Introduction

The main factors limiting the productivity of maize (Zea mays L.) are the inadequate water and nitrogen (N) supplies (Wang et al. 2020). Changes in climatic conditions have further increased the probability of drought (Wang et al. 2020) as temperatures rise and rainfall distribution changes in key traditional production areas. Drought poses a grave threat to agricultural production world-wide and is a critical abiotic prevalent environmental factor that results in a significant increase in maize yield losses (Fischer et al. 2020), which threatens the livelihoods of many people around the globe (Liedtke et al. 2020). Thus, increased drought tolerance has been a major goal of maize breeding (Liu and Qin 2021). Then, to ensure continued gains in maize breeding, new drought-stress tolerant source populations are needed (Cairns et al. 2013; Tofa et al. 2021). Drought stress strongly affects growth and N metabolism, while the application of N can contribute to drought resistance to a certain extent in many plants (Wang et al. 2016). Ribeiro et al. (2018) reported that low soil nitrogen is one of the most important abiotic stressors responsible for significant yield losses in maize. Nitrogen helps maize plants to increase leaf area, and constitute a vital component required for the synthesis of chlorophyll, which enhances photosynthesis and consequently grain yield and total biomass (Kaur et al. 2015). Consequently, the development and commercialization of low nitrogen tolerant genotypes can contribute to improved food security in developing countries (Ribeiro et al. 2018).

An urgent need for useful traits that contribute to tolerance to abiotic stresses are crucial for the improvement of new crops to achieve food security (Kamal et al. 2019). Breeding for low N and drought have common characteristics, suggesting that tolerance to either stress involves a common adaptive mechanism (Banziger et al. 1999). Particularly, delay leaf senescence (LS) contribute for yield under both conditions (Banziger et al. 1999). Therefore, the stay-green phenotype is associated with a higher drought resistance and a better performance under low nitrogen conditions in several crops (Gregersen et al. 2013).

Loss of chlorophyll is the visible symptom of leaf senescence and, by definition, the stay-green trait reflects impaired or delayed chlorophyll catabolism (Thomas and Ougham 2014). Functional stay-green (SG) is a drought adaptation phenotype expresses itself in a delayed onset of senescence, a slower senescence rate, or maintain of leaves greenness of SG-genotypes for longer after anthesis than senescent ones, allowing maintenance of photosynthesis for longer during the grain-filling period (Thomas and Howarth 2000), especially under drought and heat stress conditions (Kamal et al. 2019). Thus, SG plants have longer grain-filling period and subsequently higher yield than non-SG (Kamal et al. 2019). Liedtke et al. (2020) reported that plants expressing this phenotype show less drought-induced senescence and maintain functional green leaves for longer when water limitation occurs during grain fill, conferring benefits in both yield per se and harvest ability. Stay-green may be functional or cosmetic (Thomas and Howarth 2000). The functional SG genotypes are agronomically important, as they are able to maintain their photosynthetic capacity compared with the non-SG genotypes. The functional SG genotypes delay the onset of senescence (type-A) or the rate of senescence is initiated but leaf yellowing and the decrease in photosynthetic rate are slower (type-B) (Thomas and Ougham 2014).

Drought and low nitrogen are among the environmental conditions accentuating beneficial effects of delayed leaf senescence on yield (Gregersen et al. 2013). Breeding for functional SG has contributed to improving crop yields, particularly when it is combined with other useful traits (Kamal et al. 2019; Zhang et al. 2019). In the same way Antonietta et al. (2016) reported that breeding has developed better yielding maize hybrids for low N environments, which also have delayed leaf senescence (‘stay green’ trait).

A large genetic diversity resides in the African maize landraces which could be conserved and exploited for maize improvement (Stephen et al. 2016). However, with the exception of a few studies on Algerian maize populations from the Sahara Desert, little research on maize genetic diversity in North Africa is currently available (Djemel et al. 2012; Aci et al. 2013). Because of its adaptation to biotic and abiotic stress, Saharan maize may be a great donor of alleles of stress tolerance. Previous phenotypic (Djemel et al. 2012, 2018, 2019; Maafi et al. 2021; Akrour et al. 2021) and genetic (Aci et al. 2013, 2018; Belalia et al. 2019) studies were made after the collection of some maize populations from a subtropical area in the Algerian Sahara in 2009. These authors have shown that Algerian maize has high diversity and may provide new alleles for water stress conditions. Heterotic patterns among Algerian germplasm were studied by Cherchali et al. (2018), and suggested the incorporation of this material in breeding programs due to their high genetic divergence. Then, Riache et al. (2021) and Riache et al. (2022) evaluated a diallel among six Algerian maize population under water stress and no-nitrogen fertilization to select the most productive genotypes but they did not study the relationship with leaf senescence traits.Various techniques for plant improvement have been developed over the years, each with the aim of obtaining hybrid cultivars that express advantageous traits, particularly increased productivity, deriving from heterosis or hybrid vigor (Reis et al. 2013). Heterosis refers to the phenomenon that progeny of diverse varieties of a species or crosses between species exhibit greater biomass, speed of development, and fertility than both parents (Birchler et al. 2010). Heterosis effects and combining ability are often estimated by populations derived from special genetic models, such as the diallel (Yu et al. 2020).

