Urinary Creatinine as an Indicator of Water Intake in Sheep and Goats Sustainably Farmed in Tropical Climates

Abstract

Mathematical models are valuable tools for predicting water intake in small ruminants, enhancing water use efficiency, reducing environmental pollution, alleviating competition for water with human consumption, and improving productive performance, ultimately leading to increased revenues and promoting sustainability. This study aims to evaluate creatinine as a metabolic marker for estimating water intake in sheep and goats and to develop predictive models for tropical conditions. Five Santa Inês crossbred sheep and five Boer crossbred goats were used in a replicated 5 × 5 Latin square design. Treatments consisted of titanium dioxide supplementation at 1.0, 1.75, 2.5, 3.25, and 4.0 g/day. A species effect was observed on dry matter intake. Significant correlations were identified between water intake, urinary volume, body weight, metabolic weight, and creatinine concentration. Negative correlations were observed between water intake and both dry matter intake and metabolic measures. Five mathematical models were developed to predict water intake, all of which demonstrated good predictive capacity. Among them, the equation Ŷ_H2Og/kgBW = 164.72 − 6.60 × MBW + 0.025 × Creat (mg/L) proved most reliable. This model enables accurate estimation of water intake in sheep and goats, supporting more efficient water management and sustainability in tropical regions where water resources are limited.

1. Introduction

The continued increase in Earth’s average annual temperature, a phenomenon known as global warming, is one of today’s major environmental concerns [1]. Since the early 1980s, global temperatures have been rising by approximately 0.18 °C per decade. Projections from the Intergovernmental Panel on Climate Change (IPCC) indicate that, by 2100, this increase could reach between 3.7 °C and 4.8 °C [2]. This diagnosis could lead to reduced global water availability and, consequently, water scarcity for animals in production systems.

Global climate change poses a concrete threat to livestock production, as it raises temperatures and reduces water availability. Its effects are likely to persist in the coming decades, also impacting the physiological, immunological, and behavioral parameters of animals.

The indirect consequences of climate change impact the storage capacity of aquifers, as increased reliance on groundwater for livestock supply is anticipated [3,4]. Given the high-water demand for sheep and goat production, it is necessary to adopt strategies that maximize the use of available water [5].

In this sense, water management is vital in arid and semiarid regions, where small ruminant livestock represents the primary source of income for many families and natural resources are limited [6,7]. In these environments, water and food scarcity, exacerbated by climate change, pose a growing challenge to maintaining animal productivity and preserving both environmental and socioeconomic stability [8].

In this context, accurately estimating water intake is a crucial tool for optimizing the efficient use of this resource and promoting sustainability in small ruminant farming. Ensuring adequate water intake can enhance animal productivity, leading to improved economic returns, while simultaneously reducing environmental pollution and minimizing competition for water with human populations [9,10].

The water intake in small ruminants ranges from 5% to 25% of body weight. It is influenced by factors such as water quality and temperature, feed type and quantity, diet composition, production system, body weight, breed, sex, physiological stage, age, shade availability, ambient temperature, and relative humidity [11,12,13].

In ruminants, hydration is primarily achieved through voluntary water intake, water from feed, and metabolic water generated during oxidative metabolic processes [14]. This resource is essential for homeostasis and for the performance of physiological functions, such as body temperature regulation, digestion, metabolism, nutrient transport, and excretion [15].

Therefore, indirect methods for assessing water consumption, such as urinary creatinine, offer advantages over direct measurement, especially in extensive systems, where data collection can be limited [16]. Creatinine is a product of the irreversible degradation of creatine phosphate in skeletal muscle, the production rate of which is proportional to muscle mass and relatively constant, regardless of diet or water intake [17,18].

Studies indicate that daily creatinine excretion, expressed in milligrams per kilogram of body weight, exhibits low variation between individuals, making it a reliable endogenous marker for estimating urinary volume [19,20]. This stability was observed even under dietary variations, confirming its potential as a tool for estimating water intake in a practical and non-invasive manner [21].

In this study, we hypothesized that creatinine concentration can serve as a reliable and stable indirect marker for estimating urinary volume and water intake, eliminating the need for total urine collection or direct measurement of water intake in sheep and goats. Furthermore, this study hypothesizes that mathematical models can be developed to estimate water intake, making them feasible for use in sheep and goats.

