| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
Institute of Systematic Botany, New York Botanical Garden, Bronx, New York 10458 USA
Received for publication October 8, 2003. Accepted for publication January 30, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
Key Words: Central America • Hyospathe • multivariate statistics • Palmae • South America • species delimitation • species testing
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
). As a consequence, there has never been any question about generic boundaries. There has, however, been considerable disagreement over the number of accepted species. Burret (1929)
recognized 10 species and eight more were added subsequently (Burret, 1933
, 1936
, 1938
, 1940
; Bailey, 1943
; Moore, 1949
; Steyermark, 1951
). Skov and Balslev (1989)
, criticizing Burret's species concept as narrow and typological, reduced the number of recognized species from 18 to two. Skov and Balslev's (1989)
broader concept has itself been criticized, and Hammel et al. (2003)
considered that they, Skov and Balslev, employed a very wide concept. Hyospathe thus provides a familiar example of a taxonomic problem—how many species are there, 18, two, or some other number?
Burret (1929)
used quantitative characters to separate species, such as number of pinnae and lengths of inflorescences, rachises, rachillae, flowers, and flower parts. With the few specimens available at the time, such an approach probably worked quite well. However, with an increasing number of specimens, the differences between them soon become blurred. Skov and Balslev (1989)
clearly showed that many of the characters used by Burret could not be used to distinguish the much larger number of specimens recently available, because variation is continuous. However, Skov and Balslev continued to rely mostly on quantitative characters, from stems and inflorescences. Their use of qualitative characters, particularly from flowers, was precluded by their broad concept of one of the species they recognized, Hyospathe elegans. Within this species, floral characters vary considerably, as shown in this study, and these are taxonomically useful.
In this paper, I take a two-stage approach to the problem of Hyospathe taxonomy. First, I use multivariate statistical methods to delimit groups of specimens, using as many characters as possible given the limitations of the specimens. I also employ geographic distributions to help to delimit groups. Second, I apply a specific species concept to the groups discovered.
Luckow (1995)
reviewed species concepts currently used by systematists. I find, from this review, that the Phylogenetic Species Concept (PSC) is the only species concept that can logically be applied to groups of specimens delimited by the data and methods used here. This species concept was defined by Nixon and Wheeler (1990, p. 218)
as: "the smallest aggregation of populations.... diagnosable by a unique combination of character states in comparable individuals." Individual specimens of Hyospathe are considered comparable because all are fertile.
Two operational modifications are necessary in order to apply the PSC here. According to Davis and Nixon (1992)
, phylogenetic species are delimited by successive rounds of aggregation of local populations, based on analysis of traits (i.e., polymorphic attributes) and characters (i.e., fixed attributes). As an herbarium taxonomist dealing with specimens of a widespread, unevenly collected, tropical genus, I have no a priori method of placing specimens in populations, and consequently I cannot distinguish a priori between traits and characters. I therefore score specimens (i.e., treat specimens as populations) for all attributes (i.e., traits and characters), use cluster analysis to distinguish traits from characters, and discover groups of specimens with unique combination of character states (see Materials and Methods).
A second modification of the PSC involves variation in quantitative characters. Some groups of specimens with unique combinations of qualitative character states nevertheless vary greatly in quantitative characters and occur in disjunct geographic areas. Such subgroups may differ significantly from one another in one or more characters. Luckow (1995, p. 595)
, in her discussion of the PSC, stated that "a quantitative character in which there are ‘gaps’ (i.e., no overlap in values) could be used to distinguish phylogenetic species, but one that exhibited differences only in mean values could not." However, such nonoverlapping gaps are seldom found, at least not in Hyospathe. Luckow (p. 597) continued "groups of populations that differ not by fixed characters, but by differences in mean values would be recognized as subspecies or varieties [under the PSC]." I follow a slightly modified version of this here. I apply the PSC to groups of specimens with unique combinations of qualitative character states. Within such groups, where subgroups can be delimited geographically and can be distinguished by one or more significantly different mean value of a character, then I apply a phylogenetic subspecies concept.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
) were examined. One hundred and ten specimens were excluded because of their poor quality, leaving a total of 428 specimens sampled for this study. A search was made for qualitative attributes in which two or more states of the attribute were present among the specimens and could be scored unequivocally. This search was based on those attributes used in previous monographs and on a survey of specimens. A dissecting microscope was used to survey floral attributes. A search was also made for quantitative characters that could be measured from specimens or taken from specimen labels (in case of ranges, median values were used). Characters were counted or measured with a ruler, digital calipers, or protractor.
A data matrix was constructed with specimens as rows and attributes/characters as columns. Additional columns recorded a specimen identification number, collector, collector's number, herbarium, country, latitude, longitude, and elevation. Latitude and longitude were either taken from the specimen label or from the collection locality using either maps or an electronic gazetteer (website at http://www.calle.com/world/). The data matrix is available at the website http://sciweb.nybg.org/science2/res/henderson/hyospathe.zip.
Multivariate methods of analysis—cluster analysis (CA), principal component analysis (PCA), and discriminant analysis (DA)—were carried out using the programs NTSYS (Rohlf, 2000
) and Systat (Wilkinson, 1997
). Specimens with missing values were excluded. Analyses are thus based on subsets of the data, as noted later. Because some quantitative characters were not normally distributed, they were log10 transformed before the analysis. Discrete variables were square-root transformed. For PCA and DA, variables were assumed to be multivariate normal, although this assumption is not readily tested (Tabachnik and Fidell, 2001
). Although ordination is considered relatively insensitive to violations of normality assumptions (Tabachnik and Fidell, 2001
), results should be viewed as approximate.
Specimen group delimitation
Cluster analysis was used to distinguish between traits and characters. The SIMQUAL module of NTSYS with the simple matching coefficient was used to produce a similarity matrix. This was subjected to the unweighted pair group method, arithmetic average (UPGMA) clustering algorithm. Successive rounds of CA were used. All attributes were used in the first analysis, and by a process of backward elimination of probable polymorphisms, a suite of attributes (i.e., characters) was discovered which, when analyzed, gave groups of specimens with unique combinations of states. These groups were recognized as character groups.
