Chromosome numbers and DNA content in Bromeliaceae: additional data and critical review

bs_bs_banner Botanical Journal of the Linnean Society, 2014, 176, 349 368. With 3 figures Chromosome numbers and DNA content in Bromeliaceae: additional data and critical review JAILSON GITAÍ 1,2, JURAJ PAULE 3 *, GEORG ZIZKA 3, KATHARINA SCHULTE 3,4 and ANA MARIA BENKO-ISEPPON 1 1 Department of Genetics, Universidade Federal de Pernambuco, Av. Prof. Moraes Rego, 1235, 50.670-420, Recife, PE, Brazil 2 Universidade Federal Rural de Pernambuco, Unidade Acadêmica de Serra Talhada, 56.900-000, Serra Talhada, PE, Brazil 3 Department of Botany and Molecular Evolution, Senckenberg Research Institute, Biodiversity and Climate Research Centre (BiK-F) & Johann Wolfgang Goethe-University, D-60325 Frankfurt am Main, Germany 4 Australian Tropical Herbarium & Center for Tropical Biodiversity and Climate Change, James Cook University, Cairns, QLD 4870, Australia Received 18 February 2014; revised 19 June 2014; accepted for publication 10 August 2014 For the large Neotropical plant family Bromeliaceae, we provide new data on chromosome numbers, cytological features and genome estimations, and combine them with data available in the literature. Root-tip chromosome counts for 46 species representing four subfamilies and a literature review of previously published data were carried out. Propidium iodide staining and flow cytometry were used to estimate absolute genome s in five subfamilies of Bromeliaceae, sampling 28 species. Most species were diploid with 2n = 50 in Bromelioideae, Puyoideae and Pitcairnioideae, followed by 2n = 48 observed mainly in Tillandsioideae. Individual chromosome s varied more than tenfold, with the largest chromosomes observed in Tillandsioideae and the smallest in Bromelioideae. Genome s (2C-values) varied from 0.85 to 2.23 pg, with the largest genomes in Tillandsioideae. Genome evolution in Bromeliaceae relies on two main mechanisms: polyploidy and dysploidy. With the exception of Tillandsioideae, polyploidy is positively correlated with genome. Dysploidy is suggested as the mechanism responsible for the generation of the derived chromosome numbers, such as 2n = 32/34 or 2n = 48. The occurrence of B chromosomes in the dysploid genus Cryptanthus suggests ongoing speciation processes closely associated with chromosome rearrangements. 2014 The Linnean Society of London, Botanical Journal of the Linnean Society, 2014, 176, 349 368. ADDITIONAL KEYWORDS: B chromosomes bimodality bromeliads C-value dysploidy genome polyploidy. INTRODUCTION *Corresponding author. E-mail: juraj.paule@senckenberg.de Bromeliaceae, a large Neotropical family, comprises 58 genera and about 3352 species distributed mainly in tropical to subtropical climates (Luther, 2012). The family contains a rich diversity, including epiphytic, lithophytic and terrestrial life forms associated with a wide adaptive radiation in distinct habitats (Benzing, 2000; Crayn, Winter & Smith, 2004). Traditionally, Bromeliaceae was divided into three subfamilies (Pitcairnioideae, Bromelioideae and Tillandsioideae), but, although recent molecular phylogenetic studies confirmed the monophyly of Bromelioideae and Tillandsioideae, they showed that Pitcairnioideae was paraphyletic (Crayn et al., 2004; Givnish et al., 2004; Barfuss et al., 2005; Schulte, Horres & Zizka, 2005). Consequently, Pitcairnioideae has been divided into five monophyletic subfamilies: Pitcairnioideae s.s., Brocchinioideae, Lindmanioideae, Navioideae and Hechtioideae (Givnish et al., 2007, 2011). Although Tillandsioideae is the largest bromeliad subfamily (nine genera, c. 1337 species), Bromelioideae 349

350 J. GITAÍ ET AL. harbours the most genera (35 genera, > 940 species) (Luther, 2012). Karyological analyses of plants have the potential to address a range of evolutionary questions. In Bromeliaceae, cytogenetic studies have so far been carried out for c. 10% of species. After some historical discussion and contradictory interpretations concerning the basic chromosome number (summarized in McWilliams, 1974; Brown & Gilmartin, 1986, 1989; Bellintani, Assis & Cotias-de-Oliveira, 2005), x = 25 was accepted, as proposed by Marchant (1967). Furthermore, it has been reported that the evolution of Bromeliaceae is characterized by several hybridization, polyploidization and dysploidy events (Marchant, 1967; Brown & Gilmartin, 1986; Gitaí, Horres & Benko-Iseppon, 2005), the latter often erroneously referred to as aneuploidy. Members of Bromeliaceae possess small chromosomes (e.g. Sharma & Gosh, 1971; Cotias-de- Oliveira et al., 2004; Gitaí et al., 2005) and most studied accessions have the diploid chromosome number 2n = 50. Flow cytometry is considered to be a fast and precise method for the examination of ploidy and genome (DNA content) in plants (Doležel, Greilhuber & Suda, 2007). Genome, estimated as the 2C-value, describes the total DNA content of the unreplicated 2n nucleus (Greilhuber et al., 2005). It is strongly influenced by chromosome number and chromosome /morphology, and there is often a good correlation between genome and total chromosome length in a karyotype (e.g. Leitch et al., 2009). However, genome does not necessarily correlate positively with chromosome number. For instance, in Silene L. (Caryophyllaceae), 2C-values vary between 2.25 and 6.59 pg, although the species have the same chromosome number (Široký et al., 2001). This pattern has also been observed in Zingiberaceae, where large differences in nuclear DNA content occur among taxa with the same ploidy and chromosome number (Leong-Škorničková et al., 2007), and in Cypripedium L. (Orchidaceae), where genome varies more than tenfold between species with the same ploidy and chromosome number (Leitch et al., 2009). Therefore, the combination of chromosome number, chromosome morphology and genome may yield important insights for a better understanding of genome evolution. There have been several successful attempts to estimate genome in Bromeliaceae (e.g. Ramírez-Morillo & Brown, 2001) and the data have been added to and summarized recently by Favoreto et al. (2012). However, the available data do not sufficiently cover the taxonomic diversity of the family (< 2%) and genome evolution is only poorly understood. The present study provides new or additional karyological data for 55 species of Bromeliaceae, including chromosome numbers, morphology and genome estimations. Furthermore, the results are discussed and analysed together with all previously published reports, with the aim to answer the following questions: (1) What is the taxonomic distribution of polyploidy in the family?; (2) Is there a positive correlation between genome and ploidy?; (3) Are there other mechanisms responsible for chromosome evolution besides polyploidy?; (4) In addition to chromosome numbers, are there other useful karyomorphological features?; (5) What is the role of B chromosomes reported for some genera?; (6) Are there specific cytological trends in different subfamilies and genera when the data are viewed within a phylogenetic framework? MATERIAL AND METHODS CHROMOSOME COUNTS Collected root tips were immersed in 2 mm 8-hydroxyquinoline for 20 24 h or in 0.05 0.1% colchicine for 3 6 h at 5 8 C. Root tips were fixed in Carnoy s solution 3 : 1 (ethanol/acetic acid) at 15 C for 2 24 h and stored at 20 C. Standard chromosome preparations with HCl/Giemsa were carried out as described by Benko-Iseppon & Morawetz, (2000). As all species had small chromosomes, a minimum of eight metaphase or prometaphase spreads was examined to obtain reliable chromosome counts. Measurements of maximum and minimum chromosome s were based on photographs of two to four metaphase plates of each species. For part of the analysed material, photomicrographs were taken with Kodak Technical Pan and with Kodak T-Max 400 (Eastman Kodak Company, Rochester, NY, USA). For the remaining analyses, cell images were acquired using a Leica DM LB microscope and a Leica DFC 340FX camera with Leica CW4000 software (Leica, Wetzlar, Germany). Details of the studied material are given in Table 1. Herbarium specimens were deposited at the Herbarium Senckenbergianum (FR), Herbarium (FRP) and Herbarium Universidade Federal de Pernambuco (UFP). GENOME SIZE ESTIMATIONS Genome s (2C-values) were estimated for 28 accessions (Table 1) using a Partec CyFlow SL flow cytometer (Partec, Münster, Germany) equipped with a green solid-state laser (Cobolt Samba, 532 nm, 100 mw) at the Department of Botany, Charles University in Prague (18 samples), and a Partec CyFlow Space (Partec, 532 nm, 30 mw) flow cytometer at Centre for Organismal Studies (COS) Heidelberg (15

