Nitrogen availability and genotype affect major nutritional quality parameters of tef grain grown under irrigation

Worldwide demand for tef (Eragrostis tef) as a functional food for human consumption is increasing, thanks to its nutritional benefits and gluten-free properties. As a result, tef in now grown outside its native environment in Ethiopia and thus information is required regarding plant nutrition demands in these areas, as well as resulting grain health-related composition. In the current work, two tef genotypes were grown in Israel under irrigation in two platforms, plots in the field and pots in a greenhouse, with four and five nitrogen treatments, respectively. Nutritional and health-related quality traits were analyzed, including mineral content, fatty acid composition, hydrophilic and lipophilic antioxidative capacity, total phenolic content and basic polyphenolic profile. Our results show that tef genotypes differ in their nutritional composition, e.g. higher phenolic contents in the brown compared to the white genotype. Additionally, nitrogen availability positively affected grain fatty acid composition and iron levels in both experiments, while negatively affecting total phenolics in the field trials. To conclude, nitrogen fertilization is crucial for crop growth and productivity, however it also implicates nutritional value of the grains as food. These effects should be considered when fertilizing tef with nitrogen, to optimize both crop productivity and nutritional effects.


Results
Mineral content.  (Table 1). Compared to the brown tef, white tef had higher levels of Mg, Zn and Cu in the field experiment, while the brown genotype had higher levels of Mn in the pot experiment (Table 1). Tables 2 and 3 present the statistical analysis of the response of nutritional quality attributes in tef to N fertilization in the field and pot experiments, respectively. Field experiment results show that tef grain mineral levels were significantly affected by N fertilization. An increase was observed in Fe levels with increasing N, from 7.75 mg/100 gr DW in 0 ppm N to 10.7 mg/100 gr DW in 120 ppm N (Fig. 1A). A decrease was recorded in Zn levels, which were higher in 0 ppm (4.5 mg/100 gr DW) compared to in 60 and 120 ppm (4.15 and 4.12 mg/100 gr DW, respectively; Fig. 1B). as well as in Mg levels, decreasing from 197 mg/100 gr DW at 0 ppm N to 187 mg/100 gr DW at 120 ppm (Fig. 1C). Mn levels also decreased from 6.61 mg/100 g DW at 0 ppm to 5.78, 5.37 and 5.35 in 30, 60 and 120 ppm, respectively) (Fig. 1D). Ca and Cu levels did not change in the field experiment in response to N fertilization (Fig. 1E,F, respectively).
Fatty acid composition of tef was widely affected by N fertilization in both field and pot experiments. In the field, a significant increase was recorded in C18:2 with increasing N, from 41.3% in 0 ppm to 42.4% and 42.3% in 60 and 120 ppm, respectively ( Fig. 2A), as well as in C18:1(Z11) levels, from 0.69% in 0 ppm to 0.73% in 120 ppm (data not shown). Total PUFA contents also increased, from 46.6% in 0 ppm to 47.9% in both 60 and 120 ppm N Table 1. Average values (± standard error (S.E.)) for nutritional quality parameters in white and brown tef genotypes grown in the field and pot. Asterisk (*) stands for p < 0.05, for the specific p values please see Table 2 (for field results) or  2B). At the same time, a decrease was observed in contents of C16:0 (17.9% in 0 ppm to 16.8% in 30, 60 and 120 ppm; Fig. 2C), and C15:0 (0.024% to 0.016% and 0.017% in 0 ppm vs. 30 and 60 ppm, respectively; data not shown), as well as in total SFA, from 24.8% in 0 ppm to 23.8 in 30, 60 and 120 ppm (Fig. 2D). C20:0 fatty acid was not affected (Fig. 2E).
Antioxidative capacity. In this work, we used the method described earlier by Vinokur et al. to analyze both hydrophilic antioxidative capacity (HAC) and lipophilic antioxidative capacity (LAC) of the same sample 79 . HAC of tef grains varied between 1.28-3.26 µmole TE/mg DW in the white genotype and 1.17-3.22 µmole TE/ mg DW, with the brown genotype showing higher capacity than the white in the pot experiment. LAC varied between 1.73-2.0 and 1.52-2.66 µmole TE/mg DW in white and brown genotypes, respectively, with no difference between the genotypes (Table 1).  Fig. 3A), while lipophilic antioxidative capacity (LAC) significantly increased from 0.62 µmole TE/ mg DW to 1.58, 1.58 and 1.8 µmole TE/mg DW in 30, 60 and 120 ppm N (Fig. 3B).
