Effectiveness of using ultrasound readings to predict carcass traits and sensory quality in young bulls

1 Effectiveness of using ultrasound readings to predict carcass traits and 1 sensory quality in young bulls 2 3 M. J. Beriaina , K. Insaustia, M. Valerab, G. Induraina, A. Purroya, T. R. Carrc, and A. 4 Horcadab,* 5 6 aResearch Institute for Innovation & Sustainable Development in Food Chain 7 (ISFOOD), Universidad Pública de Navarra, 31006 Pamplona, Spain; bDepartamento de 8 Ciencias Agroforestales, ETSIA, Universidad de Sevilla, 41013, Sevilla, Spain; 9 cCollege of Agricultural, Consumers and Environmental Sciences. Urbana, Illinois 10 61801, USA. 11 12 *Corresponding author: Alberto Horcada 13 e-mail address: [email protected] (A. Horcada) 14 15 16 2 Abstract 17 The objectives of this study were to develop ultrasound-derived prediction equations in 18 young bulls before slaughter to estimate carcass traits and beef sensory parameters and 19 to determine the optimum moment pre-slaughter to estimate the sensory quality of beef 20 from young bulls. Ultrasound images were measured in twenty live young bulls at 50, 21 25 and 1 days prior to slaughter. Intramuscular fat content, longissimus lumborum 22 muscle area, carcass fatness, texture and sensory analysis were measured by a trained 23 panel after slaughter. Partial least square methodology was used to find the relationship 24 between the ultrasound measurements and the dependent variables, such as carcass 25 traits, Warner-Bratzler shear force and the sensory profile. Additionally, a stepwise 26 procedure was used to select the most informative ultrasound variables in the prediction 27 equations for carcass and beef sensory traits. The results indicate that ultrasound testing 28 in the feedlot offers promising potential. Early ultrasound scan measurements were 29 useful during fattening to predict intramuscular fat content (R2=0.619, RMSE=0.44%, 30 50 days prior to slaughter), while ultrasound scans taken close to slaughter were useful 31 to predict kidney fat content (R2=0.717, RMSE=0.96%, 1 day prior to slaughter). 32 However, the prediction of sensory beef attributes was only useful for fatty flavor 33 (R2=0.556, RMSE=0.47%) at 1 day pre-slaughter. Thus, the prediction of fat parameters 34 using ultrasound measurements could constitute a valuable tool in the process of 35 selecting beef quality traits in young bulls before slaughter. 36 37 Key words: Beef, carcass prediction, grading, palatability, ultrasound 38 39 1. Introduction 40 Using ultrasound technology has been reported previously in cattle breeding 41 programs and for production purposes (Baker et al., 2006; Castilhos et al., 2018). The 42 3 application of this technology, not only on carcasses at the slaughterhouse but also on 43 live animals in farms, has already become a useful tool for beef producers, as it enables 44 them to match selection decisions to market demands. 45 Current carcass grading systems rely on straightforward visual evaluation of 46 carcass parameters. The European Beef Carcass Grading system is based on carcass 47 conformation and fatness (OJEU, 2006), while in the USA, it is focused on carcass yield 48 and intramuscular fat level, which are likely to have an impact on the sensory 49 acceptance by consumers. In addition, until now, carcass classification methods have 50 been performed after the animals are slaughtered. However, in order to slaughter 51 animals at the optimum moment and to make an early pre-slaughter classification of the 52 carcass and meat quality, methods to predict carcass characteristics and carcass quality 53 are required. 54 Among the most widely-used techniques in live animals is ultrasound (Wall et al., 55 2004; Peña et al., 2014) combined with image analysis (Hwang et al., 1997; Cannell et 56 al. 2002), which allows for the successful prediction of carcass parameters such as 57 muscle area or back fat thickness (Hamlin et al. 1995a; 1995b). Nogalski et al. (2017) 58 reported that both selected biometric ultrasound measurements (back fat and thickness 59 of subcutaneous rump) and selected blood parameters, such as triglycerides, could be 60 used to predict intramuscular fat content with satisfactory precision and accuracy. These 61 techniques, which can be used to estimate beef marbling, can help producers to fulfill 62 industry and consumer demands (Indurain et al., 2009). 63 The ultrasound technique has also been shown to be useful in determining the 64 nutritional quality of beef. In fact, Indurain et al. (2006) reported that the ultrasound 65 readings reflect the effect of fatness on fat composition (e.g. the fatty acid profile of 66 4 subcutaneous fat). However, there is a lack of research to demonstrate the efficiency of 67 ultrasound in estimating beef palatability. 68 Regarding beef production, Spain has a number of local breeds characterized by 69 their high muscle growth and reduced fat content. These breeds include Pirenaica, 70 which is well suited for beef production. Pirenaica is an early maturing beef breed, 71 producing medium-muscled lean carcasses. It accounts for 1.1 % of the Spanish national 72 cattle stock and it is the most abundant breed for beef production under the European 73 Union's PGI Ternera de Navarra [Navarre Veal] Protected Geographical Indication 74 (DOUE, 2004). Pirenaica breed cattle are raised in a semi-extensive system near the 75 mountains in the Pyrenees (Northern Spain). The animals take advantage of the 76 mountain pastures in spring and summer, while they remain in the valleys during the 77 winter. Young bulls are weaned at six months and are fed in feedlots with concentrate 78 and cereal straw until slaughter at approximately twelve months of age to obtained lean 79 carcasses. 80 In this context, the main objective of this study was to develop ultrasound-derived 81 prediction equations before slaughter to estimate carcass traits and beef sensory 82 parameters in young bulls with lean carcasses. 83 84 2. Material and methods 85 2.1. Pre-harvest, handling, and harvesting procedure 86 A total of 20 yearling Pirenaica bulls under the Protected Geographical Indication 87 Ternera de Navarra were used. The young bulls were reared in a semi-extensive system 88 in the mountains until weaning. After weaning at approximately seven-eight months of 89 age, they were fed on a concentrate in a local feedlot (85% barley, 10% soybean meal, 90 3% vegetal fat and 2% minerals and vitamins), and barley straw, both ad libitum. The 91 5 chemical composition of the feeds corresponds to a medium-energy diet: Dry matter 92 (88.2%); Crude protein (128.5g/kg DM); Crude fiber (48.4g/kg DM); Ash (53.7g/kg 93 DM); Metabolizable energy (11.5 MJ/kg DM). The average feed consumption during 94 the young bulls’ growth period (four months) was 7.9 kg/day. The animals were 95 considered as a representative sample of the market, and they are commonly slaughtered 96 in the area. The bulls were raised according to the handling conditions in the Spanish 97 rules and regulations for animal care (Directive 2010/63/EU) and slaughtered according 98 to Spanish rules and regulations for animal care (Council Regulation EC 1099/2009). 99 The research followed the official guidelines for the humane treatment, care and 100 handling of animals. 101 102 2.2. Ultrasound data collection 103 The live animals were weighed and two cross-sectional ultrasound images (right and 104 left side) between the 12th and 13th ribs were taken in the live animals at the feedlot, as 105 described by Wall et al. (2004) and Bergen et al. (2005). The ultrasound images were 106 measured on the skin after clipping the hair at 50, 25 and 1 days prior to slaughter, using 107 Sonovet 600-real-time ultrasound equipment (Madison Co. Ltd. Korea) equipped with a 108 linear probe (3.5 MHz, 120 x 20 mm). Mineral oil (Echoultragel; Pirrone & Co.SPA, 109 Italy) at 20-25oC was used to ensure suitable acoustic contact between the probe and the 110 skin. 111 The linear ultrasound measurements recorded on the live young bulls were: 112 intramuscular fat content or marbling (IMF), longissimus muscle depth and dorsal fat 113 thickness. The resulting two-dimensional images (longitudinal and transversal) were 114 digitalized and stored. The young bulls were scanned twice and the images evaluated 115 and interpreted by a laboratory technician at the Public University of Navarra, using the 116 6 Optimas 6.5 image analysis software (Media Cybernetics Inc., USA). The ultrasound 117 measurements of longissimus muscle depth selected were longitudinal and transversal 118 (UAL and UAT respectively), and dorsal fat thickness (UFL), since this is a linear 119 measurement. Gray levels of the longitudinal and transversal (UIMFL and UIMFT, 120 respectively), ultrasound images for intramuscular longissimus lumborum fat content 121 was measured with the gray level scale in the software, with 0 black and 150 white. 122 123 2.3. Carcass evaluation 124 On the day of slaughter, the young bulls were transported a distance of 125 approximately 20 km and harvested upon arrival using standard stunning and dressing 126 methods in a licensed slaughterhouse according to Council Regulation EC Nº 127 1099/2009. The young bulls were slaughtered at 521+51.4 kg live weight and 351+15 128 days of age and yielded an average cold-carcass weight of 323+33.8 kg. 129 Immediately prior to cool storage for aging, the carcasses were graded by a licensed 130 technician for fatness and conformation according to the EU beef grading system 131 (DOUE, 2008). The SEUROP conformation scale (S superior; E excellent; U very 132 good; R good; O fair; P poor) was transformed, with 1 = P- and 18 = S+. On the fat 133 cover classification scale (5 very high; 4 high; 3 average; 2 slight; 1 low), the numerical 134 transformation of standard grades were 1 = 1- and 15 = 5+. Kidney and pelvic fat (KPF) 135 was collected from the carcasses and weighed by a trained technician. Carcass fat 136 thickness (CFT) was measured at the 6th rib, taking 3/4 of the length ventrally over the 137 longissimus muscle, with a chilled stainless-steel caliber (Renand, and Fisher, 1997). In 138 addition, the longissimus lumborum area (CA) was measured with a grid. 139 Twenty-four hours post-mortem, the longissimus lumborum muscle was removed at 140 the 6-13th rib level from the left carcass side, and was transported to the Meat Science 141 7 Laboratory at the Public University of Navarra (Pamplona-Spain), where the meat was 142 aged in the dark for seven days at 2ºC. After ageing, the muscles were cut into steaks. 