| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
2Department of Organismic and Evolutionary Biology, Harvard University Herbaria, 22 Divinity Avenue, Cambridge, Massachusetts 02138 USA; 3University of Michigan Herbarium, North University Building, Ann Arbor, Michigan 48109-1057 USA; 4Department of Ecology and Evolutionary Biology, Yale University, P.O. Box 208106, New Haven, Connecticut 06520 USA
Received for publication November 7, 2000. Accepted for publication March 15, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1250 species of predominantly New World tropical flowering plants. Infrafamilial classification has long been based on fruit characters. Phylogenetic analyses of chloroplast DNA nucleotide sequences were analyzed to help resolve the phylogeny of Malpighiaceae. A total of 79 species, representing 58 of the 65 currently recognized genera, were studied. The 3' region of the gene ndhF was sequenced for 77 species and the noncoding intergenic spacer region trnL-F was sequenced for 65 species; both sequences were obtained for the outgroup, Humiria (Humiriaceae). Phylogenetic relationships inferred from these data sets are largely congruent with one another and with results from combined analyses. The family is divided into two major clades, recognized here as the subfamilies Byrsonimoideae (New World only) and Malpighioideae (New World and Old World). Niedenzu's tribes are all polyphyletic, suggesting extensive convergence on similar fruit types; only de Jussieu's tribe Gaudichaudieae and Anderson's tribes Acmanthereae and Galphimieae are monophyletic. Fleshy fruits evolved three times in the family and bristly fruits at least three times. Among the wing-fruited vines, which constitute more than half the diversity in the family, genera with dorsal-winged samaras are fairly well resolved, while the resolution of taxa with lateral-winged samaras is poor. The trees suggest a shift from radially symmetrical pollen arrangement to globally symmetrical pollen at the base of one of the clades within the Malpighioideae. The Old World taxa fall into at least six and as many as nine clades.
Key Words: biogeography • fruit evolution • Malpighiaceae • molecular • ndhF • phylogeny • systematics • trnL-F
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1250 species in 65 genera, with 100 species belonging to the 15 Old World genera (W. Anderson, unpublished data). The only overlap between the Malpighiaceae in the two hemispheres consists of two species of predominantly New World genera that also occur in west Africa (Anderson, 1990a
).
The monophyly of the Malpighiaceae has recently been confirmed using molecular data (Chase et al., 1993
; Cameron, Chase, and Anderson, 1995
; Wurdack and Chase, 1996
) and is also supported by morphological characters (Anderson, 1979a, 1990a
). Floral morphology in neotropical Malpighiaceae is distinctive and highly conserved (Anderson, 1979a
; see Fig. 1). In contrast, fruits of Malpighiaceae are extremely diverse, providing the most important characters for delimiting genera. Niedenzu (1928)
recognized five tribes in two subfamilies, placing all genera with unwinged fruits in his wholly New World subfamily Planitorae, with two tribes separated on the dehiscence or nondehiscence of the fruits. His other subfamily, Pyramidotorae, comprised three tribes, one with bristly fruits (New World), one with lateral-winged samaras (with two subtribes, one New World, the other Old World), and one with dorsally winged samaras (also with two subtribes, one New World, the other Old World). Subsequent classifications of Malpighiaceae have deviated little from Niedenzu's treatment. For example, Hutchinson (1967)
recognized five tribes, three identical to Niedenzu's, one equivalent to Niedenzu's Planitorae (Malpighieae), and a fifth (Gaudichaudieae) resurrected from de Jussieu's 1843
monograph. Takhtajan's (1997)
classification was nearly identical to the one proposed by Hutchinson aside from changes in taxonomic rank.
|
The phylogenetic analyses of Malpighiaceae reported here were based on nucleotide sequences of the chloroplast gene ndhF and the noncoding trnL-F region. ndhF encodes a subunit of the nicotinamide dehydrogenase complex and shows approximately twice the average mutation rate of rbcL (Sugiura, 1989
; Olmstead and Sweere, 1994
). It has been used successfully to infer phylogenetic relationships within and among angiosperm families (Olmstead et al., 1992
; Olmstead and Sweere, 1994
; Clark, Zhang, and Wendel, 1995
; Kim and Jansen, 1995
; Neyland and Urbatsch, 1995, 1996
; Olmstead and Reeves, 1995
; Scotland et al., 1995
; Wagstaff et al., 1998
; Alverson et al., 1999
; Ferguson, 1999
). The noncoding trnL-F region displays a high frequency of mutations, but has also been used in assessing phylogenetic relationships within traditional angiosperm families (Palmer et al., 1988
; Clegg, Learn, and Golenberg, 1991
; Soltis and Soltis, 1998
).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Ferguson, 1999
) and 65 new trnL-F sequences for 79 accessions of Malpighiaceae (Table 1) representing 58 of the 65 genera currently recognized by W. Anderson. Broader phylogenetic analyses based on several genes (Wurdack and Chase, 1996
; K. Wurdack, University of North Carolina, personal communication) indicate that the New World genus Humiria (Humiriaceae) is relatively closely related to Malpighiaceae; Humiria was, therefore, included in our analyses for rooting purposes.
|
, with a hot cetyltrimethylammonium bromide (CTAB) extraction method (Doyle and Doyle, 1987
) used most frequently. All extractions were purified on ethidium bromide/CsCl gradients. The ndhF primers were designed from Malvales by R. Nyffeler (Harvard University, unpublished data, personal communication), and trnL-F intergenic spacer (IGS) primers were designed from tobacco, rice, and Marchantia by Taberlet et al. (1991)
. 800 base pairs (bp) of the 3' end of ndhF was amplified using primers 5.5F and 10.2R, which reside between primers 5 and 6 and 10 and 11 (Olmstead and Sweere, 1994
; Table 2), respectively. The trnL-F region was amplified using primers trnC and trnF (Table 2). The trnL-F region we amplified corresponds to two of the three noncoding spacer regions examined by Taberlet et al. (1991)
: the trnL intron and an intergenic spacer between the trnL 3' exon and trnF.
|
; Table 2). Sequences were obtained using a Model 377 automated sequencer (Perkin Elmer) and chromatograms were assembled into contiguous sequences and checked for accuracy using the software program Sequencher 3.0 (Gene Codes Corporation, Ann Arbor, Michigan, USA). Both strands were sequenced with a minimum overlap of 80% of the total sequence length. All newly generated sequences were submitted to GenBank (Table 1).
