In S&D p968-70, 1977
304. Tillandsia dasyliriifolia Baker, Jour. Bot. London 25: 304. 1887. Fig 309 J.
Tillandsia drepanoclada Baker. Handb. Bromel. 188. 1889. Type. Chiquihuite, Mexico, Bourgeau 2193 (P), 20 Mar 1866.
Tillandsia geniculata Baker, Handb. Bromel. 190. 1889. Type. Mexico, without exact locality, Morren Icon (K).
Tillandsia pulvinata Baker, Handb. Bromel. 190. 1889. Type. Mexico, without exact locality, Morren Icon (K).

Plant stemless, 5-15 dm high.
Leaves many in a dense rosette, to 7 dm long, densely and finely appressed-lepidote throughout;
Sheaths large, elliptic, dark castaneous;
Blades linear-triangular, to 6 cm wide at base.
Scape erect, stout, generally exceeding the leaves;
Scape-bracts erect, ovate or lanceolate, acute, at least the lower ones imbricate and linear-laminate.
Inflorescence central, laxly bipinnate, usually ample. sub pyramidal;
Primary bracts ovate. acute or obtuse, shorter than the sterile base of the axillary branch;
Spikes suberect or curved-ascending, to 45 cm long. laxly flowered with several sterile bracts at base;
rhachis strongly flexuous or geniculate so that the flowers are nearly or quite at right angles to each other. usually stout, sulcate. flattened next to the flowers, glabrous.
Floral bracts very broadly ovate, the lower half consequently enfolding part of the rhachis at anthesis, obtuse, 16-25 mm long, shorter than the sepals, about equaling the internodes or sometimes slightly longer, coriaceous, glabrous outside, dark-punctulate-lepidote inside, even except near the margin, ecarinate;
Flowers subsessile, erect and appressed to the rhachis.
Sepals narrowly elliptic or obovate, obtuse, to 25 mm long, coriaceous, even, glabrous outside, dark-punctulate-lepidote inside, short-connate;
Petals tubular-erect, ligulate, obtuse, 3-4 cm long, white or greenish;
Stamens and pistil long-exserted.
Capsule slenderly cylindric, acute, 4-5 cm long.
Type. Gaumer s n (holotype K, GH photo), Holbox Island, Quintana Roo, Mexico.
DISTRIBUTION. Epiphytic in dry exposed habitats, from near sea level to 1800 m alt, central Mexico to Salvador, Honduras, and the Colombian Islands east of Nicaragua.

In JBS 31(5): 223. 1981 T. dasyliriifolia became T. limbata
On the identification of Tillandsia limbata Schldl. 1845 ("1844") by WILHELM WEBER

In his bromeliad monograph of 1935 in Engler's Pflanzenreich, Carl Mez placed Tillandsia limbata Schldl. into synonymy with Tillandsia flexuosa Sw. And also in the new monograph of tillandsioideae by Lyman B. Smith in Flora Neotropica the author indicates that Till. limbata Schldl. is a synonym for Till. flexuosa Sw., however with the notation that L. B. Smith himself has not seen the type plants of Till. limbata and that identification stems from C. Mez.

While re-examining Schlechtendal's old bromeliad herbarium, which is housed in the herbarium of the University of Halle, I had the opportunity to examine the type plants of Till. limbata Schldl. They consist of a larger - ca. 80 cm. - plant whose inflorescence and rosette are mounted on two separate boards and a second, somewhat smaller plant about 72 cm tall, whose inflorescence is folded over. The notations, written in hand by Schiede, say, "Till. August 29 Hac. de la Laguna" on the larger plant and, "Tillandsia, bracteis violaceis, in arboribus, Hac. de la Laguna Aug. 29" on the smaller plant.

The accompanying photographs of the type samples show clearly that these are not identical with Till. flexuosa Sw. My re-examination shows them to be the same as Till. dasyliriifolia described later by Baker.
Mez in his time, too, had substantiated Baker's findings, because the types have been added to his identification noted. This proves that Mez had the type plants of Till. limbata in his hands, and it is simply a mystery today why he then placed them into synonymy with Till. flexuosa Sw.

