Productivity and Stability Evaluation of 12 Selected Avena magna ssp. domestica Lines Based on Multi-Location Experiments during Three Cropping Seasons in Morocco

Abstract

Avena magna (2n = 4x = 28) is a tetraploid oat with a very high protein content compared to the hexaploid common oat, A. sativa (2n = 6x = 42). The wild type of A. magna originates from Morocco; its domestication has been achieved only within the past 25 years. The present study aimed to evaluate the productivity potential of an A. magna ssp. domestica collection of 11 advanced lines and a control variety, ‘Avery’. Twelve trials were conducted during three cropping seasons at four, three, and five locations and revealed significant differences among the accessions. Data on twelve agro-morphological characters and two disease traits were collected, and they confirmed the presence of variability in this oat germplasm set. Mean grain yield was 30.76 q/ha and varied from site to site, ranging from 6.89 q/ha at Bouchane_19 to 85.5 q/ha at Alnif_21. Across experimental sites, plant height ranged from 48.93 to 120.47 cm; thousand kernel weight from 32.83 to 49.73 g; and harvest index from 20.43 to 31.33%. Line AT6 was relatively tolerant of BYDV and crown rust infections, based on disease severity scoring at the heading stage. According to AMMI analysis, 78% of the grain yield variability was due to the environment factor and 4% was explained by the genetic factor. Among the highest-yielding lines, AT5 and ATC were relatively unstable. Line AT5 was more productive at the elevated site of El Kbab_19, and ATC performed better at the oasis location of Alnif_21 under irrigation. Line AT7 showed the most stable behavior; it was high yielding across the five environments and exceeded the general mean of the experiments. The A. magna ssp. domestica lines proved their suitability for cultivation under local farming conditions. Their nutritional quality, especially their high protein content, makes them good candidates for further testing in the Moroccan breeding program and for integration into local cropping systems.

1. Introduction

An ever-increasing world population and climate change are among the major contributors to famine. Crop yields must continue to increase to attain the Sustainable Development Goals and ensure safe, nutritious, and adequate food for everyone. Cereals and derivatives should play a key role in this challenge since they are the staple food of most countries of the world and the primary food for livestock. Cereal grains constitute more than half of the food energy and half of the protein consumed on Earth [1]. With the pressure on resources and the climate change scenario, future food supplies not only need to be increased but also enriched especially in nutrients and protein to address food and nutrient security. Nutrition deficiencies have a significant impact on a country’s productivity and can reduce its gross domestic product (GDP) by as much as 7% globally and up to 9–10% in African and South Asian countries [2]. In this context, there is a need to develop new sustainable sources of protein and nutrients to avoid the 150 more people expected to be at risk of protein deficiencies by 2050 [3]. It is evident that protein plays an essential role within the human body, including building muscle and bone and brain development. Its components, amino acids, are the building blocks of this development and are critically important in the first 1000 days of life and throughout the lifecycle [4]. Animal-based foods are excellent sources of protein but contribute 14.5% of all greenhouse gas emissions [5]. In this view, development of new nutrient-rich crops such as high-protein oat, Avena magna ssp. domestica, becomes one of the pathways to diversify the sources of protein and help attack the dual challenge of increasing supply in a sustainable manner.

Oat ranks sixth in world cereal production after wheat, maize, rice, barley, and sorghum [6]. The most widespread cultivated species of oat is common oat (Avena sativa), with Abyssinian oat (A. abyssinica), diploid naked or hulless oats (A. nuda), and lopsided oat (A. strigosa) being of relatively minor importance [7]. Morocco is the center of origin and a major center of diversity of the Avena genus [8]; consequently, it represents an essential country for expanding genetic resource conservation efforts and an interesting location for expanding oat production.

Oat is a cereal that has long been prized for its nutritional attributes, particularly in terms of mixed-linkage soluble beta-glucan, protein, oil, and antioxidants. Oat grain is also free-gluten and has the highest protein content among cereals; as a consequence, it is tolerated by most celiac disease patients [9]. Several previous studies have shown oat’s health-promoting effects in preventing obesity, type II diabetes, gastrointestinal diseases, coronary heart disease, and certain types of cancer [10]. For human consumption, an increasing supply of oat-based products has been exhibited in the market, including a diverse array of breads, oatcakes, biscuits, granola bars, and ever more novel foods such as yogurt-type products and oat-based drinks [11,12,13].

The newly domesticated Moroccan tetraploid species A. magna (2n = 4x = 28, CCDD genome) was developed via sexual transfer of the domestication syndrome from the hexaploid common oat A. sativa (2n = 6x = 42, AACCDD; [14]). Subsequently, Jackson [7] produced a set of A. magna ssp. domestica lines via mutagenesis within a population of F_2:8 recombinant inbred lines (RILs) from a cross between a stable domesticated backcross line, Ba13-13, and a wild A. magna parent [15]. Oliver et al. [15] also mapped three of the domestication syndrome genes, for shattering (Ba), geniculate awn (Awn), and lemma pubescence (Lp), and measured the linkage of Awn and Ba with a terminal knob on chromosome 2C at a distance of 2.1 cM.

While the lines developed by Jackson [7] are suitable for production under standard cultivation practices, they also possess much higher protein levels than common oats, exceeding 25% in the relatively new cultivar ‘Avery’. Avena magna ssp. domestica also showed promise in improving dietary iron and zinc in comparison to other cereals, thus potentially reducing anemia [16,17]. These characteristics make A. magna a good grain resource for the development of high-quality protein products to address developing-world malnutrition, developed-world obesity, and high-quality livestock feed [11]. Since A. magna is already adapted to the semi-arid conditions of Morocco, it is a promising source of sustainable grain for a world increasingly affected by climate change. Nevertheless, there is still a need to continue breeding under Moroccan conditions for traits such as resistance to seed dehiscence, lodging, dormancy, productivity, and stability, as well as tolerance of diseases.

The main objective of this study was to assess the diversity and performance of A. magna ssp. domestica lines using ‘Avery’ as the control in seven diverse environments in Morocco. The evaluation included morphological and yield parameters, disease tolerance (crown rust, BYDV), and yield stability.

