| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2Department of Ecology and Evolutionary Biology and the Natural History Museum & Biodiversity Research Center, University of Kansas, Lawrence, Kansas 66045 USA; 3Jardín de Aclimatación de la Orotava (ICIA), Puerto de la Cruz, Tenerife, Canary Islands, Spain
Received for publication December 22, 2005. Accepted for publication May 2, 2006.
ABSTRACT
Plants of oceanic islands, often remarkably divergent morphologically from continental relatives, are useful models for studying evolution and speciation because evolution is telescoped in time and space. Prior studies revealed little DNA sequence variation within the clade of ca. 10 Canary Island species of Tolpis, which precluded resolving species relationships. The present study assessed the utility of automated analysis of inter-simple sequence repeat (ISSR) loci for resolving relationships within the clade using 264 individuals from 36 populations of all recognized species and three undescribed morphological variants. Similarity (Dice coefficient) and Fitch parsimony were used to generate neighbor-joining (NJ) and strict consensus trees (MP), respectively. All individuals of the morphologically distinct endemic species formed clusters in both trees. There is also support for clusters of two undescribed variants in the NJ tree. Individuals from a morphologically variable complex consisting primarily of two species are not well resolved at population or species levels. The NJ and MP trees are not congruent at deeper levels, including relationships among species. Results are interpreted in terms of the biology of the species, and the utility of automated analysis of ISSR markers for interpreting patterns of evolution of Tolpis in the Canary Islands is discussed.
Key Words: Asteraceae • Canary Islands • Cichorieae • inter-simple sequence repeat • ISSR • Macaronesia • Tolpis
Some of the most interesting model systems for studying plant speciation and evolution come from groups that have radiated recently because in such lineages it is less problematic to infer features associated with speciation from those that have accumulated subsequent to speciation (Templeton, 1981
; Coyne and Orr, 2004
). However, estimating phylogeny for such groups employing DNA sequence data often represents a major challenge because there usually is not sufficient variation in the sequences of commonly employed regions, such as the internal transcribed spacer regions of nuclear ribosomal DNA (ITS) and noncoding regions of chloroplast DNA (cpDNA; Pelser et al., 2003
; Crawford and Mort, 2004
; Koopman, 2005
). While there is a continual search for more variable gene regions to sequence (Mort and Crawford, 2004
; Small et al., 2004
; Shaw et al., 2005
), there has also been increasing interest in the use of hypervariable, arbitrarily amplified dominant markers (e.g., AFLP, ISSR, RAPD) to generate phylogenies (Crawford and Mort, 2004
; Bussell et al., 2005
; Koopman, 2005
; Archibald et al., in press
).
Lineages that have radiated in oceanic archipelagos provide especially attractive systems for assessing the utility of hypervariable DNA markers for resolving relationships because many represent recent, rapid radiations (Baldwin et al., 1998
). Insular lineages often rapidly evolve a wide array of morphological forms and occupy different ecological zones within a small, isolated geographic area (Carlquist, 1974
). Diversification typically occurs with minimal development of postpollination isolating factors, making it feasible to examine the inheritance of a diverse array of features through controlled crossing studies (e.g., Gillett and Lim, 1970
; Lowrey, 1986
; Whitkus et al., 2000
). A highly resolved phylogeny for an insular group, although difficult to infer from molecular data, is desirable because it could serve as the necessary framework for studying patterns of evolution and diversification (e.g., Mort et al., 2002
; Baldwin, 2003
; Haworth and Baum, 2005).
The clade of Canary Island species in the small genus Tolpis Adanson (Asteraceae) exemplifies the challenge of estimating a phylogeny for an insular group. Tolpis occurs only on the five western, relatively mesic islands of El Hierro, Gran Canaria, La Gomera, La Palma, and Tenerife (Fig. 1). The six Canarian endemic species recognized by Jarvis (1980)
are T. coronopifolia (Desf.) Biv., T. crassiuscula Svent., T. glabrescens Kämmer, T. laciniata (Sch. Bip. ex Webb and Berthel.) Webb, T. lagopoda C. Sm. ex Buch, and T. webbii Sch. Bip. ex Webb and Berthel. Two previously described species placed in synonymy under T. laciniata by Jarvis (1980)
are T. calderae Bolle and T. proustii Pitard in Pitard and Proust. One species, T. barbata (L.) Gaetn., occurs commonly on the same five islands as the endemic species, but it is also widely distributed in other Macaronesian archipelagos as well as throughout the Mediterranean basin (Jarvis, 1980
). Park et al. (2001)
reconstructed a phylogeny for 28 genera of Cichorieae using parsimony analyses of ndhF sequence data; they resolved the circumscription of Tolpis as a genus, but because only two species of Tolpis from the Canary Islands were included in their analyses, the results were uninformative with regard to relationships within this lineage. Attempts to estimate relationships among the Canarian species of Tolpis have produced topologies with almost no resolution or support; these have included studies of cpDNA restriction sites (Moore et al., 2002
), ITS sequences (R. K. Jansen, University of Texas, personal communication), and a number of the ostensibly more variable noncoding cpDNA sequences (Shaw et al., 2005
; M. Mort et al., unpublished data). Although Moore et al. (2002)
did not resolve relationships among the Canary Island endemic species, their analyses do provide 100% bootstrap support for the lineage comprising the species of Tolpis from the Canary Islands with the inclusion of T. farinulosa from the Cape Verde Islands (six Canarian endemic species were sampled in addition to three undescribed morphological variants). Mort et al. (2003)
conducted a preliminary study of five described and one undescribed species of Tolpis endemic to the Canaries to test the utility of hypervariable inter-simple sequence repeat (ISSR) markers for resolving relationships within this group. Both neighbor-joining (NJ) and maximum parsimony (MP) analyses resolved each of the six species as a distinct group, but population sampling was very limited, and the most common and variable species, T. laciniata, was not sampled.
