Chromosome-Numbers in Bromeliaceae

Eighty-three chromosome counts are reported for 72 taxa of the Bromeliaceae. Fifty-eight of these counts are the first known chromosome number reports for their respective taxa. A model of chromosomal evolution in the Bromeliaceae (n = 25) is presented. The model is parsimonious and consistent with existing data on meiotic chromosome numbers within the family and in the closely related Velloziaceae (n = 9). Two hypothesized paleodiploids (n = 8 and n = 9) hybridized to form a tetraploid that in turn hybridized with the n = 8 lineage. The resultant n = 25 is the extant base number for the family. Two alternative hypotheses could explain the unique extant base number (n = 17) for Cryptanthus: 1) Cryptanthus represents the paleotetraploid level, i.e., prior to the second round of hybridization, or 2) the lower number represents the result of a more recent series of aneuploid reductions from n = 25. Given the existence of intergeneric hybrids involving Cryptanthus, aneuploid reduction is the more likely interpretation. RECENT RESEARCH concerning Bromeliaceae systematics and evolution (e.g., Brown and Gilmartin, 1984, 1986; Gilmartin and Brown, 1985, 1986b) has sparked renewed interest in the study of Bromeliaceae chromosomes and chromosome evolution. Past chromosome number surveys in the family (i.e., Lindschau, 1933; Gauthe, 1965; Weiss, 1965; Sharma and Ghosh, 1971; Till, 1984) have relied mostly on mitotic material. The only major exception to this was Marchant (1967) who utilized meiotically active microsporocytes. There is great variability in reported mitotic chromosome numbers (Brown and Gilmartin, 1986), and lack of concordance between mitotic and meiotic numbers for some taxa within the family. This variability in mitotic number is reflected in the variable interpretations of chromosome base numbers for the family. Brown and Gilmartin (1986) summarized the previous controversy over base number determination for Bromeliaceae, and discussed the current level of knowledge concerning poly' Received for publication 13 October 1987; revision accepted 28 October 1988. We thank a dedicated group of field collaborators, without whom this project would not have been possible: James Ackerman, Puerto Rico; Stephan Beck, Bolivia; Olga Benavides, Colombia; Elizabeth Bravo, Ecuador; David Brunner, Paraguay; I. Chacon, Costa Rica; Hermes Cuadros, Colombia; Linda Escobar, Colombia; Gert Hatschbach, Brazil; Stephen Koch, Mexico; Gustavo Martinelli, Brazil; Fernando Ortiz, Ecuador; Isidoro Sanchez Vega, Peru; and Rosa Subils, Argentina. Expert technical help was supplied by Carol Annable. We thank Ron Hartman, Don Hauber, and two anonymous reviewers for their comments. This work was supported by collaborative research grants BSR-8607 187 (GKB) and BSR-8407573 (AJG) from the National Science Foundation. 3 Deceased 10 February 1989. ploidy, chromosome size bimodality, and the correlation of nonconcordance in meiotic and mitotic chromosome numbers with the epiphytic mode of growth. The purpose of this paper is to describe results of an ongoing meiotic chromosome number survey within the Bromeliaceae, and especially subfamily Tillandsioideae. We also present a model for chromosome base number evolution for the family that is consistent and parsimonious with existing data. METHODS AND MATERIALSFloral buds were collected in the field, or obtained from cultivated material at Marie Selby Botanical Gardens, Sarasota, Florida (SEL). Buds were fixed in field-mixed Farmer's solution (100% EtOH: glacial acetic acid; 3: l/v:v) to which a drop of saturated aqueous ferric chloride (FeCl3-6H20) had been added. The latter enhances chromosome stainability. After a minimum of 24 hr, fixed buds are transferred to 70% EtOH. See Gilmartin and Brown (1986a) for a complete description of the field collaborator network and its operation. For chromosome squash preparations, individual anthers were removed from the bud in 70% EtOH and transferred to a pool of 1% acetic carmine on a microscope slide. While in the stain, the anther is cut transversely in half. Using ultrafine-tipped needle and forceps, the sporogenous masses are squeezed from each microsporangium through the median transverse cut. The sporogenous masses are positioned toward the center of the stain pool and a coverslip and gentle finger pressure are applied. The preparation is further flattened by passing the slide through an alcohol flame sev-

ploidy, chromosome size bimodality, and the correlation of nonconcordance in meiotic and mitotic chromosome numbers with the epiphytic mode of growth.
