Nitrogen use efficiency and genotype-by-environment interaction in durum wheat genotypes under varying nitrogen supply

ABSTRACT

Nitrogen (N) use efficiency is important for wheat grain yield and quality. This study evaluated6 durum wheat genotypes in Ethiopia to determine the extent of nitrogen use efficiency (NUE) components and genotype-by-environment interactions under high and low N supply. The results showed that there was significant variation among the genotypes in grain yield and NUE components. Grain yield ranged from 3.30 to 6.22 t ha⁻¹ under high N and 2.30 to 3.78 t ha⁻¹ under low N conditions with an average reduction of 40.1%. Nitrogen harvest index, Nitrogen uptake efficiency (NUpE), Nitrogen utilization efficiency (NUtE) and NUE increased under low N compared to high N. NUtE varied from 28.6 to 43.9 kg kg⁻¹ under high N and 39.5 to 51.2 kg kg⁻¹ under low N while NUE increased from 26.4 under high N to 31.8 kg kg⁻¹ under low N. Grain yield showed significant and positive associations with most of NUE components under both N conditions. NUpE and NUtE are the two important traits that contribute to NUE. N-efficient genotypes were found to be the most stable genotypes. Thus, the study emphasizes the importance of selecting genotypes with improved grain yield and NUE traits under low N conditions.

KEYWORDS:

• Straw N uptake
• nitrogen utilization efficiency
• nitrogen harvest index

Introduction

Low soil fertility, especially low nitrogen, is a major abiotic stress that limits crop productivity for smallholder farmers in Ethiopia. Nitrogen (N) is the most essential mineral nutrient required abundantly for crop growth and yield formation, and determining productivity and quality of produce (Manschadi and Soltani 2021). In Ethiopia too, N was identified as the most yield limiting nutrient in different crops across most agro-ecologies (Selassie 2015; Balemi et al. 2019). Although, crop productivity can be enhanced by the application of chemical fertilizers, the escalated cost has constrained its judicious application. Thus, efficient use of N is always important because its availability and uptake strongly affect the yield and quality of wheat grain (Barraclough et al. 2014). N use efficiency is a complex trait controlled by interplaying of genetic and environmental factors (Lupini et al. 2021). Crop plants grown in various environments experience significant yield variations that affect the efficiency of selection in a breeding program, and this unstable crop performance in response to shifting growing environment is described as genotype-by-environment interaction (Getahun et al. 2022). Genotypes determine yield potential and responses to environmental factors, which can be measured using high-throughput phenotyping tools (Araus and Cairns 2014). As N stress increases, genotype by environment interaction (GEI) becomes more important in breeding of wheat under low-N conditions targeting resource poor farmers. To this end, the additive main effect and multiplicative interaction (AMMI) model and genotype main effect plus genotype × environment interaction (GGE) biplots have been widely utilized in multi-site trials analysis because they provide more precise estimates and simpler explanations of the genotype-environment interaction (Singamsetti et al. 2021).

Many researchers reported the existence of genetic variability in wheat for N efficiency under low N conditions (Barraclough et al. 2014; Belay et al. 2017; Nehe et al. 2018; Tyagi et al. 2020; Ivić et al. 2021). N use efficiency (NUE) is defined as the grain dry matter yield divided by the supply of available N from the soil and fertilizer (Moll et al. 1982). It is divided into two components: N-uptake efficiency (NUpE), which is defined as above-ground N uptake per unit N available, and N-utilization efficiency (NUtE), which is defined as grain dry matter yield per unit of above-ground N uptake (Nehe et al. 2018).

The relative importance of N uptake and utilization efficiencies for grain yield production remains a topic of debate among researchers. Some studies such as Nehe et al. (2018) have shown that genetic variations in wheat grain yield under low-N conditions are primarily attributed to differences in NUpE rather than NUtE. Others, like Sinebo et al. (2004) have found that NUpE holds greater significance than NUtE in determining NUE and grain yield in barely. Conversly, Gaju et al. (2011) reported that genetic variability in NUE/grain yield under low-N conditions in wheat genotypes was largely due to differences in NUtE rather than NUpE. However, Ortiz-MonasterioR et al. (1997), Xu et al. (2018) and Duan et al. (2019) have reported that higher NUE and grain yield in winter wheat are achieved by a combination of high N uptake and N utilization. These findings highlight the complex interplay between NUE, NUpE and NUtE in determining grain yield under varying N conditions. A better understanding of the mechanisms determining NUE associated with NUpE and NUtE could provide opportunities to increase grain yield under N-limiting conditions and/or combine high yield in wheat cultivars under optimal N conditions. However, there is limited information on the extent of N use efficiency and its components (NUpE and NUtE) in durum wheat genotypes grown under Ethiopian conditions and genotype-by-environment interactions under high and low N conditions for improved NUE and grain yield is also less explored. Therefore, this study was designed to determine the extent of NUpE and NUtE, and their relative contribution to NUE in durum wheat genotypes, and the importance of genotype-by environment interaction for improved grain yield under high and low N levels.

Materials and methods

Description of the experimental sites

The field trials were carried out at three sites (viz. Debre Zeit, Chefe Donsa, and Minjar) in the central highlands of Ethiopia during the main cropping season of 2021. Debre Zeit Agricultural Research Center lies at 8° 44’N, 38° 58’ E, and900 m above sea level, with an annual rainfall of 894 mm and maximum and minimum temperatures of 26.84°C and1.39°C, respectively. Chefe Donsa is located at 8° 57’N and 39°6’ E, 37 km north of Bishoftu town at an altitude of 2435 m above sea level, and has an annual rainfall of020 mm. The average maximum and lowest temperatures at the site are 20°C and 8°C, respectively. Minjar (specific site Memhirhager), on the other hand, is located at 8° 46’N and 39°6’ E, at an elevation of 2257 m above sea levels, and has an annual rainfall of 865 mm. The maximum and minimum temperatures of the area are 28.8°C and2.3°C, respectively. At all the three sites, the main rainy season is long and lasts from June to September and the soil is Vertisol with high clay content. The weather conditions of the locations during the growing period are shown in Figure.

