A Monograph of Sabal ( Arecaceae : Coryphoideae )

This monographic study of the New World genus Saba! (Arecaceae: Coryphoideae) recognizes 15 species. In addition to defining species limits and distributions, the study addresses broader questions concerning likely modes of speciation in the group and biogeographic radiation. The systematic treatment incorporates results from extensive field work and studies of leaf anatomy and flavonoid phytochemistry, ecology and biogeography, and morphology. Distribution maps and a key to the taxa are provided. Solutions are offered for the many nomenclatural problems that existed in the genus. A phylogenetic hypothesis, the first for the genus, is proposed. Moreover, phytochemical and anatomical features are examined in an ecological perspective, and hypotheses about their function and evolutionary significance are presented.


INTRODUCTION
One of the most common genera of palms in and around the Caribbean basin is the genus Saba/ (Arecaceae: Coryphoideae). It is widespread and often weedy, thriving in anthropogenic habitats from Bermuda to Sonora, from Texas to Trinidad. Likewise, it is common in the southeastern United States and is likely one of the palms best known to north temperate botanists. Saba/ is widely cultivated as an ornamental in gardens around the world; in its native habitats, it sustains thatch, basketry, and hat-making industries. Yet despite its familiarity, Saba/ has remained poorly studied and poorly understood.
Previous workers (Bailey 1934(Bailey , 1944Beccari 1907) confined their efforts to morphological taxonomic studies of genus. Faced with the general morphological sameness of the species and confounded by inadequate collections, they were most concerned with defining species boundaries. Saba/, the sole member of the subtribe Sabalinae of the tribe Corypheae (Uhl and Dransfield 1987), was clearly circumscribed at the genus level, but species boundaries were ill-defined. At the root of much of the past taxonomic confusion lay narrow species concepts in which nearly every separate population was recognized as a distinct species. Only with an appreciation for the ease with which Saba/ has dispersed over long distances do we begin to develop a meaningful species concept for this group.
The present monograph has incorporated morphological, anatomical, and phytochemical data in an evolutionary and ecological framework. In addition to a key to the taxa, distribution maps, species descriptions and full synonymies, a phylogenetic hypothesis is provided. It is the first phylogeny proposed for the genus. Three additional questions are addressed: What has been the likely mode of speciation in the group? What can the phylogenetic hypothesis and present day distribution reveal about past biogeographical events and patterns? What adaptations are present in Saba/ that allow it to succeed so well in a variety of environments in and around the Caribbean?

Distribution and Ecology
The distribution of Saba! is primarily Mexican, southeastern United States, and Caribbean (including Bermuda), with an outlying species found in Costa Rica, Panama, Venezuela, Colombia, and Trinidad (Fig. 1). Several disjunctions in the distribution are immediately apparent.
Saba! mauritiiformis is known from southern Mexico, southeastern Costa Rica, eastern Panama and the adjacent northern coast of South America. It is also found in southern Trinidad and has been reported by Wessels Boer (1988) from the islands of Cura9ao and Bonaire. Its present distribution is probably recent, since it grows in lowland wet tropical forests that were submerged until quite recently.
Four other disjunctions are also readily attributable to overwater dispersal of seeds: those of S. maritima, S. palmetto, S. causiarum, and S. yapa. Saba! maritima is found on both Cuba and Jamaica (and is the only species of Saba! on Jamaica). Its present distribution-on recent soils on both islands-may also be recent, or it may have moved into these soils as other soils on the islands weathered. Saba! causiarum is found on Hispaniola and Puerto Rico; it inhabits lowland disturbed areas on both islands. Saba! palmetto is found in Cuba, the Bahamas, and the southeastern United States; S. yapa occurs on the Yucatan Peninsula (in Mexico and Belize) and western Cuba.
Island endemism is common in the genus, with one quarter of the species endemic to the Greater Antilles and Bermuda. Two species of Florida, S. etonia and S. miamiensis, are endemic to islandlike areas, the Central Florida Ridge and the Everglades Keys, respectively.
Most widespread species of Saba! (S. mauritiiformis, S. mexicana, S. palmetto, and S. yapa) as well as island endemics (S. causiarum, S. domingensis, and S. maritima) are small-fruited trees of the forest canopy. They thrive in high light intensity environments and commonly persist after forests are cleared for agricultural purposes. Recruitment inS. palmetto is a case in point. The species grows readily in oak forests in northern Florida, but seedlings under a closed canopy remain suppressed and form no aboveground stem. Stem elongation and sexual maturation await gap formation in the canopy. Along forest margins, on dunes, and in fields, growth and recruitment are immediate with no suppressed stage.
These species, as well as S. bermudana, S. rosei, and S. pumos, are "weedy" species, colonizing gaps and patchy habitats. They withstand burning and thrive in anthropogenic habitats. Saba! uresana, a species ofxerophyllous woodlands of northwestern Mexico, appears to survive less well in disturbed habitats and, as noted by Gentry (1942), appears to be declining in the wild. This species never forms large stands in cleared fields as do its congeners S. rosei and S. pumos. Saba! minor is an understory species of deciduous forests, while S. etonia and S. miamiensis are understory species of pine-oak associations in Florida.

Field Studies
In the years 1984-86, I studied 13 populations of species occurring in Florida in the field. During the summers of 1986 and 1987, natural populations and cultivated individuals of Saba! were studied throughout Mexico. In 1988, field studies were undertaken in Panama, Cuba, Bermuda, the Dominican Republic, Trinidad, and Jamaica, as well as in southern Florida. At each population, complete voucher specimens were gathered and a separate collection number was given to each individual collected. Specimens collected prior to September, 1985, are deposited at FLAS, with duplicates distributed to various herbaria. Specimens collected after September, 1985, are deposited at RSA, with duplicates to be distributed.
Field observations of characteristics not readily visible from dried specimens include: species abundance, altitude, soil type, associated species and vegetation type, trunk height, diameter and surface texture, leaf number and color, petiole length, inflorescence number, length and posture, flower color and fragrance, insect visitors, fruit color, seed dispersers, and seed predators.
In addition, collections of flowers, fruits, and leaf samples were preserved in FAA and later transferred to glycerine-alcohol (Martens and Uhl 1980). These specimens were used for anatomical and morphological investigations. Dried bulk samples of leaf material were collected for phytochemical analysis. Living seed, when available, was collected and distributed to the Seed Bank of the International Palm Society, Fairchild Tropical Garden (Miami, Florida), Huntington Botanical Garden (San Marino, California), and Jardin Botcinico (Mexico City, Mexico).

Herbarium Studies
Over 500 herbarium specimens were examined in the course of this study. Four herbaria (BH, FI, MEXU, P) were visited, and numerous herbaria (see Acknowledgments) lent material for study. Study of herbarium material was essential not only for determining the range of morphological variation but also for compiling data on geographic and altitudinal distribution and common names. Bailey (1934Bailey ( , 1940Bailey ( , 1944 has written eloquently and often on the problem of preparing specimens of Saba/ for the herbarium. The large stiff leaves and inflorescences resist the press and demand special techniques. I have found the following method of preparation and storage to be suitable for Saba/: a healthy leaf is selected and removed from the tree, the petiole below the hastula is measured and then discarded (petiole length varies according to shade received), one half of the lamina is cut away taking care not to cut the hastula, and the outermost segments of the other half of the lamina (often wind-tom and the first segments to senesce) are trimmed away. Once trimmed in this fashion the leaf specimen is folded to fit the herbarium case, held in place with rubber bands, placed in a press, and dried. The inflorescence (or infructescence) is likewise trimmed of half its branches, and only the lower one or two primary branches (and all of their branches) are preserved. The inflorescence specimen is folded, held in place with rubber bands, and pressed. Specimens prepared in this fashion are bulky and are usually stored in boxes, but they have that advantage in that they can be unfolded and examined from all sides, unlike sheet-mounted specimens.

Methods for Measurement of Specimens
Measurements were taken from both living or pickled material and dried pressed specimens. Measurements of floral parts were made from herbarium specimens rehydrated by boiling. Measurements of large structures were made with either a metric scale measuring tape or ruler, and those of small structures were made with rotary dial micrometer .
Tree height was estimated visually; trunk diameter was measured on living specimens. Petiole and blade lengths were measured at the time of collection prior to pressing. All other vegetative measurements were taken from dried specimens. Petiole width was measured at the juncture of the petiole and hastula. Leaf segment measurements were taken from a segment midway along one side of the hastula. Segment width and lamina thickness were taken immediately above (distal to) the point of segment connation. Only one set of measurements was made for each collection.
Inflorescence length was either estimated visually or measured at the time of collection. Its natural position relative to the leaves was recorded. Rachilla diameter and length and bracteole length were taken from pressed specimens; all other floral measurements were made from rehydrated flowers. Rachillae length and number were measured (one for each collection) from basal branchlets (penultimate branches), and thus represent maxima for these characters; rachillae tend to be shorter and fewer in number on terminal penultimate branches. Rachilla thickness was taken midway along a rachilla from a middle rachilla; for both thickness and length, in no case was a terminal rachilla used. Petals, because their margins are involute, were measured at their widest points by folding them transversely, thus inducing their margins to unroll. Only one set of floral measurements was made for each collection.
Fruits and seeds were measured in the dry condition. From each collection, five fruits and seeds, selected at random, were measured, tabulated, and averaged; every effort was made to include only mature fruits and seeds. of the hastula may be entire or undulate, erect, involute, or revolute. The size and shape of the hastula are useful taxonomically only in the most general way (Moore 1971a).
In some populations of some species (viz., S. mexicana in Veracruz, Mexico, and S. mauritiiformis in Trinidad) the hastula is highly involute, so much so that the adaxial surface of the hastula is no longer visible. The curled abaxial edge of the hastula may bear the impressions of the underlying leaf segments giving the hastula a ridged appearance.
The leaves of Saba! are alternate and spirally arranged, flabelliform, composed of 15-120 segments (in the range of 60-7 5 for most species), and weakly to strongly costa palmate. The costa in a strongly costapalmate species typically curves downward (Fig. 2B), giving the leaf its characteristic rigid curvature. Segments are induplicate with a strong central vein, the midvein, and along their margins of connation, a strong suture vein is formed. Segments may be lax or rigid, bifid at the apex or not. Filamentous fiber extensions may be inserted between the segments (at the termination of the suture vein) and at the termination of the midvein in bifid segments. The leaves of Saba! may be glaucous or evenly green.
Segments are short (less than 100 em) in some species (S. etonia, S. minor) or long (up to 200 em), and the apical bifurcation may be shallow, deep, or absent. Lamina thickness ranges from less than 0.1 mm (in some species) to 0.5 mm. Segments may be connate for 15-50% of their length, with the least amount of connation among the outermost segments and the greatest among the terminal segments (those adjacent to the costa). The size ofthe palman (the proximal fused laminar portion of the leaf) shows some variation both within and among species.
In some species, leaf segments are grouped in twos or threes, with connation within groups nearly complete and connation between groups very slight. In S. mauritiiformis, splitting between segment groups occurs along a midvein, giving some segments a reduplicate appearance. This phenomenon is also known to occur in Licuala Thunb. (Comer 1966) and other coryphoid palms (Uhl and Dransfield 198 7).
Pel tate, multiseriate trichomes are present on young leaves of all species. They are brown with a laciniate margin and give young leaves a scurfy vesture. Usually, they are rapidly caducous. The trichomes persist longest along the abaxial side of the midveins. Only S. maritima frequently retains its trichomes for the life of the leaf.