In the pot experiment no effect on hydrophilic or lipophilic antioxidative capacity was observed (Fig. 3C,D, respectively).
Free, bound and total phenolic content. Total phenolic content (TPC), as well as free and bound phenolics content of tef grains have been published previously 43 . In the current study TPC ranged between 0.89-1.2 and 1.04-1.27 mg GAE/gr DW in grain of the white and brown genotypes, respectively (Table 1), with brown genotype showing higher content than the white genotype in both experiments (Table 1). Free phenolic content in this experiment was 2.02 to 2.24 and 2.02-2.14 mg RE/g DW in white and brown genotypes, respectively, higher than the content of bound phenolics, which was 0.25-0.73 and 0.37-0.93 mg RE/g DW in the white and brown genotypes, respectively, and did not differ between genotypes (in both field and pot experiments).
TPC decreased with increasing N in the field experiment, from 1.36 mg GAE/gr DW to 1.2 and 1.14 mg GAE/ gr DW in 30 and 120 ppm, respectively (Fig. 3E). Free and bound phenolic content was unaffected by N ( Table 2).
In the pots, TPC increased with increasing N (0.76 mg GAE/gr DW at 10 ppm vs 1.16 mg GAE/gr DW in 120 ppm; Fig. 3F), while free and bound phenolics levels were not affected by N (Table 3).
Phenolic profile. Phenolic profile of tef comprises mainly of phenolic acids, in addition to flavonoids 43 . In our samples we were able to identify six phenolic compounds, including three phenolic acids: genistic, p-cou- Table 3. Statistical analysis parameters for the response of nutritional quality attributes of tef grains to N fertilization in the pot experiment. All table results are presented as mean ± SE.    Table 1). The white genotype had higher levels of p-coumaric in both experiments, and higher level of rutin in the pot experiment, while the brown genotype had higher levels of vanillin and ferulic acid in both experiments.
In the field experiment, a significant increase was observed in the contents of p-coumaric acid with increasing N levels, from 5.83 µg/g DW in 0 ppm to 40.0 and 61.0 µg/g DW in 60 and 120 ppm (Fig. 3G), with a decrease in the contents of both rutin (172.0 and 216.3 µg/g DW in 0 and 30 ppm N to 77.0 µg/g DW in 120 ppm; Fig. 3H) and quercetin (from 1803.9, 1886.8 and 1621 µg/g DW in 0,30 and 60 ppm, respectively, to 934.2 µg/g DW in 120 ppm; Fig. 3I).

Discussion
N effect on crop performances and specifically quality parameters, has been described 40,41 , often for common commodities, e.g. potato, tomato, apple etc. At the same time, the results of the current work will be mainly compared to those reported on cereal, as these are more relevant to tef. Although millets and sorghum would make the best comparison, most of the available data refers to wheat grains, with some data on rice. Additionally, in this work tef grains from two growing platforms were tested-field plots and pots in greenhouse. It is generally accepted that both field and greenhouse results are specific to given environment and genetic background. However while the pot results reflect the crop potential, as the growing conditions are well controlled, the field results are more relevant to practical conditions. Furthermore, the field and pot experiments took place in different seasonal conditions, since seeds were sawn in different dates (see "Materials and methods" section). Therefore, their growth season was different, which can explain some of the variability in the response to N fertilization. Additionally, different response to N fertilization in perlite (pots) in comparison to soil originates in other factors, e.g. differences in ion exchange, water availability. Although in many cases the results were similar for both platforms, in some other they differed in either values or effect trends. However, it is important to mention that no contradiction was found within the two dataset. In addition, due to the different nature of these platforms, no attempt was done to statistically compare them, as such comparison will not be informative due to the wide array of possible variability sources.
