Origins of the rare Australian daisy Erigeron conyzoides and its implications for biological control research and conservation management

Field work

Field work was conducted in January 2024 in Kosciuszko National Park, New South Wales, and Alpine National Park, Victoria, in Australia. After discovering an extant population of Erigeron conyzoides at Falls Creek, Victoria (36.8746°S, 147.3001°E), young leaf tissue was collected from 15 individuals spatially distributed across the population and dried on silica gel.

Herbarium specimens and floras

To explore the possibility that Erigeron conyzoides is in fact an introduced species in Australia, we first attempted to identify specimens held at the Australian National Herbarium with keys published in the following floras from the northern hemisphere: Flora Europaea (Tutin 1976), the Flora of China (Chen and Brouillet 2011), the Flora of North America (Nesom 2006), and the Flora of the USSR (Schischkin 1999). In addition, we consulted a taxonomic treatment of South American Erigeron (Solbrig 1962).

To visualise historical collections, records of preserved specimens of Asteraceae in circles of 5 km radius around 16 historical localities of E. conyzoides in Australia were queried from the Atlas of Living Australia (https://ala.org.au, accessed 29 May 2024) by using the R package galah (v.2.0.2; Westgate et al. 2024). To allow automatic removal of duplicates with R (v.4.4.0; R Core Team 2024), collector names and localities were harmonised with OpenRefine (v.3.8.1; Ham 2013) by using the key collision method and both fingerprint and n-Gram fingerprint keying functions. Duplicates of collector name and collection number were removed except for ‘s.n.’ (specimens without collection numbers), in which case duplicates of collector name and exact locality description were removed. The data were binned into 30-year intervals and plotted with R. We also mapped specimens of Erigeron conyzoides in Australia, by using records from the Australasian Virtual Herbarium (https://avh.ala.org.au, accessed 14 January 2024).

DNA extraction and library preparation

We used existing data from our recently published phylogeny of Asteraceae tribe Astereae in Australia (Chen et al. 2024) and the Plant and Fungal Tree of Life (PAFTOL), supplementing these with newly generated sequences of Erigeron conyzoides and the morphologically most similar northern hemisphere species, E. acer (Supplementary Table S1).

We followed DNA-extraction and library-preparation methods described in Chen et al. (2024). Briefly, genomic DNA was extracted from 5–15 mg of silica dried leaf tissue or herbarium material by using the QIAGEN DNeasy Plant 96 Kit. Libraries were built using the QIAGEN Ultralow Input Library Kit, by using a protocol adapted for automation by using an Echo 525 Acoustic Liquid Handler (Beckman-Coulter). Sequence capture was conducted on pools of 16 libraries by using the myBaits Expert Angiosperms353 (v.1; Johnson et al. 2019) kit (Daicel Arbor Biosciences), following the manufacturer’s standard protocol for plants (v5 September 2020). Enriched libraries were sequenced on Illumina NovaSeq 6000 SP with v1.5 paired-end 2 × 150 cycle chemistry with 8-bp dual indexing.

Demultiplexed reads were quality filtered and paired using Trimmomatic (v.0.38; Bolger et al. 2014), with illuminaclip:adapters, fa:4:20:10, minimum length of 30, and average quality of 25. Sequences were further filtered with BBDuk (Bushnell 2022) with entropy of 0.8, entropy window of 20, and low-entropy parts of sequences were masked. The target capture data were assembled with an Asteraceae-specific target file (McLay et al. 2021) by using HybPiper (Johnson et al. 2016), implemented through the Nextflow pipeline hybpiper-nf (Jackson et al. 2023).

ITS phylogeny from GenBank Sanger data

To complement our genomic-scale capture dataset, which included a limited number of samples per species, we inferred an Erigeron phylogeny by using nuclear ribosomal internal transcribed spacer (ITS) sequences, the most widely used Sanger-sequence marker. We queried all ITS sequences of Erigeron and its synonym Conyza from GenBank (accessed 24 January 2024), aligned them by using MAFFT (v.7.453; Katoh and Standley 2013), and excluded a small number of poor-quality sequences after visual inspection of the alignment. We assembled filtered and paired reads of seven samples of E. conyzoides and two of E. acer from target capture data by using hybpiper-nf and an ITS sequence of E. acer (Genbank accession OK447875) as the reference. We obtained six high-quality ITS sequences for E. conyzoides and one for E. acer. It is likely that the target capture reaction was too efficient for the other samples, leaving insufficient off-target reads for ITS assembly. To test whether the use of E. acer as the reference introduced reference bias, we repeated the assembly by using distantly related E. varius Webb. (KR059038). The resulting sequences were identical, indicating no reference bias.

