Deciphering Genotype-by- Environment Interaction for Targeting Test Environments and Rust Resistant Genotypes in Field Pea (Pisum sativum L.)

Abstract

Rust caused by Uromyces viciae-fabae is a major biotic constraint to field pea (Pisum sativum L.) cultivation worldwide. Deployment of host-pathogen interaction and resistant phenotype is a modest strategy for controlling this intricate disease. However, resistance against this pathogen is partial and influenced by environmental factors. Therefore, the magnitude of environmental and genotype-by-environment interaction was assessed to understand the dynamism of resistance and identification of durable resistant genotypes, as well as ideal testing locations for rust screening through multi-location and multi-year evaluation. Initial screening was conducted with 250 diverse genotypes at rust hot spots. A panel of 23 promising field pea genotypes extracted from initial evaluation was further assessed under inoculated conditions for rust disease for two consecutive years at six locations in India. Integration of GGE biplot analysis and multiple comparisons tests detected a higher proportion of variation in rust reaction due to environment (56.94%) as an interactive factor followed by genotype × environment interaction (35.02%), which justified the requisite of multi-year, and multi-location testing. Environmental component for disease reaction and dominance of cross over interaction (COI) were asserted by the inconsistent and non-repeatable genotypic response. The present study effectively allocated the testing locations into various categories considering their "repeatability" and "desirability index" over the years along with "discrimination power" and "representativeness." "Mega environment" identification helped in restructuring the ecological zonation and location of specific breeding. Detection of non-redundant testing locations would expedite optimal resource utilization in future. The computation of the confidence limit (CL) at 95% level through bootstrapping strengthened the accuracy of the GGE biplot and legitimated the precision of genotypes recommendation. Genotype, IPF-2014-16, KPMR-936 and IPF-2014-13 identified as "ideal" genotypes, which can be recommended for release and exploited in a resistance breeding program for the region confronting field pea rust.

Keywords: GGE biplot; desirability index; field pea; host plant resistance; repeatability; rust.

FIGURE 1

Symptoms of rust on infected field pea plants. (a) Infected field pea plants. (b) Infected field pea leaves. (c) Infected field pea stems. (d) Aeciospores of Uromyces viciae-fabae.

FIGURE 2

Principal component analysis (PCA) illustrating significant difference among test environments. Locations are: For Year-1 (2014--2015): FZB_1, Faizabad; GDP_1, Gurdaspur; KN_1, Kanpur; PNR_1, Pantnagar; SLG_1, Shillongani; and VAR_1, Varanasi. For Year-2 (2015--2016): FZB_2, Faizabad; GDP_2, Gurdaspur; KN_2, Kanpur; PNR_2, Pantnagar; SLG_2, Shillongani; and VAR_2, Varanasi.

FIGURE 3

Frequency distribution of 23 field pea genotypes for rust assessment at six locations in India during Year-1 (2014–2015) and Year-2 (2015–2016). Locations are: For Year-1 (2014--2015): FZB_1, Faizabad; GDP_1, Gurdaspur; KN_1, Kanpur; PNR_1, Pantnagar; SLG_1, Shillongani; and VAR_1, Varanasi. For Year-2 (2015--2016): FZB_2, Faizabad; GDP_2, Gurdaspur; KN_2, Kanpur; PNR_2, Pantnagar; SLG_2, Shillongani; and VAR_2, Varanasi.

FIGURE 4

Boxplot view illustrating the distribution of rust assessment among 23 genotypes of field pea across six test locations. The box represents the area from the first quartile to the third quartile. A horizontal line goes through the box at the median. The whiskers (vertical line) go from each quartile to the minimum or maximum.

FIGURE 5

Spearman’s correlation between six test locations for field pea rust during Year – 1 (2014–2015) and Year – 2 (2015–2016). ^*P < 0.05. Locations are: FZB, Faizabad; GDP, Gurdaspur; KN, Kanpur; PNR, Pantnagar; SLG, Shillongani; and VAR, Varanasi.

FIGURE 6

Mean vs. Stability view of the GGE biplot of 23 field pea genotypes across 6 testing locations. There was no transformation of data (transform = 0), and data were centered by means of the environments (centring = 2). The biplot was based on “row metric preserving.” Numbers correspond to genotypes as listed in Table 1. Locations are: For Year-1 (2014--2015): FZB_1, Faizabad; GDP_1, Gurdaspur; KN_1, Kanpur; PNR_1, Pantnagar; SLG_1, Shillongani; and VAR_1, Varanasi. For Year-2 (2015--2016): FZB_2, Faizabad; GDP_2, Gurdaspur; KN_2, Kanpur; PNR_2, Pantnagar; SLG_2, Shillongani; and VAR_2, Varanasi.

FIGURE 7

(A) PCA score values on PC1 vs. Genotype. (B) PCA score values on PC2 vs. Genotype. (C) PC1 score-values 95% BCa CLs (B = 920), shown centered on nominal score-values. (D) PC2 score-values 95% BCa CLs (B = 920), shown centered on nominal score-values. Numbers correspond to genotypes as listed in Table 1.

FIGURE 8

Hierarchical cluster analysis showing the relationship between 23 tested field pea genotypes against rust as well as 6 testing locations. Numbers correspond to genotypes as listed in Table 1. Locations are: For Year-1 (2014–2015): FZB_1, Faizabad; GDP_1, Gurdaspur; KN_1, Kanpur; PNR_1, Pantnagar; SLG_1, Shillongani; and VAR_1, Varanasi. For Year-2 (2015--2016): FZB_2, Faizabad; GDP_2, Gurdaspur; KN_2, Kanpur; PNR_2, Pantnagar; SLG_2, Shillongani; and VAR_2, Varanasi.

FIGURE 9

“Discrimitiveness vs. Representativeness” view of test locations based on GGE biplot of 23field pea genotypes across 6 testing locations. There was no transformation of data (transform = 0), and data were centered by means of the environments (centring = 2). The biplot was based on “row metric preserving.” Numbers correspond to genotypes as listed in Table 1. Locations are: For Year-1 (2014--2015): FZB_1, Faizabad; GDP_1, Gurdaspur; KN_1, Kanpur; PNR_1, Pantnagar; SLG_1, Shillongani; and VAR_1, Varanasi. For Year-2 (2015--2016): FZB_2, Faizabad; GDP_2, Gurdaspur; KN_2, Kanpur; PNR_2, Pantnagar; SLG_2, Shillongani; and VAR_2, Varanasi.

FIGURE 10

“Which-won-where” view of the GGE biplot of 23 field pea genotypes across 6 testing locations. There was no transformation of data (transform = 0), and data were centered by means of the environments (centring = 2). The biplot was based on “row metric preserving.” Numbers correspond to genotypes as listed in Table 1. Locations are: For Year-1 (2014--2015): FZB_1, Faizabad; GDP_1, Gurdaspur; KN_1, Kanpur; PNR_1, Pantnagar; SLG_1, Shillongani; and VAR_1, Varanasi. For Year-2 (2015--2016): FZB_2, Faizabad; GDP_2, Gurdaspur; KN_2, Kanpur; PNR_2, Pantnagar; SLG_2, Shillongani; and VAR_2, Varanasi.