|
). In addition, some species are morphologically uniform, rare, and have restricted distributions on a single island, whereas other species are morphologically variable, abundant, and of widespread occurrence in a variety of habitats on more than one island (Jarvis, 1980
). The present ISSR study expands on the taxon sampling of Mort et al. (2003)
to include individuals from (usually multiple) populations of each of the species of Tolpis that occur on the Canary Islands, as well as several populations that appear to be new species. In addition, the present study employs automated analysis of ISSR markers, whereas Mort et al. (2003)
employed manual methods. The potential benefits of conducting analyses of ISSRs in an automated framework have recently been reviewed by Crawford and Mort (2004)
.
The general purpose of the present study was to assess the utility of automated analysis of ISSR loci for producing a phylogeny of Tolpis in the Canary Islands. In one sense, we are using Tolpis as a model of a recent, rapid radiation on which we can test the hypothesis of Bussell et al. (2005)
that the most appropriate level for the use of hypervariable DNA markers is below the level at which ITS is useful for resolving relationships. Given the aforementioned paucity of DNA sequence variation in Tolpis, it is an appropriate subject for a study using hypervariable DNA markers. Another, somewhat parallel, objective was to use automated analysis of ISSR markers to produce a phylogeny for Tolpis that would serve as a framework for exploring the evolution of the many intriguing features of Tolpis in the Canary Islands.
MATERIALS AND METHODS
ISSR survey
The nine described species of Tolpis that occur on the Canary Islands were included in this study as well as three undescribed morphological variants from the Canary Islands, T. farinulosa from the Cape Verde Islands, and T. succulenta from Madeira (Table 1). The circumscriptions of several species of Tolpis have been questioned (see Discussion) and several of the named species may be subdivided further in the future. Regardless, 36 populations with a total of 264 individuals were collected from the field (Table 1). Leaf material (~20 mg) was collected on silica gel from 1–18 individuals within each population. Voucher specimens were deposited in the herbarium of Jardín de Aclimatación de la Orotava (ORT). DNA was extracted from leaf samples using standard hexadecyltrimethyl-ammonium bromide (CTAB) methods (Doyle and Doyle, 1987
, as modified by Mort et al., 2001
) and cleaned using an EluKwik DNA purification kit (Schleicher and Schuell, Keene, New Hampshire, USA).
|
PCR products were electrophoresed, visualized, and scored using a CEQ 8000 Genetic Analysis System (Beckman Coulter, Fullerton, California, USA) with the Fragment Analysis Module. Primers were labeled with D4 fluorescent dye for detection by the automated sequencer and peak sizes were estimated using a custom size standard (D1 MapMarker 1000, BioVentures, Murfreesboro, Tennessee, USA) that was included in each sample prior to electrophoresis. This standard includes DNA fragments ranging in size from 50 to 1000 bp; its inclusion permits very precise sizing of the various ISSR fragments produced via PCR.
The fragment analysis software provided with the CEQ permits adjustment of several locus-scoring parameters and of the range of peaks to be included. Fragment analysis parameters designate the minimum acceptable peak slope (default = 10) and relative height (default = 10%) necessary for a given peak to be accepted as a locus. These default values performed as well or better than alternative values that were examined for our data set. An "AFLP Analysis" was conducted to produce a diallelic (1 = band present, 0 = band absent) data set for further analysis, using a maximum bin width of 3.00 nt, no Y threshold, and excluding fully populated bins. Although this function is named after AFLP markers, it is equally appropriate for use with the dominant ISSR data and simply produces a data matrix of 0's and 1's from the scored peaks. Bin size can be set between 0.05 and 10 nt; any scored peaks within one bin will be counted as one locus. The estimated peak size of scored fragments often varied by 1–2 nucleotides in our replicate runs. This could be due to a variety of factors including PCR error, power fluctuations, and the concentration of the electrophoresis buffer. We therefore used a conservative bin size of three nucleotides. A comparison of the data sets generated using either a bin width of one nucleotide or three nucleotides revealed that the latter provided cleaner results for our data.