Priority rules then make the following nomenclature changes necessary:
Tillandsia limbata Schlechtendal in Linnaea 18; 1945 (" 1844"), p. 419
Synonyms: Till. dasyliriifolia Baker 1887
Till. drepanoclada Baker 1889
Till. geniculata Baker 1889
Till. pulvinata Baker 1889

TYPUS: Schiede s.n., MEXICO, Vera Cruz;
Hacienda de la Laguna, August 1828; HAL 45626
German Democratic Republic Translated by Harvey L. Kendall
Tillandsia dasyliriifolia is now resurrected to species status by Gardner in Selbyana vol.7 p372-4 1984
Note that Gardner makes no reference to T. drepanoclada or T. pulvinata treated by Smith as synonyms of T. dasyliriifolia !! Do they follow T. dasyliriifolia OR T. limbata ? I have treated T. dasyliriifolia as a stand alone!
Tillandsia limbata Schlechtendal, Linnaea 18: 1845 ("1844"). p. 419
Type: MEXICO: VERACRUZ: Hacienda de las Lagunas, Schiede s.n. (HAL, photo seen).
Tillandsia geniculata Baker, Handb. Bromel. 1889.
Type: MEXICO: without exact locality, Morren Icon (K, photo seen).
Tillandsia flexuosa auct. nonSwartz: J. G. Baker, Handb. Bromel. 1889 and L. B. Smith & R. J. Downs, Tillandsioideae in Flora Neotropica Monograph No.14, 1977.
Tillandsia dasyliriifolia auct. non Baker: W. Weber, Herbarium Studies II, Jour. Brom. Soc. 31(5): 222-224,1981.

Additional material examined: MEXICO: VERACRUZ: Mirador, 1929, Skwarra 19 (GH); San Louis Potosi: lowland forests, Rascon, 1892, Pringle 5306 (GH); Xilitla, 1979, Gardner 835 (SEL) HIDALGO: Huejutla,1946, Moore 2219 (GH); PUEBLA: E of Xicotepec de Juarez, 1979, Gardner 866 (SEL); CHIAP AS: Hwy 195,35 km S of Tabasco state border, 1981, Gardner 1399 (SEL). BELIZE: Belize, O'Neill 8510 (GH); 1970, Liesner & Dwyer 1415 (GH); 11.5 mi. N of Belize, Croat 23274 (GH). GUATEMALA: VERAPAZ: Chiquimula, 1885, Watson 49 (GH): ALTA VERAPAZ: Cerro Chinajo, between Finca Yalpemech and Chinaja above source of Rio San Diego, alt. 150-700 m, 1942, Steyermark 45690 (GH). IZABAL: Quirigua, alt. 75-225 m, 1922, Standley 24228 (GH). PETEN: La Libertad, 1933, Lundell 2643 (GH); Lundell2912 (GH).

Distribution: Moist lowland forests of eastern Mexico, Guatemala Belize, and eastern Honduras (san Pedro Sula) (fide Luther) to 800m elevation.

Photographs of Schiede's specimen annotated as the type for T. limbata, published by Weber (1981 ), prove that this species is not a synonym for T. flexuosa. His placement of T. dasyliriifolia as a synonym of T. limbata is apparently based on Mez's citation of the Schiede specimen under T. dasyliriifolia (without mention of T. limbata) (Mez, 1935) and assignment of similar specimens to T. dasyliriifolia by Smith and Downs (1977). Tillandsia limbata and T. dasyliriifolia are, however, distinct species and are readily distinguished by a full suite of characters. Many of these characters are related to ecology as these species are ecologically separated where their distibutions overlap, T. limbata being restricted to moist forests of high rainfall and T. dasyliriifolia occuring in arid forests or as an epiphyte on cacti or Yucca in deserts.
Baker (1899) described T. dasyliriifolia as having "very thick and rigid leaves" and violet petals. Tillandsia limbata has long, arching moderately thin leaves that are slightly coriaceous, at most, and greenish white petals. Tillandsia dasyliriifolia belongs to a species complex that includes T. makoyana Baker. This complex displays a mosaic pattern of character combinations and ecologies.
Tillandsia limbata may be distinguished from either of the previous species by the following: floral bracts broadly ovate, enfolding, or nearly enfolding the rachis, quite smooth (fresh or dry), glabrous, never nerved. Floral bracts of T. dasyliriifolia complex specimens are much narrower, not enfolding the rachis (or only slightly) and are typically finely nerved. Living specimens differ in pigmentation of the scape, rachis, and flush of the floral bracts that is No.42 maroon in T. limbata but No.38 rose in T. dasyliriifolia.