Figure. Monthly average rain fall, maximum and minimum temperature of the experimental sites during the year 2021. DZ = Debre Zeit, CD = Chefe Donsa and MJ = Minjar.

Soil sampling and analysis

Pre-sowing soil analysis was performed in order to establish the experiments. To accomplish this, composite soil samples were collected from each experimental site and analyzed for soil physical and chemical properties at the Soil Laboratory of Debre Zeit Agricultural Research Center (DZARC), while available N contents were determined at Horticop Company using Devardas method.

The available N was 46.2, 21.4 and 44.7 mg kg^−l of soil (113.7, 53.5 and17.1 kg ha⁻¹ N) at Debre Zeit, Chefe Donsa and Minjar, respectively (Table 1). This indicates that the available N in the soil was in the low range (<240 kg ha^−l) based on the reports of TNAU (2016) and was less than the previous results of Ierna et al. (2016) showing that available N was about% of the total N in the soil. Soil bulk density was determined from the clay, silt, and sand content of the soils of each site soil using a computer model simulates developed by the United States Department of Agriculture (USDA), Agricultural Research Service (Saxton et al. 2006). After harvest, plot-based straw and grain samples were collected from each plot and evaluated for N content at the DZARC’s Soil Laboratory to assess the N uptake and utilization of the genotypes at both N levels.

Table 1. Soil chemical and physical properties of experimental locations.

Soil properties	Locations			Methods used
	Debre Zeit	Chefe Donsa	Minjar
pH	6.74	7.39	7.64	1:2.5 soil water ratio
Electric conductivity (ds m^−1)	0.06	0.05	0.09	Potentiometric
Organic Carbon (%)	2.82	2.80	3.60	(Walkley and Black 1934)
Phosphorus (mg kg^−l)	32.08	7.49	14.86	(Olsen et al. 1954)
Total Nitrogen (%)	0.08	0.11	0.09	Kjeldahl (Bremner and Breitenbeck 2008)
Clay (%)	57.6	65.6	55.6	Hydrometer
Silt (%)	30.8	18.8	20.8	Hydrometer
Sand (%)	11.6	15.6	23.6	Hydrometer
Bulk Density (g cm^−3)	1.23	1.25	1.31	USDA model
Soil mass at 0.2 m depth (kg ha^−l)	2,460,000	2,500,000	2,620,000	M=BD×Depth×Area
Available N (mg kg^−l of soil)	46.2	21.4	44.7	Devardas (Liao 1981)
Total N (kg ha^−1)	1,968	2,750	2,358	TN×M/100
Available N (kg ha^−1)	113.7	53.5	117.1	Available N×M

Plant materials, and experimental design and management

Sixteen durum wheat genotypes were used in the field experiments. Five of these genotypes were farmers’ varieties obtained from the Ethiopian Biodiversity Institute (EBI). Four genotypes were from CIMMYT, four from ICARDA, two from DZARC, and one was a released variety (Table 2). In our earlier trials, the grain yield of these genotypes varied under low-N growing conditions, indicating potential differences in N efficiency attributes (Aga et al. 2022). The sixteen durum wheat test genotypes were arranged in RCBD with three replications at each of the two N levels of (0 and 92 kg ha^−l) on plots measuring.2 m × 2.5 m (3 m²). This means that the same set of test genotypes were used in both the 92 kg ha^−l N fertilized (high N) and the N unfertilized (low N) experiments, which were established separately, side by side, at each location.

Table 2. Description of durum wheat genotypes used for the experiment.

No.	Genotypes	Source	Pedigree/ID/Accessions
1	Farmers’ variety	EBI	222415
2	FIGSDRYWET108	ICARDA	IRNS294/ID-98797
3	FIGSDRYWET0144	ICARDA	IRNS382/ID-119026
4	Farmers’ variety	EBI	226958
5	CD15DZELT/off/1239/2015	CIMMYT	GER0MTEL–3/8/ST0T//ALTAR84/ALD/3/THB/CEP7780//2×MUSK-4/6/EC0/CMH76A.722// …
6	CD15DZ_ELT/off/1006/2015	CIMMYT	JUPAREC2001 × 2/RBC/6/STORLOM/3/RASCON-37/TARRD-2//RASCON-37/4/D00003A/5/ …
7	CD15DZELT/off/1516/2015	CIMMYT	JUPARE C 2001 × 2/IM/5/K0FA/4/DUKEM1//PA TKA-7/YAZI-1/3/PATKA-7/YAZI-1/6/ALAS/ …
8	Farmers’ variety	EBI	231585
9	Farmers’ variety	EBI	208215
10	FIGSDRYWET078	ICARDA	AFG68:77/ID-90233
11	Variety Tesfaye	DZARC	ARMENT//SRN-NIGRIS-4/3/CANED-9.1/4/TOSKA-26RASCON-37//SNITSN/5/PLAYERO
12	Farmers’ variety	EBI	231584
13	Breeding pipleline, DWNL p#16	DZARC	BHA/15/MOHAWK/4/DUKEM-1//PATKA-7/YAZI-1/3/PATKA-7/YAZI-1/6/CF4 20S/4/YAZI-1/ …
14	FIGSDWHOTCLD015	ICARDA	ETH732:91/ID-81510
15	CD15DZELT/off/935/2015	CIMMYT	C0RDEIR0/9/GUAYACANINIA/GUANY/8/GEDIZ FG0//GTA/3/SRN-1/4/TOTUS/5/ENTE/MEXI-2// …
16	Breeding pipleline, DWNL p#5	DZARC	Yerer/UC11132 … .25Yellow/DZ2013mehF1 P#6/DZ2014meh F2 P#6–1

EBI = Ethiopian Biodiversity Institute, CIMMYT = International Wheat and Maize Improvement Center, ICARDA = International Center for Agricultural Research in the Dry Areas, and DZARC = Debre Zeit Agricultural Research Center.

