IntroductionTop

Infrared thermography (IRT) represents a non-invasive method used in numerous applications, not only in industry. Effectiveness of IRT has been proved for diagnostic purposes in animal diseases (Martins et al., 2013), to assess physiological and metabolic parameters in production animals (McManus et al., 2016) or to predict physiological variables related to the thermal stress of cattle (Viera de Sousa et al., 2018).

When an animal becomes stressed, the hypothalamic–pituitary–adrenocortical axis is activated and heat is produced due to the blood flow increase, producing changes in heat radiation from the animal (Schaefer et al., 2002). Ocular temperature is considered a good indicator of body core temperature (Johnson et al., 2011) due to eyes are close to the brain. Thus, infrared ocular thermography (IROT) could be used to assess the neuro-physiological and stress conditions of animals at slaughter (Weschenfelder et al., 2013; Losada-Spinosa et al., 2018; Rocha et al., 2019).

During pre-slaughter operations, livestock is expo­sed to a range of potential stressors, such as unfami­liar environment, human contact and transportation, which may compromise their welfare and impact meat quality traits. Furthermore, stressors factors at the slaughterhouse increase physical activity of the animals causing an increase in metabolic activity, heat production, increase of blood flow and depletion of muscle glycogen, leading to increase ultimate pH in meat (Eriksen et al., 2013).

Beef quality is increasingly important as a large supply chain, and it raises concerns of a demanding fresh meat industry. Therefore, avoiding the appearance of bovine meat with high pH is important for the meat industry to decrease the appearance of quality defects (Loredo-Osti et al., 2019). Thus, detecting stre­ssed animals in the abattoir prior to slaughter would positively impact the meat industry.

The objective of this study was to determine whe­ther the IROT estimates measured immediately before slaughter, under controlled welfare animal conditions, could be used to predict beef quality in light young bulls.

Material and methodsTop

Animals

The study was carried out in a commercial slaugh­terhouse (Salteras, Seville province, Spain). A total of 175 yearling bulls (1-year-old and 450-550 kg live weight) from 9 farms located 80 km around the slaughterhouse, were randomly chosen for this study. In order to represent the southern Spanish market, all animals were Retinta × Limousin or Charolais crossbred. Young bulls were transported by road at night in compliance with animal welfare regulations (EC, 2005). The average duration of animal’s transportation to the slaughterhouse was around 3 h including loading and unloading. After lairage (12 h) in individuals pens and non-skid floors (12 h, with water available but no food), young bulls were conducted at 8 a.m. to the point of stunning and were stunned using a captive bolt (Jarvis 25 Caliber, Standard, Captive Bolt, Pistol-Style Stunne, Koch Butchers Supply Company, USA). Then, they were dressed according to the guidelines of EC (2009).

In accordance to the commercial classification weight system, carcasses were grouped as ‘light’ (˂250 kg; n=55); ‘medium’ (250-310 kg; n=75), and ‘heavy’ (˃310 kg; n=45) (Lonja Mercamadrid, https://www.feriasymercados.net/index.php/lonja/index/12).

Collection of infrared thermography images

IROT images were obtained using a handheld in­frared camera (Therma Cam i70 0, FLIR Systems AB, Danderyd, Sweden), operated by a trained technician. Images were collected in the stunning box, just prior to stunning. The left eye of young bulls was scanned from a 90º angle at a distance of 20 cm. At least three images per animal were collected, selecting the image that provided the optimal operating conditions for analy­sis. To avoid the confounding effect of environmen­tal conditions on ocular temperature, environmen­tal temperature and relative humidity were recorded with a digital thermohygrometer (Extech 44550, Extech Instruments Corporation, Nashua, NH, USA). IROT images were analyzed with the image analysis software Therma Cam Researcher Pro 2.8 SR-2 (FLIR Systems AB, FLIR Systems, Inc., Sweden) in order to obtain the maximum temperature (°C) within the area traced around the eye, including approximately 1 cm surrounding the outside of the eye lids. The maximum eye temperature, corrected by the environmental tem­perature and relative humidity and by an emissivity value of 0.98 (Stewart et al., 2007), was used for the analyses.

Blood variables

Two blood samples (10 mL each) per animal were collected, using BD vacutainer® tubes containing potassium ethylene-diamine-tetra-acetic acid as anti­coa­gulant for the analysis of glucose levels. Glucose (mg/dL blood) level was determined by an enzymatic capillary glucometer (OneTouch Ultra; LifeScan, Inc., Milpitas, CA) made in situ at the slaughterhouse, from plasma samples (Alcalde et al., 2017).

