Genetic parameters and signatures of selection in two divergent laying hen lines selected for feather pecking behaviour

Abstract

Background: Feather pecking (FP) in laying hens is a well-known and multi-factorial behaviour with a genetic background. In a selection experiment, two lines were developed for 11 generations for high (HFP) and low (LFP) feather pecking, respectively. Starting with the second generation of selection, there was a constant difference in mean number of FP bouts between both lines. We used the data from this experiment to perform a quantitative genetic analysis and to map selection signatures.

Methods: Pedigree and phenotypic data were available for the last six generations of both lines. Univariate quantitative genetic analyses were conducted using mixed linear and generalized mixed linear models assuming a Poisson distribution. Selection signatures were mapped using 33,228 single nucleotide polymorphisms (SNPs) genotyped on 41 HFP and 34 LFP individuals of generation 11. For each SNP, we estimated Wright's fixation index (FST). We tested the null hypothesis that FST is driven purely by genetic drift against the alternative hypothesis that it is driven by genetic drift and selection.

Results: The mixed linear model failed to analyze the LFP data because of the large number of 0s in the observation vector. The Poisson model fitted the data well and revealed a small but continuous genetic trend in both lines. Most of the 17 genome-wide significant SNPs were located on chromosomes 3 and 4. Thirteen clusters with at least two significant SNPs within an interval of 3 Mb maximum were identified. Two clusters were mapped on chromosomes 3, 4, 8 and 19. Of the 17 genome-wide significant SNPs, 12 were located within the identified clusters. This indicates a non-random distribution of significant SNPs and points to the presence of selection sweeps.

Conclusions: Data on FP should be analysed using generalised linear mixed models assuming a Poisson distribution, especially if the number of FP bouts is small and the distribution is heavily peaked at 0. The FST-based approach was suitable to map selection signatures that need to be confirmed by linkage or association mapping.

Fig. 1

Phenotypic trend over 11 generations. The interrupted and continuous lines show the average number of feather pecking bouts per hen during the observation period of 180 min with respective standard errors for the high and low feather pecking line

Fig. 2

Manhattan plots of F_ST-indexes. The top panel shows the F_ST-values of each marker from chromosomes 1 to 8 and the bottom panel for chromosomes 9 to 28. The top threshold value indicates the genome-wide significance level p_{genome wide} <0.05; the middle and bottom threshold values are the nominal significance levels p_nominal ≤5 × 10⁻⁵ and p_nominal ≤5 × 10⁻⁴, respectively

Fig. 3

Manhattan plots of F_ST-indexes in a sliding window of 25 consecutive SNPs

Fig. 4

Selection response in the high (HFP) and low (LFP) feather pecking lines across generations. The mean additive hen effects were estimated with a linear mixed model (top panel) and with a Poisson model (bottom panel). The estimated intercept was added to the additive hen effects