The trial was conducted at the dairy farm of the Rafaela Experiment Station of the National Institute of Agricultural Technology (INTA) (Santa Fe, Argentina, 31˚12'S, 61˚30'W), starting on January 6th, 2021 and lasting for 12 weeks (2 weeks of pre-experimental period, 1 week of habituation to lipids and 9 weeks of data collection). The experimental protocol was evaluated and approved by the Institutional Committee for the Care and Use of Experimental Animals of the Santa Fe Regional Center of INTA (CICUAE-CERSAN).

Thirty Holstein cows were distributed in 15 blocks by parity (2.0 ± 1.1), days in milk (182 ± 80) and milk production (29.4 ± 5.7 kg∙day⁻¹) at the beginning of the trial and randomly assigned within each block to the following treatments (diets): SPF: supplementation with protected fat or WPF: without supplementation with protected fat. All the cows were kept in a dry-lot (with access to shade and free access to water) where they were given after a.m. milking a partial mixed ration (PMR) ad libitum (26.0% corn silage, 33.2% alfalfa silage, 8.5% ground corn, 18.1% soybean meal, 5.2% soybean expeller and 9.0% alfalfa hay) while in the milking parlor they received differential supplementation depending on the treatment. The SPF diet contained 4.0 kg∙day⁻¹ pelleted concentrate + 0.6 kg∙day⁻¹ ground corn grain + 0.7 kg∙day⁻¹ of protected fat, distributed individually in equal parts in each milking shift, while the WPF diet was similar to that offered to SPF group, but the protected fat was replaced isoenergetically by ground corn grain (equivalence: 1 kg DM fat = 2 kg DM corn). The fat supplement (96% DM, 84.2% EE, 15.2% NFC, 0.6% ash) contained animal and vegetable fats and microencapsulation was used in the manufacturing process. The composition of fatty acids (FA) was 32.0% palmitic, 33.4% stearic, 6.9% oleic, 19.4% linoleic, 3.0% linolenic and 5.3% others. Before the start of the trial, the degree of protection of the fat supplement was determined, for which the impact of adding it on ruminal fermentation was evaluated through an in vitro digestion system [14].

During the pre-experimental period, the cows received the WPF diet. Milk production and composition during this period were used as a covariate. All cows were equipped with neck transponders for automatic recording of daily milk production on an individual basis (ALPRO version 6.60/DeLaval, Tumba, Sweden).

Ambient temperature (AT) and relative humidity (RH) measurements were recorded daily at three different times (0900 AM, 0300 PM and 0900 PM) by the Agrometeorological Station of Rafaela Experiment Station. The temperature-humidity index (THI) was used to monitor environmental stress conditions. The following equation was used for THI calculation [15]:

THI = (1.8 × Tdb + 32) − (0.55 − 0.0055 × RH) × (1.8 × Tdb − 26),

where Tdb is dry bulb temperature.

The results referring to milk production and composition, BW, BCS, and plasma metabolites and hormones were analyzed according to a randomized complete block design with repeated observations in time adjusted by covariate (α = 0.05). The following model was used:

Y i j k l = μ + T i + B j + A ( B ) k ( j ) + W l + ( T × W ) i l + C o v + E i j k l ,

where:

Y_ijkl = dependent variable, μ = general mean, T_i = treatment effect, B_j = block effect, A(B)_k(j) = random effect of animal nested to block, W_l = sampling week effect, (T × W)_il = effect of treatment interaction × sampling week, Cov = covariate and E_ijkl = residual error.

The intake and feed efficiency data were analyzed using a model with a classification criteria (treatment):

Y i j k = μ + T i + B j + A ( B ) k ( j ) + E i j k ,

where:

Y_ijk = dependent variable, μ = general mean, T_i = treatment effect, B_j = block effect, A(B)_k(j) = random effect of animal nested to block and E_ijk = residual error.

All statistical analyzes were performed using the MIXED procedure of the SAS statistical package (2010). Trend was considered 0.05 < p < 0.10.

The average values of the chemical composition of the PMR used in the trial are presented in Table 1. The chemical composition of the pelleted concentrate was 92.2% ± 0.8% DM, 13.2% ± 0.6% CP, 28.2% ± 2.4% NDF, 14.4% ± 1.3% ADF, 3.4% ± 0.3% EE, 13.0% ± 0.6% ash and 71.5% ± 1.4% IVDMD.

