Intrabreed variability and relationships for 41 carcass and meat traits in Pirenaica cattle

A total of 125 13-month-old, entire males of the Pirenaica breed were selected from among the progeny of nine different sires. They were used to study the intrabreed variability of 41 carcass and meat traits, including weights, lengths, diameters and perimeters, grading parameters, dissection of the sixth rib, pH, colour and meat texture variables, and sensory attributes. Pearson correlations were also calculated. Coefficients of variation ranged from 3.5% for dressing percentage to 47.0% for fat red index, and all of the values could be considered within the normal range for cattle. Carcass traits showed lower variability (11.7% on average) than meat variables (14.8% on average). Hence, to achieve a standardised product, it would be desirable to include in the breeding selection programme a variable that could be measured on the carcass and that can predict meat quality. Nevertheless, from the results of the present study, none of the variables studied fulfil this requirement, and further studies would be necessary to widen our knowledge on this subject. Additional key words: beef, linear measurements, meat instrumental quality, sensory quality.


Introduction
Beef production is of great importance throughout Europe.European Agricultural Policy has been developed to encourage the use of local breeds to develop sustainable animal production (Piedrafita et al., 2003).However, the persistence of a population is partially linked to its ability to meet current market requirements, and persistence depends on carcass and meat quality.Consequently, within the scope of the European Union (EU), identifying and describing breeddependent traits related to carcass and meat quality have gained in interest over recent years because satisfying consumer requirements is essential for producers (Renand et al., 2001).Traditionally, producers have selected sires for their productive and morphological aspects, including live or carcass weight, because these aspects are considered more important than those related to meat quality.
An abundance of literature has been generated on the study of breed effect on carcass and meat characteristics (Chambaz et al., 2003;Keane, 2003;Piedrafita et al., 2003;Özlütürk et al., 2004;Albertí et al., 2005), but there is little information regarding the intrabreed variability of carcass and meat traits, especially in beef cattle.Furthermore, the practical implications of having information about quality parameters related to bull sires has not been extensively studied in the international literature addressing carcass or meat quality (Maher et al., 2004;Altarriba et al., 2005).Intrabreed variability would be of major interest to breeders; to be competitive through an identifiable brand, they have to be able to offer the market a homogeneous product.
The aim of this study was to quantify intrabreed variation in carcass and meat quality characteristics of Pirenaica cattle and also to highlight the differences between carcasses grouped by the corresponding sire's selection index.The selection index was the weight of progeny at 210 days, and it is the current criterion used by the Breeders' Association to determine whether or not to maintain the sires in the selection programme.

Animals
Pirenaica is a beef-specialised breed located on the southern slopes of the Pyrénées in NE Spain.More information on this breed's productive traits and ethnologic characteristics can be found in Sánchez-Belda (2002) and Piedrafita et al. (2003).Beef production with the Pirenaica breed is based on maximising the use of grazing lands.Calving generally occurs in the first third of the winter season, and calves suck milk directly from the mother until they are put out for spring grazing.In spring, they accompany their mothers to alternate a milk diet with the consumption of grass.On returning from the summer grazing pastures, calves are weaned (about 5 to 7 months of age) and reared indoors with an ad libitum concentrate diet, reaching the market with a 450-kg live weight at around 12 months of age (Sánchez-Belda, 2002).
A total of 125 young males were used (Table 1).All animals were descendants of nine sires, which were those being used for artificial insemination in the Breeders'Association reproduction programme at the time of the study.Once weaned, young bulls were fed at the National Centre of Animal Selection and Reproduction (Zaragoza, Spain) on a rearing concentrate until they were 11 months old, followed by a fattening period of two months with concentrate until slaughter.Concentrate and cereal straw were provided ad libitum.During the rearing period, the diet consisted of 90.3% dry matter, 17.6% crude protein, 5.4% fibre, and 5.3% crude fat, with 140.4 g kg -1 of digestible protein.For the fattening period, the diet characteristics were 89.9% dry matter, 14.2% crude protein, 4.9% fibre, 5.0% crude fat, and 114.2 g kg -1 of digestible protein.
Slaughtering was established at 13 months of age, and the day prior to slaughter, the animals were weighed (live slaughter weight).EU welfare regulations were followed when handling the animals.Animals were slaughtered at the nearest EU-licensed abattoir, at a distance of 30 km, to minimise the transport stress effect.Stunning was performed by captive bolt pistol.Just after slaughter, the left forelimb autopod was removed and weighed and its length and perimeter measured.Carcass dressing was undertaken according to standard commercial practice.Carcasses were chilled at 4 ±1ºC for 24 h.

