Effects of Air Temperature and Relative Humidity on Milk Yield of Holstein Dairy Cattle Raised in Hot-Dry Southeastern Anatolia Region of Türkiye

ABSTRACT

the effect of heat stress on milk production.The critical values at which the milk yield began to decrease due to heat stress in this study slightly deviated from the critical value of 72, which is accepted as the threshold value for the start of heat stress and determined as 77, 54, 64, 69, and 54 for THI a , THI b , THI c , THI d , and THI e , respectively.Based on these values, the loss of milk production of one cow per year was calculated as 98.25, 157.68, 207.36, 164.30, and 190.08 kg when using THI a , THI b , THI c , THI d , and THI e , respectively.This study confirmed that weather stations located away from farms provide useful information for research on heat stress in dairy cows.It can be concluded that THI d , which shows the least deviation from the critical value of 72 (only 3 unit), better reflects the stress condition that animals are exposed to due to temperature and humidity.For this reason, the highest daily air temperature and lowest daily humidity appear to be the most important factors in this investigation to assess heat stress and both variables can be combined into a THI. with a high humidity level reduces the cooling capacity and causes the body temperature to rise (West 1994), resulting in significant milk yield losses (Herbut & Angrecka 2012;Konyves et al. 2017;Gantner et al. 2017;Yazgan 2017).Moreover, heat stress negatively affects performance not only for dairy cattle but also for beef cattle (Yazgan et al. 2013).
The effect of heat stress on test-day milk yield can be described as a function of four variable groups or variables (Ravagnolo et al. 2000).The first of these groups is highest, average, or lowest temperature and humidity values during the 24 hour-period prior to milk yield recording.Second is heat stress measures (e.g.fan, shading, and sprinkler applications), third is duration of current heat stress and finally duration of previous heat stress.There are many methods to quantify heat stress and the simplest of these is the temperaturehumidity index (THI), calculated by the combination of temperature and humidity into one value and defined by several formulas (Thom 1959;Bianca 1962;NRC 1971;Leonard 1985;Mader et al. 2006).Ravagnolo et al. (2000) reported that meteorological data obtained from public weather stations contain useful information for studies on heat stress in dairy cattle, since daily yields are affected by weather conditions and they reflect the effect of weather temperature and humidity.This means that the impact of heat stress on animals can be determined when weather conditions, such as temperature and humidity, prior to the test days are recorded.Another important problem encountered while calculating the effect of heat stress on animals is deciding which values to consider as the maximum, minimum, or average temperature and humidity variables while calculating THI.Because the temperature and humidity values do not remain constant throughout the day, they change constantly, and it may not always be appropriate to only use average values.
A significant amount of cattle milk production is carried out in the Diyarbakir province of Türkiye.Since, however, the province is one of the hottest regions of Türkiye, milk production is adversely affected.In the summer seasons, in particular, temperatures can reach as high as 46 °C (Kallioglu et al. 2015).Accordingly, the average daily maximum air temperature is around 37 °C, which negatively affects milk production due to heat stress.
This study (1) investigates the relationship between milk production and air temperature and RH in Holstein dairy cattle raised in Diyarbakir province of Türkiye by using publicly available weather information and (2) calculates milk yield losses that occur due to heat stress.

Data
The milk production data were obtained from a modern commercial dairy cattle farm located in Diyarbakir.The farm is located at 37°59'03" N latitude and 40°21'37" E longitude, with the altitude of 665 meters.The cattle were kept in an open-system free stall barn, fed ad libitum, had free access to water, and were milked three times a day with their yield recorded by an automatic milking system.Each cow had at least total 270 records to be part of the analysis.Records with milk production <8 kg or >50 kg, daily records of animals during the first four days of lactation and those after 350 th day for extended lactations were eliminated from the data set.There were five parities in lactation records for daily milk yields and only one lactation record (non-repeated observation) for each cow.Weather data included daily maximum, minimum and average temperature, and humidity were obtained from the public weather station located in Diyarbakir that belongs to the Turkish State Meteorological Service authorised by Ministry of Environment, Urbanization and Climate Change of the Republic of Türkiye.While the distance between the weather station and the farm was 15.32 km as a straight line (crow flies), the altitude difference between the farm and the weather station was only 15 m.The formula proposed by Mader et al. (2006) is highly correlated with the panting score.For this reason, the formula given below (Eq. 1) was used for THI calculations in this study.According to this formula, heat stress in dairy cattle begins when the THI value reaches 72, which corresponds to 100% humidity at 22 °C, 50% humidity at 25 °C or 20% humidity at 28 °C.Using combinations of maximum, minimum or average temperature and humidity value with this equation THI a (maximum temperature and humidity), THI b (minimum temperature and humidity), THI c (average temperature and humidity), THI d (maximum temperature and minimum humidity) and THI e (minimum temperature and maximum humidity) were calculated daily.Figure 1 shows all calculated THI variants for each day of the year (averaged over 3 years) for the present data set.Each test-day record was assigned the daily THI a , THI b , THI c , THI d , and THI e values of the previous days and put together with the daily milk production data.Final data comprised 46 438 various parity daily records of milk collected from 2018 through 2020 from 185 healthy Holstein dairy cattle (Tables 1, 2).

