1. Introduction
Plants are necessary for maintaining human life and provide nutrition for an increasing population, and authentic germplasm is imperative for executing reliable and repeatable research as well as for breeding new varieties with particular characteristics. Thus, effectiveness of research and breeding depend on the use of true genotypes.
Intraspecific genetic diversity of crops represented by wild resources, landraces and bred varieties is mostly preserved ex situ. Gene banks implement standard procedures to attain accuracy and reliability [
1,
2], but human mistakes during seed increase and storage are inevitable and accumulate over a period of years. This is particularly so when handling many accessions [
3,
4,
5]. Errors such as confusing identities, duplication or misreading of accessions, all denoted here as mislabeling, result in total non-authenticity of GB accessions and mechanical genotype contamination or outcrossing in their partial non-authenticity. When these are genotyped or sequenced [
6,
7], faulty genetic data can be linked to mislabeled genotypes, resulting in incorrect or biased results or reduced power to detect real biological patterns [
8]. Therefore, control of germplasm identity and genotype purity must be a high priority in research, breeding and conservation activities [
9]. However, failure to publicize such discrepancies, especially in cereal GB accessions, means that it is often difficult to recognize aberrant genotypes.
Resistance genes against diseases present in commercial varieties have been usually identified during breeding, trialling or soon after their registration. It is often based on the gene-for-gene model [
10,
11], which uses pathogen diversity and is designated as gene postulation [
12]. It is a classical but still very effective approach that compares the responses of tested varieties with standard host lines containing known resistance gene(s), both inoculated with a set of pathogen isolates. Such genotype characteristics are designated as reaction patterns [
13], resistance spectra [
14], profiles [
15], response type arrays [
16] and, in this report, as infection response arrays (IRAs).
Europe is the largest producer of barley (
Hordeum vulgare L.) and, since powdery mildew caused by
Blumeria hordei, M. Liu and Hambl., was a predominant disease, breeding resistant varieties was a priority and these possess many major genes of specific resistance and their combinations. An effort to collect information about the genetic basis of varietal resistance led to the identification and publication of the resistance genes in almost 700 European varieties [
17]. The primary use of resistance genes is protecting plants against diseases or pests. However, knowledge of these genes in a host has wide utilization in research and breeding. For example, current results of resistance gene identification can be compared with existing original data and on this basis varietal authenticity can be detected and established.
Non-authentic genotypes were a source of complications in our study of powdery mildew resistance genes in winter barley varieties [
18], and verification of correct genotypes and detection of non-authentic genotypes became an important topic of our research. When we screened a collection of the wild progenitor of cultivated barley (
H.
vulgare subsp.
spontaneum) lodged in an international GB, a great heterogeneity of accessions was uncovered [
19]. The Czech (CZE) GB core collections of bred varieties and landraces of spring and winter barley were subsequently studied [
16,
20] and when results were compared with older data many non-authentic and heterogeneous accessions were found for reasons outlined above. In a later investigation a smaller set of spring barley varieties originating from seven foreign GBs was pathologically screened and the results compared with those from identically labeled varieties maintained in the domestic GB; 37.5% of the accessions were heterogeneous, and at least 20.0% were mislabeled [
21].
The aim of this work was to detect non-authentic genotypes in a set of winter barley varieties originating from numerous GBs. A large number of genotypes was grown, harvested and tested, their powdery mildew resistance genes were postulated, and based on these results non-authentic genotypes were identified.
3. Discussion
Among the commonest reasons for varietal non-authenticity maintained in GBs are (1) mislabeled accessions, (2) contamination with other genotypes, or (3) reselection and resultant homozygosity of heterogeneous varieties. As there are a large number of genotypes analyzed here, only a few model examples can be highlighted.
Non-authentic accessions resulting from varietal mislabeling. Mislabeling of varieties in GBs can occur during the handling of seeds and it was found in at least 40 out of 289 accessions (
Table S1). However, in some cases determining the authenticity of accessions can be difficult to disentangle as in the case of Anson. This variety was obtained from three GBs (
Table 3).
