Genetic analysis of grain yield and yield-attributing traits in navy bean (Phaseolus vulgaris L.) under drought and optimal environments


 Knowledge of the genetic basis of navy bean (Phaseolus vulgaris L.) performance under drought stress (DS) is important for planning appropriate breeding and selection strategies in DS environments. Twenty-eight F2 progenies generated from an 8 x 8 half-diallel mating design were evaluated to determine combining ability effects and mode of gene action of grain yield (GYD) and yield attributing traits in navy bean under DS and non-stressed (NS) conditions. The experiments were conducted in two locations in a 6 x 6 square lattice design with two replications during the 2020 dry season. There were significant (p < 0.001; p < 0.05) positive correlations for number of pods per plant (NPPP), number of seeds per plant (NSPP) and 100-seed weight (SW) with GYD under both DS and NS. General and specific combining ability (GCA; SCA) effects were significant (p < 0.05) under both DS and NS for most traits indicating the importance of both additive and non-additive gene effects in the expression of the traits. Parents with best combining ability for most of the studied traits were G1, G7, G6 and G8 under NS, and G3, G4, G7 and G8 under DS. The most promising progenies with high values for GYD and its component traits under DS were G2 X G3, G2 X G8, G4 X G5, G4 X G8, and G6 X G8. Good general and specific combiners with high significant positive effects under DS should be used further in breeding for moisture stress tolerance.


Introduction
Navy bean (Phaseolus vulgaris L.) is an important food legume crop in Zimbabwe, grown mostly for human consumption and income generation. It is mainly consumed as canned beans making it an essential raw material for the bean canning industry in Zimbabwe, which requires around 4000 MT of grain annually (Mukweza, personal communication, May 2018). However, local supplies can meet only 40% of the national consumption requirements (Tsiko 2020). Despite its importance in human diets and the bean canning industry, average grain yields realised by smallholder farmers in Zimbabwe have remained relatively low (less than 700 kg ha −1 ) (AGRITEX 2016) against a yield potential of 3000 kg ha −1 . The large discrepancy in yields has been attributed to the different biotic and abiotic stresses, including drought stress ( Some of the navy bean farmers in Zimbabwe produce the crop in drought-prone areas under rain-fed conditions (January -April) with no capacity for supplemental irrigation during periods of drought stress (Mutari et al. 2021). Moreover, the rainfall pattern uctuates from season to season due to climate change, consequently exposing the crop to moisture stress (intermittent or terminal) at some stage during growth. This severely affects productivity and grain quality of navy bean. The recent participatory rural appraisal study conducted by Mutari et al. (2021) in the main navy bean growing regions of Zimbabwe revealed that the crop experiences moisture stress during the reproductive stage of development. Singh (2001) and Beebe et al. (2010) reported that about 60% of cultivated beans worldwide are grown under the risk of either terminal or intermittent drought. The impact of drought stress to navy bean productivity in Zimbabwe is further worsened by lack of drought tolerant cultivars (Mutari et al. 2021). Therefore, the development, commercial release, and subsequent cultivation of drought tolerant cultivars would be an effective, sustainable, and appropriate strategy for ensuring increased navy bean productivity and stability to ensure self-su ciency and food security in Zimbabwe.
To develop improved genotypes that are adapted to drought stress with signi cant genetic gain, knowledge of genetic variability of drought stress tolerance and its genetic basis is important. Several studies have reported that genetic variability for drought stress tolerance exists in dry beans (Assefa et al. 2013;Makunde 2013; Darkwa et al. 2016), implying that genetic improvement of this trait would be possible. Considering that grain yield is a quantitative polygenic trait, selection of grain yield along with its attributing traits could be more reliable and e cient than selection on the basis of grain yield alone (Romanus et al. 2008). Thus, selection e ciency in dry bean improvement programs can be improved by generating information on the association between grain yield and yield attributing traits.
