| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
0 Department of Biology, San Francisco State University, 1600 Holloway Avenue, San Francisco, California 94132 USA
Received for publication February 26, 1999. Accepted for publication January 18, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: biogeography • Leptodactylon • Leptosiphon • Linanthus • maximum-likelihood • molecular clock • Polemoniaceae
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Patterson, 1977, 1979
; Johnson et al., 1996
; Porter, 1997
). The family has been well studied and demonstrates patterns of plant diversity in the flora of western North America (Epling and Dobzhansky, 1942
; Grant, 1959
; Taylor and Levin, 1975
; Raven and Axelrod, 1978
; Plitmann and Levin, 1983
; Galen, 1989, 1990, 1992
; Wolf and Soltis, 1992
); however, relatively few researchers have estimated phylogenies within the family. The most recent and frequently used taxonomic system for Polemoniaceae (Grant, 1959
) predates any quantitative phylogenetic analysis. Recently, investigators have begun to reconstruct phylogenies at the family level (Johnson et al., 1996
; Porter, 1997
; Porter and Johnson, 1998
). In addition, work has been completed on several of the genera within Polemoniaceae [Gilia (Johnson and Soltis, 1995
), Navarretia (Spencer and Porter, 1997
), Cobaea (Prather and Jansen, 1998
; Prather, 1999
), and Polemonium (de Geofroy, 1998
)].
Linanthus, with 44 species, is the third largest genus of Polemoniaceae (Grant, 1959
; Patterson, 1993
). It occurs throughout much of the western United States and Baja California, with one species endemic to Chile. Most species, however, are endemic to the California Floristic Province and grow in an array of habitats throughout the state. Several authors (Patterson, 1979
; Golden and Patterson, 1988
; Buxton, 1994
) have used morphological data to attempt to sort relationships among members of the genus, but their results have provided little resolution. A preliminary investigation into the genus (Bell, Patterson, and Hamilton, in press
), using ITS sequence data, suggested that not all sections within Linanthus or Leptodactylon represent monophyletic groups. In addition, recent family-level molecular phylogenetic work with the chloroplast matK region (Johnson et al., 1996
) and the nuclear ITS region (Porter, 1997
) do not support the monophyly of Linanthus.
Throughout this paper, we use the names of genera and sections in their traditional context (i.e., sensu Grant), realizing that although these names may not reflect natural groups, they are likely to be most familiar to current readers. Porter and Johnson (in press)
have proposed a classification of the entire Polemoniaceae, using monophyly as a requirement for recognition of groups. In that paper they divide the traditional Linanthus into two genera: (1) Linanthus (comprising Linanthus sects. Linanthus and Dianthoides, as well as Leptodactylon, and four species of Gilia sect. Giliastrum (G. filiformis, G. campanulata, G. inyoensis, and G. maculata); and (2) Leptosiphon (comprising sects. Leptosiphon, Dactylophyllum, Siphonella, and Pacificus). In this paper we use Linanthus and Leptosiphon when referring to these major clades.
One of the major foci of this project was to investigate further the relationships of the members traditionally placed into Linanthus and their affiliation with Phlox and Leptodactylon. Species of Linanthus, Phlox, and Leptodactylon, with few exceptions, have opposite leaves, while the rest of the species in the family, primarily, have alternate leaves. Phlox differs from Linanthus and Leptodactylon by having a base chromosome number of n = 7, where the other two genera have a base chromosome number of n = 9. The morphological similarities between perennial Linanthus and Leptodactylon have been noted (Rydberg, 1906
), namely their suffrutescent habit, mainly palmately divided leaves, and white corollas (with the exception of Le. californicum, which has a pink corolla). Wherry (1961)
, however, hypothesized that Linanthus was polyphyletic, with the members of section Linanthus and Leptodactylon being closely associated, while the remainder of the species were related to Linanthastrum (= Linanthus nuttallii and relatives). Several characters, including calyx morphology, have been used to differentiate the perennial members of Linanthus from Leptodactylon (Patterson, 1977
).
Grant's 1959
treatment of the genus recognized 37 species of Linanthus, all but two being annuals. Within the genus he recognized six sections (see Table 1) based on morphological features. Patterson (1993)
recognized Grant's sections but added additional taxa that had been described since the time of Grant's treatment. A review of the different treatments is presented by Bell (1998)
and Bell, Patterson, and Hamilton (in press)
.
|
, using the chloroplast gene matK, were the first to reconstruct a molecular phylogeny of Polemoniaceae. Although they sampled 15 of the 19 genera in the family, they sampled only one species each of Linanthus and Leptodactylon, and only two species of Phlox. These three genera did, however, form a strongly supported group (94% bootstrap).
Johnson et al. (1996)
, also using matK sequence data, sampled 59 species representing 18 of the 19 genera of Polemoniaceae. They also found an association between Linanthus, Leptodactylon, and Phlox. In addition, they found Gymnosteris and Gilia filiformis within the same, strongly supported lineage (100% bootstrap). Furthermore, these data placed species of Linanthus in two separate lineages. One lineage included sect. Dianthoides (Li. dianthiflorus and Li. demissus), sect. Linanthus (Li. dichotomus), and Leptodactylon (Le. jaegeri, Le. pungens, and Le. watsonii) along with Gilia filiformis. The remaining sections sampled, Linanthus sect. Leptosiphon (Li. ciliatus and Li. breviculus) and Linanthus sect. Siphonella (Li. pachyphyllus) grouped with Phlox and Gymnosteris. The authors did not sample members from sect. Dactylophyllum or Li. grandiflorus (sect. Pacificus).
