| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
Department of Systematic Botany, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, SE-752 36 Uppsala, Sweden
Received for publication September 21, 2000. Accepted for publication February 16, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and MethodsResultsDiscussionLITERATURE CITED |
|---|
65 million years ago (mya), Campynemataceae, Melanthiaceae, Smilacaceae + Liliaceae, and Alstroemeriaceae + Luzuriagaceae + Colchicaceae. Aspects of historical biogeography were investigated by dispersal–vicariance analysis. Several dispersal and vicariance events were found to coincide with Late Cretaceous–Early Tertiary changes in continental interconnections. The study contains the first published sequence of Campynemanthe, supporting the Campynemataceae as a monophyletic group.
Key Words: biogeography • Campynemataceae • dating • Gondwana • Liliales • molecular clock • rbcL
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionLITERATURE CITED |
|---|
; Källersjö et al., 1998
; Soltis et al., 1998
; Soltis, Soltis, and Chase, 1999
). With a robust phylogeny available, the question of dating the phylogeny comes into focus. Phylogenetic dating is important, since clades of the same age are needed for comparison in evolutionary biology and historical biogeography.
Many attempts at phylogenetic dating are concerned with the problems of unequal substitution rates (e.g., Bousquet et al., 1992
; Gaut et al., 1992
; Gaut, Muse, and Clegg, 1993
; review in Sanderson, 1998
), whereas the major impact that taxon sampling, tree topology (Baldwin and Sanderson, 1998
), and rate calibration (Bremer, 2000
) can have on the results has received much less attention. We shall comment on all these sources of error. We focus on the major clades of Liliales, i.e., families and groups of families, and do not draw any conclusions within families, where the results are particularly sensible to the sources of error mentioned.
Bremer (2000)
proposed identification of all clades of flowering plants that date back to the Early Cretaceous 100 million years ago (mya). In a phylogenetic dating based on a robust phylogeny of monocots, on variation in rbcL sequences, and on Late Cretaceous reference fossils, he tentatively identified 14 clades of monocots to be at least 100 million years (my) old. One of these clades is the Liliales, which is more thoroughly investigated in this study.
Traditionally many lilioid monocots have been assigned to one large family, Liliaceae (Cronquist, 1981
). Eventually this polyphyletic taxon was reclassified into several partly distantly related families, several of them placed in a large order Asparagales, others in a more narrowly circumscribed Liliales (Dahlgren, Clifford, and Yeo, 1985
). With molecular data, the circumscriptions of Liliales and the families of Liliales have been further refined (Rudall et al., 1995, 2000
). Today Liliales are a well-supported monophyletic group of 1300 species mostly with tepal nectaries and extrorse anthers, in contrast to the septal nectaries and introrse anthers commonly found in other monocots. Familiar representatives are lilies (Lilium), tulips (Tulipa), and autumn crocus (Colchicum).
The order includes seven families, Liliaceae, Melanthiaceae, Colchicaceae, Smilacaceae, Alstroemeriaceae, Campynemataceae, and Luzuriagaceae (Rudall et al., 2000
). The first two are restricted to the northern hemisphere, the last three mainly to the southern hemisphere (in South America, Australia, New Caledonia, and New Zealand). Colchicaceae have a largely southern hemisphere distribution in Africa, Madagascar, Malesia, New Guinea, Australia, and New Zealand, but notably excluding South America; Colchicaceae also include several species in North America and Eurasia. Smilacaceae are predominantly pantropical but with some species also in southern South America, New Zealand, and temperate regions of the northern hemisphere.
We have sampled 40 taxa of Liliales and four outgroup taxa from the related order Asparagales (Table 1). The ingroup sequences represent all families and most genera within the smaller families, and a selection of genera within the larger families. The selection was done so as to reflect adequately the total distribution of each family. The sampling should be sufficient for conclusions at and above the family level, but not at genus level.
