Introduction
Over the year 2014, there were 31 purple- or red-fleshed potato cultivars enlisted in the European Cultivated Potato Database (2015). Most of them were underutilized tetraploid cultivars, dating back at least 50 years or more, and their origin often remains unclear. Intensely red or purple tubers show significantly higher levels of phenolic compounds and antioxidant capacity (Brown, Reference Brown2005; Reyes et al., Reference Reyes, Miller and Cisneros-Zevallos2005; Hamouz et al., Reference Hamouz, Lachman, Hejtmánková, Pazderů, Čížek and Dvořák2010). Therefore, certain heirloom cultivars such as ‘Vitelotte’, ‘Blue Congo’ and ‘Highland Burgundy Red’ have been reported to be a good source of phenolics (Lachman et al., Reference Lachman, Hamouz, Orák, Pivec and Dvorak2008; Hejtmánková et al., Reference Hejtmánková, Pivec, Trnková, Hamouz and Lachman2009). As most of these coloured potatoes are not commercially registered or their economic importance is low, they are usually held by non-professional farmers (The Kenosha Potato Collection Catalogue, 2015). Despite the lack of information about these cultivars, the attractive colours and shapes of coloured potato tubers and the increasing interest in the relationship between phytochemicals and health are contributing to the expansion of coloured potato market. In addition, the characterization of high phytochemical-producing tetraploid genotypes adapted to long-day conditions may facilitate the incorporation of interesting quality traits into new potato breeding lines without problems derived from the utilization of Andean potato cultivars with different ploidy levels and non-commercial characteristics (Visser et al., Reference Visser, Bachem, Boer, Bryan, Chakrabati, Feingold, Gromadka, Ham, Huang, Jacobs, Kuznetsov, Melo, Milbourne, Orjeda, Sagredo and Tang2009; Lindhout et al., Reference Lindhout, Meijer, Schotte, Hutten, Visser and van Eck2011).
The identification of potato cultivars is important during their registration process, seed production, trade and inspection. Despite the fact that morphological descriptors constitute an important tool for the classification of potato germplasm (UPOV, Reference Anon.1984), the morphometric approach to characterization is not effective for a number of cultivars (Nováková et al., Reference Nováková, Šimáčková, Bárta and Čurn2010). In the last 20 years, molecular markers have provided a powerful tool for DNA fingerprinting, and among these markers, microsatellites (simple sequence repeat, SSR) have proved to be very useful (Ashkenazi et al., Reference Ashkenazi, Chani, Lavi, Levy, Hillel and Veilleux2001). Microsatellites are highly polymorphic co-dominant molecular markers that are generally well conserved within and between related species (Powell et al., Reference Powell, Machray and Provan1996). Because of that, SSR markers have subsequently been used to study genetic relationships among Solanum tuberosum L. cultivars (Milbourne et al., Reference Milbourne, Meyer, Bradshaw, Baird, Bonar, Provan, Powell and Waugh1997; Ashkenazi et al., Reference Ashkenazi, Chani, Lavi, Levy, Hillel and Veilleux2001; Ruiz de Galarreta et al., Reference Ruiz de Galarreta, Barandalla, Ríos, López and Ritter2011; Côté et al., Reference Côté, Leduc and Reid2013) and also between accessions belonging to different potato species (Ghislain et al., Reference Ghislain, Spooner, Rodríguez, Villamón, Núñez, Vásquez, Waugh and Bonierbale2004; Spooner et al., Reference Spooner, Núñez, Trujillo, Herrera, Guzmán and Ghislain2007; Gavrilenko et al., Reference Gavrilenko, Antonova, Ovchinnikova, Novikova, Krylova, Mironenko, Pendinen, Islamshina, Shvachko, Kiru, Kostina, Afanasenko and Spooner2010; Cadima et al., Reference Cadima, Veramendi and Gabriel2013).
