Understanding the complex relationship between animal geno- types, phenotypes and climate is a topic of increasing relevance in animal breeding, given the rapidly changing climatic condi- tions. In this work, we used 15,545 first lactation daily records on Italian Holstein dairy cows from ANAFIBJ. Recorded traits were milk production, fat and protein content, as well as climate data such as the Temperature Humidity Index (THI), maximum daily temperature (MaxTemp) and average daily percent humidity (%Hum). In addition, herd characteristics such as geographical coordinates, were available. Moreover, pedigree and genomic (Bovine 50k SNP chip) data were available for each recorded cow. Since THI is highly correlated with MaxTemp (r=0.99), THI was not further considered in the study.The first step of the study consisted of the analysis of the lactation curves for milk, fat and protein through a functional Principal Component Analysis (fPCA). The aim of fPCA is to summarize longitudinal data in a few synthetic variables (the eigenfunctions). For the three traits, three eigenfunctions were retained by the fPCA. The first eigen- function describes the average curve, and the following ones are related to the features of the curve. For instance, for milk pro- duction, the three eigenfunctions describe the production level, the production persistency and the peak production. They explain, respectively: 89.6%, 8% and 2.4% of the total variation for milk. Similar results were observed for protein content (85.2%, 11.7 % and 3.1 %), and fat content (89.2 %, 7.4%, 3.4 %). Eigenfunctions scores were then analysed in relationship with the average cli- mate variables MaxTemp and %hum, with a model that also includes a Herd effect. For instance, climate variables are signif- icant for the milk production level (First eigenfunction of milk production level) Following steps of the study will consist in the joint analysis of climate, production traits and genotypes by com- bining milk production and climate records with the SNP geno- types of dairy AI sires that have lactating daughters across different regions of Italy. This will add a genetic and a geographic component to the analysis of the relationships between pheno- types, climate and genotypes. Pre-corrected milk yields by fixed effect (lactation number, herd year season) where obtained and residuals were used to calculate the SDMilk for each animal. An animal model in a Bayesian framework was applied to estimate h2 with heterogeneous resid- uals according to the different lactations used to calculate SDMilk. Genetic correlations with other fitness and productive traits were then calculated. Finally, the ssGWAS algorithm was used in GWAS analysis to better match animals without phenotype but geno- typed and vice versa. A total of 10,538 animals presented a phe- notypic SDMilk value, and 2499 were genotyped (100,000 SNPs). Only 630 animals had both phenotype and genotype, but ssGWAS was performed in all genotyped animals with reasonable accuracy. The target trait had an h2 of 0.17 (±0.05), showing that SDMilk present a moderate h2 . Genetic correlations showed a positive correlation with other productive traits such as milk, protein, and fat production (r=0.25, on average) and a negative genetic cor- relation with other fitness traits like somatic cell score (0.15), udder health (−0.22), longevity (−0.35) and fertility (−0.25). Last, the GWAS identified 6 significant signals that combined with pathway analysis demonstrated a connection between these traits and biological processes related to the immune and nervous systems.
Integrative factorial methods to explore the relationships between genotypes, phenotypes and climate in Holstein cows
Giuseppe Conte;Roberta Ciampolini;
2023-01-01
Abstract
Understanding the complex relationship between animal geno- types, phenotypes and climate is a topic of increasing relevance in animal breeding, given the rapidly changing climatic condi- tions. In this work, we used 15,545 first lactation daily records on Italian Holstein dairy cows from ANAFIBJ. Recorded traits were milk production, fat and protein content, as well as climate data such as the Temperature Humidity Index (THI), maximum daily temperature (MaxTemp) and average daily percent humidity (%Hum). In addition, herd characteristics such as geographical coordinates, were available. Moreover, pedigree and genomic (Bovine 50k SNP chip) data were available for each recorded cow. Since THI is highly correlated with MaxTemp (r=0.99), THI was not further considered in the study.The first step of the study consisted of the analysis of the lactation curves for milk, fat and protein through a functional Principal Component Analysis (fPCA). The aim of fPCA is to summarize longitudinal data in a few synthetic variables (the eigenfunctions). For the three traits, three eigenfunctions were retained by the fPCA. The first eigen- function describes the average curve, and the following ones are related to the features of the curve. For instance, for milk pro- duction, the three eigenfunctions describe the production level, the production persistency and the peak production. They explain, respectively: 89.6%, 8% and 2.4% of the total variation for milk. Similar results were observed for protein content (85.2%, 11.7 % and 3.1 %), and fat content (89.2 %, 7.4%, 3.4 %). Eigenfunctions scores were then analysed in relationship with the average cli- mate variables MaxTemp and %hum, with a model that also includes a Herd effect. For instance, climate variables are signif- icant for the milk production level (First eigenfunction of milk production level) Following steps of the study will consist in the joint analysis of climate, production traits and genotypes by com- bining milk production and climate records with the SNP geno- types of dairy AI sires that have lactating daughters across different regions of Italy. This will add a genetic and a geographic component to the analysis of the relationships between pheno- types, climate and genotypes. Pre-corrected milk yields by fixed effect (lactation number, herd year season) where obtained and residuals were used to calculate the SDMilk for each animal. An animal model in a Bayesian framework was applied to estimate h2 with heterogeneous resid- uals according to the different lactations used to calculate SDMilk. Genetic correlations with other fitness and productive traits were then calculated. Finally, the ssGWAS algorithm was used in GWAS analysis to better match animals without phenotype but geno- typed and vice versa. A total of 10,538 animals presented a phe- notypic SDMilk value, and 2499 were genotyped (100,000 SNPs). Only 630 animals had both phenotype and genotype, but ssGWAS was performed in all genotyped animals with reasonable accuracy. The target trait had an h2 of 0.17 (±0.05), showing that SDMilk present a moderate h2 . Genetic correlations showed a positive correlation with other productive traits such as milk, protein, and fat production (r=0.25, on average) and a negative genetic cor- relation with other fitness traits like somatic cell score (0.15), udder health (−0.22), longevity (−0.35) and fertility (−0.25). Last, the GWAS identified 6 significant signals that combined with pathway analysis demonstrated a connection between these traits and biological processes related to the immune and nervous systems.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