Inflorescence
The paniculate inflorescence in Saba! is interfoliar, and its posture early in development and degree of ramification are diagnostic for some species (Fig. 3). The inflorescence may be erect (emerging 90° from horizontal), ascending (emerging less than 90° but greater than 45° from horizontal), arching (emerging ca. 45° from horizontal and arching downward), or cemuous (emerging more or less horizontally and hanging downward). Normally, ascending or arching inflorescences may sag under the weight of developing fruits, so inflorescence posture is best observed early in the development of the inflorescence before the rachillae have fully emerged. The inflorescence ranges in length from 0.4 to 3 m, and it is sparingly to densely branched. There are 2-4 orders of branching enumerated according to the system ofTomlinson and Zimmermann (1968). The inflorescence is clasped by a sheathing bicarinate prophyll and 2-5 tubular bracts, according to the vigor of the plant. Branches up to and including the penultimate branches are each subtended by a bicarinate bract. Tubular bracts, with straight or oblique openings, clasp all branches up to and including the antepenultimate branches. The bicarinate bract of the penultimate branches may be exserted or inserted within the tubular bracts of the antepenultimate branches. The ultimate branches (the rachillae) are borne in the axils of solitary small triangular bracts. Flowers are subtended by one small bract, and a pedicillar bracteole is borne obscurely on each flower (Morrow 1965).
R.achillae are more or less terete to strongly angular in cross section, and although rachilla shape has been taken as a specific character by Beccari (1907) and Bailey (1944), it has no taxonomic value over a broad range of collections. A rachilla gradually tapers from its base to its apex; however pathogenic conditions (fungal in origin?) may give rachillae a puffy or swollen appearance. Swollen and fusiform rachillae have been mistaken as a characteristic of some species (e.g., S. uresana). Various fungal infections manifest by patches of hyphae and/or reproductive structures are commonly seen on rachillae of all species of Saba!.

Flower
Flowers of Saba/ are borne singly. They are exposed in bud and open more or less acropetally along the inflorescence. Flower color is creamy white, and the flowers have a pungent sweet fragrance. They are ca. 3.5-7 mm high. Valuable taxonomic characters can been found in the flowers of some species (Fig. 4A), but there is generally a monotonous sameness to the flower morphology ( Fig.  4B, C).
The calyx is carnose at the base, usually becoming membranous and hyaline at the apex. The calyx is typically costate when dry, although inS. yapa, which has a more carnose calyx, the costae are not apparent. The calyx may be cupulate (sides more or less parallel), campanulate, or urceolate.
Petals are generally membranous with hyaline and denticulate margins and are obovate to more nearly spatulate. In S. yapa, the petals are triangular-ovate and basally connate. They are generally noncostate when dry, but S. mexicana is noteworthy for its costate petals. A pattern of papillate cells is often visible on the adaxial surface of the petals (Fig. 4A). It resembles the letter "W" in parentheses, with the base of the "W" pointing toward the base of the petal. This (W) pattern is sometimes only weakly apparent, and its presence varies from individual to individual. The pattern may play a role in pollination ecology.
Stamens are in two whorls of three, connate basally, and adnate to the petals. The filaments are generally long triangular in shape (but acuminate inS. yapa, Fig. 4A). Typically, the stamens are ascending to spreading, with the filaments weakly sigmoid, but in some species the antipetalous stamens are reflexed, and the antisepalous stamens are ascending to erect (Fig. 4C). Anthers, twice as long as they are wide, are yellow, versatile, and dehisce latrorsely. Pollen in Saba/ possesses an elliptical amb and a finely reticulate exine. It is uniform throughout the genus and has no apparent taxonomic value (Sowunmi 1972).
Gynoecia are composed of three fused carpels and are variously shaped: conical, pyriform, or lageniform. Gynoecia are oflittle taxonomic value, as their size and shape varies considerably among individuals. The stigma is obscurely three lobed and papillate. It is rounded or truncate and about 0.5 mm in diameter.

Fruit and Seed
Fruits of Saba/ are usually single-seeded berries. Occasionally, more than one ovule matures, and two-or three-lobed berries result. Fruits are spherical, oblate, or pyriform, with the style and stigmatic remains persisting basally along with the calyx and, more rarely, the perianth. Fruits range in size from 6.5 to 27.5 mm in diameter and from 6.5 to 22.5 in height. Fruits are green when immature, passing through a brownish stage, eventually becoming black in most species; in some species, however, fruits are dispersed while still in the greenish brown stage.
The epicarp is smooth and thin; the mesocarp is thick and sweet in most species but may be thin and dry in S. minor. The endocarp is dry and membranous and shiny brown, separating easily from the seed.
The seed is oblate-spherical and brown to black. The seed is concave on the funicular end, but the depression may be more or less filled with the funicular remains. Seeds are 4.5-18.8 mm in diameter by 4.0-11.2 mm in height. The testa is smooth and shiny, but immature seeds, when dry, take on a rugose appearance that has been mistaken to have taxonomic significance in S. uresana. The embryo is small ( < 2.0 mm in length) and poorly developed at the time of seed dispersal. The embryo position is betrayed by a small ringlike depression in the testa. The embryo may be located equatorially to supraequatorially. Very rarely, individuals may be found to possess seeds with subequatorial embryos. Beccari (1907) attached considerable significance to embryo position, but over a large sample size, meaningless variation in this character state is readily apparent. The endosperm is bony, white, and homogeneous.

Introduction
Palms have been the subject of much attention from plant anatomists; although this attention has most often been directed to fundamental problems of xylem or phloem transport Zimmermann 1965, 1966;Parthasarathy and Tomlinson 1967), development of the plicate leaf (Kaplan, Denger, and Denger 1982), or derivation ofthe diverse inflorescence structures within the family (Uhl and Moore 1978). The application of anatomical data to systematic problems in the palm family has most often been at the genus level or above (Tomlinson 1961). Several workers (Barfod 1988;Glassman 1972b;Read 1975;Uhl 1972Uhl , 1978a have demonstrated the value of anatomical data at the species level, and of course for palm palaeobotanists (Daghlian 1978;Dilcher 1971), anatomy is their stock in trade.
Relatively little anatomical work has been interpreted in an ecological light. Uhland Moore ( 1977), who discussed floral anatomy and pollination, and Barfod (1988), who described leaf anatomy, both drew correlations with ecology.
Saba/ has received little attention from comparative anatomists. Limited ethnobotanical data (E. Sandoval pers. com.) suggested that differences in usefulness of the species for thatch and basketry may reflect differences in their anatomy. The abundance of sterile specimens and great ecological diversity suggested that leaf anatomical studies could produce useful taxonomic, ethnobotanical, and evolutionary information.

Materials and Methods
Transverse sections were prepared for anatomical study following the methods outlined in Martens and Uhl (1980). Although numerous collections were examined (Appendix 1), quantitative data from only one specimen per species are presented in Table 1. Collection data are abbreviated in Table 1; complete data for each specimen may be found in the taxonomic treatment. Leaves were sampled from the middle segment (of one side of the leaf) at or near where the segments become free. Pickled or rehydrated lamina samples, including the midvein, were desilicified in HF, dehydrated in an alcohol series, embedded in paraffin, and sectioned at 12-14 JLm on a rotary microtome. Sections were stained with a standard safranin-fast green combination and mounted in a synthetic resin (Permount).
Lamina samples were cleared using 2.5% NaOH at 60 C for 12-24 h, then bleached with one-third strength commercial bleach for 5-10 min. Cleared sam- ples were dehydrated in an alcohol series, stained with a mixture of safranin and fast green, then destained for approximately 2 h in absolute ethanol before being transferred to ethanol-xylene ( 1: 1 ), and ultimately to xylene. Samples were mounted in synthetic resin. Anatomical measurements were made with the aid of a digitizer. The slide image was projected onto the digitizing pad, and data were quantified with the software package SigmaScan, version 3.9 (Jandel Scientific, Corte Madera, California). Statistical analysis was possible within SigmaScan.

Results
The leaf of Saba/ is generally isolateral, with cutinized epidermises, and with one exception, stomata on both surfaces. Trichomes are absent from the lamina but may be present near the hastula along the mid-and suture veins. Epidermal cells lack sinuous anticlinal walls. Stomata are restricted to the intercostal regions and are plugged by cutinous substances. Beneath each epidermis is a hypodermis one or two cell layers thick. The chlorenchyma is palisade like beneath each surface and surrounds a mesophyll comprised of large spherical cells. Septate fibers in strands are attached to the abaxial hypodermis; although these are sometimes converted into small vascular bundles. Large vascular bundles, sometimes encased in bundle sheath extensions, are present; small vascular bundles, suspended from the adaxial hypodermis by fiber bands, are interspersed among the large vascular bundles. Large transfer cells are present around the middle of bundle sheath extensions, large vascular bundles, and small vascular bundles. Phloem is present in two to three separate strands, vessels are present in the metaxylem, and parenchyma is scattered within the vascular bundle. Transverse commissures are variable in abundance and distribution, but are always present. The midvein is composed of two to many vascular bundles, often encased in a single fiber sheath. Smaller fiber bundles are arrayed around the periphery of the mid vein. On the adaxial side, there is a prominent region of bulliform cells, which function in the expansion of the leaf. The suture vein, the rib joining the margins of two leaf segments, is similar in all respects to the midvein, but is more highly variable throughout its length. Stegmata and raphides may be abundantly present throughout the leaf.
Lamina and mid vein transections and leaf clearings of all species are illustrated in Figures 5-13. The results of the comparative study are summarized in Table  1. Both quantitative and qualitative data are presented. Differences in both reflect at least two selection pressures: changes in leaf size and adaptation to aridity. Closely allied to the latter selection factor is defense against herbivores. Each character given in Table 1 will be discussed with these evolutionary constraints in mind.