143 One steak was cut for instrumental texture analysis (3.5cm thick) and another for 144 sensory analysis (2.5cm thick) from the second lumbar vertebra. The steaks were then 145 vacuum packed in polyamide/polyethylene pouches (Vaeseen Schoemarket Ind., 146 Barcelona, Spain - film thickness of 120 µm and O2 permeability of 1 cc/m2/24 h, CO2 147 permeability of 3 cc/m2/24 h, and N2 permeability of 0.5 cc/m2/24 h measured at 5 ºC 148 and 75% relative humidity; water vapor transmission rate 3 g/m2/24 h at 28 ºC and 149 100% RH; vicat softening point of sealing 97ºC, and a dart drop strength of 1300g) and 150 stored at -20ºC for subsequent analysis. 151 152 2.4. Fat extraction 153 At 24 h post-mortem, a steak from the longissimus lumborum muscle at the 12th 154 rib level was used to determine the intramuscular fat content. The total lipids were 155 extracted from each meat product with chloroform:methanol (2:1, v/v), according to the 156 Folch et al. (1957) method. The results were expressed as total intramuscular fat content 157 per 100g of fresh meat. Two replicates were taken per sample. 158 159 2.5. Warner –Bratzler Shear Force 160 The steaks for the Warner–Bratzler shear force (WBSF) test were thawed 161 overnight at 4ºC. The beef fillet (3.5 cm thick) was cooked on a preheated sheet 162 previously at 180ºC, turning the fillet every 4 minutes, until reaching an internal 163 temperature of 70ºC (AMSA, 2016). The temperature inside the steak was controlled by 164 temperature probes and a data acquirer (thermocouple probe TB 190 and data acquirer 165 Almemo 5990-2 V5; Ahlborn mess-un Regelungstechnik GMBH, Holzkircken, 166 8 Alemania). After cooling for two hours seven cores measuring 1 cm wide, 1 cm high 167 and 3 cm long were removed from the cooked steaks parallel to the longitudinal axis of 168 the muscle fibers. The Shear Force analysis was carried out using a TA-XT2i texture 169 analyzer (Stable Micro Systems, Inc., Goodming, Surrey, UK). The maximum shear 170 force (kg) was assessed with a Warner-Bratzler shearing device following the 171 methodology proposed by Beltrán (2005). The samples were then tempered at room 172 temperature. The analysis was reformed with a crosshead speed of 200 mm•min−1, 50 173 kg load cell, 40 mm distance and a calibration weight of 10 kg. The full peak shear 174 force was recorded and the maximum shear force was calculated in kg as the mean of 175 the seven measurements. 176 177 2.6. Trained sensory panel 178 The sensory analysis was performed by a seven-member trained descriptive panel 179 selected and trained as described by Cross et al. (1978) and Meilgaard et al. (1991). The 180 members of the descriptive panel were selected and trained as described by the ISO 181 8586–2014 (International Organization for Standardization) to evaluate the beef 182 samples. The definitions used for the different sensory attributes are listed in Gorraiz et 183 al. (2000). Juiciness, hardness, characteristic beef flavor, liver flavor and fatty flavor 184 were evaluated, using a 150 mm unstructured line scale, with a 10 mm mark from the 185 left, representing "low intensity", and another mark 10 mm from the right, representing 186 "high intensity". 187 The frozen steaks were thawed over night at 4ºC, and cooked on a grill to an internal 188 temperature of 70ºC measured with a TB 190 probe (data logger Almemo 5990-2 V5; 189 Ahlborn mess-un Regelungstechnik GMBH, Holzkircken, Germany). The steaks were 190 then cut into 1.5x2x1.5cm cubes, which were wrapped in aluminum foil with an 191 9 identification tag with a random three-digit code and kept in a heat-retaining container 192 or waterless food warmer before serving. The tasting took place in the sensory testing 193 room at the Meat Science Laboratory of the Public University of Navarra (Pamplona-194 Spain) under red lighting, with controlled temperature and humidity. The tasting hall 195 meets UNE 87004 (1979) standard requirements for sensory trials. The panelists were 196 provided with slices of apple and water to cleanse their palate between samples. Each 197 panelist tasted five samples per session in a random order and each sample was assessed 198 by all seven members of the panel. 199 200 2.7. Statistical Analyses 201 All the data were analyzed using SPSS software (SPSS V. 25.0, SPSS Inc. USA). 202 A statistical descriptive analysis was carried out to describe the animal and carcass 203 characteristics, ultrasound readings, shear force and sensory position and dispersion 204 parameters. The effects on carcass traits of the moment when the ultrasound 205 measurements were performed were analyzed using the repeated measures GLM 206 procedure and a post hoc Fisher comparison of means test. Carcass weight was used as 207 a linear covariate. Pearson correlation coefficients were calculated to assess the 208 relationship between the ultrasound readings at three given points in the animals’ life 209 (50, 25 and 1 days prior to slaughter) and the corresponding carcass and beef 210 measurements and sensory evaluation. A partial least square methodology was used to 211 find the relationship between the dependent variables (carcass traits, Warner-Bratzler 212 shear force and the sensory profile after 7 days of ageing) and the ultrasound 213 measurements. Additionally, a stepwise procedure (forward selection of variables, 214 significance criterion P<0.05) was used to select the most informative ultrasound 215 variables in the prediction equations for carcass and beef sensory traits. Both the 216 16 young bulls from lean breeds. In fact, Albrecht et al. (2006) reported that the connective 321 tissue in the muscle is often present in strong cords which are not streaked with fat, and 322 the highest UIMF observed in the early scans of young bulls could be attributed to the 323 relative abundance of connective tissue in the muscle, which is likely to be confused 324 with fat during ultrasound image processing. 325 326 3.3. Correlations between ultrasound measurements and carcass traits 327 The Pearson correlation coefficients between in vivo ultrasound readings (UAL, 328 UAT, UFL, UIMFL and UIMFT) taken at 50, 25 and 1 days before slaughter and the 329 values of carcass and beef parameters are presented in Tables 3 and 4, respectively. In 330 overall terms, the ultrasound measurements taken in young bulls before slaughter 331 correlated with some carcass parameters (Table 3), while few significant correlations 332 with beef traits were observed (Table 4). 333 334 Ultrasound measurementsb Antemortem scan day Carcass weight Conformation scorea Fatness scorea KPFa CAa CFTa CIMFa 50 0.452 0.490* -0.228 -0.425* 0.473* -0.117 -0.533* UAL 25 0.203 0.511* -0.111 -0.477* 0.377 -0.420 -0.451* 1 0.160 0.520* -0.084 -0.464* 0.560* -0.268 -0.491* 50 0.386 0.485* -0.174 -0.529* 0.445 -0.273 -0.616* UAT 25 0.459* 0.703* -0.315 -0.635* 0.614* -0.472* -0.288 1 0.511* 0.738* -0.389 -0.574* 0.398 -0.558* -0.139 50 -0.223 -0.091 0.270 -0.054 -0.241 0.330 -0.012 17 UFL 25 -0.372 -0.080 0.232 -0.153 0.019 0.068 0.215 1 -0.074 -0.047 0.293 -0.116 0.434 -0.021 -0.417 50 -0.193 -0.263 -0.042 0.247 -0.159 0.073 0.302 UIMFL 25 -0.578* -0.540* 0.575 0.444* -0.111 -0.123 0.324 1 -0.413 -0.435 0.420 0.529* -0.263 0.328 0.538* 50 0.029 0.037 -0.189 0.636* 0.010 -0.072 0.535* UIMFT 25 -0.536* -0.441 0.439 0.556* -0.319 0.107 0.468* 1 -0.491* -0.542* 0.383 0.718* -0.394 0.457* 0.462* aText caption, see Table 1; bText caption, see Table 2; *: P < 0.05. 335 Table 3 Pearson coefficients of correlation between carcass and ultrasound measurements at 336 different scan days before slaughter. 337 338 Carcass weight was positively correlated with UAT, but negatively with 339 ultrasound measurements for fat traits (UFL, UIMFL and UIMFT) in most scans (Table 340 3). The significant correlations between a series of ultrasound carcass measurements 341 and carcass weight were high, with values over 0.459. These findings were in agreement 342 with those described by Peña et al. (2014) between slaughter weight and a series of 343 ultrasound measurements in the muscle area (r=0.503), or the intramuscular fat 344 percentage (r=-0.335) in young bulls of Retinta, Charolais and Limousine breeds. 345 In general, the correlations between ultrasound measurements for muscle traits 346 (UAL and UAT) were positively related to carcass conformation and CA measured in the 347 carcass, while negative correlations were observed between UAL or UAT and fat traits 348 of the carcass (KPF, CFT and CIMF). The strongest significant relationships (r = 0.485–349 0.738) were observed between the ultrasound measurements of longissimus lumborum 350 18 area, at the three pre-harvest times, and the conformation score. In fact, the correlation 351 tended to increase with the duration of the fattening phase. The role of longissimus 352 muscle area for predicting carcass yield has proved to be controversial (Wall et al., 353 2004). Numerous studies have tested its usefulness in predicting cuttability; May et al. 354 (2000) and Realini et al. (2001) correlated rib eye area both to sub-primal cut yields and 355 the weights of the major primals (r-value between 0.64 and 0.45). However, Crews et al. 356 (2002) and Wall et al. (2004) stated that the USDA yield grade is not heavily influenced 357 by the longissimus lumborum area. Hodgson et al. (1992) and Tait et al. (2005) showed 358 that the loin eye area had a low correlation coefficient with retail yield. 359 The UAL readings at 50, 25 and 1-days ante-mortem correlated negatively with 360 the percentage of KFT and CIMF. The highest correlations between muscular 361 ultrasound scans (UAL and UAT) and intramuscular fat content (r=-0.533 and -0.616, 362 respectively) were observed at early scanning times. Wall et al. (2004) obtained 363 regression coefficients ranging between 0.44 and 0.42 for back fat and 0.47 for the 364 degree of marbling estimates from UAL. The cited authors found no significant 365 improvement in estimates from scans taken closer to the day of slaughter, but it should 366 be noted that the scans were taken between 2 and 3.5 months prior to animal slaughter. 