Scholarly article on topic 'Phenotypic characterisation of colour stability of lamb meat'

Phenotypic characterisation of colour stability of lamb meat Academic research paper on "Animal and dairy science"

Share paper
Academic journal
Meat Science
OECD Field of science
{"Lamb meat" / "Colour stability" / "Phenotypic trait"}

Abstract of research paper on Animal and dairy science, author of scientific article — Robin H. Jacob, Mario F. D'Antuono, Arthur R. Gilmour, Robyn D. Warner

Abstract A study was undertaken, using 2701 overwrapped loin samples aged for 5days and subjected to a simulated retail display (SRD) for 3days; sourced from lambs in the Cooperative Research Centre for Sheep Industry Innovation information nucleus flock, born 2007–2009. The ratio of reflectance of light in the wavelengths of 630nm and 580nm (oxy/met) was measured daily during the SRD, using a Hunterlab spectrophotometer. A series of linear mixed models was fitted to the oxy/met and time data to compare 4 breed types and identify relevant covariates, of 19, using a forward selection process. Breed type, pH at 24h post slaughter and Linoleic acid concentration (LA) were the most important factors and covariates, in that order. Merino breed type, high pH and high LA reduced colour stability. Fitting a spline model to predict the time for oxy/met to reach a set value, represents an alternative to comparing oxy/met at a set time, for describing colour stability.

Academic research paper on topic "Phenotypic characterisation of colour stability of lamb meat"


science science meat scienck


science science B^HSCIENCE science science


MESC-05891; No of Pages 9

Meat Science xxx (2013) xxx-xxx

Phenotypic characterisation of colour stability of lamb meat

Robin H. Jacob a,b,*I Mario F. D'Antuono b, Arthur R. Gilmourc, Robyn D. Warner d

a Cooperative Research Centre for Sheep Industry Innovation, University of New England, Armidale, NSW 2351, Australia b Department of Agriculture and Food WA, 3 Baron Hay-Court, South Perth, WA 6151, Australia c Centre for Statistical and Survey Methodology, University of Wollongong, NSW 2522, Australia d Animal Food and Health Sciences, CSIRO, Private Bag 16, Werribee, Vic 3030, Australia



Available online xxxx

Keywords: Lamb meat Colour stability Phenotypic trait

A study was undertaken, using 2701 overwrapped loin samples aged for 5 days and subjected to a simulated retail display (SRD) for 3 days; sourced from lambs in the Cooperative Research Centre for Sheep Industry Innovation information nucleus flock, born 2007-2009. The ratio of reflectance of light in the wavelengths of 630 nm and 580 nm (oxy/met) was measured daily during the SRD, using a Hunterlab spectrophotometer. A series of linear mixed models was fitted to the oxy/met and time data to compare 4 breed types and identify relevant covariates, of 19, using a forward selection process. Breed type, pH at 24 h post slaughter and Linoleic acid concentration (LA) were the most important factors and covariates, in that order. Merino breed type, high pH and high LA reduced colour stability. Fitting a spline model to predict the time for oxy/met to reach a set value, represents an alternative to comparing oxy/met at a set time, for describing colour stability.

© 2012 Published by Elsevier Ltd.

1. Introduction

The finding that meat colour stability has a genetic component for both lamb meat (Mortimer et al., 2010) and beef (King et al., 2010) provides an opportunity to use animal genetics to increase the retail display time of meat. However a phenotypic trait that identifies individual animals' superior for colour stability has yet to be defined. Consumers perceive meat colour to be a cue for freshness and prefer lamb meat to be red in colour for this reason. On the other hand, meat has a tendency to change from red to brown after slicing and in the context of retail display this characteristic is known as colour stability. Meat that remains red during display is stable in colour, and lamb meat is regarded as less stable than beef (Gutzke & Trout, 2002). This has been attributed at least in part, to subtle differences in the sequence of amino acids in the globin moiety of myoglobin that interact with aldehyde compounds produced during oxidation (Faustman, Mancini, Sun, & Suman, 2010).

The best way to quantify stability of meat colour for genetic comparisons remains unclear, despite the basic mechanisms responsible for colour change in meat during retail display being largely understood (Faustman, 1990; Faustman et al., 2010). Simplistically, colour change in meat can be attributed to either oxygenation or oxidation of the pigment myoglobin. Myoglobin becomes red (oxymyoglobin) due to oxygenation whilst deoxygenated myoglobin is purple, and turns brown (metmyoglobin) due to oxidation (Faustman, 1990). After slicing, the meat surface typically changes in colour from purple

* Corresponding author at: Department of Agriculture and Food WA, 3 Baron Hay-Court, South Perth, WA 6151, Australia. Tel.: +61 8 9368 3470; fax: +61 8 9368 2905. E-mail address: (R.H. Jacob).

0309-1740/$ - see front matter © 2012 Published by Elsevier Ltd.! 0.1016/j.meatsci.2012.11.031

to red due to oxygenation, or blooming. This 'bloom' extends a few millimetres below the surface of meat (Krzywicki, 1979). Oxidation initiates at the junction of the oxygenated and deoxygenated layers where the oxygen partial pressure is low (Forrest, Aberle, Hedrick, Judge, & Merkel, 1975). This junction moves progressively towards the meat surface and in the process the surface colour changes from red to brown.

CIE-L* (black-white), a* (red-green) and b* (blue-yellow) values have traditionally been used to quantify meat colour, but reflectance of specific wavelengths in the visible spectrum 400-700 nm can be more precise in relation to the chemical form of myoglobin. The ratio of reflectance of light in the wavelengths of 630 nm and 580 nm, known as "oxy/met" has been used to detect chemical change in meat due to either oxygenation or oxidation of myoglobin (Hunt et al., 1991) and has been associated with consumer preference for colour (Khliji, van de Ven, Lamb, Lanza, & Hopkins, 2010; Morrissey, Jacob, & Pluske, 2008).

However calibration with chemical analyses is required to quantify the different chemical forms of myoglobin due to light reflectance being affected by other factors such as texture (Hunt et al., 1991). Cost and time make chemical analyses impractical for large genetics study, so developing a simple technique based on oxy/met measurement to compare samples is warranted.