In the present study, 18 selected high anthocyanin-producing tetraploid potato cultivars adapted to long-day conditions, which have been previously classified in terms of phytochemical content (Tierno et al., Reference Tierno, Riga, Ruiz de Galarreta, Goffart, Rolot, Demeulemeester and Goeminne2014), have been characterized by both microsatellite markers (SSR) and morphological descriptors in order to select appropriate parents for use in a potato breeding programme for nutritional quality. The aim of the present study was to characterize a set of purple- or red-fleshed tetraploid S. tuberosum L. cultivars or breeding lines adapted to long-day conditions to (1) assist parental line selection and breeding strategy and (2) analyse and compare the information obtained using SSR markers and morphological descriptors.
Materials and methods
Plant material and experimental conditions
A total of 18 potato accessions held at the Neiker-Tecnalia Potato Germplasm Collection with highly or fully pigmented red or purple flesh and high levels of phenolic compounds and hydrophilic antioxidant capacity were characterized. These included accessions collected in Spain, France, The Netherlands, Sweden, Canada, Peru and Belgium (Table 1).
Table 1 Details of the collection of 18 Solanum tuberosum L. accessions with purple or red flesh
P, purple; R, red; PP, partially purple; PR, partially red; DP, deep purple.
Field trials were conducted in two different locations during the year 2014: Arkaute Research Station (42°51′ N, 2°37′ W and 518 m altitude) and Iturrieta Research Station (42°47′ N, 2°20′ W and 980 m altitude), province of Alava, North of Spain. Both sites are located in a transitional region between Atlantic and Mediterranean climates, but altitude determines to a great extent their climates. The Arkaute Research Station has a mean annual precipitation of 700 l/m2 (lpsm) and an average temperature ranging from 10 to 37°C during the cropping season, and the Iturrieta Research Station has a mean annual precipitation of 1000 lpsm and an average temperature ranging from 8 to 34°C during the cropping season. The soil, with a clay loam and a sandy loam texture in Arkaute and Iturrieta, respectively, was previously used for conventional wheat cropping. Plants were grown from mid-May to mid-October 2014 after pre-sowing fertilization with 800 kg/ha (NPK 4-8-16). Watering was performed using an automatic spray irrigation system. After harvesting, potatoes were stored at 4°C in a dark cold room for 1 month.
Morphological characterization
Morphological characterization was performed in two replicated plots of each genotype, with each replicate being composed of five plants grown on standard potato hills. Hills were separated by 0.75 m between rows and 0.30 m within rows. Experimental plots (1.2 m × 0.75 m) were arranged in a completely randomized design and comprised a total of 41 traits, including 34 quantitative, 1 qualitative and 6 pseudo-qualitative traits selected from the potato descriptors (UPOV, Reference Anon.1984). Qualitative and pseudo-qualitative traits were assessed using a continuous scale, thus being suitable to be collectively used with quantitative traits in the multivariate analysis.
ANOVA was performed with the SAS program package (SAS, 2011) to determine significant characteristics (data not shown). The multivariate principal component analysis (PCA) was performed with the NTSYS-PC software package (Rohlf, Reference Rohlf2001) to determine similar groups of accessions. After data standardization, a PCA was conducted to assess the patterns of variation, considering all the 41 traits. Cluster analysis was performed based on the Euclidean dissimilarity matrix using the weighted neighbour-joining method for interval measure (Nei and Saitou, Reference Nei and Saitou1987) with DARwin6 software (Perrier and Jacquemoud-Collet, Reference Perrier and Jacquemoud-Collet2006).
Molecular analyses
DNA extraction was carried out using the DNAeasy Plant Mini Kit (Qiagen, Valencia, CA, USA), and concentration was determined using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA). A set of 11 highly informative SSR markers were selected for the characterization of heirloom purple- and red-fleshed potato cultivars (Table 2).
Table 2 List of SSR markers, forward (F) and reverse (R) primer sequences, marker sizes based on sequence data, and optimum primer annealing temperaturesa
a Ghislain et al. (Reference Ghislain, Spooner, Rodríguez, Villamón, Núñez, Vásquez, Waugh and Bonierbale2004), Kawchuk et al. (Reference Kawchuk, Lynch, Thomas, Penner, Sillito and Kulcsar1996) and Milbourne et al. (Reference Milbourne, Meyer, Bradshaw, Baird, Bonar, Provan, Powell and Waugh1997).
