Multi-environment performance analysis identifies more productive and widely adapted chicken breeds for smallholder farmers

Abstract

Smallholder chicken production system is dominant in tropical developing countries and it contributes significantly to the livelihoods of farmers. Performance of flocks is often too low to meet growing demands for meat and eggs. Unavailability of productive and adaptive breeds that match suitably with the environment is a major limitation. Breeds developed for low- or medium-input systems elsewhere can be evaluated for their performances and introduced at scale to enhance productivity and improve socioeconomic outcomes. Such genetic interventions require conducting multi-environment performance analysis (MEPA) of candidate breeds. However, analytical frameworks and methods are not readily available to identify the best performing breeds considering agroecological differences. Methods used in plant breeding to predict productivity and yield stability of genotypes across environments are theoretically applicable to smallholder livestock systems. In the present study, we adapted two modeling approaches of MEPA to evaluate growth performance of chicken breeds across different agroecologies in Ethiopia. Contrary to the conventional classification system that relies on the types of plants grown and other agronomic variables to delineate agroecological classes, we utilized classes defined by Species Distribution Models (SDMs). SDM defined agroecologies take into account the most relevant environmental predictors that influence suitability of habitats for a livestock species and are ideal for breed performance evaluations. Additive main effects multiplicative interaction model (AMMI) and linear mixed-effects models (LMM) were fitted on three agroecologies and five improved chicken breeds to evaluate growth performance until 180-days-of-age (W180) and yield stability (environmental sensitivity). A total of 21,562 chickens were evaluated in 1,557 smallholder flocks. Our results show that LMM had the best model fit on productivity and yield stability. In both methods of MEPA, Sasso and Kuroiler dual-purpose commercial hybrid chickens were the most productive breeds for W180. Indexes based on LMM consistently identified these two breeds also as the most yield stable. Our results demonstrate that the existing methods of MEPA that are being used in plant breeding are applicable to breed performance comparisons and prediction of genotype by environment interactions (GxE). Moreover, the present study validated that SDM-defined agroecologies are useful for undertaking MEPA in smallholder livestock systems.