Introduction
In the following we provide an overview and circumscription of Tillandsia utriculata (L.) L. species complex where we place T. dasyliriifolia along with other species. A complete morphological description of T. dasyliriifolia, its geographical distribution, and characteristics to distinguish it from the other species in the complex, are included. Reproductive biology data are presented, as well as photographs of the species.
Tillandsia dasyliriifolia belongs to a group of species here referred to as the Tillandsia utriculata (L.)L. alliance. This alliance is characterized by epiphytic or subterrestrial rosettes, a monocarpic or polycarpic habit, leaves somewhat triangular, acute, never ligulate, with terminal, erect inflorescences, which are rarely racemes but more commonly 1-2 pinnate panicles, with flexuous and/or geniculate (bent like a knee), sessile flowers, with exerted stamens and stigma from a tubular corolla. The following species have been included in this complex: T. limbata Schltdl., T. makoyana Baker, T. pringlei S. Watson, T. simplexa Matuda, T. swartzii Baker, and T. dasyliriifolia Baker.
Most of the species related to Tillandsia dasyliriifolia, and thus in the T. utriculata complex, fall into Group II in Sue Gardner's proposed classification of the genus Tillandsia ( Gardner C S .1986. Preliminary classification of Tillandsia based on floral characters. Selbyana 9:76-87.). This group corresponds to subgenus Tillandsia (sensu Smith & Downs 1977) which occurs from southern Florida throughout the Caribbean, from central Mexico to Central America, and along the northern and northwestern parts of South America (Gardner C S .1986. Preliminary classification of Tillandsia based on floral characters. Selbyana 9:76-87.). Gardner's Group II contains 19 taxa, nine of which occur in Mexico. Gardner characterized the group by the presence of filaments of unequal length, round in cross section for their entire length, petal apex erect or recurved, and an open corolla throat.
A morphologically similar group, especially in floral characters (flexuous and /or geniculate flowers) is that comprised of such taxa as Tillandsia albida Mez & Purpus, T. flexuosa Sw., T. calcicola L.B. Sm. & G.R. Proctor, T. cucaensis Wittm., T. fresnilloensis W Weber & Ehlers, and T. karwinskyana Schult. f. This group is excluded from this discussion since several relevant vegetative characters (rosette shape and size, indument, inflorescence position) as well as geographical distribution set them apart from populations identified or confused with T. dasyliriifolia. We lack phylogenetic data to ascertain the relative positions of these two groups of taxa. Thus, it is not clear at this time whether the similarities in inflorescence structure are due to most recent common descent or to convergence.
Tillandsia dasyliriifolia was described by Baker in 1887 based on a collection by S. Gaumer from Mexico, Yucatan Peninsula, State of Quintana Roo, from a locality called Holbox Island "Bay of Honduras". This was once mistakenly referred to as Honduras by Smith (Smith, L.B. 1938. Bromeliaceae in T J. Yuncker. A contribution to the Flora of Honduras. Field Museu. Nat. Hist., Bot. Set. 17:318-322. ) but is actually located north of the Quintana Roo State (FIGURE 7).
In the original description Baker mentioned that the species has violet petals, but extensive field work in the Yucatan Peninsula has failed to reveal any populations of taxa in this complex with violet petals. Instead, all members of this complex in the area have chartreuse petals. Perhaps the violet or reddish color was intended to describe the floral bracts, because petals are not violet in this species (FIGURE 8). This inaccuracy in the original description caused many of the misidentifications in the species complex.
The name Tillandsia dasyliriifolia has been applied to a large number of populations of Tillandsia across Mexico and Mesoamerica including populations in Colombia (Providence Island) (Smith & Downs 1977, among others), inhabiting such diverse vegetation types as shrubby vegetation in sand dunes, oak-pine forests, low inundated forests, etc.As a consequence, the distribution of T. dasyliriifolia has been interpreted until recently as extending from Jalisco (NW Mexico) to northern Colombia, and from sea level up to 2300 m elevation.
Most of the confusion with the delimitation of taxa in this complex characterized by inflorescences with flexuous rachis and /or geniculate flowers is due to the lack of good types (herbarium specimens), incomplete locality data (most of them come from such vague localities as "Mexico without exact locality"), misinterpretations in the description of petal color (previously mentioned in Baker, 1889) or the apparent hybridization between several species. In addition, some of the types are based on illustrations (i.e., Tillandsia makoyana Baker, T. pulvinata E. Morren ex Baker, Tillandsia geniculata Baker, all based on Professor Morren's plates), which makes some important characters impossible to study (drawings are impossible to dissect!). However, a more serious problem is that most herbarium specimens of the species in the T. utriculata complex look alike since most of the important diagnostic characters (color, symmetry, textures, etc.) are lost when the plant is dried. Furthermore, being such large plants, many herbarium specimens are incomplete, i.e., lacking either portions of the inflorescences or of the rosette. Moreover, critical morphological characters are either ignored in the field descriptions provided in the labels by the collectors, or completely misinterpreted.
Following is a complete description of Tillandsia dasyliriifolia, which is based on herbarium specimens from the United States (F, MO, SEL), Mexico (CICY, CIQRO, MEXU, UAMIZ, UADY), and Central America (INBIO, TEFH). The circumscription of this species, as well as those of the other taxa in the complex, were ascertained with the aid of herbarium and cultivated material from several localities in the Yucatan Peninsula and Mexico.
Recent fieldwork in many localities in Mexico, including several type localities (Espejo-Serna, A. & A R Lopez-Ferrari. 1994. Las Monocotiledoneas Mexicanas: una sinopsis floristica. 1. Lista de referencia, parte III. Bromeliaceae, Burmanniaceae, Calochortaceae y Cannaceae. Consejo Nacional de la Flora de Mexico, A.C., Universidad Autonoma Metropolitana y Comision Nacional para el Conocimiento y Uso de la Biodiversidad, Mexico, D E , a detailed observation of population variation in the field, cultivation of several species, and a more complete documentation through herbarium material, photographs, and pickled material, greatly aided in the understanding of the Tillandsia utriculata complex.

