The study was carried out in two separate locations. Experiment 1 was conducted at the Central Queensland Innovation and Research Precinct (CQIRP) in Rockhampton, Queensland, Australia, coordinates 150°30′ E, 23°19′ S, elevation 40 m. Experiment 2 was conducted at Wilburra Downs, a commercial station in Richmond, Queensland, Australia, coordinates 143°14’ E, 20°47’ S, elevation 220 m. Animal care for Experiment 1 followed the approved research protocol of the CQUniversity Animal Ethics Committee (approval number 24410). In Experiment 2, all procedures adhered to the Australian code for the care and use of animals for scientific purposes [
7].
2.1. Experiment 1
The study had 30 Droughtmaster yearling steers with an initial liveweight (LW) of 291 ± 5.3 kg. The animals had unrestricted access to two group pens (GPs) interconnected by an outside area (
Figure 1). Each GP had 180 m
2 of covered area where feed and water were available via ten Smart Feed Pro
® (Smart Feed Pro, C-Lock Inc., Rapid City, SD, USA). The additional uncovered area had approximately 62 m
2. Cleaning was undertaken three times per week using a high-pressure hose inside, and only faeces were removed with shovels in the uncovered area.
The animals were monitored over the seven days in which the evaluations were performed. After two days, it was identified that two animals were not accessing the Smart Feed Pro
® to consume hay and drink water, and they were removed from the trial. Throughout the 7-day period, the animals were fed
ad libitum a low-quality Rhodes grass (
Chloris gayana) hay (92% DM, 5.1% CP, 71% NDF, 45% ADF) offered in seven Smart Feed Pro
® feeders. The water treatments used in this experiment consisted of one of three treatments: Treatment 1 (T1) − solution of water + uPRO ORANGE
® + Agolin Ruminant L
®; Treatment 2 (T2) − water + uPRO ORANGE
®; and Treatment 3 (T3) − water. The water supplied came from the same line for all three treatments. The proportions and chemical composition of each treatment is presented in
Table 1.
Treatments were administered via the uDOSE® automatic system from DIT AgTech, delivered into water troughs within each Smart Feed Pro®. The water troughs consisted of 100 L PVC boxes placed inside the smart feeders and were refilled three times a day (8:00 h, 13:00 h, and 16:00 h). Three uDOSE® devices controlled a dose rate of 10 mL of solution/L of fresh water, consistently meeting the target dose rate and achieving at least 99% dose delivery when administering nutrients into the pipelines. Water consumption remained ad libitum.
Dry matter intake and water intake were determined by measuring the weight difference when each animal accessed the Smart Feed Pro®. RFID tags on the right ear of each animal enabled the Smart Feed Pro® to control and quantify DMI and water intake. Additionally, despite not specifically focusing on animal liveweight gain, the initial and final liveweights (LW) were measured using a fixed scale Gallagher model TWR-1® (Gallagher Group Limited, Hamilton, New Zealand) at the beginning and end of the one-week experimental period.
Throughout the experimental period, weather data (temperature [max and min], humidity [max and min] and wind speed), was recorded daily using a portable weather station (Ecowitt model WS2910CA).
2.2. Experiment 2
The experiment used 120 yearling steers, with 30 animals from one of four genetic groups (
i.e., Angus, Brahman, Charolais, and Senepol). Breed identification was based on genotype information. The initial LWs for each breed were Angus 272 ± 9.3 kg, Brahman 274 ± 8.9 kg, Charolais 268 ± 6.4 kg, and Senepol 270 ± 6.3 kg. The animals grazed 886 ha of a mixture of native and introduced grasses and legumes traditionally found in the Central Queensland Region (Mitchell grass (
Astrebla lappacea), Flinders grass (
Iseilema spp.), Buffel grass (
Cenchrus ciliaris), Pigweed (
Portulaca oleracea), and Prickly Acacia (
Vachellia nilotica)), in a paddock divided into two main areas—the square area (SA), with three water troughs, and the corner area (CA), with one water trough (
Figure 2).
In the square area (SA), three rectangular concrete water troughs (60 cm × 300 cm × 60 cm, 1000 L capacity) were used to administer treatments in the water. Treatments included uPRO ORANGE HP
® at a regular dose (RD) of 50 mL/30 L of fresh water, uPRO ORANGE HP
® at a high dose (HD) (starting from 50 mL/30 L and increased by 10 mL weekly until reaching 80 mL/30 L), and just water as a negative control (NC). In the corner area, a water trough, with just water, served as a positive control (PC). Water, flushed weekly or fortnightly, was sourced from a bore to all four troughs. The uDOSE
® automatic system from DIT AgTech ensured precise dosages, with one device per treatment, achieving the target dose rate of at least 78% of the time when dispensing nutrients into the pipelines. The composition and chemical details of the supplemental treatments are provided in
Table 2.
The animals were kept in the area over four experimental periods of 24 days each. They had
ad libitum access to legumes and herbs, which were patchily distributed across the Mitchell grass pastures. The chemical composition of hand plucked samples was performed via the NIR method, and forage availability was calculated as total standing dry matter (TSDM, kg DM/ha) based on CiboLabs (Australian Feedbase Monitor—AFM, CiboLabs [
8]). The latter were undertaken monthly and resulted in the following: September 2023: 90.2% DM, 6.3% CP, 54.6% NDF, 48.4% ADF and 691 kg DM/ha as TSDM (measured 98% total paddock area). October 2023: 91.8% DM, 6.2% CP, 60.2% NDF, 49.5% ADF and 549 kg DM/ha as TSDM (measured 91% total paddock area). November 2023: 92.0% DM, 6.6% CP, 47.8% NDF, 49.4% ADF and 828 kg DM/ha as TSDM (measured 61% total paddock area). December 2023: 91.7% DM, 8.9% CP, 21.0% NDF, 42.2% ADF and 809 kg DM/ha as TSDM (measured 74% total paddock area).