Carcass quality
The following variables were recorded: -Hot carcass weight, measured without removing subcutaneous fat and maintaining the testicles and kidney, channel, and pelvic fat.The tail remained on the right half carcass.

Subcutaneous fat colour and meat quality
The colour of the subcutaneous fat was measured 24 h after slaughter on the carcass using a Minolta CM-200 spectrophotometer with a D65 illuminant and a 10º standard observer in the CIE L*a*b* space (Commission Internationale de l'Eclairage, 1976).Afterwards, the pH of the L. thoracis muscle at the lumbar region level was measured with a CRISON pH meter equipped with a penetration electrode.Subsequently, the L. thoracis muscle was removed from the left side of the carcass (between the seventh and ninth hemi-vertebrae).
Meat quality parameters were determined according to the guidelines of Honikel (1998) and Cañeque and Sañudo (2000).
The day after slaughter, meat colour was measured in the L. thoracis (sixth rib level) muscle, after 2 h of air exposure, with a Minolta CM-2002 spectrophotometer with a D65 illuminant and a 10º standard observer in the CIE L*a*b* space.This sample was then placed on a polystyrene tray wrapped with oxygen-permeable plastic film and kept at 2-4ºC until the seventh day postmortem.The pH at 7 days post mortem, haem pigment concentration (Hornsey, 1956), and water-holding capacity were measured using a compression method (Grau and Hamm, 1953).
The rest of the L. thoracis muscle was sliced into 3.5cm-thick steaks for instrumental analysis or 2-cm-thick steaks for sensory analysis.Steaks were vacuum packaged and aged for 7 d at 2-4ºC.All samples were frozen and stored at -18ºC until further analysis.
For texture analysis, steaks were thawed in tap water for 4 h until they reached an internal temperature of 16-19ºC.Samples with a 1-cm 2 cross-section were cut with muscle fibres parallel to their longitudinal axis.The texture of raw meat was analysed with an Instron 4301 using a modified compression device that avoids transversal elongation of the sample (Lepetit and Culioli, 1994).Maximum load (N) and stress at 20% and 80% of maximum compression (N cm -2 ) were recorded.
Sensory analyses were carried out by an 11-member panel, trained in accordance with ISO 8586-1 (1993), with additional methods specifically for meat.Steaks were thawed inside their vacuum bags with tap water at 16-19ºC and then wrapped in aluminium foil and cooked to an internal temperature of 70ºC on a double plate grill preheated to 200ºC.The internal temperature of the sample was monitored with a data logger using a thermocouple probe, inserted horizontally to the midpoint of the steak.Thawed and cooked steaks were weighed, and cooking losses were calculated.The core portion of the steaks was cut into 11 pieces.Each sub-sample was immediately wrapped in aluminium foil, codified, and kept at 60ºC.Panellists evaluated samples in individual booths under red lights.Three steaks from each animal were assessed.Each sensory evaluation session consisted of 12 randomly selected loin samples.The 12 subsamples were tasted by each panellist in a different order in each session.Panellists were asked to rate beef odour intensity, tenderness, juiciness, and beef flavour intensity, using a 1-100 unstructured line scale, in which 1 represented the lowest and 100 the highest value for the attribute under consideration.

Statistical analysis
Statistical analysis was performed using the SPSS statistical package (13.0).Means, coefficients of variation, minimum and maximum values, and Pearson's correlation coefficients were calculated for all variables.
A k-means cluster analysis was performed to study the cluster of sires based on their descendants' weight at 210 d of age.An ANOVA with the cluster grouping as the fixed effect was carried out to establish if there were any relationships between this common criterion and the meat quality of the descendants.Finally, an ANOVA procedure was performed with the group as the fixed effect, co-varying by slaughter weight.

