| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2 School of Plant Science 3 Cooperative Research Centre for Forestry, University of Tasmania, Private Bag 55, Hobart, Tasmania 7001, Australia
Received for publication 6 June 2007. Accepted for publication 25 January 2008.
ABSTRACT
Genus Eucalyptus, with over 700 species, presents a number of systematic difficulties including taxa that hybridize or intergrade across environmental gradients. To date, no DNA marker has been found capable of resolving phylogeny below the sectional level in the major subgenera. Molecular markers are needed to support taxonomic revision, assess the extent of genetic divergence at lower taxonomic levels, and inform conservation efforts. We examined the utility of 930 amplified fragment length polymorphisms (AFLPs) for analyzing relationships among Tasmanian taxa of subgenus Symphyomyrtus section Maidenaria. Phenetic and cladistic analyses resolved species into clusters demonstrating significant genetic partitioning, largely concordant with series defined in the most recent taxonomic revision of Eucalyptus. Some departures from current taxonomy were noted, indicating possible cases of morphological convergence and character reversion. Although the resolution obtained using AFLP was greatly superior to that of single sequence markers, the data demonstrated high homoplasy and incomplete resolution of closely related species. The results of this study and others are consistent with recent speciation and reticulate evolution in Maidenaria. We conclude that a combination of phylogenetic and population genetic approaches using multiple molecular markers offers the best prospects for understanding taxonomic relationships below the sectional level in Eucalyptus.
Key Words: amplified fragment length polymorphism (AFLP) • eucalypt genetics • Eucalyptus • Maidenaria • Myrtaceae • phylogeny • Tasmania
Eucalyptus L'Hérit (Myrtaceae) is a large genus of over 700 species (Brooker, 2000
) with a widespread distribution across Australia, Papua New Guinea, Timor, Sulawesi, and the Philippines. Its extraordinary adaptability has allowed it to occupy Australian habitats ranging from desert and subalpine terrain to swamps and coastal heaths and to form a dominant component of forest and woodland environments. In the past decade, molecular analyses have improved our knowledge of relationships within and among the major subgenera of Eucalyptus and its near relatives. Advances in phylogenetic analysis using DNA sequence data (Udovicic et al., 1995; Steane et al., 1999, 2002; Udovicic and Ladiges, 2000; Whittock et al., 2003; Gibbs, 2007; Parra-O. et al., 2006) have resolved longstanding questions regarding relationships among Eucalyptus and its close relatives, Corymbia and Angophora, and explored the divergence of subgenera Eucalyptus, Symphyomyrtus, and Eudesmia. These analyses have contributed to our understanding of the evolution of diversity in the present-day Australian flora in response to past periods of climatic instability (Crisp et al., 2004). Complementary analyses at lower systematic levels are needed to explore present evolutionary processes and inform taxonomic revision in Eucalyptus. Knowledge of the genetic relationships within and among lower-level taxonomic groups will support conservation efforts that aim to preserve phylogenetic diversity and evolutionary potential (Faith, 1992) in this dominant tree genus.
Despite all efforts, the majority of phylogenetic relationships between species within sections of Eucalyptus have proven impossible to resolve using standard DNA sequence markers. Steane et al. (1999, 2002) showed that variation in the internal transcribed spacer (ITS) region of nuclear ribosomal DNA was insufficient to differentiate some species currently classified into different sections in subgenus Eucalyptus or to resolve relationships within sections of subgenus Symphyomyrtus. In analyses of the more highly variable single-copy nuclear gene for cinnamoyl coA reductase (CCR; McKinnon et al., 2005; Poke et al., 2006), recombination, incomplete lineage sorting, and/or reticulation confounded the use of this gene for exploring relationships within and among sections of Symphyomyrtus. Chloroplast DNA phylogeny in subgenus Eucalyptus conflicts with past and current taxonomic treatments (McKinnon et al., 1999). Likewise, a series of detailed analyses of cpDNA variation in subgenus Symphyomyrtus section Maidenaria (Steane et al., 1998; Jackson et al., 1999; McKinnon et al., 2001, McKinnon et al., 2004) indicate chloroplast phylogeny to be incongruent with taxonomy and consistent with a history of recurrent hybridization among species. As a result, our understanding of the phylogenetic relationships among taxa within sections of subgenera Eucalyptus and Symphyomyrtus continues to be derived primarily from morphology. A recent analysis (Nicolle et al., 2006) and taxonomic revision (Nicolle and Whalen, 2006) of species from subgenus Symphyomyrtus section Bisectae used morphological characters exclusively.
Amplified fragment length polymorphism (AFLP) is a whole-genome approach to studying genetic variation that is gaining in popularity for lower-level systematics (Bussell et al., 2005). The strengths of AFLP include high levels of polymorphism and the ability to sample randomly across the genome, thus generating phylogenies based on multiple rather than single genomic regions. Although most analyses report high levels of homoplasy, AFLP has successfully resolved genetic relationships among groups having almost invariable ITS sequences in plant genera such as Macaranga (Bänfer et al., 2004), Trollius (Després et al., 2003), and Soldanella (Zhang et al., 2001). The technique appears to be least successful where species are closely related and cross-fertile, presumably because interspecific differences are blurred by both introgression and retention of ancestral polymorphisms (Koopman et al., 2001; Després et al., 2003; Pellmyr et al., 2007). Under these circumstances, variation of AFLP markers between morphologically distinct species can sometimes be exceeded by variation among populations within species (e.g., Quercus petraea and Q. robur; Kelleher et al., 2005), and conspecific individuals may be scattered throughout trees generated by phenetic or cladistic analysis (e.g., Polylepis, Schmidt-Lebuhn et al., 2006; Yucca, Pellmyr et al., 2007).
This study used AFLP to examine genetic relationships among the 17 Tasmanian species of subgenus Symphyomyrtus section Maidenaria, previously shown to have very similar or identical ITS (Steane et al., 1999, 2002) and chloroplast (McKinnon et al., 2001) DNA sequences. Maidenaria is a large section of 73 species, classified by Brooker (2000) into three subsections with 17 constituent series. Its taxa are characterized by bilobed cotyledons, axillary inflorescences, versatile anthers, seeds with a ventral hilum, and adult leaves that are moderately reticulate, with areolar and intersectional glands. The two major subsections are Triangulares with 22 species (having juvenile leaves that are petiolate and alternate, and fruits with a triangular arrangement of valves, with the disc or valves scarcely prominent) and Euryotae with 49 species (having juvenile leaves that are sessile and opposite for many pairs, a long style, leaves with oil glands, and fruit with the disc and valves usually prominent). A previous phylogenetic analysis of the entire section based on adult and seedling morphology by Chappill and Ladiges (1996) proved difficult because of high homoplasy, intergrading taxa and nondiscrete characters. The resulting cladogram, which conflicts with Brooker's (2000) taxonomy in many respects, was considered by Chappill and Ladiges (1996) to be only an approximate representation of the true phylogeny due to problems of character coding.
