Morpho-physiological and multivariate assessment of stay-green sorghum (Sorghum bicolor (L.) Moench) genotypes under drought stress in Ethiopia

Abstract

Background

Drought is a major abiotic constraint that severely limits crop growth, development, and yield worldwide. In Sorghum bicolor (L.) Moench, post-flowering drought stress is particularly detrimental, significantly reducing grain yield and biomass accumulation. Enhancing drought tolerance through the stay-green trait remains a key breeding objective to improve productivity under water-limited conditions.

Results

A field experiment was conducted at Efratana Gidim District using an irrigation scheme to impose two water regimes: post-flowering water deficit and well-watered conditions. The trial was arranged in an alpha-lattice design to evaluate performance and determine genetic variation for yield and related traits among stay-green introgression near-isogenic sorghum lines. Analysis of variance revealed highly significant (p ≤ 0.001) and significant (p ≤ 0.05) genotypic differences for most traits under both moisture regimes. Broad-sense heritability ranged from 46.6% for chlorophyll content at flowering (CHlF) to 93.1% for plant height (PHT) under water deficit, and from 67.2% (CHlF) to 94.7% (PHT) under well-watered conditions. Genetic advance as a percentage of the mean varied from 5.49% (CHlF) to 80.68% (photosynthetic rate at the hard-dough stage, AG) under water deficit, and from 4.85% (days to maturity, DTM) to 58.33% (peduncle exertion, PedEx) under well-watered conditions. Cluster analysis grouped the genotypes into four and five clusters under the respective moisture regimes. Principal component analysis (eigenvalues > 1) explained 78.05% and 74.03% of the total phenotypic variation under water-deficit and well-watered conditions, respectively.

Conclusions

Substantial morpho-physiological diversity was observed among the evaluated sorghum lines. Under drought stress, chlorophyll maintenance showed direct positive inherent effect (CHlG, 0.431) on grain yield; the stay-green genotype ETSC16140 gave the maximum grain yield (4037.9 kg ha^–1). Under well-watered set, the maximum grain yield was also obtained from the stay-green genotype ETSC16227 (5887.2 kg ha^–1). Furthermore, thirteen stay-green introgression lines exhibited consistently superior grain yield performance across both water regimes. These genotypes constitute valuable breeding materials for sorghum improvement programs aimed at enhancing drought tolerance in semi-arid environments.

Introduction

Sorghum (Sorghum bicolor (L.) Moench) is one of the crop types for which Ethiopia has been credited as being a Vavilovian center of origin and diversity [22]. According to [14], Sorghum is originated in the northeastern quadrant of Africa with large variability of wild and cultivated species, and [31] stated that sorghum is cultivated under erratic and unpredictable rain fall patterns. Sorghum grows in a wide range of agro-ecologies most importantly in the moisture stressed condition where other crops can least survive [6]. Sorghum best adapted to fertile, well drained soils at a pH between 6.0 and 6.5. It also grows in low fertility, moderately acidic and highly alkaline soils with adaptability from lowland, medium and highland altitude [16].

Based on [50] report, during 2022, sorghum is the fifth important world crop among cereals after maize, wheat, rice and barley with 62.3 million metric tons of global production, and it is also second in Africa, next to maize both in area coverage (23.81 million hectares) and total production (22.33 million metric tons). According to [11] explanation, it is the most important cereal crop in Ethiopia ranking 4th both in area coverage (1.68 million hectares) and volume of total production (45.2 million quintals) after maize, tef and wheat.

Under natural and agricultural conditions plants are often exposed to various environmental stresses of biotic and a biotic factor [7], which are major constraints that seriously reduce productivity of crops. Biotic and a biotic stress is any environmental factor that affects plant physiology, morphology and biochemical processes. Drought is among the most important a biotic factor which brings huge impact on production and productivity. Drought stress (soil water deficit) can be defined as a situation in which plant water potential and turgor are reduced enough to interfere with normal functions [15]; tissue or cell water content is below the highest water content exhibited at the most hydrated state. According to [16], the global agricultural land area is rain fed and adversely affected by drought. Due to this fact, drought reduces crop production and risks the wellbeing of livelihood as well as food security.

In sorghum, the best characterized form of drought tolerance during post-flowering crop growth stage is called non-senescence or “stay-green” trait [37], which is the ability to resist premature plant senescence (retain green leaf area), resist lodging and filling grain normally. Stay-green is an integrated drought adaptation trait that enhances the balance between the supply and demand of water, resulting in the retention of green leaves for longer periods during grain filling. When water is limited during the grain-filling period, sorghum genotypes possessing this trait maintain photosynthetically active leaf area better than genotypes that do not possess this trait [10]. As a result, stay green trait allows plants to have an improved grain-filling process even under stress conditions [37].

Improving the drought tolerance/resistance sorghum is one of the most important objectives of plant breeders [28]. Drought resistance is a primary reason that sorghum is a major crop in arid and semiarid regions. As stated by [44], two distinct drought responses namely, pre- and post-flowering drought responses have been described in sorghum and are probably controlled by different genetic mechanisms. The pre-flowering response occurs when plants are under significant moisture stress from panicle differentiation to flowering. The post-flowering response occurs when plants are under severe drought stress during grain filling, and it is the most damaging stress commonly referred to as “terminal drought stress” [28]. Plant breeding with an improvement of stay green traits helps to increase yield potential under such drought stress conditions [1], which enable to secure food availability and meet the future demand for agricultural production. In this context, moisture stress yield losses especially from climate change indicate an urgent need for improving stay-green trait genotype as a prospective climate resilient strategy. according to [17], in moisture stressed agroecology of Ethiopia, the research strategy is developing early maturing sorghum varieties, and improving late maturing landraces and elite genotypes through stay green trait introgression.

In the country, various drought escaping (early maturing) varieties have been released; however, the stay green trait introgression approach didn’t get much attention and only limited level of stay green trait introgression done through marker assisted backcrossing. As result, in Ethiopia improvement of stay green trait accessions and/or landraces for drought prone areas is still very low. Even though marker assisted backcrossing, quantitative trait loci identification and evaluation of stay green trait generations has been done by different researchers [5, 28, 46]; however, further morpho-physiological performance and variability studies of those introgression genotypes have not yet done [27]. stated that evaluation of newly developed drought tolerant genotypes through physiological and morphological approach is useful technique for understanding the response of crops. Hence, to fill this research gap, irrigation-based field experiment was conducted in line with the goal of sorghum improvement strategy of the country for drought prone areas. We conducted this experiment with the research questions: what was the impact of post-flowering drought on sorghum productivity? and what mitigation actions can be taken to raise sorghum productivity under post-flowering drought conditions? We also hypothesized that stay-green introgression lines would exhibit higher photosynthetic retention and yield stability under post-flowering drought compared to recurrent parents. The study aimed to provide base line information and possibly identify potential lines for further breeding program.

Materials and methods

Description of the study area

The experiment was conducted during 2022 by means of irrigation in the low land areas of North Shewa in Amhara region, namely, Efratana Gidim District at Jeweha Negeso station (Fig. 1). The testing site was located at 105 km North East of Debre Birhan town at 39⁰ 56’47” E longitudes, 10⁰5’49” N latitudes and 1172 m.a.s.l altitude. The site soil textural class is silty clay containing 34% clay, 47% silt and 19% sand proportion with 7.9 soil pH and 0.23 dS/m EC (source: Debre Birhan Agricultural Center). During this field experiment, no rainfall was occurred in the area, and irrigation was the sole source of water for the crop throughout the whole growth period. Soil samples were taken every two weeks interval throughout the experimental months, from 0 to 30 cm and 30–60 cm depths. Soil moisture (Fig. 2c), was determined using the gravimetric method following the standard procedure. The daily weather data including minimum and maximum temperature, relative humidity and precipitation (Fig. 2a and b) were obtained from Robi National Tobacco Enterprise (Ethiopia) planted around the area.

Fig. 1.

Map of the study area

Fig. 2.

2022 monthly rain fall (a), temperature and relative humidity (b), and soil moisture (c) of the study area

Experimental materials

The study materials consisted of 48 genotypes (39 - stay green introgression near isogenic lines, 6 – released, 2- known landraces of recurrent parents, and 1- stay green donor parent (B35)). The donor parent (B35) is internationally known for post-flowering drought tolerance and has been extensively used as source of stay green trait genes in international sorghum breeding programs. The recurrent released and local parents (Table 1), are known medium maturing genotypes, and give high yield under optimum moisture conditions. These genotypes have good agronomic traits, however unable to withstand extreme drought conditions. Stay green introgression lines were developed by a series of crosses and backcrosses to transfer stay green trait genes from B35 donor parent into cultivated varieties (seed parents) [46].

Table 1.

Description of parent materials

S/N	Variety	Pedigree	Year of release	Yield (ton/ha)	Days to flowering (days)
1	Melkam	WSV-387	2009	3.7–5.8	82
2	Teshale	3443-2-0P	2002	2.6–5.2	76
3	Gambella 1107	Gambell 1107	1976	3.0–5.0	90
4	Dekeba	ICSR 24,004	2012	3.7–4.5	75
5	Macia	Macia	2007	4.2–4.4	60
6	Meko-1	M – 36,121	1997	2.2–3.3	92
7	Tseadachimure	Local	-
8	Wediaker	Local	-

Source: National variety release book

Experimental design and treatments

Two sets i.e., well-watered and post flowering water deficit experiments have been executed side by side by using an alpha-lattice design. Planting was done at early February 2022 using a plot size of 2 rows of 3 m length with 15 cm and 75 cm plant and row spacing respectively. The plot, block and replication spacings were 1 m, 1 m and 1.5 m respectively. The seed and fertilizer rates were 13 kg ha^–1 seed, 121 kg ha^–1 NPS, and 90 kg ha^–1 urea. All NPS and Half of urea fertilizer was applied at planting, and the remaining half of urea was applied at knee height stage. Irrigation was applied soon after planting using furrow method, and continued in a week interval at establishment in response to wilting symptom through visual observation. Then, when the crop reached above knee height stage, irrigation was applied at ten days interval and continued with this schedule. Experimental plots irrigated uniformly to ensure consistent water conditions. In the water deficit experimental set, irrigation was terminated at the exact beginning of 50% flowering stage, and continued until harvesting. Depending on the flowering and maturity stage of the genotypes, on average, the stress extended from 30 to 40 days. All agronomic management practices were applied uniformly to ensure good crop stand. The crop was protected from stem borer insect pest by using the recommended chemical called Karate 5% EC (50 g/l Lamda-cyhalothrin) with 250 ml ha^–1 rate.