Therefore, this study aimed to (1) evaluate the use of urinary creatinine concentration as an indirect marker for estimating water intake in sheep and goats and (2) develop mathematical models based on variables such as body weight, creatinine concentration, and urinary volume, with applicability in intensive and extensive production systems for small ruminants in tropical climates.

2. Materials and Methods

2.1. Ethical Considerations and Experimental Location

The experiment was conducted at the Experimental Farm, belonging to the School of Veterinary Medicine and Animal Science of UFBA, located in the municipality of São Gonçalo dos Campos, Bahia, Brazil, at coordinates 12°23′49.5″ S and 38°52′43.5″ W, at an altitude of 228 m.

Based on the Köppen–Geiger climate classification, the region has a climate type As (tropical climate “A”, with average temperatures above 18 °C and “s”, with the dry season occurring during the time of highest sunlight and longest days), characterized by an average annual precipitation of 900 to 1200 mm.

2.2. Animals, Experimental Design, and Diets

Five Santa Inês crossbred sheep (8 to 10 months with an average body weight of 32 ± 4.5 kg) and five Boer crossbred goats (average age 3 to 4 months and average body weight of 22 ± 2.5 kg) were used. The animals were housed in individual, covered pens equipped with individual drinkers and feeders, measuring 1.0 m² (1.0 × 1.0 m) and featuring a suspended, wooden slatted floor. At the beginning of the experiment, the sheep and goats were identified, subjected to endoparasite and ectoparasite control, and immunized with a polyvalent vaccine against clostridial diseases.

Data related to estimated water intake were obtained concurrently with the digestibility experiment. The animals were distributed in a replicated 5 × 5 Latin square design (each square consisting of one species: goats and sheep), with five 15-day periods, during which the last three days of each period were dedicated to data collection. Experimental treatments consisted of a basal diet with titanium dioxide (TiO₂) inclusion levels of 1.0, 1.75, 2.5, 3.25, and 4.0 g/day. Titanium dioxide (TiO₂) was administered orally to each animal from days 6 to 15, in capsules, before the morning meal, and samples were collected on days 13, 14, and 15 of each experimental period.

The animals were fed twice daily, at 8:00 a.m. and 4:00 p.m., as a complete mixture. Animals were fed using a basal diet composed of 40% concentrate based on soybean meal, ground corn, and mineral mixture, and Tifton-85 hay (Cynodon sp.) was used as roughage, which comprised 60% of the diet (

Table 1. Proportion of ingredients and chemical composition of the diet provided, expressed as g/kg of dry matter (DM).

Ingredients	Inclusion Level (g/kg)
Tifton-85 hay	600.0
Soybean meal	110.0
Ground corn	270.0
Mineral mixture 1	20.0
Chemical composition (g/kg of DM)	Mean	SD
Dry matter (DM) g/kg, natural matter (NM)	841.1	0.63
Organic matter	801.0	0.64
Mineral matter	40.1	0.01
Crude protein	114.1	0.39
Ether extract	24.8	0.11
Neutral detergent fiber	481.4	0.81
Indigestible neutral detergent fiber	156.7	2.09
Neutral detergent fiber corrected for ash and protein	447.8	0.92
Acid detergent fiber	240.3	0.83
Hemicellullose	207.6	1.30
Cellullose	197.0	0.76
Lignin	43.2	0.20
Non-fibrous carbohydrates 2	339.7	0.84
Total carbohydrates 3	821.1	0.34
Total digestible nutrients 4	657.7	0.68

¹ Guaranteed levels (per kg in active elements): calcium—135 g; phosphorus—65 g; sodium—107 g; sulfur—12 g; copper—100 mg; cobalt—175 mg; magnesium—6000 mg; iodine—175 mg; manganese—1440 mg; selenium—27 mg; zinc—6000 mg; maximum fluorine—650 mg; ² Non-fibrous carbohydrates estimated by the Weiss equation [22]; ³ Total carbohydrates by the Sniffen et al. equation [23]; ⁴ Total digestible nutrients estimated by the Cruz et al. equation [24]; SD = standard deviation of the means.

).

The diet used for both species was formulated to be isonitrogenous (12% CP), as recommended by the National Research Council [15], to meet the nutritional requirements of goats and sheep for weight gains of 150 g/day.