We aimed to determine whether AFLP could contribute to taxonomic revision in this section and, by implication, in other difficult sections of Eucalyptus and other genera. We analyzed a subset of Maidenaria comprising multiple replicates of 10 endemic Tasmanian taxa, seven taxa common to both Tasmania and mainland Australia, and four mainland taxa, representing five series from the two major subsections (series Globulares, Viminales, Orbiculares, and Semiunicolores from subsection Euryotae and series Foveolatae from subsection Triangulares). This group of taxa included species reported to form clines (e.g., species of Semiunicolores; McGowen et al., 2001), zones of morphological intergradation (e.g., species of series Globulares; Jordan et al., 1993), and numerous natural intra- and interseries hybrids (Duncan, 1989). The use of multiple replicates per species enabled us to place variation among species and series in the context of variation within both rare and widespread species for Tasmanian Maidenaria.
MATERIALS AND METHODS
Sample collection
Sampling of section Maidenaria covered all 17 taxa native to Tasmania, three additional taxa of the E. globulus complex from mainland Australia (E. bicostata, E. maidenii, and E. pseudoglobulus), and the pulpwood species E. nitens which is now planted widely in Tasmania (Table 1). Where possible, we sampled three widely separated populations of each species from Tasmania and/or mainland Australia, giving six samples for species common to both regions. Samples of E. nitens were sourced from seed orchards representing core collection areas. Samples of E. archeri and E. barberi were from a eucalypt field trial planted at Boyer, southern Tasmania. All other species were field-collected for this study or sourced from the collections of McKinnon et al. (2001, 2004), McGowen et al. (2001), Jones et al. (2005), and Rathbone et al. (2007) as indicated in Table 1. Leaf tissue from the three outgroup taxa E. grandis and E. balladoniensis (subgenus Symphyomyrtus, sections Latoangulatae and Bisectae, respectively) and E. cloeziana (subgenus Idiogenes) was from the collection of Steane et al. (2002). For each individual, leaf tissue was frozen in liquid nitrogen and stored at –70°C until DNA preparation, and a herbarium specimen was deposited in the herbarium of the School of Plant Science, University of Tasmania.
|
using simultaneous restriction and ligation of DNA samples. All eucalypt DNA samples were pretested to ensure full enzymatic digestion under the conditions used for restriction–ligation reactions. Reproducibility of AFLP profiles was confirmed experimentally for replicate DNA purifications and restriction–ligation reactions containing 10, 20, 50, 100 or 200 ng of genomic DNA. Final restriction–ligation reactions (30 µL) for all samples contained 50–200 ng genomic DNA, 1x T4 DNA-ligase buffer (Promega, Madison, Wisconsin, USA), 50 mM NaCl, 50 ng/µL bovine serum albumin, 5 U MseI, 16 U Eco RI, 2 Weiss U T4 DNA ligase, 3 pmol Eco RI adapter and 30 pmol Mse I adapter. Restriction–ligation reactions were incubated for 16 h at 37°C. Preselective PCRs (30 µL) contained 1x Taq DNA polymerase buffer [67 mM Tris-HCl pH 8.8, 16.6 mM (NH4)2SO4, 0.45% Triton X-100, 0.2 mg/mL gelatin], 1.2 U Taq DNA polymerase, 1.5 mM MgCl2, 0.2 mM of each dNTP, 45 ng EcoR I-A primer, 45 ng Mse I-C primer, and 2 µL of restriction–ligation reaction as template. Thermocycler conditions were: 94°C for 30 s, followed by 28 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min; followed by 72°C for 5 min. Selective amplification reactions (20 µL) contained 1x Taq DNA polymerase buffer as described earlier, 0.8 U Taq DNA polymerase, 1.5 mM MgCl2, 0.2 mM of each dNTP, 0.25 µM of each selective primer, and 5 µL of diluted (1 in 10) preselective PCR as template. Thermocycler conditions were: 94°C for 1 min, followed by 11 cycles of 94°C for 30 s, 65°C decreasing by 0.7°C per cycle for 30 s, and 72°C for 1 min; then 23 cycles of 94°C for 30 s, 56°C for 30 s and 72°C for 1 min; followed by 72°C for 5 min. EcoRI selective primers were labeled using fluorescent WellRED dyes (Sigma Proligo, Lismore, New South Wales, Australia). Separation and binning of AFLP fragments was performed using the CEQ8000 Genetic Analysis System 8.0.52 (Beckman Coulter, Gladesville, New South Wales, Australia). Fragments (70–591 bp) were scored as present/absent. Fragments between these size limits that were poorly separated or inconsistently detected by the binning software (50% of fragments) were discarded from the analysis. (For the complete data matrix, see Appendix S1, Supplemental Data accompanying online version of this article.)
Statistical and distance analyses
Partitioning of taxa into genetic groups was investigated through principal coordinates analysis (PCo; Maidenaria taxa only) using the genetic distance measure of Nei and Li (1979), calculated by PAUP* version 4.0b10 (Swofford, 2002). This distance measure is based on the shared presence of fragments, minimizing error caused by shared absences that are not homologous. PCo was performed using the DCENTRE, EIGEN, and plotting modules from the software NTSYS-PC 2.1 (Rohlf, 2000). A neighbor-joining (NJ) phenogram (Saitou and Nei, 1987) of Maidenaria taxa with three outgroup taxa, based on the same genetic distance measure, was computed using PAUP* 4.0b10 with statistical support obtained from 10000 bootstrap replicates. The distribution of genetic variation within Maidenaria was determined by analysis of molecular variance (AMOVA; Excoffier et al., 1992), using the software Arlequin version 2.000 (Schneider et al., 2000), with significance of group partitioning tested using 1000 permutations of individuals among groups. The same software was used to calculate pairwise genetic distances among groups (F-statistics; Wright, 1951) and nucleotide diversity or average gene diversity over loci 
n, which is equivalent to the probability that two randomly chosen homologous nucleotides are different (Tajima, 1983; Nei, 1987).