Data measurements and collection

Plot and plant-based data were collected for quantitative and qualitative traits. Days to emergence (DE), days to flowering (DTF), days to maturity (DTM), panicle weight (PW), hundred grain weight (HGW), stay-green score and grain yield (GY) were measured in a plot base. Plant height (PH), panicle length (PL), peduncle exertion (PE), chlorophyll content (CHL), leaf nitrogen (LN), photosynthesis rate (A), transpiration rate (E), and stomatal conductance (gsw) were measured in a plant base.

Days to emergence is the number of days from sowing to 50% of seedlings rise in a plot. Days to flowering is the number of days from DE to 50% of plants flowered in a plot. Days to maturity is the number of days from DE to 95% of plants in a plot formed black spot or layer on the hilum of the seed. Panicle weight is weight of total panicles per plot. Hundred grain weight is weight of hundred seeds from 12.5% moisture adjusted grains. Grain yield is weight of 12.5% moisture adjusted grain measured and expressed as kg per plot and converted to kg per hectare. Plant height measured from the base of the plant to the tip of the panicle using five randomly selected plants at physiological maturity. Panicle length is measured from the base to the tip of the panicle during physiological maturity using five randomly selected plants. Peduncle exertion measured from the flag leaf to the base of the panicle at physiological maturity taken as 1- slightly exserted (< 2 cm); 2 - exerted (2–10 cm); 3 - well-exserted (> 10 cm). Chlorophyll content measured from five randomly selected plants (three readings/plant) at flowering and hard dough stage by using chlorophyll Meter “SPAD-502 plus”. Stay-green scores taken at maturity by using 0–9 visual ratings: 0 = completely green, 1 = 90–99% of leaves green, 2 = 80–89% of leaves green, 3 = 70–79% of leaves green, 4 = 60–69% of leaves green, 5 = 50–59% of leaves green, 6 = 40–49% of leaves green, 7 = 30–39% of leaves green, 8 = 20–29% of leaves green and 9 < 20% of leaves green. Leaf nitrogen content taken from oven dried and milled leaf samples destructively collected from the top three randomly selected plant leaves at physiological maturity. Photosynthesis rate (µmol m⁻² s⁻¹), Transpiration rate (mol m⁻² s⁻¹) and Leaf stomatal conductance (mol m⁻² s⁻¹) were taken by using the instrument (LI-6800 Portable Photosynthesis System, Software Version 2.0. LI-COR, Inc. (2021), Lincoln, NE, USA).

Data analyses

Before any further statistical analysis normality of distribution (by using Shapiro-Wilk test and histogram) and homogeneity of variance (by using Bartlet test and box plot test) were checked. Analysis of variance (ANOVA) was done by using R 4.2.0 statistical software, and means were compared using Tukey’s Honestly Significant Difference (HSD) test method. Estimates of variance component, heritability, genetic advance, genetic advance as percentage of mean, correlation analysis, path coefficient analysis, cluster analysis and principal component analysis were also done using R 4.2.0 statistical software.

Results and discussion

Analysis of variance

Under water deficit experimental set, mean squares obtained from analysis of variance revealed highly significant difference (P ≤ 0.001) among 48 sorghum genotypes for most of the measured traits. Significant differences (P ≤ 0.05) also obtained for panicle length, panicle weight, yield, chlorophyll content at flowering and transpiration rate at flowering, with non-significance value for days to maturity (Table 2). Under well-watered experimental set, the mean squares due to genotypes showed highly significant difference (P ≤ 0.001) for all measured traits except stomatal conductance at flowering stage that gave non-significant variation (Table 3). This result confirms the occurrence of highly significant variation for most traits among the tested sorghum genotypes under both water regimes [5, 19, 24, 48]. were also evaluated sorghum genotypes under well-watered and water deficit experimental conditions, and reported the existence of significant variation among sorghum genotypes for those different traits. This substantial variation is an indicative of adequate genetic variability for the traits among the tested sorghum genotypes, which suggest that each genotype is genetically diverse from each other [2]. This genetic variability will provide good opportunity for selecting superior and desired genotypes in drought prone areas.

Table 2.

Mean squares of traits, Mean, range and CV values for water deficit experimental set in 48 sorghum genotypes

Traits	Mean squares				Mean	Range	CV (%)
	Treatment (df = 47)	Rep (df = 2)	Block (df = 21)	Error (df = 73)
DTF	61.48***	159.65***	9.9	7.24	73.58	66.99–85.68	3.7
DTM	32.01	0.52	8.23	6.66	108.48	103.33–116	2.4
PHT	6331***	621***	20	32	208.45	126.8–316.2	2.7
PedEx	30.95***	12.01**	1.461	1.92	9.27	0.23–16.5	15
PL	10.91*	3.77	3.19	2.36	21.56	15.9–26.9	7.1
PWt	205,401*	357,651*	51,226	36,688	2252.62	1675.9–2805.1	8.5
HGW	0.28***	0.31***	0.05**	0.02	2.33	1.62–3.45	5.9
Yld	565,779*	851,474**	205,698*	118,669	3246.72	2152–4037.9	10.6
CHlF	20.35*	59.97*	3.91	5.62	56.68	48.97–62.43	4.2
CHlG	95.38***	33.26	11.81	11.46	37.69	24.96–51.72	9
LN	0.22***	0.12**	0.0076	0.014	1.37	0.83–2.01	8.6
AF	94.85***	75.87***	15.21**	6.69	27.37	14.07–40.25	9.4
EF	0.0000044*	0.00002*	0.0000035*	9.00E-07	0.0046	0.0016–0.0075	20.8
gswF	0.011***	0.0428***	0.0087***	0.002	0.171	0.026–0.324	26.3
AG	47.75***	45.19***	4.74	3.22	9.41	1.99–22.71	19
EG	0.0000038***	0.000001***	2.10E-07	2.00E-07	−0.0009	−0.003–0.002	42.6
gswG	0.002***	0.00061***	0.000068*	0.000033	−0.019	−0.069–0.056	30.1

Table 3.

Mean squares of traits, Mean, range and CV values for well-watered experimental set

Traits	Mean squares				Mean	Range	CV (%)
	Treatment (df = 47)	Rep (df = 2)	Block (df = 21)	Error (df = 73)
DTF	50.29***	6.67*	2.05	2.16	74.24	68.09–83.74	2.0
DTM	29.08***	1.51	2.73	2.5	113.81	108.66–120.84	1.4
PHT	6022.4***	33*	9	8.8	212.95	131.57–317.59	1.4
PedEx	40.8***	0.67	1.7	1.73	11.98	1.8–21	11
PL	16.27***	4	1.67	1.83	22.29	17.53–27.53	6.1
PWt	563,957***	136,517	26,222	41,271	2904.7	2082.5–3814.6	7.0
HGW	0.3947***	0.28***	0.013	0.024	2.73	1.9–3.7	5.7
Yld	1,991,580***	401,536*	101,713	119,117	4471.65	2885.7–5887.2	7.7
CHlF	19.88***	8.86	1.74	2.78	59.02	52.84–64.62	2.8
CHlG	96.32***	56.84***	2.83	3.53	44.97	31.61–57.89	4.0
LN	0.22***	0.0002	0.0098	0.01	1.51	0.97–2.24	6.7
AF	98.75***	143.48***	14.19	5.66	25.52	10.97–37.47	9.3
EF	0.0000048*	0.000019***	0.00000099*	4.10E-07	0.0044	0.002–0.0073	17.0
gswF	0.00023	0.00059	0.000056	0.000049	0.029	0.015–0.048	23.6
AG	51.39***	39.15**	11.52*	5.8	15.55	9.27–24.65	16.0
EG	0.0000032***	3.00E-07	6.80E-07	3.80E-07	0.0016	−0.0014–0.004	38.4
gswG	0.00195***	0.000051	0.000121	0.000132	0.039	−0.034–0.081	29.7

Where, *,** and *** significant at P ≤ 0.05, P ≤ 0.01 and P ≤ 0.001, respectively

CV coefficient of variation (%), DTF  Days to flowering, DTM  Days to Maturity, PHT  Plant height (cm), PedEx Peduncle Exertion (cm), PL Panicle Length (cm), PWt Panicle Weight (g plot^–1), HGW Hundred grain weight(g), Yld Yield (kg ha^–1), ChlF Chlorophyll content at flowering (SPAD value), ChlG  Chlorophyll content at grain filling (SPAD value), LN  Leaf nitrogen (%), AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), EF  Transpiration rate at flowering (mol m⁻² s⁻¹), gswF Stomatal conductance rate at flowering (mol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), EG Transpiration rate at hard dough stage (mol m⁻² s⁻¹), gswG Stomatal conductance rate at hard dough stage (mol m⁻² s⁻¹)

Mean performances of genotypes for different traits

The range and mean performance values of 17 traits were presented in the above Tables 2 and 3. Furthermore, each trait minimum and maximum scores together with its respective genotypes presented in Table 4 below. Wide ranges of mean values have been observed for most of the evaluated characters, which implies presence of variability among the tested genotypes. The mean values of sorghum genotypes showed highly significant variation for most of the phenological, morphological and physiological traits. This large yield and morpho-physiological trait variations indicate for the presence of genetic variability among the tested genotypes. Galyuon et al. [18] suggests that stay-green lines have considerable potential to increase the genetic diversity of the stay-green traits within sorghum breeding programs.

