| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2 Department of Biological Sciences, Eastern Illinois University, Charleston, Illinois 61920 USA; 3 Department of Biological Sciences, Washington State University, Pullman, Washington 99164 USA; 4 Department of Biological Sciences, Florida International University, Miami, Florida 33199 USA; and 5 Jardin Acclimacion de La Orotava, Calle Retema, La Orotava, Tenerife, Canary Islands, Spain
Received for publication February 5, 1999. Accepted for publication April 11, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: character evolution • Crassulaceae • matK • phylogenetics
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Members of the family are leaf-succulent, usually herbaceous, and often have five-parted, radially symmetrical flowers with two whorls of five stamens each. The family inhabits primarily semiarid habitats and is nearly cosmopolitan in distribution, with centers of diversity in Mexico, southern Africa, Macaronesia, and the Himalayas. The family has long been considered a natural group (e.g., Schönland, 1891
; Berger, 1930
), and recent molecular phylogenetic analyses of the angiosperms indicate that the family is monophyletic (e.g., Chase et al., 1993
; Soltis et al., 1997
). While Crassulaceae are easily recognized, defining monophyletic groups within the family has been extremely difficult because of extensive diversity in morphology, cytology, and habit. Not only are generic boundaries unclear, but relationships among genera are also poorly understood due to the frequent intergradation of morphological characters among taxa. Moran (1942)
stated that "if too much emphasis were placed on technical characters, the numerous exceptions and intergradations would necessitate the combination of genera until but six or only one genus remained." In part, this morphological complexity may represent recurrent evolution of adaptations to xeric habitats.
The most comprehensive treatment of Crassulaceae is that of Berger (1930)
who recognized 35 genera in six subfamilies (Fig. 1). These subfamilies have been placed into two lineages, a Crassula lineage including the three subfamilies (Crassuloideae, Cotyledonoideae, and Kalanchoideae) found predominantly in southern Africa, and a Sedum lineage including the three subfamilies (Echeverioideae, Sedoideae, and Sempervivoideae) found predominantly in the Northern Hemisphere (‘t Hart and Eggli, 1995
). Within these two lineages, Berger circumscribed subfamilies based primarily on floral morphology. For example, within the Crassula lineage, Crassuloideae include species possessing a single whorl of stamens (haplostemonous) and unfused corollas, whereas Cotyledonoideae and Kalanchoideae include diplostemonous species with fused corollas. Berger distinguished Kalanchoideae from Cotyledonoideae based on the number of floral parts: Cotyledonoideae have five-merous flowers, whereas Kalanchoideae have four-merous flowers. In the Sedum lineage, Sempervivoideae all possess unfused, polymerous flowers, and the Echeverioideae have typically five-merous corollas that are partially to completely fused. Berger's Sedoideae have been described as a "catch-all" taxon (Uhl, 1963
) and include the large genus Sedum, as well as the remaining genera of Crassulaceae not easily placed in other subfamilies. Importantly, Sedoideae comprise taxa that display many morphological features used to circumscribe the other five subfamilies, including haplostemonous androecia, sympetalous corollas, and polymerous flowers.
|
; Sempervivoideae—Uhl, 1961a
; and Sedoideae—Uhl, 1963
). Chromosome studies have supported the monophyly of Crassuloideae (Uhl, 1963
) and demonstrated that Kalanchoideae are a natural group closely related to Cotyledonoideae (Uhl, 1948
). Although informative in some instances, chromosomal data have not been particularly useful for estimating boundaries of other taxa (e.g., Uhl, 1961a, 1963
). For example, Echeverioideae and Mexican Sedum species possess a wide range of chromosome numbers due to rampant polyploidy and/or aneuploidy (Uhl, 1961b, 1963, 1970
).
Sedum, the largest genus of Crassulaceae, is cosmopolitan in distribution and encompasses much of the morphological diversity present in the family as a whole. A diverse array of chromosome morphologies and base chromosome numbers is also present in the genus. Due to this high degree of diversity, Sedum has a sordid taxonomic history. Berger included 
500 species in the genus; subsequent authors have named as many as 32 segregate genera (see ‘t Hart and Eggli, 1995
). While several recent studies using morphology and DNA data have supported these segregate genera as discrete groups, the systematics of Sedum is still problematic and in need of additional investigation.
The central position of Sedum with regard to the evolution of Crassulaceae has been stressed by ‘t Hart (1982)
. In his classification, Sedum is subdivided into three large, geographically defined sections, which are hypothesized to have given rise to the genera endemic to each respective region. However, this hypothesis is yet to be tested rigorously (‘t Hart and Eggli, 1995
). Clearly, to understand fully the systematics of Crassulaceae it will be necessary to define the limits of Sedum and the segregate genera already named, as well as to test more rigorously the monophyly of the lineages proposed by ‘t Hart (1982)
.
Using phylogenetic analyses of cpDNA restriction site data, Ham and ‘t Hart (1998) suggested recognizing two subfamilies, Crassuloideae and the rest of Crassulaceae (Sedoideae), as well as seven clades of "major importance" and concluded that many of the subfamilies proposed by Berger (1930)
are not monophyletic. In addition, their analyses placed the 23 species of Sedum analyzed in five of the seven major clades recovered, clearly illustrating the polyphyly of Sedum. Although their study provided initial phylogenetic insights for Crassulaceae, some caution is warranted. For example, only 19 of the 35 genera of Crassulaceae were sampled, and in many instances, only a single species of each genus was included. Furthermore, many of the nodes recovered, especially the deeper nodes, received bootstrap support below 50%.