Quantitative characters were then examined. Because these had more missing values than qualitative characters, they were divided into universal quantitative characters, with <25% missing values and local ones having >25% missing values. Character groups were analyzed by PCA of a correlation matrix of transformed universal quantitative characters. Discriminant analysis of pre-classified groups of specimens was used with the same data to test the hypothesis that group centroids were equal. Wilk's lambda, most discriminatory characters, and percentage classification success are reported. Distribution of traits within each character group was also examined.
Geographic distribution of character groups was then analyzed. Distributions were mapped with Arcview GIS 3.2 (Environmental Systems Research Institute) using latitude and longitude data for each specimen. If geographically separate groups (defined as groups separated by a horizontal distance >100 km) were found, a t test (two-sample, separate variance test on log10-transformed characters) or one-way ANOVA (on log10-transformed variables) was used to test for group differences for each universal quantitative character. The Bonferroni pairwise procedure was used to see which pairs of means differed significantly. If there was at least one significant (P < 0.01) difference in any universal character for each possible pair of subgroups, then these were recognized as geographic subgroups. These were further tested using DA.
Application of species concepts
The PSC was applied to groups of specimens with unique combinations of qualitative character states (i.e., character groups), and a PSC subspecies concept applied to subgroups that could be delimited geographically and by one or more significantly different mean values of a quantitative character (i.e., geographic subgroups; see Introduction).
Taxonomic treatment
A detailed genus description can be found in Uhl and Dransfield (1987)
and Skov and Balslev (1989)
and is not repeated here. The numbering system for species corresponds to that used for character groups. Images of types of new taxa are available at the website http:// www.nybg.org/bsci/herbarium_imaging/. In the descriptions, only those quantitative characters measured or counted in this study are reported. Qualitative character states are given in Results and also in the key to the species and are not repeated in the taxa descriptions. For each quantitative character, one measure of central tendency, the mean, and two measures of variability, the range and coefficient of variation, are given, as well as sample size (e.g., stem length: 2.0 [1.4–2.5] m, CV 0.2, N = 7). The coefficient of variation is given instead of the standard deviation so that the relative amount of variation in each taxon and each character can be compared. Distribution data are repeated from the results section. A complete list of specimens examined is available at the website http://sciweb.nybg.org/science2/res/henderson/hyospathe.zip. Several dubious names, the types of which have been destroyed, are listed by Skov and Balslev (1989)
and are not repeated here.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
Group delimitation
Cluster analysis was performed on a matrix containing 415 specimens (13 omitted because of missing data) and the five qualitative characters. Results (not shown; cophenetic correlation = 0.996) show that six major clusters are separated at the 0.8 similarity level. This is the level at which each of the clusters (i.e., character groups) is characterized by a unique combination of character states.
A PCA was carried out using 12 universal quantitative characters. Sample size is greatly reduced to 176 specimens because many more data are missing for quantitative characters. A scatterplot of the first two components (Fig. 1) shows overlap of four of the character groups, although specimens of groups 1 and 4 cluster separately. Discriminant analysis, using the same specimens pre-classified into the six character groups and the same data set, found significant differences among group centroids (Wilks' lambda = 0.000; P < 0.00). Discriminant analysis correctly classified specimens with 91% success in the classification matrix but only 84% in the jackknifed classification matrix. Peduncle length is the most discriminatory character.
|
|
|
|
Character group 3
Specimens have infrafoliar inflorescences, filiform rachillae, irregular, raised triads, pedicellate pistillate flowers, and non-tubular pistillate sepals (Fig. 2C). For traits, specimens have long, flexuous hairs on the rachillae (Fig. 2C), spirally arranged triads, and red flowers (pistillate scar not applicable). The group contains five specimens from the eastern slopes of the Cordillera Central in Colombia between 6°01'–6°55' N and 75°01'–75°04' W (Fig. 4) at a mean elevation of 1599 (1495–1750) m. Specimens occur in two localities, approximately 100 km apart, but there are too few specimens to test for differences between the two.
Character group 4
Specimens have infrafoliar inflorescences, non-filiform rachillae, regular, non-raised triads, sessile pistillate flowers, and tubular pistillate sepals (Fig. 2D). For traits, specimens have glabrous rachillae (Fig. 2D) or rarely rachillae with hairs, spirally arranged triads, and white or red flowers (pistillate scar not applicable). For universal quantitative characters, this group has only one leaf division (all specimens have simple leaves), short, narrow rachises, and few rachillae (Fig. 3). The group contains 11 specimens from the western slopes of the Cordillera Occidental in Colombia between 6°29'–6°45' N and 76°14'–76°25' W (Fig. 4) at a mean elevation of 1370 (1000–1725) m. There is no geographic discontinuity.
Character group 5
Specimens have infrafoliar inflorescences, non-filiform rachillae, irregular, raised triads, sessile or very shortly pedicellate pistillate flowers, and non-tubular pistillate sepals (Fig. 2E). For traits, specimens have rachillae that are glabrous or with crustose hairs (Fig. 2E) or with long flexuous hairs, spirally arranged triads, and red flowers (pistillate scar not applicable). For universal quantitative characters, this group contains the largest specimens of any group, especially in stem diameter, rachis width, number of leaf divisions, sterile basal width, and number of rachillae (Fig. 3). The group contains 36 specimens from montane areas in northern Venezuela and Colombia and just reaching Panama between 1°08'– 10°47' N and 66°35'–77°45' W (Fig. 4) at a mean elevation of 1459 (1100–1980) m. There is geographic discontinuity, with several localities >100 km apart, but specimens are too few and too scattered to test for differences among localities.
Character group 6
Specimens have infrafoliar inflorescences, non-filiform rachillae, regular, non-raised triads, sessile pistillate flowers, and non-tubular pistillate sepals (Fig. 2F). For traits, specimens have glabrous rachillae or with crustose hairs (Fig. 2F) or with dense crustose hairs or with long, flexuous hairs or with flattened, lacerate hairs, distichously or spirally arranged triads, low or raised pistillate scars (Fig. 2F), and white or red flowers. For almost all universal quantitative characters, this group is the most variable (Fig. 3). The group contains 360 specimens from Central and northern South America between 8°45' N–17°00' S and 46°15'–83°51' W (Fig. 4) at a mean elevation of 422 (40–1900) m.