Phylogenetic analyses
Nucleotide and amino acid sequences were aligned by eye, and primer sites were trimmed from the data sets. Individual and combined parsimony analyses were conducted using PAUP* 4.0 (Swofford, 1999
). An initial heuristic search of 100 random taxon addition replicates was conducted with tree-bisection-reconnection (TBR) branch swapping and MULPARS in effect, but retaining only ten trees after each replicate. The resulting consensus tree was then used as a backbone constraint to search for trees not consistent with the initial trees. This method was employed due to the excessive number of trees generated for unconstrained heuristic searches. This search strategy should detect that there are no shorter trees and that the strict consensus tree reflects all of the most parsimonious trees (Catalán, Kellogg, and Olmstead, 1997
). Searches using the combined trnL-F and ndhF data were conducted the same as above using both a reduced taxon data set, which included only those taxa sampled for both gene sequences, and an expanded taxon data set, including all taxa sampled for either gene. As an additional assessment of our search strategy we implemented the parsimony ratchet (Nixon, 1999) using PAUPRat (Sikes and Lewis, University of Connecticut, unpublished computer program). For each of the four data sets we implemented five searches using 200 iterations for each search and reweighting was conducted for 15% of the available characters. Bootstrap support (Felsenstein, 1985
) was estimated based on 1000 replicates using simple taxon addition, TBR, MULPARS, and holding ten trees at each replicate.
The "Templeton test" as implemented in PAUP* (Swofford, 1999
) was used to assess support for clades appearing in the most parsimonious trees (Templeton, 1983
; see also Larson, 1994
). The shortest trees in which the monophyly of a clade of interest was not supported were used as a constraint and compared to the shortest trees in which the clade of interest was present. For each constraint search we used only the reduced taxon sampling. Monophyly constraints of interest were a monophyletic Byrsonimoideae, an Old World hiraeoid clade, an Old World banisterioid clade, and three independent constraints for the monophyly of three traditionally recognized genera: Aspicarpa, Janusia, and Mascagnia. In order to assess congruence between the two independent data sets we conducted the incongruence length difference (ILD) test (Farris et al., 1994
), implemented as the partition homogeneity test in PAUP* 4.0 (see Mason-Gamer and Kellogg, 1996
). We used the heuristic search strategy presented above with 999 random repetitions to generate the distribution. As a conservative estimate of significance we present only the highest P values for each assessment of congruence.
We have overlaid Hutchinson's 1967
tribal classification of Malpighiaceae onto the consensus tree based on the combined expanded analysis using parsimony. Fruit type for each species was scored and treated as an unordered multistate character (Table 1) and mapped onto one of the most parsimonious trees from the combined expanded data matrix using MacClade version 3.0 (Maddison and Maddison, 1992
). The tree was selected at random among those trees that were most parsimonious with respect to the character of interest. Similarly, we recorded the geographic distribution of each taxon as New World or Old World (Table 1) and mapped that information onto one of the most parsimonious trees from the combined expanded data matrix.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Individual trnL-F sequences within Malpighiaceae ranged in length from 1028 bp (Blepharandra heteropetala) to 709 bp (Ptilochaeta bahiensis), and the outgroup sequence (Humiria) was 943 bp. Length variation was mainly attributable to several A–T rich regions that could not be aligned readily due to repeated motifs and numerous indels. These hypervariable regions, totaling 180 bp, were excluded from the analyses. Over the entire alignment we identified 113 indels located in the trnL-F region that were scored and added to the matrix as unordered gap characters. This coding was done using a program developed and distributed by R. Ree (Harvard University, personal communication). After exclusion of the hypervariable regions and inclusion of the gap characters, the aligned matrix was 1219 bp in length. The longest ingroup distance was between Caucanthus auriculatus and Diacidia ferruginea (88 steps; 7.22%). The longest overall distance was 150 steps (12.3%) between Caucanthus auriculatus and Humiria balsamifera (the outgroup). Of the 481 variable nucleotide positions within the ingroup, 198 were parsimony informative.
The combined data matrix was 1987 bp in length. The reduced data set contained 63 ingroup species from which complete sequences were obtained from both gene regions (see Table 1). Of the 777 variable nucleotide positions within the ingroup, 345 were parsimony informative. The expanded data set contained 79 ingroup species for which each species was sequenced for at least ndhF or trnL-F (Table 1). Of the 796 variable nucleotide positions within the ingroup, 370 were parsimony informative.