2. Materials and Methods

2.1. Plant Materials

Eleven A. magna ssp. domestica lines were used in this study along with ‘Avery’, which was registered in the Official Moroccan Catalog in 2019, as control (

Table 1. List of A. magna ssp. domestica lines used in this study.

IAV_ID	Correspond	LINE_ID	PEDIGREE
ATC (Avery)	A40	BAM_96-5-6	BAM_96-5-6
AT1	A18a	2013Y1193	BAM_34/55_1
AT2	A14	2013Y1397	BAM_6/235_44
AT3	A27	2013Y1291	BAM_55/231_22
AT4	A35	2013Y1302	BAM_55/231_33
AT5	A44	2013Y1307	BAM_55/231_38
AT6	A45	2013Y1200	BAM_34/55_36
AT7	A02	2013Y1297	BAM_55/231_28
AT9	A20	2013Y1275	BAM_55/231_6
AT13	A41	2013Y1508	BAM_34/235_21
AT14	A43	2013Y1310	BAM_55/231_41
AT15	A10	2013Y1373	BAM_6/235_20

Explanation: IAV_ID: Institut Agronomique et Vétérinaire Hassan II Identifier. ATC: Avena magna spp. domestica variety ‘Avery’.

). These lines were selected from a set of 41 lines introduced in Morocco during the 2017–2018 and 2018–2019 cropping seasons for adaptation trials. A. magna was domesticated via hybridization with common hexaploid oat A. sativa followed by a single backcross cycle to produce line Ba13-13 [14]. Subsequently, Jackson [7] produced six foundational A. magna ssp. domestica lines from a domesticated Ba13-13 × wild A. magna # 169 population [15,18], segregating for the wild-type growth habit and resistance to field races of crown rust (Puccinia coronata) in Baton Rouge, LA, USA. According to Jellen et al. [19], a virtual pedigree using a genotype/phenotype model with the JMP Genomics software package version 10 (SAS Institute Cary, NC, USA) was used as instruction to produce the 41 lines and the internal control Avery by intercrossing the foundational lines.

2.2. Experimental Locations

Field experiments were conducted in three consecutive growing seasons (2018–2019, 2019–2020, and 2020–2021), and at 4, 3, and 5 experimental sites in each of these respective winter growing seasons. The selected experimental sites belong to contrasting Emberger bioclimatic stages across the central part of Morocco including per-arid (Alnif in the Anti-Atlas/Sahara); arid (Bouchane in the Phosphate Plateau); semi-arid cold winter (El Kbab in the Middle Atlas); sub-humid (Ain Itto in the Saïs Plain); and sub-humid cold winter (Oukaimeden in the High Atlas; [20]).

Among the twelve trials conducted in this investigation, one was not successful; data from the other eleven trials were subjected to ANOVA analysis. As four tests did not reveal any productivity differences among the advanced lines, the productivity and stability evaluation comprised the seven remaining trials.

Site descriptions for the seven retained trials are reported in

Table 2. Location and description of the experiment agro-climatic sites. Climate data were provided by worldweatheronline.com. Only data covering the periods of the experiments (between October and July) are reported.

Location	Zone	Geographic Position		Alt	Soil Type	Temp		Rain	Hum	S.H	Harvest Date
		Latitude	Longitude			Min	Max
Ain Itto_21	Saïss Plain	34°05′33″ N	5°48′54″ W	855	Vertisols	8	36	609.8	61.8	3099	2 June 2021
Alnif_21	Oasis	31°06′37″ N	5°03′50″ W	1320	Fluvisols	9	43	43.2 *	26.2	3529	17 May 2021
Bouchane_21	Phosphate Plateau	32°14’35″ N	8°19’45″ W	830	Cambisols	9	38	250.2	54.1	3304	25 May 2021
Oukaimeden_21	Atlas Mounain	31°14′07″ N	7°48′42″ W	2530	Fluvisols	11	40	319.3	47.0	3369	25 June 2021
Bouchane_20	Phosphate Plateau	32°14’35″ N	8°19’45″ W	830	Cambisols	9	40	105.5	53.7	3376	22 June 2020
El kbab_19	Middle Atlas	32°42’25″ N	5°31’57″ W	1540	Fluvisols	2	35	573.0	50.9	3477	3 July 2019
Bouchane_19	Phosphate Plateau	32°14’35″ N	8°19’45″ W	830	Cambisols	9	33	313.1	53.4	3463	20 June 2019

With additional 200 m³ of irrigation during the cycle development of plants. * By simplifying the number added on each location, the cropping season is specified, and each number is used for the rest of the article: 21 (season 2020–2021), 20 (season 2019–2020), and 19 (season 2018–2019). Alt: altitude (m); Temp: temperature (°C); Rain: rainfall (mm); Hum: humidity (%); S.H: sun hour (h).

. During the 2018–2019 season, experimental trials were installed at El Kbab (mountain plateau) on 29 November and Bouchane (Phosphate Plateau) on 30 November. Experiments for the following season, 2019–2020, were conducted at only one location, Bouchane, with a sowing date of 23 November. The third season (2020–2021) consisted of four experimental stations: Ain Itto, Bouchane, Oukaimeden, and Alnif. Seeds were sown on 16 October in Oukaimeden, 18 October in Bouchane, 20 October in Ain Itto, and on 21 November in Alnif. Table 2 summarizes the experimental locations’ specifications, the type of environmental climates (warm to cool), and the altitudes at the trial sites. Photos of some experimental sites are shown in Figure 1.

In Alnif, Ain Itto, and Bouchane, mechanized moldboard plowing was used to prepare the soil for sowing, while for El Kbab and Oukaimeden, tillage was performed with a horse-drawn plow. Depending on the germination rate, the sowing dose was calculated to establish a density of 25 plants per linear meter. Trials were fertilized at the early tillering stage in the form of ammonium nitrate at the rate of 45 Kg/ha. Accessions were evaluated under rainfed conditions except for the Alnif site, where irrigation (1 m³ of water per m²) was supplied due to aridity. Weed control was performed manually at the tillering and flowering stages; if necessary, the number of weeding times exceeded two, depending on the degree of weed infestation, to avoid competition between plants.

