Inferring effects of barn emissions, housing conditions and genetics on specific dermatitis digitalis diagnoses in dairy cows

The aim of the present study was to infer effects of housing systems, cow phenotypes and genomics on specific stages of the claw disorder dermatitis digitalis (DD) of dairy cows kept in compost bedded pack barns (CBPB) and conventional cubicle barns (CCB) applying structural equation models (SEM). Housing system characterisations, herd hygiene status determination and greenhouse gas emission recordings considered 11 farms, whereas 6 farms represented the CBPB system, 2 farms represented the CCB system, and 3 farms “mixed farming” system with CBPB for sub-herd A and CCB for sub-herd B. In these 11 farms, 1,047 Holstein-Friesian and Fleckvieh-Simmental cows (1,611 observations) were phenotyped for the DD stages DD sick, DD acute and DD chronic. Cows from 4 further farms without housing and greenhouse gas emission data were considered for DD phenotyping and SNP genotyping, implying the availability of 2,980 DD observations from 1,710 cows for genomic studies of DD traits. In a first step, generalized linear mixed models were applied to identify the most relevant housing characteristics on DD sick, DD acute and DD chronic. Least-squares-means for infection probabilities were generally smaller in CBPB than in CCB. With regard to compost, barn air and barn emission characteristics in CBPB, a bedding temperature in the range < 28°C, a C:N ratio in the bedding material > 21, a pH-value in the bedding material > 8.8, small NH₃ concentrations (< 0.55) in the barn air, as well as small as moderate air humidity, were associated with the highest DD health status. The single-step GWAS indicated similar Manhattan plots for DD sick and DD acute, and respective shared potential candidate genes based on gene annotations from the Bos taurus ARS1.2 genome assembly. Three same SNPs were significantly associated (according to normative significance threshold) with DD acute and DD sick, but no overlaps in this regard were identified for other DD stages. Strong association signals in the Manhattan plots according to strict pBF were identified for DD chronic including three further SNPs, and for DD acute including the SNP Hapmap47993-BTA-56668 (HAP) located on BTA 23. These SNPs together with latent variables for the cow DD individuality (DD indiv, including phenotypes and estimated breeding values for DD stages), cow productivity before and after a DD diagnosis, and the relevant barn and housing characteristics (as identified via mixed model applications), were simultaneously considered in SEM for DD sick, DD acute and DD chronic. Housing and barn characteristics played a predominant role with regard to infection risks for DD sick and DD acute. In contrast for DD chronic, path coefficients on DD indiv were quite large for DD chronic EBVs, as well as for single SNP effects.