The experiment was conducted in the outdoor area of the Experimental Station of the Institute of Plant Nutrition and Soil Science, Kiel University, Germany (54°20′50″N, 10°6′55″E) starting in March 2021. Three oat (A. sativa L.) varieties (obtained from Saaten Union, Niedersachsen, Germany), Zorro (black oat; A. sativa var. nigra), Symphony (white oat; A. sativa var. alba), and Apollon (yellow oat; A. sativa var. aurea), were grown to maturity in large-scale containers (height, 0.9 m; area, 0.16 m²; volume, 120 L; see also Hohmann et al., 2016) with a seeding density of 300 seeds per container, which were later thinned to 90 plants after emergence. As a substrate, a subsoil (Cambisol; IUSS Working Group WRB, 2015) from the experimental station “Hohenschulen” of Kiel University, Germany, and arable topsoil from the district of Ost-Holstein (Schleswig-Holstein, Germany) were selected (see details in Table 1 ).

The containers were filled with air-dried and homogenized soil as follows (from bottom to top): 1) 20 kg gravel as a drainage layer, 2) 100 kg subsoil + sand (1:1, w/w), 3) 30 kg subsoil + topsoil (1:1, w/w), 4) 10 kg topsoil plus fertilizer according to standard application for oats [in kg/ha: 100 N (split into N1 prior to seeding and N2 at shooting stage), 55 P, 80 K]. Weed and pathogen control were applied as required.

Soil plant analysis development (SPAD) values were measured on the fifth leaf after waterlogging at BBCH 31 and on the flag leaf after stress treatment at BBCH 51 (Meier, 2001). An average of 10 readings per container was taken using a chlorophyll meter (SPAD-502, Konica Minolta Sensing Europe B.V., Wrocław, Poland).

Soil moisture was maintained at 60% water-holding capacity (WHC) until treatments started. While the respective controls (W0) were watered at 60% WHC throughout the entire crop cycle, waterlogging (100% WHC) was imposed for a total of two consecutive weeks: 1) W1 = early waterlogging at BBCH 31 and 2) W2 = late waterlogging at BBCH 51. Water treatment was checked every 2 days, and re-irrigation was performed based on weight loss if necessary. After terminating waterlogging, water was drained to achieve a target WHC of 60%, which was then retained until harvest. The experiment was set up with four replicates per treatment and oat variety in a completely randomized design (CRD). Randomization of the position of containers was performed twice a week together with the check of WHC.

Two weeks after terminating waterlogging (W1 and W2), 30 plants (including side shoots) were randomly selected and harvested, and fresh weights were recorded. At maturity, the 30 remaining plants (including side shoots) per container were harvested and separated into straw and panicles. Subsequently, the biomass of straw and panicles, grain yield, and yield parameters were quantified. Panicles per container were counted and hand-threshed to determine total grain and thousand kernel weight. The number of grains per panicle was calculated as follows (Equation 1):

To record dry weights, samples of each treatment were oven-dried at 60°C to constant weight and subsequently milled (Cyclotec 1093, Foss Tecator, Höganäs, Sweden) to fine powder for further analysis.

For mineral nutrient analysis, 200 mg of finely ground plant material of each plant part per replicate was digested with 10 mL 69% HNO₃ in a microwave oven (1800 W, MARS 6, Xpress, CEM, Matthews, MC, USA) at 190°C for 45 min and subsequently analyzed by inductively coupled plasma–mass spectrometry (ICP-MS; Agilent Technologies 7700 Series, Böblingen, Germany) according to the method described by Jezek et al. (2015).

Determination of total N was conducted using a CNS elemental analyzer (Flash EA 1112 NCS, Thermo Fisher Scientific, Waltham, MA, USA), for which 5–10 mg of finely ground plant material was weighed into tin capsules. Results were validated using a certified wheat flour standard (Isotopenstandard Weizenmehl, IVA Analysentechnik, Meerbusch, Germany) as a reference.