Phylogenetic analyses
Parsimony searches on the ndhF data resulted in 990 trees of 652 steps (Fig. 2). Searches using the trnL-F data resulted in 840 trees of 750 steps (Fig. 3). Searches on the combined reduced taxon data set including trnL-F gap coding resulted in 760 trees of 1328 steps (Fig. 4). Searches on the combined expanded taxon data set resulted in 1000 trees of 1408 steps (Fig. 5). In all cases, use of the consensus backbone constraint trees failed to find any trees of equal or less length that contradicted the respective consensus trees. Similarly, trees generated from PAUPRat were identical to those presented here.
|
|
|
|
Malpighiaceae are split into two major clades (Fig. 5), which we will refer to by the subfamilial names, Byrsonimoideae and Malpighioideae (100 and <50% bootstrap values, respectively). The Byrsonimoideae include the byrsonimoid (99%) and acmantheroid (100%) clades. The Malpighioideae comprise the bulk of diversity within Malpighiaceae and include the galphimioid (100%), acridocarpoid (100%), mcvaughioid (100%), Barnebya, and the banisterioid (100%) clades. The galphimioids are weakly placed (<50%) as sister to the rest of the Malpighioideae in the combined analyses. In trnL-F trees, the galphimioids form a weakly supported clade with Byrsonimoideae, whereas their position is unresolved by ndhF. The remainder of the Malpighioideae form a strongly supported clade (100%) with the acridocarpoids basal in this clade. The mcvaughioids are placed with moderate support (70%) as sister to the remaining Malpighioideae. The enigmatic South American genus Barnebya is weakly placed (58%) as sister to the banisterioids. This last result is also found with ndhF (53%), but is contradicted in the trnL-F trees, which find Barnebya sister (71%) to the mcvaughioids, that clade forming a clade with the acridocarpoids (<50%). Sister to the remainder of the banisterioids is a well-supported (85%) ptilochaetoid clade with two subclades, one of the Mexican genus Lasiocarpus and the southern South American genus Ptilochaeta (98%), the other containing the Chilean endemics Dinemagonum and Dinemandra (100%). The tristellateioids (<50%) are placed as sister to a large clade (83%; not supported by ndhF) containing most of the wing-fruited genera. The tristellateioid clade is poorly supported (i.e., <50%) in the ndhF trees and is not detected in the trnL-F trees. Included in the tristellateioids is a clade containing Tristellateia and Heladena (100%), which is sister to a poorly supported (<50%) clade containing Bunchosia, Thryallis, and Echinopterys. The latter clade is not present in the trnL-F trees and is poorly (<50%) supported in the ndhF trees. The rest of the banisterioid clade contains six major lineages: malpighioids, stigmaphylloids, tetrapteroids, hiraeoids, and the genera Ectopopterys and Lophopterys. The tetrapteroids are weakly supported in all analyses and are not detected in the trnL-F trees. In contrast, the stigmaphylloids form a strongly (100%) supported clade that contains the gaudichaudioid (99%) and cordobioid (100%) clades. The hiraeoid clade is not recovered in the trnL-F trees, but is moderately to poorly supported (<50% ndhF, 75% reduced, and 55% expanded) in the ndhF and combined analyses. Lophopterys is very weakly (<50%) placed as sister to the malpighioid clade, and Ectopopterys is placed but with little support in a tetrachotomy with the hiraeoids, the tetrapteroids, and the clade containing the malpighioids plus the stigmaphylloids.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
described the subfamily Byrsonimoideae, he subjected the unwinged fruit types of the Planitorae/Malpighieae to critical evaluation and broadened the taxonomic base to include characters other than those of the fruit. He justified the exclusion of several taxa from the Planitorae on the grounds that some smooth-walled fruit types represented examples of convergence and were likely derived from wing-fruited ancestors. For example, although the fruits of Byrsonima, Bunchosia, and Malpighia are all indehiscent and have a fleshy exocarp, they are only superficially similar. Fruits of Byrsonima contain a single bony stone, which results from fusion of the three endocarps. In Bunchosia each of the two or three carpels contributes a smooth, cartilaginous, seed-containing endocarp, but these are free from each other. Malpighia fruits possess fibrous stones that are connate only along a central axis, and the individual stones bear ridges or crests in the position of the wings seen in wing-fruited Malpighiaceae. Aside from Bunchosia and Malpighia, Anderson (1978)
also argued that Dicella, Heladena, and Thryallis should be excluded from Byrsonimoideae. In our analyses all five of these genera are clearly removed from the Byrsonimoideae. Anderson's tribes Acmanthereae (Acmanthera, Coleostachys, and Pterandra) and Galphimieae (Galphimia, Lophanthera, and Spachea) are monophyletic in our results. However, his tribe Byrsonimeae (including Blepharandra, Byrsonima, Diacidia, Burdachia, Glandonia, and the subsequently described Mcvaughia) is not monophyletic and the genera are divided between the byrsonimoid and mcvaughioid clades. The monophyly of Anderson's Byrsonimoideae is strongly rejected (Templeton test P = 0.0012). One perplexing outcome from our study is the phylogenetic placement of the galphimioids. Our trnL-F data detected a Byrsonimoideae that included the galphimioids (Fig. 3), and these results were duplicated with better support using matK and rbcL (Cameron et al., 2001
), but our combined analyses placed the galphimioids (without strong support) as sister to the rest of the Malpighioideae. Given the weak support for this relationship, and the evidence from chromosome numbers (see below), we expect that in future analyses the galphimioids will group with the Byrsonimoideae.
|
). When he summarized those and additional chromosome numbers (Anderson, 1993
), all of the counts for the genera retained in the Byrsonimoideae as recognized here, and for the galphimioids, were still n = 6, 12, or 24. Therefore, x = 6 does appear to be a consistent character of the subfamily in our restricted sense, but given the structure of the tree (Fig. 5), we cannot say whether that is an ancestral or a derived condition. Mcvaughia, described after 1978 but assigned to Byrsonimoideae close to Burdachia by Anderson (1979b)
, has n = 10, which is consistent with its placement here in the Malpighioideae. It would be especially interesting to know the chromosome numbers of Burdachia and Glandonia; we would expect them to have numbers based on x = 10. Barnebya has n = 30, a number that could be a multiple of either 6 or 10 (Anderson, 1993
). Given its phylogenetic placement with other Malpighioideae it most likely represents a multiple of 10. The counts known for Acridocarpus (Mangenot and Mangenot, 1958, 1962
; Carr and McPherson, 1986
; Paiva and Leitão, 1987
) indicate that its base number is x = 9, readily derived from n = 10 through aneuploidy. Thus, the available counts and the results of this study suggest that x = 10 may well be ancestral in our subfamily Malpighioideae, above the galphimioids.
