Introduction
The consumer-preferred bright cherry-red color in fresh beef occurs due to oxygen binding to myoglobin, leading to oxymyoglobin formation (King et al., 2023; Mancini et al., 2022). Oxymyoglobin is oxidized to form metmyoglobin, leading to discoloration on the surface of meat. At around 13–20% discoloration, consumers and trained panelists begin to discriminate against meat products (Hood and Riordan, 1973; Limsupavanich et al., 2008; Lybarger et al., 2023). Hence, it is important to limit and predict the formation of metmyoglobin during retail display to minimize losses due to discoloration.
Biochemical processes such as metmyoglobin-reducing activity, oxygen consumption, and lipid oxidation influence metmyoglobin formation and color stability of beef. More specifically, lipid oxidation products can bind with oxymyoglobin and promote metmyoglobin formation (Faustman et al., 2010; Lynch and Faustman, 2000). Metmyoglobin-reducing ability can extend color stability by reducing metmyoglobin to oxy- or deoxymyoglobin through Nicotinamide adenine dinucleotide plus hydrogen (NADH)-dependent reductase, non-enzymatic, and mitochondria-mediated pathways (Arihara et al., 1995; Brown and Snyder, 1969; Denzer et al., 2020; Tang et al., 2005). Various studies have demonstrated that anaerobic conditions promote greater metmyoglobin-reducing activity than aerobic conditions (Tang et al., 2005). Hence, oxygen consumption is critical to promote anaerobic conditions for metmyoglobin-reducing activity and also to determine the formation of a bright red color. Although the role of metmyoglobin-reducing activity, oxygen consumption, and lipid oxidation is well documented, there are inconsistencies in the application of these parameters in predicting color stability (English et al., 2016; Ramanathan et al., 2019; Ramanathan et al., 2020; O’Keeffe and Hood, 1982; Sammel et al., 2002). For example, some studies noted metmyoglobin-reducing activity and oxygen consumption decrease with increased storage time and lower redness (English et al., 2016; Sammel et al., 2002). Using methods recommended by the American Meat Science Association Meat Color Guidelines (King et al., 2023), oxygen consumption and metmyoglobin-reducing activity can be evaluated using retail (oxygen exposed: oxygen exposed) or interior (non–oxygen-exposed: non–oxygen-exposed) surfaces. Research has reported using the oxygen-exposed surface and the non–oxygen-exposed surface for metmyoglobin-reducing activity evaluation (Mancini et al., 2008). The a* values and the oxygen-exposed metmyoglobin-reducing activity were strongly correlated, whereas the non–oxygen-exposed metmyoglobin-reducing activity and a* values were weakly correlated (Mancini et al., 2008).
Feature selection is a process of choosing the most relevant features from a dataset while eliminating irrelevant or redundant ones (Chandrashekar and Sahin, 2014). It is widely used in machine learning and statistical modeling to improve model performance and reduce errors. Feature selection is applied in various fields to study the most relevant variables (Gredell et al., 2019;Dokeroglu et al., 2022). The objectives of this study were to evaluate the contribution of lipid oxidation, metmyoglobin-reducing activity, oxygen penetration, and oxygen consumption of oxygen-exposed and non–oxygen-exposed areas to understand the features that contribute to color stability of the longissimus lumborum muscle.
Materials and Methods
Materials
Seven USDA Low Choice strip loins (Institutional Meat Purchase Specifications (IMPS) #180; grain-finished and market-age cattle) were collected from a local processing facility 5 d postmortem and transported to Oklahoma State University Food and Agricultural Products Center on ice. On day 7 postmortem, loins were removed from their packaging, and pH was measured in 3 locations across the loin using a probe-type pH meter (Handheld HI 99163; probe FC232; Hanna Instruments). From the anterior end, loins were sliced using a meat slicer (Bizerba USA INC., Piscataway, NJ, USA) into six 1.91-cm steaks. The posterior steak (n = 1 steak per loin) was used for proximate analysis. Four steaks were randomly selected to be packaged in pairs in white Styrofoam® trays overlapped with polyvinyl chloride (PVC) (15,500–16,275 cm3 O2/m2/24 h at 23°C, E-Z Wrap Crystal Clear Polyvinyl Chloride Wrapping Film, Koch Supplies, Kansas City, MO, USA). The packaged steaks (n = 2 paired steaks per loin) were randomly selected for 3 (n = 1 paired steaks) or 6 d (n = 1 paired steaks) in retail display. One steak from each packaged pair was assigned to oxygen consumption and oxygen penetration measurements, and the second steak was assigned to lipid oxidation and metmyoglobin-reducing activity. The remaining steak from each loin (n = 7 steaks) was used for the day 0 analysis of the non–oxygen-exposed surface of the loins.
Proximate composition
Each steak (n = 1 steak per loin) was ground using a tabletop grinder (Big Bite Grinder, 4.5 mm, fine grind, LEM), and approximately 200 g of ground meat was placed in a 140-mm sample cup. A FoodScan Lab analyzer equipped with near-infrared spectrophotometry (Foss, NIRsystem Inc.; Slangerupgrade, Denmark) was used to determine the percent of moisture, fat, and protein.
Retail display
White coffin-style cases were used for a simulated retail display using continuous light emitting diode (LED) lighting (Philips LED lamps, 12 watts, 121.92 centimeters, color temperature = 3,500°K, 54 Phillips, China) for 6 d at 2 ± 1°C. Instrumental color was evaluated every day on steaks selected for 6 d in retail using a HunterLab 4500L MiniScan EZ Spectrophotometer (2.5-cm aperture, illuminant A, and 10°standard observer angle, HunterLab Associates, Reston, VA, USA). Three scans were taken on each steak in a package. From the CIE a* and b* values, the chroma and hue were calculated.
Oxygen exposure
Different terminologies used to indicate oxygen-exposed and non–exposed-surfaces are represented in Figure 1. Retail surfaces used for instrumental color analysis were considered oxygen-exposed surfaces. The interior of the muscle was the non–oxygen-exposed surface. For comparison of the surfaces, steaks (n = 7) were cut into 2.5 × 2.5 × 1.91-cm cubes and then bisected parallel to the retail surface to expose the non–oxygen-exposed surface. To evaluate the oxygen penetration of the freshly cut non–oxygen-exposed surface, 2 measurements were taken. Oxygen depth was evaluated as combined oxymyoglobin and metmyoglobin depth on the surface perpendicular to the freshly cut non–oxygen-exposed surface. Separating these 2 measurements on the non–oxygen-exposed surface ensured oxygen penetration was considered both into the top surface of the muscle (surface layer depth) and the top and side of the muscle (oxygen depth). Previous studies have focused on penetration of the retail surface (surface layer depth of oxygen exposed; McKenna et al., 2005; O’Keeffe & Hood., 1982).