Primers were screened for utility within Tolpis using manual ISSR methods (i.e., employing non-labeled oligonucleotide primers and visualizing the amplified fragments in an agarose gel stained with ethidium bromide) due to the expense of fluorescently labeled primers (see Archibald et al., 2005
, in press
). Six primers were optimized for this study: (GT)7YG, (CTC)7RC, (CA)7RY, (GA)9H, (AGA)9B, and (CT)9D. For simplicity, these primers will be referred to as primer I, II, III, IV, V, and VI. The annealing temperature used for each primer was 49°C, 55°C, 50°C, 57°C, 55°C, and 57°C, respectively. All replicates of primer I in one individual within T. proustii failed, and these loci were scored as missing; successful amplifications were achieved in all other cases. For all primers, loci smaller than 75 bp or greater than 1000 bp were excluded because visualization and sizing of these loci was less reliable than in the intervening size range. For some primers, there were very strong, artifactual peaks at the beginning of the run. These peaks were between 50 and 200 bp, depending on the primer, and in some cases disrupted the analyses. This latter occurrence was corrected by excluding the size standard peaks in that size range (using the fragment analysis parameters), after which the program will not recognize sample peaks in that range. Any remaining artifactual peaks were excluded from the "AFLP Analysis," including loci smaller than: 90 bp for II, 250 bp for III, and 200 bp for IV. Primers IV and VI had multiple runs that prematurely terminated before the 1000 bp marker peak; scored loci larger than 850 bp for IV and 950 bp for VI were thus excluded.
ISSR data analyses
These data were analyzed using a variety of distance-based and character-based techniques, including trials with different coefficients and different weighting schemes, respectively. A comparison and discussion of the concerns and benefits of these varying methods is given in Archibald et al. (in press
). The results from two methods of analysis will be discussed here—neighbor-joining (NJ; Saitou and Nei, 1987
) analyses of Dice's (1945)
coincident index and Fitch parsimony analyses (MP)—to compare results from distance and character-based trees. One benefit of NJ analyses using Dice's (1945)
index is the exclusion of "shared–absence" characters; this coefficient is equivalent to Nei and Li's (1979)
similarity coefficient: 2nXY / (nX + nY), where nXY is the number of shared loci, nX is the total number of loci in taxon X, and nY is the total number of loci in taxon Y. In contrast, parsimony analyses have the benefit of comprising only informative resolution (see Archibald et al., in press
, for further comparison of these methods). In both cases, the trees were rooted using T. succulenta as an outgroup because this species was shown to be closely related to the Canary Island taxa in a previous study of Tolpis (Moore et al., 2002
).
For the neighbor-joining analyses, a similarity matrix was constructed (using "DICE") and a NJ analysis was run using NTSYS vs. 2.02h (Rohlf, 1998
), testing for tied trees. The similarity matrix produced by NTSYS was also used to construct an identical NJ tree in PAUP* (Swofford, 2003
) because the resulting tree was easier to view. Relative bootstrap support values were produced using PhylTools (NEI coefficient, ignore missing data, 1000 replicates; http://www.dpw.wau.nl/pv/PUB/pt/, with NEIGHBOR/PHYLIP (with jumble option), and CONDENSE/PHYLIP (Felsenstein, 2004
).
Fitch parsimony analyses were conducted in PAUP* with the heuristic search option, MulTrees, no MaxTrees limit, swapping on all trees, and parsimony options set to collapse branches if minimum length is zero ("amb-"). Uninformative characters were excluded before all analyses and all remaining characters were equally weighted and unordered. One thousand quick searches with random taxon addition and NNI branch swapping were used to locate multiple islands of minimum-length trees, forming a starting pool of trees for more thorough searches employing tree bisection-reconnection (TBR; Maddison, 1991
). Relative support for clades was assessed via jackknife analyses (500 replicates, 37% deletion, emulate "Jac" resampling).
RESULTS
Number of loci and trees
Amplifications from the six primers produced 1628 scored loci (326, 293, 262, 232, 248, 267 loci for each primer, in order). Neighbor-joining analyses resulted in a single tree (Fig. 2); parsimony analyses resulted in five most parsimonious trees with 37 366 steps (Fig. 3).
|
|
Grouping of species
In both the NJ and MP trees, all individuals of T. coronopifolia, T. crassiuscula, T. glabrescens, and Tolpis sp. nov. 3 form exclusive groups (Figs. 2, 3). In the NJ tree, all of these groups have strong support (i.e., >86% bootstrap; Fig. 2). In the MP tree, T. coronopifolia also has strong (87%) jackknife support; there is weak support (60%) for T. crassiuscula; and support is less than 50% for T. glabrescens and Tolpis sp. nov. 3. In the NJ tree (Fig. 2), Tolpis sp. nov. 1 (only one population studied) receives strong bootstrap support whereas individuals of this taxon do not form an exclusive group in the MP tree (Fig. 3).
In both the NJ and MP trees, the two populations of T. barbata, each of which forms an exclusive group, do not group together (Figs. 2, 3). Also, individuals of T. calderae, T. proustii, and T. webbii occur in several places. Populations of T. laciniata and T. lagopoda are scattered throughout both the NJ and MP trees (Figs. 2, 3).