Malpighioideae
Acridocarpoids
The acridocarpoids (Fig. 5) comprise Acridocarpus and Brachylophon. Morphology supports the sister relationship of Acridocarpus and Brachylophon. Acridocarpus has a schizocarpic fruit with a large dorsal wing on each samara and no trace of lateral crests; Brachylophon has a similar fruit except that the dorsal wing has been reduced to a crest. In both genera the anthers are poricidal, a rare condition in the family, and both have subulate styles with tiny terminal stigmas, like those found in most Byrsonimoideae and the mcvaughioids (W. Anderson, personal observation). The pollen in both is tricolporate, but in Acridocarpus it is derived in that the colpi fuse at the poles (syncolporate) (Lobreau, 1967, 1968
; Lowrie, 1982
).
Mcvaughioids
The mcvaughioid clade is a well-supported (100%) lineage containing Mcvaughia, Burdachia, and Glandonia. These three genera have unwinged, indehiscent fruits. If the ancestor of the Malpighioideae (above the galphimioids) had dry schizocarpic fruits with wings (e.g., Acridocarpus), the fruits of Mcvaughia, Burdachia, and Glandonia must be derived in their indehiscent, unwinged fruits. There is little in the structure of most of those fruits to signal derivation from a schizocarpic, wing-fruited ancestor, but it is worth noting that the fruit of Burdachia prismatocarpa Adr. Juss. bears 8–9 longitudinal aerenchymatous ribs or winglets in the approximate position of the lateral and dorsal wings on many fruits in the banisterioid clade (see Anderson, 1981
).
The pollen grains in the Byrsonimoideae, the galphimioids, the acridocarpoids, and the mcvaughioids are mostly tricolporate or readily derived from a tricolporate ancestor (Anderson, 1978, 1979b
; Lowrie, 1982
). Tricolporate pollen, which is common in other rosid families, is presumably plesiomorphic in the Malpighiaceae (Anderson, 1990a
).
Barnebya
Barnebya is sister to the banisterioids and has a schizocarpic fruit, with each mericarp bearing a large dorsal wing. As Anderson and Gates pointed out when they described Barnebya (1981), its samara is very similar to the samara of Acridocarpus, and primarily on that basis, they suggested that Barnebya might be close to the ancestor of Acridocarpus, which is relatively derived in many of its characters. Our data do not support that suggestion (Fig. 5). If this study places Barnebya correctly, it is clearly derived in the characters emphasized by Anderson and Gates (1981)
, such as the alternate estipulate leaves, the perigynous hypanthium, and the samaras attached at maturity to the receptacle by two coriaceous marginal ribs. The pollen of Barnebya is different from that of most basal malpighs in that it lacks ectoapertures and has other anomalies, such that Lowrie (1982)
doubted its placement with the other genera in the mcvaughioid clade or with Acridocarpus.
Banisterioids
Ptilochaetoids
Sister to the rest of the banisterioids is the clade we have called the ptilochaetoids, comprising the three South American genera Dinemandra, Dinemagonum, and Ptilochaeta plus the Mexican genus Lasiocarpus. Those genera fall into two well-supported clades; each of those clades is internally coherent in its gross morphology, but the two clades together are quite unlike each other in most characters, including their fruits (Niedenzu, 1928
; Simpson, 1989
; W. Anderson, personal observation). It is especially intriguing that Lowrie (1982)
found them to share a similar polycolporate pollen morphology, which is so peculiar in the family that he put all four of these genera together in his "Lasiocarpus group" and suggested that they were allied phylogenetically to the byrsonimoid groups with tricolporate pollen. The ptilochaetoid clade is a striking example of a grouping supported by both DNA sequences and pollen in spite of other morphological differences.
The rest of the banisterioids are a strongly supported clade that contains most of the genera and species of the Malpighiaceae. Pollen morphology supports the clade—these banisterioids have globally symmetrical pollen grains, with the pores not all in the same plane and the ectoapertures, if present, variously oriented (Fig. 5). All nonbanisterioids plus the ptilochaetoids have radially symmetrical pollen, i.e., the pores are equatorial and the colpi, if present, are oriented at right angles to the equator (Lowrie, 1982
). Lobreau (1967)
described Rhynchophora humbertii as having tricolporate pollen, but that is probably not the case; see comments below under Rhynchophora.
Tristellateioids
The tristellateioids are a miscellany of morphologically diverse and isolated genera; several of its branches are poorly supported and may not persist in future analyses. However, the very strong support for the association of Heladena and Tristellateia is perplexing. Heladena is a South American genus that seems quite without morphological synapomorphies with the Old World genus Tristellateia (W. Anderson, personal observation). Lowrie (1982)
did not consider their pollen grains to be especially similar, although both are polyporate.
Aside from the ptilochaetoids and tristellateioids, the remainder of the banisterioids fall into a well-supported but poorly resolved clade. Because of the distribution of its representatives in that clade the genus Mascagnia merits special comment. It has already been noted that Mascagnia contains diverse species with lateral-winged samaras that do not fit readily into segregate genera (Anderson, 1981, 1990b
). In Fig. 5 the seven species sampled are scattered among at least four clades, showing that Mascagnia is indeed nonmonophyletic. The monophyly of Mascagnia is rejected by a Templeton test (P = 0.0253).