The mineral content of tef grain is higher than that of most staple cereals used in western nutrition, with high levels of Ca, Fe, Cu, Mg and Zn 9 . Although mineral composition of tef grains has been published, only a few reports show detailed data regarding both white and brown tef genotypes, with some only specifying a general range of values. Our results were generally in agreement with previously published data, with some differences between genotypes. For Ca, levels of 124 and 155 6 or 17-124 and 18-178 mg/100 gr 25 , were reported for white and brown tef grains, respectively. Our results were in this range in the pot experiment, and slightly higher in the field experiment, with no apparent difference between genotypes. For Mg, this work is the first to present contents of both white and brown tef, which is in agreement with available data of 184-200 mg/100 gr 13 , with white genotype showing higher levels than brown in the field. Fe levels were reported as 37.7 6 , 31.6 44 or 15.9 mg/100 gr 45 and 24.6 mg/100 gr 45 for white and brown grains, respectively. Previously reported high levels of above 150 mg/100 gr can probably be attributed to soil contamination, as suggested 6 . Our results are slightly below this range, with no difference between genotypes. Zn content was reported as 2.86-4.02 6 and 2.4-6.8 and 2.3-6.7 25 mg/100 gr in white and brown tef, respectively, and our results are well within this range for both genotypes in the field, and higher in white genotype compared to brown, and slightly higher in the pots. For Mn content, this is the first report on different genotypes, and in the field results both genotypes were in the range of values published for tef of 56.5 ppm 11 and 3.8 mg/100 gr 9 , in the pot experiment both genotypes showed higher levels, with brown genotype higher than white. As for Cu, white and brown genotype content was reported as 2.5-5.3 and 1.1-3.6 mg/100 gr, respectively 25 . However, we found much lower levels, with white genotype containing higher levels than brown in the field.
Grain Zn levels in wheat were reported to positively respond to N supply when Zn levels in soil and tissues are sufficient, although results in rice showed that this effect depends on initial seed Zn levels and yield capacity 46 . This might explain the observed mixed trend of a decrease in the field plants, alongside the increase in the pot plants. For Fe, N fertilization was reported to positively affect both acquisition and grain allocation in wheat 47 , showing the same trend as in our data. Likewise our results, Mn was reported to remain unchanged in wheat grains in response to increasing N. Cu was reported to increase in wheat grains 48 but decrease in rice 49 . Likewise our data, Mg levels decreased and Ca increased 49 . There seems to be an agreement across works in various cereal regarding the increase in Zn and Fe with increasing N, however in regard to other minerals published data is generally inconsistent, thus implying that other genetic and environmental factors are involved in plant mineral content responses to N fertilization.
In this work, we report for the first time a detailed fatty acid composition of tef grains, for two genotypes. Lipids in tef are nutritionally important since they play a role in baking-related qualities, binding to gluten/non gluten proteins and affecting the bioaccessibility of polyphenols in bread 50 . Fatty acids are also important for Scientific RepoRtS | (2020) 10:14339 | https://doi.org/10.1038/s41598-020-71299-x www.nature.com/scientificreports/ sensory properties of the final baked products, contributing to texture and taste. In addition, as tef is fermented to make injera, initial fatty acid composition of the fermented substance may affect its fermentation properties, as was reported for other fermented foods, e.g. beer 51 and olives 52 . Tef fatty acid profile generally resembles that of other cereal, with the PUFA C18:2 as the main fatty acid, followed by C18:1 and C16:0 53 , and trace amounts of longer C20:0 and C22:0 fatty acid like other millets 54 and quinoa 55 . Two available reports only generally described fatty acid composition of white tef grains, showing slightly different profiles: an older work 14 showed that oleic acid content (32%) was higher than that of linoleic acid (24%), while our results are in agreement with those of Hager et al. 42 , where linoleic acid content (50%) is higher than oleic acid (29.5%) 42 . High levels of unsaturated fatty acids are nutritionally desirable, due to their positive health effects 42 . Specifically, the presence of linoleic (C18:2) and α-linolenic (C18:3) acids is valuable, being essential fatty acids not synthesized by the human body. Tef profile is unique in containing higher levels of C18:3, in addition to odd-chain fatty acids, C15:0, C17:0 and C21:0, not commonly found in nature. Consumption of odd fatty acids was correlated with unfavorable health effects 56 . Nevertheless, these fatty acids can be utilized by bacteria through the a-oxidation pathway 56 , and thus fatty acid composition may be important for the fermentation process. Since non-digestible components of the cereal matrix may also serve as prebiotics 57 , these might also contribute to tef gut-microbiota benefits, working either as a pre or probiotic substance.