Deeper branches of trees
There is nearly complete resolution in the MP tree, but no jackknife support for the deeper branches (i.e., those which include more than two populations). The same pattern of low support (<50% bootstrap) for larger groups is also found in the NJ tree (Figs. 2, 3). A comparison of the NJ and MP trees shows that larger groups (clades) differ in the two trees (Figs. 2, 3). One example of this nonconcordance is that T. coronopifolia and T. crassiuscula group in the NJ tree, but these two species are not sister taxa in the MP topology. There are, however, several examples where two or more populations occur together in both trees. For example, T. proustii consistently forms a group with the two populations of T. laciniata from El Hierro, and T. webbii groups with some or all members of three populations of T. lagopoda from Tenerife (Figs. 2, 3). The more common situation, however, is for the positions of groups of populations and species to differ in the two trees.
DISCUSSION
Automated scoring of ISSR loci
Confidence in the inferences of any study of evolutionary relationships rests on the number of loci resolved and the accuracy with which those loci were scored as homologous. Automation of the scoring process saved a significant amount of time compared to manual scoring, and automated electrophoresis on polyacrylamide with an internal size standard should more accurately resolve a larger number of ISSR loci than agarose electrophoresis (Huang and Sun, 2000
; Liu and Wendel, 2001
; Crawford and Mort, 2004
; Archibald et al., in press
). Automated ISSR analyses employing six primers produced a data set of 1628 loci. This is an extremely large data set compared to previous studies of insular plants; for example, Mort et al. (2003)
employed five primers and agarose electrophoresis in their preliminary study of the Canary Island Tolpis, and the resulting data set consisted of only 48 loci. While an increase in number of discernable peaks is expected with the higher resolving power of an acrylamide gel (Crawford and Mort, 2004
), the number of scored loci reported here (an average of 271 loci per primer) is generally over three times greater than what has been reported in several other acrylamide-based ISSR studies (e.g., Blair et al., 1999
; Liu and Wendel, 2001
; Patzak, 2001
; Xu and Sun, 2001
; Nagaraju et al., 2002
; Schrader and Graves, 2004
). However, some of these studies scored loci up to only 500 bp, whereas in the present study we scored peaks up to 1000 bp. It is possible that a portion of the increased number of loci reported in the present study represent false peaks. However, the higher number of peaks we scored compared to other studies using comparable techniques seems reasonable when one considers that the majority of bands that are scored in manual ISSR studies of angiosperms are between 800 and 1600 bps (e.g., Mort et al., 2003
). Additionally, parsimony analyses of these data resulted in five trees whose strict consensus was nearly completely resolved; it appears that the signal in this data set has overwhelmed the noise.
Species delimitation in Tolpis
With the exception of T. barbata, which is widespread in Macaronesian archipelagos, southern Europe, and northern Africa (Jarvis, 1980
), Tolpis in the Canary Islands consists only of endemic species. The nonendemic T. barbata, together with the Canarian endemic species T. coronopifolia, T. crassiuscula, and T. glabrescens are easily recognizable morphologically (Jarvis, 1980
). All individuals of the two populations of T. coronopifolia and T. glabrescens, and all individuals of the population of T. crassiuscula, each form distinct groups in both the NJ and MP trees. Mort et al. (2003)
included two populations of T. crassiuscula in their ISSR study, and both populations grouped together in the NJ and MP trees. Thus, ISSR markers support the genetic cohesiveness of the three morphologically distinct endemic species of Tolpis in the Canary Islands.
In contrast to the three endemic species, the two populations of the morphologically distinct, non-endemic T. barbata do not occur together in either the NJ or MP trees (Figs. 2, 3), although each individual population does form a cohesive cluster. One of the populations is from the island of La Palma and the other from Tenerife. These same populations were shown to have different multilocus genotypes for 14 allozyme loci (Crawford et al., 2006
). Moore et al. (2002)
showed that T. barbata is in a clade sister to the Canary Island endemics, but it was equivocal whether T. barbata is native or was introduced to the islands from the continent. Results of the present study likewise are not informative on whether T. barbata is native or introduced; failure of the two populations to cluster could be explained by either scenario. If T. barbata is introduced, then the two populations could be the result of introductions from different source areas. If T. barbata arose on the islands, then selfing in this self-compatible species (D. Crawford et al., unpublished data) has fixed different combinations of loci in the two populations so that they now fail to group. Another possibility is that T. barbata was introduced from one source area, but has since diverged into several genetic types due to the accumulation of fixed differences via selfing. Regardless, T. barbata differs from T. coronopifolia, the other self-compatible species of Tolpis in the Canaries, in that the two populations of T. coronopifolia cluster together in both trees (Figs. 2, 3).
Canarian Tolpis also includes a widespread, morphologically variable, and ecologically diverse species complex (i.e., the T. laciniata–T. lagopoda complex; Crawford et al., 2006
). In addition to T. laciniata and T. lagopoda, this complex includes T. webbii, two described species (T. proustii from El Hierro and T. calderae from La Palma) not recognized by Jarvis (1980)
, and several morphological variants that may prove worthy of taxonomic recognition (A. Santos-Guerra, unpublished observations).