Not only are generic boundaries uncertain, but also the number of major groups (e.g., subfamilies) is not yet evident. Berger's (1930)
subfamilies have been shown to be polyphyletic (e.g., Uhl, 1963
; Ham and ‘t Hart, 1998); Thorne (1983)
and Takhtajan (1997)
each suggested recognizing three subfamilies in Crassulaceae (Crassuloideae, Kalanchoideae, and Sedoideae), whereas Ham and ‘t Hart (1998) recognized two (Crassuloideae and Sedoideae sensu lato). However, the topology presented by Ham and ‘t Hart (1998) is consistent with the subfamilies of Thorne (1992) and Takhtajan (1997)
. Thus, despite the use of numerous sources of data, systematic relationships within Crassulaceae remain enigmatic. We have employed comparative sequencing of the chloroplast gene matK to provide a comprehensive, family-level estimate of phylogeny for Crassulaceae. Our goals were to: (1) infer phylogenetic relationships on a broad scale across Crassulaceae and (2) investigate the distribution of several morphological and cytological characters often used to define major groups within Crassulaceae, using our phylogenetic hypothesis as an evolutionary framework.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
classification of Crassulaceae as a guide, taxa were sampled from 30 of 35 recognized genera (Table 1). For the large and systematically complex genera (e.g., Sedum, Crassula, Sempervivum, and Kalanchoe), multiple species were included, and with the exception of Bryophyllum, Thompsonella, and the monotypic Tacitus, at least two species from each genus were sampled. A total of 112 ingroup taxa, representing 33 genera (several of which were not recognized by Berger; see Table 1) of Crassulaceae was sequenced for this study. Recent phylogenetic analyses of the angiosperms (Chase et al., 1993
; Soltis et al., 1997
) indicate that Crassulaceae are a member of Saxifragales and are closely related to Penthorum, Tetracarpaea, and Pterostemon (Soltis and Soltis, 1997
). The latter three taxa were previously sequenced for matK and were used as outgroups.
|
). Leaves of Crassulaceae were desiccated on silica gel for a minimum of 1 wk prior to DNA extraction. Leaf material (0.7 g) was ground in liquid nitrogen and mixed with 5 mL of 4x CTAB (110 mmol/L) buffer; extractions were incubated at 60°C for 2 h.
The chloroplast gene matK is 1550 base pairs long and encodes a maturase used in RNA splicing (Neuhaus and Link, 1987
; Wolfe, Morden, and Palmer, 1992
). matK was chosen for analysis because many studies have documented the utility of this gene for resolving phylogenetic relationships at a variety of taxonomic levels, from closely related species to the family level (e.g., Johnson and Soltis, 1994, 1995
; Soltis et al., 1996
; Hilu and Liang, 1997
; Kron, 1997
).
PCR (polymerase chain reaction) amplification employed the primer combinations trnK-3914F and trnK-psbA-R (Johnson and Soltis, 1994
). Manual sequencing followed Johnson and Soltis (1994, 1995)
and used the sequencing primers trnK-710F, matK-1470R, matK-1470F, and matK-2000R (Johnson and Soltis, 1994
; Soltis et al., 1996
). The same primer combinations, as well as a primer designed specifically for Crassulaceae, matK-1800R (5'-AGT TGA CTC CGT ACA ACB GAA-3'), were used for automated sequencing. Automated sequencing was performed on an ABI 377 automated sequencer following the general methods outlined in Soltis and Soltis (1997)
, and employed the PRISM Ready Reaction Dye Deoxy Terminator Cycle Sequencing Kit (Applied Biosystems, Inc., Foster City, California, USA).
Phylogenetic analyses
Sequences in this matrix range from 1121 to 1145 bp in length and provide a data set of 1202 bp after alignment. Alignment was easily accomplished visually; there are occasional indels in matK, all in multiples of three. Indel length ranged from 3 to 12 bp (Table 2). All gap characters ("-") were scored as missing data ("?"), rather than a fifth character; following parsimony analyses of only base substitutions, the phylogenetic distribution of indels was explored by plotting the indels onto the shortest trees.
|
). Large data sets can pose special problems during phylogenetic analyses. To analyze as much tree space as possible, initial searches were conducted using 1000 replicates with RANDOM taxon addition, NNI branch swapping, and MULPARS, with five trees saved per replicate; all characters were equally weighted. This strategy allowed us to obtain a pool of "starting trees." The shortest trees from these initial searches were used as starting trees for subsequent searches with TBR branch swapping and saving a maximum of 5000 minimum-length trees. Relative support for clades was assessed by using bootstrap analyses (Felsenstein, 1985
), with 5000 replicates of fast bootstrapping as implemented in PAUP* (Swofford, 1998
). Fast bootstrapping has recently been shown to be a slightly more conservative estimate of internal support than full heuristic bootstrap analyses (Mort et al., 2000); this approach is well suited for analysis of large data sets.
Character evolution
Berger (1930)
defined subfamilies of Crassulaceae primarily based on three floral features, haplostemonous androecia, sympetalous corollas, and polymerous flowers, in concert with biogeography. The evolution of these floral traits was investigated by tracing the character states for terminal taxa onto the strict consensus of the shortest trees using MacClade (Maddison and Maddison, 1992
). Biogeography was similarly investigated by plotting the distribution of each species onto the strict consensus tree. Morphological data and biogeographic distributions for taxa included in our analyses were obtained from previous studies (Praeger, 1921
; Quimby, 1971
; Clausen, 1975
; Spongberg, 1978
; Stephenson, 1994
; ‘t Hart and Eggli, 1995
). Chromosomal evolution was similarly investigated. Base chromosome numbers were obtained from the extensive data collected by Uhl (1948, 1961a, b, 1963)
, as well as from numbers reported elsewhere (Baldwin, 1935, 1937
; Moore, 1973
). A simplified topology for Crassulaceae presenting only the major relationship among clades was constructed for analysis of chromosome evolution. This approach was used because not all of the species included in the phylogenetic analyses have chromosome numbers reported. In other cases, close relatives of taxa included in our analyses have base chromosome numbers that have been reported. Included on this topology as sister to Crassula is Tillaea, which was not analyzed in the present study, but was strongly supported as sister to the Crassula clade by Ham and ‘t Hart (1998). Base chromosome numbers were then traced onto this simplified topology to provide initial insights into chromosomal evolution in Crassulaceae. When several alternative base chromosome numbers are apparent, they are provided. The distribution of polyploidy was also explored using this summary topology for Crassulaceae.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
Distribution of indels
To align the matK sequences of 112 ingroup taxa and three outgroups, it is necessary to infer 16 indels (Table 2). Previous studies (e.g., Johnson and Soltis, 1994, 1995
; Plunkett et al., 1996, 1997
; reviewed by Soltis and Soltis, 1998
) indicate that indels in matK are often phylogenetically informative. Comparison of the taxonomic distribution of indels to the results of our phylogenetic analyses suggests that ten of these indels are phylogenetically informative (indels A, B, D, E, J, K, L, M, O, P); five are autapomorphic (indels F, G, H, I, and N); and one (indel C) is informative, but appears to be homoplasious. The distribution of indels, other than those that are autapomorphic, is shown by plotting them onto the strict consensus tree (Fig. 2).