367 Renand and Fisher (1997) noted that if the back-fat estimates from ultrasound 368 measurements for a population showed a standard deviation of more than 3 mm, the 369 coefficients of correlation between the final and ultrasound values can reach 0.6, while 370 for a SD of less than 2 mm, they barely reach 0.3. 371 The ultrasound longissimus muscle depth reading (UAL and UAT) also appears to 372 be correlated, albeit negatively, with various carcass fatness measurements, such as the 373 degree of marbling and the KPF score (Table 3), revealing an inverse relationship 374 between muscular development and final fatness (Mendizabal et al., 1999; Alberti et al., 375 19 2005). The range of back fat thickness in Pirenaica calves (Table 1) was between 0.1 376 and 0.45 cm, with a mean value of 0.23 cm and a standard deviation of 0.08 cm, while 377 40% of the UFL estimates deviated by more than 0.25cm from CFT. 378 The grey level value of a scan depends on tissue composition and therefore on the 379 degree of marbling in the muscle. In general, positive correlations between ultrasound 380 intramuscular fat content (UIMFL and UIMFT) and fat trait were observed in the 381 carcasses. In fact, significant correlations at all scanning times were observed between 382 UIMFT and carcass traits, such as KPF or CIMF. The correlation increased significantly 383 between the first and last scans for UIMFT and KPF, while the correlation between 384 UIMFT and CIMF decreased from the first to the last scans (Table 3). These results are 385 not in agreement with Wall et al. (2004), who reported no changes in the correlations 386 between marbling and ultrasound scans made 100 days before slaughter to preslaughter 387 in fattened bulls from the Angus, Simmental, Red Angus and Charolais breeds. 388 389 3.4. Correlations between ultrasound measurements and beef quality traits 390 Table 4 shows the Pearson coefficients of correlation between the measurements 391 obtained from the in vivo scans (UAL, UAT, UFL, UIMFL and UIMFT) taken 50, 25 and 392 1 days prior to slaughter, the Warner-Braztler shear force scores and the sensory 393 parameters of the meat as evaluated by a trained panel. Fatty flavor was the only 394 sensory parameter with significant coefficients of correlation with the ultrasound area 395 (UAL) and ultrasound intramuscular fat (UIMFT) readings. Fatty flavor correlated 396 negatively with the in vivo UAL readings taken 50, 25 and 1 days prior to slaughter, and 397 UAT at 50 days ante-mortem (r=-0.616). Fatty flavor was correlated positively with 398 UIMFT at the three measurement times and only with UIMFL at 1 day prior to slaughter 399 20 (r=0.538; P<0.05). Correlations between fatty flavor and UAL or UIMFT decreased 400 constantly from 50 days before slaughter to the pre-slaughter scans. 401 402 Ultrasound measurementa Antemortem scan day WBSF Juiciness Hardness Beef flavor Liver flavor Fatty flavor 50 0.233 0.023 0.106 0.025 -0.337 -0.533* UAL 25 -0.247 0.081 -0.235 0.311 -0.225 -0.451* 1 -0.096 -0.060 -0.175 0.105 0.060 -0.491* 50 -0.002 0.101 -0.081 0.211 -0.186 -0.616* UAT 25 0.018 0.204 -0.176 0.046 0.036 -0.288 1 0.034 -0.047 -0.050 0.107 -0.096 -0.139 50 -0.308 -0.304 0.280 0.421 -0.324 -0.012 UFL 25 -0.365 -0.031 -0.065 0.361 -0.062 0.215 1 0.055 -0.253 -0.238 -0.130 -0.040 -0.417 50 0.077 -0.131 0.328 -0.119 0.243 0.302 UIMFL 25 -0.335 0.011 -0.236 -0.107 0.307 0.324 1 0.041 -0.061 0.256 0.049 0.177 0.538* 50 0.152 0.148 0.142 -0.161 0.361 0.535* UIMFT 25 -0.162 -0.009 -0.099 0.034 0.259 0.468* 1 0.098 -0.057 0.186 0.090 0.260 0.462* aText caption, see Table 2; WBSF: Warner–Bratzler shear force test; *: P < 0.05. 403 Table 4 Pearson coefficients of correlation between beef quality traits and ultrasound 404 measurements at different scanning days before slaughter. 405 21 406 407 3.5. Prediction equations 408 With the practical aim of saving time and costs, we identified the best moment for 409 scanning to obtain an acceptable prediction. As regards the estimation of the beef 410 parameters, the best significant equations (P<0.05 and R2>0.5) to predict the percentage 411 of kidney and pelvic fat, intramuscular fat content, carcass conformation, carcass fatness 412 and the fatty flavor of the meat are shown in Table 5. The results of the regression 413 equations included the scores obtained from the in vivo ultrasound scans taken 50, 25 414 and 1 days prior to slaughter. The best prediction equations for KPF and carcass fatness 415 score were at 1 day prior to slaughter, while those for conformation carcass were 416 obtained at 25 days prior to slaughter. The most favorable prediction of intramuscular 417 fat content was 50 days prior to slaughter, since at this time the ultrasound measurement 418 was more stable than that performed one day before the animals were slaughtered. This 419 observation can be seen in Table 2 in which the ranges of dispersion between minimum 420 and maximum observed in the gray levels in UIMFL and UIMFT were higher when the 421 measurements were made one day before slaughtering of the calves than those observed 422 when the ultrasound measurements were carried out 50 days before the animals were 423 slaughtered. This finding suggests that better carcass parameter estimates result from 424 scans taken in the last month prior to slaughter, but not necessarily the day before 425 harvest. 426 22 427 Dependent Variables*, a Best antemortem scan dayb R2 P-value RMSE Best prediction equationsb KPF (%) 1 0.717 0.002 0.962 KPF (%) = -5.27-4.16E-2*UAL1+4.80E-2*UAT1+0.60*UFL1-2.91E-2*UIMFL1+0,11*UIMFT1 CIMF (%) 50 0.619 0.011 0.436 IMF (%) = 0.96-5.39E-3*UAL50-1.47E-2*UAT50+0.29*UFL50-1.51E-2*UIMFL50+4.37E-2*UIMFT50 Conformation score 25 0.738 0.001 0.556 Conformation score= 5.0-1.60E-2*UAL25+0.06*UAT25+0.14*UFL25-4.18E-2*UIMFL25+0.02*UIMFT25 Fatness score 1 0.550 0.031 0.395 Fatness score= 1.40+9.11E-3*UAL1-1.29E-3*UAT1+0.27*UFL1+6.42E-3*UIMFL1+0.01*UIMFT1 Fatty flavor 1 0.556 0.029 0.470 Fatty flavor = 2.89-4.02E-2*UAL1+2.47E-2*UAT1-4.26* E-2UFL1+9.16E-3*UIMFL1+6.45E-3* UIMFT1 *Only significant regression equations (P<0.05 and R2>0.5) are shown. 428 aText caption, see Table 1; bText caption, see Table 2; RMSE: Root mead square error. 429 430 Table 5 Best prediction equations for young bull carcass characteristics and beef quality traits from ultrasound readings at 50, 25 and 1-days ante-mortem.431 23 The regression coefficients (R2) for the fat content in carcass and beef (KPF, CIMF 432 and fatness score) were higher than 0.5 (R2=0.717, 0.619 and 0.550 respectively) as 433 reported also by Wall et al. (2004). The best prediction of the percentage of pelvic and 434 kidney fat containing UAL, UAT, UFL, UIMFL and UIMFT was observed at 1 day prior 435 to slaughter. All independent variables accounted for 71.7% of the variation in KPF. 436 The R2 values for the regression equations to estimate CIMF were in line with Aass 437 et al. (2009) and Castilhos et al. (2018), who indicated that ultrasound constitute an 438 accurate method for predicting in vivo intramuscular fat content in lean breeds. In fact, 439 in this study, the R2 values were higher (R2= 0.619) than those reported by Chambaz et 440 al. (2003) (R2 = 0.420), but in agreement with the ones reported by Aass et al. (2009) 441 (R2 = 0.670) in lean beef. It can be therefore considered that in lean breeds, 442 intramuscular fat can be predicted early in young bulls because intramuscular fat depot 443 is a late-maturing tissue and changes cannot be expected at an early slaughter age. 444 With reference to sensorial variables, the best prediction equation for fatty flavor of 445 beef (R2=0.556; P=0.029) was obtained 1 day before slaughter. In accordance with 446 Table 4, fatty flavor can be predicted from in vivo scans, using the combination of all 447 the ultrasound measurements from the scan taken 1 day prior to slaughter, including a 448 combination of UAL, UAT, UFL, UIMFL and UIMFT measurements, which account for 449 56% of the variation in the fatty flavor of beef from young bulls. 450 In early scans, fat content could determine fatty flavor, while difficulties in 451 predicting sensory parameters related to muscle development at different scanning times 452 can be observed. A higher intramuscular fat content should increase oiliness, which 453 plays a role in increased juiciness and causes meat to be perceived as more tender, as 454 noted by Savell and Cross (1988). In addition, fat tissue is the main depot for the 455 precursors of the volatile compounds produced when cooking beef, through Maillard’s 456 24 browning reaction and the oxidation of lipids (Mottram, 1998). The fat content in 457 muscle correlates significantly with aroma intensity (Savell and Cross 1988), so that 458 pleasant-tasting beef will tend to be fattier, with a higher degree of marbling (Berry et 459 al., 1980). In fact, there is some controversy over the fat percentage in muscle required 460 to achieve the optimal flavor intensity: Denoyelle (1995) defended a minimum of 4% of 461 fat in muscle, whereas Savell and Cross (1988) lowered the percentage to 3%. However, 462 Spanish beef is characterized by a low fatness level and marketed carcasses show less 463 than 3-4% intramuscular fat in the longissimus lumborum muscle (Indurain et al., 464 2006). The animals used in the current study had a low percentage of intramuscular fat 465 in the longissimus lumborum muscle (0.85 % of fresh meat) and, even at this low 466 percentage, fatness would seem to have a positive impact on beef palatability. As in the 467 estimation of the KPF and fatness scores, the best scans for predicting fatty flavor 468 sensory quality are taken just prior to slaughter, although the best prediction of 469 intramuscular fat could be propossed early. 