Cladistic analysis
Cladistic analysis based on maximum parsimony was performed by PAUP* 4.0b10 using the following options: character-state optimization by accelerated transformation, tree-bisection-reconnection (TBR) branch-swapping, MulTrees on, steepest descent, and branches collapsed if maximal branch length was zero. The search strategy was designed to ensure that no shorter trees existed (Catalán et al., 1997). First, two heuristic searches were conducted on the unweighted data matrix using (a) closest addition sequence and (b) 1000 random addition sequence replicates, saving a maximum of five trees per replicate. The strict consensus of all shortest trees combined from searches (a) and (b) was used as a constraint for a further search of 5000 replicates of random addition sequence, saving no more than five trees per replicate and setting PAUP* to save only trees that did not match these constraints. The data set was bootstrapped using 10000 replicates (Felsenstein, 1985) of the fast stepwise algorithm of PAUP*.
RESULTS
Using eight primer pairs, we scored 930 AFLP fragments, 920 of which were polymorphic across the full set of taxa. Within Maidenaria, 850 characters were polymorphic. Table 2 lists the total number of polymorphic fragments scored from different primer pairs, which varied from 59 to 177 across all taxa. All individuals had unique AFLP profiles.
|
, are shown in Fig. 1. The first three vectors accounted cumulatively for 26.2% of the total variance detected, comprising 12.4%, 8.1% and 5.7% from the first, second, and third vectors respectively. Ordination of the first two vectors (Fig. 1A) showed three distinct genetic clusters, corresponding to (1) the E. globulus complex (E. globulus, E. pseudoglobulus, E. maidenii, and E. bicostata); (2) members of series Viminales (E. viminalis, E. rubida, and E. dalrympleana) together with E. perriniana and E. nitens; and (3) all remaining taxa of series Foveolatae, Semiunicolores, and Orbiculares. Within the third cluster, there was separation of Foveolatae from the other two series, with overlap of E. archeri 1 (Orbiculares) and E. barberi 1 and 2 (Foveolatae). Whereas separation of clusters 1–3 was achieved with ease using fewer markers (primer pairs 1–5; 559 markers; not shown), separation of Foveolatae from Semiunicolores and Orbiculares required the full set of 930 markers. Addition of the third PCo vector resolved E. nitens from the cluster containing Viminales and E. perriniana (Fig. 1B).
|
. Collectively, all samples of Maidenaria formed a cluster with 99% bootstrap support. With the exception of two species (E. perriniana and E. nitens), the NJ tree placed all Maidenaria species into four main clusters congruent with their classification into Brooker's (2000) series Orbiculares, Semiunicolores, Foveolatae, Viminales, and Globulares. Cluster 1 (Fig. 2) contained all members of series Foveolatae, except one individual of E. barberi. Within Foveolatae, E. barberi was sister to a subcluster containing E. ovata, E. brookeriana, and E. rodwayi. The morphologically very similar taxa E. ovata and E. brookeriana could not be fully separated from one another, nor from E. rodwayi, which reportedly intergrades with either E. ovata or E. brookeriana near the collection site of E. rodwayi 1 (Williams and Potts, 1996).
|
Cluster 3 contained all taxa of the E. globulus complex (series Globulares), with E. maidenii resolved as sister to the three remaining taxa. The intergrading taxa (E. globulus, E. pseudoglobulus, and E. bicostata) remained incompletely resolved. Cluster 4 contained all members of series Viminales (E. viminalis, E. rubida, and E. dalrympleana) except for one anomalous outlier of E. dalrympleana, which clearly belonged within Viminales based on the subsequent cladistic analysis (described later). In addition, cluster 4 contained all six samples of E. perriniana (series Orbiculares), which formed an exclusive lineage with 68% bootstrap support. Eucalyptus perriniana was further resolved into two well-supported clusters corresponding to Tasmanian and mainland Australian samples. With the exception of cluster 3 (71% bootstrap support), the four major clusters received <50% bootstrap support, reflecting the shortness of supporting branches relative to terminal branches. Samples of E. nitens (series Globulares) formed a fifth cluster with 100% bootstrap support, distinct from all others.
Analysis of molecular variance (AMOVA; Excoffier et al., 1992) across the four major species clusters determined by PCo and NJ analysis (clusters 1–4 from Fig. 2; excluding E. barberi which was divided among clusters) indicated significant partitioning of genetic variation with 83% of the detected variation residing within species, 7% between species within clusters, and 10% between clusters (P < 0.001; Table 3). Pairwise Fst values (Table 4) further indicated that all clusters were significantly differentiated (P < 0.001), with the highest level of genetic differentiation being between Foveolatae (cluster 1) and the E. globulus complex (cluster 3; Fst = 0.189). The Tasmanian endemics (cluster 2) were genetically closest to Foveolatae, regardless of whether E. barberi was included or excluded (Fst = 0.081, Table 4). Clusters 2, 3, and 4 were approximately equidistant from one another (Fst about 0.11). Genetic diversity, as measured by the average nucleotide diversity (n, equivalent to the probability that two randomly chosen homologous nucleotides are different; Tajima, 1983; Nei, 1987), was not significantly different among clusters (Table 4).
|
|
) of 0.176 and an ensemble retention index (RI; Farris, 1989) of 0.285. The strict consensus of these trees is shown in Fig. 3. Despite the high level of homoplasy (82%), the strict consensus tree resolved species into clades consistent with the NJ and PCo analyses. Collectively, species from section Maidenaria formed a monophyletic clade with 85% bootstrap support. Within Maidenaria, E. nitens was the sister to a clade containing all other species. These were divided into two major subclades. The first subclade contained species of Foveolatae, together with endemic Tasmanian taxa of Orbiculares and Semiunicolores. Within the second subclade, E. perriniana and species of Viminales collectively formed a sister clade to species of the E. globulus complex. As with the NJ tree, supporting branches for individual trees were short relative to terminal branches (not shown), and bootstrap support was low for the major nodes within the section. Further within-clade structure was limited to the grouping of some individuals with their conspecifics. Grouping of conspecifics was violated in a few instances where clades comprised species known to intergrade across environmental gradients (E. vernicosa-E. subcrenulata; species of the E. globulus complex; and Tasmanian E. dalrympleana-E. rubida-E. viminalis) or to exhibit very low differentiation on population genetic analysis based on nuclear microsatellite markers (E. archeri-E. urnigera; C. Hudson, University of Tasmania, unpublished data).
