Fasil Kebede

Smallholder chicken production is dominant in many tropical developing countries and contributes significantly to sustainable livelihoods. As a low input-low output production system, it has low efficiency to meet growing demands for meat and eggs in these regions. The lack of productive breeds and inadequate understanding of the roles of local adaptation are major factors contributing to poor performance. Knowledge on adaptive phenotypic and genetic variation is essential to design sustainable chicken genetic improvement and development programmes.

In this thesis I aim to address two overarching research questions: What are the environmental drivers of local adaptation, and phenotypic and genetic differentiation in indigenous chickens? How do improved chicken populations that are introduced into smallholder systems respond phenotypically to environmental variations? To answer these questions, I follow a landscape genomics approach and integrate genetic, phenotypic, and environmental information in my study design and statistical analyses.

In the first part of the thesis I investigate phenotypic and genetic differentiation in indigenous chickens. In chapter 2, I perform species distribution models (SDMs) and apply correlative methods to identify environmental predictors associated with habitat suitability and phenotypic differentiation in Ethiopian indigenous chickens. I report that the presence of population differentiation among Ethiopian chickens is probably in response to environmental variation. I use habitat suitability maps produced by SDMs to show that populations went through different environmental selective pressures. Based on the matching between the presence of distinct phenotypes and availability of unique environmental niches, I classify the Ethiopian indigenous chicken populations into three ecotypes.

In chapter 3, I look for candidate genes and regions under positive selection in different agroecologies (lowland, midaltitude, and highland) and environmental gradients (clines in different geographies). I show that phenotypic differentiation in Ethiopian indigenous chickens has a genetic basis. I look at independent results and overlaps between two methods of signatures of selection ( and XP-EHH). I show that Ethiopian chicken populations differentiated the most between gradients but selection pressures leading to adaptive variation are stronger between agroecologies. These results lead to the hypothesis that evolutionary processes other than natural selection, such as gene flow and genetic drift may have contributed to genetic divergence among populations sampled from different gradients.

I perform Redundancy analysis (RDA) and show that SDM-identified environmental predictors and quantitative traits are useful to explain variations in the genome. I demonstrate that RDA can be used as an alternative approach to GWAS in random mating, indigenous livestock populations which have sufficiently interacted with the environment. I indicate that the results from RDA are supported by the outputs from signatures of selection analyses ( and XP-EHH). I demonstrate that signatures of selection analysis with the two methods ( and XP-EHH) can be used complementarily with RDA to shed light on the relationship between genomic, phenotypic, and environmental variation in local adaptation studies in indigenous chickens.

In the second part of the thesis, I evaluate the performance of improved chicken breeds introduced into smallholder systems. In Chapter 4, I apply distribution models to compare performances of improved chicken breeds introduced into smallholder systems. I show that classifying agroecologies based on environmental variables associated with habitat suitability and phenotypic differentiation of a livestock species improves model fit in GxE predictions. I demonstrate that phenotypic distribution models (PDMs) like boosted GAMs and boosted GLMs are valuable tools in animal breeding to integrate environmental and phenotypic information and predict phenotypic values.

Finally, in chapter 5, I utilize the concept of phenotypic plasticity to evaluate yield stability among improved chickens distributed to smallholder farmers. I implement two methods of multi-environment breed performance analysis (MEPA), namely, additive main effects and multiplicative interaction model (AMMI) and linear mixed-effects model (LMM) to identify and recommend more productive and stable breeds for wider dissemination into smallholder systems. I report that improved chicken breeds introduced into different agroecologies significantly vary in growth performance and yield stability probably owing to their different genetic backgrounds.

Together, I demonstrate in this thesis how adaptive phenotypic and genetic variation can be exploited to enhance performance of chickens in smallholder systems.