Botanical Journal of the Linnean Society, 2016. With 7 figures Phylogenetics and evolution of the Tillandsia utriculata complex (Bromeliaceae, Tillandsioideae) inferred from three plastid DNA markers and the ETS of the nuclear ribosomal DNA
1,2*, IVON
M. RAM
IREZ-MORILLO1, GERMAN
CARNEVALI1, JUAN P. PINZON 3 3 2 MICHAEL H. J. BARFUSS , WALTER TILL , JUAN TUN and JUAN J. ORTIZ-D
IAZ2 1
A.C., Unidad de Recursos Naturales-Herbario CICY, Centro de Investigaci
on Cient
ıfica de Yucatan, Calle 43 No. 130 Colonia Chuburna
de Hidalgo, CP 96200 Mérida, Yucatán, Mexico 2 Departamento de Botánica, Campus de Ciencias Biológicas y Agropecuarias, Universidad Autónoma de Yucatán, Carretera Mérida-Xmatkuil km. 15.5, Apdo, Postal 4-115 Itzimná, CP 97100 Mérida,
Mexico Yucatan, 3 Department of Botany and Biodiversity Research, Faculty of Life Sciences, University of Vienna, Rennweg 14, 1030 Vienna, Austria Received 31 July 2015; revised 13 February 2016; accepted for publication 4 March 2016 We performed a phylogenetic analysis using maximum parsimony and Bayesian inference of three plastid DNA markers and the external transcribed spacer (ETS) of nuclear ribosomal DNA to assess the species composition of the Tillandsia utriculata complex and their phylogenetic relationships, and to reconstruct patterns of character evolution and biogeography. The results showed that species of the T. utriculata complex are nested in a clade composed mainly of Mexican and Central American species of T. subgenus Tillandsia (Mexican Clade), and are organized in two lineages: the T. utriculata clade and the T. limbata clade. The ancestor of the core Mexican Clade was probably a T. utriculata-like epiphyte (Group II-type remote flowers and flexuous rachises). The T. utriculata clade is defined morphologically by the presence of acute petals. In this clade, there are two lineages: one of high-elevation, saxicolous, grey-leaved plants from the Mexican Plateau; and one which is more widespread and found from the Gulf of Mexico to Venezuela. The T. limbata clade probably arose in western Mesoamerica and is defined by rounded petals. These species are found mainly in tropical dry forests, but one species colonized wet environments of eastern Mesoamerica. Finally, analyses based on the ETS region allowed us to distinguish between T. utriculata and T. pringlei. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 ADDITIONAL KEYWORDS: biogeography – Central America – matK – Mexico – Neotropics – rpl32-trnL – rps16 – taxonomy. INTRODUCTION Tillandsia L. is the most diverse genus of Bromeliaceae, with > 600 species (Luther, 2012), distributed in tropical and subtropical America, and is one of the most distinctive components of the epiphytic and epilithic flora in that region (Benzing, 2000). The taxonomy of the genus is based mainly on the monograph of subfamily Tillandsioideae (Smith & Downs, 1977) and the subgeneric classification hinges on one or a *Corresponding author. E-mail: juan.pinzone@correo.uady.mx few floral characters, such as the exsertion of stamens and shape of the sepals. Gardner (1986) challenged the classification of Tillandsia subgenus Tillandsia of Smith & Downs through a detailed study of floral characters, but that classification remained provisional and has no molecular phylogenetic basis. For this reason, the phylogenetics of the genus Tillandsia need to be elucidated. To deal with such a large and diverse group, two strategies can be followed: (1) a top–down approach, sampling as many species as possible, trying to represent all the © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 1 2
ET AL. J. P. PINZON morphological and ecological variation and the geographical distribution; or (2) a bottom–up approach, sampling all species in a species complex, to deal with a manageable number of taxa with a reduced but representative sampling of the outgroup. Here, we have chosen to follow the second approach, to answer more fine-scale evolutionary and taxonomic questions than those that could be made for the entire genus. Because Tillandsia utriculata L. is the type species of the genus, it is important to correctly assess the relationships and species limits in the species complex, which is composed of morphologically similar taxa, which are difficult to diagnose. The T. utriculata complex, as defined by Ram
ırez, Carnevali & Chi (2004) (s.s.), is represented by a group of species that share vegetative and floral characteristics, including triangular leaves, spicate or paniculate inflorescences with sessile flowers, a flexuous rachis, and exserted stamens and style. The names initially included by Ram
ırez et al. (2004) and Ram
ırez & Carnevali (2007a,b), in addition to T. utriculata, are T. aesii I.Ram
ırez & Carnevali, T. calcicola L.B.Sm. & Proctor, T. cucaensis Wittm., T. dasyliriifolia Baker, T. geniculata E.Morren ex Baker, T. limbata Schltdl., T. makoyana Baker, T. pinicola I.Ram
ırez & Carnevali, T. pringlei S.Watson, T. pulvinata E.Morren ex Baker, T. simplexa Matuda, T. swartzii Baker and T. tehuacana I.Ram
ırez & Carnevali. Although not mentioned in the cited references, these species also feature remote floral bracts, which makes the rachis visible, and the flowers are appressed to it. In addition to the aforementioned taxa, there are two groups of species that, despite sharing the characteristics of the T. utriculata complex s.s., were omitted by Ram
ırez & Carnevali (2007a,b) and Ram
ırez et al. (2004). The first group includes the lithophytic Mexican species T. albida Mez & Purpus, T. fresnilloensis W.Weber & Ehlers, T. karwinskyana Schult. & Schult.f. and T. socialis L.B.Sm. The second group comprises T. extensa Mez, T. hildae Rauh, T. mima L.B.Sm., T. propagulifera Rauh and T. secunda Kunth, which are also lithophytic, but are distributed in north-western South America and are generally larger than the Mexican species. Furthermore, after the publication of the study by Ram
ırez et al. (2004), subsequent studies described additional species that possess characters similar to those in the complex, namely T. comitanensis Ehlers, T. huamelulaensis Ehlers, T. nicolasensis Ehlers (Ehlers, 2006a,b,c), T. elusiva Pinz
on, I.Ram
ırez & Carnevali and T. izabalensis Pinz
on, I.Ram
ırez & Carnevali (Pinz
on, Ram
ırez-Morillo & Carnevali Fern
andez-Concha, 2011, 2012). All of these species (T. utriculata complex s.l.) possess characteristics that agree with Gardner’s (1986) Group II of the classification of Tillandsia subgenus Tillandsia. That is, they present stamens of unequal length based on cross-sections, erect or recurved petal apices and flowers with an open corolla throat. The only exception is T. swartzii, which is a synonym of Vriesea swartzii (Baker) Mez, and is characterized by the presence of appendages at the base of the petals and secund spreading flowers (Mez, 1935). It is important to note that the T. utriculata complex s.l. is not exactly equivalent to Group II of Gardner (1986), because not all of the species in Group II agree with the characteristics of the T. utriculata complex s.l. Specifically, T. andreana E.Morren ex Andr
e and T. funckiana Baker have solitary flowers per rosette, T. argentea Griseb. and T. fuchsii W.Till have filiform leaves and spreading flowers and T. flagellata L.B.Sm. (= T. lehmannii Rauh), T. kegeliana Mez and T. paraensis Mez have imbricate floral bracts and the rachis is not regularly flexuous. Although there are a number of molecular phylogenetic studies that have included Tillandsia spp., these were aimed at either addressing taxonomic problems at the family or subfamily levels (Ranker et al., 1990; Terry, Brown & Olmstead, 1997a,b; Horres et al., 2000; Crayn, Winter & Smith, 2004; Givnish et al., 2004, 2011; Barfuss et al., 2005) or focused on understanding the evolution of different species complexes in Tillandsia (Granados, 2008; Chew, De Luna & Gonz
alez, 2010). Therefore, such studies include a limited sampling of species belonging to the T. utriculata complex. Barfuss et al. (2005) provided the most exhaustive sampling of Tillandsia conducted to date, including 58 species, but only included one species (T. utriculata) from the T. utriculata complex. One of the goals of this study is to assess the phylogenetic relationships of the species that share characteristics of the T. utriculata complex. The questions we seek to address are as follows. Do species of the T. utriculata complex constitute a monophyletic group? If so, are the Mexican lithophytic species and the South American taxa related to T. mima part of the T. utriculata complex? Are the South American species with similar characteristics part of this group? Based on these analyses, we also provide a test of monophyly of Group II proposed by Gardner (1986). By assessing the species composition of the T. utriculata complex, of Group II, and the phylogenetic relationships among their constituent species, we are also able to propose probable scenarios of evolution, biogeography and diversification of this group. In addition, the inclusion of specimens from different populations for some of the species analysed (e.g. Tillandsia karwinskyana, T. pringlei and T. utriculata © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA or T. makoyana and T. tehuacana) will contribute to resolve taxonomic issues that have remained diffuse and have hindered the delimitation of some of these taxa. METHODS TAXON 3 onym of T. makoyana, T. geniculata which is a synonym of T. limbata, T. aesii which is a synonym of T. cucaensis (Pinz
on et al., 2012), T. pulvinata which is a synonym of T. dasyliriifolia and T. lehmannii which is a synonym of T. flagellata. Tillandsia swartzii was also excluded, as we had no access to the original material and it belongs to Vriesea (Smith & Downs, 1977). SELECTION To determine the phylogenetic position of the T. utriculata complex s.l. in the genus, we conducted independent phylogenetic analyses using the matK gene and a section of the 30 end of the trnK intron (matK-trnK) and the rps16 intron (rps16), and combined analyses of the two regions (hereafter referred to as ‘broad analyses’). We selected these markers as they have been used for the largest number of Tillandsia spp. available from public databases. For the analyses of matK-trnK, we included 175 accessions which represented 122 Tillandsia spp. (169 accessions), two species of Racinaea M.A.Spencer & L.B.Sm. (two accessions), one species of Vriesea Lindl. (three accessions) and Catopsis nutans (Sw.) Griseb. as a functional outgroup (one accession), as the results reported by Barfuss et al. (2005) indicate that Catopsis Griseb. and Glomeropitcairnia Mez form the sister group of the rest of Tillandsioideae. For the analyses using rps16, we included 168 accessions representing 113 Tillandsia spp. (164 accessions), one Racinaea sp. (one accession), one Vriesea sp. (two accessions) and C. nutans (one accession). The ‘broad analyses’ combining the two regions (i.e. matK-trnK and rps16) were performed with 108 Tillandsia spp. (145 accessions), one Racinaea sp. (one accession), one Vriesea sp. (one accession) and C. nutans (one accession). Sequences were generated during this study or obtained from GenBank based on studies by Crayn et al. (2004), Barfuss et al. (2005), Granados (2008), De Castro et al. (2009) and Rex et al. (2009) (accession numbers: Appendix 1). A second set of analyses was also performed, hereafter called ‘restricted analyses’, with more characters, but fewer taxa. Here, we included