Mortimer et al. (2010) described the heritability of oxy/met values for lamb meat at discrete time points of a simulated retail display period (SRD). However, differences between samples based on a single point measurement of oxy/met might be confounded by differences in texture. Subsequently Jacob, Mortimer, Hopkins, Warner, and D'Antuono (2011) demonstrated that the rate of change of oxy/met


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx

is a function of SRD time and hypothesised that colour change during SRD was due initially to oxygenation then later to oxidation of myoglobin. Visualisation of this trend may also be influenced by frequency of measurement, and changes in texture.

Management of meat colour stability can be achieved by manipulating a range of non-genetic factors such as vitamin E concentration and chill rate (Faustman, 1990). In practice these operate at different points along the lamb meat supply chain (Jacob, D'Antuono, Smith, Pethick, & Warner, 2007; Jose, Pethick, Gardner, & Jacob, 2008) and could also confound genetic evaluation of animals for meat colour if they interacted with genotype. Colour measurement protocols can be standardised and instrument specifications such as aperture are important when comparing results between studies (Hopkins, Kerr, Lamb, & Jacob, 2008; Kropf, 1993; Tapp, Yancey, & Apple, 2011). Meat is translucent so light measured at the surface is reflected from the surface and various depths from within the meat. Factors that affect the depth of the oxygenated layer (Krzywicki, 1979) such as chilling conditions during processing as speculated by Jacob and Thomson (2012) could potentially influence light reflectance and confound comparisons between animals for oxy/met as well.

The aim of this study was to describe the change in oxy/met value during simulated retail display, for loin meat from lambs produced in the Cooperative Research Centre for Sheep Industry Innovation ( information nucleus flock (INF). The purpose was to further refine the methodology for measuring colour stability of lamb meat to be used in genetic comparisons.

2. Methods

2.1. Lambs

Samples (3088) of m. Longissimus thoracis et lumborum (LL) were collected from lambs born and raised at 5 sites of the INF; Trangie, Cowra, Rutherglen, Hamilton, and Katanning. Lambs were slaughtered at 3-4 dates in each of the years 2007, 2008 and 2009 at 3 abattoirs, 2 of which slaughtered lambs from 2 sites. Lambs were slaughtered in batches that varied in size from 26 to 173 lambs. Lambs were stratified for live weight then allocated to slaughter batch on the basis that each batch was balanced for mean live weight, sire and gender where possible at each site.

The lambs were of 4 sire breed type and dam breed type combinations; Maternal breed ram and Merino ewe (MatM), Merino ram and Merino ewe (MM), Terminal breed ram and Merino ewe (TM), and Terminal ram and Maternal breed ewe (TMat). The breed types used and details of lamb feeding and management are described elsewhere (Ponnampalam et al., in press; van der Werf, Kinghorn, & Banks, 2010). The measurement of eye muscle area, carcass fat depth, hot carcass weight, pH24LL and pre-rigor pH and temperature on the carcass are outlined by Pearce (2009), and the calculation of the predicted pH at temperature 18 °C and the predicted temperature at pH 6 is described by van de Ven et al. (in press). Samples of muscle were collected at 24 h post-slaughter and frozen for subsequent measurement of intramuscular fat and myoglobin (see Warner et al. (2010) and for subsequent measurement of long chain fatty acids and minerals (iron and zinc) (see Pannier et al. (2010) for assay method). The number of lambs included in the analyses after excluding records where data fields were missing, for each site by breed combination, is presented in Table 1.

2.2. Colour measurement

LL samples were taken from each carcass 24 h after slaughter, packed under vacuum in clear plastic, gas impermeable packaging and stored at 2 °C for 5 days. The sample was a full cross section of the LL, about 60 mm in longitudinal length. After the 5 day ageing period individual samples were removed from their packaging, re-sliced

Table 1

The number of lambs measured for oxy/met from each breed type (Maternal x Merino, Merino x Merino, Terminalx Merino and Terminal x Maternal), site (Trangie, Cowra, Rutherglen, Hamilton, and Katanning) and year of birth (2007, 2008, and 2009).

Year of birth Site Total

Trangie Cowra Rutherglen Hamilton Katanning

Maternal x Merino

2007 0 19 23 14 65 121

2008 41 22 32 28 69 192

2009 35 41 43 30 51 200

Sub total 76 82 98 72 185 513

Merino xMerino

2007 0 28 0 0 1 29

2008 39 32 42 0 1 114

2009 35 26 41 23 37 162

Sub total 74 86 83 23 39 305

Terminal x Merino

2007 0 66 0 74 131 271

2008 68 74 0 126 220 488

2009 57 45 49 42 127 320

Sub total 125 185 49 242 478 1079

Terminal x Maternal

2007 0 40 232 0 0 272

2008 62 19 139 0 0 220

2009 71 87 74 80 0 312

Sub total 133 146 445 80 0 804

Total 408 501 675 417 702 2701

to a length of 30 mm, and then placed on a black polystyrene foam tray, with the muscle fibres running transversely to the direction of slicing. Samples were allowed to bloom for 30 min at a temperature of 2 °C before wrapping with polyvinyl chloride cling wrap (Resinite "DHW" Meat AEP, 15 |jm thickness, oxygen transmission rate of 35,650-46,500 cc/m2/24 h) and exposed to 72 h of continuous simulated retail display (SRD).

Temperature and light conditions during the SRD were designed to simulate those encountered in retail stores. The SRD was conducted in a cool room with air temperature kept in the range of — 2 to 6 °C. An overhead light source consisting of 8 Nelson Fluorescent Meat Display BRB Tubes 58 W and 1520 mm in length was suspended 1.5 m above the samples to provide a light intensity of 1000 Lux, as measured with a Dick Smith Electronics Light Meter Q1367.