Diallel mating design was used for determining the utility of genotypes as parents in the development of hybrids (Mahto 2003). In addition, diallel analysis has been used in studies on inheritance and genetic control of important traits (Santos et al. 2019). Thus, several methods have been devised for analyzing half diallel data to estimate the genetic components in plant populations. Of these, Gardner and Eberhart (1966) using the set-up multiple regression approach, partitioning heterosis in terms of average, general and specific heterosis effects (EL-Satar 2016). Information about the inheritance of the stay-green trait in maize are limited. Some previous studies shows that it is controlled by only one locus with two alleles, presenting complete dominance and that its inheritance is quantitative (Gentinetta et al. 1986). However, other studies have reported that additive effects are more important than non-additive effects (Beavis et al. 1994; Guei and Wassom 1996; Bänziger et al. 2000; Costa et al. 2008).

Therefore, the objectives of this study were to (1) estimate varietal and heterosis effects of Algerian Saharan populations and their crosses for stay-green characteristics under water stress and no-nitrogen fertilization conditions and by that (2) know the relationship between stay-green and yield under stress conditions, and (3) identify the most promising populations per se and crosses to use in selection for stress tolerance across years.

Materials and methods

Germplasm materials

The six maize parental populations used in this study were selected from the collection of maize germplasm reported by Djemel et al. (2012) (Table 1). Aci et al. (2013) studied these populations based on genetic distances and geographic origin. The six Algerian populations were used as parents in a diallel mating design without reciprocals in 2013, producing 15 hybrids (Djemel et al. 2018). Cherchali et al. (2018) described the method of the diallel crossing system, where 60 pairs of plants were used to produce 60 crosses for each pair of populations, and for each hybrid, a bulk of all kernels was made.

Table 1 Name and origin of the six Algerian populations used in the diallel mating design
Full size table

Field trials

The six parents and their fifteen crosses along with the hybrid EP17 × EP42 from Spain, as a check from the European Flint germplasm, the two synthetic varieties EPS20 (originated from eight Reid inbred lines), EPS21 (originated from eight non-Reid inbred lines) and their respective cross EPS20 × EPS21], as check of the Reid × non-Reid heterotic pattern, were evaluated under water stress and no-nitrogen fertilization in Algiers, Algeria (36°43′16"N, 3°09′03"E, 36 m altitude, with 600 mm of annual rainfall). Maize seeds were sown manually the 18th May 2018 and 3rd May 2019, respectively. Plants were harvested by hand the 2nd and 7th of September in 2018 and 2019, respectively. During the cropping period, total rainfall was 189 mm (2018) and 76 mm (2019), while minimum temperature varied from 10.6 (April) to 20.3 (July) in 2018, and from 9.6 (April) to 21.2 (July). However, maximum temperature varied from 21.9 (April) to 32.4 (July) in 2018 and from 21.2 (April) to 34.1 (July) in 2019 (Infoclimat 2021).

The field trials were arranged following a split-split-plot design with three repetitions. Each repetition contained two main plots representing the two water regimes (well-watered and water stress). Each main plot was divided into two sub-subplots that include the two nitrogen treatments 0 kg/ha and 120 kg/ha of nitrogen. Finally, within each subplot, the genotypes were randomized in experimental units consisting of one 6-m row with 0.7 m row spacing. The sowing density was 70,000 plants/ha with inter-hills spaces of 0.2 m.

In order to know the characterization of experimental soil, analyses of five bulked samples following a diagonal at 30 cm depth were made before sowing. The test site was silty with a low organic matter content (1.53%) which contained 100 ppm of total nitrogen and moderately poor (20 ppm) in inorganic nitrogen (N min) in both years. These results allowed the estimation of nitrogen fertilizer rate. Plots with nitrogen supply received 120 kg of N per ha in the form of urea, which was applied in two splits with 1/3 (40 kg/ha) at three-leaf growth stage and 2/3 (80 kg/ha) at the six leaves growth stage. The total amount of water received by maize plants was 600 mm under control conditions and 300 mm under water stress conditions from sowing to post-flowering. For plant irrigation, every week, the drip irrigation method was used and the exact quantity was removed when the trials obtained water from rainfall. Weeding was done manually when necessary.