Tillandsia dasyliriifolia Baker, J. Bot. 25: 304, 1887; Emend Ramirez et al, J. Brom. Soc. 54(3): 112-121. 2004
TYPE: Mexico: Quintana Roo: Hab. Holbox Island, "Bay of Honduras". Received from E Godman, F R S. in VIII-1886, G. Gaumer s.n. (K). FIGURES 8,9.
Epiphytic rosettes, often rupicolous or sub terrestrial (on sandy soils or low inundated forests), (0.5-)0.7-1.7(-3) m tall, sometimes producing offsets ("keikis") on the inflorescence (especially populations from shrubby vegetation on sand dunes or in large, robust plants).
Leaves forming a funnel form rosette, (23)70-90(95) cm long, 2.1-2.7(3.6) cm wide;
foliar sheaths dark castaneous, densely adpressed to subadpressed white-lepidote, widely elliptic, gradually merging into the triangular lamina, 6-7 cm long, 3.5-4 cm wide;
lamina narrowly triangular, white-lepidote abaxially, green adaxially, sparsely white lepidote adaxially, usually with the mid longitudinal area of the leaf thicker, (-24)29-30 cm long.
Inflorescence usually a 1-2 pinnate panicle, 50-58 cm or much longer up to 3 m long or much reduced in small depauperate plants to a raceme 30 cm tall, erect to arcuate, usually the peduncle much longer than the leaves, especially in plants growing in shady, humid places;
peduncle 0.7-1.8(2) cm diameter at base, (30-)66-100 cm long, usually longer than the rosette, rarely as long as the rosette,
scape bracts 3.5-5.5 cm long, with a long, acuminate apex, erect, longer than internodes at the base of the peduncle but as long to shorter than internodes toward the apex of the inflorescence, margins of scape bracts adnate by 1 cm long on those at the base of peduncle, widely triangular, acute;
branches (1-3)7-12(-22) per inflorescence, 16-20 cm long; sterile portion of branches to 6 cm long, fertile portion 35-46 (67) cm long, 6-7 flowers per each 10 cm of rachis length on branches;
rachis 2-5 mm diameter, naked or covered by floral bracts, usually red or dark pink;
primary bracts 2-3 at base of each branch, 3.5-5.5 cm long, (0.8-) 1.5-2 cm wide, wide elliptic, acute, imbricate, margins hyaline.
Flowers (3.9-)4.2-5.2 cm long, 70-100 flowers on well developed inflorescences, actinomorphic at first but becoming zygomorphic after anthesis and pollinator visit when stamens and stigma get pushed toward the adaxial portion of the corolla tube and this is distorted and becomes curved, sessile or subsessile at base.
Floral bracts widely ovate to wide triangular, obtuse to acute, covering the rachis with their bases, as long as or shorter than the sepals, deeply concave, non imbricate, ecarinate, smooth or apically veined on the adaxial surface, glabrous, coriaceous, 2-2.4 x (1.4)-1.9 mm, drying roseate or sometimes the apical half maroon, red when fresh as well as the rachis and peduncle, (13)19-22 x 13-15 (21) mm.
Sepals 15-22 (25) x 7-8 mm, narrowly elliptic to elliptic or elliptic-obovate, acute to obtuse, green, white-lepidote inside, the sepals opposite to the rachis are connate by 3 mm of their length.
Petals 28-33 (37) x 4-6 mm, with a constriction caused by the apex of the sepaline tube at 17-20(25) mm from base, elliptic to oblong, acute to obtuse, concave, pale chartreuse to almost white.
Stamens exerted, filaments pale green or white, 3.6(-5) cm long;
anthers black, 4-5 mm long. Stigma exerted, longer than stamens, with three long, expanded lobes, pale chartreuse.
Ovary elliptic, (6.5-)8.5-9 mm long, 2.5-3 mm wide.
Capsules 5-6 cm long, 5-8 mm diameter; seeds 2.7-4.2 cm long, including the hairy appendages.