Data were statistically analyzed using SPSS software (version 25.0). The analysis was based on four replicate containers per treatment set up as CRD. The effects of treatments per cultivar were tested using one-way ANOVA according to Duncan’s (homogeneity of variance) or Games–Howell (heterogeneity of variance) multiple-range tests at p ≤ 0.05. Significant differences are indicated by different letters. The significance of the correlations was tested using two-tailed Pearson’s correlation coefficient at p ≤ 1%.

After 14 days of waterlogging at BBCH 31 (W1), all oat varieties clearly showed stunted growth and beginning chlorosis at older leaves (see Supplementary Figure 1 ). All oat varieties were similarly affected and showed a significant reduction in fresh weight of 58%, 57%, and 53% for black, white, and yellow oats, respectively ( Figure 1A ). Except in white oats, dry weight was not significantly reduced compared to the corresponding non-stressed control (data not shown).

At the second sampling date, 1 week after the late waterlogging event (W2), differences in the recovery potential between varieties became obvious ( Figure 1B ). While black oats were able to recover from early waterlogging, white and yellow oats still showed significantly impaired growth 6 weeks after water drainage. Interestingly, late waterlogging at BBCH 51 had no negative effect on the total fresh weight of all oat varieties and could maintain weights similar to the corresponding control ( Figure 1B ).

These results were largely confirmed at the final harvest ( Figure 1C ). While the recovery of white oats after the early waterlogging was not confirmed till maturity, fresh weight increased to the level of control for black and yellow oats ( Figure 1C ). However, the fresh weight of all oat varieties remained unaffected by late waterlogging.

SPAD values were measured always right after the termination of the waterlogging treatment. Significant differences were monitored between control plants and plants waterlogged at BBCH 31 for all three oat varieties ( Figure 2A ; Supplementary Figure 2 ). This waterlogging-induced decline in SPAD values was even more pronounced after stress treatment at BBCH 51 for white and yellow oats when compared to early waterlogging and control, while SPAD values in black oats remained unaffected by late waterlogging ( Figure 2B ; Supplementary Figure 3 ).

Whether the fresh weight was now broken down into the individual main major yield components, differences between oat varieties became obvious. While the number of panicles of black oats showed a slight but non-significant reduction at W1 and W2, white and yellow oats showed a significant reduction ( Figure 3A ). Concerning the number of grains per panicle, in black and white oats, W1 had a negative effect, leading to a reduced number of grains, while yellow oats were not influenced ( Figure 3B ). However, after W2, white oats compensated for the reduced number of panicles with the number of grains per panicle on the level of the control. Similarly, also, black and yellow oats significantly increased the number of grains after late waterlogging compared to W1 and remained on the level of their respective control ( Figure 3B ). Contrary to this, thousand kernel weight was not responsive at all to W1 and W2 in either black or white oats, while it was on a relatively high level in yellow oats compared to the other two varieties, but with a reduction after W2 ( Figure 3C ). This in turn led to an unchanged grain yield under W1 for both black and white oats, while under W2, there was even an increase in grain yield for both varieties ( Figure 3D ). However, yellow oats were the only variety that reacted sensitively to early waterlogging but could regain grain yield at least on the level of the control after late waterlogging ( Figure 3D ). Although differences in the major yield parameters were recognizable, the harvest index and the grain:straw ratio were unresponsive to both timings of waterlogging (data not shown).

Correlating yield with the single yield parameters showed that there was no correlation between yield and the number of panicles per container and thousand kernel weight for all three oat varieties, with all coefficients of determination being non-significant ( Supplementary Figures 3A–C, G–I ). However, yield significantly correlated to the number of grains per panicle at least for black oats (R ² = 0.735; Supplementary Figure 3D ) and white oats (R ² = 0.779; Supplementary Figure 3E ). Only yellow oats lacked such a correlation between yield and number of grains (R ² = 0.118, Supplementary Figure 3F ).

Early waterlogging (W1) resulted in a reduction of nitrogen (N) concentration in all three oat varieties. Hereby, the decrease was the highest in white oats (55%) followed by yellow oats (44%), and the lowest was in black oats (36%) ( Figure 4A ). Similarly to N, all oat varieties showed a strong decline in phosphorus (P) concentration, with white oats being most responsive compared to yellow oats and black oats ( Figure 4B ). Additionally, also S showed a marked decrease after early waterlogging (data not shown).