Supporting the monophyly of Crassulaceae is a 9-bp (base pair) deletion (indel B). The monophyly of the genus Crassula is supported by a 12-bp deletion (indel J). Another insertion found in Crassula (indel C) is considered homoplasious. This 6-bp insertion is shared by all species of Crassula and Umbilicus included in our analyses; these taxa are well separated from one another in the topology of Crassulaceae (Fig. 2). When the taxa possessing indel C are constrained to form a clade and phylogenetic analyses are repeated (see above), the minimum-length trees obtained are 2649 steps in length. This large increase in tree length (28 steps) supports the conclusion that indel C is homoplasious. As noted above, analyses of matK sequences do not recover the Telephium clade of Ham and ‘t Hart (1998). Rather, the three component subclades of taxa representing the Telephium clade form a polytomy (Fig. 2); hence our data do not contradict the potential monophyly of this clade. Significantly, a 6-bp insertion (indel D) is shared by all members of the Telephium clade, and if indels are included in phylogenetic analyses, the Telephium clade is recovered (Fig. 2), but receives low bootstrap support (<50%). Within the Telephium clade, a 3-bp insertion (indel E) is shared by Orostachys and Sinocrassula. Indel O, a 6-bp deletion, is shared by all members of Crassulaceae sampled except Crassula and two clades of Sedum (S. fusiforme, S. lancerotense, and S. nudum; and S. oryzafolium—S. urvillei). The sequence of six nucleotides in indel O in these two Sedum clades is identical (although the fourth position of the indel is polymorphic), but differs from the corresponding 6-bp sequence observed for Crassula (Table 2). Based on the shortest trees and base composition of the indel, we infer that deletion "O" occurred following the divergence of the Crassula clade; this was followed by a reinsertion at the same position in these two clades of Sedum. An additional 6-bp insertion (indel P) is unique to the Sedum fusiforme, S. lancerotense, S. nudum clade. Other indels that support major clades revealed in our analyses include indel L, which is shared by all members of the Acre clade, and indels A and M, which support the monophyly of the Aeonium clade.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Morgan and Soltis, 1993
) and 18S rDNA sequences (Soltis and Soltis, 1997
; Soltis et al., 1997
) indicate that the Crassulaceae are part of a Saxifragales clade. Our analyses of matK sequences for 112 species likewise indicate that Crassulaceae are a strongly supported monophyletic group (Fig. 2). Also supporting the branch leading to Crassulaceae is a 9-bp deletion (Table 2, indel B). Given these molecular data, as well as the numerous morphological features that unite the family, there is little dispute that Crassulaceae are monophyletic. Within Crassulaceae is a well-supported basal split between the Crassula clade and the rest of the family. A similar basal dichotomy was noted by Ham and ‘t Hart (1998), and these two clades correspond to the two subfamilies that they recognize: Crassuloideae and Sedoideae. We will follow this demarcation of the family into Crassuloideae and Sedoideae herein (Fig. 2). Within the Sedoideae clade, a well-supported dichotomy exists between the Kalanchoe clade and the remainder of the clade. A number of additional subclades (i.e., Leucosedum, Acre, and Aeonium clades) that largely correspond to those recovered by Ham and ‘t Hart (1998) are present in the remainder of Sedoideae. Strong support for a relationship among the Aeonium, Acre, Leucosedum, and Sempervivum subclades is also apparent (Fig. 2).
The taxonomic composition and support for these clades will be discussed in detail below. For clarity, we applied to clades the same names that have been used by Ham and ‘t Hart (1998); however, the monophyly of several of these clades (i.e., Telephium and Sempervivum) is not strongly supported by either matK or cpDNA restriction site analyses (see Ham and ‘t Hart, 1998).
Crassula clade
The matK sequence data indicate that the first-branching lineage in Crassulaceae is Crassula, in agreement with the results from cpDNA restriction site analyses (Ham and ‘t Hart, 1998). Crassula, along with four other genera (Dinacria, Pagella, Rochea, and Vauanthes), were placed by Berger (1930)
in Crassuloideae. More recently Tölken (1977, 1985)
placed all five of these genera in a broadly defined Crassula, or Crassula s.l. (sensu lato) that comprises 200 species distributed primarily in southern Africa. Morphologically, Crassula s.l. differs from the remainder of Crassulaceae by possessing haplostemonous flowers.
We sampled nine species of Crassula, including a species formerly placed within Rochea. Samples of species placed within the other closely related genera (Pagella, Dinacria, and Vauanthes) could not be obtained. Based on the samples employed, our sequence data strongly support the monophyly of Crassula. Also supporting the monophyly of this clade is a unique 12-bp deletion (indel J). The position of Crassula as sister to the remainder of Crassulaceae is also supported by a 6-bp deletion (indel O) shared by all Crassulaceae except Crassula and two small clades of Sedum (see above for discussion); Crassula shares the same base composition as the outgroup taxa in this 6-bp region.
Tillaea, which Berger (1930)
recognized as a section of Crassula, was not included in our analyses. The analyses of Ham and ‘t Hart (1998) place Tillaea as a lineage distinct from Crassula; cpDNA divergence between Tillaea and Crassula is greater than the divergence among many other taxa of undisputed generic status. However, a long branch does not necessarily imply that Tillaea is distinct from Crassula. As currently circumscribed, Tillaea includes 20 species, with a nearly worldwide distribution. This genus of diminutive, semi-aquatic plants differs from Crassula in a number of morphological characters, including fruit dehiscence and ovule number (‘t Hart and Eggli, 1995
). However, Schönland (1916, cited in Spongberg, 1978
) concluded that "no sharp line" can be drawn between section Tillaea and other sections of Crassula. Although we obtained plant material of T. erecta, attempts to obtain DNA from this material were unsuccessful. Thus, it remains to be demonstrated whether Tillaea is distinct from Crassula.
