Genotype × Environment Interaction for Resistance to Early Leaf Spot of Groundnut Mini Core Collections in the Savannas of Nigeria

The research was carried out at Bayero University, Kano (BUK, Latitude 11°58' N and Longitude 8°25' E) in 2016, 2017 and 2019; and Institute for Agricultural Research, Samaru Zaria (SMR, Latitude 11°11' N and Longitude 7°38' E) in 2019. In 2019, one field in each location was artificially inoculated with ELS to increase the inoculum load of the disease. The laboratory culture spores of ELS were sprayed on groundnut germplasm grown in the field 30 days after sowing (DAS). The weather information recorded for the different environments is presented in Fig 1.
 

 
One hundred and eighty-two groundnut mini core accessions including five check varieties were evaluated using 14 ´ 13 randomized incomplete block design with two replications. The description of the mini core collection along with the checks has been published previously (Shaibu et al., 2020). Single row plots measuring 4 m in length with inter- and intra-row spacing of 0.75 and 0.1 m, respectively were used. A basal dose of Nitrogen, Phosphorous and Potassium was applied to all the plots @ 20:40:40 kg ha-1, of N:P₂O₅:K₂O at planting. Hand weeding was done using hoes at 3^rd, 8^th and 12^th weeks after sowing (WAS) to prevent weed infestation and competition between plants and weeds.
 
The genotypes were categorized using a disease severity scale of 0-9 (Subrahmanyam et al., 1995) into resistant (£ 3), moderately resistant (4-5), susceptible (6-7) and highly susceptible (>7) as described by Sudini et al., (2015).
 
Data analysis
 
To adequately capture the effect of genotype (G), environment (E) and GEI, a mixed model analysis was done using PROC MIXED in SAS (SAS Institute, 2015). The genotypes that had £ 3 disease severity scores for ELS in each location were selected to check their stability for their reaction against ELS and pod weight across the environments. The stability analysis was done using all the integrated stability methods in STABILITYSOFT online program (Pour Aboughadareh et al., 2019).
 
Also, GGE biplot analysis was done using “R” packages to visualize the GEI based on the following model:
 
Y_ij - b_j = b_ja_i + l₁η_j1 + Σ_ij
 
Where,
Y_ij is the average yield of genotype i in environment j, b_j is the average yield of all genotypes in environment j, a_i is the main effect of genotype i, bj is the regression coefficient of the environment-centered yields (i.e., Y_ij-b_j) within environment j on the genotype main effects (ai). l₁ Σ_i1η_j1 is the first principal component (PC) from singular value decomposition (SVD) of the residual. The values of λ₁, Σ_j1 and η_j1 are simultaneously obtained by subjecting the environment-centered yield (i.e. Y_ij-b_j) to SVD.

Phenotypic variation for ELS and pod weight
 
The variance component analysis revealed highly significant (p <0.0001) genotype and GEI effects for ELS and pod weight (kg ha^-1) indicating that the genetic components and the GEI are more important in the reaction of groundnut genotypes against ELS and pod weight (Table 1). High variability in groundnut against ELS reaction has been previously reported (Shaibu et al., 2020; Zanjare et al., 2020). The significant G and GEI effects for resistance to ELS suggested the possibility of identifying resistant genotypes adapted specifically to a target environment and the need to deploy specifically adapted varieties in the future for more effective genetic control of ELS. The percent contribution of genotype to the total variability for ELS was 4.07% while the environment and GEI accounted for 55.84 and 5.11%, respectively. For pod weight, the genotype contributed 6.43% of the total variability while, the environment and the GEI contributed 5.95 and 18.20%, respectively.
 

       
The phenotypic and genotypic coefficient of variation (PCV and GCV) were moderate in each environment and the genetic advance as percent of the mean (GAM) ranged from 6.97% (BUK 2017) to 23.85% (BUK 2016) (Table 2). Similar PCV, GCV and GAM for late leaf spot (LLS) resistance in groundnut have been previously reported  (John et al., 2006; Vishnuvardhan et al., 2013; Chaudhari et al., 2019).
 

 
Disease reactions of genotypes against ELS
 
The number of genotypes in each environment for resistant and moderately resistant categories was highly variable (Fig 2). This highlighted the important role of GEI and the polygenic nature of ELS. This result was similar to that of Chaudhari et al., (2019) who showed that the reaction of groundnut to LLS was highly variable in different environments owing to the complex nature of the disease resistance, which is governed by polygenes with additive gene effects. The number of genotypes in the resistant and moderately resistant categories respectively was 62 and 73 (BUK 2016), 50 and 71 (BUK 2017), 25 and 100 (BUK 2019I), 5 and 18 (BUK 2019U), 8 and 22 (SMR 2019I) and 8 and 12 (SMR 2019U). ICG 3240 and ICG 4540 had ELS score ≤ 3 in all the environments. One of the improved varieties used (Samnut 22) also had an ELS score of £ 3 in all environments except at BUK 2017 where it had a score of 4. The other improved varieties used had varied responses to ELS in different environments. In 2019, at both BUK and SMR, six genotypes including Samnut 22 had ELS score of ≤ 3.
 