2.3. Nutrient Intake

Nutrient intake was estimated between days 13 and 15 of each experimental period by calculating the difference between the total amount of each nutrient contained in the diets offered and the total amount of each nutrient present in the leftovers. Daily at 8:00 a.m., prior to the first meal, the leftovers were collected and weighed using a digital scale to determine the animals’ dry matter intake. Dry matter intake was obtained by adjusting the amount of feed offered to the animals of both species to allow for between 10 and 15% leftovers. Samples of ingredients, diet, and leftovers were frozen for subsequent analysis of their chemical composition.

Dry matter intakes were expressed in g/animal/day, g/kg of body weight (BW), and g/kg of metabolic body weight (MBW). They were calculated using the following equations: DMI (g/animal/day) = amount of DM offered − amount of DM in leftovers.

Intake as a percentage of body weight (%BW) was calculated using the equation: DMI (%BW) = amount of DM consumed × 100/BW (kg). Furthermore, metabolic weight DM intakes were estimated using the following equation: DMI (g/kg MBW) = amount of DM consumed)/MBW.

2.4. Water Intake

Water consumption was measured on days 13, 14, and 15 of each experimental period. Water was supplied in standardized 10-L plastic buckets. Water consumption was calculated by the difference in bucket weights before and after animal consumption, at 24-h intervals.

All animals’ buckets were washed and refilled daily. Furthermore, at the same time, two buckets were positioned in different locations in the barn to measure evaporation losses. This allowed for the correction of water consumption [25].

IH₂O = (IW − FW) − EW

where IH₂O = water intake; IW = Initial weight of the bucket with water; FW = Final weight of the bucket with water after 24 h; EW = Average difference in bucket weights for estimating evaporation.

2.5. Urine Collection and Urine Volume Estimation

On day 14 of each experimental period, four hours after offering the morning diet, spot urine samples were collected from each animal during spontaneous urination using sterile plastic containers.

Immediately after collection, 10 mL aliquots of urine were filtered through gauze and mixed with 40 mL of 0.036 N sulfuric acid (H₂SO₄) [26] to preserve urinary compounds from microbial degradation [27]. The samples were then placed in labeled plastic containers and stored at −20 °C for later creatinine analysis at the Animal Nutrition Laboratory (LANA) of the Federal University of Bahia.

Creatinine concentrations were estimated using a commercial kit (ID: 35-100, Labtest^®, Lagoa Santa, Brazil), and absorbance values were read on a spectrophotometer. Creatinine was used as a marker to estimate urinary volume and was assumed to be excreted at a rate of 19.82 mg/kg body weight per day. In the current study, the same value of creatinine excretion was used for goats and sheep, as it was noted in our previous study [18] that this value is not influenced by the species evaluated. Thus, the following formula was used:

Daily urine excretion (mL) = [Body weight (kg) × 19.82]/[Creatinine concentration in urine sample (mg/dL) × 10]

2.6. Sample Collection and Laboratory Analyses

The ingredient, diet, and leftover samples were pre-dried in a forced-air oven (55 °C for 72 h). Then, samples were ground in a Willey knife mill with a 1-mm sieve, packaged in labeled plastic bags, and stored for later laboratory analysis according to protocols described by the National Institute of Science and Technology in Animal Science [28].

The samples were then subjected to evaluations of dry matter (DM; method G-003/1), ash (MM; method M-001/2), crude protein (CP; method N-001/2), ether extract (EE; method G-005/2), neutral detergent fiber (NDF; method F-013/1), acid detergent fiber (ADF; method F-015/1), and lignin (method F-005/2). The organic matter (OM) content was calculated from the difference between the DM and ash contents [28]. For NDF determinations, samples were treated with heat-stable alpha-amylase without sodium sulfite (Prodooze^®—Vila Suissa, Mogi das Cruzes, Brazil) and eight molar urea to reduce starch contamination. Neutral detergent fiber (NDF) was corrected for ash and protein (NDFap) according to the AOAC (method 973.18) [29].

Non-fibrous carbohydrates (NFC) were calculated using the equation proposed by Weiss [22], where NFC = 100 − (% CP + % EE + % Ash + % NDF); and total carbohydrates (TC) were calculated according to Sniffen et al. [23]: TC = 100 − (% CP + % EE + % Ash). Total digestible nutrient levels were estimated according to Cruz et al. [24].

