5, Juan Arroyo1 and Marcial Escudero1, 1Department of Plant Biology and Ecology, Faculty of Biology, University of Seville, Seville, 41012, Spain, 2Department of Organisms and Systems Biology, University of Oviedo, Oviedo, 33071, Spain, 3Department of Molecular Biology and Biochemical Engineering, Pablo de Olavide University, Seville, 41013, Spain, 4Institute of Natural Resources and Territorial Planning (INDUROT), Campus de Mieres, Mieres, 33600, Spain, and 5Real Jardín Botánico (RJB), CSIC, Plaza de Murillo 2, Madrid, 28014, Spain *For correspondence. E-mail [email protected] Received: 16 January 2024 Returned for revision: 11 April 2024 Editorial decision: 23 April 2024 Accepted: 29 April 2024 • Background and Aims Polyploidy is considered one of the main mechanisms of plant evolution and speciation. In the Mediterranean Basin, polyploidy has contributed to making this region a biodiversity hotspot, along with its geological and climatic history and other ecological and biogeographical factors. The Mediterranean genus Centaurium (Gentianaceae) comprises ~25 species, of which 60 % are polyploids, including tetraploids and hexaploids. To date, the evolutionary history of centauries has been studied using Sanger sequencing phylogenies, which have been insufficient to fully understand the phylogenetic relationships in this lineage. The goal of this study is to gain a better understanding of the evolutionary history of Centaurium by exploring the mechanisms that have driven its diversification, specifically hybridization and polyploidy. We aim to identify the parentage of hybrid species, at the species or clade level, as well as assessing whether morphological traits are associated with particular ploidy levels. • Methods We sequenced RADseq markers from 42 samples of 28 Centaurium taxa, and performed phylogenomic analyses using maximum likelihood, summary coalescent SVDquartets and Neighbor-Net approaches. To identify hybrid taxa, we used PhyloNetworks and the fastSTRUCTURE algorithm. To infer the putative parental species of the allopolyploids, we employed genomic analyses (SNIPloid). The association between different traits and particular ploidy levels was explored with non-metric multidimensional scaling. • Key Results Our phylogenetic analyses confirmed the long-suspected occurrence of recurrent hybridization. The allopolyploid origin of the tetraploid C. serpentinicola and the hexaploids C. mairei, C. malzacianum and C. centaurioides was also confirmed, unlike that of C. discolor. We inferred additional signatures of hybridization events within the genus and identified morphological traits differentially distributed in different ploidy levels. • Conclusions This study highlights the important role that hybridization has played in the evolution of a Mediterranean genus such as Centaurium, leading to a polyploid complex, which facilitated its diversification and may exemplify that of other Mediterranean groups. Key words: Allopolyploidy, centauries, hybridization, Mediterranean, plant evolution, polyploidy, RADseq. INTRODUCTION Polyploidy, originally defined as the possession of three or more chromosome sets in each cell (Grant, 1981), is considered a key driver of plant evolution and speciation (Stebbins, 1947, 1950; Grant, 1981; Ramsey and Schemske, 1998; Soltis et al., 2009). Indeed, almost 35 % of angiosperms have been reported to be recent polyploids (Stebbins, 1971; Grant, 1981; Wood etal., 2009). Meanwhile, genomic data have revealed that most flowering plants include in their evolutionary history rounds of polyploidization followed by post-polyploid diploidization (Wendel, 2015; Escudero and Wendel, 2020). Polyploids have been generally classified into two categories: autopolyploids, which form from unreduced gametes of the same species, and allopolyploids, which result from hybridization between different species (Stebbins, 1947; Grant, 1981; Tate et al., 2005). It has been suggested that polyploidy and hybridization have been critical evolutionary processes shaping the evolution and diversification of the Mediterranean flora (Thompson, 2005; Marques et al., 2018; Nieto Feliner et al., 2023). The geological and climatic history of the Mediterranean area has allowed these evolutionary mechanisms to play an important role, contributing to the establishment of this area as a biodiversity hotspot (Marques et al., 2018). Processes such as the Messinian salinity crisis (Duggen et al., 2003), the onset of the Mediterranean climate (Suc, 1984) and the climatic changes © The Author(s) 2024. Published by Oxford University Press on behalf of the Annals of Botany Company. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/ by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 248 during the Pleistocene (Hewitt, 2000) provided a suitable arena for these evolutionary mechanisms (Nieto Feliner et al., 2023). On one hand, hybridization events could have resulted from the contact between previously isolated lineages when climate regime oscillation processes combined with successive changes in land connections during different geological events led to changes in species’ distribution ranges (Hewitt, 2000; Thompson, 2005; Nieto Feliner, 2014). On the other hand, polyploidization and hybridization have been associated with the emergence of new traits that could facilitate the colonization of new areas or confer different abilities to cope with climate change (Levin, 2002; Marques et al., 2016; Vallejo-Marín et al., 2016). Finally, an increase in the production of unreduced gametes has been associated with environmental stress from the climatic changes recorded in the Mediterranean (Ramsey and Schemske, 1998; Brownfield and Köhler, 2011; Mason and Pires, 2015). Mediterranean genera such as Narcissus (Santos- Gally et al., 2012), Phlomis (Albaladejo and Aparicio, 2007) and Centaurium (Mansion et al., 2005; Jiménez-Lobato et al., 2019; Maguilla et al., 2021) have been proposed to have experienced these processes. The genus Centaurium (Gentianaceae), commonly known as centauries, comprises ~25 species (Mansion and Struwe, 2004; Díaz-Lifante, 2012; Plants of the World Online, 2023). This genus is distributed in temperate and arid climate regions of Asia, Europe, north-central Africa and North America (Supplementary Data Fig. S1), and its centre of diversity is the Mediterranean Basin (Maguilla et al., 2021). Zeltner (1970) found two basic chromosome numbers, x = 9 and x = 10, and three ploidy levels: diploid (2n = 2x = 18, 20), tetraploid (2n = 4x = 36, 40) and hexaploid (2n = 6x = 54, 56, 60) (Table 1). There are a few species that have more than one ploidy ploidy level. For instance, there are diploid and