470 Table 6 shows the best ultrasound variables to contribute to an acceptable prediction 471 of carcass and beef parameters in young bulls from the Pirenaica breed. To predict KPF, 472 the conformation and fatness score from two ultrasound measurement scans accounted 473 for the highest variation, while for all the other parameters, only one ultrasound 474 measurement accounted for the highest variation. 475 476 Dependent variables *, a Independent variablesb R2 P-value RMSE KPF (kg) UIMFT1; UFL1 0.621 <0.001 1.010 CA (cm2) UAT25 0.408 0.002 11.134 CFT (cm) UIMFT1 0.210 0.042 0.078 CIMF (%) -UAT50 0.306 0.011 0.519 Conformation score UAT25; -UIMFL25 0.689 < 0.001 0.550 25 Fatness score UIMFL1; UFL1 0.462 0.005 0.390 WBSF (kg/cm2) -UFL25 0.225 0.035 1.394 Fatty flavor UIMFL1 0.279 0.017 0.529 *Only significant variables (P<0.05) are showing. 477 aText caption, see Table 1; bText caption, see Table 2; Represented values scan at 50, 25 and 1-days 478 before slaughter; RMSE: Root mead square error. 479 Table 6 Regression to predict young bull carcass and beef quality traits corresponding to the 480 selected ultrasound measurements. 481 To predict KPF, CFT, carcass conformation, carcass fatness, WBSF and fatty 482 flavor, ultrasound measurements from fat are the main requirements. However, to 483 predict CA, CIMF and carcass conformation, what is needed is mainly ultrasound 484 measurements from the muscular area (UA). 485 In general, the best ultrasound measurement to determinate fat traits (KPF, CFT, 486 fatness score and fatty flavor) were determined close to slaughter (1 day pre-slaughter), 487 while the best scan time to determine CA, conformation and intramuscular fat content 488 was a longer time before slaughter. In KPF prediction, the contribution of 489 measurements related to fat content at an early pre-slaughter time (UIMFT and UFL) 490 were positive (R2=0.621; P<0.001). In CFT, the contribution of UIMFT prediction was 491 positive too (R2=0.210; P<0.042). In the case of carcass fatness prediction, UIMFL and 492 UFL both contributed positively (R2=0.462; P<0.005). The contribution of UIMFL to 493 predict fatty flavor in beef at 1 day pre-slaughter of the young bulls was positive 494 (R2=0.279; P<0.017) (Table 6). At 25 days pre-slaughter, the contribution of UAT was 495 positive (R2=0.408; P<0.002) to predict the longissimus lumborum area. To predict 496 carcass conformation, UAT (positively) and UIMFL (negatively) both contributed to 497 explaining the best prediction equation (R2=0.689; P<0.001). As regards the prediction 498 of WBSF, the UFL measurement at 25 days pre-slaughter contributed to explaining 499 32 de Vacuno del País Vasco» o «Euskal Okela», «Ternera Asturiana» y «Carne de 611 Cantabria») 612 DOUE. 2008. Commission Regulation (EC) No 1249/2008 of 10 December 2008 laying 613 down detailed rules on the implementation of the Community scales for the 614 classification of beef, pig and sheep carcases and the reporting of prices thereof. 615 L 337/3. 616 Eriksson, S., Näsholm, A., Johansson, K., Philipsson, J., 2002. Genetic analysis of post-617 weaning gain of Swedish beef cattle recorded under field conditions and at 618 station performance testing. Livest. Prod. Sci. 76, 91–101. 619 https://doi.org/10.1016/S0301-6226(02)00007-6 620 Folch, J., Less, M., Sloane, G.H., 1957. A simple method for the isolation and 621 purification of total lipids from animal tissues. J Biol. Chem. 226, 497–509. 622 PMID: 13428781 623 Gorraiz, C., Berian, M.J., Chasco, J., Iraizoz, M., 2000. Descriptive analysis of meat 624 from young ruminants in Mediterranean systems. J. Sens. Stud. 15, 137–150. 625 doi: 10.1111 / j.1745-459X.2000.tb00261.x 626 Greiner, S.P., Rouse, G.H., Wilson, D.E., Cundiff, L.V., Wheeler, T.L., 2003. 627 Prediction of retail product weight and percentage using ultrasound and carcass 628 measurements in beef cattle. J. Anim. Sci. 81, 1736–1742. doi: 10.2527 / 629 2003.8171736x 630 Hamlin, K.E., Green, R.D., Perkins, T.L., Cundiff, L.V., Miller, M.F., 1995a. Real-time 631 ultrasonic measurement of fat thickness and longissimus muscle area: I. 632 Description of age and weight effect. J. Anim. Sci. 73, 1713–1724. doi: 10.2527 633 / 1995.7361713x 634 33 Hamlin, K.E., Green, R.D., Cundiff, L.V., Wheeler, T.L., Dikeman, M.E., 1995b. 635 Realtime ultrasonic measurement of fat thickness and longissimus muscle area: 636 II. Relationship between real-time ultrasound measures and carcass retail yield. 637 J. Anim. Sci. 73, 1725–1734. doi: 10.2527 / 1995.7361713x 638 Hodgson, R.R., Belk, K.E., Savell, J.W., Cross, H.R., Williams, F.L., 1992. 639 Development of a multivariate yield grade equation to predict compositional 640 traits in mature cow carcasses. J. Anim. Sci. 70, 2159–2166. doi: 10.2527 / 641 1992.7072159x 642 Hwang, H., Park, B., Nguyen, M., Chen, Y.R., 1997. Hybrid image processing for 643 robust extraction of lean tissue on beef cut surfaces. Comput. Electron. Agric. 644 17, 281–294. https://doi.org/10.1016/S0168-1699(97)01321-5 645 Indurain, G., Beriain, M.J., Goñi, M.V., Arana, A., Purroy, A. 2006. Composition and 646 estimation of intramuscular and subcutaneous fatty acid composition in Spanish 647 young bulls. Meat Sci. 73, 326–334. 648 https://doi.org/10.1016/j.meatsci.2005.12.007 649 Indurain, G., Carr, T.R., Goñi, M.V., Insausti, K., Alfonso, L., Beriain, M.J., 2009. The 650 relationship of carcass measurements to carcass composition and intramuscular 651 fat in Spanish beef. Meat Sci. 82, 155–161. 652 https://doi.org/10.1016/j.meatsci.2009.01.005 653 ISO 8586-2014. Sensory analysis- General guidelines for the selection, training and 654 monitoring of selected assessors and expert sensory assessors. 655 May, S.G., Mies, W.L., Edwards, J.W., Harris, J.J., Morgan, J.B., Garrett, R.P., 656 Williams, F.L., Wise, J.W., Cross, H.R., Savell, J.W., 2000. Using live estimates 657 and ultrasound measurements to predict beef carcass cuttability. J. Anim. Sci. 658 78, 1255–1261. doi: 10.2527 / 2000.7851255x 659 34 Meilgaard, M., Civille, G.V., Carr B.T., 1991. Sensory evaluation techniques. 3 th 660 edition. Boca Raton, Florida, CRC Academia Press. 416 p. 661 https://doi.org/10.1201/9781003040729 662 Mendizabal, J.A., Albertí, P., Eguinoa, P., Arana, A., Soret, B., Purroy A., 1999. 663 Adipocyte size and lipogenic enzyme activities in different adipose tissue depots 664 in steers of local Spanish breeds. Anim. Sci. 69, 115–121. 665 doi: https://doi.org/10.1017/S1357729800051146 666 Mottram, D.S. 1998. Flavour formation in meat and meat products: a review. Food 667 Chem. 62, 415–424. https://doi.org/10.1016/S0308-8146(98)00076-4 668 Nogalski, Z., Pogorzelska-Przybyłek, P., Białobrzewski, I., Modzelewska-Kapituła, M., 669 Sobczuk-Szul, M., Purwin, C., 2017. Estimation of the intramuscular fat content 670 of m. longissimus thoracis in crossbred beef cattle based on live animal 671 measurements. Meat Sci. 125, 121–127. 672 https://doi.org/10.1016/j.meatsci.2016.11.026 673 OJEU. 2006. (Official Journal of the European Union). 2006. L 17/6 (January 21st, 674 2006) 675 Peña, F., Molina, A., Juárez, M., Requena, F., Avilés, C., Santos, R., Domenech, V., 676 Horcada, A., 2014. Use of serial ultrasound measures in the study of growth- 677 and breed-related changes of ultrasonic measurements and relationship with 678 carcass measurements in lean cattle breeds. Meat Sci. 96, 247–255. 679 https://doi.org/10.1016/j.meatsci.2013.07.012 680 Perkins, T.L., Green, R.D., Hamlin, K.E., 1992a. Evaluation of ultrasonic estimates of 681 carcass fat thickness and longissimus muscle area in beef cattle. J. Anim. Sci. 682 70, 1002–1010. doi: 10.2527 / 1992.7041002x 683 35 Perkins, T.L., Green, R.D., Hamlin, K.E., Shepard, H.H., Miller, M.F., 1992b. 684 Ultrasonic prediction of carcass merit in beef cattle: Evaluation of technical 685 effects on ultrasonic estimates of carcass fat thickness and longissimus muscle 686 area. J. Anim. Sci. 70, 2758–2765. https://doi.org/10.2527/1992.7092758x 687 Piedrafita, J., Quintanilla, R., Sañudo, C., Olleta, J.L., Campo, M.M., Panea, B., 688 Renand, G., Turin, F., Jabet, S., Osoro, K., Olivan, M.C., Noval, G., Garcıa, P., 689 Garcıa, M.D., Oliver, M.A., Gispert, M., Serra, X., Espejo, M., García, S., 690 López, M., Izquierdo, M., 2003. Carcass quality of 10 beef cattle breeds of the 691 Southwest of Europe in their typical production systems. Livest. Product. Sci. 692 82, 1–13. https://doi.org/10.1016/S0301-6226(03)00006-X 693 Realini, C.E., Williams, R.E., Pringle, T.D., Bertrand J.K., 2001. Gluteous medius and 694 rump fat as additional live animal ultrasound measurements for predicting retail 695 product and trimmable fat in beef carcasses. J. Anim. Sci. 79, 1378–1385. 696 doi: 10.2527 / 2001.7961378x 697 Renand, G., Fisher, A.V., 1997. Comparison of methods for estimating carcass fat 698 content of young Charolais bulls in performance testing station. Livest. Product. 699 Sci. 51, 205–213. https://doi.org/10.1016/S0301-6226(97)00060-2 700 Savell, J.W., Cross, H.R., 1988. The role of fat in the palatability of beef, pork and 701 lamb. In: Designing foods: animal product options in the marketplace (pp. 345-702 355). Washington DC: National Academic Press. 703 Smith, M.T., Oltjen, J.W., Dolezal, H.G., Gill, D.R., Behrens, B.D., 1990. Evaluation of 704 real-time ultrasound for predicting carcass traits of feedlot steers. Oklahoma 705 Agric. Exp. Sta. Res. Rep. MP-129, p. 374 706 36 Tait Jr., R.G., Wilson, D.E., Rouse, G.H., 2005. Prediction of retail product and 707 trimmable fat yields from the four primal cuts in beef cattle using ultrasound or 708 carcass data. J. Anim. Sci. 83, 1353–1360. doi: 10.2527 / 2005.8361353x 709 UNE 87004-1979. Guide for the installation of a tasting room. CTN 87 - Sensory 710 analysis. UNE Spanish Standardization. 711 Wall, P.B., Rouse, G.H., Wilson, D.E., Tait Jr., R.G., Busby, W.D., 2004. Use of 712 ultrasound to predict body composition changes in steers at 100 ad 65 days 713 before slaughter. J. Anim. Sci. 82, 1621–1629. 714 https://doi.org/10.2527/2004.8261621x 715 Waldner, D.N., Dikeman, M.E., Schalles, R.R., Olson, W.G., Houghton, P.L., 716 Unruh, J.A., Corah, L.R., 1992. Validation of real-time ultrasound technology 717 for predicting fat thicknesses, longissimus muscle areas, and composition of 718 Brangus bulls from 4 months to 2 years of age. J. Anim. Sci. 70, 3044–3054. 