all the species that exhibited morphological characteristics present in the T. utriculata complex s.l., most of the species belonging to Group II (Gardner, 1986) and belonging to the clades that were more closely related to species of the T. utriculata complex based on results from the broad analyses. For the ‘restricted analyses’, we used matK-trnK, rps16 and the rpl32-trnL region combined and the external transcribed spacer (ETS) of the nuclear ribosomal (nr) DNA region alone. Of the names included in the T. utriculata complex s.l. (see Introduction) and Gardner’s Group II, we excluded the following: T. simplexa which is a syn- DNA EXTRACTION, AMPLIFICATION AND SEQUENCING For the DNA extraction, we used dried (with silica gel) or fresh plant material, obtained from the field or from exchange with the Botanical Garden of the University of Vienna (Austria) or the Marie Selby Botanical Garden (Florida, USA). The herbarium vouchers are listed in Appendix 1. DNA extraction was performed following the cetyltrimethylammonium bromide (CTAB) protocol (Doyle & Doyle, 1987). To amplify the plastid DNA regions, we used the following reagents and final concentrations: buffer (19), MgCl2 (5 mM), deoxynucleoside triphosphates (dNTPs) (200 lM), ‘forward’ and ‘reverse’ primers (0.4 lM), Taq DNA polymerase (1 U), 1 lL DNA dilution and the remaining volume of distilled H2O. For the amplification of rpl32-trnL, we modified the MgCl2 concentration to 1.5 mM) and added bovine serum albumin (BSA) (0.2 lg/lL) (Shaw et al., 2007) and, for ETS, we used MgCl2 at 2.25 mM and added dimethylsulphoxide (DMSO) at 2.7%. The pairs of primers used to amplify the matKtrnK region were matK-19F (Molvray, Kores & Chase, 2000) with trnK2R (Johnson & Soltis, 1995) and matK-19F with matK1520R (Whitten, Williams & Chase, 2000), or the pairs matK-19F/matK966rBRO and matK808fBRO/trnK2R* (Barfuss, 2012). For rps16, we used the primers rpsF and rpsR2 (Oxelman, Lid
en & Berglund, 1997). For rpl32-trnL, we used trnL(UAG) and rpl32-F (Shaw et al., 2007). For ETS, we used the primers Till2 (Chew et al., 2010) and 18S-IGS (Baldwin & Markos, 1998). The PCR conditions for matK-trnK and rps16 were the same as in Barfuss et al. (2005) and, for rpl32-trnL, we followed Shaw et al. (2007). For ETS, we used the following protocol: initial denaturation at 97 °C for 2 min, 15 cycles at 99 °C for 2 min, annealing at 68 °C for 30 s and extension at 72 °C for 1 min, followed by 20 cycles under the same conditions, but with an increment of 5 s/cycle during the extension step; subsequently, a final extension at 72 °C for 7 min and hold at 4 °C. To verify that DNA extraction and amplification were successful, we performed electrophoresis on 1% agarose gel stained with ethidium bromide. The purification was performed with a QIAquick © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 4
ET AL. J. P. PINZON (QIAGEN) purification kit following the manufacturer’s instructions. Sequencing was performed using the Sanger method with the same primers as used for the amplification on an ABI3730XL (Applied Biosystems) sequencer. SEQUENCE ASSEMBLY AND ALIGNMENT AND CODING OF INSERTIONS/DELETIONS Sequences were assembled with Geneious 4.1.4 (Biomatters Ltd., Auckland, New Zealand) and aligned using the algorithm MUSCLE 3.6 (Edgar, 2004) as implemented in the platform eBioTools (www.ebioinformatics.org), through eBioX 1.5.1 (Lagercrantz, 2008), and checked visually. Insertion/deletions (indels) were coded following the simple coding method of Simmons & Ochoterena (2000). PHYLOGENETIC ANALYSES We conducted separate analyses with the matrices of matK-trnK and rps16 and with the matrix of both regions combined (broad analysis), including indels. The restricted analyses included the combined analysis of three regions of the plastid DNA (matK-trnK, rps16 and rpl32-trnL) and indels, and also the analysis with the ETS nrDNA. All analyses were performed using the parsimony algorithm of Fitch with equal weight for all characters. The most-parsimonious trees (MPTs) were retrieved from heuristic searches with 10 000 replicates, retaining ten trees per replicate and using tree bisection–reconnection (TBR) as the branch swapping algorithm. The maximum number of trees was fixed at 100 000 (Max. trees). To assess branch support, we performed a bootstrap (BS) analysis with 10 000 iterations employing heuristic searches with ten replicates, and retained ten trees per replicate using the support levels as in Sung et al. (2007) for the interpretation of the results. Given that we obtained multiple MPTs in all the analyses, we calculated strict consensus trees. All of these analyses were performed with the program TNT 1.1 (Goloboff, Farris & Nixon, 2003). The consistency index (CI) and retention index (RI) of the MPTs were calculated with the WinClada 1.00.08 platform (Nixon, 2002). We also conducted Bayesian analyses of all the matrices explained above with MrBayes 3.1 (Ronquist & Huelsenbeck, 2003). The nucleotide substitution model for each DNA partition was selected under the Akaike information criterion (AIC) with three substitution schemes, in program jModelTest 0.1.1 (Posada, 2008). For all analyses, data partitions were set corresponding to each DNA region and indels. For the broad analysis, the nucleotide substitution model used for partitions of matK and rps16 was GTR + I + Γ and the model for trnK was GTR + Γ. For the restricted analysis with the three plastid DNA regions combined and the indels, we used the models GTR + I + Γ for the matK and rps16 partitions, GTR + Γ for the trnK partition and HKY + Γ for the rpl32-trnL partition. Finally, the model used for ETS was HKY + I + Γ. In all cases, the partitions of indels were treated under the binary model, using type of data as ‘restriction’ and establishing the coding option as ‘variable’. For all the analyses, we unlinked the estimation of the parameters of each partition (except for topology and branch length), and the global rate was allowed to vary independently for each partition. The broad analysis consisted of three simultaneous but independent runs, each consisting of 5 000 000 generations produced by the Metropolis-coupled Markov chain Monte Carlo (MCMCMC), with a sampling every 100 generations using one cold chain and four hot chains with a temperature of 0.17, whereas, for the remaining parameters, we used the default values given by the program MrBayes 3.1. The restricted analyses of the four regions of the plastid DNA plus indels and of the ETS region were performed using the same parameters specified in the previous analyses, but in this case with 10 000 000. Convergence of parameters between runs was considered as reached when the ‘average standard deviation of split frequencies’ was < 0.01, as recommended by Ronquist, Huelsenbeck & Teslenko (2011), and also by visual examination of the plot of generation vs. log likelihood, considering the convergence achieved when the dots that represented different runs were mixed. For the estimation of parameters and posterior probabilities (PPs), in all cases we discarded 25% of the initial generations. The clades of interest were labelled with letters in the tree that resulted from the broad analysis with matK-trnK, rsp16 and indels. For the trees produced by the other analyses, we repeated letters for clades that shared species and were congruent with the clades from the first analysis (although tree internal topologies and numbers of species were not necessarily identical between these analyses). To assess the suitability of analysing the plastid DNA and nrDNA (ETS) data together, we performed the incongruence length difference (ILD) test (Farris et al., 1994). The infrageneric allocation of Tillandsia spp. to the trees shown was performed following the circumscription of Smith & Downs (1977), with the exception of T. tortilis Klotzsch ex Baker and T. lepidosepala L.B.Sm. Although the last two species were considered as part of T. subgenus Tillandsia by Smith & Downs (1977), subsequent studies found that they belong to T. subgenus Allardtia (A.Dietrich) Baker (Gardner, 1982; Ehlers (2009). © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA CHARACTER EVOLUTION AND BIOGEOGRAPHICAL ANALYSIS To explain the evolution of the studied group, we conducted a parsimony-based reconstruction with unordered character states for several morphological and ecological characters with Mesquite 2.75 (Maddison & Maddison, 2011), using the strict consensus tree generated from the parsimony analysis with the three plastid DNA regions and indels. We reconstructed five groups of morphological characters: (1) the T. utriculata complex syndrome, i.e. the combination of characters that define the complex, such as the inflorescence in a spike or panicle, a flexuous rachis, flowers appressed to the rachis, remote floral bracts and exserted stamens and style; (2) the Group II syndrome, i.e. the combination of open corolla throat, filaments in series of two lengths, round and of the same width throughout their entire length; (3) the presence or absence of vegetative reproduction and the position of propagules when present: monocarpic genet, axillary propagules, basal propagules, caespitose growth and propagules originating from the inflorescence; (4) inflorescence colour (including the peduncle), the main axis of a compound inflorescence, the rachis and the floral bracts; and (5) petal colour. The ecological characters that have been reconstructed are the type of substrate in which the species grows as an epiphyte, lithophyte or terrestrial. We also performed an analysis for the reconstruction of the ancestral distribution areas with maximum parsimony in the same way as for the characters above and with the Bayesian binary MCMC method (BBM) (Ronquist & Huelsenbeck, 2003), as implemented in RASP (Yu et al., 2015) using the default configuration, on one of the 63 MPTs obtained from the restricted analysis of the plastid DNA markers. Both analyses were based on the phytogeographical regions proposed by Gentry (1982): Mexico and Central America; West Indies; northern Venezuela and Colombia; northern Andes; southern Andes; and the Amazon Basin. The region of Mexico and Central America was subdivided into three areas, because most of the studied species are distributed in this region and the use of a finer geographical subdivision was helpful to describe the biogeographical patterns appropriately. This subdivision consisted of: (1) Gulf of Mexico and Caribbean coast; (2) Pacific Ocean coast and mountainous region; and (3) the Mexican Plateau. The subdivision of this phytogeographical region along an east–west (1 and 2) axis, taking, as the division line, the Sierra Madre Oriental and the mountains of northern Oaxaca and Chiapas, was based on the cladistic biogeographical study by Escalante et al. (2007), which recognized biogeographical affinities between the combined pro- 5 vinces of the Gulf of Mexico and the Yucatan Peninsula and the combined Pacific coast and the mountains of Oaxaca and Chiapas provinces. The biogeographical province of eastern Central America was included in the Gulf of Mexico coast and the Caribbean. The mountainous zone of Central America (Guatemala, Honduras and Nicaragua) was grouped with the Pacific coast, as both are found in the same province as the mountains of Chiapas (Morrone, 2001). The Mexican Plateau zone was considered as a third subdivision because it has been classified as part of the Nearctic region (Morrone, 2001, 2005) and