The newly assembled ITS sequences were added to the alignment of sequences from GenBank by using MAFFT (v.7.453). The resulting alignment had 867 sequences with 984 characters, of which 363 were parsimony informative, 162 variable but uninformative, and 459 constant. Phylogenetic analysis was conducted with IQ-TREE (v.1.6.12; Nguyen et al. 2015) without partitioning, under automatic model testing, and by using the approximate likelihood-ratio test (aLRT) (Anisimova and Gascuel 2006) as support values with 1000 replicates. SYM+R4 was chosen as the preferred model. The phylogeny was outgroup rooted on E. rhizomatus Cronquist (AF046992).

Target-sequence capture (Angiosperm353) phylogeny

We took three different approaches to generate the phylogeny. First, two-step coalescent analysis was performed directly from the paralog sequences retrieved by hybpiper-nf. Sequences were aligned with MAFFT (v.7.490; Katoh and Standley 2013) and gene trees were generated with IQ-TREE 2 (v.2.2.0.5; Nguyen et al. 2015; Minh et al. 2020). The best substitution model was estimated with ModelFinder (Kalyaanamoorthy et al. 2017) and support values were estimated with SH-aLRT (Guindon et al. 2010) and 10,000 ultrafast bootstrap replicates (Hoang et al. 2018). The gene trees were combined into a species tree with ASTRAL-Pro (v.1.15.1; Zhang et al. 2020), which allows for multi-copy genes and therefore deals with paralogs. Branch supports were measured as local posterior probabilities, and branch lengths for non-terminal branches were measured in coalescent units.

Second, the paralog sequences retrieved by hybpiper-nf were processed with paragone-nf (v.1.0.1; Jackson et al. 2023) to resolve paralogy (Yang and Smith 2014). The monophyletic-outgroups (MO) algorithm was used in favour of both the rooted-subtrees (RT) and maximum-inclusion (MI) algorithms, as MO produces a more complete matrix. The multiple gene alignments were cleaned and checked to be in-frame by using a custom Python script (https://bitbucket.csiro.au/projects/NRCA/repos/bioinformatics-and-phylogenetics/browse/check_odd_indels_and_frames2.py). For the concatenated analysis, which produced the main results discussed in this paper, concatenation was performed with a custom Python script (https://bitbucket.csiro.au/projects/NRCA/repos/bioinformatics-and-phylogenetics/browse/concat7.py), with a minimum of 20 genes and five taxa per alignment. One sample (Erigeron_conyzoides_SL2138H) was removed because it had fewer than 20 genes, whereas four genes (6064_2, 6448_1, 6705_1, and 6958_1) were removed. With an input of 33 taxa and 260,349 total sites (15.5% missing data), a phylogeny was constructed with IQ-TREE 2 (v.2.2.0.5; Nguyen et al. 2015; Minh et al. 2020) with partition models (Chernomor et al. 2016). There were 8004 parsimony-informative sites, 11,199 singleton sites, and 161,151 constant sites. The data were partitioned by gene, with a total of 332 partitions. The best substitution model for each partition was estimated with ModelFinder (Kalyaanamoorthy et al. 2017), and support was estimated with SH-aLRT (Guindon et al. 2010) and 10,000 ultrafast bootstrap replicates (Hoang et al. 2018).

Third, for other two-step coalescent analyses after resolution of paralogy, the MO outputs were aligned with MAFFT (v.7.490: Katoh and Standley 2013) and gene trees were constructed with IQ-TREE 2 (v.2.2.0.5; Nguyen et al. 2015; Minh et al. 2020) by using ModelFinder (Kalyaanamoorthy et al. 2017), SH-aLRT (Guindon et al. 2010), and 10,000 ultrafast bootstraps (Hoang et al. 2018) with partition models (Chernomor et al. 2016). The gene trees were combined into a species tree with ASTRAL (v.5.7.8; Zhang et al. 2018).

Variant calling and population genomics

The trimmed and quality- filtered target-sequence capture reads for Erigeron species were aligned to the nine chromosomes of the Erigeron canadensis reference genome (Laforest et al. 2020) by using BWA v.0.7.17 (Li 2013), and variants were called with freebayes (v.1.3.7; Garrison and Marth 2012).

