Understanding the genetic diversity of breeding germplasm is one of the fundamental tools for a plant breeder, to make informed crosses for the development of new and improved cultivars. By understanding the genetic diversity of in vitro collections of potato accessions in genebanks, plant breeders may utilize it wisely in their crossing schemes. In this study, we used SNPs from an array containing 22K SNP markers to assess the genetic diversity of accessions used for breeding table potatoes for the Swedish market. The accessions included breeding clones from the SLU breeding program; germplasm from NordGen made up of old breeding clones, obsolete cultivars and farmer’s cultivars; and a few cultivars grown today often used as breeding parents in the SLU breeding program.
The number of genotypes (population size) are unevenly distributed among the groups being compared throughout all analysis in this study. The largest difference in population size was found when comparing the three groups based on source of germplasm (NordGen, SLU and cultivars grown today), the group cultivars grown today contain only nine accessions (five with phenotypic data avaliable), and NordGen contain 75 accessions. There are about 25 cultivars commonly grown for the potato market today in Sweden (personal communication: Anders Andersson, Potatisodlarna). Among these 25 cultivars are ‘King Edward’, ‘Bintje, ‘Connect’ and ‘Carolus’, which we have included in this study. The most commonly grown early maturing cultivar in Sweden by far is ‘Solist’. Both the phenotypic and genotypic diversity could be limited to the number of individuals, which must be considered when discussing the results from this study. The grouping based on clonal type has a more even distribution of population size, spanning from 29 (farmer’s cultivars) to 57 (breeding clones) genotypes.
The phenotypic data from the NordGen accesssions has been extensively described in a previous publication (Veteläinen et al.2005). We combined these old phenotypic records with new data recorded for clones from the SLU breeding program to compare the phenotypic diversity among both germplasm sources. Phenotypic traits such as percent dry matter are affected by environmental factors, for example post-harvest storage conditions and soil type (Wilson and Lindsay 1969; Schrippers 1976). There were no cultivars used as checks in common for the study by Veteläinen et al.and the field trial described here, hence it was not possible to determine any environmental effects on the three phenotypic traits. The results show that on the group level, all measured phenotypic traits differ among the three groups based on source of the germplasm (Fig. 1). Tuber shape uniformity significantly differs among the groups, and the highest variation for this trait was noted for SLU accessions, and more specifically, among the breeding clones. An explanation to why the genotypes from SLU exhibits a large variation in tuber shape uniformity may be that this trait is not considered as important compared to other breeding traits during the early selection cycles of the breeding program (T1-T4). A larger proportion of the genotypes from NordGen has a higher percent dry matter content compared to the other two groups, a character that is correlated to mealy potato tubers (Burton 1989). When studying the phenotypic variation among the different groups based on clonal type (Fig. 2), it becomes clear that the higher scores for percent dry matter from NordGen originated from the farmer’s cultivars within this population. The farmer’s cultivars have, on average, higher percent dry matter than the breeding clones and cultivars grown today in Sweden. The largest spread of percent dry matter scores was noticed among the breeding clones. This result suggests that this trait was not the target when making selections in the potato breeding program. The farmer’s cultivars also differ from the two other clonal types regarding tuber eye depth. The farmer’s cultivars seem to have much deeper tuber eyes compared to the two other clonal types. Potato breeders are often selecting for shallow eyes as this makes tubers easier to peel for the consumers of table potato. These farmer’s cultivars did not result from breeding efforts, but instead they reflect what characters potato growers and consumers have favoured. Hence, the tuber characters favoured by end users differ geographically and over time. In Sweden today, ‘King Edward’ remains the favourite table potato mainly due to its appreciated flavour and good properties for the production of mashed potatoes as well for other uses in gastronomy. Such a finding shows that flavour and possibility to use a cultivar for many purposes should be considered when breeding new cultivars.
In general, potato exhibits a low degree of population structure (D’hoop et al. 2010: Ortiz 2015), with European cultivars in particular stemming from a very narrow genetic base (Glendinning 1983; Srivastava et al. 2016). The results from our study suggest that there is a very limited population structure among potato cultivars and breeding clones in the Nordic region. In line with what previously was reported by Veteläinen et al. (2005), no population structure was observed based on country of origin (Fig. 3A). The NordGen definition of country of origin is, however, limited to and defined as the country in which the accession was collected, and it is known that several of the farmer’s cultivars have been grown in more than a single Nordic country. This as well as the limited genetic base of European potato cultivars in general may have affected the lack of structure based on country of origin (Fig. 3A).
Population structure was examined using a PCoA based on Nei’s pairwise distance among genotypes (Fig. 3A-C) and STRUCTURE (Pritchard et al. 2000) with K ranging from 0 to 10 (Fig. 4A and Fig. 4C). The K with the maximum likelihood was 2, thus suggesting two subpopulations in the data. This theoretical number of subpopulations is to a limited extent explained in the results from the PCoA and heatmap. A group of breeding clones from SLU does cluster separately from the other material. When studying this group more closely, it appears that clones from the early cycles of selection in SLU breeding program are represented in this outlying group (Fig. 3B, Fig. 4C). Breeding clones from SLU which have undergone a larger number of cycles of selection are in the same cluster with the rest of the material, thereby suggesting that the potato breeders are actively making selections towards clones that are similar to cultivars grown in Sweden today and farmer’s cultivars previously grown in the Nordic region.