Discussion
Lamina thickness. -A clear trend in Saba/ is toward thinner lamina in smaller leaves, but this trend is countered by the need for thicker leaves in more arid habitats. Saba/ minor and S. etonia have the smallest leaves in the genus, but the leaves of the latter, a xerophytic species, are considerably thicker. Saba/ mauritiiformis has large leaves overall, but the segments are clustered into groups of 2-3, so structural support is modified, and the lamina is thin as if the leaf were small. Saba/ yapa also has clustered leaf segments, but life in drier habitats has selected for medium thick leaves.

Bundle sheath extensions (BSE's).
-Nearly all species have BSE's ( Fig. 5A, D, 6A, D, E, 7 A-D), and their presence is the unspecialized state within the genus. Six species (e.g., S. yapa, Fig. 7E) possess large veins sheathed by fibers, but the sheaths seldom extend from hypodermis to hypodermis. The lack of well developed BSE's is taken to have great phylogenetic importance.
The number of BSE's or large veins is also given for each species in Table 1. The range is from 18 to 56 per segment, with most species having 30-34. The evolutionary significance of either reduction or proliferation is unclear, as they appear to have no correlation with size (i.e., support) or drought adaptation.
Small adaxial vascular bundles (SVB's) between BSE's. -Generally, between a pair of BSE's, one finds three SVB's attached to the adaxial surface by a thick sheath of fibers. The central SVB is larger than the other two. Although one can easily imagine three as the "base number" on which reductions and elaborations in SVB number are made, outgroup comparison with Washingtonia and Brahea points to seven as the unspecialized state. Transformation of SVB's into BSE's produces the vasculature seen in S. minor (Fig. 6E), in which one SVB alternates with each BSE. In two species, S. domingensis (Fig. 5C) and S. guatemalensis (Fig. 5E), reduction of BSE's has led to more than seven SVB's between each BSE pair. In S. bermudana, S. mexicana, S. pumos, and S. uresana, the vasculature pattern is uneven within the leaf.
As with the preceding character, the functional or ecological significance ofSVB number is not known.
Cuticle thickness-What is measured here is actually the cuticle together with the heavily cutinized outer wall of the epidermis. In those cases (indicated in Table  1) in which the adaxial cuticle differs in thickness from the abaxial cuticle, the adaxial cuticle is always thicker. This is a character which clearly shows ecological adaptation. Not unexpectedly, the species with the thickest cuticle isS. uresana ( Fig. 7D) from the dry thorn scrub of Sonora.
Average stomatal depth. -Like the preceding character, this one shows ecological adaptations. For this character, the epidermis is measured from the upper lip of the guard cells to the surface of the leaf. If the average abaxial cuticle thickness is subtracted from the average stomata depth, one is left with the "true" depth of the guard cells below the epidermis. The data thus transformed, S. uresana (7 JLm), followed by S. minor (6 JLm) and S. maritima (5 JLm), have the most sunken stomata. Saba! minor is a bit anomalous here in that it is not a palm of xeric or even seasonally dry areas. Saba/ mauritiiformis and S. rosei have the shallowest abaxial stomata. If the adaxial data are similarly transformed, S. mexicana has the deepest stomata.
Fiber bundle height. -Fiber bundles attached to the abaxial hypodermis vary in height according to the size of the leaf and thickness of the lamina (Fig. 5-7). Although fiber bundles may play a role in herbivore defense, they are more likely acted upon by constraints in leaf size and lamina thickness. The broad overlap in range of values limits their systematic usefulness. Large vessel diameter within BSE.-Width of the large vessels of the metaxylem is likely a characteristic under strong selection pressure. While large vessels are more efficient conductors of water, they are more susceptible to cavitation than narrow vessels. Lamina thickness is also an important constraint. Saba! domingensis, S. guatemalensis, S. maritima, S. rosei, S. uresana, and S. yapa have wider vessels than the remaining species. This group includes mesophytes (S. guatemalensis and S. rosei) as well as the most xerophytic species of the genus, S. uresana. Another xerophyte, S. etonia (Fig. 5d), has small vessels, similar in size to those of S. minor (Fig. 6e), a mesophyte. These results suggest that leaf size and vessel efficiency are stronger selection constraints than vessel safety.
Vessels per group within BSE. -Only three species are exceptional in having more than one wide metaxylem vessel per BSE bundle. They areS. minor (2.8 v/gr), S. palmetto (2.1 v /gr), and S. uresana (3.4 v/gr). In petiole vascular bundles, vessel number is known to have systematic significance at high taxonomic levels (Klotz 1978), and the Coryphoideae is the most variable in this character. Its ecological significance is not clear.
Palisade layers. -A leaf of Saba! does not posses a palisade layer in the traditional dicotyledonous sense of that word. Rather, there is both an upper (adaxial) and lower (abaxial) layer of compact, somewhat elongated chlorophyllous cells surrounding and intergrading with a mesophyll of larger (by a factor of 2-4) isodiametric cells. In no species of Saba! are the palisades clearly demarcated from the mesophyll; however, an important systematic feature is whether the palisades are equal in height (similar number of cell layers) or unequal in height (different number of cell layers). In Saba!, some species possess adaxial palisades that are 1.5 to 2 times as large as the abaxial palisades. These species are the Antillean species plus S. mexicana, S. guatemalensis, and S. mauritiiformis.
Transverse commissures (Fig. 8-9).-The pattern and distribution of transverse commissures in the leaves are very important systematic characters. In Saba!, patterns of transverse commissures tend to be either long-looping and prominent (as in S. maritima, Fig. 8F, or S. yapa, Fig. 9G) or short and straight and often obscure in the dry leaf (as in S. etonia, Fig. 8D, or S. minor, Fig. 9B). In addition, transverse commissures when long-looping tend to run below the middle of the mesophyll; short and straight transverse commissures tend to be found in the middle of the mesophyll. These two characters, while of uncertain ecological value, have great systematic importance and are not readily modified by environmental factors.
Midvein shape.-The shape of the midvein in transection, whether triangular, rectangular, or trapezoidal, is apparently correlated with leaf size. Species with small leaves most often have triangular midveins, and those with larger leaves have rectangular midveins. Medium size leaves have trapezoidal midveins. In this instance, structural support is the most important evolutionary constraint for Saba!.  Fig. 12A) typically lack fibers altogether, a fact that suggests the fibers may play a major role in structural support of larger leaves. In other species, the fibers are either scattered throughout the expansion region (as in S. domingensis, Fig. 1 OC) or arrayed along the axil of the fold, i.e., along the uppermost side ofthe midvein (as inS. bermudana, Fig. lOA).
Midvein vessel diameter.-The wide metaxylem vessels of the midvein show considerable variation in average diameter. Once again, diameter appears more closely correlated with leaf size than with constraints of ecology. It is oflimi ted systematic importance, and its use in the phylogenetic analysis would be redundant.
Number of vascular bundles.-Between 2 and 11 vascular bundles containing wide metaxylem vessels are present in the mid veins of Saba!. This character is strongly correlated with leaf size, or at least, functional size.
Tannin deposits.-There is great variation in the distribution and abundance of tannin deposits within the leaf (Fig. 5-7). These cells are often idioblastic and contain dark-staining substances, most likely procyanidin. They are present to varying degrees in the epidermis, hypodermis, mesophyll, around or within bundle sheath extensions, around transverse commissures, and/or around or in the bulliform cells of the midvein. There is great taxonomic and systematic value to the pattern of tannin deposits; although there is some intraspecific variation. Tannins are more readily observed in sections offresh or pickled material than in sections made from old, dried collections.
Other cellular inclusions. -Saba! has both raphides and stegmata. The raphides are found in idioblastic cells in the mesophyll. Silica bodies (stegmata), roughly spherical and of varying sizes, are present in linear files along vascular bundles and/or transverse commissures (Fig. 9H). Undoubtedly, these inclusions arose as adaptations against herbivory; however, interpopulational variation has been observed in the abundance ofboth types of inclusions. These differences are thought to be random, and they probably do not reflect differences in herbivore pressure.

Introduction
Although palms are rich in flavonoids and other so-called secondary compounds (Bate-Smith 1962;Harbome, Williams, Greenham, and Moyna 1974;Williams, Harbome, and Clifford 1973), rarely has flavonoid chemistry been used at the --+ specific level to resolve taxonomic questions. Exceptions are the studies by Balick and Cooper-Driver (in Balick 1986) on Oenocarpus and Jessenia (Arecoideae), Madulid (1980) on Plectocomia (Lepidocaryoideae), Williams, Harbome, and Glassman (1985) on Attalea and its allies (Arecoideae), and Zona and Scogin (1988) on Washingtonia (Coryphoideae). In all cases, differences and similarities in flavonoid profiles assisted in delimiting species or species groups; although, in the case of Attalea, a certain amount of infraspecific variation was detected.
Saba! has received only cursory examination by phytochernists (Harbome et al. 1974;Williams et al. 1973). An in-depth examination of flavonoid aglycones and C-glycosides was undertaken with the hope that variation in flavonoid constituents would shed light on certain taxonomic problems and phylogenetic relationships.

Materials and Methods
Leaf samples were obtained from wild and cultivated plants (Appendix 1); material from two taxa (S. guatemalensis and S. miamiensis) was not available. Samples were dried prior to flavonoid extraction and the analysis followed the methods outlined in Zona and Scogin (1988).
A presumptive test for negatively charged flavonoids was performed in the following way: flavonoids were extracted in 85% methanol for 1 h at room ternperature, reduced in volume, and chrornatographed in two dimensions in TBA and HOAc. Spots were visualized under ultraviolet light with and without ammonia vapor. Negatively charged flavonoids were recognized by their distinctive cornet shaped spots and by their low Rf values in the TBA.
Saponins were presumed present if a stable foam persisted in an aqueous solution for more than 20 minutes. This test was performed on most species during aglycone preparation and extraction.
More than one individual was sampled for most of the species. In many instances intraspecific variation in the flavonoid profiles became apparent. In these cases it was assumed that the greater number of compounds were present but that some of the compounds were not present in detectable amounts.
The replacement of flavonols by flavones is thought to be a specialized characteristic within angiosperms (Bate-Smith 1962;Harbome 1966), as is 0-rnethylation (as in tricin). Isoorientin and orientin are not thought to be interconvertible because C-glycosylation is an early biogenetic step, not merely a late or terminal C-glycosylation of luteolin (Wallace, Mabry, and Alston 1969). For this reason, orientin and isoorientin, although similar, are considered independent characters.

Results
The results are presented in Table 2. Only four aglycones and C-glycosides were detected from the leaves of Saba!. Tricin, a methylated flavone, is present in all samples of all 13 species. Orientin and isoorientin, both flavone C-glucosides, +- were detected in 13 and 11 of the taxa, respectively. A fourth and apparently rare compound, vitexin, another flavone C-glycoside, was seen in two taxa. Procyanidin was detected, either strongly or in trace amounts, in about half the species. Saponins and negatively charged ftavonoids were detected in 11 and 10 species, respectively. Two-dimensional chromatography of methanol extracts (viz., glycosides) demonstrated the abundance of ftavonoids in Saba!. The spots were too numerous to allow further analysis of the glycoside profile of each species.