Stay-green trait was recorded as leaf senescence and plant coloration. Plants in each experimental unit were rated every 10 days, starting from 70 days after sowing (7 days after full flowering approximately), with a scale of 1–9 for leaf senescence (otherwise called leaf death), where 9 = 0–10% dead leaf area is taken upwards from the base of the plant and 1 = 90–100% dead leaf area. For coloration trait, 1–9 green coloration scale was used, where 9 = 90–100% of the plant is with normal green leaf coloration, and 1 = 0–10% of the plant leaves remain green. Finally, grain yield was previously measured and published by Riache et al. (2021) in t/ha, and used in this study to explain any correlation with stay-green traits where five ears harvested from each single-row plot, were shelled to determine percentage moisture and grain weight per hectare adjusted to 14% moisture by recording fresh and dry grain weight.

Statistical analyses

Plant senescence and leaf coloration data for each genotype within all the treatments of water regimes and nitrogen rates were regressed against days of measurement (linear regression) to estimate the slope, which has been used in all analyses of variance including diallel. The regression analysis using SAS 9.4 allows the use of the slope as an indicator of stay-green variation.

Most important traits of commercial crops are controlled by polygenes with various kinds of genetic effects that are affected by the environment. Thus, the variety trials commonly involve several environments. Replicated yield trials involving several environments (MET) are used in breeding programs to select genotypes based on yield and other economically important traits. One of the focus of Multi-environment Trials (MET) is the “Broad” (across environments) inference (Balzarini 2002).

A mixed effects model contains both fixed and random effects. Fixed effects are the variables that we’re interested in making conclusions about. They have levels that are of primary interest and would be used again if the experiment were repeated. In contrast, random effects are parameters that are themselves random variables and have levels that are not of primary interest, but rather are thought of as a random selection from a much larger set of levels (Pinheiro and Bates 2006).

In our combined analyses of variance, years (environments) and repetitions were considered as random effects whilst the genotypes (populations per se and crosses) and treatments (nitrogen and water regime) were considered as fixed effects.

The combined analyses of variance over years and conditions and individual analysis by each condition over the years were made by using the PROC MIXED procedure of SAS 9.4 software. The mean values of maize genotypes under water stress and no-nitrogen fertilization were compared by using Fisher’s protected Least Significant Difference (LSD) with LSD tests at P = 0.05.

The final model used by SAS was:

$$\begin{aligned} {\text{Y}} & = \mu + {\text{E}} + {\text{Rep }}\left( {\text{Y}} \right) + {\text{treatment I}} + {\text{Y}} \times {\text{treatment I}} + {\text{treatment II}} \\ & \quad + {\text{Y}} \times {\text{treatment I}} + {\text{treatment I}} \times {\text{treatment II}} \\ & \quad + {\text{Rep }}\left( {{\text{Y }} \times {\text{ treatment I }} \times {\text{ treatment II}}} \right) + {\text{Entry}} \\ & \quad + {\text{Y}} \times {\text{Entry}} + {\text{treatment I}} \times {\text{Entry}} + {\text{treatment I}} \\ & \quad \times {\text{Entry}} + {\text{treatment I}} \times {\text{treatment II}} \times {\text{Entry}} + {\text{Y}} \\ & \quad \times {\text{treatment I}} \times {\text{Entry}} + {\text{Y}} \times {\text{treatment II}} \times {\text{Entry}} \\ & \quad + {\text{Y}} \times {\text{treatment I}} \times {\text{treatment II}} \times {\text{Entry}} + {\text{Error}} \\ \end{aligned}$$

where Y: is the observed measurement. μ: is the mean of the values observed in the experiment. Y, treatment I, treatment II, Entry: are the effects of years, irrigation regimes, nitrogen rates and genotype (15 hybrids, six populations per se and the 4 checks), respectively, Rep (Y) is the effect of repetition within years, Error is the residual term and treatment I + Y × treatment I, Y × treatment I, treatment I × treatment II, Y × Entry, treatment I × Entry, treatment I × Entry, treatment I × treatment II × Entry, Y × treatment I × Entry, Y × treatment II, and Y × treatment I × treatment II × Entry are the interactions effects between the different factors.

Method II of Gardner and Eberhart was used to estimate varietal effects and heterosis effects (average heterosis, varietal heterosis and specific heterosis) in the diallel crosses for each water × nitrogen treatment combination, excluding the checks.