The following four multiplexed SSR markers were utilized based on the protocol used by Reid et al. (Reference Reid, Hof, Esselink, Vosman and Burns2009, Reference Reid, Hof, Felix, Ruecker, Tams, Milczynska, Esselink, Uenk, Vosman and Weitz2011): set 1 – STM0019, STM3009 and STSS1; set 2 – STM2005, STM3012 and STM3023; set 3 – STM5136 and STM5148; set 4 – STM1024, STM1052 and STM1064. PCRs were performed in a 20 μl volume containing 2 μl 10 × PCR Buffer, 250 μM dNTPs, 2.5 mM MgCl2, 20 μM of each primer (forward and reverse), 5 units/μl Taq polymerase (LINUS, Teknovas, Spain) and 40 ng genomic DNA. Forward primers were labelled in their 5′ ends with the fluorescent dyes 6-FAM, VIC and NED (Applied Biosystems, Foster City, CA, USA). PCR was carried out in a Robocycler Gradient 96 thermocycler (Stratagene, La Jolla, CA, USA), using the following cycling profile: one cycle of 5 min at 95°C, 15 touch-down cycles of 45 s at 95°C, 30 s at 58°C and 90 s at 72°C, and 25 cycles of 45 s at 95°C, 30 s at 55°C and 90 s at 72°C followed by one final elongation cycle of 7 min at 72°C. Each 1 μl PCR product was mixed with 10 μl deionized formamide, labelled with 0.3 μl LIZ dye (Applied Biosystems) and denaturalized at 95°C. SSR fragments were detected on an ABI 3100 Genetic Analyzer (Applied Biosystems).
Each allelic peak was scored for presence (1) and absence (0). These data values were used to compute dissimilarity coefficients between the 18 accessions using the Dice coefficient (Dice, Reference Dice1945). A weighted neighbour-joining cluster analysis (Nei and Saitou, Reference Nei and Saitou1987) was performed with DARwin6 software (Perrier and Jacquemoud-Collet, Reference Perrier and Jacquemoud-Collet2006). A cophenetic matrix was computed from the tree matrix and compared with the original dissimilarity matrix in order to measure the goodness-of-fit (Rohlf, Reference Rohlf2001) based on Mantel statistics (Mantel, Reference Mantel1967). These analyses were performed with the NTSYS-PC software package (Rohlf, Reference Rohlf2001). The polymorphic information content (PIC) was calculated according to Nei's statistic (Nei, Reference Nei1973) as follows:
$$\begin{eqnarray} PIC = 1 - \sum ( p _{ i }^{2}), \end{eqnarray}$$
where p i is the frequency of the ith allele detected in the germplasm. Allele frequency distribution was analysed with the PROC UNIVARIATE procedure of the SAS program package (SAS, 2011).
Results
Morphological characterization
The dendrogram derived from the Euclidean distance coefficient using the UPGMA clustering method is shown in Fig. 1. Only significant traits were included in the statistical analyses. Differences were not found to be statistically significant at P≤ 0.05 in the following traits (data not shown): shape of light sprout (trait 2; UPOV, Reference Anon.1984); intensity of anthocyanin colour of the base of light sprout (trait 3; UPOV, Reference Anon.1984); proportion of blue in anthocyanin colour of the base of light sprout (trait 4; UPOV, Reference Anon.1984); size of tip in relation to the base of light sprout (trait 6; UPOV, Reference Anon.1984); habit of the tip of light sprout (trait 7; UPOV, Reference Anon.1984); number of root tips in light sprout (trait 10; UPOV, Reference Anon.1984); length of lateral shoots of light sprout (trait 11; UPOV, Reference Anon.1984); foliage structure (trait 12; UPOV, Reference Anon.1984); growth habit (trait 13; UPOV, Reference Anon.1984); outline size of leaves (trait 15; UPOV, Reference Anon.1984); openness of leaves (trait 16; UPOV, Reference Anon.1984); presence of secondary leaflets (trait 18; UPOV, Reference Anon.1984); depth of veins of leaflets (trait 24; UPOV, Reference Anon.1984); glossiness of the upper side of leaflets (trait 25; UPOV, Reference Anon.1984); pubescence of blade at the apical rosette of leaflets (trait 26; UPOV, Reference Anon.1984); plant height (trait 28; UPOV, Reference Anon.1984). The cophenetic matrix derived from the cluster analyses of morphological traits was in good agreement with the distance matrix (r 2= 0.830).
