| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
New York Botanical Garden, Bronx, New York 10458 USA
Received for publication January 15, 2002. Accepted for publication April 19, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
Key Words: Central America • multivariate analysis • Palmae • phenetics • phylogenetics • Reinhardtia
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
). Because of this Reinhardtia has been regarded as an isolated genus, with no obvious close relatives among other neotropical arecoid palms (Moore, 1957
). Uhl and Dransfield (1987)
placed the genus in its own subtribe, the Malortieinae.
Recent molecular studies have clarified relationships of Reinhardtia. Lewis and Doyle (2001)
placed it as sister group to the chamaedoreoid palms, although with less than 50% bootstrap support. Two other possible sister genera (Podococcus and Roystonea, see below) were not included in this study. Asmussen and Chase (2001)
placed Reinhardtia as sister genus to Podococcus, with less than 50% bootstrap support, and this clade as sister to Roystonea and the chamaedoreoids. Most recently, Lewis and Doyle (in press)
placed Reinhardtia and Roystonea as sister genera, with 89% bootstrap support.
Reinhardtia contains six species. Five were recognized by Moore (1957)
, and these are widely distributed from Mexico to northwestern Colombia. A sixth species, from the Dominican Republic, was added by Read, Zanoni, and Mejía (1987)
. All species occur in lowland to montane moist forests, from sea level to 1600 m elevation. Moore divided the five species known to him into two subgenera, Reinhardtia and Malortiea. The former contained two species of larger plants, and the latter three species of small or very small plants. One of these, R. gracilis (Wendl.) Burret, is the most widely distributed and exhibits considerable morphological variation. Moore divided it into four varieties, noting that there were still problems associated with this division.
Moore (1957
, p. 543) wrote that in Reinhardtia "reduction in size, foliage, inflorescences, flowers and fruits closely approaches a linear series." He considered R. elegans Liebm., with tall, solitary stems, much divided leaves, and inflorescences with numerous rachillae, to be the most "primitive" species, followed by R. latisecta (Wendl.) Burret, with smaller, clustered stems and less divided leaves. Further reduction occurs in R. gracilis and R. simplex (Wendl.) Burret, and in the species with the smallest stems, R. koschnyana (Wendl. & Dammer) Burret, simple leaves and spicate inflorescences are found. Reinhardtia koschnyana has been cited as an example of the smallest palm known (e.g., Moore, 1957
; Henderson et al., 1995
). Moore was not aware of R. paiewonskiana Read, Zanoni, & Mejía, but this species, with tall, solitary stems, much divided leaves, and numerous rachillae, fits at the primitive end of his series. Another point of interest in this supposed series is that the largest species are in the northern part of the range and the smallest in the southern part (Henderson, Galeano, and Bernal, 1995
).
Moore's (1957)
work is the most recent systematic treatment of the genus. He considered that his conclusions were at best preliminary and that they reflected inadequate material and suggested a problem for further study. Since 1957 much new material has been collected. Moore examined a total of 117 specimens; here 476 have been examined. The present study has several objectives. The first is to provide a review of the species boundaries in the genus, based on a morphometric study of data taken from herbarium specimens. The second is to reconsider the polymorphic Reinhardtia gracilis using similar methodology. These two objectives are a precursor to a treatment of the genus for Flora Mesoamericana. The third objective, in contrast to the phenetic approach, is to examine phylogenetic relationships among species, based on cladistic analysis of qualitative characters. The resulting phylogeny is then used to test Moore's ideas on the linear series of size reduction in the genus.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
). A list of specimens examined is available from the author. Latitude, longitude, and elevation were also recorded. Most specimens are mapped, and each dot on the maps represents one or more specimens.
|
as the umbo, is a nipple-shaped structure at the apex of the fruit, 1–2.5 mm long, that remains whitish-brown as the fruit matures to purple-black. It is found only in some varieties of R. gracilis.