Although fatty acid amount might seem small and non-significant, cereals actually make a significant contribution to essential fatty acid consumption, being consumed in large amounts 58 . Throughout our data, the white genotype was consistently higher in C18:0 and C20:0, while the brown genotype had higher levels of C18:3 and PUFA (Fig. 1). Nutritionally, the consumption of saturated fatty acids such as C18:0 and C20:0 is undesired, correlating with adverse effects such as heart disease and metabolic syndrome, while consumption of PUFA, and mainly C18:3, is recommended to maintain good diet and health. Thus, although from a cultural point of view white tef is preferred over brown tef, from a nutritional point of view brown tef seems to be superior in regard to fatty acid composition, presenting a higher contents of essential and health-beneficial acids. Additionally, white tef may be preferred due to the presence of higher levels of more palatable saturated fatty acids.
Antioxidative capacity (AOC) is an important health-related trait of foods, reflecting not only the chemical and phytochemical composition, but also the biological activity. Much data is available regarding antioxidative capacity of cereals in general and of tef, however not much of it address the differences between white and brown tef genotypes. In addition, this work shows for the first time the hydrophilic and hydrophobic AOC of tef. Antioxidative capacity of cereals and millets was reported, showing 0.5-0.9 µM TE/g for rice and amaranth, 1.4 for quinoa and 2.4 µM TE/g for buckwheat 59 , 8.5 µmol TE/g for wheat, 15 for barley, 13 for rye, 21.4 for pearl millet and 52 µmol TE/g for sorghum 60 . In different reports, white and brown tef total AOC was found to be 2.9-3 and 4.6-6 µM TE/g 43 , 40 and 50 µM TE/g 22 and 9.3 and 10.3 mmol TE/kg 61 , respectively, and 35 µmol TE/ gr for brown tef 62 . Clearly, not all sources are comparable, however when summing up our HAC and LAC for calculation of total AOC, values are in accordance with other works, which show that when compared to other staple western cereals and gluten free cereals and millets, tef has a higher AOC than most 62,63 .
High levels of N fertilization was reported to decrease AOC in wheat 64 , similarly to our HAC field results. HAC is the main antioxidative capacity (AC) in tef, being higher than LAC, implying that most of the AC in the grain originate in hydrophilic compounds, e.g. polyphenols, rather than lipophilic antioxidants, e.g. tocopherols, as was demonstrated for tef 43 . In addition, in the field experiment N fertilization response trends were similar for HAC and TPC, and correlated well (R 2 = 0.40, p < 0.012), which may imply that polyphenolic compounds are at least partially attributed to the antioxidative activity in tef.
In comparison to other cereal, tef is relatively rich in phenolic compounds 22,62 . TPC values of tef grains are presented in several works: 1.4 to 1.6 mg GAE/g for white and 1.9-2.2 mg GAE/g for brown tef 43,61 , 263-500 mg catechin equivalent (CE)/100 gr in white and 409-700 mg CE/100 gr in brown 22,65 . Our results are in agreement with previous data, i.e. within the same range and follow the clear trend of higher TPC values for brown vs white tef genotypes as shown in other works, and in general for pigmented cereal 43 . As this is the first report presenting the phytochemical and nutritional composition of tef grains grown in Israel, it was of interest to compare current results to existing information regarding teff grain health-related composition.
Free and bound phenolics in white tef range 0.9-1.2 and 0.4-0.5, respectively, and 1.4 and 0.5-0.8, respectively, in brown genotypes 43 . Our field and pot results for both genotypes are within this range. It is important to mention that free and bound phenolics do not sum up to total phenolics measurement due to different extraction methods.
