Random regression of Hereford percentage intramuscular fat on geographical coordinates

Authors: DELGADILLO LIBERONA, JOSE - Texas A&M University; LANGDON, JOHN - Texas A&M University; HERRING, ANDY - Texas A&M University; Blackburn, Harvey; SPEIDEL, SCOTT - Colorado State University; SANDERS, STACY - Natureserve; RILEY, DAVID - Texas A&M University

Submitted to: Journal of Animal Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/23/2019
Publication Date: 11/26/2019
Citation: Delgadillo Liberona, J.S., Langdon, J.M., Herring, A.D., Blackburn, H.D., Speidel, S.E., Sanders, S., Riley, D.G. 2019. Random regression of Hereford percentage intramuscular fat on geographical coordinates. Journal of Animal Science. 98(1):skz359. https://doi.org//10.1093/jas/skz359.
DOI: https://doi.org/10.1093/jas/skz359

Interpretive Summary: We have previously shown cattle of the same breed raised in varying geographic regions have different gene frequencies for genes associated with environmental stress. This analysis evaluates how geographic, and thereby environmental differences, impact estimates of heritability of the trait intramuscular fat using the statistical method of random regression. Data from 227,902 animals provided by the American Hereford Association were used. Results suggest that in harsher environments heritability is decreased, which in turn can impact estimation of genetic merit. Furthermore, the differences in heritability underscore the presence of genetic by environmental interactions. Knowledge and quantification of these differences can be used by breed association in their evaluation of an animal's genetic merit with improved accuracy.

Technical Abstract: Accounting for genotype-environment interactions may improve genetic prediction and parameter estimation. The objective was to use random regression (RR) to estimate variances and heritability for intramuscular fat (IMF) across longitude and latitude within continental US. Records from the American Hereford Association (n = 169,440) and pedigree with 227,902 animals were used. Analyses were conducted across the continental US, and with 2 or 4 subdivisions. Animal models, linear and quadratic RR on longitude or latitude (separately) were assessed. With subdivision, analyses employed linear RR unique to regions (quadratic polynomial was not accomplished). Regions were North and South (separated at 40°N), or West and East (separated at 99°W) using longitude or latitude, respectively. Boundaries at 44.46°N 34 and 36.46°N, and 104.55°W and 92.22°W were set to subdivide in 4 regions. Heritability from animal models was 0.19 ± 0.004. Quadratic RR best fit the data based on likelihood ratio tests using longitude or latitude (P < 0.01). Heritability from quadratic RR on latitude ranged from 0.12 to 0.27 (from South to North). From quadratic RR on longitude, heritability ranged from 38 0.17 (central United States) to 0.37 (far West and East longitudes). When RR within 2 regions was modeled, heritability from RR on latitude in the East region was similar to that from analyses without regions (0.09 to 0.32), but the West region was lower (0.14 to 0.27 from South 41 to North). Heritability from 2-region RR on longitude was similar to that from analyses without regions. Heritability for the South region was somewhat lower with a lower range (0.15 to 0.31) than for the North region (0.19 to 0.47). When modeling RR within 4 regions, estimation of only a subset of covariances among RR coefficients was possible (within-region covariances of 45 intercept and linear terms with latitude; those and covariances of all linear RR coefficients with longitude). With 4 regions, heritability was high in low latitudes in the furthest West and high latitudes in the furthest East region, with approximate difference of 0.3 and 0.2 between West and East regions, respectively. Higher heritability (RR on longitude) was estimated in the North region, especially at the furthest east longitudes of the most northern region. Substantial additive genetic variance and heritability differences appear to correspond to different geographical environments as modeled by RR on within-region geographical coordinates.