2.3. Experimental Design

The field experiment design was a randomized complete block with three replications. The sowing plan varied from one site to another according to land availability for the trial. For the 2018–2019 season, the elementary plot included three six-meter rows at Bouchane and four three-meter rows at El Kbab with an inter-row spacing of 0.5 and 0.35 m, respectively. The space between elementary plots was 1.2 m for Bouchane and 1 m for El Kbab. Plots at Bouchane during the 2019–2020 season had four rows of 3.6 m length and were spaced 0.30 m apart. The 2020–2021 experimental plan was carried out in six rows of 12.5 m length and 0.3 m spacing. However, there was an exception at the Alnif and Oukaimeden sites, where eight rows of 8 m and four rows of 3 m were used, respectively, because of the land space limitations. The seeding rate per row was adjusted accordingly in those two sites.

2.4. Notations and Measurements

Crown rust (RC) and barley yellow dwarf virus (BYDV) assessments were made under natural infection at the flowering stage. The percentage of leaf area showing pathogen symptoms was scored for each accession and site where diseases were noticeable. At maturity, 12 agro-morphological traits were scored on harvested individual plants. Plant height (PH, cm) was measured at maturity from the soil surface to the tip of the panicle. After hand-harvesting of the four plants per elementary plot, the panicles were threshed and dried at 35 °C for seeds and 70 °C for other parts of the plant (stem weight SW (g) and root weight RW (g)). Grain yield per plant YP (g) and dry matter per plant DMP (g) were measured using a weighing scale. Harvest index (HI, %) was derived as the ratio of grain yield to total biomass by the formula: HI = (YP/(YP + DMP)) × 100. At harvest, plant density was estimated by counting the number of plants on a one-meter row length. Subsequently, the yield in quintals per hectare (Yield q/ha) and dry matter in quintals per hectare (DM q/ha) were estimated according to plant density, yield per plant, and the plot area which varied between experimental sites. The root length RL (cm), number of fertile tillers NFT, and number of spikelets per panicle (NSP) were also assessed. Furthermore, the weight of a thousand unshelled kernels (TKW g) was estimated using three samples of 250 seeds.

2.5. Statistical Analysis

Collected data were subjected to a variety of statistical analyses. Descriptive statistics, two- and three-way analyses of variance (ANOVA), and multivariate analyses such as principal component analysis (PCA) and additive main effects and multiplicative interaction (AMMI) were performed to assess the genetic diversity in terms of productivity, yield stability, and disease tolerance among lines and experimental sites. The R software version 3.5.1 was used for these tests. The ANOVA tests were conducted to assess the variability among lines within and between sites. For significant ANOVAs, determination of homogeneous groups of lines was performed using the Newman–Keuls post hoc test. Associations among pertinent traits, which were useful for line classifications, as well as significant correlations between those traits, were performed through multivariate principal component analysis (PCA) with Factominer package version 2.4 [21]. The AMMI analysis via R [22] combines both ANOVA and PCA to determine the main effects and genotype-by-environment interactions (GEI) in multi-location trials. The interaction principal component axes, IPCA1 and IPCA2, are calculated and displayed as biplots to project GEI patterns graphically. Initially, adaptability and phenotype stability across the locations were performed by the AMMI method described by Zobel et al. [23]. Subsequently, Purchase [24] developed the AMMI stability value (ASV) based on the AMMI model’s principal components axis 1 and 2 scores for each genotype, respectively. Genotypes with small values of IPCA1, IPCA2, and ASV are more stable across environments. Of course, yield stability is a critical factor to protect small-holding subsistence farmers from starvation in adverse growing seasons.

3. Results

3.1. Disease Assessments

Crown rust (RC; Puccinia coronata f. sp. avenae), barley yellow dwarf virus (BYDV), and powdery mildew (Blumeria graminis f. sp. avenae) are the three most common diseases found on oats in Morocco [19]. Powdery mildew was not noticeable in the experimental trials; consequently, only RC and BYDV infections were assessed at the flowering stage in Bouchane_19, El Kbab_19, Bouchane_21, Ain Itto_21, and Oukaimeden_21.

Table 3. Summary one-way ANOVA for BYDV and RC infestations.

Site	Season	Trait	F	Mean (%)	p_Value
Bouchane_19	2018–2019	BYDV	2.109 NS	6.11	0.061
		RC	6.811 ***	20.28	0.000
El Kbab_19	2018–2019	BYDV	0.731 NS	4.09	0.699
Bouchane_21	2020–2021	BYDV	7.928 ***	14.25	0.000
Ain Itto_21	2020–2021	BYDV	5.437 ***	14.62	0.000
Oukaimeden_21	2020–2021	BYDV	5.377 ***	9.38	0.000

NS = not significant; *** significant at p = 0.001. BYDV: barley yellow dwarf virus; RC: Crown rust.

shows that BYDV symptoms were more frequent than the two other diseases in five out of seven trials. The ANOVA test for BYDV revealed significant differences among oat lines at three sites: Bouchane_21, Ain Itto_21, and Oukaimeden_21. The average leaf area covered by the virus was equal to 14.62 and 14.25% in Ain Itto_21 and Bouchane_21, respectively. El Kbab_19 BYDV infection was the lowest at only 4.09% and with no significant differences among lines. Regarding RC, the ANOVA test was significant in Bouchane_19 with an average infection value of 20.28%.

The Student–Newman–Keuls test allowed ranking of the oat lines in locations where disease infections were significant (

Table 4. Disease rankings of the 12 lines using the Student–Newman–Keuls test.

