Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

A box on the river: The phylogenetics and phylogeography of Eucalyptus baueriana (Eucalyptus sect. Adnataria ser. Heterophloiae)

  • Patrick S. Fahey ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing – original draft, Writing – review & editing

    Current address: Queensland Alliance for Agriculture and Food Innovation, University of Queensland, Brisbane, QLD, Australia; Research Centre for Ecosystem Resilience, Royal Botanic Gardens, Australian Institute of Botanical Sciences, Sydney, Australia

    Affiliation School of BioSciences, The University of Melbourne, Parkville, Victoria, Australia

  • Frank Udovicic,

    Roles Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing

    Affiliation Royal Botanic Gardens Victoria, South Yarra, Victoria, Australia

  • David J. Cantrill,

    Roles Conceptualization, Methodology, Supervision, Writing – review & editing

    Affiliation Royal Botanic Gardens Victoria, South Yarra, Victoria, Australia

  • Michael J. Bayly

    Roles Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – review & editing

    Affiliation School of BioSciences, The University of Melbourne, Parkville, Victoria, Australia


We present a phylogeographic study of the tree species Eucalyptus baueriana Schauer, which occurs in disjunct areas on the near coastal plains and ranges of the south-east Australian mainland. DArTseq data are used to build a phylogeny including E. baueriana and closely related taxa to test its monophyly, test the genetic distinctness of the three subspecies of E. baueriana, and investigate relationships between its disjunct populations. Additionally, we use population structure analysis to investigate the genetic distinctness of populations, and MaxEnt to investigate the environmental factors potentially influencing the species’ distribution. We show E. baueriana is monophyletic and most closely related to three other Blue Box eucalypt species: E. conica H.Deane & Maiden, E. dalveenica T.L.Collins, R.L.Andrew & J.J.Bruhl and E. magnificata L.A.S.Johnson & K.D.Hill, with some evidence for genetic introgression between these taxa. Within E. baueriana, the deepest genetic breaks do not correspond with the subspecies classification as the two geographically restricted subspecies, together with samples of the more widespread E. baueriana subsp. baueriana from west of the Gippsland lowlands, form a south-western clade with that is sister to other populations of subsp. baueriana. The oldest genetic break in the species occurs in far eastern Gippsland (Victoria), corresponding to one of the shortest geographic disjunctions in the species’ distribution. Genetic breaks in other species have been observed in this region which is broadly referred to as the southern transition zone. Both total annual rainfall and the seasonality of this rainfall are hypothesised to affect the species’ distribution; gaps in its distribution are in areas of higher rainfall that support closed forest and in regions with more winter dominated rainfall.


The eastern coast of the Australian mainland stretches from the continent’s southernmost point at Wilsons Promontory in Victoria to the northernmost point on Cape York Peninsula in North Queensland, crossing a latitudinal gradient of more than 3500 km. This gradient covers climatic regions from tropical in the north to cool temperate in the south [1], leading to dramatic changes in ecosystems and vegetation along the coast [2]. Previous work has identified many transition zones and biogeographical barriers that help to shape the diversity of these changes [25].

The Southern Transition Zone (STZ) is one such transition zone and corresponds to areas of lower elevation plains that separate the upland areas of the Great Dividing Range (GDR) and the lowland coastal areas in south-east New South Wales [2, 4]. Additionally, the STZ sits near the boundary of the hot summer temperate and warm summer temperate climatic zones [1]. While many species have been shown to have a genetic break in this region, these breaks are often not concordantly located across species. Several barriers to closed forest taxa have been identified in this region (e.g. the Illawarra district, a low-lying coastal plain [2] and the Shoalhaven River [4]). Patterns in sclerophyllous open forest and woodland species have been less obvious in the region and minimal genetic structuring has been observed in many plant taxa that form part of these communities including Xanthorrhoea spp. [6], Correa spp. [7] and Callitris rhomboidea R.Br. ex A.Rich. & Rich [8]. Different patterns have been observed in Hardenbergia violacea (Schneev.) Stearn, which shows high haplotype diversity on the south coast of NSW and evidence of more recent expansion southward from this region [9].

In this study we set out to further elucidate patterns of genetic diversity in sclerophyllous plant taxa of the STZ by investigating patterns of genetic diversity in Eucalyptus baueriana Schauer, Blue Box, a member of Eucalyptus section Adnataria series Heterophloiae (Fig 1A). Eucalyptus ser. Heterophloiae contains nine species that predominately occur in woodlands and open forests in south-eastern Australia (Fig 1A). The only species to occur outside this area, the mallee (multi-stemmed) species E. lucens Brooker & Dunlop that occurs on arid slopes in rocky ranges west of Alice Springs in central Australia [10] (Fig 1A), has been suggested to be the most distinct member of the complex [11]. Eucalyptus fasciculosa F.Muell. is also geographically distinct from all other members of the complex, occurring on Kangaroo Island and much of south-eastern South Australia from the Mount Lofty Ranges south-east across the Victorian border (Fig 1A). A pair of morphologically similar and closely related species, considered by Nicolle [11] to be potentially conspecific, E. rudderi Maiden and E. hypostomatica L.A.S.Johnson & K.D.Hill are restricted to central NSW [10]. The former is limited to a single coastal area around Taree while the latter occurs in two distinct areas in the ranges around the Sydney Basin: around Lake Burragorang and in the hills that form the southern border of the lower Hunter Valley (Fig 1A). The southern population of E. hypostomatica is parapatric with E. baueriana populations on the Cumberland Plain.

Fig 1. Maps of south-eastern Australia based on records from the Atlas of Living Australia (accessed 30 November 2020) showing (A) distribution of the members of Eucalyptus sect.

Adnataria including E. baueriana, and (B) the distribution of E. baueriana including locations of samples used in this study. Bioregions, rivers and the Southern Transition Zone (STZ) discussed in the text are shown in panel B. Bioregions are those of the Interim Biogeographic Regionalisation for Australia Version 7 [12] subregion classification (adapted from [13]). While the boundaries of the STZ are not defined, we have based our understanding of it on the thoughts of previous authors [2, 4]. Eucalyptus baueriana records downloaded from the Atlas of Living Australia are indicated with blue triangles and points representing collections of the species are coloured and labelled to match Figs 2 and 3. The base map is adapted from [14] under a CC BY 3.0 AU license.

Another species in the group that overlaps in distribution with E. baueriana is E. polyanthemos Schauer, which occurs in woodlands on the inland slopes of the GDR from central Victoria to the southern edge of the Brigalow Belt between Dubbo and Goulburn River National Park and on the coastal side of the GDR in Gippsland from the Traralgon region into far south-east NSW [10]. These two species are differentiated where they co-occur by several features including the parts of the landscape they occupy; E. baueriana is a species of riparian vegetation whereas E. polyanthemos most commonly occurs on ridges and slopes away from riparian zones. Morphologically, the species are differentiated by their habit and degree of rough bark coverage, with E. baueriana typically a more robust species that always has rough, box-type bark [10] covering the trunk and large branches, a trait that is sometimes absent in E. polyanthemos, and by the fruit and leaves, with E. baueriana having more obconical fruit and glossy green leaves which are noticeably thinner and more delicate than the leathery leaves of E. polyanthemos [10].