Kalanchoe clade
Our analyses reveal a strongly supported Kalanchoe clade that includes six genera, Adromichus, Tylecodon, Cotyledon, Bryophyllum, Kitchingia, and Kalanchoe (Fig. 1). This Kalanchoe clade encompasses Berger's Kalanchoideae, which consists of three genera (Bryophyllum, Kitchingia, and Kalanchoe), as well as three genera of Berger's Cotyledonoideae (Adromichus, Cotyledon, and Tylecodon). Hence, this single clade illustrates well the lack of correspondence between monophyletic groups and the six traditionally recognized subfamilies. The limits of Bryophyllum, Kitchingia, and Kalanchoe have been much debated among systematists, with some recognizing three genera (e.g., Berger, 1930
) and others two genera (e.g., ‘t Hart and Eggli, 1995
) or even a single genus (Baldwin, 1938
). Analyses of matK sequence data place Kitchingia and Bryophyllum within Kalanchoe with strong support, suggesting that it may be more appropriate to recognize the single genus Kalanchoe. However, greater taxon density is needed to resolve fully the boundaries of these genera.
Representatives of the Kalanchoe clade all possess flowers with fused corollas and are mostly distributed in southern Africa. The close relationship between Kalanchoe and certain genera of Cotyledonoideae, such as Adromischus and Cotyledon, was suggested by Baldwin (1938)
, who, based on cytotaxonomy, hypothesized that the Kalanchoe subfamily was an allopolyploid derivative involving "Cotyledon-like" and "Crassula-like" ancestors. Uhl (1948)
similarly suggested that Kalanchoideae are polyploids derived from "Cotyledon-like" ancestor(s), but he excluded the role of a "Crassula-like" taxon. Uhl further suggested that Adromischus and Cotyledon were closely related to Kalanchoe, but that several other genera (e.g., Umbilicus, Mucizonia, and Pistorinia) placed within the Cotyledonoideae by Berger (1930)
were not closely related to Cotyledon or Adromischus. While the proposed polyploid origin of Kalanchoideae remains to be tested more rigorously, our data do suggest that a Cotyledon-like taxon could have been the maternal parent, contributing the chloroplast genome to Kalanchoideae.
Telephium clade
When only base substitutions are considered, matK sequence data are inconclusive with regard to monophyly of the Telephium clade. However, taxa in the Telephium clade share a 6-bp insertion (indel D, Table 2), and if this indel is coded as an additional character and included in parsimony analyses, the Telephium clade is monophyletic, but receives bootstrap support <50%. This clade was also weakly supported (bootstrap value of 25%) using cpDNA restriction site data (Ham and ‘t Hart, 1998).
Our analyses of matK sequence data resolve and strongly support three subclades within the Telephium clade: (1) Umbilicus, a genus placed by Berger (1930)
in Cotyledonoideae; (2) Orostachys and Sinocrassula, both of Berger's Sedoideae, and Hylotelephium, a segregate genus of Sedum; and (3) Phedimus, still another segregate of Sedum. Hence, this clade again illustrates well the problems of the traditional delineation of subfamilies in Crassulaceae.
Umbilicus is primarily Mediterranean in distribution and, like other members of Berger's Cotyledonoideae, has five-parted flowers with fused corollas. However, this genus has a base chromosome number of x = 24, whereas other Cotyledonoideae have x = 9. In addition, the chromosome morphology and biogeography of Umbilicus differ from Cotyledonoideae (Uhl, 1948
). Although the closest relative of this genus within the Telephium clade is unresolved, our analyses support Uhl's conclusion that Umbilicus is not allied with other Cotyledonoideae.
A subclade within the Telephium clade comprising Hylotelephium, Sinocrassula, and Orostachys is also strongly supported. Within this subclade a sister-group relationship between Orostachys and Sinocrassula is supported by a 3-bp insertion (indel E) that is unique to these genera. Taxa in this subclade share a primarily Asian distribution, ranging from southwestern China (Sinocrassula) to central Asia (Orostachys and Hylotelephium). Orostachys is particularly noteworthy in that most species were originally described as members of Cotyledon, but various authors have placed these taxa within Sempervivum, Sedum, Umbilicus, and Crassula (reviewed by Uhl, 1948
).
Sempervivum clade
Ham and ‘t Hart's analyses of cpDNA restriction sites recovered a clade comprising Sempervivum and several Eurasian Sedum species, including S. sediforme, S. mooneyi, and S. assyriacum. Although their Sempervivum clade was only weakly supported (bootstrap of 17%), it suggested a close relationship between Sempervivum and Sedum section Rupestre, a relationship previously suggested by Jacquin (1770)
and Uhl (1961a)
. While our analyses do not contradict the monophyly of the Sempervivum clade, it is not resolved on our strict consensus topology.
Jovibarba and Sempervivum form a clade (bootstrap of 100%); and each is monophyletic (bootstrap of 98 and 99%, respectively). Berger (1930)
considered Jovibarba to be a section of Sempervivum. In contrast, Parnell (1991)
recognized Jovibarba as a genus distinct from Sempervivum. Jovibarba and Sempervivum have similar chromosome morphology and overlapping base chromosome numbers (Uhl, 1961a
) suggesting that they are closely related, but these data are inconclusive regarding whether they should be treated as distinct genera.
Leucosedum clade
Taxa in the Leucosedum clade are distributed throughout the arid southwestern United States, Mexico, and Europe. This biogeographically widespread clade is weakly supported (bootstrap <50%) by our analyses. In contrast, it is one of the most strongly supported clades (bootstrap of 82%) in analyses of cpDNA restriction sites (Ham and ‘t Hart, 1998). In part, this difference in support may be due to differences in taxon sampling. For example, our data set includes 12 species with nearly equal sampling of Mexican and European taxa, whereas that of Ham and ‘t Hart sampled 11 species, primarily from Europe.
Although the Leucosedum clade does not receive strong support in our analyses, two subclades were recovered, each with moderate to high bootstrap support. One subclade includes Dudleya, the monophyly of which is strongly supported (bootstrap of 100%), and its sister taxa Parvisedum followed by Sedum gracile (bootstrap of 96%). In addition, a subclade of mostly European Sedoideae, including Rosularia, several species of Sedum (e.g., S. dasyphyllum, S. lydium, and S. hispanicum), and Sempervivella was recovered, with bootstrap support of 66%. Rosularia appears polyphyletic, which was also suggested by Ham and ‘t Hart (1998).