Fig. 1 Neighbour-joining tree constructed based on morphological data using the Euclidean dissimilarity matrix. Bootstrap values at the inner nodes were calculated using 100 bootstrap samples. In bold: accession numbers and major groups. In bold-italics: sub-groups.
Figure 1 shows four mean groups at a cut-off point for the bootstrap values (BV) of 50. Most North European, one North American and two Spanish potato accessions comprised the first group (I), which could be divided into two different subgroups (Ia and Ib). The partially purple cultivars ‘Bleu de La Manche’, ‘Blue Congo’, ‘Fenton’, ‘Valfi’ and ‘British Columbia Blue’ formed the first subgroup (Ia; BV = 50). The accessions ‘Bleu de La Manche’ and ‘Blue Congo’ were particularly close based on the results obtained from morphological traits (BV = 98). The second subgroup included the Spanish cultivars ‘Morada’ and ‘Entzia’ (Ib; BV = 100). The two fingerling and highly pigmented accessions ‘Purple Peruvian’ and ‘Vitelotte’ were clustered together in the second group (II; BV = 100). The third group consisted of three red-fleshed cultivars from Central Europe (‘Highland Burgundy Red’ and ‘Rouge de Flandes’) and South America (‘Rosa Roter’) (III; BV = 100). Considering the high BV of 86, the European genotypes formed a subgroup (IIIa). The fourth group consisted of two Spanish advanced clones: ‘NK-08/360’ and ‘NK-08/362’ (IV; BV = 60). A weak support was found for the cultivars ‘Blue Star’, ‘Jesus’, ‘NK-08/349’ and ‘Violet Queen’, considering the obtained BVs that were < 50.
PCA identified the variables most responsible for the pattern of relationships observed among the accessions, based on morphological descriptors. Table 3 presents eigenvalues and vectors for first three principal component (PC) axes calculated for 41 morphological traits of the studied clones. The first three PC axes explained half (63.8%) of the total variation. The first PC accounted for 37.0% of the total variation, with an eigenvalue of 10.3 (Table 3). Pubescence of base (trait 5; UPOV, Reference Anon.1984) and pubescence of tip (trait 9; UPOV, Reference Anon.1984) of light sprout showed positive coefficients in PC1 and PC2, explaining 21.2% (trait 5; UPOV, Reference Anon.1984) and 18.0% (trait 9; UPOV, Reference Anon.1984) of variation in PC1, and 38.6% and 18.3% in PC2, respectively. In contrast, green colour of leaves (trait 18; UPOV, Reference Anon.1984) and anthocyanin colour on the midrib of the upper side of leaves (trait 19; UPOV, Reference Anon.1984) were inversely correlated in PC1, PC2 and PC3. The time of maturity of plant (trait 36) was positively correlated in PC1. Besides green colour of leaves (trait 18; UPOV, Reference Anon.1984), anthocyanin colour on the midrib of the upper side of leaves (trait 19; UPOV, Reference Anon.1984), waviness of margin of leaves (trait 23; UPOV, Reference Anon.1984), anthocyanin coloration of flower buds (trait 27; UPOV, Reference Anon.1984), anthocyanin coloration on the peduncle of inflorescence (trait 31; UPOV, Reference Anon.1984), intensity of anthocyanin coloration on the inner side of flower corolla (trait 33; UPOV, Reference Anon.1984), proportion of blue in anthocyanin coloration on the inner side of flower corolla (trait 34; UPOV, Reference Anon.1984), extent of anthocyanin coloration on the inner side of flower corolla (trait 35; UPOV, Reference Anon.1984) and tuber shape (trait 37; UPOV, Reference Anon.1984) were highly but inversely correlated in PC1.