Information on the combining ability and genetic control of economic quantitative traits such as grain yield and its associated traits under drought stress environments is cardinal when improving crops for drought tolerance. Combining ability is the breeding value of parental genotypes to produce superior hybrids based on the performance of their offspring (Romanus et al. 2008). General combining ability refers to the average performance of a parental line in a series of cross-combinations (Sprague and Tatum 1942;Gri ngs 1956). On the other hand, speci c combining ability is the average performance of certain cross-combinations relatively better or poorer than would be expected on the basis of the average performance (GCA) of the other parental lines involved (Su et al. 2017). Combining ability analysis assists bean breeders in the identi cation of the best hybrid combinations with high SCA and parental lines with high GCA, increasing the chances of selecting superior transgressive segregants in the subsequent segregating populations. It also provides information on the type of gene action governing the expression of different quantitative agronomic traits of interest.
Evidence for the expression of grain yield and its components in beans under drought stressed (DS) and non-drought stressed (NS) environments is contradictory. Phiri (2015) reported the predominance of additive gene action over the non-additive gene action in controlling the number of pods per plant, number of days to 50% owering, number of seeds per pod, hundred seed weight except for grain yield under DS environments. Asadi et al. (2010) and Amongi et al. (2015) reported that drought tolerance is governed by both additive and non-additive genes with predominance of additive gene effects for grain yield, pod weight, number of seeds per pod and number of pods per plant. On the contrary, Makunde et al. (2007) reported the predominance of dominance effects for grain yield, number of days to 95% owering and physiological maturity under both DS and NS conditions. These differences necessitate the need to conduct genetic analysis studies for germplasm to be used for speci c breeding programmes to elucidate the nature of gene action governing the inheritance of grain yield and yield related traits under moisture stressed conditions. The objectives of this study were to (i) estimate the combining ability effects and gene action for grain yield and its attributing traits among F2 progenies under DS and NS conditions and (ii) determine association between grain yield and yield attributing traits under DS and NS conditions, in order to assess the feasibility of indirect selection for grain yield.

Experimental materials and study locations
Eight parental lines comprising of three drought tolerant genotypes (ZABRA16575-73F22, G37 and DAB550), three drought susceptible genotypes (CZ113-13, G6 and G97) and two other genotypes (CIM-NAV02-17-3 and Navy Line60) with desirable agronomic traits that include high grain yield, early maturity, resistance to diseases and pod shattering were used (Table 1). Generation of lial generations, trial establishment, and management At owering, the parents were cross pollinated inside a glasshouse in all possible cross-combinations using an 8*8 half diallel mating design, Method II, following the arti cial hybridization procedure of Makunde (2007). The twenty-eight F 1 crosses, along with eight parents, were left to selfpollinate in a glasshouse from January 2020 to April 2020 to produce su cient seed for genetic analysis at the F 2 generation in replicated eld trials. A total of eight parents and twenty-eight F 2 progenies were evaluated at CRS and CES under DS and NS (control) environments in a 6*6 square lattice design with two replications, in two locations. Each genotype was planted in four row plots, 3 m long and 0.45 m wide. Intra-row spacing was 0.20 m. In the DS treatments, the genotypes were irrigated with overhead sprinkler irrigation system up to when 80% of the plants had owered and thereafter moisture stress was imposed until physiological maturity to induce terminal drought stress (Assefa et al. 2013;Beebe et al. 2013). This was done by withholding irrigation to 30% of the eld capacity before re-watering based on tensiometer reading (Darkwa et al. 2016). Under the NS treatments, soil moisture was maintained at eld capacity from planting until physiological maturity.

Data collection
The data for number of days from planting to 50% owering (DFW), number of days of seed ll (DSF), and number of days from planting to physiological maturity (DPM) were recorded from two middle rows of each plot excluding the boarder plants on the ends of each row. At physiological maturity, plant height (PH; cm), number of pods per plant (NPPP), number of seeds per pod (NSP), number of seeds per plant (NSPP), and 100-seed weight (SW; g) were recorded as average of the ten randomly selected plants. Additionally, grain yield (GYD; g) was recorded at 14% moisture content based on ten plants per plot.