TPC, as well as the content of specific polyphenols, was reported to decrease in wheat in response to high N fertilization 64,66 , which was hypothesized to result from stress condition imparted by high N levels, thus consuming polyphenols to scavenge the resulting reactive oxygen species (ROS) 64 . Nevertheless, some other works reported an increase in wheat TPC in response to N availability 67 . Interestingly, It was postulated that while free soluble phenolics increase with increasing N supply, conjugated soluble compounds decrease, and bound forms are not affected 66 . The invers trends observed in the field vs. pot experiments may reflect a different TPC composition of the grains, i.e. higher levels of conjugated phenolics in the field and higher levels of free phenolics in the pots. In agreement with mentioned report, bound phenolics were not affected in both experiments.
A detailed phenolic profile of tef have been published for both grain colors, and includes flavonoids, stilbenes and phenolic acids 22,43,61,65,68,69 . In this work, we focused on the main free polyphenolic compounds, and our results are concomitant with those of Kotaskova and co-workers 43 , showing that the white genotype was higher in rutin, while the brown genotype higher in ferulic acid (Table 1). Among the compounds we detected, quercetin and rutin were the major phenolic compound, with much higher levels compared to other reports. However, since genotype and environmental conditions greatly affect polyphenolic profile 43 , these may be among the major reasons for the observed difference between our results ad previous reports. In the same manner, we were also able to identify vanillin in our samples, not previously reported in tef. These phenolic compounds are Scientific RepoRtS | (2020) 10:14339 | https://doi.org/10.1038/s41598-020-71299-x www.nature.com/scientificreports/ highly abundant in nature, and many cereals present a similar phenolic profile 70 , with ferulic acid as a major phenolic acid 71 . Furthermore, the trends for rutin and quercetin response to N fertilization was similar to that of TPC in both filed and pot experiments, which might support their presence as major phenolic compounds in our samples (Fig. 3). Numerous reasons may account for the observed variety effects, including inherent genetic variation in N-use efficiency as was previously reported for tef 33,36 , possibly due to a rhizobacteria effects, as was indicated for wheat 72 and tef 73 . Moreover, tef varieties may differ in N-utilization through biosynthetic pathways, e.g. grain production, as implied by differences in their yield 74 . As for possible reasons for observed N effects, N fertilization has been reported to affect mineral accumulation in grain through remobilization of micronutrients within plants , influencing the translocation of metals like Fe and Zn 48 . N availability was also reported to affect fatty acid synthesis, although no mechanism has been suggested 75 . Polyphenolic compound biosynthesis is also affected by N, through increasing amino acid content, including those which are precursors of phenolic acids 41,76 .
As for the possible dilution effect of high yield on reducing seed quality, in our experiments trends of quality parameter levels were not concomitant with those of yield 74 . We thus assume that the possible negative effect of high yield on quality was non-significant under these conditions.
In conclusion, interest in tef is increasing worldwide thanks to its beneficial health effects and gluten-free properties. In this work, we show for the first time the effect of N availability on brown and white tef grain, in addition to a detailed phytochemical profile of both genotypes. The results presented here can be implied as part of a biofortification tool for functional food, aiming at producing healthier and more nutritional food by means of agrotechnology rather than by addition of artificial additives.
While fertilization is crucial for crop managements and high yield, it also affects nutritional value of the food. N fertilization affects tef health and nutritional value, including mineral content, fatty acid profile, anti-oxidative capacity and polyphenol levels and composition. These effects should be considered when deciding on fertilization regime, to optimize both plant physiology, productivity and food-related effects. Of specific consideration is Fe and Zn content, since many health-aware consumers who consume tef and are vegetarian or vegan, low in these nutrients. In addition, being gluten-free, tef in consumed in large amounts by celiac patients, who already have a problematic mineral absorption due to colon inflammation, and if N fertilization management lowers the mineral content this should be noted and acknowledged. At the same time, it should be mentioned that conclusions from the current study are naturally limited, the results being based on only a 1-year field trial and one greenhouse experiment, and are not always consistent for some of the nutritional parameters (Tables 2, 3). Hence, more research is required in order to elucidate the effects of crop cultivar and management on tef grown under irrigation in a dry region like Israel.