Genotypes were randomly assigned to each plot within the blocks. Hand-sorted clean, uniform-sized seed was used for the experiments. Manual sowing was carried out on finely prepared seedbeds. For the N treated experiments, split applications of granule form (UREA) of N fertilizer was applied in rows, half at the time of sowing and half at tillering stages. Phosphorus fertilizer in the form of TSP (triple supper phosphate) was applied at the recommended rate0 kg P ha⁻¹ uniformly to both the N fertilized and unfertilized experiments.

Recommended crop management practices were employed uniformly to all plots. Repeated hand weeding was done as necessary, and the plots were kept free of weeds. Rust infestations were controlled by using sprays of Nativo 300SC (200 g l⁻¹ Tebuconazole +100 g l⁻¹ Trifoxystrobin) fungicide twice at a rate of 0.75 liter with50 liters of water per hectare.

Data collection

The four central rows were harvested at physiological maturity for dry matter determination and plant N analysis. Data on above-ground biomass (BM) and grain yield (GY) were recorded and converted to a hectare basis after plants were manually harvested. BM was determined using a spring balance in the field during harvesting, whereas GY was measured by weighing the threshed grain on an analytical balance and adjusted to2.5% moisture content. N uptake (NUP), N uptake efficiency (NUpE), N utilization efficiency (NUtE), N use efficiency (NUE) and N harvest index were calculated from the data of pre-planting soil analysis and post-harvest straw and grain N content analysis of the durum wheat genotypes under low and high N growth conditions. These variables are computed following the methods of Moll et al. (1982) as follows:

NUP was designed to be equivalent to the total N (kg) in the aboveground biomass or plant N.

(1) $\mathit{Nitrogen}\mathit{uptake}\mathit{efficiency}\left(\mathrm{NUpE}\right)=\frac{\mathit{Plant}\mathit{N}}{\mathit{Fertilizer}\mathit{N}+\mathit{Soil}\mathit{N}}\times 00$(1)

Where, plant N= amount of N in the grain + in the straw.

(2) $\mathit{Nitrogen}\mathit{utilization}\mathit{efficiency}(\mathit{NUtE})=\frac{\mathit{Yield}}{\mathit{Plant}\mathit{N}}$(2)

Where, yield=grain yield per hectare, plant N= amount of N in the grain + the straw.

Nitrogen use efficiency (NUE)= NUpE x NUtE

(3) $\mathit{Nitrogen}\mathit{harvest}\mathit{index}(\mathit{NHI}\text{)}=\frac{\mathit{Yield}\mathit{N}}{\mathit{Plant}\mathit{N}}\times 100$(3)

Where, Yield N=amount of N in the grain, plant N= amount of N in the grain + the straw.

Statistical analysis

Statistical analyses for grain yield and N efficiency components (NUPE, NUtE, NUE, and NHI) were carried out using SAS Software Version 9.4 (SAS 2013). The analysis of variance (ANOVA) was performed, followed by multiple comparisons of means using ANOVA-protected least significant difference (LSD) at p ≤ 0.05. Correlation analysis was done to assess the relationship between grain yield and N efficiency components.

The mean grain yield for each genotype in each environment was determined, and the data were subjected to a combined analysis of variance to evaluate the effects of environment (E), genotypes (G), and their interactions. The additive main effects and multiplicative interaction (AMMI) model were used to examine similarity and dissimilarity between testing environments, stability of genotypes as well as the interaction patterns between genotypes and environments. The AMMI analysis combines ANOVA and PCA into a single model with additive and multiplicative parameters. AMMI analysis was performed using the GGE biplot Packages in R Software Version 4.2.3 (R Core team 2013).

Results and discussion

Analysis of variance (ANOVA)

The combined ANOVA over the three locations showed that durum wheat genotypes varied significantly in grain yield (GYD) and N use efficiency under both high and low N supplies (Table 3). The genotypes differed significantly across all locations for all traits, except for N uptake efficiency (NUpE) under low N and N harvest index (NHI) under both N supplies. Similarly, the genotype by location interaction was significant for GYD and N use efficiency (NUE) under high N, above ground biomass N yield (BNY) and NHI under low N, and N utilization efficiency (NUtE) under both N conditions (Table 3).

Table 3. Mean squares, means and coefficients of variation from the combined analysis of variance of grain yield (GYD) and N use efficiency components of durum wheat genotypes under high and low N conditions over three locations.

Traits	N levels	Mean squares					Grand mean	LSD (0.05)	CV (%)
		Geno. (df = 15)	Loc. (df = 2)	Rep. (df = 2)	Geno*Loc. (Df = 30)	Error (df = 94)
GYD	LN	1.9**	58.2**	0.6^ns	0.4^ns	0.3	2.9	0.5	18.1
	HN	5.6**	47.4**	0.9^ns	0.7*	0.4	4.9	0.6	13.1
BNY	LN	827.6**	38170.3**	612.7^ns	413.9*	260.6	70.7	15.1	22.8
	HN	693.0**	23307.5**	781.8^ns	338.7^ns	291.6	131.6	15.9	12.9
NUpE	LN	0.08**	0.07^ns	0.02^ns	0.04^ns	0.02	0.7	0.1	21.1
	HN	0.02**	0.03*	0.03^ns	0.01^ns	0.01	0.7	0.1	12.5
NHI	LN	39.5**	4.2^ns	29.7^ns	62.8**	10.9	79.6	3.09	4.1
	HN	261.4**	35.2^ns	10.6^ns	31.3^ns	15.8	76.2	3.7	5.2
NUtE	LN	66.9**	2532.8**	6.4^ns	36.2**	4.5	43.1	1.9	4.9
	HN	203.9**	1284.1**	10.1^ns	13.1 **	4.6	37.2	2.0	5.8
NUE	LN	194.1**	1684.1**	12.5^ns	25.9^ns	28.8	31.8	5.0	16.8
	HN	156.9**	509.2**	29.1^ns	17.8*	10.7	26.4	3.1	12.4

** and *= significant at p< 0.001 and p< 0.05 respectively, ns = non-significant, GYD = grain yield, BNY = above ground biomass N yield, NUPTE = N uptake efficiency, NHI = N harvest index, NUTE = N utilization efficiency and NUE = N use efficiency.