|
Multivariate methods of analysis were carried out using the programs Systat (Wilkinson, 1997
) and NTSYS (Rohlf, 2000
). Principle component analysis (PCA), cluster analysis (CA), and discriminant analysis (DA) were performed in sequence. Principal component analysis was used to detect groups of specimens and to estimate the contribution of each variable to the analysis. Because the variables are both continuous and discrete and because some are not normally distributed, the data were log10-transformed. Principal component analysis, using Systat, was based on correlation matrices of transformed quantitative variables. Only those axes corresponding to components with eigenvalues greater than 1.0 were extracted and plotted. Cluster analysis was used to assess the groupings found in PCA. Cluster analysis is better at representing distances among similar specimens, whereas PCA is better at representing distances among groups of specimens (Sneath and Sokal, 1973
). Cluster analysis was carried out using NTSYS. Standardized data were used to compute a distance matrix based on average taxonomic distance, and this was subjected to the unweighted pair-group method, arithmetic average (UPGMA) clustering algorithm. Discriminant analysis was used to test differences among the groups found in PCA and CA and to find the most useful discriminant variables. Discriminant analysis, using Systat, was performed on transformed variables of pre-classified groups of specimens. The automatic forward stepping option of Systat was used to select the variables that contributed most to the separation of groups. Wilk's lambda, best variables, F-to-remove statistics, and the jackknife classification tables are reported. The F-to-remove statistics give an indication of the relative importance of the variables used in the analysis. Discriminant analysis was also used to assign unclassified specimens to groups. In ordination and clustering, specimens with missing values (using the listwise deletion option in Systat) were excluded, as were variables with few entries. Analyses are thus based on subsets of the data, as noted below.
For phylogenetic analysis of the species of Reinhardtia, three outgroups were used (the chamaedoreoid Hyophorbe lagenicaulis, Podococcus barteri, and Roystonea oleracea), based on recent molecular data (Asmussen and Chase, 2001
; Lewis and Doyle, 2001
, in press
, see INTRODUCTION). The ingroup contains the six species recognized by Moore (1957)
and Read, Zanoni, and Mejía (1987)
. Twelve characters were used in this analysis (Appendix 1), taken from Moore (1957)
, Read, Zanoni, and Mejía (1987)
, and from examination of specimens. The data matrix was constructed and edited with Winclada (Nixon, 1999
). Parsimony analyses were conducted with Nona (Goloboff, 1993
). The mult*max* search strategy was used with the following settings: hold = 1000, mult x N = 100, and hold/ = 10.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
|
A scatterplot of the first two components (Fig. 2) shows that specimens from each species cluster together, with little overlap. Reinhardtia paiewonskiana, R. elegans, R. latisecta, and R. gracilis are separated along PC axis 1, reflecting differences in overall size. Reinhardtia simplex, R. koschnyana, and R. gracilis are separated on PC axis 2, reflecting the contrast between wider angprox and longer lenrach in R. simplex and R. koschnyana and longer lendist in R. gracilis. Although R. simplex and R. koschnyana do not overlap, there is very little separation between them.
|
Discriminant analysis was carried out using the same set of specimens and variables, with the six species recognized by Moore (1957)
and Read, Zanoni, and Mejía (1987)
as grouping variables. Results (Table 3) show that specimens are classified with 100% success.
|
Reinhardtia paiewonskiana occurs in western Dominican Republic, where it is known from two nearby localities in a restricted area of the Sierra de Baoruco (Fig. 3). It was destroyed at a third, the type locality prior to 1987 (T. Zanoni, New York Botanical Garden, personal communication). Its elevation range is 800–1200 m (mean = 905 m). It is the species of tallest plants in the genus, with a mean plant height and stem diameter of 9.6 m and 26.3 cm, respectively. It is a rare species, and only five specimens have been examined in this study.
|
Reinhardtia latisecta has a patchy distribution, occurring in Belize, Honduras, Nicaragua, and Costa Rica (Fig. 3). Its elevation range is 30–950 m (mean 282 m). It is a species of larger plants, with a mean plant height and stem diameter of 4.2 m and 4.3 cm, respectively. Despite its wide range, it appears, like R. elegans, to be an uncommon and seldom-collected species. Only 21 specimens have been examined in this study. It has a wide range of habitats; in Belize it can occur in montane forest on steep slopes or in tidally flooded mangrove forests (J. Janovec, New York Botanical Garden, personal communication).