The purpose of this paper is to describe results of an ongoing meiotic chromosome number survey within the Bromeliaceae, and especially subfamily Tillandsioideae.We also present a model for chromosome base number evolution for the family that is consistent and parsimonious with existing data.
METHODS AND MATERIALS-Floral buds were collected in the field, or obtained from cultivated material at Marie Selby Botanical Gardens, Sarasota, Florida (SEL).Buds were fixed in field-mixed Farmer's solution (100% EtOH: glacial acetic acid; 3: l/v:v) to which a drop of saturated aqueous ferric chloride (FeCl3-6H20) had been added.The latter enhances chromosome stainability.After a minimum of 24 hr, fixed buds are transferred to 70% EtOH.See Gilmartin and Brown (1986a) for a complete description of the field collaborator network and its operation.
For chromosome squash preparations, individual anthers were removed from the bud in 70% EtOH and transferred to a pool of 1% acetic carmine on a microscope slide.While in the stain, the anther is cut transversely in half.Using ultrafine-tipped needle and forceps, the sporogenous masses are squeezed from each microsporangium through the median transverse cut.The sporogenous masses are positioned toward the center of the stain pool and a coverslip and gentle finger pressure are applied.The preparation is further flattened by passing the slide through an alcohol flame sev-657 [Vol.76 eral times.Heating the slide helps to rupture the callose that encapsulates the microsporocyte.Squashes were examined with phase contrast microscopy, and documented using Kodak Technical Pan 2415 and drawings.As standard practice, a minimum of five microsporocytes with unambiguous meiotic figures (usually in diplotene, diakinesis, metaphase I, or metaphase II) serve as the basis for chromosome number determination.In cases where a new chromosome number (e.g., Tillandsia leiboldiana) or abnormality (e.g., B-chromosomes or fragments) was encountered, as many as 18 unambiguous meiotic figures were documented.Voucher herbarium specimens are at WS unless otherwise indicated (see Table 1).All graphic documentation of chromosomes is at RM.The nomenclature followed here is that of Smith and Downs (1974, 1977, 1979).RESULTS-Eighty-three chromosome counts are reported for 72 taxa (Table 1).For the most part, reports are either the first known published chromosome number, or represent a previously unreported number for a taxon.Representative squash preparations are shown in Fig. 1-4 Except for Cryptanthus and Aechmea tillandsioides (Martius ex Schultes f.) Baker, all Bromelioideae genera thus far studied appear to have a meiotically established extant base number of x = 25 (see Brown and Gilmartin, 1986).Cryptanthus is anomalous in having a base of x = 17 (Marchant, 1967) and the possible significance of this is discussed later.Aechmea tillandsioides (n = 21; Marchant, 1967) would appear to be an aneuploid derivative.
Pitcairnioideae-Chromosome counts are available for six pitcairnioid genera (Deuterocohnia, Dyckia, Fosterella, Hechtia, Pitcairnia, Puya).Members of the subfamily are thus far homogeneous for the base number of x = 25 (also see Brown and Gilmartin, 1986).All repeat chromosome number reports made here for Pitcairnioideae taxa corroborate one or more previous counts (i.e., Lindschau, 1933;Di Fulvio, 1967;Marchant, 1967;Brown et al., 1984) Tillandsiafasciculata is a polymorphic species that includes at least ten varieties (Smith and Downs, 1977).Its geographical range includes Mexico, Central America, Florida, Caribbean Islands, and northern South America.All varieties are epiphytic, with two (laxispica and venosispica) that are sometimes saxicolous.
A similar lack of concordance between meiotic and mitotic chromosome numbers is encountered in T. imperialis.Gauthe (1965) reported a mitotic 2n = 64, while we report a gametic number of n = 25.Tillandsia imperialis typically is epiphytic in forests and may be saxicolous (Smith and Downs, 1977).
