Meta-analysis to predict sweating and respiration rates for Bos indicus, Bos taurus, and their crossbreds1

The overall objective of this work was to develop empirical equations from a meta-analysis study to be used to implement initial values in a mechanistic heat balance model. The meta-analysis was conducted to 1) develop prediction equations for sweating and respiration rate (SR, g·m^sup -2^·h^sup -1^ and RR, breaths·min^sup -1^, respectively) based on skin and body temperature (T^sub s^ and T^sub b^, °C, respectively) for different breed types: Bos indicus, Bos taurus, and their crossbreds, and 2) evaluate the fit of existing SR equations and the SR and RR equations (from objective 1) against independent data sets. Fourteen studies were collected for the SR analysis, 12 for fitting and 2 for evaluation. The fitted SR equations (Thompson model) for the 3 breeds types were B. indicus, SR = 0.085e^sup 0.22·T^sub s^^; B. taurus, SR = 0.75e^sup 0.15·T^sub s^^ and crossbreds, SR = 0.015e^sup 0.25·T^sub s^^. Twenty-three studies were collected for the RR analysis, 20 for fitting and 3 for evaluation. The fitted RR equations for the 3 breed types were B. indicus, RR = -1,660 + 43.8·T^sub b^; B. taurus, RR = -1,385 + 37·T^sub b^; and crossbreds, RR = -2.226 + 59·T^sub b^. Three SR equations (Maia, McArthur, and Gatenby models) from the literature were evaluated against the Thompson model using the 14 studies. The McArthur model predicted SR within the correct range, but with an increased slope bias because the equation was linear and not the correct shape. The Maia model overpredicted SR for all breed types with the greatest overprediction being for crossbreds. The Gatenby model overpredicted SR for B. taurus (root mean square error of prediction = 506 g·m^sup -2^·h^sup -1^), but was the best predictor for B. indicus. The Thompson model overpredicted SR for B. indicus (root mean square error of prediction ranged from 134 to 265 g·m^sup -2^·h^sup -1^), but was the best predictor for B. taurus and crossbreds. The Thompson model was a good predictor for RR across all breed types. The meta-analysis showed that the Thompson model outperformed previous models for both RR and SR with the exception of the SR of B. indices, which was best predicted by the Gatenby model. [PUBLICATION ABSTRACT]