The DM content of PMR was within the range commonly used in high-producing dairy herds (40% to 60% DM; [31]). The contents of NDF and energy of the PMR were within the ranges recommended by the NRC (2001) [22] for cows in mid-lactation, that is, 33% - 35% NDF and 1.55 - 1.60 Mcal∙NEL∙kg∙DM⁻¹, while the protein content (20.2%) was above the recommended range (16% - 17% CP; [22]). However, the final protein concentration of the experimental diets (18%) was slightly higher than the indicated for cows in mid-lactation.

PMR and total DMI were similar (p > 0.05) between treatments, while concentrate DMI was significantly higher in the control group (Table 2). This latter result is because the cows in the control group received +0.8 kg∙day⁻¹ of concentrate due to the type of trial design (isoenergetic concentrates).

¹Values expressed as the average ± standard deviation. DM = dry matter; IVDMD = in vitro dry matter digestibility; CP = crude protein; NDF = neutral detergent fiber; ADF = acid detergent fiber; ADL = acid detergent lignin, EE = ether extract.

¹Values expressed as least square means (LSMeans) and the standard error of the LSMeans (SEM). ²SPF concentrate: 4.0 kg∙day⁻¹ pelleted concentrate + 0.6 kg∙day⁻¹ ground corn grain + 0.7 kg∙day⁻¹ protected fat; WPF concentrate: 4.0 kg∙day⁻¹ pelleted concentrate + 2.1 kg∙day⁻¹ ground corn grain. ³Estimated NEL values for SPF concentrate, WPF concentrate and PMR: 2.01, 1.76 and 1.55 Mcal∙kg∙DM⁻¹, respectively. ⁴Feed conversion efficiency.

Decreases in concentrate DMI when using supplemental fats in rations have been associated with a lower rate of intake and size of each meal [32], which would be a relevant problem in dairies where the concentrate is supplied for a limited time in the milking parlor, as in this study. Australian researchers [33] included increasing doses from 0% to 40% of free FA (as feed) in the concentrate fed to grazing dairy cows twice daily at each milking, observing decreases in concentrate intake when the fat concentration was higher than 22%. They concluded that the negative effects observed were a consequence of the concentration rather than the amount of fat supplied. In the present study, in agreement with that reported by [33], the inclusion of protected fat in the experimental concentrate was 14% and did not affect its palatability.

The inclusion of fat supplements in the diet is generally associated with a reduction in DM intake in thermal neutral cows [34], but this effect is commonly not observed in heat-stressed cows [35]. Coincidentally, in this study, the total DM intake was similar between treatments (Table 2).

The estimated total NEL intake was similar between treatments (Table 2), in agreement with [32], who reported that the inclusion of fat supplements on TMR-based diets would have a null or slightly positive effect (5% - 6%) on energy intake in thermal neutral cows.

In the present study, a possible negative effect of the fat supplement on DM and NEL intake could not be detected.

Cows receiving by-pass fat produced significantly more than 4% FCM and ECM per kilogram of total DMI and tended to produce more 4% FCM and ECM per Mcal of NEL intake (Table 2). Similar results were obtained in previous studies [11] [13] [36] and are in agreement with [37] suggestion that increasing diet energy density by supplementation with non-fermentative nutrients might improve the conversion efficiency of feed to milk in heat-stressed cows.

The energy conversion efficiency (ECM Mcal NEL⁻¹) was 0.030 higher in the cows of the SPF group. This difference was above 0.012, which is the value of the potential improvement in energy use efficiency when fat replaces dietary carbohydrate and is added to 3% of dietary DM [11], as was observed in the present study.

The existing information on the productive response of dairy cows supplemented with fat under heat stress conditions is limited (reviewed by [35]). An analysis of the results of the reviewed studies (n = 8) indicated a positive average effect of protected fat on milk production (+1.16 kg∙day⁻¹, p < 0.01), FCM (+1.22 kg∙day⁻¹, p = 0.03) and fat (+0.08 kg∙day⁻¹, p = 0.01) for average intake levels of 0.57 ± 0.21 kg∙day⁻¹, with no significant effects on protein content (+0.03 g∙100g⁻¹, p = 0.47) and production (+0.01 kg∙day⁻¹, p = 0.62). In addition, a trend to a higher fat content was observed in the supplemented cows (+0.14 g∙100g⁻¹, p = 0.08).