Statistical analysis
The following statistical model (equation 2) was used to calculate the least square means of daily milk yield by THI variants.Since there were no repeated observations of any animals in the dataset, an element of the individual effects of animals was not added to the model. Where: Y ijkl : Daily milk yield for parity i, year-month j, days in milk class k and THI variant class l; In order to calculate the milk yield losses, a similar approach was used as reported by Ravagnolo et al. (2000) and formulated for all THI variants as follows (equation 3); Where: All analyses were conducted with the GLM procedure of SAS (2000).

Results and Discussion
The number of observations, mean milk yield, and standard deviation for each month of the three years are shown in   As shown in Figure 2 and Table 4, there were fluctuations in the least square means of the milk yields, ranging from 38 to 77.When the THI a value exceeded 77, the milk yield began to decrease, but increased slightly after 87.In this range, the milk yields decreased from 26.64±0.318kg to 25.33±0.385kg and the difference was 1.31 kg (p<0.05).However, least square means of milk yields began to decrease at the point THI 77 instead of the critical value previously stated 72.

THI b
As indicated in Figure 2 and Table 4, all possible THI b values lies in the range of 33-71 since daily minimum temperature and minimum humidity values were used in its calculation.The lowest milk yield was obtained (24.82±0.432kg) when THI b was equal to 34. Between 34 and 54 THI b values, continuous fluctuations were observed in the milk yields.However, after the 54 THI b value, there was a continuous decrease in milk yields to 67 THI b value.When THI b was 54, the milk yield was 26.95±0.408kg; however, when THI b value reached to 67 the milk yield decreased to 25.19±0.533kg and the difference was 1.76 (p<0.05)due to heat stress.

THI c
As indicated in Figure 2 and Table 4, THI c values were calculated accepting that daily average temperature and humidity values ranged between 35-78.Accordingly, the threshold THI c value at which the milk yield started to decrease continuously was determined as 64, far behind the critical value (THI=72).The milk yield tended to increase despite fluctuating values from the point where the THI c was 35 to 64, but after this point it decreased rapidly and reached minimum at the 78 point.While the THI c value was equal to 64, the milk yield was 27.44±0.510kg.When the THI c value increased to 78, decreased to 25.19±0.486kg and the difference was 2.25 (p<0.05).

THI d
THI d values calculated by considering daily maximum temperature and minimum humidity values were in the range of 43-82 as observed in Figure 2 and Table 4.This range (39 units) was not found to be larger than THI a .However, the threshold THI d value at which the milk yield began to continuously decrease was determined as 69 and was very close to the critical value (THI=72).In other words, THI d had the least deviation from the critical value by 3 units.Although the THI d values fluctuated from 43 to 69, the milk yield tended to remain constant in this range, but after this point, it decreased rapidly and reached a minimum at the 82.When the THI d value reached 69, the milk yield was 27.08±0.398kg.When the THI d value increased to 82, the milk yield decreased to 25.43±0.442kg and the difference was 1.65 (p<0.05).

THI e
As shown in Figure 2 and Table 4, THI e values calculated based on average temperature and humidity values ranged between 31 and 74.
The threshold THI e value at which the milk yield started to continuously decrease was determined as 54, and it was below the critical value (THI=72), similar to THI c and THI d .However, THI e had the greatest deviation from the critical value among all variants by 18 units.Even though the THI e had fluctuating values of 31 to 54, the milk yield tended to increase in this range.After that point, however, it rapidly decreased and reached a minimum at 70 points.While the THI e value was equal to 54, the milk yield was 27.21±0.551kg.
When the THI e value increased to 70, the milk yield decreased to 25.09±0.509kg and the difference was 2.12 (p<0.05).

