Inter- and intra-rater reliability of linear scoring in warmblood breeding

Reliable phenotypes are essential for successful animal breeding programmes. Subjective phenotypic assessment may lead to under- or overestimation of heritability estimates and thereby affect the power of genome-based analyses. In Warmblood horses, conformation traits are commonly assessed using linear scoring, but its reliability has not been sufficiently quantified.

In this study, we investigated inter- and intra-rater reliability of linear scoring of conformation traits in German warmblood horses.

Professional equine breeding experts (e.g. breeding judges) completed an online questionnaire in which they scored 17 linear conformation traits in 34 warmblood stallions using side-view images. The participants were asked to repeat the questionnaire after about one month. The images, sourced from the archive of the Hanoverian State Stud Celle, showed stallions born between 1929 and 1993. Reliability was assessed using intraclass correlation coefficients (ICCs).

Mean linear scores were close to zero. Low average absolute deviations indicated limited use of the -3 to ＋3 scale. Overall, inter-rater reliability between the 11 participants was lower than intra-rater reliability. Trait-specific patterns were identified across both analyses: The trait Head achieved moderate to good inter- and intra-rater agreement (inter-rater: Round 1 ICC = 0.51; 95% Confidence Interval (CI) [0.38, 0.65], Round 2 ICC = 0.50; 95% CI [0.37, 0.65], intra-rater: mean ICC = 0.71 ± 0.23), whereas traits such as Shoulder Length and Carpal Joints showed ICCs close to zero and CIs including zero.

Our findings imply the possible need for improved standardisation, training, or alternative phenotyping approaches to enhance reliability in equine conformation assessment.