Table 4.

Minimum and maximum score summary of traits with associated sorghum genotypes

Traits	Water deficit set				Well- watered set
	Genotypes	Min.	Genotypes	Max	Genotypes	Min.	Genotypes	Max.
DTF	ETSC16214	66.98	ETSC16213	85.68	ETSC16239	68.09	ETSC16230	83.74
DTM	ETSC16145	103.33	ETSC16210	116	Meko-1	108.66	ETSC16210	120.84
PHT	ETSC16211	126.8	ETSC16244	316.2	ETSC16211	131.57	ETSC16244	317.59
PedEx	ETSC16221	0.23	ETSC16257	16.5	ETSC16221	1.8	ETSC16257	21
PL	ETSC16238	15.9	ETSC16221	26.9	ETSC16238	17.53	ETSC16244	27.53
PWt	ETSC16244	1675.9	ETSC16140	2805.1	ETSC16238	2082.5	ETSC16220	3814.6
HGW	ETSC16212	1.62	ETSC16254	3.45	ETSC16212	1.9	ETSC16254	3.7
Yld	ETSC16244	2152	ETSC16140	4037.9	ETSC16231	2885.7	ETSC16227	5887.2
CHlF	ETSC16257	48.97	ETSC16244	62.43	ETSC16210	52.84	ETSC16244	64.62
CHlG	ETSC16230	24.96	B35	51.72	ETSC16227	31.61	B35	57.89
LN	ETSC16215	0.83	ETSC16242	2.01	ETSC16140	0.97	ETSC16217	2.24
AF	ETSC16258	14.07	ETSC16142	40.25	ETSC16248	10.97	ETSC16211	37.47
EF	ETSC16230	0.0016	ETSC16248	0.0075	ETSC16242	0.002	ETSC16211	0.0073
gswF	ETSC16230	0.026	ETSC16256	0.324	Dekeba	0.015	ETSC16220	0.048
AG	Melkam	1.99	ETSC16220	22.71	ETSC16239	9.27	ETSC16248	24.65
EG	ETSC16211	−0.003	Tseadachim	0.002	ETSC16221	−0.001	ETSC16220	0.004
gswG	Melkam	−0.069	ETSC16220	0.056	ETSC16221	−0.034	ETSC16145	0.081

Where, Min. minimum, Max. Maximum, DTF Days to flowering, DTM Days to maturity, PHT Plant height (cm), PedEx Peduncle exertion (cm), PL panicle length (cm), PWt Panicle weight (g plot^–1), HGW Hundred grain weight(g), Yld Yield (kg ha^–1), ChlF Chlorophyll content at flowering (SPAD value), ChlG Chlorophyll content at grain filling (SPAD value), LN Leaf nitrogen (%), AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), EF Transpiration rate at flowering (mol m⁻² s⁻¹), gswF Stomatal conductance rate at flowering (mol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), EG Transpiration rate at hard dough stage (mol m⁻² s⁻¹), gswG Stomatal conductance rate at hard dough stage (mol m⁻² s⁻¹)

Phenological traits of DTF and DTM showed highly significant variations (p ≤ 0.001) among genotypes under well-watered experimental set. Under water deficit set only DTF showed highly significant variation. This result is in agreement with the findings of [52], who reported the presence of highly significant variation for DTF and DTM among various sorghum lines. Morphologically growth-related traits of PH, PedEx and PL showed significant (P ≤ 0.05) and highly significant variations (p ≤ 0.001) under both water regimes. Similar findings also reported by different authors [34, 40, 41]. From this analysis result, peduncle exertion showed large variations, and it is an important trait associated with drought tolerance and should be taken in to consideration for effective selection among sorghum genotypes [3, 45].

The mean performances of grain yield (Yld), panicle wight (PWt) and hundred grain weight (HGW) showed highly significant variations (p ≤ 0.001) under both water regimes, except Yld that showed significant variations (p ≤ 0.05) under water deficit set. These results were in agreement with the finding of [5, 40], who reported variations of those traits under the two water regimes. Panicle weight and grain yield showed substantial variations between well-watered and water deficit experimental sets. The average grain yield and panicle weight obtained from the well-watered set was above the maximum yield and panicle weight gained under the water deficit set. Although sorghum has developed different drought stress tolerance mechanisms, however the yield obtained under dry land conditions is much less than those under irrigated condition [27]. This implies that drought tolerance in sorghum is not absolute [53]. A general reduction in panicle weight and grain yield was obtained under drought stress than those under well-watered condition [5]. From the water deficit experimental set, the maximum grain yield was gained from the stay-green gene introgression genotype ETSC16140 (4037.9 kg ha^–1). (Paul et al., 2001) stated that stay-green sorghum varieties outperform conventional varieties under drought conditions. Under the water deficit set, 66.7% of genotypes gave above mean grain yield. Under the well-watered setup, the maximum yield was also obtained from the stay-green gene introgression genotype ETSC16227 (5887.2 kg ha^–1). Among the tested genotypes, 50% of them gave grain yield above the mean value. It is also important to underline that, in stay-green introgression lines, high yield has been obtained under both drought stressed and well-watered conditions [26].

Among the tested 48 sorghum genotypes, 18 of them exhibited consistence above mean yield performance under both water regimes. ETSC16224, ETSC16217, ETSC16214, ETSC16220, ETSC16226, ETSC16248, ETSC16216, ETSC16145, ETSC16219, Melkam, ETSC16146, B35, ETSC16213, ETSC16227, ETSC16140, Macia, Wodiaker and Dekeba were genotypes which showed consistence above mean yield performance. Out of these eighteen genotypes, thirteen of them were stay green-gene introgression RILs obtained from successive back cross of senescent parents and the stay-green donor B35. Under both water regimes, the top six above mean yield was obtained from the stay-green introgression lines. This yield advantage is obtained due to the fact that stay-green genotypes accumulate more soluble sugar than the senescent type [33]. It also influences the root architecture and increases the water accessibility during grain filling under water-limited conditions [32]. Furthermore, this trait possibly increases adaptation of sorghum to drought due to its regulation in leaf anatomy, root growth, water uptake and utilization [9]. Genotypes possessing the stay-green trait have significant yield advantage under post-anthesis drought stress conditions compared with genotypes not possessing this trait [28]. Moreover, under drought conditions, the introgressed lines outperform the recurrent parents in yield, and they maintain stay-green trait comparable to the donor parent (B35). This result implies that, stay-green trait is one of the best climate resilience traits, and should be considerd in sorghum breeding strategies for drought prone agro-ecologies of the country. Drought stressed dry lowland agroecology is the vast majority sorghum growing area (80%) of the country [17]. For this agroecology, developing stay-green sorghum genotypes is one of the primary product concepts in the country’s breeding pipeline. In this area, sorghum is predominantly cultivated by smallholder farmers. Hence, stay-green trait is considered as feasible prospective approach for the livelihood of those small holder farmers.

Chlorophyll content at flowering (CHlF, SPAD value), chlorophyll content at grain filling (CHlG, SPAD value), leaf nitrogen (LN) (%), photosynthesis rate at flowering (AF, µmol m⁻² s⁻¹), transpiration rate at flowering (EF, mol m⁻² s⁻¹), stomatal conductance at flowering (gswF, mol m⁻² s⁻¹), photosynthesis rate at hard dough stage (AG, µmol m⁻² s⁻¹), transpiration rate at hard dough stage (EG, mol m⁻² s⁻¹), and stomatal conductance at hard dough stage (gswG, mol m⁻² s⁻¹) were physiological parameters evaluated under both water regimes. Most of these traits exhibited highly significant (p ≤ 0.001) variations under both well-watered and water deficit experimental sets. CHlF from water deficit set and EF from both water regimes showed significant (p ≤ 0.05) variations. However, gswF tested under well-watered set displayed non-significant variation. Since photosynthesis and transpiration rates closely corelated with stomatal conductance, under drought stress conditions, stomatal conductance have practical implications interims of adjusting transpiration loss at the expense of reduced carboxylation rate [35]. Presence of such physiological variations will be exploited in advancing efficient genotypes under drought stress. These specific physiological parameters require much attentions for the improvement of drought tolerances in sorghum. Asfaw et al. [5] reported the presence of significant variation in leaf photosynthetic rate, transpiration rate and chlorophyll fluorescence parameters for sorghum genotypes evaluated under both water regimes.

All physiological parameters showed high mean performance reductions as the measurement goes from flowering stage to hard dough stage under both experimental sets, in which the drastic change was highly pronounced under the water deficit set. This result is in agreement with the findings of [5, 39], who reported reduction of photosynthesis rate, leaf chlorophyll content and transpiration rate under drought stress condition. Since water was terminated at flowering stage, both well-watered and water deficit sets didn’t show performance variations for flowering stage physiological traits. However, at hard dough stage high mean performances obtained for all physiological traits under well-watered experimental set. Under both experimental sets, CHlG showed extensive variations in which the maximum chlorophyll content was obtained from genotype B35, which is a so-called donor parent for the stay-green trait. Isaac et al. [23] reported that chlorophyll retention measured by SPAD was highest in B35 with slow rate of chlorophyll loss. Majority of the introgression lines showed high level of chlorophyll content at both water regimes. This result confirms the finding of [28], who reported that the majority of the QTL introgression lines had higher leaf chlorophyll content both before and during leaf senescence. Stay-green scores also visually taken in a plot base at maturity stage. Under drought stressed experimental set, 75% of the genotypes exhibited an intermediate to high (50–89%) green leaves. However, under well-watered experimental set, all genotypes exhibited 60% and above green leaf scores. Under water deficit condition, the stay-green introgression lines were better as compared with the senescent parents. Throughout the experiment, B35 had the highest percent green leaf area (%GLA) under both water regimes. The leaf nitrogen measurements from water deficit and well-watered sets showed slight changes as compared between the two-water regimes in which the leading score was obtained from the well-watered set.