After late waterlogging, all oat varieties were able to recover, showing a N concentration similar to their respective control ( Figure 5A ). Furthermore, no effect of W2 on shoot N concentration could be determined. Similarly, the P status could be restored to the control level (black oats) or even increased (white oats and yellow oats) till BBCH 51 ( Figure 5B ). However, similar to early waterlogging, W2 led to a significant decrease in P in all three oat varieties ( Figure 5B ).

At the timepoint of maturity, all plants were harvested, and nutrient concentrations in straw and grains were determined. With respect to straw N, it was observed that both black oats and white oats showed no changes in N concentration ( Figure 6A ). Only in yellow oats was a significant difference between W1 and W2 measurable, whereas no significant difference between W2 and the respective control was obvious. Likewise, also in grains of black and white oats, no effect of either W1 or W2 could be detected on N concentration ( Figure 6B ). However, yellow oats showed an increase in N at maturity when waterlogged at BBCH 51.

A similar pattern was measured for P concentration. Black oats were unable to regenerate the straw P level at W2, while in white and yellow oats, P concentration increased to the level of the well-drained control ( Figure 6C ). However, in grains of black and yellow oats, no waterlogging effect was measured ( Figure 6D ), but a significant increase in P in white oats at W1 and W2.

Furthermore, N and P concentrations in total shoot biomass as well as in grain at harvest were not effective on grain yield, with all coefficients of the determination being non-significant ( Supplementary Figure 4 ).

Waterlogged plants usually show wilting and development of chlorosis especially of the older basal leaves (Arbona et al., 2008). Also, Wollmer et al. (2018a) reported chlorosis formation on older leaves and even spot necrosis in winter wheat, which they explain as the oxidation of cell membranes by reactive oxygen species (ROS) formation and their reduced detoxification under waterlogging (Tan et al., 2008). Generally, chlorophyll reduction can be accredited to oxygen deficiency-induced changes in plant metabolism, promoting overproduction of ROS, mainly H₂O₂, and thus photooxidative damage of chloroplast (Yordanova et al., 2004; Zhang et al., 2015; Ren et al., 2016; Hasanuzzaman et al., 2017). As a consequence, photosynthesis will be decreased and thus biomass production (Zeng et al., 2020; Pais et al., 2023). However, even though SPAD measurements confirmed a decrease in chlorophyll in this study after early waterlogging ( Figure 2 ), all tested oat varieties showed no distinct chlorosis but slightly brighter color compared to non-stressed plants (see Supplementary Figure 1 ). This is in contrast to wheat or barley, as oats have the capability to become less chlorotic and stay green even under waterlogged conditions, which gives an advantage to tolerate transient water stress (Setter and Waters, 2003). However, even though reduced biomass production under waterlogging is associated with lower photosynthetic activity (Ashraf, 2012), it is more likely a consequence of disturbed water and mineral uptake, rather than a photosynthesis effect (Colmer and Greenway, 2010; de San Celedonio et al., 2017).

As other crops (e.g., de San Celedonio et al., 2014; Ploschuk et al., 2018; Arduini et al., 2019; Hussain et al., 2022), oats also respond with an initial reduction in shoot growth especially when waterlogged in an early developmental phase ( Figure 1A ; Supplementary Figure 1 ). In agreement with Watson et al. (1976), growth reduction in oats under waterlogging was more pronounced when applied in an early growth stage and must be ascribed to a reduced or damaged root system. Notably, a loss in seminal roots and death of seminal root apical meristem were described, e.g., wheat (see Herzog et al., 2016). Ploschuk et al. (2023) also showed that root mass density was significantly reduced after early waterlogging in wheat, barley, oilseed rape, and pea, triggered by a lack of oxygen and the formation of ethylene. Likewise, also in oats, an initial decrease in shoot dry weight of 40% was explained by a decline in root dry weight of 50% (Watson et al., 1976; Cannell et al., 1985).

