RT-qPCR is a widely used method for gene expression analysis because of its relative simplicity and high sensitivity. However, the reliability of the results is greatly determined by reference gene selection used for normalization [5,8]. Ideally, the stability of the reference gene should not be influenced by the experimental conditions. Nevertheless, several studies point out, that the mRNA transcript levels could be affected by tissue type [11,12], developmental stage  treatment type [19,20] and genotype  as well. Furthermore, the expression stability of the same reference gene may also depend on the investigated species . Therefore, it is always necessary to analyse and validate the potential reference genes prior to their applications.
In our study, 9 candidate reference genes (ACT, TUB, CYP, GAPD, UBC, EF1, TBP, ADPR, PGD) were screened from the leaves and roots of two oat genotypes under different abiotic stresses (drought, salt, heavy metal, cold and heat), and their expression stabilities were analysed by 4 different statistical programs (GeNorm, NormFinder, Bestkeeper, RefFinder). The general rating by the different programs had a substantial agreement, which were the least stable genes for each treatment/tissue/genotype combinations so they could be easily excluded. However, the most stable gene determined by the applied programs were not always the same due to their different calculations. It was especially remarkable when using BestKeeper, which gave a higher ranking to a certain gene compared to the other programs in some cases. That was especially true for TBPII in drought stressed roots of Mv Hópehely, and for PGD in cold stressed leaves of both cultivars, respectively. The difference between the ranking order of Bestkeeper and the rest of the softwares is also mentioned by other studies [22–24]. The stability of the reference genes were influenced by all the experimental conditions, such as tissue type, genotype and treatment type, but the applied stress treatments had the most pronounced effect on stability in general.
For drought stress, all the stability values were very low, independently from the applied program which indicates that gene expression stabilities of the reference genes was not particularly affected by this treatment in general. The suitability of ADPR under drought stress was confirmed by our experiments and it was also mentioned earlier in barley (Hordeum vulgare, L.) . Furthermore, CYP was found to be stable in barley when exposed to drought, while in our experiment it was one of the least stable genes. Interestingly, a study in durum wheat (Triticum durum, L.) identified GAPDH as a stable reference gene under drought , although, under our experimental conditions it only showed high stability in roots, while it was unstable in the leaves of the investigated oat cultivars. In addition to our study, the applicability of ACT as reference gene in drought stressed leaves was also suggested in barley . Interestingly, the stability values of UBC were genotype and tissue type dependent. Namely, it was the least stable gene in the leaves of Mv Pehely, while it was one of most stable ones in the leaves of Mv Hópehely. Furthermore, it had moderate stability in the roots of Mv Pehely while it was stably expressed in the roots of Mv Hópehely.
When applying salt stress, tissue type greatly affected the expression stability. GAPDH, CYP and PGD had high stability in leaves, while in roots UBC and ADPR were the most stably expressed genes in both cultivars. GAPDH was a proposed reference gene in Triticum durum  under salt stress as well. However, Duan et al. (2020) also investigated the effects of salt stress in oat and they found the expression of GAPDH unstable. In contrast, the same study identified EF1 and TBP as the best combination for normalization, but in our experimental setup both genes had relatively low stabilities. ACT exhibited low expression stability in oat both in our study and in the work of Duan et al. Furthermore, α-TUB was also in the end of ranking order in our analysis and in the above mentioned study as well. The expression stability of UBC showed genotype and tissue type dependence under salt stress similarly to drought stress, which is also an osmotic stress.
Under Cd stress, GAPDH was found to be stably expressed according to the different evaluating softwares in our study. However, in giant reed (Arundo donax L.), which also belongs to the Poaceae family, GAPDH was amongst the least stable genes when the plants were treated with Cd . ACT1 showed high stability in leaves of both investigated oat cultivars. In agreement with this, another members of the Actin gene family showed high expression stabilities under Cd exposure, ACT12 in switchgrass (Panicum virgatum L.)  and ACT3 in soybean (Glycine max L.). PGD was a stable reference gene in the roots of both oat genotypes, however, in soybean its expression was greatly influenced by Cd stress . TUB exhibited low expression stability in our oat samples and in Cd-treated soybean plants  as well.
When looking for stably expressed genes under temperature stresses, two oat homologs of wheat ADPR (Ta2291) and PGD (Ta30797) were also tested, since they were suggested as suitable reference genes candidates by Paolacci et al. (2008). The overall applicability of ADPR for cold and heat stresses in different genotypes, tissue types and growing media was confirmed since it was in the first three places of the ranking order calculated by every software with the exception of Bestkeeper. The high expression stability of ADPR under cold and heat stress was also indicated in Hordeum brevisubulatum  and it was the best reference gene in cold stressed barley . However, PGD only exhibited high expression stability in the roots of Mv Pehely under cold stress, while it was less stable in leaves and under heat stress. The stability of ACT was influenced by genotype, tissue type and growing medium, namely, it showed the high stability in the leaves, when cold stress was applied in Mv Pehely, and under heat stress, but only when the plants were grown hydroponically. ACT is a commonly used reference gene in different plant species  but these differences suggest a caution approach when choosing this gene as reference gene. EF1 was usually amongst the least stable genes under temperature stresses with the exception of the heat stressed roots of Mv Pehely where it was the most stable gene. EF1 was also ranked the highest in terms of stability in Hordeum brevisubulatum under heat stress .
In order to validate the stability of the investigated reference genes, PAL expression analysis was performed. Our study revealed, as long as a reference gene with very high stability is chosen (Fig. 5), one reference gene may be sufficient for accurate normalization. However, according to the GeNorm V/V value calculation the application of 2 reference genes are more suitable in leaves or in roots exposed to a certain abiotic stress. As it is shown in Additional file 2 on Fig. S1-S4, the expression of PAL was highly inducible by most of the treatments, especially in leaves. However, drought stress might be considered as a mild stress in this study, which is supported by the fact, that PAL expression was not induced by PEG treatment nor in leaves or roots of the two genotypes. Correspondingly, the reference gene stability values calculated by the different software were very low in the case of all reference genes tested compared to the values under other abiotic stresses. In contrast, salt treatment, heavy metal exposure and low temperature caused a remarkable elevation of PAL transcript levels in both genotypes, which indicates increased stress effect for the plants. Accordingly, the stability of the reference gene candidates changed to a higher extent depending on the individual genes. Interestingly, heat stress only induced PAL expression in the leaves of Mv Hópehely when grown in soil but not in hydroponic culture; however, this treatment greatly affected reference gene stability in general. In agreement with this observation, the growing media influenced the stability of certain reference genes as well. While ADPR kept its high stability under both growing conditions as indicated by the different software, the ranking order of the rest of the reference gene candidates changed according to the growing medium. For example, ACT showed high stability in the leaves of heat stressed Mv Pehely in hydroponic culture, however, in soil it was less stable. Furthermore, when investigating the stability in the cold stressed leaves of Mv Hópehely, CYP was only stable in soil but its expression was influenced by cold temperature under hydroponic growing conditions.