tetraploid individuals in C. portense Butcher and C. serpentinicola Carlström, and tetraploid and hexaploid individuals in C. scilloides (L. fil.) Samp. and C. turcicum (Velen.) Ronniger. Two species include three ploidy levels (2x, 4x, 6x): C. tenuiflorum (Hoffmanns. & Link) Fritsch and C. pulchellum (Sw.) Druce, although there is only one known diploid population of C. pulchellum, located in Israel (Zeltner, 1985). Around 60 % of the taxa in Centaurium are polyploids (Zeltner, 1970; Mansion et al., 2005), which suggests that polyploidy has been a significant force in the evolutionary history of this genus. Interestingly, ploidy levels seem to follow a geographical pattern. Diploids (2x) are mainly distributed in the Mediterranean Basin, tetraploids (4x) in Northern Europe and Eastern Asia, and hexaploids (6x) in the south-western Mediterranean Basin and on the Arabian Peninsula (Mansion et al., 2005; Prieto et al., 2012; Maguilla et al., 2021). At a finer scale within the Mediterranean Basin, ploidy levels are also not randomly distributed, with tetraploids dominating at upper latitudes, hexaploids more common at lower latitudes, and diploids inhabiting the core of the Mediterranean Basin (Mansion et al., 2005). Thus, the biogeographical study of Centaurium supports the hypothesis that polyploidy played an important role in the spread of the genus, suggesting that ancestral diploid species remained in the likely area of origin (i.e. the Mediterranean Basin), whereas polyploids expanded into new areas (Maguilla et al., 2021). Both allopolyploidy and autopolyploidy have been reported among Centaurium species using karyological and phylogenetic evidence, and it has been tentatively suggested that the hexaploids are allopolyploids while tetraploids are autopolyploids (Zeltner, 1970; Mansion et al., 2005). However, inferring the auto- or allopolyploid origin of a polyploid species is challenging, because of the intermediate levels of differentiation in genomes from conspecific populations and the dynamism of merged genomes following whole-genome duplication. In this genus, only one allopolyploid origin is documented with enough certainty: Guggisberg et al. (2006) reported C. discolor (Gand.) Ronniger as an allotetraploid derived from C. maritimum (L.) Fritsch and C. tenuiflorum using molecular analyses (RAPD analyses) and flow cytometry. Several phylogenetic studies have reconstructed the evolutionary history of the genus using Sanger sequencing data from several nuclear and plastid regions. Mansion and Struwe (2004) and Mansion et al. (2005) used the nuclear internal transcribed spacer (ITS) and the plastid regions from the trnL intron and trnL-F spacer. More recently, Jiménez-Lobato et al. (2019) used two other nuclear DNA regions [external transcribed spacer (ETS) and NADPH-cytochrome P450 reductase, CPR1] in addition to the ITS, as well as several plastid regions: trnL, matK, rpoB, rpoC, trnL-F, psbA-trnH and atpF- atpH. Both nuclear and plastid reconstructions supported the monophyly of Centaurium. The topologies obtained in the phylogenetic and biogeographic studies by Jiménez-Lobato et al. (2019) and Maguilla et al. (2021) are consistent with those obtained by Mansion et al. (2005) in reconstructing two main clades: the ‘western’ clade, which includes most of the western Mediterranean species, and the ‘widespread’ clade, composed of the more widely distributed species. To study polyploidy in the genus, Mansion et al. (2005) performed phylogenetic analyses on datasets containing only diploid species and on datasets containing both diploid and polyploid species. With the diploid dataset there was no incongruence between nuclear and plastid DNA trees. However, the incongruence length difference (ILD) test found a lack of congruence between plastid and nuclear trees when diploid and polyploid species were analysed together. The authors concluded that intensive reticulation plus polyploidy was the most reasonable explanation for their results, supporting the idea that polyploidy and hybridization play an essential role in the evolutionary history of the genus (Mansion et al., 2005). Although these studies have shed light on the relationships among species, the phylogenetic relationships at shallow nodes remained poorly resolved (Mansion and Struwe, 2004; Mansion et al., 2005; Prieto et al., 2012; Jiménez-Lobato et al., 2019; Maguilla et al., 2021). This highlights the need for genomic approaches to better understand the evolutionary history of Centaurium, especially if complex mechanisms such as polyploidy and hybridization have contributed to shaping its current diversity. Restriction site-associated DNA sequencing (RADseq) allows the sequencing of millions of single-nucleotide polymorphisms (SNPs) using restriction enzymes. The large number of SNPs have proven useful for reconstructing the evolutionary history and polyploid origin of species in phylogenetic frameworks, including non-model organisms (Dufresne et al., 2014). However, conflicts among gene trees Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 249 Table 1. Studied Centaurium taxa, locality, collector, voucher, collection date, internal code, ploidy level, source of the sample (i.e. herbarium material or silica-gel dried) and code used for PhyloNetworks analysis. Species Locality Collector Voucher Collection date Code Ploidy level Source Code for PhyloNetworks analyses C. capense Broome Mexico: Baja California, Bocana Rio L. & N. Zeltner NEU-2605 06/06/1998 CENT-53 4xHerbarium material Cap C. centaurioides R.S. Rao & Hemadri India: Maharashtra, Ramtek L. & N. Zeltner NEU-1674 25/03/1991 CENT-51 6xHerbarium material Cent C. chloodes (Brot.) Samp. Spain: Coruña, Muxia Louzán & Casais SANT-45049 23/05/2001 CENT-26 4xHerbarium material Port C. discolor (Gand.) Ronniger Spain: Mallorca, Algaida C. Navarro & al. MA-618284 02/06/1998 CENT-25 4xHerbarium material Dis C. erythraea Rafn subsp. erythraea France: Cevennes, Vebrori J. Arroyo SEV-270216 06/06/2015 CENT-64 4xHerbarium material Ery2 C. erythraea Rafn subsp. erythraea Spain: Los Molinos, Sierra de Guara J. Viruel SEV-241216 23/06/2009 CENT-3 4xHerbarium material Ery2 C. erythraea subsp. rhodense (Boiss. & Reut.) Melderis Italy: Sicilia, San Fratello L. & N. Zeltner NEU-3517 19/06/2002 CENT-77 4x Herbarium material Ery2 C. erythraea subsp. rhodense (Boiss. & Reut.) Melderis Sicily: Portella Z. Díaz – 04/06/2017 CAV-2.2 4xSilica-gel dried Ery2 C. erythraea subsp. rumelicum (Velen.) Melderis Bulgary: Fargovo J. Aldasoro VAL-163793 06/07/2004 CENT-36 2xHerbarium material Maj C. erythraea var. subcapitatum (Corb.) Ubsdell Ireland: Galway, Carraroe, Connemara L. & N. Zeltner NEU-2869 04/08/1996 