Porter (1997)
undertook an investigation into the Polemoniaceae using the nuclear ribosomal internal transcribed spacer (ITS) region, providing an independent source of phylogenetic information from the previous chloroplast sequence data. Porter's study included 53 species from 17 genera of the Polemoniaceae. Results were in strong agreement with those from the chloroplast gene matK (Johnson et al., 1996
), both in the position of Linanthus within Polemoniaceae and its affiliation with Leptodactylon and Phlox. The Porter study did include a sample from sect. Dactylophyllum (Li. aureus) that grouped with members of sect. Leptosiphon (Li. bicolor) and sect. Siphonella (Li. nuttallii), along with members of Phlox. In addition, Porter found four species of Gilia (G. campanulata, G. inyoensis, G. filiformis, and G. maculata) in the Linanthus clade. Gilia maculata had been placed in Linanthus, but was removed by Patterson (1989)
and placed in Gilia based on morphological similarities with G. campanulata, G. inyoensis, and G. filiformis. In order to better understand the relationships and character evolution of the members of Linanthus, Leptodactylon and Phlox we felt a more complete sampling of the group needed to be undertaken.
Biogeography and age of Linanthus
Of the 44 species of Linanthus, 36 occur in the California Floristic Province or adjacent desert communities. Likewise, many other genera in the "temperate" Polemoniaceae (Steele and Vilgalys, 1994
; Johnson et al., 1996
) show a high degree of endemism in this region. This distributional pattern is similar to those reported by Baldwin (1996)
for the subtribe Madiinae (Asteraceae). The diversification of many plant groups found in the California Floristic Province and adjacent areas may have occurred in response to a shift from a summer-wet to a summer-dry (Mediterranean) climate that occurred abruptly in the westernmost part of North America, 
15 million years ago (mya; Axelrod, 1992
; Baldwin, 1997
; Baldwin and Sanderson, 1998
). Given that some species of Linanthus range across western North America from the Rocky Mountains to California, it has also been suggested that this group, as well as the entire tribe Gilieae (Polemoniaceae) including Gilia, Ipomopsis, Eriastrum, Langloisia, Navarretia, Leptodactylon, and Linanthus, may have had an earlier association with the broad sclerophyll Madrean vegetation (20 mya) (Raven and Axelrod, 1978
). Grant (1959)
hypothesized that Polemoniaceae represent a lineage of tropical origin that dispersed to and radiated in the temperate regions independently two separate times—one within the tribe Polemonieae and the other within the tribe Gilieae.
Recent molecular work (Johnson et al., 1996
; Porter, 1997
; Porter and Johnson, 1998
), however, does not support the monophyly of the tribe Gilieae or of tribe Polemonieae. Mitochondrial, chloroplast, and nuclear DNA evidence suggest that the tribes Gilieae and Polemonieae are polyphyletic assemblages with some of its members (Navarretia, Ipomopsis, and Linanthus) being associated with at least three separate lineages in Polemoniaceae. The chloroplast data (Steele and Vilgalys, 1994
; Johnson et al., 1996
) do, however, support the presence of a "tropical" (including the genera Cantua, Cobaea, and Bonplandia) and "temperate" (the remainder of the genera with the exception of Acanthogilia) lineage within Polemoniaceae. Recent phylogenies of Polemoniaceae based on sequence data from the mitochondrial nad1b intron also support three lineages within the family: (a) "tropical" lineage containing Cantua, Cobaea, and Bonplandia; (b) a lineage made up of the genus Acanthogilia; and (c) a "temperate" lineage composed of the genera of Gilieae and Polemonieae, in addition to the genus Loeselia [from the "tropical" tribe Bonplandiae (Grant, 1959
)]. Given its center of diversity in western North America, it is possible that the current distribution and diversity of the entire "temperate" Polemoniaceae are tied to the development of this summer-dry, Mediterranean style climate in western North America. However, it is important to note that while the members of this lineage are predominantly temperate, it includes several lineages that are either completely or partly tropical (e.g., Loeselia) in their distribution (Porter, 1997
). During the Pliocene, western North America continued to experience the spread of drought, and Axelrod (1977)
hypothesized that many species may have arisen during this time in response to an additional drying trend. The purpose of this investigation was to estimate a phylogeny of Linanthus using molecular sequence data from ITS and matK with a more complete sampling than has previously been undertaken. These gene tree estimates of the species phylogeny were then used to (1) further examine the integrity of Grant's sections and (2) to study the biogeography of a diverse array of perennial and annual taxa by using paleoclimatic data to establish an "external" calibration point in order to estimate the age of divergences within this group of plants.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Baldwin et al., 1995
; Johnson et al., 1996
; Porter, 1997
). During this project we were unable to extract DNA and/or amplify and sequence PCR products from three species of Linanthus (Li. serrulatus, Li. oblanceolatus, and Li. latisectus).
DNA extraction and amplification
Total DNAs were extracted using the hexadecyltrimetnylammonium bromide (CTAB) methods of Doyle and Doyle (1987)
and Cullings (1992)
. The DNA extracts were then further purified with the Prep-A-Gene® DNA Purification Kit (Bio-Rad, Hercules, California, USA).