Evaluation of oxygen penetration through oxygen exposure of oxygen-exposed and non–oxygen-exposed surfaces. 1Depth of oxymyoglobin and metmyoglobin after retail display. 2Depth of oxymyoglobin and metmyoglobin after 1-h bloom of the lateral surface to the freshly cut non–oxygen-exposed surface. 3Depth of oxymyoglobin and metmyoglobin after 1-h bloom and re-exposure of the lateral surface to evaluate oxygen penetration into the non–oxygen-exposed surface.
Lipid oxidation
On day 0, day 3, and day 6 of display, lipid oxidation was evaluated using the thiobarbituric acid reactive substances value method (Witte et al., 1970; Denzer et al., 2022). From each steak (n = 7), 3 grams of both the non–oxygen-exposed and oxygen-exposed surface was removed and blended with 27 mL of trichloroacetic acid. The homogenate was filtered through a Whatman No. 42 filter paper. One mL of filtrate was combined with 1 mL of thiobarbituric acid, heated for 10 min at 100°C, and cooled for 5 min. At 532 nm, the absorbance was evaluated using a UV-Vis Spectrophotometer (UV-2600, UV-Vis Spectrophotometer, Shimadzu, Columbia, MD, USA) and converted to mg of malondialdehyde per kg of sample.
Metmyoglobin-reducing activity
On 0, 3, and 6 d of retail display, steaks (n = 7) assigned to metmyoglobin-reducing activity were cut parallel to the retail surface to create a non–oxygen-exposed and oxygen-exposed surface. Both non–oxygen-exposed and oxygen-exposed surfaces were evaluated for metmyoglobin-reducing activity by a modified procedure from Sammel et al. (2002). Samples were submerged in 0.3% sodium nitrite solution with the non–oxygen-exposed and oxygen-exposed surfaces facing upward in the solution. The samples were submerged for 20 min to evaluate nitric oxide-induced metmyoglobin-reducing activity. Initial metmyoglobin formation was determined using a HunterLab 4500L MiniScan EZ Spectrophotometer with 3 scans on each surface. The K/S ratio of K/S572 ÷ K/S525 was used to calculate the amount of metmyoglobin formed, with a greater number indicating a lower metmyoglobin formation and greater metmyoglobin-reducing activity.
Oxygen consumption
Oxygen consumption was calculated using 2 approaches: (1) as changes in oxymyoglobin levels pre- and post-incubation and (2) metmyoglobin formation during oxygen consumption. The same procedure to expose the oxygen-exposed and non–oxygen-exposed surfaces was used on the steak assigned to oxygen consumption. Oxygen consumption was determined by oxygen exposure of both the non–oxygen-exposed and oxygen-exposed surface for 1 h at 4°C. The bloomed samples were vacuum packaged, incubated at 30°C for 60 min, and read using a HunterLab 4500L MiniScan EZ Spectrophotometer at 0, 30, and 60 min. To determine the oxygen consumption, the change in oxymyoglobin formation during incubation was determined by the preincubation K/S610 ÷ K/S525 – post incubation K/S610 ÷ K/S525 (King et al., 2023). A greater change in oxymyoglobin indicates a greater oxygen consumption.
The conversion of bloomed steaks (predominant oxymyoglobin) to metmyoglobin formed during incubation was determined by as K/S ratio of K/S572 ÷ K/S525 at 30 min and 60 min of incubation (Denzer et al., 2025). By evaluating metmyoglobin, the reducing capacity of samples was evaluated in the transition from oxymyoglobin to metmyoglobin to deoxymyoglobin in anaerobic conditions, in addition to the oxygen consumption (King et al., 2023; Denzer et al., 2025). Low oxygen consumption could extend the period of low oxygen partial pressures, leading to greater oxidation of myoglobin and the formation of metmyoglobin.
Oxygen penetration
The oxygen penetration was determined using a modified procedure based on Limsupavanich et al. (2008) and Denzer et al. (2023). Oxygen penetration was considered as the depth of oxygen diffused into the muscle, forming metmyoglobin and oxymyoglobin. This penetration was separated into 2 terms for the 2 different surfaces of oxygen-exposed and non–oxygen-exposed. Oxygen depth was the oxy- and metmyoglobin layers of the non–oxygen-exposed surface formed after 1 h bloom at 4°C. A handheld caliper was used to measure the depth of oxymyoglobin and metmyoglobin 3 times across the surface of the steak lateral to the cut surface. Additionally, the thickness of the steak piece was measured to determine the percent oxygen depth based on the size of the steak piece. Oxygen depth measurements are further detailed in Figure 1.
Surface layer depth was analyzed for non–oxygen-exposed and oxygen-exposed surfaces as the depth of oxymyoglobin and metmyoglobin layers. For the oxygen-exposed surface, the surface layer depth was measured laterally to the retail surface. The surface layer depth of the non–oxygen-exposed surface was analyzed after 1 h bloom at 4°C. To expose the surface layer of the non–oxygen-exposed surface, approximately 2 mm was sliced transversely to the cut surface to re-expose the lateral surface. At 3 locations, the depth was measured lateral to the cut surface. Percent surface depth was determined based on surface layer oxygen depth divided by the thickness of the steak piece.
Statistical analysis
This experiment was a split-plot design with a randomized block. The loin (n = 7) was considered a block and a random effect. The whole plot experimental unit was the steak (n = 7) with the whole plot factor as the pull day (0, 3, 6 d) in retail display. The sub-plot factor was the oxygen exposure, with the experimental unit as the non–oxygen-exposed (n = 7) and oxygen-exposed steak pieces (n = 7). The fixed effects were the pull day, oxygen exposure, and their interactions. Repeated measures were evaluated for retail display with a covariance structure of first-order autoregressive based on the Akaike’s information criterion (AICC) values. PROC GLIMMIX of SAS was used to determine significant effects and interactions with a protected type III F-test. Least-squares means were determined with the GLIMMIX procedure and separated using the PDIFF option with a P < 0.05 considered significant.