BYDV (%)									RC (%)
Bouchane_21			Ain Itto_21			Oukaimeden_21			Bouchane_19
AT6	9.00	a	AT3	8.33	a	AT6	5.00	a	ATC	10.00	a
AT4	10.00	ab	AT6	10.00	ab	AT13	7.50	b	AT6	13.33	ab
AT13	11.00	ab	AT14	11.67	abc	AT1	8.75	b	AT2	13.33	ab
AT5	12.00	ab	AT1	13.33	abcd	AT5	8.75	b	AT5	13.33	ab
ATC	12.00	ab	ATC	13.33	abcd	AT14	10.00	bc	AT1	16.67	ab
AT3	12.50	ab	AT2	14.17	abcd	AT15	10.00	bc	AT3	16.67	ab
AT15	13.50	ab	AT9	15.83	bcd	AT2	10.00	bc	AT4	16.67	ab
AT14	16.00	bc	AT15	16.25	cd	AT3	10.00	bc	AT14	23.33	ab
AT2	16.00	bc	AT7	17.08	cd	AT7	10.00	bc	AT15	23.33	ab
AT1	19.00	c	AT13	17.50	cd	AT9	10.00	bc	AT13	26.67	bc
AT9	19.00	c	AT4	18.33	d	ATC	10.00	bc	AT7	30.00	c
AT7	21.00	c	AT5	19.58	d	AT4	12.5	c	AT9	40.00	d

Values within a column followed by a common (a–d) are not significantly different (p < 0.05). BYDV: barley yellow dwarf virus; RC: crown rust.

). There was an irregular ranking of the BYDV infection across sites apart from line AT6. This accession tended to show more tolerance than the others with a value of 9% at Bouchane_21 (Rank 1), 10% at Ain Itto_21 (Rank 2), and 5% at Oukaimeden_21 (Rank 1). It was also ranked second for RC at Bouchane_19 (13.33%).

3.2. Analyses of Agro-Morphological Traits

The average values (means) and coefficients of variation (CVs) of the different morphological (PH, RL) and agronomic (NFT, Yield, DM, HI, etc.) traits across sites are presented in

Table 5. Descriptive statistics of the agro-morphological traits of domesticated A. magna for the three cropping seasons.

	Season 2018–2019				Season 2019–2020		Season 2020–2021
	El Kbab_19		Bouchane_19		Bouchane_20		Ain_Itto_21		Alnif_21		Bouchane_21		Oukaimeden_21
Traits	Mean	CV	Mean	CV	Mean	CV	Mean	CV	Mean	CV	Mean	CV	Mean	CV
PH	72.86	13.92	48.93	21.18	87.91	14.70	118.47	15.65	116.89	11.44	64.61	14.08	120.47	9.97
RL	15.41	19.84	16.16	22.28	15.66	21.83	14.19	20.05	12.58	19.69	14.42	22.6	11.87	14.10
NFT	4.35	69.81	3.10	34.25	4.80	44.98	4.42	43.24	11.09	32.09	3.59	51.99	5.60	36.07
DMP	15.36	56.65	9.87	74.84	9.35	88.19	12.54	41.16	35.51	40.92	4.08	58.74	13.59	48.41
YP	9.69	59.6	5.63	86.16	3.18	65.59	4.74	52.71	16.1	51.14	1.83	69.37	5.92	51.28
HI	25.74	79.56	20.43	86.38	27.23	27.33	27.12	22.53	30.68	23.67	31.33	24.14	30.37	17.32
Yield	51.96	77.79	23.19	85.76	11.85	62.79	20.17	55.77	85.5	55.78	6.89	87.64	18.99	60.83
DM	78.33	67.19	42.76	82.94	34.41	80.98	53.5	46.01	188.15	45.77	15.34	81.72	43.85	59.30
NSP	53.89	58.81	29.74	76.45	16.17	38.40	25.3	52.15	43.43	34.23	14.24	53.21	24.83	32.50
TKW	45.73	11.84	37.66	55.58	40.61	7.58	43.94	6.82	42.76	6.98	34.29	45.11	40.04	7.95
SW	13.74	56.7	8.55	77.51	NA	NA	10.86	41.64	26.63	41.7	2.83	60.01	11.61	46.57
RW	1.62	72.66	1.33	74.59	NA	NA	1.68	48.65	8.88	51.83	1.25	63.9	1.98	66.73

Explanations: CV: coefficient of variation; PH = plant height (cm); RL = root length (cm); NFT = number of fertile tillers; DMP = dry matter per plant (g); YP = grain yield per plant (g); HI = harvest index (%); Yield = grain yield per hectare (q); DM = dry matter per hectare (q); NSP = number of spikelets per panicle; TKW = thousand-seed weight (g); SW = stem weight (g); RW = root weight (g); NA: not available.

. Plant height (PH) was less variable than the other traits, with CVs of 9.97% at Oukaimeden_21 and 21.18% at Bouchane_19. The variable TKW also possessed small CVs among samples but was quite variable from site to site (6.82% and 55.58%, respectively). As expected, traits related to grain yield or biomass showed large variation. For example, the CV for yield was over 55% across sites. The Alnif_21 trial yield was the highest at 85.5 q/ha, while the Bouchane_21 trial was the least productive at just 6.89 q/ha. Bouchane_19 plants, in comparison to the other sites, were the shortest (48.93 cm) and had the deepest roots (16.16 cm).

The two-way ANOVA detected significant differences among lines for all the investigated traits except for DMP, TKW, and SW at Bouchane_19.

Table 6. Ranking of the 12 lines for grain yield per hectare at each site where it was significant across the three cropping seasons.