A portion of the matK region of the chloroplast DNA was amplified with the primers matK-1176F and matK-1800R (Johnson and Soltis, 1995
; Steele and Vilgalys, 1994
). Amplification of the ITS1, 5.8S, and ITS2 regions of the nuclear ribosomal DNA was done with the primers ITS4 and ITS5 (White et al., 1990
). Double-stranded copies of both regions were amplified using standard Polymerase Chain Reaction (PCR) in 50-µL volume reactions. Amplified PCR products were cleaned prior to sequencing using a PEG precipitation. Amplification primers, along with ITS2 and ITS3 (White et al., 1990
), were then used to sequence each corresponding region.
Sequencing
All sequencing was done via dye terminator cycle sequencing on a Catalyst 800 Molecular Biology LabStation using the protocol specified by the ABI PRISM® Dye Primer Cycle Sequencing Ready Reaction Kit (Revision B, August 1995, Perkin-Elmer, Foster City, California, USA).
Alignment
Sequence fragments were aligned using the computer package Sequencher (Gene Codes Corporation, Inc., 1994
) to build contig sequences. All sequences were then aligned visually. An aligned matrix is available on TreeBase (www.herbaria.harvard.edu/index.html) or is available on request from the authors. The computer program MacClade version 3.06 (Maddison and Maddison, 1992
) was also used to translate DNA sequences to protein sequences to help with the alignment of matK sequences.
Preliminary sequence analysis
Aligned sequences were evaluated for overall base composition bias and among-taxon base composition. The base composition bias statistic was calculated according to Irwin, Kocher, and Wilson (1991)
and ranges in value from 0 to 1, with 0 indicating no bias and 1 showing complete base composition bias. These results are important when choosing appropriate maximum likelihood or distance models because an extreme overabundance of one nucleotide state can increase the tendency for sites to become saturated (Irwin, Kocher, and Wilson 1991
). In addition, a strongly skewed mutation bias can violate the assumptions of parsimony (Perna and Kocher, 1995
; Spicer, 1995
; Yoder, Vilgalys, and Ruvolo, 1996
). Examination of base composition and base composition bias in the data sets are summarized in Table 2.
|
). This test asks whether the best phylogenetic estimates from one data set are significantly different from those of another. Rejection of the null hypothesis favors not combining the data sets. In addition, the partition-homogeneity test (Farris et al., 1995
) [also known as the incongruence length difference (ILD) test] as implemented in PAUP* 4.0 b2 (Swofford, 1998) was used to assess the congruence between ITS and matK data sets. The null hypothesis of this test is that each pair of loci are as congruent as two randomly generated partitions of equal size. This test compares the length of the most parsimonious tree(s) from the original pair of partitions with the lengths of randomly generated data sets. The character columns in the random data sets are rearranged, but the size of each partition is held constant. One thousand random partitions were generated to create a null distribution to test the statistical significance of the tree lengths from the original partitions. Because incongruence between chloroplast and nuclear phylogenies has been previously reported in Linanthus (Johnson and Soltis, 1998
), these tests were also performed on a combination of culled data sets (where taxa were systematically removed) to try and identify the areas of incongruence.
Phylogenetic analysis
Because parsimony has been shown to be inconsistent under some situations when dealing with molecular sequence data (Felsenstein, 1983
; Hasegawa and Fujiwara, 1993
; Kuhner and Felsenstein, 1994
; Huelsenbeck, 1995
; Swofford et al., 1996
), a variety of model-based methods, in addition to maximum parsimony, were employed to search for phylogenetic relationships and then evaluated using maximum likelihood. Baum, Sytsma, and Hoch (1994)
suggested that analyzing data with a variety of algorithms (e.g., maximum parsimony, minimum evolution, and maximum likelihood) is desirable because it can detect potentially weakly supported lineages. Because different methods are sensitive to different biases in the data set, elements in topology that are supported in different analyses may be considered more robust than those supported strongly by one search method but contradicted by another. All analyses were performed using the computer software PAUP* 4.0 b2 (Swofford, 1998).
Maximum parsimony searches were conducted using heuristic search methods with tree bisection reconnection (TBR) branch swapping, collapse of zero-length branches, and weighting all characters equally. The analyses were repeated 100 times with the RANDOM ADDITION option to minimize problems of multiple islands of most parsimonious trees (Maddison, 1991
). Sets of equally most parsimonious trees were summarized by a strict consensus tree. To assess confidence in resulting tree topologies, bootstrap tests (Felsenstein, 1985
; Hillis and Bull, 1993
) were performed using 300 replicates. Likewise, a decay analysis (Bremer, 1988
; Donoghue et al., 1992
) was also performed using AutoDecay version 3.0 (Eriksson, 1995
).
The heuristic search option was used to invoke minimum evolution criteria (Saitou and Imanishi, 1989
) with tree bisection reconnection (TBR) branch swapping and collapse of zero-length branches. The analyses were repeated 100 times with the RANDOM ADDITION option. A general time-reversible (GTR) model of base substitution was used to take into account multiple hits (multiple base substitution events). To account for among-site rate heterogeneity an alpha (
) shape parameter of the gamma (
) distribution was estimated according to the methods of Yang (1994)
.
To test the significance of suboptimal tree topologies, constraint trees were generated in MacClade 3.06. Constraint trees were made to hold the monophyly of Linanthus sensu Patterson (1993)
as well as holding the monophyly of the sections within Linanthus sensu Patterson. Competing tree topologies, both optimal and suboptimal, were evaluated with a Kishino–Hasegawa test (Kishino and Hasegawa, 1989
). This parametric test uses the differences in likelihood values of the trees as a test statistic, under the assumption that nucleotide sites are independent.