The results were calculated following this statistical model:

$${\text{Yij}} = {\text{E}} + {\text{b}}\left( {\text{e}} \right) + \mu {\text{v}} + {1}/{2}\left( {{\text{vi}} + {\text{vj}}} \right) + \raise.5ex\hbox{$\scriptstyle 1$}\kern-.1em/ \kern-.15em\lower.25ex\hbox{$\scriptstyle 2$} \left( {{\text{evi}} + {\text{evj}}} \right) + {\text{k}}\left( {{\text{hij}} + {\text{ehij}}} \right) + {\text{Error}}$$

where Yij is the average value obtained for each variety (i = j) or for a cross (i ≠ j); E is the year effect; b(e) is the effect of repetition within year; μv is the mean of n parental genotypes; vi and vj are the varietal effects for i and j respectively; k = 0 when i = j and k = 1 when i ≠ j; hij is the overall heterosis effect; evi is the interaction effect of year and population i and ehij is the interaction of year and heterosis of populations i and j. Error is the experimental error. Varietal effect was calculated as the difference between the mean performance of each parent and the mean of all parents, whereas heterosis effect was calculated as the difference between the mean of two parental populations and their cross. In addition, hij is the deviation from mean heterosis observed in the cross of populations i and j; partitioned into these components:

$${\text{hij}} = {\text{h}} + {\text{hi}} + {\text{hj}} + {\text{sij}}$$

where h is the average heterosis of all crosses calculated as the difference between the mean of all crosses and the mean of all parents; hi and hj are the parental heterosis contributed by the variety i and j in its crosses measured as a deviation from the average heterosis, and sij is the specific heterosis effect of the cross between ith and jth parents. The DIALLEL-SAS05 program of Zhang et al. (2005) was used to analyze all data.

Finally, linear regression of yield on leaf senescence and plant coloration were calculated for the mean of six Algerian maize populations and their diallel crosses under each stress combination, namely water stress plus no-nitrogen supply, water stress plus nitrogen supply, well-watered plus no-nitrogen supply, and well-watered plus nitrogen supply.

Results

Analyses of variance and comparisons of means

Combined analysis of variance revealed non-significant differences between the years, and most interactions between years and other factors were not significant (Table 2). The effects of genotypes were more important than irrigation, nitrogen, years and most interactions involving years. However, irrigation and nitrogen effects were not significant for stay-green traits. There were significant differences among maize populations under control and stress conditions (Table 2). In line with the current results, Cherchali et al. (2018) and Riache et al. (2021) found in the Algerian maize populations and their crosses a large amount of variability. Furthermore, African maize landraces are an important class of genotypes distinguished by wide diversity in phenology, plant growth, grain yield, and leaf photosynthesis, the majority of which indicate a diversity of farmer preferences and adaptive characteristics to a diverse range of environments in which they have evolved (Stephen et al. 2016). The presence of genetic variability among the genotypes for grain yield and other agronomic traits under drought, low N or drought combined with low N stress conditions suggested that significant progress could be achieved in selecting for improved grain yield and other traits under these stress conditions among genotypes (Oyekunle and Badu-Apraku 2018, Naggar et al. 2020). Such high variability suggested that the germplasm was adapted to a wide range of environmental conditions (Naggar et al. 2020), and could provide valuable alleles for maize improvement in temperate environments since Algerian maize has a wide adaptability to temperate regions and a high degree of genetic diversity (Djemel et al. 2012; Aci et al. 2013), and can be identified as potential sources of gene diversity for developing varieties with tolerance to abiotic stresses (Twumasi et al. 2017; Nelimor et al. 2020).

Table 2 Mean squares from the analysis of variance combined across years for leaf senescence and plant coloration analyzed in a diallel among six Algerian maize populations evaluated along with four checks in two years in Algiers under water stress and nitrogen deficiency
Full size table

Our populations are divided into three genetic pools, based on Cherchali et al. (2018): Cluster I (IGS: from the most southern area (Tamanrasset province); Cluster II (accessions from the center of the prospected area, Adrar (AOR, IZM, and MST), with a small group from the north of Algeria; and Cluster III from Bechar province (BAH) as described by Cherchali et al. (2018) who studied the heterotic patterns among Algerian germplasm. Furthermore, the distance between their collection sites was greater than 50 km (Cherchali et al. 2018).

The interactions were significant for years × irrigation and irrigation × genotypes (for all traits) and for irrigation × Nitrogen and years × irrigation × genotypes for plant coloration only (Table 2). Differences among genotypes were significant for all traits under most treatments (except under water stress for leaf senescence with nitrogen and plant coloration without nitrogen). Furthermore, years × genotypes interactions were not significant (Table 3). The non-significant genotypes × years interactions for stay-green traits under all treatments are evidence of the stability of genotypes and indicated that the response patterns were consistent across years (Oyekale et al. 2020). Exploitation of hybrid vigor and selection of suitable parents require knowledge on the significance and magnitude of varietal and specific heterosis. Information on the type of genetic action governing the inheritance of traits in a population guides the breeder on what objectives to design for designing optimized breeding programs for that population (Talabi et al. 2017).