Table 3 Eigenvalues and vectors for first three principal component (PC) axes estimated for 24 morphological traits among 18 Solanum tuberosum L. accessions with purple or red flesh
a Only significant morphological characteristics were derived from UPOV (Reference Anon.1984) (P≤ 0.05).
b Morphological characteristic of light sprout.
c Morphological characteristic of stem.
d Morphological characteristic of leaf.
e Morphological characteristic of second pair of lateral leaflets.
f Morphological characteristic of terminal and lateral leaflets.
g Morphological characteristic of leaflet.
h Morphological characteristic of flower bud.
i Morphological characteristic of plant.
j Morphological characteristic of inflorescence.
k Morphological characteristic of flower corolla.
l Morphological characteristic of tuber. In bold: highest variance value of each principal component.
The second PC axis explained 15.4% of the total variation. Anthocyanin coloration of the tip of light sprout (trait 8; UPOV, Reference Anon.1984), intensity of anthocyanin coloration on the inner side of flower corolla (trait 34; UPOV, Reference Anon.1984), extent of anthocyanin coloration on the inner side of flower corolla (trait 35; UPOV, Reference Anon.1984) and colour of tuber flesh (trait 41; UPOV, Reference Anon.1984) showed positive correlations in PC2. Size of light sprout (trait 1; UPOV, Reference Anon.1984), green colour of leaves (trait 18; UPOV, Reference Anon.1984), anthocyanin colour on the midrib of the upper side of leaves (trait 19; UPOV, Reference Anon.1984), size of the second pair of leaflets (trait 20; UPOV, Reference Anon.1984), frequency of flowers (trait 29; UPOV, Reference Anon.1984), size of inflorescences (trait 30; UPOV, Reference Anon.1984), tuber shape (trait 37; UPOV, Reference Anon.1984) and tuber depth of eyes (trait 38; UPOV, Reference Anon.1984) showed high negative weights on PC2.
The third PC accounted for 11.3% of the variation unexplained by PC1 and PC2 (Table 3). Size of light sprout (trait 1; UPOV, Reference Anon.1984), frequency of coalescence in terminal and lateral leaves (trait 22; UPOV, Reference Anon.1984), frequency of flowers (trait 29; UPOV, Reference Anon.1984) and size of inflorescence (trait 30; UPOV, Reference Anon.1984) explained 20.4, 19.9, 29.3 and 24.2% of PC3 variation, respectively. Pubescence of the base of light sprout (trait 5; UPOV, Reference Anon.1984), width in relation to the length of the second pair of lateral leaflets (trait 21; UPOV, Reference Anon.1984), waviness of margin of leaves (trait 23; UPOV, Reference Anon.1984) and time of maturity (trait 36; UPOV, Reference Anon.1984) showed high negative weights on PC2.
Molecular analysis
All accessions were genotyped and well separated using this set of 11 SSR markers. Table 4 summarizes the information about the observed polymorphisms generated by each SSR marker. The number of alleles detected per SSR locus across the set of 18 tetraploid potato accessions with highly or fully pigmented purple- or red-flesh ranged from 3 to 11. A total of 61 SSR alleles were observed within the collection, ranging from 1 (monomorphic) to 4 (polymorphic) alleles per SSR locus. Of these total alleles detected, only three were present in all the clones. The average allele frequency was 18.0% and their coefficient of variation was 111%. According to the Jarque–Bera test (Jarque and Bera, Reference Jarque and Bera1987), allele frequencies were not normally distributed (P≤ 0.05).
Table 4 SSR polymorphisms in a collection of 18 Solanum tuberosum L. accessions with purple or red flesh
NAL, total number of alleles; Poly Al, number of polymorphic alleles; NPAT, number of SSR patterns; NHZ, number of homozygous accessions; Chr, chromosome location of the SSR marker in the potato genome; PIC, polymorphism information content.
a Origin of markers: All STM markers were employed by Milbourne et al. (Reference Milbourne, Meyer, Bradshaw, Baird, Bonar, Provan, Powell and Waugh1997). SSR1, STM5136 and STM5148 markers were utilized by Reid et al. (Reference Reid, Hof, Esselink, Vosman and Burns2009).