|
Level of resolution afforded by AFLP
This study is the first to find molecular markers capable of resolving series and species within this complex section of Eucalyptus. The use of AFLP for phylogenetic analysis is somewhat controversial, and previously noted drawbacks of the technique include (1) use of total DNA, which may include contaminant DNA from other sources, (2) lack of knowledge regarding the identity of AFLP fragments, (3) homoplasy caused by nonhomologous comigrating fragments or independent losses of fragments, and (4) potential lack of independence between amplified fragments (Bussell et al., 2005; Koopman, 2005). Homoplasy due to nonhomology of fragments increases with taxonomic divergence (O'Hanlon and Peakall, 2000; Althoff et al., 2007), seriously limiting the use of the technique for phylogenetic analysis of distantly related taxa. While AFLP appears to be most appropriate for resolving relationships among closely related taxa, at lower systematic levels there is the possibility that population genetic effects may swamp any phylogenetic signal present in arbitrarily amplified DNA markers (Bussell et al., 2005). The majority of genetic variation uncovered by AFLP in studies of closely related outcrossing taxa resides within, rather than among, species (e.g., > 90% for four species of Quercus; Dodd and Kashani, 2003). Rather than having species-specific AFLP fragments, recently diverged taxa may have only small frequency differences in fragments, similar to or even less marked than the differences among populations within taxa. For example, Kelleher et al. (2005) found that species differentiation between Quercus petraea and Q. robur, based on AFLP markers, accounted for only 13% of the observed genetic variation and was obscured by population differentiation, which accounted for 27%.
Despite these difficulties, plant species phylogenies generated using AFLP and ITS are generally in good agreement, and there is good bootstrap support for clades recovered using AFLP, where corresponding variation within ITS is 
10–30 nucleotides (Koopman, 2005). Furthermore, AFLP has proven an effective technique for resolving relationships among species where ITS sequences are almost invariable (Zhang et al., 2001; Després et al., 2003; Bänfer et al., 2004), although homoplasy is typically high, with CI values of <0.2 reported for studies of Macaranga (Bänfer et al., 2004), Polylepis (Schmidt-Lebuhn et al., 2006), and Minthostachys (Schmidt-Lebuhn, 2007). The results of our study are concordant with these findings. Resolution of Maidenaria from the outgroup taxon E. grandis, which is supported by 12 or more nucleotide changes in ITS sequence data (Steane et al., 1999, 2002) received good bootstrap support in the AFLP phenetic and cladistic analyses. Within Maidenaria, among species that previously demonstrated ITS divergence of 0–4 nucleotides, and in some cases demonstrated higher intraspecific than interspecific ITS variation (Steane et al., 1999, 2002), the AFLP analysis provided resolution of significantly differentiated genetic groups corresponding, with few exceptions, to morphological series. The separation of series using the whole-genome approach also presents a strong contrast to previous studies using cpDNA, which showed greater correlation with geographical location than with species or series boundaries (Steane et al., 1998; Jackson et al., 1999; McKinnon et al., 2001, 2004). We conclude that any contribution of cytoplasmic DNA markers or introgressed nuclear markers to AFLP profiles was generally insufficient to mask nuclear genomic divergence among morphological series. Although the clades resolved by AFLP showed significant genetic partitioning, homoplasy was high (CI < 0.2) and bootstrap support in the NJ and cladistic analyses was low, reflecting the shortness of supporting branches relative to terminal branches. At the intrasectional level, 83% of the genetic variation found resided within species, only 10% separated the major species groups, and only 7% resided among species within groups. These figures illustrate the low percentage of total genetic variation that contributes to recognizable morphological divergence among series and species in section Maidenaria.
Within series, many species could not be fully resolved using AFLP, as has been noted before for complexes containing closely related or hybridizing taxa (e.g., Quercus, Dodd and Kashani, 2003; Macaranga, Bänfer et al., 2004; Populus, Cervera et al., 2005; Polylepis, Schmidt-Lebuhn et al., 2006; Minthostachys, Schmidt-Lebuhn, 2007; Yucca, Pellmyr et al., 2007). In most cases, lack of resolution in this study occurred among taxa that are reported to intergrade morphologically. For these taxa, both phenotypic and AFLP evidence suggest either (1) very recent divergence, such that lineage sorting of ancestral polymorphisms remains incomplete or (2) incomplete speciation with continuing gene flow. There was also a lack of resolution among some endemic Tasmanian taxa of series Orbiculares that appear morphologically and ecologically distinct. Single individuals of E. gunnii and E. morrisbyi clustered with one another rather than with their conspecifics in the NJ analysis, and individuals of E. urnigera were scattered within the series. There was some tendency of these species to cluster according to geographic location rather than to species affiliation. While this finding is consistent with either reticulate evolution or nonmonophyletic origins for endemic Tasmanian taxa of Orbiculares, it is equally likely that the signal-to-noise ratio of the AFLP data is simply too low to discriminate among species at this level. We recently obtained population genetic data from eight microsatellite loci showing that parapatric populations of E. archeri, E. gunnii, and E. urnigera from central Tasmania have extremely low levels of genetic differentiation from one another (mean pairwise Fst = 0.02) and from allopatric populations of E. morrisbyi (mean pairwise Fst = 0.08; C. Hudson, University of Tasmania, unpublished data). This level of differentiation is exceeded by that found between the two largest extant populations of E. morrisbyi (Fst > 0.1; C. Hudson, unpublished data; Jones et al., 2005). These results are analogous to those of Craft and Ashley (2006), who analyzed genetic variation within and among Quercus alba, Q. macrocarpa, and Q. bicolor using five microsatellite loci and found higher differentiation between two populations of Q. alba than between any pair of species. Detailed population genetic approaches that combine morphological and molecular data would thus appear to offer the best hope of understanding species relationships and limits at the within-series level in Maidenaria.