Milk yield loses for THI variants
THI a values, obtained by using the maximum temperature and maximum humidity, were in the range of 38-91 and the stress zone in the range of 77-87 (Figure 2, Table 4).The difference between the two values is 10 units.If a THI a of 77 was considered to cause heat stress, then the cattle would be under heat stress for more than one third of the year.On average, Diyarbakir had 125 THI a days per year with values ranges from 77 to 87 (Figure 1), and the mean THI a on these days was 83.The difference between the two values is 6 units.This means that a lactating cow during that entire period would be exposed to 750 units (6x125) of THI a over the comfort zone (Equation 3).As a result, cows lose a production equal to 750 units.So, the loss of milk production of one cow per year because of heat stress would be 98.25 kg with a loss of 0.07 kg (98.25/125=0.78and 0.78/10=0.07)per unit of THI a greater than 77.Similarly, the losses of milk yields for one cow per year during heat stress periods were calculated as 157.68, 207.36, 164.30 and 190.08 for THI b , THI c , THI d , and THI e , respectively.In addition, the losses of milk production per unit of THI greater than the threshold were 0.08, 0.09, 0.07, and 0.08 kg for THI b , THI c , THI d , and THI e , respectively (Table 5).As mentioned earlier in Table 3, all R 2 values were around 0.45, indicating that almost half of the yield variation was explained by the model, including weather variables.However, THI d and THI e combinations containing extreme values together (maximum and minimum) have slightly higher coefficients of determination than the others.Ravagnolo et al. (2000) reported that while the amount of moisture in the air was constant, the lowest humidity occurred when the temperature was highest.This is consistent with the findings in this study.While the coefficients of determination values for THI variants were higher than the values reported by Ravagnolo et al. (2000), West et al. (2003), Freitas et al. (2006), they were close to the values reported by Dikmen & Hansen (2009) and Yazgan (2017).
Due to the different combinations of temperature and humidity levels (maximum, minimum, or average) being used when calculating THI variants (Figure 2, Table 4), the highest and lowest values of THI variants are different.For this reason, the highest THI b value obtained was only 71 using the minimum temperature and humidity values.This showed that THI b was insufficient to determine the effect of heat stress under the conditions in which this study was conducted.This indicates that it is not practical to use minimum temperature and minimum humidity variables when calculating THI values in Diyarbakir conditions.
According to the THI formula (Equation 1), heat stress in dairy cattle starts at a THI of 72 and is called the critical value; after this point the milk yield continuously decreases.In this study, deviations from critical values (THI=72) were observed for all THI variants.In comparison to 72, the point at which milk yields began to decline continuously for the THI a , THI b , THI c THId and THI e variants were 77, 54, 64, 69 and 54 respectively (Figure 2).While the deviation value of the THI a variant was greater than the critical value of 72, all other variants (THI c , THI d and THI e ) had values less than the critical level.The least deviation was observed in the THI d variant by 3 units.The reason why other the THI variants deviate more from the critical 72 value than THI d may be due to the temperature and humidity variables used in THI a , THI c and THI e calculations.In other words, combining the variables of maximum temperature and minimum humidity into one THI seems to better reflect the stress conditions to which animals are exposed in Diyarbakir conditions.Another reason for deviations from the critical value may be the distance between the farm and the weather station.
Yazgan (2017) reported the critical THI values where milk yield started to decrease as 68, 76, 80, and 70 for minimum temperature and humidity, average temperature and humidity, maximum temperature and minimum humidity, and minimum temperature and maximum humidity combinations, respectively.These results differ from the values reported in this study.However, deviations from critical values (THI=72) when combinations of mean temperature and humidity (THI c ) and maximum temperature and minimum humidity (THI d ) were used in this study were similar to the values reported by Bouraoui et al. (2002) and Bohmanova et al. (2007).
As shown in Figure 2, for all THI variants, fluctuations were observed during the comfort zone, which corresponds to the range from the starting THI values to dashed vertical lines.THI d showed the minimum deviation from the starting point of heat stress and the minimum fluctuation during the comfort zone when compared with others.Some fluctuations may be caused by the use of fans, shading, and sprinkler equipment.When such equipment is activated, heat stress may appear to be lower at higher temperatures.This also causes the THI to appear not only linear but also of zigzag shape (Figure 2).Fluctuations in all of the THI curves could also be caused by an insufficient number of daily milk yield records with a given THI, by partial confounding with other effects in the model (Equation 2) and by ignoring herd management practices (e.g.change in feeding regimen in some animals) and other conditions (e.g.prolonged exposure of animals to direct sunlight or strong wind when animals were in the paddock).Ravagnolo et al. (2000), reported that the maximum daily air temperature and minimum daily humidity were the most critical variables to quantify heat stress.Similar results were obtained from this study as shown in Figure 2 where THI d seems to be less affected by the factors causing fluctuations and showed a deviation of only 3 units from the heat stress beginning point (THI=72).Moreover, THI d performed better than other THI variants in quantifying the heat stress in this study.Therefore, it can be said that the results obtained from the use of THI d in calculating of milk yield losses are more reliable than the other THI variants.
Considering the THI d , the amount of milk yield loss (0.07 kg) for each unit of THI d increase obtained from this study was similar to that reported by Konyves et al. (2017) and Gantner et al. (2017).It was, however, slightly lower than that reported by Igono et al. (1992) and Ravagnolo et al. (2000), and much lower than that reported by Ingraham (1979), Her et al. (1988), Bouraoui et al. (2002), West et al. (2003), Bohmanova et al. (2007), Zimbelman et al. (2009), Herbut & Angrecka (2012) and Yazgan (2017).Various causes might have contributed to these discrepancies.For instance, the results may be sensitive to distances between the meteorology station and the farm, 15 km in the present study and the distances were different in all studies.In addition, while in some studies (Her et al. 1988;Igono et al. 1992), the weather conditions were measured on the farm, in this study they were obtained away from the farm.Different measures towards alleviation heat stress levels (e.g., fans, shading, and sprinkler application systems) may have been used in all studies.
In addition, the use of other weather variables in some of the studies may be another reason.Apart from these, some researchers' (West et al. 2003) use of temperature and humidity values 2 or 3 days before milk yield may explain the differences in milk yield loss between this study and others.Lastly, to obtain the least squares mean of milk yields, daily yield records were used in this research.Most of the other studies mentioned above, however, used monthly data.
Temperature and humidity values vary during the day, and the characteristics of this variation are different for each region.For example, while the RH level in one region is 70% during the day for 3 hours, it may remain at this level for 14 hours in another region.This is valid for the air temperature and is a determining factor for heat stress on animals.That is, if the minimum humidity level during the day does not change for a long period of time and then suddenly drops, it would not be correct to use the maximum humidity value when calculating the THI value.This also applies to other temperature and humidity variables (maximum or minimum).Findings from this study confirm this.For this reason, this type of research should be carried out when the effect of heat stress on animal production in a region is to be determined.

