| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
Department of Botany, Natural History Museum, London SW7 5BD, UK
Received for publication October 15, 2005. Accepted for publication December 1, 2005.
ABSTRACT
The revision of species-rich genera underpins research and supports the sustainable use and monitoring of biological diversity. One fifth to one quarter of the diversity of all seed plant species occurs in such genera, but difficulties with the revision of species-rich genera has resulted in many of them being ignored since the late 1800s. Pilea, with 600–715 species is in need of revision. The only realistic approach is in manageable subunits, which requires confirmation of monophyly and identification of monophyletic subdivisions. Parsimony analyses of trnL-F, ITS, and morphology data were used to test the monophyly of, and explore intrageneric relationships within, Pilea. Analysis of trnL-F data confirms and recovers two morphologically diagnosable monophyletic clades that include all of the taxa within Pilea. Overlaying geographic distribution on a most parsimonious tree indicates a strong association between geography and phylogenetic relatedness. It is suggested that a strategic revision within the framework of morphologically and geographically diagnosable units might enable the revision of the group using an iterative approach. Analysis of the outgroup taxa supports the inclusion of Poikilospermum within the Urticaceae and suggests that the Urticaceae tribes could be placed into two clades that are supported by floral morphology.
Key Words: Achudemia • ITS • Pilea • Poikilospermum • species-rich genera • trnL-F • Urticaceae
Pilea, with 600–715 taxa (Adams, 1970
; Burger, 1977
; Monro, 2004
) is the largest genus in the Urticaceae and one of the larger genera in the Urticales and Eudicot Rosids. It is distributed throughout the tropics, subtropics, and warm temperate regions (with the exception of Australia and New Zealand). The majority of taxa are succulent shade-loving herbs or shrubs, which are easily distinguished from other Urticaceae by the combination of opposite leaves (with rare exceptions) with a single ligulate, intrapetiolar stipule in each leaf axil and cymose or paniculate inflorescences (again with rare exceptions). Pilea is of little economic importance; four species are of minor horticultural importance (P. cadierei, P. involucrata, P. microphylla, and P. peperomioides) and one species is used in Chinese traditional medicine (P. plataniflora). The genus has attracted little monographic attention since Weddell (1869)
, and the majority of taxonomic contributions have come from floristic treatments (Killip, 1939
; Standley and Steyermark, 1952
; Adams, 1972
; Burger, 1977
; Van Royen, 1982
; Chen, 1982
, 1995
; Monro, 2001
, in press
; Chen and Monro, 2003
; Pool, 2001
; Hopkins and Monro, in press
). To date, 787 species names have been published (International Plant Names Index, 2003) and estimates for the species number range from 250 (Friis, 1989
), 715 (Monro, 2004
) to 1000 (C. D. Adams, BM, personal communication). Based on previous floristic treatments (Table 1), ca. 30% of the species from regions not yet covered by contemporary floristic treatments may be undescribed. As more species are discovered, the logistical problems of undertaking a monographic revision will grow.
|
). Difficulties associated with the revision of such large genera have resulted in many of these groups being ignored since the 19th century. This has impeded research into broad-ranging questions in systematics such as the reasons for the radiation in species-rich genera and the accumulation of species richness in the tropics (Axelrod, 1983
; Pelser et al., 2002
; Bramley et al., 2004
; Frodin, 2004
). Robust and stable classifications of such species-rich groups also underpin the effective assessment, monitoring, and conservation of biodiversity within the context of the Convention on Biological Diversity (Secretariat of the Convention on Biological Diversity, 2005b
).
Given the number of taxa, that almost all are nondescript understory herbs of minimal economic use, and the current global taxonomic capacity (Secretariat of the Convention on Biological Diversity, 2005a
), it is unlikely that Pilea will be the subject of a global monographic revision in the near future. One approach to tackling taxonomic problems in Pilea would be to revise geographically diagnosable units, as happens in floristic treatments, or morphologically diagnosable units, as happened in the classical works of the 19th and 20th centuries. Under this approach, pragmatically defined units could be revised by different authors at different times on a pragmatic basis. The limitations of such an approach would be that it is unlikely to result in monophyletic groupings that would represent best estimates of evolutionary relationships between taxa. They would not therefore result in a classification that underpins both broader research in systematics and science and the conservation of biological diversity. An alternative approach would be to confirm generic monophyly first and then identify manageable monophyletic infrageneric groupings that could be revised independently. However, the identification of such monophyletic subgroupings from existing classifications can be problematic in large genera in the absence of a robust phylogenetic framework as has been demonstrated in numerous groups (Wojciechowski et al., 1993
, 1999
, Astragalus; Bohs and Olmstead, 1997
, Solanum; Pelser et al., 2002
, Senecio; Plana, 2003
, Begonia; Moylan et al., 2004
, Strobilanthes; Berry et al., 2005
, Croton).
In the context of generic monophyly, Pilea appears to be a well-circumscribed group within the tribe Lecantheae (Friis, 1993
). However, the majority of the diagnostic generic characters within the Urticaceae relate to inflorescence and flower structure, characters for which homology is difficult to assess. The "reduced" flowers of Urticaceae, possessing a single perianth whorl and the frequency of "reduced" inflorescence types (e.g., Cecropia, Rousselia, Parietaria, Phenax, Myriocarpa, Forrskaolea) has made the classification of Urticaceae notoriously difficult, and genera such as Cecropia and Poikilospermum have been repeatedly excluded from, and included within, the family (Weddell, 1856
; Berg, 1977
, 1978
; Friis, 1989
, 1993
; Sytsma et al., 2002
). As a consequence, the morphological delimitation of genera and subfamily groupings can be problematic and may not be monophyletic. Hadiah et al. (2003)
demonstrated that Elatostema, for example, is very likely to be polyphyletic, requiring the inclusion of Procris to form a monophyletic group. The monophyly of Pilea, a large and morphologically diverse genus, also remains to be critically examined, an important step in defining units for taxonomic revision. Investigating its tribal position will also help to support future studies of intrageneric relationships within the tribe and family.