Grain yield

In this study, the means grain yield of the test genotypes ranged from 3.30 t ha⁻¹ to 6.22 t ha⁻¹ under high N and from 2.30 t ha⁻¹ to 3.78 t ha⁻¹ under low N. The average reduction in GYD for all durum wheat genotypes under low N compared to high N supply was.96 t ha⁻¹ (40.1%), with the yield reductions of the individual genotypes ranging from 0.34 t ha⁻¹ (10.3%) for genotype 2 (FIGSDRYWET108) from the N efficient group (1–8) to 3.28 t ha⁻¹ (58.8%) for genotype0 (FIGSDRYWET078, source from ICARDA) from the N inefficient group (9–16), which experienced the greatest yield decrease (Table 4). Comparing the two durum wheat genotypes groups, the average yield reduction for the N-efficient genotypes was.53 t ha⁻¹ (31.50%) whereas it was 2.40 t ha⁻¹ (48.5%) for the N-inefficient groups (Table 4). Genotype 5 (CD15DZELT/off/1239/2015 from CIMMYT) from the N efficient group achieved the highest GYD (6.22–3.78 t ha⁻¹) under both N conditions. This demonstrates its high N efficiency and responsiveness to N application. Moreover, the two farmers’ varieties genotypes and 8 and 7 (CD15DZELT/off/1516/2015 from CIMMYT) were among the top performing durum wheat genotypes under low N conditions without significant differences among each other.

Table 4. Mean grain yield and above ground biomass N yield of6 durum wheat genotypes over the three locations.

Durum wheat genotypes	Grain Yield (t ha^−1)		Above ground biomass N yield (kg ha^−1)
	Low N	High N	Low N	High N
1 (Farmers; variety)	3.57^abc	4.86^cd	86.1^a	136.4^abc
2 (FIGSDRYWET108)	2.96^d-^h	3.30^g	73.9^a-^e	114.4^e
3 (FIGSDRYWET0144)	3.24^b-^e	5.67^ab	76.1^a-^d	143.7^a
4 (Farmers’ variety)	3.08^c-^g	4.14^ef	76.6^abc	137.7^abc
5 (CD15DZELT/off/1239/2015)	3.78^a	6.22^a	76.2^a-^d	142.0^ab
6 (CD15DZ_ELT/off/1006/2015)	3.16^b-^f	5.25^bc	77.6^ab	135.2^a-^d
7 (CD15DZELT/off/1516/2015)	3.30^a-^d	5.13^bc	73.0^a-^e	126.8^b-^e
8 (Farmers’ variety)	3.41^ab	4.12^ef	86.9^a	130.9^a-^d
9 (Farmers’ variety)	2.78^e-^i	3.67^fg	76.1^a-^d	120.2^cde
10 (FIGSDRYWET078)	2.30^i	5.58^b	54.9^f	142.7^ab
11 (Variety Tesfaye)	2.63^ghi	4.89^cd	64.3^b-^f	122.6^cde
12 (Farmers; variety)	2.52^hi	4.40^de	59.4^ef	130.1^a-^e
13 (Breeding pipleline, DWNL p#16)	2.45^i	4.91^cd	61.3^def	123.3^cde
14 (FIGSDWHOTCLD015)	2.70^f-^i	5.29^bc	62.5^b-^f	138.3^abc
15 (CD15DZELT/off/935/2015)	2.55^hi	5.17^bc	61.7^c-^f	125.7^cde
16 (Breeding pipleline, DWNL p#5)	2.49^hi	5.71^ab	64.3^b-^f	136.3^abc
Grand mean	2.9	4.9	70.7	131.4
LSD(<0.05)	0.5	0.6	15.11	15.98
CV	18.1	13.1	22.8	12.9

Means in the same columns followed by the same letter(s) are not significantly different at p ≤ 0.05.

Under both N supplies, the highest grain yield was harvested at Minjar site, while the lowest grain yield was obtained at Chefe Donsa site. This might be due to the high rainfall at Chefe Donsa that leaches the available and applied N from the root zone of the crop (Figure 2(a)). All of the N efficient durum wheat genotypes group produced higher grain yields than the N inefficient group under low N, with an average yield advantage of 0.76 t ha⁻¹ (22.9%) over the inefficient ones (Table 4). In line with our finding, Gaju et al. (2011) reported significant variation in grain yield performance of wheat cultivars under high and low N conditions. Moreover, Mariem et al. (2020) also observed grain yield variation among durum wheat genotypes under high and low N conditions.

Figure 2. Average grain yields (a) and above ground biomass N yield (b) of durum wheat genotypes at the three experimental sites under low and high N growth conditions (the different small and capital letters indicate the presence of significant difference between sites under low N and high N, respectively).

Total plant N uptake

The total N up-taken by the durum wheat genotypes in the above-ground biomass varied from14.4 kg ha⁻¹ to43.7 kg ha⁻¹ under high N, and from 54.9 kg ha⁻¹ to 86.9 kg ha⁻¹ under low N conditions (Table 4). The highest N was taken up by genotype 3 (FIGSDRYWET0144) under high N and genotype 8 (Farmers’ variety) under low N, both from the N efficient group, while genotype 2 (FIGSDRYWET108) under high N and genotype0 (FIGSDRYWET078) under low N took up the least N in the aboveground biomass (Table 4). Genotypes and 8, both farmers’ varieties, were among the top N uptake efficient genotypes under low N, which corresponds to the high grain yield obtained from the same genotypes. Genotype0 (FIGSDRYWET078) exhibited significantly higher N uptake by the aboveground biomass under high N but least under low N supply, indicating its inefficiency but responsiveness to N application. Such genotypes cannot be selected as suitable for low N environments (Table 4).