is limited to the east by the Sierra Madre Oriental, to the west by the Sierra Madre Occidental and to the south by the Trans-Mexican Volcanic Belt. In addition, we included the peninsula of Florida as part of the West Indies region. The areas were assigned to the terminals in a presence/ absence scheme, in accordance with the observed distribution of specimens observed in the field, registered in herbaria CICY, WU, MEXU and XAL, or cited in Smith & Downs (1977). When several accessions of the same species were included, the distribution of the whole species was assigned to each accession. RESULTS OF DNA CHARACTERIZATION REGIONS Table 1 shows the characteristics of the DNA regions used in the parsimony analyses, such as size and percentage, and number of variable and potentially informative sites. The most variable plastid DNA region with the greatest percentage of potentially parsimony-informative characters was trnK (partial) in both the broad and restricted analyses, followed by matK. The rps16 intron was the least informative region. Although trnK was the most variable and informative region in terms of percentage of informative sites, matK provided a greater absolute number of variable and informative characters. For the restricted analysis of plastid DNA regions, the most variable and informative region was again trnK, followed by rpl32-trnL, matK and, lastly, rps16. The level of variability in ETS was more than double that observed for trnK, and the percentage of potentially parsimony-informative characters was almost four times greater relative to this region. PHYLOGENETIC RELATIONSHIPS Broad analyses (Fig. 1) The parsimony analysis with the matK-trnK region yielded 54 MPTs with CI = 0.73 and RI = 0.93, whereas that of the rps16 region and indels resulted © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 6
ET AL. J. P. PINZON Table 1. Size, variability and level of information for the parsimony of the DNA markers used for the phylogenetic analyses Marker Aligned size (bp) Variable sites (number, %) Parsimonyinformative characters (number, %) matK matK matK 1438 1438 1438 222, 15.4% 205, 14.3% 146, 10.2% 119, 8.3% 103, 7.2% 53, 3.7% trnK intron (partial) trnK intron (partial) trnK intron (partial) 137 137 137 38, 27.7% 36, 26.3% 30, 21.9% 20, 14.6% 19, 13.9% 9, 6.6% rps16 intron rps16 intron rps16 intron 873 873 858 105, 12.0% 105, 12.0% 82, 9.6% 47, 5.4% 44, 5.0% 25, 2.9% 1003 135, 13.5% 52, 5.2% rpl32-trnL intergenic spacer External transcribed spacer (partial) (ETS) 440 (423) 255, 58.0% (229, 54.1%) in 13 360 MPTs with CI = 0.73 and RI = 0.92. In addition, the parsimony analysis of the combined matrices generated 2196 MPTs with CI = 0.73 and RI = 0.92. The strict consensus tree based on these trees and the majority rule consensus tree from the Bayesian analysis (Fig. 1) did not exhibit incongruence, although the latter had a higher resolution. The individual analyses of matK-trnK and rps16 (not shown) and the combined analysis yielded a clade composed mainly of taxa of Tillandsia subgenus Tillandsia (Fig 1, clade A) (BS = 57, PP = 1), which also included the T. utriculata complex s.l. However, some species inserted in clade A belong to T. subgenus Allardtia (e.g. T. guatemalensis L.B.Sm.) or to T. subgenus Pseudalcantarea Mez [e.g. T. paniculata (L.) L]. Clade A consists of a trichotomy (clades B, C and D). Clade B received high support (BS = 93, PP = 1), whereas clade C had weak support (BS = 73, PP = 1). Within these two clades, some species of the Tillandsia utriculata complex s.l. were found, such as T. secunda, T. propagulifera and T. mima (clade B) and T. hildae (clade C). In clade B, we also found T. adpressiflora Mez and T. marnier-lapostollei Rauh (Allardtia), whereas, for clade C, we had Vriesea malzinei E.Morren and T. paniculata (subgenus Pseudalcantarea). Clade D (Mexican clade) was also strongly supported (BS = 98, PP = 1) and included a larger number of species (44). The species of the T. utriculata 137, 31.1% (108, 25.5%) Matrix matKtrnK matKtrnK + rps16 + indels matKtrnK + rps16 + rpl32trnK + indels matKtrnK matKtrnK + rps16 + indels matKtrnK + rps16 + rpl32trnK + indels rps16 + indels matKtrnK + rps16 + indels matKtrnK + rps16 + rpl32trnK + indels matKtrnK + rps16 + rpl32trnK + indels ETS Number of species/ specimens 126/175 111/148 62/88 126/175 111/148 62/88 116/168 111/148 62/88 62/88 72/100 complex s.s. were placed here and distributed mainly in two clades: clade E, which we named the T. utriculata clade, received moderate to high support (BS = 80, PP = 1), and clade F, which we named the T. limbata clade, also received moderate to high support (BS = 88, PP = 1). Tillandsia socialis also exhibits a morphology similar to species of the T. utriculata complex, but its relationship with the clades of the complex remains unclear, as it is part of a polytomy at the base of the clade containing clades E, F, G and H. Tillandsia tehuacana and T. nicolasensis were grouped with the T. limbata clade in the majority rule consensus tree from the Bayesian analysis, albeit without statistical support. This relationship was not observed in the strict consensus tree from the parsimony analysis (Fig. 4). The internal relationships of the T. utriculata clade showed a dichotomy formed by the Mexican Plateau clade (T. albida, T. fresnilloensis and T. karwinskyana) (BS = 62; PP = 1) and the Gulf-Antillean clade (BS = 74; PP = 1), comprising T. calcicola, T. elusiva, T. pringlei and T. utriculata. In the T. limbata clade, two lineages can be observed, one called here the western Mesoamerican clade (T. comitanensis, T. cucaensis, T. huamelulaensis, T. pinicola and T. makoyana) and the other named here the eastern Mesoamerican clade (T. izabalensis, T. limbata, T. may-patii and T. dasyliriifolia). In clade D, another lineage can be observed, which is composed of species from subgenus Allardtia © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA 7 Figure 1. Majority rule consensus phylogram resulting from the Bayesian inference analysis of species of the Tillandsia utriculata complex s.l. and the outgroup, using the plastid DNA regions matK, trnK, rps16 and indels for the latter (broad analysis). Above and below each branch, we indicate the bootstrap and posterior probability values, respectively. For a description of the clades labelled with letters, see text. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 8
ET AL. J. P. PINZON (clade G) and a clade in which none of the species exhibits the morphology of the T. utriculata complex (clade H). Clade G received weak support (BS < 50, PP = 1), whereas clade H had strong support (BS = 93, PP = 1). Restricted analyses with plastid DNA (Fig. 2) Parsimony analyses of matK-trnK, rps16 and indels, and rpl32-trnL and indels, yielded 63 MPTs with CI = 0.76 and RI = 0.88. Clades A–H (from the broad analyses) were also recovered in strict consensus to the MPTs and the majority rule consensus tree of the Bayesian analysis. There were some incongruences between the topologies of these two trees, but these were only present outside clade A. This clade also received weaker support (BS = 49, PP = 1) in comparison with the same clade in the broad analyses. In contrast, clades B and C received improved support (BS = 97 and 88, respectively) and the latter also showed better resolution. Clade D also received improved support (BS = 99, PP = 1). In clade D, clades G (subgenus Allardtia), H and F (T. limbata clade) received stronger support with BS values of 71, 99 and 95, respectively (with PP = 0.98 and 1). Tillandsia nicolasensis and T. tehuacana were not found in sister group position to clade F, whereas clade E (the T. utriculata clade) showed a slightly lower support (BS = 78, PP = 1). The internal relationships of clades E and F did not change. Based on the Bayesian analysis, T. fuchsii and T. socialis were grouped together in a clade (PP = 0.91), whereas, for the parsimony analysis, their relationships in clade D were not resolved. Restricted analyses with ETS (Fig. 3) The parsimony analysis produced 19 169 MPTs with CI = 0.60 and RI = 0.77. The strict consensus of this latter analysis (not shown) and the majority rule consensus tree from the Bayesian analysis (Fig. 3) exhibited a few incongruences in the earlier divergent clades, but none of these was well supported (BS < 50, PP < 0.85). For clades A–H resulting from the plastid DNA analysis, only clade G was recovered; all the rest exhibited incongruences. With respect to the phylogenetic relationships of the T. utriculata complex s.s., only two clades were recovered: one with weak support (BS = 73, PP = 1), which included T. calcicola, T. elusiva and T. utriculata, and another with moderate support based on the Bayesian analysis (PP = 0.98), which included species of the T. limbata clade (according to the plastid DNA data) and all specimens of T. pringlei. Tillandsia fuchsii and T. socialis formed a group with stronger support than in the analyses based on plastid DNA regions (BS = 87, PP = 1). Test of incongruence The ILD test showed that the matrices of plastid DNA and ETS are significantly incongruent (P = 0.0909). CHARACTER EVOLUTION AND BIOGEOGRAPHICAL ANALYSES (FIG. 4) Tillandsia utriculata syndrome The reconstruction of ancestral states indicated that this set of characters coincided together in clade A at least three times independently. In clade B, they were found together at least once, although it is not clear whether there are two reversions or three gains. All species of this clade have in common many features of the T. utriculata complex, with the exception of T. adpressiflora and T. marnier-lapostollei which have included stamens and T. spiraliflora which has polystichous flowers. In the core Mexican clade (excluding the clade formed by T. punctulata, T. gymnobotrya and T. prodigiosa), these characters are again found together. Most of the species have stamens and style exserted, but clade H has lost the Group II floral morphology and changed to Group I floral morphology, whereas, in clade G, there is a reversion to included stamens. Floral morphology The Group II-type floral morphology presumably emerged at least four times: once in clade B, with one reversion; one to three times in clade C; and one to four times in clade D. The reconstruction placed this morphology as ancestral for the clade formed by clades E, F, G and H and T. fuchsii, T. tehuacana and T. nicolasensis. The evolution of violet petal colour is ambiguous for clade A, but ancestral for clades B and D. The ancestral state of clade E is whitish, whereas the ancestral state for clade F is ambiguous. For one subclade of clade F, composed of T. izabalensis, T. limbata, T. dasyliriifolia, T. comitanensis and T. may-patii, the ancestral petal colour was whitish. Red petal colour evolved independently twice, once in clade C and another in clade D, with T. nicolasensis. Vegetative reproduction The ancestral form of vegetative reproduction in clade A was the production of axillary propagules. The change to monocarpic plants presumably occurred independently at least seven times. The ancestral state of clade E is ambiguous, although monocarpy evolved at least once in this clade (in T. utriculata and T. elusiva). In this clade, caespitose growth emerged at least once, in T. pringlei and © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA 9 Figure 2. Majority rule consensus phylogram from the Bayesian inference analysis of species of the Tillandsia utriculata complex s.l. and the outgroup, using the plastid DNA regions matK, trnK, rps16, rpl32-trnL and indels from the last two (restricted analysis). Above and below each branch, we indicate the bootstrap and posterior probability values, respectively. For a description of the clades labelled with letters, see text. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 10