Acre clade
This clade comprises Echeverioideae and species of Sedum from around the world (Fig. 2), including S. nudum, S. fusiforme, and S. lancerotense from Macaronesia; S. furfuraceum, S. burito, and S. clavatum from Mexico; S. urvillei, and S. oryzafolium from Asia; and S. multiceps from Europe. Placed within this clade are taxa that are among the most variable and confusing taxa based on chromosome data, including Pachyphytum, Graptopetalum, Echeveria, Lenophyllum, and the Mexican species of Sedum (see below).
The support for the Acre clade in our analyses is low (bootstrap of 66%) when compared to the analyses of Ham and ‘t Hart (bootstrap of 100%). This is likely attributable to the small taxon sampling of Ham and ‘t Hart compared to the present study (13 vs. 32 taxa). Ham and ‘t Hart included a single species of each of only three genera of Echeverioideae, whereas our data set includes eight genera of Echeverioideae (four of five genera recognized by Berger), four additional genera described since Berger's treatment (Dudleya, Graptopetalum, Thompsonella, and Tacitus), and a greater density of Mexican Sedum species. The topology recovered in our analyses is not in conflict with that of Ham and ‘t Hart (1998), but contains a number of unresolved polytomies. That matK provides few characters to resolve many of the relationships within this clade might indicate that this clade is of relatively recent origin. Also relevant to this low resolution is the substantial intergeneric and inter-subfamilial crossability found among many of the taxa in the Acre clade (Spongberg, 1978
; Uhl, 1989, 1994
). It is, therefore, likely that frequent chloroplast exchanges have occurred among members of this clade, affecting a plastid-based phylogeny.
Several well-supported subclades within the Acre clade were recovered by our analyses. There is strong support (bootstrap of 100%) for a clade of Sedum species (S. nudum, S. fusiforme, and S. lancerotense) from Macaronesia and Africa and a clade comprising S. ternatum and S. hemsleyanum (bootstrap of 100%), both native to North America. Also well supported is the monophyly of Pachyphytum (bootstrap of 99%). The Acre clade comprises approximately one-third of the taxonomic diversity of Crassulaceae, and while our study expands upon the phylogenetic hypothesis of Ham and ‘t Hart (1998), much more phylogenetic work is needed to resolve fully the generic boundaries and relationships within this large clade.
Aeonium clade
This clade comprises four genera of Sempervivoideae (Aeonium, Aichryson, Greenovia, and Monanthes) that are largely endemic to Macaronesia. These genera were placed by Berger in Sempervivoideae because, like Sempervivum, these taxa all possess, in varying degrees, polymerous flowers. Our analyses strongly support the monophyly of the Aeonium clade and place two African species of Sedum, S. modestum and S. jaccardianum, as sister to the four Macaronesian genera. In addition to analyses of base substitutions, two indels (A and M) support the monophyly of this clade. The sister relationship between the Macaronesian genera and some African Sedum species was suggested by Uhl (1961a)
and has recently been supported by cpDNA restriction sites (Ham and ‘t Hart, 1998), as well as RAPD analyses and ITS sequence data (Mes, 1995
). Analyses of matK sequences also indicate that the Macaronesian genera are not closely related to the Sempervivum clade, as proposed by Berger (1930)
.
Most authors (e.g., Lems, 1960
; Uhl, 1961a
; Liu, 1989
; Mes, Wijers, and ‘t Hart, 1997
; Ham ‘t Hart, 1998) agree that the four Macaronesian genera are closely related, but the relationships among and limits of these genera have been debated. Aichryson differs from the rest of these genera in having base chromosome numbers of x = 15, 16, or 17, and, with one exception, the genus comprises annuals. The other genera, with one exception, Monanthes icterica, are perennials. Monanthes shares a basic number of x = 18 with Aeonium and Greenovia, but differs by having large, petaloid nectary scales. Greenovia differs from Aeonium in several respects, including 20–35 merous flowers (vs. 12–18 merous flowers), carpels partially sunken into the receptacle, and placentation type.
Mes (1995)
suggested that Greenovia is derived from within Aeonium. However, there is little support for these conclusions as measured by their bootstrap and decay analyses. Also, these conclusions must be considered tentative as the authors included only a single species of Greenovia. The placement of Monanthes icterica (annual habit) has also been debated. Recent phylogenetic analyses (Mes, Wijers, and ‘t Hart, 1997
) indicate that this species is imbedded within a clade of Aichryson species.
Bootstrap support for relationships within the Aeonium clade is generally low. However, there is moderate support for the association of Monanthes icterica with species of Aichryson (bootstrap of 77%). In addition, several small clades are resolved and receive moderate bootstrap support. The results of our analyses are in agreement with a recent radiation of the Aeonium clade in Macaronesia. Phylogenetic analyses of the Aeonium clade employing both additional taxa and gene sequences are currently in progress (Mort et al., unpublished data).
Character evolution
Berger (1930)
relied largely on three floral characters to define subfamilies in Crassulaceae: haplostemonous androecia, sympetalous flowers, and polymerous flowers. It has been suggested elsewhere (e.g., Ham and ‘t Hart, 1998) that these characters have evolved independently a number of times in Crassulaceae. However, the distribution of these characters has not been investigated in a broad phylogenetic context. Therefore, we traced the distribution of these features onto our strict consensus cladogram (Fig. 3) using MacClade (Maddison and Maddison, 1992
). We chose to use the strict-consensus topology because we recovered a large number of minimum-length trees (5000 trees). However, this approach is not without problems, for the strict-consensus tree is a summary of all trees obtained and does not reflect the exact relationships portrayed by any one of the minimum-length trees. Since our goal is to examine the distribution of floral features, and not necessarily the patterns of evolution, we feel that the strict-consensus tree best summarizes the overall results of our phylogenetic analyses. In addition, biogeography was investigated by plotting broadly defined distributions for terminal taxa onto this same cladogram.
|
Crassuloideae. A haplostemonous androecium has arisen once and is confined to the Crassula clade (Fig. 3). However, haplostemy has been reported in Sedum (see Uhl, 1963
) and Hypagophytum (Uhl, 1961a
). Species of these genera that display this floral feature were not included in our analyses. Thus, it remains uncertain whether this condition truly arose only once. Furthermore, the ontogeny and homology of this character have yet to be investigated. Additional taxonomic sampling, as well as detailed morphological investigation of this character, are necessary to address fully the evolution of haplostemy.