CENT-54 4xHerbarium material Ery2 C. grandiflorum subsp. boissieri (Willk.) Z. Díaz Spain: Málaga, Ardales C. Andrés & Z. Diaz SEV-299926 23/06/2008 CENT-4 2xHerbarium material Boi C. grandiflorum subsp. boissieri (Willk.) Z. Díaz Spain: Sevilla, Universidad Pablo de Olavide Z. Díaz - 16/04/2015 CAV-5 2xSilica-gel dried Boi C. grandiflorum (Pers.) Ronniger subsp. grandiflorum Spain: Castellón, Morella Z. Díaz SEV-241188 02/07/2009 CENT-6 2x Herbarium material Grand C. grandiflorum (Pers.) Ronniger subsp. grandiflorum France: Corbières, Les Sauzils J. Arroyo SEV-270215 02/07/2015 CENT-68 2xHerbarium material Grand C. grandiflorum subsp. majus (Hoffmanns. & Link) Z. Díaz Italy: Cerdeña, Nuoro M.A. García & al. VAL-173304 06/06/2003 CENT-31 2xHerbarium material Maj C. grandiflorum subsp. majus (Hoffmanns. & Link) Z. Díaz Spain: Huelva, Aroche F. Pina SEV-224140 15/05/2007 CENT-9 2xHerbarium material Maj C. littorale (Turner) Gilmour subsp. littorale Denmark: Jutland, Faerfer Vig, Fur, Ile L. & N. Zeltner NEU-2980 14/07/2001 CENT-55 4xHerbarium material Lit2 C. littorale (Turner) Gilmour subsp. littorale Denmark: Jutland, Skallingen D27 L. & N. Zeltner NEU-2988 16/06/2001 CENT-76 4xHerbarium material Lit2 C. littorale subsp. uliginosum (Waldst. & Kit.) Melderis Denmark: Jutland, Skallingen D27 L. & N. Zeltner NEU-2989 16/07/2001 CENT-56 4xHerbarium material Lit2 C. littorale subsp. uliginosum (Waldst. & Kit.) Melderis Sweden: Gotland, Herrvik prope Katthammarsvik L. & N. Zeltner NEU-1699 06/08/1990 CENT-71 4xHerbarium material Ulig1 C. mairei Zeltner Niger: Aïr, Monts des Bagzanes L. & N. Zeltner NEU-1740 28/03/1986 CENT-70 6xHerbarium material Cap C. malzacianum Maire Oman: Al Akhdar, At tayyib, Wadi Beni Ghafir G. Mansion & N. Zeltner NEU-2844 04/04/2000 CENT-72 6xHerbarium material Mal Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 250 Species Locality Collector Voucher Collection date Code Ploidy level Source Code for PhyloNetworks analyses C. maritimum (L.) Fritsch Portugal: Cruz de João Mendes Z. Díaz & V. Girón SEV-224120 30/04/2008 CENT-29 2xHerbarium material Mar C. maritimum (L.) Fritsch Spain: Cádiz, Algeciras, Zanona Z. Díaz MAR-551 22/05/2018 CENT-29.2 2xSilica-gel dried Mar C. portense Butcher Spain: Piélagos, Cantabria Ceballos & Fdez. Prieto FCO-32680 10/09/2011 CPO-1.2 2xSilica-gel dried Port C. pulchellum (Sw.) Druce Croatia: Trogir island J. Arroyo & A. Castro SEV-270218 19/06/2015 CENT-13 4xHerbarium material Pul C. pulchellum (Sw.) Druce Portugal: Aveiro, Puente de Barra Z. Díaz & A. Castro SEV-270223 07/05/2015 CENT-22 4xHerbarium material Pul C. quadrifolium subsp. barrelieri (L.M. Dufour) G. López Spain: Valencia, Els Corrals Z. Díaz SEV-241173 01/07/2009 CENT-14 2xHerbarium material Qua C. quadrifolium subsp. barrelieri (L.M. Dufour) G. López Spain: Valencia, Barx Z. Díaz – 21/06/2019 CAV-11.2 2xSilica-gel dried Qua C. quadrifolium subsp. linariifolium (Lam.) G. López Spain: Castellón, Mosqueruela Z. Díaz SEV-241185 02/07/2009 CENT-15 2xHerbarium material Lin C. quadrifolium subsp. linariifolium (Lam.) G. López Spain: Granada, Frigiliana B. Cabezudo & al. MGC-57199 23/05/2003 CENT-32 2xHerbarium material Lin C. quadrifolium subsp. parviflorum (Willk.) Pedrol Spain: Segovia, Hinojosa del Cerro Z. Díaz SEV-270224 21/07/2015 CENT-60 2xHerbarium material Qua C. quadrifolium subsp. quadrifolium (L.) G. López & C.E. Jarvis Spain: Madrid, Villaconejos-Chinchón Z. Díaz & F. García – 23/07/2015 CAV-15 2xSilica-gel dried Qua C. quadrifolium subsp. quadrifolium (L.) G. López & C.E. Jarvis Spain: Toledo Z. Díaz – 19/07/2016 12 quadri (TO) 2xSilica-gel dried Qua C. scilloides (L. fil.) Samp. Spain: Coruña, Porto do Son R. Carballal SANT-57987 20/10/2008 CENT-37 2xHerbarium material Port C. scilloides (L. fil.) Samp. Azores, Ilha do Pico Bueno & Fdez. Prieto FCO-32672 16/06/2011 CSC-1 (CAV-27) 2xSilica-gel dried Port C. serpentinicola Carlström Turkey: Mugla, Carie, Köycegiz L. & N. Zeltner NEU-2998 07/07/2000 CENT-74 4xHerbarium material Serp C. somedanum M. Laínz Spain: Oviedo, Pola de Somiedo Z. Díaz SEV-270225 21/07/2015 CENT-59 4xHerbarium material Port C. somedanum M. Laínz Spain: Oviedo, Valle de Lago C. Andrés SEV-270226 20/08/2015 CENT-63 4xHerbarium material Ulig1 C. suffruticosum (Griseb.) Ronniger Spain: Cádiz, Vejer V. Girón SEV-224201 01/05/2009 CENT-16 2xHerbarium material Maj C. tenuiflorum (Hoffmanns. & Link) Fritsch Croatia: Peljesac Peninsula, Zabrde J. Arroyo & A. Castro SEV-270219 19/06/2015 CENT-18 2xHerbarium material Ten C. turcicum (Velen.) Ronniger Turkey: Sinop Valcárcel & al. VAL-146633 22/06/2001 CENT-28 4xHerbarium material Ery2 Blackstonia perfoliata Spain: Huesca, Los Molinos, Sierra de Guara J. Viruel SEV 241216 23/06/2009 CENT-19 – Herbarium material Out Exaculum pusillum Spain: Huelva, Embalse Grande Z. Díaz & V. Girón SEV-270228 03/06/2010 CENT-24 – Herbarium material Out Schenkia spicata Spain: Albacete, Cordovilla Z. Díaz SEV-270220 11/08/2010 CENT-20 – Herbarium material Out Table 1. Continued Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 251 are common in groups that have suffered from processes such as incomplete lineage sorting, hybridization and/or introgression (Maddison, 1997). The study of the relationships among polyploid taxa – especially allopolyploids – has been limited because the subgenomes convey different phylogenetic signals. In these cases, using only classical bifurcating phylogenetic trees is not the best approach to unravel their evolutionary history, and network approaches may be more informative. Polyploid genera such as Salix, Fothergilla and Dactylorhiza (Qi et al., 2015; Brandrud et al., 2020; He et al., 2020; Wagner et al., 2020) have been insightfully studied using RADseq approaches. Differences in morphological traits between diploids and polyploids have been documented (Stebbins, 1947, 1950; Husband and Schemske, 2000). For example, several studies indicate a positive correlation between genome size and cell size (Müntzing, 1936; Otto and Whitton, 2000; Gregory, 2001; Beaulieu et al., 2008), so that as ploidy level increases there is a corresponding increase in the size of cells and structures. Specifically, it has been proposed that polyploids exhibit larger leaves and flowers compared with diploids (e.g. Husband and Schemske, 2000; Li et al., 2010; Balao et al., 2011; Kim et al., 2012; Laport and Ramsey, 2015; Etterson et al., 2016). However, important reproductive traits (floral size and herkogamy) have been failed to be associated with ploidy levels in genus Centaurium (Jiménez-Lobato et al., 2019). In this context, we use RADseq markers to (1) reconstruct the phylogenetic relationships among Centaurium