Statistical analyses
The collected data were subjected to analysis of variance (ANOVA) per environment and also across environment (combined ANOVA) using the Breeding Management Systems (BMS) statistical analysis software version 18 (The Integrated Breeding Platform's BMS 2021). Bartlett's test of homogeneity of error variance across the two locations was conducted for all traits (Bartlett 1947). The means were separated using the Least Signi cant Difference (LSD). In the combined ANOVA a mixed linear model was followed, genotypes were considered as xed effects and the environment, blocks and replications were considered as random effects. Pearson's correlation analysis was performed using Genstat 18th edition (Payne et al. 2018) to determine the degree of trait association within separate moisture treatments. The GCA and SCA effects were determined separately per environment according to Gri ng's (1956) Method II, Model I using the Analysis of Genetic Designs in R software, version 3.0 (Rodriguez et al. 2015). The xed model for combining ability analysis was as follows: Where Y ijk is the mean phenotypic value of a character measured on cross i x j in kth replication, µ is the general/population mean effect, g i and g j are the GCA effects of i th and j th parental lines, respectively, Sij is the SCA effect of i x j crosses, rk is the replication effect and e ijk is the environmental effect associated with the ijk th individual observation (Gri ng 1956; Dabholkar 1992). The signi cance of variance due to GCA and SCA effects was tested using 't' test. To make inferences on the type of gene action involved in the expression of GYD and yield components, the relative importance of GCA and SCA was determined using Baker's ratio (Baker 1978) as follows: where: MS GCA is mean square for GCA, and MS SCA is mean square for SCA. The GCA: SCA ratio was estimated by dividing GCA with SCA using the sum of squares of the respective trait. To identify parental lines and F 2 progenies that combine high GYD with drought tolerance, percentage grain yield reduction (% GYR) or drought tolerant index was calculated as the percentage of GYR due to moisture stress on the GYD obtained under NS environments as:

%GYR = meangrainyieldofgenotypeunderNS − meangrainyieldofthesamegenotypeunderDS meangrainyieldofthesamegenotypeunderNS X100
Genotypes with good stability across both moisture stress and non-stress environments will have lower values of the % GYR, while high values will indicate poor stability.

Results
Variation under moisture stressed and non-moisture stressed environments The mean square values and signi cant tests for the nine traits of 28 F 2 progenies and 8 parents evaluated across two water regimes and two locations are presented in Table 2. The mean square estimates for genotypes (G) were signi cant (p < 0.05) for DFW, DMP, NPPP, NSP, GYD and SW under both DS and NS environments. On the other hand, the mean square for genotypes was only signi cant (p < 0.05) for DSF and NSPP under DS conditions. Mean squares for location (L) were signi cant (p < 0.05) for NPPP, NSP and GYD under both DS and NS. Combined ANOVA showed high signi cant effects (p < 0.001) of the genotypes and locations on DPM, NPPP, and GYD. The effects of environmental conditions (water regime -WR; DS vs. NS) were signi cant (p < 0.05) for all the studied traits except for NSPP. Non-signi cant effects of the genotype x location interaction (G x L) were observed for all the traits except for NSP. However, the L X WR interaction signi cantly (p < 0.001) affected PH, DFW, DSF, NSP, and NSPP, while the G x WR interaction effects were only signi cant (p < 0.05) for DSF, GYD, and SW.