|
dating of the major clades of monocots. A robust phylogeny of Liliales with branch lengths from variation in rbcL sequences is dated using the mean branch-length method of Bremer and Gustafsson (1997)
as well as the nonparametric rate smoothing approach by Sanderson (1997)
. The datings are relevant for interpretation of the historical biogeography of the Liliales, with their many intercontinental distributions indicating a connection to Gondwanan continental formations. We also address the questioned relationship of the two genera within Campynemataceae. |
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionLITERATURE CITED |
|---|
and Doyle and Doyle (1987)
. The DNA was purified with QIAquick PCR (polymerase chain reaction) kit (Qiagen, Santa Clarita, California, USA) according to the instructions provided by the manufacturer. Polymerase chain reaction amplifications were performed with Taq extender and Taq polymerase (Promega, Madison, Wisconsin, USA) according to the manufacturer's protocol. Primer sequences are given in Table 2. The PCR cycles consisted of one initial denaturation step of 6 min at 94°C followed by 30–35 cycles of 1 min of denaturation at 94°C, 1 min of annealing at 50–60°C, and 2 min of extension at 72°C. The PCR was terminated with a final step of 7 min at 72°C. The PCR products were purified with QIAquick PCR kit (Qiagen) according to the instructions provided by the manufacturer (using ddH2O as the eluting agent). Automated sequencing was performed on an ABI PRISM 377 Sequencer (Perkin Elmer Applied Biosystems, Foster City, California, USA) with ABI PRISM® BigDyeTM Terminator Cycle Sequencing Ready reaction kit (Perkin Elmer Applied Biosystems) according to the manufacturer's instructions.
|
), with all characters weighted equally. One hundred random addition sequence replicates and TBR (tree bisection reconnection) branch swapping were used to search for most parsimonious reconstructions. Branch lengths for one tree were obtained by PAUP using ACCTRAN (accelerated transformation) optimization. Alternative optimization strategies, e.g., DELTRAN (delayed transformation), yield similar branch lengths and do not significantly influence the results. Bootstrap (Felsenstein, 1985
) and jackknife support analyses (Farris et al., 1996
) were made in PAUP with 1000 replicates and TBR branch swapping for each replicate. Both analyses were heuristic, with five random addition sequences and the multrees option not in effect. The jackknife analysis was set to delete 37% of the characters in each replicate. Mean branch lengths from the terminals to each node were calculated as described by Bremer and Gustafsson (1997)
. An analysis of variance (ANOVA) was used to investigate if there are added variances in branch lengths among the seven family clades. Scheffé's test was used to make pair-wise comparisons among the families.
Phylogenetic dating requires at least one reference node dated by at least one reference fossil. The reference node(s) is used for rate calibration, and with an observed change rate for the tree (or part of the tree), the age of other nodes of the tree may be estimated. Furthermore, the reference fossils must be sufficiently well identified to be attached to the phylogeny and they need to be sufficiently old to be relevant for the phylogeny to be dated. For our tree of Liliales, Early Tertiary or Cretaceous fossils would be suitable. Younger fossils would necessitate a much larger sampling of genera and species to enable a correct assignment of the fossil to the phylogeny. Since either the dating or the identification of the existing Liliales fossils are questioned (see DISCUSSION) the only available approach is to use the previous dating of monocots (Bremer, 2000
), where the split between Liliales and their sister-group was estimated to 112 my and the age of the basal node within Liliales to 82 ± 10 my. The mean branch length from the terminals to the basal node, 65 steps, was divided by the minimum age of 82 my to obtain an estimate of the overall maximum change rate along the branches, 65/82 = 0.79 steps per my. This change rate was used to calculate the minimum age for each node, by dividing the mean branch length from the terminals to the node by the change rate. The tree with its branch lengths was also analyzed with Sanderson's r8s program (1999) implementing his nonparametric rate smoothing approach to phylogenetic dating (Sanderson, 1997
). The idea behind Sanderson's method is to accept local differences of rate in the tree. The program smoothes the differences over the tree by minimizing the rate changes among adjacent branches. The age of the basal node was again set to 82 my following Bremer (2000)
.
Confidence intervals (95%) on the age estimate for each node were also calculated, for both the mean branch-length method (Bremer and Gustafsson, 1997
) and for the nonparametric rate smoothing (Sanderson, 1997
). Using the SEQBOOT program in the PHYLIP package (Felsenstein, 1993
), 100 bootstrap data matrices were generated. With PAUP all 100 matrices were optimized on the original tree and branch lengths and mean branch lengths were calculated. Thus, for each node 100 mean branch lengths from the terminals to the node were obtained from the 100 matrices. The standard deviation of the 100 mean branch lengths for each node provides the standard error in mean branch length due to character sampling, and the corresponding confidence interval for the age estimate is calculated as ±1.96 standard errors in mean branch length divided by the change rate (0.79). The 100 matrices were also analyzed by Sanderson's r8s program (1999), which then provides similar confidence intervals for the age estimates according to nonparametric rate smoothing.