ET AL. J. P. PINZON Figure 3. Majority rule consensus phylogram from the Bayesian inference analysis of species of the Tillandsia utriculata complex s.l. and the outgroup, using the external transcribed spacer (ETS) region from the nuclear ribosomal DNA (restricted analysis). Above and below each branch, we indicate the bootstrap and posterior probability values, respectively. Green: species of the T. utriculata clade complex according to results using plastid DNA; dark green, Gulf-Antillean Clade; light green, Mexican Plateau Clade; salmon pink, T. limbata clade. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA in the clade formed by T. albida, T. karwinskyana and T. fresnilloensis. Propagation via basal propagules, but without caespitose growth, arose at least four times in clade A, once in clade B (T. mima), once in clade C (T. hildae) and at least five times in clade D. In the T. limbata clade (clade F), monocarpy evolved at least three times, in T. comitanensis, T. aff. comitanensis, T. izabalensis and T. huamelulaensis. In contrast, the production of propagules in the inflorescence arose independently at least three times, once in clade B, once in clade C (T. flexuosa) and once in clade F (T. dasyliriifolia). Epiphytism Epiphytism is the ancestral state in clades A, C, D, F, G and H. The ancestral states of clades B and E are ambiguous. The invasion of the saxicolous habitat occurred at least six times in clade A, once in clade B, three times in clade C, at least once in clade E and at least once in clade H. The invasion of terrestrial habitats occurred only once in clade A, with T. dasyliriifolia (clade F). Biogeographical analysis The parsimony-based character state reconstruction indicated that the northern zone of the Andean Region was the ancestral distribution for clades A and B, and this is congruent with the BBM ancestral state reconstruction, which reports a probability of 86.1% and 76.0%, respectively, for the same area. In the latter clade, there was one colonization to the Amazonian region (T. adpressiflora). The ancestral distribution area of clade C is ambiguous with parsimony, but BBM analysis showed a probability of 63.9% for the West Indies as the ancestral area for this node. This clade exhibits a broad distribution and is represented in the southern and northern Andes, in northern Venezuela, in the West Indies, in the Amazonian region and in the eastern Mesoamerican Zone. Conversely, the ancestral area of distribution of clade D, according to both parsimony-based reconstruction and BBM, was the western Mesoamerican Zone (97.6%). From this point, there were two colonizations of the Mexican Plateau, one with T. tehuacana and another with clade E (the latter at 78.9%), at least two colonizations of the eastern Mesoamerican Zone (one in clade F and one in clade H), and at least one colonization of the West Indies and Florida in clade E (T. utriculata and T. calcicola). DISCUSSION GENERAL CONSIDERATIONS To date, the broad analysis presented in our study includes the largest number of Tillandsia spp. (108 11 species of > 620 species in this genus; Luther, 2012). The number of species used in this analysis represented 17.4% of the species of this genus, in contrast with the 58 species (9.3%) analysed by Barfuss et al. (2005). It is important to note that the sampling of taxa conducted in our study was designed to assess the phylogenetic position of the species of the T. utriculata complex and of species with similar morphology in Tillandsia, and to determine the phylogenetic relationships among these species. As a result of the bias in our sampling scheme, any conclusions about the results from phylogenetic analyses at the generic or subgeneric level should be taken with caution. Having said this, we proceed to make observations for some of the most important results from these analyses. Clade A, or the clade of Tillandsia subgenus Tillandsia s.s., is equivalent to clade K plus T. paniculata in the study of Barfuss et al. (2005). According to our results, this clade presumably originated in the northern Andes (at 86.1% probability; this and all further probabilities are based on BBM analysis) (Fig. 4) from an epiphytic ancestor with red inflorescences. All the species with the T. utriculata complex syndrome are found in clade A, although the ancestor of this clade presumably did not exhibit this morphology (Fig. 4). In clade A, the species with the T. utriculata complex syndrome do not form a monophyletic group; rather this combination of characters arose in at least four independent events (Fig. 4). THE TILLANDSIA UTRICULATA COMPLEX S.L. Early-diverging clades Clades B and C are composed mostly of South American species, some of which exhibit the morphology of the T. utriculata complex, but were excluded by Ram
ırez et al. (2004) based on their definition of the complex, and have not been associated with these species in any other study. In clade B (clade of T. secunda), which originated in the northern Andes, the species that share the T. utriculata syndrome are T. secunda, T. propagulifera and T. mima (Fig. 4). The rest of the species are similar, but differ in some characters. For example, T. adpressiflora and T. marnier-lapostollei differ from this syndrome only in that they have stamens that are included in the corolla (subgenus Allardtia), whereas the only character that separates T. spiraliflora is the polystichous flowers. Conversely, species of clade C (clade of T. paniculata) exhibit morphological variation and a broader geographical distribution. In this clade, we find Vriesea malzinei, which is morphologically strikingly dissimilar (mesic species, imbricate floral bracts, appendices in the petals) and a clade that includes species with red petals (T. funckiana, © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 12
ET AL. J. P. PINZON © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA 13 Figure 4. Parsimony-based reconstruction of the ancestral states of five morphological characters, one ecological character and the areas of distribution of the Tillandsia utriculata complex and the outgroup. On the branches the series of transformations are indicated by symbols: (1) the solid black rectangular tick indicates the emergence of the T. utriculata morphological syndrome, the white rectangular tick indicates its loss; (2) the solid black arrow indicates the emergence of the Group II floral morphology, the white arrow indicates its loss; (3) the ellipse represents the different methods of vegetative reproduction (or absence) indicated by colours: monocarpic genet (white), axillary propagules (green), basal propagules (blue), caespitose growth (red), propagules in the inflorescence (violet); (4) the inflorescence colour is indicated by the colour of the symbol ‘flower with stem’; (5) the petal colour is indicated by the colour of the symbol ‘corolla’; (6) the growth substrate is represented by a tree and the states indicated by colour: epiphyte (green), lithophyte (grey), terrestrial (orange). The areas of distribution are represented by the colour of the branches and the regions are indicated in the map in the top left corner. These characters were mapped on the strict consensus of 63 most-parsimonious trees (MPTs) from the parsimony analysis of the T. utriculata complex s.l. and the outgroup, using the plastid DNA regions matK, trnK, rps16, rpl32-trnL and indels from the last two (restricted analysis). The pie diagrams show the probabilities of ancestral distribution areas for selected nodes from an analysis of the Bayesian binary Markov chain Monte Carlo (MCMC) method obtained from one of the 63 MPTs from the analysis described above; colour grey indicates an uncertain area or two or more areas. For a description of the clades labelled with letters, see text. T. argentea, T. flexuosa, T. kegeliana and T. juruana) (Figs 1, 2, 4). Only T. hildae and T. paniculata exhibit the T. utriculata complex syndrome. Tillandsia paniculata is considered to be part of Tillandsia subgenus Pseudalcantarea because of its stamen and petal morphology (Smith & Downs, 1977), but Beaman & Judd (1996) concluded that this species is more closely related to subgenus Tillandsia, and this is consistent with our findings. The ancestral distribution of this clade is uncertain, but the BBM shows a slight preference for the West Indies geographical zone. Tillandsia socialis, which shows a morphology coherent with the T. utriculata complex, is found in the Mexican clade (D). However, it does not group with the T. utriculata clade, but with T. fuchsii, albeit with relatively low support. These two species share the floral morphology of Group II as a symplesiomorphy. Nonetheless, the presence of scales on the floral bracts represents a synapomorphy of this clade. Tillandsia fuchsii has lost some of the typical characteristics of the T. utriculata complex, given that the flowers of this species are spreading with respect to the rachis (not appressed) and it has undergone a reduction in size, growing as small, globose rosettes with filiform leaves. THE TILLANDSIA UTRICULATA CLADE This lineage is supported by three homoplasious morphological characters, all of which are associated with the petals, namely spathulate shape, acute apex and the loss of violet pigment (petals in these species are whitish or greenish) (Figs 4–6). As a result of the lack of resolution in clade D, the interpretation of the evolution of the ancestral characters is ambiguous in many cases. However, it is possible to infer that the ancestor of this clade already had a morphology simi- lar to the T. utriculata complex and exhibited an inflorescence with red tinges and, as mentioned previously, whitish petals. What remains uncertain, however, is whether this ancestor was epiphytic, had vegetative reproduction or was monocarpic. The distribution of this ancestor could have been restricted to the western Mesoamerican Zone, from where some species presumably invaded the eastern Mesoamerican Zone, the Antilles and Florida in one direction and the Mexican Plateau in another direction (Fig. 4). The Gulf-Antillean clade (T. utriculata, T. calcicola, T. elusiva and T. pringlei) was named because it has a distribution that is limited to the west by the Sierra Madre Oriental and occupies the Gulf of Mexico, the Continental Caribbean shore (except Panama), the Antilles, Florida and northern Venezuela. The Mexican Plateau clade (T. albida, T. karwinskyana and T. fresnilloensis) is restricted to this dry and high area. The Gulf-Antillean clade is formed by species distributed from eastern Mesoamerica and the Antilles, which share several morphological characteristics: paniculate inflorescences; a zygomorphic corolla with a lateral opening; and warty wing cells of the foliar scales, which have an entire or crenate margin (Fig. 5). The ancestral area analysis indicated that the most probable ancestral distribution area of this clade was the western Mesoamerican Zone. This ancestor presumably colonized warm montane and humid lowland areas with xeric T. calcicola in the Antilles and with mesic T. utriculata, which has the broadest distribution in this complex, as it is found from arid zones of the Yucatan Peninsula (Mexico) and the Antilles, to warm and humid zones in Mesoamerica, the Gulf of Mexico and the Continental Caribbean slopes and subtropical areas in Florida. Tillandsia elusiva occupied a zone restricted to intermediate elevations of warm and subhumid © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 14