Pearson’s correlation and simple linear regression analysis were analyzed for the non–oxygen-exposed and oxygen-exposed surface parameters and a* values during retail display. Parameters evaluated include lipid oxidation, metmyoglobin-reducing activity, oxygen consumption after 30 min, oxygen consumption after 60 min, metmyoglobin formation during oxygen consumption after 30 min, metmyoglobin formation during 60 min, surface layer oxygen depth, and oxygen depth of the non–oxygen-exposed surface. Significant correlation and regression were considered with a P < 0.05. The correlation coefficient values were considered weak at r < |0.35|, moderate at |0.36| ≤ r < |0.67|, and strong at r ≥ |0.68| (Bohrer and Boler, 2017). The coefficients of determination (R2) were considered weak at R2 < 0.12, moderate at 0.13 ≥ R2 < 0.45, and strong at R2 ≥ 0.46 (Bohrer and Boler, 2017). The feature selection method was also used to understand the contribution of each variable to color stability. Various models, such as random forest, lasso regression, and recursive feature elimination, were tested, and the lowest mean cross-validation score was used to select the best model. The feature selection was conducted in Python.
Results
Proximate composition, pH, and lipid oxidation
The proximate compositions of loins are included in Table 1. The mean pH for the strip loins was 5.55, indicating pH in the normal range for beef. There were no significant (P > 0.05) effects on lipid oxidation of the steaks for all retail days (Day 0: 0.40±0.04; Day 3:0.29±0.04; Day 6: 0.36±0.04 mg MDA/kg). Further, oxygen exposure surfaces did not impact (P > 0.05) lipid oxidation (NOE: 0.35±0.03; OE: 0.35±0.03 mg MDA/kg).
Proximate compositions (%) and pH least square means for the longissimus lumborum muscles (n = 7)
| Parameter | Mean % | SEM |
|---|---|---|
| Protein | 22.57 | 0.10 |
| Moisture | 73.72 | 0.21 |
| Fat | 4.52 | 0.24 |
| pH | 5.55 | 0.01 |
SEM = standard error of mean.
Retail color
Retail day had a significant (P < 0.05) effect on L* values, a* values, chroma, and hue of the steaks (Table 2). As retail time increased, color stability decreased, represented by a decrease (P < 0.05) in L* values, a* values, and chroma. Therefore, the steaks appeared less red and darker as retail display time increased. There was a minimal decrease (P < 0.05) in hue angle during retail display, indicating a slight shift in color from true red. The limited changes in hue angle align with the color stability of the longissimus lumborum during retail display.
Effects of retail display day on color attributes of longissimus lumborum steaks (n = 7) displayed for 6 d
| Retail display day | L* | a* | Chroma | Hue |
|---|---|---|---|---|
| 0 | 43.23ab | 35.36a | 46.79a | 40.89a |
| 1 | 43.70ab | 31.34b | 39.14b | 36.73c |
| 2 | 44.42a | 30.10c | 37.38c | 36.32d |
| 3 | 43.16ab | 29.84c | 37.12c | 36.51cd |
| 4 | 42.19b | 28.18cd | 34.97de | 36.32cd |
| 5 | 43.05ab | 29.02cd | 36.02cd | 36.40cd |
| 6 | 40.29c | 27.15e | 34.10e | 37.28b |
| P = 0.0002 | P < 0.0001 | P < 0.0001 | P < 0.0001 | |
| SEM = 1.63 | SEM = 0.59 | SEM = 0.81 | SEM = 0.23 |
a–eLeast squares means with different letters are significantly different (P < 0.05).
SEM = standard error of mean.
Metmyoglobin-reducing ability
There was a significant (P < 0.05) oxygen exposure × retail day effect on the metmyoglobin-reducing activity of the steaks (Table 3). The oxygen-exposed surface had a lower (P < 0.05) metmyoglobin-reducing activity compared with the non–oxygen-exposed surface on day 3 and day 6 of retail. There were limited changes (P > 0.05) in metmyoglobin-reducing activity of the non–oxygen-exposed surface through retail display.
Effect of oxygen exposure1 on the metmyoglobin-reducing ability2 (metmyoglobin-reducing activity) of the longissimus lumborum steaks (n = 7)
| Retail display day | |||
|---|---|---|---|
| Oxygen exposure | 0 | 3 | 6 |
| Not exposed | 0.92a | 0.93a | 0.93a |
| Exposed | 0.92a | 0.77b | 0.76b |
| P < 0.0001 | SEM = 0.02 | ||
Least-squares means with different letters are significantly different (P < 0.05).
Exposure of muscle to oxygen as display surface indicated as exposed or interior of the muscle as not exposed.
Metmyoglobin formation after submersion in sodium nitrite solution, determined by K/S ratio of metmyoglobin (K/S 572/K/S525), where a greater number indicates greater reduction.
SEM = standard error of mean.
Oxygen consumption
There was a significant (P < 0.05) oxygen exposure × retail day effect on the oxygen consumption of steaks after 30 min of incubation (Table 4; when oxygen consumption was calculated as changes in oxymyoglobin). The non–oxygen-exposed surface of the steaks had greater (P < 0.05) oxygen consumption than the oxygen-exposed surface on day 3 of retail display, whereas there was no (P > 0.05) difference on day 6 of display. The oxygen exposure did not greatly impact the oxygen consumption of the longissimus lumborum. Oxygen exposure and retail display did not have a significant (P > 0.05) impact on oxygen consumption after 60 min.
Least-squares means of oxygen consumption2 (retail display day × oxygen exposure1) of longissimus lumborum steaks (n = 7) displayed for 6 d
| Retail display day | |||
|---|---|---|---|
| Oxygen exposure | 0 | 3 | 6 |
| Not exposed | 0.14a | 0.15a | 0.12ab |
| Exposed | 0.14a | 0.09b | 0.12a |
| P = 0.0012 | SEM = 0.01 | ||
Least squares means with different letters are significantly different (P < 0.05).
Exposure of muscle to oxygen as display surface indicated as exposed or interior of the muscle as not exposed.
Change in oxymyoglobin formation before and after 30 min of incubation determined by the change in K/S ratio of oxymyoglobin (pre-incubation K/S610/K/S525 – post-incubation K/S610/K/S525).
There was no difference (P > 0.05) in a* values after the 1-h bloom of the non–oxygen-exposed surface and the oxygen-exposed surface after retail display. There was a significant (P < 0.05) decrease in bloom a* values of the non–oxygen-exposed surface as retail display increased (data not included).