A Hunter Lab Mini Scan(tm) XE Plus (Cat. No. 6352, model No. 45/ 0-L, 31.8 mm reading port) was used to measure light reflectance. The light source was set at "D65" with the observer set to 10°. The instrument was calibrated on a black glass then a white enamel tile as directed by the manufacturer's specifications. Colour measurements were taken in the cool room where the SRD was conducted and samples were left wrapped during measurement. Each measurement was repeated after rotating the spectrophotometer 90° in the horizontal plane, and the mean of these two measurements was used. Oxy/met was calculated by dividing the percentage of light reflectance at wavelength 630 nm by the percentage of light reflectance at wavelength 580 nm (Hunt et al., 1991). Samples with an oxy/met value greater than or equal to 3.5 were considered to be acceptable (red) and those less than 3.5 unacceptable (brown) in colour. Pure forms of oxymyoglobin and metmyoglobin have oxy/met values of 5 and 1 (Hunt et al., 1991) respectively and the midway point between these values is 3. However studies suggest that the oxy/met value at which consumers discriminate between red and brown meat colour varies from 3.5 (Morrissey et al., 2008) to 3.3 (Khliji et al., 2010). For our study, the more conservative estimate of 3.5 was used given that consumer responses are variable. Oxy/met may need to be greater than 6.3 to be confident that 95% of consumers selected at random will perceive meat colour to be acceptable (Khliji et al., 2010).


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx

2.3. Statistics

2.3.1. Statistical analyses

The R Statistical System (version R-2.15.0) (http://www.r-project. org/) was used to fit a series of linear mixed models with the ASReml-R package (version 3.0) ( to estimate the relationships between various factors and covariates with the oxy/ met response over time. A principal component analysis (PCA) of the standardised covariates was also done using the R Statistical System to test for correlations between the 19 covariates. This was done to assist with identifying covariates to include in the model. As no grouping of covariates was apparent (Table 4), all of the 19 covariates were included in the selection process.

Preliminary preparation of the database included removal of any records with missing data for any covariate. To avoid inconsistencies with model selection, since missing data was not missing at random, only complete records were used for the fitting of covariates in the model and the base model. The final data set consisted of 10,808 records with oxy/met values collected from 2701 lambs at 4 time points (days 0,1, 2, and 3) during the SRD. The six factors and 19 covariates in the data are listed in Table 2. For ease of summarizing the numerical computations, each of the covariates (x) was standardised (x—x)/Sx; x is the sample mean, and Sx is the standard deviation of x. The predictions shown in Fig. 1, Tables 5 and 6 were back-transformed to indicate the mean responses on the raw scale.

Table 2

The abbreviated names and descriptions of the factors (F terms) and covariates (x terms) used in the model selection.

Abbreviated name Description Term

DAM Unique identification number for the lambs F

dam (n = 1915)

DROP The calendar year in which the lamb was F

born (2007, 2008, 2009)

DATE The date the lamb was slaughtered (3-4 F

dates per year) in the given year

SITE The INF site from where the lamb was F

sourced (5 sites)

SIRE The lamb father's unique identification F

number (n = 325)

BRDX The breed type of the lambs sire and dam F

(MM, MatM, Tmat, TM )

TIME The time in days for the SRD (0, 1, 2, 3) F

AGEATOBS Age of the lamb at the time of slaughter x


AGEOFDAM Age of the lambs dam (years) x

EMA Cross sectional area of the LL (mm2) x

FAT5 Thickness of subcutaneous fat at the 5th x

rib (mm)

ALAa a Linolenic acid concentration in LL x

(mg/100 g)

EPAa Eicosapentaenoic acid in LL (mg/100 g) x

DHAa Docosahexaenoic acid concentration in LL x

(mg/100 g)

LAa Linoleic acid concentration in LL (mg/100 g) x

AAa Arachidonic acid concentration in LL x

(mg/100 g)

HCWT Hot carcass weight kg) x

HGRFAT Tissue depth at the GR site (mm) x

IMF Intramuscular fat concentration in LL (%) x

IRONWETa Iron concentration in LL (mg/kg) x

MYOGLOBINa Myoglobin concentration in LL (mg/kg) x

pH24LL pH of the LL 24 h postmortem x

SHEARF5 Shear force of the LL 5 days post slaughter (N) x

SPLINEPH18 Predicted pH of the LL when its temperature x

reached 18 °C

SPLINEPH6TEMP Predicted temperature of the LL when the x

pH was 6

ZINCWETa Zinc concentration in LL (mg/kg) x

a Concentrations based on wet muscle (as received) basis.

The factors and covariates used in the models were selected from those listed in Table 2. Although data for 37 long chain fatty acids (LCFA) were available, only 5 LCFA were included in the covariate list used for model selection due to practical difficulties associated with data processing. The LCFA included were those with a human health claim: a Linolenic acid (ALA, =C18_3n3), Eicosapentaenoic acid (EPA, =C20_5n3) and Docosahexaenoic acid (DHA, =C22_6n3), and omega 6 fatty acids associated with colour stability in previous studies (Faustmanetal.,2010): Linoleicacid (LA, =C18_2n6) andAra-chidonicacid (AA, = C20_4n6). SPLINEPH18 and SPLINEPH6TEMP were calculated using the method of van de Ven et al. (in press).

2.3.2. Notes on selection of model terms

The base model used to predict colour stability using the oxy/met values at each of 4 time points, involved fixed and random model terms. The fixed model terms were Intercept + time+ BRDX +time: BRDX where time is the time of SRD measurement (0, 1, 2 or 3 days) as a covariate. Curvature terms spl(time) and spl(time):BRDX were fitted as random along with the factors SITE, DROP and DATE. Note that interactions of 2 terms in the model are represented by a colon (:) as this is the convention for ASReml-R that differs in this regard from other statistical packages. The spline approach, as described in Verbyla, Cullis, Kenward, and Welham (1999) was used instead of a parametric form to allow a more 'flexible' construction of the response curves over time, because an exponential-type decay was not always suited to the data.

SITE, DROP and DATE were included as random rather than fixed or "treatment" effects because in the context of colour stability, the design was unbalanced and various factors were confounded. DATE effectively was the same as batch because batches of lambs were slaughtered on different dates. SITE and DROP were confounded with abattoir; with some abattoirs receiving lambs from more than one but no more than two SITEs, and some SITEs consigning lambs from different drops to different abattoirs. In relation to oxy/met an affect attributed to SITE could have been due to a number of factors from finishing diet through to processing conditions, none of which were balanced or necessarily held constant within the experimental design. Inclusion of the term "SITE/DROP/DATE" in the random term effectively removed the effects of these factors, all of which could have influenced colour stability. In ASReml-R notation 'SITE/DROP/ DATE' is equivalent to 'SITE +SITE: DROP +SITE:DROP:DATE'.