Reinhardtia gracilis is the most widespread species, occurring from southeastern Mexico to northwestern Colombia (Fig. 3). It is a species of smaller plants, with a mean plant height and stem diameter of 1.3 m and 0.6 cm, respectively. It is by far the most commonly collected species, and 248 specimens have been examined. Variation within this species is discussed below.
Reinhardtia simplex is found in southeastern Mexico, Guatemala, Belize, Honduras, Nicaragua, Costa Rica, Panama, and northwestern Colombia (Fig. 3). It is a species of smaller plants, with a mean stem height and diameter of 1.0 m and 0.5 cm. It is a common species; 135 specimens have been examined in this study, most of which are from Nicaragua and Costa Rica.
The elevation range of Reinhardtia simplex is 5–1000 m (mean 296 m). There are significant (P < 0.01) negative associations between elevation and six variables. Squared multiple r for the regression of petiole length on elevation is 0.087, of length of proximal margin 0.094, rachis length 0.135, number of veins 0.118, number of rachillae 0.284, and inflorescence rachis length 0.315. As elevation increases, leaves and inflorescences become smaller.
There is a gap in the range of the species in central Panama. Based on a two-sample, separate variance t test on log10-transformed variables (P < 0.01), specimens from the eastern part of the range in Darién Province, Panama, and adjacent Colombia have significantly longer leaf rachises with more veins than other specimens.
Reinhardtia koschnyana has a disjunct distribution and occurs in Nicaragua, Costa Rica, eastern Panama, and northwestern Colombia (Fig. 3). It is the species of smallest plants, with a mean stem height and diameter of 0.7 m and 0.5 cm, respectively. It is a fairly common species, and 57 specimens have been examined. Its elevation range is 10–830 m (mean 178 m). There are no significant associations between elevation and any variable. There are no significant differences between the Nicaraguan/Costa Rican specimens and the Panamanian/Colombian specimens in any variable.
Variation within Reinhardtia gracilis
Moore (1957)
divided Reinhardtia gracilis into four varieties: var. gracilis, var. gracilior (Burret) H. E. Moore, var. rostrata (Burret) H. E. Moore, and var. tenuissima H. E. Moore.
Moore (1957)
based var. tenuissima on a single specimen from Mexico, differing from the sympatric var. gracilior by its 5 mm versus 3–3.5 mm staminate flowers and 16–17 versus 8–10 stamens. There are now several more specimens available with staminate flowers, and it is apparent that ranges of stamen number both within a single specimen and within tenuissima and gracilior are wider than reported by Moore. For example, Soto 37 has flowers with 12–15 stamens; Hernández 1291 (the type of var. tenuissima) has flowers with 14–15 stamens (my count) or 16–17 (Moore's count); Calderón 1747 has 10–15 stamens. Other specimens of var. gracilior have a range of 8–12 stamens. For staminate flower length, estimated from the staminate petal length, there is also continuous variation. Given these ranges in stamen number and staminate flower length, var. tenuissima is not recognized here.
Specimens of Reinhardtia gracilis were assigned to the remaining three varieties (var. gracilis, var. gracilior, and var. rostrata) using the key provided by Moore (1957)
. In order to test the distinctness of these three varieties, two PCAs were carried out. The first used 113 specimens with complete data for 15 variables with >175 entries (lenpeti was excluded because it is seldom recorded for var. gracilis). Variables with <120 entries (stemdia, lenstam, stamens, fruitle, fruitdi, rostrum) were also excluded. The second PCA used these 15 variables plus the three fruit variables (fruitle, fruitdi, rostrum). This led to a considerably smaller sample size of 40 specimens, because few specimens have fruits.
The first PCA produced four components with eigenvalues greater than one. The first component explains 50% of the total variance. It has high to moderate loadings for most leaf and inflorescence variables, and low loadings for nodivis, angdist, and angprox. The second component explains a further 10% of the variance. It has moderately high loadings for norachi and infrach. The third and fourth components explain a further 8% and 7%, respectively, of the variance. A scatterplot (not shown) of the first two components shows some overlap among the varieties, with var. gracilior tending to occur on the negative side of PC axis 1, var. rostrata in the middle, and var. gracilis on the positive side, reflecting an increase in size. Similar results were obtained with the second analysis, including fruits. Fruit variables had low loadings on the first component and high on the second. In the scatterplot (not shown), there was better separation between the three varieties.