719 https://doi.org/10.2527/1992.70103044x 720 Yang, X.J., Albrecht, E., Ender, K., Zhao, R.Q., Wegner, J., 2006. Computer image 721 analysis of intramuscular adipocytes and marbling in the longissimus muscle of 722 cattle. J. Anim. Sci. 84, 3251–3258. doi: 10.2527 / jas.2006-187 723 1 Effectiveness of using ultrasound readings to predict carcass traits and 1 sensory quality in young bulls 2 3 M. J. Beriaina , K. Insaustia, M. Valerab, G. Induraina, A. Purroya, T. R. Carrc, and A. 4 Horcadab,* 5 6 aResearch Institute for Innovation & Sustainable Development in Food Chain 7 (ISFOOD), Universidad Pública de Navarra, 31006 Pamplona, Spain; bDepartamento de 8 Ciencias Agroforestales, ETSIA, Universidad de Sevilla, 41013, Sevilla, Spain; 9 cCollege of Agricultural, Consumers and Environmental Sciences. Urbana, Illinois 10 61801, USA. 11 12 *Corresponding author: Alberto Horcada 13 e-mail address: [email protected] (A. Horcada) 14 15 16 2 Abstract 17 The objectives of this study were to develop ultrasound-derived prediction equations in 18 young bulls before slaughter to estimate carcass traits and beef sensory parameters and 19 to determine the optimum moment pre-slaughter to estimate the sensory quality of beef 20 from young bulls. Ultrasound images were measured in twenty live young bulls at 50, 21 25 and 1 days prior to slaughter. Intramuscular fat content, longissimus lumborum 22 muscle area, carcass fatness, texture and sensory analysis were measured by a trained 23 panel after slaughter. Partial least square methodology was used to find the relationship 24 between the ultrasound measurements and the dependent variables, such as carcass 25 traits, Warner-Bratzler shear force and the sensory profile. Additionally, a stepwise 26 procedure was used to select the most informative ultrasound variables in the prediction 27 equations for carcass and beef sensory traits. The results indicate that ultrasound testing 28 in the feedlot offers promising potential. Early ultrasound scan measurements were 29 useful during fattening to predict intramuscular fat content (R2=0.619, RMSE=0.44%, 30 50 days prior to slaughter), while ultrasound scans taken close to slaughter were useful 31 to predict kidney fat content (R2=0.717, RMSE=0.96%, 1 day prior to slaughter). 32 However, the prediction of sensory beef attributes was only useful for fatty flavor 33 (R2=0.556, RMSE=0.47%) at 1 day pre-slaughter. Thus, the prediction of fat parameters 34 using ultrasound measurements could constitute a valuable tool in the process of 35 selecting beef quality traits in young bulls before slaughter. 36 37 Key words: Beef, carcass prediction, grading, palatability, ultrasound 38 39 1. Introduction 40 Using ultrasound technology has been reported previously in cattle breeding 41 programs and for production purposes (Baker et al., 2006; Castilhos et al., 2018). The 42 3 application of this technology, not only on carcasses at the slaughterhouse but also on 43 live animals in farms, has already become a useful tool for beef producers, as it enables 44 them to match selection decisions to market demands. 45 Current carcass grading systems rely on straightforward visual evaluation of 46 carcass parameters. The European Beef Carcass Grading system is based on carcass 47 conformation and fatness (OJEU, 2006), while in the USA, it is focused on carcass yield 48 and intramuscular fat level, which are likely to have an impact on the sensory 49 acceptance by consumers. In addition, until now, carcass classification methods have 50 been performed after the animals are slaughtered. However, in order to slaughter 51 animals at the optimum moment and to make an early pre-slaughter classification of the 52 carcass and meat quality, methods to predict carcass characteristics and carcass quality 53 are required. 54 Among the most widely-used techniques in live animals is ultrasound (Wall et al., 55 2004; Peña et al., 2014) combined with image analysis (Hwang et al., 1997; Cannell et 56 al. 2002), which allows for the successful prediction of carcass parameters such as 57 muscle area or back fat thickness (Hamlin et al. 1995a; 1995b). Nogalski et al. (2017) 58 reported that both selected biometric ultrasound measurements (back fat and thickness 59 of subcutaneous rump) and selected blood parameters, such as triglycerides, could be 60 used to predict intramuscular fat content with satisfactory precision and accuracy. These 61 techniques, which can be used to estimate beef marbling, can help producers to fulfill 62 industry and consumer demands (Indurain et al., 2009). 63 The ultrasound technique has also been shown to be useful in determining the 64 nutritional quality of beef. In fact, Indurain et al. (2006) reported that the ultrasound 65 readings reflect the effect of fatness on fat composition (e.g. the fatty acid profile of 66 4 subcutaneous fat). However, there is a lack of research to demonstrate the efficiency of 67 ultrasound in estimating beef palatability. 68 Regarding beef production, Spain has a number of local breeds characterized by 69 their high muscle growth and reduced fat content. These breeds include Pirenaica, 70 which is well suited for beef production. Pirenaica is an early maturing beef breed, 71 producing medium-muscled lean carcasses. It accounts for 1.1 % of the Spanish national 72 cattle stock and it is the most abundant breed for beef production under the European 73 Union's PGI Ternera de Navarra [Navarre Veal] Protected Geographical Indication 74 (DOUE, 2004). Pirenaica breed cattle are raised in a semi-extensive system near the 75 mountains in the Pyrenees (Northern Spain). The animals take advantage of the 76 mountain pastures in spring and summer, while they remain in the valleys during the 77 winter. Young bulls are weaned at six months and are fed in feedlots with concentrate 78 and cereal straw until slaughter at approximately twelve months of age to obtained lean 79 carcasses. 80 In this context, the main objective of this study was to develop ultrasound-derived 81 prediction equations before slaughter to estimate carcass traits and beef sensory 82 parameters in young bulls with lean carcasses. 83 84 2. Material and methods 85 2.1. Pre-harvest, handling, and harvesting procedure 86 A total of 20 yearling Pirenaica bulls under the Protected Geographical Indication 87 Ternera de Navarra were used. The young bulls were reared in a semi-extensive system 88 in the mountains until weaning. After weaning at approximately seven-eight months of 89 age, they were fed on a concentrate in a local feedlot (85% barley, 10% soybean meal, 90 3% vegetal fat and 2% minerals and vitamins), and barley straw, both ad libitum. The 91 5 chemical composition of the feeds corresponds to a medium-energy diet: Dry matter 92 (88.2%); Crude protein (128.5g/kg DM); Crude fiber (48.4g/kg DM); Ash (53.7g/kg 93 DM); Metabolizable energy (11.5 MJ/kg DM). The average feed consumption during 94 the young bulls’ growth period (four months) was 7.9 kg/day. The animals were 95 considered as a representative sample of the market, and they are commonly slaughtered 96 in the area. The bulls were raised according to the handling conditions in the Spanish 97 rules and regulations for animal care (Directive 2010/63/EU) and slaughtered according 98 to Spanish rules and regulations for animal care (Council Regulation EC 1099/2009). 99 The research followed the official guidelines for the humane treatment, care and 100 handling of animals. 101 102 2.2. Ultrasound data collection 103 The live animals were weighed and two cross-sectional ultrasound images (right and 104 left side) between the 12th and 13th ribs were taken in the live animals at the feedlot, as 105 described by Wall et al. (2004) and Bergen et al. (2005). The ultrasound images were 106 measured on the skin after clipping the hair at 50, 25 and 1 days prior to slaughter, using 107 Sonovet 600-real-time ultrasound equipment (Madison Co. Ltd. Korea) equipped with a 108 linear probe (3.5 MHz, 120 x 20 mm). Mineral oil (Echoultragel; Pirrone & Co.SPA, 109 Italy) at 20-25oC was used to ensure suitable acoustic contact between the probe and the 110 skin. 111 The linear ultrasound measurements recorded on the live young bulls were: 112 intramuscular fat content or marbling (IMF), longissimus muscle depth and dorsal fat 113 thickness. The resulting two-dimensional images (longitudinal and transversal) were 114 digitalized and stored. The young bulls were scanned twice and the images evaluated 115 and interpreted by a laboratory technician at the Public University of Navarra, using the 116 12 average 104.44 cm2 and 0.23 cm, respectively) were in line with those reported by Peña 244 et al. (2014) in the Retinta breed and other continental beef breeds (Charolais and 245 Limousine). Moreover, the longissimus lumborum intramuscular fat content (average 246 0.85%) was in the range of light carcasses on the South European market. These 247 findings agree with those of Piedrafita et al. (2003) and Albertí et al. (2008) in calves 248 reared under the same production systems. 249 Table 1 shows the sensory attributes of the meat as determined by the trained panel. 250 From these results, the meat from these young bulls could be described as moderately 251 juicy and tender, with a not very intense beef flavor and a slight taste of liver. The 252 assessments given by the trained panel were similar to those reported in previous studies 253 (Beriain et al., 2016), which reported Pirenaica beef as having a medium degree of 254 juiciness, tenderness, flavor and a reduced fat content. 255 256 3.2. Ultrasound measurements 257 Table 2 shows the measurements of the longissimus lumborum muscle area (UAL 258 and UAT) and UFL obtained by means of the scans on live animals 50, 25 and 1 days 259 prior to slaughter. The scans were also used to measure gray level (UIMFL and UIMFT) 260 to determine the longissimus lumborum intramuscular fat content at the 12th rib (grey 261 scale 0-150). The influence of the moment the ultrasound measurements were taken on 262 differences was significant (P <0.05) for UAT. In fact, the lowest value of UAT was 263 observed at 50 days ante-mortem, while after 25 days ante-mortem, no differences in 264 UAT measurements were observed. 