species and subspecies, (2) infer the incidence of hybridization in the genus by identifying hybrid species and their parentage at the species or lineage level, and (3) explore morphological traits that may be associated with specific ploidy levels. In doing so, we aim to provide further insight into specific case studies that may contribute to estimating the evolution of polyploidy and hybridization in the Mediterranean flora. MATERIALS AND METHODS Plant material A total of 42 samples belonging to 28 taxa of Centaurium (15 species and 13 subspecies) encompassing the whole range of distribution were included in this study (Table 1). Exaculum pusillum Caruel, Schenkia spicata (L.) G. Mans. and Blackstonia perfoliata (L.) Huds. were included as outgroups (Table 1). We collected samples from the field and from different herbaria: University of Seville (SEV), University of Oviedo (FCO), Royal Botanic Gardens, Madrid, CSIC (MA), University of Santiago de Compostela (SANT), University of Valencia (VAL), University of Málaga (MGC) and University of Neuchâtel (NEU). We followed the taxonomic treatment proposed by Mansion et al., (2005) and reviewed by Díaz- Lifante (2012). Several Centaurium hybrids have received a formal name (Centaurium × aschersonianum (Seemen) Hegi, Centaurium × cicekii Yıld. & Yaprak, Centaurium × joliv etinum P. Fourn. and Centaurium × litardierei Ronniger, among others). However, these taxa represent occasional hybrids and have not been included in this study, as it focuses on hybridization events that have evolutionary implications. DNA extraction, sequencing, and data treatment DNA extraction was carried out using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA), following the manufacturer’s instructions. DNA concentration was evaluated with a Qubit 2.0 Fluorometer (Life Technologies, Grand Island, NY, USA) and its quality checked on a 1 % agarose gel. RADseq libraries and barcoding were prepared following Baird et al. (2008), using the restriction enzyme Pstl, by Floragenex Inc. (Beaverton, OR, USA), sequenced on the Illumina HiSeq 2000 platform. Once we had the raw data, we used ipyrad 0.9.65 (Eaton and Overcast, 2020) under a de novo assembly for demultiplexing and clustering. The filtering step was performed with a maximum of four low-quality base calls per read, and the minimum length of reads after the adapter trim was set at 35 bp. For the alignment, we used a clustering threshold of 90 %, we set a minimum number of samples per locus of 20, and we allowed at most two alleles per site in consensus sequences. Phylogenetic analyses Phylogenetic relationships were inferred using a maximum likelihood (ML) approach with IQ-TREE 1.6.11 (Minh et al., 2020) for all ploidy levels within Centaurium (diploid, tetraploid and hexaploid). IQ-TREE analyses were run with the GTR + I + G substitution model (Swofford et al., 1996). Statistical support was estimated by 1000 ultrafast bootstrap (UFBoot) replicates (Hoang et al., 2018) and by 1000 SH-like approximate likelihood ratio test (SH-aLRT) replicates (Guindon et al., 2010). We used the quartet-based method of quartet sampling (QS) to examine the potential phylogenetic discordances of internal and terminal branches with the obtained IQ-TREE topology (Pease et al., 2018). We estimated the quartet concordance score (QC), the quartet differential score (QD) and the quartet informativeness score (QI) for each internal node and the quartet fidelity score (QF) for each terminal node by conducting an analysis consisting of four parallel threads and 10 000 replicates. The QC measures the concordance between the QS-derived topology and the test topology (here, the IQ-TREE topology), returning positive values if a given branch is concordant between the two topologies and a negative value if they are discordant. The QD estimates whether the frequencies of two alternative topologies of a discordant branch are similar or favour one over the other; when QD values are close to 1, none of the alternative topologies is favoured, while values close to 0 mean that one of the alternatives is favoured (Pease et al., 2018). The QI shows the proportion of informativeness of each replicate; when its value is close to 1, the replicates are informative and when close to 0, the informative replicates descend. QS values (QC/QD/QI) are shown with bootstrap support (BS) values along the phylogeny. We used the obtained IQ-TREE topology and 10 000 randomly chosen SNPs to obtain 3060 quartets, using four threads and 1000 bootstrap repetitions. As tree topologies cannot completely describe complex evolutionary scenarios such as hybridization (Huson and Bryant, 2006), we performed a Neighbor-Net analysis in SplitsTree4 v. 4.18.3 (Huson and Bryant, 2006) including all ploidy levels within the genus. The genetic distance was inferred by the Uncorrected P method, while statistical support for branches was estimated by 10 000 Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 252 BS replicates. The recovered network was built by compatible partitions of sets of taxa (or splits; Bryant and Moulton, 2004). The split weight (i.e. branch length) indicates the depth of divergence between taxa, with shorter branches corresponding to genetically similar taxa and longer branches corresponding to more genetically distinct taxa. Since our dataset included Centaurium species with different ploidy levels (2x, 4x, 6x) and they are susceptible to having experienced incomplete lineage sorting, we also used the summary coalescent approach SVDquartets (Chifman and Kubatko, 2015) as implemented in PAUP* 4.0b10 (Swofford, 2002). Thus, a consensus species tree was inferred by analysing 1 000 000 randomly chosen quartets. Branch support was estimated by 10 000 BS repetitions. Exploring genetic diversity: identification of hybridization events Potential hybridization events between lineages occurring in different nodes from the IQ-TREE topology were identified using the SNaQ function of the PhyloNetworks Julia package (Solís-Lemus and Ané, 2016; Solís-Lemus et al., 2017). The maximum pseudolikelihood-based SNaQ approach (Solís- Lemus and Ané, 2016) simultaneously infers the main topology (i.e. species relationships) and the underlying historical reticulations (i.e. ancient hybridizations events). We estimated the observed quartet concordance factors (CFs) of our initial alignment in the R package SNPs2CF (Olave and Meyer, 2020), setting the maximum number of SNPs to be used to 15 000 and the maximum number of quartets to 1 000 000, and performing bootstrap replicates to obtain the lower and upper limit of the credibility intervals. For the analyses we used an Imap file (the code used in this