Within Pilea there have been two main infrageneric classifications (Table 2). The first (Weddell, 1856
, 1869
) saw Pilea divided into three sections: Integrifoliae, Heterophyllae, and Dentatae, based on leaf isomorphy and margin morphology. This was expanded by Killip (1939)
, working on a revision of species from the Andes, who recognized a total of 12 groups, through the inclusion of leaf nervation, pubescence, and staminate inflorescence characters. The second classification (Chen, 1982
), based largely on the Chinese and Asian taxa, saw the genus divided into seven sections: Achudemia, Smithiella, Tetrameris, Pilea, Dimeris, Urticella, and Lecanthoides, based on pistillate tepal number, pistillate tepal isomorphy, leaf nervation, and staminate inflorescence structure (Table 2). The two classifications are incompatible with each other (Table 2) suggesting the nonmonophyly of subdivisions in at least one of them. Furthermore, both Weddell's and Chen's classifications are based on morphological characters for which an understanding of ontogeny and focus on homoplasy is absent; what anatomists would call "morphology of systematists" (Weber, 2003
). Given the patchy success of this approach to identifying morphological characters in the identification of monophyletic infrageneric groups in other genera, it is reasonable to question the monophyly of the subdivisions of both classifications.
|
and Chen (1982)
. MATERIALS AND METHODS
Nomenclature
Taxon names follow Weddell's (1869)
classification except where a more recent floristic treatment exists (Adams, 1972
; Van Royen, 1982
; Monro, 2001
; Chen and Monro, 2003
and Fortune-Hopkins and Monro, in press
). Eight undescribed taxa (Pilea sp. ‘Nepal', P. sp. ‘Peru 2'., P. sp. ‘Peru 3', P. sp. ‘Peru 5', P. sp. ‘Peru 6', P. sp. ‘Peru 9', P. sp. ‘Peru 10', P. "aff. pteridophylla") were also included.
Taxon sampling
A morphologically and geographically diverse sample of Pilea was included in all three data sets with three or more representatives taxa from all of Weddell's (1856)
three sections and five of Chen's (1982)
seven sections sampled. To ensure a geographically diverse sample, at least three representative taxa were sampled from each floristic region (Takhtajan, 1986
) in which Pilea occurs (floristic region and number code in parantheses): Boreal subkingdom (Eastern Asiatic, 2); African subkingdom (Guineo-Congolian, 10; Sudano-Zambesian, 12); Indomalesian subkingdom (Indian, 16; Indochinese region, 17; Malesian region, 18) and the Neotropical kingdom (Caribbean region, 23; Brazilian region, 26; Andean region, 27). The only exception was the taxon-poor Madagascan region for which two species were sampled.
Ingroup delimitation and outgroup choice
The trnL-F sequence data were generated for 112 accessions, representing 109 taxa (95 ingroup, 14 outgroup) (Appendix). Outgroup selection for the trnL-F analysis was directed by a combined analysis of rbcL, trnL-F, and ndhF by Sytsma et al. (2002)
as well as an analysis of ndhF by Datwyler and Weiblen (2004)
. Both analyses recovered a strongly supported monophyletic Urticaceae. In light of this fact, and combined with other evidence for both Urticaceae monophyly and a Celtidaceae sister group (Sytsma et al., 2002
), an initial analysis to investigate the monophyly and sister group relationship of Pilea within Urticaceae was performed using trnL-F data from 23 taxa at generic rank, including multiple representatives of all of the Urticaceae tribes sensu Friis (1989)
. Trema integerrima (Celtidaceae) was selected as an outgroup for this analysis.
The ITS analysis included sequence data from 89 accessions, representing 85 taxa (84 ingroup, 1 outgroup) (Appendix). ITS sequence data is highly variable between genera, making alignment of the whole sequence difficult. As a result, ingroup taxa were aligned with a single outgroup, Lecanthus peduncularis. This taxon was selected as outgroup as the trnL-F analysis recovered it as sister group to Pilea. The ITS region included a single region (positions, 619–669) that was difficult to align. Two phylogenetic analyses were undertaken with this region alternately excluded then included. Support values were also calculated for each. The resultant strict consensus trees were entirely congruent and did not differ markedly in support values. As a result the variable region was included in the analyses presented here (Fig. 3; Appendix S1, see Supplemental Data accompanying online version of this article).
|
and Chen (1982)
to define existing generic sections (Table 2), those used by Killip (1939)
to subdivide Weddell's sections, and those used for the revision of Pilea for Flora Mesoamericana (Monro, 2001
), together with additional traits identified during the course of this study. Morphological data were not scored for the outgroup taxon because assessments of homology are problematic for many characters (see Introduction).
Characters were scored from herbarium collections at BM, BRU, C, F, GH, K, L, LL, MEXU, MO, NY, P, PE, PMA, TEX, and US. Traits not amenable to cladistic analysis were screened out using the following criteria: they showed overlapping variation that could not be partitioned into discrete states (Stevens, 1991
), they were polymorphic within species, they were unique to a single taxon and hence uninformative of relationships, and they could be scored for less than 50% of taxa. Characters excluded by these criteria did not include any of those used by Weddell to define his sections, but resulted in the exclusion of two characters used by Chen: the presence/absence of a receptaculate staminate inflorescence, which was restricted to a single taxon, and the presence/absence of a spicate pistillate inflorescence, for which material could not be obtained. Field characters (e.g., habit, presence/absence of tubers, presence/absence of succulent leaves) were also excluded, due to their sporadic and inconsistent reporting in collector's field notes. Where characters exhibited non-independence (the so-called red-tail/blue-tail scenario of Maddison, 1993
), the conventional coding protocol was followed and the variation was coded as two characters (Hawkins, 2000
). Thus in the case of trichome structure the variation was coded as pubescence present or absent and trichomes single celled, multicellular, or inapplicable where pubescence was absent.