ET AL. J. P. PINZON B A C D Figure 5. Morphology of the species of the Tillandsia utriculata clade, Gulf-Antillean Clade. A, Inflorescence of T. elusiva. B, Petal of T. utriculata (note acute apex). C, Foliar trichome of T. utriculata (note entire margin). D, Flower of T. pringlei (note the lateral opening of the corolla). climatic conditions in western Chiapas, at the limit of the Gulf of Mexico and Pacific provinces (Pinz
on et al., 2011). This species is the only one in the T. utriculata clade that has a pink inflorescence (Fig. 4). The species of the Mexican Plateau clade (T. albida, T. fresnilloensis and T. karwinskyana) share simple inflorescences and foliar scales with a dentate margin, in addition to having reddish inflorescences with whitish petals and spreading petal tips (Fig. 6). In this group, T. albida (caulescent, with reticulate ornamentation in the wing cells of foliar scales) is the earliest diverging species and subtends the clade formed by T. fresnilloensis and T. karwinskyana (acaulescent, with smooth wing cells of foliar scales). The ancestor of these three species was probably distributed in the Mexican Plateau, growing on rocks and exhibiting caespitose growth (Fig. 4). The aspect of this ancestor may have been similar to that of T. albida but acaulescent, as it presumably had conspicuous foliar sheaths and a dense indumentum, but with scales appressed to the leaf, without the tomentose aspect found in T. fresnilloensis and T. karwinskyana, which lack conspicuous foliar sheaths. This ancestor presumably was adapted to rocky environ- ments in south-eastern areas of the Mexican Plateau, in the states of Hidalgo, Quer
etaro and Guanajuato, where it gave rise to T. albida, and to more northern areas with gypsum-rich outcrops, where it gave rise to T. karwinskyana. Towards the western side of the plateau, this ancestor gave rise to T. fresnilloensis, where it adapted to volcanic rocks present in the
Sierra de Organos and related systems in the states of Zacatecas, Durango and Jalisco. THE TILLANDSIA LIMBATA CLADE The T. limbata clade (F) is composed almost exclusively of species restricted to or including Mexico in their distribution range, the only exception being T. izabalensis which is distributed from Honduras to Nicaragua (Pinz
on et al., 2012). The inclusion of T. nicolasensis and T. tehuacana in this complex is weakly supported and only evident in the broad analysis of matK-trnK and rps16 (Fig. 1). Nonetheless, all species of clade F can be differentiated from the T. utriculata clade in that the apex of the petal is rounded and they have a constriction of the corolla at the height of the ovary apex (Fig. 7). In any case, T. nicolasensis and T. tehuacana appear to have © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA 15 B A C Figure 6. Morphology of the species of the Tillandsia utriculata clade, Mexican Plateau Clade. A, Inflorescence of T. karwinskyana. B, Foliar trichome of T. albida (note the dentate margin). C, Flower of T. fresnilloensis. diverged earlier than the rest of the species belonging to the complex. The ancestor of the T. limbata clade, including T. nicolasensis and T. tehuacana, was presumably distributed in the western Mesoamerican Zone. From there, it migrated and gave rise to T. tehuacana in the high-elevation and arid eastern zone of the Trans-Mexican Volcanic Belt Province (Morrone, 2005) and adjacent areas, or in the Valle de Tehuac
an-Cuicatl
an Province in the phytogeographical scheme of Rzedowski (1978). Tillandsia nicolasensis remained in the lowlands and eventually occupied (as at present) coastal areas in southern Mexico. An autapomorphic change that appeared in this species is the red pigment in the petals, which is a unique characteristic in this complex and is rare in the Mexican clade and in Tillandsia as a whole (Smith & Downs, 1977) (Fig. 4). The western Mesoamerican clade (T. comitanensis, T. cucaensis, T. huamelulaensis, T. pinicola and T. makoyana) is unresolved, except for the position of one early-diverging specimen of T. cucaensis, which is separated from the rest of the species, which themselves form a polytomy that includes the remaining specimens of T. cucaensis. This early-diverging specimen could represent a cryptic species, but phylogeographical analyses are needed to test this hypothesis. Although the eastern Mesoamerican clade (T. izabalensis, T. limbata, T. may-patii and T. dasyliriifolia) has moderate to low support, it exhibits geographical, morphological and ecological congruence. The inclusion of T. may-patii in this clade is remarkable because this taxon does not exhibit the characteristics of the T. utriculata complex, instead having a cylindrical and compact paniculate inflorescence and imbricate bracts. Tillandsia may-patii is probably a natural hybrid for which T. dasyliriifolia is the maternal parent, as this species is the only species in this clade that is sympatric with the former (Ram
ırez & Carnevali, 1999). The ancestor of the T. limbata clade presumably colonized lowlands with a warm subhumid climate present in the Gulf of Mexico and Gulf of Honduras coming from the west, from the other side of the mountains in Mexico and Central America. The invasion of this biogeographical zone presumably occurred once in the T. limbata clade, but it is not clear whether the ancestral area of distribution was the actual eastern Mesoamerican Zone (42.68%) or a broader area, including both eastern and western Mesoamerican Zones (42.93%). This ancestor had, according to the parsimony-based reconstruction, reddish inflorescences, whitish petals, was an epiphyte and produced © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 16
ET AL. J. P. PINZON B A C D Figure 7. Morphology of the species of the Tillandsia limbata clade. A, Inflorescence of T. cucaensis. B, Petal of T. cucaensis (note the rounded apex). C, Foliar trichome of T. dasyliriifolia. D, Corolla, androecium and gynoecium of T. pinicola (note the constriction towards the base of the corolla). axillary propagules (Fig. 4). The three species of the eastern Mesoamerican clade invaded different environments: T. dasyliriifolia became established on the Yucatan Peninsula, in warm subhumid environments, and in the arid north-western zone of this region as an epiphytic or terrestrial species with the capacity to produce propagules in the inflorescence (Fig. 4); T. limbata occupies the warm and humid region of the Gulf of Mexico and the temperate subhumid mountainous zone of the Sierra Madre Oriental and northern Chiapas (this colonization to midelevations was secondary); and T. izabalensis occupies the warm humid zone of the Gulf of Honduras, of southern Belize, Guatemala, Honduras and northern Nicaragua. Based on this information, the ancestor of the Gulf-Caribbean clade could have been similar in aspect to T. izabalensis. THE ETS NRDNA The most interesting finding of this analysis is that the Mexican Plateau species of the T. utriculata clade and T. pringlei are grouped in a lineage together with species of the T. limbata clade and not with T. utriculata, T. calcicola and T. elusiva (Fig. 3). This incon- gruence could have been caused by homoplasious characters (which probably resulted in low support), but could also be indicative of reticulate evolution for which species of the Mexican Plateau clade and T. pringlei would have shared a maternal parent of the T. utriculata clade and a paternal parent of the T. limbata clade. Nonetheless, further exploration using more nuclear molecular markers is needed to reach stronger conclusions in this regard. What is clear is that T. pringlei is different from T. utriculata, as it is located outside the Gulf-Antillean clade, with up to seven different positions in the alignment. With regard to the remaining species of the GulfAntillean clade, we observed a grouping that included T. utriculata specimens from the humid zone of the Gulf and continental Caribbean slopes (Chiapas and Guatemala) and T. elusiva, which is found in subhumid and semiarid environments of the transition zone of the Gulf of Mexico Province and the Pacific Province (sensu Morrone, 2005). From these results, we did not find evidence that T. elusiva is a hybrid between T. utriculata and any species of the T. limbata clade, as suggested by Gardner (1984). The specimens of T. utriculata from the Antilles and T. calcicola formed a polytomy at the © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA base of the Gulf-Antillean clade. Because of the low resolution of the clade, it is not possible to determine whether the populations of the continental tropical area form a species that is different with respect to Antillean populations, as there were insufficient morphological differences to separate them. The only difference we detected was the inflorescence colour, which is dark purple in the continental populations from humid zones and red or green in the populations from the Antilles and the Yucatan Peninsula. INCONGRUENCE OF PLASTID DNA AND ETS PHYLOGENETIC TREES It is important to mention that results based solely on plastid DNA data, as used primarily in this study, only allow the discussion of maternal-side phylogenetic relationships. In a group with no reticulate evolution, the maternal and paternal phylogenetic history should be identical, but we have evidence that natural hybridization in Tillandsia is, if not ubiquitous, at least possible, and there are several reports of putative natural hybrids (Gardner, 1984). Furthermore, there is evidence of reticulate evolution and probably plastid capture in other genera of Bromeliaceae, e.g. in Puya Molina, in which plastid data strongly support a Chilean clade, whereas the PHYC marker splits Chilean Puya into two clades, one of them sister to the core Puya clade (Jabaily & Sytsma, 2010). A similar pattern occurs in the Deuterocohnia Mez/Abromeitiella Mez alliance, which, with nuclear DNA data, forms a monophyletic group, but, with plastid DNA, forms a paraphyletic group, with one of the clades sister to Dyckia Schult.f. and Encholirium Mart. ex Schult.f. (Sch€ utz, 2012). The author interprets this pattern as plastid capture from a Dyckia/Encholirium ancestor through hybridization and introgression of a Deuterocohnia ancestor through pollination (Sch€ utz, 2012). Although we found that the matrices with plastid DNA and ETS are not congruent, as the ILD test shows, there are not hard incongruences in the phylogenetic trees, i.e. the incongruent clades in the analysis with ETS have low support. Hence, these incongruences could be a result of plastid capture, but also could be an effect of high homoplasy in the ETS data. To assess this, it is necessary to explore other nuclear DNA markers for comparison with the phylogenetic trees obtained with plastid data. COMPARISON WITH OTHER PHYLOGENETIC STUDIES Previous phylogenetic studies included only a few species of the T. limbata and T. utriculata clades obtained here. One of the first phylogenetic studies of Bromeliaceae (Terry et al., 1997b) only included 17 T. utriculata, which was located in a clade with T. secunda and Vriesea espinosae (L.B.Sm.) Gilmartin. Excluding V. espinosae, this clade would be equivalent to clade A in our study. It seems likely that there was an error in assigning the sequence to V. espinosae, as this species is located outside clade A, with other grey-leaved xeric Vriesea spp. (Barfuss, 2012). The study of Barfuss et al. (2005) only included two accessions of T. utriculata which were located in a clade that is equivalent to clades D (clade K in Barfuss et al., 2005) and A (equivalent to clade K plus T. paniculata in Barfuss et al., 2005) in our study, and therefore results are consistent. In addition, the phylogenetic study of the T. macdougallii L.B.Sm. complex by Granados (2008) included T. utriculata and T. makoyana. These species formed a polytomy in a clade equivalent to clade D in our study. Also, the phylogenetic analysis with ETS by Chew et al. (2010) for species of T. subgenus Tillandsia with pseudobulbs did not resolve the relationships of T. utriculata, which formed a polytomy at the base of their cladogram (excluding T. deppeana Steud.); on the other hand, T. dasyliriifolia and T. makoyana were grouped in a clade with low support (BS = 62), which is consistent with clade F in our study. In the combined analysis of 5.8S, ITS2, ETS nrDNA and coded indels as a fifth state, T. makoyana was grouped with T. filifolia Schltdl. & Cham., although this relationship is unsupported. However, the coding of indels as a fifth character state is controversial and has not been used often, because it can be redundant in indelrich markers, giving excessive weight to indels during the phylogenetic reconstruction. This relationship is also not consistent with our analyses, even in the topology obtained here with the ETS nrDNA (Fig. 3). CONCLUSIONS Based on our phylogenetic analyses, we conclude that the species that share characteristics of the T. utriculata complex do not constitute a monophyletic group, and we instead suggest that this syndrome has been gained and lost repeatedly throughout the evolution of T. subgenus Tillandsia. However, all the species with this morphology are located in a clade dominated by species of T. subgenus Tillandsia. The South American species with this morphology are found in two lineages in a trichotomy with the Mexican clade in T. subgenus Tillandsia and are not closely related to T. utriculata. The species originally proposed as part of the complex (T. utriculata s.l.) are found in a predominantly Mexican clade, forming two lineages: the T. utriculata clade and the T. limbata clade. Based on the available information, it is not possible to determine © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 18
ET AL. J. P. PINZON whether these two complexes represent a monophyletic group. The origin of both lineages appears to be western and central Mesoamerica and the T. utriculata complex is symplesiomorphic. In this zone, there were several colonizations of different habitats. The Mexican Plateau clade underwent a diversification in this area and gave rise to lithophytic species with caespitose growth and simple inflorescences; the Gulf-Antillean clade presumably migrated to the Gulf of Mexico region and Antilles, whereas monocarpy arose in T. utriculata and T. elusiva. Conversely, the western Mesoamerican clade radiated in its ancestral distribution area, where it originally occupied an epiphytic niche and was distributed in tropical and subtropical zones, and, lastly, the eastern Mesoamerican clade colonized lower, warm and humid or subhumid areas in the eastern Mesoamerican zone, adapting to mesic conditions. The analysis with ETS resulted in low resolution, but allowed us to distinguish T. utriculata and T. pringlei, which were previously considered to be subspecies of the same species. ACKNOWLEDGEMENTS The first author acknowledges Consejo Nacional de Ciencia y Tecnologı́a (National Counsil on Science and Technology) for providing a scholarship during his doctoral studies at the CICY and postdoctoral stay at the University of Vienna. We are especially indebted to Peter Tristram for organizing funding that partially covered the costs of this project and all the organizations that provided funds: the Bromeliad Societies of Australia, Cairns, Hunter, Illawarra, New South Wales, Queensland, and South Australia, and the German Bromeliad Society. We also thank the following persons who have made donations: Greg Aizlewood, Peter Bak, Margaret and Derek Butcher, Brenton Cadd, Ray Clark, Len Colgan, Nanette Collingwood, Terry Davis, Joe DeGabriel, Laurie Dorfer, Renate Ehlers, Barry Genn, Brad Gillis, Ian Hook, Paul Isley III, Maurice Kellett, Chris Larson, Justin Lee, Ross Little, Kerry McBurnie, Steve Morgan, G. and J. Newell, George Nieuwenhoven, John Olsen, Grant Paterson, Bob Reilly, Dave Sheumack, Mark Supple, Peter Tristram, Paul Turvey, Shane Weston and Dawn Williams. We also thank Gregorio Am
ılcar Castillo (CICY), Francisco Chi May (CICY), Rodrigo Duno (CICY), Gustavo A. Romero (AMES) and Jos
e Luis Tapia (CICY) for helping with the field work, and Lilia Can and Silvia Hern
andez (CICY) for herbarium specimen management. We acknowledge Bruce Holst (SEL), Helmut and Lieselotte Hromadnick, and Lyidia and Gerahard K€ores, who allowed the senior author access to and to take samples from the living collection of the Marie Selby Botanical Gardens and their private collections. We thank Carolina Granados (UNAM) for providing sequences of Tillandsia, Luis Abdala Roberts for the translation of the manuscript and Mario Mart
ınez Cordero for helping with the editing of the figures. Finally, we acknowledge the two anonymous reviewers who helped to improve this article significantly. REFERENCES Baldwin BG, Markos S. 1998. Phylogenetic utility of the external transcribed spacers (ETS) of 18S–26S rDNA: congruence of ETS and ITS trees of Calycadenia (Compositae). Molecular Phylogenetics and Evolution 10: 449–463. Barfuss MH. 2012. Molecular studies in Bromeliaceae: implications of plastid and nuclear DNA markers for phylogeny, biogeography, and character evolution with emphasis on a new classification of Tillandsioideae. DPhil Thesis, University of Vienna. Barfuss MH, Samuel R, Till W, Stuessy TF. 2005. Phylogenetic relationships in subfamily Tillandsioideae (Bromeliaceae) based on DNA sequence data from seven plastid regions. American Journal of Botany 92: 337–351. Beaman RS, Judd WS. 1996. Systematics of Tillandsia subgenus Pseudalcantarea. Brittonia 48: 1–19. Benzing DH. 2000. Bromeliaceae: profile of an adaptive radiation. Cambridge: Cambridge University Press. Chew T, De Luna E, Gonz
alez D. 2010. Phylogenetic relationships of the pseudobulbous Tillandsia species (Bromeliaceae) inferred from cladistic analyses of ITS2, 5.8 ribosomal RNA gene, and ETS sequences. Systematic Botany 35: 86–95. Crayn DM, Winter K, Smith JAC. 2004. Multiple origins of the crassulacean acid metabolism and the epiphytic habit in the Neotropical family Bromeliaceae. Proceedings of the National Academy of Sciences 101: 3703–3708. De Castro O, Cennamo P, V
azquez-Torres M, De Luca P. 2009. Molecular studies about two rare species of the genus Tillandsia L. (T. califanii Rauh and T. tomasellii De Luca, Sabato, Balduzzi). Journal of the Bromeliad Society 59: 206–218. Doyle JJ, Doyle JL. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin, Botanical Society of America 19: 13–15. Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research 32: 1792–1797. Ehlers R. 2006a. Tillandsia huamenulaensis: a new species from the State of Oaxaca, Mexico. Journal of the Bromeliad Society 56: 56–59. Ehlers R. 2006b. Tillandsia nicolasensis: a new species from coastal Jalisco, Mexico. Journal of the Bromeliad Society 56: 70–72. Ehlers R. 2006c. Tillandsia comitanensis: a new highland species from Chiapas, Mexico. Journal of the Bromeliad Society 56: 116–119. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 PHYLOGENY IN TILLANDSIA UTRICULATA Ehlers R. 2009. The green-blooming, small, grey Tillandsias from Mexico. Die Bromelie, Sonderheft, Vol. 6. Escalante T, Rodr
ıguez G, Cao N, Ebach MC, Morrone JJ. 2007. Cladistic biogeographic analysis suggests an early Caribbean diversification in Mexico. Naturwissenschaften 94: 561–656. Farris JS, K€ allersj€ o M, Kluge AG, Bult C. 1994. Testing significance of incongruence. Cladistics 10: 315–319. Gardner CS. 1982. Tillandsia ehrenbergii (K. Koch) Klotzsch ex Mez – an ancient case of mistaken identity. Journal of the Bromeliad Society 32: 17. Gardner CS. 1984. Natural hybridization in Tillandsia subgenus Tillandsia. Selbyana 7: 380–393. Gardner CS. 1986. Preliminary classification of Tillandsia based on floral characters. Selbyana 9: 130–146. Gentry AH. 1982. Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny? Annals of the Missouri Botanical Garden 69: 557–593. Givnish TJ, Barfuss MHJ, Van Ee B, Riina R, Schulte K, Horres R, Gonsiska PA, Jabaily RS, Crayn DM, Smith JAC, Winter K, Brown GK, Evans TM, Holst BK, Luther H, Till W, Zizka G, Berry PE, Sytsma KJ. 2011. Phylogeny, adaptive radiation, and historical biogeography in Bromeliaceae: insights from an eight-locus plastid phylogeny. American Journal of Botany 98: 872–895. Givnish TJ, Millam KC, Evans TM, Hall JC, Pires JC, Berry PE, Sytsma KJ. 2004. Ancient vicariance or recent long-distance dispersal? Inferences about phylogeny and South American–African disjunctions in Rapataceae and Bromeliaceae based on ndhF sequence data. International Journal of Plant Sciences 165(4 Suppl.): S35–S54. Goloboff P, Farris J, Nixon K. 2003. T.N.T.: tree analysis using new technology. Program and documentation, available from the authors, and at www.zmuc.dk/public/phylogeny Granados C. 2008. Sistem
atica del complejo Tillandsia macdougallii. Master’s Thesis, Universidad Nacional Aut
onoma de M
exico. Horres R, Zizka G, Kahl G, Weising K. 2000. Molecular phylogenetics of Bromeliaceae: evidence from trnL (UAA) intron sequences of the chloroplast genome. Plant Biology 2: 305–315. Jabaily RS, Sytsma KJ. 2010. Phylogenetics of Puya (Bromeliaceae): Placement, major lineages, and evolution of Chilean species. American Journal of Botany 97: 337–356. Johnson LA, Soltis DE. 1995. Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae) using matK sequences. Annals of the Missouri Botanical Gardens 82: 149–175. Lagercrantz E. 2008. eBioX. Version 1.5.1. Available at: http://www.ebioinformatics.org Luther H. 2012. An alphabetical list of bromeliad binomials, 13th edn. Sarasota, FL: Marie Selby Botanical Gardens & Bromeliad Society International. Maddison WP, Maddison DR. 2011. Mesquite: a modular system for evolutionary analysis. v 2.75. Available at: http:// mesquiteproject.org 19 Mez HC. 1935. Bromeliaceae. In: Engler HGA, ed. Das Pflanzenreich. Regni Vegetabilis Conspectus, Leipzig: Verlag von Wilhelm Engelmann (Grossdruckerei Paul D€ unnhaupt, K€othen (Anhalt)). Molvray M, Kores PJ, Chase M. 2000. Polyphyly of mycoheterotrophic orchids and functional influences on floral and molecular characters. In: Wilson KL, Morrison DA, eds. Monocots: systematics and evolution. Collingwood, Vic.: CSIRO Publishing, 441–448. Morrone JJ. 2001. Biogeograf
ıa de Am
erica Latina y el Caribe. Zaragoza: Manuales y Tesis SEA 3. Morrone JJ. 2005. Hacia una s
ıntesis biogeogr
afica de M
exico. Revista Mexicana de Biodiversidad 76: 207–252. Nixon KC. 2002. WinClada ver. 1.00.08. Available at: http:// www.cladistics.com/wincDownload.htm Oxelman B, Lid
en M, Berglund D. 1997. Chloroplast rps16 intron phylogeny of the tribe Sileneae (Caryophyllaceae). Plant Systematics and Evolution 206: 393–410. Pinz