Meat quality assessment

After 24 h at 4°C, pH was measured at the caudal segment of the right longissimus thoracis (LT) muscle using a pH-meter Crison 507 equipped with a penetrating electrode (Crison Instruments S.A., Barcelona, Spain) associated with a temperature compensation probe. To evaluate carcass colour, lightness (L*), redness (a*), and yellowness (b*) were then recorded in the CIELAB space (CIE, 1986) at the rectus abdominis (RA) muscle using a Minolta spectrophotometer (CM-700d; Minolta Sensing Inc., Singapore, Japan) with a 25-mm aperture, a specular component included, 10º view in angle, 0% UV and D65 standard illuminant. Chroma (C*) and hue (H*) indexes were calculated as C*= (a*2+b*2)0.5 and H*=tan−1 (b*/a*) and were expressed in degrees (MacDougall, 1986).

Seven days after slaughter, around 50 g of RA muscle was removed from the carcass and water holding capacity (WHC) was measured using the filter paper press method (Grau & Hamm, 1953). WHC was reported as % of juice expelled from fresh meat.

Statistical analyses

As physiological and meat quality parameters sa­tisfied assumptions of normal distribution (results not shown), parametrical statistical analyses were used. In order to test the influence of the carcass weight level on the IROT measurements, a previous main effects ANOVA test was developed (model included weight as a fixed effect and no random effects), with this parameter showing statistically significant (p<0.05) differences for all the variables. According to that, differences due to the weight level of the animal within IROT measurements, blood glucose levels and meat quality parameters were assessed using a Duncan’s post-hoc test. Pearson’s correlations between IROT measurements and meat quality parameters were cal­culated for each weight level considered, in order to ascertain a relationship within them. Finally, in order to predict meat quality values from IROT measurements, a quadratic regression analysis was developed for those previous variables that resulted statistically significant (p<0.05) on the correlation analysis, accounting for the weight as a continuous variable. The quadratic equation with the coefficient of determination (R2), and the root mean squared error (RMSE) and p value of the analysis were provided for each parameter. Statistical analyses were performed using the Statistica v.12.0 (Statistica software, v.12.0. Statsoft, Inc. 1984-2014) package.

Results and discussionTop

Means values of IROT and meat quality parameters according to three carcasses weight levels are shown in Table 1. Overall, IROT values obtained for the three carcass weight ranges were lower (33±4ºC) than repor­ted by others authors (37.5±1.2ºC, Church et al., 2014), probably due to lower environmental temperatures at the slaughterhouse (5-10ºC in this study vs 20ºC reported by Church et al., 2014) and breed differences (Retinta x Limousin or Charolais crossbred vs Angus crossbred and Holstein cows).

A higher (p<0.05) ocular temperature was found in heavier carcasses compared with light and intermediate carcasses. These differences in IROT could be due to vessels’ diameter, which is smaller in light young bulls, resulting in smaller blood flow in these animals compared to heavier young bulls. Behavioral reactions to stress suffered prior to slaughter induce a fast de­pletion of glycogen from the liver, increasing blood glucose levels and, indirectly, pH values (Lacourt & Tarrant, 1985). High pH and glucose levels in blood are associated with dark-cutting beef (Lu et al., 2018).

In our study, glucose levels (53-194 mg/dL blood) and pH (5.4 to 5.9) were in the range of unstressed animals (Lu et al., 2018). Morgan et al. (1995) reported that cattle whose IROT measures were outside the test temperature range (28-32 ± 2 °C) could be rejected as having a high probability of producing poor meat quality. In our study (Table 1), higher IROT ranges (29.2-37.3) would, a priori, predict a stress response that should affect meat quality. However, according to pH and glucose values, no stress signs were found in carcasses. This could be due to differences in animal responses to stress according to breed and weight factors that would condition the emission of temperature by the caruncle and hence, the temperature measured by the infra­red camera. Amakiri (1976) reported a more superficial vascular arrangement in rustic cattle breeds, that could certainly bias the temperature captured by this tool due to the deeper vessels.

Regarding physical parameters of meat quality, colour values were within the ranges reported for com­mercial carcasses of young bulls from Southwest of Europe (Peña et al., 2014). No significant differences (p>0.05) were found in the beef luminosity between the carcass weight groups. However, differences between means due to carcass weight were observed for a*, b*, C* and H* attributes due to a lower fat content expected in light carcasses compared with heavier carcasses (Gagaoua et al., 2018).

