320 Can We define production environments more effectively by combining climate and management data?

Abstract The U.S. sheep industry operates under diverse environments, with differences in climatic and management conditions. A consequence may be expression of genotype-by-environment interactions (G×E), necessitating their incorporation into breeding strategies. Traditional methods for defining production environments may fail to capture the complexity of real-world management and climatic conditions, risking a loss in reliability of genetic evaluations. This study integrated climate and management data to define eco-management clusters, potentially offering a more accurate representation of production environments for G×E analyses in U.S. sheep. Data were collected from 97 flocks representing 5 sheep breeds: Katahdin (50 flocks), Polypay (21 flocks), Suffolk (13 flocks), Rambouillet (8 flocks), and Targhee (5 flocks). Climate data were sourced from NASA POWER based on flock-specific latitude and longitude coordinates and included elevation, seasonal precipitation, and soil moisture. Other meteorological variables (ambient temperature, relative humidity, wind speed, and solar radiation) were combined into a Comprehensive Climate Index and used as another climatic variable. Management data were obtained via a producer survey consisting of 60 questions, covering general husbandry, lambing, feeding, culling practices, and strategies for parasite control and mitigating adverse climatic conditions. Clustering analyses were conducted in R using the climate and management data either separately or together to delineate production environments. Climate clusters were derived using Principal Component Analysis (PCA). Management clusters were generated using Multiple Correspondence Analysis (MCA). Climate and management were combined into eco-management clusters using Factorial Analysis of Mixed Data (FAMD). Additionally, Linear Discriminant Analysis was applied to refine cluster separation. Only the top 5% of variables contributing to differentiation among clusters were retained to enhance interpretability. Clusters were validated using silhouette scores within each clustering method to confirm distinct groupings. Cross-tabulation of cluster memberships revealed a significant overlap between climate-based (PCA), management-based (MCA), and eco-management (FAMD) clusters. Climate-based clusters captured broad environmental differences, while management-based clusters reflected operational diversity. Eco-management clusters provided a more realistic classification, incorporating both factors, as animals thrive in environments shaped by both climatic conditions and management practices. Across all breeds, climate-based clustering explained the largest proportion of total variation (94 to 99%), while eco-management and management-based clusters accounted for less and similar amounts of variation (30 to 55%). Though the degree of improvement varied by breed, eco-management clusters offered a more refined classification of production environments. By improving the characterization of production environments, analyses of G×E may become more reliable improving predictions of genetic merit. Aligning selection for robustness and climatic resilience with flock-specific conditions can lead to more targeted breeding decisions. Future research will evaluate use of eco-management clusters in G×E analyses to identify climate-resilient sheep genotypes.