Results
Means, coefficients of variation (CV), and minimum and maximum values for the variables studied are shown in Table 1 for carcass traits and dissection data and in Table 2 for meat quality characteristics.The average CV for the pooled data was 14.0%, but the spread of the CV depended on the variable considered.The CV for carcass weight was 11.2%, whereas dressing percentage had a low CV (3.5%).The pH showed the lowest CV, both at 24 h and at 7 d (1.8 and 1.4%, respectively).Regarding meat colour, the L* of the muscle was less variable than a* or b*, and muscle b* had a CV that was nearly twice that of a*.Considering water losses, it can be seen that the CV was higher for cooking losses than are shown in Table 4.There appear to be statistical differences among the groups for the majority of the variables studied.Thus, animals with a positive selection index had greater carcass traits (conformation score, muscle percentage, morphological measurements, and loin surface) than those from the other two groups.
Correlation coefficients between carcass traits are shown in Table 5, and Table 6 shows the correlation between meat quality characteristics.In general, several significant correlations were found among carcass traits, but they were scarce among meat quality characteristics.The highest correlation coefficient was established between maximum load and compression stress at 80% of compression.
Table 7 shows correlations between carcass and meat characteristics.There were a number of significant correlations, but in general coefficients were low.Animals with heavier and better-conformed carcasses tended to have a lower pH, both at 24 h and at 7 d.On the other hand, hot carcass weight was significant and negatively correlated with L* and positively with the a* of the muscle, although this fact is not so clear in a comparison of the bull sire groups according to the weight of their progeny at 210 days.

Discussion
The current results for all the parameters studied were normal within the Pirenaica breed, and they are in for water holding capacity (WHC).A higher CV was found for stress at 20% (43.3%) than for stress at 80% (27.8%) and maximum load (23.8%).
Table 3 shows the selection index of each sire and the centroid of each cluster.After clustering, bulls remained in three groups (Table 3).Group 1 included sires with negative indexes, lower than average; group 2 included sires with average indexes; and group 3 included sires with positive indexes.Average values for carcass, subcutaneous fat, and meat quality traits for every group agreement with the data from several studies of the same breed (Albertí et al., 1995;Campo et al., 1998;Sañudo et al., 2001) or of other European beef breeds (Crouse et al., 1985, in Angus and Simmental;Jurie et al., 1995, in Limusin;Destefanis et al., 1996, in  Table 4. Means, standard error (SED), and significance (p value) for all carcass and meat variables studied, with the index selection group as the fixed effect.Adjusted p value was the significance when data were co-varied by slaughter weight several British breeds; Piedrafita et al., 2003).The similar CV for live and carcass weights and the low CV for dressing percentage suggest homogeneity in live breed performances, considering that the animals were slaughtered at a similar age.
In general, morphological measurements showed a low CV, and the results obtained agree with those reported by other authors (Piedrafita et al., 2003;Özlütürk et al., 2004;Albertí et al., 2005).In the dissection variables, it can be seen that the CV for bone and fat percentages was around three and five times greater than the muscle percentage, respectively.The same pattern has previously been described by other authors (Subrt and Divis, 2002;Piedrafita et al., 2003;Farmer et al., 2004).
The high variability found for conformation and fatness scores was not surprising; Albertí et al. (2005) reported a CV of around 30% for both traits, and Piedrafita et al. (2003) reported a CV of 16% for conformation and a CV of 20.5% for fatness scores, even when intrabreed variability in carcass weight was low (5%).Nevertheless, the high variability in grading parameters 2003, in Angus, Simmental, Charolais and Limousin;and Albertí et al., 2007, in 15 European breeds).Compared with other Spanish cattle breeds, the Pirenaica is a medium-sized breed, with a well-shaped, lean carcass, light subcutaneous fat, light pink meat, and low values for texture variables.It is also characterised by having a tender, mild-flavoured meat (Albertí et al., 1999;Campo et al., 2000;Gil et al., 2001;Panea, 2002;Albertí et al., 2003;Piedrafita et al., 2003).