Aavonoids and saponins
are narrow endemics (S. causiarum, S. maritima, and S. yapa less so), yet these appear to show as much intraspecific variation as the wide ranging species. These discrepancies may represent quantitative or qualitative intraspecific (i.e., interpopulational) differences, but the exact nature of the variation cannot be discerned at this level of inquiry. Because ofthe significant amount ofvariation in the flavonoid profiles of many species, systematic conclusions based on flavonoid data are made only tentatively. Lack of variation in tricin and abundant unaccountable infraspecific variation in orientin and isoorientin all but excludes these data from the phylogenetic analysis. Vitexin (present in only 3 of 3 7 samples) seems to be rare in Saba/, but its rarity may reflect the difficulty in the reliable detection of compounds present in small quantities. A similar difficulty in applying variable flavonoid data to phylogenetic questions was encountered by Williams et al. (1985), who when faced with such variation were unable to draw systematic conclusions from their data. Saponins, procyanidin, and negatively charged flavonoids have been incorporated into the phylogenetic analysis.
A matter of particular interest, and one not apparent in Table 2, is the localization ofprocyanidin within the leaf tissues as confirmed by anatomical studies. Procyanidin was detected in S. causiarum, S. domingensis, S. maritima (all of the Greater Antilles), S. rosei, S. pumas, and S. uresana (all of western Mexico), as well as S. bermudana and S. palmetto. In the Antillean species, S. palmetto, and S. bermudana, procyanidin is localized in tannin sacs scattered through the mesophyll and surrounding the bundle sheath extension. In Saba! rosei, procyanidin is present in abundant tannin sacs within the mesophyll. Saba/ pumas, in which procyanidin is present in only trace amounts, has far fewer tannin sacs, compared with S. rosei. In contrast, S. uresana has a mesophyll essentially devoid of tannin sacs but has large amounts of tannin deposited in the bulliform cells (expansion cells) found along the mid vein and suture vein (i.e., in the axil of each plication in the leaf). These tannin deposits can be shown to be procyanidin by extracting lamina tissue minus the veins, whole lamina tissue, and vein tissue. Only the latter two samples yield detectable levels of procyanidin.
Much has been written concerning the function of flavonoids in plants as herbivore deterrents (Levin 1971 and references therein). Certainly, if the major function of flavonoids is herbivore deterrence, Saba! palms are well protected. Flavonoids, in conjunction with anatomical/chemical protection mechanisms (viz., silica bodies, raphides, fiber}, would then form a seemingly impenetrable protective phalanx around vital tissues. Recent work in the area of phytochemical ecology suggests a correlation between the duration of leaves and the chemicals invested in their defense (Chabot and Hicks 1982;Cooley, Bryant, and Chapin 1985;Mooney and Gulmon 1982). Leaves of Saba! palms are evergreen and persist for more than one year (Zona pers. obs.). Large leaves of palms represent a considerable investment of photosynthates and are not rapidly replaced, even in areas of seemingly abundant resources. The prediction by Mooney and Gulmon ( 1982) that plants in resourcelimited environments will defend their leaves rather than replace them seems justified in the case of Saba!. Levin (1971), however, suggested that the most heavily defended plants are not just those with long-lasting leaves but rather those which are late successionary, tropical K strategists, i.e., those plants of predictable habitats. Plants of nonpredictable habitats (weeds, temperate plants, and early successionary r strategists) are less likely to invest heavily in defense. Apparentness to herbivores is implicit in Levin's argument. Saba! does not comfortably fit Levin's model. Saba! is typically a weed of tropical grasslands, wetlands, or pastures-all unpredictable habitats-and appears to have many characteristics of an r strategist (early succession or canopy gap colonizer, high annual rate of fruit set, small seeds). Unlike many herbaceous or perennial weeds, Saba! has large, long-lived leaves. Flavonoid data would seem to support the hypothesis that defense ofleaves is more positively correlated with the longevity of the leaves (predictability in time) than with habitats (predictability in space).

Pollination
Despite its abundance and relative accessibility, Saba! has been largely ignored by biologists interested in the interactions between plants and their pollen vectors. To date there are only two published accounts of pollination in Saba! in its native habitat; Saba! palmetto was studied by Brown in several localities in the southeastern U.S. (Brown 1976), and S. etonia was studied by Zona (1987) in southern Florida. Knuth (1904) reported observations made on Saba! (spp. unknown) cultivated in Indonesia. Knuth (1909) cited work by Delpino who studied the pollination of S. minor, but apparently Delpino made his observations on palms cultivated in Europe. What follows are observations by the author of pollinators and flower visitors along with a discussion of pollination ecology as it relates to the reproductive isolation of species of Saba/.
The pollination biology of S. bermudana is quite readily understood: the principal pollen vector is the introduced European honeybee, Apis mellifera. Prior to the widespread naturalization of the honeybee, Megachile pruina pruina, an Augachlora species, and a Halictus species may have had important roles in the pollination of this palm; however the latter two species have not been seen in this century and are thought to be extinct. The Megachile is thought still to exist in small numbers on Nonsuch I. but is absent from the main island.
The principal pollinator of S. etonia is a member of the Megachilidae, Megachile albitarsus (Zona 1987). Other solitary bees are important, viz., Megachile xylocopoides, Augachloropsis metallica, Xylocopa micans, and Colletes mandibularis, as is Apis mellifera. Flies of the families Syrphidae and Bombyliidae play a minor role in pollen transport. This species is slightly protandrous. Brown (1976) reported the major pollinators of S. palmetto to be the halictid bees Augachlora pura pura, Agapostemon splendens, and Dialictus spp. The introduced honeybee is also an active pollinator. Brown (1976) stated that the species is protogynous.
Saba/ mauritiiformis was observed in Panama, where its flowers are visited and likely pollinated by bees of the genera Dialictus and Augachloropsis, both of the Halictidae. In Trinidad, this species is visited by numerous bees.
Saba/ palmetto and S. maritima growing in the Jardin Botanico Nacional de Cuba, Havana, are visited by numerous species and individuals of Hymenoptera, viz., bees and wasps. Flowers of S. causiarum were collected in the Dominican Republic also with numerous bees.
These observations suggest that Hymenoptera, especially solitary bees of the Megachilidae and Halictidae, are probably the principal pollinators for the genus. Saba/ has many morphological traits that suit it to bee pollination. Several specializations for bee pollination, as listed by Henderson (1986), are apparent in all species of Saba/: a loose, open paniculate inflorescence exserted well beyond any sheathing or appressed bracts, sweet fragrance, and copious nectar production. To these criteria can be added hermaphroditic flowers that are short lived and that function during the daylight hours when bees are active and floral parts thin in texture.
The pattern of papillate cells found on the petals of Saba/ may serve as nectar guides. The pattern may differentially reflect light and thus guide visitors to the septal nectaries. The petals of Saba/ have not been examined under ultraviolet light to see if they show nectar guide patterns.
Saba/ minor is reportedly protogynous (Knuth 1909) as is S. palmetto (Brown 1976), but S. etonia is weakly protandrous (Zona 1987). Morrow (1965) characterized the genus as "perhaps slightly protandrous." Further research is needed to resolve the contradictions in the literature.
Virtually nothing is known about whether hybridization in Saba/ is possible and the relationship between hybridization and speciation in Saba/. Hybridization has been implicated (Zona 1985(Zona , 1987 in the origin of one species, but evidence is purely circumstantial. Mixed populations of two or three species can be found in the wild (Bataban6, Cuba, for example), but such populations appear to contain no hybrid intermediates. Isolation barriers, beyond those of ecology, phenology, and pollinator specificity, are likely in play.