The three remaining species in the series, E. conica H.Deane & Maiden, E. magnificata and E. dalveenica T.L.Collins, R.L.Andrew & J.J.Bruhl, do not have overlapping distributions with E. baueriana but together with it were termed the Blue Boxes by Collins et al. [14]. Eucalyptus conica is widespread on the inland slopes of the GDR and adjacent plains from the West Wyalong-Young-Cowra region north to the south Burnett region and Barakula State Forest in Queensland, with an outlying population to the north-west in and around Carnarvon National Park. Maiden [15] considered this species to be conspecific with E. baueriana, although this taxonomy has not been followed by any authors since. The distribution of E. conica completely encompasses that of both E. magnificata, which is only found on the south-east Armidale Plateau in the New England region of NSW, and the recently described E. dalveenica known initially from a single, extremely restricted population near the town of Dalveen in the Granite Belt region of southern Queensland [16], although the Stanthorpe Rare Wildflower Consortium has confirmed via email (, May 24th 2022) that a second population of the species has subsequently been discovered on private land approximately 3 km from the initial population. Prior to the description of E. magnificata [17], and E. dalveenica [16], populations of these taxa were included in a broader circumscription of E. baueriana [18].

In 2011, Rule [19] described two new subspecies of E. baueriana, both restricted to isolated populations in Victoria. Eucalyptus baueriana subsp. thalassina includes all populations in the Werribee River catchment west of Melbourne and was considered to be morphologically distinct based upon its “smaller habit, smaller buds with shorter pedicels and smaller fruits” [19]. Populations on the Deddick River in eastern Victoria were classified as E. baueriana subsp. deddickensis Rule and were described as having a smaller stature than E. baueriana subsp. baueriana, along with darker green adult leaves and smaller buds and fruit [19]. Since these publications, these subspecies have not been recognised by many other authors [e.g. 10, 11], but are considered valid by the Australian Plant Census [20].

Eucalyptus baueriana occurs in several discrete areas on the south-east coast of Australia between the Sydney Basin and Werribee River catchment west of Melbourne (Fig 1B). The species primarily occurs in fertile soils in riparian corridors along creeks and rivers in open forests and woodlands [19], and it is absent from many river catchments in the south-east corner with higher rainfall that supports closed forest. The westernmost populations occur in the Werribee River catchment west of Melbourne (Fig 1B) and are those described as E. baueriana subsp. thalassina by Rule [11]. The largest remaining population in this catchment occurs along Pyrites and Djerriwarrh creeks in the Long Forest Flora and Fauna Reserve; a biogeographical oddity which supports the only mallee vegetation south of the GDR [21]. These isolated E. baueriana populations are separated by more than 300 km from the nearest populations to the east on the boundary of the Gippsland plain and the East Gippsland Lowlands (Fig 1B). Much of this disjunction lies across the Gippsland Plain (Fig 1B), which has been hypothesised to be a biogeographical break in closed forest taxa [2]. However, as E. baueriana is not a closed forest species the environmental factors maintaining this disjunction likely differ from these taxa. North of these populations, several extremely restricted populations along the Deddick River, a tributary of the Snowy River (Fig 1B), represent the second geographically restricted subspecies, E. baueriana subsp. deddickensis Rule [11]. The populations of this subspecies are the most inland known of the species and occur in a landscape of rough, steep terrain with deep river gorges that is very different to the coastal plains and hills typical of the species.

East of the eastern Gippsland Plains/western East Gippsland Lowlands and Deddick River populations, the species inhabits the upper reaches of the Genoa River/Wallagaraugh River catchment, also in the East Gippsland Lowlands region (Fig 1B); with consistent occurrences along many water courses north from here as far as the Deua and Moruya rivers at Moruya. A further restricted population on the south coast of NSW occurs in the Bendalong region just south of the Illawarra district. This population is not associated with a particular watercourse, rather inhabiting a small area of coastal woodland on Bendalong Point. These populations are within the STZ as defined by both Milner et al. [4] and Bryant and Krosch [2]. The species is absent from the Illawarra region and ranges north of there but occurs along several watercourses on the Cumberland Plains in the Sydney Basin. While no previous phylogenetic nor distribution-wide population genetic studies have been undertaken on this species, populations were sampled from the Cumberland Plain and Bendalong Point in a study of E. magnificata L.A.S.Johnson & K.D.Hill, and these populations were shown to be somewhat genetically and morphologically distinct from one another [16].

In this study, we aim to establish the monophyly of E. baueriana within Eucalyptus series Heterophloiae and whether there is any genetic support for the recognition of the subspecies of Rule [19]. We then investigate patterns of genetic relatedness between the disjunct populations of E. baueriana and build hypotheses regarding the environmental processes that have established and maintained this unique distribution. We do this by developing a reduced representation genetic dataset utilising Diversity Arrays Technologies sequencing (DArTseq) [22, 23] on both samples of all species in E. series Heterophloiae along with outgroups of various phylogenetic distance, and population level sampling from across the range of E. baueriana. We present phylogenetic analyses of the complete dataset and a STRUCTURE analysis on our E. baueriana samples.



We had initially aimed to collect ninety total samples of E. baueriana, ten samples from nine sites across the species distribution. However, due to small population sizes and limits to travel due to COVID-19, actual sample numbers per population ended up ranging from one to ten, and we added two further sampling sites bringing the total number of sites to eleven. As we could only access a single stand of fourteen individuals on a ~150 m stretch of the Deddick River in the Snowy River National Park due to the steep landscape, we limited our sample to the eight most spread-out individuals at the site to limit sampling of siblings, leaving a total of eighty-three samples of the species. All field collections consisted of approximately ten leaves placed into individually labelled coffee filters and then into silica desiccant beads for rapid drying. At least one representative specimen was taken from each collection site for accessioning in the University of Melbourne Herbarium (Table 1). All collecting was undertaken under scientific collecting permits granted by the Victorian Department of Sustainability and Environment (permit number 10008557), New South Wales National Parks and Wildlife Service (permit number SL102100), and South Australian Department of Environment and Water (permit number Q26766-2).

Table 1. Collection details for samples used in this study.

For E. baueriana collection sites, where multiple samples were collected, geographic coordinates are the mean of the individual sample coordinates.

We supplemented our sampling of E. baueriana with sampling of all other members of series Heterophloiae. Field collected samples in this group included: two samples of E. polyanthemos from Victoria, two samples of E. dalveenica from near the township of Dalveen and a sample of E. conica from the town of Warwick in southern Queensland. We sampled from six herbarium specimens from the Australian National Herbarium and the National Herbarium of Victoria: two samples of E. magnificata, and one each of E. rudderi, E. hypostomatica, E. conica and E. lucens. We were also able to take advantage of a previous DArTseq run to supplement our data and provide outgroups, which included two further samples from E. series Heterophloiae: a sample each of E. polyanthemos and E. fasciculosa, completing our sampling of all species in the series. The additional E. sect. Adnataria samples from this previous work included single samples of twenty-one species from E. ser. Melliodorae, E. ser. Buxeales and E. ser. Subbuxeales. For outgroups we included one member of the sister section Dumaria, and two samples from E. section Bisectae.