Mean performance of genotypes under moisture stressed and non-moisture stressed environments The means of parental lines and F 2 progenies with respect to GYD and its components are presented in Table 3   Under DS, the best performing progeny (G6 X G8) with respect to GYD, SW and DPM ranked seventh in terms of NPPP (29.50), however, this was not signi cantly different from the NPPP of the top performer with 37.00 (G1 X G5). This progeny also ranked second and third best with respect to GYD (2616 g) and SW (32.75 g), however, these were not signi cantly different from the GYD (2580 g; G4 X G8) and SW (41.50 g; G3 X G8) of the top performers. The progeny G4 X G8 ranked second best in SW under both DS (48.25 g) and NS (40.25 g), however, the SW was not signi cantly different from the SW of the top performers. Among the parents under DS, the best performing genotypes with respect to GYD, NPPP, NSP, NSPP and SW were G3, G6, G7, and G8. Under NS, the parent G8 ranked best for DPM, GYD, and SW and second for GYD under DS. Generally, the parents G3, G7, G4, and G8 were the top performing genotypes in terms of NSP, NSPP, GYD, SW and NPPP under NS.
Combining ability analysis of F 2 progenies and their parents for grain yield and yield attributing traits under moisture stressed and non-moisture    The parent G8 was the best general combiner for GYD as revealed by its signi cant (p < 0.001), positive and high GCA effects, followed by G6. The same trend was realized under DS conditions. Therefore, both G8 and G6 would be good general combiners for grain yield under DS and NS conditions. On the other hand, signi cant (p < 0.05) and negative GCA effects for GYD were observed on parents G2 and G5. G1 and G7 were high general combiners for NSPP with signi cant (p < 0.05) and positive GCA effects, G4 and G8 for SW with signi cant (p < 0.05) positive GCA effects, and G4 for NSP with signi cant (p < 0.05) positive GCA effects.
Speci c combining ability estimates for grain yield and yield attributing traits under moisture stressed conditions The estimates of SCA of 28 F 2 families for GYD and its components are presented in Table 6 and Online Resource 3. Four crosses, namely G3 X G5, G3 X G7, G4 X G8, and G1 X G5 exhibited signi cant (p < 0.05) and positive SCA effects for NPPP under DS conditions. On the other hand, four crosses, namely G4 X G5, G2 X G8, G2 X G3, and G6 X G8 showed signi cant (p < 0.05) and positive SCA effects for GYD. However, signi cant (p < 0.05) and negative SCA effects on GYD were exhibited by G2 X G6 and G5 X G8. For DPM, signi cant (p < 0.05) negative SCA effects were exhibited by G1 X G4, G1 X G7, G4 X G6, G5 X G7, and G6 X G8. Thus, parent combination G6 X G8 combined signi cant (p < 0.05), negative and positive SCA effects for DPM and GYD respectively. Good speci c combiners for NSP with signi cant (p < 0.05) SCA effects were G1 X G5, G2 X G3, and G3 X G5. Cross-combinations of G4 X G8, G5 X G7, and G6 X G8 showed signi cant (p < 0.05) and positive SCA effects for SW. Signi cant (p < 0.05) and positive SCA effects for NSPP were exhibited by G1 X G2, G4 X G8, G5 X G6, and G6 X G7.
Speci c combining ability estimates for grain yield and yield attributing traits under non-moisture stressed conditions Higher positive and signi cant SCA values were considered desirable for GYD and its components under NS environments ( Table 6). The parent combination G4 X G5 showed non-signi cant (p > 0.05) and negative SCA estimates for GYD under NS despite exhibiting signi cant SCA estimates for GYD under DS conditions. In addition, signi cant (p < 0.05) and negative SCA effects on GYD were exhibited by crosses G1 X G7 and G5 X G8.  See footnote in Table 1 for genotype codes. DS moisture stressed environments, NS non-stressed environments, PH plant height, DFW days to owering, DPM days to physiological maturity, DSF days to seed ll, NPPP number of pods per plant, NSP number of pods per pod, NSPP number of seeds per plant, GYD grain yield, SW 100 seed weight, * p < 0.05, ** p < 0.01, *** p < 0.001 Two crosses, namely G4 X G8 and G6 X G8 showed signi cant (p < 0.05) and positive SCA effects for GYD, NSPP, and SW. Other good combiners with signi cant (p < 0.05) SCA effects for SW under NS conditions were G3 X G8 and G5 X G7. Good speci c combiners for NSP were G1 X G5, G2 X G3 and G2 X G6. For NPPP, signi cant (p < 0.05) and positive SCA effects were exhibited by the crosses G2 X G3, G4 X G5, and G4 X G8. Crosses G2 X G8, G3 X G5, G3 X G7, G5 X G6 and G5 X G8 showed signi cant (p < 0.05) and positive SCA for DPM. On the other hand, signi cant (p < 0.05) and negative SCA effects for DPM were exhibited by the crosses G1 X G7, G2 X G6, G5 X G7 and G6 X G8.