Season 2018–2019						Season 2019–2020			Season 2020–2021
Bouchane_19			El Kbab_19			Bouchane_20			Ain_Itto_21			Alnif_21			Bouchane_21			Oukaimeden_21
AT5	39.45	a	AT5	103.17	a	AT14	22.13	a	AT3	31.64	a	ATC	114.9	a	AT4	12.26	a	AT9	34.83	a
AT9	28.86	ab	ATC	73.8	b	AT3	15.41	b	AT15	27	b	AT1	103.7	ab	AT7	11.84	a	AT5	27.11	ab
AT2	27.21	ab	AT1	68.52	bc	AT15	15	bc	AT13	22.29	bc	AT7	97.1	abc	AT1	9.28	ab	AT7	24.67	bc
AT3	25.09	ab	AT3	62.7	bc	AT13	14.14	bc	AT6	22.1	bc	AT14	93.5	abc	AT5	9.13	ab	AT14	19	bc
AT13	24.85	ab	AT6	59.01	bc	AT1	13.4	bc	AT1	21.69	bc	AT4	90.7	abc	AT2	8.88	ab	AT4	18.96	bc
AT7	24.17	ab	AT2	58.09	bc	AT4	11.23	bc	AT14	21.14	bc	AT15	90.3	abc	AT3	6.9	abc	AT13	18.48	bc
AT1	23.99	ab	AT7	44.8	bcd	AT9	10.82	bc	AT9	20.49	bc	AT6	88.2	abc	AT6	5.19	bc	AT15	17.24	bc
ATC	23.19	ab	AT13	38.34	bcd	AT5	9.99	bc	ATC	16.82	cd	AT13	87.7	abc	AT9	4.75	bc	AT2	15.47	bc
AT4	19.3	ab	AT4	36.12	bcd	AT7	9.99	bc	AT7	16.54	cd	AT5	86	abc	AT14	4.59	bc	AT6	14.24	c
AT14	15.24	b	AT14	30.76	cd	AT6	9.19	bc	AT5	16.46	cd	AT3	61.6	bc	AT13	4.2	bc	AT3	13.59	c
AT6	15.15	b	AT15	13.47	d	AT2	7.4	bc	AT4	15.93	cd	AT2	58.7	bc	ATC	2.83	c	AT1	12.6	c
AT15	9.08	b	AT9	10.35	d	ATC	6.67	c	AT2	9.99	d	AT9	53.7	c	AT15	2.81	c	ATC	11.64	c

Values within a column followed by a common (a–d) are not significantly different (p < 0.05).

summarizes the line rankings for seed yield at the experimental sites based on the Student–Newman–Keuls test. Table 6 shows that there was no dominant line across all locations. For example, the ATC control performed well in El Kbab_19 (73.8 q/ha) and Alnif_21 (114.9 q/ha), while it ranked last at Bouchane_20 (6.67 q/ha) and Oukaimeden_21 (11.64 q/ha). Line AT5 had the highest yield at Bouchane_19 (39.45 q/ha) and El Kbab_19 (103.17 q/ha); it yielded relatively well at Bouchane_21 (9.13 q/ha) and Oukaimeden_21 (27.11 q/ha) but was less productive at the other locations.

Merely considering line performance through all sites combined does not allow for their correct ranking because they showed contrasting productivity potential. The three-way ANOVA was conducted to take into account line × environment interaction. The ANOVA_3 test revealed significant differences among genotypes for most of the traits except TKW. Furthermore, it showed a high location effect on genotype morphology and productivity.

The ranking of the experimental sites for productivity parameters showed that average Yield varied from 6.89 to 85.5 q/h; DM from 15.34 to 188.15 q/ha; and HI ranged from 20.43 to 31.33%. Alnif_21 and El Kbab_19 produced both high grain and biomass yields in comparison to the other sites. During the 2019_2020 and 2020_2021 seasons, the arid Bouchane site produced less than the other locations.

The Student–Newman–Keuls test results of the three-way ANOVA are reported in

Table 7. Ranking of 12 lines for principal agro-morphological traits with ATC (Avery) as control.

Agronomic Traits								Morphological Traits
Yield		DM		HI		TKW		PH		RL		NTF		NSP
AT5	43.41 a	AT1	80.05 a	AT5	34.76 a	AT5	49.73 a	AT15	94.92 a	AT15	15.61 a	AT1	5.93 a	AT7	36.06 a
AT1	35.93 ab	AT5	69.24 ab	AT4	31.78 b	AT4	42.94 ab	AT13	94.57 a	AT13	15.27 ab	AT14	5.69 ab	AT13	34.34 ab
ATC	35.66 ab	ATC	67.33 abc	AT3	31.33 b	AT9	41.93 abc	AT7	92.71 ab	AT14	15.23 ab	AT9	5.64 ab	AT14	33.13 ab
AT3	31.74 bc	AT13	66.75 abc	ATC	31.20 b	AT1	41.62 abc	AT5	89.93 bc	AT7	14.96 abc	AT4	5.6 ab	AT5	31.56 abc
AT7	31.64 bc	AT6	65.85 abc	AT6	30.35 b	AT7	41.27 abc	AT4	89.76 bc	AT4	14.83 abcd	AT6	5.51 ab	AT15	30.48 abc
AT6	30.15 bc	AT3	63.36 abc	AT2	29.67 b	AT6	40.99 abc	AT9	88.94 bcd	AT1	14.31 bcd	AT5	5.17 abc	ATC	30.47 abc
AT13	29.78 bc	AT4	62.83 abc	AT1	29.47 b	AT2	40.29 bc	AT1	88.58 bcd	AT6	14.13 bcd	ATC	4.98 bc	AT3	30.39 abc
AT14	28.75 bc	AT14	61.77 bc	AT7	22.57 c	AT14	40.19 bc	AT3	88.56 bcd	AT3	14.11 bcd	AT2	4.94 bc	AT1	28.77 abc
AT4	28.07 bc	AT2	61.28 bc	AT9	22.17 c	AT15	39.60 bc	AT14	88.07 cd	AT9	13.85 cd	AT13	4.86 bc	AT4	28.50 abc
AT2	27.32 bc	AT7	58.33 bc	AT14	20.33 c	AT13	38.72 bc	ATC	85.06 de	AT2	13.81 cd	AT3	4.58 cd	AT6	28.04 bc
AT15	23.90 c	AT15	52.64 bc	AT15	20.28 c	ATC	37.97 bc	AT6	84.64 de	ATC	13.65 d	AT7	4.45 cd	AT9	25.23 c
AT9	23.11 c	AT9	49.99 c	AT13	19.88 c	AT3	32.83 c	AT2	84.07 e	AT5	13.52 d	AT15	4.03 d	AT2	24.85 c

Values within a column followed by a common (a–e) are not significantly different (p < 0.05). Yield = grain yield per hectare (q); DM = dry matter per hectare (q); HI = harvest index (%); TKW = thousand-seed weight (g); PH = plant height (cm); RL = root length (cm); NFT = number of fertile tillers; NSP = number of spikelets per panicle.