Molecular method
Total genomic DNA was extracted from ca 0.5 g of leaf material (herbarium specimens or silica-gel dried) using a modified cetyltrimethyl ammonium bromide (CTAB) microextraction protocol (Doyle and Doyle, 1987
) and purified, without precipitation, using GFX cleaning columns (Amersham Bioscience, Little Chalfont, UK) according to manufacturer's protocols. Standard polymerase chain reaction (PCR) procedures were used to amplify the target gene regions using a Techgene Thermal Cycler (Techgene, Cambridge, UK).
The trnL-F region was amplified (30 cycles of 94°C for 15 s, 50°C for 30 s, 72°C for 1 min) in one fragment using the universal primers trnLc and trnLf (Taberlet et al., 1991
). Where reactions repeatedly failed, trnLc was replaced with trnLFern-1 (Trewick et al., 2002
). The ITS-1 and ITS-2 regions, including the 5.8S spacer (White et al., 1990
) were amplified (30 cycles of 94°C for 30 s, 50°C for 30 s, 72°C for 2 min) in one fragment, using primers 16 SE and 23 SE (Sun et al., 1994
). Betaine (1.2 mol/L) was added to prevent the formation of secondary structures in difficult templates (Chakrabarti and Schutt, 2001
). Amplified products were purified using GFX cleaning columns according to the manufacturer's protocols.
Dideoxy cycle sequencing (28 cycles: 30 s of denaturation at 95°C, 15 s of annealing at 50°C, 4 min of extension at 60°C) with big dye terminators was performed in 10 µL volumes using a Techgene Thermal Cycler (Techgene). Excess dye-labeled nucleotides from the sequence reactions were removed by standard ethanol/sodium acetate precipitation. Sequence products were subsequently resuspended and run on an ABI 377 DNA sequencer (Applied Biosystems, Foster City, California, USA).
Sequence alignment and indels
Sequence data were edited and assembled using Lasergene Navigator (DNAStar, Madison, Wisconsin, USA). Verified sequences were then aligned by eye in Se-Al (version 1.0al; Rambaut, 1996
). Gaps were coded as missing data.
Phylogenetic analyses
Parsimony analyses were performed on the following data sets using PAUP* 4.0b10 (Swofford, 2002
): morphology, trnL-F (including all ingroup and outgroup taxa), ITS (including all ingroup and a single outgroup taxon). The trnL-F data set was pruned to exclude accessions not included in the ITS data set, and the ITS data set was pruned to exclude accessions not included in the trnL-F data set for the combined trnL-ITS-morphology analysis.
Parsimony analyses were conducted with all characters unordered. Given the large size of the data sets, there are large numbers of possible topologies (Felsenstein, 1978
). Furthermore, multiple islands of equally parsimonious trees might occur (Maddison, 1991
). To overcome these potential problems, the search strategy protocol of Catalán et al. (1997)
was adopted. An initial heuristic search, comprising 10 000 replicates of random stepwise addition using tree bisection and reconnection (TBR) and branch swapping (with MULTREES option on), saving five trees at each replicate, was performed. A strict consensus tree was then calculated from the resulting most-parsimonious trees. The strict consensus tree was then loaded as a constraint tree in a further round of heuristic searches during which 20 000 replicates of random stepwise addition were performed with TBR branch swapping saving only trees not compatible with the strict consensus tree.
For each analysis, the consistency index (CI) (Kluge and Farris, 1969
), the CI excluding uninformative characters (CI*), and retention index (RI) (Farris, 1989
) were calculated. The robustness of clades recovered in the strict consensus trees was evaluated using nonparametric bootstrap analysis (Felsenstein, 1985
) and by computing decay value indices (Bremer, 1988
). Bootstrap values were obtained from 100 pseudoreplicates, each comprising 1000 heuristic searches of random stepwise addition with TBR branch swapping. Bootstrap support values were categorized as follows: poor, <50%; weak, 50–74%; moderate, 75–84%; strong, 85–100% (Chase et al., 2000
). Decay value indices (DI) were obtained using DNA Stacks 1.3 (Eernisse, 2000
). Those clades with DI values 
4 were considered well supported (Marcilla et al., 2001
). Neither support value logically provides accurate confidence levels (Sanderson, 1995
; Lee, 2000
), and there is clearly a difficulty in the interpretation of bootstrap values (Lee, 2000
). A combination of different support values were used, because incongruencies between support values can provide insights into the size of the suites of characters corroborating and contradicting support for a particular node (Lee, 2000
) and therefore help in the interpretation of the support for that node.
To investigate the evolution of morphological characters and identify morphologically diagnosable infrageneric groups, all morphological characters were optimized onto a randomly chosen most parsimonious tree (MPT) of the combined data set (Table 3, analysis 4) using MacClade 4 (ACCTRAN optimization; Maddison and Maddison, 2000
)
|
) have been proposed to test for congruence. The ILD test is widely used, but there is increasing evidence that it is a poor indicator of data set combinability under biologically realistic parameters (Dolphin et al., 2000
; Barker and Lutzoni, 2002
; Darlu and Lecointre, 2002
). Furthermore, conflict between molecular data sets may be the result of sample error, resulting from the small number of data sets analyzed and not incongruence in phylogenetic signal (Taylor and Piel, 2004
). For this reason, topographic congruence was used to assess combinability.