BROMELIAD CHROMOSOMES AND DNA CONTENT 351 Table 1. List of studied accessions, karyological features and 2C-values. Approximate chromosome counts are indicated by Taxon Provenance and material number Chromosome numbers C-value Metaphase chromosome Present study Previous reports 2C (pg) ± SD 2C (pg) previous reports length maximal (μm) Karyotype architecture BROCCHINIOIDEAE Brocchinia uaipanensis (Maguire) Givnish BROMELIOIDEAE Aechmea aquilega (Salisb.) Griseb. A. bracteata (Sw.) Griseb A. caudata Lindm. var. caudata A. eurycorymbus Harms A. fendleri André ex Mez 9510; originally collected on Auyantepui in Venezuela 16872 21586 12334 16942; originally collected in Pernambuco, Brazil 14217 A. ornata Baker Botanical Garden of the University Leipzig, Leg. Nr. 16929 2n = 46 0.92 ± 0.014 0.86 1.62 0.81 ± Homogeneous 2n = 50 2n = 50 1.08 ± 0.013 1.78 0.71 ± Homogeneous 2n = 50 2n = 50 2.24 0.71 ± Homogeneous 2n = 50 2n = 50 1.78 0.17 ± Homogeneous 2n 88* 2.19 ± 0.013 1.92 0.76 ± Homogeneous 2n = 50* 1.25 ± 0.018 1.78 0.20 ± Homogeneous 2n = 50 2n = 54 1.25 0.71 ± Homogeneous

352 J. GITAÍ ET AL. Table 1. Continued Taxon Provenance and material number Chromosome numbers C-value Metaphase chromosome Present study Previous reports 2C (pg) ± SD 2C (pg) previous reports length maximal (μm) Karyotype architecture A. sphaerocephala (Gaudich.) Baker Billbergia horrida Regel var. tigrina hort. ex Baker B. pallidiflora Liebm. 19145 16800; originally collected in Espirito Santo, Brazil (s.n.) B. rosea Beer 16813 Bromelia antiacantha Bertol. Cryptanthus bahianus L.B.Sm. C. marginatus L.B.Sm. C. praetextus E.Morren ex Baker Brazil, Sao Paulo, Leg. 2n = 50* 2.14 0.17 ± Homogeneous 2n = 50* 0.85 ± 0.014 0.77 ± 0.005 2.14 0.89 ± Homogeneous 2n = 50* 1.03 ± 0.009 1.78 0.71 ± Homogeneous 2n = 50 2n = 50 2.50 0.71 ± Homogeneous Nr. Bro-003 W 1 2n = 100 2n =4x 1.07 0.53 ± Homogeneous Botanical Garden of the University Leipzig, Leg. Nr. F008 14242 15191 2n =34+ 1 3B 2n =34+ 1 4B, n =17 0.75 ± 0.068 1.78 0.71 ± Homogeneous 2n = 32* 1.26 ± 0.005 1.78 0.17 ± Homogeneous 2n 32 + 1 2B 2n = 34 1.35 ± 0.014 1.96 1.07 ± Homogeneous

BROMELIAD CHROMOSOMES AND DNA CONTENT 353 Edmundoa lindenii (Regel) Leme Fascicularia bicolor (Ruiz & Pav.) Mez ssp. bicolor E.C.Nelson & Zizka F. bicolor canaliculata E.C.Nelson & Zizka 16988; originally collected in Brazil 16846 17118 Greigia sp. 19040; originally collected in Cundinamarca, Colombia G. sphacelata (Ruiz & Pavon) Regel Neoglaziovia variegata Mez Ochagavia elegans R.Philippi Portea petropolitana Mez Quesnelia arvensis Mez 16855; originally collected Forest at Hualqui, Conception, Chile Brazil, Pernambuco, Camocin de São Félix, Leg. Nr. Bro-1121 16852 18000 Botanical Garden of the University Leipzig, Leg. Nr. ZG 245100 2n = 50 2n = 50 1.32 ± 0.010 2.14 1.07 ± Homogeneous 2n = 50 1.05 ± 0.010 1.47 0.59 Decreasing 2n = 50 1.12 ± 0.008 2.06 0.88 ± Homogeneous 2n = 50 1.35 ± 0.019 1.14 0.57 ± Homogeneous 2n = 50 1.56 ± 0.004 1.71 0.86 Decreasing 2n = 100 2n = 100 2.24 0.71 ± Homogeneous 2n = 50 1.12 ± 0.015 1.38 0.83 ± Homogeneous small, spheroid or rod shaped 2n = 50* 1.31 ± 0.011 1.75 0.75 ± Homogeneous 2n = 50 2n = 50 0.88 ± 0.001 2.14 0.71 ± Homogeneous

354 J. GITAÍ ET AL. Table 1. Continued Taxon Provenance and material number Chromosome numbers C-value Metaphase chromosome Present study Previous reports 2C (pg) ± SD 2C (pg) previous reports length maximal (μm) Karyotype architecture PITCAIRNIOIDEAE s.s. Deuterocohnia lorentziana (Mez) Spencer & Smith Encholirium spectabile Mart. Pitcairnia atrorubens (Beer) Baker P. breedlovei L.B.Sm. P. chiapensis Miranda 130007 (ex BG HD) Brazil, Pernambuco, Petrolandia, Leg. Nr. Bro-1304 16095; originally collected in Costa Rica Botanical Garden of the University Leipzig, Leg. Nr. 52598 19495; originally collected south from Tuxtla Gutiérrez, Chiapas, Mexico P. flammea Lindl. Brazil, Rio de Janeiro, Cabo Frio, Leg. Nr. Bro-1413 2n = 50 1.73 ± 0.021 2.29 1.14 Bimodal, with 6 2n 100 small and 19 larger pairs, mainly submeta- to metacentric 2n 100 Bimodal, with 1 larger pair 2n = 50 2n = 50 1.25 1.00 ± Homogeneous 2n = 50 1.29 ± 0.013 1.20 1.43 1.14 ± Homogeneous 2n = 48/ (2n 94)* 1.50 0.75 ± Homogeneous 2n = 50* 1.31 ± 0.005 1.22 1.50 0.75 ± Homogeneous 22n = 50/ (2n = 100) n =25 2n =50 1.28/1.44 1.25 0.75 ± Homogeneous

