Genetic, environmental and management contributions to ratoon decline in sugarcane

•Yield and quality decline of sugarcane was expressed as a linear function of ratoon number for various agronomic field trials.•Environment and cultivar×environment interaction accounted for more variation in ratoon decline compared to effects of cultivar under irrigated conditions.•Switching of rankings of cultivars between early and late ratoons were more pronounced under rainfed compared to irrigated conditions.•Sharper ratoon decline trends were observed in non-fertilized compared to fertilized treatments in two trials, while nematicide application did not influence ratoon decline rates.•The linear coefficient captured statistically significant differences in ratoon decline and is recommended for further use in sugarcane. Studies of ratoon decline (RD) in sugarcane (Saccharum spp.) are primarily limited to investigations of cultivar ratooning ability (RA) and selection for this trait. The relative effects of cultivar, environment, and management on RD are rarely considered. This study (i) explored the relative effects of cultivar, environment, and management practices on RD patterns, (ii) evaluated the appropriateness of linear coefficients to describe differential RD trends, and (iii) determined if there are statistically significant differences in RD (as defined by a linear coefficient) between various agronomic treatments. Six irrigated cultivar trials from South Africa and Swaziland, three rainfed cultivar trials, two cycles of a long-term burning and trashing trial, and two cultivar×nematicide trials were analyzed. Cane yields (TCANE), estimated recoverable crystal percent (ERC) and ERC yields (TERC) were fit as linear functions of ratoon number (six or more crops) for individual plots in each trial. In the irrigated dataset, variance components analysis showed that environment (trial) contributed 83%, 59%, and 76% of total variation in RD for TCANE, ERC and TERC, respectively. The effect of cultivar accounted for less variation in RD compared with the effect of the cultivar×trial interaction for all three traits. However, significant (p