DNA extraction and DArT-seq

A standard CTAB protocol per McLay [24] modified per Fahey et al. [25] was undertaken on 70–80 mg of dried leaf material per sample. DNA pellets were resuspended in 50 μL of 10 mM Tris HCl pH 7 buffer before extraction quality and concentration were quantified using a Nanodrop 2000 spectrophotometer (ThermoFisher) and Qubit 2.0 fluorometer (Invitrogen). Based upon Qubit estimates of DNA concentrations, samples were standardised to 100 ng DNA/μL and sent to Diversity Arrays Technology Pty Ltd, Canberra for DArTseq data generation [22, 23]. Extractions from one field collection and three herbarium specimens were below the target concentration of 100 ng/μL and were included at the concentration of the extraction, the lowest being ~60 ng/μL. While all field collected samples passed initial quality controls for DArTseq, five of the six herbarium specimens (E. magnificata CANB540728, E. magnificata CANB875142, E. rudderi MEL0706391, E. conica CANB480492 and E. lucens MEL0278390) initially failed due to low concentrations and short DNA fragments. These samples were rerun at a higher concentration and, while still producing fewer reads than field collections, four of the five returned usable data, with only the sample of E. lucens (MEL0278390) not producing sufficient reads to be included in our phylogenetic analyses. All datasets used in this study are available for download from Dryad ( [26]).


As the SNP dataset returned by DArT and used in later analyses was not in a format that could easily be used in rooted phylogenetic analyses and had bias in missing data towards outgroup taxa, a known issue in DArTseq datasets, for phylogenetic analyses we reconstructed loci from the raw sequencing reads using ipyrad [27]. These reads were first trimmed to remove adaptor and barcode sequences using CutAdapt [28], rendering them 75 bp in length on average. The reads were treated as single end RAD data and assembled using reference mapping to the eleven chromosomes of the E. grandis reference genome [29]. This approach was taken to remove plastid and mitochondrial loci from the dataset, as it has been shown that the cpDNA phylogeny of the eucalypts often has discordance with nDNA phylogenies [30, 31], which may have influenced our results had these loci not been removed. Default ipyrad settings were changed as follows: maximum low-quality bases = 2, minimum samples per locus = 4, cluster threshold = 0.9, maximum SNPs per locus = 10%, maximum indels per locus = 5 and maximum shared heterozygous sites = 0.25. A maximum likelihood (ML) analysis was performed on the PHYLIP file containing the full data alignment produced by ipyrad using RAxML [32]. The rapid bootstrap analysis and ML tree search algorithm was used for tree generation under a GTR model with BFGS rate optimisation, and one hundred bootstrapping replicates were performed. A maximum parsimony (MP) analysis was undertaken on the SNP alignment generated by ipyrad in PAUP* [33], with ambiguous bases treated as polymorphisms and one hundred bootstrap replications performed. Resulting ML and MP trees were viewed in TreeGraph 2 [34] rooting on the two section Bisectae samples (E. oleosa F.Muell. ex Miq. PSF41B and E. leptophylla F.Muell. ex Miq. PSF22I), and then compared for incongruence by eye. Samples with unexpected placement in resultant phylogenies were tested for introgression due to hybridisation between taxa using ABBA-BABA tests [35] run in the ipyrad toolbox.

To investigate patterns of genetic relatedness and diversity across the sampling sites of E. baueriana, the original SNP dataset generated by DArT was used to perform STRUCTURE analysis [36, 37] as outgroups are not included in this analysis. Data were filtered in R [38] utilising the dartR package [39] to include only eighty-one of the E. baueriana samples and one SNP per marker with a call rate of ≥ 90% and a reproducibility of 1. We excluded two E. baueriana samples from this analysis: the sample PSF117D from Ropes Creek that did not form part of the monophyletic E. baueriana clade in the phylogeny, and the single Redbank Creek sample PSF116A. The latter was removed as it was not supported as closely related to any other samples in the phylogeny, thus representing a lone sample from a unique population, and uneven sampling of populations can lead to incorrect inferences of K in STRUCTURE analyses [40]. Additionally, when we did include it in our analyses, at the optimal value of K it showed relatively even contributions from multiple clusters that corresponded to other populations rather than a unique cluster by itself, which we could not meaningfully interpret. Basic population statistics including observed and expected heterozygosity, Fis and Fst values were calculated on these SNPs using the relevant functions with dartR.

The filtered dataset was exported from R into a.str formatted file for STRUCTURE analysis. This analysis was undertaken using the StrAuto Python program that facilitates automation and parallelisation of runs [41]. Ten replicate runs for values of K one through ten were completed with a burnin of 100 000 followed by a run of length 500 000, with the upper limit of K being chosen based upon the ten sampling sites represented in the dataset. Once the StrAuto run was completed, the probability and delta K was examined to find the optimal values of K (Table 2). Results from these values of K were then summarised across runs using CLUMPP [42] and plotted using the online implementation of STRUCTURE PlotV2 [43].

Table 2. Key statistics used to determine optimal values of K (number of clusters) for structure analysis.

All values are derived from ten replicate runs at each K value. LnP(K) is minimised at K = 8, while Delta K is highest at K = 2, reflecting the two values of K shown in Fig 3.

Species distribution modelling.

MaxEnt [44] was used to build species distribution models to assist in building hypotheses regarding the environmental factors that influence the distribution of E. baueriana. Thirty initial environmental factors were chosen as potential predictors for these models: the nineteen bioclimatic variables available from WorldClim [45], ten soil characteristics (organic carbon content, clay content, silt content, sand content, nitrogen content, phosphorus content, pH, available water capacity, depth of soil and depth of regolith) available from the Soil and Landscape Grid of Australia measured at a depth of 30–60 cm where relevant [46] and a 1 arc-second digital elevation model of Australia [47]. Bioclimatic and elevation rasters were resampled and aligned to match the resolution of the soil data rasters (3 arc-seconds) using a bilinear (2x2 kernel) resampling scheme, and all predictors were clipped to an area of south-east Australia defined by the following IBRA 7 regions [12] in which the species occurs, or which are within 50 km of a known population of the species: Australian Alps, South East Coastal Plain, South East Corner, South Eastern Highlands, Southern Volcanic Plain, Sydney Basin and Victorian Midlands. Occurrence records of E. baueriana were downloaded from the Atlas of Living Australia [48: accessed June 2020] and initially filtered manually to remove records that were outliers, likely misidentifications or incorrectly georeferenced, and to only include records with unique coordinates.

The VariableSelection function from the MaxentVariableSelection package [49] in R [38] was used to test and optimise both the environmental variable set and beta multiplier for our MaxEnt model. We tested beta multipliers between 1 and 15 using a 0.5 step change and predictors variables were chosen using a contribution threshold of 5% and a correlation threshold of 0.9. The optimised model under these parameters included four predictors (Bio04 temperature seasonality, Bio08 mean temperature of wettest quarter, Bio19 precipitation of coldest quarter and elevation) and a beta multiplier of 1 and was run in the standalone implementation of MaxEnt to produce the final model presented here.


Data overview

Reads per sample generated by DArTseq ranged from 1,598,695 to 5,256,113, excluding the failed E. lucens sample (MEL0278390) which returned only 54,343 reads. A total of 43,054 loci were reconstructed by ipyrad, which created a total data alignment of 2,811,766 bases (79.6% missing data) used for RAxML ML analysis and a matrix of 108,777 SNPs (72.1% missing data) used for MP analysis. The filtered DArTseq SNP dataset (one SNP per locus) used in the STRUCTURE analysis contained 4323 SNPs (6.7% missing data).