Concordance between AFLP and current taxonomy
The most recent taxonomy of Eucalyptus (Brooker, 2000) is based conceptually on the classification of Pryor and Johnson (1971) and draws on extensive observation of heritable phenotypic characters in seedling and adult plants. Brooker's (2000) two major subsections of section Maidenaria, namely Triangulares and Euryotae, are allied closely to Pryor and Johnson's (1971) two informal series, Ovatae and Viminales, respectively. A full phylogenetic analysis of morphological characters in Maidenaria by Chappill and Ladiges (1996) found that Pryor and Johnson's Ovatae was a paraphyletic group recognizable only by the sharing of plesiomorphic (primitive) features, whereas Viminales was a monophyletic group defined by juvenile leaves being opposite, sessile, and cordate for many nodes. Both seedling and adult characters demonstrated high homoplasy (overall homoplasy index = 82%). Other problems noted were intergrading taxa and difficulties in coding continuously varying characters. The present AFLP analysis also encountered high homoplasy (82%), and there was low bootstrap support for clades recovered by the NJ and parsimony analyses, meaning that the clustering order of taxa could not be determined with a high level of certainty. However, AFLP could be used to define groups of species that had significant genetic differentiation from one another and to assess the relative levels of genetic divergence among groups. The results, while mostly congruent with morphology, highlighted possible cases of both convergence and reversion of phenotypic characters within Maidenaria.
Phenotypic convergence appears to have occurred between E. perriniana (the type species of series Orbiculares) and the five endemic Tasmanian species classified into Orbiculares (E. archeri, E. gunnii, E. urnigera, E. cordata, and E. morrisbyi). Species of this series have conspicuous opposite, orbicular, often glaucous juvenile leaves (Brooker, 2000) that persist in the mature crown in both E. perriniana and E. cordata (Nicolle, 2006). The AFLP analysis placed all samples of E. perriniana in a single clade that was genetically distinct from endemic Tasmanian Orbiculares and had a closer affinity to three nonendemic species of series Viminales (E. viminalis, E. dalrympleana, and E. rubida). Although this grouping received low bootstrap support, it is consistent with the geographical distribution of E. perriniana, which has disjunct populations in southeastern mainland Australia and central-eastern Tasmania (Rathbone et al., 2007). The AFLP evidence suggests that E. perriniana has evolved from a lineage other than that of endemic Tasmanian members of Orbiculares and subsequently dispersed into Tasmania. Persistence of juvenile foliage in the adult tree is one of several convergent characters that occur independently in quite unrelated species of Eucalyptus (Pryor and Johnson, 1971; Potts and Wiltshire, 1997), and it is reasonable to conclude this character has arisen more than once in Maidenaria. An affinity to Viminales is plausible on the basis of the bud and adult leaf morphology of E. perriniana. Additional research is needed to confirm the affiliations of the remaining mainland Australian members of Orbiculares: E. glaucescens, E. saxatilis, E. pulverulenta, and E. chapmaniana. Based on their distributions, which are mostly associated with that of mainland E. perriniana, it is probable that they belong to the same evolutionary branch as E. perriniana.
The relationships between E. perriniana, E. rubida, E. dalrympleana, E. viminalis, and the endemic Tasmanian Orbiculares have been subject to differing interpretations since Pryor and Johnson (1971) placed all except E. viminalis in their subseries Cordatinae. The analysis of Chappill and Ladiges (1996) also separated E. viminalis from the other species on the basis of juvenile leaf width. However, E. dalrympleana intergrades clinally with both E. viminalis and E. rubida in Tasmania, and the distinction between the three species becomes obscure at altitudes between 200 m and 600 m a.s.l. (Williams and Potts, 1996). Phillips and Reid (1980) considered that clinal variation in juvenile leaf and capsule morphology between Tasmanian E. dalrympleana and E. viminalis resulted from selection acting on a continuous group of potentially interbreeding populations. The AFLP analysis supports Brooker's (2000) classification in which these three species are placed in the same series and is consistent with either low genetic divergence or incomplete reproductive isolation among them.
Within the E. globulus complex, regarded by different authors as a complex of either four closely related species (Brooker, 2000) or subspecies (Kirkpatrick, 1974), separation of taxa in the phenetic analysis was consistent with morphology and records of intergradation (Jordan et al., 1993), which indicate that E. maidenii, characterized by seven buds per inflorescence, is the most distinct taxon while E. pseudoglobulus intergrades with the two remaining species. In the cladistic analysis, individuals of E. bicostata and E. maidenii formed clades with their conspecifics, whereas E. globulus and E. pseudoglobulus had mixed affinities. A poor correlation between phenotype and nuclear microsatellite markers has been previously described for species and intergrades belonging to this complex (Jones et al., 2002; Steane et al., 2006). A comparative analysis of microsatellite markers and quantitative traits within E. globulus found cases of marked phenotypic divergence of parapatric races demonstrating low levels of microsatellite differentiation, as well as phenotypic convergence of races that were well-differentiated using microsatellites (Steane et al., 2006). This finding was considered indicative of selection maintaining quantitative traits that did not necessarily correlate with evolutionary affinities (Steane et al., 2006). Eucalyptus nitens, a species native to mainland Australia but used in Tasmanian pulpwood plantations, is usually considered to be closely related to species of the E. globulus complex and has been classified in the same series (Globulares; Brooker, 2000). Against expectations, the AFLP results indicated E. nitens to be genetically highly differentiated from the E. globulus complex. The PCo separated this species from all others along the third axis, and all of the most parsimonious cladograms placed E. nitens outside all other species of Maidenaria. A fuller analysis of Maidenaria is required to confirm these findings and to clarify the relationship of E. nitens to the remainder of the section.
Another unexpected outcome of the analysis was the grouping, in both the distance- and parsimony-based analyses, of Foveolatae and endemic Tasmanian members of Orbiculares and Semiunicolores. On the basis of morphology, series Orbiculares and Semiunicolores have been classified into subsection Euryotae with Viminales and Globulares, while series Foveolatae is part of subsection Triangulares (Brooker, 2000). This classification is supported mainly by a change in seedling leaf character, from alternate and petiolate in Triangulares to opposite and sessile for many pairs in Euryotae. The alternate, petiolate form appears to be plesiomorphic because species that were sisters to section Maidenaria in the ITS analysis of Steane et al. (2002) have this character (Chippendale, 1988). Chappill and Ladiges (1996) considered the retention of opposite, sessile, and cordate leaves for many nodes to define a monophyletic group (broadly concordant with Euryotae) within Maidenaria. Based on this character, their cladistic analysis placed members of Triangulares as basal to Euryotae. In the AFLP parsimony analysis, the strict consensus of all shortest trees placed E. nitens as basal to all other species. If this topology is correct, the primitive juvenile leaf form of Triangulares would represent a reversion. While analysis of more taxa is needed to resolve this question and better describe the whole section, overall our results are consistent with (1) a close evolutionary relationship between all endemic Tasmanian taxa of Orbiculares and Semiunicolores and (2) lower genetic divergence of these taxa from Foveolatae than from either Viminales or Globulares.
