Genetics of tolerance to heat stress in milk yield of dairy buffaloes assessed by a reaction norm model

The objectives of this study were to assess the effects of heat stress on the milk yield and investigate the presence of genotype × environment interaction (G × E) in Brazilian Murrah buffaloes reared under tropical conditions. With this, 58,070 test-day (TD) records for milk yield from 3,459 first lactations of buffaloes collected between 1987 and 2018 were evaluated. A mixed model considering days in milk (DIM) and temperature–humidity index (THI) was applied to quantify milk yield losses due to heat stress. The most detrimental effect of THI on TD milk yield was observed in the mid-stages of lactation, after lactation peak, in DIM 105–154 and 155–204 days (−0.020 and −0.015 kg/day per THI, respectively). The least-squares means of TD milk yield were used to identify a heat stress threshold using a piecewise linear regression model. A substantial reduction in TD milk yield due to heat stress was observed for THI values above 77.8 (−0.251 kg/day per increase of 1 THI unit). An analysis using a single-trait random regression animal model was carried out to estimate variance components and genetic parameters for TD milk yield over THI and DIM values. Increased additive genetic variance and heritability estimates were observed for extreme THI values (THI = 60 and 80) combined with mid-lactation stages. The lowest genetic correlation (0.50) was observed between TD records at opposite extremes of the THI scale (THI = 60 vs. THI = 80). The genetic trends observed for the regression coefficients related to the general level of production (0.02) and specific ability to respond to heat stress (−0.002) indicated that selection to increase milk yield did not affect the specific ability to respond to heat stress until the present moment. These trends reflect the low genetic correlation between these components (0.05 ± 0.14). In this sense, monitoring trends of genetic components related to response to heat stress is recommended.