on JP, Ram
ırez-Morillo IM, Carnevali Fern
andezConcha G. 2011. Morphometric analyses within the Tillandsia utriculata L. complex (Bromeliaceae) allow for the recognition of a new species, with notes on its phylogenetic position. Journal of the Torrey Botanical Society 138: 353–365. Pinz
on JP, Ram
ırez-Morillo IM, Carnevali Fern
andezConcha G. 2012. The re-establishment of Tillandsia cucaensis (Bromeliaceae), a good species formerly confused with a new species from the Gulf of Honduras. Phytotaxa 61: 1–16. Posada D. 2008. jModel Test: phylogenetic model averaging. Molecular Biology and Evolution 25: 1253–1256. Ram
ırez I, Carnevali G. 1999. New species of Tillandsia, some new records, and a checklist of the Bromeliaceae from the Yucatan Peninsula. Harvard Papers in Botany 4: 185– 194. Ram
ırez I, Carnevali G. 2007a. Two new species in the Tillandsia utriculata complex (Bromeliaceae) from Mexico. Novon 17: 72–78. Ram
ırez I, Carnevali G. 2007b. A new species in the Tillandsia utriculata complex (Bromeliaceae) from Mexico. Novon 17: 383–385. Ram
ırez I, Carnevali G, Chi F. 2004. Portraits of Bromeliaceae from the Mexican Yucatan Peninsula IV: Tillandsia dasyliriifolia Baker: taxonomy and reproductive biology. Journal of the Bromeliad Society 54: 112–121. Ranker TA, Soltis DE, Soltis PS, Gilmartin AJ. 1990. Subfamilial phylogenetic relationships of the Bromeliaceae: evidence from chloroplast DNA restriction site variation. Systematic Botany 15: 425–434. Rex M, Schulte K, Zizka G, Weising K. 2009. Phylogenetic analysis of Fosterella L.B. Sm. (Pitcairnioideae, Bromeliaceae) based on four chloroplast DNA regions. Molecular Phylogenetics and Evolution 51: 472–485. Ronquist F, Huelsenbeck JP. 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. Ronquist F, Huelsenbeck JP, Teslenko M. 2011. MrBayes version 3.2 manual: tutorials and model summaries. Program manual. Available at: www.mrbayes.net © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 20
ET AL. J. P. PINZON Rzedowski J. 1978. La vegetaci
on de M
exico, 1a ed. digital. Mexico: Comisi
on Nacional para el Conocimiento y Uso de la Biodiversidad. Sch€ utz N. 2012. Systematics and evolution of the genus Deuterocohnia Mez (Bromeliaceae). Dr. Rer. Nat. Dissertation, University of Kassel. Shaw J, Lickey EB, Schilling EE, Small RL. 2007. Comparison of whole chloroplast genome sequences to choose noncoding regions for phylogenetic studies in angiosperms: the tortoise and the hare III. American Journal of Botany 94: 275–288. Simmons MP, Ochoterena H. 2000. Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49: 369–381. Smith LB, Downs RJ. 1977. Tillandsioideae (Bromeliaceae). In: Flora Neotropica, Monograph 14, Part 2. New York: Haffner Press, 661–1492. Sung GH, Sung JM, Hywel-Jones NL, Spatafora JW. 2007. A multigene phylogeny of Clavicipitaceae (Ascomy- cota, Fungi): identification of localised incongruence using a combinational bootstrap approach. Molecular Phylogenetics and Evolution 44: 1204–1223. Terry RG, Brown GK, Olmstead RG. 1997a. Examination of subfamilial phylogeny in Bromeliaceae using comparative sequencing of the plastid locus ndhF. American Journal of Botany 84: 664–670. Terry RG, Brown GK, Olmstead RG. 1997b. Phylogenetic relationships in subfamily Tillandsioideae (Bromeliaceae) using ndhF sequences. Systematic Botany 22: 333–345. Whitten WM, Williams NH, Chase M. 2000. Subtribal and generic relationships of Maxillarieae (Orchidaceae) with emphasis on Stanhopeinae: combined molecular evidence. American Journal of Botany 87: 1842–1856. Yu Y, Harris AJ, Blair C, He XJ. 2015. RASP (reconstruct ancestral state in phylogenies): a tool for historical biogeography. Molecular Phylogenetics and Evolution 87: 46–49. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Species Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality Catopsis nutans (Sw.) Griseb. var. nutans Racinaea fraseri (Baker) M.A.Spencer & L.B.Sm. R. fraseri B0002 AY614392† AY614148† KU848418 KU848264 Costa Rica GB2910 AF539977*,§ AF537914§ NS NS FRP90 EU681906‡‡ EF643192‡‡ NS NS T. achyrostachys E.Morren ex Baker T. achyrostachys T. achyrostachys DTA1 FM210787** FM211650** NS NS LTA2 ALF6532 FM210788** NS FM211653** NS NS NS NS FJ666937‡ T. adpressiflora Mez T. aeranthos (Loisel.) L.B.Sm. T. albertiana Verv. T. albida Mez & Purpus B0597 B0111 KU848347 AY614131† KU848508 AY614253† KU848440 NS KU848284 NS B0033 JP016 AY614117† KU848380 AY614239† KU848509 NS KU848458 NS KU848321 T. andrieuxii (Mez) L.B.Sm. T. atroviridipetala Matuda T. argentea Griseb. T. argentina C.H.Wright T. ariza-juliae L.B.Sm. & J.Jim
enez. Alm. T. balbisiana Schult.f. T. baliophylla Harms T. barclayana Baker T. barthlottii Rauh B0063 AY614088† AY614210† NS NS E. Trauner s.n. (WU) G. Brown 2910 (RM) G. Zizka 1582 (FRP) Dotterer TA1 (NAP) Larson TA2 (NAP) A. Espejo et al. 6532 (UAMIZ) W. Till 21158 (WU) Coll. M.H.J. Barfuss s.n. (WU) HBV B387/90 (WU) I. Ram
ırez & S. Zamudio 1414 (CICY) HBV B 256/95 (WU) TC089 NS NS NS FJ666932‡ T. Chew 89 (XAL) Mexico JP082 B0087 PKT504 KU848359 AY614124† NS KU848568 AY614246† NS KU848431 NS NS KU848289 NS Fj666939‡ K. Willinger s.n. (SEL) H. Till 88-45 (WU) Bird Rock Tropical Koide T504 Cuba: Oriente Argentina: Catamarca – TC167 B0101 B0028 B0035 NS AY614114† AY614079† AY614076† NS AY614236† AY614201† AY614198† NS NS NS NS EU126833‡ NS NS NS – Dominican Republic: La Vega Ecuador Ecuador: Loja T. barthlotti B0716 NS NS KU848427 NS T. bergeri Mez B0097 AY614134† AY614256† NS NS T. bergeri B0110 AY614133† AY614255† NS NS T. Chew 167 (XAL) W. Till 17025 (WU) HBV B518/96 (WU) H. & L. Hromadnik 4078 (WU) H. & L. Hromadnik 4078 (WU) W. Papsch & G. Hold 89-060/074 Coll. M.H.J. Barfuss s.n. (WU) – – Mexico Mexico – Ecuador: Napo – Argentina: Salta Mexico: Quer
etaro Mexico Ecuador: Loja Argentina: Buenos Aires – PHYLOGENY IN TILLANDSIA UTRICULATA © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Appendix 1 List of taxa, code, GenBank accession number, voucher and locality of the samples used for this study (NS, not sequenced). 21 22 Appendix 1. Continued Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. bermejoensis L. Hrom. T. biflora Ruiz & Pav. T. brachyphylla Baker T. brevilingua Mez ex Harms T. bulbosa Hook. T. cacticola L.B.Sm. T. calcicola L.B.Sm. & Proctor T. califanii Rauh T. califanii T. caput-medusae E. Morren T. caput-medusae T. carlos-hankii Matuda T. carnosa L.B.Sm. T. caulescens Brong. ex Baker T. chlorophylla L.B. Sm. T. coinaensis Ehlers T. comitanensis Ehlers B0034 AY614123† AY614245† NS NS W. Till 144 (WU) Bolivia: Santa Cruz B0090 B0082 B0056 AY614123† AY614105† AY614113† AY614245† AY614227† AY614235† NS NS NS KU848281 KU848280 NS F.-G. Gruber s.n. HBV B99B16-1 (WU) W. & S. Till 2097 (WU) Venezuela: Lara Brazil: Rio de Janeiro Peru: San Martin TC126 B0044 JP105 NS AY614070† KU848367 NS AY614192† KU848539 NS KU848426 KU848445 FJ666933‡ NS KU848308 T. Chew 126 (XAL) W. Till 2133 (WU) Rutschmann s.n. (WU) – Peru: Piura Jamaica WR36219 WTC5 B0046 FM210789** FM210790** AY614098† FM211651** FM211652** AY614220† NS NS KU848500 NS NS KU848307 W.Rauh 36219 (HEID) Wrinkle TC5 (NAP) W. Till 7117 (WU) Mexico: Puebla Mexico Costa Rica: Puntarenas TC100 B0062 NS AY614089† NS AY614211† NS NS FJ666934‡ KU848296 – Mexico: Oaxaca B0755 B0071 KU848356 AY614126† KU848572 AY614248† KU848430 NS KU848269 NS JP139 NS KU848564 KU848498 KU848299 B0091 JP074 AY614102† KU848387 AY614224† KU848513 NS KU848467 NS KU848327 T. aff. comitanensis JP075 KU848386 KU848514 KU848468 KU848317 T. cucaensis Wittm. JP029 KU848388 KU848532 KU848471 KU848342 T. cucaensis JP030 KU848389 KU848530 KU848469 KU848341 T. cucaensis JP056 KU848390 KU848524 KU848470 NS T. cucaensis JP076 KU848392 KU848526 KU848472 KU848340 T. dasyliriifolia Baker JP001 KU848405 KU848534 NS NS T. dasyliriifolia JP003 KU848406 KU848503 KU848488 KU848331 T. Chew 100 (XAL) L. Hromadnik 15169 (WU) W. Till 2066 (WU) E. Vitek 820812/72-1 (WU) J.P. Pinz
on et al. 119 (CICY) E. Zecher 21/76 (WU) J.P. Pinz
on et al. 97 (CICY) J.P. Pinz
on et al. 98 (CICY) J.P. Pinz
on et al. 1 (CICY) J.P. Pinz
on & G. Carnevali 77 (CICY) J.P. Pinz
on et al. 67 (CICY) J.P. Pinz
on et al. 99 (CICY) I. Ram
ırez et al. 785 (CICY) G. Carnevali s.n. (CICY) Peru: Amazonas Peru: Apurimac Mexico: Chiapas Peru: Cajamarca Mexico: Chiapas Mexico: Chiapas Mexico: Oaxaca Mexico: Oaxaca Mexico: Chiapas Mexico: Chiapas Mexico: Yucat
an Mexico: Campeche
ET AL. J. P. PINZON © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Species Species Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. dasyliriifolia T. dasyliriifolia JP083 JP084 KU848407 KU848408 KU848521 KU848517 KU848487 KU848486 KU848332 NS Belize Belize T. dasyliriifolia JP085 KU848409 KU848519 KU848489 KU848339 T. demissa L.B.Sm. B0075 AY614115† AY614237† NS NS T. deppeana Steud. T. didisticha (E.Morren) Baker T. diguetii Mez & Rol.-Goss. T. disticha Kunth T. disticha TC051 B0038 NS AY614127† NS AY614249† NS NS FJ666926‡ NS W. Berg s.n. (SEL) Berg & Cathcart s.n. (SEL) Carnevali et al. s.n. (SEL) K.-D. & R. Ehlers EE84 s.n. (WU) T. Chew 51 (XAL) W. Till 10130 (WU) ALF2972 NS NS NS FJ666923‡ – B0048 B0233 AY614068† KU848346 AY614190† NS NS KU848422 NS KU848265 T. dodsonii L.B.Sm. T. dodsonii Tillandsia duratii Vis. var. duratii T. eizii L.B.Sm. T. elusiva Pinz
on, I.Ram
ırez & Carnevali T. elusiva B0016 B0127 B0088 AY614072† KU848344 AY614119† AY614194† NS AY614241† NS KU848505 NS KU848273 KU848282 NS Lopez-Ferrari et al. 2972 (UAMIZ) K. Oppitz s.n. (WU) H. & L. Hromadnik 17063 (WU) W. Rauh 34183 (WU) C. H. Doson 5225 (WU) W. Till 5072 (WU) JC1374 JP111 NS KU848373 NS KU848540 NS KU848451 EU126830‡ KU848310 Ceja et al. 1374 (MEXU) J.P. Pinz
on et al. 104 (CICY) – Mexico: Chiapas JP120 KU848374 KU848541 KU848452 KU848311 Mexico: Chiapas T. erubescens Schltdl. T. espinosae L.B.Sm. TC84 B0143 NS NS NS NS NS NS EU126831‡ KU848266 T. esseriana Rauh & L.B.Sm. T. exserta Fernald T. exserta T. fasciculata Sw. var. fasciculata T. fasciculata T. fendleri Griseb. var. fendleri T. flabellata Baker B0069 AY614120† AY614242† NS NS J.P. Pinz
on et al. 105 (CICY) T. Chew 84 (XAL) BGBM Berlin-Dahlem 021-03-74-83 16926 (B) HBV B342/90 (WU) LTE2 B0390 B0076 B0717 WTF2 B0009 FM210791** KU848414 AY614100† FM211654** KU848562 AY614222† NS KU848497 NS NS KU848306 KU848305 Larson TE2 (NAP) Schatzl 51/77 (WU) W. & S. Till 7050 (WU) Mexico Mexico: Nayarit Costa Rica: San Jos
e FM210792** AY614116† FM211655** AY614238† NS NS NS NS Mexico Peru: La Libertad JP069N KU848416 KU848559 NS NS Wrinkle TF2 (NAP) H. & L. Hromadnik 2082 (WU) J.P. Pinz
on et al. 64 (CICY) Mexico: Quintana Roo Ecuador: Loja – Argentina: Jujuy Ecuador: Azuay Ecuador Ecuador Ecuador: Pichincha Argentina: La Rioja – – Paraguay: Amambay Mexico: Chiapas PHYLOGENY IN TILLANDSIA UTRICULATA © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Appendix 1. Continued 23 24 Appendix 1. Continued Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. flabellata ALF6419 NS NS NS FJ666928‡ – T. flexuosa Sw. JP002 KU848361 KU848569 KU848434 NS T. fresnilloensis W.Weber & Ehlers T. fuchsii W.Till JP018 KU848381 KU848537 KU848460 KU848329 A. Espejo et al. 6419 (UAMIZ) J.P. Pinz
on & G. Carnevali 230 (CICY) I. Ram
ırez 1310 (CICY) JP017 KU848366 KU848556 KU848461 KU848285 T. funckiana Baker var. recurvifolia Blass ex Rauh T. funebris A.Cast. T. gardneri Lindl. var. gardneri T. grandis Schltdl. T. guatemalensis L.B.Sm. T. guatemalensis JP046 KU848357 KU848565 KU848432 B0089 B0041 AY614118† AY614104† AY614240† AY614226† B0124 JP071 NS NS JP072 T. guatemalensis T. guatemalensis Venezuela: Aragua Mexico: Zacatecas Mexico: Chiapas KU848286 J.P. Pinz