Maximum likelihood was then used to evaluate competing phylogenetic hypotheses obtained from the distance and parsimony searches, along with constraint trees generated in MacClade 3.06. A series of likelihood ratio tests were performed (on a variety of tree topologies) to determine which model of sequence evolution best fit the data (Huelsenbeck and Rannala, 1997
). We found the GTR- (as implemented in PAUP*, with a rate matrix estimated using maximum likelihood) to be the best fitting model. Empirical base frequencies were used and among-site rate heterogeneity was accounted for using the -distribution (Yang, 1994
). A total of ten rate categories were used, and the average rate category was represented by the mean. In addition, tree topologies were searched for using maximum likelihood, employing the parameters outlined above using heuristic search methods with TBR branch swapping and collapse of zero-length branches. These analyses were repeated 100 times with the RANDOM ADDITION option.
Estimation of lineage diversification times
Given the absence of a fossil record in the Polemoniaceae, the dating of the lineage becomes problematic. We assume that the present-day distribution of species traditionally classified as Linanthus and its relatives (i.e., the "temperate" members of Polemoniaceae) is tied to the development of a summer-dry (Mediterranean) climate in westernmost North America that began 15 mya (Axelrod, 1977, 1992
; Flowers and Kennet, 1994
; Baldwin, 1997
), therefore, a minimum age for the most recent common ancestor Polemonium and the Linanthus/Phlox/Leptosiphon clade is set at 15 million years. A gene tree estimate of the species phylogeny can be used to estimate the times of divergence within the ingroup. Branch lengths were then estimated by maximum likelihood with constraint to a rate-constant model of molecular evolution (molecular clock enforced). A GTR- model of sequence evolution was used. Because the total base frequency does not reflect the base frequency of the positions that are free to vary (Spicer, 1995
; Spicer and Pitnick, 1996
), base frequencies for variable positions were used in the analysis. The rate substitution matrix, as well as the -shape parameter, were calculated using maximum likelihood.
The likelihood ratio test (Felsenstein, 1993
; Goldman, 1993
; Yang, 1996a
) was used to test whether the clock could be rejected by the ITS and matK data sets. The null hypothesis of this test is that the rate of base pair substitution is constant over all branches of a given tree. First, a maximum log-likelihood value under an unconstrained GTR- model (log L0) was calculated using the maximum likelihood settings outlined above. Then a log-likelihood value was calculated for the constraint (clock enforced) tree (log L1). It is assumed that the difference, 2(log L1-log L0), is 
2 distributed with n-2 degrees of freedom (where n = number of taxa). If the maximum log-likelihood value without the clock enforced is significantly greater than the constrained model (e.g., P < 0.05), then the hypothesis of constant rate across all branches of the tree is rejected. The standard errors of the branch lengths were calculated using the computer program PAML (Yang, 1996b
). Branch lengths were then taken as being proportional to time in order to test Axelrod's (1977)
hypothesis of active speciation during the Pliocene.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
base substitution model (Kimura, 1980
) with an -shape parameter of 0.28 (estimated according to Yang, 1994
) resulted in the same 24 tree topologies uncovered with maximum parsimony and had a minimum evolution (ME) score of 0.29237.
|
|
). When constraint trees holding the monophyly of individual sections of Linanthus were tested using the Kishino–Hasegawa test, several were not significantly different from the most parsimonious trees (Table 4).
|
Phylogenetic analyses of ITS nuclear rDNA
A total of 650 bases were sequenced for the ITS-1, 5.8S, and ITS-2 regions. Using published sequences of the region to demarcate ITS boundaries, the length of ITS1 (248–253; C + G 54%), 5.8S (164; C + G 53%), and ITS2 (221–228; C + G 53%) are similar to those reported in the Polemoniaceae (Baldwin et al., 1995
; Porter, 1997
) (Table 3). Within ITS1, 5.8S, and ITS2 there were 122, 4, and 96 variable sites, respectively (70, 0, and 58 positions were potentially parsimony informative). Because of the lack of any informative characters, the 5.8S region was removed from all phylogenetic analyses.
Phylogenetic analysis of the ITS 1 and 2 data resulted in 397 minimum-length trees of 795 steps, with a CI of 0.5082 (0.4438 excluding uninformative characters) and a RI of 0.7588. A strict consensus tree of the 397 most parsimonious trees is presented in Figure 2. A distance analysis using a GTR- model of base substitution with an -shape parameter of 0.54 (estimated according to Yang, 1994
) resulted in an additional tree topology with a ME score of 1.63546. The major difference between the maximum parsimony and minimum evolution tree topologies is the position of Li. jonesii.
|
|
|
|
Linanthus clade
An ILD test with 1000 replicates was performed on the 16 taxa of the Linanthus clade yielded a nonsignificant result (P = 0.788), suggesting the incongruence was not due to the members in this clade. A maximum parsimony analysis for the combined matK and ITS data with these 16 taxa resulted in nine most parsimonious trees, which were identical to the trees recovered when trees were searched for using only the ITS data.