The measurement in photosynthesis rate showed drastic reduction as it goes from flowering to hard dough growth stage under water deficit experimental set. Asfaw et al. [5] reported a decrease in carboxylation rate that was observed from all genotypes evaluated under drought stressed condition. From water deficit set, at hard dough stage, the minimum (1.99 µmol m⁻² s⁻¹) and maximum (22.71 µmol m⁻² s⁻¹) photosynthesis rate was obtained from genotypes Melkam and ETSC16220, respectively. Genotype ETSC16220 is a stay-green gene introgression RIL developed from genotype Melkam and donor B35. This result indicates that variety Melkam showed high response for the functional expression of the received stay-green trait obtained from donor B35. However, this senescent variety should be further re-checked by crossing with donor B35 to confirm for the reliability of functional stay green trait improvement. Functional stay-green is characterized by the maintenance of leaf photosynthesis during grain filling [8]. This genotype also showed a slight yield increase (1.4% yield advantage) over its senescent parent (Melkam). Moreover, from this experiment, an increased level of photosynthesis showed enhanced levels of yields. This outcome is supported by the report of [8], who stated that the contribution of stay-green gene to higher grain yield appears due to the extension of the photosynthetically active phase of the leaf and also possibly to higher photosynthetic rates. Photosynthesis is the fundamental factor of dry matter production and grain yield, so leaf photosynthesis and its related physiological traits are the major targets to be improved in physiological breeding [47].

Transpiration rate and stomatal conductance showed high level of associations in which both traits exhibited similar trends under both experimental sets. This implies that transpiration rate is highly dependent on the opening and closing levels of the stomata. Transpiration per unit leaf area can be limited by stomatal density or aperture, timing of stomatal opening, and hydraulic factors [8]. The specific physiological relationships among photosynthesis, transpiration, and stomatal conductance are of particular interest for breeding [38]. Due to senescence and water deficit stress, both physiological traits showed high rate of reduction as the measurement goes from flowering to hard dough growth stage under both experimental sets, in which the drop level is highly pronounced in the water deficit set. This result is in agreement with [5], who reported that the rate of transpiration was significantly lower under water stressed treatments than with irrigation, and showed a general decreasing pattern with drought stress. Due to the combined effects of drought stress and senescence, in the water deficit set, negative scores of transpiration rate and stomatal conductance were obtained from majority of the tested genotypes measured at hard dough growth stage. As a result, the water deficit experimental set, showed negative mean values for these traits. Negative measurement implies that the stomata shut down and leads for the cease of transpiration.

Estimation of phenotypic and genotypic variance components

The estimated parameters of phenotypic variances (σ²p), genotypic variances (σ²g), phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV) are shown in Tables 5 and 6. The phenotypic variance gave values in the range between 0.007 and 302931.9 for gswF and Yld, respectively under water deficit set, and 0.08 and 743271.6 for LN and Yld, respectively under well-watered set. The genotypic variance also placed in the range between 0.0035 and 165171.8 for gswF and Yld, respectively under water deficit set, and 0.07 and 624154.6 for LN and Yld, respectively under well-watered set. From both experimental sets, results of variance components revealed the presence of higher genotypic and phenotypic variance for majority of the traits. This result indicates for the presence of low environmental influence for those traits. The existence of this genotypic variability can facilitate effective selection process on improvement of these traits through pedigree breeding method.

Table 5.

Estimates of genetic components of traits for post flowering water deficit set

Trait	Range	Mean	δ²g	δ²p	GCV%	PCV%	H² %	GA	GAM %
DTF	65–87	73.58	18.17	25.46	5.79	6.86	71.3	7.42	10.08
PHT	120.9–320	208.45	1938.57	2081.48	20.87	21.74	93.1	91.21	43.75
PedEx	0.2–17	9.27	9.68	11.6	33.56	36.75	83.4	5.85	63.15
PL	14.4–28.6	21.56	3.26	5.81	8.38	11.18	56.1	2.79	12.93
PWt	1376.2–2925.9	2252.6	59226.2	99087.8	10.8	13.97	59.8	387.6	17.21
HGW	1.53–3.51	2.33	0.098	0.12	13.41	15.07	79.1	0.57	24.57
Yld	1721.16–4473.74	3246.72	165171.8	302931.9	12.52	16.95	54.5	618.2	19.04
ChlF	46.34–64.72	56.68	4.91	10.53	3.91	5.72	46.6	3.12	5.49
ChlG	21.06–57.76	37.69	28.08	39.59	14.06	16.69	70.9	9.19	24.39
LN	0.73–2.02	1.37	0.068	0.082	19.1	20.94	83.2	0.49	35.89
AF	10.25–43.98	27.37	39.74	48.29	23.03	25.39	82.3	11.78	43.03
gswF	0.026–0.51	0.17	0.0035	0.007	34.64	48.98	50	0.086	50.46
AG	1.61–24.26	9.41	16.5	20.04	43.16	47.56	82.4	7.59	80.68

Table 6.

Estimates of genetic components of sorghum traits for well-watered set

Trait	Range	Mean	δ²g	δ²p	GCV%	PCV%	H² %	GA	GAM %
DTF	67–86	74.24	16.86	19.33	5.53	5.92	87.3	7.9	10.65
DTM	107–123	113.81	9.35	12.17	2.69	3.07	76.9	5.52	4.85
PHT	128.8–319.5	212.95	1996.89	2108.51	21.09	22.42	94.7	96.76	45.43
PedEx	1.7–21.2	11.98	13.03	14.75	30.13	32.06	88.3	6.99	58.33
PL	16.4–28.4	22.298	4.81	6.64	9.84	11.56	72.4	3.85	17.25
PWt	1825.4–3987.7	2904.7	174228.7	215499.5	14.37	15.98	80.9	773.1	26.62
HGW	1.8–3.9	2.73	0.12	0.15	12.87	14.09	83.5	0.66	24.23
Yld	2645.5–6187.3	4471.65	624154.6	743271.6	17.67	19.28	84	1491.4	33.35
ChlF	50.5–65.5	59.02	5.7	8.48	4.05	4.93	67.2	4.03	6.83
ChlG	29–59.24	44.97	30.93	34.46	12.37	13.05	89.8	10.85	24.14
LN	0.86–2.32	1.51	0.07	0.08	17.27	18.54	86.8	0.5	33.15
AF	7.17–42.59	25.52	38.78	46.37	24.4	26.68	83.6	11.73	45.97
AG	5.35–27.45	15.55	19.23	26.32	28.19	32.99	73	7.72	49.63

Where, σ2g genotypic variance, σ2p phenotypic variance, GCV  Genotypic Coefficient of Variation, PCV Phenotypic Coefficient of Variation, H2 Broad sense heritability, GA Genetic Advance, GAM  Genetic Advance as Percent of Mean, DTF Days to flowering, DTM  Days to maturity, PHT Plant height (cm), PedEx  Peduncle exertion (cm), PL  Panicle Length (cm), PWt  Panicle Weight (g plot^–1), HGW  Hundred grain weight(g), Yld Yield (kg ha^–1), ChlF Chlorophyll content at flowering (SPAD value), ChlG Chlorophyll content at grain filling (SPAD value), LN Leaf nitrogen (%), gswF Stomatal conductance rate at flowering (mol m⁻² s⁻¹) AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage

PCV analysis results found in the range between 5.72% and 48.98% for CHlF and gswF, respectively under water deficit set, and 3.07% and 32.99% for the traits DTM and AG, respectively from the well-watered set. GCV values also showed values in a range between 3.91% and 43.16% for CHlF and AG, respectively under water deficit set, and 2.69% and 30.13% for DTM and PedEx, respectively under well-watered set. Values of the PCV and GCV were considerd low (< 10%), moderate (10–20%) and high (> 20%) as classified by [43]. Under water deficit set, the higher estimates of PCV obtained from gswF (48.98%) followed by AG (47.56%) PedEx (36.75%), AF (25.39%), PHT (21.74%) and LN (20.94%). From the well-watered set, higher PCV were also obtained from AG (32.99%) followed by PedEx (32.06%), AF (26.68%) and PHT (22.42%). In the same way, the higher estimates of GCV were obtained from AG (43.16%) followed by gswF (34.64%), PedEx (33.56%), AF (23.03%) and PHT (20.87%) under water deficit set, and from PedEx (30.13%) followed by AG (28.19%), AF (24.4%) and PHT (21.09%) under well-watered set. These results are in agreement with the findings of [3, 48], who reported presence of high PCV and GCV values for majority of sorghum landraces and stay-green introgression genotypes evaluated under moisture deficit condition. The high GCV implies that, these traits highly contribute for the total variability in the tested genotypes. Hence, there is a good opportunity for making selection using these traits. It also showed for the presence of wide genetic variability, which provide feasible responses to selection.

Heritability and genetic advance

The estimated heritability and genetic advance parameters were provided in Tables 5 and 6. Higher estimates of broad sense heritability obtained for most of the traits. This deduction is in agreement with the findings of [52, 54], who reported presence of high broad sense heritability for most of the examined traits in sorghum genotypes. This high heritability implies that the environmental influence is insignificant on these traits, and the traits can be potentially used for further improvement via selection. Since most of the genotypes developed through introgression breeding, this result indicates that heritable components were highly transferred from parents to offspring during breeding program. The proportion of genetic variability transmitted from parents to offspring is reflected by heritability [54]. The heritability estimates ranged from 46.6% (CHlF) to 93.1% (PHT) under water deficit set, and from 67.2% (CHlF) to 94.7% (PHT) under well-watered set. According to [25], the heritability values were classified as low (< 30%), moderate (30–60%) and high (> 60%). Based on this categorization, medium to high heritability values has been obtained from all considered traits. Under water deficit set, the characters including PHT, PedEx, LN, AG, AF, HGW, DTF, CHlG, and under well watered set al.l considered characters exhibited high heritability. This finding indicates that, simple selection via using such traits will be sufficient for genetic improvement of desirable characters. This result is in agreement with the findings of [3, 34, 49], who reported presence of high heritability for the above considered traits.