2.7. Statistical Analyses

The results obtained for dry matter intake, water intake, urinary volume, and creatinine excretion were analyzed using PROC MIXED (Statistical Analysis System—SAS Institute Inc., Cary, NC, USA) in a double Latin square design (5 × 5) according to the model below:

Ŷijkl = μ + Ti + QLj + Anik (QL) + Perl (QL) + Ti × QLj + Ɛijkl

Ŷijkl = response of treatment i, in Latin square j, in animal k, in period l; μ = general constant; Ti = effect relative to treatment i; QLj = relative effect of Latin square j (sheep and goats); Anik (QLj) = effect relative to animal k within QLj; Per = effect relative to period l; QLj × Ti = effect relative to the interaction between treatment and Latin square; Ɛijkl = assumed random error NID (0; σ²);

The effect of the marker dose (TiO₂) was assessed by adjusting orthogonal linear (−2 −1 0 +1 +2) and quadratic (+2 −1 −2 −1 +2) polynomial contrasts. Simple or multiple linear regression models were estimated using SAS PROC REG (SAS 9.1 version).

Pearson’s correlation analysis was performed to evaluate the relationship between water consumption and the other independent variables adopted as possible predictors. For all evaluations, a 5% probability level for Type I Error was considered.

3. Results

3.1. Descriptive Statistics

The database used to analyze Person’s correlation coefficients and develop the equations revealed dispersion among the variables, as observed in the descriptive statistics (

Table 2. Descriptive statistics of the database used to develop prediction equations for water intake in sheep and goats.

Variable	N	Mean	SD	Minimum	Maximum
IH2O
L/day	42	1.87	0.63	0.93	3.13
g/kg BW	42	60.13	28.48	20.31	155.74
g/kg MBW	42	141.32	60.90	53.38	329.36
DMI
g/day	48	969.77	235.22	474.66	1498.00
g/kg BW	48	29.91	5.29	19.13	47.75
g/kg MBW	48	71.00	11.62	46.92	97.81
Urinary volume
L/day	43	1.45	0.60	0.49	2.98
mL/kg BW	43	46.76	26.34	16.90	149.02
mL/kg MBW	43	109.93	56.32	40.07	315.14
Body weight (kg)	48	32.81	7.45	17.60	47.70
Metabolic body weight	48	13.64	2.35	8.60	18.15
Creatinine (mg/L)	48	523.61	246.23	133.00	1173.00

IH₂O = H₂O intake; L/day = liters/day; g/kg BW = grams per kilogram of body weight; g/kg MBW = grams/kilogram of metabolic body weight; DMI = dry matter intake; g/day = grams/day; mL/kg BW = milliliters/kilogram of body weight; mL/kg MBW = milliliters/kilogram of metabolic body weight; mg/L = milligrams/liters; SEM = standard error of the mean.

).

3.2. Effects of Titanium Doses and Animal Species on Water and Dry Matter Intake, Creatinine, and Urinary Volume

There was no interaction (p > 0.05) between titanium doses and species for any of the studied variables (

Table 3. Effects of titanium doses and animal species on water and dry matter intake, creatinine concentration, and urinary volume.

Item	Marker (Grams/Day)					SEM	Specie		SEM	p Value
	1.0	1.75	2.50	3.25	4.0		Sheep	Goats		Linear	Quadratic	Specie	Marker × Specie
IH2O
L/day	1.66	1.94	1.80	1.33	2.14	0.286	2.02	1.52	0.282	0.512	0.227	0.146	0.220
g/kg BW	53.22	62.22	57.85	46.46	67.41	9.862	54.78	60.1	10.32	0.424	0.333	0.657	0.268
g/kg MBW	125.14	146.34	135.8	106.8	159.66	22.397	134.89	134.61	23.00	0.431	0.279	0.991	0.253
DMI
g/day	1011.11	926.71	925.89	980.55	1007.05	63.782	1128.38	812.15	57.35	0.733	0.084	<0.001	0.226
g/kg BW	31.122	28.027	29.192	30.22	31.51	1.606	30.10	29.93	1.28	0.416	0.059	0.871	0.104
g/kg MBW	74.07	67.03	69.72	72.12	74.54	3.515	74.64	68.35	2.68	0.484	0.082	0.034	0.124
Creatinine (mg/L)	455.72	495.66	520.85	583.55	571.28	86.997	589.63	461.19	71.78	0.200	0.812	0.229	0.927
Urinary volume (L/day)	1.571	1.562	1.459	1.240	1.280	0.194	1.422	1.425	0.17	0.103	0.993	0.992	0.737

IH₂O = H₂O intake; L/day = liters/day; g/kg BW = grams/kilogram of body weight; g/kg MBW = grams/kilogram of metabolic body weight; DMI = dry matter intake; g/day = grams/day; mg/L = milligrams/liters; SEM = standard error of the mean.