PIC values per SSR marker ranged from 0.518 (STM3023) to 0.849 (STM5148) and the number of homozygotes varied from 5 to 9, with an average of 7.64 (45%). The number of patterns generated by each SSR marker varied between 1 and 13. Allele numbers (NAL), pattern numbers (NPAT) and PIC values were all significantly and positively correlated with each other (
$$r _{PIC - NAL}^{2} = 0.718 $$
,
$$r _{PIC - NPAT}^{2} = 0.805 $$
,
$$r _{NAL - NPAT}^{2} = 0.903 $$
). However, from Table 4, we can observe that STM3009 has a relatively high PIC value (0.695) and reveals a maximum of seven alleles, but generates only six different patterns. Other SSR alleles, such as STM1052 (PIC = 0.725) and STM1064 (PIC = 0.602), have only four and three alleles, respectively, but reveal six different patterns and hence show a higher discriminative power in this case. Of the 11 SSR markers used in our study, four were previously utilized by Ruiz de Galarreta et al. (Reference Ruiz de Galarreta, Barandalla, Lorenzo, González, Ríos and Ritter2007, Reference Ruiz de Galarreta, Barandalla, Ríos, López and Ritter2011) and five were employed by Ghislain et al. (Reference Ghislain, Spooner, Rodríguez, Villamón, Núñez, Vásquez, Waugh and Bonierbale2004) to fingerprint cultivated potato. The base-pair range limits in our collection of pigmented potato accessions for these five SSR markers were comparable with those obtained by Barandalla et al. (Reference Barandalla, Ruiz de Galarreta, Ríos and Ritter2006), and 34% of the alleles obtained with these markers were detected in our study. Despite the fact that our collection was reduced, we found higher NALs for STM1052 than did Ruiz de Galarreta et al. (Reference Ruiz de Galarreta, Barandalla, Lorenzo, González, Ríos and Ritter2007, Reference Ruiz de Galarreta, Barandalla, Ríos, López and Ritter2011) and also higher NALs for STM0019, STM3012, STM3023, STM1052 and STM 1064 than those observed by Barandalla et al. (Reference Barandalla, Ruiz de Galarreta, Ríos and Ritter2006).
We detected several accession- and group-specific alleles (Table 4). Of these alleles detected, thirteen were accession specific: ‘Fenton’ (STM5136_247); ‘Jesus’ (SSR1_244); ‘Morada’ (STM5148_425); ‘Entzia’ (STM2005_162); ‘NK-08/349’ (SSR1_249); ‘NK-08/362’ (STM3009_143, STM3023_198 and STM5148_458); ‘Purple Peruvian’ (STM2005_166); ‘Violet Queen’ (STM2005_168 and STM5148_471); ‘Vitelotte’ (STM0019_236 and STM5148_429). Furthermore, two alleles (SSR1_250 and STM1024_151) were specific for the partially red-fleshed cultivars ‘Highland Burgundy Red’ and ‘Rouge de Flandes’ (group III), one allele (SSR1_245) for the partially coloured accessions ‘NK-08/349’ and ‘NK-08/362’ (group II) and two alleles (STM5148_428 and STM1052_228) for the partially purple-fleshed Spanish accessions ‘NK-08/360’, ‘Entzia’, ‘Morada’ and ‘Jesus’. They also shared another allele (SSR1_253) with five members of group I (‘Bleu de La Manche’, ‘British Columbia Blue’, ‘Blue Congo’, ‘Valfy’ and ‘Blue Star’). Accession and group-specific alleles were found mostly with the SSR markers STM2005 (16.7%), SSR1 (22.2%) and STM5148 (27.8%). High levels of heterozygosity were found among the collection of tetraploid potato accessions with purple or red flesh.