BROMELIAD CHROMOSOMES AND DNA CONTENT 355 P. macrochlamys Mez 12596 P. sceptrigera Mez Botanical Garden of the University Leipzig, Leg. Nr. F009 PUYOIDEAE Puya coerulea Miers Botanical Garden of the University Leipzig, Leg. Nr. F012 P. mirabilis (Mez) L.B.Sm. TILLANDSIOIDEAE Alcantarea brasiliana (L.B.Sm.) J.R.Grant Catopsis morreniana Mez Guzmania monostachia (L.) Rusby ex Mez var. variegata M.B.Foster Racinea ropalocarpa (André) M.A.Spencer & L.B.Sm. Tillandsia aff. aequatoriales L.B.Sm. 18507 Botanical Garden of the University Leipzig, Leg. Nr. F001 2891; originally collected in Mexico 1502 2652; originally collected in Peru 8809 22n = 50* 1.25 ± 0.014 1.20 1.75 1.00 ± Homogeneous 22n = 50/ (2n 100) n = 25 1.20 1.50 0.50 ± Homogeneous 2n = 50* 1.02 ± 0.013 1.50 0.75 ± Homogeneous 2n = 50 0.99 ± 0.010 0.88 1.71 0.86 Bimodal, with 2 pairs of larger chromosomes 2n = 50* 1.71 0.86 ± Homogeneous 2n = 50 2n 42 1.15 ± 0.013 1.78 0.71 Decreasing 2n =50 n =25 2n =48 1.18 ± 0.010 1.75 1.25 ± Homogeneous 2n = 50 n = 25 1.27 ± 0.022 2.00 1.00 ± Homogeneous 2n = 48* 1.75 0.75 Bimodal (10 small and 14 larger pairs)

356 J. GITAÍ ET AL. Table 1. Continued Taxon Provenance and material number Chromosome numbers C-value Metaphase chromosome Present study Previous reports 2C (pg) ± SD 2C (pg) previous reports length maximal (μm) Karyotype architecture T. argentina C.H.Wright 6322 T. bourgaei Baker 9962; originally collected in Ecuador T. brachycaulos Schltdl. T. cyanea E.Morren T. humilis C.Presl. Botanical Garden of the University Leipzig, Leg. Nr. AD 158/01 12612 867; originally collected in Peru T. lajensis André 1435; originally collected in Colombia T. latifolia Meyen Botanical Garden of the University Leipzig, Leg. Nr. 6263 T. mollis H. Hrom. & W.Till 872; originally collected east of Tarija, Peru 2n = 50* 1.50 0.50 Bimodal (6 small and 18 larger pairs) 2n = 50 n = 25 2.50 1.07 Bimodal (6 small and 19 larger pairs) 2n = 48* 1.51 ± 0.003 2.50 1.23 Bimodal (6 small and 18 larger pairs) 2n = 48 n = 25 2.21 3.00 1.25 Bimodal (6 small and 18 larger pairs) 2n = 48 n = 25 2.00 1.00 Bimodal (7 small and 18 larger pairs) 2n = 50* 2.50 1.23 Bimodal (6 small and 18 larger pairs) 2n = 48 n = 25 1.04 ± 0.020 1.50 0.75 Decreasing 2n = 50* 2.50 1.25 Bimodal (6 small and 18 larger pairs)

BROMELIAD CHROMOSOMES AND DNA CONTENT 357 T. pohliana Mez 2046; originally collected in Brazil T. polystachia L. Brazil, Pernambuco, IPA station, Caruaru, Leg. Nr. Bro-1287 T. streptophylla Scheidw. T. brachycaulos Schltdl. (putative hybrid) T. stricta Sol. var. stricta Botanical Garden of the University Leipzig, Leg. Nr. AD 216/01 1481 T. utriculata L. 2786; originally collected in Guatemala T. xiphioides Ker Gawl. Vriesea splendens (Brongn.) Lem. Werauhia viridiflora (Regel) J.R.Grant 1801; originally collected in Argentina 10796 10757 2n = 48* 3.00 1.30 Bimodal (7 small and 18 larger pairs) 2n = 50 n = 25 1.17 0.86 ± Homogeneous 2n = 50* 1.50 0.75 Decreasing 2n = 48* 1.20 2.50 1.00 ± Homogeneous 2n = 48 n = 25 2.50 1.30 Bimodal (6 small and 18 larger pairs) 2n = 48* 2.40 1.07 ± Homogeneous 2n =48 n =25 2n =48 2.23 ± 0.010 2.50 1.20 Bimodal (7 small and 17 larger pairs) 2n = 50* 1.75 1.25 ± Homogeneous *First report; Gitaí et al. (2005); See Supporting Information Table S1; See Supporting Information Table S2.