To examine congruence, parsimony analyses were performed and a strict consensus tree generated for taxonomically equivalent data sets for trnL-F, ITS, and morphology. Strict and semi-strict consensus trees of the two partition strict consensus trees were then generated to identify (1) nodes present in both analyses, and (2) nodes present in one that were not contradicted by the other analysis. Where incongruent nodes were identified, support for these nodes was assessed through the examination of bootstrap support values following the guidelines of Taylor and Piel (2004)
. Incongruence between nodes was considered conflicting phylogenetic signal when each node had bootstrap support values 95% (Taylor and Piel, 2004
).
RESULTS
Herbarium material as a source of DNA for molecular study
Herbarium material represented a good source of DNA for the amplification of the trnL-F and ITS regions in Urticaceae. Extraction was attempted from a total of 198 herbarium collections, representing 24 genera and was successful in 141 instances (a success rate of 72%). DNA was successfully obtained for samples collected between 1811 (P. verbascifolia) and 2002, although the majority of the collections used were collected after 1930.
Morphology and evolution
A description of the morphological variation, characters, and character states used in this analysis is presented in Appendix S2 (see Supplemental Data with online version of this article). Of the 27 variable characters coded, four were multistate, 23 were binary characters, and five were of the red tail/blue tail type (3, 11, 13, 17, and 18). Parsimony analysis (Table 3, analysis 1) of the morphological matrix resulted in a strict consensus tree (Fig. 1) that yielded 28 clades, of which two had >50% bootstrap support.
|
|
Assessment of congruence between data sets
The strict consensus trees for the taxonomically equivalent trnL-F and ITS data partitions (Fig. 4) recovered 42 and 47 nodes respectively. A semi-strict consensus of these two consensus trees identified 24 nodes that were either recovered in both partitions or recovered in one partition and not contradicted in the other. No incongruencies were recovered between analyses of the data partitions for the primary, secondary or tertiary clades, the main focus of this study. Of the nodes that were incongruent between the two partitions, only three had bootstrap values >95%, and all three were terminal (Fig. 4, clades I-1, I-2, I-3). Much of the incongruence between data partitions was a consequence of the positioning of five taxa (P. dauciodora, P. daguensis, P. hyalina, P. purulensis, and P. verbascifolia). Between them, these taxa accounted for 14 incongruent nodes between the partitions, of which only two (Fig. 4, clades I-1, I-2) had support (>95%). Given the lack of well-supported incongruence between data sets, the decision was taken to combine the data sets for analysis. No conflict between the strict consensus of the morphology partition, and the two molecular data partitions were identified. The strict consensus of the morphology data partition did not resolve any nodes with bootstrap values greater than 60%, a topographic comparison was therefore not undertaken as incongruence could not be considered conflicting.
|
|
Monophyly of the genus, outgroup relationships, and existing subgeneric classifications
This study represents the first phylogenetic analysis of Pilea. The analysis of trnL-F sequence data provides strong support for the monophyly of the genus (bootstrap 100%, DI 16, Fig. 2), confirming its morphological delimitation by Lindley (1821)
and subsequent authors. The analysis of trnL-F data identified Lecanthus peduncularis as the most closely related taxon to Pilea, but with moderate support (bootstrap 78%, DI 5), followed by Pellionia with poor support (bootstrap <50%, DI 1). This supports Weddell's (1856)
and Friis's (1989)
placing of these genera within the Lecantheae but contradict Sytsma et al.'s (2002)
recovery of a paraphyletic Lecantheae in their combined analysis of ndh-F and trnL-F data.
The analyses of ITS data placed P. japonica outside of the monophyletic Pilea (Fig. 3). Unfortunately, it was not possible to amplify either accession of this species for trnL-F. Pilea japonica was originally placed in the genus Achudemia by Maximovicz (1876)
, Handel-Mazzetti (1929)
later placed it in synonymy with Pilea. Pilea section Achudemia was recognized by Chen (1982)
. It is characterized by isopentamerous flowers. Examination of material for this study suggests that, as circumscribed for the Flora of China (Chen and Monro, 2003
), P. japonica includes a mixture of isomerous and hetermerous individuals. Heteromerous material, determined for this study as P. japonica 1, is recovered in clade 2, whilst isomerous material conforming to type material of P. japonica was recovered outside of the monophyletic Pilea clade. Given these results, the position of Achudemia relative to Pilea and the delimitation of P. japonica needs further study. If Achudemia were to be separated from Pilea, then this may suggest that the isomerous character state in Pilea is basal to the heteromerous state.
The 16 Urticaceae genera included in the trnL-F analysis (Fig. 2) represents the largest phylogenetic analysis of Urticaceae published to date. Previous phylogenetic studies that include Urticaceae genera (Sytsma et al., 2002
; Hadiah et al., 2003
; Datwyler and Weiblen, 2004
) have sampled cpDNA from 6–10 genera. These studies have however, been difficult to compare because there are few taxa in common between studies.
The strict consensus tree rooted on Trema integerrima (Fig. 2), resolved two clades within the Urticaceae, one weakly (Fig. 2, clade I, bootstrap 64%, DI 2) and the other strongly supported (Fig. 2, clade II, bootstrap 100%, DI 16). This analysis supports Sytsma et al.'s (2002)
conclusion that Poikilospermum belongs within the Urticaceae. In contrast, Datwyler and Weiblen's (2004)
analysis of ndhF data place Poikilospermum as sister to the Urticaceae. With respect to the relationship between the tribes, this study suggests that the tribes could be placed into two clades, one corresponding to clade I in Fig. 2, that would include the Parietarieae, Boehmerieae (excluding Myriocarpa), and the Forsskaoleae, and the other, which would include the Urticeae and Lecantheae (including Myriocarpa), corresponding to clade II in Fig. 2. Based on Friis's 1993
circumscription of the tribes and morphology within the Urticaceae, the following groupings may be supported by the combination of two morphological characters: clade I, pistillodes present and glabrous; II); clade II, pistillodes absent or present, where present pubescent. Clade II is also congruent with Friis's tentative cladogram for the family (Fig. 16.9A of Friis, 1993
).
