Introduction

Hair sheep are well adapted to tropical climatic changes and the variation in forage availability throughout the year (^{Chay-Canul et al. 2011}). However, these changes induce periods of weight loss in animals (^{Chay-Canul et al. 2017}), which give rise to variations in the proportion of their tissues; as well as in the chemical composition of the body and the carcass (^{Chay-Canul et al. 2017}). Pelibuey sheep is a breed that shows fat depot variations due to the accumulation of a large fat quantity in their internal cavity (^{Chay-Canul et al. 2016}, ^{Morales-Martinez et al. 2020}).

The body of livestock has four chemical components in decreasing amounts: water, protein, fat, and minerals (^{Maeno et al. 2013}, ^{Tedeschi et al. 2017}, ^{Tedeschi 2019}, ^{Chay-Canul et al. 2019}). However, there is a lack of in-depth information about body composition for farm animals. The evaluation of the body chemical composition of productive animals is essential because this knowledge can aid in the assessment of energy and protein requirements and could improve feeding efficiency (^{Maeno et al. 2013}, ^{Tedeschi et al. 2017}, ^{Tedeschi 2019}). However, due to the sophisticated equipment required by other methods to evaluate body chemical composition (^{Eisenmann et al. 2004}, ^{Achamrah et al. 2018}, ^{Chay-Canul et al. 2019}), it is necessary to find a similar non-invasive method that allows evaluating internal body energy reserves in vivo.

Body mass index (BMI) is an indirect and non-invasive method commonly used to identify obesity (^{Doak et al. 2013}) or as an indicator of energy status in humans (^{Okorodudu et al. 2010}, ^{Kinge 2016}, ^{Ortega et al. 2016}). Due to the relationship between BMI and adiposity, BMI has been proposed for use in farm animals (^{Ptáček et al. 2018}, ^{Salazar-Cuytún et al. 2020}). In recent years, relationships between BMI and several parameters in farm animal were reported: milk and meat productivity of goats and sheep’s (^{Randby et al. 2015}, ^{Ptáček et al. 2018}), hormone production in the adipose tissues of adult goats and growing kids (^{Vilar-Martínez et al. 2009}, ^{Habibu et al. 2016}), body condition score of prepuberal sheep (^{Monteiro et al. 2010}), and in adult Pelibuey ewes (Chavarría-Aguilar et al. 2016, ^{Salazar-Cuytún et al. 2020}). Most recent studies showed the potential of BMI as an alternative tool to predict body composition in sheep (^{Ptáček et al. 2018}). Nevertheless, there is no information about the relationship of BMI with the chemical body composition in sheep. The present study aimed to investigate the relationship between BMI and body chemical components to predict CP, Fat, Energy, and Ash in Pelibuey ewes.

Materials and methods

Results

The mean, minimum and maximum values of the variables are presented in Table 1. The chemical body components that showed a higher variation were total body energy (TBE), carcass energy (CE), and visceral energy (VE). The correlation coefficients (r) between variables are shown in Table 2. BMI was moderately correlated (p < 0.001) with carcass crude protein (CCP; r = 0.51) and visceral crude protein (VCP); r = 0.48). It showed from moderate to high correlations (p < 0.001) with carcass fat (CF; r = 0.81) and visceral fat (VF; r = 0.71). However, the correlation between BMI was not significant (p > 0.05) with total body fat (TBF), TBE, and CA.

Table 3 shows the regression equations describing the relationship between BMI and body chemical components in Pelibuey ewes. The coefficient of determination (r²) for the equations involving BMI and body chemical components ranged from 0.62 to 0.97. There was a quadratic relationship between total body crude protein and total body ash (RSD = 0.572 and 0.027, respectively). Meanwhile, total body fat and total body energy adjusted to a linear trend (RSD = 3.11 and 129.3, respectively). Regarding chemical carcass components, CCP, CF, and CE had a linear relationship with BMI, with an r² ranging from 0.67 (RSD = 66.27) for CE to 0.96 (RSD = 0.511) for CCP. The regression equations describing the relationship between BMI and visceral composition had r² values ranging from 0.23 for visceral crude protein (VCP; RSD = 0.14 kg) to 0.97 for visceral ash (VA; RSD = 0.008 kg). Finally, for VF and VE, the r² values ranged from 0.63 for VF (RSD: 1.76) to 0.64 for VE (RSD: 611.47). Because the intercepts of equations 1, 3, 5, 7, and 11 were not significant (p > 0.05), we fitted a linear regression through the origin. Table 4 shows the regression equations describing the relationship between BMI corrected (using the empty body weight for calculating BMI, BMIc) and body chemical components in Pelibuey ewes. The coefficient of determination (r²) for the equations involving BMI and body chemical components ranged from 0.58 to 0.69, and for carcass and visceral components, the r² ranged from 0.37 to 0.98.