The distinction between Orbiculares and Foveolatae is particularly blurred in the case of E. barberi, a rare, midaltitude mallee species that occurs in small, disjunct populations in eastern Tasmania. This species shows affinities to both of these series based on AFLP analysis. Morphologically, it is close to E. brookeriana (Ladiges et al., 1984), but a phenetic analysis of morphological variation within E. barberi revealed high variability among populations (McEntee et al., 1994). Our study used E. barberi from populations identified as "northern" (the type locality, Cherry Tree Hill) and "southern" (Ravensdale Hill and Ringrove Razorback) morphotypes in the study of McEntee et al. (1994). All three populations had morphological affinities to E. ovata and E. brookeriana, but the population at Ringrove Razorback varied in bud number between seven (typical of Foveolatae) and three (typical of Orbiculares). The sample from Ringrove Razorback nevertheless clustered with Foveolatae in the AFLP analysis, while the sample from Ravensdale Hill had affinities to Orbiculares in both the PCo and NJ analyses. McEntee et al. (1994) also recorded two anomalous populations of E. barberi that appeared intermediate with E. gunnii and were considered either introgressed or representative of high genetic variation within E. barberi. The species has high cpDNA diversity, in common with E. gunnii, but unlike other endemics such as E. cordata, E. morrisbyi, and E. rodwayi (McKinnon et al., 2001, 2004; Jones et al., 2005; G. McKinnon, unpublished data). Collectively, these results indicate that the gene pool of E. barberi is highly diverse and may combine genes from two separate evolutionary branches.
Conclusions
Based on their observations of phenotypic variation, Chappill and Ladiges (1996) noted the possibility that evolution within Maidenaria may not have been predominantly divergent. This concept is supported both by patterns of cpDNA variation that are consistent with localized genetic exchange among species belonging to different series (Steane et al., 1998; Jackson et al., 1999; McKinnon et al., 2001, 2004) and by observations of natural interspecific hybridization. Duncan (1989) recorded all 17 species of Tasmanian Maidenaria as being involved in at least one natural hybrid combination and most species as being involved in several, including combinations across series. Given these observations, we think it unlikely that any bifurcating tree, produced by standard cladistic analysis, can adequately represent the complex evolution of the section. The high level of homoplasy encountered in this analysis is probably due, in part, to genetic exchange among species and series, although incomplete lineage sorting across loci is also likely to be important. Despite these difficulties, the analysis was able to identify significantly differentiated genetic groups within Tasmanian Maidenaria. These groups were mostly concordant with morphological series defined by Brooker (2000). However, the results highlight the need for potential revision of (1) the current division of series among the two major subsections, Triangulares and Euryotae; (2) the classification of species of series Orbiculares; and (3) the classification of E. nitens.
The results of this study indicate that AFLP is a valuable technique suitable for supporting taxonomic revision as low as the series level in Eucalyptus. At the intraseries level, more detailed population genetic approaches will be necessary to delimit species and to understand the relationships between morphology and genetic variation. We conclude that a combination of these approaches will best inform conservation efforts that aim to preserve phylogenetic diversity and evolutionary potential in Eucalyptus. In addition, identification and genomic mapping of the AFLP fragments that differentiate species and series may shed light on the genetic changes that underpin species divergence (Scotti-Saintagne et al., 2004).
FOOTNOTES
1 The authors thank D. Nicolle, D. Rathbone, M. McGowen, R. Jones, C. Grosser, J. Marthick, and H. Jackson for their kind assistance with samples, and R. Wiltshire for expert information on eucalypt morphology. They also thank their anonymous reviewers for their constructive suggestions. This research was supported by the Australian Research Council grant DP0664923. 
4 Author for correspondence (e-mail: Gay.McKinnon{at}utas.edu.au)
LITERATURE CITED
Althoff, D. M., M. A. Gitzendanner, AND K. A. Segraves. 2007. The utility of amplified fragment length polymorphisms in phylogenetics: A comparison of homology within and between genomes. Systematic Biology 56: 477–484.[CrossRef][Medline]
Bänfer, G., B. Fiala, AND K. Weising. 2004. AFLP analysis of phylogenetic relationships among myrmecophytic species of Macaranga (Euphorbiaceae) and their allies. Plant Systematics and Evolution 249: 213–231.[CrossRef][ISI]
Brooker, M. I. H. 2000. A new classification of the genus Eucalyptus L'Her. (Myrtaceae). Australian Systematic Botany 13: 79–148.[CrossRef][ISI]
Bussell, J. D., M. Waycott, AND J. A. Chappill. 2005. Arbitrarily amplified DNA markers as characters for phylogenetic inference. Perspectives in Plant Ecology, Evolution and Systematics 7: 3–26.[CrossRef][ISI]
Catalán, P., E. A. Kellogg, AND R. G. Olmstead. 1997. Phylogeny of Poaceae subfamily Pooideae based on chloroplast ndh F gene sequences. Molecular Phylogenetics and Evolution 8: 150–166.[CrossRef][ISI][Medline]
Cervera, M. T., V. Storme, A. Soto, B. Ivens, M. Van Montagu, O. P. Rajora, AND W. Boerjan. 2005. Intraspecific and interspecific genetic and phylogenetic relationships in the genus Populus based on AFLP markers. Theoretical and Applied Genetics 111: 1440–1456.[CrossRef][ISI][Medline]
Chappill, J. A., AND P. Y. Ladiges. 1996. Phylogenetic analysis of Eucalyptus informal subgenus Symphyomyrtus section Maidenaria. Australian Systematic Botany 9: 71–93.[CrossRef][ISI]
Chippendale, G. M. 1988. Eucalyptus, Angophora (Myrtaceae). Flora of Australia. Australian Government Publishing Service, Canberra, Australia.