Morphological characters
Morphological characters, as used in this study, were found to be homoplastic (Table 3, analysis 1: CI* (consistency index excluding uninformative characters) of 0.177; Fig. 1). There are three explanations for the high level of homoplasy found here. (1) Morphological characters may be subject to frequent reversals and convergence. (2) Morphological characters may have been studied insufficiently to fully distinguish primary homologies from nonhomologies (Weber, 2003
), and/or (3) morphological characters are selected and coded in an "atomistic" way (Kirchoff et al., 2004
). Despite being homoplastic, morphological characters still remain essential to a functional classification of Pilea as they do for most organisms (Scotland et al., 2003
), and in this study, unique combinations of leaf incision, pubescence, inflorescence, and flower characters can be used to diagnose several monophyletic clades (Table 4). Leaf margin incision formed part of unique combinations corresponding with two monophyletic clades (Table 4, Fig. 5, clades G and H), and isomorphy can be used to diagnose three clades in combination with other characters (Table 4, Fig. 5, clades A, G, H). The presence or absence of pubescence on the leaves formed part of unique combinations corresponding to three monophyletic clades (Table 4, Fig. 5, clades A, G, H) and staminate tepal sub-apical appendage structure formed part of unique combinations corresponding to terminal lades A, F, and G (Table 4, Fig. 5).
|
) and six of seven of Chen's (1982)
sections were included in this study and are plotted alongside one optimal tree from the combined analysis (Fig. 6). All of the sections included were para- or polyphyletic, with the exception of Chen's section Tetrameris, which together with section Lecanthoides forms a monophyletic group corresponding to clade 1–1 (Fig. 6). In summary, whilst Weddell's and Chen's subdivisions are valuable as artificial classifications to facilitate the identification of taxa, both are ill-suited as the basis of a revision of Pilea based on the strategic revision of monophyletic groups. Similar incongruence with respect to the subgeneric classifications of the 19th and early 20th century subdivisions have been found in Astragalus (Wojciechowski et al., 1999
), Croton (Berry et al., 2005
), Strobilanthes (Moylan et al., 2004
), and Quercus (Nixon, 1993).
|
Congruence with geography
Geographic distribution for the major clades is overlaid on a phylogram from the combined analysis in Fig. 6. It is apparent from Fig. 6 that there is a strong geographical signal in the phylogeny. It is worth noting that both clades 1 and 2 (Fig. 6) are relatively cosmopolitan, each including taxa from four or five of the five Takhtajan (1986)
biogeographic kingdoms/subkingdoms where Pilea is known to occur. Basal subclades within each of these two clades (1–1, 2–1, and 2–2) are dominated by Indomalesian taxa. This concurs with high levels of morphological diversity in the region. Taxa distributed in Indomalesia exhibit the full range of variation in stipule, leaf, and inflorescence morphology, as well as a number of unique states: divided and lineate stipules (not sampled here), alternate and peltate leaves, and receptaculate and spicete (not included here) inflorescences.
Subclade 1–2 includes both neotropical and African taxa. Subclade 2–3 is dominated by neotropical taxa, which account for 90% of the species in the clade (45 of 50) and 67% of the species of Pilea sampled (45 of the 67) (Fig. 6). Large radiations in the neotropics are well documented in other plant genera (Wojciechowski et al., 1999; Richardson et al., 2001
; Berry et al., 2005
), but this is the first time that it has been documented in the Urticaceae, sensu Friis (1989)
. Within subclade 2–3, there is further geographical structure, clade H (Fig. 5) containing only taxa from the Antilles and clade A (Fig. 5) containing taxa only from Mesoamerica.
Toward a classification of Pilea
Morphological characters fail to define all nodes, either as synapomorphies or in unique combinations. Given the strong geographical signal evident in the phylogeny, a combination of morphologically and geographically circumscribed groups may provide a pragmatic way to identify units for a revision. Frodin (2004)
in his review of the history and concepts of big plant genera suggests two strategies for taxonomists wanting to revise the classification of species-rich genera: (1) rapid reviews of current knowledge, and (2) further study combining traditional and contemporary methods. These two approaches, if they are to be effective, rely on extensive congruence between existing subdivisions based on classical works of the 19th and early 20th centuries and those resulting from contemporary phylogenetic studies. Whilst this is the case for a number of species-rich genera, such as Ficus and Solanum (Frodin, 2004
), it is clearly not the case for Pilea, where existing subdivisions are unsuitable as a framework for the strategic revision of the genus. An alternative strategy would be to base a revision entirely on monophyletic groupings recovered as a result of phylogenetic analysis. However, as in the case of Astragalus (Wojciechowski et al., 1999
), many of the monophyletic groupings recovered in this study were cryptic clades and as such unsuitable for a strategic revision (it would not be possible to identify which taxa to include at any stage of the revision).
Given the strong association between geographical distribution and phylogenetic relationships in clade 2 (Fig. 6), a pragmatic subdivision of the clade into units for future revision is therefore proposed that utilizes both morphology and geography. Where morphologically diagnosable clades can be identified these will be used for future revisionary studies. However, where it is not possible to identify morphologically diagnosable clades, then geographical units will be used. Effectively this strategy would combine phylogenetic and floristic approaches to enable the strategic revision of the group over an undetermined (but not indefinite) period of time. It is hoped that this approach will become iterative, the process of revising the pragmatically defined monophyletic groups identified in this study (units 1–6, see also Fig. 6), resulting in the discovery of new morphological characters, enabling the diagnosis of clades that are currently cryptic. Ultimately the repetition of this process could lead to the inclusion of all taxa in diagnosable clades.