Table 3 Mean squares from the analysis of variance of leaf senescence and plant coloration traits analyzed in the diallel systems with six Algerian maize populations evaluated along with four checks in two years in Algiers under both managed water stress and nitrogen deficiency
Full size table

For both leaf senescence and plant coloration, the regression coefficient is always negative. Small absolute values of b mean plants stay-green longer (Table 4). Under well-watered conditions, IZM had the most delayed leaf senescence value − 1.63 and − 1.9 under no-nitrogen supply and with nitrogen supply, respectively, though other populations were not significantly different. Under water stress with no-nitrogen fertilizer, IGS had the most delayed leaf senescence (− 1.73) while AOR had the fastest senescence (− 1.95), though they were not significantly different. EPS20 check had on average the fastest leaf senescence value and EP17 × EP42 the slowest leaf senescence in all treatments. Under water stress, general mean leaf senescence varied from − 1.75 under no-nitrogen supply to − 1.79 with nitrogen supply respectively, and from − 1.90 with no nitrogen supply to − 2.05 with nitrogen supply under well-watered conditions. Under water stress with no-nitrogen fertilizer, AOR and MST × BAH were not significantly different from the check EPS20 for leaf senescence. Under water stress with nitrogen fertilizer, IZM × BAH was the cross with the lowest leaf senescence value (− 1.9), followed by crosses involving MST with SHH (− 1.87) and BAH (− 1.85), and most crosses are not significantly different.

Table 4 Means a of leaf senescence and plant coloration traits analyzed in the diallel systems with six Algerian maize populations evaluated along with four checks in two years in Algiers under water stress and nitrogen deficiency
Full size table

Plant coloration of EP17 × EP42 was the best under all the treatments and among all the other genotypes and the worst value was found for EPS20 under well-watered conditions with both nitrogen levels. Most populations do not differ significantly. Under water stress and no-nitrogen fertilizer, EP17 × EP42 (− 0.88), AOR × IZM (− 1.05), SHH (− 1.12), IGS (− 1.17) were among the genotypes with high plant coloration values for more time, while under water stress only, EP17 × EP42 (− 0.9), SHH × BAH (− 1.23), IZM × IGS (− 1.27) were found among the best values. Under well-watered with no-nitrogen supply, the genotypes with the highest plant coloration values for more time were EP17 × EP42 (− 0.15), AOR × IGS (− 0.8), IZM × IGS (− 0.93) and the parent IGS. Finally, under optimum conditions, EP17 × EP42 (− 0.58), IZM × IGS (− 1.33) and SHH × IGS (− 1.75) were among the best plant coloration values.

Varietal and heterosis effects among Algerian Maize Populations

Analysis of diallel crosses was made separately for each treatment (Table 5). Combined analyses over years revealed significant differences among years under all the treatments. Entries (genotypes) were significantly different for all treatments except under water stress for leaf senescence with nitrogen supply, and for plant coloration with no-nitrogen supply.

Table 5 Mean squares from the Analysis II of (Gardner and Eberhart, 1966) of the diallel made among with six Algerian maize populations evaluated along with four checks in two years in Algiers under water stress and nitrogen deficiency
Full size table

The source of variation treatments was non-orthogonally divided in varietal effect and heterosis, and heterosis was divided into varietal, average, specific, and their corresponding interactions. Under well-watered conditions, varietal effects were significant for the two traits except for leaf senescence with nitrogen. Heterosis was significant for leaf senescence only with nitrogen supply. Variety × years interactions were significant for leaf senescence under both nitrogen levels. Years × entry interaction was significant for plant coloration with nitrogen supply. Heterosis × years interaction was significant for plant coloration under both nitrogen levels. Variety heterosis × years interaction was significant for plant coloration with nitrogen. Specific heterosis × years interaction was significant for plant coloration with no-nitrogen.

Under water stress, heterosis was significant for leaf senescence with no-nitrogen supply. Under no-nitrogen supply, specific heterosis was significant for leaf senescence. Average heterosis × years interaction was significant for leaf senescence with nitrogen supply. Average heterosis was significant and positive only for leaf senescence under water stress with no-nitrogen, indicating the existence of heterosis in this set of diallel crosses (Table 6).

Table 6 Genetic parameters from the Analysis II of (Gardner and Eberhart 1966) (Varietal effect, Heterosis effect, specific heterosis and average heterosis) for leaf senescence and plant coloration traits in the diallel made among six Algerian maize populations evaluated in two years in Algiers under both managed water stress and nitrogen deficiency
Full size table

For leaf senescence, under water stress, significant and negative specific heterosis were found in SHH × AOR and positive in AOR × IZM with no-nitrogen supply. Under well-watered conditions, MST had a significant and negative varietal effect (− 0.47) and (− 0.33) with no-nitrogen fertilizer and with nitrogen, respectively. However, IZM exhibited a significant and positive varietal effect (0.35) and (0.26) with no-nitrogen fertilizer and with nitrogen, respectively. Moreover, IGS had a significant and positive value (0.27) with no-nitrogen fertilizer. With nitrogen supply, significant and positive varietal heterosis for IGS (0.16) and MST (0.17) was observed.