Leptosiphon clade
When an ILD test with 1000 replicates was performed on the 28 taxa of the Leptosiphon clade, a significant result (P = 0.001) was obtained, suggesting the incongruence was due to some of the members in this clade. Twenty-eight additional data sets were created by systematically removing one terminal taxon from each data matrix to try and identify which taxon/taxa may be causing the incongruence. An ILD test with 1000 replicates was performed on each of these culled data matrices, each time resulting in a significant value (P < 0.01), suggesting no single taxon was causing the observed incongruence. Next, additional data sets were created by removing all combinations of two taxa (378 data sets in all). Once again, all ILD tests resulted in significant values (P < 0.01). Upon visual inspection of the matK and ITS strict consensus trees, several taxa (Li. nudatus, Li. liniflorus, Li. harknessii, and Li. septentrionalis A and B) appear to be more likely candidates to be causing the observed incongruence than others. When these five taxa (two samples of Li. septentrionalis) were removed from the data set, an ILD test did yield a nonsignificant value (P = 0.156). Since this nonsignificant result may have been an artifact of removal of a significant number of taxa, we wanted to test whether the removal of any five (random) taxa would also yield a nonsignificant value. Because there are a possible 98 280 possible combinations of data sets with five taxa removed from them, this was only explored heuristically. A total of 100 data sets were created, by removing five random taxa, and then an ILD test was performed on them. In all cases a significant value (P = 0.01) was obtained. A maximum parsimony analysis was then performed on a combined matK and ITS data set (minus Li. nudatus, Li. liniflorus, Li. harknessii, and Li. septentrionalis A and B). A total of six tree topologies, of 340 steps, were uncovered. Once again, tree topologies recovered from a "culled" data set on the combined data were no different than the trees obtained from the ITS data set by itself. When an ILD test was performed on the original data set minus Li. nudatus, Li. liniflorus, Li. harknessii, and Li. septentrionalis A and B a nonsignificant result was also obtained (P = 0.067). A maximum parsimony analysis on this culled data set, however, did not retrieve trees that differed in topologies than the original ITS data set with the five taxa being pruned.
Rate-constant evolution in Linanthus
Using the log-likelihood ratio test, the assumption of a molecular clock is strongly rejected (P < 0.0001) for the ITS data set; we failed to reject a clock for the matK data set (P = 0.0664) (Table 7). Although we rejected the clock for the ITS data, we estimated divergence times from the rate-constant tree to compare with dates estimated from the matK data set. Only dates for nodes showing congruence between the ITS and matK data were estimated (Figs. 4 and 5). If the lineage containing all the members of the ingroup and outgroup taxa is estimated to be 15 million years old, the following can be suggested: (1) a common ancestor of the ingroup taxa arising 11–13 mya, (2) a common ancestor of the Linanthus clade arising 6–8.5 million years, and (3) a common ancestor of the lineage uniting the Phlox–Leptosiphon lineage arising 11 mya. Even though ITS did not appear to show clock-like evolution, the estimated times of divergence are surprisingly similar to that of the matK data set for these lineages (Figs. 4 and 5). Standard errors for the divergence times inferred from the matK region are presented in Table 8. Due to the nonclock-like nature of the ITS data, PAML was unable to calculate standard errors for the rate-constant constrained GTR- model.
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Porter, 1997
). Given the lack of resolution of the matK gene data (nearly half of all variable sites are parsimony uninformative), little can be inferred about the species relationships within either the Linanthus clade or the Leptosiphon clade (Fig. 1). There is, however, moderate to substantial support for some groupings. Within the Linanthus clade, the members of Leptodactylon form a weakly supported group (64%), as does the species pair of Li. dianthiflorus and Li. uncialis (70%). The matK data also support the taxon Gilia filiformis falling outside or basal to this lineage (81%) (Figs. 1 and 4).
Within Leptosiphon clade, matK does support sect. Siphonella (Li. nuttallii, Li. pachyphyllus, Li. floribundus, and Li. laxus) as a monophyletic group, although weakly (65%). In addition, there is also support for the Li. bolanderi complex (Li. bolanderi, Li. rattanii, and Li. ambiguus) and a group consisting of Li. jepsonii, Li. bicolor, and Li. androsaceus (all members of the Li. androsaceus complex). One of the more interesting groupings is that of Li. aureus and Li. breviculus, two morphologically very different taxa. There is also support for grouping Li. harknessii and Li. septentrionalis as well as Li. montanus and Li. ciliatus. Sister to the Leptosiphon clade is a strongly supported lineage (100%) consisting of the species of Phlox.
The ITS data, from a more rapidly evolving region, show similar support for some of the same groups (Figs. 2 and 3); there is strong support for the monophyly of sect. Siphonella; the members of the Li. bolanderi complex form a well-supported group, and Li. septentrionalis and Li. harknessii appear as sister taxa. The ITS data also show strong support for the members of Phlox being sister to the Leptosiphon clade.
Within the Linanthus clade, the ITS data also support Li. dianthiflorus and Li. uncialis as sister taxa as well as support for both Li. bigelovii and Li. dichotomus and Li. arenicola and Li. killipii as sister taxa. Decay values support a lineage consisting of the members of sect. Linanthus (Li. bigelovii, Li. jonesii, Li. dichotomus, and Li. arenicola) and Li. parryae, as well as Li. killipii. Although there is little to no bootstrap support, there is decay branch support for placing the members of Gilia included here (G. maculata and G. filiformis) outside and basal to rest of the species in the Linanthus clade.