Malpighioids
One of the two well-supported large clades in the banisterioids is the one we are calling malpighioids. That clade comprises the genus Malpighia, our only representative of Mascagnia section Mascagnia (the probably monophyletic part of Mascagnia that includes the type of the genus), the West Indian genus Triopterys, and the Old World genera Aspidopterys, Caucanthus, Rhynchophora, and Triaspis. Rhynchophora is a special case that will be discussed separately below. Except for Malpighia, which has a fleshy fruit, the other genera have samaras with the lateral wings dominant and continuous at the base (as do several genera in other clades). W. Anderson has suggested (1987, 1990b) that, given its other morphological characteristics and the rudimentary winglets on the endocarp concealed by the fleshy exocarp, Malpighia is likely to have had its origin in or near Mascagnia sect. Mascagnia, so it is not surprising to see those two taxa together in this clade. The distribution of New World and Old World genera within the clade is not what we would have predicted, but those branches are weakly supported. The pollen is informative in this case. Lowrie (1982)
considered the pollen of Malpighia and Mascagnia section Mascagnia to be a derived type that he called "mascagnioid," and he called that of Triopterys "near mascagnioid," saying it could easily be derived from a mascagnioid grain. Those data suggest that we should expect the four American taxa in this clade to come together in future analyses. Lowrie found the pollen grains of the Old World genera Aspidopterys and Triaspis to be very similar to each other (his "aspidopteroid" type), suggesting a close relationship of these genera. Caucanthus auriculatus has problematic pollen. Lowrie found it to fit his tetrapteroid type, quite different from that of C. albidus Nied., which has aspidopteroid pollen. While that does not help us to interpret this clade, it is worth noting that Caucanthus auriculatus also has an indehiscent fruit (see discussion of fruit evolution); with anomalous fruits and pollen, C. auriculatus clearly merits closer study.
Rhynchophora is a genus of Madagascar with only two species. Its peculiar fruit is indehiscent, and each carpel bears a single horizontal to ascending wing. In R. phillipsonii the wings are borne around the apex of the fruit and give the whole fruit the appearance of a helicopter. The homology of that unique wing is not entirely obvious, but it seems most likely to be derived from a displaced lateral wing (W. Anderson, 2001). Lobreau (1967, 1968)
initially described the pollen of R. humbertii as tricolporate, later as inaperturate. We have not had the opportunity to study pollen of that species, but W. Anderson (2001) reports that in R. phillipsonii the pollen of morphologically bisexual flowers is inaperturate (and presumably nonfunctional), while that of staminate flowers is globally symmetrical, with (5) 6 (7) nonequatorial pores, and lacks ectoapertures. That pollen resembles what Lowrie (1982)
called his aspidopteroid type, suggesting that Rhynchophora is quite at home with the other Old World genera in the malpighioids.
Lophopterys is placed as sister to the malpighioids, but with low bootstrap support. Lophopterys is a South American genus with two long narrow lateral wings on each samara and a triangular dorsal winglet (except for the type species, which has lost the lateral wings). Aside from its fruit, the genus is distinguished by having the paired calyx glands of other neotropical genera replaced by a single large gland, a condition found otherwise only in some species of Jubelina and Mezia (W. Anderson, personal observation), both of which fall into the unresolved mass of tetrapteroids in our analysis.
Stigmaphylloids
The other large and strongly supported clade in the banisterioids is the stigmaphylloids. The following comments will deal with all the genera except Gallardoa, which is discussed separately. Most genera in the clade have apical (terminal) stigmas vs. stigmas on the internal angle of the apex of the style, which is the condition found in Ectopopterys and most of the tetrapteroids and hiraeoids, the sisters to the stigmaphylloids. Cordobia has internal stigmas; given its other character states and its position in Fig. 5, the condition is probably secondary. The other genus in the stigmaphylloids with internal stigmas is Stigmaphyllon. Given its position near the base of this unresolved clade, one cannot help speculating that Stigmaphyllon may eventually prove to be sister to the rest of the clade, with terminal stigmas derived in the other genera, including its Old World sister, Ryssopterys. It is also worth noting that the enigmatic genus Ectopopterys has folioles on the anterior style, similar to those found in Stigmaphyllon (and nowhere else in the family). In other characters those two genera are very dissimilar (Anderson, 1980b
), but the placement of Ectopopterys in a tetrachotomy with the stigmaphylloids in Fig. 5 offers the interesting possibility that they may have shared a common ancestor with folioles.
Most genera of the stigmaphylloids have mericarps with the dorsal wing dominant and thickened along the adaxial edge; such a fruit occurs only in this clade and in the acridocarpoids (Anderson, personal observation). In many cases where the dorsal wing has been reduced (Aspicarpa, Diplopterys, and Mionandra), the derivation from a dorsal wing is obvious. In Peregrina, and even more strongly in Gaudichaudia, the dorsal wing has been reduced while the lateral wing has become dominant, surely a secondary phenomenon. It is also worth noting that most genera in this clade have mericarps that are suspended from the torus by a carpophore (Hutchinson, 1967
), except for Aspicarpa and Diplopterys, both of which have lost the dorsal wing, so that the presumed adaptive advantage of having a samara dangle from a cartilaginous thread until the wind is strong enough to break it and carry the samara away is no longer relevant. That carpophore is known in no other clade of the family (W. Anderson, personal observation). Most members of this clade, and only those species, have a derived pollen type that Lowrie (1982)
called "banisterioid," and the "Banisteriopsis group" in his classification of the family based on pollen agrees perfectly with the stigmaphylloids in Fig. 5.
Gallardoa is the one somewhat discordant element in the stigmaphylloids. This monotypic genus is endemic to an area in Argentina farther south than that of any other neotropical member of the family. Its samara has narrow dorsal and lateral wings with the lateral wing slightly larger, and lacks a carpophore. Its stigmas are internal. All of these characteristics would suggest an origin in the tetrapteroids or malpighioids. On the other hand, Gallardoa has its stipules connate in interpetiolar pairs and the stamens opposite the sepals reduced to staminodes, just as in the other Argentinian endemics Cordobia and Mionandra, and as in Peixotoa, which may be sister to the cordobioids (W. Anderson, personal observation). Lowrie (1982)
found its pollen to be unusual but most likely derived from that of Cordobia and Mionandra. Therefore, in spite of its anomalies, the morphology of Gallardoa is consistent with its placement in Fig. 5.