The group has considerable geographic discontinuity. Specimens are divided by the Andes into two groups, separated at their nearest point in Ecuador by a distance of approximately 200 km. These two groups are here referred to as Western Andean (Pacific coast of Ecuador and Colombia, and Central America) and Amazon (eastern Andean foothills and Amazon).
The Western Andean specimens are not continuously distributed, but can again be divided into five geographic groups (Fig. 5). Each is separated from the other by a distance of at least 100 km. Although the ranges of these groups are discrete, it is not possible to know if the gaps among them are collecting artifacts, although many of the intervening areas are well collected botanically. Analysis of variance shows that for pairwise comparison probabilities, all 12 universal variables differ significantly (P < 0.01), and 4–8 pairs of groups differ for each quantitative character, although no character differed among all groups. Discriminant analysis, using the same universal characters, separates these five subgroups with 98% success in the classification matrix and 91% success in the jackknife classification matrix (Wilks' lambda = 0.000; P < 0.00). Sterile basal length and number of leaf divisions are the most discriminatory characters. In both DA and ANOVA, two suspected hybrids (see later) are excluded. These five western Andean groups are therefore treated as geographic subgroups. Some of these are also supported by their unique combinations of traits, as described later.
|
|
The San Blas subgroup contains 17 specimens from the western end of the Serranía de San Blas, Panama, between 9°14'–9°25' N and 78°34'–79°37' W (Fig. 5) at a mean elevation of 358 (80–750) m. For traits, specimens have glabrous rachillae, spirally arranged triads, and white flowers. For quantitative characters, this subgroup has few leaf divisions (16 specimens have simple leaves) and few rachillae (Fig. 6). Inflorescence bracts are present on more specimens than in other subgroups, and these bracts appear to persist longer after the subtending leaf falls. These bracts are unequal in size—in contrast to the Talamanca group. Prophylls have a mean length of 7.85 cm (N = 13) and peduncular bracts 21.75 cm (N = 4).
The Tacarcuna subgroup contains five specimens from Cerro Tacarcuna in Panama and adjacent Colombia between 8°04'–8°10' N and 77°14'–77°15' W (Fig. 5) at a mean elevation of 1360 (1300–1400) m. For traits, specimens have glabrous rachillae, distichously arranged triads, and white flowers. For quantitative characters, this group has the lowest values for most variables (Fig. 6). One additional specimen (de Nevers 8404) from the same locality appears intermediate between the Tacarcuna subgroup and character group 5 and may represent a hybrid.
The Chocó subgroup contains 36 specimens from western Colombia and Ecuador between 0°37' S–7°20' N and 76°13'– 79°20' W (Fig. 5) at a mean elevation of 290 (40–900) m. For traits, specimens have dense crustose hairs on the rachillae, spirally arranged triads, and white or red flowers. This subgroup has a different inflorescence size than other groups. It has longer sterile basal parts, interbract distances, peduncles, and rachises (Fig. 6), but the distal rachilla is shorter than the proximal one (vs. more or less equal in other subgroups). A few specimens from the central part of the range of this subgroup are smaller than others and have simple leaves. One of these (Rubiano 121) is smallest of all and very similar in leaf size to specimens of the Tacarcuna subgroup.
The Amazon subgroup contains 279 specimens from throughout the Amazon region between 17°00' S–6°12' N and 46°15'–78°38' W (Fig. 7) at a mean elevation of 385 (40– 1900) m. For traits, specimens have rachillae that are glabrous or with crustose hairs or with flattened, lacerate hairs, spirally or distichously arranged triads, low or raised (Fig. 2F) pistillate scars, and white or red flowers. The Amazon subgroup shows geographical discontinuities (Fig. 7). These gaps, however, are more likely to be collecting artifacts, given the size of the area and unevenness of collecting (Henderson, 1995
).
|
Along the eastern Andean foothills of Ecuador and Peru (Fig. 7), some specimens (N = 33, here called Raised Scars) have glabrous rachillae and raised pistillate scars (Fig. 2F). There are other specimens that appear intermediate with Widespread (glabrous rachillae and low scars) or are difficult to score for these characters, and delimitation of this group is also arbitrary. Leaf size and division are highly variable among these specimens.
From five scattered localities at higher elevations (800–1550 m) on eastern Andean slopes in Ecuador and Peru (Fig. 7), there are small or very small-sized plants (N = 7, here called High Elevation). These have a mean leaf rachis length of 12.4 cm and often simple leaves and a mean inflorescence rachis length of 2.9 cm. They are not homogeneous, and at different localities there are specimens with different combinations of traits. One specimen has flattened, lacerate hairs on the inflorescence unlike those of any other specimen examined. There are only one or two specimens from each locality, which precludes further statistical analysis.
Because of the difficulty of delimiting groups and/or small sample size, statistical analysis of these Amazon specimens (Narrow-leafed, Raised Scars, High Elevation) is not attempted here, and no geographic subgroups are recognized.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
The Amazon geographic subgroup is problematic, morphologically and taxonomically. This subgroup has considerable variation, but it cannot be further divided because of either intermediate specimens between potential groups or small sample size of other potential groups. Complexity in this group, especially in the western Amazon region and on eastern Andean slopes, may be attributable to several different factors—a zone of variation in pistillate scar morphology along eastern Andean slopes (i.e., Raised Scar); local differentiation in isolated localities at higher elevations on eastern Andean slopes (i.e., High Elevation); and sympatric, possibly ecological differentiation of leaf shape at low elevations in the western Amazon (i.e., Narrow-leafed).
In particular, the zone of variation in pistillate scars along eastern Andean slopes may be attributable to a hybrid zone between the Amazon subgroup and character group 5. There are several lines of evidence in support of this. First, there is overlap between character group 5 and the Amazon subgroup on eastern Andean slopes in southern Colombia (Fig. 4). Second, specimens of character group 5 have large, much-divided leaves, as do several specimens of the Amazon subgroup from eastern Andean slopes (represented by the far outside values in leaf division for the Amazon subgroup shown in Fig. 6). There is, in general, great variation in leaf size and number of divisions along eastern Andean slopes. Third, the raised pistillate scar (Fig. 2F) could be interpreted as intermediate between that of character group 5 (Fig. 1E) and widespread Amazon specimens (Fig. 2F). However, a hybrid zone hypothesis cannot be tested with herbarium specimens as the data source.