The neighbour-joining tree derived from the Dice coefficient of dissimilarity is shown in Fig. 2. The cophenetic matrix derived from the cluster analyses of SSR data was in good agreement with the original dissimilarity matrix (r 2= 0.896). Setting the cut-off point for the BVs to 50 allowed us to distinguish a total of six clusters, including two major groups and four subgroups. The first group (I), which included ten partially purple- and red-fleshed accessions from Spain, Central Europe and South America, could be divided into three subgroups (Ia, Ib and Ic). The first one consisted of the Spanish cultivars ‘Jesus’, ‘Morada’ and ‘Morea’, which showed similar SSR patterns (Ia; BV = 79). The accessions ‘Jesus’ and ‘Morada’ were grouped closer and showed a BV of 85. The next subgroup included the red-fleshed cultivars from Central Europe: ‘Highland Burgundy Red’ and ‘Rouge de Flandes’ (Ib; BV = 87). The almost totally pigmented cultivars ‘Purple Peruvian’ and ‘Vitelotte’ clustered together in the last subgroup (Ic; BV = 100). The breeding line ‘NK-08/349’ was also included in the first group. A total of five partially purple-fleshed accessions from Europe and North America comprised the second group: ‘Bleu de La Manche’, ‘Blue Congo’, ‘British Columbia Blue’, ‘Fenton’ and ‘Valfi’ (II; BV = 72). A high BV of 66 was also found in a subgroup of three accessions: ‘Bleu de La Manche’; ‘Blue Congo’; ‘Valfi’ (IIa). A weak support was obtained for the accessions ‘Blue Star’, ‘Rosa Roter’ and ‘Violet Queen’ based on bootstrapping (BV < 50).
Fig. 2 Neighbour-joining tree based on SSR data using the Dice dissimilarity matrix. Bootstrap values at the inner nodes were calculated using 100 bootstrap samples. In bold: accession numbers and major groups. In bold-italics: sub-groups.
Discussion
The collection of 18 accessions with purple- or red-fleshed potato cultivars or lines displayed a large amount of variation for most morphological traits (Table 3 and Fig. 1). Both the PCA and cluster analyses showed the distinctness, similarities and overlap of morphological traits within and among the groups. Analysis of the data from the 11 SSR markers showed the genotypic relatedness of the material (Fig. 2). The existence of broad morphological diversity among the 18 coloured potato accessions was further substantiated by PCA, which indicated that the total variation was distributed across 39 of the 41 observed morphological traits. The grouping pattern observed for the collection of 18 purple- or red-fleshed potato accessions showed consistency for morphological descriptors and at genotypic relatedness (based on SSR data). Most of the North Central European or North American potato accessions, such as ‘Bleu de La Manche’ (France), ‘British Columbia Blue’ (Canada), ‘Blue Congo’ (Sweden), ‘Valfi’ (Sweden) and ‘Fenton’ (Canada), clustered together using both approaches. The cultivars ‘Purple Peruvian’ and ‘Vitelotte’, two fingerling potatoes with deep eyes and deep purple flesh, also clustered together. The red-fleshed potato accessions ‘Highland Burgundy Red’ and ‘Rouge de Flandes’ also grouped closely in both cases (Figs 1 and 2).