Seed Dispersal and Predation
The fruits of Saba! are typically black with a generous sweet pericarp surrounding 1-3 seeds. A few species, viz., the species ofwestern Mexico, have large fruits which are often greenish brown rather than black and which have a very thick pericarp. Saba! minor has a small fruit with a notably thin, dry pericarp. Hemsley (1885, p. 49) suggested that bird dispersal was the most likely explanation for the arrival of Saba! bermudana ["S. blackburniana"] to Bermuda, and drew a parallel between the dispersal and arrival ofBermuda's indigenous juniper (Juniperus bermudiana L.) and its indigenous palmetto. Guppy (1917, p. 16) noted that the Jamaican palmetto, S. maritima ["S. umbraculifera"] has fruits that are not buoyant and are seldom, if ever, found among beach drift. Guppy concluded that evidence weighed in favor of bird dispersal for Saba! and that the present distribution of the genus indicated past dispersal events.
In contrast, Brown (1973) suggested that water dispersal, hydrochory, was the principal mode oflong-distance dispersal for S. palmetto. His experiments, floating mature dry fruit in 3.5% NaCl solution, showed that buoyancy varies among populations from low values of 0.0-3.0% floating after 3 weeks to 45.4% seeds afloat. Seed viability after 8 weeks in salt water ranged from 30% to 60%. Brown suggested that South Carolina and North Carolina coastal populations were derived from more southerly populations. He noted that northern populations had more buoyant fruits than southern populations, and suggested that this trend would be the expected outcome if northern populations were established by watertransported fruits from southern populations. Although Brown's hypothesis may be correct, Brown admits that confirmation awaits more rigorous testing. Saba! mexicana, another wide-ranging continental species, also has buoyant fruits (Zona pers. obs.).
Hydrochory may play a role in the dispersal of S. minor. This palmetto grows along stream banks and seasonally flooded areas where flooding corresponds with the fruiting season. Its pericarp is notably scanty. In a flotation experiment performed with 109 fresh S. minor fruits collected from Gainesville, Florida (Perkins & Herring 987), 43 seeds (39%) remained floating in distilled water after three days. After 7 days, only 6 seeds (6%) remained floating. Although post-flotation germination tests were not performed, the seeds appeared healthy and viable. These data would suggest that short term hydrochory is at least possible.
Mammals too play a large role in the dispersal of Saba! seeds in Florida; known dispersers are the Florida black bear, Ursus americanus, and the raccoon, Procyon lotor (Maehr and Brady 1984;Martinet al. 1951). Seeds of S. palmetto and S. etonia, in apparently viable condition, have been found in bear dung (Zona pers. obs.). The bat Artibeus jamaicensis is reported to feed on fruits of S. palmetto ["S. parviflora"] in Cuba (Silva 1979). Mammals, rather than birds, may play a greater role in the dispersal of the large-fruited Saba! of western Mexico, S. rosei, S. pumas, and S. uresana. Their fruits are more often greenish brown rather than black. Dull coloring and large size are suggestive of mammal dispersal (van der Pijl 1982).
The long-distance dispersal of Saba! by animals, notably birds, would agree with the biogeographical data, i.e., insular Saba! distributions. For example, the activities of the white crowned pigeon, Columba leucocephala, a nomadic frugivore found throughout the Antilles, Florida, and eastern Mexico, may contribute to the dispersal of Saba!.
Like many good colonists, Saba! is readily dispersed and probably does so by both hydrochory and zoochory. Dispersal is, and probably always has been, unpredictable and stochastic. The survival of most species of Saba! depends on exploiting new and disturbed environments, which are themselves unpredictable.
Saba! has been remarkably successful in this regard. In addition, the patchy distribution of S. mauritiiformis in Central and South America supports a long-distance dispersal explanation. This species skips over large areas of apparently suitable habitat in Honduras, Nicaragua, and El Salvador, only to reappear in extreme southeastern Costa Rica, eastern Panama and the Perlas Archipelago, north coastal South America and Trinidad. There is nothing in its present day distribution that suggests widespread extinction, rather its unpredictability suggests that the distribution is the result of chance dispersal events, most likely by birds. Not coincidentally, the most widely distributed species of Saba! (S. causiarum, S. mauritiiformis, S. minor, S. palmetto, S. maritima, S. yapa) are those with small fruits (less than 12 mm in diameter). Species of Saba! are hosts to beetles of the genus Caryobruchus (Coleoptera: Bruchidae: Pachymerinae): adults feed on the nectar (Brown 1973) and larvae feed on the seed endosperm. The taxonomy of Caryobruchus is not settled, but clearly more than one species of the genus can be found throughout the range of Saba!. Caryobruchus g!editsiae is known from the southern United States (Brown 1973;Paxson 1969), the Gulf coast ofMexico (Olvera 1981), the Greater Antilles, and Bermuda (J. Kingsolver in litt.). A second, much larger species, tentatively referable to C. curvipes, is known from the larger fruited Saba! of western Mexico.
It is not clear, however, if size of the adult bruchid is the direct result of a larger food source as a larva. Larvae of both species feed on Saba! in the wild but are known from seeds of cultivated coryphoid palms of other genera, e.g., Pritchardia Seem. &H. Wendl., Serenoa Hook. f., CoccothrinaxSarg., andPhoenixL. (Olvera 1981).
Adult Caryobruchus have been taken from the following species of Saba!: S. bermudana, S. causiarum, S. domingensis, S. etonia, S. mauritiiformis, S. mexicana, S. minor, S. palmetto, S. maritima, S. pumas, S. rosei, S. uresana, and S. yapa. Brown (1973) discussed aspects of the life history of C. gleditsiae on S. palmetto, and found levels of predation as high as 92%. High levels (ca. 50%) are also known from S. uresana (Zona pers. obs.). Generation time is not known but is apparently short. Caryobruchus has the potential of being a highly efficient predispersal predator; however, Brown (1973) noted that level of predation can vary wildly from year to year (92% in one population in 1972, and 4% the following year). The causes for these fluctuations are not known.
In Florida, larvae of C. gleditsiae are parasitized by a wasp, Heterospilus sp. nov. (Hymenoptera: Braconidae). It has not been observed on other species of Caryobruchus and seems confined to Florida. Its life history is poorly known.

HISTORiCAL BIOGEOGRAPHY
The modem distribution of Saba! is very different from its historical distribution. Fossil Saba! and Sabalites are known from the Soviet Union (Takhtajan 1958), Great Britain (Reid and Chandler 1933), Alaska (Wolfe 1972), Vancouver Island, and Japan (Kryshtofovich 1918), as well as New Jersey, Delaware, Maryland, South Carolina, Kentucky, Tennessee, Arkansas, Texas, Montana, Wyoming, Colorado, New Mexico, and California (Daghlian 1978;Noe 1936;Read and Hickey 1974). Given this distinctly north temperate distribution of fossils, how can we reconcile the presumed origin of the Arecaceae in West Gondwanaland (Moore and Uhl 1973) with the north temperate origin of Saba!?
Saba! is probably of Laurasian origin (Moore in Raven and Axelrod 197 4). If palms arose in West Gondwanaland, then the progenitors of Saba! probably spread to Laurasia, where the genus evolved into recognizable form. The coryphoid palms of Laurasia (including Saba!) diversified independently from those taxa that remained in Gondwanaland (Dransfield 1987;Uhland Dransfield 1987). Radiation in Laurasia followed by Neogene or Pleistocene extinction is a likely and parsimonious explanation for the modem and historical distribution of Saba!.
An alternate hypothesis, that Saba! evolved in West Gondwanaland and is recentlybeginningtoinvade NorthAmerica (Comer 1966;Long 1974), completely ignores the fossil record that demonstrates: 1) that Saba! existed in North American long before a land connection was established between North and South America; and 2) that Saba! existed in Europe and Asia after the North Atlantic and Bering land bridges were severed. Furthermore, this hypothesis does not account for the absence of Saba! in some parts of Central America and in most of South America.
Saba! was a component of what Wolfe ( 197 5) called the "boreotropical flora." The equable climate of the Tertiary (Buchardt 1978;Wolfe 1975) favored the rapid spread of a mixed flora with modem counterparts from temperate deciduous hardwood forest (e.g., Juglans L., Carpinus L., Betula L., Liquidambar L.) and tropical (especially paleotropical) rain forests (e.g., Mastixia Blume and members of the lcacinaceae). The classic London Clay flora, of which Saba! is an element (Reid and Chandler 1933), represents the boreotropical flora. There is no reason to suppose that the ecological requirements of Saba! in the Tertiary were any different than those of the genus today. In fact, Saba! is known from European fossil assemblages that contain many of the same genera that can today be found growing alongside Saba! in eastern North America, such as Serenoa Hook. f., Leitneria Chapm., and Asimina Adanson (Tiffney 1985).
A preponderance of evidence (Daghlian 1978;Dilcher 1971) suggests that the paleoecology of Saba! was not appreciably different from its modem ecology: Saba! is likely to have grown in warm temperate to cool tropical regions with continually moist to seasonally dry moisture regimes, growing in broadleafwoodland, riparian, or perhaps even swamp communities. Axelrod (1975)

Geologic History of North America
Several phenomena figure prominently in the geologic history ofNorth America (including Mexico) since the origin of angiosperms in the Cretaceous: 1) land connection with Europe across the North Atlantic via the North Atlantic land bridge until approximately 49 MYBP (Eocene) and connection to eastern Asia via Beringia periodically throughout the Tertiary (Tiffney 1985); 2) isolation from South America until 5.7 MYBP (Upper Miocene) at which time the Panamanian isthmus arose; 3) massive orogeny in western North America; and 4) extensive glaciation during the Pleistocene.
Fragmentation of the range of Saba! began in the Eocene as the Rocky Mountains began to form (Early Eocene) and the North Atlantic land bridge was severed (Late Eocene) (Tiffney 1985). Orogeny of the Sierra Madre Occidental and Sierra Madre del Sur in the Miocene (Dressler 1954) further fragmented the range of Saba!, effectively isolating the western North American species from those in eastern North America. Western North American elements of the boreotropical flora retreated southward (Leopold and MacGinitie 1972), and uplift of the Sierra Madre Oriental in the Pliocene further isolated the species of western North America from those of the southern coastal plain. Residual populations of Saba! in the Central Plains, Great Basin, and Altiplana would have succumbed to the gradual climatic deterioration caused by inland rain shadows and cooler temperatures (Leopold and MacGinitie 1972).
Climatic deterioration in the Oligocene resulted in continual cooling of the Northern Hemisphere, eventually resulting in the glaciation of the Pleistocene (Tiffney 1985;Wolfe 1975). Such cooling and subsequent glaciation would have severely diminished the extensive distribution of Saba! in the Northern Hemisphere. As the populations of Saba! retreated southward in Eurasia, they were likely pinned against east-west running mountain ranges and eventually extirpated. In North America, Saba! was extirpated from the northern regions but was able to retreat to Mexico, the southeastern coastal plain, and the Antilles, as did Nyssa L., Celtis L., Carya Nutt., and many other genera (Dressler 1954;Graham 1973). The modem distribution of Saba! (with the possible exceptions of S. mauritiiformis and S. palmetto) is very similar to its distribution during the last glaciation.