To assign distributions to the internal nodes in the tree the DIVA (dispersal–vicariance analysis) program was used (Ronquist, 1996, 1997
). The program optimizes distributions for each node of the tree by favoring vicariance events and minimizing the number of assumed dispersals and extinctions. Between the nodes of the given tree DIVA assigns a cost to changes in distribution interpreted as extinctions or dispersals but no cost to changes interpreted as vicariance. Optimizations minimizing the cost are favored. The assigned distributions were then compared with the age estimates and if possible correlated to geological history (see below). Eight main areas of distribution were used; (1) Eurasia, (2) North America, (3) South America, (4) Africa, (5) Tropical Asia, (6) Australia together with New Guinea and Tasmania, (7) New Caledonia, and (8) New Zealand. The areas Australia, New Guinea, and Tasmania were treated together since they are historically more closely related and in order to reduce the number of alternative solutions (the nexus files are available upon request from the first author). The number of areas allowed at each node may be restricted to less than the maximum of eight. The analysis was run repeatedly with successively fewer areas allowed at each node.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionLITERATURE CITED |
|---|
|
|
|
nonparametric rate smoothing are both given in Table 3. Confidence intervals are also reported in Table 3. In Fig. 2 nodes are arranged according to age estimated by the mean branch-length method and confidence intervals are indicated for the basal nodes. Four clades are considered to date back to the Cretaceous 65 mya, Campynemataceae, Melanthiaceae, Smilacaceae + Liliaceae, and Alstroemeriaceae + Luzuriagaceae + Colchicaceae.
|
) are shown in Fig. 3. Ancestors represented by basal nodes were generally found to be widespread, in both the northern and southern hemisphere. For five of the basal nodes, marked with asterisks in Fig. 3, there were numerous alternative optimizations (including the basal node of the Smilacaceae, due to the wide distribution of the genus Smilax). Alternative optimizations remained numerous with the number of areas allowed at each node restricted from eight to seven, six, and five, respectively. With four areas allowed at each node, the number of alternative optimizations at four of the asterisk-marked nodes decreased considerably, as shown in Fig. 3.
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionLITERATURE CITED |
|---|
, p. 1225) score better than a clock-based method if the sequences are not evolving in a clock-like fashion. Since we are interested in the biogeography of the families of Liliales and their diversification is in the basal part of the tree, it is only this part, viz. below the level of families, that is tested for a molecular clock. There are only two comparisons, between Colchicaceae-Melanthiaceae and Melanthiaceae-Smilacaceae, that have greater variance among than within the families (Table 4). These pairs are not sister-groups, however. We also made sister-group comparisons to reveal sister-groups significantly different in branch lengths (Tom Britton, University of Uppsala, unpublished data), and there are no such significant differences among the larger clades discussed in this paper. There are, however, eight nodes more terminal in the tree (nodes 5, 11, 18, 19, 20, 22, 25, and 29 in Fig. 2) that show significant differences in branch lengths. The comparison of Liliaceae and Smilacaceae (node 24 in Fig. 2) also indicates differences in branch lengths. Since our main interest is in the clades dating back to the Cretaceous, viz. the clade containing Liliaceae and Smilacaceae together, this does not influence the results. In two cases we do refer to ages within Smilacaceae and Liliaceae (nodes 26 and 29). In these circumstances the estimated dates of both clock and nonclock methods support the conclusions. A global rate test does indeed indicate deviations from the molecular clock, however, as shown by our analysis, it is due to inequalities in the more terminal parts of the tree. We only discuss age and biogeography of the major clades, and the procedure of calculating mean branch lengths reduces the effects of rate variation towards the base of the tree, where estimates are based on an increasing number of branch lengths (Bremer and Gustafsson, 1997
). Since the data are not completely clock-like, we chose, however, to compare our results also with those from the nprs method (Sanderson, 1997
). The two different methods used gave similar results (Table 3). The mbl method yielded somewhat younger estimates in more than half of the nodes. Generally the differences lie within the confidence intervals calculated. In nodes 17, 18, 19, 26, and 30 (Fig. 2) the differences are larger (>10 my) and presumably due to unequal branch lengths in these parts of the tree. The differences in dating between mpl and nprs in no case affect the biogeographical conclusions drawn in this paper.
