Quantitative review of fungicide efficacy trials for managing soybean rust in Brazil

With the arrival of Asian soybean rust (caused by Phakopsora pachyrhizi) in the Western Hemisphere in 2001, field research to optimize chemical control of this important yield-limiting disease has proliferated. We present a meta-analytical synthesis of the results of 71 uniform fungicide trials containing 930 entries (specific fungicidal treatments) conducted in Brazil from 2003/2004 to 2006/2007. Our objectives were to: (1) quantify the overall efficacy of fungicidal treatments in reducing disease and yield loss; (2) determine to what extent fungicide efficacy depends on overall disease pressure, the number of spray applications, and the amount of disease present at the time of the first application; and (3) test for differences in efficacy among fungicide classes and specific active ingredients. Weighted median response ratios for disease severity (R"S) and yield (R"Y) were 0.413 and 1.439, respectively, indicating that, on average, fungicide treatments reduced disease by 58.7% (range: -38.9-100%) and increased yield by 43.9% (range: -21.8-458%). Response ratios were dependent on disease pressure (expressed as disease severity of the untreated check), with the greatest reduction in rust severity (i.e., lowest R"S values) observed for low disease pressure and the best yield response (i.e., highest R"Y values) observed for high disease pressure. In trials where both one and two application schedules were included, R"S and R"Y were better for entries receiving two applications than for one application. In ~65% of entries across all trials, disease was present at the time of the first application, albeit at low levels (median=0.20% severity). Only disease severities of up to 0.05% at the time of the first application could be tolerated without affecting R"S negatively, whereas presence of any disease at the first application had a negative effect on R"Y, even when disease pressure was low. In general, triazole fungicides applied alone performed better than strobilurins alone, but there was a wide range in efficacy among individual triazoles, with prothioconazole and tebuconazole performing best and fluquinconazole and difenoconazole being least effective. Combinations of strobilurins with triazoles (especially those containing cyproconazole) improved disease and yield loss control compared with either class alone. In contrast, combinations of triazoles with a benzimidazole fungicide did not improve R"S or R"Y compared with triazoles alone. Across fungicides, R"Y and R"S were correlated negatively (r=-0.6296, P=0.0017), indicating that treatments with better disease control also had higher yields.