There were significant (P < 0.05) main effects of retail display duration and oxygen exposure on the metmyoglobin formation during oxygen consumption (Table 5 and Table 6, respectively). There was less metmyoglobin (P < 0.05) formed during oxygen consumption on day 0 compared to days 3 and 6 of display. The non–oxygen-exposed surface had less metmyoglobin (P < 0.05) formed during oxygen consumption compared with the oxygen-exposed surface. Incubation for 60 min resulted in no significant (P > 0.05) effect of retail display day and oxygen exposure on metmyoglobin formation during oxygen consumption.
Effects of retail display on oxygen consumption of longissimus lumborum steaks (n = 7) displayed for 6 d
| Retail display day | Metmyoglobin1 |
|---|---|
| 0 | 2.04a |
| 3 | 1.67b |
| 6 | 1.58b |
| P = 0.0028 | SEM = 0.09 |
Least-squares means with different letters are significantly different (P < 0.05).
Oxygen consumption measured as metmyoglobin formation after anaerobic incubation. It was calcuated as K/S ratio (K/S 572 divided by K/S525), where a greater number indicates greater oxygen consumption.
SEM = standard error of mean.
Effect of oxygen exposure1 on myoglobin oxidation by oxygen consumption of longissimus lumborum steaks (n = 7) displayed for 6 d
| Oxygen exposure | Metmyoglobin2 |
|---|---|
| Not exposed | 1.86a |
| Oxygen exposed | 1.67b |
| P = 0.0282 | SEM = 0.08 |
Least-squares means with different letters are significantly different (P < 0.05).
Exposure of muscle to oxygen as display surface indicated as exposed or interior of the muscle as not exposed.
Metmyoglobin formation after anaerobic incubation, determined by K/S ratio of metmyoglobin (K/S 572 ÷ K/S525), where a greater number indicates lower metmyoglobin.
SEM = standard error of mean.
Oxygen penetration
The surface layer depth of the oxygen-exposed surface was measured as the visual layer of oxy- and metmyoglobin after retail display. There was a significant effect (P < 0.05) of the retail display day on the oxygen-exposed surface layer depth (Table 7). On day 6 of retail, there was a larger (P < 0.05) surface layer of oxy- and metmyoglobin compared with day 0 of retail display. There was not a significant (P > 0.05) day effect on the oxygen depth of the non–oxygen-exposed surface after a 1-h bloom. Furthermore, the surface layer of the non–oxygen-exposed surface was measured after a 1-h bloom at 4°C, and there was not a significant (P > 0.05) day effect on the surface layer of the non–oxygen-exposed surface.
Effect of retail display day on percent surface layer depth1 of longissimus lumborum steaks (n = 7) displayed for 6 d
| Retail display day | Surface layer (%) after retail |
|---|---|
| 0 | 0.00b |
| 3 | 60.23a |
| 6 | 66.23a |
| P < 0.0001 | SEM = 5.82 |
Least-squares means with different letters are significantly different (P < 0.05).
Depth of the layer of oxymyoglobin and metmyoglobin on the oxygen-exposed surface of muscle after retail display divided by the size of the piece where the surface layer was measured. % Surface layer = (Depth after retail/Piece size) × 100.
SEM = standard error of mean.
Correlation analysis
There was a significantly strong correlation (r = −0.86; P < 0.05) between retail display day and a* values as expected (Table 8). The oxygen-exposed surface layer oxygen depth was strongly correlated (r = −0.83; P < 0.05) with a* values. Therefore, the development of a larger oxygen penetration layer of oxymyoglobin and metmyoglobin under the retail surface correlates with a decrease in redness. The a* values and non–oxygen-exposed oxygen depth were strongly correlated (r = 0.54, P < 0.05), indicating a greater a* value aligned with a larger oxygen penetration upon bloom. The a* values and oxygen-exposed metmyoglobin-reducing activity had a strong correlation (r = 0.77; P < 0.05). The decline in redness of the retail surface correlated with a decrease in the metmyoglobin-reducing activity of the oxygen-exposed surface. However, the non–oxygen-exposed metmyoglobin-reducing activity was weakly correlated (r = −0.12; P > 0.05) with a* values. Oxygen consumption at all time periods and surfaces was weakly to moderately correlated with a* values. Metmyoglobin formed during oxygen consumption (after 30 min) was strongly correlated (r = 0.68; P < 0.05) with a* values, indicating a higher value or lower metmyoglobin content aligned with a higher a* value during retail display. However, traditional oxygen consumption methodology (measured as changes in oxymyoglobin levels) was limited in connecting to changes in the color of the longissimus lumborum muscle during retail display. Conversely, metmyoglobin formed during oxygen consumption measurements demonstrated viability to connect to retail color stability.
Pearson’s correlation between a* values and pull day, lipid oxidation, metmyoglobin-reducing activity, oxygen consumption, metmyoglobin formation, and oxygen depth
| Parameter | Correlation | P value |
|---|---|---|
| Pull day | −0.86* | <0.0001 |
| Not exposed lipid oxidation | 0.14 | 0.54 |
| Exposed lipid oxidation | 0.16 | 0.48 |
| Not exposed metmyoglobin-reducing activity | −0.12 | 0.60 |
| Exposed metmyoglobin-reducing activity | 0.77* | <0.0001 |
| Not exposed oxygen consumption at 30 min (changes in oxymyoglobin) | 0.18 | 0.43 |
| Exposed oxygen consumption at 30 min (changes in oxymyoglobin) | 0.25 | 0.28 |
| Not exposed oxygen consumption at 60 min (changes in oxymyoglobin) | 0.42 | 0.06 |
| Exposed oxygen consumption at 60 min (changes in oxymyoglobin) | 0.36 | 0.11 |
| Not exposed metmyoglobin formed during oxygen consumption at 30 min | 0.30 | 0.19 |
| Exposed metmyoglobin formed during oxygen consumption at 30 min | 0.68* | 0.0006 |
| Not exposed metmyoglobin formed during oxygen consumption at 60 min | 0.24 | 0.29 |
| Exposed metmyoglobin formed during oxygen consumption at 60 min | −0.09 | 0.70 |
| Not exposed surface layer oxygen depth (%) | −0.15 | 0.51 |
| Exposed surface layer oxygen depth (%) | −0.83* | <0.0001 |
| Not exposed oxygen depth (%) | 0.54* | 0.01 |
*Significant correlation values indicated.