Conclusion
This study confirmed that weather stations located away from the farms contain information useful for research on heat stress in dairy cows.Using the combination of maximum daily air temperature and minimum daily humidity in the THI formula (THI d ) performed better than other THI variants in quantifying the heat stress in this study.This combination was affected less by other environmental factors, and the results obtained from this combination appear to be more biologically meaningful.As a result, it can be used to quantify heat stress in farms with conditions similar to those in this study.However, the performance of these weather variable combinations can be different in other geographic areas.The distance between a farm and weather station is another important factor for the accurate measurement of the heat stress effect.Therefore, similar studies should be carried out on farms located in different regions.

:
Milk yield loss (kg) in stress zone; THI m : Average THI value of the interval when milk yield starts to decrease and reaches the minimum value; THI cr : Critical THI value at which milk yield starts to decrease, d is the number of days over the critical THI value (calculated from Figure 1); Y 1 : Least squares mean of milk yield at the critical THI value; Y 2 : Least squares mean of milk yield corresponding to THI m ; u : Total THI unit between heat stress periods.

Figure 2
Figure2shows the change of least square means of the milk yields by the values of THI variants.The THI a values obtained by using the maximum temperature and maximum humidity were in the range of 38-91, and this range was the largest of all THI variants (54 units).As shown in Figure2and Table4, there were fluctuations in the least square means of the milk yields, ranging from 38 to 77.When the THI a value exceeded 77, the milk yield began to decrease, but increased slightly after 87.In this range, the milk yields decreased from 26.64±0.318kg to 25.33±0.385kg and the difference was 1.31 kg (p<0.05).However, least square means of milk yields began to decrease at the point THI 77 instead of the critical value previously stated 72.

Figure 2 -
Figure 2-Least square means of milk yields by THI variants.Vertical lines show the critical THI value (72) while the dashed vertical lines show the THI value where milk yield starts to decrease continuously

Table 2 -Means and standard deviations of milk yield and THI on the farm by months between 2018 and 2020 Milk yield (kg) THI a THI b THI c THI d THI e Month n Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
n: Number of observations, SD: Standard deviation

Table 2 .
The milk yield average is the lowest (17.81±5.96kg) in August when THI a , THI b , THI c , THI d and THI e values reach their highest values.The estimated values for the coefficients of determination (R 2 ), sums of squares (SS), and mean square errors (MSE) for the THI variants are provided in Table3.All fixed effects in the model (equation 2) were statistically significant (p<0.05) for all analyses.As shown in Table3, R 2 , SS and MSE values of THI variants were determined to be very close to each other.The R 2 value of THI e was the highest (0.4599) whereas the R 2 value of THI a was the lowest (0.4544).Furthermore, THI b had the lowest MSE value (43.19).