Tolpis webbii is a species particularly common in the volcanic crater of Las Cañadas in central Tenerife, with rare occurrences outside that area (Jarvis, 1980
). The species is distinct morphologically over most of its range with the defining characters of erect, linear-lanceolate, coarsely pubescent leaves restricted to basal rosettes and erect inflorescences (Jarvis, 1980
). However, where the species overlaps geographically with T. lagopoda, it is more variable with more lax, broader, and less pubescent leaves, and the inflorescence is less erect, all of which suggest hybridization with T. lagopoda (Jarvis, 1980
). The three populations of T. webbii examined in this study have the suite of features typical of the species, do not come from the area of overlap with T. lagopoda, and thus appear to represent "pure" T. webbii. Individuals from the same populations do not form exclusive groups in either NJ or MP, and the species does not form an exclusive group in either tree (Figs. 2, 3). However, in both trees, individuals of T. webbii do group together closely and group with several populations of T. lagopoda from Tenerife.
Tolpis proustii, which Jarvis (1980)
considered a variant of T. laciniata, was named to accommodate plants from the mountains of western El Hierro having large, glabrous, pinnatifid leaves and large capitula on short peduncles. Our results provide some support for the genetic distinctiveness of T. proustii because most individuals of this species form a group in the NJ tree and in turn cluster with plants of T. laciniata from El Hierro. The MP topology is similar to the NJ topology, with the majority of individuals of T. proustii occurring in a clade with T. laciniata from El Hierro (Figs. 2, 3). Thus, the ISSR data indicate that plants from El Hierro are more closely related to each other than they are to other Tolpis in the Canary Islands.
Tolpis calderae was described from plants with bipinnatifid, densely white-tomentose leaves with linear pinnae and elongated stems growing in the Caldera de Taburniente on La Palma. As with T. proustii, Jarvis (1980)
did not consider T. calderae worthy of taxonomic recognition and viewed it as a variant of T. laciniata. In the NJ tree, individuals of T. calderae form three distinct clusters. In the MP tree, individuals of T. calderae occur at four different places. ISSR data provide no support for the genetic cohesiveness of what has been called T. calderae.
Jarvis (1980)
presented an extensive discussion of morphological variation in T. laciniata, emphasizing the variation in size, dissection, and pubescence of the leaves. Jarvis (1980)
also commented that morphological variation sometimes occurs within as well as among populations of T. laciniata. In contrast to T. laciniata, Jarvis (1980)
viewed T. lagopoda as "... very similar in all its localities." The only exception noted was at localities on Tenerife where T. lagopoda appears to intergrade morphologically with T. webbii (Jarvis, 1980
). While Jarvis (1980)
commented on the level of variation within T. laciniata and T. lagopoda, he did not discuss features distinguishing the two species except in his key to the species. In the key, Jarvis (1980)
separated T. lagopoda from T. laciniata (and several other species) by "flowering axes markedly leafy, pendant to ascending" as contrasted with "flowering axes not markedly leafy, erect." The type specimens for the two species are clearly distinct, and plants matching the two extremes in development of leaves along the flowering axes occur. However, our field observations show that there is extensive variation in both leafiness and orientation of the floral axes of both species; this variation segregates among progeny from some natural populations when grown in the greenhouse (D. Crawford et al., unpublished observations). The populations assigned to either T. laciniata or T. lagopoda do not form clusters in either the NJ or MP trees; rather, these populations are scattered throughout the trees (Figs. 2, 3). The ISSR markers do not portray either T. laciniata or T. lagopoda as distinct, cohesive genetic entities, and this is concordant with the overlapping morphological variation between the species.
One morphological variant worthy of consideration for taxonomic recognition (A. Santos-Guerra, unpublished observations) includes populations Tolpis sp. nov. 3 A and B growing on the walls of two isolated canyons in the Adeje region, which is one of the three paleo-islands of Tenerife (Fig. 1). These large plants have highly branched stems, and the inflorescences also branch extensively and bear numerous capitula. All sampled individuals of Tolpis sp. nov. 3 form an exclusive group that is subdivided into two population-specific subclusters in both the NJ and MP trees (with one or both populations having strong jackknife/bootstrap support), thus supporting the interpretation of the populations from Adeje as distinct genetic entities. One subcluster comprises individuals collected from Barranco del Infierno (Tolpis sp. nov. 3 B); Bramwell and Bramwell (1974
, 1990
) considered morphological variants from this same barranco to be T. crassiuscula. Our results do not place Tolpis sp. nov. 3 B as a close relative of T. crassiuscula in either the NJ or MP topologies.
Tolpis sp. nov. 1, collected from northeast La Palma, is recognizable morphologically by its thin, glabrous leaves, and few, large capitula (A. Santos-Guerra, unpublished observations). In the NJ tree, all individuals from this population occur together with strong support, whereas in the MP tree one of the six individuals sampled is widely separated from the strongly supported clade of the other five (Figs. 2, 3). The ISSR data indicate that this morphological variant may be genetically distinct and perhaps worthy of taxonomic recognition. Morphological and biosystematic studies of this population and similar populations are in progress.