It is also important to note that that in addition to the nutritional quality traits we chose to evaluate in this work, there are also some very important organoleptic quality characteristics, including taste, aroma and texture of the final fermented product (e.g. injera bread). In addition, other nutritional aspects may also play a role in quality, i.e. protein and fiber content, as well as the presence of anti-nutrients previously reported in tef, e.g. phytate.

Materials and methods
Chemicals. All chemicals and commercial standards were purchased from Sigma (Sigma, St. Louis, MO, USA). Acetone, methanol, hexane, acetic and hydrochloric acid were from BioLab (BioLab, Jerusalem, Israel). Ethanol was from Gadot (Gadot, Netanya, Israel). Rutin and trolox were purchased from Acros Organics (Acros Organics, New Jersey, USA). Lanthanum chloride was from EMD (Millipore Darmstadt, Germany). Twenty five days after sowing, differential N treatments were initiated, with all other nutrients kept throughout the entire experiment at their initial concentrations. Pots were irrigated with final solutions, according to treatments, via a drip system. In the pot experiment, the yield recorded for the white variety was 10-25 gr/pot, while the brown variety yielded 8-16 gr/ pot. Detailed experimental information for both field and pot is described by Gashu et al. 74 .
Field experiment. Field experiment was conducted during summer 2016. The soil type was a typic Haploxeralf, sandy loam loess, containing 55% sand, 30% silt and 15% clay, and initial soil N availability was 0.18, 0.34 and 0.25 mg/L NO 3 at 0-30, 30-60 and 60-90 cm respectively, and 3.03, 0.9 and 0.72 mg/kg soil of NH 4 at 0-30, 30-60 and 60-90 cm, respectively. Minimum temperature during the experiment ranged between 12 and 22 and maximum temperatures were 29-39 °C (min/max), with no rain events recorded. Experiment included Scientific RepoRtS | (2020) 10:14339 | https://doi.org/10.1038/s41598-020-71299-x www.nature.com/scientificreports/ four N levels: (0, 30, 60 and 120 ppm in the irrigation water). N was provided as 70% NO 3 − and 30% NH 4 in all treatments . Concentrations of all other minerals were identical to those used in the pot experiment. A factorial (N treatments × genotypes) split plot block design was used. Each plot (5 m × 2.1 m) consisted of 28 rows (14 rows per genotype). Each main plot was irrigated by 14 drip lines (one line between each pair of rows). Seeds were directly sown on 13 July 2016 into well-prepared dry soil at a depth of ~ 1 cm. Two weeks after sowing, fertigation treatments were started by injecting 1L of custom-made fertilizer solutions to 100 L of water. Fertigation was applied daily via a drip system. In the field experiment, the yield recorded for the white variety was 26-61 gr/m 2 , while the brown variety yielded 79-121 gr/m 274 . Seed sample preparation. For sampling, three replicates of 5 gr. seeds each from each treatment were freeze-dried in a lyophilizer (Martin Christ, Germany). The dry seeds were powdered by a bead-beater (Zeleniki, Slovenia) at 30 Hz for 1 min with two 1 mm stainless steel beads, and the powder was kept at −20 °C.
Mineral content. For mineral content determination, 1 gr of lyophilized grain powder was baked at 350 °C for 60 min and then at 550 °C for another 5 h. After cooling, 5 ml of HCl was added, and after 60 min the samples were filtered through 42 Whatman filter paper, and 20 ml DD water was added, to a total volume of 25 ml. One ml of lanthanum chloride 98% solution was added to 0.1 ml of the solution, and brought to a final volume of 10 ml with DD water. Minerals were quantified by atomic absorption spectrometry (Perkin Elmer Precisely Analyst 200).
Fatty acid composition. Fatty acid profiling by gas chromatography was performed as we described earlier 77 . The relative composition of fatty acids was determined as percentage of total fatty acid content.