Tillandsia dasyliriifolia inhabits xerophytic shrub lands on coastal dunes, deciduous forest, mangroves, low inundated forests, and evergreen forest, from 0-250 m above sea level.
Geographical distribution: Mexico (Yucatan, Quintana Roo, Campeche,Tabasco); Belize (Districts of Belize,Toledo).

Phenology:
Plants have been recorded with flowers in April, May, June, and October; fruits in February, May, August, September, November, and December. Our knowledge of the phenology of the species in likely incomplete and will improve as more collections are incorporated into herbaria.

Diagnostic characters:
We have carried out studies to estimate the variation among populations of several vegetative and floral traits in Tillandsia dasyliriifolia, particularly comparing plants from the low inundated forests and the shrubby vegetation on sand dunes associated with the coasts in the Yucatan Peninsula. There is variation in size of vegetative and floral traits-plants from the coastal shrubby vegetation are taller, with larger inflorescences and higher fruit set are compared to those in low inundated forests. We suggest that larger-sized plants are correlated with the humidity received from the sea, and the differences in fruit set is probably due to fewer visits by pollinators in the tintales, and not necessarily due to fewer pollinators. This might be due to a more restricted flower "menu" in the coastal dune compared to the tintales, thus forcing the pollinators to use more efficiently what is available in the season.
Another variant of this taxon was observed in populations from more humid and/or shady places south of the Peninsula of Yucatan. In tall evergreen forests in Quintana Roo, the plants tend to have softer, longer leaves and longer peduncles on inflorescences and branches located toward the apex. Although the species was described as having violet petals (Baker 1887), none of the populations from the type locality and vicinities present this feature. This mistake on the floral description was noted by several authors whose interpretations of the species were distorted by this claim (Smith & Downs 1977, McVaugh 1989, Utley & Utley 1994). McVaugh, R. 1989. Bromeliaceae in Flora Novo-Galiciana, vol. 15: 4-79. Ann-Arbor, MI: University of Michigan Press. Utley, J.F & K. Burt-Utley. 1994. Bromeliaceae in Flora Mesoamericana 6:89-156.
Tillandsia dasyliriifolia can be identified by the presence of a funnelform rosette, and is mostly epiphytic or sometimes subterrestrial on sandy soils. In most cases the leaves are straight, but sometimes they curve downwards (especially in plants from humid, shady places), white lepidote abaxially, dark green adaxially, narrowly triangular, with a wide base that becomes castaneous upon drying; the peduncle, rachis, primary and floral bracts are red, the sepals are dark green and petals are light chartreuse. Flowers are initially actinomorphic but eventually become zygomorphic after visits of the pollinating hummingbirds, Doricha eliza and Amazilia yucatanensis.