For plant coloration, under water stress, AOR had a significant and positive varietal heterosis (0.15) with no-nitrogen supply. IZM × BAH showed a significant and positive specific heterosis (0.20) with no-nitrogen. Under the same water regime, with nitrogen supply, SHH × BAH recorded a significant and positive value (0.25), while significant and negative for SHH × MST (− 0.18). Under well-watered conditions, MST showed a significant and negative varietal effect, (− 0.57) and (− 0.44), with no-nitrogen fertilizer and with nitrogen, respectively.

A comparison between stay-green traits and yield calculated in Riache et al. (2021) was made using the estimations of genetic parameters revealed the presence of similar patterns of significant effects between leaf senescence and yield. Under well-watered conditions, a significant and positive varietal effect was recorded for IGS with no-nitrogen and negative for MST with nitrogen. Furthermore, under water stress conditions, a significant and negative specific heterosis was found for SHH × AOR but differed in the nitrogen rate. However, IZM recorded a significant and positive varietal effect under well-watered conditions with no-nitrogen supply, whilst for yield the varietal effect was negative.

Regression of yield on leaf senescence and plant coloration

Under water stress and no-nitrogen, leaf senescence (LS) has a positive significant effect on yield (R2 = 0.2607), while it was not significant for plant coloration (PC) (Fig. 1). Conversely, under water stress and nitrogen fertilization, PC had a negative significant effect on yield (R2 = 0.2946) and leaf senescence had not significant effect. Under well-watered and no-nitrogen, both LS and PC had positive significant effects on yield, being the effect of PC more important. Finally, under well-watered conditions and nitrogen fertilization, only PC had positive significant effects on yield.

Fig. 1
figure 1

Linear regression of yield on leaf senescence and plant coloration a for six Algerian maize populations and their diallel crosses evaluated under water stress and nitrogen stress during two years in Algiers (A, B: Water stress + No-nitrogen fertilizer WSN-; C, D: Water stress + Nitrogen fertilizer WSN + ; E, F: Well-watered + No-nitrogen WWN-; G, H: Well-watered + Nitrogen WWN+)

Full size image

Discussion

The analysis of variance proposed by Gardner and Eberhart (1966) included the main sources of variation: repetitions, entries (parental lines and hybrids), error, and total. The source of variation “entries” (genotypes) was broken down into effects of varieties (parental lines), and heterosis. On the other hand, total heterosis effect was broken down into average heterosis, variety heterosis, and specific heterosis. This method is useful in evaluating n varieties and their n(n- 1)/2 F1 crosses (Gardner and Eberhart 1966; Murray et al. 2003). Contribution of additive effects was assessed by estimating variance due to varieties (the variance within parents). Variety heterosis was estimated to judge overall contribution of a variety to its array heterosis. Specific heterosis helps identify hybrids with better characteristics than those of their parental lines. The significance of this heterosis indicates presence of dominance effects, probably attributable to favorable gene complementation in the crosses in which they form part (Valdés et al. 2017).

Understanding the genetic architecture controlling stay‐green traits and yield will aid in selecting future water stress and nitrogen deficiency tolerant maize genotypes. For leaf senescence, under well-watered conditions with nitrogen supply, the mean squares of heterosis were significant which suggests that non-additive effect was important in the inheritance of this trait (Table 5). However, with no-nitrogen supply, mean squares of varietal effects were significant, indicating the importance of additive effect for this trait. Under water-stress conditions with no-nitrogen, the significant mean squares of heterosis indicate that non-additive effects controlled this trait. For plant coloration, varietal effects were significant under well-watered conditions with both nitrogen levels, which suggested that additive effects controlled this trait (Table 5). However, Badu-Apraku et al. (2016) founded a greater contribution of general combining ability (GCA) sum of squares than specific combining ability (SCA) in stay-green characteristic under low N and across drought and low-N environments. Although the analysis of 55 inbred maize lines revealed additive and non-additive effects on the governance of stay-green, a predominance of additive effects was found (Costa et al. 2008). Annor et al. (2019) reported the dominating effect of SCA stay-green sum of squares over that of GCA which suggests that the non-additive effects regulated the inheritance of the trait and that SCA was the main constituent, under both low N and drought conditions. Average heterosis was significant and positive for leaf senescence under water stress with no-nitrogen (Table 6), and for yield under all the treatments (Supplementary table 1) (Riache et al. 2021), indicating that the crosses had, on average, higher values than the populations per se and the presence of adequate genetic diversity among the parental array which resulted in valuable heterosis in the first-generation hybrids. For leaf senescence, under well-watered conditions, significant and positive varietal effects were found for IZM (under both nitrogen levels) and for IGS (with no-nitrogen). Varietal heterosis was significant and positive under well-watered conditions with nitrogen supply for IGS and MST. For coloration, AOR presented a significant and positive varietal heterosis under water-stress conditions with no-nitrogen supply (Table 6). The implication is that these populations would be useful in breeding programs for delayed leaf senescence by transmitting their characteristics to the progeny. Therefore, appreciable progress could be made through recurrent selection. Recurrent selection is a population improvement strategy which utilizes multiple parents in the crossing program to accumulate favorable alleles while maintaining genetic diversity (Ramya et al. 2016).For leaf senescence, under water stress conditions with no-nitrogen supply, the most promising cross was AOR × IZM, with a significance and positive specific heterosis. For plant coloration, under water-stress conditions, based on specific heterosis, the most promising crosses were IZM × BAH (with no-nitrogen supply) SHH × BAH (with nitrogen fertilizer) indicated that these hybrids’ parents were good specific combiners for developing high-yielding hybrids which implied that it could be promising for developing drought-tolerant or/and low nitrogen hybrids through reciprocal recurrent selection to improve stay-green characteristics. Furthermore, none of the crosses had significant specific heterosis under well-watered conditions for the stay-green traits and water-stress conditions with nitrogen supply for leaf senescence (Table 6).