The tree topology is generally robust. More than half of the nodes are supported by jackknife values >90%. Four nodes in the tree of Fig. 2 are not present in the strict consensus of the 12 equally parsimonious trees. These nodes (13, 22, 23, 36) received at least some jackknife support and the tree of Fig. 1 was chosen for dating (nonparametric rate smoothing requires fully resolved trees). The four nodes not present in the consensus are furthermore all within larger families in the upper part of the tree and do not influence relationships of the major clades. Our results are thus not affected by the choice among the 12 parsimonious trees. There are, however, some uncertainties in the topology of the basal part of the tree. Nodes 3 and 16 (Fig. 2) received no jackknife support and node 14 a jackknife value of 52% only. Hence, interrelationships remain uncertain among Alstroemeriaceae, Luzuriagaceae, and Colchicaceae, and among the four hypothesized Cretaceous clades (Alstroemeriaceae + Luzuriagaceae + Colchicaceae, Campynemataceae, Melanthiaceae, and Smilacaceae + Liliaceae), although their sister-group relationships are identical in all 12 equally parsimonious trees. Changes in these interrelationships affect the estimated age of the families. The differences are minor, however, and the four major clades are, whatever their interrelationships, always estimated to date back to the Cretaceous, 65 mya.
Rudall et al. (2000)
analyzed morphology, rbcL, and trnL-F sequences of Liliales. Their sample of genera is different from ours but the results are largely congruent. Both studies provide strong support for the families, except for Smilacaceae, which are weakly supported in both analyses. Nevertheless, we follow the classification of Rudall et al. (2000)
and include Ripogonaceae and Philesiaceae in Smilacaceae. They also found strong support for the group of Alstroemeriaceae, Luzuriagaceae, and Colchicaceae and for the pair of Smilacaceae and Liliaceae. Their tree is somewhat different with respect to other family interrelationships, illustrating the uncertainties mentioned in the previous paragraph. Thus, Luzuriagaceae are sister to a pair of Alstroemeriaceae and Colchicaceae, and Melanthiaceae assume a position as sister to these three families rather than to Smilacaceae and Liliaceae as in our Fig. 2. Several other studies show similar patterns, e.g., Chase et al. (2000)
analyzed rbcL, atpB, and 18s rDNA. Their configuration of families is identical to ours but the analysis does not include Luzuriagaceae and Campynemataceae. A robust phylogeny is also important for the estimation of the ancestral distributions. The uncertainties discussed above have no major influence of the DIVA result since the nodes involved (1, 2, 14, 16) are assigned many equally parsimonious distributions, and different groupings of these major clades will not change this situation. There are other studies that show minor differences also in more terminal parts of the tree, e.g., by Tamura (1998)
, and Fay and Chase (2000)
. Additional analyses with their alternative topologies show ancestral distributions at the nodes involved and later discussed (28, 29, 30, 33) to be the same.
John Conran (University of Adelaide, personal communication; Conran, Christophel, and Scriven, 1994
) questions the position of Petermannia within Colchicaceae. He maintains that phytochemical and morphological data (e.g., lack of colchicine alkaloids and U-shaped petals encircling the stamens) indicate a more basal position for Petermannia (as a separate family, Petermanniaceae) in Liliales. The rbcL sequences, however, place Petermannia as sister-group to Tripladenia within Colchicaceae with a jackknife support of 98%. Rudall et al. (2000)
also found Petermannia in the same position based on rbcL and trnL-F sequences. Another questioned relationship within Liliales is that of the two genera of Campynemataceae. Morphologically there are some differences, e.g., in the number of ovule locules. A sequence of the rbcL gene has so far not been available for Campynemanthe. We have for this analysis sequenced Campynemanthe viridiflora, one of the three species, and the two genera of Campynemataceae do group together with a jackknife support of 100%.
There are only a few different fossils known for Liliales. Petermanniopsis angleseaënsis is dated to Late Eocene. Judging from its name it seems related to Petermannia, but Conran, Christophel, and Scriven (1994)
, who named and described the fossil, considered its relationship to Petermannia as speculative. This latter view is actually supported by our analysis, given that Petermannia is related to Tripladenia as in our tree (but see above). Petermanniopsis has no resemblance to Tripladenia and if the fossil is related to Petermannia, node 11 (Fig. 2) would be pushed back into the Eocene, >34 mya, and the age of the Liliales becomes largely incongruent with the dating of monocots by Bremer (2000)
, who used eight reliably identified reference fossils. Hence, if Petermannia is related to Tripladenia, it is unlikely that Petermanniopsis is related to Petermannia. Given the uncertainties about Petermannia and Petermanniopsis, we have not used this fossil for dating Liliales.