on & G. Carnevali 231 (CICY) M. Speckmaier s.n. (WU) NS NS NS NS HBV B35/94 (WU) W. Till 11134 (WU) Bolivia: Cochabamba Brazil: Rio de Janeiro NS NS NS NS KU848271 KU848300 Mexico: Veracruz Mexico: Chiapas KU848363 KU848563 KU848491 NS B0008 B0103 AY614092† AY614094† AY614214† AY614216† NS NS NS KU848301 T. guatemalensis B0104 AY614093† AY614215† KU848490 KU848302 T. gymnobotrya Baker JP045 KU848363 KU848551 KU848494 KU848295 T. heterophylla E.Morren T. heterophylla T. heterophylla B0047 AY614111† AY614233† NS KU848276 TC052 JP068R NS NS NS NS NS KU848428 FJ666927‡ KU848277 T. eubergeri Ehlers T. hildae Rauh T. huamelulaensis Ehlers T. huamelulaensis B0042 JP040 JP142 AY614106† KU848355 KU848393 AY614228† KU848571 KU848520 NS KU848429 KU848474 NS KU848292 KU848325 JP143 KU848394 KU848518 KU848473 NS T. intermedia Mez TC189 NS NS NS FJ666935‡ E. Zecher s.n. (WU) J.P. Pinz
on et al. 89 (CICY) J.P. Pinz
on et al. 390 (CICY) HBV B 260/96 (WU) H. & L. Hromadnik 14257 (WU) L. Hromadnik 15127 (WU) R. Ehlers EM031403 (WU) L. Hromadnik 15191 (WU) T. Chew 52 (XAL) J.P. Pinz
on & V. Rebolledo 73 (CICY) F. Fuchs. s.n. (WU) HBV B148/82 (WU) J.P. Pinz
on et al. 227 (CICY) J.P. Pinz
on et al. 228 (CICY) T. Chew 189 (XAL) Venezuela: Carabobo Mexico: Chiapas Mexico Mexico: Chiapas Mexico: Chiapas Mexico Mexico: Veracruz – Mexico: Veracruz Brazil: Bahia Peru: Cajamarca Mexico: Oaxaca Mexico: Oaxaca –
ET AL. J. P. PINZON © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Species Species Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. ionantha Planch. var. ionantha T. ionantha T. ixioides Griseb. B0084 AY614099† AY614221† NS NS Mexico: Puebla TC038 B0043 NS AY614129† NS AY614251† NS NS FJ666931‡ KU848275 – Argentina: Catamarca JP080 KU848401 KU848515 KU848482 KU848328 H. & L. Hromadnik s.n. (WU) T. Chew 38 (XAL) G. Neuhuber GN 96-936/3084 (WU) R. Foster s.n. (SEL) B0732 B0073 KU848402 AY614097† KU848522 AY614219† KU848481 NS NS KU848304 W. Rauh 70802 (HEID) W. & S. Till 7033 (WU) Guatemala: Izabal Costa Rica: Limon TC057 JP112 JP044 NS KU848358 KU848382 NS KU848567 KU848538 NS KU848435 KU848459 EU126832‡ KU848288 KU848330 T. Chew 57 (XAL) Hromadnik 23176 (HBV) Schatzl 76/77 (WU) – Peru: Junin Mexico: Hidalgo B0734 KU848379 NS NS NS Mexico: Tamaulipas T. kauffmannii Ehlers T. kegeliana Mez T. klausii Ehlers B0074 JP064 B0085 AY614103† KU848360 AY614096† AY614225† KU848570 AY614218† NS KU848433 KU848501 KU848279 KU848287 KU848298 T. latifolia Meyen var. divaricata (Benth.) Mez T. leiboldiana Schltdl. B0068 AY614108† AY614230† NS NS R. Ehlers & L. K€ ohres s.n. (HEID) E. Trauner s.n. (WU) M. Speckmaier s.n. (WU) K.-D. & R. Ehlers EM851801 W. Till 13069 (WU, QCA) JP140 KU848411 KU848553 KU848492 KU848303 Mexico: Chiapas T. lepidosepala L.B.Sm. T. lepidosepala KHTL001 B0219 FM210793** NS FM211656** NS NS KU848423 NS KU848293 T. limbata Schltdl. JP020 KU848403 KU848504 KU848483 KU848334 T. limbata JP055 KU848404 KU848528 KU848484 NS T. 9 duvalii L. Duval T. 9 duvalii T. macbrideana L.B.Sm. var. macbrideana T. macdougallii L.B.Sm. B0023 B0746 B0070 AY614080† NS AY614109† AY614202† NS AY614231† KU848419 NS NS KU848274 KU848283 NS J.P. Pinz
on et al. 120 (CICY) Kak.Haa TL001 (NAP) L. Hromadnik 15195 (WU) I. Ram
ırez et al. 1464 (CICY) J.P. Pinz
on et al. 70 (CICY) HBV B91/80 (WU) G€ ottingen s.n. (WU) HBV B249/87 (WU) HSSN FM956440†† NS NS NS S.H. Salas s.n. (MEXU) T. izabalensis Pinz
on, I.Ram
ırez & Carnevali T. izabalensis T. juncea (Ruiz & Pav.) Poir T. juncea T. juruana Ule T. karwinskyana Schult. & Schult.f. T. cf. karwinsyana Honduras: Cayos Peru: La Libertad Panama Mexico: Chiapas Ecuador: Chimborazo Mexico Mexico: Puebla Mexico: Veracruz Mexico: Chiapas – – Peru: Lima PHYLOGENY IN TILLANDSIA UTRICULATA © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Appendix 1. Continued Mexico: Oaxaca 25 26 Appendix 1. Continued Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. macdougallii T. macropetala Wawra JP100 B0742 NS NS KU848550 KU848573 KU848496 KU848425 NS KU848272 Mexico: Oaxaca Mexico: Oaxaca T. magnusiana Wittm. T. makoyana Baker TC130 JP008 NC KU848397 NC KU848533 NC KU848479 FJ666941‡ NS T. makoyana JP028 KU848398 KU848529 KU848480 NS T. makoyana JP048 KU848399 KU848531 KU848478 KU848333 T. makoyana JP051 KU848400 KU848527 KU848477 NS T. marconae W.Till & Vitek T. marnier-lapostollei Rauh T. matudae L.B.Sm. T. may-patii I.Ram
ırez & Carnevali T. mima L.B.Sm. B0098 AY614069† AY614191† NS NS D. Mondrag
on 28 (CICY) J. Lautner 05/17 (GOET, WU) T. Chew 130 (XAL) J.P. Pinz
on et al. 28 (CICY) I. Ram
ırez et al. 1519 (CICY) J.P. Pinz
on et al. 109 (CICY) J.P. Pinz
on et al. 110 (CICY) W. Till 234 (WU) JP113 KU848352 KU848511 KU848442 NS Hromadnik 4125 (WU) Ecuador: Azuay KHTM001 JP054 FM210794** KU848410 FM211657** KU848536 NS KU848485 NS NS Kak.Haa TM001 (NAP) J.P. Pinz
on et al. 76 (CICY) Mexico Mexico: Quintana Roo JP091 KU848351 KU848574 KU848438 NS Ecuador: Azuay T. multicaulis Steud. B0107 AY614112† AY614234† NS NS T. multicaulis T. narthecioides C.Presl T. nicolasensis Ehlers TC047 B0060 NS AY614071† NS AY614193† NS NS EU126829‡ KU848278 Cathcart & Berg s.n. (SEL) H. & L. Hromadnik 1087 (WU) T. Chew 47 (XAL) HBV B8/90 (WU) JP010 KU848384 KU848554 KU848465 KU848319 T. nicolasensis JP077 KU848385 KU848516 KU848466 KU848320 T. novakii H.Luther T. paniculata (L.) L. JP092 B0102 KU848415 AY614086† KU848561 AY614208† KU848499 KU848444 NS KU848294 T. paucifolia Baker DL0109 FN550871** FN550873** NS NS T. pinicola I.Ram
ırez & Carnevali JP027 KU848395 KU848535 KU848476 KU848323 J.P. Pinz
on et al. 51 (CICY) I. Ram
ırez et. al 1108 (CICY) A.J. Novak s.n. (SEL) W. Till 17057 (WU) De Luca & VazquezTorres 01.2009 (NAP, HEID) G. Carnevali et al. 7353 (CICY) – Mexico: Guerrero Mexico: Mexico Mexico: Oaxaca Mexico: Oaxaca Peru: Ica Mexico: Veracruz – Ecuador Mexico: Jalisco Mexico: Guerrero Mexico: Veracruz Dominican Republic: Distrito Nacional Mexico: Veracruz Mexico: Oaxaca
ET AL. J. P. PINZON © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Species Species Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. pinicola JP070 KU848396 KU848525 KU848475 KU848326 Mexico: Oaxaca T. plumosa Baker B0086 AY614075† AY614197† NS NS T. pohliana Mez T. pringlei S.Watson B0080 JP004 AY614128† KU848375 AY614250† KU848548 NS KU848457 NS KU848324 T. pringlei T. pringlei T. pringlei T. pringlei T. prodigiosa (Lem.) Baker T. prodigiosa JP096 B0733 B0735 B0736 CG320 NS KU848376 KU848377 KU848378 FM956437†† KU848543 KU848545 KU848542 KU848544 NS KU848455 KU848456 KU848454 KU848453 NS KU848335 NS KU848336 NS NS JP098 NS KU848552 KU848495 NS JP043 KU848350 KU848575 KU848443 KU848268 B0036 AY614110† AY614232† NS NS J.P. Pinz
on & G. Carnevali 136 (CICY) K.-D. & R. Ehlers EM 881905 (WU) W. Till 11004 (WU) I. Ram
ırez & S. Zamudio 1435 (CICY) G. Newhouse s.n. (SEL) W. Rauh 21345 (HEID) A. Lau s.n. (HEID) W. Rauh 21340 (HEID) C. Granados 320 (MEXU) A.R. L
opez-Ferrari et al. 3069 (CICY) H. & L. Hromadnik 2139 (WU) W. Rauh 53774 (WU) JP049 KU848417 KU848560 KU848502 NS Zecher s.n. (WU) Mexico B0061 AY614087† AY614209† KU848493 KU848297 H.-H. Deissl s.n. (WU) Costa Rica TC049 B0092 NS AY614101† NS AY614223† NS NS FJ666930‡ NS T. Chew 49 (XAL) W. Rauh 69417 (WU) – Peru: Cajamarca B0072 AY614095† AY614217† NS NS Honduras: Cop
an JP063 TC121 B0064 KU848348 NS AY614039† KU848577 NS AY614161† KU848441 NS NS KU848318 FJ666929‡ NS H. & I. Seethaler s.n. (WU) W. Till 21022 (WU) T. Chew 121 (XAL) W. Till 15023 (WU) JP062 JP094 JP104 KU848365 KU848364 KU848349 KU848557 KU848558 KU848576 KU848462 KU848463 KU848439 KU848290 KU848291 NS B0081 AY614130† AY614252† NS NS T. propagulifera Rauh T. pseudomacbrideana Rauh T. pueblensis L.B.Sm. T. punctulata Schltdl. & Cham. T. punctulata T. rauhii L.B.Sm. var. rauhii T. remota Wittm. T. secunda Kunth T. seleriana Mez T. singularis Mez & Werckl
e T. socialis L.B.Sm. T. socialis T. spiraliflora Rauh T. stricta Sol. ex Sims var. stricta HBV B271/96 (WU) D. Cathcart s.n. (SEL) L. Hromadnik 2114 (WU) E. Markus s.n. (WU) Mexico: Oaxaca Brazil: S~ ao Paulo Mexico: Quer
etaro Mexico: Mexico: Mexico: Mexico: Mexico: Tamaulipas San Luis Potos
ı Quer
etaro San Luis Potos
ı Oaxaca Mexico: Oaxaca Peru: Amazonas Peru: Cajamarca Ecuador: Imbabura – Costa Rica: Alajuela Mexico Mexico: Chiapas Peru: Amazonas Brazil: Minas Gerais PHYLOGENY IN TILLANDSIA UTRICULATA © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Appendix 1. Continued 27 28 Appendix 1. Continued Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. rupicola Baker B0039 AY614073† AY614195† NS NS Ecuador: Azuay T. tehuacana I.Ram
ırez & Carnevali T. tenuifolia L. var. tenuifolia T. tomasellii De Luca, Sabato & Balduzzi T. tortilis Klotzsch ex Baker ssp. tortilis T. triglochinioides C.Presl T. usneoides (L.) L. T. usneoides JP050 KU848383 KU848555 KU848464 KU848322 B0026 AY614132† AY614254† NS NS W. Till 13081 (WU, QCA) J.P. Pinz
on et al. 47 (CICY) W. Till 131 (WU) PA3777 FM210795** FM211658** NS NS B0049 AY614074† AY614196† NS B0725 KU848345 KU848506 B0083 B0109 AY614122† AY614121† T. usneoides T. utriculata L. TC050 JP006 T. utriculata Mexico: Puebla Bolivia: Santa Cruz Mexico: Oaxaca NS P. de Luca et al. 3777 (PAV) HBV B218A/88 (WU) KU848420 NS W. Rauh 34378 (HEID) Ecuador: Manabi AY614244† AY614243† NS NS NS NS Venezuela – NS KU848370 NS KU848547 NS KU848450 FJ666938‡ NS JP060 KU848372 KU848549 KU848449 KU848314 T. utriculata JP061 NS KU848507 KU848448 KU848313 T. utriculata T. utriculata JP095 B0027 NS AY614091† NS AY614213† KU848447 NS KU848309 NS T. utriculata T. utriculata T. cf. utriculata T. venusta Mez & Werckl
e T. viridiflora (Beer) Baker T. wagneriana L.B.Sm. T. wagneriana L.B.Sm. T. werneriana J.R.Grant B0100 B0807 TC143 B0007 AY614090† KU848368 NS AY614081† AY614212† KU848546 NS AY614203† NS KU848446 NS NS KU848315 KU848316 FJ666940‡ NS G. Palim s.n. (WU) Coll. M.H.J. Barfuss s.n. (WU) T. Chew 50 (XAL) J.P. Pinz
on et al. 233 (CICY) J.P. Pinz
on et al. 56 (CICY) J.P. Pinz
on et al. 206 (CICY) H.B. Rinker s.n. (SEL) G. Neuhuber 98-982/3296 (WU) W. Till 17007 (WU) W. Janetzky 22 (WU) T. Chew 143 (XAL) HBV B98B136-1 (WU) Dominican Republic: Espaillat Jamaica: Middlesex – – B0006 AY614066† AY614188† NS NS HBV B87/80 (WU) – B0058 AY614067† AY614189† KU848421 NS HBV B222/93 (WU) Peru: Amazonas B0217 KU848343 KU848579 NS KU848270 Peru: Amazonas B0067 AY614078† AY614200† NS NS H. Prinsler s.n., 1990-09 (WU) H. & L. Hromadnik 2142 (WU) Mexico: Oaxaca – Mexico: Yucat
an Mexico: Tabasco Mexico: Chiapas USA: Puerto Rico USA: Florida Peru: Amazonas
ET AL. J. P. PINZON © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Species Species Code matK-trnK 30 rps16 intron rpl32-trnL ETS Voucher Locality T. xerographica Rohweder T. xiphioides Ker Gawl. var. xiphioides Vriesea malzinei E.Morren LOSN FM210797** FM211660** NS NS Lozada s.n. (NAP) Mexico B0040 AY614125† AY614247† NS NS F. Strigl FO 275 (WU) Argentina: Santiago del Estero B0145 KU848353 KU848510 KU848437 KU848267 BGBM 109-37-74-83 (B) Mexico *Partial matK sequence, without non-coding part of 30 end of trnK intron. †Barfuss et al. (2005). ‡Chew et al. (2010). §Crayn et al. (2004). **De Castro et al. (2009). ††Granados (2008). ‡‡Rex et al. (2009). PHYLOGENY IN TILLANDSIA UTRICULATA © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016 Appendix 1. Continued 29