Accessions from CZE and USA GBs carried Mla8, while the accession from the British GB contained Mla7. Thus, accessions from two GBs carrying an identical resistance gene (Mla8) could be considered as authentic. However, Anson is a British variety and the accession from its domestic GB had a different resistance. The pedigree of Anson in the Czech GB is Triumph × Yamina, while in the GB of the USA is Volbar × Jefferson. Therefore, they must be different varieties both named Anson and with the same resistance gene. To confuse matters further, Anson obtained from the British and Czech GBs have identical pedigrees but different resistance genes. Therefore, based on resistance genes and pedigree it is not possible to state whether the accessions, their pedigrees or both are incorrect. Similarly, MlaLo was detected in 15 SSPs of Slaski II accessions from Czech, German and USA GBs, but in the heterogeneous accession of this Polish variety sent from the Polish GB MlaLo was not present.
Four SSPs of the Japanese variety Nakaizumi-zairai from the Czech GB were characterized by the absence of resistance genes (
none) and one SSP by the presence of
Mla8, while all SSPs from the German GB contained
MlRu2. Based on these results it would be impossible to decide which one is authentic. However, this variety is the standard for
Mlk2 [
22] and this gene was detected only in all three SSPs derived from the USA accession (
Table S1). Thus, both the CZE and DEU accessions cannot be accepted as authentic. A similar example relates to an accession of the Ethiopian spring barley landrace Abyssinian 1102—a known donor of durable Mlo resistance [
23]. All homogeneous accessions from GBR and USA collections contained Mlo, while in a heterogeneous CZE accession only lines with
Mla8 and without the resistance gene (
none) were found. Conversely, in two of five SSPs of the CZE accession of Diamant, and in three of five SSPs of this variety from a Slovakian (SVK) GB, Mlo was detected (CZE and SVK GBs have close cooperation), whereas in five SSPs of Diamant from the British GB the expected
Mla8 allele was present [
21].
It was surprising that, although cereal accessions rank among the most numerous in plant GBs, we found no references dealing with mislabeled varieties. Nevertheless, results from GBs of other plant species showed the problem is common and concerning. In yam (
Discorea), whose tubers are used as a staple food or for natural medicine, 20.6% of the total 3156 GB accessions were not true to type, i.e., misidentified individuals [
3]. Af Sätra et al. [
24] genotyped apples (
Malus domestica) and confirmed the identity of multiple accessions with the same variety name but also identified several mislabeled accessions. Shan et al. [
25] described the possibility of mislabeled accessions in the world germplasm collections of wild Cicer (chick-pea). Jreisat and Laten [
1] reported that 4 of the 13 specimens of clover (
Trifolium) appear to be mislabeled or misidentified. Van de Wouw et al. [
26] found a high degree of non-authentic lettuce (
Lactuca sativa) varieties in GB collections; this was especially true for the oldest varieties, but even for varieties released from the 1960s to 1990 it was estimated that approximately 10% were not authentic. Zhang et al. [
27] applied molecular methods to determine species’ identities in rice (
Oryza) and revealed that 17% of 53 seed accessions from GBs or field collections were mislabeled. Akpertey et al. [
28] analyzed 400 coffee (
Coffea canephora) genotypes and 18.6% were mislabeled. Parentage analysis showed that 33.3% of the progenies derived from controlled crossing and none of the progenies originating from an open pollinated bi-clonal seed garden had parents (both parents) corresponding to the breeders’ records. Most of results conducted in the plant kingdom and reporting mislabeling of genotypes relate to woody plants such as trees [
29,
30,
31] and shrubs [
32,
33,
34]. Although the situation in GBs of cereals is similar, relevant studies are rare.
Non-authentic genotypes through contamination. The degree of genotypic contamination of accessions can differ and range from just a single seed or plant to the complete replacement of the original variety with another genotype(s). Accession heterogeneity may be welcomed in order to maintain the natural variation within original landraces or varieties bred by crossing that have not undergone line selection. However, in barley GBs the number of modern and almost exclusively single-line varieties has increased and therefore their heterogeneity was not originally present.
