Sources of material - herbarium and field

A total of 86 samples of Phalaris arundinacea were obtained from herbarium material and additional targeted sampling carried out for this study (Tables 1, 2, Suppl. materials 1, 3). Herbarium samples, from modern to 130 years old and in relatively good condition, were selected for sampling from the University of British Columbia Herbarium (UBC) and the Herbarium of the Bell Museum, University of Minnesota (MIN). Further dried leaf samples (used in a previous study; Kettenring et al. (2019)) were kindly provided by Professor Karen Mock of Utah State University. In addition, extensive field sampling was carried out in Elk Island National Park, Alberta, where park authorities were concerned about the ecologically harmful spread of, as well as appropriate control methods for, Phalaris arundinacea. Further recent samples were sourced from Greater Vancouver. Voucher specimens are deposited in UBC. As outgroups for the phylogenetic analyses, we used eight individuals obtained from herbarium samples of P. aquatica Guss., P. canariensis L., P. caroliniana Walter, P. coerulescens Desf. and P. paradoxa L. (Suppl. material 2).

DNA extraction, PCR and sequencing

Dried leaf material was ground to a slurry in liquid nitrogen and the DNA extracted using a modified CTAB method (Doyle and Doyle 1987). Full length PCR (ITS1-5.8S-ITS2) was performed using primers ITS-A (forward) (Blattner 1999) and ITS4 (reverse) (White et al. 1990) and PCR conditions 94°C/4 min, followed by 30 cycles of 94°C/30 sec, 50°C/1 min, 72°C/1 min and final extension of 72°C/10min. In cases of highly degraded DNA from older herbarium specimens, ITS1 and ITS2 were amplified separately using primers ITS3P (forward) (Möller and Cronk 1997) and the reverse complement ITS2P (reverse). Bidirectional Sanger sequencing was performed by Eurofins (Kentucky, USA) and sequences were checked using Sequencher version 4.8 (Gene Codes).

Sequence alignment and phylogenetic analysis

Sequences of 60 individuals were aligned manually using Sequencher and Se-Al (Rambaut 2002). Subunit boundaries follow those determined for Oryza (Takaiwa et al. 1985, Yokota et al. 1989) as follows: 18S/ITS1 CATTG/TCGTG; ITS1/5.8S AAATC/CACAC; 5.8S/ITS2 CACGC/CAAAA; ITS2/26S GGACC/GCGAC (an example of a full Oryza sequence for location is GenBank accession MF029734). Eight putative hybrids (between the different ribotypes) were excluded from the phylogenetic analysis due to sequence superposition. We included one sample from GenBank (KF753778) as the only previously databased sequence with the ITS-short genotype. Phylogenetic analysis was performed using three approaches: a Neighbour-joining (NJ) analysis with uncorrected (p) distances and 1000 bootstrap replicates, a Maximum Parsimony (MP) analysis with heuristic search (random addition of taxa and TBR branch swapping), both methods being performed in PAUP* (Swofford 2003); and a Maximum Likelihood (ML) analysis using RAxML (v. 8.2.4) with GTRCAT, 1000 rapid bootstraps and Gamma optimisation of tree space run on the CIPRES Science Gateway (Miller et al. 2010, Stamatakis 2014). The MP analysis also included a gap code matrix (for nine gaps: three in P. arundinacea and six in outgroup taxa). Sequences are deposited in GenBank under accession numbers: OQ740187-OQ740255.

Structural analysis of ITS2

Structural analyses were performed using the ITS2 database (Ankenbrand et al. 2015). We used the Phalaris arundinacea ITS2 structure of GenBank accessions FJ821785 (MFE -66.8 kcal/mol) in the ITS2 database for homology modelling (Wolf et al. 2005) of our common variant (ITS-long) as it had a near identical sequence. As homology modelling of the rare variant (ITS-short) fails on FJ821785, alternative templates for homology modelling were investigated. Plausible configurations for ITS2-short were obtained using Arctagrostis latifolia (EU792351) and Phalaris canariensis (FJ377670) as templates.

Capillary sizing assay

A primer was designed using the NCBI Primer-BLAST tool (Ye et al. 2012) ITS2AindelR: 5’-GCAGCCATATCTTCGGC-3’ for use in conjunction with ITS primer ITS3P to allow an accurate sizing assay on an ABI 3730 automated DNA Sequencer (Applied Biosystems). The primer was combined with a M13 tail (5'-TGTAAAACGACGGCCAGT-3') on the forward primer to facilitate fluorescent dye labelling and a further PIG tail (5’-GTTTCTT-3’) on the reverse primer to promote terminal adenylation. We used a hot start touchdown PCR protocol with 95°C/3 min, followed by 10 cycles of 94°C/30 sec, 65°C/30 sec (-1°C per cycle, R 3°C/sec), 72°C/45 sec, followed by a further 30 cycles of 94°C/30 sec, 55°C/30 sec, 72°C/45 sec and a final extension at 72°C/4 min. PCR products were loaded into the capillary machine at 1:30 dilution and traces read using the programme Geneious 8.1.9 (Biomatters Ltd.). The PCR assay was designed to give products of 238 or 245 bp depending on the presence of the 7 bp deletion.