Antioxidative capacity. For evaluating hydrophilic and lipophilic AO capacity of tef grains, 2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS)-Trolox equivalent antioxidant method was employed, as described earlier 78,79 . Free and bound phenolic content. Free and bound phenolic content was measured according to the method described by Kotaskova et al. 43 . For free phenolics, 0.1 gr lyophilized tef seed powder was weighed in an Eppendorf tube, and 500 µl of 75% (v/v) acetone in water solution were added, and centrifuged for 5 min at 17 KG at room temperature, three times. The supernatant was pooled and brought up to a final volume of 1.5 ml. For bound phenolic content, the residues after the free phenolic extraction were dried using a speed-vac (Gemini BV, Netherlands) for 15 min at 15 mbar. 1.5 ml of 2 M NaOH were added the dried residue and vortexed for 30 s, and then mixed at a thermoshaker (AccuTherm, NJ, USA) at 70 °C, 500 rpm for 70 min. 200 µl of the extract were read in a 96-well plate on 360 nm (wavelength optimized to solvent using rutin).
Total phenolic content. Total phenolic content was measured as described earlier 78 , modified to 96 well plates. Briefly, 0.1 gr of tef seed powder was weighed and 500 µl of 80% ethanol solution (v/v) were added, mixture vortexed and centrifuged for 5 min at 17 KG at room temperature, three times. Supernatant was pooled and brought up to 1,500 µl. 13 µl of the extract were then added to 750 µl double distilled H 2 O (DDW) and 63 µl Folin-Ciocalteu reagent and vortexed. 188 µl of 20% Na 2 CO 3 (v/v) and 238 µl of DDW were added and the mixture was let stand for 75 min, after which 200 µl were read in a 96 well plate at 765 nm in a spectrophotometer. Gallic acid (0-600 mg/l) was used as a standard and the results were expressed as mg of gallic acid equivalent (GAE)/g DW of the sample.
Basic polyphenol profile. Polyphenols were extracted from 0.3 gr of tef powder by adding 1 ml of 80% ethanol solution (v/v), mixing, centrifuging (17 KG for 5 min). The extract was filtered through 0.45 µm polytetrafluoroethylene (PTFE) filter and 1.5 ml were aliquoted in a vial. Basic polyphenolic profiling of tef seeds was done using an Ultra Performance Liquid Chromatography (UPLC) system (ACQUITY UPLC H class, Waters, Millford, MA, USA), consisting of a photo diode array (PDA) detector, vacuum degasser, an auto sampler, a binary pump and a reversed phase Benzene hexachloride (BHC) C18 analytical column (2.1 mm × 100 mm, 1.7 µm, Waters). The mobile phases consisted of Milli Q water (0.1% formic acid, v/v) (A) and 100% methanol (UPLC grade) (B). Flow rate was 0.3 ml/min and column temperature maintained at 35 °C. The program was as follows: 1 min at 98% A, decreasing to 95% A over 1 min, decreasing to 30% over another 5 min, further decreasing to 5% A over the next 3 min and final decrease to 0% A over the next 2 min, followed by an increase to 95% A over another 2 min and re-equilibrate to 98% A over the last minute. Identification and quantification of quinic, gallic, protocathchic, caffeic, vanillic syringic, trans-cinnamic, genistic, ferulic, and p-coumaric acids, vanillin, rutin, catechin and quercetin was done with commercial standards. For identification, multiple reaction monitoring (MRM) was used in positive and negative mode, in 5-1,200 mass range, and the MS data was processed by Masslynx software (Waters, Millford, MA, USA). Quantification was performed using calibration curves based on PDA. Wavelength and MRM information for each standard are presented in Table S1. Statistical analysis. Statistical analysis was performed by JMP 13. For Tables 1, 2 and 3 a multifactorial model using analysis of variance (ANOVA) was used, with N fertilization level (continuous) and genotype (character) as predictor variables. For Figs. 1, 2 and 3 a second analysis was performed, using a different model, defining N as ordinal and as a single predictor variable, followed by Tukey pairwise comparisons. The resulting treatment level means are presented as bar graphs, in order to show specific levels of quality parameters in the Scientific RepoRtS | (2020) 10:14339 | https://doi.org/10.1038/s41598-020-71299-x www.nature.com/scientificreports/