). There was an effect of species on dry matter intake (p < 0.001). Higher intakes (g/day) were observed in sheep compared to goats.

Similarly, there was no effect of titanium doses or species (p > 0.05) on water intake (L/day; g/kg BW and g/kg MBW), dry matter intake (g/kg BW and g/kg MBW), creatinine concentrations, or urinary volume (Table 3).

3.3. Pearson Correlation

A positive correlation was observed between water intake (g/kg BW) and urinary volumes expressed in mL/kg BW (r = 0.569; p < 0.001) and in mL/kg MBW (r = 0.524; p < 0.001). A similar correlation was observed between water intake (g/kg MBW) and the same urinary volume variables expressed in mL/kg BW (r = 0.526; p < 0.001) and in mL/kg MBW (r = 0.488; p < 0.001) (

Table 4. Linear correlations between water intake and the variables of dry matter intake, creatinine concentration, body weight and metabolic body weight, and urinary volume.

Variable	DMI						Creatinine		Body Weight		Metabolic Body Weight		Urinary Volume
	g/day		g/kg BW		g/kg MBW								L/day		mL/kg BW		mL/kg MBW
	r	p	r	p	r	p	r	p	r	p	r	p	R	p	r	p	r	p
L/day	0.043	0.784	0.203	0.197	0.164	0.299	−0.284	0.068	−0.111	0.481	−0.111	0.481	0.262	0.093	0.297	0.056	0.294	0.056
g/kg BW	−0.373	0.014	0.226	0.151	0.007	0.961	−0.373	0.015	−0.586	<0.001	−0.593	<0.001	0.297	0.055	0.569	<0.001	0.524	<0.001
g/kg MBW	−0.296	0.056	0.221	0.145	0.042	0.792	−0.365	0.017	−0.504	<0.001	−0.509	<0.001	0.297	0.055	0.526	<0.001	0.488	<0.001

L/day = liters/day; g/kg BW = grams/kilogram of body weight; g/kg MBW = grams/kilogram of metabolic body weight; DMI = dry matter intake; mL/kg BW = milliliters/kilogram of body weight; mL/kg MBW = milliliters/kilogram of metabolic body weight; BW = body weight in kilograms; MBW = metabolic body weight; Creatinine in mg/L = milligrams/liters.

).

A negative correlation was observed between water intake (g/kg BW) and dry matter intake (g/day) (r = −0.373; p = 0.014). Similarly, there was a negative correlation between water intake (g/kg BW) and creatinine concentration (r = −0.373; p = 0.015), body weight (r = −0.586; p = 0.015), and metabolic (r = −0.593; p < 0.001). Negative correlations were also observed for water intake (g/kg MBW) and creatinine concentration (r = −0.365; p = 0.017), body weight (r = −0.504; p < 0.001), and metabolic (r = −0.509; p < 0.001) (Table 4).

No correlation was observed between water consumption (L/day), dry matter intake, creatinine concentration, body and metabolic weight, and urinary volume when expressed in L/day, mL/kg BW, and mL/kg MBW (p > 0.05). Similarly, no correlation was noticed between water intake (g/kg BW) and dry matter intake (g/kg BW and g/kg MBW) and urinary volume (L/day) (p > 0.05). No correlation was detected between water intake (g/kg MBW) and dry matter intake (g/day, g/kg BW, and g/kg MBW) and urinary volume (L/day) (p > 0.05) (Table 4).

3.4. Mathematical Models for Predicting Water Intake in Small Ruminants

Five equations were obtained to predict water intake in sheep and goats (p < 0.001) (

Table 5. Regression equations for predicting water intake in goats and sheep.