The variant call format (VCF) file was filtered using vcftools (v.0.1.16; Danecek et al. 2011) and vcflib (v.1.0.1; Garrison et al. 2022). Low-quality sites were removed (QUAL/DP > 0.25 and --minQ 20 --min-meanDP 4 --max-meanDP 40 --mac 2 --max-missing-count 32 –remove-indels). Filtering was then performed for allelic balance (AB > 0.25 and AB < 0.75| AB < 0.01) and discrepancy in mapping quality between reference and alternative alleles (MQM/MQMR < 2 and MQMR/MQM < 2). The adegenet package (Jombart 2008) was used to perform principal-component analysis (PCA) in (R v.4.0.4; R Core Team 2024).

Field observations

During field surveys by the New South Wales National Parks and Wildlife Service (NPWS) and its contractors in 2019, Erigeron conyzoides was recorded from several sites in Kosciuszko National Park (bionet.nsw.gov.au, accessed 22 December 2023). We visited one site north of Thredbo, just east of Kosciuszko Walk. At and around the exact coordinates recorded, we were unable to find the species. The local habitat was predominantly alpine heath. No species were present that could plausibly have been misidentified as E. conyzoides, except perhaps two species of Pappochroma Raf. (formerly classified as part of Erigeron), with very different inflorescence structure.

There has been a historical decline in collections of Erigeron conyzoides (Fig. 1). The most recent herbarium collection of Erigeron conyzoides was made in 2011 (P.A. French s.n., MELUD106780a, MELUD106792a), along a road directly south of Falls Creek, Victoria. The population was described as ‘growing in gravel up to 60 cm [tall]. Five specimens growing closely together but not occurring anywhere else in surroundings.’. We were unable to find this population. Given its small size and the time since it was last observed, it may be extinct in that location.

The second most recent herbarium collection with precise location data was made in 1998, also near Falls Creek, but at Rocky Valley Dam (N.G. Walsh 4720, MEL 2044671A). This was the only population we were able to find in January 2024. Erigeron conyzoides grew in large numbers on a terraced rock slope of the northern side of the dam in a weed-dominated community (Fig. 2a) with Euchiton japonicus (Thunb.) Holub, Hypochaeris radicata L., and a Verbascum species. It also grew in smaller numbers along the creek under the dam, but not in undisturbed, native vegetation further away from the dam. We did not count individuals, but the population is likely to have onstituted at least a few hundred individuals, including sterile, young rosettes, mid-sized plants producing one flowering stem, and old plants producing multiple flowering stems each year, indicating a stable, reproducing population and a perennial habit.

Fig. 2.

(a) Erigeron conyzoides in Victoria, Australia, Schmidt-Lebuhn 2138 (CANB 997869), photographer Alexander Schmidt-Lebuhn. (b) Erigeron acer subsp. politus in Russia, iNaturalist observation 38471468, reproduced with permission from Igor Pospelov (taimyr).

The largest specimens were ~50 cm tall, smaller than the largest herbarium specimens deposited at CANB, perhaps because of the exposed, often windy location. Stems were reddish-purple, and flowering capitula showed a striking contrast between reddish-purple phyllaries and white ray florets. We revisited the population on 5 February 2024 and found that older ray florets had turned light purple (Fig. 2a).

During the second visit in February, fertilised capitula were bagged. They were harvested by the Victorian Conservation Seedbank on 20 February 2024, resulting in the retrieval of ~17,000 seeds, of which an estimated 75% were filled, as estimated from X-ray imaging of a sample (Rebecca Miller, pers. comm.). Seeds were divided between the Victorian Conservation Seedbank and the National Seed Bank, and some were set aside to raise plants for host-specificity testing in weed biocontrol research.

Morphological affinities of Erigeron conyzoides

In the Flora Europaea (Tutin 1976), Erigeron conyzoides somewhat doubtfully keyed out to Erigeron acer subsp. politus (Fries) H.Lindberg. Couplet 11 in the key to the species of Erigeron and Couplet 2 in the key to subspecies of E. acer are problematic. The first lead of the former requires the plant to be an ‘annual or short-lived perennial’, with basal leaf rosette ‘usually absent at anthesis’, which is not the case for E. conyzoides, but the lead matches its growth form and capitulescence. The latter requires phyllaries to be ‘glabrous or almost so’ to lead to E. acer subsp. politus, whereas E. conyzoides has glandular and/or hispid phyllaries. However, the implied morphology of E. acer subsp. politus in the lead is contradicted by that described in the Flora of China.