Geologic History of the Antilles
The geologic history of the Antilles during the Cenozoic is exceedingly complex and has been the subject of much speculation (Hedges 1982;MacFadden 1980;Rosen 1975;Tarling 1980). The existence of the Caribbean Plate is now accepted, as is its eastward movement to its present position (Malfait and Dinkelman 1972). Post-Miocene marine transgressions, periodic uplift, and extensive subsidence, however, confound efforts to interpret geological data in a biologically meaningful way.
There is strong geologic evidence that the Greater Antilles arose de novo and that they were never attached to any continent (Malfait and Dinkelman 1972;Pregill 1981;Tarling 1980). The Greater Antilles arose in the late Cretaceous from subduction of oceanic lithosphere of the Caribbean Plate beneath ancestral southern Guatemala, Honduras, Nicaragua, the Nicaraguan Plateau, and the Cayman Ridge. The beginnings of Jamaica, Hispaniola, and eastern Cuba consequently arose as a volcanic arc parallel to the subduction zone (Pregill 1981). At the end of the Eocene, the Cayman Ridge broke away from the Nicaraguan Plateau and carried eastern Cuba, Jamaica, and much of Hispaniola to the northeast relative to North America. Subduction of the Atlantic oceanic crust beneath the Caribbean Plate resulted in volcanism that led to the formation of central Cuba, eastern Hispaniola, and Puerto Rico (Pregill 1981). Hispaniola, Jamaica, and Puerto Rico are moving eastward along the Puerto Rican Trench, although the rate of movement is in dispute (Hedges 1982).
Of greater interest to biogeographers are the dates at which the land masses became emergent. There is strong evidence that western and central Cuba have been emergent since at least the Eocene, although uplift of eastern Cuba did not take place until the late Miocene. The Virgin Islands, Puerto Rico, and part of Hispaniola may have been contiguous and emergent in the Oligocene (Graham and Jarzen 1969), although the evidence for this hypothesis is thin. Southern Hispaniola has been emergent since Pliocene times. Jamaica arose by seafloor uplift during the Miocene, but was largely inundated in the Oligocene (Buskirk 1985). As young as the Antillean land masses are, their coastal habitats (where Saba! is found) are even younger, since sea level changes during the Pleistocene ranged from about + 20 m to -100 m or more (Mann, Taylor, Burke, and Kulstad 1984).
There is absolutely no geologic evidence for an Antillean-Guatemalan Land Bridge, as envisioned by Asprey and Robbins (1953) and more recently imagined by Borhidi (1985). The geologic interpretation presented by Malfait and Dinkelman (1972) does suggest that the Yucatan Peninsula has always been closest to Cuba, thus emphasizing the importance of the Yucatan-to-Cuba migration route. This migration route has had the greatest influence in the establishment of the Caribbean flora (Howard 1973;Raven and Axelrod 1974).
The Lesser Antilles arose quite independently from the Greater Antilles via volcanism in the late Eocene. The Lesser Antilles have not served as a northward migration route for Saba! (Saba! is not well-represented in South America), nor have they served as a southward migration route into South America. Their steep topography, with few coastal plains or swamps, may explain why Saba! has not become established in the Lesser Antilles.
Saba! certainly existed in North America prior to the uplift of the Antilles. As  (Bailey 1944, p. 293). It is quoted here in its entirety: Although Saba! minor is the "oldest" species in terms of taxonomy and basis for the genus Saba!, one does not conclude that it is genetically primeval, or that it represents the main or dominant evolution in the group. Probably we should have had a more correct estimate of the genus if S. palmetto had happened to have been the descriptive starting point.
This statement, implying that S. minor is somewhat specialized and that S. palmetto is not, was the only clue Bailey gave to his concept of the evolution of the genus. His treatments (Bailey 1934(Bailey , 1944, as well as those of Cook (1901) and Beccari (1907), are purely taxonomic in scope.
The taxonomic units used in this phylogenetic analysis are species of Saba/ as Table 4. Data matrix for cladistic analysis of Saba/ and outgroups, Brahea (BRAH) and Washingtonia (WASH). however, because Saba! is placed in its own subtribe (Uhl and Dransfield 1987), the outgroups are more closely related to each other than they are to Saba!. Phylogenetic hypotheses were tested using PAUP (version 2.4 by D. Swofford). The apomorphic and plesiomorphic conditions of the characters used in the analysis are presented in Table 3. The evolutionary direction for all characters was inferred by regarding the outgroup condition as ancestral. Polarization was also guided by the major trends in evolution outlined by Moore and Uhl (1982). The data matrix for 1 7 taxonomic groups and 22 synapomorphic characters is given in Table 4. The plesiomorphic condition was scored as "0," and the apomorphic condition was coded as "1." Missing data were scored as "9." The cladogram was constructed manually and checked against a consensus tree. The cladogram is presented in Figure 14. It has 45 steps (character state changes) and several reversals and parallelisms. Figure 14 shows five major clades: the MINO, BERM, MARl, MEXI, and URES clades. The MINO clade consists solely of S. minor. Its lack ofsaponins isolates it from the remainder ofthe genus. Saba! minor has several apomorphies (1, 10, 15, and 18); however none is unique to this species. Its erect inflorescence position is autapomorphic.
The BERM clade consists of S. bermudana, S. palmetto, S. etonia, and S. miamiensis. It possesses synapomorphies for characters 18 and 21, both tannin location characters that are thought to have evolved several times on separate clades. The three continental species of this clade are joined by the presence of tannins peripheral to the expansion region of the midvein. This synapomorphy shows neither reversals nor parallelisms within the genus. Within the clade, S. palmetto is clearly isolated from S. etonia and S. miamiensis. Saba! palmetto is  Table 3. Negative numbers are character state losses. relatively unspecialized and, as Bailey (1944, p. 293) noted, may well be similar morphologically to the ancestor of the genus. Saba/ etania lacks negatively charged ftavonoids (15), but as material of S. miamiensis was unavailable, presence or absence of negatively charged ftavonoids in S. miamiensis cannot be scored with certainty. The remaining clades (URES, MEXl, MARl} share the presence of scattered fiber bundles (or fiber bundles absent) in the expansion region of the midvein (synapomorphy for character 8).
The MEXl clade and the MARl clade form sister groups, defined by synapomorphies for characters 4, 5, and 6. Within the clades, there are reversals in both 5 and 6, but synapomorphy 4 (long-looping transverse commissures) is unaffected by parallelisms or reversals. The MARl clade is differentiated by the presence of tannins within the bundle sheath surrounding the vascular bundles (21). Within the clade, S. damingensis and S. causiarum are sister groups sharing a reversal to the ancestral condition, fruits longer than wide ( 12). The MEXl clade is defined by the loss of bundle sheath extensions (7). Saba/ mexicana and S. guatemalensis share the derived condition petals costate when dry (11). This character appears nowhere else in the cladogram. Saba/ mauritiifarmis and S. yapa possess synapomorphies for leaf segmentation (clustered) and lamina texture (thin or papery}, as well as inflorescence posture (ascending). Only the latter synapomorphy appears elsewhere in the cladogram. Each of these species has several morphological autapomorphies, not shown in Figure 14, that distinguish it from all other species of Saba/.
The final major clade is the URES clade comprising S. uresana, S. rasei, and S. pumas. These three species form an unresolved trichotomy and are united by the presence of tannins in the parenchyma of the mid vein (16). Saba/ pumas is the least specialized of the three and, in fact, is the least specialized in the entire genus. Of the characters considered here, the acquisitions of only three (8, 14, and 16) separate S. pumas from the hypothetical ancestral species.
The phylogenetic hypothesis proposed above is remarkable in its congruence with the biogeographic data, illustrated in Figure 15. The URES clade is clearly ' ·.f. isolated from the other Mexican species. The Antillean species share a common ancestor with the southern Mexican species and echo a well-documented floristic relationship between tropical Mexico and the Greater Antilles (Howard 1973).
Speciation in the genus has likely occurred by allopatric means. Allopatric speciation best explains the high correlation between the geographic distribution and the phylogenetic hypothesis. For example, orogeny of the Sierra Madre Occidental (and the Rocky Mountains to the north) may have isolated the common ancestor of the URES clade in western Mexico; subsequent ecological specialization could have led to further speciation in allopatry. In eastern Mexico, increasing aridity in the Tehuacim Valley or Isthmus ofTehuantepec may have fragmented the once continuous population of the ancestor of both S. mexicana and S. guatemalensis. Ecological allopatry may have been responsible for differentiation among S. etonia, S. miamiensis, and S. palmetto. Likewise, S. yapa and S. mauritiiformis, although geographically sympatric over portions of their ranges, are ecologically isolated-S. yapa inhabits slightly more arid habitats.
Since Saba! seeds are readily dispersed by birds and perhaps ocean currents, there is no need to believe that the Antillean species arose through vicariance events as have been proposed for other less vagile organisms (Buskirk 1985;McFadden 1980;Rosen 1975). Saba! palmetto (of Florida, Cuba, and the Bahamas) and S. bermudana (Bermuda) share a common ancestor but inhabit areas that were never contiguous, a fact which suggests that dispersal, rather than vicariance events, has played a greater role in the evolution of this clade and in the Antillean species. Historical events of dispersal, as an explanation for the origin of the Antillean species, are in agreement with the geological evidence and the known modes of dispersal in the genus.