The combined analysis of trnL-F, ITS, and morphological characters recovered two monophyletic clades, which are morphologically diagnosable and thus easily identifiable; these should form the basis of any future revision of Pilea. In contrast, the smaller clade (Fig. 5, clade 1) constitutes a manageable unit for revision as it comprises ca. 75 species. In contrast, the larger of these clades (Fig. 5, clade 2) contains over 85% of the taxa sampled here, approximately 600 species or more. As such, it is still too large to be revised by a single author or in a single study. Adopting this strategy results in the delimitation of the following six monophyletic groupings as the basis for future revision:
Recognition of these units leaves a residual group of cryptic clades, isolated taxa, and other biogeographic areas (for example Africa) for which phylogenetic relationships remain unresolved. These unresolved groups represent ca. 50% of the species in the genus. It is hoped that the revision of the units outlined will result in the ultimate inclusion of these residual groups in new or existing diagnosable monophyletic groupings.
APPENDIX
Voucher information, Weddell (1869)
section, Chen 1982
) section, and GenBank accession numbers for all sequences for the taxa analyzed. For outgroup taxa, Urticaceae tribe sensu Friis (1993)
replaces reference to section. A dash indicates the region was not sampled. A number or locality in bold after the species epithet indicates that the sample is one of a multiple accession for this taxon. Voucher specimens are deposited in the following herbaria: BM = The Natural History Museum, London; MO = Missouri Botanical Garden; PE = Chinese Academy of Sciences Institute of Botany, Beijing; PMA = Universidad de Panamá.
Taxon—voucher; locality; Weddell section; Chen (section; GenBank accession nos.: trnL-F; ITS.
Boehmeria nivea Gaudich.—Togasi 1253 (BM); Japan; Boehmerieae; DQ179366; —.
Cecropia obtusifolia Bertol.—Monro 3767 (BM); El Salvador; tribe unknown; DQ179377; —. Cypholophus cf. trapula H. Winkler—Pullen 5962 (BM); Papua New Guinea; Boehmerieae; DQ179365; —.
Discocnide mexicana (Liebm.) Chew—Gereau et al. 2205 (BM); Mexico; Urticeae; DQ179369; —. Droguetia iners (Forssk.) Schweinf.—Wood Y-74–382 (Yemen); Forrskaoleae; DQ179371; —.
Forsskaolea tenacissima L.—Thesiger s.n. (BM); Saudi Arabia; Forrskaoleae; DQ179376; —.
Gesnouinia arborea Gaudich.—Evrard 12088 (BM); Spain: Canary Islands; Parietarieae; DQ179372; —.
Lecanthus peduncularis Wedd.—Chiang & Hu 172 (BM); Taiwan; Lecantheae; DQ179370; DQ175619.
Myriocarpa longipes Liebm.—Monro 3998 (BM); Peru; Boehmerieae; DQ179364; —.
Parietaria pensylvanica Muhl. ex Willd.—Hinton 2901 (BM); Mexico; Parietarieae; DQ179375; —. Pellionia repens (Lour.) Merr.—Villacorta & Aranivea 98 (BM); El Salvador, cultivated; Lecantheae; DQ179373; —. Pilea alpestris Fawcett & Rendle—Adams 10626 (BM); Jamaica; Integrifoliae; Urticella; DQ179307; DQ175544. P. alpina Urb—Tuerckheim 3426 (BM)—Santo Domingo; Integrifoliae; Pilea; DQ179309; DQ175543. P. alsinifolia Wedd.—Killip 35585 (BM); Colombia; Integrifoliae; Urticella; DQ179322; —. P. angustifolia Killip—Davidse & G. Herrera Ch. 26254 (BM); Costa Rica; Dentatae; Urticella; DQ179289; DQ175556. P. anisophylla Wedd.—Ludlow 7081 (BM); China; Heterophyllae; Urticella; DQ179343; —. P. aphrophila Killip—Killip 34825 (BM); Colombia; Dentatae; Urticella; DQ179323; DQ175589. P. basicordata W.T. Wang ex C. J. Chen—Qing 76 (PE); China; Dentatae; Tetrameris; DQ179361; DQ175614. P. bassleriana Killip; Mexia 6360 (BM); Peru; Heterophyllae; Urticella; —; DQ175592. P. benguetensis C.B. Robinson—Merrill 9696 (BM); Phillipines; Dentatae; Urticella; DQ179337; DQ175554. P. cadierei Gagnep. & Guillaumin—RBGE 19697470 (E); cultivated; Dentatae; Tetrameris; DQ179359; DQ175608. P. caribaea Urb.—Howard et al. 19931 (BM); St. Lucia; Dentatae; Urticella; DQ179305; —. P. carnulosa Wedd.