The purpose of the design that generated the data used in this study was primarily to determine genetic estimates presented elsewhere (Mortimer et al., in press). Testing for phenotypic affects, the subject of this paper was secondary. The covariates were added in a forward selection manner allowing for interactions with time and BRDX. In the first stage the random model terms "x+x:time +x:BRDX +x:spl(time)+ x: time:BRDX +x:spl(time):BRDX' were added where x was one of the 19 covariates. The covariate explaining most variation (pH24LL) was then added to the model and the process repeated with the remaining 18 covariates to identify the second most important covariate (LA). This was process was repeated for 9 x terms in total. The 19 covariates were treated as 'random' to obtain valid tests as explained in the ASReml manual section 2.5.1, when using the REML likelihood ratio test (REMLRT) with nested models (

2.3.3. Model description for prediction of oxy/met change with SRD time

In ASReml-R terminology (Butler, Cullis, Gilmour, & Gogel, 2009),

the structure of the base model (Fit0) can be described as follows:

Fit0 = asreml(oxy/met~time*BRDX,

random = ~SITE/DROP/ DATE +sire +sire:dam +id +spl(time)+ spl (time):BRDX), where dam is nested within sire and "id" is the animal identification variable that represents sire:dam:lamb.


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx

2 3 0 1

SRD time (in days)

Fig. 1. Predicted oxy/met values versus SRD time (in days) for MatM (MaternalxMerino), MM (MerinoxMerino), TM (TerminalxMerino) and TMat (TerminalxMaternal) breed types; predictions made using the base model. The solid lines indicate predicted mean values, the shaded area the 95% upper and lower prediction limits, and the dashed line the benchmark level for oxy/met of 3.5.

Similarly, we can add one covariate at a time to fit the extended base model (Fit1)

Fitl = asreml(oxy/met~time*BRDX,

random =~SITE / DROP / DATE +id +sire +sire:dam +spl(time) +spl (time):BRDX +x1 +x1:time +x1:BRDX +x1:spl(time) +x1:time:BRDX + x1:spl(time):BRDX).

Having found the best covariate x1 that maximised the log-likelihood, we then retain this (x1) in the next model called Fit2 and find the next best covariate x2

Fit2 = asremlioxy/met~time*BRDX,

random =~ SITE/DROP/DATE+id +sire +sire:dam +spl(time) +spl (time):BRDX +x1 +x1:time+x1:BRDX+x1:spl(time) +x1:time:BRDX + x1:spl(time):BRDX +x2 +x2:time +x2:BRDX +x2:spl(time) +x2:time: BRDX +x2:spl(time):BRDX).

Whilst the difference in the REML log-likelihood was still significant (6 degrees of freedom) after 9 'x' terms were added, only 2 'x' terms were included in the final model used for predictions, pH24LL and LA. After 2 'x' terms the difference in the percentage variance accounted for changed by less than 2% (Table 3) and this we considered was too low to warrant inclusion of further terms.

The inverse calibration estimate of SRD time taken for the oxy/met value to reach the benchmark value of 3.5 was done by using the method described in Verbyla et al. (1999). Prediction curves and 2 prediction intervals (± 1.96*standard error) were constructed and the inverse estimate of time corresponding to the Y (oxy/met) value = 3.5 was derived from the intersection of these 3 curves.

The predictions from the model Fit2 with the 2 best covariates x1 = pH24LL and x2 = LA with the base model were constructed and 3 representative points of the joint distribution of x1, x2 namely, x1 = (5.6, 5.7, 5.8) by x2= (100,125,150) combinations.

Table 3

The order that the first 9 covariates were selected for inclusion in the model.

Order of selection x Terms Log-likelihood % vara Residual animal variance (id term) Sigma2 Sire variance Sire: dam variance

0 BASE -2843 71.10 0.24 0.48 0.07 0.10

1 pH24LL -2405 73.60 0.11 0.44 0.03 0.05

2 LA -2010 76.10 0.11 0.39 0.03 0.06

3 DHA -1859 77.10 0.12 0.38 0.03 0.06

4 FAT5 -1648 78.30 0.12 0.36 0.03 0.06

5 AGEATOBS -1487 79.20 0.12 0.34 0.03 0.05

6 HGRFAT -1333 79.90 0.12 0.33 0.03 0.05

7 AA -1215 80.50 0.13 0.32 0.03 0.05

8 MYOGLOBIN -1124 81.00 0.12 0.31 0.03 0.05

9 SPLINEPH6TEMP -1058 81.30 0.12 0.31 0.03 0.05

a % var = percentage variance = 1 — Sigma2/variance of Y (1.65).


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx 5

3. Results

3.1. Comparison to benchmark values

The proportion of lambs within a batch (mean± SEM) exceeding the benchmark value of 3.5 was 99 ±0.89%, 75 ±3.46%, 41 ±3.91% and 23 ±3.37% for day 0, 1, 2 and 3 SRD times respectively.

3.2. Model predictions for oxy/met during the SRD

Oxy/met was significantly associated with time, breed type and all of the 19 covariates tested (P>0.05). The log likelihood values between successive models were not significant (P>0.05). However, the base model accounted for 71.1% of the variance and after 2 covar-iates were added, adding further terms increased the variance accounted for by less than 2% for each term added (Table 3). For the purpose of making predictions, we considered that only two covari-ates were necessary; pH24LL and LA, despite all 19 covariates being significant (P<0.05) according to the log likelihood comparisons.

A principal component analysis of the 19 standardised covariates at the animal level indicated no dominant pattern of the relationship between these covariates as shown in Table 4.

3.2.1. The relationship between oxy/met and time

Oxy/met decreased over time (Fig. 1), and this decrease was typically characterised by a monotonic exponential decay curve (always decreasing). However the nature of this relationship varied and in some individual samples there was little or no change in oxy/met between days 0 and 1 as shown in the example plot of the 2009 drop lambs from Rutherglen (Fig. 2). In some combinations of breed type, pH24LL and LA a sigmoid shaped curve was seen for predicted values. For cases with a sigmoid type curve (particularly TM breed types), the oxy/met benchmark of 3.5 was reached relatively late in the SRD indicating a more stable colour compared to those with a monotonic exponential decay curve (particularly the MM breed type).

