LAUSR

Abstract Limited sample sizes imply parametric assumptions could be violated, even if traits have been reported to fulfil parametric assumptions. Parametric studies have addressed a non-significant influence of CSN1S1 genes on Murciano-Granadina milk yield, fat, protein and dry extract. We used non-parametric categorical tests to find alternative statistical methods to analyse the power to explain the variability found in the population regarding milk yield and its components. We analysed 2090 records for milk yield, and its components from 710 Murciano-Granadina CSN1S1-genotyped goats. Categorical regression equations were issued to predict which and at what level these factors may determine milk yield (kg), fat (kg), protein (kg) and dry extract (kg). All environmental (farm and parturition year) and animal-inherent factors (genotype, birth type and age) resulted statistically significant (p < .05) except for birth season and month. CSN1S1 genotype was highly statistically significant and explained from 8.3% to 9.2% of protein and fat content variability, resembling the values for highly selected French breeds. Seasonal peaks and lows resembled other breeds’. Heterozygote advantage of certain combinations of E allele with those alleles strongly or weakly influencing milk components and yield such as A, B, B2, F and homozygote BB genotype reported the highest statistically significant effects on milk components and yield. Our results suggest that non-parametric tests may report contextually valid results when having a large sample size is not possible. Selecting for certain CSN1S1 genotypes may promote the efficient production of better-quality milk in greater amounts, improving the international competitiveness and profitability of local breeds. Highlights Non-parametric tests are crucial if normality and heteroskedasticity analyses fail. Murciano-Granadina milk traits compared with highly selected international breeds’. E allele combinations and BB reported highest effects on milk components and yield.