Taxonomically, the Amazon subgroup becomes a subspecies by default. Of the two possible taxonomic options for character group 6 (one species, or one species and six subspecies), I have chosen the latter. Even though the Amazon subgroup cannot be distinguished from the Western Andean subgroups (apart from geographic criteria), it automatically becomes a subspecies because I have applied a PSC subspecies concept to the other five Western Andean subgroups.
The two-stage approach to taxonomy used in this study— multivariate statistical analysis to delimit specimen groups and subsequent application of a specific species concept—has two implications; one is for the Palmae and the second is more general. First, this approach leads to a more realistic and certainly more scientific estimate of taxonomic diversity in the Palmae than do traditional herbarium methods. It also appears to lead to a higher number of taxa (e.g., Henderson, 2002
; Henderson and Ferreira, 2002
). Using similar methods to those used here, I recognize 18 species of Calyptrogyne (Henderson, unpublished manuscript), whereas de Nevers (1995)
recognized eight. Here I recognize six species of Hyospathe, whereas Skov and Balslev (1989)
recognized two. Based on these few studies and on these methods, the number of palm species may be more than double the currently accepted 2300.
On the other hand, multivariate studies of morphological data from more complex types of variation have failed to resolve taxonomic problems. Examples are potential hybrid zones (the Amazon subgroup in this study) and species complexes in the palm genus Geonoma (e.g., Borchsenius, 1999
; Henderson and Martins, 2002
). Henderson et al. (1995) estimated that 10% of palm species may be species complexes, and I now estimate that the percentage of species having complex variation (e.g., hybrid zones, species complexes, etc.) may be higher.
The second, general implication of this study concerns species as hypotheses. Species concepts function as hypotheses and these can be tested (e.g., Wheeler and Platnick, 2000
). In the two-stage approach used in this study—multivariate statistical analysis to delimit groups and subsequent application of a specific species concept—it is the first stage that is tested, the groups of specimens. Because these depend in turn on distinguishing traits from characters, the test is that characters are not traits, and traits are not characters. In the former case, a supposed character may turn out to be distributed as a trait. Such a misinterpretation would give an underestimation of the number of species. In the latter case, a supposed trait may be distributed as a character, giving an overestimation of the number of species. In the case of Hyospathe elegans sensu Skov and Balslev (1989)
, I reject this species concept because I find that traits are distributed as characters (e.g., pistillate flower type is a character with two states—sessile or pedicellate—not a trait); therefore, Skov and Balslev underestimated the number of species. The advantage of the multivariate approach as used in this study is that the data matrix is available for future studies and can be easily used to test the species hypotheses presented here.
|
TAXONOMIC TREATMENT |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
1. Hyospathe macrorachis Burret, Notizbl. Bot. Gart. Berlin-Dahlem 15: 34. 1940. TYPE. ECUADOR. Pastaza: Mera, 1000 m, 7 Sep 1938, H. Schultze-Rhonhof 2789 (holotype: B, destroyed). ECUADOR. Pastaza: shore of Río Pastaza, opposite Mera, 1000 m, 6 Feb 1956, E. Asplund 19249 (neotype (Skov and Balslev, 1989
): S; photo at AAU!).
Stem length: 0.6 (0.2–1.0) m, CV 0.6, N = 7. Stem diameter: 1.1 (0.8–1.4) cm, CV 0.2, N = 6. Internode length: 1.2 (0.5–1.6) cm, CV = 0.4, N = 6. Sheath length: 13.1 (9.0– 16.0) cm, CV 0.2, N = 5. Petiole length: 17.7 (6.5–36.5) cm, CV 0.6, N = 10. Rachis length: 26.4 (17.0–37.5) cm, CV 0.2, N = 10. Rachis width: 2.8 (1.9–3.4) mm, CV 0.2, N = 10. Number of divisions: 2.3 (1–4), CV 0.5, N = 11. Proximal pinna angle: 37.0° (20°–50°), CV 0.2, N = 11. Distal pinna angle: 23.0° (17°–29°), CV 0.2, N = 10. Distal pinna length: 23.9 (17.5–29.5) cm, CV 0.1, N = 10. Sterile basal part length: 3.7 (2.2–5.0) cm, CV 0.3, N = 5. Sterile basal part width: 4.6 (3.2–5.6) mm, CV 0.2, N = 5. Prophyll length: 19.7 (12.0–24.5) cm, CV 0.3, N = 3. Peduncular bract length: 44.0 (39.0–50.0) cm, CV 0.1, N = 3. Interbract distance: 6.5 (4.5–8.0) cm, CV 0.2, N = 5. Peduncle length: 25.9 (21.0–30.0) cm, CV 0.1, N = 5. Rachis length: 22.1 (15.0–36.0) cm, CV 0.3, N = 9. Rachis width: 1.9 (1.2– 2.6) mm, CV 0.2, N = 9. Rachillae number: 22.1 (11–34), CV 0.4, N = 9. Proximal rachilla length: 10.2 (8.0–12.4) cm, CV 0.2, N = 9. Distal rachilla length: 4.3 (1.7–9.5) cm, CV 0.6, N = 9. Fruit length: 7.7 (6.6–8.8) mm, CV 0.2, N = 2. Fruit diameter: 5.5 (4.0–6.9) mm, CV 0.4, N = 2.
Distribution. Eastern Andean slopes in Ecuador and northern Peru between 0°43'–5°50' S and 77°34'–78°55' W at a mean elevation of 1627 (1100–2100) m (Fig. 4).
2. Hyospathe peruviana Henderson, sp. nov.—TYPE. PERU. Huánuco: Prov. Leoncio Prado, road between Tingo María and Pucallapa, km 35, 9°10' S, 75°48' W, 1500 m, 3 Jun 1981, G. Sullivan & K. Young 1180 (holotype: NY!; isotype: MO!).
Ab omnibus speciebus generis his notulius differt: inflorescentiis infrafoliis, rachillis filiformibus, triadibus elevatis, irregularibus, floribus pistillatis pedicellatis, et sepalis pistillatis tubularibus.