How to recognize the species from herbarium specimens?
Leaves tend to dry light brown, the foliar sheaths are castaneous, inflorescences only rarely are racemes, usually 1-2 pinnate panicles, the floral bracts covering the rachis, the rachis flexuous, the flowers geniculate, the floral bracts shorter than to rarely equal to the sepals, the rachis and floral bracts red, the sepals green, petals light chartreuse. Tillandsia dasyliriifolia is known from all three states in the Yucatan Peninsula (Campeche, Quintana Roo, and Yucatan), the state of Tabasco, Peten of Guatemala and Belize. There is another species similar to Tillandsia dasyliriifolia in these states, such as T. utriculata, with rachis, floral bracts and sepals green, petals white with an asymmetric corolla, leaves shiny green above, slightly white lepidote abaxially, forming a utriculate rosette, inflorescences are larger (with a proportionally shorter peduncle) and less dense (fewer flowers per branch) and slightly flexuous rachis and slightly geniculate flowers. Both species produce new branches if a portion of the inflorescence is damaged, a character absent in other related species such as T. limbata and T. makoyana. T. utriculata and relatives, however, are almost always monocarpic plants as opposed to the usually serially monocarpic plants of T. dasyliriifolia.
The species was also reported for Guatemala by L.B. Smith (Smith L B 1958. Bromeliaceae in E C. Standley and J.A. Steyermark. Flora de Guatemala. Fieldiana, Bot. 24:380-476. {Aug 29}) who mentioned that the flowers are pedunculate (5 mm long) and with green or violet petals. Thus, if Smith was right about the floral characters, he must have been referring to an entirely different taxon and the Guatemalan plants in question are neither Tillandsia dasyliriifolia nor any other of its relatives since the flowers in the T. utriculata complex are nearly sessile.
McVaugh in Flora Novo Galiciana (1989) reported the species from several states in Mexico (Nayarit, Jalisco, Michoacan, Guerrero, Mexico, Morelos, Puebla, Quintana Roo). This cited distribution indicates a disjunction, from the Pacific Coast to the Caribbean Coast of Quintana Roo, and from high elevations to sea level. Our studies indicate that there are more than one species in the Pacific Coast in this complex. Gardner (1984) correctly stated (and we agree) that many of the specimens from that geographical area previously identified as Tillandsia dasyliriifolia actually represent T. limbata, which occurs in moist forests, with thinner and less coriaceous leaves, and is known from the states of Veracruz, Puebla, and Chiapas, and south to at least Honduras.
In the group of taxa with violet petals, several are involved. Some of the morphospecies fit the Tillandsia makoyana concept. Another entity with violet petals, flexuous rachis and geniculate flowers, rosette with triangular, erect leaves, that inhabits mostly oak-pine forest above 1000 m above sea level, is probably new to science; while a third one with geniculate, green flowers and a flexuous rachis, from lower elevations in the state of Oaxaca is also undescribed.

Reproductive Aspects of Tillandsia dasyliriifolia
In order to determine the breeding system of the species, we conducted a series of controlled crosses in the field-a shrubby vegetation area, north of the Yucatan state, in the San Benito area (FIGURE 7). We used 60 plants and 1662 flowers in order to perform the following crosses:
agamospermy (fruit production without pollination by removing anthers and thus pollen from flowers - a process called emasculation);
manual selfing (pollinating the flower with its own pollen);
unassisted selfing (by bagging the flowers and observing fruit formation through pollination by one's own pollen);
cross pollination (pollinating flowers with exogenous pollen);
anemophily (bagging the emasculated flowers with a material able to let pollen through but not animal pollen vectors);
natural pollination (leaving flowers unprotected and observing pollination).

The percentages of fruit set for each controlled cross is shown on TABLE 1.

TABLE 1: Controlled crosses, number of plants used, total number of flowers used for each cross, number of fruits produced, and fruiting percentage in Tillandsia dasyliriifolia.

Controlled cross	# Plants	# Flowers	# Fruits	% Fruiting
Agamospermy	10	42	0	0
Manual Selfing	9	110	102	92.7
Unassisted selfing	9	617	574	93
Manual cross pollination	15	42	38	90.5
Anemophyly	9	42	0	0
Natural pollination	8	809	707	87.4

A first approach to know the breeding system of a species is to get results from controlled crosses. According to our results:
1. An absence of fruits by agamospermy indicates that the flowers will produce fruits by pollination only.
2. The absence of fruits in anemophily indicates that in order for pollination to occur, the pollen must be moved by a vector other than wind or water (such as insects, birds, etc).
3. The species is self-compatible since we obtained 92.7% of fruit setting by selfing, but also a similar result by cross pollination (90.5% of fruit setting).The plant will produce fruits by pollination with its own pollen and exogenous pollen.
4. The species also produced fruits by pollinating itself, since protected flowers produced fruits without pollinator assistance (unassisted selfing with 93% or fruit setting). This indicates that even though anthers release pollen before the stigma is receptive, there is a moment when both are functional and pollination occurs (flowers are not dychogamous) and both structures, anthers and stigma touch each other at some time (flowers are not herkogamous).
5. A high percentage (87.4%) of fruit setting was also obtained by natural pollination. Usually when we record lower percentages of fruit setting in natural pollination compared with selfmg or cross pollination, it suggests a lack of pollinators or some sort of pollinator's inefficiency. The differences in percentage between natural pollination and the other crosses with fruit set are very small for this species suggesting that pollinators are very efficient moving pollen from one flower to another.
A more accurate way to determine the breeding system is to observe seed formation; an even more accurate method is to observe viable seeds. In other words, fruit can be formed but with few or no seeds, and seeds can be formed but some of them may not be viable, indicating inbreeding depressions and /or hybrid vigor, as many plant growers are familiar with.
In accordance with this idea, when we observed results based on seed numbers, the data tendency is similar to the results based on fruit setting (TABLE 1).Thus, we have the same story told by both data sets.
A statistical test, ANOVA, was performed in order to determine whether there are significant differences between the number of seeds produced among all controlled crosses, {F(3,354)=16.23; p<0.0001*}. Once this result was determined from this ANOVA; we wanted to know in which of the controlled crosses the number of seeds was the higher and so on. For this, we proceeded to perform a second statistical test, LSD, a comparison of medias. Results indicated that the major amount of seeds produced occurred in manual cross pollination (letter a), this value being much higher than the other treatments. Manual selfing rendered the second largest amount of seeds (letter b); third was natural pollination, and the lowest value was for unassisted selfing (letter d) (TABLE 2). Since manual cross pollination produced the largest number of seeds, we suggest that there is some sort of hybrid vigor, as well as an inbreeding depression or defective pollination as suggested by the lowest amount of seeds in selfing.