3.2.2. Effects of breed type

There was a significant effect of breed type on oxy/met change in time. Using the base model, the predicted time for the oxy/met value to reach 3.5 for MM was lower (P<0.05) than all other breed types (Table 5), but there was no difference between MatM, TM and TMat breed types.

3.2.3. Effect of pH24LL and LA on the relationship between oxy/met and time

The predicted change in oxy/met with time was affected by pH24LL and LA and the magnitude of these effects depended on breed type (Table 6). The shortest predicted time for oxy/met to come down to 3.5 (least stable) was for Merino lambs when pH24LL was 5.8 and LA was 150 mg/100 g WM. The minimum, mean, maximum and standard deviation values were 5.28, 5.66, 6.67, and 0.14 for pH24LL and 61.60, 130.20, 325.00 and 40.57 for LA respectively. The correlation between pH24LL and LA was poor (Fig. 3).

Table 4

Percentage variance (% var) for the first 9 principal components.

Principal component 1 2 3 4 5 6 7 8 9

% var S % var 22.21 22.21 14.25 36.46 11.65 48.11 7.66 6.09 5.59 5.47 55.77 61.86 67.45 72.92 4.42 77.34 4.04 81.38

4. Discussion

4.1. External validity of the SRD

If the SRD used in this study is considered realistic, then a need clearly exists to improve the colour stability of Australian lamb meat; as the genotypes, production and processing systems used broadly represent the range expected in the Australian industry. Anecdotal evidence suggests that supermarkets aim to display overwrapped lamb meat for 2 days before applying discounts to prevent rejection by consumers due to browning. The results show that the SRD used in our study was conducive to browning within this commercial time frame; hence we suggest this system was in fact externally valid.

Whilst the distribution of genotypes and production systems would be different in the wider population, the percentage of samples reaching the benchmark value for consumer acceptability of 3.5 seemed high. After 1 and 2 days of SRD, 27% and 57% of samples respectively may have been perceived to be brown in colour by consumers and therefore unacceptable for purchase. If the lower value of 3.3 was used (Khliji et al., 2010) as a benchmark then the values for unacceptability would still be high; being 17% and 49% for days 1 and 2 respectively. Furthermore the LL is only intermediate in terms of colour stability and other muscles, particularly M. semimembranosus, could be expected to change colour from red to brown more quickly than for LL (Jacob & Thomson, 2012; Jose et al., 2008).

Further work to make a direct comparison between results from the SRD and meat in retail stores could be warranted. Retailers tend to make decisions on display time rather than colour or colour change so no comparable oxy/met data exists from Australian retail stores. Various authors have demonstrated that temperature and light intensity have significant effects on the rate of colour change (Faustman, 1990; Ledward, 1985; Rosenvold & Wiklund, 2011). Gutzke and Trout (2002) found that the autoxidation rate of myoglobin increased 5 fold for every 10 °C increase in temperature. The ambient temperatures during SRD were likely to have been lower and more consistent than those seen in retail display cabinets that are often open to ambient conditions and undergo regular defrost cycles. However the light conditions may have been more intense than for retail conditions as the meat was exposed to light constantly for the entire SRD period.

Another consideration is that the SRD used was designed to simulate current conditions for meat sold domestically in Australia. However meat display methods vary in commercial practice from the standardised 5 day ageing period and oxygen permeable overwrapping system we used. Modified atmosphere packaging (MAP) is being used and extended ageing periods of 30 days or more occur with meat exported. Moore and Young (1991) demonstrated the importance of ageing time and packaging system for colour stability. Some comparison may be warranted to determine if animals will be ranked similarly for colour stability when different ageing and packaging methods are used.

4.2. Describing the phenotypic trait

As reported previously (Jacob et al., 2011), the rate of change for oxy/ met depended on time during the SRD, so comparing rates would be difficult in large scale genetic comparisons. Predicting the time taken for oxy/met value to reach a benchmark value reduces this complexity and is in a form readily understood by retailers. The spline approach used in this study allows the relationship between oxy/met and time to vary from an exponential decay to a sigmoid like response as happens under standard conditions. This therefore represents an alternative method for describing colour stability to the one reported by (Mortimer et al., 2010), based on oxy/met measured at one time point. As mentioned previously, oxy/met values are not commensurate with pigment concentrations unless calibrated with chemical analyses (Hunt et al., 1991). Differences between samples based on a single time point measurement might therefore be influenced by factors such


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx

0.0 0.5 1.0 1.5 2.0 2.5 3.0



0.0 0.5 1.0 1.5 2.0 2.5 3.0

SRD time (in days)

Fig. 2. Example plot of actual oxy/met values versus SRD time (in days) for MatM (Maternal x Merino), MM (MerinoxMerino), TM (Terminal x Merino) and TMat (Terminalx Maternal) breed types for lambs from the 2009 drop at the Rutherglen site.

as texture and bloom depth as well as the formation of metmyoglobin. This alternative method might provide some relative adjustment for other factors given that measurements are repeated on the same meat sample, but requires more measurement points and a more complex statistical method for analyses compared to single point measurement. Interestingly Mortimer et al. (2010) found that the estimated heritabil-ity of oxy/met measured at one time point, increased with length of

the SRD, with heritability at time point day 0 being extremely low. Variation in oxy/met during the first day of SRD may therefore be less important in a genetic sense and a single measurement taken at the end of the SRD could be sufficiently accurate to rank animals for selection purposes. Further investigation is warranted to compare the heritability estimates of oxy/met value at one time point (day 3) with that for the predicted time for oxy/met value to reach the benchmark value of 3.5,


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx 7

Table 5

Predicted time (days) with lower and upper prediction limits (95%) taken for the oxy/ met value to reach 3.5 during the SRD for Maternal x Merino, Merino x Merino, Terminal x Merino and Terminal x Maternal breed types.