Sympetaly
Sympetaly, in concert with biogeography and the number of floral parts, was used by Berger (1930)
to define three distinct subfamilies: Kalanchoideae, Cotyledonoideae, and Echeverioideae. Sympetaly, however, is not entirely confined to these taxa, but is also present in several genera of Berger's (1930)
Crassuloideae and Sedoideae, as well as several different lineages of Sedum (Uhl, 1963
). Taxa displaying any fusion of petals were scored as sympetalous. This strategy likely underestimates the complexity of floral fusion as not all fused corollas are necessarily the end result of the same developmental pathway (e.g., Erbar, 1991
). However, this approach illustrates the distribution of this character and serves as a starting point to further phylogenetically based investigations of the evolution of sympetaly in Crassulaceae.
The results of our analyses indicate that sympetaly has arisen in a minimum of five separate lineages of Crassulaceae with no apparent reversals (Fig. 3a). A single origin of sympetaly can be hypothesized for the Kalanchoe clade. Additional origins are inferred for Umbilicus (Telephium clade), as well as two separate origins in the Leucosedum clade. Within the Acre clade sympetaly is widespread. Our analyses indicate a minimum of two origins of sympetaly in this clade; however, many of these relationships are unresolved or weakly supported.
Not included in our analyses are several taxa known to display sympetaly (e.g., Rochea). Therefore, it is possible that this character is even more widespread in Crassulaceae than suggested by our analyses. We conclude that sympetaly is a widespread condition in Crassulaceae that has evolved in parallel in at least five lineages and represents a grossly homoplasious character.
Polymerous flowers
The typical number of floral parts in Crassulaceae, as well as the outgroups, is four or five. Members of our Crassuloideae, as well as the early-branching taxa in Sedoideae, possess five-parted flowers. MacClade reconstructs five-parted flowers as the ancestral floral condition in Crassulaceae. We define polymerous flowers as those flowers possessing at least twice this number of floral parts (i.e., ten or greater). Species with polymerous flowers were assigned by Berger (1930)
to Sempervivoideae. Our analyses indicate that this subfamily is polyphyletic; two genera (Jovibarba and Sempervivum) form a well-supported clade that is removed from the remainder of polymerous taxa (Aeonium clade). Thus, polymerous flowers appear to have evolved in two separate and well-supported lineages (Fig. 3a).
Biogeography
Crassulaceae are nearly cosmopolitan in distribution due to several widespread taxa (e.g., Tillaea), but the family has discrete centers of taxonomic diversity, including southern Africa, Mexico, Macaronesia, the Mediterranean region, and the Himalayas (Spongberg, 1978
). Within each of these centers of diversity are found genera of Crassulaceae endemic to that region. In addition, in many of these regions (especially those in the Northern Hemisphere), there are species that have been assigned to Sedum. As noted above, ‘t Hart (1982)
has hypothesized that Sedum is a geographically widespread, paraphyletic genus. He further hypothesized that certain species of Sedum appear to have given rise to other genera in the family that are confined to specific biogeographic regions.
Ham and ‘t Hart (1998) noted two major biogeographic inferences from their cpDNA-based tree. First, Crassulaceae appear to have originated in either southern Africa or the Mediterranean region. Second, Crassulaceae have reached the Macaronesian islands a minimum of three times. Furthermore, review of their topology indicates that Sedum species are placed in five of the seven major clades removed.
Our analyses also place species of Sedum in five of the seven major clades recovered (Fig. 3); only the Crassula and Kalanchoe clades lack Sedum species. Thus, Sedum as currently defined is grossly polyphyletic and in need of taxonomic revision. As predicted by ‘t Hart (1982)
, many of the species of Sedum included in our analyses generally group together with genera that are confined to a single biogeographical region (Fig. 3b). For example, S. modestum and S. jaccardianum from north Africa are sister to the genera endemic to Macaronesia, and the Mexican species of Sedum are placed in a clade of genera endemic to this same region (viz. Pachyphytum and Lenophyllum). However, it is important to note that matK is maternally inherited. Because of the potential for hybridization and chloroplast capture, additional sequences from the nuclear genome (or other sources of data) should be analyzed before too many generalizations concerning Sedum are made.
As noted, phylogenetic analyses of matK sequences strongly support the Crassula clade (Crassuloideae) as sister to a large Sedoideae clade; within the latter the Kalanchoe clade is sister to the remaining Sedoideae. Taxa comprising the Crassula and Kalanchoe clades are confined to southern Africa and Madagascar (Fig. 3b), a distribution that suggests a southern African origin of Crassulaceae. However, to test this hypothesized southern African origin of the familiy, additional analyses employing a broader sampling of taxa from Saxifragales is needed. Regardless of the origin of the family, it appears that the first major diversification event occurred in southern Africa. We hypothesize that from southern Africa the family spread through the Mediterranean region and into eastern Europe and Asia (e.g., Sempervivum and Leucosedum clades), and species from northern Africa dispersed to Macaronesia, where they subsequently diversified (Aeonium clade). North America was reached by Crassulaceae at least two times: once by the ancestor of the clade comprising Parvisedum and Dudleya, and at least once by the ancestor of a core subclade in the Acre clade. These biogeographic patterns are in agreement with those described by Ham and ‘t Hart (1998).
Included in the Macaronesian Crassulaceae flora are four genera of Sempervivoideae (Aeonium, Aichryson, Greenovia, and Monanthes), several species of Sedum, and one species of Umbilicus (Santos-Guerra, 1983
; Bramwell and Bramwell, 1990
). Ham and ‘t Hart (1998) suggest, based on cpDNA restriction site analyses, that Crassulaceae arrived in Macaronesia a minimum of three times. Likewise, our analyses indicate that Crassulaceae have reached the Macaronesian islands three times (Fig. 3b): once by the ancestor of a clade of three species of Sedum (S. nudum, S. lancerotense, and S. fusiforme), once by the ancestor of the core of the Aeonium clade, and by Umbilicus. It is noteworthy that the progenitor of the Macaronesian Sedum species appears to be from Mexico, whereas Umbilicus and the core of the Aeonium clade is of northern African/Mediterranean origin.