TABLE 2:Average seed number and aborted seed number on each individual controlled cross performed in Tillandsia dasylirtifolia

Controlled cross	Seed number
Manual selfing	164.1 +/- 10.3 b
Unassisted selfing	154.8 +/- 16.4 d
Cross pollination	188.1 +/- 15.9 a
Natural pollination	159.7 +/- 11.0 c

Viable seeds provide more information about possible effects of selfing (inbreeding), out crossing (hybrid vigor), etc. We selected seeds from each fruit type (from selfing, out crossing, etc.) and germinated seeds in the lab on Petri dishes with a filter paper and controlling humidity, light, and temperature. A statistical test called Xi^2 was applied in order to detect differences among the results (TABLE 3).
According to these results, there are no differences among number of viable seeds among the crosses (all have same literal, a), suggesting no inbreeding depression and/or hybrid vigor at this level. Results of germination of seeds suggest that all seeds will be equally successful in germinating, an important advantage for the species.

TABLE 3: Seed germination percentage for each individual controlled cross performed in populations of Tillandsia dasyliriifolia. A Xi^2 test was performed in order to detect differences.

Species/Cross	Selfing	Unassisted Selfing	Cross pollination	Natural pollination
T. dasyliriifolia	96.00 a	96.00 a	97.30 a	96.30 a

We recorded several flower visitors, bees and birds, some of them being the pollinators, such as the hummingbird Doricha enicura male (COVER), which takes nectar from the flowers and gathers pollen on the forehead, pollinating other flowers in successive visits. This hummingbird usually visits several flowers in the same inflorescence before moving to the next plant, but geitonogamy (pollination among flowers of the same inflorescence or genet) is minimized by the fact that only a few flowers per inflorescence open each day, increasing the rate of cross pollination.
Populations of Tillandsia dasyliriifolia in the Yucatan peninsula, and specifically in the study site, have very high reproductive values. They are able to set fruit without pollinators assistance, as well as with pollination by hummingbirds; it is a self compatible species. Moreover, they produce new rosettes in the inflorescences ("keikis", FIGURE 10) and the genet is polycarpic. All these factors added to the fact that the species grow in coastal shrub lands, low caducifolious, forest, tall evergreen forest and low inundated forest, indicate that the species, thus far, is not in danger.

Tillandsia dasyliriifolia inhabits xerophytic shrub lands on coastal dunes, deciduous forest, mangroves, low inundated forests, and evergreen forest, from 0-250 m above sea level.
Geographical distribution: Mexico (Yucatan, Quintana Roo, Campeche,Tabasco); Belize (Districts of Belize,Toledo).

Phenology:
Plants have been recorded with flowers in April, May, June, and October; fruits in February, May, August, September, November, and December. Our knowledge of the phenology of the species in likely incomplete and will improve as more collections are incorporated into herbaria.

Diagnostic characters:
We have carried out studies to estimate the variation among populations of several vegetative and floral traits in Tillandsia dasyliriifolia, particularly comparing plants from the low inundated forests and the shrubby vegetation on sand dunes associated with the coasts in the Yucatan Peninsula. There is variation in size of vegetative and floral traits-plants from the coastal shrubby vegetation are taller, with larger inflorescences and higher fruit set are compared to those in low inundated forests. We suggest that larger-sized plants are correlated with the humidity received from the sea, and the differences in fruit set is probably due to fewer visits by pollinators in the tintales, and not necessarily due to fewer pollinators. This might be due to a more restricted flower "menu" in the coastal dune compared to the tintales, thus forcing the pollinators to use more efficiently what is available in the season.