—Killip & Varela 34659 (BM); Colombia; Dentatae; Urticella; DQ179324; DQ175593 P. centradenoides Seem. 1—Dressler 5884 (PMA); Panama; Heterophyllae; Urticella; —; DQ175573. P. centradenoides Seem. 2—Henshold 1114 (BM); Panama; Heterophyllae; Urticella; —; DQ175576. P. ciliata Blume—Webster & Wilson 5163 (BM); Jamaica; Heterophyllae; Urticella; DQ179300; DQ175538. P. clementis Britton—Gonzáles 690 (BM); Cuba; Dentatae; Urticella; DQ179310; DQ175550. P. consanguinea Wedd.—Tuerckheim 3597 (BM); Santo Domingo; Integrifoliae; Urticella; DQ179312; DQ175539. P. cornuto-cucullata Cufod.—D'Arcy & B. Hammel 12448 (BM); Panama; Dentatae; Urticella; —; DQ175577. P. costaricensis Donn. Sm.—Burger & J. Gentry 9069 (BM); Costa Rica; Heterophyllae; Urticella; —; DQ175566. P. craspedodroma A.K. Monro—Johns 10522 (BM); Indonesia: Irian Jaya; Dentatae; Urticella; DQ179336; —. P. daguensis Killip—Ibarra M. et al. 3844 (BM); Mexico; Heterophyllae; Urticella; DQ179332; DQ175567. P. dauciodora Mexico Wedd. ex Pav. 1—Martínez 13170 (BM); Mexico; Dentatae; Urticella; DQ179285; —. P. dauciodora Mexico Wedd. ex Pav. 2; Martínez 20771 (BM); Mexico; Dentatae; Urticella; DQ176857; DQ175562. P. dauciodora Peru Wedd. ex Pav. 3—Monro 4018 (BM); Peru; Dentatae; Urticella; —; DQ175564. P. digitata A.K. Monro—McPherson 7169 (MO); Panama; Dentatae; Urticella; DQ179326; DQ175559. P. dolichocarpa C. J. Chen—Chien 768 (PE); (China); Dentatae; Tetrameris; —; DQ175609. P. dominguensis Urb.—Tuerckheim 3180 (BM); Santo Domingo; Integrifoliae; Urticella; DQ179313; DQ175541. P. ecboliophylla Donn. Sm.—Servín O. 1011 (BM); (Mexico); Heterophyllae; Urticella; DQ179292; DQ175531. P. elegantissima C. J. Chen—Sino-American Ghuizhou Expedition 1575 (BM); China; Dentatae; Urticella; DQ179348; —. P. foliosa Killip—Killip & Smith 24400 (BM); Peru; Heterophyllae; Urticella; DQ179291; DQ175571. P. forgetii N.E. Br.—Folsom 3498A (BM); Panama; Dentatae; Urticella; DQ179333; DQ175585. P. forsythiana Wedd.—Ernst 1196 (BM); Dominica; Integrifoliae; Urticella; DQ179311; DQ175546. P. fruticosa Hook. f.—Clemens & Clemens 27865 (BM); Borneo; Integrifoliae; Urticella; DQ179353; DQ175604. P. glaberrima Blume—Gardner 220 (BM); Nepal; Integrifoliae; Urticella; DQ179352; DQ175600. P. glabra S. Watson—Servín O. 1415 (BM); Mexico; Integrifoliae; Urticella; DQ179354; DQ175584. P. grandifolia Blume—Morley & Whitefoord 867 (BM); Jamaica; Dentatae; Urticella; DQ179303; DQ175551. P. harrisii Urb.—Harris 11220 (BM); Jamaica; Dentatae; Urticella; DQ179302; DQ175537. P. herniarioides (Sw.) Wedd.—Bellingham 1280 (BM); Jamaica; Dentatae; Urticella; DQ179321; DQ175572. P. holstii Engl.—Taylor 2563 (BM); Uganda; Dentatae; Urticella; DQ179319; DQ175580. P. hyalina Fenzl—Monro et al. 3284 (BM); Belize; Dentatae; Urticella; DQ179331; DQ175586. P. imparifolia Wedd.—Grayum 9775 (BM); Costa Rica; Heterophyllae; Urticella; DQ179270; —. P. inaequalis Wedd.—Broadway 9268 (BM); Trinidad; Dentatae; Urticella; DQ179304; DQ175552. P. irrorata Donn.Sm.—Wendt et al. 3919 (BM); Mexico; Integrifoliae; Urticella; DQ179294; DQ175535. P. aff. japonica (Maxim.) Hand.-Mazz. 1—Chen 86409 (PE); China; Dentatae; Achudemia?; —; DQ175574. P. japonica (Maxim.) Hand.-Mazz. 2—Robinson & Kloss 58 (BM); Indonesia; Dentatae; Achudemia; —; DQ175618. P. jayaensis A.K. Monro—Sands 7049 (BM); Indonesia: Irian Jaya; Dentatae; Urticella; DQ179334; —. P. johniana Stapf—Edwards 4240 (BM); Indonesia: Irian Jaya; Integrifoliae; Urticella; DQ179340; —. P. johnsii A.K. Monro—Barker et al. 2 (BM); Indonesia: Irian Jaya; Dentatae; Urticella; DQ179335; —. P. johnstonii Oliver—Taylor 1076 (BM); Kenya; Dentatae; Urticella; DQ179275; —. P. krugii Urb.—Wagner 964 (BM); Puerto Rico; Heterophyllae; P.; DQ179315; DQ175581. P. lacorum P. van Royen—Shea 71031 (BM); Indonesia: Irian Jaya; Integrifoliae; Urticella; DQ179339; —. P. lapestris Chew ex A.K. Monro—Johns 9979 (BM); Indonesia; Integrifoliae; Pilea; DQ179341; DQ175598. P. latifolia Wedd.—Dressler 5884 (MO); Panama; Dentatae; Urticella; —; DQ175594 P. lindeniana Wedd.—Wright 1449 (BM); Cuba; Integrifoliae; Urticella; DQ179314; DQ175547. P. lippioides Killip—Killip & Smith 15559 (BM); Colombia; Dentatae; Urticella; —; DQ175616. P. longicaulis Hand.