Ripogonum scandens from early to middle Miocene (Pole, 1993
) was also rejected as too young for being relevant for our tree. More Ripogonum species would need to be sampled to decide where, in the phylogeny of the genus, the correct assignment of the fossil would be. Since the analysis has a limited sample of only one Ripogonum species, attaching this fossil to node 26 (Fig. 2) would grossly overestimate the change rate and underestimate the node ages, again becoming largely incongruent with the monocot dating by Bremer (2000)
. This leaves us with a preliminary report of Smilax from Eocene (Sun and Dilcher, 1988
). Again, we refrain from using this fossil, since we are uncertain about the identification. Furthermore, using a single reference fossil increases the danger of error in rate calibration. The currently best available approach for Liliales is to accept the dating of the basal node within Liliales made by Bremer (2000)
. Using eight reference fossils from different monocots, he estimated the age of major clades of monocots and that of the basal node within Liliales to 82 my. The mean branch length to node 1 in Fig. 2 is 65 steps, and thus we obtained an observed change rate of 65/82 = 0.79 steps per my for the rbcL gene. For monocots in general, Bremer observed a similar change rate for the rbcL gene, 0.73 steps per my.
Biogeography
The intercontinental southern hemisphere distributions of Liliales point to possible ancient Gondwana connections. A first step in corroborating such a relationship is phylogenetic dating of the major clades, attempted here. The break-up of Gondwana into a western part, comprising Africa and South America, and an eastern part, comprising Australia, Antarctica, Madagascar, and India, is dated to 180–150 mya (Scotese, Gahagan, and Larsen, 1988
; Hallam, 1994
). Due to rotation of the continental plates, South America and Antarctica were subsequently brought together, enabling floral exchange between South America and Australia via a habitable Antarctica well into the Tertiary <65 mya (Hallam, 1994
). Africa separated from South America in the Late Cretaceous 95–80 mya (Scotese, Gahagan, and Larsen, 1988
; Hallam, 1994
) and final break-up of Africa and South America was completed 80 mya. Liliales diverged from their sister-group in the Early Cretaceous, >100 mya, and the extant lineages diverged from each other 82 mya (Bremer, 2000
). We have hypothesized that four clades of the Liliales date back to the Cretaceous 65 mya. It should be remembered also, that since fossils provide minimum ages and observed rates thus are maximum rates, our age estimates are minimum ages. The possible intercontinental vicariance events, suggested by the current distributions of the taxa in the tree, are not all easily aligned with the separation of the continents. We here hypothesize some possible connections.
Raven and Axelrod (1974)
discussed the biogeography of Liliiflorae and concluded that the ancestor of the orders would have been old enough to have existed while South America and Australia were connected via Antarctica. Our dating supports this and shows that the ancestor of the Liliales, i.e., the lineage that diverged from other monocots in the Early Cretaceous >100 mya and persisted until it split 82 mya, is certainly old enough to have existed throughout the interconnected African and South American–Antarctic–Australian parts of Late Cretaceous Gondwana. The present distributions do not clearly indicate whether the ancestor was distributed in the northern hemisphere, in the southern hemisphere, or both (cf. Conran, 1995
). Liliaceae and Melanthiaceae are restricted to the northern hemisphere, while Alstroemeriaceae, Campynemataceae, Luzuriagaceae, and Smilacaceae (except Smilax, which is pantropical, and with a few species extending into North America, Europe, and Eastern Asia) occur in the southern hemisphere. Colchicaceae seem to have one mainly North American and one mainly African–Australasian lineage. The DIVA analysis postulates a widespread Liliales ancestor, distributed in many areas both in the northern and southern hemispheres. If the number of areas at the nodes is constrained to four, as is done for the nodes marked with an asterisk in Fig. 3, DIVA postulates the ancestral distribution to involve North and South America, Australia, and New Caledonia. If the number of areas is constrained to fewer than four, there are several equally parsimonious optimizations involving either one or two to three of these areas, demonstrating a remaining uncertainty about the ancestral distribution of the Liliales.
The ancestral distribution of the clade Alstroemeriaceae + Luzuriagaceae + Colchicaceae is also unclear. Following the DIVA results (Fig. 3), the ancestor was distributed either in both North and South America, and possibly also in Australia, or in the southern hemisphere only, in Australia or in South America and Australia. According to Raven and Axelrod (1974)
, Luzuriagaceae (then including the asparagalian genus Behnia) might have reached Australia from South America via Antarctica and Drymophila (then in Convallariaceae) was thought to have reached Australia by long-distance dispersal. The DIVA analysis indicates the ancestor of Alstroemeriaceae and Luzuriagaceae to have been distributed in South America and possibly also Australia and New Zealand. The latter alternatives imply that the Alstroemeriaceae-Luzuriagaceae split may represent a vicariance event due to the isolation of South America from Australia and New Zealand following the break-up of the Antarctic connection during the end of the Eocene 55–34 mya (Hallam, 1994
). The split is estimated to 48 ± 10.5 my (node 3), and the ancestors of these two families may have become isolated in the mid-Eocene, following gradually deteriorating connections across Antarctica.
