Meta‐analysis of corm yield responses to chemical and biological approaches for managing taro leaf blight disease

Abstract

Abstract BACKGROUND Taro leaf blight (TLB), caused by Phytophthora colocasiae , inflicts yield losses of up to 90% and annual economic damage exceeding USD 1.4 billion in West Africa. Despite a growing body of literature on chemical and biological disease management, no quantitative synthesis has compared their efficacy for protecting taro corm yield. This meta‐analysis quantified the effects of chemical fungicides and biological control treatments on taro yield. RESULTS Literature searches of Scopus and Google Scholar (December 2024) identified 21 eligible studies published between 1980 and 2022, yielding 168 treatment comparisons (91 chemical, 77 biological) across five countries. A multilevel random‐effects meta‐analysis with cluster‐robust variance estimation (RVE) was applied to account for correlated effect sizes within studies. Chemical treatments produced a large, statistically significant pooled effect (Hedges' g = 1.072, 95% CI: 0.650–1.494, P < 0.001; k = 74 effect sizes from 15 studies). This result was robust across all sensitivity analyses, including outlier removal ( g = 0.983), complete‐case standard deviations ( g = 0.968), and bias adjustment via trim‐and‐fill ( g = 0.653). Multi‐component fungicide formulations outperformed single‐active‐ingredient products, and effect sizes increased modestly over four decades. Biological treatments, based on only six unique studies contributing 73 effect sizes, yielded a non‐significant pooled effect under RVE ( g = 0.508, 95% CI: −0.497 to 1.514, P = 0.249). Extreme heterogeneity was observed ( I 2 = 93.7%), and trim‐and‐fill adjustment further reduced the estimate to near zero ( g = 0.011). Median biological treatment yields were 7.9% below untreated control yields. CONCLUSION Chemical fungicides provide robust, reproducible yield protection against TLB across geographic regions and experimental conditions. The evidence base for biological control of TLB is currently insufficient to support conclusions about efficacy: the available studies are few, the data are heterogeneous, imputation rates for standard deviations are high, and accounting for study‐level clustering eliminates the apparent statistical significance of the pooled estimate. Standardized reporting protocols and targeted trials in major taro‐producing regions are needed before biological or integrated management strategies can be recommended on an evidence‐based basis. © 2026 Society of Chemical Industry.