-Mazz.—Báise Expedition 01909 (PE); China; Dentatae; Tetrameris; DQ179363; DQ175611. P. longifolia Baker—Hildebrandt 4044 (BM); Madagascar; Integrifoliae; Urticella; DQ179318; DQ175553. P. lucida Blume—Adams 11432 (BM); Jamiaca; Heterophyllae; Urticella; DQ179298; —. P. magnicarpa A.K. Monro—Monro 3532 (BM); (Panama); Heterophyllae; Urticella; DQ179277; —. P. manniana Wedd.—Monod 12039 (BM); Sao Tomé; Integrifoliae; Urticella; DQ179276; —. P. melastomoides Wedd.—Beaman 12229 (BM); Indonesia: Irian Jaya; Dentatae; Urticella; DQ179345; DQ175596. P. mexicana Wedd.—Monro 3512 (BM); Panama; Dentatae; Urticella; DQ179278; DQ175579. P. microphylla (L.) Liebm.—Monro 2716 (BM); Belize; Integrifoliae; Pilea; DQ179279; —. P. sp. ‘Nepal’—RBGE 19892544 (E); (Nepal, cultivated at RBGE); Heterophyllae; Urticella; DQ179344; DQ175601. P. nigrescens Urb.—Morley & Whitefoord 543 (BM); Jamaica; Dentatae; Urticella; DQ179301; DQ175582. P. notata—C.H. Wright; Chow 76038 (BM); China; Dentatae; Urticella; —; DQ175615. P. nummularifolia (Sw.) Wedd.—Whitefoord 7312 (BM); Dominica; Dentatae; Urticella; DQ179317; DQ175561. P. nummularifolia (Sw.) Wedd.—Monro 3989 (BM); Peru; Dentatae; Urticella; DQ179316; DQ175588. P. pansamalana Donn. Sm.—Méndez G. 7731 (BM); Mexico; Heterophyllae; Urticella; DQ179296; DQ175533. P. parietaria (L.) Blume—Grayum 5155 (BM); Costa Rica; Integrifoliae; Urticella; DQ179273; —. P. pelonae Urb. & Ekman—Howard 9015 (BM); Dominican Republic; Integrifoliae; Urticella; DQ179327; DQ175540. P. cf. peltata Hance—Tsang 21849 (BM); China; Dentatae; Urticella; DQ179338; —. P. peperomiifolia Liebm.—D'Arcy 295 (BM); Virgin Islands; Integrifoliae; Pilea; DQ179281; DQ175569. P. peperomioides Diels—Monro 4182 (BM); (China, cultivated in UK); Integrifoliae; Urticella; DQ179350; DQ175605. P. peplioides (Gaudich.) Hooker—Tanaka 1744 (BM); Taiwan; Integrifoliae; Dimeris; DQ179342; —. P. sp. ‘Peru 2’—Monro 3991 (BM); Peru; Integrifoliae; Urticella; —; DQ175590. P. sp. ‘Peru 3’ —Monro 4009 (BM); Peru; Dentatae; Urticella; DQ179328; DQ175560. P. sp. ‘Peru 5’—Monro 4015 (BM); Peru; Heterophyllae; Urticella; DQ179290; DQ175595. P. sp. ‘Peru 6’—Monro 4017 (BM); Peru; Dentatae; Urticella; DQ179286; —. P. sp. ‘Peru 9’ —Monro 4032 (BM); Peru; Dentatae; Urticella; DQ179288; DQ175591. P. plataniflora C. H. Wright—Gressitt 509 (BM); Japan; Integrifoliae; Urticella; DQ179349; DQ175599. P. pleuroneura Donn. Sm.—Forther 10110 (BM); Guatemala; Heterophyllae; Pilea; DQ179297; DQ175532. P. plumulosa A.K. Monro—Hammel 14645 (MO); Panama; Dentatae; Pilea; DQ179295; —. P. pseudonotata C.J. Chen—Chen 95398 (PE); China; Dentatae; Tetrameris; DQ179360; DQ175613. P. aff. pteridophylla A.K. Monro—Hammel & Merello 15570 (MO); Mexico; Dentatae; Pilea; —; DQ175534. P. pubescens Liebm.—Monro 2663 (BM); Belize; Heterophyllae; Urticella; DQ179325; DQ175558. P. purulensis Donn. Sm.—Forther et al. 10120 (BM); Guatemala; Heterophyllae; Urticella; DQ179272; DQ175555. P. quadrata A.K. Monro—Monro 3531 (BM); Panama; Heterophyllae; Urticella; DQ179271; —. P. quercifolia Killip—Torres & Ramírez 13348 (BM); Mexico; Dentatae; Urticella; DQ179320; DQ175557. P. racemosa (Royle) Tuyama—Ho et al. 2686 (BM); China; Heterophyllae; Achudemia; DQ179347; DQ175602. P. receptacularis C.J. Chen—Shennong Jia Expedition 32530 (PE); China; Dentatae; Lecanthoides; DQ179362; DQ175612. P. rhizobola Miq.; Mexia 5365 (BM); Brazil; Heterophyllae; Urticella; DQ179287; —. P. rhombea Liebm.—Hawkes & García-Barriga 82 (BM); Colombia; Integrifoliae; Urticella; DQ179274; DQ175549. P. rivularis Wedd. Tanzania—Bruce 935 (BM); Tanzania; Dentatae; Urticella; DQ179358; DQ175606. P. rivularis Wedd. Comoros Islands—Hildebrandt & Rensch 1685 (BM); Comora Islands; Dentatae; Urticella; —; DQ175607. P. rotundinucula Hayata—Wang 130 (BM); China; Dentatae; Urticella; DQ179356; —. P. rubriflora C. H. Wright—Chen 4944 (PE); China; Dentatae; Tetrameris; —; DQ175610. P. rufa Wedd.—Webster & Wilson 5147 (BM); Jamaica; Heterophyllae; Urticella; DQ179299; DQ175578