However, heritability only indicates for the magnitude of inheritance of quantitative and qualitative traits. Heritability should be supported with estimates of genetic advance for efficient improvement of the characters. Heritability values along with estimates of genetic advance would be more reliable than heritability alone [52]. Hence, in this study, the genetic advance values ranged from 0.086 (gswF) to 618.2 (Yld) under water deficit set and 0.5 (LN) to 1491.4 (Yld) under well-watered set. Genetic advance as percentage of mean (GAM) also ranged from 5.49 (CHlF) to 80.68% (AG) under water deficit set, and 4.85% (DTM) to 58.33% (PedEx) under well-watered set. GAM were classified as low (< 10%), moderate (10–20%) and high (> 20%), which is recommended by [25]. Based on this classification, high GAM was obtained from most of the evaluated traits. Under water deficit experimental set, high GAM were obtained from PHT (43.75%), PedEx (63.15%), HGW (24.57%), CHlG (24.39%), LN (35.89%), AF (43.03%), gswF (50.46%) and AG (80.68%). And under well-watered set, high GAM obtained from PHT (45.43%), PedEx (58.33%), HGW (24.23%), PWt (26.62%), Yld (33.35%), CHlG (24.14%), LN (33.15%), AF (45.97%), and AG (49.63%). Similar findings also reported by [3], who evaluated sorghum genotypes under moisture deficit condition.

Traits with high estimates of heritability along with high genetic advance over mean can be improved using simple selection without difficulty. Higher estimates of heritability coupled with high genetic advance as percentage of mean was obtained for the characters PHT, PedEx, HGW, CHlG, LN, AF and AG under water deficit set, and PHT, PedEx, PL, PWt, Yld, CHlG, LN, AF and AG under well-watered set. These traits predominantly governed by additive gene actions; selecting desirable genotypes using these characters would be more effective and reliable. Similar findings also obtained by [29, 34, 36, 42, 52], who reported the presence of higher estimates of heritability along with high genetic advance over mean from those traits. High value of heritability along with low genetic advance as percentage of mean were also observed for the trait CHlF; implying that variation is obtained mainly due to the non-additive or dominant gene effects. Improving such characters through hybridization or heterosis breeding should be better. Tsegau et al. [49] also reported similar results.

Genotypic and phenotypic correlation of grain yield with other traits under water deficit

The results of genotypic and phenotypic correlation coefficients are provided in Table 7. Grain yield displayed highly significant (p ≤ 0.01) positive correlations with PWt (rg = 0.96, rp = 0.89), and HGW (rp = 0.22), and significant (p ≤ 0.05) positive associations with gswF (rg = 0.31). These results indicate that increasing in one or more of these characters will further increase grain yield. Similar finding also reported by [39] from stay-green introgression sorghum lines evaluated under post-flowering drought stress. Genotypic correlation coefficient of PWt was higher in magnitude than its corresponding phenotypic correlation coefficient, indicates for the presence of high degree inherent associations between this character and grain yield. Moreover, DTF (rp = −0.32) exhibited highly significant (p ≤ 0.01) negative associations with grain yield. DTF (rg = −0.36) and PHT (rg = −0.31, rp = −0.21) also showed significant (p ≤ 0.05) negative associations with grain yield. The strong negative correlations among these traits implies, an existence of high and inverse relationships between the traits and grain yield. Such traits should be used as indirect selection criteria for improving grain yield through selection breeding. The negative associations of yield with DTF and PHT suggests to shift the direction of selection towards early maturing and short statured sorghum genotypes, leads for better improvement of grain yield under drought prone areas. Furthermore, grain yield also displayed nonsignificant but positive associations with EF (rg = 0.17, rp = 0.06) and EG (rg = 0.11, rp = 0.09), which implies that yield slightly increases as transpiration rate increases. Transpiration rate during flowering stage is much higher than those in terminal drought at hard dough stage. This result confirms that plants lower E and save water at a time when drought occurred, but do not preserve water earlier in the vegetative stage. Moreover, stay-green trait improves the root to access water at depth and lower transpiration, which contribute to produce more under drought condition. Stay-green contribute in the regulation of leaf anatomy, root growth, water uptake and utilization [9] and increases the water accessibility during grain filling under water-limited conditions [32].

Table 7.

Genotypic (above diagonal) and phenotypic (below diagonal) correlation coefficients under post flowering water deficit condition

	DTF	PHT	PedEx	PL	PWt	HGW	Yld	ChlF
DTF	1	0.17	−0.18	0.31*	−0.38**	−0.36*	−0.36*	0.10
PHT	0.13	1	0.42**	-0.04	−0.34*	0.29*	−0.31*	0.18
PedEx	−0.13	0.37**	1	-0.39**	−0.08	0.24	−0.04	−0.26
PL	0.14	−0.02	−0.28**	1	0.13	−0.32*	0.11	0.51**
PWt	−0.34**	−0.26**	−0.05	0.11	1	0.19	0.96**	0.05
HGW	−0.29**	0.26**	0.16	-0.15	0.16	1	0.27	0.03
Yld	−0.32**	−0.21*	−0.01	0.03	0.89**	0.22**	1	−0.09
ChlF	−0.02	0.13	−0.20*	0.35**	0.06	0.04	−0.05	1
ChlG	0.003	−0.34**	−0.07	0.08	0.17*	−0.04	0.08	0.31**
LN	0.02	−0.02	−0.07	0.11	−0.01	−0.15	−0.06	0.01
AF	−0.12	−0.39**	−0.16	-0.11	0.08	−0.02	0.08	−0.24**
EF	−0.19*	−0.42**	−0.04	-0.18*	0.10	−0.11	0.06	−0.15
gswF	−0.09	−0.53**	−0.07	-0.17*	0.13	−0.24**	0.07	−0.15
AG	−0.17*	−0.19*	0.03	-0.25**	0.20*	0.09	0.16	−0.09
EG	−0.13	−0.16	−0.05	-0.14	0.12	0.08	0.09	−0.09
gswG	−0.09	−0.24**	−0.14	-0.09	0.18*	0.09	0.14	−0.08

	ChlG	LN	AF	EF	gswF	AG	EG	gswG
DTF	0.03	0.02	−0.14	−0.30*	−0.21	−0.24	−0.19	-0.14
PHT	−0.40**	−0.01	−0.42**	−0.55**	−0.73**	−0.21	−0.17	-0.24
PedEx	−0.10	−0.07	−0.19	−0.03	−0.13	0.04	−0.01	-0.15
PL	0.17	0.11	−0.07	−0.09	−0.03	−0.23	−0.15	-0.11
PWt	0.18	0.02	0.09	0.17	0.33*	0.23	0.16	0.26
HGW	−0.13	−0.19	−0.02	−0.12	−0.25	0.09	0.09	0.09
Yld	0.07	−0.06	0.12	0.17	0.31*	0.16	0.11	0.21
ChlF	0.38**	−0.03	−0.31*	−0.27	−0.28	−0.15	−0.20	-0.12
ChlG	1	0.15	0.21	0.46**	0.64**	0.39**	0.49**	0.42**
LN	0.11	1	0.05	0.16	0.08	0.19	0.29*	0.25
AF	0.21*	0.02	1	0.74**	0.76**	0.35*	0.15	0.25
EF	0.34**	0.10	0.64**	1	0.90**	0.41**	0.71**	0.34**
gswF	0.34**	0.04	0.54**	0.75**	1	0.54**	0.48**	0.47**
AG	0.37**	0.17*	0.34**	0.35**	0.39**	1	0.93**	0.84**
EG	0.45**	0.27**	0.14	0.49**	0.33**	0.81**	1	1.00**
gswG	0.38**	0.23**	0.24**	0.27**	0.34**	0.78**	1.00**	1

Where; * and ** significant at P ≤ 0.05 and P ≤ 0.01, respectively

DTF  Days to flowering, PHT Plant height (cm), PedEx Peduncle exertion (cm), PL panicle length (cm), PWt Panicle weight (g plot^− 1), HGW Hundred grain weight(g), Yld Yield (kg ha^− 1), ChlF Chlorophyll content at flowering (SPAD value), ChlG Chlorophyll at grain filling (SPAD value), LN Leaf nitrogen (%), AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), EF Transpiration rate at flowering (mol m⁻² s⁻¹), gswF Stomatal conductance rate at flowering (mol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), EG Transpiration rate at hard dough stage (mol m⁻² s⁻¹), gswG Stomatal conductance rate at hard dough stage (mol m⁻² s⁻¹)

Genotypic and phenotypic correlation among other traits under water deficit

Significant correlations were obtained between pairs of yield related traits. Days to flowering exhibited highly significant (p ≤ 0.01) negative associations viz., PWt (rg = −0.38, rp = −0.34), HGW (rg = −0.36, rp = −0.29). The negative associations with PWt and HGW implies, under drought condition late maturing genotypes exhibited high reduction for panicle and hundred seed weight. This result is in agreement with [12]. Plant height displayed highly significant (p ≤ 0.01) positive and negative associations with PedEx (rg = 0.42), PWt (rp = −0.26), HGW (rp = 0.26), CHlG (rg = −0.4, rp = −0.34), AF (rg = −0.42, rp = −0.39), EF (rg = −0.55, rp = −0.42), gswF (rg = −0.73, rp = −0.53), gswG (rp = −0.24). Mofokeng et al. [34] reported presence of significant association between peduncle exertion and plant height. This result suggests that long plants have inverse associations with physiological traits. Highly significant (p ≤ 0.01) positive associations were observed between chlorophyll content at hard dough stage and others physiological traits, viz., CHlF (rg = 0.38, rp = 0.31), EF (rg = 0.46, rp = 0.34), gswF (rg = 0.64, rp = 0.34), AG (rg = 0.39, rp = 0.37), EG (rg = 0.49, rp = 0.45), gswG (rg = 0.42, rp = 0.38). In general, genotypes with high chlorophyll level displayed enhanced level of photosynthesis rate, transpiration rate and stomatal conductance. Photosynthesis rate at hard dough stage also showed highly significant (p ≤ 0.01) strong positive correlations with EG (rg = 0.93, rp = 0.81), gswG (rg = 0.84, rp = 0.78). This result suggests that photosynthesis rate increased as transpiration rate and stomatal conductance increased. At hard dough stage, leaf stomatal conductance and transpiration rate showed perfect correlations (rg = 1, rp = 1). Furthermore, many traits exhibited significant positive and negative correlations among each other as provided in Table 7.