. Combining all the sites’ data showed clearly that AT5 was the most productive line. It was ranked first for grain yield (43.41 q/ha), HI (34.76%), and TKW (49.73 g). For the DM, it was ranked second after AT1 with a value of 69.24 q/ha. On the other hand, the least productive lines were AT15 (23.90 q/ha) and AT9 (23.11). Regarding the morphological traits, the accession PH ranged between 84.07 cm (AT2) and 94.92 cm (AT15). Lines ATC (85.06 cm), AT6 (84.64 cm), and AT2 (84.07 cm) were the shortest. Root length (RL), NFT, and NSP varied between 13.52 (AT5) and 15.61 cm (AT15); between 4.03 (AT15) and 5.93 (AT1); and between 24.85 (AT2) and 36.06 (AT13), respectively.

3.3. Principal Components Analysis

The principal component analysis (PCA) was conducted to identify the main traits that contribute to differentiation among lines. PCA considers all the variables at the same time to cluster the oat lines through their similarities. The PCA outputs showed that 84.57% of the variability was explained by the first four principal components’ axes. This high percentage reflects strong discrimination among the assessed lines. The main contributors to the first principal component (PC1) were PH, RL, HI, and RC (35.73%). Thus, PC1 can be considered an indicator of the plant morphology behind the RC disease. Both PC2 and PC3 accounted for 37.71% of the total variation; they were mainly linked to the grain and biomass yield components: NTF, NSP, SW, DMP, YP, DM, and Yield. The fourth axis was essentially explained by the TKW and BYDV degree of susceptibility.

According to the Pearson correlations matrix (

Table 8. Pearson correlation with traits from principal component analysis results.

	PH	RL	NFT	NSP	RW	SW	DMP	YP	Yield	DM	HI	TKW	RC	BYDV
PH	1
RL	0.72	1
NFT	−0.48	−0.31	1
NSP	0.60	0.54	−0.38	1
RW	0.22	0.41	0.03	0.63	1
SW	0.27	0.32	0.62	0.10	0.15	1
DMP	0.36	0.55	0.51	0.39	0.49	0.82	1
YP	0.43	0.34	0.36	0.63	0.65	0.67	0.81	1
Yield	−0.15	−0.47	0.13	0.32	−0.07	0.00	−0.06	0.30	1
DM	−0.26	−0.26	0.42	0.12	−0.27	0.29	0.25	0.19	0.75	1
HI	−0.59	−0.74	0.25	−0.36	−0.50	−0.11	−0.41	−0.24	0.63	0.53	1
TKW	0.06	−0.21	0.36	−0.04	0.10	0.28	0.10	0.35	0.40	0.14	0.23	1
RC	0.54	0.38	−0.05	0.14	0.36	0.24	0.31	0.31	−0.56	−0.63	−0.78	−0.01	1
BYDV	0.52	0.24	−0.10	0.23	0.13	0.17	0.10	0.30	−0.09	−0.29	−0.45	0.44	0.62	1

Bolded values are significant Pearson correlation. PH = plant height (cm); RL = root length (cm); NFT = number of fertile tillers; NSP = number of spikelets per panicle; RW = root weight (g); SW = stem weight (g); DMP = dry matter per plant (g); YP = grain yield per plant (g); Yield = grain yield per hectare (q); DM = dry matter per hectare (q); HI = harvest index (%); TKW = thousand-seed weight (g); RC: crown rust (%); BYDV: barley yellow dwarf virus (%).

), DMP exhibited both highly significant and positive correlations with YP and SW (R² = 0.81; R² = 0.82). Yield per plant (YP) was also positively correlated with NSP and RW (R² = 0.63 and R² = 0.65). Grain yield per hectare (Yield) and DM were positively correlated at 75%. A negative correlation was found between the HI and PH (R² = −0.59), RL (R² = −0.74), and RC (R² = −0.78). On the other hand, the HI showed a positive relationship with grain yield (R² = 0.63). The susceptibility to BYDV and RC diseases appeared together in oat plants up to 62%.

The biplot PC1 × PC2 revealed four accessions’ clusters (Figure 2). Two accessions, AT5 and AT1, formed cluster I, which included high biomass and grain production potential. Cluster II gathered AT13 and AT14, which had good individual plant performance (grain yield and dry matter per plant). Cluster III included ATC and AT6, short size, and high harvest-index lines with better tolerance to BYDV and RC infection than AT7 and AT9. Cluster IV included AT4 with its high TKW.

3.4. AMMI Analysis

An AMMI analysis was performed to scrutinize yield stability of the experimental lines across environments. The analysis of variance (ANOVA) associated with the AMMI model revealed highly significant site, line, and their interaction (S × L) effects (p-value ≤ 0.001). The environment explained the largest grain yield variability (78%), followed by the interaction (18%;

Table 9. Additive main effects and multiplicative interaction analysis (AMMI) analysis of variance for grain yield of 12 lines for three cropping seasons.

Source	Df	Sum Sq	Mean Sq	F Value	Pr (>F)	% Variation
Site	6	596,996	99,499	12.32 ***	6.92 × 10−5	78
Line	11	27,740	2522	6.04 ***	1.40 × 10−9	4
S × L	66	136,793	2073	4.96 ***	2.20 × 10−16	18
IPCA1	16	19,821.10	1238.82	2.97 ***	0.0001	59.7
IPCA2	14	10,539.33	752.81	1.80 *	0.0344	31.7
Residuals	897	374,620	418

* significant at p = 0.05; *** significant at p = 0.001.

). Only 4% of the total sum of squares was assigned to the genetic factor of the A. magna ssp. domestica lines—the large contrast between the sites was intended to assess the specific adaptation of the experimental lines through the GxE interactions. Differences among lines were more expressed within the experimental sites. The two first axes, IPCA1 and IPCA2, explained 59.7% and 31.7%, respectively, of the genotype×environment interaction (GEI) sum of squares.