Tolpis sp. nov. 2, from the paleo-island of Teno on Tenerife (Fig. 1), has a combination of distinctive morphological features including a very woody base and more or less entire leaves along the stem (A. Santos-Guerra, unpublished observations). However, plants from this population occur scattered in four different places in the MP tree and at two locations in the NJ tree (Figs. 2, 3). Thus, ISSR markers offer no support for the recognition of this as a distinct entity.
Populations of T. laciniata from the island of La Gomera (T. laciniata A, C, and G) become very woody at the base, have coarse, deeply lobed leaves, several bracts immediately subtending the inflorescence, and large capitula. This combination of characters distinguishes these plants from other elements in the T. laciniata–T. lagopoda complex (A. Santos-Guerra, unpublished observations). In addition, these populations group together by similarity at allozyme loci (Crawford et al., 2006
). However, the three populations of T. laciniata from La Gomera occur scattered in both the NJ and MP trees (Figs. 2, 3). Despite morphological and allozyme evidence to the contrary, the ISSR data do not indicate that populations of T. laciniata from La Gomera are distinct from other members of the T. laciniata–T. lagopoda complex.
The strict-consensus MP tree is almost fully resolved. However, both the MP and NJ trees show essentially no support for the larger groups (i.e., those with more than two populations; Figs. 2, 3). Also, there are some notable differences between the two trees in the grouping of species and populations. For example, T. coronopifolia and T. crassiuscula form a group in the NJ tree but are in widely separated clades within the MP tree (Figs. 2, 3).
The utility of ISSR markers for resolving relationships
In a recent review, Bussell et al. (2005
, p. 19) concluded that the appropriate level for the application of arbitrarily amplified dominant markers, which include ISSRs, "appears to be below taxonomic levels at which ITS and other variable sequences can provide sufficient information ... and may be at the interface between phylogeny and population biology." These authors further stated that the markers might be useful for phylogenetic analyses involving species that are closely related and represent recent radiations. The Canarian clade of Tolpis is an ideal system for testing these hypotheses because the species are below the taxonomic level where sequences of ITS and cpDNA spacer regions are phylogenetically informative (Moore et al., 2002
; M. Mort et al., unpublished data; R. K. Jansen, University of Texas, personal communication). Also, it is clear that the endemic Tolpis species constitute a monophyletic, recently radiated lineage (Moore et al., 2002
). How well do our results for Tolpis support the hypotheses of Bussell et al. (2005)
regarding the utility of hypervariable DNA markers?
In general, ISSR markers are most useful in Tolpis for grouping individuals into populations and species. In most instances, those populations or species that form exclusive clusters have at least one of a combination of attributes, including small population size, isolated distribution, self-compatibility, and heterozygote deficiency at allozyme loci (D. Crawford et al., unpublished data). At both the population and species levels, it appears as if certain processes such as drift and inbreeding are acting to fix either diagnostic ISSR markers (grouping in MP tree) or combinations of markers (NJ tree). In contrast, ISSR markers fail to group populations or species that are self-incompatible, morphologically variable, and have large population sizes. It is possible that this reflects a true lack of evolutionary relatedness among populations of species such as T. laciniata and T. lagopoda, which is at least partially supported by their lack of morphological cohesiveness. The lack of clustering of these members of Tolpis may also be the result of the presence of a suite of ancestral polymorphic markers segregating within populations referable to more than one species in the current taxonomy of the genus (Jarvis, 1980
). Gene flow between populations could also serve to slow or prevent the sorting of particular loci or combinations of loci within populations. These processes would result in more ISSR loci segregating within populations than there are markers or arrays of markers that define populations or groups of populations.
ISSR markers were least successful in inferring relationships among species, even those species that were resolved as distinct groups of individuals (e.g., T. coronopifolia, T. crassiuscula, T. glabrescens). Because there is no phylogeny available for comparison with our results, our assessment of the limited utility of ISSR markers for inferring relationships among species is made on the basis of the different topologies in the NJ and MP trees, and the lack of relative branch support greater than 50% for the deeper branches in both trees. This situation contrasts with the resolution at the population level where the same 12 populations form exclusive groups in both the NJ and MP trees. This indicates that there is stronger signal in the ISSR data within populations than there is at the interspecific levels, at least for one third of the populations. The lack of stronger signal for interspecific comparisons could be caused by errors in assigning band homology; the rapid radiation of the early lineages, thus precluding the generation of markers in the common ancestors of the lineages; and mutations causing the loss of diagnostic loci within lineages. Our results for Tolpis suggest that for any group of plants it will be difficult to predict a priori at what taxonomic level ISSR or other arbitrarily amplified DNA markers will be useful characters for phylogenetic inference. Even within a small, monophyletic group such as Tolpis, the utility of the markers varies and is influenced by historical and biological factors.
