Variability of feed ingredient quality is a challenge for precise nutrition. A commercial dataset of NIRS predictions (AB Vista) containing different sources of corn, wheat, and soybean meal (SBM) was used to analyze the impact of nutrient and antinutrient levels on feed formulation. Data from the United States (USA) and Brazil (BR), obtained from 2019 to 2024, included 54,801 corn (NUSA = 17,390 and NBR = 37,411), 1,430 wheat (NUSA = 564 and NBR = 866), and 60,196 SBM samples (NUSA = 16,170 and NBR = 44,026). All analytical variables for each ingredient were used to partition the data within each country into three clusters of similar nutrient profiles using k-means. One-way ANOVA was used to compare nutrient and antinutrient (Phytic P; PP and Total Non- Starch Polysaccharide; tNSP) content among clusters within each country and ingredient. Mean separation was done using Tukey’s HSD. Stepwise differences (P < 0.05) were observed for tNSP and PP clusters within each country and ingredient, with few exceptions. Two clusters for BR wheat, comprising 81.80% of the samples, had the same PP content (0.23%), lower (P < 0.05) than the highest PP cluster (0.27%). Similarly, two clusters for USA SBM, grouping 92.5% of the samples, had the same mean PP (0.41%), lower (P < 0.05) than the highest PP cluster (0.43%). Diets, following Ross 708 recommendations for three feeding phases were formulated. Nutrient values from clusters with the highest PP and tNSP per country were compared with country means. Formulating with the highest PP clusters, representing 8.53 and 49.95% of corn, and 25.69 and 7.50% of SBM samples (BR and USA, respectively), increased feed PP by 0.02% pts to 0.26% for BR and 0.01% pt to 0.26% for USA. For the high tNSP clusters, congregating 37.02 and 15.70% of corn and 29.65 and 7.50% of SBM samples, increased mean tNSP content 1.17% pts to 10.26% for BR and 2.12% pts to 10.76% for USA. In conclusion, clustering ingredients based on nutrient profile can help assess patterns in nutrient variability. Identifying feedstuff cluster characteristics can aid in making better formulation decisions. For example, ensuring accurate PP and tNSP content estimation to decide enzyme inclusion and potentially minimizing gut health issues.

Marshall, C., V. Blanvillain and  E. Oviedo-Rondón. 2025. Exploration of the antinutrient variability of corn, wheat, and soybean meal.2025 International Poultry Scientific Forum, Abstract M106.