358 J. GITAÍ ET AL. samples). Sample preparation followed the standard two-step (Otto) protocol of Doležel et al. (2007), with an internal standard of Glycine max (L.) Merr. Polanka (Doležel, Doleželová & Novák, 1994). Isolated nuclei were stained with propidium iodide and the fluorescence intensities of 5000 particles (nuclei) were recorded. Sample/standard ratios were calculated from the means of the sample and standard fluorescence histograms, and only histograms with coefficients of variation (CVs) of less than 5% for the G 0/G 1 sample peak were considered. Event (nuclei) counts for both the samples and the internal standard were approximately the same. Three replicates of each sample were measured on different days to minimize potential random instrumental error. If the between-day variation in fluorescence intensity was more than 3%, the most extreme value was discarded and the sample was re-analysed. To assess the reproducibility between the two different instruments (Prague/Heidelberg), four samples were measured on both machines. LITERATURE REVIEW For the review of previously published chromosome numbers (Supporting Information Table S1), taxa were critically assessed and the classification system of Luther (2012) was followed, with some adjustments from The Plant List (2013). Additional genome data were extracted from the Plant DNA C-values database (Bennett & Leitch, 2012) and three publications (Sgorbati et al., 2004; Favoreto et al., 2012; Nunes et al., 2013; Supporting Information Table S2). STATISTICAL ANALYSES Statistical calculations were performed in R v2.15.1 (R Core Team, 2013). Because the variables did not fulfil the normality assumption (Shapiro Wilk test), differences in 2C-values between subfamilies and pairs of groups were assessed by the Kruskal Wallis test and non-parametric Wilcoxon rank sum test, respectively. Correlations between chromosome counts and 2C-values were assessed using nonparametric Spearman s rank correlation coefficient. The data were visualized using boxplots. RESULTS Altogether, we present new data for chromosome numbers, additional karyological features and/or 2C-values for 46 species representing five subfamilies: Brocchinioideae, Bromelioideae, Pitcairnioideae, Puyoideae and Tillandsioideae (Table 1, Figs 1, 2). For 22 species, chromosome counts are reported for the first time (Table S1). 2C-values are given for 28 taxa, 22 of which have not been studied previously [including eight genera: Catopsis Griseb., Fascicularia Mez, Greigia Regel, Guzmania Ruiz & Pav., Ochagavia Phil., Portea Brongn. ex K.Koch, Quesnelia Gaudich. and Racinaea M.A.Spencer & L.B.Sm.; Table S2]. The most frequent diploid chromosome number of 2n = 50 was found in 27 species analysed, followed by 2n = 48 in 10 taxa of Tillandsioideae (Table 1) [Tillandsia brachycaulos Schltdl., T. cyanea Linden ex K.Koch (Fig. 2A), T. humilis C.Presl, T. latifolia Meyen, T. pohliana Mez, T. stricta Sol. ex Ker Gawl. (Fig. 2E), T. utriculata L., T. xiphioides Ker Gawl., T. aff. aequatorialis L.B.Sm. and Vriesea splendens (Brongn.) Lem. (Fig. 2F)] and in one species of Pitcairnioideae, Pitcairnia breedlovei L.B.Sm. (Fig. 1N, O). For three species of Pitcairnioideae, root tips with several cells having doubled chromosome numbers (i.e. 2n = 100) were found (Fig. 1M). The lowest diploid numbers were found in Cryptanthus Otto & A.Dietr. (subfamily Bromelioideae): C. bahianus L.B.Sm. with 2n = 34 (Fig. 1F), C. marginatus L.B.Sm. with 2n = 32 and C. praetextus E.Morren ex Baker (Fig. 1G) with 2n = 32. Cryptanthus bahianus and C. praetextus had B chromosomes (one to three and one or two, respectively; Fig. 1F, G); they were not found in any other analysed species. Polyploidy was observed in Aechmea eurycorymbus Harms with 2n 88, Bromelia antiacantha Bertol. with 2n = 100 and Neoglaziovia variegata (Arruda) Mez with 2n = 100 (Fig. 1I), all belonging to Bromelioideae. Individual chromosome s varied from 3.00 to 0.17 μm (Table 1). Karyotypes with small chromosomes were observed in different subfamilies, e.g. in Bromelia antiacantha (Bromelioideae, 2n = 100, 0.53 1.07 μm), Pitcairnia sceptrigera Mez (Pitcairnioideae, 0.5 1.5 μm, Fig. 1M) and Tillandsia argentina C.H.Wright (Tillandsioideae, 0.5 1.5 μm). Tillandsia pohliana had the largest chromosomes (1.3 3.0 μm). In general, larger chromosomes were found in Tillandsioideae and smaller ones in Bromelioideae. The identification of constrictions (which allows a clear-cut characterization of chromosomal morphology) proved to be challenging because of the minute chromosome, and was successful in only a few cases. The presence of one pair of secondary constrictions was found in three species of Bromelioideae [Aechmea ornata (Gaudich.) Baker, A. sphaerocephala (Gaudich.) Baker and Billbergia rosea Beer, Fig. 1D], in four species of Pitcairnioideae {Encholirium spectabile Mart. ex Schult. & Schult.f., Pitcairnia flammea Lindl., P. macrochlamys Mez and P. sceptrigera, Fig. 1H, K, L, M} and in four samples of Tillandsioideae Racinea ropalocarpa (André) M.A.Spencer & L.B.Sm., T. cyanea, T. streptophylla Scheidw. T. brachycaulos and T. stricta, Fig. 2A, D, E). As a

BROMELIAD CHROMOSOMES AND DNA CONTENT 359 Figure 1. Mitotic chromosomes of Bromelioideae (A H) and Pitcairnioideae species (I O). A, Aechmea caudata (2n = 50); B, A. fendleri (2n = 50); C, A. ornata (2n = 50); D, A. sphaerocephala (2n = 50); E, Billbergia horrida var. tigrina (2n = 50); F, Cryptanthus bahianus (2n = 34 + 1 3B); G, C. praetextus (2n = 32 + 2B); H, Encholirium spectabile (2n = 50); I, Neoglaziovia variegata (2n = 100); J, Portea petropolitana (2n = 50); K, Pitcairnia flammea (2n = 50); L, P. macrochlamys (2n = 50); M, P. sceptrigera (2n = 100); N and O, P. breedlovei (2n = 48). Bar (in J and M) corresponds to 10 μm. The scale bar in J refers to all images, except in L, where the same scale bar as in M applies. Inserts at the top left part of D, H, K, L and M highlight the satellite chromosome pair. Arrows (in C and E) indicate secondary constrictions, whereas arrowheads (F and G) show B chromosomes. result of the larger chromosomes in Tillandsioideae, the morphology of some pairs was characterized as meta-, submeta- and subtelocentric (Fig. 2D). Most species possessed chromosomes with decreasing or similar s, especially in Bromelioideae, Puyoideae and Pitcairnioideae (Table 1) and some Tillandsioideae [Catopsis morreniana Mez, Guzmania monostachia (L.) Rusby ex Mez, Racinea ropalocarpa, T. streptophylla brachycaulos and T. polystachia (L.) L.]. Contrasting chromosome s within the same complement were detected in T. argentina, T. bourgaei Baker, T. brachycaulos, T. cyanea (Fig. 2A), T. streptophylla brachycaulos (Fig. 2D), T. humilis, T. latifolia, T. mollis H.Hrom. & W.Till, T. utriculata and Vriesea splendens, indicating a tendency of bimodality in this subfamily. In total, 28 taxa were investigated by flow cytometry, including nine samples studied by Gitaí et al. (2005) (Table 1). Mean 2C-values ranged from 0.85 pg in Billbergia horrida Regel var. tigrina Baker to 2.23 pg in Vriesea splendens. The genome of four samples analysed on both flow cytometers differed by a maximum of 2.4%. The CVs for the G 0/G 1 peak of all analysed samples ranged from 1.63 to 4.43. In tetraploid Aechmea eurycorymbus (2n 88), the genome (2C-value = 2.19 pg) was approximately twice that of the diploid members of the subfamily (2C-value = 0.64 1.56 pg, median = 1.04 pg). Furthermore, in samples in which two different chromosome numbers were detected, only one peak (ploidy) was recovered. When taking data from the literature into account, 2C-values in Bromeliaceae ranged from 0.64 to 3.34 pg (Fig. 3, Tables 2, S2) highlighting differences between subfamilies. The Kruskal Wallis test indicated significant differences in genome between Bromelioideae, Pitcairnioideae and Tillandsioideae (Kruskal Wallis chi-squared = 9.29, d.f. = 2, P = 0.010). (N.B. Puyoideae, Navioideae, Hechtioideae, Lindmanioideae and Brocchinioideae were omitted from the test because of insufficient data.)