The correlation coefficient values were considered weak at r < |0.35|, moderate at |0.36| ≤ r < |0.67|, and strong at r ≥ |0.68| (Bohrer and Boler, 2017).
Regression analysis
The simple linear regression analysis evaluated the ability of parameters to predict a* values during retail display. There was a significant (R2 = 0.29; P < 0.05) R-squared value for the non–oxygen-exposed oxygen depth (Table 9). However, the R-squared value was below 0.46, meaning the non–oxygen-exposed depth indicated a moderate relationship between a* values and non–oxygen-exposed oxygen depth. Additionally, a significant (R2 = 0.59; P < 0.05) R-squared between the oxygen-exposed surface metmyoglobin-reducing activity and a* values demonstrated a strong relationship between the 2 parameters. Oxygen-exposed metmyoglobin-reducing activity was a strong predictor (R2 = 0.59) of a* values in retail. There was a strong relationship (R2 = 0.47; P < 0.05) between the metmyoglobin formed during oxygen consumption after 30 min of incubation and a* values during retail display. The R-squared value between the oxygen-exposed surface layer oxygen depth and a* values was the strongest (R2 = 0.69; P < 0.05), indicating that oxygen-exposed surface layer oxygen depth was a strong predictor of a* values. Overall, the oxygen-exposed surface color stability parameters best represented the changes in color during retail display.
Regression between a* values and lipid oxidation, metmyoglobin-reducing activity, oxygen consumption, metmyoglobin formation, and oxygen depth
| Parameter | B0 (intercept) | B1 (slope) | R2 | P value |
|---|---|---|---|---|
| Not exposed lipid oxidation | 29.63 | 3.31 | 0.02 | 0.54 |
| Exposed lipid oxidation | 29.13 | 4.72 | 0.03 | 0.47 |
| Not exposed metmyoglobin-reducing activity | 38.30 | −8.11 | 0.02 | 0.60 |
| Exposed metmyoglobin-reducing activity | 3.14 | 33.92 | 0.59* | <0.0001 |
| Not exposed oxygen consumption at 30 min (changes in oxymyoglobin) | 27.75 | 22.34 | 0.03 | 0.43 |
| Exposed oxygen consumption at 30 min (changes in oxymyoglobin) | 28.27 | 21.60 | 0.06 | 0.27 |
| Not exposed oxygen consumption at 60 min (changes in oxymyoglobin) | 23.81 | 38.45 | 0.18 | 0.06 |
| Exposed oxygen consumption at 60 min (changes in oxymyoglobin) | 24.18 | 35.89 | 0.13 | 0.11 |
| Not exposed metmyoglobin formed during oxygen consumption at 30 min | 25.12 | 3.06 | 0.09 | 0.18 |
| Exposed metmyoglobin formed during oxygen consumption at 30 min | 20.16 | 6.35 | 0.47* | 0.0006 |
| Not exposed metmyoglobin formed during oxygen consumption at 60 min | 25.12 | 3.06 | 0.01 | 0.18 |
| Exposed metmyoglobin formed during oxygen consumption at 60 min | 26.88 | 2.38 | 0.06 | 0.28 |
| Not exposed surface layer oxygen depth (%) | 31.62 | −0.02 | 0.02 | 0.51 |
| Exposed surface layer oxygen depth (%) | 34.49 | −0.09 | 0.69* | <0.0001 |
| Not exposed oxygen depth (%) | 21.91 | 0.10 | 0.29* | 0.01 |
*Significant regression values indicated.
Feature selection
Feature selection is the process of choosing the most relevant variables (features) for a statistical model. In the current analysis, lasso regression has the highest mean cross-validation score, followed by random forest and recursive feature elimination. Hence, lasso regression was used to understand the importance of predicting redness (a* values). The top 4 features were oxygen penetration on the exposed surface, metmyoglobin formed during oxygen consumption on the exposed surface, oxygen consumption calculated as changes in oxymyoglobin in the interior, and metmyoglobin-reducing activity measured on the oxygen-exposed surface as initial metmyoglobin formation (Figure 2).
Lasso regression feature importance in predicting the redness of beef longissimus steaks during retail display. A greater value indicates a better predictor.
Discussion
It has been estimated that surface discoloration costs the meat industry approximately $3.73 billion annually (Ramanathan et al., 2022). Therefore, determining parameters to predict color stability and mitigate surface discoloration will reduce economic loss. Various biochemical parameters are quantified to better understand meat color changes. Understanding the contribution of each factor helps to predict color stability. In the current research, a color-stable muscle was used to determine the relationship between oxygen consumption, oxygen penetration, and metmyoglobin-reducing activity on color stability. Research has shown aerobic packaging is less color stable in comparison with anaerobic packaging (Griffin et al., 1982; Lavieri and Williams, 2014; Mohan et al., 2010; Reyes et al., 2022). In addition, aerobic packaging such as PVC has more metmyoglobin formation on the surface during retail display than anaerobic vacuum packaging of beef longissimus lumborum steaks (Mohan et al., 2010; Reyes et al., 2022). High regression and strong negative correlation values of the retail a* and oxygen-exposed surface layer depth support the greater penetration of oxygen during retail display as an indicator for the loss of redness and color stability. Anaerobic vacuum packaging has been reported to have a higher bloom of the longissimus lumborum muscle after storage compared to PVC packaging (Fu et al., 1992). The current study contrasted the impact of oxygen on bloom, as the non–oxygen-exposed and oxygen-exposed surfaces had no difference in bloom a* values and retail a* values. No differences in bloom might be due to the color-stable nature of longissimus lumborum muscle. Oxygen penetration during retail provides insight into color stability but has a limited impact on the oxygenation of the color-stable longissimus lumborum muscle.
Research has revealed that the oxidation of lipids and myoglobin is interconnected with increased lipid oxidation connected to increased myoglobin oxidation (Faustman et al., 2010; Lynch and Faustman, 2000). Previous research has reported greater lipid oxidation in PVC-packaged longissimus lumborum steaks compared with vacuum packaging with greater display time (Fu et al., 1992; Reyes et al., 2022). The present study did not indicate differences in lipid oxidation and limited changes in hue angle when a* values declined. As previous research has indicated, the longissimus lumborum muscle is considered color stable based on its predominant glycolytic metabolism (McKenna et al., 2005; Seyfert et al., 2006). In addition, increased antioxidant capacity (Joseph et al., 2012) observed in the longissimus lumborum muscle may relate to the limited changes during retail display. The limited effects of oxygen exposure on lipid oxidation and color stability are supported by the low regression and correlation values for a* values and lipid oxidation of both surfaces, as well as minimal impact on oxygen consumption in the present study. The impact of atmospheric oxygen levels on protein oxidation was more variable, with some studies reporting higher protein oxidation levels (Bao and Ertbjerg, 2015) and others with minimal impact on protein oxidation compared to anaerobically packaged pork chops (Spanos et al., 2016). Although protein oxidation was not evaluated in the present study, its effects could explain the differences reported in metmyoglobin-reducing activity based on oxygen exposure.