	Equation	Coefficient of Determination (R2)	p
1	ŶH2Og/kgBW = 112.42 − 5.17 × MBW + 0.40 × UV	46	<0.001
2	ŶH2Og/kgBW = 164.72 − 6.60 × MBW + 0.025 × Creat	40	<0.001
3	ŶH2Og/kgBW = 139.53 + 0.016 × DMI − 2.48 × BW − 0.025×Creat	40	<0.001
4	ŶH2Og/kgBW = 92.78 + 0.01 × DMI − 2.02 × BW + 0.39 × UV	46	<0.001
5	ŶH2Og/kgBW = 116.22 + 0.017 × DMI − 6.65 × MBW + 0.39 × UV	47	<0.001

BW = body weight per kilogram; MBW = metabolic body weight; UV = urinary volume per mL/kilogram body weight; DMI = dry matter intake per grams/day; Creat = creatinine per mg/L.

). The first equation for predicting water intake was: Ŷ_H2Og/kgBW = 116.22 + 0.017 × DMI − 6.65 × MBW + 0.39 × UV, which considered dry matter intake (DMI), metabolic body weight (MBW), and urinary volume (UV) as predictor variables.

Of the two subsequent equations for predicting the animals’ water intake, the first considered metabolic weight and urinary volume Ŷ_H2Og/kgBW = 112.42 − 5.17 × MBW + 0.40 × UV, and the second used, in addition to body weight and urinary volume, the animals’ dry matter intake: Ŷ_H2Og/kgBW = 92.78 + 0.01 × DMI − 2.02 × BW + 0.39 × UV. Finally, two other water intake prediction equations were also obtained, considering metabolic weight and creatinine as predictor variables: Ŷ_H2Og/kgBW = 164.72 − 6.60 × MBW + 0.025 × Creat and creatinine together with dry matter intake and body weight of the animals: Ŷ_H2Og/kgBW = 139.53 + 0.016 × DMI − 0.025 × Creat − 2.48 × BW.

4. Discussion

4.1. Effects of Titanium Doses and Animal Species on Water and Dry Matter Intake, Creatinine, and Urinary Volume

The administration of different levels of titanium dioxide (TiO₂), ranging from 1.00 to 4.00 g in the diet, did not result in significant effects on the physiological and nutritional indicators evaluated in sheep and goats: water intake (IH₂O), dry matter intake (DMI), creatinine concentration, and urinary volume (UV).

The absence of interaction between the levels shows that TiO₂ does not interfere with the water or nitrogen metabolism of the animals, preserving their physiological and metabolic responses. Characteristics such as physiological inactivity and the absence of intestinal absorption are relevant to the study, given their non-interaction with the observed variables. These characteristics ensure that the marker transits through the gastrointestinal tract without undergoing metabolic transformations or influencing the physiological functioning of the animals [30,31].

Furthermore, the lack of significant interaction between the indicator and the variables reinforces that the response pattern was similar in sheep and goats, suggesting that metabolic behavior when using the marker is stable in small ruminants. These findings are supported by several authors, who demonstrate the safety and efficiency of TiO₂ as an external marker in digestibility experiments and its non-interference with the physiological activities studied [32,33,34].

Thus, the observed responses in the digestive and metabolic variables can be more confidently attributed to the intrinsic characteristics of the animals and the diets provided, rather than to possible interference from the marker [35,36].

Sheep showed higher dry matter intake compared to goats, corroborating the literature that adaptive differences linked to feeding habits, selectivity, and body size impact DMI [2,37]. Therefore, the results corroborate those found by Dos Santos et al. [18] when comparing nutrient intake in feedlot goats and sheep. As noted by the authors, it was possible to suggest a lower intake in goats than in sheep, which is partially attributed to the time spent by the animals’ selecting nutrients from their diets. Therefore, considering that sheep are characterized as typical grazers and goats as intermediate selective grazers, sheep have a larger rumen-reticular space according to body weight compared to goats [15]. Consequently, it is possible to expect higher dry matter and fiber intake according to body weight in sheep.

4.2. Correlation Between Water Consumption, Food Intake, and Physiological Parameters

Water intake was positively correlated with urinary volume, corroborating the maintenance of water homeostasis and the need for proportional replacement of losses [17,38]. This physiological variable, therefore, acts as an integrated indicator of renal function and protein metabolism.

However, a negative correlation was observed between water intake and urinary creatinine concentration, which may indicate that creatinine, a muscle metabolite, can be widely used as a marker of glomerular filtration rate and hydration status. High concentrations of this metabolite indicate less urinary dilution, requiring greater water intake for the excretion of nitrogenous waste [16,20,39,40].