The bar plot from the STRUCTURE output suggests a possible subpopulation division based on farmer’s cultivars versus the remaining materials (Fig. 4A). This division of subpopulations is unclear when studying the PCoA (Fig. 3C). However, there is a cluster represented by farmer’s cultivars appearing in the heatmap (Fig. 4B), which is based on the same genetic data as the PCoA. A similar population structure, where farmer’s cultivars grouped in a separate subpopulation was found in Chinese potato collections (Wang et al. 2019). It is still important to keep in mind that almost every genotype was an admixture of both theoretical subpopulations, which suggests a weak population structure.
No structure based on country of origin was observed, which could be explained by the narrow genetic base of the sampled accessions. The pedigree information is limited for potato in general, but it is not inconceivable that the farmer’s cultivars included in this study would appear in the pedigree of the breeding clones either from SLU or among the NordGen accessions. Several of the cultivars grown in Sweden today are actively used as breeding parents at the SLU potato breeding program. It would have been interesting to include genotypes with a wider geographical background, especially outside Europe to get an estimation if the Nordic region has been uniquely differentiated compared to potato grown elsewhere in the world.
Previous research found that market class was a good biological explanation to subpopulations in cultivated tetraploid potato (Hirsch et al. 2013; Igarashi et al. 2019; Pandey et al. 2021). In this study, all potato accessions are classified as table potato with one exception, the cultivar ‘Kuras’ from the population of cultivars grown in Sweden today. ‘Kuras’ is grown for starch production (Bauw et al. 2006). We did not, however, observe this as an outlier in our genetic structure analysis, but instead it grouped close to the other Dutch cultivars in the study (data not shown).
The genetic diversity of the potato germplasm from NordGen has been examined previously by Veteläinen et al. (2005). Their study, focusing on finding morphological characters to identify duplicates among accession, included a fraction of the accessions used in our study. All of them were farmer’s cultivars. Our results using a 22K SNP array supports what Veteläinen et al. (2005) noticed; i.e., duplicated accessions of potato did not exist in the NordGen genebank. The study also included a genetic diversity study assessment using 63 AFLPs. The dendrogram from their study based on AFLPs was different than ours based on SNPs (Supplementary Fig. S1). The dendrogram by Veteläinen et al. (2005) based on their 57 morphological characters did not match with the dendrogram from our study (Supplementary Fig. S1). A separate dendrogram containing all accessions from the NordGen population was also drawn using the SNP data (Fig. 5). As in the PCoA (Fig. 3A), no population structure was revealed based on country of origin from the dendrogram. When the accessions were assigned as type of clone, the structure could be explained by the clustering in the dendrogram to a certain degree. Two subgroups were revealed, where most of the released cultivars and breeding clones branched separately from the farmer’s cultivars.
Crop yield potential has successfully been improved through heterosis for other crops (Breiger 1950; Duvick 2001; Lippman and Zamir 2007). It has been theorized that over time, potato breeding will increase the level of heterozygosity due to heterosis related to increased tuber yield (Mendoza and Haynes 1974; Hirsch et al. 2013). In line with what was found by Hirsh et al. (2003) studying US potato cultivars, the level of heterozygosity did not change significantly over time in our study. However, the number of cultivars were significantly larger in the later years compared to the early years, making the estimations skewed.
The average percent heterozygosity ranged from 48 to 50 % for each population included in this study. This is a bit lower than what was recorded previously in other germplasms. The average percent heterozygosity was 56 % or 57% in germplasm from the USA or Japan respectively (Hirsh et al. 2013, Igarashi et al. 2019). Another US germplasm set was investigated by Pandey et al. (2021) with an average percent heterozygosity of 60 %. In both studies in the USA as well as in the study in Japan, a very similar or identical SNP array was utilized for genotyping as in our study. Hence, the genetic variability of the Nordic germplasm is slightly lower. The percent heterozygosity did not vary between populations or types of clones, thus contradicting the theory that breeding would increase the germplasm’s level of heterozygosity, which may otherwise suggest limited genetic gains for tuber yield in potato breeding as already noted in the USA (Douches et al. 1996).
Employing SNP arrays for genotyping may lead to ascertainment bias in population research such as ours (Nielsen 2004; Albrechtsen et al. 2010; Heslot et al. 2013; Geibel et al.2021). SNP arrays often show an underrepresentation of SNPs with extreme allele frequency and are limited to the heterozygosity of the loci found using a limited panel of genotypes included in the development of the array. The 22K SNP array used in this study was developed using 569 unique accessions selected from all over the world, but with an emphasis on European cultivars (Vos et al. 2015). Several of the cultivars grown in Sweden today were included in the development of the SNP array (‘Bintje’, ‘Kuras’, ‘Sarpo Mira’, ‘Desirée’ and ‘Bionica’), and some of these cultivars are used as parents in the SLU breeding program. However, only one non-Nordic accession kept at NordGen was included when the array was developed (the US cultivar ‘Early Rose’). Hence, a lot of the genetic diversity present in this group may be unavailable when using this genotyping method.
In conclusion, the germplasm from NordGen and the potato breeding program at SLU seems to be closely related. There is a slightly larger variation spread among the breeding clones from SLU according to our genetic analyses. While the spread of phenotypes might be larger among the farmer’s cultivars kept at NordGen. The results generated will be of interest to potato breeders in Sweden and other countries of the Nordic region as they consider introducing accessions from NordGen to expand the genetic diversity in their breeding programs.