Another variant of this taxon was observed in populations from more humid and/or shady places south of the Peninsula of Yucatan. In tall evergreen forests in Quintana Roo, the plants tend to have softer, longer leaves and longer peduncles on inflorescences and branches located toward the apex. Although the species was described as having violet petals (Baker 1887), none of the populations from the type locality and vicinities present this feature. This mistake on the floral description was noted by several authors whose interpretations of the species were distorted by this claim (Smith & Downs 1977, McVaugh 1989, Utley & Utley 1994). McVaugh, R. 1989. Bromeliaceae in Flora Novo-Galiciana, vol. 15: 4-79. Ann-Arbor, MI: University of Michigan Press. Utley, J.F & K. Burt-Utley. 1994. Bromeliaceae in Flora Mesoamericana 6:89-156.

Tillandsia dasyliriifolia can be identified by the presence of a funnelform rosette, and is mostly epiphytic or sometimes subterrestrial on sandy soils. In most cases the leaves are straight, but sometimes they curve downwards (especially in plants from humid, shady places), white lepidote abaxially, dark green adaxially, narrowly triangular, with a wide base that becomes castaneous upon drying; the peduncle, rachis, primary and floral bracts are red, the sepals are dark green and petals are light chartreuse. Flowers are initially actinomorphic but eventually become zygomorphic after visits of the pollinating hummingbirds, Doricha eliza and Amazilia yucatanensis.

How to recognize the species from herbarium specimens?
Leaves tend to dry light brown, the foliar sheaths are castaneous, inflorescences only rarely are racemes, usually 1-2 pinnate panicles, the floral bracts covering the rachis, the rachis flexuous, the flowers geniculate, the floral bracts shorter than to rarely equal to the sepals, the rachis and floral bracts red, the sepals green, petals light chartreuse. Tillandsia dasyliriifolia is known from all three states in the Yucatan Peninsula (Campeche, Quintana Roo, and Yucatan), the state of Tabasco, Peten of Guatemala and Belize. There is another species similar to Tillandsia dasyliriifolia in these states, such as T. utriculata, with rachis, floral bracts and sepals green, petals white with an asymmetric corolla, leaves shiny green above, slightly white lepidote abaxially, forming a utriculate rosette, inflorescences are larger (with a proportionally shorter peduncle) and less dense (fewer flowers per branch) and slightly flexuous rachis and slightly geniculate flowers. Both species produce new branches if a portion of the inflorescence is damaged, a character absent in other related species such as T. limbata and T. makoyana. T. utriculata and relatives, however, are almost always monocarpic plants as opposed to the usually serially monocarpic plants of T. dasyliriifolia.
The species was also reported for Guatemala by L.B. Smith (Smith L B 1958. Bromeliaceae in E C. Standley and J.A. Steyermark. Flora de Guatemala. Fieldiana, Bot. 24:380-476. {Aug 29}) who mentioned that the flowers are pedunculate (5 mm long) and with green or violet petals. Thus, if Smith was right about the floral characters, he must have been referring to an entirely different taxon and the Guatemalan plants in question are neither Tillandsia dasyliriifolia nor any other of its relatives since the flowers in the T. utriculata complex are nearly sessile.

McVaugh in Flora Novo Galiciana (1989) reported the species from several states in Mexico (Nayarit, Jalisco, Michoacan, Guerrero, Mexico, Morelos, Puebla, Quintana Roo). This cited distribution indicates a disjunction, from the Pacific Coast to the Caribbean Coast of Quintana Roo, and from high elevations to sea level. Our studies indicate that there are more than one species in the Pacific Coast in this complex. Gardner (1984) correctly stated (and we agree) that many of the specimens from that geographical area previously identified as Tillandsia dasyliriifolia actually represent T. limbata, which occurs in moist forests, with thinner and less coriaceous leaves, and is known from the states of Veracruz, Puebla, and Chiapas, and south to at least Honduras.

In the group of taxa with violet petals, several are involved. Some of the morphospecies fit the Tillandsia makoyana concept. Another entity with violet petals, flexuous rachis and geniculate flowers, rosette with triangular, erect leaves, that inhabits mostly oak-pine forest above 1000 m above sea level, is probably new to science; while a third one with geniculate, green flowers and a flexuous rachis, from lower elevations in the state of Oaxaca is also undescribed.