The American Meat Science Association Meat Color Guidelines (King et al., 2023) recommends either the retail surface or the interior surface for metmyoglobin-reducing activity and oxygen consumption determination. Past research evaluating metmyoglobin-reducing activity and oxygen consumption in relation to color stability has limited focus on oxygen exposure, and many studies do not indicate the surface used. Several studies have reported using the oxygen-exposed surface for metmyoglobin-reducing activity (Canto et al., 2016; Joseph et al., 2012; Salim et al., 2019; Seyfert et al., 2006; Wu, Han, et al., 2020; Wu, Luo, et al., 2020) and the non–oxygen-exposed surface for oxygen consumption (Wu, Han, et al., 2020; Wu, Luo, et al., 2020) in longissimus lumborum steaks. Studies using the non–oxygen-exposed surface reported a decrease in oxygen consumption during retail display for steaks (Wu et al., 2020). An oxygen-exposed surface to evaluate metmyoglobin-reducing activity has demonstrated a decline in metmyoglobin-reducing activity during retail display (Canto et al., 2016; Seyfert et al., 2006) or no change in metmyoglobin-reducing activity (Salim et al., 2019; Wu, Han, et al., 2020) for longissimus lumborum steaks. Some previous studies used 2 non–oxygen-exposed surfaces for metmyoglobin-reducing activity and oxygen consumption, reporting decreased metmyoglobin-reducing activity and oxygen consumption during retail display (Abraham et al., 2017; Ke et al., 2017). By evaluating oxygen exposure in the present study, the variations in oxygen consumption and metmyoglobin-reducing activity may be mitigated. Furthermore, the implications of oxygen on the predictability of color stability of metmyoglobin-reducing activity and oxygen consumption should be considered moving forward.
Current results indicate the potential of oxygen-exposed metmyoglobin-reducing activity to represent changes in color stability when compared with oxygen consumption and metmyoglobin-reducing activity using non–oxygen-exposed and oxygen-exposed surfaces. Mancini et al. (2008) evaluated the differences between the metmyoglobin-reducing activity of the non–oxygen-exposed and oxygen-exposed surfaces. These researchers reported the non–oxygen-exposed portions had greater metmyoglobin-reducing activity after retail display than the oxygen-exposed portions of longissimus lumborum steaks, supporting results presented in the current study. This result is supported by the general loss of metmyoglobin-reducing activity during retail display seen in past research (Canto et al., 2016; Seyfert et al., 2006). In addition, the non–oxygen-exposed surface metmyoglobin-reducing activity had a weak correlation to the color stability in retail (Mancini et al., 2008), supporting the present results. Mancini et al. (2008) speculated the limited changes in the non–oxygen-exposed surface metmyoglobin-reducing activity could be due to lower oxygen penetration and limited photooxidation. Based on the results of the present study and the results of Mancini et al. (2008), the oxygen-exposed surface metmyoglobin-reducing activity would represent the color stability of the longissimus lumborum muscle in retail compared to the metmyoglobin-reducing activity of the non–oxygen-exposed surface. To our knowledge, no study has compared the use of the non–oxygen-exposed and oxygen-exposed surface on the results of oxygen consumption to develop a better understanding of the oxygen exposure and determine the surface representative of the retail color stability. However, the weak relationship and weak to moderate correlation between a* value during display and the oxygen consumption of both oxygen-exposed and non–oxygen-exposed surfaces indicates that the oxygen consumption determined by oxymyoglobin changes may not represent color stability. Metmyoglobin formed via oxygen consumption of the oxygen-exposed surface may be a more suitable method for evaluating color stability during storage, as supported by the strong relationship with a* values. This methodology represented a reduction of formed metmyoglobin under anaerobic conditions as well as oxygen consumption of the longissimus lumborum as the steaks transitioned from oxymyoglobin to deoxymyoglobin via metmyoglobin. Therefore, greater metmyoglobin formation via oxygen consumption indicated both low reduction capacity and low oxygen consumption, increasing oxidation of myoglobin. Metmyoglobin formation in oxygen-exposed areas via oxygen consumption and metmyoglobin-reducing activity in oxygen-exposed surfaces represented the changes in color stability of the longissimus lumborum during retail display better than the non–oxygen-exposed surface. Oxygen exposure is a key attribute to consider when evaluating color stability in retail.
Conclusion
The longissimus lumborum muscle is classified as a color-stable muscle, and the limited effects of oxygen exposure on the longissimus lumborum during retail continue to support color stability. Although oxygen exposure had a detrimental impact on oxygen consumption and metmyoglobin-reducing activity, the non–oxygen-exposed surface reported consistent oxygen penetration and no difference in a* values upon bloom compared with the oxygen-exposed surface. The new method for assessing metmyoglobin formation during oxygen consumption has the potential to better represent color changes in retail settings compared to the current approach, which relies on measuring changes in oxymyoglobin levels. In addition, changes in a* values during retail display were best represented by the oxygen exposed by oxygen consumption, oxygen-exposed surface layer depth, non–oxygen-exposed oxygen depth, and oxygen-exposed metmyoglobin-reducing activity. The influence of oxygen on color stability attributes should be considered in future studies.
Conflict of Interest
The authors declare no conflicts of interest regarding the content of this manuscript.
Acknowledgements
This research was supported, in part, by Ranjith Ramanathan’s Leo and Kathy Noltensmeyer Endowed Research Chair funds.
Author Contributions
Morgan Denzer conducted the study, collected data, analyzed data, and wrote the draft manuscript; Gretchen Mafi edited the manuscript; Morgan Pfeiffer edited the manuscript; Ranjith Ramanathan edited the manuscript, secured funding, and provided supervision.