The first meiotic and diploid level count for T. juncea is reported here.Both Lindschau (1933) and Gauthe (1965) reported (as T. juncifolia) 2n = 96 for this species.DISCUSSION-The chromosome numbers presented here continue to support Marchant's (1967) proposal that x = 25 is the base number for the family.We view this as the extant base number, since, according to criteria put forth by Grant (1963Grant ( , 1981)) For the interpretation of chromosome number evolution, the use of somatically determined (holdfast-root tips) chromosome number reports (e.g., Lindschau, 1933;Gauthe, 1965;Weiss, 1965;Sharma and Ghosh, 1971) that do not agree with meiotically determined numbers would seem to be an unwise practice.Whether apparent meiotic and mitotic chromosome number nonconcordance in epiphytic bromeliads is real (e.g., unstable B-chromosomes), an artifact, or is the resultant combination of bias and/or inaccurate observation by some previous workers has not yet been clearly established.Research is now in progress to examine this problem.At present, we feel that the most prudent approach is to base our interpretations and hypotheses about chromosome evolution for the Bromeliaceae on meiotic chromosome numbers (e.g., Marchant, 1967;Brown and Gilmartin, 1983;Brown et al., 1984;Varadarajan and Brown, 1985; the reports here).

The first two chromosome number reports for T. polystachia merit mention because of B-chromosomes found in the Colombian collections (Escobar et al., 3659). The number of B-chromosomes varied between floral buds from an observed low number of two, to as many as six
. With the addition of new genera Brewcaria and Steyerbromelia published since Smith and Downs (1974), Pitcairnioideae contains 15 genera.Chromosome data are lacking for the following genera: Abromeitiella (2 spp.), Ayensua (1 sp.), Brewcaria (1 sp.), Brocchinia (18 spp.), Connellia (4 spp.), Cottendorfia (24 spp.), Encholirium (12 spp.), Navia (74 spp.), and Steyerbromelia (1 sp.).Tillandsioideae-With over 800 species in six genera (Catopsis, 19 spp.; Glomeropitcairnia, 2 spp.; Guzmania, 126 spp.; Mezobromelia, 3 spp.;Tillandsia, 410 spp.; Vriesea, 250 spp.), this is the largest of the subfamilies.Published chromosome number information now is available for all but Mezobromelia.The primary focus of this research is Tillandsia, the largest genus in the family.A comparison of published chromosome number data for Tillandsia (Brown and Gilmartin, 1986) has revealed a striking discrepancy between mitotic (root tip) and meiotic chromosome numbers.Prior attempts to discover trends of chromosomal evolution within Tillandsioideae (e.g., Lindschau, 1933; Gauthe, 1965) had been hindered by this variability in mitotic chromosome number reports (see Brown and Gilmartin, 1986, for additional discussion).An explicit goal of our research has been to determine the level of meiotic chromosome number variability within Tillandsia.The count of n = 22 for T. complanata (subg.Allardtia) differs from an earlier report (dysploidy from an ancestral n = 25.We view such chromosome numbers as being derived.The two reports for T. complanata and the remaining 13 reports for species of the subg.Allardtia presented here are the only known chromosome number reports for this subgenus.The chromosome number reports for T. lorentziana and T. vernicosa (n = 25) are the third and fourth counts for subgenus Anoplophytum.The earlier reports are for T. aeranthos (2n = 64; Gauthe, 1965, reported as T. dianthoidea Rossi) and T. tenuifolia L. (n = 25 + 1-2 fragments; Marchant, 1967).For the first time, tetraploidy in Tillandsia is documented from meiotic material in T. cap-illaris (subg.Diaphoranthema).Till (1984) reported mitotic "tetraploid level" numbers (2n = 84-96) for this same species.The reports for T. recurvata and T. tricholepis are the first meiotic counts for these species.They corroborate the ploidy levels deT.tricholepis where Till (1984) reported 2n = 50 (diploid) for variety macrophylla and 2n = 90 and 94 ("tetraploid-level") for variety tricholepis.The count of n = 25 + 2-10 fragments for T. scaligera is the second for this species.This report notes the variable number of chromosome fragments present at metaphase IT. fasciculata without varietal designation, and 2n = 56 for variety venosispica (reported as T. compressa Bert.ex Schultes).

Figure
model is dibasic and involves hybridization and polyploidy of paleodiploid base numbers x = 8 and 9 to yield a paleotetraploid (n = 17).This was followed by hybridization between a paleodiploid (n = 8) and the paleotetraploid lineage (n = 17) with polyploid stabilization at the hexaploid level (n = 25).The most recent common ancestor for the three subfamilies was hexaploid.Electropho-gia nutans H. Wendl.ex Regel).A complete chromosomal characterization of the parental species and hybrid individuals may provide more definitive clues as to the true chromosomal nature of Cryptanthus (i.e., ancient tetraploid or aneuploid derivative).