Genotypic path coefficient analysis of grain yield with other traits under water deficit

Estimates of genotypic direct and indirect effects of the selected traits on grain yield are provided in Table 8. Since correlation coefficient specifies only association between any two traits, partitioning of those correlation coefficient results into direct and indirect effects will provide reliable estimates of trait contributions towards yield improvement. According to [30], path coefficients classified as: negligible (0.00–0.09.00.09), low (0.10–0.19), moderate (0.20–0.29), high (0.30–0.99) and very high (> 1.0). The genotypic path coefficient analysis revealed direct and positive effects on grain yield; obtained from the characters viz., PedEx (0.03), PL (0.171), PWt (1.046), CHlG (0.431), AF (0.371), AG (0.027). In breeding program, for sorghum yield improvement, these traits should be considerd as direct selection criteria. The results were in agreement with the findings of [2, 13, 49]. However, negative direct effects on grain yield were obtained from DTF (−0.068), PHT (−0.115), HGW (−0.024), CHlF (−0.467), LN (−0.139), gswF (−0.818). Stomatal conductance at flowering showed positive significance correlation with grain yield, however it exhibited negative direct effect. This negative effect might be due to the low direct contribution of this variable towards grain yield improvement. Unlike correlation, path analysis identifies the indirect causes behind the positive association of the trait with grain yield. Therefore, the negative direct effect of this trait might be compensated via its positive indirect effect through traits viz., DTF (0.014), PHT (0.084), PWt (0.342), HGW (0.006), CHlF (0.13), CHlG (0.274), AF (0.281), AG (0.015). Traits exhibited positive inter-association with other traits and positive indirect effects on grain yield. Positive genotypic indirect effects were obtained from the characters viz., DTF through PL (0.049), HGW (0.009), CHlG (0.017), gswF (0.163); PHT via PedEx (0.012), LN (0.001), gswF (0.594); PWt via DTF (0.028), PHT (0.039), PL (0.023), CHlG (0.079), AF (0.034), AG (0.006); gswF through DTF (0.014), PHT (0.084), PWt (0.342), HGW (0.006), CHlF (0.13), CHlG (0.274), AF (0.281), AG (0.015). Similar results were reported by [42, 49]. The estimated residual effect was 0.05, and such low effect indicates that the considered attributes significantly contribute for grain yield.

Table 8.

Direct (bold diagonal) and indirect (off diagonal) genotypic effects of traits on yield of sorghum genotypes from post flowering water deficit experimental condition

	DTF	PHT	PedEx	PL	PWt	HGW	ChlF	ChlG	LN	AF	AG	gswF	rg
DTF	−0.068	−0.020	−0.005	0.049	−0.423	0.009	−0.055	0.017	−0.006	−0.039	−0.006	0.163	−0.36*
PHT	−0.012	−0.115	0.012	−0.006	−0.357	−0.007	−0.084	−0.174	0.001	−0.157	−0.006	0.594	−0.31*
PedEx	0.011	−0.049	0.030	−0.067	−0.083	−0.006	0.121	−0.043	0.010	−0.071	0.001	0.104	−0.04
PL	−0.019	0.004	−0.012	0.171	0.140	0.008	−0.237	0.075	−0.016	−0.024	−0.006	0.022	0.11
PWt	0.028	0.039	−0.002	0.023	1.046	−0.004	−0.023	0.079	−0.003	0.034	0.006	−0.268	0.96**
HGW	0.025	−0.033	0.007	−0.055	0.197	−0.024	−0.013	−0.054	0.027	−0.008	0.002	0.201	0.27
ChlF	−0.008	−0.021	−0.008	0.087	0.051	−0.001	−0.467	0.163	0.005	−0.113	−0.004	0.228	−0.09
ChlG	−0.003	0.047	−0.003	0.030	0.193	0.003	−0.177	0.431	−0.020	0.078	0.010	−0.521	0.07
LN	−0.003	0.001	−0.002	0.019	0.023	0.005	0.016	0.063	−0.139	0.017	0.005	−0.067	−0.06
AF	0.007	0.049	−0.006	−0.011	0.095	0.001	0.143	0.090	−0.007	0.371	0.009	−0.620	0.12
AG	0.015	0.025	0.001	−0.040	0.241	−0.002	0.070	0.166	−0.026	0.128	0.027	−0.441	0.16
gswF	0.014	0.084	−0.004	−0.005	0.342	0.006	0.130	0.274	−0.011	0.281	0.015	−0.818	0.31*

Residual = 0.05

Where; DTF Days to flowering, PHT Plant height (cm), PedEx Peduncle exertion (cm), PL panicle length (cm), PWt Panicle weight (g plot^− 1), HGW Hundred grain weight(g), ChlF Chlorophyll content at flowering (SPAD value), ChlG Chlorophyll content at grain filling (SPAD value), LN  Leaf nitrogen (%), AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), gswF Stomatal conductance rate at flowring stage (mol m⁻² s⁻¹), rg genotypic correlation

Phenotypic path coefficient analysis of grain yield with other traits under water deficit

Estimates of phenotypic direct and indirect effects of the selected traits are provided in Table 9. The phenotypic path coefficient analysis revealed direct and positive effects on grain yield; obtained from characters viz., DTF (0.002), PedEx (0.017), PWt (0.895), HGW (0.0064), AF (0.042). The results were in accordance with findings of [13]. From this experiment, at both genotypic and phenotypic level, panicle weight exhibited highly significant positive correlation and high positive direct effect on grain yield; signifying for the desirability of the trait in yield improvement program. In addition, negative direct effects were obtained from PHT (−0.033), PL (−0.029), CHlF (−0.077), CHlG (−0.034), LN (−0.031), AG (−0.02), gswF (−0.065). Traits showed positive phenotypic inter-association with other traits and positive indirect effects on grain yield. Positive phenotypic indirect effects were obtained from characters viz., DTF via CHlF (0.004), AG (0.003), gswF (0.006); PHT via PedEx (0.006), PL (0.001), HGW (0.017), CHlG (0.012), LN (0.001), AG (0.004), gswF (0.034); PWt through PHT (0.008), HGW (0.01), AF (0.003); HGW via PedEx (0.003), PL (0.004), PWt (0.142), CHlG (0.001), LN (0.005), gswF (0.015). Similar result was also obtained by [42]. The residual effect was 0.19; this large effect indicates that there were other traits that could have contributed to yield and were not yet considered in this experiment.

Table 9.

Direct (bold diagonal) and indirect (off diagonal) phenotypic level effects of traits on yield of sorghum genotypes for post flowering water deficit experimental condition

	DTF	PHT	PedEx	PL	PWt	HGW	ChlF	ChlG	LN	AF	AG	gswF	rp
DTF	0.002	−0.004	−0.002	−0.004	−0.277	−0.019	0.004	0.000	−0.001	−0.004	0.003	0.006	−0.32*
PHT	0.000	−0.033	0.006	0.001	−0.228	0.017	−0.010	0.012	0.001	−0.016	0.004	0.034	−0.21*
PedEx	0.000	−0.012	0.017	0.008	−0.048	0.010	0.015	0.002	0.002	−0.007	0.000	0.005	−0.01
PL	0.000	0.001	−0.005	−0.029	0.096	−0.010	−0.027	−0.003	−0.003	−0.004	0.005	0.011	0.03
PWt	−0.001	0.008	−0.001	−0.003	0.895	0.010	−0.004	−0.006	0.000	0.003	−0.004	−0.009	0.89**
HGW	−0.001	−0.009	0.003	0.004	0.142	0.064	−0.003	0.001	0.005	−0.001	−0.002	0.015	0.22**
ChlF	0.000	−0.004	−0.003	−0.010	0.052	0.003	−0.077	−0.010	0.000	−0.010	0.002	0.010	−0.05
ChlG	0.000	0.011	−0.001	−0.002	0.152	−0.002	−0.023	−0.034	−0.003	0.009	−0.007	−0.022	0.08
LN	0.000	0.001	−0.001	−0.003	−0.006	−0.010	−0.001	−0.004	−0.031	0.001	−0.003	−0.003	−0.06
AF	0.000	0.013	−0.003	0.003	0.060	−0.001	0.019	−0.007	−0.001	0.042	−0.007	−0.035	0.08
AG	0.000	0.006	0.000	0.007	0.180	0.006	0.007	−0.013	−0.005	0.014	−0.020	−0.026	0.16
gswF	0.000	0.017	−0.001	0.005	0.118	−0.015	0.012	−0.012	−0.001	0.023	−0.008	−0.065	0.07

Residual = 0.19

Where; Where; Where; DTF Days to flowering, PHT Plant height (cm), PedEx Peduncle exertion (cm), PL panicle length (cm), PWt Panicle weight (g plot^− 1), HGW Hundred grain weight(g), ChlF Chlorophyll content at flowering (SPAD value), ChlG Chlorophyll content at grain filling (SPAD value), LN Leaf nitrogen (%), AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), gswF Stomatal conductance rate at flowring stage (mol m⁻² s⁻¹), rp phenotypic correlation

Cluster analysis

Based on the similarity among morphological and physiological traits, forty-eight (48) genotypes grouped into four genetic clusters from the post-flowering water deficit experimental set (Fig. 3), and into five clusters from the well-watered experimental set, done using hierarchical clustering method (Fig. 4). All clusters except cluster five under well-watered set, had more than two sub-clusters. Under the water deficit set, majority of the assessed genotypes were allocated in cluster II (33.3%) followed by cluster III (29.2%), cluster I (22.9%) and cluster IV (14.6%). Under well-watered set, most of the genotypes were also allocated in cluster II (37.5%) followed by cluster III (20.8%), cluster IV (18.7%), cluster I (16.7%) and cluster V (6.3%). Clusters with least number of genotypes were observed from cluster IV (14.6%) and cluster V (6.3%) under water deficit and well-watered sets, respectively. From small to medium number of assessed genotypes, large cluster numbers were reported by different authors. Chala et al. [12] reported grouping of 36 sorghum genotypes into thirteen distinct clusters [13]., reported three clusters from 28 sweet sorghum genotypes.