Stem length: no data. Stem diameter: 0.8 (0.6–1.0) cm, CV 0.2, N = 5. Internode length: 6.4 (3.7–9.0) cm, CV = 0.4, N = 5. Sheath length: 13.3 (13.0–13.5) cm, CV 0.0, N = 2. Petiole length: 7.6 (5.5–9.5) cm, CV 0.2, N = 4. Rachis length: 17.9 (14.0–21.0) cm, CV 0.1, N = 5. Rachis width: 2.5 (2.1–3.0) mm, CV 0.2, N = 5. Number of divisions: 2.8 (1–5), CV 0.5, N = 5. Proximal pinna angle: 41.6° (30°– 55°), CV 0.2, N = 5. Distal pinna angle: 26.4° (24°–30°), CV 0.1, N = 5. Distal pinna length: 20.9 (18.0–23.4) cm, CV 0.1, N = 4. Sterile basal part length: 2.2 (1.0–3.4) cm, CV 0.5, N = 5. Sterile basal part width: 3.3 (2.3–4.5) mm, CV 0.3, N = 5. Prophyll length: 11.5 cm, N = 1. Peduncular bract length: 22 cm, N = 1. Interbract distance: 2.1 (1.1–3.3) cm, CV 0.4, N = 5. Peduncle length: 3.4 (2.0–5.0) cm, CV 0.4, N = 5. Rachis length: 8.0 (5.4–11.0) cm, CV 0.3, N = 5. Rachis width: 2.4 (1.5–3.5) mm, CV 0.3, N = 5. Rachillae number: 25.4 (18–34), CV 0.2, N = 5. Proximal rachilla length: 10.1 (5.5–13.3) cm, CV 0.3, N = 5. Distal rachilla length: 7.6 (3.9–11.0) cm, CV 0.4, N = 5. Fruit length: 8.8 (8.0–9.5) mm, CV 0.1, N = 2. Fruit diameter: 7.1 (6.8–7.3) mm, CV 0.1, N = 2.
Distribution. Eastern Andean slopes in northern Peru between 5°45'–9°15' S and 75°45'–77°45' W at a mean elevation of 1640 (1500–1850) m (Fig. 4).
3. Hyospathe wendlandiana Dammer ex Burret, Notizbl. Bot. Gart. Berlin-Dahlem 10: 855. 1929. TYPE. COLOMBIA. Antioquia: Dos Quebradas, 1435 m, 21 Jan 1880, W. Kalbreyer 1348 (holotype: B, destroyed). COLOMBIA. Antioquia: carretera Granada-San Luis, 5.5 km adelante de El Chocó, 1750 m, 20–21 Sep 1987, R. Bernal & L. Tobón 1386 (neotype (Bernal et al., 1989
): COL).
Stem length: 2.5 m, N = 1. Stem diameter: 0.9 (0.8–0.9) cm, CV 0.1, N = 2. Internode length: 3.7 (2.0–5.3) cm, CV = 0.6, N = 2. Sheath length: 21.0 (16.0–26.0) cm, CV 0.3, N = 2. Petiole length: 10.1 (4.5–16.5) cm, CV 0.6, N = 4. Rachis length: 32.9 (26.5–39.9) cm, CV 0.2, N = 4. Rachis width: 3.7 (2.6–4.8) mm, CV 0.2, N = 5. Number of divisions: 6.6 (3–11), CV 0.5, N = 5. Proximal pinna angle: 57.4° (54°–65°), CV 0.1, N = 5. Distal pinna angle: 26.8° (25°–30°), CV 0.1, N = 5. Distal pinna length: 20.6 (18.5– 22.2) cm, CV 0.1, N = 3. Sterile basal part length: 1.6 (1.0– 3.0) cm, CV 0.5, N = 5. Sterile basal part width: 5.2 (4.3– 6.8) mm, CV 0.2, N = 5. Prophyll length: no data. Peduncular bract length: no data. Interbract distance: 1.4 (0.9– 1.9) cm, CV 0.3, N = 5. Peduncle length: 3.3 (2.5–4.1) cm, CV 0.2, N = 5. Rachis length: 7.5 (6.0–9.4) cm, CV 0.2, N = 5. Rachis width: 3.3 (2.9–3.6) mm, CV 0.1, N = 5. Rachillae number: 23.8 (19–26), CV 0.1, N = 5. Proximal rachilla length: 16.2 (15.0–18.5) cm, CV 0.1, N = 4. Distal rachilla length: 13.7 (13.7–14.5) cm, CV 0.1, N = 3. Fruit length: 11.3 (9.8–12.7) mm, CV 0.2, N = 2. Fruit diameter: 6.0 (5.5–6.5) mm, CV 0.1, N = 2.
Distribution. Eastern slopes of the Cordillera Central in Colombia between 6°01'–6°55' N and 75°01'–75°04' W at a mean elevation of 1599 (1495–1750) m (Fig. 4).
Discussion. Although an isoneotype is reported to be at NY (Bernal et al., 1989
), it is not there. Other isoneotypes supposedly at AAU and BH have not been received on loan for this study. This name is therefore interpreted from R. Bernal & L. Tobón 1389 (COL, FTG), from the same locality and with the same collection date as the neotype.
4. Hyospathe frontinensis Henderson, sp. nov.—TYPE. COLOMBIA. Antioquia: Municipio de Frontino, corregimiento de Murrí, camino desde la quebrada Peñitas hacia Cieneguetas, 1200–1500 m, 19 Sep 1983, R. Bernal, G. Galeano & I. Turner 706 (holotype: COL!; isotype: NY!).
Ab omnibus speciebus generis his notulius differt: inflorescentiis infrafoliis, rachillis non filiformibus, triadibus non elevatis, regularibus, floribus pistillatis sessilibus, et sepalis pistillatis tubularibus.