Contamination of accessions arises from mechanical admixtures with other genotypes during operations associated with seed multiplication or accidental cross-pollination, both resulting in undesirable heterogeneity. Examples of genotypic contamination in three out of six accessions of three varieties are illustrated in
Table 4. In the first case (Freya) five SPPs from the domestic GB possess
Mla6, whereas two SSPs derived from the DEU GB do not contain this gene. Similarly, nine SSPs derived from two accessions of Virgo carry
MlLa,
Mlra but in one SSP a different combination of
Ml genes was found (a8, h). All these cases (SSPs nos. 2 and 4 of Freya from DEU and no. 2 of Virgo from CZE accession) are considered as random mechanical admixtures (genotype contaminants). Also, one of 10 SSPs of Duet (CZE and GBR) differs from the rest in one of four genes of an unusual gene combination. Therefore, it is probable that this different genotype could have arisen by accidental cross-pollination during seed increase, although a line of the original breeding population cannot be excluded.
A similar case to the above is the heterogeneous variety Bordia (
Table 5). Three accessions (DEU, NLD and USA) contained a genotype without a resistance gene (
none), which could prove the possible authenticity of at least a proportion of seeds of these accessions. Also, some SSPs in the accessions from the USA and GBR could be considered authentic, as
MlaLo predominated in both. However, if only accessions from German and British GBs were tested, the conclusion would be that these are different varieties as no common genotype was detected in them.
Landjeva et al. [
35] studied 91 Bulgarian wheat (
Triticum aestivum L.) genotypes mostly with microsatellite markers and approximately 74% of the varieties showed heterogeneity. Varshney et al. [
36] found heterogeneity in an average of 12.4% of 38 wild barley accessions (
H.
vulgare subsp.
spontaneum) and 5.7% of 185 cultivated genotypes. Ayala et al. [
37] analyzed 46 wheat landraces from Andalusia and up to 77 genotypes were found in them. Ertiro et al. [
38] studied 265 maize (
Zea mays) inbred lines from three GBs; only 22% of accessions were considered “pure” with <5% heterogeneity, while the remaining 78% had a heterogeneity ranging from 5.1 to 31.5%. In the germplasm collection of the ornamental crop
Hydrangea macrophylla, the authors found that 36% of the tested plants were mislabeled [
39]. Bairwa et al. [
40] analyzed 92 accessions of rice (
Oryza sativa L.) landrace germplasm with 23 phenotypic traits and 48 microsatellite markers and recorded a large degree of heterogeneity and numbers of heterozygotes in the germplasm.
To uncover heterogeneity depends on the reliability of the method used and may be underestimated here because Dreiseitl [
19] found 40.2% heterogeneous wild barley accessions and later more than 85% heterogeneous GB accessions of cultivated winter barley [
41] when tested with different numbers of powdery mildew isolates. In this report most of these varieties were tested and only 35.3% of accessions were heterogeneous. This great difference can be explained because in the previous work each accession was represented by about 50 plants, whereas less genotype contamination was uncovered herein since only five SSPs were tested. One might even hypothesize that if an accession of a homogeneous variety of the set is represented with about 100 SSPs and tested with the same set of the pathogen isolates then very few accessions will show full homogeneity, i.e., identical genotypes among all SSPs.
To achieve complete and durable genotype purity of GB accessions of plant varieties is a difficult task while frequent occurrence of non-authentic genotypes in GBs is a significant problem for germplasm users [
42]. Therefore, instead of direct use of seed or randomly selected plants from GB accessions it is preferable to prepare and verify authenticity of genotypically pure lines before use in research or breeding. A small number of seeds (10 to 20) should then be made available to ensure reproducibility of experiments [
43].
To detect heterogeneity of accessions may be easier than to uncover mislabeled varieties because individual plants or derived lines can be compared concurrently and differences easily seen, whereas accessions incorrectly labeled must be determined using analytical methods similar to those current at the time the variety was bred.