 
Stability of ELS reaction across the environments
 
An important objective of resistance breeding is to identify genotypes with durable resistance irrespective of the environment (Chaudhari et al., 2019). To identify consistent sources of ELS resistance, 48 genotypes that had ELS scores of ≤ 3 (except for the six checks) were subjected to stability analysis using different parametric and non-parametric statistics (Table S1). Wricke’s ecovalence (Wi²) had positive correlations with all other stability parameters except for the regression coefficient (bi) and the GE variance component (Fig 3). Shukla’s stability variance (σ²i) also had a similar trend with Wi². Based on Wi², σ²i and GE variance component, ICG 14106 was the most stable variety for ELS resistance. Six of the stability parameters ranked ICG 14106 as the most stable genotype while four ranked it as the second stable genotype. ICG 4540 was identified as the most stable genotype based on deviation from regression (S²di) and coefficient of variance (CVi) stability parameters. The few discrepancies observed in the ranking of the genotypes among the parametric and non-parametric statistics could have arisen due to the differences in the analytical methods (Pour Aboughadareh et al., 2019).
 

 
The first two PCs of the GGE biplot accounted for 58.27% of the total variation. This shows that most of the variabilities from the GEI were explained by the first two PCs. From the biplot of the relationship among the environments, BUK19I and SMR19I were related (Fig 4a). The similarity observed in the GGE biplot between BUK19I and SMR19I might be due to the creation of artificial epiphytotics by ELS spore inoculation in these two environments. The which won where/what biplot of the data showed that line 48 (TAG24) was the most resistant genotype among the selected genotypes at BUK in 2016 and 2017 (Fig 4b). Lines 1 (ICG 4540) and 3 (ICG 12991) had superior performances (resistant) in all the environments. The discriminativeness and repres entativeness of BUK19I and SMR19I (Fig 4c) were largely due to the inoculation with ELS carried out in these environments. Therefore, for cultivar evaluation against ELS resistance, the test environments should contain the right inoculum. Line 2 (ICG 3240) was the most resistant and stable genotype (Fig 4d) and other genotypes such as J11 and Samnut 24 were also stable but had ELS scores >5 across the environments. Samnut 22 (line 4) which showed resistance to ELS was, however, not stable. All the stability model used consistently identified ICG 1519 as a stable genotype. Therefore, ICG1519 can be used as an elite source of resistance against ELS in groundnut breeding programs in Nigeria.
 

 
Stability of pod weight across the environments
 
The correlations between the parametric and non-parametric statistics were majorly negative (Fig 5). Wricke’s ecovalence had positive correlations with regression coefficient (bi), deviation from regression (s²di) and Shukla stability variance (σ²i). Based on Wricke’s ecovalence, ICG 7897 was the most stable genotype followed by ICG 8494, ICG 8896, ICG 7463 and ICG 9449. Kang’s rank-sum statistic ranked ICG 9449 and ICG 8494 as the most stable genotypes followed by ICG 8896, ICG 12991 and ICG 7897.
 

 
The first two PCs of the GGE biplot explained 61.17% of the observed variations. The relationship among environments biplot (Fig 6a) showed that BUK16 and BUK17 are highly related. From the which won where/what biplot, lines 1 (ICG 4540) and 7 (ICG 11542) were ideal for BUK16 and BUK17 environments (Fig 6b). Lines 8 (ICG 8494) and 2 (ICG 3240) were ideal for BUK19U and SMR19I environments. SMR19U was the most discriminative and representative environment for the evaluation of the genotypes for pod weight (Fig 6c). The stability of the genotypes was determined by their projections onto the average-tester coordinate y-axis single-arrow line. The greater the absolute length of the projection of a genotype, the less stable it is. Lines 4 (Samnut 22), 30 (ICG 3584) and 44 (J11) were stable (Fig 6d). ICG 9449 and ICG 4540 were identified to be stable for both ELS and pod weight.
 

The incidence of ELS in the savannas of Nigeria is an ongoing challenge that necessitated the efforts to identify stable ELS resistant groundnut genotypes. The genetic composition and GEI played important roles in the reaction of the genotype against ELS and expression of pod weight. Here, we identified elite stable sources against ELS resistance under natural and artificial ELS epiphytotic environments. These identified sources of resistance can be used in the development and deployment of ELS resistant groundnut varieties that are high yielding. Furthermore, the presence of high genetic variability in the mini core also makes them important materials for genome-wide association studies for the identification of markers and genomic regions/genes that are linked to ELS resistance and pod weight of groundnut.