The ancestor of Colchicaceae had, according to DIVA, a rather peculiar distribution, in North America and Australia (Fig. 3). According to our dating, the lineage leading to Colchicaceae existed throughout Eocene when Australia was connected to South America via Antarctica, and the family ancestor may have been present also in South America. Since there are no extant South American members of Colchicaceae, DIVA cannot trace that ancestral distribution. The basal split of the Colchicaceae may represent a vicariance event between a North American lineage (comprising the former Uvulariaceae genera Uvularia and Disporum) and a southern hemisphere lineage. It should be noted that our sample includes only nine of 20 genera of Colchicaceae, and we do not know where the other genera attach to the tree, but they are all from the southern hemisphere and have not been classified in the former Uvulariaceae (Dahlgren, Clifford, and Yeo, 1985
). Colchicaceae are strongly represented in Africa. The African genera included in our analysis are nested inside the family and the DIVA result supports Raven and Axelrod's conclusions (1974) and indicates that they reached Africa in the Late Tertiary (Fig. 3), long after the separation of Africa from South America during the Cretaceous. Present distribution of Colchicaceae is more convoluted than that of the other families and deserves further study.
The ancestral distribution of the second main clade of the Liliales, Campynemataceae + Melanthiaceae + Smilacaceae + Liliaceae, is unclear, like that of all Liliales. The DIVA postulates a wide ancestral distribution involving many areas from both hemispheres, including North and South America. If the number of areas at the basal nodes is constrained to four or less, as in Fig. 3, DIVA postulates the ancestral distribution to involve North America and New Caledonia, or North America, Australia, and New Caledonia, a rather unlikely distribution (Fig. 3). Similar uncertainties in ancestral distribution pertain to the ancestor of Melanthiaceae + Smilacaceae + Liliaceae and to the ancestor of Smilacaceae + Liliaceae. If the number of areas at the basal nodes is constrained to four or less, DIVA postulates the ancestral distribution to involve North America and, for the ancestor of Smilacaceae + Liliaceae, possibly also South America or Australia (Fig. 3).
Campynemataceae apparently represent an ancient southern hemisphere lineage, with today one Australian and one New Caledonian genus. Melanthiaceae are a northern hemisphere family, in our analysis with one Eurasian and one North American lineage. Figure 3 indicates that the basal split in Melanthiaceae could represent a North American–Eurasian vicariance event, although it should be noted that only half of the genera are represented in the tree and adding taxa may affect the conclusions. Three of the missing genera have a distribution restricted to Eurasia, four to North America, and one genus occurs in both areas. Liliaceae are also a northern hemisphere family with Eurasian (including some North African representatives) and North American lineages (cf. Conran, 1995
). The DIVA result indicates that Liliaceae are originally North American and that they expanded from North America to Eurasia sometime 30–40 mya (the lineage between nodes 29 and 33; Fig. 2). This would have been possible via Beringia where North America and Eurasia have been connected intermittently during the Tertiary (Enghoff, 1995
). In our analysis the three genera Gagea, Cardiocrinum, and Notholirion are not represented. They all belong within the Lloydia to Nomocharis clade (node 35) (Tamura, 1998
; Chase et al., 2000
; Fay and Case, 2000
; Rudall et al., 2000
). The three genera are restricted to Eurasia, and their inclusion would only reinforce the interpretation of the Lloydia to Nomocharis clade (node 35) as ancestrally Eurasian. The preceding split (node 33) between Clintonia + Medeola and the Lloydia to Nomocharis clade then represents a possible North American–Eurasian vicariance event.
The ancestral distribution of Smilacaceae is very unclear due to the currently widespread genus Smilax, and DIVA has numerous alternative solutions. It is possible that the presence of Smilax in Africa is a relict from a Cretaceous ancestral distribution of the Liliales involving also Africa before it separated from South America. The basal nodes in Fig. 3 all lack Africa in the DIVA optimizations. This is because the program was constrained to assign a maximum of four areas to these nodes, a constraint imposed to reduce the number of alternative optimizations. The nodes that are constrained are marked with an asterisk in Fig. 3. Without the constraint, all asterisk-marked nodes do contain Africa in at least some of the optimizations.