In this study, milk production was similar between treatments, while ECM production tended (p = 0.06) to be higher (+1.0 kg∙day⁻¹) in cows supplemented with protected fat (Table 3).

The inclusion of protected fat in the rations for dairy cows would increase the efficiency of energy use for milk production due to a decrease in energy losses in the form of methane, direct use of long-chain FA for milk fat secretion and a higher efficiency of ATP generation from long-chain FA rather than acetate [38] [39]. In this sense, it has been postulated that the contribution of 8% of the absorbed energy requirement in the form of protected lipids predisposes to achieve a better global productive response of the dairy cow [40] [41]. The total ME requirement calculated according to [22] for a cow of 653 kg of body weight, with a body weight gain of 0.526 kg∙day⁻¹, producing 26.6 kg of milk with 4.04% fat, 3.32% of protein and 5.04% of lactose (average values obtained in the SPF group during the experimental period) and under heat stress conditions would be 56 Mcal∙cow⁻¹∙day⁻¹. Taking into account that the intake of protected lipids recorded in the SPF group was 0.672 kg DM∙cow⁻¹∙day⁻¹ and assuming that their energy value is 6.3 Mcal∙kg∙DM⁻¹ [22], the amount of ME contributed by the lipids in SPF group can be calculated as 4.23 Mcal∙cow⁻¹∙day⁻¹, which is equivalent

¹Values are expressed as least square means (LSMeans) and the standard error of the LSMeans (SEM). ²Effects of treatment (Treat), sampling (S) and treatment × sampling interaction (Treat × S). 4% FCM = 4% fat corrected milk; ECM = energy corrected milk.

to 7.55% of the total ME requirement. This latter value was slightly lower than that proposed by [40] and [41] in order to improve the global productive response in dairy cows.

When correcting milk yields for the same fat content (4% FCM), differences between treatments were detected. The supply of protected fat significantly increased the production of 4% FCM (+1.1 kg∙day⁻¹) compared with the control group (Table 3). This increase was very similar to the average response observed in experiments conducted under heat stress conditions (+1.22 kg∙day⁻¹). As cows had similar total NEL intake (Table 2), the increase in 4% FCM yields (Table 3) after fat feeding were not apparently explained by higher energy absorption.

Fat supplementation did not significantly modify milk fat content (Table 3), but it tended (p = 0.06) to increase fat production (+0.06 kg∙day⁻¹). This increase was similar to that observed in studies conducted under heat stress conditions (+0.08 kg∙day⁻¹). The net effect of supplementation with protected lipids on the concentration and production of butterfat depends on the balance between increased uptake of dietary FA by the mammary gland and decreased de novo synthesis [32]. In this study, the increase (p = 0.02) in preformed FA content (>C17:0) recorded in lipid-supplemented cows (42.5 vs. 39.7 g 100 g∙FA⁻¹, for SPF and WPF, respectively) would not have compensated the reduction (p < 0.01) observed in the content of FAs synthesized de novo (C4:0 - C15:1) with respect to the control group (22.0 vs. 24 .6 g 100 g∙AG⁻¹, for SPF and WPF, respectively), since the fat content of the milk was not significantly modified.

Neither the protein concentration nor the amount of protein secreted was affected by the treatments (Table 3), in agreement with the average null effect observed on these parameters in studies carried out under thermal stress conditions. The absence of negative effects on the protein content of milk is an important aspect since this parameter determines the price of milk and affects the speed and quality of coagulation for subsequent transformation into cheese.

Lactose content was not affected by lipid supplementation (Table 3), a result compatible with the similar milk production observed between treatments. A reduction in the availability of fermentable carbohydrates in the rumen could be expected with the isoenergetic replacement of corn grain by lipids in the SPF concentrate. However, the average concentrations of urea in milk were similar between treatments (Table 3), a result consistent with the similar concentrations of urea in plasma observed (Table 4). The average values of urea in milk (Table 3) were very close to the upper limit of the range (21.4 - 34.2 mg∙100ml⁻¹) proposed by [42], which would reflect a high efficiency of N use and a low excretion of N.