265 13 50 days ante-mortem 25 days ante-mortem 1 day ante-mortem P-value Mean SD Min. Max. Mean SD Min. Max. Mean SD Min. Max. SEM UAL (cm2) 74.78 10.07 58.73 88.36 75.52 11.27 45.99 95.52 77.39 9.73 56.81 98.14 6.32 0.367 UAT (cm2) 86.28a 13.24 63.41 104.16 93.62b 13.12 72.84 117.47 94.88b 14.36 76.15 127.30 8.52 0.008 UFL (cm) 0.47 0.10 0.26 0.98 0.48 0.09 0.35 0.71 0.46 0.13 0.25 0.60 0.46 0.331 UIMFL (grey level)† 59.72a 16,81 29.42 95.03 49.72b 16,78 27.62 87.97 50.80b 24,52 7.53 97.71 13.90 0.038 UIMFT (grey level) † 69.89a 13,79 39.37 92.04 64.09b 14.77 35.53 109.24 64.08b 21,95 20.23 101.11 10.99 0.016 SD: Standard deviation; SEM: Standard error of the mean; UAL: ultrasound of longissimus lumborum area at the 12th rib from longitudinal ultrasound measurement; UAT: 266 ultrasound of longissimus lumborum area at the 12th rib from transversal ultrasound measurements; UFL = ultrasound of 12th fat thickness from longitudinal ultrasound 267 measurement; UIMFL: ultrasound of longissimus lumborum intramuscular fat content at the 12th rib from longitudinal ultrasound measurement; UIMFT: ultrasound of 268 longissimus lumborum intramuscular fat content at the 12th rib from transversal ultrasound measurement; ns: not significant; sig: significant; a,b: Marked differences are 269 significant at P <0.05 in the same line; †Grey level of ultrasound image scale 0-150. 270 Table 2 Least mean squares and fisher-test means for ultrasound data collected in young bulls at 50, 25- and 1-days ante-mortem. 271 14 The longissimus lumborum area at the 12th rib measured with a grid (104.44 cm2; 272 Table 1) was greater than that predicted by the in vivo ultrasound images (which ranged 273 from 74.78 to 94.88cm2; Table 2). This is in line with findings published by Bergen et 274 al. (2005), who reported a tendency for the ultrasound measurements of the loin area to 275 be lower than the final carcass values. In contrast, Greiner et al. (2003) stated that in 276 vivo ultrasound measuring techniques can overestimate carcass loin areas by between 277 3.31 and 6.76 cm2 in young bulls with low musculature. On the other hand, Smith et al. 278 (1990) reported that the CA of bulls with a longissimus lumborum area around 104 cm2 279 was generally underpredicted, whereas in cattle with a longissimus lumborum area 280 lower than 85 cm2, it was usually overpredicted. The fact that the cited authors 281 performed their measurements on carcasses with a longissimus lumborum area of 282 approximately 80 cm2, while the Spanish cattle used in this study had bigger muscles 283 and therefore a larger loin area (> 100 cm2), might, at least partially, explain the greater 284 differences found in the present study. 285 The UFL means across the scanning time are presented in Table 2. The mean values 286 across the scanning time (0.47 cm) were in the range of those reported by Peña et al. 287 (2014) in a long scanning period from 212 days ante-mortem for continental breeds 288 such as Retinta, Charolais and Limousine, reared under similar conditions to the 289 Pirenaica breed. No variations in the average UFL values between the first and the last 290 scans were observed (P>0.05). The range of UFL values for carcass back fat thickness 291 observed by ultrasound (from 0.46 to 0.48 cm; Table 2) was greater than carcass 12th-rib 292 fat thickness measured with a caliber (0.23 cm; Table 1). These results are in agreement 293 with findings by Bergen et al. (2005) who found that the in vivo ultrasound values were 294 higher than carcass measurements when the differences were >0.5 cm. In fact, Peña et 295 15 al. (2014) showed an overestimation of ±0.2 cm in 98.9% of the bulls’ UFL vs CFT in 296 young bulls of continental breeds. 297 In contrast, Greiner et al. (2003) and Baker et al. (2006) found an underestimation 298 of UFL values in young bulls when the back fat thickness was greater (0.7 cm). 299 According to Waldner al. (1992), an improvement in the accuracy of the relationship 300 between ultrasound and carcass measurements during the animals’ maturing period up 301 to 16 months can be observed. Perkins et al. (1992a) stated that the accuracy of the 302 ultrasound measurements of back fat thickness is highly dependent on the degree of rib 303 fatness during the animals’ growth. On the one hand, the misinterpretation of this 304 accuracy could occur in fatter cattle, in contrast to the improved interpretation in lighter 305 carcasses. In the present study, in 60% of the animals, the differences between the last 306 UFL measurement and CFT were < 0.2 cm. This accuracy is greater than that recorded 307 by May et al. (2000) in fatter young bulls. This result may be explained because the 308 animals used in the present research were lean and had a very thin layer of back fat 309 (Perkins et al., 1992b; Charagu et al., 2000). 310 Ultrasound grey level for the three pre-harvest periods was used to compare 311 intramuscular fat content in beef during the growth period (Table 2). A significant 312 decrease in the grey level in longitudinal and transversal images (around 15%) was 313 observed between the first and last scans. No significant differences in grey levels were 314 observed (P>0.05) during the last scanning period (between 25 and 1-days ante-315 mortem). This finding was in agreement with Albrecht et al. (2006), who observed a 316 relative decrease in intramuscular fat in young bulls during the fattening period up to 317 the age of twelve months in Angus steers and a marked increase in intramuscular fat 318 deposition from 12 to 14 months of age. However, Yang et al. (2006) affirmed that 319 there was no clear relationship between intramuscular fat content and body weight in 320 16 young bulls from lean breeds. In fact, Albrecht et al. (2006) reported that the connective 321 tissue in the muscle is often present in strong cords which are not streaked with fat, and 322 the highest UIMF observed in the early scans of young bulls could be attributed to the 323 relative abundance of connective tissue in the muscle, which is likely to be confused 324 with fat during ultrasound image processing. 325 326 3.3. Correlations between ultrasound measurements and carcass traits 327 The Pearson correlation coefficients between in vivo ultrasound readings (UAL, 328 UAT, UFL, UIMFL and UIMFT) taken at 50, 25 and 1 days before slaughter and the 329 values of carcass and beef parameters are presented in Tables 3 and 4, respectively. In 330 overall terms, the ultrasound measurements taken in young bulls before slaughter 331 correlated with some carcass parameters (Table 3), while few significant correlations 332 with beef traits were observed (Table 4). 333 334 Ultrasound measurementsb Antemortem scan day Carcass weight Conformation scorea Fatness scorea KPFa CAa CFTa CIMFa 50 0.452 0.490* -0.228 -0.425* 0.473* -0.117 -0.533* UAL 25 0.203 0.511* -0.111 -0.477* 0.377 -0.420 -0.451* 1 0.160 0.520* -0.084 -0.464* 0.560* -0.268 -0.491* 50 0.386 0.485* -0.174 -0.529* 0.445 -0.273 -0.616* UAT 25 0.459* 0.703* -0.315 -0.635* 0.614* -0.472* -0.288 1 0.511* 0.738* -0.389 -0.574* 0.398 -0.558* -0.139 50 -0.223 -0.091 0.270 -0.054 -0.241 0.330 -0.012 17 UFL 25 -0.372 -0.080 0.232 -0.153 0.019 0.068 0.215 1 -0.074 -0.047 0.293 -0.116 0.434 -0.021 -0.417 50 -0.193 -0.263 -0.042 0.247 -0.159 0.073 0.302 UIMFL 25 -0.578* -0.540* 0.575 0.444* -0.111 -0.123 0.324 1 -0.413 -0.435 0.420 0.529* -0.263 0.328 0.538* 50 0.029 0.037 -0.189 0.636* 0.010 -0.072 0.535* UIMFT 25 -0.536* -0.441 0.439 0.556* -0.319 0.107 0.468* 1 -0.491* -0.542* 0.383 0.718* -0.394 0.457* 0.462* aText caption, see Table 1; bText caption, see Table 2; *: P < 0.05. 335 Table 3 Pearson coefficients of correlation between carcass and ultrasound measurements at 336 different scan days before slaughter. 337 338 Carcass weight was positively correlated with UAT, but negatively with 339 ultrasound measurements for fat traits (UFL, UIMFL and UIMFT) in most scans (Table 340 3). The significant correlations between a series of ultrasound carcass measurements 341 and carcass weight were high, with values over 0.459. These findings were in agreement 342 with those described by Peña et al. (2014) between slaughter weight and a series of 343 ultrasound measurements in the muscle area (r=0.503), or the intramuscular fat 344 percentage (r=-0.335) in young bulls of Retinta, Charolais and Limousine breeds. 345 In general, the correlations between ultrasound measurements for muscle traits 346 (UAL and UAT) were positively related to carcass conformation and CA measured in the 347 carcass, while negative correlations were observed between UAL or UAT and fat traits 348 of the carcass (KPF, CFT and CIMF). The strongest significant relationships (r = 0.485–349 0.738) were observed between the ultrasound measurements of longissimus lumborum 350 18 area, at the three pre-harvest times, and the conformation score. In fact, the correlation 351 tended to increase with the duration of the fattening phase. The role of longissimus 352 muscle area for predicting carcass yield has proved to be controversial (Wall et al., 353 2004). Numerous studies have tested its usefulness in predicting cuttability; May et al. 354 (2000) and Realini et al. (2001) correlated rib eye area both to sub-primal cut yields and 355 the weights of the major primals (r-value between 0.64 and 0.45). However, Crews et al. 356 (2002) and Wall et al. (2004) stated that the USDA yield grade is not heavily influenced 357 by the longissimus lumborum area. Hodgson et al. (1992) and Tait et al. (2005) showed 358 that the loin eye area had a low correlation coefficient with retail yield. 359 The UAL readings at 50, 25 and 1-days ante-mortem correlated negatively with 360 the percentage of KFT and CIMF. The highest correlations between muscular 361 ultrasound scans (UAL and UAT) and intramuscular fat content (r=-0.533 and -0.616, 362 respectively) were observed at early scanning times. Wall et al. (2004) obtained 363 regression coefficients ranging between 0.44 and 0.42 for back fat and 0.47 for the 364 degree of marbling estimates from UAL. The cited authors found no significant 365 improvement in estimates from scans taken closer to the day of slaughter, but it should 366 be noted that the scans were taken between 2 and 3.5 months prior to animal slaughter. 