Gaudichaudioids
Embedded within the stigmaphylloids is the strongly supported clade we are calling the gaudichaudioids. That clade corresponds to Hutchinson's tribe Gaudichaudieae (1967), which is the only monophyletic tribe in his classification of the family (Fig. 6). (As an historical note, Hutchinson was not the first to recognize the tribe Gaudichaudieae. Adrien de Jussieu did so in the first monograph of the family [1843, p. 588, pl. 23]. He used that name and the same characteristics mentioned by Hutchinson [reduced androecium and style number, distinct carpels, carpophore, dimorphic flowers], and his work was undoubtedly the source of Hutchinson's tribe, even though Hutchinson did not cite de Jussieu.) That is also the only tribe in Hutchinson's system that was not based primarily on fruit type, which demonstrates how misleading such overreliance on one set of morphological characters has been in this family. In this case Hutchinson focused mainly on shared floral characteristics, specifically apocarpous gynoecia and the production in many of what he called dimorphic flowers, which were subsequently shown by W. Anderson (1980a)
to be chasmogamous and cleistogamous flowers. Additionally, this clade exhibits a trend toward herbaceous habit; its species are among the least woody members of the family (W. Anderson, personal observation). Within the gaudichaudioid clade Janusia and Aspicarpa are probably not monophyletic as traditionally circumscribed. A monophyletic Janusia is not rejected (Templeton test P = 0.0578), but in all analyses Janusia anisandra and J. mediterranea of South America are not directly linked with Janusia californica and J. linearis of northwestern Mexico. A monophyletic Aspicarpa is strongly rejected (Templeton test, P = 0.0028). The Mexican species of Aspicarpa (A. brevipes and A. hirtella) are united with the Mexican species of Gaudichaudia, rather than with the South American A. pulchella. These results call for a reexamination of the generic taxonomy of the gaudichaudioids.
Hiraeoids/Tetrapteroids
The remaining taxa included in our analysis are in the hiraeoid and tetrapteroid "clades," both of which are poorly supported with little or no resolution. Given their weakness, neither group merits detailed comment at this time, but a few points are worth mentioning. The fairly strong support for a clade comprising the representatives of the two major groups in the genus Hiraea was to be expected because that genus is coherent in its morphology, with all its species sharing several synapomorphies—elongated stipules borne on the petiole, short-stalked axillary umbels, and butterfly-shaped samaras. Excluding the pair of Dicella and Tricomaria for the moment, the rest of the tetrapteroids and hiraeoids have the lateral wing(s) of the samara dominant and the dorsal wing more or less reduced, with the exception of Heteropterys, one of the two largest genera in the family. In the latter the dorsal wing is dominant and the lateral wings are reduced to crests or often lost. That dorsal wing is thickened on the abaxial edge and bends upward, exactly the opposite of what is found in the stigmaphylloids. That samara type has always made Heteropterys inconsistent in its placement with Stigmaphyllon and its relatives, where it was placed by Niedenzu (1928)
and Hutchinson (1967)
. Its internal stigmas make it difficult to distinguish Heteropterys from Tetrapterys in flower (W. Anderson, personal observation), and its pollen also favors a tetrapteroid affinity (Lowrie, 1982
). The placement of Heteropterys with this group of genera is therefore not a surprise, considering all the morphological evidence and setting aside the radical shift in samara type.
The lateral-winged tetrapteroids and hiraeoids are very diverse in their morphology, even more diverse than the number of generic names in Fig. 5 suggests, because those four species of Mascagnia probably do not constitute a clade, and Tetrapterys microphylla fits poorly in a genus with Tetrapterys discolor (W. Anderson, personal observation). In fact, when we consider the weak support for all the clades with lateral-winged samaras (excepting the gaudichaudioid genera Gaudichaudia and Peregrina and the malpighioids), we can only conclude that more and better data are needed before we can say anything with confidence.
The neotropical genera Dicella and Tricomaria are paired with fairly strong support in Fig. 5. Dicella is a small Central and South American genus of woody vines in which the fruit proper is a large, indehiscent, smooth-walled nut. Dicella appears to have transferred the wind dispersal function of wings on the fruit to an accrescent calyx of five winglike sepals, which subtend the fruit proper. Tricomaria is a monotypic shrub of very dry areas in Argentina; its fruit (apparently also indehiscent) is covered by long, stiff, vascularized bristles. The two genera have a generally similar pollen type that is widespread among tetrapteroids, although Tricomaria is derived in having diorate colpi (Lowrie, 1982
). Both genera have the petals abaxially sericeous. Aside from the indehiscent fruit and the hairy petals, there are no obvious morphological synapomorphies uniting those two genera (W. Anderson, personal observation).
Evolution of fruits
We have mapped basic fruit types onto one of the most parsimonious trees used to generate Fig. 5 (Fig. 7). Fruits have long been a major criterion for recognition of genera in the Malpighiaceae, with floral and vegetative characters contributing additional (in some cases primary) characters. With a few exceptions, most notably Mascagnia, those genera are supported in our analyses, i.e., where we included two or three species of the same genus, they formed a clade (Acridocarpus, Gaudichaudia, Hiraea, Malpighia, Ptilochaeta, Stigmaphyllon, and Tristellateia). It is also the case that similar fruits often characterize clades of two or more genera; examples are Blepharandra/Diacidia, the acmantheroids, the galphimioids, Acridocarpus/Brachylophon, Ryssopterys/Stigmaphyllon, the hiraeoids, Lasiocarpus/Ptilochaeta, and Burdachia/Glandonia. However, patterns of fruit evolution in the Malpighiaceae are not entirely simple. They have caused problems for past efforts to construct an infrafamilial classification and continue to present challenges today. Several such cases were mentioned above; others will be discussed here in terms of trends in the evolution of fruit types.
|
; W. Anderson, personal observation). The most puzzling genus among the banisterioids, with respect to its wingless fruits, is Bunchosia. In that genus the cartilaginous endocarps, buried in flesh, are quite smooth and show no sign of a winged ancestry. Given the placement of Bunchosia in our tree, embedded in a clade in which most fruits are bristly or winged, it is very probably derived in its winglessness; perhaps developmental-anatomical study of its fruits would shed light on this problem. In summary, the earliest Malpighiaceae may have had unwinged fruits, with winged fruits evolving at or very near the base of the Malpighioideae. Within Malpighioideae wings have apparently been lost in several lineages.