A sequencing survey and phylogenetic analysis reveals intraspecific divergence in ITS including a 7 bp deletion

Initial results of an ITS sequencing survey of Phalaris arundinacea from western Canada revealed two distinctive sequences. One is full length with fixed heterozygosity characteristic of polyploids; the other is shorter, with a 7 bp deletion in ITS2 and with no fixed heterozygous base positions. The differences are summarised in Table 3. The tree topologies recovered from the different phylogenetic approaches were nearly identical. The matrix length was 603 bp (612 characters with gap coding) and the MP search recovered two trees with length 117 (CI: 0.93, RI: 0.98); we present the strict consensus topology in Fig. 2 showing majority rule consensus values as well as NJ and ML bootstrap support values. The best ML model fit for the data (AIC) was GTR+G (-lnL 1611.45). Use of outgroups showed that the full length sequence (which we call ITS-long) was likely the ancestral one and the deletion (ITS-short) is a putatively-derived character so far known only from plants in north-western North America (Fig. 3). When compared with all available world-wide sequences from GenBank (including Asia, Europe, North and South America), only one sequence was found to have the ITS-short genotype (KF753778) from Cook Inlet, Alaska; all other GenBank samples are the ITS-long genotype and ITS-long sequences found in North America are highly similar or identical to European genotypes. Tables 1, 2 show to which clade (ITS-long/-short) historical herbarium specimens can be assigned.

The 7 bp deletion alters the secondary structure of helix I of ITS2. The predicted secondary structure of the common variant (ITS2-long) of Phalaris arundinacea ITS2 is the usual eukaryotic four helix model (Fig. 4). Homology modelling of the structure of the ITS2-short sequence against this structure fails, as helix I, which has the 7 bp deletion, is not a suitable model. However, homology modelling with a related grass of similar ITS2 sequence suggests a plausible model for helix I despite the deletion (Fig. 4).

A PCR-based capillary sizing assay allows rapid detection of the 7 bp deletion clade

In order to genotype individuals without sequencing, we developed a primer that amplifies a 238 vs. 245 bp amplicon (short enough to size accurately to a single base pair on a capillary machine). ITS-long gave a clear peak at 245 bp and a complete absence of a peak at 238 bp. Despite using a design to promote terminal adenylation (see Methods), if there is a large amount of starting DNA, this peak was split, showing a peak or shoulder at 244 bp. However, in all cases, the fully adenylated peak was unambiguous and as strong or stronger than the unadenylated peak. ITS-short samples gave a strong, unambiguous peak at 238 bp. Product without terminal adenylation sometimes showed as a shoulder, but never a separate peak. ITS-short samples sometimes showed a small peak at 245 bp, but the 238 peak was, in all cases, much stronger. A total of 68 individuals were sized with this method, providing clade (ITS-long/-short/hybrid) affiliation for an additional 34 individuals. Putative hybrids (10 individuals) were identified either by both sequencing and sizing assay (seven individuals), sequence data only (one individual) and two specimens identified as ITS-short in the length assay, but were determined as a putative hybrid with sequence data (Suppl. material 1).

A survey of Elk Island National Park, Alberta, reveals presence of both ITS ribotypes

Using the molecular tools detailed above, we were able to conduct extensive sampling of Elk Island National Park (EINP), Alberta. Phalaris arundinacea is extremely abundant at EINP and the material in the Park tends to be strongly spreading-rhizomatous and invasive. EINP is bisected into a northern and southern portion by the east-west highway 16. These portions have different management histories, with the northern portion experiencing much greater public access and road development. We refer to these portions as north EINP and south EINP. In all sampled localities of north EINP, ITS-long was the only genotype detected (except a few possible hybrids at Tawayik Lake). In south EINP the situation is very different. Of the 12 individuals genotyped from south EINP, five were ITS-short (DPQC10A and DPQC11A-D).

North-western North America

Phalaris arundinacea, P. aquatica, P. canariensis, P. caroliniana, P. coerulescens and P. paradoxa.

Specimen details for all 94 samples genotyped (86 Phalaris arundinacea and eight outgroup taxa sampled).

1