Perennial Arabis alpina has a wide geographic distribution and is adapted to local environments. However, the traits that underlie adaptation are unknown. Flowering phenology is an adaptive trait in other species, but its geographic variation has not been systematically studied in herbaceous perennials.

Accessions of A. alpina were collected across the European range. Their flowering behavior was tested in controlled conditions, in experimental common-garden plantations at native sites and in-situ in natural populations. Also, genetic diversity within and among populations was examined.

French Alpine and Scandinavian accessions varied in timing and duration of flowering. By contrast, in controlled conditions and in-situ, all Spanish accessions were obligate vernalization-requiring with a short duration of flowering. Nevertheless, Spanish populations were as genetically diverse as French Alpine populations and more so than Scandinavian populations. Furthermore, perpetual flowering 1, a mutant that shows no vernalization requirement and a long duration of flowering, showed higher mortality and poorer performance than local accessions at Spanish experimental sites.

We propose that in this perennial species, the vernalization requirement and short duration of flowering are under selection in Spain as a strategy to survive exposure to longer, warmer growing seasons.

Genotyping and whole-genome sequencing

Genetic variation and differentiation of A. alpina populations were analyzed with genotype and whole-genome sequence data. First, DNA was extracted from fresh leaves of 31 individuals from 9 populations representing Spanish, French Alpine and Scandinavian populations (using Qiagen DNeasy Plant Maxi Kit, Qiagen) and whole genomes were sequenced on an Illumina HiSeq2000 at the Max Planck Genome Centre Cologne and at FASTERIS sequencing service in Switzerland (www.fasteris.com). From these sequences, 35788 unambiguous biallelic SNPs were identified (details in Supporting Information). A subset of 253 widely spread, intergenic SNPs was used for genotyping on a Custom GoldenGate SNP-panel. The SNP set was chosen to be informative across the entire sampled geographic range and/or within regions and populations. Then, 892 accessions were genotyped using DNA extracted from leaf material collected in-situ (Supporting Information). These accessions include: i) a range-wide collection; ii) an evenly subsampled set of 50 accessions from each of three populations from each of the three main regions; iii) a total of 187 Spanish accessions from 10 populations to analyze in detail population structure in Spain. 

In addition, genetic variation was analyzed genome-wide on an evenly subsampled set of the whole genome sequences, comprising eight individuals from two populations in both Spain and the French Alps, and eleven individuals from four populations in Scandinavia, for a total of 27 individuals. Average coverage for these samples varied between 7.2X and 23.1X (mean across samples = 15.9X).

Phenotyping experiments

Flowering and fitness traits were scored by phenotyping plants in controlled greenhouse conditions and at experimental sites established within or next to native populations of A. alpina. Some traits were also scored in situ. Details are given in Supporting Information. 

Greenhouse experiments

Flowering behavior was scored under greenhouse conditions following two different experimental treatments. i) Experiment without vernalization: Plants were grown under long days (LD: 16h/8h day/night cycle, 22°/16°C) and scored for flowering traits twice a week, for 26 to 51 weeks after germination across four replicates of the experiment (Table S1) ii) Experiment with vernalization: Plants were grown in long days for 6 weeks, then vernalized for about 14 weeks (short days, SD: 8h/16h day/night cycle, 4°C), and then transferred back into LD and scored for 19 weeks. In each experiment, most natural populations were represented by at least 20 maternal families and at least 8 siblings per family, except experiment 4 in which smaller family sizes were used (Table S1). As controls, the reference accession Pajares from Spain and the early-flowering pep1-1 mutant were used in all experiments. All plants were randomized in all experiments.

Experimental sites   

Flowering behavior and fitness of plants grown at experimental sites were scored in Spain, the French Alps and Scandinavia, close to or within native populations, and in Germany outside the natural range. Plants were pre-grown in small plugs for about 4 weeks in a greenhouse to ensure high establishment success. The seedlings were then transplanted into natural soil at the experimental sites in spring. Apart from watering on the first few days after planting, the plants grew under natural environmental conditions and in native soil, and they were scored at the end of the growing season the year of plantation and the following year. Each population was represented by about 20 maternal families and 10 siblings per family (Table S1). As controls, we used the reference accession Pajares from Spain and the early-flowering pep1-1 mutant. The individuals were planted in a randomized block design.

Observations in-situ

Plants in natural populations were labelled, photographed and scored for various traits 2-5 times per year over several years. To estimate the duration of flowering the photographic documentation was evaluated for a period of five years (Supporting Information).

Phenotypic traits

For flowering behavior under controlled greenhouse conditions, the following traits were scored: i) the onset of flowering without vernalization, which was the time between germination and appearance of the first open flower, ii) the onset of flowering after vernalization, which was the time from the end of the vernalization treatment to the appearance of the first open flower, iii) the duration of flowering after vernalization, which was the time between appearance of the first and the last open flower and iv) the proportion of flowering plants, which was the proportion of plants that produced visible open flowers. Plants were assigned to three categories for the onset of flowering without vernalization: “Early-onset” plants flowered within four months of germination (this reflects the average snow-free period in the Scandinavian and French Alpine habitats). “Late-onset” plants required more than four months from germination to flowering. “Not flowering” plants produced no open flowers without vernalization during the experiment (up to 51 weeks).

At experimental sites, we scored survival, as the percentage of surviving plants, and the proportion of flowering plants, inferred by the proportion of plants bearing siliques. Scoring was done at the end of the growing season the year of plantation and the following year. At the Spanish site we additionally scored the number of rosettes for vegetative performance, and the average number of siliques as an indication of fecundity and fitness (Supporting Information).

In natural populations (in-situ) individual plants were labeled, visited 2-5 times per year, photographed and scored for whether or not they were flowering. The proportion of plants showing an extended (perpetual) flowering duration was estimated at the end of the growing season, typically late September to early October.

Weather data

Air temperature was measured with logging intervals ≤ 45 minutes by sensors installed about 20 cm above the soil surface in a subset of the natural populations and at all experimental sites. Records were standardized to hourly means from which daily means were calculated. Periods of snow coverage were estimated based on oscillations ≤ 1°C in daily temperature, indicating that the sensor was covered by snow. The growing season was defined as the number of days between the midpoints of the first and last snow-free, 5-day period with mean temperature > 5°C, following the definition of (Sletvold & Agren, 2015).

Temperature data were analyzed using custom made scripts and the R-package ‘seas’ (Toews et al., 2007). Precipitation data was obtained from public weather stations near the field sites.