Chromosome numbers
Chromosome morphology and base chromosome numbers have been extensively studied in Crassulaceae (e.g., Baldwin, 1935, 1937
; Uhl, 1948, 1961a, 1963, 1995
). Although chromosome numbers are highly variable in most genera and a few species, the patterns of chromosomal evolution are not the same in all groups of Crassulaceae, and thus may have "major phylogenetic significance" (Uhl, 1961). Previously published chromosome counts were used to infer the base number for the major clades recovered in our phylogenetic analyses; these numbers were then plotted onto a simplified topology that we produced to examine in a preliminary fashion the distribution and possible evolution of both base chromosome number and polyploidy (Fig. 4).
|
MacClade also reconstructs x = 8 as the base number for the large Sedoideae clade (Fig. 4). Within the Sedoideae clade, the base chromosome number increases to x = 9 in the Kalanchoe clade. Several genera of the Kalanchoe clade (Adromischus, Cotyledon, Tylecodon) have 2x = 18. However, all species of Kalanchoe, Bryophyllum, and Kitchingia have either x = 17 or 18 (or a multiple thereof). Base chromosome number, as well as chromosome morphology suggest that the subclade of Kalanchoe, Bryophyllum, and Kitchingia is of polyploid origin (Baldwin, 1938
) and is likely derived from a tetraploid "Cotyledon-like" taxon (Uhl, 1963
). Our data are in agreement with this hypothesis, suggesting that either the ancestor or some extinct member of the Cotyledon/Tylecodon subclade was the maternal parent of these polyploids. Another base chromosome number increase from x = 8 to x = 12 and higher has occurred in the Telephium clade. Chromosome evolution in this clade is complex and still unclear. The clade comprises three subclades (Fig. 2). The first subclade comprises Hylotelephium, Orostachys, and Sinocrassula; these taxa share a base chromosome number of x = 12. A second subclade corresponds to Phedimus, which has a base number of x = 16. Umbilicus (x = 24) forms the third subclade. Because analyses of matK sequences do not resolve the relationships among these subclades, it is not possible to infer the ancestral chromosome number for this clade. Regardless, the base chromosome numbers of x = 16 and x = 24 are likely the result of polyploidy. However, the origin of x = 12 is unclear. It could represent aneuploid increase from x = 8 or decrease from a polyploid ancestor with x = 16. The Telephium clade, therefore, represents still another episode of polyploidy in Crassulaceae.
Base chromosome numbers for the Sempervivum clade are x = 16, 17, 18, 19, 28. Within this clade, Jovibarba and Sempervivum form a clade and have base chromosome numbers of x = 19 and x = 16–19, respectively. The Sempervivum clade likely represents another instance of polyploid increase from x = 8, coupled with aneuploidy. Another episode of polyploidy is suggested for Sedum section Rupestre, which has x = 28. Relationships within the Sempervivum clade are unclear, and additional work is needed to resolve the cytological evolution of this clade.
The Acre and Leucosedum clades are sister groups; their ancestral base chromosome number is equivocal based on our MacClade reconstruction (Fig. 4). Within the Leucosedum clade most taxa are diploid with 2x = 12, 14 (x = 6, 7). Two subclades are found within this clade (Fig. 2). The first comprises taxa that have base chromosome numbers of either x = 6 or 7, with the exception of S. dasyphyllum, which has x = 14. Likewise, the first branching members of the second subclade have x = 6. However, Dudleya, also in this subclade, has x = 17 (or a multiple thereof), which suggests another polyploid event in Crassulaceae.
Taxa within the Acre clade display a wide array of chromosome numbers, ranging from x = 6 to x = 270 (or greater); polyploidy appears to be widespread in this clade (e.g., Uhl, 1963, 1970, 1993
; Federov, 1969
; Moore, 1973
; Uhl and Moran, 1973
). Because of this common polyploidy, the base number of many genera is often not apparent, which greatly complicates inferences of cytological evolution. However, when examined in a phylogenetic context, it is possible to make several general inferences regarding chromosomal evolution in the Acre clade.
The early-branching members of the Acre clade comprise a number of small clades that are primarily Mexican species of Sedum. These groups are highly variable in base chromosome number. Uhl (1961b)
reported 45 different base numbers for 60 Sedum species from this region; however, x = 10 is most common. In the remainder of the Acre clade polyploidy appears to be very common, and chromosomal evolution is especially complex. For example, species of Pachyphytum have x = 31–33, with polyploids of x = 66 and x = 128 reported (Uhl and Moran, 1973
). The base number for Graptopetalum is x = 30–35, with two species forming polyploids of x = 240 to 275 (Uhl, 1970
). Polyploidy and aneuploidy have likely played a major role in the evolution of this clade.
In the Aeonium clade, Sedum modestum and S. jaccardianum are sister to the Macaronesian Sempervivoideae. These species of Sedum share a base chromosome number of x = 8; the base number for many of the remaining species in the Aeonium clade is x = 18, indicating that the core of this clade is polyploid, apparently derived from diploid ancestors with x = 8.
Summary
Relationships in Crassulaceae have been a focus of study for a number of years, and students of the family have employed a variety of tools to unravel these relationships. Analyses of matK sequence data provide additional insights into the evolution of this family. The results of our analyses are in agreement with those of Ham and ‘t Hart (1998) in that two major clades are recovered: a Crassula clade (Crassuloideae) and a large clade comprising six subclades (Sedoideae). Thorne (1983)
recognized three subfamilies: Crassuloideae, Kalanchoideae (including Cotyledonoideae), and Sedoideae. Our topology is also in basic agreement with this treatment in that within the large Sedoideae clade there are two clades, one corresponding to Thorne's Kalanchoideae, and one (with five subclades) that corresponds to Thorne's Sedoideae. Five of the seven clades of major interest named by Ham and ‘t Hart (1998) are recovered by our analyses: the Crassula, Kalanchoe, Acre, Leucosedum, and Aeonium clades. In many instances these clades are also supported by other sources of data (e.g., cpDNA restriction sites, base chromosome number, biogeography). Two clades recognized by Ham and ‘t Hart (1998), the Telephium and Sempervivum clade, are not contradicted by our analyses, but receive bootstrap support below 50%. If indels are included in our analyses, the Telephium clade is recovered, but no indels support the Sempervivum clade. Phylogenetic relationships are largely unresolved in the Acre and Aeonium clades, possibly suggesting a relatively recent radiation of these lineages.