FOOTNOTES
1 This research was supported by the Department of Ecology and Evolutionary Biology and the Natural History Museum & Biodiversity Research Center at the University of Kansas; a Kansas NSF EPSCoR Ecological Genomics postdoctoral award to J. K. A. and D. J. C.; and NSF DEB-0344883 to M. E. M. The authors thank Sr. D. Manuel Fernández-Galván, Instituto Canario de Investigaciones de Agrarias, and the Jardín de Aclimatación de la Orotava for logistical support in the Canary Islands, K. Nus for greenhouse assistance, and H. Cartwright, N. Levsen, and T. R. O'Leary for laboratory assistance. G. Ortiz produced Fig. 1 and C. P. Randle assisted with Figs. 2 and 3. B. Baldwin and J. Francisco-Ortega provided constructive comments that improved the manuscript. 
4 Author for correspondence (jkarchibald{at}yahoo.com
; present address: Department of Biology, Indiana University of Pennsylvania, Indiana, PA 15705 USA
LITERATURE CITED
Archibald J. K. Mort M. E. Crawford D. J. Santos-Guerra A. In press Evolutionary relationships within recently radiated taxa: comments on methodology and analysis of inter-simple sequence repeat data and other hypervariable, dominant markers. Taxon.
Archibald J. K. Mort M. E. Wolfe A. D.. 2005. Phylogenetic relationships within Zaluzianskya (Scrophulariaceae s.s., tribe Manuleeae): classification based on DNA sequences from multiple genomes and implications for character evolution and biogeography. Systematic Botany 30: 196-215.[CrossRef][ISI]
Baldwin B. G.. 2003. A phylogenetic perspective on the origin and evolution of Madiinae. In S. Carlquist, B. G. Baldwin and G. D. Carr [eds.] Tarweeds and silverswords: evolution of the Madiinae (Asteraceae) 193-228 Missouri Botanical Garden Press, St. Louis, Missouri, USA.
Baldwin B. G. Crawford D. J. Francisco-Ortega J. Kim S. Sang T. Stuessy T.. 1998. Molecular phylogenetic insights into the origin and evolution of island plants. In D. E. Soltis, P. S. Soltis and J. J. Doyle [eds.] Molecular systematics of plants II 410-441 Kluwer, Boston, Massachusetts, USA.
Blair M. W. Panaud O. McCouch S. R.. 1999. Inter-simple sequence repeat (ISSR) amplification for analysis of microsatellite motif frequency and fingerprinting in rice (Oryza sativa L). Theoretical Applied Genetics 98: 780-792.[CrossRef]
Bramwell D. Bramwell Z. I.. 1974. Wildflowers of the Canary Islands Cabildo Insular de Tenerife, Santa Cruz de Tenerife, Spain.
Bramwell D. Bramwell Z. I.. 1990. Flores silvestres de las Islas Canarias Editorial Rueda, Madrid, Spain.
Bussell J. D. Waycott M. Chappill J. A.. 2005. Arbitrarily amplified DNA markers as characters for phylogenetic inference. Perspectives in Plant Ecology, Evolution and Systematics 7: 3-26.[CrossRef][ISI]
Carlquist S.. 1974. Island biology Columbia University Press, New York, New York, USA.
Coyne J. A. Orr H. A.. 2004. Speciation Sinauer, Sunderland, Massachusetts, USA.
Crawford D. J. Archibald J. K. Santos-Guerra A. Mort M. E.. 2006. Allozyme diversity within and divergence among species of Tolpis (Asteraceae-Lactuceae) in the Canary Islands: systematic, evolutionary, and biogeographical implications. American Journal of Botany 93: 656-664.
Crawford D. J. Mort M. E.. 2004. Single-locus molecular markers for inferring relationships at lower taxonomic levels: observations and comments. Taxon 53: 631-635.[CrossRef][ISI]
Dice L. R.. 1945. Measures of the amount of ecological association between species. Ecology 26: 297-302.[CrossRef][ISI]
Doyle J. J. Doyle J. L.. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15.
Felsenstein J.. 2004. PHYLIP (phylogeny inference package), version 3.6 Computer program distributed by the author, website http://evolution.genetics.washington.edu/phylip.html. Department of Genome Sciences, University of Washington, Seattle, Washington, USA.
Gillett G. W. Lim E. K. S.. 1970. An experimental study of the genus Bidens in the Hawaiian Islands. University of California Publications in Botany 56: 1-63.
Howarth D. G. Baum D. A.. 2005. Geneological evidence of homoploid speciation in an adaptive radiation of Scaevola (Goodeniaceae) in the Hawaiian Islands. Evolution 59: 948-961.[CrossRef][ISI][Medline]
Huang J. Sun M.. 2000. Fluorescein PAGE analysis of microsatellite-primed PCR: a fast and efficient approach for genomic fingerprinting. BioTechniques 28: 1068-1072.[ISI][Medline]
Jarvis C. E.. 1980. Systematic studies in the genus Tolpis Adanson. Ph.D. dissertation University of Reading, Reading, UK.
Koopman W. J. M.. 2005. Phylogenetic signal in AFLP data sets. Systematic Biology 54: 197-217.[CrossRef][ISI][Medline]
Liu B. Wendel J. F.. 2001. Intersimple sequence repeat (ISSR) polymorphisms as a genetic marker system in cotton. Molecular Ecology Notes 1: 205-208.[CrossRef][ISI]
Lowrey T. K.. 1986. A biosystematic revision of Hawaiian Tetramolopium (Compositae: Asterae). Allertonia 4: 203-265.