Adjustments where water was provided were made during the experimental periods, aiming to eliminate any pre-existing influence on animal behaviour. The protocol aimed to identify the animals’ most preferred water trough under different conditions, with consistent free and permanent access granted throughout the experimental periods. During the first experimental period (11 September−5 October 2023), water was provided into water troughs dosed with treatments NC, PC, and just water, with or without weekly flushing. In the next experimental period (6−30 October 2023), all four water troughs provided water for the animals (RD, HD, NC, and PC). The following period (31 October−24 November 2023) the water was provided only in two water troughs in the SA (RD and HD). In the final experimental period (25 November−19 December 2023), water alone was delivered into all four water troughs again (RD, HD, NC, and PC). The approach was developed to simulate the transitional period between the wet season and dry season in a beef cattle extensive grazing system. At all times, water was provided ad libitum.
Dry matter intake was not evaluated in Experiment 2; however, water intake was measured using the uDOSE
® automatic system, factoring in the environmental changes in each trough through Equation (1) by Coimbra et al. [
9] and accounting for evaporation and rainfall effects:
where
Wd is water disappearing (L),
Wflow is water flowing into the water trough (L), E is evaporation (mm),
R is rainfall (mm), and
A is water trough surface area (m
2).
Liveweight changes were measured throughout the experiment using a Walker overweight, WoW (Tru-Test Remote WoW; Tru-Test® by Datamars Australia Pty Ltd., Banyo, Queensland, Australia), installed in the SA, where three water troughs were placed (NC, RD, HD). When the animals accessed SA to drink water, it was achieved by traversing the WoW system.
During all experimental periods, the weather data (temperature [max and min], humidity [max and min], rainfall, wind [direction and speed] and pressure) was recorded using raw data measured by the Bureau of Meteorology [
10] at the closest site, located 18 km from the experiment site, on a daily basis (
Supplementary Materials Table S1).
2.3. Chemical Analyses
The hay [Experiment 1] and forage [Experiment 2]) were sampled to represent the feed available. In Experiment 1, approximately one hand full of hay offered was collected daily, with a 300–500 g bulk sample being collected to represent the entire seven days. In Experiment 2, hand plucked samples were frozen (−18 °C) until further lab analysis. Samples from both trials were dried in an air forced oven 60 °C during 72 h and ground using a 2 mm sieve. The water samples were sampled in the field, in both experiments, using sterile plastic containers and stored frozen at −18 °C.
All forage chemical analyses were conducted using a Near Infra-Red (NIR) Instrument model Perkin Elmer—DA 7250® (Waltham, MS, USA). The output included dry matter (DM), nitrogen as crude protein (CP), neutral detergent fibre (NDF), acid detergent fibre (ADF). For the water analysis, a Multiparameter Photometer with COD for Water and Wastewater model Hanna Instruments HI83399® was used to determine either Nitrogen or phosphorus, pH, electrical conductivity, total dissolved solids, and salinity.
2.5. Statistical Analysis
All data were submitted to previous normality (Kolmogorov–Smirnov) and homoscedasticity tests. When necessary, the raw data transformation was applied to achieve the normality and/or homoscedasticity. The significance was considered by a Tukey Test at 5%.
In Experiment 1, the results (total water intake [L/head], total time drinking [min/head] and water intake ratio [L water consumed/min]) were analysed in a completely randomized design. The model can be expressed as:
where, Y
ij is the observation for the j-th unit in the i-th treatment group, μ is the overall population mean, τ
i is the effect of the i-th treatment group (deviation from the overall mean due to treatment), and ε
ij is the random error term representing individual variability and experimental error.
The individual water intake (L/day) was evaluated by considering a completely randomized design with repeated measures to verify the effects across the experimental days. For this variable, the model applied was:
where, Y
ij is the observation for the i-th individual at the j-th time point, μ is the overall mean. τ
i is the effect of the i-th treatment, β
j is the effect of the j-th time point, and ε
ij is the random error associated with the observation for the i-th individual at the j-th time point.
In addition to the previous analysis, a multivariable analysis was performed, known as a principal component analysis (PCA), in both experiments. This was conducted using individual water intake, dry matter intake (DMI), weather data (air temperature [max and min] and humidity [max and min], water temperature [max and min]), and average liveweight. The data were standardized before the analysis.
The statistical analysis developed for Experiment 2 to evaluate water intake (L/day) on a mob level for each experimental period, evaluating treatments and between periods, was developed as a completely randomized design. Some tests were evaluated using THI and rainfall as covariates in the model; however, both variables were not significant in this model. The model applied for the analysis was the same from Experiment 1. A similar multivariable analysis was performed using water intake in a herd level, the TDS of water (WaterTDS), pH of water (WaterpH), the temperature humidity index (THI) and weather data (rainfall [Rain], relative humidity [RH], air temperature [TMin, TMax], and wind speed [WindMax]) from Experiment 2. The data were standardized before the analysis.
All statistical analyses were performed using software R (version 4.3.2) [
13].