Craft, K. J., AND M. V. Ashley. 2006. Population differentiation among three species of white oak in northeastern Illinois. Canadian Journal of Forest Research 36: 206–215.[CrossRef]
Crisp, M., L. Cook, AND D. Steane. 2004. Radiation of the Australian flora: What can comparisons of molecular phylogenies across multiple taxa tell us about the evolution of diversity in present-day communities? Philosophical Transactions of the Royal Society of London, B, Biological Sciences 359: 1551–1571.[CrossRef]
Després, L., L. Gielly, B. Redoutet, AND P. Taberlet. 2003. Using AFLP to resolve phylogenetic relationships in a morphologically diversified plant species complex when nuclear and chloroplast sequences fail to reveal variability. Molecular Phylogenetics and Evolution 27: 185–196.[CrossRef][ISI][Medline]
Dodd, R. S., AND N. Kashani. 2003. Molecular differentiation and diversity among the California red oaks (Fagaceae; Quercus section Lobatae). Theoretical and Applied Genetics 107: 884–892.[CrossRef][Medline]
Duncan, F. 1989. Systematic affinities, hybridisation and clinal variation within Tasmanian eucalypts. Tasforests 2: 13–24.
Excoffier, L., P. E. Smouse, AND J. M. Quattro. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: Application to human mitochondrial DNA restriction data. Genetics 131: 479–491.[Abstract]
Faith, D. P. 1992. Conservation evaluation and phylogenetic diversity. Biological Conservation 61: 1–10.[CrossRef][ISI]
Farris, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5: 417–419.[ISI]
Felsenstein, J. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution; International Journal of Organic Evolution 39: 783–791.
Gibbs, A. 2007. Phylogeny and biogeography of Eucalyptus subgenus Eudesmia. Ph.D. dissertation, University of Melbourne, Australia.
Jackson, H. D., D. A. Steane, B. M. Potts, AND R. E. Vaillancourt. 1999. Chloroplast DNA evidence for reticulate evolution in Eucalyptus (Myrtaceae). Molecular Ecology 8: 739–751.[CrossRef]
Jones, R. C., G. E. McKinnon, B. M. Potts, AND R. E. Vaillancourt. 2005. Genetic diversity and mating system of an endangered tree, Eucalyptus morrisbyi. Australian Journal of Botany 53: 367–377.[CrossRef][ISI]
Jones, R. C., D. A. Steane, B. M. Potts, AND R. E. Vaillancourt. 2002. Microsatellite and morphological analysis of Eucalyptus globulus populations. Canadian Journal of Forest Research 32: 59–66.[CrossRef]
Jordan, G. J., B. M. Potts, J. B. Kirkpatrick, AND C. Gardiner. 1993. Variation in the Eucalyptus globulus complex revisited. Australian Journal of Botany 41: 763–785.[CrossRef][ISI]
Kelleher, C. T., T. R. Hodkinson, G. C. Douglas, AND D. L. Kelly. 2005. Species distinction in Irish populations of Quercus petraea and Q. robur: Morphological versus molecular analyses. Annals of Botany 96: 1237–1246.
Kirkpatrick, J. B. 1974. The numerical intraspecific taxonomy of Eucalyptus globulus Labill. (Myrtaceae). Botanical Journal of the Linnean Society 69: 89–104.[CrossRef][ISI]
Kluge, A. G., AND J. S. Farris. 1969. Quantitative phyletics and the evolution of anurans. Systematic Zoology 18: 1–32.
Koopman, W. J. M. 2005. Phylogenetic signal in AFLP data sets. Systematic Biology 54: 197–217.[CrossRef][ISI][Medline]
Koopman, W. J. M., M. J. Zevenbergen, AND R. G. Van den Berg. 2001. Species relationships in Lactuca s.l. (Lactuceae, Asteraceae) inferred from AFLP fingerprints. American Journal of Botany 88: 1881–1887.
Ladiges, P. Y., M. B. Dale, D. R. Ross, AND K. G. Shields. 1984. Seedling characters and phylogenetic relationships in the informal series Ovatae of Eucalyptus, subgenus Symphyomyrtus. Australian Journal of Botany 32: 1–13.[Medline]
McEntee, A. C., B. M. Potts, AND J. B. Reid. 1994. Variation in Eucalyptus barberi L. Johnson &Blaxell. Proceedings of the Royal Society of Tasmania 128: 21–30.
McGowen, M. H., R. J. E. Wiltshire, B. M. Potts, AND R. E. Vaillancourt. 2001. The origin of Eucalyptus vernicosa, a unique shrub eucalypt. Biological Journal of the Linnean Society 74: 397–405.[CrossRef][ISI]
McKinnon, G. E., B. M. Potts, D. A. Steane, AND R. E. Vaillancourt. 2005. Population and phylogenetic analysis of the cinnamoyl coA reductase gene in Eucalyptus globulus (Myrtaceae). Australian Journal of Botany 53: 827–838.[CrossRef][ISI]
McKinnon, G. E., D. A. Steane, B. M. Potts, AND R. E. Vaillancourt. 1999. Incongruence between chloroplast and species phylogenies in Eucalyptus subgenus Monocalyptus (Myrtaceae). American Journal of Botany 86: 1038–1046.
McKinnon, G. E., R. E. Vaillancourt, H. D. Jackson, AND B. M. Potts. 2001. Chloroplast sharing in the Tasmanian eucalypts. Evolution; International Journal of Organic Evolution 55: 703–711.[Medline]
McKinnon, G. E., R. E. Vaillancourt, D. A. Steane, AND B. M. Potts. 2004. The rare silver gum, Eucalyptus cordata, is leaving its trace in the organellar gene pool of Eucalyptus globulus. Molecular Ecology 13: 3751–3762.[CrossRef][Medline]
Nei, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York, New York, USA.
Nei, M., AND W.-H. Li. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proceedings of the National Academy of Sciences, USA 76: 5269–5273.
Nicolle, D. 2006. Eucalypts of Victoria and Tasmania. Bloomings Books, Melbourne, Australia.