Stem length: 2.0 (1.4–2.5) m, CV 0.2, N = 7. Stem diameter: 0.7 (0.5–0.9) cm, CV 0.2, N = 8. Internode length: 6.9 (3.7–9.5) cm, CV = 0.3, N = 7. Sheath length: 11.5, N = 1. Petiole length: 10.5 (7.0–14.5) cm, CV 0.2, N = 10. Rachis length: 18.0 (13.0–26.0) cm, CV 0.2, N = 10. Rachis width: 2.9 (2.9–4.1) mm, CV 0.2, N = 10. Number of divisions: 1, N = 11. Proximal pinna angle: 31.5° (25°–40°), CV 0.2, N = 10. Distal pinna angle: 23.6° (20°–27°), CV 0.1, N = 10. Distal pinna length: 23.9 (20.0–33.0) cm, CV 0.2, N = 10. Sterile basal part length: 3.3 (1.7–7.2) cm, CV 0.5, N = 10. Sterile basal part width: 3.0 (2.1–4.0) mm, CV 0.3, N = 10. Prophyll length: 10.9 (6.0–19.5) cm, CV 0.4, N = 6. Peduncular bract length: 32.5 (21.5–50.0) cm, CV 0.3, N = 6. Interbract distance: 3.8 (2.3–6.5) cm, CV 0.4, N = 10. Peduncle length: 14.0 (7.9–18.0) cm, CV 0.2, N = 10. Rachis length: 1.9 (0.1–5.5) cm, CV 0.8, N = 11. Rachis width: 1.6 (1.2–2.0) mm, CV 0.2, N = 11. Rachillae number: 3.2 (2–5), CV 0.3, N = 11. Proximal rachilla length: 18.6 (12.5– 23.0) cm, CV 0.2, N = 8. Distal rachilla length: 17.4 (11.0– 22.5) cm, CV 0.2, N = 7. Fruit length: 8.7 (8.2–9.2) mm, CV 0.1, N = 2. Fruit diameter: 6.1 (5.9–6.3) mm, CV 0.1, N = 2.
Distribution. Western slopes of the Cordillera Occidental in Colombia between 6°29'–6°45' N and 76°14'–76°25' W at a mean elevation of 1370 (1000–1725) m (Fig. 4).
5. Hyospathe pittieri Burret, Notizbl. Bot. Gart. Berlin-Dahlem 14: 137. 1938. TYPE. VENEZUELA. Aragua: Valle de Ocumare, 1000 m, 17 Sep 1937, H. Pittier 14146 (holotype: B, destroyed; lectotype (Stauffer and Stauffer, 1996
): VEN; photo in Stauffer and Stauffer, 1996
!).
Stem length: 4.7 (2.0–8.0) m, CV 0.4, N = 14. Stem diameter: 2.3 (1.3–5.0) cm, CV 0.4, N = 19. Internode length: 3.2 (1.5–7.0) cm, CV = 0.6, N = 10. Sheath length: 35.3 (24.0–39.0) cm, CV 0.1, N = 9. Petiole length: 21.8 (11.5– 41.0) cm, CV 0.3, N = 19. Rachis length: 76.1 (41.0–105.0) cm, CV 0.3, N = 12. Rachis width: 8.0 (3.9–11.3) mm, CV 0.2, N = 28. Number of divisions: 19.5 (10–27), CV 0.2, N = 14. Proximal pinna angle: 49.9° (30°–90°), CV 0.3, N = 24. Distal pinna angle: 16.5° (11°–30°), CV 0.3, N = 15. Distal pinna length: 28.7 (23.0–37.0) cm, CV 0.1, N = 11. Sterile basal part length: 1.9 (1.2–3.3) cm, CV 0.3, N = 31. Sterile basal part width: 10.8 (5.8–15.8) mm, CV 0.2, N = 22. Prophyll length: 27.5 (14.0–45.0) cm, CV 0.5, N = 4. Peduncular bract length: 32.7 (20.0–31.0) cm, CV 0.3, N = 3. Interbract distance: 2.2 (1.4–3.4) cm, CV 0.2, N = 31. Peduncle length: 4.0 (1.2–8.2) cm, CV 0.4, N = 29. Rachis length: 13.5 (4.5–20.5) cm, CV 0.3, N = 20. Rachis width: 5.3 (2.9–10.1) mm, CV 0.3, N = 22. Rachillae number: 38.7 (18–51), CV 0.3, N = 16. Proximal rachilla length: 23.6 (9.4–34.5) cm, CV 0.3, N = 16. Distal rachilla length: 12.6 (7.5–17.5) cm, CV 0.3, N = 5. Fruit length: 11.6 (8.4–14.2) mm, CV 0.2, N = 8. Fruit diameter: 6.0 (4.0–7.3) mm, CV 0.2, N = 8.
Distribution. Montane areas in northern Venezuela and Colombia and just reaching Panama between 1°08'–10°47' N and 66°35'–77°45' W at a mean elevation of 1459 (1100–1980) m (Fig. 4).
6. Hyospathe elegans Mart., Hist. nat. palm. 2: 1. 1823. TYPE. BRAZIL. Amazonas: Rio Negro, no date, C. Martius 3122 (holotype: M!; isotype: P!; F neg. 18528a).
Stem length: 2.7 (1.0–7.0) m, CV 0.5, N = 91. Stem diameter: 1.0 (0.3–2.3) cm, CV 0.4, N = 291. Internode length: 3.9 (1.0–11.0) cm, CV = 0.4, N = 238. Sheath length: 16.8 (4.5–35.0) cm, CV 0.4, N = 149. Petiole length: 15.0 (1.0–41.0) cm, CV 0.5, N = 286. Rachis length: 41.0 (5.5–125.0) cm, CV 0.5, N = 264. Rachis width: 4.0 (1.4– 8.5) mm, CV 0.3, N = 321. Number of divisions: 4.5 (1– 28), CV 1.1, N = 290. Proximal pinna angle: 40.4° (10°– 90°), CV 0.4, N = 297. Distal pinna angle: 23.4° (15°–35°), CV 0.2, N = 232. Distal pinna length: 26.4 (13.0–46.0) cm, CV 0.3, N = 155. Sterile basal part length: 1.6 (0.3–6.0) cm, CV 0.6, N = 351. Sterile basal part width: 5.3 (1.4– 16.7) mm, CV 0.4, N = 286. Prophyll length: 13.1 (1.0– 37.0) cm, CV 0.6, N = 49. Peduncular bract length: 24.5 (7.0–50.0) cm, CV 0.4, N = 34. Interbract distance: 1.7 (0.5–4.2) cm, CV 0.4, N = 349. Peduncle length: 3.6 (0.9– 12.5) cm, CV 0.5, N = 347. Rachis length: 6.3 (0.1–22.0) cm, CV 0.6, N = 332. Rachis width: 3.3 (1.1–8.3) mm, CV 0.4, N = 335. Rachillae number: 15.9 (2–45), CV 0.5, N = 329. Proximal rachilla length: 18.0 (1.4–40.0) cm, CV 0.4, N = 234. Distal rachilla length: 15.1 (1.4–37.0) cm, CV 0.4, N = 199. Fruit length: 10.0 (1.3–15.7) mm, CV 0.3, N = 72. Fruit diameter: 6.2 (3.0–8.4) mm, CV 0.2, N = 72.