Although there were some topological differences between the ML and MP phylogenies, there were no supported conflicts between the two analyses. The resulting phylogeny (Fig 2) shows that E. series Heterophloiae is the monophyletic sister group to the rest of our section Adnataria samples from E. series’ Melliodorae, Buxeales and Subbuxeales, though the later clade is not supported at an 80% bootstrap threshold in the MP analysis (Fig 2). Two clades form at the base of the series; an unsupported one consisting of the E. fasciculosa sample sister to E. rudderi and E. hypostomatica, and a clade only supported in the ML analysis containing E. polyanthemos and the Blue Boxes including E. baueriana. The three samples of E. polyanthemos formed a supported clade, with the Gippsland sample sister to the two samples from central Victoria. The Blue Box species, E. baueriana, E. conica, E. magnificata and E. dalveenica, form a supported clade sister to E. polyanthemos, and apart from a single sample of E. baueriana from the Ropes Creek site (PSF117D), these species were each supported as monophyletic in the MP analysis, although in the ML analysis support for the E. baueriana clade falls just short of the 80% bootstrap threshold. Relationships between these Blue Box species are less supported, with the topology suggesting E. baueriana is sister to the other three taxa, with E. conica and E. dalveenica being sister and the E. baueriana sample PSF117D and E. magnificata being sister to these species in turn. The ABBA-BABA test (D-statistic = 0.223, D-statistic bootstrapped standard deviation = 0.089, Z-score = 2.500) showed bias in shared derived alleles between the E. baueriana sample in this clade and E. conica compared to the remaining samples collected at Ropes Creek but did not reach the commonly used significance threshold of a Z-score > 3 [50].

Fig 2. Maximum likelihood phylogeny including all samples of Eucalyptus sect. Adnataria included in this study.

The phylogeny is rooted on two samples from E. sect. Bisectae and support values shown are ML and MP bootstrap values and branches with bootstrap values above 80 for both analyses are thickened. Members of E. ser. Heterophloiae are indicated by tip bubbles coloured to correspond to Fig 1A and tip labels of E. baueriana are coloured by their collecting site of origin matching Fig 1B.

Within the E. baueriana clade two geographic clades form, one of which is supported in both ML and MP analyses and which we call the south-western clade contains samples from the Boggy Creek and Tambo River sites from the eastern Gippsland Plain/western East Gippsland Lowlands region, the Deddick River site, and the two sites in the Werribee River catchment at Long Forest and Werribee Gorge (Fig 2). The second, which lacks support in both analyses, contains all samples from Maramingo Creek in the Genoa River/Wallagaraugh River catchment at the eastern side of the East Gippsland Lowlands, and all sites to the north of this in NSW excluding the Ropes Creek sample PSF117D, and which we call the north-eastern clade. Within the south-western clade, three main supported clades are present corresponding to the three catchments sampled from, with the ML but not the MP analysis supporting the Deddick River and Werribee River catchments as sister clades. Few relationships are supported within these clades; however, several samples in the Deddick River clade are likely closely related to one another, possibly even siblings, given their short tip branches.

The sole Redbank Creek sample is sister to the rest of the north-eastern clade. The remainder of the north-eastern clade is supported as a single lineage and four of the five sampling sites form supported clades, with only the Tuross River plants not being monophyletic. The Bendalong samples were supported as the sister lineage to the four other sites, which is unexpected given the location is between the Ropes Creek site on the Cumberland Plain and the three other sites to the south. Relationships between the four remaining sites are not resolved with the Maramingo Creek clade, Doctor George Mountain clade and a clade containing the Tuross River and Ropes Creek samples. Within the Tuross River and Ropes Creek clade, the Ropes Creek samples show very limited genetic divergence from one another and form a supported clade that is part of a polytomy with the Tuross River samples between which relationships are largely unresolved.

Eucalyptus baueriana population genetics and structure analysis

Fst and heterozygosity values for E. baueriana populations (Table 3) show lower observed heterozygosity than expected heterozygosity and a high overall Fis (0.286), providing evidence for inbreeding at most collection sites. The overall Fst value (0.218) for the species is high, indicating strong genetic structuring of the species, which reinforces the lack of geneflow between populations. Two isolated populations, Deddick River and Ropes Creek, show consistently high Fst values when compared to all other populations. Of the remaining populations, those that formed the south-western clade in the phylogeny (discussed below), bar the Deddick River, shared lower reciprocal Fst values, which is also true of all sites from the NSW south coast and Maramingo Creek in far eastern Victoria that are members of the north-eastern clade in the phylogeny.

Table 3. Population statistics for the ten sampling sites of E. baueriana investigated in this study.

Cells showing interpopulation Fst values are coloured by the magnitude of the value, with redder cells having higher Fst values and bluer cells having lower.

Two K values showed high delta K and likelihood scores, two and eight. While delta K skews towards favouring K = 2 [51], this value of K is still presented here as it fits with our a priori knowledge of the two major clades in the phylogeny, while K = 8 matches the eight main sampling regions included in the dataset. Indeed, when the results for these two K-values are plotted these expectations are realised. As shown in Fig 3, the two clusters in the K = 2 analysis correspond broadly to the two clades in the phylogeny, a south-eastern cluster containing the Werribee River catchment (Werribee River Gorge and Pyrites and Djerriwarrh creeks samples), Deddick River, and Tambo River samples. The Tambo River samples are suggested to share greater ancestry with the north-eastern lineage than the first two. Reciprocally, we see the highest share of south-western ancestry of any north-eastern population in samples from Maramingo Creek, the site nearest the south-western populations.

Fig 3. Structure plots showing clustering of E. baueriana samples at the two optimal values of K: K = 2 (top) which correspond to the two main clades found in the phylogeny (Fig 2) and K = 8 (bottom) where clusters correspond to the eight major areas sampled.

In the K = 8 analysis, the 8 ancestry’s largely correspond to the eight main regions our samples are sourced from; the Werribee River catchment (including both the Werribee River Gorge and Pyrites and Djerriwarrh creeks sampling sites), the Deddick River, the eastern Gippsland Plain/western East Gippsland Lowlands (including both the Tambo River and Boggy Creek sites), Maramingo Creek (Genoa River/ Wallagaraugh River catchment), Doctor George Mountain (Bega River catchment), the Tuross River, Bendalong, and Ropes Creek (Cumberland Plain). The ancestry that corresponds to the Deddick River samples is largely absent from other samples, however there is a significant contribution of the Werribee River catchment lineage and the eastern Gippsland Plain/western East Gippsland Lowlands lineage into the Deddick River samples. There is also a small reciprocal contribution between samples in these two other lineages as well, however, in this analysis, there is very limited sharing of ancestry between the south-eastern and north-western clades. There was less complete separation of samples from the Maramingo Creek, Doctor George Mountain and Tuross River into distinct lineages, with contributions from each of the other’s ancestry being present in most samples, along with small contributions from the Bendalong and Ropes Creek lineages. The Bendalong samples mainly formed a single lineage but had small contributions from the Tuross River and Doctor George Mountain lineages. The Ropes Creek samples formed their own distinctive lineage.

Species distribution models

As shown in Fig 4, with only the four predictor variables MaxEnt returned a modelled suitability that strongly reflects the known distribution of the species. We see breaks in environmental suitability corresponding to three disjunctions, the Strzelecki Ranges and western Gippsland Plain, the Gippsland Lowlands, and the Illawarra district, however we also see areas of high suitability outside where the species is known to occur. There is a large extension of suitable habitat extending across a much greater portion of the eastern half of the Gippsland Plain than the known range of the species, no break in suitability corresponding to the disjunction between the south-east coast and Bendalong populations, and a large area of higher suitability north of the Cumberland Plain.

Fig 4. Map showing predicted environmental suitability in south-east Australia for E. baueriana based on MaxEnt model.

Four environmental predictor variables were employed in the model: Bio04 temperature seasonality, Bio08 mean temperature of wettest quarter, Bio19 precipitation of coldest quarter and elevation. Greener areas show higher predicted suitability while redder areas have lower predicted suitability, with blue triangles representing records of E. baueriana downloaded from Atlas of Living Australia (accessed 30 November 2020) and used to build model. The base map is adapted from [14] under a CC BY 3.0 AU license.