Nicolle, D., AND M. A. Whalen. 2006. A taxonomic revision and morphological variation within Eucalyptus series Subulatae subseries Spirales (Myrtaceae) of southern Australia. Australian Systematic Botany 19: 87–112.[CrossRef][ISI]
Nicolle, D., M. A. Whalen, AND D. A. Mackay. 2006. Morphological variation and phylogenetic relationships within Eucalyptus series Subulatae (Myrtaceae) of southern Australia. Australian Systematic Botany 19: 59–86.[CrossRef][ISI]
O'Hanlon, P. C., AND R. Peakall. 2000. A simple method for the detection of size homoplasy among amplified fragment length polymorphism fragments. Molecular Ecology 9: 815–816.[CrossRef][Medline]
Parra-O., C., M. Bayly, F. Udovicic, AND P. Ladiges. 2006. ETS sequences support the monophyly of the eucalypt genus Corymbia (Myrtaceae). Taxon 55: 653–663.[ISI]
Pellmyr, O., K. A. Segraves, D. M. Althoff, M. Balcázar-Lara, AND J. Leebens-Mack. 2007. The phylogeny of yuccas. Molecular Phylogenetics and Evolution 43: 493–501.[CrossRef][Medline]
Phillips, R. L., AND J. B. Reid. 1980. Clinal variation between Eucalyptus viminalis Labill. and E. dalrympleana Maiden. Australian Journal of Botany 28: 329–342.[CrossRef][ISI]
Poke, F. S., D. P. Martin, D. A. Steane, R. E. Vaillancourt, AND J. B. Reid. 2006. The impact of intragenic recombination on phylogenetic reconstruction at the sectional level in Eucalyptus when using a single copy nuclear gene (cinnamoyl CoA reductase). Molecular Phylogenetics and Evolution 39: 160–170.[CrossRef][ISI][Medline]
Potts, B. M., AND R. J. E. Wiltshire. 1997. Eucalypt genetics and genecology. In J. Williams, and J. Woinarski [eds.], Eucalypt ecology: Individuals to ecosystems, 56–91. Cambridge University Press, Cambridge, UK.
Pryor, L. D., AND L. A. S. Johnson. 1971. A classification of the eucalypts. Australian National University, Canberra, Australia.
Rathbone, D. A., G. E. McKinnon, B. M. Potts, D. A. Steane, AND R. E. Vaillancourt. 2007. Microsatellite and cpDNA variation in island and mainland populations of a regionally rare eucalypt, Eucalyptus perriniana (Myrtaceae). Australian Journal of Botany 55: 513–520.[CrossRef][ISI]
Rohlf, F. J. 2000. NTSYS-pc: Numerical taxonomy and multivariate analysis system, version 2.1. Exeter Software, Setauket, New York, USA.
Saitou, N., AND M. Nei. 1987. The neighbor-joining method: A new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4: 406–425.[Abstract]
Schmidt-Lebuhn, A. N. 2007. Using amplified fragment length polymorphism (AFLP) to unravel species relationships and delimitations in Minthostachys (Labiatae). Botanical Journal of the Linnean Society 153: 9–19.[CrossRef][ISI]
Schmidt-Lebuhn, A. N., M. Kessler, AND M. Kumar. 2006. Promiscuity in the Andes: Species relationships in Polylepis (Rosaceae, Sanguisorbeae) based on AFLP and morphology. Systematic Botany 31: 547–559.[ISI]
Schneider, S., D. Roessli, AND L. Excoffier. 2000. ARLEQUIN, version 2.000: A software for population genetics data analysis. University of Geneva, Geneva, Switzerland. http://anthro.unige.ch/software/arlequin/
Scotti-Saintagne, C., S. Mariette, I. Porth, P. G. Goicoechea, T. Barreneche, C. Bodénès, K. Burg, AND A. Kremer. 2004. Genome scanning for interspecific differentiation between two closely related oak species. [Quercus robur L. and Q. petraea (Matt.) Liebl.] Genetics 168: 1615–1626.
Steane, D. A., M. Byrne, R. E. Vaillancourt, AND B. M. Potts. 1998. Chloroplast DNA polymorphism signals complex interspecific interactions in Eucalyptus (Myrtaceae). Australian Systematic Botany 11: 25–40.[CrossRef][ISI]
Steane, D. A., N. Conod, R. C. Jones, R. E. Vaillancourt, AND B. M. Potts. 2006. A comparative analysis of population structure of a forest tree, Eucalyptus globulus (Myrtaceae), using microsatellite markers and quantitative traits. Tree Genetics &Genomes 2: 30–38.[CrossRef][ISI]
Steane, D. A., G. E. McKinnon, R. E. Vaillancourt, AND B. M. Potts. 1999. ITS sequence data resolve higher level relationships among the eucalypts. Molecular Phylogenetics and Evolution 12: 215–223.[CrossRef][ISI][Medline]
Steane, D. A., D. Nicolle, G. E. McKinnon, R. E. Vaillancourt, AND B. M. Potts. 2002. Higher-level relationships among the eucalypts are resolved by ITS-sequence data. Australian Systematic Botany 15: 49–62.[CrossRef][ISI]
Swofford, D. L. 2002. PAUP*: Phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sinauer, Sunderland, Massachusetts, USA.
Tajima, F. 1983. Evolutionary relationship of DNA sequences in finite populations. Genetics 105: 437–460.
Udovicic, F., AND P. Y. Ladiges. 2000. Informativeness of nuclear and chloroplast DNA regions and the phylogeny of the eucalypts and related genera (Myrtaceae). Kew Bulletin 55: 633–645.[CrossRef]
Udovicic, F., G. I. Mcfadden, AND P. Y. Ladiges. 1995. Phylogeny of Eucalyptus and Angophora based on 5S rDNA spacer sequence data. Molecular Phylogenetics and Evolution 4: 247–256.[CrossRef][Medline]
Vos, P., R. Hogers, M. Bleeker, M. Reijans, T. Van de lee, M. Hornes, A. Frijters, J. Pot, J. Peleman, M. Kuiper, AND M. Zabeau. 1995. AFLP: A new technique for DNA fingerprinting. Nucleic Acids Research 23: 4407–4414.
Whittock, S., D. A. Steane, R. E. Vaillancourt, AND B. M. Potts. 2003. Molecular evidence shows that the tropical boxes (Eucalyptus subgenus Minutifructus) are over-ranked. Transactions of the Royal Society of South Australia 127: 27–32.[ISI]
Williams, K. J., AND B. M. Potts. 1996. The natural distribution of Eucalyptus species in Tasmania. Tasforests 8: 39–165.
Wright, S. 1951. The genetical structure of populations. Annals of Eugenics 15: 323–354.[ISI]
Zhang, L. B., H. P. Comes, AND J. W. Kadereit. 2001. Phylogeny and quaternary history of the European montane/alpine endemic Soldanella (Primulaceae) based on ITS and AFLP variation. American Journal of Botany 88: 2331–2345.
| |||||||||||||||
50%).