Literature Cited
Abraham, A., J. W. Dillwith, G. G. Mafi, D. L. VanOverbeke, and R. Ramanathan. 2017. Metabolite profile differences between beef longissimus and psoas muscles during display. Meat Muscle Biol. 1. doi: https://doi.org/10.22175/mmb2016.12.0007
Arihara, K., R. G. Cassens, M. L. Greaser, J. B. Luchansky, & P. E. Mozdziak. 1995. Localization of metmyoglobin-reducing enzyme (NADH-cytochrome b5 reductase) system components in bovine skeletal muscle. Meat Sci. 39:205–213. doi: https://doi.org/10.1016/0309-1740(94)p1821-c
Bao, Y., and P. Ertbjerg. 2015. Relationship between oxygen concentration, shear force and protein oxidation in modified atmosphere packaged pork. Meat Sci. 110:174–179. doi: https://doi.org/10.1016/j.meatsci.2015.07.022
Brown, W., and H. Snyder. 1969. Nonenzymatic reduction and oxidation of myoglobin and hemoglobin by nicotinamide adenine dinucleotides and flavins. J. Biol. Chem. 244:6702–6706.
Bohrer, B. M., and D. D. Boler. 2017. Review: Subjective pork quality evaluation may not be indicative of instrumental pork quality measurements on a study-to-study basis. The Professional Animal Scientist 33:530–540. doi: https://doi.org/10.15232/pas.2017-01644
Canto, A., B. R. C. Costa-Lima, S. P. Suman, M. L. G. Monteiro, F. M. Viana, A. Salim, C. A. Conte-Junior. 2016. Color attributes and oxidative stability of longissimus lumborum and psoas major muscles from Nellore bulls. Meat Sci. 121:19–26. doi: https://doi.org/10.1016/j.meatsci.2016.05.015
Chandrashekar, G., and F. Sahin. 2014. A survey on feature selection methods. Comput. Electr. Eng. 40(1):16–28. doi: https://doi.org/10.1016/j.compeleceng.2013.11.024
Denzer, M., C. Mowery, H. Comstock, N. B. Maheswarappa, G. G. Mafi, D. L. VanOverbeke, and R. Ramanathan. 2020. Characterization of the cofactors involved in non-enzymatic metmyoglobin/methemoglobin reduction in vitro. Meat Muscle Biol. 4. doi: https://doi.org/10.22175/mmb.9507
Denzer, M. L., G. G. Mafi, D. L. VanOverebeke, and R. Ramanathan. 2022. Effects of glucono delta-lactone enhancement and nitrite-embedded packaging on fresh color, cooked color, and sensory attributes of dark-cutting beef. Applied Food Research 2:100189. doi: https://doi.org/10.1016/j.afres.2022.100189
Denzer, M. L., M. Pfeiffer, G. G. Mafi, and R. Ramanathan. 2023. Metabolomics and bioinformatic analyses to determine the effects of oxygen exposure within longissimus lumborum steak on beef discoloration, J. Anim. Sci. 101. doi: https:/doi.org/10.1093/jas/skad155
Denzer, M. L., M. Pfeiffer, G. G. Mafi, and R. Ramanathan. 2025. The importance of including metmyoglobin levels in reflectance-based oxygen consumption measurements. Meat Sci. 219. doi: https://doi.org/10.1016/j.meatsci.2024.109651
Dokeroglu, T., A. Deniz, and H.E. Kiziloz. 2022. A comprehensive survey on recent metaheuristics for feature selection. Neurocomputing 494:269–296. doi: https://doi.org/10.1016/j.neucom.2022.04.083
English, A. R., K. M. Wills, B. N. Harsh, G. G. Mafi, D. L. VanOverbeke, & R. Ramanathan. 2016. Effects of aging on the fundamental color chemistry of dark-cutting beef. J. Anim. Sci. 94:4040–4048. doi: https://doi.org/10.2527/jas.2016-0561
Faustman, C., Q. Sun, R. Mancini, and S. P. Suman. 2010. Myoglobin and lipid oxidation interactions: mechanistic bases and control. Meat Sci. 86:86–94. doi: https://doi.org/10.1016/j.meatsci.2010.04.025
Fu, A.-H., R. A. Molins, and J. G. Sebranek. 1992. Storage quality characteristics of beef rib eye steaks packaged in modified atmospheres. J. Food Sci. 57:283–287. doi: https://doi.org/10.1111/j.1365-2621.1992.tb05477.x
Gredell, D. A., A. R. Schroeder, K. E. Belk, C. D. Broeckling, A. L. Heuberger, et al. 2019. Comparison of machine learning algorithms for predictive modeling of beef attributes using rapid evaporative ionization mass spectrometry (REIMS) data. Sci. Rep. 9(1):5721. doi: https://doi.org/10.1038/s41598-019-40927-6
Griffin, D. B., J. W. Savell, G. C. Smith, C. Vanderzant, R. N. Terrell, K. D. Lind, and D. E. Galloway. 1982. Centralized packaging of beef loin steaks with different oxygen-barrier films: physical and sensory characteristics. J. Food Sci. 47:1059–1069. doi: https://doi.org/10.1111/j.1365-2621.1982.tb07621.x
Hood, D. E., and E. B. Riordan. 1973. Discolouration in pre-packaged beef: measurement by reflectance spectrophotometry and shopper discrimination. Int. J. Food Sci. Tech. 8:333–343. doi: https://doi.org/10.1111/j.1365-2621.1973.tb01721.x
Joseph, P., S. P. Suman, G. Rentfrow, S. Li, and C. M. Beach. 2012. Proteomics of muscle-specific beef color stability. J. Agr. Food Chem. 60:3196–3203. doi: https://doi.org/10.1021/jf204188v
Ke, Y., R. M. Mitacek, A. Abraham, G. G. Mafi, D. L. VanOverbeke, U. DeSilva, and R. Ramanathan. 2017. Effects of muscle-specific oxidative stress on cytoxygen consumption hrome c release and oxidation-reduction potential properties. J. Agr. Food Chem. 65:7749–7755. doi: https://doi.org/10.1021/acs.jafc.7b01735
King, D. A., M. C. Hunt, S. Barbut, J. R. Claus, D. P. Cornforth, P. Joseph, Y. H. B. Kim, G. Lindahl, R. A. Mancini, M. N. Nair, K. J. Merok, A. Milkowski, A. Mohan, F. Pohlman, R. Ramanathan, C. R. Raines, M. Seyfert, O. Sørheim, S. P. Suman, and M. Weber. 2023. American Meat Science Association guidelines for meat color measurement. Meat Muscle Biol. 6. doi: https://doi.org/10.22175/mmb.12473
Lavieri, N., and S. K. Williams. 2014. Effects of packaging systems and fat concentrations on microbiology, sensory and physical properties of ground beef stored at 4±1°C for 25 days. Meat Sci. 97:534–541. doi: https://doi.org/10.1016/j.meatsci.2014.02.014
Limsupavanich, R., D. Kropf, M. C. Hunt, E. Boyle, D. Boyle, and T. Loughin. 2008. Myoglobin layer depth, surface myoglobin and visual colour-life of beef Longissimus and Psoas major steaks from various postmortem times. Paper presented at: 54th International Congress of Meat Science and Technology (54th ICoMST), Cape Town, South Africa, August 10–15.