To evaluate the equations for predicting body composition from BMI the null hypothesis was accepted with an intercept = 0 and slope = 1 (Table 5). The results for equations 1 and 3 had low precision (r² = 0.32 and 0.26, respectively), a low reproducibility index and concordance with the observed data (CCCs = 0.48 and 0.42, respectively; < 0.50), and a low efficiency of prediction (MEF = 0.23 to 0.30). Meanwhile, equations 2 and 4 presented moderate precision (r² = 0.61; Table 5), high accuracy (Cb > 0.80; Table 5), and moderate CCCs (0.76) and MEFs (0.61). For all equations, the CDs ranged from 1.60 to 3.06, indicating high variability in the predicted data (Table 4). The partition of the MSEP (% MSEP) indicated that the largest proportion (> 94%) of the error was associated with random error (Table 5). In general, the equations overpredicted the body chemical components by around 14.88 to 33.58%.

On the other hand, the equations (5 to 8) for predicting carcass composition from BMI had an r² indicating low to moderate precision (0.25 to 0.66) and moderate to high accuracy (Cb > 0.61; Table 5). Nonetheless, the CCCs were 0.31, 0.80, 0.44 and 0.80 for equations 5, 6, 7 and 8, respectively. The CDs ranged from 0.43 to 2.68, indicating high variability in the predicted data (Table 5). Except for equation 5, the partition of the MSEP (% MSEP) indicated that the largest proportion (> 91%) of the error was associated with random error (Table 5). In general, the equations overpredicted the carcass chemical composition (ranged from 17.30 to 34.41%).

Finally, for equations developed to predict visceral composition from BMI, the null hypothesis was accepted in all of them (Table 5). Equations 9 and 11 had low precision (r² of 0.23 and 0.08, respectively), a low reproducibility index in concordance with the observed data (CCCs = 0.22 and 0.15, respectively; < 0.50) and low efficiency prediction (MEF = -0.23 to 0.08). However, equations 10 and 12 presented moderate precision (r² > 0.62; Table 6), high accuracy (Cb > 0.92) and moderate CCCs (> 0.73) and MEFs (> 461; Table 5). The partition of the MSEP (% MSEP) indicated that the largest proportion (> 60%) of the error was associated with random error (Table 5). In general, the equations overpredicted the body chemical components around 16.38 to 32.43%.

Discussion

To the best of our knowledge, the present study is the first to evaluate the relationship between the BMI and body chemical components of adult sheep. The determination of the body chemical composition of productive animals can aid in the assessment of energy and protein requirements and improve feeding efficiency (^{Tedeschi et al. 2017}, ^{Tedeschi 2019}). Given the positive relationships found by Chavarría-Aguilar et al. (2016) between BMI and BCS (r² = 0.80) and body energy reserves (total body fat), the present study explored BMI as a predictor of the body chemical composition of Pelibuey ewes. The calculation of BMI in sheep involves the measurement of mass (body weight) and body size and shape (withers heights and body length) (^{Salazar-Cuytún et al. 2020}). Body weight was found to be a function of body size (skeletal development), body fat (BCS), and gut fill in lactating dairy cows (^{Yan et al. 2009}). Notably, these variables would be ideal for the assessment of body chemical composition because they can be easily measured (body size) or estimated. Besides, ^{Muliyono et al. (2009)} reported that body size could be estimated from chest deep, and that shape could be described from several body measurements in sheep. Body size and shape together could provide a better description of an animal’s body conformation, and their inclusion in the estimation of BMI could generate a more accurate estimate (^{Yan et al. 2009}, ^{Muliyono et al. 2009}). However, there is a lack of information about the relationship between body chemical composition, body size measurements, and weight in sheep (^{Chay-Canul et al. 2017}, ^{Salazar-Cuytún et al. 2020}).

