Application of random regression models to infer the genetic background and phenotypic trajectory of binary conception rate by alterations of temperature × humidity indices

Abstract

Data for conception rate (CR) and somatic cell score (SCS) from 2055, 1344 and 2212 Holstein cows in first parity from the three production systems “indoor”, a continental pasture based system “pasture” and pasture based systems located at the coast “p_coast”, respectively, were used to identify heat stress thresholds, and to study genetic parameters in the course of THI (=temperature × humidity index). The production systems were defined within the federal state of Lower Saxony, Germany, which is characterized by temperate climate conditions. First and second inseminations were considered, resulting in 2959 observations in the production system “indoor”, 2123 observations in the production system “pasture”, and 3249 observations in the production system “p_coast”. For longitudinal data analyses, random regression sire models with threshold methodology for CR were applied. In all production systems, a THI of 65 was identified as a heat stress threshold for CR, because for THI>65, a substantial decline in CR was observed. Decline in CR was most pronounced in high yielding cows kept in the pasture based system. THI measurements from the day of insemination were more relevant compared to more distant THI measurements (recorded either before and after the day of insemination). Heritabilities for the functional traits CR and SCS were as low as expected (h2<0.10) for THI values in the range from THI=35 to THI=65. Heritabilities and additive-genetic variances for both traits increased under heat stress as well as under cold stress conditions. Therefore, results for genetic analyses confirmed a more pronounced genetic differentiation for functional traits in harsher environments. Genetic correlations between CR and SCS were consistently close to zero or negative in the continental pasture-based production system and in the production systems “indoor” and “p_coast” for THI=50 and higher, and indicating the detrimental impact of increasing SCS on CR on the genetic scale. Finally and from a statistical perspective, random regression animal models including threshold methodology can be applied to study genetic effects on binary traits in the course of a temperature x humidity dependent covariate. Nevertheless, some extreme values for genetic parameters at the extreme end of the THI scale might be attributed to a limited number of observations.