In this study, higher N uptake was observed under high N compared to low N. Under low N, the amount of N taken up by the efficient durum wheat genotypes was statistically comparable to one another (Table 4). The highest above-ground biomass N yield was obtained at Minjar, although there was no statistically significant difference with the Debre Zeit site. The lowest N yield was found at Chefe Donsa, which might be related to the limited availability of N in the soil and other climatic factors (Figure 2(b)). In agreement with this study, Barraclough et al. (2010) found variation in total N uptake among wheat varieties, and he observed more N uptake under high N than low N rate. Furthermore, López-Bellido and López-Bellido (2001) reported a significant difference in total N uptake by wheat crop between 0 and50 kg ha⁻¹ N rates, but lower in total N uptake compared to the current study, which could be due to differences in the wheat genotype and growth conditions.

Nitrogen uptake efficiency

N uptake efficiency (NUpE) is the amount of available soil N that is utilized by the plants, and indicates increased synchronization between N availability and plant demand (Congreves et al. 2021). The mean NUpE for all genotypes tested increased from 0.71 under high N to 0.75 under low N (Table S1). This is consistent with the findings of Gaju et al. (2011) depicting higher NUpE under low N than under high N. Similarly, NUpE increased from 0.72 under high N to 0.82 under low N for the N efficient group of durum wheat genotypes; whereas it declined from 0.70 under high N to 0.66 under low N for the N inefficient group (Table S1).

NUpE under high N varied from 0.63 (Genotype 2) to 0.77 (Genotype0) and from 0.58 (Genotype0) to 0.89 (Genotype 8) under low N. Based on mean values across locations, genotype 8,, 2, 4 and 5 were among the best N uptake efficient genotypes under low N conditions, this holds true for genotype0, 5, 3, 4 and4 under high N conditions (Figure 3). This shows that genotype 5, 3 and 4 can adapt under both N growth conditions, while genotype0 and 8 can be selected for high and low N growth conditions, respectively. Similarly, Belay et al. (2017) found significant variation in NUpE among Ethiopian durum wheat varieties. On the other hand, genotype 2, 9 and1 under high N and genotype0,2 and5 under low N performed poorly in NUpE (Figure 3). The mean NUpE was higher under low N than high N for all N-efficient genotypes, but the N-inefficient durum wheat genotypes showed inconsistent performance in NUpE under high and low N conditions (Figure 3).

Figure 3. Average N uptake efficiency (NUpE) of durum wheat genotypes across the three locations under high and low N (the different small and capital letters indicate the presence of significant difference between sites under low N and high N, respectively).

The results showed differences in NUpE among the three test sites, with the over-all mean NUpE being highest at Chefe Donsa and lowest at Minjar under high N and highest at Minjar and lowest at Debre Zeit under low N (Table S2). This agrees with the findings of Rahimizadeh et al. (2010), who found increased NUpE in soils with low original N concentration.

Nitrogen harvest index

N harvest index (NHI) is described as the ratio between N accumulated in grain to the amount of N accumulated in the grain plus straw (Fageria 2014). NHI is useful to identify plants with high N translocation to the economic component or grain formation (Congreves et al. 2021) but with the limitations that the results may be confounded by soil nutrient status (i.e high NHI values may indicate N deficiency rather than increased N use efficiency). In the present study, NHI increased from 76.2% to 79.6% under high and low N, respectively. Compared to the result of this study López-Bellido and López-Bellido (2001) reported 71–78% of NHI and Ivić et al. (2021) found 83.2% and 82.4% mean NHI values for 48 wheat cultivars under low and high N environments. The overall mean values of NHI boosted from 75.2% to 79.8% for the N efficient group and from 77.2% to 79.3% for the N inefficient durum wheat group under high and low N conditions, respectively (Table S1). The average value of NHI was higher under low N than under high N for all durum wheat genotypes studied, except for genotypes 6, 7,4 and6 (Figure 4). Under high N supply, the highest NHI (83.2%) was recorded for genotype6, and this was statistically at par with that of the genotypes 6,4, 5, 7 and 3. In contrast, the lowest NHI (65.6%) was recorded for genotype 2. On the other hand, under low N condition, the maximum NHI (82.7%) was recorded for genotype0 followed by genotype 5 (82.3%), 3 (81.8%) and 4 (81.4%) (Figure 4 and Table S1). In line with the current study, Belete et al. (2018) observed variation in NHI values between bread wheat varieties and N levels. In contrast, NHI values declined as the N rate decreased in Pearl millet (Pujarula et al. 2021) and dry beans (Fageria 2014).

Figure 4. Average nitrogen harvest index (NHI) values across the three locations of durum wheat genotypes under high and low N growth conditions (the different small and capital letters indicate the presence of significant difference between sites under low N and high N, respectively).

Nitrogen utilization efficiency (NUtE)

N utilization efficiency (NUtE) is defined as the ratio of grain yield to total above-ground biomass N, and it is useful to identify plants that have superior ability in producing yield relative to plant tissue N (Congreves et al. 2021). The mean values for NUtE among tested durum wheat genotypes under high N ranged from 28.6 to 43.9 kg kg⁻¹ N and from 39.5 to 51.2 kg kg⁻¹ N under low N. This indicates an increase of 53.5% under high N and 29.6% under low N between the highest and lowest N efficient durum wheat genotypes. Overall, NUtE increased from 37.3 under high N to 43.1 kg kg⁻¹ N under low N growth conditions. This implies that durum wheat genotypes produced about 37.3 kg grain yield per kg of total above-ground N uptake under high N and about 43.1 kg grain yield per kg N in the above ground biomass under low N, indicating that utilization efficiency increased under low N than high N. The utilization efficiency of the above-ground N at N zero was higher by3.5% compared to high N (Table 5).

Table 5. Mean nitrogen utilization efficiency (NUtE) and nitrogen use efficiency (NUE) of6 durum wheat genotypes over the three locations.