Fig. 3.

Cluster dendrogram for post flowering water deficit experimental set

Fig. 4.

Cluster dendrogram for well-watered experimental set

Genetic distance of sorghum genotypes

The inter and intra-cluster distances of sorghum genotypes are presented in Tables 10 and 11. Since most of the assessed genotypes shared common stay-green trait from doner (B35), the inter-cluster distance of the genotypes were moderately divergent. The inter and intra-cluster distance ranged from 30.64 to 39.5 and 4.2 to 5.63, respectively under water deficit experimental set, and from 28.34 to 37.82 and 4.05 to 5.17, respectively under well-watered experimental set. The maximum inter-cluster distance was observed between clusters II and IV (39.25) followed by clusters II and III (38.15) under water deficit experimental set. It is also obtained between clusters III and IV (37.82), followed by between clusters II and V (36.96) under well-watered experimental set. crossing of genotypes from distant clusters could be vital to assemble desirable traits with high heterotic potential [21]. Hence, materials from the maximum inter-cluster distance could help to identify good parental lines for effective use in breeding programs. Since the study area located at drought prone lowland agro-ecology, parent materials from clusters of moisture deficit experimental set should be highly preferred. Based on this, long stature and low yielder parents from cluster II and short stature and high yielder parents from cluster III could be used as crossing materials so as to obtain long stature and high yielder food and feed dual purpose variety. The diversity that observed among genotypes could be useful in the improvement of sorghum for various traits [34]. The significant genetic variability offers an excellent opportunity for genetic improvement of sorghum through hybridizing genotypes from different clusters [21].

Table 10.

Intra (bold diagonal) and inter (off-diagonal) cluster distances of sorghum genotypes evaluated under water deficit set

Clusters	I	II	III	IV
I	4.20	30.64*	30.84*	32.08**
II		4.77	38.15**	39.25**
III			4.83	33.64**
IV				5.63

Table 11.

Intra (bold diagonal) and inter (off-diagonal) cluster distances of sorghum genotypes evaluated under well-watered set

Clusters	I	II	III	IV	V
I	5.05	30.57*	28.34*	29.01*	35.21**
II		5.17	31.49*	33.47**	36.96**
III			4.05	37.82**	32.05**
IV				4.64	34.99**
V					5.16

Cluster mean analysis

Cluster mean values of morphological and physiological traits were provided in Tables 12 and 13. The study showed considerable mean variation among the clusters for the measured traits. The mean values of different traits in different clusters ranged as DTF (70.15–79.24), PHT (176.83–250.93), PedEx (5.8–10.17), PL (20.67–23.56), PWt (2059.98–2468.99), HGW (2.13–2.52), YLD (2898.75–3590.12), CHlF (54.49–58.28), CHlG (34.61–45.22), LN (1.35–1.45), AF (22.49–33.05), EF (0.0036–0.0061), gswF (0.119–0.233), AG (6.89–11.18), EG (−0.0015 – −0.0004) and gswG (−0.036 – −0.012) under water deficit experimental set, and DTF (71.58–81.99), DTM (111.88–118.24), PHT (186.26–238.27), PedEx (6.53–13.74), PL (21.25–23.94), PWt (2586.78–3298.3), HGW (2.54–2.92), YLD (3889.04–5316.89), CHlF (58.41–61.32), CHlG (39.96–52.64), LN (1.38–1.87), AF (20.79–31.73), EF (0.003–0.0054), AG (11.1–21.2), EG (−0.0007–0.0027) and gswG (−0.017–0.055) under well-watered experimental set.

Table 12.

Cluster mean analysis for post flowering water deficit experimental set

Traits	Cluster I	Cluster II	Cluster III	Cluster IV
DTF	70.15	75.69	71.02	79.24
PHT	206.13	250.93	177.53	176.83
PedEx	9.45	9.86	10.17	5.80
PL	21.56	21.46	20.67	23.56
PWt	2468.99	2059.98	2301.87	2297.26
HGW	2.52	2.33	2.30	2.13
Yld	3590.12	2898.75	3398.85	3198.17
ChlF	57.52	57.33	54.49	58.29
ChlG	36.72	34.61	38.22	45.22
LN	1.36	1.35	1.36	1.45
AF	22.49	24.45	33.05	30.36
EF	0.0041	0.0036	0.0061	0.0051
gswF	0.146	0.119	0.233	0.204
AG	10.08	6.89	10.90	11.18
EG	−0.0007	−0.0015	−0.0006	−0.0004
gswG	−0.014	−0.036	−0.012	−0.002
No. of Genotypes	11	16	14	7

Table 13.

Cluster mean analysis for well-watered experimental set

Traits	Cluster I	Cluster II	Cluster III	Cluster IV	Cluster V
DTF	73.08	74.85	74.13	71.58	81.99
DTM	113.07	114.89	112.85	111.88	118.24
PHT	238.27	203.71	233.12	186.26	213.63
PedEx	13.28	13.74	11.65	9.47	6.53
PL	23.94	21.60	21.25	23.13	23.09
PWt	3298.3	2586.78	2797.85	3254.5	2735.53
HGW	2.92	2.65	2.86	2.63	2.54
Yld	5233.30	3889.04	4272.34	5316.89	4064.85
ChlF	58.41	59.24	57.43	60.11	61.32
ChlG	43.00	46.69	39.96	46.25	52.64
LN	1.72	1.40	1.57	1.38	1.87
AF	31.73	25.72	20.79	24.08	27.86
EF	0.0054	0.0047	0.003	0.0046	0.0043
AG	14.27	15.95	12.11	21.20	11.10
EG	0.0015	0.0019	0.001	0.0027	−0.0007
gswG	0.037	0.048	0.026	0.055	−0.017
No. of Genotypes	8	18	10	9	3

Where; DTF Days to flowering, DTM Days to maturity, PHT Plant height (cm), PedEx  Peduncle exertion (cm), PL panicle length (cm), PWt Panicle weight (g plot^–1), HGW Hundred grain weight(g), Yld Yield (kg ha^–1), ChlF Chlorophyll content at flowering (SPAD value), ChlG Chlorophyll content at grain filling (SPAD value), LN Leaf nitrogen (%), AF Photosynthesis rate at flowering (µmol m⁻² s⁻¹), EF Transpiration rate at flowering (mol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), EG Transpiration rate at hard dough stage (mol m⁻² s⁻¹), gswF Stomatal conductance rate at flowering (mol m⁻² s⁻¹), gswG Stomatal conductance rate at hard dough stage (mol m⁻² s⁻¹)

As mentioned above, genotypes evaluated under post-flowering water deficit experiment set were grouped in to four clusters. Cluster I comprised 11 genotypes characterized by relatively medium stature, exerted peduncle, medium panicle length, very high panicle weight, large seed size, very high yield, moderate leaf nitrogen, moderate photosynthesis and transpiration rate. Cluster II contained 16 genotypes characterized by long stature, low panicle weight, medium seed size, low yielder, low chlorophyll content, low nitrogen content, low photosynthesis rate, low transpiration and stomatal conductance. Cluster III included 14 genotypes which are characterized by short stature, well exerted peduncle, short panicle length, high grain yield and panicle weight, medium grain size, high photosynthesis and transpiration rate and high stomatal conductance. Cluster IV gathered 7 genotypes which characterized as relatively short stature, exerted peduncle, long panicle length, medium grain yield and panicle weight, small grain size, very high nitrogen content, very high chlorophyll content, high photosynthesis and transpiration rate and high stomatal conductance. From this moisture deficit set, cluster IV contains drought tolerant genotypes, which is indicated by the presence of enhanced physiological and drought related traits. The cluster mean interpretation of well-watered experimental set also described as follows. Cluster I contained 8 genotypes which are characterized by relatively medium maturing, long stature, well exerted peduncle, long panicle length, high grain yield, high panicle weight, large grain size, low chlorophyll content, medium nitrogen content, medium photosynthesis rate and medium transpiration rate. Cluster II embraced 18 genotypes which are characterized by relatively medium maturing, medium plant height, well exerted peduncle, medium panicle length, low grain yield and panicle weight, medium grain size, medium (chlorophyll content, nitrogen content, transpiration rate, photosynthesis rate and stomatal conductance). Cluster III included 10 genotypes which are characterized by relatively early maturing, long stature, well exerted peduncle, medium panicle length, medium grain yield and panicle weight, large grain size, low (chlorophyll content, nitrogen content, transpiration rate, photosynthesis rate and stomatal conductance). Cluster IV holds 9 genotypes that characterized by relatively early maturing, short stature, exerted peduncle, long panicle length, very high grain yield and panicle weight, medium grain size, high chlorophyll content, high photosynthesis and transpiration rate, high stomatal conductance. Cluster V comprised 3 genotypes characterized by relatively late maturing, medium stature, exerted peduncle, long panicle, low grain yield and panicle weight, very high chlorophyll content, high leaf nitrogen, low (photosynthesis rate, transpiration rate and leaf conductance). From the well-watered experimental set, cluster IV holds early maturing genotypes with better physiological and drought tolerance related traits, which possibly scape drought while maintaining its yield. Thus, under both water deficit and well-watered conditions, from cluster IV, drought tolerant and early maturing genotypes should be selected for effective use in breeding programs of the country.