The ITS data neither support nor reject a monophyletic Leptodactylon, which has been traditionally recognized. Since all the members of Leptodactylon were not included, this might be due to sampling error. A monophyletic Leptodactylon was, however, uncovered in the minimum evolution search of the ITS data. Regardless of the monophyly of Leptodactylon, the members appear basal to the lineage consisting of sect. Linanthus, plus Li. parryae, and Li. killipii, and is nested inside the Linanthus clade. A tree topology constraining the monophyly of Leptodactylon was not significantly different than the most parsimonious tree topologies (Table 4).
Incongruence between nuclear and chloroplast gene trees
Incongruence between gene trees may arise from a variety of circumstances, including sampling (insufficient data collection or sequencing errors), organismal-level phenomenon (hybridization, lineage sorting, convergence at the nucleotide level, and horizontal gene transfer), or genetic-level factors (selection, rate heterogeneity, and base/codon composition biases) (Pamilo and Nei, 1988
; Wendel and Doyle, 1998
). In this study, both the matK and ITS support two distinct lineages of Linanthus. In addition, both loci lend support for the position of Phlox nested within Linanthus (in the traditional sense) and basal to members of sects. Leptosiphon, Siphonella, Dactylophyllum, and Pacificus (the Leptosiphon clade), as well as grouping members of Leptodactylon with sects. Linanthus and Dianthoides in the Linanthus clade.
Results from a series of ILD tests suggest that the majority of the conflict between the two regions is in the species relationships within the Leptosiphon clade. Removal of several "problematic" taxa (Li. nudatus, Li. liniflorus, Li. harknessii, and Li. septentrionalis A and B) did result in a nonsignificant result (P = 0.067) when the IDL test was performed on the culled data set (entire data set minus these five taxa). However, the combined matK and ITS data set did not retrieve trees that differed from the ITS data by itself. This is not surprising based on the number of phylogenetically informative characters between the ITS and matK data sets (194 vs. 63 for matK). It appears that the ITS data are swamping the phylogenetic signal of the matK region.
Taxonomic implications
Based on sequence data, Linanthus, as traditionally recognized, is polyphyletic and consists of two distinct clades. One (the Leptosiphon clade) contains sections Leptosiphon, Siphonella, Dactylophyllum, and Pacificus and is sister to Phlox. The other (the Linanthus clade) contains sects. Dianthoides and Linanthus, the genus Leptodactylon plus several species of Gilia, and is sister to the Leptosiphon–Phlox clade. Despite the apparent lack of monophyly in four of the six sections of Linanthus, none is fragmented across the two major linanthoid clades. The phylogenetic position of Phlox nested within Linanthus will likely require the two clades to be recognized as distinct genera, as proposed by Porter and Johnson (in press)
. The Linanthus clade contains the type species, Li. dichotomus, and retains the name Linanthus. The Leptosiphon clade contains members of sect. Leptosiphon. Leptosiphon was originally recognized as a genus in 1833 (Leptosiphon androsaceus Benth., as adopted by Porter and Johnson, in press
); accordingly, it is the appropriate name for the genus consisting of the group that is sister to Phlox.
Traditionally recognized sections of Linanthus defined by suites of morphological characters (Grant, 1959
) show varying degrees of cohesion based on molecular evidence. Section Siphonella appears to be monophyletic and nested within a clade composed of some, but not all members of sect. Leptosiphon. Section Pacificus is monospecific, thus its monophyly is moot. Its alignment with Li. liniflorus and Li. lemmonii, however, is of interest. It shares with Li. liniflorus a similar habit—tall, generally unbranching annuals—and relatively large, open, funnelform corollas; whether these similarities have evolutionary significance remains unclear.
Section Leptosiphon is characterized primarily by long-salverform corollas clustered into head-like inflorescences; however, it is polyphyletic based on molecular data. One lineage, consisting of Li. androsaceus, Li. bicolor, Li. jepsonii, and Li. parviflorus, forms a monophyletic group sister to sect. Siphonella. Another Leptosiphon clade member, Li. acicularis, is basal to the Li. androsaceus–Siphonella clade. All five of these species are characterized by calyces with relatively broad green sepals united by very narrow hyaline membranes. The remaining four species of sect. Leptosiphon sampled (Li. montanus/ciliatus/nudatus/breviculus) possess calyces with broader membranes connecting the relatively narrower green sepals. They occur in a phylogenetically distant clade that includes two species of sect. Dactylophyllum (Li. aureus and Li. jamauensis). The appearance of sect. Leptosiphon in two distinct clades based on molecular data and calyx morphology calls into question the integrity of the section, despite the seeming uniformity of corolla morphology. Accordingly, long-salverform corollas appear to have evolved twice in this group.
Section Dactylophyllum traditionally has been distinguished by having flowers borne on thread-like pedicels (with one exception, Li. lemmonii, having sessile flowers), thread-like stems, and short, thread-like leaf lobes. The molecular data, however, suggest that members of this section belong to as many as four distinct lineages. One of these lineages consists of Li. harknessii and Li. septentrionalis. These two species have very small corollas, often included within the calyx, which Grant and Grant (1965)
associated with autogamy. Biogeographically these two species are distinct within the section in having essentially montane and intermountain distributions. A second lineage, sister to all of the above, includes species common to foothill and grassland localities in California and Oregon (Li. filipes, Li. pygmaeus, Li. bolanderi, Li. rattanii, and Li. ambiguus), with one species (Li. pusillus) in similar habitats in Chile. Members of this lineage also tend to have corollas larger than those of Li. harknessii and Li. septentrionalis, implying fundamental differences in pollinator systems. Floral features and biogeography both support the monophyly of the harknessii–septentrionalis lineage and the woodland lineage of sect. Dactylophyllum.