Supplementation with protected fat did neither affect any of the parameters associated with the variation of body reserves evaluated, nor did it modify the metabolic profile of the cows (Table 4).

The absence of treatment effect on BW and BC was compatible with the similar levels of βHB observed (Table 4). Apparently, in the cows of the SPF group,

¹Values expressed as least square means (LSMeans) and the standard error of the LSMeans (SEM). ²Effects of treatment (Treat), sampling (S) and treatment × sampling interaction (Treat × S). BW = body weight; BC = body condition; βHB = beta hydroxybutyrate; GH = somatotrophin; IGF-I = somatomedin C.

the energy consumed was not directed to a differential accumulation of body reserves, but was diverted to the production of 4% FCM and ECM (Table 3). Similar results were obtained by [36] when replacing energy in the form of starch (corn grain) with energy in the form of lipids (calcium salts of fatty acids) in the diet of high-producing cows in mid-lactation and under heat stress conditions.

Uremia and glycemia were similar between treatments (Table 4), a result compatible with the similar values of urea and lactose observed in milk (Table 3). Fat supplementation had no consistent effects on circulating glucose concentrations [43]. However, as fat replaced starch in the concentrate, a decrease in the entry of propionic acid to maintain hepatic glucose synthesis could be expected. Fatty acids absorbed from the by-pass fat may have also contributed to maintaining glycemia by reducing total (CO₂) or partial (to NADPH2) glucose oxidation [38].

In cattle, the GH/IGF-I system plays a critical role in the control of lactation [44]. Effects of fat feeding on plasma GH concentration are highly variable, which depends on the sources of fat supply and the physiological state of the animals [45]. In this study, GH concentrations were not affected by fat feeding to dairy cows in mid-lactation, a result consistent with the similar milk yields observed (Table 3).

No differences in IGF-I plasma concentrations were observed (Table 4). The lack of effects on IGF-I concentrations is consistent with the similar GH plasma concentrations (Table 4) and energy intake (Table 2) recorded in both treatments, since these two parameters and hepatic production of IGF-I would be positively correlated [46].

It can be concluded that in the present study, the increase in production observed in the supplemented group could not be explained by changes in the hormonal profile of the animals.

Plasmatic metabolites and hormones such as glucose, non-esterified fatty acids (NEFA), insulin, GH and IGF-I act as dynamic indicators of energy balance [47]. Decreases in plasma glucose, insulin, and IGF-I concentrations and increases in NEFA and GH circulating levels are observed in cows with a negative energy balance [48]. Taken together, the results obtained in this study (similar plasma concentrations of glucose, βHBA, GH and IGF-I) suggest that supplementation with by-pass fat in replacement of corn as an energy source did not affect the energy metabolism of dairy cows in mid-lactation.

The daily average THI during the data collection period was 74.0 (range 80.8 to 68.6), which was 2.0 units above the value of 72 which was determined as being critical for milk production [49] [50].

The average daily respiratory rate and the one recorded at 2:30 p.m. were significantly higher in the cows of the SPF group, without differences at the rest of the times recorded (Table 5).

¹Values expressed as least square means (LSMeans) and the standard error of the LSMeans (SEM). ²Effects of treatment (Treat), sampling (S) and treatment × sampling interaction (Treat × S).

No significant differences in body temperature between treatment groups at any of the recorded times during the experimental period were observed (Table 5). Treatment × sampling interaction was detected for average daily body temperature (Table 5). During the first 39 days of the data collection period, the average body temperature was significantly higher (p < 0.05) or tended to be higher (p < 0.10) in the cows of the SPF group, without differences in the rest of the period.

Most of the experiments reviewed by [35] reported little or no difference in body temperatures and respiratory rates. In fact, only one report, in line with our results, indicated that cows fed additional fat had increased body temperatures and respiratory rates, even though the estimated metabolic heat production was lower compared with cows in the control group [36].

Feeding fat does not appear to be an appropriate nutritional strategy to improve heat balance in heat-stressed cows. To explain these results, [35] suggested that it could be possible that small reductions in a thermal load would be difficult to detect at limited time points, although these minor changes would accumulate over time into a significant improvement. Consequently, the authors recommend measuring body temperatures in heat-stressed cows with a continuous thermometer system, such as the one used in the present study.