Performance thresholds, processing requirements, and circular bioeconomy potential of agro-industrial by-products in African broiler nutrition: A systematic review and meta-analysis.

Feed ingredients account for most broiler production costs in Africa, yet large volumes of agro-industrial residues remain underutilized. The practical question is not whether these by-products can replace maize or soybean meal, but the inclusion levels, processing conditions, and evidence quality required for safe, effective use. This systematic review synthesised the available controlled-feeding evidence and pooled it by random-effects meta-analysis. A structured search of Web of Science and PubMed identified 273 unique records published between 1976 and 2026, from which 13 primary controlled broiler feeding studies with quantitative performance data were selected. Mean differences in feed conversion ratio (FCR) between each by-product diet and its within-study control were pooled under a random-effects model with robust variance estimation clustered on study, and regressed on dietary inclusion level; a 33-treatment castor bean cake database was analyzed to identify minimum effective ricin-removal conditions; and study quality was appraised with a SYRCLE-adapted instrument before pooling. Thirty-two contrasts from seven studies entered the primary analysis. Each percentage point of by-product inclusion raised FCR by 0.0116 units (95% CI +0.0007 to +0.0224; P = 0.041). Feedstuff functional class accounted for a substantial part of the heterogeneity (Q between classes = 41.1, df = 4, P < 0.001). Pooled across the fruit and oil processing residues the effect on FCR was -0.006 units (95% CI -0.096 to +0.085), indicating no detectable class-level effect despite the 11-12% FCR reduction reported for orange peel meal. For the bulk energy substitutes the pooled mean difference in FCR was +0.200 units at inclusions of 20% of the diet or less (95% CI -0.025 to +0.425) and +0.453 above 20% (+0.306 to +0.600), locating the provisional threshold at 20%. Processing pigeon pea seed meal at 26% inclusion improved average daily gain by 33-38% and FCR by 34-36% relative to raw inclusion, exceeding any inclusion-level effect. Only autoclaving at 15 psi for 60 min and calcium hydroxide at 40 g/kg achieved complete ricin destruction. Within a small, geographically concentrated evidence base with no eligible East African trials, these findings tentatively support provisional inclusion thresholds, indicate functional roles for fruit residues at low doses, and establish minimum processing standards, but require independent replication. Priority needs are standardized multi-site trials with full variance reporting, East African trials, and simultaneous cost-performance analysis.