Cluster analysis was carried out using the same number of specimens (113) and variables (15) as for the first PCA analysis. In the phenogram (not shown; cophenetic correlation r = 0.83), there is overlap among all three varieties, particularly var. gracilior and var. rostrata.
Two DAs were carried out with the varieties as grouping variables, without fruit and with fruit, as in the PCA. Results show that specimens can be classified into three varieties with 94% success using quantitative variables excluding fruits and with 98% success including fruits (Table 3). All three varieties are geographically separate from one another, except for two outlying populations of var. gracilior (see below). Within each of these three varieties it is possible to detect further variation.
Variation within var. gracilior
Variety gracilior occurs in Mexico with outlying populations in Honduras and Nicaragua (Fig. 4). Moore (1957)
considered that this variety also occurred in Belize. The only specimen from that country he cited was the type, which was destroyed in Berlin in 1944. Based on Burret's (1932)
description, the type appears to represent var. gracilis, not var. gracilior (which will cause nomenclatural changes). There appear to be some differences among the three populations (Mexico, Honduras, Nicaragua) of var. gracilior, although sample size is small for Nicaragua (N = 6) and Honduras (N = 3). ANOVA (P < 0.05) shows that Honduran specimens have significantly shorter apical rachillae than Mexican specimens, and Nicaraguan specimens have significantly shorter petioles, fewer windows, shorter distal margins, and shorter proximal margins than Mexican specimens. The elevation range of Honduran plants is 600 m (only one record) and of Nicaraguan plants is 600–1100 m (mean 954 m).
|
Variation within var. gracilis
Moore (1957)
distinguished var. gracilis by its 16–22 stamens and larger leaves and inflorescences. In this study five specimens have staminate flowers present, and 16–18 stamens were counted. (Moore reported 17–19 stamens, but based his figure of 16–22 on an earlier count by Wendland.) Variety gracilis occurs in Belize, Guatemala, and Honduras (Fig. 4). Its range is split by a gap in western Honduras, near the border with Guatemala (Fig. 4). Moore (1957)
noted that the Belize/Guatemalan specimens were somewhat different from the Honduran ones, but he did not find any clear-cut distinction between them. The elevation range of Belize/Guatemalan plants is 75–940 m (mean 487 m) and of Honduran plants 100–770 (mean 326 m). Sample size in these two populations is too small to test for associations between elevation and variables.
Principal component analysis was carried out using 24 specimens with complete data for nine leaf variables with >25 entries. All inflorescence variables were excluded because the large size of the inflorescences of Honduran plants means that they are seldom collected in their entirety and are consequently seldom present on specimens. Other variables with <17 entries (plheigh, stemdia, lenstam, stamens, fruitle, fruitdi, rostrum) were also excluded.
The first PCA produced three components with eigenvalues greater than one. The first component explains 52% of the total variance. It has high to moderate loadings for most variables and low, negative loadings for angdist and angprox. The second component explains a further 16% of the variance. It has high positive loadings for angprox contrasted with moderately high negative loading for angdist. The third component explains a further 13% of the variance. A scatterplot (not shown) of the first two components shows that the two populations are separated along principal component axis 1, based on leaf size.
Cluster analysis was carried out using the same number of specimens and variables as for the PCA. In the phenogram (not shown, cophenetic correlation r = 0.84), the two populations cluster separately except for two specimens of Honduran gracilis, which cluster with Belize/Guatemalan gracilis.
Discriminant analysis was carried out using the same number of specimens (24) and variables (nine), and with the two populations as grouping variables (Belize/Guatemalan and Honduran). Results (Table 3) show that specimens of the two populations are classified with 96% success. A further distinguishing variable is found in the fruits, although sample size is small. In Honduran specimens (N = 2), the mean rostrum length is 1.4 mm; in Belize/Guatemalan specimens (N = 13), the mean rostrum length is 1.0 mm.
Variation within var. rostrata
Moore (1957)
distinguished var. rostrata by its distichous or partially distichous arrangement of the triads and prominently rostrate fruit. Another variable, fruiting rachillae diameter, appears to be greater in var. rostrata than in the other two varieties. This variable was not included in this study because rachillae thicken during fruit development, thus making comparable measurements difficult. Var. rostrata occurs in Honduras, Nicaragua, Costa Rica, Panama, and northwestern Colombia (Fig. 4), although it is patchily distributed and rare in Panama and Colombia.