Reciprocal recurrent selection (RRS) method is an alternative for the improvement of maize yield under water stress, which simultaneously improved two populations (Valadares et al. 2022); by selecting for performance in the interpopulation cross (Schnicker and Lamkey 1993). Valadares et al. (2022) added that genetic variability could be explored in the next cycles of reciprocal recurrent selection for drought tolerance in maize progenies. Selection is based on the performance of the interpopulation cross, in which almost any type of genetic effect is expressed (Keeratinijakal and Lamkey 1993). Improvement of the cross between two populations by complementary changes in allelic frequencies between populations could be achieved (Keeratinijakal and Lamkey 1993), because this method favors the achievement of gains due to the additive effects, through the concentration of the favorable alleles in both populations, as well as the dominance (non-additives) deviations, since the genetic distance between the populations is maintained, which allows exploiting the heterosis in crosses between the populations and/or lineages deriving from them (Souza Jr and Pinto 2000; Santos et al. 2007; Vieira et al. 2021).Furthermore, the goal of RRS is to improve the mean performance of the interpopulation cross while maintaining the variability within populations for continued selection (Hinze et al. 2005). Through RRS, improved populations are used as a source of parental lines for the production of high-performance or superior hybrids (Reis et al. 2013).

The visual stay-green phenotype is defined as a delay in the beginning of leaf senescence and it is visually characterized by the maintenance of green leaf area during the grain-filling period (Thomas and Howarth 2000). The selection for this character is generally considered important for increasing the duration of kernel filling, and also for maintaining water use efficiency and root health—though the relationship between stay-green scores and grain yield under drought is often weak (Chapman and Edmeades 1999). Assimilating carbohydrates and then remobilizing them to newly developing organs or storage sections is the responsibility of green leaves with photosynthetic activity. In order to achieve a higher yield, it is critical for plants to retain adequate green leaf area at the final stage of plant development process (Borrell et al. 2000a,b).

Under water-stress conditions with no-nitrogen fertilizer, SHH × AOR exhibited significant and negative specific heterosis (− 0.23) for leaf senescence whilst it was the cross with the highest yield (1.31 t/ha, although other crosses were not significantly different) (Supplementary table 2) (Riache et al. 2021). This result can be explained by the mobilization of nitrogen from leaves to the grains in the grain-filling period. Besides, Pommel et al. (2006) reported that senescence is not simply a degenerative process because the N released from senescent leaf is recycled in the plant to storage organs or developing seeds. It is possible that increased demand of N by the larger ear due to selection is met by mobilization of N from the leaves, associated with mobilization to the grain and resulting in senescence (Muchow 1994). Both soil N uptake during grain filling and mobilization of N from vegetative tissues contributed to grain N under low N supply (Muchow 1994).

Comparing the estimations of genetic parameters revealed similar patterns between grain yield (Riache et al. 2021) (Supplementary table 1) and plant senescence (Table 6). Under well-watered conditions with nitrogen, a significant and negative varietal effect was recorded for MST for both plant senescence and yield. As reported by Yang et al. (2017), senescence process involves a rapid chlorophyll degradation and accompanying decline in photosynthetic capacity. Hence, Gregersen et al. (2013) showed that early senescence translates to reduced green leaf area and photosynthesis and significantly decreased grain yields.