Association of grain yield with yield attributing traits under moisture stressed and optimal environments Results of correlation coe cients among different traits under DS and NS conditions are presented in Table 7. Most of the correlations observed ranged from weak to strong. Signi cant and positive correlations were observed for NPPP (r = 0.19, p < 0.05) PH (r = 0.21, p < 0.05) and SW (r = 0.36, p < 0.001) with GYD under DS conditions. In addition, the NPPP was signi cantly and positively correlated with NSP (r = 0.17, p < 0.05), NSPP (r = 0.60, p < 0.001), PH (r = 0.40, p < 0.001) and SW (r = 0.17, p < 0.05) under DS conditions. Furthermore, signi cant and positive correlations were observed for NSP with NSPP (r = 0.26, p < 0.01) under DS conditions. Under NS conditions, signi cant and positive correlations were observed for DPM (r = 0.96), NPPP (r = 0.33), NSPP (r = 0.42) and SW (r = 0.29) with grain yield. In addition, the NPPP was signi cantly (p < 0.001) and positively correlated with NSP (r = 0.32) and NSPP (r = 0.72) under NS conditions. Furthermore, signi cant (p < 0.001) and positive correlations were observed for NSP with NSPP (r = 0.34) under NS conditions.  On the other hand, the signi cant mean squares of SCA observed under both test environments for traits such as DFW, DPM, NPPP, GYD, and SW indicate that non-additive gene action was also important in accounting for the expression of the aforementioned traits. This implies that arti cial selection of these characters for further genetic improvement could be possible through arti cial hybridization and recurrent selection methods (Owusu et al. 2017 The signi cant interaction of genotypes, GCA and SCA with location for some of the traits indicates that different gene combinations, alleles and genes may be present for developing improved cultivars of navy bean that are adapted to moisture stress. For instance, "G1" exhibited positive and negative GCA effects for GYD under NS and DS environments respectively; thus, the gene/s in uencing the expression of GYD under optimal conditions in "G1" did not contribute to moisture stress tolerance. Conversely, the parents "G3" and "G4" had positive (tolerant to moisture stress) and negative (poorly adapted to optimal conditions) GCA effects under DS and optimal environments. Similar ndings were reported by Rainey and DPM, indicating that they could also be useful in breeding for earliness in drought prone regions. Therefore G3, G4, and G8 could be useful in navy bean breeding programs to improve GYD and its components under both DS and NS environments. Parental lines such as G1, G7, and G8 showed higher GCA effects on GYD under optimal conditions than under DS conditions suggesting that the ability of genotypes to combine well depends on action, interaction, and linkage relationships of genes (So et al. 2006 three-way crosses, and four-way crosses) other than single crosses in order to enhance genetic recombination. Mutation breeding through mutation induction and backcrossing breeding can also be used to enhance genetic recombination. The signi cant, positive, and high SCA effects for GYD observed under DS environments for cross-combinations G2 X G3, G2 X G8, G4 X G5, and G6 X G8 was an indication of transfer of favourable alleles for improved GYD performance from the parents to the progenies. Such crosses may result in transgressive segregants that could be selected for DS environments. Thus, these crosses represent potential breeding material to further select for improved GYD and yield attributing traits under DS and NS environments. The superior performance of these crosses may be attributed to the involvement of additive x dominance (high/low -G2 X G8; G4 X G5; G6 X G8), and additive x additive (high/high -G4 X G8) type of gene action interactions for expression of GYD and its components. In studies by Iqbal et al. (2012) on dry bean, most of the superior cross-combinations under optimal environments involved high GCA x low GCA, average GCA x low GCA, high GCA x average GCA and average GCA x average GCA general combiners.