Principal component analysis

The linear transformation of the original variables was done by considering the evaluated morpho-physiological traits simultaneously. According to [20], principal components having eigenvalues greater than unity and component loadings greater than ± 0.3 are valuable. Under both water regimes, the first six principal components gave eigenvalues greater than one. Those principal components with eigenvalues greater than unity explained 78.05% and 74.03% of the total variations under water deficit and well-watered experimental sets, respectively (Tables 14 and 15). Andiku, Wedajo Gebre et al. [4, 51] reported the presence of 75% and 76.5% variability, respectively from six principal components (PCs) with eigenvalues greater than one. The first principal component (PC1) explained a variation of 27.09% followed by PC2 (14.06%), PC3 (12.83%), PC4 (11.83%), PC5 (6.55%) and PC6 (6.07%) under water deficit experimental set. under the well-watered experimental set, PC1 accounts 19.8% of the total variation followed by PC2 (15.82%), PC3 (13.96%), PC4 (9.23%), PC5 (8.58%) and PC6 (6.64%). The most contributing traits to the total variation in PC1 were gswF, EF, AG, gswG and EG under water deficit set, and EG, gswG, AG, DTF, PWt, Yld and DTM under well-watered set. Hamidou et al. [21] also reported similar traits that largely contribute to PC1. The traits with a significant contribution in PC2 were PL, DTF, PedEx, HGW, CHlF and CHlG under water deficit set, and CHlG, PL, CHlF, PedEx and AF under well-watered set. In PC3 most of the variation was contributed by PWt, Yld, and CHlF under water deficit set, and Yld, PWt, PHT and EF under well-watered set. The highest contributing traits to the variation accounted in PC4 were EG, GswG, AG, and PHT under water deficit set, and AF, EF, HGW, and LN under well-watered set. High contribution to the variation in PC5 was obtained by LN, CHlG and HGW under water deficit set, and PHT, HGW, CHlF, PedEx and DTM under well-watered set. Variations in PC6 mainly contributed by CHlF, DTF and LN under water deficit set, and HGW and DTF under well-watered experimental set. The two principal components (PC 1) and PC2) accounted for most of the variability under both water regimes. under water deficit experimental set, the variation in PC1 and PC2 were remarkably explained by physiological and drought related traits of gswF, EF, AG, gswG, EG, CHlF, CHlG, PL, DTF and PedEx, which had the greatest loadings and contributed as major source of variation. These physiological and drought related traits will be the most discriminative traits in developing drought tolerant sorghum genotypes under drought stress environments. Therefore, during improvement for drought tolerant stay-green genotypes, parental line selection should be caried out by using the main contributing traits from PC1 and PC2.

Table 14.

Principal component analysis of measured traits of sorghum genotypes evaluated under post flowering water deficit set

Traits	PC1	PC2	PC3	PC4	PC5	PC6
DTF	−0.16	−0.37	−0.22	−0.10	−0.04	0.43
PHT	−0.29	0.16	−0.01	−0.33	−0.02	−0.18
PedEx	−0.07	0.37	−0.09	−0.15	−0.21	−0.17
PL	−0.05	−0.40	0.22	0.13	0.21	−0.26
PWt	0.19	0.05	0.56	0.10	0.13	0.16
HGW	0.00	0.36	0.26	−0.17	−0.34	−0.11
Yld	0.18	0.12	0.54	0.15	0.13	0.23
ChlF	−0.11	−0.33	0.31	−0.11	−0.26	−0.47
ChlG	0.25	−0.32	0.08	−0.13	−0.39	−0.23
LN	0.10	−0.15	−0.11	−0.15	0.57	−0.35
AF	0.29	−0.01	−0.19	0.25	−0.24	0.08
EF	0.37	0.00	−0.18	0.24	−0.14	−0.23
GswF	0.39	−0.08	−0.14	0.24	−0.10	−0.05
AG	0.34	0.04	−0.02	−0.37	−0.03	0.09
EG	0.32	0.00	−0.04	−0.45	0.13	0.02
GswG	0.34	−0.05	0.01	−0.41	0.11	0.14
Eigenvalue	4.61	2.39	2.12	2.01	1.11	1.03
Variance (%)	27.09	14.06	12.83	11.83	6.55	6.07
Cumulative variance (%)	27.09	41.15	53.61	65.43	71.98	78.05

Table 15.

Principal component analysis of measured traits of sorghum genotypes evaluated under well-watered experimental set

Traits	PC1	PC2	PC3	PC4	PC5	PC6
DTF	−0.31	−0.14	0.23	−0.16	−0.02	0.38
DTM	−0.30	−0.24	0.22	−0.10	0.31	0.07
PHT	−0.14	0.19	−0.31	0.16	0.48	0.19
PedEx	0.06	0.34	0.04	0.22	0.36	0.27
PL	0.04	−0.39	−0.22	−0.07	0.05	0.25
PWt	0.30	−0.23	−0.41	−0.05	−0.08	0.14
HGW	−0.01	0.01	−0.25	0.32	0.40	−0.41
Yld	0.30	−0.21	−0.42	−0.03	−0.11	0.16
ChlF	−0.02	−0.37	−0.10	−0.20	0.39	−0.18
ChlG	−0.05	−0.41	0.20	−0.20	0.25	−0.24
LN	−0.22	−0.13	−0.21	0.31	−0.14	−0.19
AF	0.08	−0.30	0.21	0.58	−0.01	0.05
EF	0.20	−0.29	0.31	0.46	−0.12	0.05
AG	0.37	0.01	0.18	−0.23	0.04	−0.18
EG	0.43	0.07	0.16	−0.10	0.19	−0.05
GswG	0.38	0.12	0.22	0.01	0.18	−0.09
Eigenvalue	3.37	2.69	2.37	1.57	1.46	1.13
Variance (%)	19.8	15.82	13.91	9.23	8.58	6.64
Cumulative variance (%)	19.8	35.63	49.58	58.81	67.39	74.03

Where; PC Principal Component, DTF Days to flowering, DTM  Days to maturity, PHT Plant height (cm), PedEx  Peduncle exertion (cm), PL panicle length (cm), PWt Panicle weight (g plot^− 1), HGW  Hundred grain Weight(g), Yld Yield (kg ha^− 1), ChlF Chlorophyll Content at Flowering (SPAD value), ChlG Chlorophyll content at grain filling (SPAD value), LN Leaf Nitrogen (%), AF Photosynthesis rate at Flowering (µmol m⁻² s⁻¹), EF Transpiration rate at flowering (mol m⁻² s⁻¹), gswF Stomatal conductance rate at flowering (mol m⁻² s⁻¹), AG Photosynthesis rate at hard dough stage (µmol m⁻² s⁻¹), EG Transpiration rate at hard dough stage (mol m⁻² s⁻¹), gswG Stomatal conductance rate at hard dough stage (mol m⁻² s⁻¹)

Conclusions and recommendation

The analysis of variance displayed presence of significant differences for most of the evaluated traits which revealed wide range of genetic variability among the tested genotypes. From this experiment, in line with the objective of the study, outstanding stay-green expression results has been obtained from stay-green gene introgression lines developed by crossing between parents of senescent Meko-1, Melkam and donor B35.

Higher estimates of heritability coupled with high genetic advance as percentage of mean was obtained from the characters PHT, PedEx, HGW, CHlG, LN, AF and AG from the water deficit set, and PHT, PedEx, PL, PWt, YLD, CHlG, LN, AF and AG from the well-watered set. These are the most promising traits, predominantly governed by additive gene actions. Selecting desirable genotypes using these characters would be more effective and reliable. From phenotypic and genotypic path coefficient analysis result, PWt showed direct and high positive effect on grain yield. Thus, using such trait as direct selection criteria would be vital for improvement of grain yield through selection breeding.

In cluster analysis, forty-eight genotypes grouped in to four and five clusters from the water deficit and well-watered sets, respectively. Under both water regimes, cluster IV holds drought tolerant and early maturing stay green genotypes. These materials should be effectively utilized in breeding programs of the country. Under water deficit set, all the physiological traits loading in PC1 and PC2 were major source of variations. Improvement of drought tolerance stay-green genotypes, and parental line selection should be caried out by using those main contributing traits found in PC1 and PC2.

Among the tested genotypes, thirteen stay-green introgression genotypes showed consistence above mean grain yield performance under both water regimes. The maximum yield were also obtained from the stay-green introgression lines of ETSC16140 (4037.9 kg ha^–1) and ETSC16227 (5887.2 kg ha^–1) under water deficit, and well-watered sets, respectively.

Over 80% of sorghum in Ethiopia is produced under severe to moderate drought stress conditions. In this huge drought prone areas, developing drought-resilient stay green sorghum genotypes is one of the national breeding strategies. To address the drought stress problem of the country, these potentially high yielder stay-green genotypes can be further utilized in sorghum breeding program of the country. Based on its high yield and photosynthesis rate under drought, genotype ETSC16220 should be used as a donor parent to transfer the stay-green trait into the genetic background of high- yielding but drought susceptible varieties like Melkam, Meko-1. However, these promising genotypes should be evaluated across multiple locations and seasons to confirm their stability and adaptability before being promoted for use in breeding programs.

Authors’ contributions

S.C. initiated, designed and conducted the experiment. And also collected, organized and analyzed the data. Did the result interpretation and finally wrote the manuscript.T.D. contributed in reviewing, advising and editing from proposal development to the final manuscript write up.

Funding

Amhara Agricultural Research Institute provides financial and logistic supports during the entire research fieldwork.

Data availability

The data used to support these findings will available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.