360 J. GITAÍ ET AL. Figure 2. Mitotic chromosomes of Tillandsioideae species. A, Tillandsia cyanea (2n = 48); B, T. mollis (2n = 50); C, T. polystachia (2n = 50); D, T. streptophylla T. brachycaulos (2n = 50); E, T. stricta (2n = 48); F, Vriesea splendens (2n = 48). The abbreviations MT, SM and ST (in D) correspond to metacentric, submetacentric and subtelocentric chromosomes. Bar (in F, for all photographs) corresponds to 10 μm. DISCUSSION CHROMOSOME NUMBERS The majority of analysed species (29 taxa) exhibited 2n = 50, 14 of which were studied for the first time. Our counts confirmed the diploid level (2n = 50) for 12 previously studied taxa (Tables 1, S1). These data further confirm the suggestion that the basic chromosome number for Bromeliaceae is x = 25 (Marchant, 1967; Brown, Palací & Luther, 1997). The second most frequent number at the diploid level was 2n = 48, found in nine members of Tillandsioideae and in Pitcairnia breedlovei (Table 1). A count of 2n = 48 was reported previously for Ananas ananassoides (Baker) L.B.Sm., Bromelia alta L.B.Sm., Puya chilensis Molina, Guzmania monostachia, G. musaica (Linden & André) Mez and Vriesea splendens (Table S1). For V. splendens, our data confirmed the count of Doutreligne (1939) and Weiss (1965), but were in conflict with the meiotic count of Marchant (1967), whereas the report for G. monostachia of 2n = 48 (Gauthé, 1965) and the meiotic counts of McWilliams (1974) were not confirmed by our findings. Furthermore, in Tillandsia cyanea, T. humilis and T. utriculata, our counts (2n = 48) differed from the previously reported 2n = 50 or n = 25 based on meiotic bivalents (Brown, Varadarajan & Gilmartin, 1984; Brown & Gilmartin, 1989).

BROMELIAD CHROMOSOMES AND DNA CONTENT 361 Dominant 2n Ploidy level Genome [pg] Bromelioideae 34, 50 2x, 4x, 6x 2C = 0.64 2.19 (34) Puyoideae 50 2x 2C = 0.88 1.13 (5) Pitcairnioideae 50 2x, 4x, 6x 2C = 0.60 1.86 (42) Navioideae 2C = 1.42 (1) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Chromosome count discrepancies and conflicts could be attributed to three major factors: (1) taxonomic uncertainties and misidentifications; (2) difficulties in chromosome counting and the application of methods which may yield erroneous results; and (3) discrepancies between mitotic and meiotic counts. Bromeliaceae is a taxonomically challenging family and herbarium vouchers for material investigated in cytogenetic studies are often not available. Hence, straightforward comparisons between newer and older reports are often difficult because of possible misidentifications or inextricable synonymy (for conflicts, see Table S1). A large number of relatively small chromosomes and floral anatomical features (e.g. secondary wall thickenings of the anther endothecium in the case of meiotic counts) were identified as major inconveniences and possible sources of discrepancies between studies (e.g. Taylor, 1925; McWilliams, 1974; Till, 1984; Brown & Gilmartin, 1986; see also Table S1). Therefore, some authors provide just the ploidy (e.g. Till, 1992) or approximate counts (e.g. Marchant, 1967). Moreover, several discrepancies may arise when comparing newer data with older reports (e.g. Lindschau, 1933; Gauthé, 1965). Within our sampling, all four species of Billbergia Thunb. had 2n = 50, agreeing with the data of Cotias-de-Oliveira et al. (2000, 2004). These contrast with most previous counts for the genus, which deviated from the base number x = 25 (e.g. 2n = 54, 72 and 108; Lindschau, 1933; Table S1). However, it is noted that Lindschau (1933) mainly used the manual section technique, which should be considered with caution, as highlighted by McWilliams (1974), Brown & Gilmartin (1986) and Gitaí et al. (2005). The major concern regarding the section technique relies on the fact that Hechtioideae 50 2x 2C = 0.94 0.96 (2) Tillandsioideae 48, 50 2x, 4x 2C = 1.04 3.34 (12) Lindmanioideae Brocchinioideae 46 2x 2C = 0.76 0.92 (4) Figure 3. Dominant diploid chromosome number, ploidy level and genome estimate range for each subfamily of Bromeliaceae plotted on a phylogenetic tree adapted from Givnish et al. (2011). Numbers in parentheses after genome estimate range indicate the number of available data. metaphase chromosomes are often in different planes in three-dimensional nuclei. Thus, the evaluation of stained sections may lead to over- or underestimation of the chromosome number. Nevertheless, using a smear method (as used here), Matsuura & Sutô (1935) reported 2n = 54 for Quesnelia liboniana (DeJonghe) Mez and 2n = 108 for Billbergia sp., and there are also reports of 2n = 54 for Neoregelia L.B.Sm. and Nidularium Lem. (Table S1). In addition, Gauthé (1965) confirmed several results of Lindschau (1933). Thus, to be certain about variable basic chromosome numbers in these genera, additional studies are required. Discrepancies between mitotic and meiotic counts have been discussed previously by Brown & Gilmartin (1986), and have been attributed to a possible higher tolerance against somatic chromosome number alterations in primarily mechanical root. In addition, meiotic studies carried out by Palma-Silva et al. 2004; 2008) reported meiotic irregularities in Vriesea and Aechmea, including irregular pairing and observations of univalents, which may lead to a misinterpretation of the bivalent number during meiosis, especially in taxa with small chromosomes. To summarize, on the basis of our results and previous reports, we confirm that the basic chromosome number for Bromeliaceae is x = 25, with the most common diploid number being 2n = 50, followed by occasional occurrences of 2n = 48 found in Tillandsioideae and possibly other subfamilies (Table 2). GENOME SIZE To examine the relationship between ploidy and genome, chromosome numbers were combined with available 2C-value data. After excluding of