In contrast, animals that maintain stable creatinine levels despite lower water intake demonstrate adaptive renal efficiency, a characteristic particularly advantageous in semiarid environments [41,42]. Creatinine indirectly reflects renal excretion capacity and is helpful in assessing the balance between water intake and metabolic waste elimination, especially in animals with more intense protein metabolism or higher glomerular filtration rates.

It is also observed that body weight and metabolic body weight negatively correlate with water consumption, suggesting that larger animals tend to have greater relative water efficiency, which should be considered in management strategies.

4.3. Mathematical Models for Predicting Water Intake

The use of mathematical models has become an indispensable tool for supporting decision-making processes. Complex phenomena, such as the effects of global warming on terrestrial biology, public health, and pandemic management, rely heavily on modeling studies. In this context, the application of models has expanded to various fields of knowledge, including animal science and production, medicine, physics, chemistry, economics, engineering, and others.

These models are derived from observational data and are applied in experimental studies that examine variable-response relationships to understand how one element influences another and its associated variables.

These models are developed based on observational data and applied in experimental studies that evaluate variable-response relationships, explaining how one element relates to another and impacts its variations.

Multiple regression models, with coefficients of determination ranging from 40% to 47%, demonstrated a moderate integration of physiological (MBW, urinary volume, creatinine) and nutritional (DMI) variables. However, in biological terms, the coefficients found justify their use to predict water intake in small ruminants. In the context of the Brazilian semiarid region, marked by water scarcity and exacerbated by climate change, these equations offer valuable tools for rational water use and support for management strategies and genetic selection for water efficiency [43,44,45].

Water intake prediction models in goats and sheep have historically been based on variables related to environmental and physiological conditions, such as ambient temperature, relative humidity, wind speed, solar radiation, body weight, production level (milk yield and weight gain), dry matter intake, and diet composition [44,46,47,48,49]. Although these variables are relevant, they exhibit high temporal and spatial variability, necessitating continuous monitoring or complex measurements that can limit their practical applicability in extensive production systems, particularly in tropical regions.

The inclusion of creatinine in predictive models enables the estimation of volume and the association of water intake with the intensity of protein catabolism and excretion efficiency, which is crucial for ruminants to adapt to hot and dry climates. Studies in sheep have shown that daily creatinine excretion remains relatively constant between treatments, allowing for the estimation of urinary volume from point samples [18,50]. In cattle, Silva et al. [51] determined creatinine excretion equations as a function of body weight, showing that heavier animals release more absolute creatinine, which may be related to greater muscle mass.

In this context, the inclusion of urinary creatinine as an additional variable in the models is proposed, due to its relatively constant endogenous production, which reflects both muscle mass and the efficiency of nitrogenous waste excretion [52]. This characteristic, combined with the ease and low cost of quantification by routine laboratory methods, makes creatinine a stable marker that is less influenced by environmental fluctuations, offering a promising alternative for estimating water intake of small ruminants in hot and dry tropical environments, where water availability and quality often limit animal productivity.

These relationships make creatinine a strategic physiological variable in prediction equations. In the models evaluated, creatinine appears as a variable with a significant effect, indicating that its role impacts water balance, intake, and excretion. This behavior aligns with the literature, which discusses creatinine not only as a marker of renal function but also as an indirect indicator of metabolic efficiency and water management in ruminants [53].

The negative association between dry matter intake (kg/day) and water intake confirms that diets with lower moisture content induce greater voluntary water intake. The equations that consider urinary volume as a parameter were also significant, indicating its role in water intake variation, in accordance with physiological principles. Thus, the combination of DMI, MBW, creatinine, and urinary volume yielded greater accuracy in predicting water intake in small ruminants.

However, the prediction models published by international committees are primarily obtained in temperate countries and, as such, are designed specifically for their own environmental characteristics, breeds, and feed components, which contrast sharply with the reality found in tropical regions [54]. Therefore, equations that are more suitable for our climatic conditions and can more accurately predict animal needs are crucial to reducing the environmental impact and high economic costs of animal production in the tropics.

