DNA methylation can be environmentally modulated and play a role in phenotypic plasticity. To understand the role of environmentally induced epigenetic variation and its dynamics in natural populations and ecosystems, it is relevant to place studies in a real-world context. Our experimental model is the wild potato Solanum kurtzianum, a close relative of the cultivated potato S. tuberosum. It was evaluated in its natural habitat, an arid Andean region in Argentina characterised by spatial and temporal environmental fluctuations. The dynamics of phenotypic and epigenetic variability (with Methyl Sensitive Amplified Polymorphism markers, MSAP) was assayed in three genotypes across three growing seasons. These genotypes were cultivated permanently and also reciprocally transplanted between experimental gardens (EG) differing in ca. 1000 m of altitude. In two seasons, the genotypes presented differential methylation patterns associated to the EG. In the reciprocal transplants, a rapid epigenomic remodelling occurred according to the growing season. Phenotypic plasticity, both spatial (between EGs within season) and temporal (between seasons), was detected. The epigenetic and phenotypic variability were positively correlated. The lack of an evident mitotic epigenetic memory would be a common response to short-term environmental fluctuations. Thus, the environmentally induced phenotypic and epigenetic variation could contribute to populations persistence through time. These results have implications for understanding the great ecological diversity of wild potatoes.

The spreadsheet has four sheets. The first titled MSAP_epiallele presents raw MSAP data with methylation patterns (0,1,2 and 3) for three MSAP markers. The second one titled MSAP_epiloci presents the binary data of the MSAP markers. The third and fourth show all the phenotypic features measured in leaves and tubers.

Experimental gardens and reciprocal transplants approach

We evaluated morphological, biochemical, and phenological phenotypes and analysed the dynamics of DNA methylation patterns in clones of three S. kurtzianum genotypes. These clones were cultivated during three successive years, both permanently and as reciprocally transplanted, in two Andean Experimental Gardens (EGs), located at 1141 and 2113 m a.s.l. in the VNR. To establish the EGs, and according to the number of available tubers, one genotype from each of the three populations collected in the VNR and studied by Marfil and Masuelli (2014) -1228M, QH5 and 2166M- were selected and labelled as G-1, G-2 and G-3, respectively. Twelve tubers (clones) per genotype were cultivated in pots at the campus of Facultad de Ciencias Agrarias, UNCuyo (33° 00′ 24´´ S, 68° 52′ 19´´ W, 940 m a.s.l.) from November 2013 to January 2014. When the pot-grown plants presented three fully expanded leaves, they were transplanted into 10 L pots with a mixture of autoclaved local soil and organic compost (3:1), to prevent seed contamination. These pots were used to establish the field experiment in January 2014, in two EGs located in the VNR. One EG (hereinafter called "EG-1100") was located at 1141 m a.s.l. (S32° 34' 39.57” S, 68° 56' 45.65” W), and the other (hereinafter called "EG-2100") at 2113 m a.s.l. (32° 35' 06.47'' S, 69° 05' 56.85'' W). Half of the clones of each genotype were placed in each EG following a completely random design (Ibañez et al. 2017). Overall, six clones × three genotypes × two EG were evaluated. Plants were weekly irrigated with KSC II PHYT-actyl fertiliser (Timac Agro, USA). In September 2014, tubers formed during the growing season (November 2013- April-May 2014) were collected and conserved in the laboratory at 4º C until dormancy breakdown. The sprouted tubers were used as starting material for the next growing season. The cultivation conditions were repeated with the modifications detailed below. The cultivation conditions in 2015, 2016, and 2017 were modified by incorporating reciprocal transplants between EGs (Fig. 1). In each season, clones were divided into two groups. Three clones per genotype were cultivated repeatedly in the same EG as in the previous season (permanent clones). For example, clones of G-1, G-2 and G3 cultivated in EG-1100 in 2014 were cultivated again in EG-1100 in 2015. The other three clones were transplanted (transplanted clones) into the other EG (reciprocal transplants). For example, clones of G-1, G2 and G3 cultivated in EG-1100 in 2014 were transplanted into EG-2100 in 2015, whereas those grown in EG-2100 in 2014 were transplanted into EG-1100 in 2015 (Fig. 1). In summary, in 2015, 2016 and 2017, three biological replicates of each of the three genotypes and condition (permanent or transplanted) were cultivated in each EG, that is, three permanently grown clones and three transplanted clones of each of the three genotypes were cultivated in each EG, totalling 18 plants per EG .

MSAP analysis

MSAP markers were used to estimate epigenetic variability according to Cara et al. (2013). Three selective primer combinations without fluorescence were assayed: EcoRI+AGC⁄HpaII-MspI+ATC, EcoRI+AGA⁄HpaII-MspI+ATC and EcoRI+ACA⁄HpaII-MspI+ATC. The amplification products were separated by electrophoresis on a 6% denaturing polyacrylamide gel, silver stained, and scanned for manual scoring. To calculate the error rate, a subset of 12 samples were analysed in duplicate for each season, using the same extracted DNA. From the EcoRI/HpaII and EcoRI/MspI raw data, each locus could present multistate information (MSAP_epiallele). To construct binary epigenetic matrices, each locus in the multistate matrix was split into three subloci (or subepiloci; MSAP_epiloci) following the codification ‘Mixed Scoring 2’ described by Schulz et al. (2013): fragments present in both the HpaII and MspI lanes were codified as Type I, fragments present in the HpaII lane but absent in MspI lane were codified as Type II, and fragments present in the MspI lane but absent in HpaII lane were codified as Type III. Then, two types of methylation states were considered: non-methylated (Type I) and methylated (types II +III). In the methylated state, both amplification patterns (II + III) were clustered in one category according to the findings of Fulneček and Kovařík (2014), who indicated that the hemi-methylated state (fragment presence in the HpaII lane) could not be determined with accuracy. The absence of fragments on both lanes was not used for calculating the methylation state because it could be the result of genetic changes.

Phenotypic characterisation

All variables described below were measured for all the plants cultivated in 2015, 2016 and 2017 at both EGs. The 6^th and 7^th fully expanded leaves were removed from plants in anthesis, wrapped with aluminium foil and conserved on ice. Once in the laboratory, leaves were photographed and the total leaf area, terminal leaflet area and rachis length were measured with AxioVison software (Carl Zeiss MicroImaging GmbH, Jena, Deutschland); leaflet number was also registered. Leaflet thickness was estimated through the area/weight ratio in two 2 cm² leaflet discs dried at 40°C until reaching a constant weight. Photoprotective pigments as ultraviolet absorbing compounds (UVAC) and anthocyanins were determined in the leaflet discs according to Ibañez et al. (2017). At harvest, the total tuber number and total and mean tuber weight were recorded for each plant, except in the 2017 season in which data from EG-2100 had to be excluded due to the harmful action of stray animals on both substrate and tubers. Harvested tubers were stored at 4ºC and periodically examined for sprouting (dormancy breakdown). Sprouted tubers were maintained at room temperature in darkness, and the sprouting percentage and sprout length were recorded every two weeks for 60 days.