Non-authentic genotypes due to selection. In some GBs accessions can be strictly maintained in their original state, while in others a homogenization of accessions by negative or even positive selection is performed and can lead to gene erosion. Furthermore, a change of curator may modify the approach to maintaining seeds in a GB.
Examples of particular accessions comprising different lines probably through selection are presented in
Table 6, which shows resistance genes in nine accessions of two homogeneous varieties. In accessions of Frost from GBs CZE and DEU, a combination of two
Ml genes (
a6,
h) was postulated, while in accessions from GBs SVK and SWE one more gene (
a6,
h,
VIR) was found. This suggests that the original Frost was composed of two lines and progenies of both are maintained in different GBs.
Similarly, the accession Protidor (DEU) was characterized by Mla12, MlVIR, while the FRA, GBR and USA accessions and three of the five SSPs of the CZE accession additionally contained Mlg. In two other SSPs of the CZE accession only Mlg, i.e., the gene that was absent in the DEU accession, was detected. It is, therefore, likely that Protidor was originally formed by several lines carrying different resistance genes and their combinations, and different genotypes of this variety are maintained in GBs.
Genetically pure lines can be better characterized than heterogeneous accessions. However, selection results in loss of genes, reduced genetic diversity and increased genetic erosion [
44]. Since each gene is linked with other genes, then even selections within a variety might contain or lack powdery mildew resistance genes as well as other alleles that determine different traits. Therefore, if a variety was originally heterogeneous and a pure line was selected for the GB, then such an accession might have lost some of its original characteristics and authenticity.
We can conclude that use of resistance genes against diseases such as mildews and rusts in cereals is a very powerful and precise way for verifying the identity of accessions. Despite that, this method, as for most methods generally, has some limits and detection of non-authentic accessions can be underestimated. Here four groups of accessions carrying exclusively or predominantly genotypes with single Ml resistance genes a8, aLo and Ch, and also none (no resistance gene), were most frequent. If, for example, Cenad 450 and Oksamyt (both have MlaLo) are mixed together, then it is not possible to detect non-authenticity of these varieties using the resistance gene postulation method.
The underestimation of non-authentic accessions can also be partly attributed to the fact that genotypes which were more frequent were usually considered as authentic. For example, the 10 predominant genotypically identical lines in Anson were thought to be authentic whereas three lines from its domestic GB were designated as non-authentic. However, the converse interpretation would indicate that the number of non-authentic genotypes would be higher. Similarly, heterogeneity of accessions may be higher because we studied a maximum of only five lines (SSPs) from each accession and accessions with fewer different components may not have been detected.
Human activities including those in plant GBs are subject to errors resulting in confusion of items. Despite the excellent systems of reducing such mistakes in human medicine there are similar occasional errors in medical data [
45], vaccines [
46], blood accessions [
47] and patients’ identities, organs or body parts [
48] and inadvertent exchanges of eggs and sperm [
49], embryos [
50] and even born children [
51]. Thus, human mistakes cannot be avoided even in well-secured systems and GBs are no exception.
In this study we found a high proportion of non-authentic genotypes in barley GB accessions. This finding confirms our previous research as well as results of studies of GB accessions of some other plants. Mislabeled accessions or randomly selected incorrect genotypes from heterogeneous accessions result in the faulty characterization of varieties which can drastically devalue the results of subsequent research and cannot efficiently combine the required traits in breeding.
The problem with non-authentic accessions is general and GB staff must be scrupulous and consistently follow updated methodological procedures [
52] for maintaining authentic varieties. In the effort to reduce the number of non-authentic accessions, national gene banks could act as guarantors of the authenticity of their domestic varieties. To this end, they could request respective accessions from other gene banks and use standard varietal characteristics, appropriate methods [
53,
54] and available information to review their authenticity. No samples should be kept in only one gene bank, but conversely, no samples should be held in multiple gene banks. Nevertheless, it is unreasonable to expect complete authenticity of GB accession at least for the foreseeable future. Therefore, the responsibility for using authentic genotypes in research and breeding projects rests with scientists and breeders. Before doing experiments, they should obtain accessions from different sources (GBs) and verify the required varietal attributes with suitable methods.