Studies of genotype by environment (GE) interactions in sugar cane

Abstract

Sugar cane is a perennial grass but in most countries productionnis based on annual cycles which involve the harvest and regrowth of thenabove-ground parts of the plant. Each annual cycle is referred to as ancrop. Sugar yield tends to decrease over a sequence of crops but thenextent of this trend differs among genotypes. Differential responses ofngenotypes to crops are manifest as genotype by crop interactions. Thesenare a subset of genotype by environment interactions (GE), wherein thenrelative value of genotypes is influenced by environmental effects such asncrops, years and locations. The interaction of genotypes with crops andnother classes of environments affects parental evaluation and the earlynstages of selection in sugar cane.The objective of this project was to investigate the magnitudenand nature of genotype by crop interactions in the early stages ofnselection in sugar cane, and to consider their impact on breeding strategies,ngenetic parameters, and response to selection. The effect of interspecificnhybridisation and the selectional origins of genotypes on these interactionsnwas also examined.The experiments involved a sequence of 3 annual plantings andnharvests at a single location which allowed comparisons between cropsnwithin years (crops and years are completely confounded in most sugarncane experiments). Sugar and cane yield, sugar content, and stalk weightnand number were studied. The population was genetically diverse andnincluded parents from several countries and from domestic programs aimednat increasing the genetic diversity of sugar cane through interspecific hybridisation with S. spontaneum.GE interaction was sufficiently large to affect selection.nSecond order interaction among genotypes, crops and years (GCY) accountednfor most GE interaction indicating that it was largely unpredictable andnthat genotype by crop interactions (GC) were not repeatable in differentnyears. All crop and year combinations were analysed as random environmentsnas most GE was unpredictable.It is unlikely that selection for adaptation to ratoon (i.e.,nregrowth) crops would be successful during the early stages of selectionnbecause GC interactions were not repeatable in different years. A largernsample of years (or locations) would be required to characterise thenratoon performance of genotypes by estimating a repeatable GC effect.nAlternative strategies to select for ratoon yields during the early stagesnof selection should be evaluated, such as frequent harvests or testing innplant and ratoon crops in several locations at the earliest feasible stagenin the program.The nature of genotypic responses to environments was examinednby grouping genotypes according to their pedigrees or according tonpattern analysis of genotype and environment means. Pedigree groups werenbased on selection histories and the minimum number of generations throughnthe pedigree to a S. spontaneum ancestor. Joint linear regression analysesnwere also examined but this method was unsuitable because GE interactionsndid not show a linear association with environmental means.Responses to environments by groups based on pedigrees werenconsistent with expectations based on preceding studies of interspecific hybridisation and evolution in sugar cane. Groups of genotypes morenclosely related to S, spontaneum performed well in environments with lownmean yield but were unable to exploit environments with high mean canenyield as effectively as other groups. This trend was most marked fornsugar content. n n n n