Except for pH and H* colour values, no significant correlations were found between IROT and meat quality traits on each carcass weight group (Table 2). The IROT value was moderately and positively correlated with pH measured in the LT muscle of light carcasses (r=0.66; p<0.05). Thus, in case of light carcasses, a higher pH value in meat was expected in animals showing higher IROT values previous to slaughter. On the other hand, the correlation between IROT and H* value was weak and negative, although statistically significant (r=-0.22; p<0.05) in each weight group tested. Thus, regardless the slaughter weight, low H* values could be expected in beef from bulls showing high IROT values before slaughter. This observation can be explained due to the intraclass variability in each weight group tested was small, whereas the interclass variability observed was high.

Figure 1 shows the quadratic regression analyses for both parameters that showed statistically signi­ficant (p<0.05) correlations with IROT (pH and H*). For this analysis, weight was considered as a continuous variable. Results shown in Figure 1a indicate that carcass pH after slaughter could be pre­dicted from IROT measurements obtained before slaughter, following the quadratic equation: pH = 7.231-0.0109*x-0.0507*y+2.8455E-5*x*x-3.2054E-5*x*y+0.0011*y*y (R2=0.52; RMSE=0.032; p<0.01), where y is the weight of the animal and x is the IROT measured before slaughter. As regards to Hue value, it could be predicted following the quadratic equation showed in Figure 1b: H* =-25.4711­

Figure 1. Quadratic regression analyses between infrared ocular temperature (x axis), carcass weight measurements (y axis) and meat quality parameters that resulted statistically significant on the previous correlation analysis (z axis): (a) for pH value; (b) for Hue (H*) value. The square in the bottom left side of each figure showed the quadratic equation that predicts pH and H* values, whereas the square in the upper right side of each figure showed the model fit parameters R-squared (R2), root mean square error (RMSE) and p-value.

+0.0831*x+4.2168*y+0.0003*x*x-0.0084*x*y-0.0334*y*y (R2=0.41; RMSE= 3.793; p<0.001), where y is the weight of the animal and x is the IROT measured before slaughter.

Morgan et al. (1995) already found IROT as a useful tool for the early detection of poor meat quality in cattle before slaughter. They proposed to reject animals with IROT measurements differing at least 1.20 times from the mean IROT measurement calculated for the whole group slaughtered together. The quadratic regression analysis showed differences in the effectiveness of this tool due to animal weight. Furthermore, this analysis supported previous correlation results showing that the higher the IROT, the higher the pH value within the lightest animals’ weight group (<250 kg), with a medium-strong relation (R2=0.52). This could be due to the characteristics of the Spanish native breeds used for the analysis, in general lean and more heterogeneous in weight than those used by previous studies (mean body weight of 283±100kg in this study, vs 500±2kg reported by Morgan et al., 1995, in Canadian bull breeds or 309±5kg reported by Church et al., 2014, in Angus cross steers). Thus, the IROT interval previously proposed should be taken with caution in rustic animals. These results are supported by previous studies that found differences in meat quality and muscle structure of rustic compared with more selected cattle breeds (Vieira et al., 2007).

Regarding meat colour, our results seem to suggest that low Hue values, related to redder carcass tone, could be linked to a greater blood flow through the muscles during slaughter, as a result of the HPA axis activation after a stress response, indicated indirectly by the high IROT values obtained (Stewart et al., 2007). Despite, this relation was weak when just a linear correlation was considered (r=-0.22), a medium relation between H* and IROT value was found (R2=0.41), when carcass weight was considered as a continuous variable, so that, the higher the weight, the higher the H* value and the lower the IROT tem­perature, thus supporting previous results obtained from the negative linear correlation.

The pre-slaughter use of IROT in bulls from lean breeds could be a potential low-cost and quick de­cision-making method to reduce meat quality de­fects due to pre-slaughter stress. Our preliminary re­sults indicate a tendency of this instrument to change its effectiveness depending on carcass weight, showing better prediction potential with H* value regardless the weight and with pH on light lean young bulls. However, this technique will need further validation to confirm the correlations observed in the present study (including variability in pre-slaughter conditions and data from stressed animals) and to achieve the level of automation required to operate at line speed in a commercial slaughterhouse.

AcknowledgmentsTop

We wish to thank Matadero del Sur S. A. (Salteras, Spain) staff for their technical support for the development of this work and for providing the animals and the slaughter facilities.