Hampered root growth is always accompanied by a decline in nutrient uptake, subsequently contributing to growth reduction. This effect is further triggered by a drop of redox potential and changes in pH in soil, also affecting nutrient transformation and availability, i.e., N and P (Patrick and Mahapatra, 1968; Hasanuzzaman et al., 2017, and literature within; Kaur et al., 2020b, and literature within). Such an effect is also shown in this study: nitrogen concentration dropped in W1 plants in all oat varieties ( Figure 4A ), corresponding to SPAD data ( Figure 2 ), indicating a transient undersupply in the shooting stage. These results are in agreement with Arduini et al. (2019) for oats, Ren et al. (2017) for maize, Zhou et al. (1997) for oilseed rape, and Wollmer et al. (2018a) for wheat. In soil, nitrogen concentration, i.e., nitrate, will be decreased by several processes under waterlogging, such as runoff, denitrification, or nitrate leaching (see Kaur et al., 2020b, and literature within). However, decreasing nutrient concentrations in the vegetative shoot tissues can be explained not only by reduced root growth but also by inhibited uptake mechanisms. As plants switch to anaerobic respiration, they lack ATP, a necessity to drive ion uptake and xylem loading mediated by H⁺-ATPases (Colmer and Greenway, 2010; Elzenga and van Veen, 2010). Such decline in N concentration was not prominent in W2 plants ( Figure 5A ), which is attributed to the N fertilization (see Section 2.1) performed right after drainage of W1. This N dose served as a “post-waterlogging rescue N fertilizer” (Watson et al., 1976; Rasaei et al., 2012; Kaur et al., 2020a), leading to a regeneration of N status, which could be maintained until maturity in both straw and grain ( Figure 6A ). However, among all temperate cereals, oats must be considered as the crop with the greatest ability to regenerate from waterlogging (Watson et al., 1976; Cannell et al., 1985; Solaiman et al., 2007). Setter and Waters (2003) suggested that this is due to an extensive formation of aerenchyma, which coincides with increased root porosity (Herzog et al., 2016). Also, Solaiman et al. (2007) described an increase in root porosity from 6% (well drained) to approximately 20% (waterlogged) in adventitious roots of oats compared to 2% in seminal roots. By this, O₂ in roots is kept high, allowing the roots to maintain aerobic respiration and high ATP levels and thus improve nutrient uptake characteristics (Colmer and Greenway, 2010; Takahashi et al., 2014).

As the redox potential drops, the solubility of P increases due to a loss of sorption sites. This in turn leads to a higher pore water concentration (Patrick and Mahapatra, 1968) and thus plant availability and uptake. However, like N concentration, P concentration also declined at W1 ( Figure 4B ), indicating a period of deficiency. As this was not expected, it must be reasoned that 1) P either leached down (Smith, 2020; Rupngam et al., 2023), 2) P retention in soil was increased due to sorption and/or precipitation with free Fe (Patrick and Mahapatra, 1968; Smith, 2020; Rupngam et al., 2023), or 3) uptake is inhibited, as the limited internal energy under waterlogging is directed to internal pH regulations and transport of solutes involved in anaerobic respiration (Greenway and Gibbs, 2003). This effect was reversed at W2 for all oat varieties ( Figure 5B ), indicating a full regeneration of the P status in W1 plants, which can be ascribed to an increased uptake due to higher available P and uptake in submerged soils. However, submergence at W2 again led to an undersupply of P in all three oat varieties. However, these were only of a transient nature in white and yellow oats ( Figure 6C ). Even though it seems that the time span till maturity was not enough for full recovery in black oats, a dilution effect must be assumed, as P content was on the level of control for all waterlogging treatments in all oat varieties (data not shown), though grains were not affected at all by all waterlogging events ( Figure 6D ). This is in line with Cannell et al. (1985), who also found no differences in N and P concentrations at harvest between treatments.

Although much research was conducted on various crops, such as wheat (e.g., de San Celedonio et al., 2017; Ploschuk et al., 2018; Wollmer et al., 2018a; Pais et al., 2023), oilseed rape (e.g., Wollmer et al., 2019; Hussain et al., 2023; Zhu et al., 2023), barley (e.g., Masoni et al., 2016; de San Celedonio et al., 2017), or maize (e.g., Tian et al., 2019; Liang et al., 2020), there are hardly any data available on oats’ response to waterlogging, especially regarding cultivar differences, the existence of flooding-related quantitative trait loci (QTLs), or -omics data on waterlogging and associated O₂ deprivation (Mustroph, 2018).