Breed type Predicted time (days) Corresponding inverse prediction limits (95%) Lower Upper

Maternal x Merino 2.00 1.70 2.45

Merino x Merino 1.30 1.00 1.65

Terminal x Merino 2.10 1.80 2.55

Terminal x Maternal 2.40 1.95 2.95

to determine whether the extra effort required to predict time to reach a benchmark value using the spline method, is needed for genetic evaluation purposes.

4.3. Phenotypic associations with oxy/met

Of the covariates tested, pH24LL and LA were the two most strongly associated with oxy/met and the change in oxy/met with SRD time, but the effect of these covariates depended on the lamb breed type. These associations potentially have implications for trait description as well as being important findings for managing colour stability in their own right. This study showed that these covariates have the potential to change the oxy/met by time response from an exponential decay to a sigmoid type response. Standardizing animal production and meat processing systems might therefore be considered, as done with measurement instruments and protocols, when comparing genotypes.

Breed type was aliased in the design with lamb age in that MM lambs tended to be older than lambs of other breed types. Being relatively older at the time of slaughter is an inherent characteristic of the Merino prime lamb system as Merino lambs grow slower and are later maturing (Hopkins, Stanley, Martin, & Gilmour, 2007). The greater

Table 6

Predicted time (days) for the oxy/met value to reach 3.5 for each breed type Maternal x Merino, Merino x Merino, Terminal x Merino and Terminal x Maternal and two levels of pH24LL (5.6,5.8) and LA concentration (100,150 mg/100 g WM); predictions made with the base model plus pH24LL and LA added. NA indicates no prediction made.

pH24LL LA Corresponding Predicted time Corresponding (mg/100gWM) lower inverse (days) upper inverse

prediction limit prediction limit

(days) (days)

Maternal x Merino

5.6 100 1.80

5.6 150 1.85

5.8 100 1.05

5.8 150 1.25


5.6 100 1.35

5.6 150 1.10

5.8 100 1.05

5.8 150 0.90


5.6 100 1.90

5.6 150 1.90

5.8 100 1.15

5.8 150 1.30


5.6 100 2.35

5.6 150 1.85

5.8 100 1.15

5.8 150 0.95

change in oxy/met value for MM lambs over the SRD period included higher starting and lower finishing values (Fig. 1) and a shorter period of time taken to reach the benchmark value of 3.5 for browning (Table 5). The higher starting value for oxy/met could have been indicative of higher myoglobin concentrations in MM lambs due to lamb age as myoglobin concentration increased with age (Pannier et al., 2010), or alternatively a more rapid blooming period.

Other studies have found that meat from Merino lambs had poor colour stability relative to other breeds (Warner, Ponnampalam, Kearney, Hopkins, &Jacob, 2007). Since lamb age was different between Merino and other breeds then reducing the age at slaughter by improving growth rate might improve the colour stability of meat from Merino lambs. If blooming time was breed dependent then factors expected to influence this such as mitochondrial density could be worthy of investigation (Tang et al., 2005). Studies have shown that Merino lambs tend to have a higher pHu than other breeds (Gardner, Kennedy, Milton, & Pethick, 1999; Hopkins, Fogarty, & Menzies, 1996; Young, Reid, & Scales, 1993) but pH24LL would have provided adjustment for pHu in the model.

pH24LL was ranked the highest of all the covariates so presumably was primarily more important than LA. The literature about the effect of pH on colour stability suggests a complex relationship dependent on the age of meat and species. Our finding was contrary to that of Ledward, Dickinson, Powell, and Shorthose (1986) that a high ultimate pH (>5.8) had less tendency to form metmyoglobin during retail display for beef. They attributed this effect to the rate of autoxidation of myoglobin decreasing with increasing pH and the enzymatic reducing system being more active at a high pH. Gutzke and Trout (2002) demonstrated that for the myoglobin au-toxidation reaction, the activation energy (Ea) increased and the reaction rate coefficient (k) decreased with increasing pH. However this effect of pH on k was temperature dependent and negligible when temperature was below 20 °C. So whilst the effect of low pH would be expected to increase the rate of autoxidation, the low temperature during the SRD may have limited this effect. Furthermore Bekhit, Geesink, Morton, and Bickerstaffe (2001) suggested that metmyoglobin reductase activity had little effect on colour stability of ovine longissimus muscle and this may represent a difference between lamb and beef.

Tang et al. (2005) found that mitochondrial oxygen consumption was inhibited in vitro when the pH was 5.6 compared to 7.2, and that mitochondrial respiration favoured conversion from oxymyoglobin to deoxymyoglobin or metmyoglobin. A low pH might therefore result in stable colour in the absence of autoxidation and metmyoglobin reduc-tase having strong effects. If an effect of pH on mitochondrial activity was the reason for low pH24LL being associated with stable colour in our study, then conceivably the age of the meat would be crucial for comparing phenotypes. Mitochondrial activity declines with time post slaughter and mitochondria become inactive after 60 days (Tang et al., 2005). Ledward et al. (1986) found that beef muscle changed colour more rapidly during a 3 day display period when aged for 33 days compared to 5 days ageing. Meat aged for 5 days in a domestic scenario might therefore be ranked differently to meat aged for greater than 30 days in an export scenario and further investigation is warranted in this regard.

Generally pHu is measured 48 h postmortem, but this is difficult when carcasses are deboned at 24 h as occurred with INF lambs. pH24LL may have been higher than the ultimate pH (pHu) for some samples and influenced by the rate of pH decline in the first 24 h. Jacob and Thomson (2012) found that rapid chilling reduced colour stability in lamb meat and speculated this might be due to an effect on blooming depth. Interestingly the estimate of rate of pH decline (SPLINEPH6TEMP) was ranked 9th in the order of selection for covari-ates in the model and accounted for only 0.3% of the variance (Table 3), suggesting that the effect of pH at 24 h was in fact a pHu effect.

2.30 2.20 1.40 1.55

2.85 2.75 1.95 1.95

1.65 1.35 1.45 1.15

2.10 1.75 1.95 1.50

2.35 2.25 1.55 1.65

2.9 2.8 2.2 2.15

2.85 2.30 1.50 1.20

NA 2.90 2.05 1.50


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx

6.0 PH24LL

Fig. 3. Scatter plot of pH24LL and LA (mg/100 g).