Bio08 (mean temperature of the wettest quarter) was the most important variable in the model with the modelled response to this variable suggesting that the species favours areas which are not winter rainfall dominated (Table 4). The inclusion of Bio04 (temperature seasonality) and Bio19 (precipitation of coldest quarter) in the optimised model (Table 4) corroborates this hypothesis that seasonality of temperature and rainfall are more important in determining the species distribution than annual averages. Elevation was the second most important predictor (Table 4), which is not unexpected given the species is mostly found on coastal plains and lowlands but is a complex variable which correlates with many environmental factors that may affect the distribution of E. baueriana more directly.

Table 4. Contribution and importance of variables in the optimised MaxEnt model for E. baueriana.


Phylogeny of E. ser. Heterophloiae and placement of E. baueriana

Our results support the monophyly of E. series Heterophloiae as currently defined, although we could not get data from our E. lucens sample to confirm its placement in the series. The subsectional classification of Brooker [52], which places series Heterophloiae together with series Melliodorae in subsection Terminales, is not supported (Fig 2). Rather, we find that series Melliodorae is grouped with members of subsection Apicales (series Buxeales and Subbuxeales) in the phylogeny (Fig 2).

Within E. series Heterophloiae inferred relationships between taxa are only somewhat congruent with what has been hypothesised in previous classification schemes. The relationship between E. fasciculosa and other members of the series was unclear, and the single sample of this species formed part of a polytomy at the crown of E. series Heterophloiae along with the E. rudderi/E. hypostomatica and E. polyanthemos/Blue Box clades. We find E. polyanthemos to be sister to the Blue Box clade, which fits with the thoughts of previous authors [11, 16, 53]. While Collins et al. [16] considered E. conica to be a member of the Blue Box group with E. baueriana, E. magnificata and E. dalveenica matching our phylogeny, both Brooker [52] and Nicolle [11] considered E. hypostomatica, E. rudderi and E. conica to form a group. Our results suggest that E. baueriana is the sister lineage to the other three species of Blue Box, E. conica, E. dalveenica and E. magnificata, with the former two possibly being sister taxa. However, the bootstrap support values on several key branches do not reach the 80% support threshold in one or both analyses, and a sample of E. baueriana from the northernmost sampled population at Ropes Creek falls in the clade containing the three other Blue Box taxa.

Collins et al. [14] identified stable populations of putative E. conica × E. magnificata hybrids, although work to confirm the exact identity of this entity is ongoing. This suggests hybridisation between the Blue Box species may occur where there is sympatry or parapatry. While there is no distributional overlap between E. baueriana and E. conica, the Ropes Creek site was the closest sampled to E. conica populations. We hypothesise the placement of the E. baueriana sample from this site outside the main E. baueriana clade may be due to historical hybridisation leading to introgression of E. conica genetic material into this population of E. baueriana. While our ABBA-BABA test does not show significant support for this hypothesis, there is weak evidence to reinforce it in the form of the D-statistic of 0.223 which indicates a slightly higher number of shared derived alleles between E. conica and the E. baueriana sample in question than between E. conica and other E. baueriana samples collected at Ropes Creek. Phantom hybrids have been observed in other Australian plant taxa, where there is a genetic signal for introgression from a species which does not currently co-occur with the introgressed population [54], including in eucalypts [5557], Additionally, this introgressed sample can be isolated as the cause of the low bootstrap values on the branches between the Blue Box species, as subsequent phylogenetic analyses with this sample removed from the dataset resulted in supported relationships placing E. conica and E. dalveenica as sister taxa, with E. magnificata sister to this clade (S1 Fig).

Subspecific classification.

Despite the two geographically restricted subspecies of E. baueriana being supported as sister clades in our phylogeny, they are nested within E. baueriana subsp. baueriana. The eastern Gippsland Plain/western East Gippsland Lowlands populations of the nominotypical subspecies are supported as more closely related to the two isolated subspecies than to other populations of E. baueriana subsp. baueriana. This suggests that if subspecies that represent distinct and monophyletic lineages are desired the current infraspecific circumscription needs to be reassessed. While it may be the case that monophyletic subspecies are not desired, nor necessarily expected given the divergence of a monophyletic peripheral population can occur without a dichotomous split with the remaining populations of a taxon which are rendered paraphyletic [58, 59], we support the approach of previous authors to synonymise all subspecies [10, 11]. While the subspecies of E. baueriana were erected using an implied morphological subspecies concept, the overlapping and poorly defined morphological character states used to distinguish the existing subspecies (plant stature, leaf size and colour, bud size and pedicel length, and fruit size) [10, 19] makes identifying the subspecies on any basis other than geography difficult. This lack of morphological differentiation is only reinforced by our findings that there is no greater genetic divergence between the three subspecies in the south-western clade than the level of genetic differentiation amongst populations of E. baueriana subsp. baueriana which occurs in both the south-western and north-eastern clades.

If we wished to maintain an infraspecific classification scheme we could accept that the nominotypical subspecies is not monophyletic and maintain the current classification, which is often considered a valid approach at the infraspecific level [60]. However, if we wish to apply a phylogenetic subspecies concept that emphasises monophyly, further work could be undertaken to determine if there are morphological differences between the Gippsland populations and other populations of E. baueriana subsp. baueriana that form the north-eastern clade in our phylogeny. If consistent morphological differences can be identified, either a new subspecies could be erected to include the Gippsland populations, or these populations could be united with the other two subspecies in the south-western clade. The latter would result in a classification scheme that reflects the two genetic clades we found in our analyses, with a disjunction between them across the central Gippsland Lowlands. A further option not explored here, but given preliminary support by previous findings of seedling leaf morphology differences between populations on the Cumberland Plain and at Bendalong [16], is an infraspecific classification scheme more in line with the result of our K = 8 STRUCTURE analysis with different taxa occurring in each major catchment area.

Phylogeography of E. baueriana

We report the highest Fst value (0.218) for a Eucalyptus species we are aware of to date (0.04 in E. melliodora A.Cunn. ex Schauer [61], 0.017–0.018 in E. albens Benth. and E. sideroxylon A.Cunn. ex Woolls [62], 0.066 in E. behriana F.Muell. [25], E. globulus Labill. sensu lato = 0.08 [63]), and despite wide acceptance of a strong inbreeding depression in the eucalypts [64], we find evidence of high inbreeding in E. baueriana. This indicates that there is stronger geographic structuring of the genetic diversity of the species and more limited gene flow between populations than in previously studied species. This is corroborated by the strong support for relationships between sampling sites in the phylogeny, despite previous studies using similar data failing to resolve interspecific relationships in E. sect. Adnataria [65]. There are multiple potential variables that could lead to this pattern, although there are two main ones that are relevant here and it is likely both have played a role in the evolution of the current pattern of genetic diversity: firstly, that the isolation and restriction of geneflow between populations of the species is comparatively old, and secondly that the ecology of the species and the nature of its populations are playing a role. The species occurs as localised, restricted populations in specific river catchments and, being closely associated with waterways, the populations are relatively linear and constrained in their occupancy of the landscape; both factors likely limit the potential for geneflow between populations.

The largest area of occurrence of E. baueriana on the south coast of NSW is within the region included in the STZ by previous authors [2, 4], and we find the basal divergence of lineages of the species is near this region, corresponding to the disjunction between populations in the east and west of the East Gippsland Lowlands. This finding is unexpected, as this disjunction is across a significantly smaller distance and more similar ecosystems than other disjunctions in the species range, especially that between the Werribee River Catchment populations and the eastern Gippsland Plains/western East Gippsland Lowlands populations.