There are some specimens (N = 14) from a few areas in Panama and adjacent Colombia (Fig. 4) that have much larger fruits than normal (mean length and diameter 19.0 and 9.2 mm, respectively, versus 13.5 and 6.0 mm). Only one of these larger-fruited specimens has staminate flowers and has 13 stamens (versus the 7–10 found in normal-fruited var. rostrata).
To test for differences between these two groups (large-fruited, normal-fruited), PCA was carried out using 41 specimens with complete data for 17 variables with >53 entries. Other variables were excluded (plheigh, stemdia, lenpeti, lenstam, stamens), either because of small sample size or because they were seldom recorded for large-fruited rostrata. Inclusion of fruit variables considerably reduced sample size. Principal component analysis produced five components with eigenvalues greater than one. The first component explains 33% of the total variance. It has high to moderate loadings for few variables. The second component explains a further 17% of the variance. It has moderately high positive loadings for fruitle and fruitdi, contrasted with high negative loadings for norachi and infrach. The third to fifth components explain a further 11, 9, and 6%, respectively, of the variance. A scatterplot (not shown) of the first two components shows that the two groups are clearly separated from one another, except for one specimen of large-fruited rostrata.
Cluster analysis was carried out using the same number of specimens and variables as for the PCA. In the phenogram (not shown; cophenetic correlation r = 0.84) large-fruited specimens cluster together, except for one specimen clustering among normal-fruited specimens.
Discriminant analysis was carried out using the same number of specimens (41) and variables (17) and with the two groups (large-fruited and normal-fruited) as grouping variables. Results (Table 3) show that specimens of the two groups are classified with 98% success. Discriminant analysis indicated that the two specimens from Panama without fruits should be included in the large-fruited group, as mapped (Fig. 4).
The elevation range of the normal-fruited plants is 5–1100 m (mean 466 m) and of large-fruited plants is 200–950 m (mean 560 m). In normal-fruited rostrata, there are significant (P < 0.05) negative associations between elevation and four variables. Squared multiple r for the regression of length of rachis on elevation is 0.075, number of veins 0.067, length of distal margin 0.098, and length of basal margin 0.170. As elevation increases, leaves become smaller.
Phylogeny
Twelve qualitative characters were used (Appendix 1) for species-level analysis, with Hyophorbe lagenicaulis, Podococcus barteri, and Roystonea oleracea as outgroups. The data matrix is shown in Table 4. NONA found one tree of 23 steps, with a consistency index of 0.69 and a retention index of 0.73 (Fig. 5).
|
|
two subgenera, only Malortiea is monophyletic. The three species (R. gracilis, R. simplex, R. koschnyana) share one synapomorphy: petiole wings present. It is of interest to note that one of the outgroups, Roystonea oleracea, has windows in the apical part of the leaf, in the same manner as R. paiewonskiana and R. elegans, supporting a sister group relationship between Roystonea and Reinhardtia.
Moore (1957)
considered that the species of Reinhardtia formed a linear series of reduction in size. Figure 6 shows box-and-whisker plots of quantitative variables from plant size and leaves, with species arranged according to the phylogeny presented in Fig. 5. Plant height and stem diameter decrease from the largest-stemmed, basal species, R. paiewonskiana, to the smallest, most derived species, R. simplex and R. koschnyana. For leaves, a similar decrease is evident in some variables (rachis length, rachis width, number of veins, number of divisions), showing an overall decrease in leaf size. In other variables there is no such decrease (petiole length, number of windows, distal margin length, proximal margin length), but a marked increase in R. latisecta. Distal margin angle shows little change, and proximal margin angle shows no particular order except for an increase from R. gracilis to R. koschnyana.
|
Leaves of the next three species in the series, R. gracilis, R. simplex, and R. koschnyana, follow a similar reduction in size as seen between R. paiewonskiana and R. elegans. The leaf of R. gracilis is a smaller version of that of R. latisecta. Between R. gracilis and R. simplex, there is a "joining" of the two distal segments of the leaf and a concomitant narrowing of leaf angles, to give a large distal segment with only a short distal margin. There is also a loss of windows. Between R. simplex and R. koschnyana there is a reduction in length of distal margin, to give a simple leaf, but also a slightly longer distal margin.