Under well-watered conditions with no-nitrogen, IGS recorded a significant and positive varietal effect for leaf senescence and yield, indicating that IGS was tolerant to nitrogen stress and had a good delay for leaf senescence, with a long duration of active photosynthesis, which allows having a good yield as shown in our results. In line with our results, Riache et al. (2021) reported that based on early vigor and yield, IGS was tolerant to nitrogen stress because its relative performance was improved under no-nitrogen fertilization, compared to standard nitrogen fertilization. Furthermore, this result is an indication that the functional stay-green extends the period of photosynthesis to enhance the grain-filling rate as a way to increase the crop yield rather than increasing the size of the grain (Hörtensteiner 2009; Yang et al. 2017). Therefore, it is critical for the plant to maintain sufficient green leaf area until the final stage of its development process to achieve a higher yield (Yang et al. 2017; Zhang et al. 2019). Delayed leaf senescence in stay-green phenotype can enhance crop yields by remobilizing nutrients from source to sink under various stresses and nutrient-limited conditions (Munaiz et al. 2020). Chibane et al. (2021) reported that the stay-green genotypes had chlorophyll content and photosynthetic values higher than the corresponding values in the non-stay-green, during the whole grain-filling period. However, IZM recorded a significant and positive varietal effect for leaf senescence under well-watered conditions with no-nitrogen supply, while yield was negative. The presence of non-functional stay green can explain this opposition.

As reported by Hörtensteiner (2009), the photosynthetic rate in some of the stay-green genotypes decreases at the normal rate, although the chlorophyll content decreases much slower or later than traditional hybrids, and decreased grain production. This type of stay-green genotype is known as a ‘‘non-functional stay green genotype’’, or cosmetic, where the primary lesion is confined to pigment catabolism, while those genotypes with delayed loss of pigment and decrease in the photosynthetic rate are denominated as ‘‘functional stay-green genotypes’’, because the entire senescence syndrome, of which chlorophyll catabolism is part, is delayed or slowed down, or both (Thomas and Howarth 2000; Hörtensteiner 2009).

The onset and rate of senescence may be accelerated by stresses such as drought or N deficiency (Wolfe et al. 1988). An exposure of water deficit stress at flowering and grain filling stage brought severe negative effects on phenological and yield traits attributes of the maize lines (Sah et al. 2020). Drought stress results in a dramatic decline in the photosynthesis rate by reducing the chlorophyll content in ear leaf, which greatly contributes to the source for ear and kernel development as a photosynthesis factory in maize plants, and led to a final yield loss compared with the normal conditions (Song et al. 2019). As noted before by Edmeades (2008), under irrigation, gains in weight per kernel and stay-green were large but not under terminal stress, implying that kernels failed to fill fully under drought due to a lack of assimilates. However, in the studied varieties an acceleration of senescence was not detected when they were evaluated under stress conditions, indicating that it is an ideal material as a source of favorable alleles for stay-green trait.

Concerning the effects of stay-green on yield under water stress and no-nitrogen supply (Fig. 1), our results indicate that there were significant, though low, effects; indeed, slowing senescence improves yield under water stress and nitrogen stress. Conversely, improving the ability of maintaining color had negative effects on yield under water stress and nitrogen fertilization, and the effects on yield were positive under well-watered conditions and nitrogen fertilization. The ability of maintaining color increased yield under well-watered and nitrogen fertilization. Finally, the ability of maintaining senescence and color increased yield under well-watered and no-nitrogen fertilization. These results agree with previous publications showing the stay-green provides drought tolerance (Kamal et al. 2019; Zhang et al. 2019). Interestingly enough, our results suggest that stay-green could be particularly beneficial for improving tolerance to low nitrogen alone or in combination with water stress.

Conclusions

The present results revealed the existence of significant genetic variability for stay-green and stress tolerance in the material studied under stress. This knowledge indicated the importance of Algerian maize populations in future breeding programs to develop water and nitrogen stress-tolerant crosses.

Our results revealed the importance of additive and non-additive effects based on estimations of varietal and specific heterosis and the inheritance of traits. Breeding programs could capitalize additive effects by using the populations IZM or IGS with favorable significant varietal effects for leaf senescence under nitrogen stress, or dominance effects by using the population AOR with favorable significant heterotic effects for plant color under water and nitrogen stresses. Under water-stress conditions with no-nitrogen supply, the most promising cross AOR × IZM, based on specific heterosis, will be useful for delayed leaf senescence. However, under water-stress conditions, the most promising crosses IZM × BAH with no-nitrogen supply, and SHH × BAH with nitrogen fertilizer, will allow a longer maintenance of the plant greenness.

These parents could be promising for developing drought-tolerant or/and low nitrogen hybrids through reciprocal recurrent selection to improve stay-green characteristics and may result in significant genetic progress for attributes such as high yield. Finally, those populations and crosses could be selected and classified into useful heterotic groups, maximum heterosis could therefore be exploited.