The ancestor of Ripogonum, Lapageria, and Philesia of the Smilacaceae was according to DIVA distributed in South America and New Zealand and possibly also in Australia (Fig. 3). Break-up of the Antarctic link occurred in several steps. The separation between Australia and the Antarctic took place at 38 mya. During the Paleogene 65–24 mya the connection between South America and the Antarctic started to deteriorate and was broken towards the end of the Eocene 34 mya (Hallam, 1994
). This isolation of South America from Australia and New Zealand corresponds to the split of the South American Lapageria and Philesia from the Australian–New Zealand Ripogonum estimated to 47 ± 8.4 my (node 26). As discussed above, the same interpretation is possible for the split between Alstroemeriaceae and Luzuriagaceae, so it may be that termination of the Antarctic link during Eocene resulted in two vicariance events within Liliales.
|
FOOTNOTES |
|---|
2 Author for reprint requests (annika.vinnersten{at}ebc.uu.se
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionLITERATURE CITED |
|---|
Baldwin B. G. M. J. Sanderson 1998 Age and rate of diversification of the Hawaiian silversword alliance (Compositae). Proceedings of the National Academy of Sciences USA 95: 9402-9406
Bousquet J. S. H. Strauss A. D. Doerksen R. A. Price 1992 Extensive variation in evolutionary rate of rbcL gene sequences among seed plants. Proceedings of the National Academy of Sciences USA 89: 7844-7848
Bremer K. 2000 Early Cretaceous lineages of monocot flowering plants. Proceedings of the National Academy of Sciences USA 97: 4707-4711
Bremer K. M. H. G. Gustafsson 1997 East Gondwana ancestry of the sunflower alliance of families. Proceedings of the National Academy of Sciences USA 94: 9188-9190
Brummitt R. K. C. E. Powell 1992 Authors of plant names. Royal Botanic Gardens, Kew, UK
Chase M. W. D. E. Soltis P. S. Soltis P. J. Rudall M. F. Fay W. H. Hahn S. Sullivan J. Joseph M. Molvray P. J. Kores T. J. Givnish K. J. Sytsma J. C. Pires 2000 Higher-level systematics of the monocotyledons: an assessment of current knowledge and a new classification. In K. L. Wilson and D. A. Morrison [eds.], Monocots: systematics and evolution, Proceedings of the Second International Conference on the Comparative Biology of Monocotyledons, 3–16. CSIRO Publishing, Melbourne, Australia
Conran J. G. 1995 Family distributions in the Liliiflorae and their biogeographical implications. Journal of Biogeography 22: 1023-1034[CrossRef][ISI]
Conran J. G. D. C. Christophel L. Scriven 1994 Petermanniopsis angleseaënsis gen. & sp. nov.: an Australian fossil net-veined monocotyledon from Eocene Victoria. International Journal of Plant Sciences 155: 816-827[CrossRef][ISI]
Cronquist A. 1981 An integrated system of classification of flowering plants. Columbia University Press, New York, New York, USA
Dahlgren R. M. T. H. T. Clifford P. F. Yeo 1985 The families of monocotyledons. Springer-Verlag, Berlin, Germany
Doyle J. J. J. L. Doyle 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15
Enghoff H. 1995 Historical biogeography of the Holarctic: area relationships, ancestral areas, and dispersal of non-marine animals. Cladistics 11: 223-263
Farris J. S. V. A. Albert M. Källersj D. Lipscomb A. G. Kluge 1996 Parsimony jackknifing outperforms neighbour-joining. Cladistics 12: 99-124[CrossRef][ISI]
Fay M. F. M. W. Case 2000 Modern concepts of Liliaceae with a focus on the relationships of Fritillaria. Curtis's Botanical Magazine 17: 146-149[CrossRef]
Felsenstein J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783-791[CrossRef][ISI]
Felsenstein J. 1993 PHYLIP: phylogeny inference package, version 3.5c. Department of Genetics, University of Washington, Seattle, Washington, USA
Gaut B. S. S. V. Muse W. D. Clark M. T. Clegg 1992 Relative rates of nucleotide substitution at the rbcL locus of monocotyledonous plants. Journal of Molecular Evolution 35: 292-303[CrossRef][ISI][Medline]
Gaut B. S. S. V. Muse M. T. Clegg 1993 Relative rates of nucleotide substitution in the chloroplast genome. Molecular Phylogenetics and Evolution 2: 89-96[CrossRef][Medline]