This Gippsland disjunction corresponds to the area of higher rainfall in the central area of the East Gippsland Lowlands. Given that E. baueriana most commonly occurs in riparian areas in open forest and woodland communities associated with drier sites [10], this may indicate the species is outcompeted by other species in the area that are better suited to the closed rainforest habitat associated with the higher rainfall. This is reinforced by the findings of White et al. [66], who showed that in Victoria, gallery rainforest is most widespread in riparian habitats in the central parts of the East Gippsland Lowlands, roughly corresponding to the disjunction in E. baueriana’s distribution. The results of our SDMs hint at the relationship with rainfall being less about absolute annual rainfall, rather the amount of winter rainfall received as the areas E. baueriana occurs have limited seasonality to summer dominated rainfall, and the two southern disjunctions correspond to areas with more winter dominated rainfall. We also see evidence that the species distribution may be limited by how cold winters get on average.

However, we must make sure to distinguish between an environmental condition maintaining a disjunction and a possible vicariance event. Long distance colonisation is unlikely in eucalypts given their lack of seed dispersal mechanisms [67] and has been assumed to not be a major factor in other eucalypt species with disjunct distributions [25, 68, 69]. This leads to the hypothesis that the populations in the eastern and western East Gippsland Lowlands must have had a more or less continuous area of occurrence at some point in the past and vicariance is responsible for the current disjunction between them.

The most broadscale cause of recent natural vegetation change in south-east Australia were the climatic shifts between glacial and interglacial climates during the Quaternary, with glacial periods likely resulting in lower temperatures and decreased rainfall [70, 71], although some environmental records support there being higher rainfall in at least some parts of south-east Australia under glacial climates [72]. If it is high rainfall maintaining the disjunction in the species distribution, it can be hypothesised lower rainfall under glacial climates may have allowed the species to be more widespread, with continuous population from the Sydney Basin to the Werribee River catchment, or some points in between given we cannot rule out northward or westward expansion of the species since then due to the limited genetic distinction of these peripheral populations.

Our data suggest that isolation of the Werribee River Catchment populations is relatively recent given their close genetic relationship to those on the Deddick River. Previous findings of Fahey et al. [25] showed that the isolation of a population of another eucalypt species, E. behriana, at Long Forest in the Werribee River Catchment was one of the most recent vicariance events across the distribution of that species which otherwise occurs inland of the Great Dividing Range. While E. behriana occurs in very different environments to E. baueriana, together these patterns suggest that there may have been major changes of vegetation patterns in this part of Victoria in the recent past, possibly tied into glacial-interglacial climate fluctuations or volcanic activity [21, 25].

Our data do little to explain the presence of the populations on the Deddick River, other than providing strong evidence they do represent outlying populations of E. baueriana. Being most closely related to the populations in the Werribee River catchment is an unexpected result, given the substantial geographic separation and the presence of intervening populations in the Gippsland Plain and East Gippsland Lowlands, but may fit with the descriptions of these two subspecies sharing some morphological traits [19]. The geographic changes that may have led to this relatedness are less clear. While we see limited genetic divergence in many of the eight samples from the Deddick River in our phylogeny (Fig 2), this largely reflects our restricted sampling. We sampled from the smaller of the two populations known at the time of publication of the subspecies, representing only 14 individuals of the estimated total 200 individuals on the river, as the steep terrain along the Deddick River makes accessing the river banks difficult, likely reflecting why the populations weren’t represented by collections prior to 2005 [19].

In the north-eastern clade, the population structuring appears to predate the isolation of the populations on the Cumberland Plain, at Bendalong, and the far south coast. Both the seedling morphology and genetic data examined by Collins et al. [16] suggested a level of divergence between populations on the Cumberland Plain and at Bendalong, which our data corroborate, with the Bendalong population diverging earlier than the populations on the Cumberland Plain and the far south-coast. This is despite the Bendalong population being the most central, siting nearly halfway between the other two. It seems unlikely that there was geographic connectivity between the Cumberland Plain and the far south coast populations post isolation of the Bendalong population as the major inland barrier between these populations is the higher elevations to the north and west of the Illawarra district, a barrier not likely to have been permeable in the past, and therefore historical connectivity was likely along the coast which would have included the Bendalong area. Due to this, we find ourselves unable to put forward any solid biogeographical hypotheses to explain why the Bendalong population diverges early within the north-eastern clade. It is possible that the cause is not biogeographical, but rather the result of genetic processes such as incomplete lineage sorting between the populations, founder effects or greater genetic drift due to the small population size at Bendalong.

The isolation of these three areas of occurrence (Cumberland Plain, Bendalong and the South Coast) might be maintained by the higher elevation of the Illawarra escarpment and higher rainfall in the intervening areas. There is a narrower coastal plain in the Illawarra district than further south on the NSW coast, limiting suitable habitat for E. baueriana, and the higher rainfall produces more closed forests. While the coastal shelf off eastern Australia is narrow (~20 km to ~45 km within the study region) [70, 73], under the modal sea levels of glacial climates, including the last glacial maximum (LGM), there would have been a slightly wider coastal plain [74]. Under a hypothesised lower rainfall regime [70] this may have benefitted E. baueriana, providing more flat land habitat and leading to more continuous occupancy of the south-east corner of the Australian mainland than at present. This wider, more continuous population could have then undergone vicariance due to the narrowing of the coastal plain and expansion of closed forests as rainfall and sea level increased moving into an interglacial period.


In this study we have shown that there is strong genetic structuring corresponding to the disjunct populations of E. baueriana, with levels of inbreeding that are high for a eucalypt species, potentially playing a role in establishing this structure. The earliest divergence between populations of the species is in the Gippsland Lowlands, forming two genetic groupings corresponding to south-western population and north-eastern populations. These two main genetic groups do not correspond to the three described subspecies of E. baueriana, as the nominotypical subspecies is represented in both groups while the two geographically restricted subspecies forming part of the south-western clade. Additionally, we have shown the Blue Box group, consisting of E. baueriana, E. conica, E. magnificata and E. dalveenica, represents a monophyletic grouping with E. ser. Heterophloiae most closely related to E. polyanthemos.

Supporting information

S1 Fig. Maximum likelihood phylogeny excluding E. baueriana sample that shows evidence for introgression from E. conica rooted on two samples from E. sect. Bisectae.

Support values shown are ML and MP bootstrap values and branches with bootstrap values above 80 for both analyses are thickened. Members of E. ser. Heterophloiae are indicated by tip bubbles coloured to correspond to Fig 1A and tip labels of E. baueriana are coloured by their collecting site of origin per Fig 1B.



We are grateful to Harvey Orel (The University of Melbourne) for assistance during fieldwork, and Rachel Fowler (The University of Melbourne) and Todd McLay (The University of Melbourne and National Herbarium of Victoria) for insights into data use and analysis.