While yield decreases in a range of a few percent up to an almost total loss are reported, a meta-study by Tian et al. (2021) revealed that approximately 3% of all database samples showed contrasting behavior, thus increasing yield. One explanation for this phenomenon is the capability of such crop varieties to tolerate time periods of waterlogging. Thereby, it plays a crucial role in which developmental stage waterlogging occurs and for how long crops are submerged. In this study, it was found that oats, for the most part, are characterized by a high tolerance to both early and late waterlogging. While the grain yield of black and white oats was unaffected by early waterlogging and even increased after late waterlogging ( Figure 3D ), only yellow oats were sensitive to early waterlogging. However, the grain yield reduction of yellow oats after late waterlogging was only slightly but non-significantly reduced ( Figure 3D ). Such high recovery rates, at least as shown for black and white oats, are also consistent with the few published data for oats. For example, Watson et al. (1976) showed that when waterlogging ceased, oats recovered better than, e.g., wheat or barley. They reported that especially ear emergence was more delayed in these crops, which was even more pronounced at very early waterlogging (2 weeks after seeding) or when seeding was already delayed, shortening the recovery phase and leading to less grain per ear. In contrast, similarly to waterlogging at BBCH 51 in this study, waterlogging 6 weeks after seeding or at ear emergence was of minor effect, which was also confirmed for winter wheat (Watson et al., 1976; Cannell et al., 1985). Also, other studies, e.g., on wheat, report that early reproductive states are more adversely affected than tillering stages (Setter and Waters, 2003). However, this contradicts the results of Wollmer et al. (2018a), who showed the highest yield reduction of wheat after waterlogging in the generative phase.

In oats, the by far largest reduction in grain yield was observed when plants were waterlogged at the tillering stage, caused by the formation of smaller grains. This effect was almost completely eliminated, when N was applied (Watson et al., 1976). This is in agreement with this study, in which the grain yield of W1 and W2 plants was similar or even increased in the case of black and white oats ( Figure 3 ). Only in yellow oats was the speed of grain yield recovery somewhat slower but could reach a value comparable to the control after W2. Similar results were found for winter oats (Cannell et al., 1985), where tillering was reduced but could be reversed by N application. Reductions in tillers though were not found in this study at W1, rather than an increase for white oats (data not shown), probably compensation grain yield reductions. Only white and yellow oats at W2 showed a reduced number of panicle-bearing tillers, but this was also reversed and did not affect grain yield due to compensation by a distinct increase in grain number per panicle in yellow oats ( Figure 3 ). A reduction in kernel weight of 9% in oats, as reported by Cannell et al. (1985), could thereby only be confirmed in yellow oats ( Figure 3 ), while the other two varieties did not show any change compared to the well-drained treatment. In comparison, under similar conditions, for wheat, a reduction of 10%–30% was reported (Cannell et al., 1985; Ploschuk et al., 2018), indicating the high recovery potential of the tested oat varieties in this study. However, the observed reduction is not caused by a reduced number of grains per panicle rather than the reduced number of panicles in total.

In complete contradiction to the already discussed studies are data by Arduini et al. (2019). Similar to this study, oats were waterlogged at tillering after seeding in spring, differently from Watson et al. (1976), who used winter oats. This difference has of large effect on the regeneration period; while winter oats had a prolonged phase of 118 days after draining, the recovery phase in spring oats is much shorter. Therefore, Arduini et al. (2019) argued that higher temperatures of 20°C during waterlogging could in part be responsible for the higher sensitivity observed in their study. Although after 14 days of waterlogging a not yet significant decrease in harvest index became obvious, A. sativa compared to Avena byzantina showed a 79% and 83% reduction in grain yield, respectively, resulting in a decrease in harvest index of 8% and 10%, respectively, after 35 days of submergence (Arduini et al., 2019). Similarly, in the present study, the harvest index remained on the level of the respective controls for all three oat varieties (data not shown) at W1 and W2 after only 14 days of waterlogging, which may go back to an increased tiller fertility (Watson et al., 1976; Cannell et al., 1985).