Hallam A. 1994 An outline of phanerozoic biogeography. Oxford University Press, Oxford, UK
Källersjö M. J. S. Farris M. W. Chase B. Bremer M. F. Fay C. J. Humphries G. Petersen O. Seberg K. Bremer 1998 Simultaneous parsimony jackknife analysis of 2538 rbcL DNA sequences reveals support for major clades of green plants, land plants, seed plants, and flowering plants. Plant Systematics and Evolution 213: 259-287[CrossRef][ISI]
Kubitzki K. 1998 The families and genera of vascular plants, vol. 3. Springer-Verlag, Berlin, Germany
Pole M. 1993 Early Miocene flora of the Manuherika Group, New Zealand. 5. Smilacaceae, Polygonaceae, Elaeocarpaceae. Journal of the Royal Society of New Zealand 23: 289-302[ISI]
Raven P. H. D. I. Axelrod 1974 Angiosperm biogeography and past continental movements. Annals of the Missouri Botanical Garden 61: 539-673[CrossRef][ISI]
Ronquist F. 1996 DIVA: dispersal–vicariance analysis, Vers. 1.1. Uppsala University, Uppsala, Sweden
Ronquist F. 1997 Dispersal–vicariance analysis: a new approach to the quantification of historical biogeography. Systematic Biology 46: 195-203[CrossRef][ISI]
Rudall P. J. P. J. Cribb D. F. Cutler C. J. Humphries 1995 Monocotyledons: systematics and evolution I–II. Royal Botanic Gardens, Kew, UK
Rudall P. J. K. L. Stobart W.-P. Hong J. G. Conran C. A. Furness G. Kite M. W. Chase 2000 Consider the lilies: systematics of Liliales. In K. L. Wilson and D. A. Morrison [eds.], Monocots: systematics and evolution, Proceedings of the Second International Conference on the Comparative Biology of Monocotyledons, 347–359. CSIRO Publishing, Melbourne, Australia
Saghai-Maroof M. A. K. M. Soliman R. A. Jorgensen R. W. Allard 1984 Ribosomal DNA spacer length polymorphism in barley: Mendelian inheritance, chromosomal location, and population dynamics. Proceedings of the National Academy of Sciences USA 81: 8041-8081
Sanderson M. J. 1997 A nonparametric approach to estimating divergence times in the absence of rate constancy. Molecular Biology and Evolution 14: 1218-1231[ISI]
Sanderson M. J. 1998 Estimating rate and time in molecular phylogenies: beyond the molecular clock?. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II, DNA sequencing, 242–264. Kluwer, Boston, Massachusetts, USA
Sanderson M. J. 1999 r8s ("rates"), version 0.9. Software for the analysis of rates of evolution and miscellaneous other stuff. University of California, Davis, California, USA
Scotese C. R. L. M. Gahagan R. L. Larsen 1988 Plate tectonic reconstructions of the Cretaceous and Cenozoic ocean basins. Tectonophysics 155: 27-48
Soltis D. E. P. S. Soltis M. E. Mort M. W. Chase V. Savolainen S. B. Hoot C. M. Morton 1998 Inferring complex phylogenies using parsimony: an empirical approach using three large DNA data sets for angiosperms. Systematic Biology 47: 32-42[CrossRef][ISI][Medline]
Soltis D. E. P. S. Soltis M. W. Chase 1999 Angiosperm phylogeny inferred from multiple genes as a tool for comparative biology. Nature 402: 402-404
Sun Z. D. L. Dilcher 1988 Fossil Smilax from Eocene sediments in western Tennessee. American Journal of Botany 75 : (Supplement 6) 118
Swofford D. L. 1998 PAUP*: phylogenetic analysis using parsimony, vers. 4. Sinauer, Sunderland, Massachusetts, USA
Tamura M. N. 1998 Liliaceae. In K. Kubitzki [ed.], The families and genera of vascular plants, vol. 3, 343–353. Springer-Verlag, Berlin, Germany.
This article has been cited by other articles:
|
|
 |
|
  M. R. Duvall, J. W. Robinson, J. G. Mattson, and A. Moore Phylogenetic analyses of two mitochondrial metabolic genes sampled in parallel from angiosperms find fundamental interlocus incongruence Am. J. Botany, July 1, 2008; 95(7): 871 - 884. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Sanderson, J. L. Thorne, N. Wikstrom, and K. Bremer Molecular evidence on plant divergence times Am. J. Botany, October 1, 2004; 91(10): 1656 - 1665. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Vinnersten and G. Reeves Phylogenetic relationships within Colchicaceae Am. J. Botany, October 1, 2003; 90(10): 1455 - 1462. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