Durum wheat genotypes	Nitrogen utilization efficiency (kg kg^−1)		Nitrogen use efficiency (kg kg^−1)
	Low N	High N	Low N	High N
1 (Farmers; variety)	44.7^bc	35.5^e	37.8^abc	26.2^ef
2 (FIGSDRYWET108)	40.2^fg	28.6^g	32.7^d-^g	18.2^i
3 (FIGSDRYWET0144)	44.1^bc	39.7^cd	34.2^b-^e	30.1^bc
4 (Farmers’ variety)	41.6^def	30.4^fg	33.9^b-^f	22.6^g
5 (CD15DZELT/off/1239/2015)	51.2^a	43.9^a	40.3^a	33.3^a
6 (CD15DZ_ELT/off/1006/2015)	43.0^cde	39.2^d	33.8^c-^f	28.1^b-^e
7 (CD15DZELT/off/1516/2015)	45.4^b	41.3^bc	34.8^bcd	27.5^cde
8 (Farmers’ variety)	44.3^bc	31.6^f	39.1^ab	22.1^gh
9 (Farmers’ variety)	39.5^g	30.6^f	29.5^e-^h	19.5^hi
10 (FIGSDRYWET078)	43.2^cd	39.1^d	25.7^h	30.0^bcd
11 (Variety Tesfaye)	41.6^def	40.2^bcd	29.1^fgh	26.5^ef
12 (Farmers; variety)	42.9^cde	34.3^e	27.3^h	23.6^fg
13 (Breeding pipleline, DWNL p#16)	41.2^efg	39.9^cd	27.5^h	27.0^de
14 (FIGSDWHOTCLD015)	43.1^cde	38.2^d	29.1^fgh	28.5^b-^e
15 (CD15DZELT/off/935/2015)	43.1^cde	41.5^ab	27.9^gh	27.9^b-^e
16 (Breeding pipleline, DWNL p#5)	40.7^fg	41.9^ab	26.6^h	30.7^ab
Grand mean	43.1	37.3	31.8	26.4
LSD(<0.05)	1.99	2.01	5.02	3.06
CV (%)	4.9	5.8	16.8	12.4

Means in the same columns followed by the same letter(s) are not significantly different as judged by LSD at p ≤ 0.05.

N utilization efficiency (NUtE) increased from 36.3 under high N to 44.3 kg kg⁻¹under low N for N efficient durum wheat genotypes in which the NUtE under low N was higher by 22.2% than under the high N. This increment for the inefficient genotypes was from 38.2 to 41.9 kg kg⁻¹, which was0% greater under low N than under high N conditions. This result is in line with the findings of Ali et al. (2022). Under high N conditions, genotype 5 had the highest NUtE, followed by genotypes6,5, and 7. Genotype 2, on the other hand had the lowest NUtE. All N efficient genotypes had better performance than the N inefficient group under low N conditions. Genotype 5 followed by genotype 7,, 8, and 3 were among the genotypes with high NUtE under low N while genotype 9 had the lowest NUtE under similar N supply (Table 5). In agreement with the results of the current study, Lupini et al. (2021) found significant variation in NUtE among durum wheat genotypes. Haile et al. (2012) also reported that NUtE increased when the N rate decreased. Gaju et al. (2011) observed differences in NUtE amongst winter wheat cultivars with higher NUtE under low N than high N conditions. Furthermore, Ali et al. (2022) found substantial variations in NUtE (87.9 and 67.9 kg kg⁻¹ N) among 32 barley genotypes. However, the values of NUtE are greater than those in our results, which could be attributable to differences in crop and soil N condition. The highest NUtE was obtained at Minjar site followed by Chefe Donsa, while Debre Zeit had the lowest value under high N. On the other hand, under low N the highest NUtE was obtained at Chefe Donsa followed by Minjar, and the least was from Debre Zeit (Figure 5). Similarly, Hitz et al. (2016) reported variation in NUtE of soft red winter wheat between locations. Contrary to our results, Silva et al. (2020) found no differences in wheat NUtE across a wide range of experimental environments.

Figure 5. Average nitrogen utilization efficiency (a) and nitrogen use efficiency (b) of durum wheat genotypes at the three experimental sites under low N and high N growth conditions (the different small and capital letters indicate the presence of significant difference between sites under low N and high N, respectively).

Nitrogen use efficiency (NUE)

NUE is the ratio of grain yield to the total amount of available N or it is the product of NUpE and NUtE, which enables comparisons of yield potential among crop genotypes (Congreves et al. 2021). In the present study, the negative effect of high N on NUE was marked as it increased from 26.4 under high N to 31.8 kg kg⁻¹ N under low N conditions, which was consistent with the findings of Rahman et al. (2011). The highest NUE value (33.3 kg kg⁻¹ N) was recorded for genotype 5 followed by genotype6 (30.7 kg kg⁻¹), 3 (30.1 kg kg⁻¹) and0 (30.0 kg kg⁻¹) while genotype 2 had the lowest NUE under high N growth conditions. According to the current findings, the best yielding durum wheat genotypes also had the highest NUE, suggesting that there is genetic diversity in NUE, which represents the range of yields. Under low N supply, NUE ranged from 25.7 kg kg⁻¹ for genotype0 to 40.3 kg kg⁻¹ for genotype 5 (Table 5).

In agreement with these results, Hawkesford and Riche (2020) reported that NUE progressively decreased as the N input increased. Higher NUE values were observed for N-efficient durum wheat genotypes than for the inefficient genotypes group under low N, while they were comparable under high N conditions as reported by Swain et al. (2014). The N-inefficient durum wheat genotypes were not statistically different from each other in NUE under low N. The Mean value of NUE for genotype 5 was at par with that of genotypes 8 and under low N (Table 5). This indicates that the genotype with better NUE under low N can be selected as a source of gene pool in breeding of durum wheat genotypes for low N environments. The average increase in NUE under low N compared to high N was 37.7% for N-efficient, and it was 4.2% for the N-inefficient durum wheat genotypes (Table 5). The mean NUE values significantly varied between experimental sites under both N conditions. Lower NUE value was recorded at Debre Zeit compared to Minjar and Chefe Donsa under high and low N conditions. The value of NUE was higher under low N than high N, and there was no significant difference between Minjar and Chefe Donsa under both N conditions (Figure 5). Similarly, higher NUE under low N than under high N, and differences among experimental sites was reported by Hitz et al. (2016).