Lynch, M. P., and C. Faustman. 2000. Effect of aldehyde lipid oxidation products on myoglobin. J. Agr. Food Chem. 48:600–604. doi: https://doi.org/10.1021/jf990732e
Lybarger, K. R., E. S. Beyer, K. J. Farmer, L. A. Egger, L. N. Drey, M. C. Hunt, J. L. Vipham, M. D. Zumbaugh, M. D. Chao, and T. G. O’Quinn. 2023. Determination of consumer color and discoloration thresholds for purchase of representative retail ground beef. Meat Muscle Biol. 7:1–19. doi: https://doi.org/10.22175/mmb.16757
Mancini, R. A., M. Seyfert, and M. C. Hunt. 2008. Effects of data expression, sample location, and oxygen partial pressure on initial nitric oxide metmyoglobin formation and metmyoglobin-reducing-activity measurement in beef muscle. Meat Sci, 79:244–251. doi: https://doi.org/10.1016/j.meatsci.2007.09.008
Mancini, R. A., R. Ramanathan, M. C. Hunt, D. H. Kropf, and G. G. Mafi. 2022. Interrelationships between visual and instrumental measures of ground beef color. Meat Muscle Biol. 6:1–8. doi: https://doi.org/10.22175/mmb.14040
McKenna, D. R., P. D. Mies, B. E. Baird, K. D. Pfeiffer, J. W. Ellebracht, and J. W. Savell. 2005. Bichemical and physical factors affecting discoloration characteristics of 19 bovine muscles. Meat Sci. 70:665–682. doi: https://doi.org/10.1016/j.meatsci.2005.02.016
Mohan, A., M. C. Hunt, T. J. Barstow, T. A. Houser, and D. M. Hueber. 2010. Near-infrared oximetry of three post-rigor skeletal muscles for following myoglobin redox forms. Food Chem. 123:456–464. doi: https://doi.org/10.1016/j.foodchem.2010.04.068
O’Keeffe, M., and D. E. Hood. 1982. Biochemical factors influencing metmyoglobin formation on beef from muscles of differing colour stability. Meat Sci. 7:209–228. doi: https://doi.org/10.1016/0309-1740(82)90087-0
Ramanathan, R., L. Lambert, M. N. Nair, B. Morgan, R. Feuz, G. Mafi, and M. M. Pfeiffer. 2022. Economic loss, amount of beef discarded, natural resources wastage, and environmental impact due to beef discoloration. Meat Muscle Biol. 6:1–8. doi: https://doi.org/10.22175/mmb.13218
Ramanathan, R., M. N. Nair, M. C. Hunt, and S. P. Suman. 2019. Mitochondrial functionality and beef colour: A review of recent research. S. Afr. J. Anim. Sci. 49:9–19. doi: https://doi.org/10.4314/sajas.v49i1.2
Ramanathan, R., S. P. Suman, and C. Faustman. 2020. Biomolecular interactions governing fresh meat color in post-mortem skeletal muscle: A Review. J. Agr. Food Chem. 68:12779–12787. doi: https://doi.org/10.1021/acs.jafc.9b08098
Reyes, T. M., M. P. Wagoner, V. E. Zorn, M. M. Coursen, B. S. Wilborn, T. Bonner, J. T. Sawyer. 2022. Vacuum packaging can extend fresh color characteristics of beef steaks during simulated display conditions. Foods 11. doi: https://doi.org/10.3390/foods11040520
Salim, A., S. P. Suman, A. Canto, B. R. C. Costa-Lima, F. M. Viana, M. L. G. Monteiro, C. A. Conte-Junior. 2019. Muscle-specific color stability in fresh beef from grain-finished Bos indicus cattle. Asian Austral. J. Anim. 32:1036–1043. doi: https://doi.org/10.5713/ajas.18.0531
Sammel, L. M., M. C. Hunt, D. H. Kropf, K. A. Hachmeister, and D. E. Johnson. 2002. Comparison of assays for metmyoglobin reducing ability in beef inside and outside semimembranosus muscle. J. Food Sci. 67:978–984. doi: https://doi.org/10.1111/j.1365-2621.2002.tb09439.x
Seyfert, M., R. A. Mancini, M. C. Hunt, J. Tang, C. Faustman, and M. Garcia. 2006. Color stability, reducing activity, and cytochrome c oxidase activity of five bovine muscles. J. Agr. Food Chem., 54:8919–8925. doi: https://doi.org/10.1021/jf061657s
Spanos, D., M. A. Tørngren, M. Christensen, and C. P. Baron. 2016. Effect of oxygen level on the oxidative stability of two different retail pork products stored using modified atmosphere packaging (MAP). Meat Sci. 113:162–169. doi: https://doi.org/10.1016/j.meatsci.2015.11.021
Tang, J., C. Faustman, R. A. Mancini, M. Seyfert, and M. C. Hunt. 2005. Mitochondrial reduction of metmyoglobin: dependence on the electron transport chain. J. Agr. Food Chem. 53:5449–5455. doi: https://doi.org/10.1021/jf050092h
Witte, V. C., G. F. Krause, and M. E. Bailey. 1970. A new extraction method for determining 2-thiobarbituric acid values of pork and beef during storage. J. Food Sci. 35:582–585. doi: https://doi.org/10.1111/j.1365-2621.1970.tb04815.x
Wu, S., J. Han, R. Liang, P. Dong, L. Zhu, D. L. Hopkins, X. Luo. 2020. Investigation of muscle-specific beef color stability at different ultimate pHs. Asian Austral. J. Anim. 33:1999–2007. doi: https://doi.org/10.5713/ajas.19.0943