TAXONOMIC HISTORY
The first mention in the literature of a palm referable to Saba! is that of P. Browne (1756), whose polynomial description of the Jamaican "Corypha (?)pal-macea assurgens, foliis flabelliformibus semipinnatis, petiolis majoribus compressis" can be reliably ascribed to Saba/ on the basis of the costa palmate ("flabelliformibus semipinnatis") leaves.
Species now assigned to Saba/ entered the early botanical literature as Corypha L., Chamaerops L., and Rhapis L. f., genera now known to be native strictly to the Old World. Not surprisingly, eastern North American species first attracted the attention of post-Linnaean botanists, so Corypha minor Jacq., described in 1776, is the earliest binomial for a species now included in Saba/.
The genus Saba/ was first proposed by Adanson (1763) in his "Families des Plantes." The derivation of the name was not stated. Adanson clearly rejected the Linnaean system ofbinomial nomenclature when he described a genus/species of palm from the Carolinas (U.S.A.) with the uninomial Saba/. Parkinson (1987) has argued convincingly, at least in the case of Saba/ and other genera first described as monotypic, that Adanson's use of unitary genus/species designations is in clear violation oflnternational Code of Botanical Nomenclature Art. 20.4(b) which states that unitary designations of species are not to be regarded as generic names.
The genus name was validated by Guersent, who in 1804 published a description of Saba/ adansonii and gave Corypha minor Jacq. as a synonym. As there were no alternate generic names proposed for Saba/ between Adanson's description and its validation by Guersent, the genus name can correctly be attributed to Guersent.
Saba/ appeared in various North American and Antillean floras (e.g., Chapman 1883; Grisebach 1864; Small 1903), and additional species were described or transferred to Saba/ (Grisebach 1864; Hemsley 1885; Martius 1853; Nash 1896), but no comprehensive monograph existed. In 1901, 0. F. Cook erected the genus !nodes to accommodate arborescent species of Saba/ with strongly costapalmate leaves. The distinction between strongly costapalmate and weakly costapalmate was by no means clear, but this detail did not prevent Cook from describing five new species of !nodes and transferring to !nodes three more. In addition, between the inception of Saba/ and its division by Cook, over 30 nomina nuda entered the botanical literature.
In 1907, the first monograph of the genus Saba/ was published by 0. Beccari. Beccari did not accept Cook's !nodes, transferred all of its species of Saba/, and described eight new taxa, bringing the total number of taxa of Saba/ recognized to 18 species and one variety. Subsequent to Beccari's monograph, several species were described as new (e.g., Beccari 1908Beccari , 1931  In H. E.  checklist of palms and in a subsequent (Moore 1971b) addendum, he reduced a number of species to synonymy and recognized a total of 15 species. The current treatment includes three additional species long recognized by earlier botanists, synonymizes a few names, and replaces one name with an earlier, validly published name. A total of 15 species is recognized, and more than 45 nomina nuda are treated as such.
Cook's !nodes, despite its inherent artificiality, has not died quietly. Small (1933) reduced !nodes to a subgenus within Saba!. This treatment was followed by subsequent students of the genus, including myself (Zona 1985 Solitary, pleonanthic, hermaphroditic palms with aerial or subterranean woody unarmed trunks. Stem covered with leatbases or clean, obscurely to strongly ringed, becoming more or less smooth or striate and bare with age. Leaves few to numerous, alternate and spirally arranged, blade weakly to strongly costapalmate, glaucous or paler on the abaxial surface or not; petiole unarmed, convex abaxially, more or less concave adaxially, splitting at the base; hastula usually well developed on adaxial surface, obtuse to acuminate triangular, with peltate trichomes (these often caducous); hastula margin entire or undulate, erect, involute, or revolute; plication induplicate; leaf segments lanceolate, basally connate to connate for half their length or groups of two or three segments connate for almost their entire length, glabrous, glabrescent, or !epidote on abaxial surface of mid veins, usually filiferous between leaf segments, apices acute or bifid and bearing a filament in each cleft; midveins prominent, transverse commissures obscure to conspicuous; stomata anomocytic, present on both surfaces or only the abaxial surface, plugged with cutin.
Saba/ bermudana flowers during June and July. The sweet fleshy fruit are produced in the fall and are consumed by birds including the introduced Kiskadee, Pitangus su/phuratus.
Discussion. -Watts and Hansen (1986) reported that Saba/ pollen, presumed to be from S. bermudana, is a common element in sediment cores dating between ca. 10,000 and 9000 yr. B.P., a fact suggesting that Saba/ has been part of the Bermuda flora for at least that long.
The taxonomic history of S. bermudana is also long. A provisional name, "Saba/ blackburnia" was used by Glazebrook (1829), for a palm of unknown origin cultivated in England. Glazebrook (1829) illustrated globose fruits 19.2 mm in diameter and seeds bearing a beaklike funicular remnant. This latter characteristic is somewhat suggestive of the Saba/ of Bermuda, but this epithet as used by Glazebrook must be rejected as a provisional name under Article 43.3 of the ICBN.
In 1830, the nameS. blackburniana was validated with a brief description and reference to Glazebrook (1829) when it was included in Schultes and Schultes' Systema Vegetabilium. Although the provenance of the species was still unknown, the name was used by subsequent authors (e.g., Hemsley 1885) to refer to the Saba/ ofBermuda. The protologue, however, includes a number of characteristics inconsistent with the Bermuda species: trunk with leaf scars, inflorescence shorter than the leaves, and fruit globose and 22.2 mm in diameter. The first two characteristics suggest S. bermudana, but the last two clearly do not. The fruits of S. bermudana are 12.9-17.9 mm in diameter and are strongly pyriform. Given the importance of fruit size and shape in recognizing species of Saba/, we cannot easily overlook this part of the description of S. b/ackburniana. Many specimens labeled S. blackburniana from European gardens representS. bermudana. Nevertheless, in the absence of type specimens, the name S. blackburniana remains a nomen ambiguum and must be rejected.
The lectotype of S. bermudana was chosen from among the specimens seen by Bailey; it is one of the most complete specimens.
Saba/ bermudana is unique in that age and growth rate of some individuals can be calculated with some degree of accuracy. The practice of tapping the stem of the palm just below the terminal bud for its sap (which was fermented to produce an alcoholic beverage, "bibby") has left visible scars. The practice was outlawed by the Governor of Bermuda in 1627 (Hodge 1960), yet trees with tapping scars are still common on the island. Trees so scarred are probably over 300 years old and yet appear to have grown only 3 m or so during that time, giving a growth rate of ca. 1 em per year. Also visible in some palm stems are the bore holes of a species of woodpecker that is no longer found on the island (J. Madieros pers. com.). Massive palm to ca. 10m tall; trunk 35-60 em DBH, smooth and gray. Leaves 20-30, evenly green or glaucous, strongly costapalmate, filiferous; petiole 2.1-4.7 em wide, 1-2 m long; hastula acute, 5.5-21 em long, glabrous or glabrescent, margin revolute, flat, erect, or involute, entire or undulate, sometimes ridged abaxially; segments 60-120 per leaf, connate for ca. 40% of their length, middle segment 75-175 em long, 2.6-5.8 em wide, 0.2-0.4 mm thick, transverse commissures long looping and conspicuous, apex bifurcate for 20-43 em. Inflorescence arcuate with 3 orders ofbranching, nearly exceeding the leaves in length, sheathing bracts usually glabrous or glabrescent, rachillae 11-20 per branchlet, 0.5-1.1 mm in diameter, 4.5-11 em long, with (7-)8-9(-10) flowers per em. Flower 3.7-5.2 mm long; calyx cupulate, strongly costate when dry, 1.3-2.0 mm long, 1.2-2.1 mm wide, sinuses 0.3-0.8 mm deep; petals obovate-long obovate, noncostate when dry, membranous, 3.0-4.0 mm long, 1.1-2.0 mm wide; stamens spreading, filaments 2.8-4.5 mm long, adnate to the corolla for 0.6-1.5 mm, anthers ca. 1.4 mm long and 0.7 mm wide; gynoecium 2.7-3.8 mm long, ovary 0.7-1.2 mm high, 0.8-1.1 mm in diameter. Fruit spherical or occasionally oblate-pyriform, black, 7.1-10.8 mm in diameter, 7.5-10.4 mm high; seed oblate concave, 5.9-7.8 mm in diameter, 4.3-5. 7 mm high; embryo supraequatorial, rarely equatorial or subequatorial. (Fig. 3C, 5B  Distribution and ecology (Fig. 17) as Haiti and the Dominican Republic (Hispaniola). Its presence on the island of Hispaniola was first recognized by Moscoso (1943). It is reported by Questel ( 1941) to be naturalized on St. Barthelemy. It has been introduced on Guadeloupe, where it persists after cultivation.
Saba! causiarum flowers in the months of April through August. The species in not endangered.
Discussion. -Dammer and Urban ( 1903) recognized another entity from Puerto Rico but stopped short of giving it a name. The description was based on a specimen in the Berlin herbarium (Sintenis 3765) and was questionably assigned to S. causiarum by Beccari (1907), who remarked that without more material he was unable to decide with certainty if this was indeed another species. A duplicate specimen (at GH) consists of a portion of an old infructescence with only two orders of branching. Dammer and Urban gave the seed size as 8 mm x 6 mm (only slightly large for S. causiarum). Given the morphological plasticity of Saba! and the lack of other similar specimens, we must conclude that Sintenis 37 65 represents a depauperate or otherwise aberrant individual of S. causiarum. Beccari, Webbia 2:49. 1907 Massive palm to ca. 10m tall; trunk ca. 60 em DBH, smooth and gray. Leaves 20-30(?), evenly green, strongly costapalmate, filiferous; petiole ca. 3. 7 em wide and 1 m long; hastula acute, ca. 15.5 em long, glabrous, margin erect, entire; segments ca. 90 per leaf, connate for ca. 30% of their length, middle segment ca. 106 em long, ca. 2.5 em wide and 0.2 mm thick, transverse commissures long and conspicuous, apex bifurcate for ca. 39 em. Inflorescence arcuate with 3 orders of branching, equalling or slightly exceeding the leaves in length, sheathing bracts glabrous(?), rachillae 11-18 per branchlet, 1.0-1.2 mm in diameter, 7.5-12 em long, with 10-11 flowers per em. Flower (based on Eggers 1678) 4.5 mm long; calyx cupulate, strongly costate when dry, 1.7 mm long, 1.3 mm wide, sinuses 0.6 mm deep; petals obovate, noncostate when dry, membranous, 3.5 mm long, 1.3 mm wide; stamens spreading, filaments 3.8 mm long, adnate to the corolla for 1.1 mm, anthers 1.2 mm long and 0.8 mm wide; gynoecium 3.0 mm long, ovary 1.1 mm high, 1.1 mm in diameter. Fruit pyriform, black, 11.5-14.1 mm in diameter, 11.0-14.4 mm high; seed oblate concave, 8.0-10.4 mm in diameter, 5.1-7.1 mm high; embryo supraequatorial, rarely equatorial or subequatorial. (Fig. 5C, 8C Distribution and ecology (Fig. 17).-Sabal domingensis is found on the interior of the island of Hispaniola at ca. 150-1000 m in elevation. It is a common component of secondary successional vegetation. Bisse (19 81) reported that a Hispaniolan palm (S. domingensis or S. causiarum) is present on the southeastern coast of Cuba; however, I have seen no specimens of either S. domingensis or S. causiarum from Cuba.

SABAL DOMINGENSIS
The species apparently flowers during the summer, from March through August.
Discussion. -Saba! domingensis is poorly represented in herbaria and poorly known, probably because of the confusion between it and S. causiarum. The above floral description is based solely on one set of measurements from a fragment of the holotype at FI and therefore does not represent the entire range of variation. When Beccari described Saba! domingensis he indicated that examples of the type specimen may be found at Berlin, Paris, Munich, etc., but explicitly designated no holotype. Although Glassman (1972a) designated the Berlin specimen as a lectotype, the holotype is the specimen at P, fragments of which are in Beccari's herbarium at FI and are annotated in Beccari's hand.
Saba! etonia flowers from late May through July. This species is not presently endangered; however, most of its habitat on the Atlantic Coastal Ridge in southeastern Florida has been destroyed by urban growth. In central Florida, its habitat has been largely cleared for agriculture. Large populations of S. etonia remain protected in the Ocala National Forest in the north and Archbold Biological Station to the south.
Discussion. -Saba! etonia is one of the few species in the genus that is not "weedy" but rather is characteristic of an undisturbed vegetation community, the sand pine scrub, a community rich in endemic plants and animals.
Collections of flowering material are known from December through May.