Figure 3 displays a scatter plot formed by genotypes and experimental locations according to their coordinates on the two first axes of GEI. The values of these coordinates indicate the accessions’ contribution to GEI. In this case, the lower the coordinate is, the lower its contribution to GEI and therefore the more stable the genotype is. Thus, lines AT7, AT6, AT13, and AT4 being closer to the biplot’s center have more stable grain production across the experimental sites. In contrast, grain yields of the AT5, AT9, AT15, and ATC lines are unstable from site to site, as their position is far away from the biplot center. The other accessions are ranked intermediate with respect to their grain yield stability (AT1, AT2, AT3, and AT14). Furthermore, the length of vectors assigned to the experimental locations shows their ability to discriminate among the oat lines. El Kbab_19 and Alnif_21 sites showed more yield variation among lines. When a genotype projection is close to a site position on the biplot, that means the line has a specific adaptation to it; for example, AT5 was better suited to El Kbab_19 (103.17 q/ha), and ATC interacted positively with the Alnif_21 environment (114.9 q/ha), as did AT9 (34.83 q/ha) with Oukaimeden_21.

AMMI analysis provided comparisons between productivity and stability (Figure 4). We observed two main categories of the studied lines. The first group consisted of high-yielding lines that exceeded the experimental site average yield (30.76 q/ha). Among these lines, two were the most productive, AT5 and ATC, though they were quite unstable (ASV = 9.62 and 6.37, respectively). Line AT1 had intermediate stability and a yield equal to 36.17 q/ha, while AT7, which was the most stable (ASV = 1.69), yielded 32.73 q/ha. The other lines produced below the mean yield, with AT6 and AT13 showing good yield stability and a slightly lower yield than average (30.44 q/ha and 29.24 q/ha, respectively).

4. Discussion

The goal of this present work was to determine the agro-morphological variability of 11 A. magna ssp. domestica advanced lines and a control cultivar named Avery (ATC). Avena sativa (2n = 6x = 42) is the principal species of oat cultivated in Morocco, usually for fodder [25]. At the beginning of the local oat research program of the National Institute of Agricultural Research, varieties derived from the native hexaploid A. byzantina (red oat), which has a low water requirement and the ability to adjust its growth cycle to Moroccan environmental variability, were also released [26]. The array of wild Avena species native to Morocco—the center of origin of the genus—unfortunately lacks the required traits for broad cultivation such as resistance to shattering, semi-dwarfism, lodging resistance, erect growth habit, resistance to seed dormancy, etc. The neo-domesticated A. magna material first described by Ladizinsky [14] and later modified by Jackson [7] should have the potential to diversify the cultivated germplasm of oats. This new oat diversity contributes to the conservation of genetic resources of the genus and plays a positive role in maintaining crop diversity [6,17]. In addition, the significantly higher protein content of ‘Avery’ previously reported by Jackson [7] could encourage farmers and manufacturers to harness this asset to improve the crop’s profitability for both human and animal feed. Over the course of the past 40 years, interest in oats for human consumption has gradually increased, concomitant with mounting clinical evidence regarding its nutritional benefits for cardiovascular health [27].

The disease assessment study had focused only on RC and BYDV, since there were no noticeable symptoms of powdery mildew at the heading stage when scoring was carried out. Several authors agree that RC (Puccinia coronata) is the major disease threatening global oat production [28]. This pathogen can drastically affect oat production by causing 10–40% yield loss [29]. Crown rust (RC) was only recorded at Bouchane in the 2018–2019 cropping season because at the other locations there were no noticeable symptoms in the experimental plots. The cause may be due to relatively dry and cool conditions at the experimental sites during 2020–2021. Nazareno et al. [28] reported that crown rust epidemics usually occur in warm areas as temperatures fluctuate between 20 and 25 °C and humidity surpasses 50–60%. At the El Kbab high-altitude site and in Alnif’s dry environment, weather conditions would be expected to suppress crown rust development. The incidence of BYDV was more common than RC across the experimental locations. BYDV is probably the most economically important small-grain virus transmitted by insects, with 25 aphid species being involved as potential BYDV vectors [30]. High light intensity and low temperatures (15–18 °C) also promote increased intensity of BYDV symptoms [31]. All of these considerations and A. magna’s well-known susceptibility to BYDV [14] may explain the high incidence of the virus observed in our experiments. The trials’ results did not show consistent tolerance to diseases of tested lines at all sites, except for line AT6, which appeared to be more tolerant to BYDV and RC infections. Combining data from all sites, PCA revealed that the control variety Avery and AT6 were relatively less susceptible to the diseases, which was not the case for AT7 and AT9. Jellen et al. [19] had already noted A02 (line AT7 in this study) among the lines showing RC symptoms in the Lahri experiment in 2018.

Data analyses revealed significant agro-morphological diversity between the domesticated A. magna lines. Plant height (PH) varied between 48.93 and 120.47 cm across the sites. Through the different sites, PH fluctuated between 84.07 cm for AT2 and 94.92 cm for AT15. Previous studies reported similar plant height values. Dumlupinar et al. [32], who studied A. sativa pure-line varieties, reported variation for plant height between 46 and 112 cm. Among our investigated traits, PH gave the lowest CV, between 9.97 and 21.18% across the five sites. This finding is in line with Solange et al. [33] and Dumlupinar et al. [32], who reported CVs of 11.16 and 8.7%, respectively, for this trait. A possible explanation for these observations is that plant height is affected by a relatively small number of major genes, which in this case might be largely fixed through selection. PCA analysis showed that PH was positively correlated with RL (0.72) and NSP (0.60); on the other hand, PH was significantly negatively correlated with HI. Other authors reported a significant negative correlation between PH and grain yield and explained the relationship as being due to lodging [34,35]. Plant size has to be examined closely in varietal selection to avoid lodging, in our case with lines AT13 and AT15.

Root length across locations ranged from 11.87 to 16.16 cm at Oukaimeden_21 and Bouchane_19, respectively. The longest roots were observed in Bouchane in three successive seasons. This behavior is likely due to the sandy soil at Bouchane, which facilitates and encourages deep root penetration as capillary water drains downward through the soil profile. Finer roots were also more easily retained as the plants were uprooted in the light-textured soil to take RL measurements. The higher clay content at the other experimental sites allowed for better soil water retention. This supports the previous observation that A. magna ssp. domestica has the ability to deeply penetrate the soil under water-stress conditions [19].