362 J. GITAÍ ET AL. Table 2. Overview of the chromosome numbers and 2C-values for the genera in Bromeliaceae. Subfamily Bromelioideae is divided into the early-diverging and core groups according to Silvestro et al. (2014). N sp, number of species in the genus according to Luther (2012); N chr, number of chromosome counts for the genus; 2n, reliable chromosome counts (B, B chromosomes; numbers in parentheses represent rare counts; approximate counts are indicated by ); N2C, number of species with genome estimates Subfamily Genus N sp N chr 2n N2C Range of 2Cvalues reported Mean Brocchinioideae Brocchinia 20 1 46 3 0.76 0.92 0.83 Bromelioideae core Acanthostachys 2 1 50 Aechmea 276 31 50/ 100 5 0.78 2.19 1.33 Androlepis 2 Araeococcus 9 3 50 Billbergia 63 12 50 5 0.75 1.03 0.87 Canistropsis 11 4 50 1 1.00 Canistrum 13 2 50 Edmundoa 3 1 50 1 1.32 Eduandrea 1 Hohenbergia 65 8 50 B Hohenbergiopsis 1 Hohenmea 1 Lymania 9 2 50 Neoglaziovia 3 1 100 Neoregelia 120 20 (46)/50 1 0.98 Nidularium 47 9 50 Niduregelia 3 1 50 Portea 9 1 1.31 Pseudoaechmea 1 Quesnelia 23 3 50 1 0.88 Ronnbergia 14 Ursulaea 2 Wittrockia 6 2 50 Bromelioideae early diverging Ananas 7 5 (48)/50/75/100 2 0.92 1.09 1.00 Bromelia 60 6 (48)/50/ 100/ 150 B 1 0.81 Cryptanthus 72 13 34/36 B 6 0.71 1.46 1.18 Deinacanthon 1 1 160 Disteganthus 2 Fascicularia 1 1 50 1 1.05 1.12 1.09 Fernseea 2 Greigia 36 2 50 1 1.35 1.56 1.46 Lapanthus 2 1 50 Ochagavia 4 2 50 1 1.12 Orthophytum 68 11 50/100/150 1 0.64 Pseudananas 1 1 100 1 1.00 Hechtioideae Hechtia 62 2 50 2 0.94 0.96 0.95 Lindmanioideae Connelia 6 Lindmania 39 Navioideae Brewcaria 6 Cottendorfia 1 Navia 93 1 1.42 Sequencia 1 Steyerbromelia 6

BROMELIAD CHROMOSOMES AND DNA CONTENT 363 Table 2. Continued Subfamily Genus N sp N chr 2n N2C Range of 2Cvalues reported Mean Pitcairnioideae Deuterocohnia 18 3 50/100 3 0.74 1.73 1.09 Dyckia 147 11 50/100 2 1.58 1.60 1.59 Encholirium 28 1 50 1 1.74 Fosterella 31 4 50/100/150 2 1.86 1.86 Pepinia 57 5 50 1 1.24 Pitcairnia 342 31 50 31 0.60 1.44 1.18 Puyoideae Puya 218 12 50 4 0.88 1.13 0.99 Tillandsioideae Alcantarea 32 4 50 Catopsis 18 6 50 1 1.15 Glomeropitcairnia 2 1 50 Guzmania 211 10 48/50/(98) 1 1.18 Mezobromelia 9 Racinea 74 6 50 1 1.27 Tillandsia 622 91 48/50/ 100 B 6 1.04 3.34 1.97 Vriesea 281 29 50 3 1.11 2.23 1.51 Werauhia 88 4 50 several outliers [Bromelia antiacantha, Cryptanthus spp., Deuterocohnia lorentziana (Mez) M.A.Spencer & L.B.Sm.] as a result of possible intraspecific ploidy variation and/or the presence of B chromosomes, the correlation analysis showed a weak positive linear relationship (Spearman s rho = 0.23, P = 0.010). However, the weak relationship was caused mainly by extreme values of Tillandsioideae. Hence, we assume that, at least for Brocchinioideae, Bromelioideae, Pitcairnioideae and Puyoideae, counts of 2n = 46/50 are associated with 2C-values in the range 0.64 1.56 pg (Table S2). Accordingly, in tetraploid Aechmea eurycorymbus, 2C = 2.19 pg was approximately twice that of its diploid congeners (2C = 1.08 1.25 pg for Aechmea). These results agree with previous observations showing that the total length of the chromosome complement was positively correlated with the increase or decrease in chromosome number (Sharma & Gosh, 1971). Tillandsioideae stands out in terms of genome and chromosome as this subfamily has the largest range (2C = 1.04 3.34 pg, Fig. 3, Table 2) and largest genomes and chromosomes (see also Weiss, 1965) in Bromeliaceae. The 2C-values in this subfamily showed a distinct trend towards being significantly different from Bromelioideae and Pitcairnioideae (t = 2.58, d.f. = 12.40, P = 0.023 and t = 1.96, d.f. = 11.80, P = 0.0741, respectively). DYSPLOIDY The observation of 2n = 48 and of haploid numbers N = 18, 20, 21 and 22 (Marchant, 1967; Brown et al., 1984; Brown & Gilmartin, 1989; Table S1) in Tillandsioideae (especially in Tillandsia) and several confirmed counts for Cryptanthus (2n = 32 to 36) indicate that dysploidy (erroneously often referred to as aneuploidy) may represent an important evolutionary process. Dysploidy is an increase or decrease in chromosome number as a result of structural rearrangements; it differs from the non-balanced gain or loss of whole chromosomes (known as aneuploidy), and is considered to be a relatively frequent cytoevolutionary strategy in Neotropical plant groups (e.g. Benko-Iseppon & Wanderley, 2002; Souza & Benko-Iseppon, 2004; Salles-de-Melo et al., 2010). In Bromeliaceae, genome estimates (Tables 1, S2) provide support for the dysploidy hypothesis as, in Cryptanthus (2n = 32 and 34), the 2C-values were not significantly different (Ramírez-Morillo & Brown, 2001; this study: Wilcoxon rank sum test, W = 114.5, P = 0.311) from those in other members of Bromelioideae with 2n = 50 (Table S2). Bellintani et al. (2005) pointed out that the larger chromosomes of Cryptanthus compared with other Bromelioideae species reflected the impact of dysploidy, and this is also supported by the comparable total chromosome lengths between Cryptanthus and other diploid species of Bromeliaceae (Sharma & Gosh, 1971). As a result of its distinct chromosome number, Cryptanthus was once considered a common ancestor for the whole family Bromeliaceae (Brown & Gilmartin, 1989). However, its phylogenetic position is in the early-diverging eu-bromelioids (Schulte et al., 2005; Schulte, Barfuss & Zizka, 2009), with other genera associated with x = 25 (including the sister genus