The regression equation Ŷ_H2Og/kgBW = 112.42 − 5.17 × MBW + 0.40 × UV estimates water consumption in goats and sheep based on metabolic body weight (MBW) and urinary volume, indicating that animals with higher MBW tend to consume less water per kilogram of body weight. In contrast, higher urinary volume increases this consumption. Highlights the negative contribution of metabolic body weight (MBW), consistent with the literature, which indicates that larger animals have lower proportional water consumption due to a lower relative metabolic rate [15].

The equations Ŷ_H2Og/kgBW = 92.78 + 0.01 × DMI − 2.02 × BW + 0.39 × UV and Ŷ_H2Og/kgBW = 116.22 + 0.017 × DMI − 6.65 × MBW + 0.39 × UV, which had urinary volume (UV) in their composition, showed a positive relationship, which is physiologically justifiable, since greater urinary excretion requires greater water intake to maintain homeostasis [13,17]. They confirm that the integration of nutritional (DMI), physiological (BW or MBW), and excretory (UV) parameters results in a better ability to predict water intake. These findings corroborate previous studies demonstrating the importance of a multifactorial approach to more accurately estimate water intake in ruminants [17,20].

In the equation Ŷ_H2Og/kgBW = 164.72 − 6.60 × MBW + 0.025 × Creat, the inclusion of urinary creatinine concentration as an explanatory variable is associated with the excretion of nitrogen metabolites, reflecting the need for greater water intake to dilute these compounds, as discussed by Chen and Gomes [16] and Gonda and Lindberg [53].

The equation Ŷ_H2Og/kgBW = 139.53 + 0.016 × DMI − 0.025 × Creat − 2.48 × BW introduces dry matter intake (DMI) as a positive factor, in line with the principle that diets with lower moisture content induce greater voluntary water intake [15,54,55,56]. Creatinine, on the other hand, presents a negative correlation in this model, possibly indicating a lower need for water intake in situations of greater renal efficiency or lower protein catabolism.

Improving nutrition efficiency in sheep and goats is more challenging than in other species [57]. Because of their nutritional and environmental adaptability, sheep and goats are managed in very varied farming (from extensive to highly intensive) and feeding (from grazing and browsing to total mixed diets) systems, in widespread geographical areas and by using different breeds, populations, and crosses.

Given this scenario, the production efficiency of small ruminants is significantly more variable and challenging to predict and improve upon than that of cattle. Therefore, it is relevant to develop statistical models of water intake in sheep and goats using variables that require less information to be obtained.

The development of accurate prediction models for drinking water intake (DWI) may serve as a tool to optimize the use of this resource, thereby contributing to the improvement of the animal breeding system [44]. Due to the great variety of breeds and differences in performance between them in previous studies to estimate water intake in small ruminants, we highlight the relevance of our study since it is necessary to search for factors that allow better integration of the models in different breeds and environmental conditions that require less information to be obtained and are considered stable. In this manner, the development of accurate prediction models for drinking water intake (DWI) may serve as a tool to optimize the use of this resource, contributing to improved efficiency in the animal breeding system and sustainability in water usage for production systems.

Finally, from a practical perspective, the integration of creatinine into the models represents an important advance, as it enables not only the estimation of water intake but also the monitoring of the relationship between diet, protein metabolism, and renal function. This can be explored as a management tool to prevent dehydration or metabolic overload in herds. Therefore, the adoption of creatinine as a physiological marker, combined with mathematical models adapted to tropical conditions, constitutes an essential tool for addressing the challenges of livestock farming in the semiarid region. It also supports feed and water management. The results obtained can inform climate resilience policies and genetic improvement programs focused on water efficiency, strengthening the sustainability of animal production in hot environments, such as tropical, arid, and semiarid regions.

The consolidation of this methodology depends on additional studies that validate its application under different livestock conditions, with an emphasis on environmental, dietary, and physiological variations. However, creatinine is a promising variable for increasing the accuracy and practical applicability of predicting water intake in small ruminant production systems.

5. Conclusions

The models developed in this study accurately predicted the water intake of sheep and goats managed under tropical conditions. However, among the five models developed, the following equation, Ŷ_H2Og/kgBW = 164.72 − 6.60 * MBW + 0.025 * Creat, was considered the most appropriate and easiest to use for predicting water intake. Therefore, it can be used for both small ruminants managed in extensive systems and in confinement, as it uses only the predictive variables of metabolic body weight and urinary creatinine concentration.