Maddison D. R.. 1991. The discovery and importance of multiple islands of most-parsimonious trees. Systematic Zoology 40: 315-328.[CrossRef]
Moore M. J. Francisco-Ortega J. Santos-Guerra A. Jansen R. K.. 2002. Chloroplast DNA evidence for the roles of island colonization and extinction in Tolpis (Asteraceae: Lactuceae). American Journal of Botany 89: 518-526.
Mort M. E. Crawford D. J.. 2004. The continuing search: low-copy nuclear sequences for low-level plant molecular phylogenetic studies. Taxon 53: 257-261.[CrossRef][ISI]
Mort M. E. Crawford D. J. Santos-Guerra A. Francisco-Ortega J. Esselman E. J. Wolfe A. D.. 2003. Relationships among the Macaronesian members of Tolpis (Asteraceae: Lactuceae) based upon analyses of inter simple sequence repeat (ISSR) markers. Taxon 52: 511-518.[CrossRef][ISI]
Mort M. E. Soltis D. E. Soltis P. S. Francisco-Ortega J. Santos-Guerra A.. 2001. Phylogenetic relationships and evolution of Crassulaceae inferrred from matK sequence data. American Journal of Botany 88: 76-91.
Mort M. E. Soltis D. E. Soltis P. S. Francisco-Ortega J. Santos-Guerra A.. 2002. Phylogenetics and evolution of the Macaronesian clade of Crassulaceae inferred from nuclear and chloroplast sequence data. Systematic Botany 27: 271-288.[ISI]
Nagaraju J. Kathirvel M. Kumar R. R. Siddiq E. A. Hasnain S. E.. 2002. Genetic analysis of traditional and evolved Basmati and non-Basmati rice varieties by using fluorescence-based ISSR-PCR and SSR markers. Molecular and Cellular Probes 99: 5836-5841.
Nei M. Li W.-H.. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proceedings of the National Academy of Sciences, USA 76: 5269-5273.
Park S. J. Korompai E. J. Francisco-Ortega J. Santos-Guerra A. Jansen R. K.. 2001. Phylogenetic relationships of Tolpis (Asteraceae: Lactuceae) based on ndhF sequence data. Plant Systematic Evolution 226: 23-33.[CrossRef]
Patzak J.. 2001. Comparison of RAPD, STS, ISSR and AFLP molecular methods used for assessment of genetic diversity in hop (Humulus lupulus L). Euphytica 121: 9-18.
Pelser P. B. Gravendell B. van der Meijden R.. 2003. Phylogeny reconstruction in the gap between too little and too much divergence: the closest relatives of Senecio jacobaea (Asteraceae) according to DNA sequences and AFLPs. Molecular Phylogenetics and Evolution 29: 613-628.[CrossRef][ISI][Medline]
Rohlf F. J.. 1998. NTSYS-pc. Numerical taxonomy and multivariate analysis system, version 2.0 Exeter Software, Setauket, New York, USA.
Saitou N. Nei M.. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4: 406-425.[Abstract]
Schrader J. A. Graves W. R.. 2004. Systematics of Alnus maritima (seaside alder) resolved by ISSR polymorphisms and morphological characters. Journal of the American Society of Horticultural Science 129: 231-236.
Shaw J. Lickey E. B. Beck J. T. Farmer S. B. Liu W. Miller J. Siripun K. C. Winder C. T. Schilling E. E. Small R. L.. 2005. The tortoise and the hare II: relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. American Journal of Botany 92: 142-166.
Small R. L. Cronn R. C. Wendel J. F.. 2004. L. A. S. Johnson Review No. 2. Use of nuclear genes for phylogeny reconstruction in plants. Australian Systematic Botany 17: 145-170.[CrossRef][ISI]
Swofford D. L.. 2003. PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4 Sinauer, Sunderland, Massachusetts, USA.
Templeton A. R.. 1981. Mechanisms of speciation—a population genetic approach. Annual Review of Ecology and Systematics 12: 23-48.
Whitkus R. Doan H. Lowrey T. K.. 2000. Genetics of adaptive radiation in Hawaiian species of Tetramolopium (Asteraceae). III. Evolutionary genetics of sex expression. Heredity 85: 37-42.
Xu F. Sun M.. 2001. Comparative analysis of phylogenetic relationships of grain amaranths and their wild relatives (Amaranthus; Amaranthaceae) using internal transcribed spacer, amplified fragment length polymorphism, and double-primer fluorescent intersimple sequence repeat markers. Molecular Phylogenetics and Evolution 21: 372-387.[CrossRef][ISI][Medline]
This article has been cited by other articles:
|
|
 |
|
  M. E. Mort, J. K. Archibald, C. P. Randle, N. D. Levsen, T. R. O'Leary, K. Topalov, C. M. Wiegand, and D. J. Crawford Inferring phylogeny at low taxonomic levels: utility of rapidly evolving cpDNA and nuclear ITS loci Am. J. Botany, February 1, 2007; 94(2): 173 - 183. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