In the Flora of China, Erigeron conyzoides keyed out to Erigeron acer subsp. politus. The only difference between the two taxa is that E. acer subsp. politus is described as having pink to lilac rays, whereas those of E. conyzoides are pure-white when young. However, as observed during field work, the rays of E. conyzoides turn purplish-pink as the capitula age, and the two taxa share deep purple stems and phyllaries, sparsely hispid stem indumentum, leaves that are more or less glabrous on the surfaces and ciliate on the margins, and phyllaries short-stipitate glandular to hispid.

In the Flora of the USSR, which applied an extremely narrow (splitting) species concept to Erigeron, E. conyzoides keyed out to E. elongatus Ledeb., a synonym of E. acer subsp. politus. In the Flora of North America, Erigeron conyzoides keyed out to E. acer. No match could be made to South American species in Solbrig’s taxonomic treatment (Solbrig 1962). The morphologically closest species were arguably E. pazensis Sch.Bip. ex Rusby and E. pratensis Phil. However, the former does not have a leaf rosette, and the latter has ray florets that are too few and too long, and no glands on the phyllaries.

Morphological comparison of herbarium specimens of Erigeron conyzoides and E. acer (some of which were collected by co-author A. N. Schmidt-Lebuhn in Germany) confirmed their affinity. Erigeron acer is a highly variable species complex, as reflected in the number of subspecies and segregate species variously recognised by taxonomists. Plants varied in overall size from ~7 to ~70 cm, annual to perennial life cycle, colouration of stem and phyllaries from green across brown to deep purple, and indumentum type and density from subglabrous to glandular and/or hispid. Crucially, however, the morphology of Australian E. conyzoides fell into the variation exhibited by E. acer. Except for the relatively small size of the plants, a specimen of E. acer subsp. politus that was available to us (G. Samuelson s.n., CANB 130739) was most similar to specimens of E. conyzoides in phyllary and stem colouration and indumentum (Fig. 3).

Fig. 3.

(a) Phyllaries of Erigeron conyzoides. (b) Phyllaries of E. acer subsp. politus. (c) Stem of E. conyzoides. (d) Stem of E. acer subsp. politus. A.N. Schmidt-Lebuhn 2138 (CANB 997869) (a, c). G. Samuelson s.n. (CANB 130739) (b, d). Scale bars indicate 1 mm. Note matching indumentum and purple colouration of phyllaries and stem.

Images of living specimens of E. acer subsp. politus present a match both to stem and phyllary colour of E. conyzoides and to its frequent occurrence on rocky substrate (iNaturalist taxon ID 235867 and iNaturalist observations 35498956, 38471468, and 28487382; Fig. 2b). However, they differ from E. conyzoides as observed in the field in having considerably fewer cauline leaves.

ITS phylogeny

ITS sequence data did not provide sufficient resolution to place Erigeron conyzoides confidently either within or outside of E. acer. The samples of E. conyzoides were placed in a clade containing all samples of E. acer but also those of several related species (Supplementary Fig. S1). Within this clade, all except one were placed in the basal polytomy of a subclade comprising most sequences of E. acer, E. altaicus Popov, E. atticus Vill., E. borealis (Vierh.) Simmons, E. elongatus (a synonym of E. acer subsp. politus), most sequences of E. thunbergii A.Gray, and one sequence of E. uniflorus L. The sixth sample of E. conyzoides was placed in a subclade of sequences mostly of E. acer and E. elongatus.

Target-sequence capture phylogeny

Mapped genes from HybPiper averaged 312 of 353 in the bait kit (n = 34, s.d. = 64), with on-target reads ranging from 1.4% to 37.4%, with an average of 15.4% (s.d. = 12.4%) (Table S2). This translates to the percentage of on-target base-pairs recovered ranging from 0.13% (sample excluded) to 80.7%, with an average of 36.3% (s.d. = 25.7%) (Table S2).

The phylogeny of available target-sequence data for species of Erigeron showed E. conyzoides and E. acer forming one clade with maximum support (100 UFBoot and 100 SH-aLRT; Fig. 4a). Within that clade, samples of both species were intermingled without any discernible geographic or population structure, implying that the two taxa cannot be genetically distinguished with these data. The same pattern was shown by coalescent analysis with both ASTRAL-Pro (Fig. S2) and ASTRAL (Fig. S3), despite several samples with <10% on-target base pairs (Erigeron_acer_CANB340701, Erigeron_acris_CANB361138, Erigeron_conyzoides_SL2138B, Erigeron_conyzoides_SL2138G, Erigeron_conyzoides_SL2138H, and Erigeron_conyzoides_SL2138M).