Multiple molecular markers have been successfully applied to potato genetic resources at low taxonomic levels, including RFLP (Görg et al., Reference Görg, Schachtschabel, Ritter, Salamini and Gebhardt1992), RAPD (Ghislain et al., Reference Ghislain, Zhang, Fajardo, Huamán and Hijmans1999), AFLP (Kim et al., Reference Kim, Joung, Kim and Lim1998), ISSR (Bornet et al., Reference Bornet, Goraguer, Joly and Branchard2002), SNPs (Hardigan et al., Reference Hardigan, Bamberg, Buell and Douches2015) and SSR markers (Veilleux et al., Reference Veilleux, Shen and Paz1995). We have chosen SSR markers for our study since they are highly polymorphic and show a good reproducibility and simplicity of use (Ghislain et al., Reference Ghislain, Spooner, Rodríguez, Villamón, Núñez, Vásquez, Waugh and Bonierbale2004). According to Milbourne et al. (Reference Milbourne, Meyer, Bradshaw, Baird, Bonar, Provan, Powell and Waugh1997) and Ghislain et al. (Reference Ghislain, Spooner, Rodríguez, Villamón, Núñez, Vásquez, Waugh and Bonierbale2004), SSR markers may be adequate tools for analysing genetic distances between potato genotypes. However, the different sets of SSR markers applied in different cases may difficult the comparison of results between various studies. According to Barandalla et al. (Reference Barandalla, Ruiz de Galarreta, Ríos and Ritter2006), NAL, PIC values and NPAT indicate the discriminative power of individual SSR primers for a particular study. A total of 76 SSR alleles were found by Ruiz de Galarreta et al. (Reference Ruiz de Galarreta, Barandalla, Ríos, López and Ritter2011) among a collection of 105 Spanish accessions, while 62 SSR patterns were reported in a set of 22 potato landraces from the Islands of La Palma and Tenerife by using a set 19 SSR markers (Ruiz de Galarreta et al., Reference Ruiz de Galarreta, Barandalla, Lorenzo, González, Ríos and Ritter2007). Average PIC values reported by these authors were 0.589 and 0.518, respectively. In this study, we found a total of 61 alleles, 73 different SSR patterns and an average PIC value of 0.675. Several accession- and group-specific alleles were also detected. Considering the reduced set of accessions, we found a wide genetic variability using a small set of highly discriminatory SSR markers. The most informative SSR markers were SSR1, STM2005, STM5136 and STM5148. Contrary to expectations, no redundancies were identified between different accessions via SSR analysis.
Findings based on morphological and SSR data implied that the accessions of clusters Ia (Fig. 1) and II (Fig. 2), IIIa (Fig. 1) and Ib (Fig. 2), and II (Fig. 1) and Ic (Fig. 2) were morphologically and genetically related. Despite the fact that Spanish accessions were not equally clustered, most of them formed separated groups in both cluster analyses (Ib and IV in Fig. 1 and I in Fig. 2). The reduced genetic and phenotypic variation among the various groups of accessions can be attributed to a closely related pedigree or to somaclonal variations (Larkin and Scowcroft, Reference Larkin and Scowcroft1981). Morphological traits associated with this grouping are pubescence of the base and tip of light sprout (traits 5 and 9), green colour of leaves (trait 18) and anthocyanin coloration on the midrib of the upper side of leaves (trait 19). Based on SSR data, two pairs of accessions (‘Blue Congo’-‘Valfi’ and ‘Purple Peruvian’-‘Vitelotte’) showed high genetic relatedness. These accessions clustered together, considering both for SSR markers and morphological traits (Figs 1 and 2). However, the variation at the morphological level was significantly higher. On the other hand, the pair of accessions ‘Highland Burgundy’ and ‘Rouge de Flandes’ was morphologically related, although it showed wide variation at the genotypic level.
This study allowed us to distinguish at least three groups of potato accessions with purple or red flesh, with distinctive genetic profiles and morphological traits. The first group constitutes the intermediate maturing North Central European and North American accessions with partially purple flesh (Figs 1 and 2). The second group includes the red-fleshed accessions, which are morphologically very different from the other groups since they have small leaves, red stems and small white flowers. The third group includes late-maturing and low-yielding accessions with long-shaped, deep-eyed and intensely purple-fleshed tubers. Most Spanish accessions were also genetically and morphologically related.
More detailed studies are needed to examine potato genetic diversity and taxonomy at low taxonomic levels. Despite the complexity of tetrasomic inheritance, tetraploid potato cultivars with purple or red flesh should be considered for potato breeding since they are well adapted to long-day conditions, show high levels of anti-ageing compounds and do not present some undesirable characteristics that are found in several native accessions or Andean potato landraces. However, the lack of information about most of these cultivars, which are usually grown as exotic products for non-commercial purposes, and even the plant material shortage make their inclusion in breeding programmes difficult. The characterization of a small set of 18 accessions suitable for breeding for nutritional quality by means of morphological descriptors and SSR markers has been helpful to assist parental line selection, to analyse genetic and morphological variability of the collection and also to shed some light on the relationships between different accessions and groups using the two approaches.
Acknowledgements
This work was financed within the frame of INIA's project RTA2013-00006-C03-01.