Although we remain uncertain about the ancestral condition in Malpighiaceae, it seems clear that fleshy, presumably bird-dispersed fruits evolved three times in the family, in Byrsonima, Malpighia, and Bunchosia, all of which are nested in dry-fruited clades. This was suggested by Anderson (1978)
on the basis of morphology.
Niedenzu (1928)
and Hutchinson (1967)
recognized the tribe Tricomarieae to accommodate five neotropical genera with fruits that bear many long vascularized bristles scattered over the surface, even though those genera are not especially similar in other aspects of morphology (W. Anderson, personal observation). We were able to include four of those five genera in our study, and those genera (Tricomaria, Echinopterys, Lasiocarpus, and Ptilochaeta) fall into three well-separated clades (Fig. 7), suggesting that bristly fruits evolved at least three times in the family. We were not able to obtain Henleophytum, a monotypic genus of Cuba. Henleophytum shares several morphological synapomorphies with Heladena, Echinopterys, and Bunchosia (W. Anderson, personal observation), so when we are able to add its sequences we expect it to be placed among the tristellateioids, and it may turn out that the bristly fruit evolved only once in that clade.
Indehiscent fruits are scattered through the tree in Fig. 7. We note also that several species of Malpighia not included in our study [e.g., M. albiflora (Cuatrec.) Cuatrec. and M. verruculosa W. R. Anderson] have dehiscent fleshy fruits, and while Caucanthus auriculatus has an indehiscent fruit, C. albidus has a dehiscent fruit (W. Anderson, personal observation).
The traditional criterion for grouping wing-fruited genera into tribes has been whether the dominant wings were dorsal or lateral, and that difference can be very informative, but overreliance on it led Niedenzu (1928)
to erect polyphyletic tribes and to misplace some genera. Most genera with the dorsal wing dominant fall into our stigmaphylloids, but dorsally winged samaras are also found in four other clades, those containing Barnebya, Acridocarpus and Brachylophon, Heteropterys, and Dinemagonum. Niedenzu forced them all into his tribe Banisterieae [the species later segregated as Barnebya dispar was treated as Banisteria dispar (Griseb.) Nied. by Niedenzu]. Dinemagonum was a particularly egregious case, because in all characters except the fruit it is very like its Chilean sister genus, Dinemandra, as recognized by de Jussieu (1843
, p. 585) and Simpson (1989)
. Niedenzu also excluded from his Banisterieae the genera Diplopterys, Gaudichaudia, and Mionandra because they lack a dominant dorsal wing, even though all share morphological synapomorphies with Banisteriopsis, Aspicarpa, and Cordobia, respectively (W. Anderson, personal observation) and are placed in the stigmaphylloids in our analysis. As noted above, de Jussieu (1843)
correctly placed Gaudichaudia near Aspicarpa and Janusia. Loss of the dorsal wing and elaboration of the lateral wings likely happened three times in the stigmaphylloids.
Biogeography
Most genera of Malpighiaceae are limited to the New World or the Old World. The only exceptions are two New World genera with one species each also in coastal western Africa; they are Heteropterys (H. leona; C. Anderson, 2001) and Stigmaphyllon [S. bannisterioides (L.) C. Anderson; Anderson, 1997
]. All of the Old World groups have winged fruits. Niedenzu (1928)
placed those Old World species with the lateral wings dominant in a subtribe Aspidopteryginae within his lateral-winged tribe Hiraeeae, while those with the dorsal wing dominant were placed in subtribe Sphedamnocarpinae within his dorsal-winged tribe Banisterieae. We have plotted geographic distribution on one of the most parsimonious trees from Fig. 5 and added Niedenzu's subtribal assignment for each Old World taxon (except Rhynchophora Arènes, which was described since 1928 and which we assigned to the subtribe in which he surely would have placed it, and Heteropterys leona, which Niedenzu classified with the rest of Heteropterys in his New World subtribe) (Fig. 8). Niedenzu's subtribes are polyphyletic, implying that he relied too heavily on geography in constructing his classification for the family. A monophyletic Old World Banisterieae and Hiraeeae are strongly rejected (Templeton test: P = 0.0001 for both).
|
). A third possibility is that the radiation of Malpighiaceae predates the separation of South America and Africa (cf. Vogel, 1990
). This would require a series of selective extinctions in the two hemispheres, leaving no representatives of any genus from either hemisphere in the other except for the two shared species of Heteropterys and Stigmaphyllon mentioned above. This is possible, of course, and can eventually be tested with information on the absolute age of splitting events in the tree.
The acridocarpoids consist of Acridocarpus, a genus of perhaps 25 species distributed in Africa, Madagascar, the Arabian Peninsula, and New Caledonia, and Brachylophon, a genus of one or several species of the Malay Peninsula (Niedenzu, 1928
). These two genera form an especially interesting Old World clade, because their phylogenetic position provides evidence that some Old World–New World disjunctions involved more ancient lineages and perhaps occurred soon after the evolution of winged fruits.