6a. Hyospathe elegans subsp. elegans
Hyospathe filiformis H. Wendl. ex Drude in Mart., Fl. bras.: Palmae II fasc. 86 vol. 3(2): 522. 1882. TYPE. BRAZIL. Amazonas: Pará: Yaburu, Rio Yapurá, no date, C. Martius s.n. (holotype: M; isotype: P!; F neg. 18527).
Hyospathe gracilis H. Wendl. ex Drude in Mart., Fl. bras.: Palmae II fasc. 86 vol. 3(2): 523. 1882. TYPE. PERU. San Martín: Río Huallaga, Tocache, no date, E. Poeppig 2057 (holotype: W, destroyed; F neg. 29888).
Hyospathe brevipedunculata Dammer, Verh. Bot. Vereins Prov. Brandenburg 48: 126. 1907. TYPE. BRAZIL. Acre: Juruá Miry, Belem, Sep 1901, E. Ule 5881 (holotype: B, destroyed; lectotype (Skov & Balslev, 1989
): G; isotype, MG!).
Hyospathe schultzeae Burret, Notizbl. Bot. Gart. Berlin- Dahlem 13: 340. 1936. TYPE. ECUADOR. Pastaza: Río Pastaza, near Puyo, 850 m, 12 May 1935, H. Schultze-Rhonhof 1857 (holotype: B, destroyed). ECUADOR. Pastaza: Puyo-Macas road km 8 and 2 km to the east, 1°30' S, 77°57' W, 1080 m, 22 Mar 1987, H. Balslev 62429 (neotype (Skov & Balslev, 1989
): AAU!; isoneotype: NY!).
Hyospathe pallida H. E. Moore, Gentes Herb. 8: 197. 1949. TYPE. COLOMBIA. Putumayo: Uchupayaco, between Urcusique and Umbria, Río Uchupayaco, 300 m, 22–23 Feb 1942, R. Schultes 3291 (holotype: BH!; isotype: GH).
Hyospathe maculata Steyerm., Fieldiana Bot. 28: 89. 1951. TYPE. VENEZUELA. Bolívar: Quebrada O-paru-má, between Santa Teresita de Kavanayén and Río Pacairao, 1065– 1220 m, 20–21 Nov 1944, J. Steyermark 60406 (holotype: F!).
Rachis length: 43.4 (9.0–125.0) cm, CV 0.4, N = 196. Number of divisions: 4.3 (1–25), CV 1.0, N = 219. Prophyll length: 15.0 (3.3–27.0) cm, CV 0.4, N = 23. Peduncular bract length: 25.0 (8.0–42.0) cm, CV 0.4, N = 20. Rachillae number: 16.3 (5–45), CV 0.5, N = 253.
Distribution. Throughout the Amazon region between 17°00' S–6°12' N and 46°15'–78°38' W at a mean elevation of 385 (40–1900) m (Fig. 7).
6b. Hyospathe elegans subsp. costaricensis Henderson, subsp. nov.—TYPE. COSTA RICA. Limón: Cordillera de Talamanca, between headwaters of Río Madre de Dios and Quebrada Barreal, 10°02' N, 83°27' W, 400–440 m, 5 Sep 1988, M. Grayum, G. Herrera & R. Robles 8784 (holotype: NY!; isotypes: F!, MO!).
Ab omnibus subspeciebus his notulius differt: pinnis paribus 8–28 et rachillis 20–37.
Rachis length: 82.4 (61.0–102.0) cm, CV 0.2, N = 7. Number of divisions: 23.3 (8–28), CV 0.3, N = 7. Prophyll length: 31.9 (26.7–37.0) cm, CV 0.3, N = 2. Peduncular bract length: 29.3 (26.0–32.5) cm, CV 0.2, N = 2. Rachillae number: 28.1 (20–37), CV 0.3, N = 7.
Distribution. Atlantic slope of Cordillera Talamanca and Pacific slope of Fila Costeña in Costa Rica between 8°45'– 10°02' N and 82°55'–83°51' W at a mean elevation of 967 (420–1200) m (Fig. 5).
6c. Hyospathe elegans subsp. concinna (H. E. Moore) Henderson, stat. nov. Hyospathe concinna H. E. Moore, Gentes Herb. 8: 195. 1949. TYPE. PANAMA. Coclé, El Valle de Anton, 600–1000 m, 8 Dec 1938, P. Allen 1202 (holotype: MO!).
Chamaedorea falcaria L. H. Bailey, Gentes Herb. 6: 254. 1943. TYPE. PANAMA. Coclé: El Valle de Anton, 800 m, 10 May 1942, P. Allen 2949 (holotype: BH!).
Rachis length: 31.2 (16.0–47.5) cm, CV 0.4, N = 12. Number of divisions: 2.9 (1–7), CV 0.8, N = 13. Prophyll length: 21.8 (15.5–28.0) cm, CV 0.4, N = 2. Peduncular bract length: 24.7 (19.0–31.0) cm, CV 0.2, N = 3. Rachillae number: 9.5 (5–15), CV 0.3, N = 13.
Distribution. Cerro Gaital and the eastern end of the Cordillera Central in Panama between 8°32'–8°42' N and 80°05'– 81°05' W at a mean elevation of 755 (500–1000) m (Fig. 5).
6d. Hyospathe elegans subsp. sanblasensis Henderson, subsp. nov. TYPE: PANAMA. San Blas: Llano-Cartí road, km 12, 16 Dec 1987, A. Henderson & H. Herrera 729 (holotype: NY!; isotypes: AAU!, COL!).
Ab omnibus subspeciebus his notulius differt: rachillis glabris; triadibus spiralibus; et floribus staminatis albis.
Rachis length: 14.9 (9.0–27.5) cm, CV 0.3, N = 17. Number of divisions: 1.1 (1–2), CV 0.2, N = 17. Prophyll length: 7.9 (5.4–10.0) cm, CV 0.2, N = 13. Peduncular bract length: 21.8 (17.5–27.0) cm, CV 0.2, N = 4. Rachillae number: 4.4 (2–6), CV 0.3, N = 14.