In a previous study, ^{Yan et al. (2009)} found that body size measurements and BCS were able to accurately predict empty body masses as well as lipid, crude protein, dry matter, water, and total gross energy contents in lactating dairy cows. Also, ^{Sanson et al. (1993)} found a high correlation between body weight (BW) and BCS (r = 0.89) in western-range ewes. However, these latter authors reported that the inclusion of both BW and BCS in regression models did not increase their accuracy and BW was the single best predictor. Even so, BCS was highly related to carcass lipids and was suggested as a possible descriptor of available energy reserves in ewes and cows. In the present study, the inclusion of BMI in the equation for estimating the body chemical composition of ewes showed good results (Table 3). ^{Maeno et al. (2013)} studied the accretion rates of chemical components in the body of domestic animals (cattle, goats, pigs, sheep, dogs, mice and rats) using allometric equations (Y = aX ^b ) describing the relationships of empty body weight (EBW), fat-free weight (FFW) and protein (PRO) with the weights of each chemical component (water, fat, and ash). The allometric growth coefficients (slope values) for PRO and water (WAT) did not differ (p > 0.2) among farm animals, although FAT and ASH showed differences (p < 0.01). The highly positive relationship found between PRO, WAT, and FFW in the evaluated farm animals, enabling the following equations to be generated: PRO = 0.1513 FFW ^1.085 and WAT = 0.8303 FFW ^0.972. However, these equations did not fit the data for laboratory animals (dogs, mice, and rats). Those relationships in farm animals confirm that protein and water weights could be estimated from equations. These results have practical implications for the estimation of body composition. It reduced the time and cost of laboratory analysis. Besides, it confirms that models to predict body composition can be built from these components due to their high association with the weight and size of many animals. It is important to highlight that these predictive equations are not generalized to animals of different sex, species, or physiological state. Meanwhile, in the present study, the eleven equations generated to estimate the body chemical composition of a single species (sheep) were able to predict body composition based on BMI to a certain extent.

In humans, ^{Maynard et al. (2001)} reported moderate to high correlations between BMI and the percentage of body fat (PBF) and total body fat (TBF), which had r values ranging from 0.64 to 0.85 and 0.83 to 0.94, respectively. Also, these correlations indicated that BMI accounted for 41 to 88% of the variation in PBF and TBF. These results coincide with those found in the present study, wherein a moderate association was found between BMI and TBF (r = 0.79). However, the equations for estimating body chemical composition showed a low to moderate precision (r² = 0.08 to 0.66; Table 5); hence, our equations were not better than those used in humans and dogs to estimate body fat (^{Speakman et al. 2001}).

In the present study, the regression equations 2, 6, and 10 developed to explain the relationship of BMI with the fatty component, or TBF, CF, and VF, respectively, presented a moderate relationship (r² = 0.62 to 0.67) and high variability. These results could be explained for the variation in body tissues (fat and muscle) according to the animal age, sex, and climatic condition or food availability in tropical conditions (^{Tedeschi et al. 2013}, ^{Bautista-Díaz et al. 2017}). Overall, the results of the present study support the use of body measurements such as live weight and body size (BMI) as indicators of body chemical composition in Pelibuey ewes. This relationship has already been confirmed in humans and other animals. However, the results should not be considered applicable to both sexes or all species and physiological states, and this method should be evaluated under different management and physiological conditions. Also, Tables 4 and 6 shows that the use of empty body weight as part of a corrected body mass index (BMIc) improves the values of the estimation of the body chemical composition. It showed a better adjustment of the coefficient of regression value (r²), meaning that the linear relationship between the predicted and observed values were improved, and the MSEP values were low, which resulted in the best models in the evaluation, with the exception for visceral ash. Therefore, we suggest the use of empty body weight to make more accurate estimates of body chemical components of adult sheep.

Conclusions

The present study showed that BMI could be used as predictors of body chemical composition in non-pregnant and non-lactating Pelibuey ewes. The use of empty body weight for calculating body mass index (BMI) yielded more accurate estimates of the chemical components of the body of adult sheep. The BMI as a predictor of body composition should be evaluated in animals with different physiological states raised under different management systems.