Figure 7 shows box-and-whisker plots of quantitative variables from inflorescences. Size and number tend to decrease phyletically, except for Reinhardtia latisecta. Inflorescences of R. paiewonskiana have a mean rachis length of 7 cm and 8 rachillae. Those of R. elegans are very similar, with a rachis 9 cm long and 9 rachillae. A large change takes place in R. latisecta. Inflorescences have a 23 cm long rachis with 39 rachillae. Inflorescences of R. gracilis and R. simplex are much reduced in size, with a rachis 2 cm long and 4–5 rachillae, and that of R. koschnyana is spicate. Staminate flower and fruit variables also exhibit a phyletic decrease in size and number (box-and-whisker plots not shown).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
; Read, Zanoni, and Mejía, 1987
) is supported by this study. The species are clearly distinguished from one another based on both quantitative variables and qualitative characters. There is no evidence of hybridization despite areas of sympatry of several species. However, it is not evident from specimen labels if sympatric species occupy the same habitat. Phylogenetic analysis does not support Moore's (1957)
recognition of two subgenera, although Malortiea is monophyletic.
Distributions of all species are now known to be much wider than recorded by Moore (1957)
. For example, Moore knew R. latisecta only from Belize, but it has now been collected in Honduras, Nicaragua, and Costa Rica. New collections also show range disjunctions in several species (e.g., R. elegans, R. latisecta, R. koschnyana), either the result of dispersal or fragmentation of former wide ranges.
Within some species there are small but significant correlations between elevation and some leaf and inflorescence variables. Leaves and inflorescences become smaller with increasing elevation. This contrasts with the situation among species, in which there is a phyletic decrease in size associated with a decrease in elevation.
Phylogenetic relationships, latitudinal distribution, and elevation range of species suggest speciation has taken place from large- to small-size plants, from north to south, and from higher to lower elevations. This is consistent with a hypothesis of an origin and early diversification in the late Cretaceous or Paleocene, when there may have been land connections between the Maya Terrane and proto-Antillean islands, including Hispaniola (Donnelly, 1992
; Coates and Obando, 1996
; Iturralde-Vinent and MacPhee, 1999
). Subsequent spread southward may have taken place through the Chortis Block during Eocene-Oligocene, finally reaching southern Central America as it became emergent during Miocene and Pliocene. However, given the lack of a fossil record for Reinhardtia, incomplete knowledge of the geological history of Central America (Donnelly, 1992
), and an unknown sequence of climatic changes in the region (Graham, 1992
), this biogeographic hypothesis remains tentative.
Phylogenetic analysis supports Moore's (1957)
suggestion of a size reduction series among species, although it is not linear. There are some notable changes in leaf size and shape among species. The apex of the leaf can be taken as a reference point for examining these changes (Fig. 1). The proportions of the leaves of R. paiewonskiana and R. elegans are very similar, but the leaves of R. elegans are smaller. This indicates a slight change in leaf ontogeny between the two species leading to a difference in size only. In the next species, R. latisecta, there is a marked change, in both leaf size and shape, indicating a larger ontogenetic change between R. elegans and R. latisecta.
In R. latisecta, compared with R. elegans, the petiole is longer, the rachis shorter, the veins fewer and closer together, and the windows more numerous. The most notable difference is the much longer distal and proximal margins. Also notable is the long vestigial rachis at the apex of the blade found in R. latisecta (but not in R. elegans). These differences between R. elegans and R. latisecta, taken together, suggest that the leaf of R. latisecta represents the leaf of R. elegans in which a major ontogenetic change has led both to the suppression of pinnae development at the apex and base of the blade (hence the vestigial rachis, long petiole, and long distal and proximal margins) and to a compression of pinnae in the central part of the rachis (hence the closer veins). In other words, the leaf of R. latisecta represents only the compressed, central section of the leaf of R. elegans.