Over dominance, wide genetic base and complimentary epistatic gene effects (non-allelic interaction) at heterozygous loci could be attributed to the superiority of progenies from parents with low GCA effects (Girase and Deshmukh 2000; Nayak et al. 2018). Similar ndings were observed by Goncalves-Vidigal et al. (2008) and Senbetay et al. (2015) who reported that parental genotypes with low GCA effects resulted in superior hybrids in dry bean under optimal conditions. These ndings imply that parental genotypes should not be discarded from the breeding pipeline based on only unfavourable and negative GCA estimates. Interestingly, some of the parents with high and positive GCA effects for some of the traits produced progenies with unfavourable negative SCA effects under both DS and NS environments. Similar ndings were reported by Mwije et al. (2014) and Musembi et al. (2015) in sweet potato (Ipomea batatas L. Lam.) and Chiipanthenga et al. (2021) in soybean. These ndings could be attributed to the lack of genetic complementation between genes of the parents involved and the presence of modi er genes which may act in combination resulting in large phenotypic variability. This suggests that high GCA effects of the parental lines indicate their relative superiority but does not necessarily mean that crossing of parental lines with high GCA effects will result in progenies with signi cant, positive, and high SCA effects.
The signi cant positive correlations that were observed between GYD and yield components (NPPP, NSPP, and SW) under both environments suggest that these traits can be improved simultaneously. Therefore, GYD improvement in navy bean under both DS and NS environments can be achieved through both direct and indirect selection for NPPP, NSPP and SW. Similar ndings were reported by Romanus et al. (2008) in cowpeas with respect to DF, DSF, and DPM under optimal conditions.

Conclusions, Recommendations, And Implications For Breeding
There is potential for breeding progenies with superior drought tolerance, which can improve GYD performance under DS environments in Zimbabwe. Both additive and non-additive gene effects were important in the inheritance of GYD and its attributing traits with preponderance of additive gene action under both test environments. This implies that there is need to incorporate breeding schemes that exploit both additive and non-additive genes in navy bean breeding. Parents G3, G4, G7, and G8 were the best combiners for GYD and its components under DS. These parents could be utilized in navy bean improvement programs to form base populations with improved tolerance to drought. The best performing speci c combiners with consistent high values for most of the studied traits under both environments were G2 X G3, G2 X G8, G4 X G5, G4 X G8, and G6 X G8. Among these, G4 X G8 and G6 X G8 had high DTI under both DS and NS environments. Thus, potential lines with improved tolerance to drought can be selected from these promising crosses for further evaluation before release. The signi cant and positive correlations of GYD with NPPP, NSPP, and SW suggest the feasibility of indirect selection for GYD through secondary traits. Grain yield was signi cantly and positively correlated with NPPP, NSPP, and SW under both optimal and moisture stressed environments suggesting that these traits can be improved simultaneously. Grain yield, NPPP, NSPP, and SW were identi ed as best selection criteria for utilization in navy bean breeding. Breeding for superior grain yield under DS should involve high GCA x high GCA or high GCA x low GCA parental combinations. Parental genotypes should not be discarded from the breeding pipeline based on only unfavourable and negative GCA estimates. High GCA effects of the parental lines indicate their relative superiority but do not necessarily mean that crossing of parental lines with high GCA effects will result in progenies with signi cant, positive, and high SCA effects.