Distribution. Western end of the Serranía de San Blas, Panama between 9°14'–9°25' N and 78°34'–79°37' W at a mean elevation of 358 (80–750) m (Fig. 5).
6e. Hyospathe elegans subsp. tacarcunensis Henderson, subsp. nov. TYPE—PANAMA. Darién: Parque Nacional del Darién, slopes of Cerro Mali, ca. 22 km E of Pucuro, 8°04' N, 77°14' W, 700–1400 m, G. de Nevers, H. Cuadros, B. Hammel & H. Herrera 8391 (holotype: NY!; isotypes: CAS!, MO!).
Ab omnibus subspeciebus his notulius differt: rachillis glabris; triadibus dichotomis; et floribus staminatis albis.
Rachis length: 7.8 (5.5–9.0) cm, CV 0.2, N = 5. Number of divisions: 1, N = 6. Prophyll length: 2.6 (1.0–4.2) cm, CV 0.9, N = 2. Peduncular bract length: 8.0, N = 1. Rachillae number: 3.2 (2–5), CV 0.3, N = 5.
Distribution. Cerro Tacarcuna in Panama and adjacent Colombia between 8°04'–8°10' N and 77°14'–77°15' W at a mean elevation of 1360 (1300–1400) m (Fig. 5).
6f. Hyospathe elegans subsp. sodiroi (Dammer ex Burret) Henderson, stat. nov. Hyospathe sodiroi Dammer ex Burret, Notizbl. Bot. Gart. Berlin-Dahlem 10: 857. 1929. TYPE. ECUADOR. Pichincha: Santo Domingo, Oct 1887, Sodiro 187/4 (holotype: B, destroyed; lectotype (Skov & Balslev, 1989
): P).
Rachis length: 38.6 (15.5–85.0) cm, CV 0.4, N = 21. Number of divisions: 4.6 (1–16), CV 0.7, N = 22. Prophyll length: 12.2 (8.0–17.0) cm, CV 0.3, N = 5. Peduncular bract length: 33.0 (22.0–50.0) cm, CV 0.5, N = 3. Rachillae number: 17.7 (4–32), CV 0.4, N = 29.
Distribution. Western Colombia and Ecuador between 0°37' S–7°20' N and 76°13'–79°20' W at a mean elevation of 290 (40–900) m (Fig. 5).
|
FOOTNOTES |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONTAXONOMIC TREATMENTLITERATURE CITED |
|---|
Bernal R. G. Galeano A. Henderson 1989 Neotypification of Colombian palms collected by W. Kalbreyer. Taxon 38: 98-106
Borchsenius F. 1999 Morphological variation in Geonoma cuneata in western Ecuador. Memoirs of the New York Botanical Garden 83: 131-139
Burret M. 1929 Die gattung Hyospathe Mart. Notizblatt des Botanischen Gartens und Museums zu Berlin–Dahlem 10: 854-859[CrossRef]
Burret M. 1933 Palmae Neogeae IV. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 11: 857-866[CrossRef]
Burret M. 1936 Palmae Neogeae X. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 13: 339-347[CrossRef]
Burret M. 1938 Eine interessante neue Hyospathe-Art von Venezuela. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 14: 137-138[CrossRef]
Burret M. 1940 Palmae. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 15: 23-28[CrossRef]
Davis J. K. Nixon 1992 Populations, genetic variation, and the delimitation of phylogenetic species. Systematic Biology 41: 421-435[CrossRef]
de Nevers G. 1995 Notes on Panama palms. Proceedings of the California Academy of Sciences 48: 329-342
Hammel B. M. Grayum C. Herrera N. Zamora 2003 Manual de plantas de Costa Rica, vol. II, Gimnospermas y monocotiledóneas (Agavaceae-Musaceae). Monographs in Systematic Botany from the Missouri Botanical Garden 92: 1-694
Henderson A. 1995 The palms of the Amazon. Oxford University Press, New York, New York, USA
Henderson A. 2002 Phenetic and phylogenetic analysis of Reinhardtia (Palmae). American Journal of Botany 89: 1491-1502
Henderson A. E. Ferreira 2002 A morphometric study of Synechanthus (Palmae). Systematic Botany 27: 693-702[ISI]
Henderson A. R. Martins 2002 Classification of specimens in the Geonoma stricta (Palmae) complex: the problem of leaf size and shape. Brittonia 54: 202-212[CrossRef][ISI]
Henderson A. G. Galeano R. Bernal 1995 Field guide to the palms of the Americas. Princeton University Press, Princeton, New Jersey, USA
Holmgren P. N. Holmgren L. Barnett 1990 Index herbariorum, part I: the herbaria of the world, 8th ed. Regnum Vegetabile 120: 1-693
Luckow M. 1995 Species concepts: assumptions, methods, and applications. Systematic Botany 20: 589-605[CrossRef][ISI]
Moore H. 1949 Palmae incertae et novae. Gentes Herbarum 8: 195-199
Nixon K. Q. Wheeler 1990 An amplification of the phylogenetic species concept. Cladistics 6: 211-223
Rohlf F. 2000 NTSYS. Numerical taxonomy and multivariate analysis system, version 2.1. Exeter Software, New York, New York, USA
Skov F. H. Balslev 1989 A revision of Hyospathe (Arecaceae). Nordic Journal of Botany 9: 189-202[ISI]
Stauffer F. L. Stauffer 1996 Tipos de palmas (Arecaceae) en el herbario nacional de Venezuela. Acta Botanica Venezuelica 19: 60-75
Steyermark J. 1951 Palmae. Contributions to the flora of Venezuela. Fieldiana Botany 28: 71-92
Tabachnik B. L. Fidell 2001 Using multivariate statistics, 4th ed. Allyn & Bacon, Boston, Massachusetts, USA
Uhl N. J. Dransfield 1987 Genera palmarum. Allen Press, Lawrence, Kansas, USA
Wilkinson L. 1997 Systat statistics, version 7.0 for Windows. SPSS, Chicago, Illinois, USA
Wheeler Q. N. Platnick 2000 The phylogenetic species concept (sensu Wheeler and Platnick). In Q. Wheeler and R. Meier [eds.], Species concepts and phylogenetic theory, 55–69. Columbia University Press, New York, New York, USA
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