Distribution and ecology
This species blossoms from March through September.
Discussion.-Corypha maritima Kunth is one of the oldest basionyms for a species now included in the genus Saba!. The type is sterile, and hence, the epithet has been treated as a possible synonym of Saba! yapa by Beccari (1912), a doubtful species by Bailey (1944), and, in violation of the ICBN rule of priority, as a synonym of S. parvif/ora (=S. palmetto) by Muniz and Borhidi (1982). Thanks toP. Morat of the Laboratoire de Phanerogamie, Paris, the type was located and a fragment provided for anatomical study. Once its identity was established by anatomical study, the epithet was again available for use. On Cuba, S. maritima has continually been confused with S. palmetto. The two species are immediately distinguished by the number and spacing of major veins about the mid vein in the leaf (Fig. 21). This character is best seen in fresh or rehydrated material viewed with transmitted light. The veins with bundle sheath extensions appear translucent; veins without bundle sheath extensions are not visible. Saba! palmetto has uniform spacing between the veins, while in S. maritima, the spacing is greatest around the midvein (Fig. 21). Near the margin or suture vein, the patterns of venation for the two species appear similar. 1.4-2.4 mm long, 1. 1-2.0 mm wide, sinuses 0.6-0.9 mm deep; petals ovate (rarely obovate), noncostate when dry, membranous, 2.4-3.9 mm long, 1.5-1.8 mm wide; antipetalous stamens spreading-reflexed, antisepalous stamens erect, filaments 2.7-3.5 mm long, basally connate and adnate to the corolla for 0.6-3.5 mm, anthers ca. 1.2 mm long and 0.6 mm wide; gynoecium 2.5-3.1 mm long, ovary 0.8-1.3 mm high, 0.8-1.1 mm in diameter. Fruit spherical to pyriform, blackish, 8.8-11 mm in diameter, 8.5-11 mm high; seed oblate spheroidal, 6.6-7.9 mm in diameter, 4.9-6.2 mm high, with rounded or bulging funicular remains; embryo supraequatorial or rarely equatorial. (Fig. 3E, 4C, 6B, 8G, llC, 22.) Common names. -Botan, carat, carata, palma amarga, palma de guagara, palma de vaca.
Distribution and ecology (Fig. 23).-The distribution ofthis species is noteworthy for its patchiness. It occurs abundantly in isolated populations in southern Mexico, Belize and Guatemala, extreme southeastern Costa Rica, eastern Panama and the Perlas Is. in the Bay of Panama, north coastal South America (Colombia and Venezuela) and Trinidad. It grows at elevations up to 1000 m but is generally found at 0-400 m, often on soils derived from limestone.
Phenological records for Saba! mauritiiformis show that it flowers from March to October, but flowering is probably sporadic throughout the year.
Distribution and ecology (Fig. 19).-Saba! mexicana is distributed in extreme southern Texas, much of gulf coastal Mexico, Oaxaca, and the Yucatan Peninsula. A single collection is known from El Salvador.
Its presence in Yucatan may be attributed to the activities of pre-Columbian peoples (J. Caballero pers. com.). In Yucatan, it is grown in plantations and is the mainstay of the thatch industry (Fig. 24).
This species is most often collected with flowers during the early part of the year, January through May, but a few specimens with flowers are known from other months as well.
Herbarium records are scant, but collections with flowers are known from throughout the year.
The species was proposed for listing as an Endangered Species by the U.S. federal government; however, the proposal was withdrawn owing to disagreement concerning the validity of the taxon. Federal protection, however, would be in name only, as the species is likely already extinct. Its habitat in Dade County has been urbanized and utterly destroyed.
Discussion.-The taxonomic history of this species has been given elsewhere (Zona 1983(Zona , 1985. The presence of both dwarfed S. palmetto and S. etonia in south Florida undoubtedly has led to some confusion which in turn has contributed to the debate concerning the validity of this taxon. Undoubtedly, S. miamiensis is more closely related to S. etonia than was previously believed (Zona 1985). Anatomically, S. miamiensis shares many features with S. etonia; although, S. etonia has more adaptations to arid environments. The morphological characteristics given previously (Zona 1985) are still useful in distinguishing the species, i.e., lax arching inflorescence with three orders of branching and large fruits and seeds. The fruits of S. miamiensis are 15.7-19.0 (16.9 ± 1.1) mm in diameter, versus 9.0-15.4 (12.9 ± 1.9) mm inS. etonia. Habitat differences are critical.  Saba/ etonia has often been confused with S. miamiensis, but the former grows on white sand, not oolite. The above description of flowers is based on only two specimens and probably does not fully account for all the variation in this species.  Distribution and ecology (Fig. 26).-Saba! minor has a wide distribution in the southeastern United States and is the most northernly ranging species in the genus. Throughout its range, it is a palm of the rich soils of floodplains, levees, river banks, and swamps where it is associated with broadleaf deciduous trees of genera suchasAcer, Betula, Carpinus, Carya, Celtis, Crataegus, !lex, Liquidambar, Quercus, and Ulmus. Nixon, Chambless, and Malloy (1973) present a detailed ecological study of S. minor in Texas. Like other north temperate species of Saba!, S. minor shows strong seasonality in flowering. It blossoms in the warm months of April through August, with peak activity in June.
Discussion. -Saba! minor has a most colorful taxonomic history. It is a conspicuous element of the vegetation of the southeastern United States and is relatively easy to collect and press, so S. minor was included, under various names, in the floras of nearly every early American and European botanist. Glassman (1972a) designated plate 8 ofJacquin's publication as the lectotype; however, a specimen (2 sheets) deposited at BM and bearing labels in Jacquin's own hand (D'Arcy 1970) appear to satisfY the definition ofholotype.
The species is highly variable, and most troublesome to early botanists was the presence or absence of an above ground stem. Palms of this species with conspicuous aerial stems have been described as S. louisiana and S. deeringiana. At the western edge of its range, S. minor is often arborescent and large, but over the entirety of its range, the species varies along a continuum in both si]:e and arborescence. Furthermore, plants of S. minor growing in rich soils can attain unusually large dimensions, but this variation in size appears to be environmen- tally induced. Since the arborescent individuals have not been shown to be genetically isolated from the suffrutescent individuals, I treat both as variants of a single variable species.
In the northern portion of its range, S. palmetto blossoms mostly in July with little or no flowering during the remainder of the year. In central Florida, it flowers from June through August, but in southern Florida and the Bahamas, it flowers throughout the year. In Cuba, it seems to flower most abundantly in the spring.
Discussion.-When S. parviflora is compared with S. palmetto, it becomes apparent that there is little reason, other than tradition, to keep them apart. Beccari (1912) described the leaf segment apices of S. parviflora as very acuminate and rigid and again in 19 31 described the segments as "acuminate with stiff apices." The segment apices of the Cuban palms sometimes appear rigid in the field, but this difference is hardly reason to recognize a separate species. In fact, the leaf segments of the isotype of S. parviflora at NY are long, flexible, and acuminate. Beccari (1912Beccari ( , 1931 used the presence ofterete (when dry) rachillae in S.jlorida (=S. maritima) to distinguish it from S. parviflora, which was said to have angular rachillae. Bailey ( 1944) included S. florida ( =S. maritima) in his circumscription of S. parviflora and consequently distinguished S. parviflora by the presence of thin, terete rachillae and narrow threadlike leaf segment apices versus irregular or angled rachillae and less attenuated apices in S. palmetto. The rachilla characteristic simply does not hold up in a large number of collections, not even for S. maritima. Another difference used by Bailey (1944), that the inflorescences of S. parviflora "seldom if ever" exceed the leaves, likewise is of limited usefulness and questionable validity.
The Cuban population has somewhat larger fruit and seed dimensions, but they are broadly overlapping with those of mainland S. palmetto. Other similarities between them are readily apparent in leaf anatomy, flavonoid chemistry, and ecology. A case might be made for recognizing the Cuban population at the infraspecific level; however, a more conservative approach is taken here.
Typification is required for S. palmetto and many of its synonyms. In 1927, Small described S. jamesiana to include an adult palm bearing juvenile foliage. He designated no types, so I have chosen as a lectotype a specimen collected by him from the type locality two years after describing the species. Bailey named two syntypes when he described S. viatoris in 1944. I have chosen the more complete of the two specimens as the lectotype. Likewise, I have chosen from among Becarri's three syntypes to typify S. palmetto var. bahamensis; the fertile specimen bearing Beccari's annotation is the lectotype. Walter's specimen of Corypha palmetto is probably no longer extant (Fernald and Schubert 1948). The neotype that I have chosen closely agrees with the protologue and is reasonably complete, and isoneotypes are widely distributed and available for study.
As circumscribed above, S. palmetto is a wide-ranging, weedy, and highly variable species.
Saba! rosei flowers from December to July.
Discussion.-The description of the flowers given above is based on only three specimens and may not fully describe the range of variation of floral morphology for this species. The assignment of Jones' Erythea loretensis to this taxon is somewhat speculative, given the fact that Jones collected no specimen and designated no type (Blake 19 57). The photograph published by Jones is unquestionably of a Saba!, but the broad range offruit size given by Jones for his species could accommodate either S. uresana or S. rosei. In Jones' key to Erythea, the leaves of E. loretensis are described as green (as in S. rosei), not glaucous as in S. uresana. For these reasons, E. loretensis is assigned to S. rosei, and Bailey's topotype specimen, collected only three years after Jones' publication, is chosen here as a neotype.
Sabat rosei clearly shares a common ancestry with the two other western Mexican species, S. pumas and S. uresana. Of the three, S. rosei is the most widespread.
Distribution and ecology (Fig. 29).-Sabal uresana occurs in thorn forest and oak forest along watercourses and valleys in the foothills of the Sierra Madre Occidental in Sonora and Chihuahua, Mexico. It can be found from sea level to 1500 m, with most populations found above 650 m (Gentry 1942). Associated species include Acacia cochliacantha Humb. & Bonpl. ex Willd., A. pennatula (Schl. & Cham.) Benth., Conzattia sericea Standi., Guazuma ulmifolia Lam., Jacquinia pungens A. Gray, Lycium exsertum A. Gray, Prosopis juliflora (Sw.) DC., and Quercus chihuahuensis Trel. A complete account of the vegetation of the region can be found in Gentry (1942).
Saba! uresana apparently flowers in mid-summer, but phenological records are scanty.
Discussion. -Herbarium records of S. uresana are poor. Precious little flowering material is available, and many collections in North American herbaria represent sterile seedlings. Consequently, the above description does not reflect the full range of variation found within this species.
Populations of S. uresana are not abundant and are never large. Gentry (1942) stated that the species was declining in abundance and assigned cause variously to drought, over-exploitation by the indigenous people, and bruchid beetle predation upon seeds. Historical records are not sufficient to document its decline, although its present rarity stands in stark contrast to the abundant stands of S. rosei to the south. If populations of S. uresana are dwindling, over-exploitation for timber, thatch, and fiber is probably the reason. Distribution and ecology (Fig. 31).-Sabal yapa is widespread in the Yucatan Peninsula from sea level to 100 m on well-drained, limestone soils, often on the steep banks of swamps or sinkholes (Lundell 1937). It grows in upland forests associated with Manilkara zapota (L.) Van Royen, Brosimum alicastrum Swartz, Callophyllum brasiliense Camb. var. rekoi Standi., Lucuma campechiana Kunth, and Swietenia macrophylla King (Bartlett 1935;Lundell 1937). In Cuba, it is found on both swampy and dry soils, also on limestone. It persists after forests are cleared and burned for agricultural use. In Yucatan, Saba! yapa grows sympatrically with S. mexicana, a species probably introduced to the peninusula by pre-Columbian peoples. In Cuba, S. yapa can be found growing with S. palmetto and S. maritima in Bataban6.
Saba! yapa flowers in the first half of the year (January-July) with sporadic flowering at other times.
Discussion.-This species is the most specialized in the genus. Several floral characteristics are unique to S. yapa, including campanulate calyx, ovate petals without hyaline margins, and basally connate petals. It also shares a number of derived features with S. mauritiiformis, such as overall growth habit, leaf venation and texture, and clustered leaf segments. Anatomical features suggest, however, that S. yapa is more drought-adapted than S. mauritiiformis, and thus, the two are probably ecologically separated.
EXCLUDED NAMES AND NOMINA NUDA The following list of names does not include herbarium names that were never published, nor does it include those names appearing in  that were