364 J. GITAÍ ET AL. Orthophytum Beer, Table S1). Hence, Cryptanthus can be considered a product of descending dysploidy from x = 25 as proposed in the alternative hypothesis (Brown & Gilmartin, 1989; Ramírez-Morillo & Brown, 2001). Descending dysploidy may also be considered for other genera in Bromelioideae and Pitcairnioideae, and for Puya Molina (Tschischow, 1956) and Brocchinia Schult. & Schult.f. (Gitaí et al., 2005), but seems to be particularly important in Tillandsioideae, as several species with 2n = 48 (see also Till, 1984, 1992) and available 2C-values show a genome comparable with taxa with 2n = 50. In this case, possibly a single chromosome pair underwent structural rearrangements, involving tandem fusion or reciprocal translocations, leading to a decrease in chromosome number. Such fusion and fission processes were recognized by Jones (1998) and may be accompanied by duplication and inversions of chromosome arms. An example of this process was demonstrated in Cephalanthera Rich. (Orchidaceae), which includes species with 2n = 32, 36 or 44, which have originated by tandem fusions, resulting in the presence of interstitial telomeres (Moscone et al., 2007). In the Triticeae (Poaceae), a detailed analysis of syntenic regions by Luo et al. (2009) has shown how the basic number x = 7 has been derived from x = 12 in the ancestral species (represented by rice and sorghum), not through end-to-end chromosome fusions or translocations and loss of microchromosomes, but by the insertion of four whole chromosomes into breaks in the centromeric region of four other chromosomes, with a further fifth fusion and translocation event. However, to fully support dysploidy as an important evolutionary mechanism in Bromeliaceae, chromosome banding techniques and/or in situ hybridization, which enables particular chromosome rearrangements to be tracked, need to be employed. BIMODALITY Bimodal karyotypes were observed in diploid species of 12 taxa of Tillandsia (Table 1). In contrast with Marchant (1967), who considered bimodality to be absent in Pitcairnioideae, bimodal karyotypes were also observed in Deuterocohnia lorentziana (Gitaí et al., 2005) and in Puyoideae [Puya mirabilis (Mez) L.B.Sm.], suggesting that this type of karyotype organization is not exclusive to Tillandsioideae. It has been suggested that bimodal karyotypes can originate following hybridization between parental taxa with chromosomes of contrasting s (Greilhuber, 1995). Analysis of the chromosomes of the intergeneric F1 hybrid Cryptbergia meadii hort. (= Cryptanthus beuckeri E.Morren Billbergia nutans H.Wendl. ex Regel), with 2n = 42, Marchant (1967) recognized two groups of chromosomes with contrasting s and observed that they had a tendency to remain separated during meiosis. This separation was more evident during chromosome segregation, as the larger chromosomes migrated before the smaller ones during anaphase. In addition to whole genome associations, some chromosome rearrangements (e.g. tandem fusions, translocations), sometimes associated with dysploidy, may lead to the genesis of some larger chromosome pairs, as observed for Cephalanthera (Moscone et al., 2007). POLYPLOIDY Polyploidy is considered to be by far the most common cause of chromosome number and karyotype variation in the evolution of plants, and consists of the duplication of an entire chromosome complement (e.g. Grant, 1981; Guerra, 2008; Heslop-Harrison & Schwarzacher, 2011). In the present study, polyploidy occurred in three members of Bromelioideae: Aechmea eurycorymbus (2n 88), Bromelia antiacantha (2n 100) and Neoglaziovia variegata (2n 100). In combination with previously published data, polyploidy has been reported in Bromelioideae (mainly early-diverging bromelioids: Ananas Mill., Bromelia L., Deinacanthon Mez, Orthophytum Beer), Pitcairnioideae (Deuterocohnia Mez, Dyckia Schult. & Schult.f and Fosterella L.B.Sm.) and Tillandsioideae (mainly Tillandsia subgenus Diaphoranthema Beer) (Tables 2, S1). In core Bromelioideae (sensu Silvestro, Zizka & Schulte, 2014), polyploid counts were also reported previously for Neoglaziovia Mez (Cotias-de-Oliveira et al., 2000, 2004) and possibly Billbergia (Matsuura & Sutô, 1935). The distribution of chromosome numbers in the phylogenetic tree of Bromeliaceae (Fig. 3) suggests that polyploidy has occurred several times independently and, in Bromelioideae, the more frequent occurrence of polyploidy in the earlier diverging lineages (Table 2) suggests that phylogenetic age and habitat might be the relevant factors. Older phylogenetic lineages not only have more time for stabilized polyploids to evolve, but may also have developed larger interspecific genetic differences which may promote the stabilization of allopolyploids following hybridization (Paun et al., 2009). Certainly, according to recent phylogenetic evidence, core bromelioids are a comparatively young group, which underwent a rapid radiation during the last 9 Myr (Givnish et al., 2011). Although hybridization may occur at a similar rate as in the earlier diverging lineages, in core bromelioids it may not lead to the formation of stable polyploids as frequently, because their genomes are still sufficiently similar to enable chromosome pairing.

BROMELIAD CHROMOSOMES AND DNA CONTENT 365 Recent molecular phylogenetic studies have revealed that most early-diverging bromelioids are characterized by a terrestrial or lithophytic habitat and an absence of a tank habit (Schulte et al., 2009). They are often highly xeromorphic and the majority of species occur in harsh xeric environments, such as the Brazilian cerrados and restingas, especially the polyploids (e.g. Bromelia, Deinacanthon, Orthophytum, Ananas, Pseudananas Hassl. ex Harms; Smith & Downs, 1979). In contrast, the core bromelioids, in which only a few polyploids have been reported (Table S1), are characterized by a central tank that allows the plant to collect water externally and absorb it via leaf trichomes. This strategy facilitates the epiphytic life form (Schulte et al., 2009; Silvestro et al., 2014) and is often found in more mesic habitats, such as the Atlantic rainforest of south-eastern Brazil (Smith & Downs, 1979). Such an infrageneric ploidy diversification pattern has also been reported in Orthophytum, in which the polyploid species are closely associated with xeric microhabitats and possess xeromorphic traits (e.g. coriaceous, densely lepidote leaves), whereas the majority of diploid taxa are found in more mesic to wet microhabitats (Louzada et al., 2010). The higher general number of heterozygous alleles in polyploids compared with their diploid progenitors may contribute to a higher physiological plasticity and the development of new morphological features, and thus can increase the tolerance to harsh environmental conditions (Levin, 1983). ENDOPOLYPLOIDY Individuals bearing root tip cells with two different ploidies were observed in Pitcairnia flammea (2n = 50/2n 100), P. sceptrigera (2n = 50/2n 100) and P. breedlovei (2n = 48/2n 94). This could be attributed to endoreduplication (endopolyploidization), a process whereby the mitotic DNA replication in somatic cells is not followed by cell division. Endopolyploidization is considered to be a mechanism that can facilitate plant growth under certain conditions (Barow & Meister, 2003) and has been reported previously in root cells of several monocots, such as cultivars of Hosta Tratt. (Asparagaceae) (Zonneveld & Van Iren, 2000), species of Dendrobium Sw. (Orchidaceae) (Jones & Kuehnle, 1998) and species of Allium L. (Amaryllidaceae) (Barow & Meister, 2003), and across a phylogenetically diverse set of angiosperms (Barow & Meister, 2003, Bainard et al., 2012). However, we assume that this feature in bromeliads was most probably induced by specific plant growth conditions under glasshouse cultivation as observed, for example, in Solanum tuberosum L. (Solanaceae) (Uijtewaal, 1987). B CHROMOSOMES B chromosomes were observed in Bromelia karatas L., Hohenbergia aff. utriculosa Ule (Cotias-de- Oliveira et al., 2000), Cryptanthus bahianus and C. praetextus (Cotias-de-Oliveira et al., 2000; this study). This contrasts with previous studies in C. praetextus (Sharma & Gosh, 1971) and other Cryptanthus, Bromelia and Hohenbergia spp., where no B chromosomes were found (Table S1). Moreover, accessions of Tillandsia polystachia of different geographical origins had zero to six B chromosomes (Brown & Gilmartin, 1989). It has been suggested (e.g. Jones & Houben, 2003; Jones, 2012) that B chromosomes can arise from errors during meiosis that generate A-chromosome fragments. Therefore, the occurrence of B chromosomes in dysploid Cryptanthus perhaps indicates ongoing speciation processes closely associated with chromosome rearrangements. In addition, the majority of Cryptanthus spp. are narrow endemics (Krapp et al., 2013), for which B chromosomes may contribute to favourable phenotypes or be associated with particular habitats (reviewed by Jones & Houben, 2003). CONCLUSIONS Chromosome evolution in Bromeliaceae involves two main mechanisms: polyploidy and dysploidy. The combination of new and previously published data revealed that polyploidy occurs in subfamilies Bromelioideae, Pitcairnioideae and Tillandsioideae. Polyploidy in Bromelioideae is more prevalent in the early-diverging groups than in core bromelioids. With the exception of Tillandsioideae, ploidy is positively correlated with genome. However, to estimate ploidy based on the genome estimations, we suggest that a chromosome count is made for at least one species per genus. Dysploidy is suggested to be the mechanism responsible for the derived chromosome numbers 2n = 32, 34 (Cryptanthus) and 2n = 48 (frequent in Tillandsia). In addition, the occurrence of B chromosomes in dysploid Cryptanthus suggests ongoing speciation processes closely associated with chromosome rearrangements. Some additional features were observed in taxa of Tillandsioideae, such as a tendency towards bimodal karyotypes, the largest chromosomes and the largest genomes. Future directions of karyological research in Bromeliaceae should focus on the early-diverging Bromeliaceae and subfamilies in which chromosome data and C-values are virtually lacking (e.g. Navioideae, Lindmanioideae). To understand the origin and evolutionary significance of dysploidy, in situ hybridization methods should also be employed.