367 Renand and Fisher (1997) noted that if the back-fat estimates from ultrasound 368 measurements for a population showed a standard deviation of more than 3 mm, the 369 coefficients of correlation between the final and ultrasound values can reach 0.6, while 370 for a SD of less than 2 mm, they barely reach 0.3. 371 The ultrasound longissimus muscle depth reading (UAL and UAT) also appears to 372 be correlated, albeit negatively, with various carcass fatness measurements, such as the 373 degree of marbling and the KPF score (Table 3), revealing an inverse relationship 374 between muscular development and final fatness (Mendizabal et al., 1999; Alberti et al., 375 19 2005). The range of back fat thickness in Pirenaica calves (Table 1) was between 0.1 376 and 0.45 cm, with a mean value of 0.23 cm and a standard deviation of 0.08 cm, while 377 40% of the UFL estimates deviated by more than 0.25cm from CFT. 378 The grey level value of a scan depends on tissue composition and therefore on the 379 degree of marbling in the muscle. In general, positive correlations between ultrasound 380 intramuscular fat content (UIMFL and UIMFT) and fat trait were observed in the 381 carcasses. In fact, significant correlations at all scanning times were observed between 382 UIMFT and carcass traits, such as KPF or CIMF. The correlation increased significantly 383 between the first and last scans for UIMFT and KPF, while the correlation between 384 UIMFT and CIMF decreased from the first to the last scans (Table 3). These results are 385 not in agreement with Wall et al. (2004), who reported no changes in the correlations 386 between marbling and ultrasound scans made 100 days before slaughter to preslaughter 387 in fattened bulls from the Angus, Simmental, Red Angus and Charolais breeds. 388 389 3.4. Correlations between ultrasound measurements and beef quality traits 390 Table 4 shows the Pearson coefficients of correlation between the measurements 391 obtained from the in vivo scans (UAL, UAT, UFL, UIMFL and UIMFT) taken 50, 25 and 392 1 days prior to slaughter, the Warner-Braztler shear force scores and the sensory 393 parameters of the meat as evaluated by a trained panel. Fatty flavor was the only 394 sensory parameter with significant coefficients of correlation with the ultrasound area 395 (UAL) and ultrasound intramuscular fat (UIMFT) readings. Fatty flavor correlated 396 negatively with the in vivo UAL readings taken 50, 25 and 1 days prior to slaughter, and 397 UAT at 50 days ante-mortem (r=-0.616). Fatty flavor was correlated positively with 398 UIMFT at the three measurement times and only with UIMFL at 1 day prior to slaughter 399 20 (r=0.538; P<0.05). Correlations between fatty flavor and UAL or UIMFT decreased 400 constantly from 50 days before slaughter to the pre-slaughter scans. 401 402 Ultrasound measurementa Antemortem scan day WBSF Juiciness Hardness Beef flavor Liver flavor Fatty flavor 50 0.233 0.023 0.106 0.025 -0.337 -0.533* UAL 25 -0.247 0.081 -0.235 0.311 -0.225 -0.451* 1 -0.096 -0.060 -0.175 0.105 0.060 -0.491* 50 -0.002 0.101 -0.081 0.211 -0.186 -0.616* UAT 25 0.018 0.204 -0.176 0.046 0.036 -0.288 1 0.034 -0.047 -0.050 0.107 -0.096 -0.139 50 -0.308 -0.304 0.280 0.421 -0.324 -0.012 UFL 25 -0.365 -0.031 -0.065 0.361 -0.062 0.215 1 0.055 -0.253 -0.238 -0.130 -0.040 -0.417 50 0.077 -0.131 0.328 -0.119 0.243 0.302 UIMFL 25 -0.335 0.011 -0.236 -0.107 0.307 0.324 1 0.041 -0.061 0.256 0.049 0.177 0.538* 50 0.152 0.148 0.142 -0.161 0.361 0.535* UIMFT 25 -0.162 -0.009 -0.099 0.034 0.259 0.468* 1 0.098 -0.057 0.186 0.090 0.260 0.462* aText caption, see Table 2; WBSF: Warner–Bratzler shear force test; *: P < 0.05. 403 Table 4 Pearson coefficients of correlation between beef quality traits and ultrasound 404 measurements at different scanning days before slaughter. 405 21 406 407 3.5. Prediction equations 408 With the practical aim of saving time and costs, we identified the best moment for 409 scanning to obtain an acceptable prediction. As regards the estimation of the beef 410 parameters, the best significant equations (P<0.05 and R2>0.5) to predict the percentage 411 of kidney and pelvic fat, intramuscular fat content, carcass conformation, carcass fatness 412 and the fatty flavor of the meat are shown in Table 5. The results of the regression 413 equations included the scores obtained from the in vivo ultrasound scans taken 50, 25 414 and 1 days prior to slaughter. The best prediction equations for KPF and carcass fatness 415 score were at 1 day prior to slaughter, while those for conformation carcass were 416 obtained at 25 days prior to slaughter. The most favorable prediction of intramuscular 417 fat content was 50 days prior to slaughter, since at this time the ultrasound measurement 418 was more stable than that performed one day before the animals were slaughtered. This 419 observation can be seen in Table 2 in which the ranges of dispersion between minimum 420 and maximum observed in the gray levels in UIMFL and UIMFT were higher when the 421 measurements were made one day before slaughtering of the calves than those observed 422 when the ultrasound measurements were carried out 50 days before the animals were 423 slaughtered. This finding suggests that better carcass parameter estimates result from 424 scans taken in the last month prior to slaughter, but not necessarily the day before 425 harvest. 426 28 In general, the R2 values for the best ultrasound measurements to determine the 514 sensory parameters of beef were low (range R 2 = 0.089 to 0.279; data not shown in 515 table) in reference to carcass parameters (Figure 1). Only the significant variable 516 regression to predict young bull sensorial quality traits corresponding to the selected 517 ultrasound measurements for fatty flavor was observed. 518 519 4. Conclusions 520 521 While it is possible to use ultrasound to predict carcass traits in young bulls of 522 lean breeds, ultrasound testing to predict pre-slaughter sensory beef parameters is not 523 common. In order to predict intramuscular fat content in young bulls from lean cattle 524 breeds, early ultrasound scans of the longissimus lumborum area measurement are 525 recommended during fattening. However, to predict kidney and subcutaneous fat 526 depots, ultrasound scans taken close to slaughter including intramuscular fat content and 527 measurement of fat thickness are recommended. The only sensory variables which 528 showed a relationship with the ultrasound measurements performed on the live animals 529 at different times during the fattening of the young bulls were those related to the fat 530 content in meat from young bulls of the Pirenaica breed. In this context, it is advisable 531 to predict the fatty flavor of beef ultrasound scan of intramuscular fat measurement 532 close to moment when the young bulls are slaughtered. However, further studies with 533 other fatness bovine breeds would be needed to determine the validity of the use of 534 ultrasound to predict the sensory fat parameters of bovine meat. 535 Acknowledgments 536 The authors wish to thank the Spanish Ministry of Science and Technology 537 (PETRI 95-0703-OP) and the University of Illinois for funding this study. 538 29 539 5. References 540 Aass, L., Fristedt, C. G., Gresham, J. D., 2009. Ultrasound prediction of intramuscular 541 fat content in lean cattle. Livest. Sci. 125, 177–186. 542 https://doi.org/10.1016/j.livsci.2009.04.005 543 Alberti, P., Ripoll, G., Goyache, F., Lahoz, F., Olleta, J.L., Panea, B., Sañudo, C., 2005. 544 Carcass characteristics of seven Spanish beef breeds slaughtered at two 545 commercial weights. Meat Sci. 71, 514–521. doi: 10.1016 / 546 j.meatsci.2005.04.033 547 Albertí, P., Panea, B., Sañudo, C., Olleta, J.L., Ripoll, G., Ertbjerg, P., Christensen, M., 548 Gigli, S., Failla, S., Cooncetti, S., Hocquette, J.F., Jailler, R., Rudel, S., Renand, 549 G., Nute, G.R., Richardson, R.I., Williams, J.L., 2008. Live weight, body size 550 carcass characteristics of young bulls of fifteen European breeds. Livest. Sci. 551 114, 19–30. https://doi.org/10.1016/j.livsci.2007.04.010 552 Albrecht, E., Teuscher, F., Ender, K., Wegner, J., 2006. Growth- and breed-related 553 changes of marbling characteristics in cattle. J. Anim. Sci. 84, 1067–1075. 554 doi: 10.2527 / 2006.8451067x 555 AMSA. 2016. Research guidelines for cookery, sensory evaluation, and instrumental 556 tenderness measurements of meat. American Meat Science Association. 557 Champaign, Illinois USA 2sd edition. Version 1.02. Available 558 http://www.meatscience.org/sensory 559 Baker, M.J., Tedeschi, L.O., Fox, D.G., Henning, W. R., Ketchen, D.J., 2006. Using 560 ultrasound measurements to predict body composition of yearling bulls. J. Anim. 561 Sci. 84, 2666–2672. doi:10.2527/jas.2006-006 562 30 Beltrán, J.A., 2005. Estandarización de las Metodologías para Evaluar la Calidad Del 563 producto (Animal vivo, Canal, Carne y Grasa) en los Rumiantes; Cañeque, V., 564 Sañudo, C., Ed.; Instituto Nacional de Investigación y Tecnología Agraria y 565 Alimentaría (INIA): Madrid, Spain, 2005; ISBN 84-7498-509-9. 566 Bergen, R.D., Miller, S.P., Mandell, I.B., Robertson, W.M., 2005. Use of live 567 ultrasound, weight and linear measurements to predict carcass composition of 568 young beef bulls. Can. J. Anim. Sci. 85, 23–35. doi: 10.4141 / A04-011 569 Beriain, M.J., Sánchez, M., Insausti, K., Sarries, M.V., Soret, B., 2016. A comparison of 570 sensory acceptance, purchase intention and willingness to pay for Pirenaica beef 571 from two different slaughter weight groups, under different consumer 572 information scenarios. J. Sens. Stud. 31, 453–464. 573 https://doi.org/10.1111/joss.12236 574 Berry, B.W., Maga, J.A., Calkins, C.R., Wells, L.H., Carpenter, A.L., Cross, H.R., 575 1980. Flavour profile analyses of cooked beef loin steaks. J. Food Sci. 45, 1113–576 1115. https://doi.org/10.1111/j.1365-2621.1980.tb06499.x 577 Cannell, R.C., Belk, K.E., Tatum, J.D., Wise, J.W., Chapman, P.L., Scanga J.A., Smith, 578 G.C., 2002. Online evaluation of a commercial video image analysis system 579 (Computer Vision System) to predict beef carcass red meat yield and for 580 augmenting the assignment of USDA yield grades. J. Anim. Sci. 80, 1195–1201. 