The three-way ANOVA detected significant differences among lines according to their tillering ability: 4.03 NFT/plant for AT15 and 5.93 for AT1 on average (Table 7). The genetic control of tillering capacity has been discussed previously in oat genotypes [36,37]. The mean NFT was mostly equivalent among our test locations except for Alnif_21, where it reached an average of 11.09. Good crop management practices such as irrigation and fertilization have a positive influence on the tillering ability of plants; this may explain the significantly higher NFT value at this irrigated site.

The NSP varied among lines, ranging from 24.85 to 36.06 for AT2 and AT7, respectively. Solange et al. [33] observed an NSP mean of 26.63 for A. sativa, which is close to our findings. However, Jan et al. [6] counted 66 spikelets per panicle on average in their experiment. The number of spikelets is influenced by soil fertility, humidity, and photoperiod.

Thousand-kernel weight (TKW) is an important productivity and grain-quality trait for oats. According to our three-way ANOVA, TKW was more significantly influenced by growing location than genotype. Considering separately the experimental sites, two-way ANOVA showed a significant effect of the accessions at all testing sites except Bouchane_19. Our line TKW values, similar to previous results in A. sativa, varied between 32.83 and 49.73 g for AT3 and AT5, respectively. Our results correspond well with A. sativa TKW values [6,32,33].

Seed partitioning on the total biomass is evaluated through the HI. Statistical analyses showed that HI was influenced by both the line and the environment; it varied between 20.43 and 31.33% across the five testing locations. Regarding the effect of the lines, the HI varied between 19.88% for AT13 and 34.76% for AT5. Jan et al. [6] and Li et al. [38] reported on variation for HI among A. sativa genotypes and between environments and years; their HI values were comparable to ours.

It has historically been a major challenge for plant breeders to consistently identify superior-yielding genotypes across environments when GEI is highly significant. Several researchers testing crops in multi-location conditions have reported larger environmental than genotypic effects on grain yield [39,40,41,42]. In our case, the environmental effect represented 78% of the total grain yield variation, while the genotype explained only 4%. Grain yield varied extensively among our sites: from 6.89 q/ha at Bouchane_19 to 85.50 q/ha at Alnif_21. Previously, within a larger set of A. magna lines, Jellen et al. [19] identified the environment as the most important factor influencing yield, with genotype being the second most significant factor. Our AMMI analysis allowed classification of the A. magna lines based on their ASVs. Combining ASV and yield into a single index should represent a better way to evaluate a genotype’s potential across locations. The yield stability index has been used by several authors to rank their germplasm [40,43,44]. Stable lines are usually less productive than low-stability ones. Stable genotypes have large adaptation and intermediate yield potential, while low-stability lines yield the maximum in the environment in which they are best adapted while ranking among the lowest in other locations. In this study, we presented a hybrid graph to display the A. magna line performance and separated them into two groups according to their ASV and grain yield. The first group included A. magna lines whose grain yield across locations exceeded the average yield of 30.76 q/ha. Two of the genotypes produced relatively higher grain yields, but they were unstable: AT5 (mean yield of 43.31 q/ha) Avery (35.66 q/ha). Line AT5 performed the best at El Kbab_19 while Avery produced the most at Alnif_21, with respective yields of 103.17 q/ha and 114.9 q/ha. Gadisa et al. [40], while investigating germplasm consisting of 15 bread wheat lines, noticed that the two highest-yielding genotypes were among the most unstable. Each of these two wheat lines had interacted positively at a given site. These results are in agreement with our findings. Among the tested A. magna lines, AT1 was moderately stable, while AT7 had the highest stability in the collection. Using the yield stability index, AT7 ranked first as being the superior genotype among the tested germplasm in terms of stability and grain yield. This accession, however, appeared to be less tolerant to diseases. Among the A. magna lines that produced below the mean grain yield, AT6 exhibited good yield stability and will be further scrutinized for its apparent tolerance to the main common diseases and its yield, which approached 30.44 q/ha.

The average grain yield of this present study across five sites and three years of experimentation was 30,76 q/ha. It may be recalled when Tam [45] tested 101 oat varieties of A. sativa from Germany, Sweden, Russia, Canada, the USA, and other countries, where the average yields ranged between 32.88 and 58.24 q/ha. In our trials, Bouchane’s relatively low average yield of 6.89 q/ha is explained by the semi-arid climate conditions; moreover, Morocco is an arid to semi-arid country, unlike Estonia where Tam carried out his experiment. At the most favorable Moroccan site, Ain Itto_21, rainfall barely reached 609.8 mm during the cropping season. However, it should be noted that with certain lines at certain locations, grain yield greatly exceeded the above-mentioned yield range: for example, AT5 at El Kbab_19 and Avery at Alnif_21. Tests of various A. sativa varieties selected for northern Morocco reached 45 q/ha under experimental conditions and 24 q/ha in farmers’ production fields [46]; these previous results confirm that grain yield tends to decrease under working farm conditions, and some A. magna lines might express their best yield potential under good cultivation and environmental conditions.

5. Conclusions

This present study focused on the agronomic performance of advanced A. magna ssp. domestica lines under working farm conditions at five contrasting locations in Morocco. The results measured diversity among the investigated lines for agro-morphological traits. Grain yield was most highly influenced by the contrasted environments, followed by GEI, with genotype explaining very little of the variation for yield. The mountainous site of El Kbab seems to be more suitable for growing A. magna under obligate rainfed conditions. The average grain yield obtained is in the range usually found for common oat production around the world, which justifies the adaptability of these A. magna ssp. domestica lines to their native zone. Line AT5 had the highest grain yield and is therefore expected to attain higher grain yields in different locations in Morocco experiencing ample rainfall. Line AT7 was interesting for its relatively high yield stability value, as measured by AMMI. We hope to be able to select one of these advanced lines for registration in Morocco along with the existing control variety, Avery. Further investigation of these lines’ nutritional differences and molecular genotyping of these lines are the next steps for facilitating further improvement of domesticated A. magna.