Grain and straw N uptake under low and high N

ANOVA demonstrated that durum wheat genotypes differed significantly in grain and straw N uptake under low and high N supply, whereas the influence of location was not significant under both N supplies (data not shown). Under low N, grain N uptake by durum wheat genotypes ranged from 75.40 to 82.68% while the rest7.32 to 24.60% was stored in straw (Figure 6(a)). Genotypes2, 2, 9, and had the highest N uptake by straw; whereas genotypes0, 5, 3, and 4 had lower straw N uptake under this N supply (Figure 6(a)).

Figure 6. Relative proportions of grain and straw N uptake by durum wheat genotypes under (a) Low N and (b) High N supply.

Under high N supply, the performance of genotypes in grain N uptake ranged from 65.60 to 83.23%, and the rest proportion being stored in the straw (Figure 6(b)). Durum wheat genotypes 2, 9, 8, and 4 had the highest straw N uptake, whereas genotypes6, 6,4, and 5 had lower straw N uptake (Figure 6(b)). The average N uptake for all genotypes in the grain of durum genotype was 79.6% under low N and 76.2% under high N, with the remaining proportion in straw under both N supplies (Table S3). In comparison to high N supply, the proportions of N in the straw are smaller under low N supply. This could be related to a better genetic response of genotypes under low available N in the soil. The variation in grain and straw N uptake by wheat cultivars reported by Belete et al. (2018) and Tufa et al. (2022) is consistent with the current finding.

Inter-traits correlations

Table 6 depicts the relationship between grain yield (GYD), above-ground biomass yield (BM) and N use efficiency components under both high and low N conditions. The study showed that there was a significant and positive correlation of GYD with all examined variables under high N. Accordingly, above-ground biomass N yield (BNY), N harvest index (NHI), N utilization efficiency (NUtE) and N use efficiency (NUE) were strongly correlated, and as such in agreement with the findings of Ivić et al. (2021). Similarly, N uptake efficiency (NUpE) was strongly, positively and significantly associated with BNY and NUE under high N conditions (Table 6).

Table 6. Simple correlation coefficients among yield and nitrogen use efficiency component variables of durum wheat genotypes under high N (upper diagonal) and low N (below diagonal) conditions.

Traits	BNY	NUpE	NHI	NUtE	NUE	BM	GYD
BNY	1.00	0.59**	0.24**	−0.04ns	0.32**	0.43**	0.71**
NUpE	0.68**	1.00	0.20*	0.01ns	0.61**	0.01ns	0.42*
NHI	−0.05ns	−0.11ns	1.00	0.64**	0.62**	−0.38**	0.59**
NUtE	−0.47**	−0.11ns	0.30**	1.00	0.79**	−0.20*	0.66**
NUE	0.26**	0.75**	0.09ns	0.56**	1.00	−0.16ns	0.78**
BM	0.89**	0.64**	−0.07ns	−0.22**	0.40**	1.00	0.19*
GYD	0.90**	0.63**	0.06ns	−0.09ns	0.49**	0.91**	1.00

* and ** = significant at p < 0.05 and p < 0.001 respectively, ns = non-significant, GYD = Grain yield, BM = Above-ground biomass yield, BNY = above-ground biomass N yield, NUpE = N uptake efficiency, NHI = N harvest index, NUtE = N utilization efficiency, and NUE=N use efficiency.

Under low N supply, GYD correlated positively and significantly with BNY, NUpE, NUE and BM, while it is negatively correlated with NUtE. Significant positive associations were detected between BM, BNY, NUpE and NUE under low N conditions. NUtE and NUE had strongly positive and significant association under both N conditions. NUpE and NUtE are strongly and positively correlated with NUE under both N conditions, which indicates that both traits contribute to NUE. Negative correlation between NUtE with BNY, non-significant correlation between NUpE and NUtE, and significant but negative correlations between BM and NUtE were observed under both N supplies (Table 6). In line with our results, Rahimizadeh et al. (2010) reported significant correlation between grain yield and traits such as NUpE, NUtE and NHI. On the other hand, Gaju et al. (2011) in contrast to the present results found strong correlation between grain yield and NUtE under low N conditions.

Genotype by environment interaction

The results of the additive main effects and multiplicative interaction (AMMI) variance analysis for grain yield of6 durum wheat genotypes tested in six environments revealed substantial (p < 0.001) effects of environments (E), genotypes (G), and G × E interaction, demonstrating the existence of genetic variation and the potential for selection of stable genotypes (Table S4). The variance (sum of squares) partitioning indicated that the environmental effect was the primary source of variance, followed by genotype × environment interaction and then genotype effects, which is consistent with the findings of Yan (2001) and Sardouei-Nasab et al. (2019).

Genotype and genotype by environment biplot analysis

AMMI biplot analysis for grain yield revealed distinct trends across the testing environments, with AXIS and AXIS 2 explaining 62.03% and 25.65% of the genotype-by-environment interaction, respectively (Figure 7). Environments with N un-fertilized (DZLN, CDLN, and MJLN) and those with N fertilized (DZHN, CDHN, and MJHN) were clearly distinguished by the AMMI analysis; however, there was no difference between environments with the same N levels. In particular, at Chefe Donsa, the interaction between low and high N environment was higher, and AMMI biplots indicated an increased dissimilarity between high and low N environments (Figure 7). Based on the AMMI biplot, genotypes 6, 7, 3, 5 and were among those correlated to the low N environments, while genotypes5,0,6 and4 were associated with high N environments. On the other hand, genotypes2,3 and1 were negatively correlated to all environments (Figure 7).

Figure 7. GGE biplot for six environments and grain yield of6 durum wheat genotypes tested under low and high N conditions at three locations. The numbers–16 indicate the genotype codes. DZLN = Debre Zeit under low N, DZHN = Debre Zeit high N, CDLN = Chefe Donsa low N, CDHN = Chefe Donsa high N, MJLN = Minjar low N and MJHN = Minjar high N.

Similarly, Worku et al. (2007) identified genotypes that correlate to different environments with varying N levels. Genotypes are regarded as widely adaptable to all environments if they have means above the grand mean and AXIS scores close to zero. However, genotypes with high mean performance and high AXIS scores are regarded as having specific adaptations to the environment. To this effect, genotypes 6 and 7 are widely adapted while genotype0 was specifically adapted genotype (Figure 7). In a comparable manner to that of the present study, Amare et al. (2020) employed this model to examine the adaptability of bread wheat genotypes to drought tolerance.