366 J. GITAÍ ET AL. ACKNOWLEDGEMENTS In Memoriam Karl-Heinz Schulmeyer (1955 2013), gardener and custodian of the bromeliad collection of the Frankfurt, and Wilfried Morawetz (1951 2007), former Director of the Botanical Garden of the University Leipzig. The authors gratefully acknowledge the Frankfurt for permission to access their collections. We are thankful to T. Urfus (Department of Botany, Charles University, Prague) and M. A. Koch (Centre for Organismal Studies, Heidelberg) for enabling us to carry out the flow cytometric measurements, as well as to anonymous reviewers for their valuable comments. This study was supported by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil), DAAD (German Academic Exchange Service), CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brazil), UNESCO (United Nations Educational, Scientific and Cultural Organization) Fellowship for Molecular Biology and the Deutsche Forschungsgemeinschaft (DFG, Zi557/6-2). The authors declare no conflicts of interest. REFERENCES Bainard JD, Bainard LD, Henry TA, Fazekas AJ, Newmaster SG. 2012. A multivariate analysis of variation in genome and endoreduplication in angiosperms reveals strong phylogenetic signal and association with phenotypic traits. New Phytologist 196: 1240 1250. Barfuss MHJ, 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. Barow M, Meister A. 2003. Endopolyploidy in seed plants is differently correlated to systematics, organ, life strategy and genome. Plant Cell and Environment 26: 571 584. Bellintani MC, Assis JGA, Cotias-de-Oliveira ALP. 2005. Chromosomal evolution of Bromeliaceae. Cytologia 70: 129 133. Benko-Iseppon AM, Morawetz W. 2000. Viburnales: cytological features and a new circumscription. Taxon 49: 5 16. Benko-Iseppon AM, Wanderley MGL. 2002. Cytogenetic studies on Brazilian Xyris species (Xyridaceae). Botanical Journal of the Linnean Society 138: 245 252. Bennett MD, Leitch IJ. 2012. Plant DNA C-values database. Release 6.0. Available at: http://data.kew.org/cvalues [accessed 10 January 2013]. Benzing DH. 2000. Bromeliaceae: profile of an adaptive radiation. Cambridge: Cambridge University Press. Brown GK, Gilmartin AJ. 1986. Chromosomes of the Bromeliaceae. Selbyana 9: 88 93. Brown GK, Gilmartin AJ. 1989. Chromosome numbers in Bromeliaceae. American Journal of Botany 76: 657 665. Brown GK, Palací CA, Luther HE. 1997. Chromosome numbers in Bromeliaceae. Selbyana 18: 85 88. Brown GK, Varadarajan GS, Gilmartin AJ. 1984. Chromosome numbers reports LXXXV. Bromeliaceae. Taxon 33: 756 760. Cotias-de-Oliveira ALP, Assis JGA, Bellintani MC, Andrade JC, Guedes MLS. 2000. Chromosome numbers in Bromeliaceae. Genetics and Molecular Biology 23: 173 177. Cotias-de-Oliveira AL, Assis JGA, Ceita GO, Palmeira ACL, Guedes MLS. 2004. Chromosome numbers for Bromeliaceae species occurring in Brazil. Cytologia 69: 161 166. Crayn DM, Winter K, Smith A. 2004. Multiple origins of crassulacean acid metabolism and the epiphytic habit in the Neotropical family Bromeliaceae. Proceedings of the National Academy of Sciences of the United States of America 101: 3703 3708. Doležel J, Doleželová M, Novák FJ. 1994. Flow cytometric estimation of nuclear DNA amount in diploid bananas (Musa acuminata and M. balbisiana). Biologia Plantarum 36: 351 357. Doležel J, Greilhuber J, Suda J. 2007. Estimation of nuclear DNA content in plants using flow cytometry. Nature Protocols 2: 2233 2244. Doutreligne J. 1939. Les divers types de structure nucléaire et de mitose somatique chez les Phanérogames. La Cellule 48: 191 215. Favoreto FC, Carvalho CR, Lima ABP, Ferreira A, Clarindo WR. 2012. Genome and base composition of Bromeliaceae species assessed by flow cytometry. Plant Systematics and Evolution 298: 1185 1193. Gauthé J. 1965. Contribution a l étude caryologique des Tillandsiées. Mémories du Muséum National d Histoire Naturelle 16: 39 59. Gitaí J, Horres R, Benko-Iseppon AM. 2005. Chromosome features and evolution of Bromeliaceae. Plant Systematics and Evolution 253: 65 80. Givnish TJ, Barfuss MHJ, Van Ee B, Riina R, Schulte K, Horres R, Gonsiska PA, Jabaily RS, Crayn DM, Smith JA, 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, Berry PE, Systsma KJ. 2007. Phylogeny, adaptive radiation, and historical biogeography of Bromeliaceae inferred from ndhf sequence data. Aliso 23: 3 26. Givnish TJ, Millam KC, Evans M, Hall JC, Pires JC, Berry PE, Systsma KJ. 2004. Ancient vicariance or recent long-distance dispersal? Inferences about phylogeny and South American African disjunctions in Rapateaceae and Bromeliaceae based on ndhf sequence data. International Journal of Plant Sciences 165: 35 54. Grant V. 1981. Plant speciation, 2nd edn. New York: Columbia University Press. Greilhuber J. 1995. Chromosomes of the monocotyledons (general aspects). In: Rudall PJ, Cribb PJ, Cutler DF, Humphries CJ, eds. Monocotyledons: systematics and evolution, volume 2. Kew: Royal Botanic Gardens, 379 414.