Invasion, isolation and evolution shape population genetic structure in Campanula rotundifolia

Abstract The distribution and genetic structure of most plant species in Britain and Ireland bear the imprint of the last ice age. These patterns were largely shaped by random processes during recolonization but, in angiosperms, whole-genome duplication may also have been important. We investigate the distribution of cytotypes of Campanula rotundifolia, considering DNA variation, postglacial colonization, environmental partitioning and reproductive barriers. Cytotypes and genome size variation from across the species’ range were determined by flow cytometry and genetic variation was assessed using cpDNA markers. A common garden study examined growth and flowering phenology of tetraploid, pentaploid and hexaploid cytotypes and simulated a contact zone for investigation of reproductive barriers. Irish populations were entirely hexaploid. In Britain, hexaploids occurred mostly in western coastal populations which were allopatric with tetraploids, and in occasional sympatric inland populations. Chloroplast markers resolved distinct genetic groups, related to cytotype and geographically segregated; allopatric hexaploids were distinct from tetraploids, whereas sympatric hexaploids were not. Genome downsizing occurred between cytotypes. Progeny of open-pollinated clones from the contact zone showed that maternal tetraploids rarely produced progeny of other cytotypes, whereas the progeny of maternal hexaploids varied, with frequent pentaploids and aneuploids. The presence of distinctive hexaploid chloroplast types in Ireland, Scottish islands and western mainland Britain indicates that its establishment preceded separation of these land masses by sea-level rise c. 16 000 years BP. This group did not originate from British tetraploids and probably diverged before postglacial invasion from mainland Europe. The combination of cytotype, molecular, contact zone and common garden data shows an overall pattern reflecting postglacial colonization events, now maintained by geographic separation, together with more recent occasional local in situ polyploidisation. Reproductive barriers favour the persistence of the tetraploid to the detriment of the hexaploid.

Supplementary Table S2 Variable nucleotide sites in the chloroplast regions trnH-trnK and trnC-ycf6, and trnD-trnT score. Genbank accession numbers are given (Acc #) for each chloroplast region for every haplotype. Nucleotide position is indicated vertically and refers to the location of each variable site in each sequence. Dashes (-) indicate deletions. Question marks (?) signify missing data. Sequences are grouped into haplotypes based on sequence similarity (H1-29). Colour coding corresponds to the groups in Fig. 5b and 6. Haplotypes coloured grey correspond to those unresolved between groups B and C, in Fig. 5b.

T C T G G T -T G T C G G C T T T G A T A T T T A G A A C C C A G 1
H28 4, 5, 6 and aneuploid  APPENDIX S1

Collections
During field collections, leaf samples and stem cuttings were stored in polythene bags and kept cool for transport. Seeds were germinated and stem cuttings were rooted in a glasshouse for further study. Samples collected by volunteers were sent by post and analysed upon receipt.

Cytotype determination
Samples were analysed with either a Becton Dickinson FACSCalibur™ or a BD Accuri™ C6 flow cytometer with a red 488 nm wavelength laser, calibrated daily and run on the low flow rate (12 -14 µl min -1 ). Consistency between their outputs was checked by running duplicate samples. Analyses were repeated if full peak coefficients of variation exceeded 5 % or where DNA contents appeared to be outliers. To save resources, some samples were combined into groups of three. They were reanalysed separately if multiple peaks indicated mixed cytotypes.

Cx values
Samples from Britain were partly collected by volunteers who submitted them by post. We examined whether the time which elapsed between collection and processing and condition of the sample on receipt (fresh vs. desiccated; green vs. red/brown) affected the results.
Desiccation and red/brown colouration both produced some erratic results and these data were excluded from Cx analyses.

Molecular investigations
Using universal primers (including those for microsatellites and other regions) (Demesure et al. 1995;Shaw et al. 2005;Weising and Gardner 1999), a subset of samples covering the range of cytotypes and full distribution of the collection were screened for variation in the chloroplast genome. Initial tests of seven primers for the chloroplast regions trnC-trnD, matK, trnC-ycf6, trnH-trnK, trnD-trnT, ccmp2 and rbcL were tested on a screening panel of eight samples. The primer sets for trnC-trnD and matK failed to amplify during PCR, whilst the remaining five regions yielded a product. The successfully amplified regions were then tested on a larger panel of 24 samples and sequenced. Polymorphism was detected at all loci. Of these, trnC-ycf6, trnH-trnK, trnD-trnT were non-redundant and informative and were used to screen the full set of samples for variation.
Leaf tissue was taken from frozen samples. For each sample, roughly 1 cm² of tissue was ground to a fine powder using a Retsch Mixer Mill. DNA extraction was then carried out using QIAGEN DNeasy 96 Plant kits following the manufacturer's protocol. Extracted DNA was assessed for quality and concentration on a 1% agarose gel before being frozen at -20 °C.
PCR amplification of all fragments was carried out in 25 μl reactions containing 2 μl genomic DNA, 200 µM each dNTP (Promega), 0.2 µM each primer (MWG Biotech), 2.5 μl of 10X PCR buffer (New England Biolabs), 0.5 U Taq DNA polymerase (New England Biolabs) and 1.6 % (v/v) bovine serum albumin. Reactions were prepared in 96-well plates and run on a Thermo MBS thermal cycler following published protocols (Demesure et al., 1995) except for annealing temperatures of 62 °C for trnH-trnK, and 48 °C for trnD-trnT. Amplification of the trnC-ycf6 locus differed from the published protocol (Shaw et al., 2005), with an initial denaturation step of 94 °C for three min, followed by 35 cycles of 94 °C, 55 °C and 72 °C each for 30s, and a final extension step of 72 °C for 10 min.
PCR products of trnC-ycf6 and trnH-trnK were cleaned using 0.4 U Shrimp Alkaline Phosphatase (New England Biolabs) and 1 U Exonuclease I (New England Biolabs) to remove excess dNTPs and primers. Forward primers were added and the samples were then sent for sequencing at the NERC Biomolecular Analysis Facility sequencing service (Edinburgh Genomics) at the University of Edinburgh. Sequences were aligned using CodonCode Aligner (CodonCode Corporation) and manually checked.