analysis is shown in Table 1) considering the presence of multiple individuals per species and, in some clades, several species per clade. This allowed us to establish the individual–clade association. The SNaQ analyses were based on the IQ-TREE topology and the CFs of 3060 quartets. This method consisted of inferring 10 independent topologies, each one repeated 10 times, with different hmax values (i.e. number of different hybridization events). Hmax values ranged from 0 (no hybridization events) to 9. The hmax value was considered the most adequate when the addition of another event did not improve the likelihood score (loglik function). The bootstrap analyses were performed with 100 replicates and 5 runs per replicate. Our Centaurium taxa included three ploidy levels (2x, 4x and 6x), which could include recent and ancient hybridizations or autopolyploid events. Therefore, we aimed to determine which type of polyploidization events (autopolyploid vs allopolyploid) explained both the variety of ploidy levels and the observed phylogenetic relationships. We explored the genetic structure of the Centaurium data using Bayesian clustering analysis with fastSTRUCTURE (Raj et al., 2014), an algorithm designed to infer population structure from large SNPs in a Bayesian framework. The study was performed for all studied taxa, for K values (i.e. the number of genotypic groups) from 2 to 30. We also determined the optimal partition of the data (the K value that best explains its genetic structure) with the tool chooseK implemented in fastSTRUCTURE (Raj et al., 2014). We used SNIPloid, developed by Peralta et al. (2013), to analyse and classify the SNPs of allopolyploid species to trace them with their putative parentals. This tool was originally designed to classify RNA-Seq SNPs, although Wagner et al. (2020) developed a pipeline for RADseq data, by using biallelic SNPs instead of sequence data. This software classifies the SNPs into different categories: categories 1 and 2 correspond to interspecific SNPs that match one of the parental genomes (to parental1 or parental2, respectively) (e.g. parental1 A/A, parental2 G/G, and hybrid G/G). Category 3or4 corresponds to SNPs that do not match either of the diploid parental genomes, as mutations may have occurred in one of the subgenomes of the allopolyploid after the polyploidization event (e.g. parental1 A/A, parental2 A/A, and hybrid A/G). Category 5 comprises the putative homoeo-SNPs (i.e. polymorphisms that occurred in the hybrid species and in the parental genomes) (e.g. parental1 A/A, parental2 G/G, and hybrid A/G). Any SNPs that do not fall into any of the previous categories are classified as other (Peralta et al., 2013). We studied here the allopolyploid origin and parentage of five polyploid species: C. discolor, C. serpentinicola, C. malzacianum Maire, C. mairei Zeltner and C. centaurioides R.S. Rao & Hemadri. Putative parental species of C. discolor (C. maritimum and C. tenuiflorum) were suggested by Guggisberg et al. (2006). To perform our analyses, we used the hypothetical parents proposed by Mansion et al. (2005) for the other four polyploids. However, for C. malzacianum, C. centaurioides and C. mairei we also used partially different parentage suggested by our own previous results from SNIPloid or fastSTRUCTURE. Specifically, we compared the SNPs of C. serpentinicola with those of their putative parentals suggested by Mansion et al. (2005), i.e. C. erythraea subsp. rumelicum (Velen.) Melderis and C. tenuiflorum. For the hexaploid C. malzacianum we considered C. maritimum and C. pulchellum, both from the widespread clade, as putative parentals according to Mansion et al. (2005). Additionally, we also used C. pulchellum and C. grandiflorum subsp. boissieri (Willk.) Z. Díaz (from the western clade) as potential parental taxa. This followed our previous SNIPloid analysis including C. maritimum and C. pulchellum, which suggested that C. pulchellum and one species from the western clade, not C. maritimum, could be the parental species. Centaurium grandiflorum subsp. boissieri was randomly selected since the parental species from the western clade was not identified. Regarding the hexaploid C. centaurioides, we considered C. pulchellum and C. tenuiflorum as putative parental species according to Mansion et al. (2005) in addition to an alternative hypothetical parentage based on the results of our previous SNIPloid analyses consisting of C. tenuiflorum and C. erythraea Rafn subsp. erythraea. SNIPloid suggested that C. tenuiflorum, not C. pulchellum, and another species from the western clade (C. erythraea subsp. erythraea was randomly selected) could be the parental taxa. Regarding C. mairei, Mansion et al. (2005) proposed this taxa to be an autopolyploid from tetraploid populations of C. pulchellum. In view of our genetic results, this hexaploid could not be an autopolyploid, as the parental species are included in different genetic clusters. Thus, another species from the western clade was randomly selected: C. erythraea subsp. erythraea. Linking morphology and ploidy level To explore phenotypic consequences of whole-genome duplication in the evolution of Centaurium, we examined whether Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 253 there is an association between ploidy level and several morphological traits. These were taken from Jiménez-Lobato et al. (2019) and consisted of eight morphological traits all coded as binary: flower display, flower size, anther length, androecium symmetry, style position, herkogamy and stigma length or life cycle (annual/biennial vs perennial). We used non-metric multidimensional scaling (NMDS; Kruskal, 1964) to ordinate samples based on these traits (Supplementary Data Table S1). First, we calculated distances between variables under the Jaccard distance method for binary values with the R package vegan, and then we performed the NMDS analysis with the metaMDS function in RStudio (v. 2021.09.2). RESULTS DNA sequencing The average number of raw reads from the sequencing was 3.8 million reads per sample. After optimization, we recovered a total of 215 894 filtered RADseq loci. The recovered aligned sequence matrix with 45 accessions had a length of 1 011 716 bp and the SNP matrix a length of 139 464 bp, with 57 437 parsimony-informative sites. Phylogenetic inference The topology recovered from IQ-TREE confirmed the sister relationship of Exaculum and Schenkia (BS 100, 1/NA/0.98) (Fig. 1). Two main well-supported clades were found within the ingroup: the ‘western’ and the ‘widespread’ clades (Fig. 1), as previously identified in Maguilla et al. (2021). The widespread clade (BS 98, 0.29/0.15/0.87) comprises eight species: the hexaploids C. centaurioides and C. mairei Zeltner, the tetraploids C. capense Broome, C. serpentinicola, C. pulchellum and C. discolor; and the diploids C. maritimum and C. tenuiflorum. Among the western clade (BS 99, 0.039/0.68/0.88), two subclades can be recognized, subclades A and B, as well as an accession of the hexaploid C. malzacianum. Subclade A (BS 100, 0.12/0.46/0.94) is composed of tetraploid subspecies of C. littorale (Turner) Gilmour (subsp. littorale and C. littorale subsp. uliginosum (Waldst. & Kit.) Melderis), the diploids C. portense and C. scilloides, the tetraploids C. somedanum M. Laínz and C. chloodes (Brot.) Samp. and, finally, the diploid subspecies of C. quadrifolium (subsp. parviflorum (Willk.) Pedrol, subsp. quadrifolium (L.) G. López & C.E. Jarvis and subsp. barrelieri (L.M. Dufour) G. López). Subclade B (BS 100, 0.092/0.8/0.95) is composed of the diploid subspecies of C. grandiflorum (subsp. boissieri, subsp. grandiflorum (Pers.) Ronniger and subsp. majus (Hoffmanns. & Link) Z. Díaz), the diploid C. erythraea subsp. rumelicum, the tetraploids C. erythraea subsp. erythraea, C. erythraea subsp. rhodense (Boiss. & Reut.) Melderis and C. erythraea var. subcapitatum (Corb.) Ubsdell, as well as the tetraploid C. turcicum and the diploid C. suffruticosum (Griseb.) Ronniger and two accessions of the diploid C. quadrifolium subsp. linariifolium. (Lam.) G. López. Similar to the phylogenetic reconstructions (Fig. 1), the quartet-based tree identifies two main clades within the genus in which most of the nodes had significant support (Fig. 2). However, some polytomies are observed across the tree: one within the western clade affecting the C. somedanum CENT59 accession; two involving hexaploid accessions with a haploid number of 28 (C. malzacianum and C. centaurioides), which were not located in any of the main clades; four within the widespread clade affecting the polyploids C. discolor and C. serpentinicola, the putative parental species C. maritimum, and the hexaploid C. mairei accessions. Furthermore, the widespread clade, which had the most polytomies, had no significant support. The polytomies affecting hexaploids and tetraploids may be caused by hybridization events, and in this case the quartet-based method is unable to conclusively resolve the evolutionary relationships among these taxa. The polytomies affecting the accessions of C. malzacianum and C. somedanum (CENT59) are not unexpected. The first one is one of the main incongruences of our phylogenetic tree and the previous reconstructions of the genus (Jiménez-Lobato et al., 2019; see Discussion section), and the latter is reconstructed as paraphyletic in the IQ-TREE reconstruction (Fig. 1). The Neighbor-Net of diploid, tetraploid and hexaploid Centaurium species (Fig. 3A) allows recognition of two clusters that coincide with the two main clades displayed in the phylogenetic reconstruction (Fig. 1). The depth of divergence between taxa is consistent with the branch lengths in the network (Fig. 3A). Short branches in the network indicate low divergence of species and parallel edges (representing alternative splits) indicate uncertainty potentially resulting from hybridization events that have occurred within Centaurium taxa. Identification of reticulation and hybridization events The loglik scores obtained from the different SNaQ analyses based on different hmax values indicate that three hybridization events is the most probable scenario (Supplementary Data Fig. S2). One of the three hybridization events involved species from the two main clades whereas the other two occurred within the western clade (Fig. 3B). Centaurium maritimum (Mar), from the widespread clade hybridized with C. malzacianum (Mal) and currently 35.2 % of its genome (estimated from SNPs) comes from C. malzacianum. The ancestor of C. grandiflorum s.s. (Grand) and C. grandiflorum subsp. boissieri (Boi) hybridized with C. quadrifolium subsp. linariifolium (Lin), and currently the C. grandiflorum lineage has 49 % of C. quadrifolium subsp. linariifolium’s genome. The third identified hybridization event involves the clade formed by the two accessions of C. quadrifolium subsp. barrelieri, the one of C. quadrifolium subsp. parviflorum and both accessions of C. quadrifolium s.s. (Qua), which seems to have hybridized with C. somedanum or C. littorale subsp. uliginosum accession CENT71 (Ulig1), so that 24.3 % of the genome in taxa of the clade Qua comes from either of these two species (Fig. 3B). Bayesian clustering analyses of Centaurium with fastSTRUCTURE revealed that the best number of genetic clusters to explain the structure of the dataset is two (Fig. 3C, Supplementary Data Fig. S3), which matches the western and widespread clades. These analyses identified four taxa showing admixed ancestry, with similar proportions of the two genetic groups: the tetraploid C. serpentinicola and the hexaploids C. mairei, C. malzacianum and C. centaurioides. One of the Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 254 two accessions of the diploid C. maritimum (CENT29.2) also showed a percentage of admixture. Allopolyploid origin of Centaurium species The allopolyploid origin of the tetraploid C. discolor proposed by Guggisberg et al. (2006), involving C. maritimum and C. tenuiflorum, was tested with SNIPloid (Fig. 4A). 44.5 % of the C. discolor's SNPs were classified in category 5 (the variations found in the genomes occur both in the hybrid and in one of the parental genomes), 19.2 % in category 3or4 (i.e. variations in SNPs occurring in C. discolor that could not be identified with certainty in either of the parent genomes), 17.9 % were classified as other (not falling into any of the defined categories), 12.6 % in category 2 and 5.8 % in category 1 (i.e. some alleles were specific to the genomes of C. tenuiflorum and C. maritimum, respectively). Exaculum pusillum 100 1/NA/0.98 100 0.76/0.4/0.99 100 0.51/0.5/0.98 100 –0.2/0.22/0.94 100 0.76/0.46/0.99 100 0.56/0.33/0.98 100 0.66/0/0.98 85 -0.33/0.055/0.98 95 0.3/0.12/1 100 0.46/0.76/0.99 100 0.38/0.3/0.98 100 0.092/0.8/0.95 100 0.092/0.8/0.95 98 0.22/0.49/0.96 100 0.83/0/0.93 100 1/NA/1 87 –0.21/0.2/0.96 100 0.59/0.8/0.98 98 0.19/0.5/0.97 85 –0.6/0.093/0.97 84 –0.3/0.28/0.96 98 0.2/0.52/0.97 83 –0.32/0.04/0.96 100 0.12/0.46/0.94 100 0.12/0.46/0.94 99 0.039/0.68/0.88 100 0.43/0/0.99 100 0.12/0.94/0.98 100 0.62/0.016/0.98 100 0.83/0/0.98 100 0.95/0/0.94 100 0.95/0/0.94 98 0.25/0.27/0.89 98 0.29/0.15/0.87 89 0.28/0.13/0.94 98 -0.25/0.14/0.96 89 0.28/0.13/0.94 98 –0.25/0.14/0.96 100 0.28/0.28/0.94 98 0.27/0.13/0.93 98 0.39/0/0.97 97 1/NA/1 100 0.44/0.75/1 100 0.3/0/0.96 100 1/NA/1 100 1/NA/0.99 100 1/NA/0.99 