  1. 1. Beck HE, Zimmermann NE, McVicar TR, Vergopolan N, Berg A, Wood EF. Present and future Köppen-Geiger climate classification maps at 1-km resolution. Sci Data. 2018;5:180214. pmid:30375988
  2. 2. Bryant L, Krosch M. Lines in the land: a review of evidence for eastern Australia’s major biogeographical barriers to closed forest taxa. Biol J Linnean Soc. 2016;119(2):238–64.
  3. 3. Edwards RD, Crisp MD, Cook DH, Cook LG. Congruent biogeographical disjunctions at a continent-wide scale: Quantifying and clarifying the role of biogeographic barriers in the Australian tropics. PLOS ONE. 2017;12(4):e0174812. pmid:28376094
  4. 4. Milner ML, Rossetto M, Crisp MD, Weston PH. The impact of multiple biogeographic barriers and hybridization on species‐level differentiation. American Journal of Botany. 2012;99(12):2045–57. pmid:23221499
  5. 5. Chapple D, Hoskin C, Chapple S, Thompson M. Phylogeographic divergence in the widespread delicate skink (Lampropholis delicata) corresponds to dry habitat barrier in eastern Australia. BMC Evolutionary Biology. 2011;11:191. pmid:21726459
  6. 6. McLay TGB, Ladiges PY, Doyle SR, Bayly MJ. Phylogeographic patterns of the Australian grass trees (Xanthorrhoea Asphodelaceae) shown using targeted amplicon sequencing. Australian Systematic Botany. 2021;34(2):206–25.
  7. 7. French PA, Brown GK, Bayly MJ. Incongruent patterns of nuclear and chloroplast variation in Correa (Rutaceae): introgression and biogeography in south-eastern Australia. Plant Systematics and Evolution. 2016;302(4):447–68.
  8. 8. Worth JRP, Sakaguchi S, Harrison PA, Brüniche-Olsen A, Janes JK, Crisp MD, et al. Pleistocene divergence of two disjunct conifers in the eastern Australian temperate zone. Biol J Linnean Soc. 2018;125(3):459–74.
  9. 9. Larcombe MJ, McKinnon GE, Vaillancourt RE. Genetic evidence for the origins of range disjunctions in the Australian dry sclerophyll plant Hardenbergia violacea. J Biogeogr. 2011;38(1):125–36. PubMed PMID: edsjsr.40996137.
  10. 10. Brooker MIH, Slee AV, Connors JR, Duffy S, West J. EUCLID: Eucalypts of Australia. Canberra: CSIRO; 2015.
  11. 11. Nicolle D. Classification of the eucalypts (Angophora, Corymbia and Eucalyptus) Version 4 2019. Available from:
  12. 12. Department of Agriculture, Water and the Environment. Australia’s bioregions (IBRA) 2017. Available from:
  13. 13. Department of Agriculture, Water and the Environment. Interim Biogeographic Regionalisation for Australia (Subregions—States and Territories) v. 7 (IBRA). 2020.
  14. 14. Government of the Commonwealth of Australia—Australian Bureau of Statistics. ABS—ASGS—State (STE) 2016. 2016.
  15. 15. Maiden JH. On Eucalyptus baueriana Schauer. Proceedings of the Linnean Society of New South Wales. 1902;27(2):216–9.
  16. 16. Collins TL, Andrew RL, Bruhl JJ. Morphological, phytochemical and molecular analyses define species limits in Eucalyptus magnificata (Myrtaceae) and lead to the discovery of a new rare species. Australian Systematic Botany. 2019;32(1):12–28.
  17. 17. Johnson LA, Hill K. New taxa and combinations in Eucalyptus and Angophora (Myrtaceae). Telopea. 1990;4(1):37–108.
  18. 18. Chippendale GM. Flora of Australia: Myrtaceae—Eucalyptus, Angophora. Canberra: Australian Government Publishing Service; 1988.
  19. 19. Rule K. Six new infraspecific taxa in Eucalyptus (Myrtaceae) for Victoria. Muelleria. 2011;29(1):3–15.
  20. 20. Australian Plant Census, IBIS database [Internet]. Centre for Australian National Biodiversity Research, Australian Government. 2016. Available from:
  21. 21. Myers BA, Ashton DH, Osborne JA. The ecology of the mallee outlier of Eucalyptus behriana F.Muell near Melton, Victoria. Aust J Bot. 1986;34(1):15–39. WOS:A1986A137200002.
  22. 22. Cruz VMV, Kilian A, Dierig DA. Development of DArT Marker Platforms and Genetic Diversity Assessment of the U.S. Collection of the New Oilseed Crop Lesquerella and Related Species. PLOS ONE. 2013;8(5):e64062. pmid:23724020
  23. 23. Sansaloni C, Petroli C, Jaccoud D, Carling J, Detering F, Grattapaglia D, et al. Diversity Arrays Technology (DArT) and next-generation sequencing combined: genome-wide, high throughput, highly informative genotyping for molecular breeding of Eucalyptus. BMC Proceedings. 2011;5(7):P54.
  24. 24. McLay TG. High quality DNA extraction protocol from recalcitrant plant tissues 2017. Available from:
  25. 25. Fahey PS, Fowler RM, McLay TG, Udovicic F, Cantrill DJ, Bayly MJ. Divergent lineages in a semi-arid mallee species, Eucalyptus behriana, correspond to a major geographic break in south-eastern Australia. Ecol Evol. 2021;11(1):664–78. pmid:33437459
  26. 26. Fahey PS, Udovicic F, Cantrill DJ, Bayly MJ. Data for: A box on the river: The phylogenetics and phylogeography of Eucalyptus baueriana (Eucalyptus sect. Adnataria ser. Heterophloiae). Dryad. 2021.
  27. 27. Eaton DAR, Overcast I. ipyrad: Interactive assembly and analysis of RADseq datasets. Bioinformatics. 2020;36(8):2592–4. pmid:31904816
  28. 28. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnetjournal. 2011;17:10.
  29. 29. Myburg AA, Grattapaglia D, Tuskan GA, Hellsten U, Hayes RD, Grimwood J, et al. The genome of Eucalyptus grandis. Nature. 2014;510(7505):356–62. pmid:24919147
  30. 30. Alwadani KG, Janes JK, Andrew RL. Chloroplast genome analysis of box-ironbark Eucalyptus. Molecular Phylogenetics and Evolution. 2019;136:76–86. pmid:30954587
  31. 31. Steane D, Byrne M, Vaillancourt RE, Potts BM. Chloroplast DNA polymorphism signals complex interspecific interactions in Eucalyptus (Myrtaceae). Australian Systematic Botany. 1998;11(1):25–40.
  32. 32. Stamatakis A. RAxML Version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics. 2014;30(9):1312–3. pmid:24451623
  33. 33. Swofford D. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4.0b102002.
  34. 34. Stöver BC, Müller KF. TreeGraph 2: Combining and visualizing evidence from different phylogenetic analyses. BMC Bioinformatics. 2010;11(1):7. pmid:20051126
  35. 35. Durand EY, Patterson N, Reich D, Slatkin M. Testing for ancient admixture between closely related populations. Molecular biology and evolution. 2011;28(8):2239–52. pmid:21325092
  36. 36. Pritchard JK, Stephens M, Donnelly P. Inference of Population Structure Using Multilocus Genotype Data. Genetics. 2000;155(2):945–59. pmid:10835412
  37. 37. Novembre J. Pritchard, Stephens, and Donnelly on Population Structure. Genetics. 2016;204(2):391–3. pmid:27729489
  38. 38. R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2016.
  39. 39. Gruber B, Unmack PJ, Berry OF, Georges A. DARTR: An R package to facilitate analysis of SNP data generated from reduced representation genome sequencing. Mol Ecol Resour. 2018;18:691–9. pmid:29266847