The rest of sect. Dactylophyllum does not show any particular pattern of phylogenetic uniformity. One pair of pedicellate species, Li. aureus and Li. jamauensis, are part of a clade that contains the remainder of the members of sect. Leptosiphon. There are no morphological similarities shared by these two species, other than those that characterize sect. Dactylophyllum. The only feature that they have in common is their general desert habitat, although this may be coincidental. The remaining two members of sect. Dactylophyllum (Li. liniflorus and Li. lemmonii) appear basal within the Leptosiphon clade along with Li. grandiflorus (sect. Pacificus).
Bell, Patterson, and Hamilton (in press)
provide evidence that neither sect. Linanthus nor sect. Dianthoides is monophyletic, but rather represent, along with Leptodactylon, one large monophyletic lineage (i.e., the Linanthus clade). Overall we feel there is no reason to try to maintain the currently recognized sections in either Linanthus or Leptosiphon, since the monophyly of most are not supported by these molecular data. We argue further that there is no compelling reason to define any section-level taxa within Linanthus or Leptosiphon (as proposed by Porter and Johnson, in press
).
Biogeography of Linanthus
Within the present study the three major clades show various degrees of fidelity to biogeographic patterns. The Linanthus clade shows the strongest pattern, occurring primarily in southern California. With the exception of northern populations of Li. dichotomus (ranging to the northern Coast Ranges of California) the annual members of this group are distributed throughout the mountains of southern and Baja California and the deserts of southwestern United States and northern Mexico. The perennial members of the Linanthus clade (i.e., Leptodactylon) have diversified and expanded their range to arid and montane regions throughout western North America. This may be due to their suffrutescent habit, which could have allowed them to take advantage of colder climate during the Pleistocene. The molecular clock trees, however, suggest that these species evolved the perennial habit and began to diversify prior to the beginning of the Pleistocene. This could have been in response to an increase in drying experienced in the Pliocene and not necessarily to cooler temperatures.
The Phlox clade (±60 spp.) is distributed throughout North America and extends into Siberia. Given the limited sampling of this clade, it is difficult to draw many conclusions about the diversification of this widespread group. Further investigation into this clade and Phlox in particular is warranted.
The Leptosiphon clade is distributed throughout most of California with most species being present north of the Transverse Ranges; however, few conclusions can be drawn herein that relate phylogenetic relationships to biogeography. Most species have modest distributions, although some (e.g., Li. nuttallii, Li. bicolor, Li. ciliatus) occur across wide ranges. Some groups, such as the Li. septentrionalis/harknessii clade and the Li. filipes/pusillus/bolanderi/rattanii/ambiguus/pygmaeus clade, appear to have radiated within more local geographic and ecological settings. Species such as Li. lemmonii and the group containing Li. aureus, Li. breviculus, and Li. jamauensis may have secondarily returned back to the desert communities of southern California. The Chilean distribution of Li. pusillus appears to be due to a dispersal event. Aside from these conclusions little else can be inferred about biogeographic patterns in the Leptosiphon clade.
Within both lineages (the Linanthus clade and the Leptosiphon clade) the molecular clock trees suggest a considerable amount of diversification during the Pliocene (between 5 and 1.2 mya). Long terminal branches, connected by short internal ones, suggest that members of both groups may have undergone relatively rapid radiations. This pattern of diversification is consistent with Axelrod's hypothesis of active speciation during this time in response to spreading drought. This pattern is also in agreement with geological evidence of considerable change in California's landscape, including the final elevations of the Sierra Nevada–Cascade axis, the Transverse Ranges, and the Peninsular Ranges, which occurred approximately at this time. Development of new microhabitats, due to geologic activity, in conjunction with increased drying, could have caused an adaptive radiation of annual taxa throughout California.
The dearth of dependable, easily measured morphological characters in Linanthus has hampered attempts to reconstruct the phylogeny of this group and to make conclusions about its biogeographical distribution. The molecular phylogeny presented herein has been useful in elucidating relationships within and among the taxa under study; yet this study should be regarded as only an initial phase. Clearly Linanthus as has been traditionally regarded since the last century is not monophyletic, but rather consists of two monophyletic lineages. Within each clade species relationships are hypothesized, and these hypotheses can be tested using additional sequence data from different sources. Of equal importance is the potential for the phylogenetic relationships presented here to serve as models to examine the evolution of morphological features. Only through this approach is it possible to assess which morphological characters are homologous and which accurately reflect evolutionary history in this group.