The associations of LA with oxy/met change during the SRD were consistent with myoglobin oxidation being linked to peroxidation of lipid postmortem (Granit et al., 2001; Ponnampalam, Trout, Sinclair, Egan, & Leury, 2001), notwithstanding the possibility that vitamin E concentration, that wasn't available, might alter this association (Ponnampalam, Butler, McDonagh, Jacobs, & Hopkins, 2012). Furthermore, in our study, LA was ranked highest in its effect on oxy/met of the 5 LCFA tested, consistent with the evidence from Faustman et al. (2010) that a specific link exists between LA and myoglobin oxidation via 4-hydroxy-2-nonenal (HNE) formed from LA. Interestingly DHA was ranked the next most important covariate in our model although it accounted for only 1.0% of the variance.

5. Conclusions

The SRD caused sufficient colour change for phenotypic evaluation purposes, but further comparison with retail conditions is warranted. Predicting the time for the oxy/met value to reach the benchmark value of 3.5 using a spline model is an alternative to comparing oxy/ met values at a set time for characterising colour stabilty, although potentially more complex and expensive to do. Colour was least stable when the lamb breed type was Merino, pH at 24 h postmortem was high, and LA concentration was high. There were no differences between other breed types. The associations between oxy/met, lamb breed type, pH at 24 h postmortem and LA may be important for making genetic comparisons as well as predicting colour stability under different market scenarios.


The authors gratefully acknowledge the financial support of the Cooperative Research Centre for Sheep Industry Innovation. The authors also wish to thank the many people who assisted with site supervision, sample collection and measurement as well as abattoir staff. For the New South Wales sites; Dr. D. Hopkins, Dr. G. Refshauge, Mr. D. Stanley, Ms. S. Langfield, Mrs. T. Bird-Gardiner, Ms. T. Lamb, Mrs. E. Toohey and Mr. M. Kerr. For the Victorian sites; Dr. E. Ponnampalam, Mr. W. Brown, Mr. A. Naththarampatha and Mr. M. Kerr. For the Western Australian sites included were; Dr. Kelly Pearce, Mr. K Hart, Mr. M. Boyce, and Mr. A. Williams.


Bekhit, A. E. D., Geesink, G. H., Morton, J. D., & Bickerstaffe, R (2001). Metmyoglobin reducing activity and colour stabilty of ovine longissimus muscle. Meat Science, 57, 427-435.

Butler, D. G., Cullis, B. R., Gilmour, A. R, & Gogel, B.J. (2009). ASReml estimates variance components under a general linear mixed model by residual maximum likelihood (REML). ASReml-R reference manual. Brisbane, Australia: The State of Queensland, Department of Primary Industries and Fisheries.

Faustman, C. (1990). The biochemical basis for discolouration in fresh meat: A review. Journal of Muscle Foods, 1, 217-243.

Faustman, C., Mancini, R A., Sun, Q., & Suman, S. P. (2010). Myoglobin and lipid oxidation: Mechanistic basis and control. In S. K. Lee (Ed.), International Congress of Meat Science and Technology. Jeju, Korea.

Forrest, J. C., Aberle, E. D., Hedrick H. B., Judge, M. B., & Merkel, R A. (1975). Principles of meat science. San Francisco: W. H. Freeman.

Gardner, G. E., Kennedy, L., Milton, J. T. B., & Pethick, D. W. (1999). Glycogen metabolism and ultimate pH of muscle in merino, first-cross, and second-cross wether lambs as affected by stress before slaughter. Australian Journal of Agricultural Research, 50,175-181.

Granit, R., Angel, S., Akiri, B., Holzer, Z., Aharoni, Y., Orlov, A., & Kanner, J. (2001). Effects of vitamin E supplementation on lipid peroxidation and color retention of salted calf muscle from a rich diet in polyunsaturated fatty acids. Journal of Agricultural and Food Chemistry, 49(12), 5951-5956.

Gutzke, D. A., & Trout, G. R. (2002). Temperature and pH dependence of the autoxida-tion rate of bovine, ovine, porcine and cervine oxymyoglobin isolated from three different muscles — Longissimus dorsi, Gluteus medias, and Biceps femoris. Journal of Agricultural and Food Chemistry, 50,2673-2678.

Hopkins, D. L., Fogarty, N. M., & Menzies, D.J. (1996). Muscle pH of lamb genotypes. 477 Proceedings of the Australian Society of Animal Production, 21. (pp. 347).

Hopkins, D. L., Kerr, M. J., Lamb, T. A., & Jacob, R H. (2008). Important considerations when measuring colour stabilty of lamb cuts. In R Venter (Ed.), International Congress ofMeat Science and Technology. Cape Town SA: The congress.

Hopkins, D. L., Stanley, D. F., Martin, L. C., & Gilmour, A. R (2007). Genotype and age effects on sheep meat production. 1. Production and growth. Australian Journal of Experimental Agriculture, 47,1119-1127.

Hunt, M. C., Acton, J. C., Benedict, R C., Calkins, C. R., Cornforth, D. P., Jeremiah, L. E., Olsen, D. G., Salm, C. P., Savell, J. W., & Shivas, S. D. (1991). Guidelines for meat color evaluation (pp. 1-17). Kansas State University, Manhattan, KS: American Meat Science Association.

Jacob, R. H., D'Antuono, M. F., Smith, G. M., Pethick D. W., & Warner, R D. (2007). The effect of age and electrical stimulation on the colour stability of fresh lamb meat. Australian Journal of Agricultural Research, 58, 374-382.

Jacob, R H., Mortimer, S. I., Hopkins, D. L., Warner, R D., & D'Antuono, M. F. (2011). Opportunities for genetic management of the retail colour stability of lamb meat. In P. Vercoe (Ed.), Association for the Advancement of Animal Breeding and Genetics, Vol. 19. (pp. 446-449)Perth, WA: Association for the Advancement of Animal Breeding and Genetics.

Jacob, R H., & Thomson, K. L. (2012). The importance of chill rate when characterising colour change of lamb meat during retail display. Meat Science, 90, 478-484.

Jose, C. G., Pethick, D. W., Gardner, G. E., & Jacob, R H. (2008). The colour stability of aged lamb benefits from vitamin E supplementation. In R. Venter (Ed.), International Congress of Meat Science and Technology. Capetown, South Africa: University of Stellenbosch.