  40. 40. Puechmaille SJ. The program structure does not reliably recover the correct population structure when sampling is uneven: subsampling and new estimators alleviate the problem. Mol Ecol Resour. 2016;16(3):608–27. pmid:26856252
  41. 41. Chhatre VE, Emerson KJ. StrAuto: automation and parallelization of STRUCTURE analysis. BMC Bioinformatics. 2017;18(1):192. pmid:28340552
  42. 42. Jakobsson M, Rosenberg NA. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics. 2007;23(14):1801–6. pmid:17485429
  43. 43. Ramasamy RK, Ramasamy S, Bindroo BB, Naik VG. STRUCTURE PLOT: a program for drawing elegant STRUCTURE bar plots in user friendly interface. SpringerPlus. 2014;3(1):431. pmid:25152854
  44. 44. Phillips SJ, Anderson RP, Schapire RE. Maximum entropy modeling of species geographic distributions. Ecol Model. 2006;190(3–4):231–59. WOS:000233859600001.
  45. 45. Fick SE, Hijmans RJ. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology. 2017;37(12):4302–15.
  46. 46. Viscarra Rossel RA, Chen C, Grundy MJ, Searle R, Clifford D, Campbell PH. The Australian three-dimensional soil grid: Australia’s contribution to the GlobalSoilMap project. Soil Res. 2015;53(8):845–64.
  47. 47. Gallant J, Wilson N, Dowling T, Read A, Inskeep C. SRTM-derived 1 Second Digital Elevation Models Version 1.0. Record 1. In: Geoscience Australia, editor. Canberra2011.
  48. 48. Atlas of Living Australia. Atlas of Living Australia 2019 [9 October 2019]. Available from:
  49. 49. Jueterbock A, Smolina I, Coyer JA, Hoarau G. The fate of the Arctic seaweed Fucus distichus under climate change: an ecological niche modeling approach. Ecol Evol. 2016;6(6):1712–24. pmid:27087933
  50. 50. Zheng Y, Janke A. Gene flow analysis method, the D-statistic, is robust in a wide parameter space. BMC bioinformatics. 2018;19(1):1–19. pmid:29310567
  51. 51. Janes JK, Miller JM, Dupuis JR, Malenfant RM, Gorrell JC, Cullingham CI, et al. The K = 2 conundrum. Molecular Ecology. 2017;26(14):3594–602. pmid:28544181
  52. 52. Brooker MIH. A new classification of the genus Eucalyptus L’Hér. (Myrtaceae). Australian Systematic Botany. 2000;13(1):79–148.
  53. 53. Pryor LD, Johnson LAS. A classification of the eucalypts. Canberra: The Australian National University; 1971.
  54. 54. Rossetto M, Allen CB, Thurlby KAG, Weston PH, Milner ML. Genetic structure and bio-climatic modeling support allopatric over parapatric speciation along a latitudinal gradient. BMC Evolutionary Biology. 2012;12:149. pmid:22906180
  55. 55. Kirkpatrick J, Simmons D, Parsons R. The relationship of some populations involving Eucalyptus cypellocarpa and E. globulus to the problem of phantom hybrids. New Phytologist. 1973;72(4):867–76.
  56. 56. Parsons R, Kirkpatrick J. Possible phantom hybrids in Eucalyptus. New Phytologist. 1972;71(6):1213–9.
  57. 57. Larcombe MJ, Vaillancourt RE, Jones RC, Potts BM. Assessing a Bayesian Approach for Detecting Exotic Hybrids between Plantation and Native Eucalypts. International Journal of Forestry Research. 2014;2014:650202.
  58. 58. Carnicero P, Schönswetter P, Garcia-Jacas N, Galbany-Casals M. Is there a need for accepting paraphyletic taxa? A case study in the Sardinian endemic Cymbalaria muelleri (Plantaginaceae). Botanical Journal of the Linnean Society. 2019;191(3):325–38.
  59. 59. Anacker BL, Strauss SY. The geography and ecology of plant speciation: range overlap and niche divergence in sister species. Proceedings of the Royal Society B: Biological Sciences. 2014;281(1778):20132980. pmid:24452025
  60. 60. Patten MA. Subspecies and the philosophy of science. The Auk. 2015;132(2):481–5.
  61. 61. Supple MA, Bragg JG, Broadhurst LM, Nicotra AB, Byrne M, Andrew RL, et al. Landscape genomic prediction for restoration of a Eucalyptus foundation species under climate change. eLife. 2018;7:e31835. pmid:29685183
  62. 62. Murray KD, Janes JK, Jones A, Bothwell HM, Andrew RL, Borevitz JO. Landscape drivers of genomic diversity and divergence in woodland Eucalyptus. Molecular Ecology. 2019;28(24):5232–47. pmid:31647597
  63. 63. Jones RC, Steane DA, Potts BM, Vaillancourt RE. Microsatellite and morphological analysis of Eucalyptus globulus populations. Canadian Journal of Forest Research. 2002;32(1):59–66.
  64. 64. Nickolas H, Harrison PA, Tilyard P, Vaillancourt RE, Potts BM. Inbreeding depression and differential maladaptation shape the fitness trajectory of two co-occurring Eucalyptus species. Annals of Forest Science. 2019;76(1):10.
  65. 65. Woodhams M, Steane DA, Jones RC, Nicolle D, Moulton V, Holland BR. Novel Distances for Dollo Data. Systematic Biology. 2013;62(1):62–77. pmid:22914977
  66. 66. White MD, Batpurev K, Salkin O, Newell G. Primary Rainforest Mapping in Victoria 2018-Extent and Type. Heidelberg, Victoria: Arthur Rylah Institute for Environmental Research, Department of Environment, Land, Water and Planning; 2019.
  67. 67. Booth TH. Going nowhere fast: a review of seed dispersal in eucalypts. Aust J Bot. 2017;65(5):401–10.
  68. 68. Marginson JC, Ladiges PY. Geographical variation in Eucalyptus baxteri s.l. and the recognition of a new species, E. arenacea. Australian Systematic Botany. 1988;1(2):151–70. edsagr.US201302707819.
  69. 69. Byrne M, Hopper SD. Granite outcrops as ancient islands in old landscapes: evidence from the phylogeography and population genetics of Eucalyptus caesia (Myrtaceae) in Western Australia. Biol J Linnean Soc. 2008;93(1):177–88. WOS:000251738300017.
  70. 70. Williams M, Cook E, van der Kaars S, Barrows T, Shulmeister J, Kershaw P. Glacial and deglacial climatic patterns in Australia and surrounding regions from 35 000 to 10 000 years ago reconstructed from terrestrial and near-shore proxy data. Quaternary Science Reviews. 2009;28(23):2398–419.
  71. 71. Mills K, Gell P, Hesse PP, Jones R, Kershaw P, Drysdale R, et al. Paleoclimate studies and natural-resource management in the Murray-Darling Basin I: past, present and future climates. Australian Journal of Earth Sciences. 2013;60(5):547–60.
  72. 72. Ellerton D, Shulmeister J, Woodward C, Moss P. Last Glacial Maximum and Last Glacial-Interglacial Transition pollen record from northern NSW, Australia: evidence for a humid late Last Glacial Maximum and dry deglaciation in parts of eastern Australia. Journal of Quaternary Science. 2017;32(6):717–28.
  73. 73. Heap AD, Harris PT. Geomorphology of the Australian margin and adjacent seafloor. Australian Journal of Earth Sciences. 2008;55(4):555–85.
  74. 74. Ribó M, Goodwin ID, O’Brien P, Mortlock T. Shelf sand supply determined by glacial-age sea-level modes, submerged coastlines and wave climate. Scientific reports. 2020;10(1):462–. pmid:31949172