581 doi: 10.2527 / 2002.8051195x 582 Castilhos, A.M., Francisco, C.L., Branco, R.H., Bonilha, S.F., Mercadante, M.E., 583 Meirelles, P.R., Pariz, C.M., Jorge, A.M., 2018. In vivo ultrasound and 584 biometric measurements predict the empty body chemical composition in 585 Nellore cattle. J. Anim. Sci. 96,1678–1687. https://doi.org/10.1093/jas/sky081 586 31 Chambaz, A., Scheeder, M.R., Kreuzer, M., Dufey, P.A., 2003. Meat quality of Angus, 587 Simmental, Charolais and Limousin steers compared at the same intramuscular 588 fat content. Meat Sci. 63, 491–500. doi: 10.1016 / s0309-1740 (02) 00109-2 589 Charagu, P.K., Crews Jr., D.H., Kemp, R.A., Mwansa, P.B., 2000. Machine effects on 590 accuracy of ultrasonic prediction of back fat and rib eye area in beef bulls, 591 steers, and heifers. Can. J. Anim. Sci. 80, 19-24. doi: 10.4141 / A99-044 592 Council Regulation EC Nº 1099/2009 of 24 September 2009 on the protection of 593 animals at the time of killing. Official Journal of the European Communities, L 594 303, 1–30, 2009 595 Crews Jr., D.J., Shannon, N.H., Crews, R.E., Kemp, R.A., 2002. Weaning, yearling, and 596 preharvest ultrasound measures of fat and muscle area in steers, bulls, and 597 heifers. J. Anim. Sci. 80, 2817–2824. doi: 10.2527 / 2002.80112817x 598 Cross, H.R., Moen, R., Stanfield, M.S., 1978. Training and testing judges for sensory 599 analysis of meat quality. Food Technol. 32, 48–54. 600 Denoyelle, C. 1995. Evolution de la flaveur de la viande bovine en fonction de la teneur 601 en lipides intra-musculaires. Viandes et Produits Carnés, 16, 89–92. 602 Directive 2010/63/EU of the European Parliament and of the Council of 22 September 603 2010 on the Protection of Animals Used for Scientific Purposes. Official Journal 604 of the European Union, L 273, 33-79. 605 DOUE. 2004. Reglamento (CE) No 1483/2004 de la comisión de 20 de agosto de 2004 606 por el que se completa el anexo del Reglamento (CE) No 2400/96, relativo a la 607 inscripción de determinadas denominaciones en el Registro de Denominaciones 608 de Origen Protegidas y de Indicaciones Geográficas Protegidas («Carne de la 609 Sierra de Guadarrama», «Ternera de Navarra» o «Nafarroako Aratxea», «Carne 610 32 de Vacuno del País Vasco» o «Euskal Okela», «Ternera Asturiana» y «Carne de 611 Cantabria») 612 DOUE. 2008. Commission Regulation (EC) No 1249/2008 of 10 December 2008 laying 613 down detailed rules on the implementation of the Community scales for the 614 classification of beef, pig and sheep carcases and the reporting of prices thereof. 615 L 337/3. 616 Eriksson, S., Näsholm, A., Johansson, K., Philipsson, J., 2002. Genetic analysis of post-617 weaning gain of Swedish beef cattle recorded under field conditions and at 618 station performance testing. Livest. Prod. Sci. 76, 91–101. 619 https://doi.org/10.1016/S0301-6226(02)00007-6 620 Folch, J., Less, M., Sloane, G.H., 1957. A simple method for the isolation and 621 purification of total lipids from animal tissues. J Biol. Chem. 226, 497–509. 622 PMID: 13428781 623 Gorraiz, C., Berian, M.J., Chasco, J., Iraizoz, M., 2000. Descriptive analysis of meat 624 from young ruminants in Mediterranean systems. J. Sens. Stud. 15, 137–150. 625 doi: 10.1111 / j.1745-459X.2000.tb00261.x 626 Greiner, S.P., Rouse, G.H., Wilson, D.E., Cundiff, L.V., Wheeler, T.L., 2003. 627 Prediction of retail product weight and percentage using ultrasound and carcass 628 measurements in beef cattle. J. Anim. Sci. 81, 1736–1742. doi: 10.2527 / 629 2003.8171736x 630 Hamlin, K.E., Green, R.D., Perkins, T.L., Cundiff, L.V., Miller, M.F., 1995a. Real-time 631 ultrasonic measurement of fat thickness and longissimus muscle area: I. 632 Description of age and weight effect. J. Anim. Sci. 73, 1713–1724. doi: 10.2527 633 / 1995.7361713x 634 33 Hamlin, K.E., Green, R.D., Cundiff, L.V., Wheeler, T.L., Dikeman, M.E., 1995b. 635 Realtime ultrasonic measurement of fat thickness and longissimus muscle area: 636 II. Relationship between real-time ultrasound measures and carcass retail yield. 637 J. Anim. Sci. 73, 1725–1734. doi: 10.2527 / 1995.7361713x 638 Hodgson, R.R., Belk, K.E., Savell, J.W., Cross, H.R., Williams, F.L., 1992. 639 Development of a multivariate yield grade equation to predict compositional 640 traits in mature cow carcasses. J. Anim. Sci. 70, 2159–2166. doi: 10.2527 / 641 1992.7072159x 642 Hwang, H., Park, B., Nguyen, M., Chen, Y.R., 1997. Hybrid image processing for 643 robust extraction of lean tissue on beef cut surfaces. Comput. Electron. Agric. 644 17, 281–294. https://doi.org/10.1016/S0168-1699(97)01321-5 645 Indurain, G., Beriain, M.J., Goñi, M.V., Arana, A., Purroy, A. 2006. Composition and 646 estimation of intramuscular and subcutaneous fatty acid composition in Spanish 647 young bulls. Meat Sci. 73, 326–334. 648 https://doi.org/10.1016/j.meatsci.2005.12.007 649 Indurain, G., Carr, T.R., Goñi, M.V., Insausti, K., Alfonso, L., Beriain, M.J., 2009. The 650 relationship of carcass measurements to carcass composition and intramuscular 651 fat in Spanish beef. Meat Sci. 82, 155–161. 652 https://doi.org/10.1016/j.meatsci.2009.01.005 653 ISO 8586-2014. Sensory analysis- General guidelines for the selection, training and 654 monitoring of selected assessors and expert sensory assessors. 655 May, S.G., Mies, W.L., Edwards, J.W., Harris, J.J., Morgan, J.B., Garrett, R.P., 656 Williams, F.L., Wise, J.W., Cross, H.R., Savell, J.W., 2000. Using live estimates 657 and ultrasound measurements to predict beef carcass cuttability. J. Anim. Sci. 658 78, 1255–1261. doi: 10.2527 / 2000.7851255x 659 34 Meilgaard, M., Civille, G.V., Carr B.T., 1991. Sensory evaluation techniques. 3 th 660 edition. Boca Raton, Florida, CRC Academia Press. 416 p. 661 https://doi.org/10.1201/9781003040729 662 Mendizabal, J.A., Albertí, P., Eguinoa, P., Arana, A., Soret, B., Purroy A., 1999. 663 Adipocyte size and lipogenic enzyme activities in different adipose tissue depots 664 in steers of local Spanish breeds. Anim. Sci. 69, 115–121. 665 doi: https://doi.org/10.1017/S1357729800051146 666 Mottram, D.S. 1998. Flavour formation in meat and meat products: a review. Food 667 Chem. 62, 415–424. https://doi.org/10.1016/S0308-8146(98)00076-4 668 Nogalski, Z., Pogorzelska-Przybyłek, P., Białobrzewski, I., Modzelewska-Kapituła, M., 669 Sobczuk-Szul, M., Purwin, C., 2017. Estimation of the intramuscular fat content 670 of m. longissimus thoracis in crossbred beef cattle based on live animal 671 measurements. Meat Sci. 125, 121–127. 672 https://doi.org/10.1016/j.meatsci.2016.11.026 673 OJEU. 2006. (Official Journal of the European Union). 2006. L 17/6 (January 21st, 674 2006) 675 Peña, F., Molina, A., Juárez, M., Requena, F., Avilés, C., Santos, R., Domenech, V., 676 Horcada, A., 2014. Use of serial ultrasound measures in the study of growth- 677 and breed-related changes of ultrasonic measurements and relationship with 678 carcass measurements in lean cattle breeds. Meat Sci. 96, 247–255. 679 https://doi.org/10.1016/j.meatsci.2013.07.012 680 Perkins, T.L., Green, R.D., Hamlin, K.E., 1992a. Evaluation of ultrasonic estimates of 681 carcass fat thickness and longissimus muscle area in beef cattle. J. Anim. Sci. 682 70, 1002–1010. doi: 10.2527 / 1992.7041002x 683 35 Perkins, T.L., Green, R.D., Hamlin, K.E., Shepard, H.H., Miller, M.F., 1992b. 684 Ultrasonic prediction of carcass merit in beef cattle: Evaluation of technical 685 effects on ultrasonic estimates of carcass fat thickness and longissimus muscle 686 area. J. Anim. Sci. 70, 2758–2765. https://doi.org/10.2527/1992.7092758x 687 Piedrafita, J., Quintanilla, R., Sañudo, C., Olleta, J.L., Campo, M.M., Panea, B., 688 Renand, G., Turin, F., Jabet, S., Osoro, K., Olivan, M.C., Noval, G., Garcıa, P., 689 Garcıa, M.D., Oliver, M.A., Gispert, M., Serra, X., Espejo, M., García, S., 690 López, M., Izquierdo, M., 2003. Carcass quality of 10 beef cattle breeds of the 691 Southwest of Europe in their typical production systems. Livest. Product. Sci. 692 82, 1–13. https://doi.org/10.1016/S0301-6226(03)00006-X 693 Realini, C.E., Williams, R.E., Pringle, T.D., Bertrand J.K., 2001. Gluteous medius and 694 rump fat as additional live animal ultrasound measurements for predicting retail 695 product and trimmable fat in beef carcasses. J. Anim. Sci. 79, 1378–1385. 696 doi: 10.2527 / 2001.7961378x 697 Renand, G., Fisher, A.V., 1997. Comparison of methods for estimating carcass fat 698 content of young Charolais bulls in performance testing station. Livest. Product. 699 Sci. 51, 205–213. https://doi.org/10.1016/S0301-6226(97)00060-2 700 Savell, J.W., Cross, H.R., 1988. The role of fat in the palatability of beef, pork and 701 lamb. In: Designing foods: animal product options in the marketplace (pp. 345-702 355). Washington DC: National Academic Press. 703 Smith, M.T., Oltjen, J.W., Dolezal, H.G., Gill, D.R., Behrens, B.D., 1990. Evaluation of 704 real-time ultrasound for predicting carcass traits of feedlot steers. Oklahoma 705 Agric. Exp. Sta. Res. Rep. MP-129, p. 374 706 36 Tait Jr., R.G., Wilson, D.E., Rouse, G.H., 2005. Prediction of retail product and 707 trimmable fat yields from the four primal cuts in beef cattle using ultrasound or 708 carcass data. J. Anim. Sci. 83, 1353–1360. doi: 10.2527 / 2005.8361353x 709 UNE 87004-1979. Guide for the installation of a tasting room. CTN 87 - Sensory 710 analysis. UNE Spanish Standardization. 711 Wall, P.B., Rouse, G.H., Wilson, D.E., Tait Jr., R.G., Busby, W.D., 2004. Use of 712 ultrasound to predict body composition changes in steers at 100 ad 65 days 713 before slaughter. J. Anim. Sci. 82, 1621–1629. 714 https://doi.org/10.2527/2004.8261621x 715 Waldner, D.N., Dikeman, M.E., Schalles, R.R., Olson, W.G., Houghton, P.L., 716 Unruh, J.A., Corah, L.R., 1992. Validation of real-time ultrasound technology 717 for predicting fat thicknesses, longissimus muscle areas, and composition of 718 Brangus bulls from 4 months to 2 years of age. J. Anim. Sci. 70, 3044–3054. 719 https://doi.org/10.2527/1992.70103044x 720 Yang, X.J., Albrecht, E., Ender, K., Zhao, R.Q., Wegner, J., 2006. Computer image 721 analysis of intramuscular adipocytes and marbling in the longissimus muscle of 722 cattle. J. Anim. Sci. 84, 3251–3258. doi: 10.2527 / jas.2006-187 723