Fig. 4.

(a) Phylogenetic tree based on concatenated analysis (260,349 total sites with 15.5% missing data and 332 partitions) of target-sequence capture data (Angiosperms353) provides evidence that Erigeron conyzoides (purple) is not distinct from E. acer (teal). Branch support values are from 10,000 replicates for SH-aLRT and UFboot. Branch lengths represent estimated nucleotide substitutions per site. (b) Principal-component analysis (PCA; 19,508 single-nucleotide polymorphisms) also demonstrated the lack of distinction between E. conyzoides and E. acer.

Mapping the sequence data for 32 individuals to the E. canadensis reference genome (Laforest et al. 2020) yielded 7,243,408 sites in the raw VCF file, and the mean depth of coverage for each site was 3.98×, and the mean depth for individuals was 8.31× (s.d. = 4.11). Following filtering, 19,508 sites remained. The result from the phylogeny is corroborated by the PCA where E. conyzoides and E. acer are indistinct (Fig. 4b). The first two PCs explained 34.6% and 11.6% of the variance respectively.

We did not undertake further population analyses nor calculate population statistics due to the poor quality of the data. The average missingness of the filtered VCF across 32 samples was 52% (s.d. = 30%). Further, data quality was particularly poor for the ingroup with 79% (s.d. = 13%) missingness for E. acer (n = 6) and 63% (s.d. = 20%) missingness for E. conyzoides (n = 17).

Is Erigeron conyzoides a distinct species?

Several independent lines of evidence suggest that Erigeron conyzoides is conspecific with E. acer in its broader circumscription, i.e. unless adopting a splitting approach to the E. acer complex as in the Flora of the USSR (Schischkin 1999). Although ITS sequences do not provide sufficient phylogenetic resolution and branch support to place E. conyzoides with high confidence (aLRT < 50), its samples were placed in the same clade as those of E. acer and some of its synonyms such as E. elongatus. Target capture data from hundreds of low-copy nuclear genes showed that E. conyzoides formed a clade with E. acer samples from various locations in the northern hemisphere with maximum (100%) branch support, and both species were intermingled in that clade.

The intermingling of geographically dispersed specimens in the E. acer clade suggests that these protein-coding, slowly evolving genes are likely to not provide sufficient phylogenetic support within this species complex to resolve subspecies or geographic provenance, and, indeed, support within the E. acer–E. conyzoides clade is often low for a dataset of tens of thousands of nucleotide characters. Placement at the shallowest taxonomic levels will therefore have to depend on morphological characters as discussed below, or a future combination of broader sampling of E. acer across the northern hemisphere and a more finely resolving genetic approach, for example, genotyping-by-sequencing approaches such as restriction site-associated DNA sequencing (RAD-seq) (Narum et al. 2013). The data obtained from mapping raw capture sequencing to the E. canadensis reference genome (Laforest et al. 2020) and a method specifically devised for Angiosperm353 data (Slimp et al. 2021) were insufficient for understanding genetic diversity and divergence between lineages because of the high missingness, sequencing bias, and the limited number of single-nucleotide polymorphisms (SNPs) produced. For now, our conclusions regarding infraspecific taxonomy are based on morphology.

Erigeron conyzoides shows strong morphological affinity with E. acer. The two species share overall growth form, habit, and inflorescence structure, leaf arrangement, shape, and size, phyllary shape and number of phyllary rows, trimorphic florets, and ray florets narrow and in many rows. Although E. conyzoides keys out to E. acer subsp. politus in older floras, its subspecific placement within the E. acer complex is less clear. Two recent, regional taxonomic revisions, one for Fennoscandia (Olander and Tyler 2017) and one for Murmansk, Russia (Sennikov and Kozhin 2023), circumscribed the subspecies or, in Russia, species, more narrowly, with E. acer subsp. politus being characterised by fewer cauline leaves and a more corymbose capitulescence than for E. conyzoides. However, none of the taxonomic entities recognised in the complex by these two treatments matches the character combination exhibited by E. conyzoides, i.e. numerous cauline leaves, more or less racemose appearance of the capitulescence, phyllaries hairy on their entire length, stem and phyllaries intensely purple, and ray florets pure-white when young (Table 1).

Therefore, we conclude that Erigeron conyzoides is not very likely to be a distinct species endemic to Australia, but a disjunct population of E. acer, and also that its placement at subspecies level must await a comprehensive revision of the E. acer complex across its entire range.

Is Erigeron conyzoides native to Australia?