100 0.36/ 0.008/ 0.97 Schenkia spicata C. grandiflorum subsp. boissieri CENT4 (2x) C. grandiflorum subsp. boissieri CAV5 (2x) C. grandiflorum subsp. grandiflorum CENT6 (2x) C. grandiflorum subsp. grandiflorum CENT68 (2x) C. grandiflorum subsp. majus CENT31 (2x) B A Widespread clade C. grandiflorum subsp. majus CENT9 (2x) C. erythraea subsp. erythraea CENT64 (4x) C. erythraea var. subcapitatum CENT54 (4x) C. erythraea subsp. erythraea CENT3 (4x) C. erythraea subsp. rhodense CENT77 (4x) C. erythraea subsp. rhodense CAV2.2 (4x) C. quadrifolium subsp. linariifolium CENT15 (2x) C. quadrifolium subsp. linariifolium CENT32 (2x) C. littorale subsp. littorale CENT55 (4x) C. littorale subsp. uliginosum CENT71 (4x) C. quadrifolium subsp. parviflorum CENT60 (2x) C. quadrifolium subsp. quadrifolium CAV15 (2x) C. quadrifolium subsp. quadrifolium 12quadriTO (2x) C. quadrifolium subsp. barrelieri CENT14 (2x) C. quadrifolium subsp. barrelieri CAV11.2 (2x) C. malzacianum CENT72 (6x) C. centaurioides CENT51 (6x) C. mairei CENT70 (6x) C. capense CENT53 (4x) C. serpentinicola CENT74 (4x) C. tenuiflorum CENT18 (2x) C. pulchellum CENT13 (4x) C. pulchellum CENT22 (4x) C. discolor CENT25 (4x) C. maritimum CENT29.2 (2x) Blackstonia perfoliata 0.02 C. maritimum CENT29 (2x) C. littorale subsp. littorale CENT76 (4x) C. littorale subsp. uliginosum CENT56 (4x) C. portense CPO1.2 (2x) C. scilloides CENT37 (2x) C. scilloides CAV27 (2x) C. somedanum CENT59 (4x) C. chloodes CENT26 (4x) C. somedanum CENT63 (4x) C. turcicum CENT28 (4x) C. erythraea subsp. rumelicum CENT36 (2x) C. suffruticosum CENT16 (2x) Fig. 1. Maximum likelihood phylogenetic reconstruction of all the samples for the genus Centaurium performed with RAD markers. Bootstrap and quartet sampling support are displayed at the nodes: bootstrap support, quartet concordance/quartet differential score/quartet informativeness. Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 255 Testing the hybrid ancestry of the tetraploid C. serpentinicola proposed by Mansion et al. (2005) using SNIPloid, 49 % of SNPs were found to be specific to C. tenuiflorum (category 2), 34 % fell in category 5, 10 % were classified as other and 7 % were specific to C. erythraea subsp. rumelicum (category 1) (Fig. 4B). Testing the hybrid ancestry of the hexaploid C. malzacianum proposed by Mansion et al. (2005), which involves C. maritimum and C. pulchellum, 50 % of C. malzacianum’s SNPs were found in the genome of C. pulchellum (category 2), 25 % were specific to C. maritimum (category 1), 21 % were referred to category other and 4 % were classified in category 5 (Fig. 4C). When we considered our previous results and tested the ancestry of the same hexaploid with C. pulchellum and a species from the western clade (C. grandiflorum subsp. boissieri), 53 % of the SNPs were specific to C. grandiflorum subsp. boissieri (category 2), 28 % were specific to C. pulchellum (category 1), 13 % fell in the category other and 6 % in category 5 (Fig. 4D). Testing the hybrid ancestry of the hexaploid C. centaurioides according to Mansion et al. (2005), using C. pulchellum and C. tenuiflorum as parents, 42 % of the SNPs were also found in the genome of C. tenuiflorum (category 2), 23 % in the C. pulchellum genome (category 1), 31 % classified in the category other, and 4 % in category 5 (Fig. 4E). When we considered our previous results and tested the ancestry of this hexaploid species with C. tenuiflorum and a species from the western clade (C. erythraea subsp. erythraea) 57 % of C. centaurioides SNPs were specific to C. erythraea subsp. erythraea (category 2), 21 % were specific to C. tenuiflorum (category 1), 19 % were classified as other, and 3 % fell in category 5 (Fig. 4F). Finally, testing the ancestry of C. mairei considering C. pulchellum and C. erythraea subsp. erythraea, 54 % were specific to C. erythraea subsp. erythraea (category 2), 17 % of C. mairei SNPs were specific to C. pulchellum (category 1), 26 % were classified as other, and 3 % fell in category 5 (Fig. 4G). Ploidy level and morphology The stress score of the NMDS analysis for two dimensions was 0.0689427, indicating a low discrepancy between the original distances and those obtained after dimensionality reduction (Kruskal, 1964), as well as a correct fit of the graphical representation in the model. The NMDS ordination analysis revealed differences among ploidy levels for some morphological traits (Fig. 5). Specifically, based on the data provided in Supplementary Data Table S1, it can be inferred that most diploid Centaurium species had larger flowers and anthers than tetra- and hexaploid species. In addition, most diploid species showed no contact between stigmas and anthers, contrary to polyploid (both 4x and 6x) species, Exaculum pusillum 99.2 99.9 99.9 99.9 99.9 98.9 99.7 99.7 99.7 99.3 57.8 98.4 61.2 96.6 69.9 99.8 93.7 99.9 64.9 84.1 92.2 99.8 92.8 99.4 97.2 77.3 55.3 79.8 99.6 54.0 2.0 53.5 53.7 94.6 75.1 Schenkia spicata C. g. boissieri_CENT4 (2x) C. g. boissieri_CAV5 (2x) C. g. grandiflorum_CENT68 (2x) C. g. grandiflorum_CENT6 (2x) C. e. erythraea_CENT64 (4x) C. e. subcapitatum_CENT54 (4x) B Western clade Widespread clade A C. e. erythraea_CENT3 (4x) C. turcicum_CENT28 (4x) C. e. rhodense_CENT77 (4x) C. e. rhodense_CAV2.2 (4x) C. suffruticosum_CENT16 (2x) C. e. rumelicum_CENT36 (2x) C. g. majus_CENT31 (2x) C. g. majus_CENT9 (2x) C. q. linariifolium_CENT15 (2x) C. q. linariifolium_CENT32 (2x) C. l. littorale_CENT55 (4x) C. l. uliginosum_CENT71 (4x) C. somedanum_CENT59 (4x) C. somedanum_CENT63 (4x) C. portense_CPO1.2 (2x) C. scilloides_CENT37 (2x) C. q. parviflorum_CENT60 (2x) C. q. quadrifolium_CAV15 (2x) C. q. quadrifolium_12quadri(TO) (2x) C. q. barrelieri_CENT14 (2x) C. q. barrelieri_CAV11.2 (2x) C. l. littorale_CENT76 (4x) C. l. uliginosum_CENT56 (4x) C. scilloides_CAV27 (2x) C. malzacianum_CENT72 (6x) C. centaurioides_CENT51 (6x) C. discolor_CENT25 (4x) C. maritimum_CENT29.2 (2x) C. tenuiflorum_CENT18 (2x) C. pulchellum_CENT13 (4x) C. pulchellum_CENT22 (4x) C. capense_CENT53 (4x) C. mairei_CENT70 (6x) C. serpentinicola_CENT74 (4x) C. maritimum_CENT29 (2x) Blackstonia perfoliata C. chloodes_CENT26 (4x) Fig. 2. Consensus species tree inferred under the coalescent approach SVDquartets. Taxa, internal code and ploidy level are shown at the tips of the tree. Bootstrap support is shown at the nodes. Downloaded from https://academic.oup.com/aob/article/134/2/247/7660149 by guest on 27 September 2024 Valdés-Florido et al. ― Unravelling the evolutionary history of Centaurium (Gentianaceae) 262 Marques I, Montgomery SA, Barker MS, et al. 2016. Transcriptome-derived evidence supports recent polyploidization and a major phylogeographic division in Trithuria submersa (Hydatellaceae, Nymphaeales). New Phytologist 210: 310–323. Marques I, Loureiro J, Draper D, Castro M, Castro S. 2018. How much do we know about the frequency of hybridisation and polyploidy in the Mediterranean region? Plant Biology 20: 21–37. Mason AS, Pires JC. 2015. Unreduced gametes: meiotic mishap or evolutionary mechanism? Trends in Genetics 31: 5–10. McKain MR, Johnson MG, Uribe-Convers S, Eaton D, Yang Y. 2018. Practical considerations for plant phylogenomics. Applications in Plant Sciences 6: 15. Minh BQ, Schmidt HA, Chernomor O, et al. 2020. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Molecular Biology and Evolution 37: 1530–1534. Müntzing A. 1936. The evolutionary significance of autopolyploidy. Hereditas 21: 363–378. Nieto Feliner G. 2014. Patterns and processes in plant phylogeography in the Mediterranean Basin. A review. Perspectives in Plant Ecology, Evolution and Systematics 16: 265–278. Nieto Feliner G, Casacuberta J, Wendel JF. 2020. Genomics of evolutionary novelty in hybrids and polyploids. Frontiers in Genetics 11: 792. Nieto Feliner G, Cellinese N, Crowl AA, Frajman B. 2023. Understanding plant diversity and evolution in the Mediterranean Basin. Frontiers in Plant Science 14: 1152340. Ning G, Shi X, Hu H, Yan Y, Bao M. 2009. Development of a range of polyploid lines in Petunia hybrida and the relationship of ploidy with the single‐/double‐flower trait. HortScience 44: 250–255. Olave M, Meyer A. 2020. Implementing large genomic single nucleotide polymorphism data sets in phylogenetic network reconstructions: a case study of particularly rapid radiations of cichlid fish. Systematic Biology 69: 848–862. Otto SP, Whitton J. 2000. Polyploid incidence and evolution. Annual Review of Genetics 34: 401–437. Pease JB, Brown JW, Walker JF, Hinchliff CE, Smith SA. 2018. Quartet Sampling distinguishes lack of support from conflicting support in the green plant tree of life. American Journal of Botany 105: 385–403. Peralta M, Combes MC, Cenci A, Lashermes P, Dereeper A. 2013. SNiPloid: a utility to exploit high-throughput SNP data derived from RNA-Seq in allopolyploid species. International Journal of Plant Genomics 1687– 5370: 890123. Plants of the World Online. Royal Botanic Gardens, Kew. http://www. plantsoftheworldonline.org (April 2023). Prieto JAF, Cires E, Pérez R, Bueno A. 2012. A new endemism for the Azores: the case of Centaurium scilloides (L. f.) plant systematics and evolution. Plant Systematics and Evolution 298: 1867–1879. Qi ZC, Yu Y, Liu X, et al. 2015. Phylogenomics of polyploid Fothergilla (Hamamelidaceae) by RAD-tag based GBS—insights into species origin and effects of software pipelines. Journal of Systematics and Evolution 53: 432–447. Raj A, Stephens M, Pritchard JK. 2014. fastSTRUCTURE: variational inference of population structure in large SNP data sets. Genetics 197: 573–589. Ramsey J, Schemske DW. 1998. Pathways, mechanisms, and the rate of polyploid formation in flowering plants. Annual Review of Ecology, Evolution, and Systematics 29: 467–501. Rothfels CJ. 2021. Polyploid phylogenetics. New Phytologist 230: 66–72. Santos-Gally R, Vargas P, Arroyo J. 2012. Insights into Neogene Mediterranean biogeography based on phylogenetic relationships of mountain and lowland lineages of Narcissus (Amaryllidaceae). Journal of Biogeography 39: 782–798. Segraves K. 2017. The effects of genome duplications in a community context. New Phytologist 215: 57–69. Solís-Lemus C, Ané C. 2016. Inferring phylogenetic networks with maximum pseudolikelihood under incomplete lineage sorting. PLoS Genetics 12: e1005896. Solís-Lemus C, Bastide P, Ané C. 2017. PhyloNetworks: a package for phylogenetic networks. Molecular Biology and Evolution 34: 3292–3298. Soltis DE, Albert VA, Leebens‐Mack J, et al. 2009. Polyploidy and angiosperm diversification. American Journal of Botany 96: 336–348. Stebbins GL. 1947. Types of polyploids: their classification and significance. Advances in Genetics 1: 403–429. Stebbins GL. 1950. Variation and evolution in plants. New York: Columbia University Press. Stebbins GL. 1971. Chromosomal evolution in higher plants. London: Edward Arnold. Suc JP. 1984. Origin and evolution of the Mediterranean vegetation and climate in Europe. Nature 307: 429–432. Swofford DL. 2002. PAUP* phylogenetic analysis using parsimony (*and other methods). Sunderland: Sinauer Associates. Swofford DL, Olsen GJ, Waddell PJ, Hillis DM. 1996. Phylogenetic inference. In: Hillis DM, Moritz C, Mable BK. eds. Molecular systematics. Sunderland: Sinauer Associates, 407–514. Tate JA, Soltis PS, Soltis DE. 2005. Polyploidy in plants. In: Gregory TR. ed. The evolution of the genome. San Diego, Californa, USA: Elsevier Science & Technology, Academic Press. Thompson JD. 2005. Plant evolution in the Mediterranean. Oxford: Oxford University Press. Vallejo-Marín M, Cooley AM, Lee MY, Folmer M, McKain MR, Puzey JR. 2016. Strongly asymmetric hybridization barriers shape the origin of a new polyploid species and its hybrid ancestor. American Journal of Botany 103: 1272–1288. Vamosi J, Goring S, Kennedy B, et al. 2007. Pollination, floral display, and the ecological correlates of polyploidy. Functional Ecosystems and Communities 1: 1–9. Van de Peer Y, Ashman TL, Soltis PS, Soltis DE. 2021. Polyploidy: an evolutionary and ecological force in stressful times. Plant Cell 33: 11–26. Wagner ND, He L, Hörandl E. 2020. Phylogenomic relationships and evolution of polyploid Salix species revealed by RAD sequencing data. Frontiers in Plant Science 11: 1077. Wang X, Morton JA, Pellicer J, Leitch IJ, Leitch AR. 2021. Genome downsizing after polyploidy: mechanisms, rates and selection pressures. Plant Journal 107: 1003–1015. Wendel JF. 2015. The wondrous cycles of polyploidy in plants. American Journal of Botany 102: 1753–1756. Wood TE, Takebayashi N, Barker MS, Mayrose I, Greenspoon PB, Rieseberg LH. 2009. The frequency of polyploid speciation in vascular plants. Proceedings of the National Academy of Sciences of the USA 106: 13875–13879. Zeltner L. 1970. Recherches de biosystématique sur les genres Blackstonia Huds. et Centaurium Hill (Gentianaceae). Bulletin de la Societe des sciences naturelles de Neuchatel 93: 1–164. Zeltner L. 1985. Étude cytotaxonomique et cytogéographique du Centaurium pulchellum (Swartz) Druce, sensu lato. Botanica Helvetica 95: 47–57. Zeltner L. 1991. Contribution à l’étude cytogéographique des genres Blackstonia Huds. et Centaurium Hill. (Gentianaceae) en Turquie, à Rhodes et à Chypre. Bulletin de la Société des sciences naturelles de Neuchâtel 114: 77–103. 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/aob/article/134/2/247/7660149 by guest on 27 September 2024