R.H. Jacob et al. / Meat Science xxx (2013) xxx-xxx 9

Khliji, S., van de Ven, R., Lamb, T. A., Lanza, M., & Hopkins, D. L. (2010). Relationship between consumer ranking of lamb colour and objective measures of colour. Meat Science, 85, 224-229.

King, D. A., Shackleford, S. D., Kuehn, L. A., Kemp, C. M., Rodriguez, A. B., Thallman, R M., & Wheeler, T. L. (2010). Contribution of genetic influences to animal to animal variation in myoglobin content and beef lean color stability. Journal of Animal Science, 88,1160-1167.

Kropf, D. H. (1993). Colour stability — Factors affecting the colour stability of fresh meat. Meat Focus International, 2, 269-275.

Krzywicki, K. (1979). Assessment of relative content of myoglobin, oxymyoglobin, and metmyoglobin at the surface of beef. Meat Science, 3(1), 1-10.

Ledward, D. A. (1985). Post-slaughter influences on the formation of metmyoglobin in beef muscles. Meat Science, 15, 149-171.

Ledward, D. A., Dickinson, R. F., Powell, V. H., & Shorthose, W. R (1986). The colour and colour stability of beef Longissimus dorsi and semimembranosus muscles after effective electrical stimulation. Meat Science, 16, 245-265.

Moore, V. J., & Young, O. A. (1991). The effects of electrical stimulation, thawing, ageing and packaging on the colour and display life of lamb chops. Meat Science, 30,131-145.

Morrissey, E. R., Jacob, R. H., & Pluske, J. M. (2008). Perception of red brown colour by consumers. In R Venter (Ed.), International Congress of Meat Science and Technology. Cape Town SA: The congress.

Mortimer, S. I., van der Werf, J. H. J., Jacob, R H., Pannier, L., Pethick D. W., Pearce, K. L., Warner, R D., Geesink, G. H., Edwards, J. E. H., Ponnampalam, E. N., Ball, A. J., Gilmour, A. R., & Hopkins, D. L. (in press). Genetic parameters for meat quality traits of Australian lamb meat. Meat Science.

Mortimer, S. I., van der Werf, J. H. J., Jacob, R. H., Pethick, D. W., Pearce, K. L., Warner, R D., Geesink, G. H., Edwards, J. E. H., Gardner, G. E., Ponnampalam, E. N., Kitessa, S. M., Ball, A. J., & Hopkins, D. L. (2010). Preliminary estimates of genetic parameters for carcass and meat quality traits in Australian sheep. Animal Production Science, 50,1135-1144.

Pannier, L., Ponnampalam, E. N., Gardner, G. E., Hopkins, D. L., Ball, A. J., Jacob, R H., Pearce, K. L., & Pethick, D. W. (2010). Prime Australian lamb supplies key nutrients for human health. Animal Production Science, 50,1115-1122.

Pearce, K. L. (2009). Sheep CRC program 3: next generation meat quality. (1 ed.). Project 3.1.1 Phenotyping the Information Nucleus Flocks: Operational Protocol Series, Vol. 1. (pp. 107)Perth, Western Australia: Murdoch University.

Ponnampalam, E. N., Butler, K. L., Pearce, K. L., Mortimer, S. I., Pethick, D. W., Ball, A. J., & Hopkins, D. L. (in press). Sources of variation of health claimable long chain omega-3 fatty acids in meat from Australian lamb slaughtered at similar weights. Meat Science.

Ponnampalam, E. N., Butler, K L., McDonagh, M. B., Jacobs, J. L., & Hopkins, D. L. (2012). Relationship between muscle antioxidant status, forms of iron, polyunsaturated fatty acids and functionality (retail colour) of meat in lambs. Meat Science, 90,297-303.

Ponnampalam, E. N., Trout, G. R., Sinclair, A. J., Egan, A. R., & Leury, B. J. (2001). Comparison of the color stability and lipid oxidative stability of fresh and vacuum packaged lamb muscle containing elevated omega-3 and omega-6 fatty acid levels from dietary manipulation. Meat Science, 58,151-161.

Rosenvold, K., & Wiklund, E. (2011). Retail colour display life of chilled lamb as affected by processing conditions and storage temperature. Meat Science, 88, 354-360.

Tang, J., Faustman, C., Hoagland, T. A., Mancini, R. A., Seyfert, M., & Hunt, M. C. (2005). Post mortem oxygen consumption by mitochondria and its effects on myoglobin form and stability. Journal of Agricultural and Food Chemistry, 53(4), 1223-1230.

Tapp, W. N., Yancey, J. W. S., & Apple, J. K. (2011). How is the instrumental color of meat measured. Meat Science, 89(1), 1-5.

van de Ven, R J., Pearce, K. L., & Hopkins, D. L. (in press). Post-mortem modelling of pH and temperature in related lamb carcases. Meat Science.

van der Werf, J. H.J., Kinghorn, B. P., & Banks, R G. (2010). Design and role of an information nucleus in sheep breeding programs. Animal Production Science, 50(11-12), 998-1003.

Verbyla, A. P., Cullis, B. R., Kenward, M. G., & Welham, S. J. (1999). The analysis of designed experiments and longitudinal data by using smoothing splines. Journal of the Royal Statistical Society: Series C (Applied Statistics), 48(Part 3), 269-311.

Warner, R D., Jacob, R H., Edwards, J. E. H., McDonagh, M., Pearce, K., Geesink G., Kearney, G., Allingham, P., Hopkins, D. L., & Pethick D. W. (2010). Quality of lamb meat from the Information Nucleus Flock Animal Production Science, 50,1-12.

Warner, R D., Ponnampalam, E. N., Kearney, G. A., Hopkins, D. L., & Jacob, R. H. (2007). Genotype and age at slaughter influence the retail shelf-life of the loin and knuckle from sheep carcasses. Australian Journal of Experimental Agriculture, 47,1190-1200.

Young, O. A., Reid, D. H., & Scales, G. H. (1993). Effect of breed and ultimate pH on the odour and flavour of sheep meat. New Zealand Journal of Agricultural Research, 36, 363-370.