Improvement of Yield per se in Sugarcane

Abstract

Sugarcane is the main source of sucrose in the world and has also become one of the main sources of renewable energy, thanks to ethanol and electricity production. The main objectives of breeding programs are improving cane biomass and qualities. The genetics of current sugarcane cultivars (Saccharum spp.) are extremely complex, owing to a high polyploid genome of 10 to 12 homeologous chromosome sets, resulting from the interspecific origin of two polyploid ancestral genomes. A century of breeding efforts based on elite domesticated progenitors improved by wild introgressions has helped build one of the most productive biomass crops. However, in recent decades the percentage increase in annual sugarcane yield has been lower than in other crops. Until now, cultivar improvement has relied on traditional breeding programs that require 12 to 15 years of expensive field evaluations. This article discusses the most recent advances in genomics applications to support traditional breeding. In the past two decades, efforts have been made to construct genetic maps, study quantitative trait loci (QTL), and start association mapping studies. Today, genomic selection (GS) approaches to select individuals for advancement in the breeding process is believed to be appropriate for complex traits such as yield, thanks to improved estimates of marker effects combined with a better grasp of small-effect QTLs. GS is particularly attractive in the highly polyploid context of sugarcane, where the nature of yield genetic determinism is presumably highly quantitative. Frequent genotype x environment interactions add to the challenges associated with QTL detection. The focus in this article is on research areas that are likely to result in concrete improvements through advancing the incorporation of association genetics-based approaches. A strategy to design a breeder-friendly marker system is also presented. Plant growth modeling could provide sounder ecophysiological parameters for describing the complex biological process underlying yield and sucrose elaboration. Such models could overcome traditional problems caused by G x E interactions and consequently improve both QTL detection and GS approaches. (Resume d'auteur)