Conclusions
Our phylogenetic results call into question many previous taxonomic conclusions. Too much reliance has been placed on similar fruit characters, which now appear to be homoplastic, and with respect to the Old World taxa, which have been revealed to represent numerous New World–Old World disjunctions, too much emphasis has been placed on geography. In future studies it will be especially desirable to add the seven genera that are not represented in the present study: from South America, Camarea, Clonodia, and Verrucularia; from Cuba, Henleophytum; from Madagascar, Digoniopterys, Philgamia, and Microsteira. In addition, the results reported here should be compared and probably combined with data from other genes and morphological characters. Nuclear genes may be helpful, especially in view of possible conflicts resulting from lineage sorting and "chloroplast capture." Low-copy-number nuclear genes have provided resolution in several phylogenetic studies (e.g., Mathews and Donoghue, 1999
), and we are now assembling Phytochrome C sequences for the accessions used in the present study. Detailed analyses of fruit morphology and development, especially of seemingly homoplastic features, are also clearly needed to improve our understanding of the evolution of Malpighiaceae.
|
|
FOOTNOTES |
|---|
5 Author for correspondence (e-mail: cdavis{at}oeb.harvard.edu
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Anderson C. 1995 Revision of Thryallis (Malpighiaceae). Contributions from the University of Michigan Herbarium 20: 3-14
Anderson C. 1997 Monograph of Stigmaphyllon (Malpighiaceae). Systematic Botany Monographs 51
Anderson C. 2001. The identity of two water-dispersed species of Heteropterys (Malpighiaceae): H. leona and H. platyptera. Contributions from the University of Michigan Herbarium 23: 35-47
Anderson W. R. 1978 ["1977"] Byrsonimoideae, a new subfamily of the Malpighiaceae. Leandra 7: 5-18
Anderson W. R. 1979a Floral conservatism in neotropical Malpighiaceae. Biotropica 11: 219-223
Anderson W. R. 1979b Mcvaughia, a new genus of Malpighiaceae from Brazil. Taxon 28: 157-161[CrossRef][ISI]
Anderson W. R. 1980a Cryptic self-fertilization in the Malpighiaceae. Science 207: 892-893
Anderson W. R. 1980b Ectopopterys, a new genus of Malpighiaceae from Colombia and Peru. Contributions from the University of Michigan Herbarium 14: 11-15
Anderson W. R. 1981 Malpighiaceae. In The botany of the Guayana Highland—Part XI. Memoirs of the New York Botanical Garden 32: 21-305
Anderson W. R. 1987 Notes on neotropical Malpighiaceae—II. Contributions from the University of Michigan Herbarium 16: 55-108
Anderson W. R. 1990a The origin of the Malpighiaceae—the evidence from morphology. Memoirs of the New York Botanical Garden 64: 210-224
Anderson W. R. 1990b Notes on neotropical Malpighiaceae—III. Contributions from the University of Michigan Herbarium 17: 39-54
Anderson W. R. 1993 Chromosome numbers of neotropical Malpighiaceae. Contributions from the University of Michigan Herbarium 19: 341-354
Anderson W. R. 2001. Observations on the Malagasy genus Rhynchophora (Malpighiaceae). Contributions from the University of Michigan Herbarium 23: 53-58
Anderson W. R. B. Gates 1981 Barnebya, a new genus of Malpighiaceae from Brazil. Brittonia 33: 275-284[CrossRef][ISI]
Cameron K. M. M. W. Chase W. R. Anderson 1995 Comparison of molecular trees from rbcL and matK analyses of Malpighiaceae. American Journal of Botany 82 : (supplement) 117 (Abstract)
Cameron K. M. M. W. Chase W. R. Anderson H. G. Hills 2001 Molecular systematics of Malpighiaceae: evidence from plastid rbcL and matK sequences. American Journal of Botany 88:1847–1862
Carr G. D. G. McPherson 1986 Chromosome numbers of New Caledonian plants. Annals of the Missouri Botanical Garden 73: 486-489[CrossRef][ISI]
Catalán P. E. A. Kellogg R. G. Olmstead 1997 Phylogeny of Poaceae subfamily Pooideae based on chloroplast ndhF gene sequences. Molecular Phylogenetics and Evolution 8: 150-166[CrossRef][ISI][Medline]
Chase M. W. et al 1993 Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcL. Annals of the Missouri Botanical Garden 80: 528-580[CrossRef][ISI]
Clark L. G. W. Zhang J. F. Wendel 1995 A phylogeny of the grass family (Poaceae) based on ndhF sequence data. Systematic Botany 20: 436-460
Clegg M. T. G. H. Learn E. M. Golenberg 1991 Molecular evolution of chloroplast DNA. In R. K. Selander, A. G. Clark, and T. S. Whittam [eds.], Evolution at the molecular level, 135–149. Sinauer, Sunderland, Massachusetts, USA
de Jussieu A. 1843 Monographie de la Famille des Malpighiacées. Archives du Muséum d'Histoire Naturelle 3: [5]-151, [255]-616, pl. 1–23
Doyle J. J. J. L. Doyle 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15
Farris J. S. M. Källersjö A. G. Kluge C. Bult 1994 Testing significance of incongruence. Cladistics 10: 315-319[CrossRef][ISI]
Felsenstein J. 1985 Phylogenies and the comparative method. American Naturalist 125: 1-15
Ferguson D. M. 1999 Phylogenetic analysis and relationships in Hydrophyllaceae based on ndhF sequence data. Systematic Botany 23: 253-268
Hutchinson J. 1967 The genera of flowering plants, vol. 2. Oxford University Press, Oxford, UK
Kim K.-J. R. K. Jansen 1995 ndhF sequence evolution and the major clades in the sunflower family. Proceedings of the National Academy of Sciences, USA 92: 10379-10383
Larson A. 1994 The comparison of morphological and molecular data in phylogenetic systematics. In B. Schierwater, B. Streit, G. P. Wagner, and R. DeSalle [eds.], Molecular ecology and evolution: approaches and applications, 371–390. Birkhauser Verlag, Basel, Switzerland