The differences between leaves of R. latisecta and R. elegans are associated with little change in plant size between the two species, so the changes may not be attributed to selection on stem size and subsequent allometric constraint. The most important habitat difference between the two species is the large difference in mean elevation, 283 m versus 1121 m, respectively.
Leaves of Reinhardtia gracilis are very similar in shape to those of R. latisecta. In fact, leaves of Honduran gracilis closely approach those of R. latisecta in size. Changes in leaf size and shape among R. gracilis, R. simplex, and R. koschnyana are more gradual, indicating slight ontogenetic changes. There is a gradual transition from longer to shorter petioles, shorter to longer rachises, increasing proximal margin angles, fewer divisions and windows, and a closing of the distal split to give, in R. koschnyana, a simple leaf.
Unlike leaves, the inflorescences of R. latisecta are much larger than those of R. elegans. Whereas leaves have undergone truncated development, leading to a smaller leaf, inflorescences have, independently, undergone extended development leading to a larger inflorescence.
Variation within Reinhardtia gracilis is greater than supposed by Moore (1957)
. Recognition of var. tenuissima is not supported. Moore (1957)
placed considerable reliance on single variables, especially stamen number, in his treatment of R. gracilis. With many more specimens available, stamen number is known to be more variable than previously thought. Consideration of several variables simultaneously gives a different resolution to the R. gracilis complex. Seven groups may be recognized as distinct: (1) Mexican gracilior; (2) Honduran gracilior; (3) Nicaraguan gracilior; (4) Belize/Guatemalan gracilis; (5) Honduran gracilis; (6) normal-fruited rostrata; (7) large-fruited rostrata. The small sample size of some of these precludes analysis of all groups together, and larger sample sizes are needed to further test this classification.
|
APPENDIX 1 |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
|
|
|
FOOTNOTES |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1LITERATURE CITED |
|---|
Burret M. 1932 Die Palmengattung Reinhardtia Liebm. und Malortiea H. Wendl. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 11: 551-556
Coates A. J. Obando 1996 The geologic evolution of the Central American isthmus. In J. Jackson, A. Budd, and A. Coates [eds.], Evolution and environment in tropical America, 21–56. The University of Chicago Press, Chicago, Illinois, USA
Donnelly T. 1992 Geological setting and tectonic history of Mesoamerica. In D. Quintero and A. Aiello [eds.], Insects of Panama and Mesoamerica, 1–13. Oxford University Press, Oxford, UK
Goloboff P. 1993 Nona. Software and documentation by the author. Tucumán, Argentina
Graham A. 1992 Utilization of the isthmian land bridge during the Cenozoic—paleobotanical evidence for timing, and the selective influence of altitudes and climate. Review of Palaeobotany and Palynology 72: 119-128[CrossRef]
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. Eighth edition. Regnum Vegetabile 120: 1-693
Iturralde-Vinent M. R. MacPhee 1999 Paleogeography of the Caribbean region: implications for Cenozoic biogeography. Bulletin of the American Museum of Natural History 238: 1-95
Lewis C. J. Doyle 2001 Phylogenetic utility of the nuclear gene malate synthase in the palm family (Arecaceae). Molecular Phylogenetics and Evolution 19: 409-420[CrossRef][ISI][Medline]
Lewis C. J. Doyle In press A phylogenetic analysis of tribe Areceae (Arecaceae) using two low-copy nuclear genes. Plant Systematics and Evolution
Moore H. 1957 Reinhardtia. Gentes Herbarum 8: 541-576
Nixon K. 1999 Winclada (BETA) ver. 0.9.99. Published by the author, Ithaca, New York, USA
Read R. T. Zanoni M. Mejía 1987 Reinhardtia paiewonskiana (Palmae), a new species for the West Indies. Brittonia 39: 20-25[CrossRef][ISI]
Rohlf F. 2000 NTSYS. Numerical taxonomy and multivariate analysis system. Version 2.1. Exeter Software, New York, New York, USA
Sneath P. R. Sokal 1973 Numerical taxonomy. W. H. Freeman, San Francisco, California, 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
This article has been cited by other articles:
|
|
 |
|
  A. J. Henderson A multivariate analysis of Hyospathe (Palmae) Am. J. Botany, June 1, 2004; 91(6): 953 - 965. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