Grain yield performance of genotypes with respect to environments (which won where)

The genotype and genotype by environment (GGE) polygon helps to identify winning genotypes in different environments Yan and Tinker (2005). The vertex genotypes (5, 8,0,3 and 2) with the largest distance from the origin were connected to generate the polygon view. These genotypes were the highest yielding in their respective environments. The biplot is divided into seven sections by the seven rays with genotypes falling in four sections while the environments fall in six areas (Figure 8).

Figure 8. Adaptability and performance of durum wheat genotypes with respect to test environments. The numbers–16 indicate the genotype codes. DZLN = Debre Zeit under low N, DZHN = Debre Zeit high N, CDLN = Chefe Donsa low N, CDHN = Chefe Donsa high N, MJLN = Minjar low N and MJHN = Minjar high N.

In this study, genotype 5 is the only genotype found in the polygon’s vertices that include all of the environments, while genotypes 8 and 2 are contained at Chefe Donsa low N condition and genotype0 is contained at Chefe Donsa high N environment. In the six test environments, genotype3 located on the vertices of the polygon devoid of any of the six environments is unfavorable. This study revealed that, genotype 5 performed best in all environments, genotypes 8 and 2 in low N environments, and genotype0 in high N environments in terms of grain yield, Hence, these genotypes can be chosen for those respective growing environments (Figure 8).

AMMI stability value

In the AMMI model, the AXIS and AXIS 2 scores are measures of stability (Amare et al. 2020). Genotypes with lower AMMI stability values (ASV) are considered more stable, whereas genotypes with greater ASV are deemed unstable.

Table 7 depicted that the most stable genotype for grain yield as measured by the ASV was genotype2 with an ASV of 0.13, followed by genotypes 7 (0.36), 6 (0.40),1 (0.76),3 (1.12) and 5 (1.22), while the least stable genotypes were genotypes 2 (3.69), 8 (3.06),0 (2.88), and genotype6 (2.67). Among the most stable genotypes, 7, 6, and 5 had mean grain yields above the grand mean and this result is in agreement with that of Lemma and Mekbib (2021) who employed ASV to evaluate grain yield stability of durum wheat genotypes.

Table 7. Mean grain yield, AMMI stability value and yield stability index for the6 durum wheat genotypes tested across the six environments.

Genotypes	Mean Yield	ASV	Rank of ASV	YSI	Rank of YSI
1 (Farmers; variety)	4.22	1.59	10	13	3
2 (FIGSDRYWET108)	3.14	3.69	16	32	16
3 (FIGSDRYWET0144)	4.45	1.42	8	10	2
4 (Farmers’ variety)	3.61	2.02	11	24	13
5 (CD15DZELT/off/1239/2015)	5.00	1.22	6	7	1
6 (CD15DZ_ELT/off/1006/2015)	4.22	0.40	3	7	4
7 (CD15DZELT/off/1516/2015)	4.20	0.36	2	7	5
8 (Farmers’ variety)	3.88	3.06	15	24	9
9 (Farmers’ variety)	3.23	2.13	12	27	15
10 (FIGSDRYWET078)	3.94	2.88	14	22	8
11 (Variety Tesfaye)	3.76	0.76	4	15	11
12 (Farmers; variety)	3.46	0.13	1	15	14
13 (Breeding pipleline, DWNL p#16)	3.68	1.12	5	17	12
14 (FIGSDWHOTCLD015)	4.01	1.54	9	16	7
15 (CD15DZELT/off/935/2015)	3.86	1.34	7	17	10
16 (Breeding pipleline, DWNL p#5)	4.09	2.67	13	19	6

ASV = AMMI stability value and YSI = Yield stability index.

Yield stability index

The yield stability index (YSI) method combines yield and stability into a single index, reducing the challenge of selecting genotypes solely based on yield stability. Thus, genotypes with lower YSI are preferred because they combine high mean grain yield performance with stability. Depending on the YSI, genotypes 5, 6, 7, 3, and (all from the N-efficient durum wheat genotype group) were chosen as the most stable durum wheat genotypes in this study because they combine high grain yield with stability, which renders them ideal for the subsequent stage of breeding for variety development (Table 7 and Figure S1). Despite the fact that genotypes0,6, and4 were high yielding genotypes with high ASV and YSI scores, they can still be suggested for specific environments (Table 7 and Figure S1). In line with our study, Wardofa et al. (2019 and Temesgen et al. (2015) employed this method to examine G x E in bread wheat.

Conclusions

Significant variations among durum wheat genotypes in grain yield and N use efficiency (NUE) components under high and low N conditions were demonstrated. The highest grain yield was demonstrated from N efficient groups which also exhibited higher N uptake under both N environments where some perform under low N condition. N uptake efficiency (NUpE) and N harvest index increased under low N compared to high N conditions for most of the genotypes and with higher values observed for the N efficient group. N utilization efficiency (NUtE) and NUE were higher under low N than under high N. Thus, the results suggest the presence of diversity in NUE components among durum wheat genotypes was with better performing ones under low N conditions. Grain yield showed positive correlations with all NUE components under high N supply exhibiting a strong correlation with the above-ground biomass N yield, biomass yield; NUpE and NUE under low N. NUpE and NUtE are the two key traits that contribute significantly to NUE. The N-efficient genotypes are found to be the most stable genotypes chosen for subsequent breeding stage. The result of this study highlights the potential for improving NUE and grain yield in durum wheat through selection of genotypes that exhibit high NUpE and NUtE under low N conditions.

Acknowledgments

The authors are grateful to the Ethiopian Institute of Agricultural Research and Debre Zeit Agricultural Research Center for funding and providing an experimental materials and fields.

Data availability statement

The data presented in this study are available upon request from the corresponding authors.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/03650340.2024.2317339

Additional information

Funding

The field and laboratory activities of this research was funded by Ethiopian Institute of Agricultural Research (EIAR).