|
FOOTNOTES |
|---|
2 Author for correspondence, current address: Harvard University Herbaria, 22 Divinity Avenue, Cambridge, Massachusetts 02138 USA (e-mail: cbell{at}oeb.harvard.edu)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
———. 1992 Miocene floristic change at 15 MA, Nevada to Washington, USA. The Palaeobotanist 41: 234–239
Baldwin, B. G. 1996 Phylogenetics of the California tarweeds and the Hawaiian silversword alliance (Madiinae; Heliantheae sensu lato). In D. J. N. Hind, H. Beentje, and G. V. Pope [eds.], Proceedings of the international Compositae conference, Kew, 1994, vol. 1, Systematics, Royal Botanic Gardens, Kew, UK
———. 1997 Adaptive radiation of the Hawaiian silversword alliance: congruence and conflict of phylogenetic evidence from molecular and non-molecular investigations. In T. J. Givnish and K. J. Sytsma [eds.], Molecular evolution and adaptive radiation, 103–128. Cambridge University Press, Cambridge, UK
———, and 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
———, ———, J. M. Porter, M. F. Wojciechowski, C. C. Campbell, and M. J. Donoghue. 1995 The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247–277[CrossRef][ISI]
Baum, D. A., K. J. Sytsma, and P. C. Hoch. 1994 A phylogenetic analysis of Epilobium (Onagraceae) based on nuclear ribosomal DNA sequence. Systematic Botany 19: 363–388[CrossRef][ISI]
Bell, C. D. 1998 Molecular phylogeny of Linanthus (Polemoniaceae). M.A. thesis, San Francisco State University, San Francisco, California, USA
———, R. Patterson, and L. A. Hamilton. In press Sectional integrity in Linanthus (Polemoniaceae): a molecular phylogeny of section Dianthoides. Systematic Botany.
Bremer, K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803[CrossRef][ISI]
Buxton, E. G. 1994 Linanthus latisectus (Polemoniaceae), a newly described species from the California North Coast Range. Madroño 41: 178–184
Cullings, K. W. 1992 Design and testing of a plant-specific PCR primer for ecological and evolutionary studies. Molecular Ecology 1: 233–240
de Geofroy, I. 1998 Systematics of alpine Polemonium (Polemoniaceae). M.A. thesis, San Francisco State University, San Francisco, California, USA
Donoghue, M. J., R. G. Olmstead, J. F. Smith, and J. D. Palmer. 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345
Doyle, J. J., and J. L. Doyle. 1987 A rapid isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11–15
Epling, C., and T. Dobzhansky. 1942 Genetics of natural populations. VI. Microgeographic races in Linanthus parryae. Genetics 27: 317–332
Eriksson, T. 1995 AutoDecay version 3.0. Distributed by the author. Department of Botany, University of Stockholm, Sweden
Farris, J. S., M. Källersjö, A. G. Kluge, and C. Bult. 1995 Testing significance of incongruence. Cladistics 10: 315–319[CrossRef][ISI]
Felsenstein, J. 1983 Statistical inference of phylogenies. Journal of the Royal Statistical Society A 146: 246–272[CrossRef]
———. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791[CrossRef][ISI]
———. 1993 Phylip (Phylogenetic Inference Package), version 3.5c. Department of Genetics, University of Washington, Seattle, Washington, USA
Flowers, B. P., and J. P. Kennet. 1994 The middle Miocene climate transition: East Antarctic ice sheet development, deep ocean circulation and global carbon cycling. Palaeogeography, Palaeoclimatology, Palaeoecology 108: 537–555
Galen, C. 1989 Measuring pollinator-mediated selection on morphometric floral traits: bumblebees and alpine skypilot Polemonium viscosum. Evolution 43: 882–890
———. 1990 Limits to the distribution of alpine tundra plants: herbivores and the alpine skypilot, Polemonium viscosum. Oikos 59: 335–358
———. 1992 Pollen dispersal dynamics in an alpine wildflower, Polemonium viscosum. Evolution 46: 1043–1051
Gene Codes Corporation, Inc. 1994 Sequencher® 2.1 reference, advanced, user friendly software tools for DNA sequencing. Gene Codes, Corporation, Inc., Madison, Wisconsin, USA
Golden, M. J., and R. Patterson. 1988 Taxonomic implications of pollen surface morphology in Linanthus and Leptodactylon (Polemoniaceae). American Journal of Botany 75 pt 2: 176
Goldman, N. 1993 Statistical tests of models of DNA substitution. Journal of Molecular Evolution 36: 182–198[CrossRef][ISI][Medline]
Grant, V. 1959 Natural history of the phlox family. Martinus Nijhoff, The Hague, The Netherands
———. 1998 Primary classification and phylogeny of the Polemoniaceae with comments on molecular cladistics. American Journal of Botany 85: 741–742[Abstract]
———, and K. A. Grant. 1965 Flower pollination in the Phlox family. Columbia University Press, New York, New York, USA
Hasegawa, M., and M. Fujiwara. 1993 Relative efficiencies of the maximum likelihood, maximum parsimony, and neighbor joining methods for estimating protein phylogenies. Molecular Phylogenetics and Evolution 2: 1–5[CrossRef][Medline]
Hillis, D. M., and J. J. Bull. 1993 An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Systematic Biology 42: 182–192[CrossRef][ISI]
Huelsenbeck, J. P. 1995 Performance of phylogenetic methods in simulation. Systematic Biology 44: 17–48
———, and B. Rannala. 1997 Phylogenetic methods come of age: testing hypotheses in an evolutionary context. Science 276: 521–566
Irwin, D. M., T. D. Kocher, and A. C. Wilson. 1991 Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32: 128–144[CrossRef][ISI][Medline]
Johnson, L. A., and D. E. Soltis. 1995 Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae): using matK sequences. Annals of the Missouri Botanical Garden 82: 149–175[CrossRef][ISI]
———, and ———. 1998 Assessing congruence: empirical examples from molecular data. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II, Chapman and Hall, New York, New York, USA
———, J. L. Schultz, D. E. Soltis, and P. S. Soltis. 1996 Monophyly and generic relationships of Polemoniaceae based on matK sequences. American Journal of Botany 83: 1207–1224[CrossRef]
