When subjected to environments with air temperatures above the thermoneutral zone, homeothermic animals employ several thermoregulatory mechanisms to compensate for heat gain, per equivalent loss, and maintain internal body temperature within narrow limits of variation and achieve thermal equilibrium (McKinley et al., 2017). The present study shows that PT, HT, ST, HQ and IRMax increased (P < 0.05) according to the elevation of the air temperature, suggesting that this fact is due to the significant increase (P < 0.05) in the body core representative temperature (RT) as a function of the heat generation due to the metabolic reactions and the reduction of the thermal gradient between the animal and the environment, which reduces the ability of the animals to dissipate heat in a sensitive way, being this a primary physiological mechanism responsible for the dissipation of body core heat through the bloodstream to the peripheries and subsequently to the environment (Rizzo et al., 2017).
In the three thermal conditions evaluated, it was possible to observe the higher surface temperature (IRMax) in the inferior mucosal region of the animals' eyes, and this fact can be justified by the existence of thinner layers of skin and greater vascularization of this region, when compared to other evaluated regions (Rizzo et al., 2017). The areas around the eye, especially around the posterior border of the eyelid and lacrimal caruncle, have a rich capillary vascularization that responds sensitively to changes in blood flow resulting from changes in the internal temperature of the animal (Stewart et al., 2008).
All the surface temperatures evaluated (PT, HT, ST, HQ and IRMax) correlated significantly (P < 0.05) with the rectal temperature showing that these are sensitive responses and can detect with a high degree of precision the changes in body core temperature of the goats when submitted to thermal stress conditions. In addition, based on the results presented in Fig. 3, it can be observed that the temperature of PT and IRMax presented a higher correlation with RT than the other surface temperatures (HT, ST and HQ), with Pearson correlation coefficient of r = 0.956 and 0.951, respectively.
According to Steck et al. (2011) the replacement of the rectal thermometry method with non-invasive methods for the measurement of thermal stress is potentially beneficial, since it dispense the direct contact of the evaluator with the animal, avoiding external influences, which may mask the results. Based only on the analysis of the correlations we have that the RT of the goats can be measured accurately and remotely through the PT and IRMax temperatures.
The correlation coefficient cannot be used alone to evaluate the relationship between two methods since it does not provide an indication of the equality discrepancy between the data values. Second, the correlation coefficient does not reveal information about the presence of a systematic difference between the methods (van Stralen et al., 2008).
The surface temperatures at the points collected did not show agreement with the RT, however with the exception of IRMax, it can be seen that the biases did not correlate significantly with the averages of the measurements and, despite significant differences between the methods evaluated (ie, PT, HT, ST and HQ did not reproduce values statistically equal to RT), they show the same behavior, keeping the error statistically constant, even with the increase in air temperature.
The RT can be estimated accurately, through the responses PT, HT, ST and HQ, as long as its measurements are increased by the respective biases, and the RT can be represented by any of the following mathematical expressions: RT ≅ PT + 2.03; RT ≅ HT + 1.35; RT ≅ ST + 1.39; or RT ≅ HQ + 1.38, with statistically insignificant mean errors. Also observing the limits of agreement between the RT measurements and the estimated averages based on surface temperatures, such estimates have a high level of precision, and there may be errors in estimates ranging from ± 0.27, ± 0.41, ± 0.47 and ± 0.55 ° C, for PT, HT, ST and HT, respectively, highlighting the expression RT ≅ PT + 2.03, as it has a smaller range of error variation.
Observing the Bland-Altman diagram presented in Fig. 4E that deals with the analysis of agreement between the RT and IRMax measurements, it can be seen that in the air temperatures between the range of 26 and 30°C the error value (difference) between the methods was approximately 0.7°C for both air temperatures, however when the air temperature increased to 34°C, the average error increased to 0.9°C, thus causing the significant correlation between the bias and the averages of the methods' measures, which reflects in the consequent increase in the error between the readings, according to the elevation of the air temperature, being the IRMax therefore, inadequate for the prediction of the RT of goats in conditions of thermal stress.