This study also provides initial insights into character evolution and biogeography in Crassulaceae. Two floral features often used to define subfamilies of Crassulaceae, sympetaly and polymerous flowers, have arisen independently in several lineages. A third floral character, haplostemy, is confined to the Crassula clade. Crassulaceae appear to have arisen in southern Africa, from where the family spread northward into the Mediterranean region. From there, the family spread to Asia/eastern Europe and northern Europe, giving rise to a number of biogeographically confined genera. Two separate lineages of European Crassulaceae subsequently dispersed to North America and underwent substantial diversification. The northern African Crassulaceae subsequently dispersed to the Macaronesian islands where the genera Aeonium, Aichryson, Greenovia, and Monanthes arose. In addition, Crassulaceae reached the Macaronesian islands at least two additional times. Finally, these analyses support ‘t Hart's (1982)
hypothesis that Sedum is a geographically widespread, polyphyletic taxon, with species having close affinities for genera that are confined to a single biogeographic region. However, while the current study includes 29 species of Sedum as well as several genera recently segregated from Sedum (e.g., Hylotelephium, Phedimus), additional taxon sampling, as well as sequence data from the nuclear genome, are needed to address relationships within this genus.
Our studies suggest that the base chromosome number for Crassulaceae was x = 8 with a reduction to x = 7 in Crassula and x = 6, 7 in the Leucosedum clade. Polyploidy is widespread in the family and has played a role in the evolution of seven major clades of Crassulaceae. Three of these clades are exclusively polyploid (Sempervivum clade, and subclades of the Kalanchoe and Aeonium clades), whereas four clades (Crassula, Telephium, Leucosedum, and Acre clades) comprise both diploid and polyploid taxa. Polyploidy has been particularly common and cytological evolution especially complex in the Acre clade.
|
|
|
FOOTNOTES |
|---|
6 Author for correspondence (e-mail: cfmem2{at}eiu.edu
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
———. 1937 The cyto-taxonomy of the Telephium sections of Sedum. American Journal of Botany 24: 126–132
———. 1938 Kalanchoe: the genus and its chromosomes. American Journal of Botany 25: 572–579[CrossRef][ISI]
Berger, A. 1930 Crassulaceae. In A. Engler and K. Prantl [eds.], Die Natürlichen Pflanzenfamilien, 2nd ed. 18A: 352–483
Bramwell, D., and Z. I. Bramwell. 1990 Flores Silvestres de las Islas Canarias. Editorial Rueda, Madrid, Spain
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]
Clausen, R. T. 1975 Sedum of North America North of the Mexican Plateau. Cornell University Press, Ithaca, New York, USA
Doyle, J. J., and J. L. Doyle. 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemistry Bulletin 19: 11–15
Erbar, C. 1991 Sympetaly—a systematic character? Botanische Jahrbuecher fuer Systematik Pflanzengeschichte und Pflanzengeographie 112: 417–451
Federov, A. 1969 Chromosome numbers of flowering plants. Academy of Sciences of the USSR, Komarov Botanical Institute, Leningrad, Russia
Felsenstein, J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791[CrossRef][ISI]
Hilu, K., and H. Liang. 1997 The matK gene: sequence variation and application in plant systematics. American Journal of Botany 84: 830–839[Abstract]
Jacquin, N. J. 1770 Hortus Botanicus Vindobonesis. 1. Vienna, Austria
Johnson, L. A., and D. E. Soltis. 1994 matK DNA sequences and phylogenetic reconstruction in Saxifragaceae sensu stricto. Systematic Botany 19: 143–156
———, and ———. 1995 Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae) using matK sequences. Annals of the Missouri Botanical Garden 82:149–175
Kron, K. A. 1997 Phylogenetic relationships of Rhododendroideae (Ericaceae). American Journal of Botany 84: 973–980[Abstract]
Lems, K. 1960 Botanical notes from the Canary Islands II. The evolution of plant life forms in the islands: Aeonium. Ecology 41: 1–17
Liu, H. Y. 1989 Systematics of Aeonium (Crassulaceae). Special Publication Number 3, National Museum of Natural Science, Taichung, Taiwan
Maddison, W. P., and D. R. Maddison. 1992 MacClade: analysis of phylogeny and character evolution. Sinauer, Sunderland, Massachusetts, USA
Mes, T. H. M. 1995 Origin and evolution of the Macaronesian Sempervivoideae. Ph.D. dissertation, Utrecht University, Utrecht, The Netherlands
———, G. J. Wijers, and H. ‘t Hart. 1997 Phylogenetic reconstruction in Monanthes (Crassulaceae) based on morphological, chloroplast, and nuclear DNA variation. Journal of Evolutionary Biology 10: 193–216
Moran, R. V. 1942 The status of Dudleya and Stylophylum. Desert Plant Life 14: 149–157
Morgan, D. R., and D. E. Soltis. 1993 Phylogenetic relationship among members of Saxifragaceae sensu lato based on rbcL sequence data. Annals of the Missouri Botanical Garden 80: 631–660[CrossRef][ISI]
Mort, M. E., D. E. Soltis, P. S. Soltis, and M. L. Mabry. 2000 An empirical test of three methods for estimating internal support on phylogenetic trees. Systematic Biology 49: 160–171[CrossRef][ISI][Medline]
Moore, R. J. 1973 Chromosome numbers 1967–1974. Oosthoock's Uitgeversmaatschappij B. V., Utrecht, The Netherlands
Neuhaus, H., and G. Link. 1987 The chloroplast tRNA (UUU) gene from mustard (Sinapis alba) contains a class II intron potentially coding for a maturase-related polypeptide. Current Genetics 11: 251–257