The lack of morphological and genetic divergence of Australian Erigeron conyzoides from E. acer of the northern hemisphere suggests very recent dispersal to Australia. However, to decide on its native or introduced status, the question arises whether the dispersal occurred during European colonisation of Australia in the past ~200 years, or naturally, for example, via seeds attached to a migratory bird, in the last tens or at most hundreds of thousands of years.

Under the hypothesis of European introduction, the potentially most surprising observation is the presence of Erigeron conyzoides as early as 1854 in the two different locations where Ferdinand Mueller found it according to his publication notes (‘sources of the Murray and Snowy Rivers’) (Mueller 1855). Although this would require earlier introduction and spread than has been documented in many other introduced species (Kloot 1987), it is plausible in the light of how quickly European graziers reached, and thus likely affected, the mountains of Australia.

Graziers brought livestock into the mountain areas by the early 1820s (The Australian Alps National Parks 2013), and colonists visited alpine areas in the 1830s (Costin et al. 2000). A well-referenced study of the history of human impacts in the Monaro Region, which it defines to include the alpine areas of New South Wales, provides more detail (Hancock 1972). ‘Squatting runs’ had been well-established across most of the region by 1848–1850, including in montane areas such as Jindabyne, Adaminaby, and Cooma. Alpine areas were not permanently occupied but primarily used when lower elevations experienced droughts, but as early as 1834 a grazier lost livestock to a surprise blizzard at what is now Kiandra in Kosciuszko National Park. Another grazier established outstations at ‘the headwaters of the snow-fed rivers’ west of the Goodradigbee River from 1839. All of this implies that by the time Mueller found Erigeron conyzoides, the alpine landscape had experienced at least 20 years of grazing impacts and the possibility of introduction of weeds through contaminated equipment or produce.

Most early colonists were British or Irish, and a complicating factor is that subspecies showing ecological and morphological affinities with Erigeron conyzoides, such as E. acer subsp. droebachiensis and E. acer subsp. politus, do not occur in the British Isles. However, they occur in Scandinavia, the European Alps, and Russia, so that immigrants from these areas could be potential vectors of introduction, as could be products imported from them. A precedent for a weed introduction early in British colonisation but from a source region other than the British Empire is Gypsophila tubulosa (Jaub. & Spach) Boiss. (Caryophyllaceae). The species is native to western Turkey, but was first collected in Australia in 1843 and was found by Mueller at Mount Timbertop (Victoria, 1259 masl) by 1853, a year before he encountered E. conyzoides.

It may also be asked why Mueller, a European immigrant to Australia, would not recognise a northern hemisphere species and instead described it as an Australian species. Even though the subspecific identity of the Australian populations remains unresolved, their ecology suggests that it would likely have to be an alpine and/or subarctic subspecies, and Mueller grew up in temperate, lowland areas of northern Germany in what are now the states of Mecklenburg-Vorpommern and Schleswig-Holstein. It is thus plausible that he would not have been familiar with the northern hemisphere source populations of Erigeron conyzoides.

The decline of Erigeron conyzoides in recent decades is another important piece of the puzzle. The species has not been collected in New South Wales since 1978, but has persisted in Victoria (Fig. 1b). This parallels the removal of livestock from the Alps in both states, which happened much earlier in New South Wales. Grazing was removed from the alpine areas of Kosciuszko National Park in 1958 (Costin et al. 2000), but only much later from many parts of the Victorian Alps, such as from the Howitt Plains in 1988 and from the Bogong High Plains in 1991 (The Australian Alps National Parks 2013).

This pattern of decline suggests that in Australia, Erigeron conyzoides may be an opportunistic, disturbance-dependent weed. Further evidence for this proposition is some collectors’ notes on its habitat; the species has often been collected in uncompetitive habitats such as rocky slopes or boulder fields (B.G. Briggs 2547, NSW 569129; G.W. Carr s.n., MEL 0670729A; N.G. Walsh 2430, CANB 721424, HO 562220, MEL 2339440) but also on a weedy road verge (N.G. Walsh 4720, MEL 2044671A) and even a playground (Rogers s.n., MEL 600723). The population we were able to study in the field occupied an engineered structure, a dam, and edges of a creek as part of a community of other weeds, but was absent from undisturbed native vegetation in the surrounding area.

In summary, Erigeron conyzoides appears dependent on low-competition and/or anthropogenic habitats and has been declining in parallel with the recovery of native vegetation since the removal of grazing from alpine environments in Australia. This behaviour is consistent with the hypothesis that it is a European introduction.