Variability of carcass quality traits, morphological measurements, and dissection data
The identified CV for carcass weight was very similar to data published by various authors with regard to several European beef breeds (11.2% for Albertí et al., 2005;6.8% for Biaggini and Lazzaroni, 2005;and 10% for Jurie et al., 1995).In the same way, the present result for CV in dressing percentage was in accordance with those of most of authors (Barton and Pleasants, 1997, in (2006) and Albertí et al. (2007).
According to Albertí et al. (2005), these inter-animal differences in shape and morphological measurements are mainly due to variability in transversal measurements of the carcass more than to the variability of longitudinal measures.Thus, it could be assumed that transversal measurements are those that actually define conformation scores.Additionally, the most variable morphological measurement was specifically hind limb width, which could also help to explain the high variability found for the conformation score.Diameters of the L. thoracis muscle are scarce in the literature, but the area of muscle is often measured.A CV of around 14% for the rib area was described by Jones and Tatum (1994) in commercial cattle; a CV of 5% to 8% was shown by Özlütürk et al. (2004); and values of around 17% were calculated from data presented by Piedrafita et al. (2003).In some European markets such as Spain, carcass price depends on conformation; well-rounded, well-conformed carcasses are the most sought by wholesalers (Bello and Calvo, 2000), who use roundness of limb and shoulder and thickness of loin to estimate the proportion of prime cuts of saleable meat.Carcass fat colour is an important attribute for the beef market.It can be observed that the L* of fat was much less variable than the a* or b* parameters.In lean carcasses, such as in the present study, it is difficult to find areas with enough fat thickness to reliably measure the fat colour.

Physical, chemical, and sensory variables
The very low CV for pH represents good pre-slaughter management and underlines the fact that the Pirenaica breed is rarely affected by stress and pH problems (Albertí et al., 1991(Albertí et al., , 1995)).
The same ratio calculated in the present study among the L*, a*, and b* parameters can be calculated from data presented by other authors (Albertí et al., 1999;Gil et al., 2001;Serra et al., 2004).The b* values depend on diet (Bidner et al., 1986), pH (Albertí et al., 1999), and the chemical state of myoglobin (Lindahl et al., 2001).Furthermore, the CV for myoglobin content was higher  than for the a* parameter, which is in accordance with Albertí et al. (1999) but contrary to Serra et al. (2004).Because the samples did not have problems with pH and all animals employed in the experiment were fed on the same diet, there appear to have been greater differences between samples in relation to sensitivity to the oxidation process during the blooming period, in agreement with Maher et al. (2004), who reported that there are individual differences in the blooming process.
The current results for water losses are in accordance with those of most authors (Albertí et al., 1999;Serra et al., 2004).Comparison of water losses is not easy because of differences in methodology between laboratories.Thus, different authors have employed different methods to measure WHC (Albertí et al., 1991;Irie et al., 1996;Failla et al., 2004).WHC is a result of events both prior to slaughter and following the post mortem changes, and it depends on several factors, including the nature of the force applied, which causes the displacement of the water (Rao et al., 1989;Palka and Daun, 1999), the extent of post mortem myofibrillar shrinkage, and changes in the extracellular water compartments (Offer and Knight, 1988).Furthermore, different cooking methods are described in the literature: water bath (Destefanis et al., 1996;Failla et al., 2001;Lopes et al., 2003), grill (Chambaz et al., 2003;Lopes et al., 2003;Failla et al., 2004), pan (Jeremiah andGibson, 2003), and oven (Crouse et al., 1985;Jeremiah and Gibson, 2003;Lopes et al., 2003;Serra et al., 2004), as well as different cooking times.It is clear that cooking losses are influenced by the same factors as WHC, plus cooking method and time, which could explain the higher variability for cooking losses than for WHC.
The current CV results for texture variables were higher than those found in the literature (Campo et al., 1999;Monson et al., 2004;Failla et al., 2004).Stress at 20% is related to the myofibrillar component (Lepetit and Culioli, 1994), and it is concerned with pre-and mainly post-slaughter handling, such as cooling or storage conditions.Improvement at this stage would thus be desirable to reduce variability.
Finally, the variability reported in the present study for sensory analysis variables is in accordance with results reported by Campo et al. (1999) or Sañudo et al. (2003) in their studies of Spanish cattle.Furthermore,   Campo et al. (1999) showed that the evolution of sensory characteristics depends on the genotype of the animals and that each breed required a specific ageing period to reach its optimum state.

General effects of selection index
Animals with a positive selection index had greater carcass traits than those from the other two groups, in accordance with the conclusions offered by Altarriba et al. (2005).Weight at 210 d could therefore be an indicator of carcass quality.Also, meat quality parameters were significantly different, but the tendencies were not so clear because on most occasions, there was no increase or decrease of the variables in line with the selection index.Thus, sires with the best selection index were found to have the lowest rates at 20% of compression, but in other meat quality parameters, they showed average values.Furthermore, descendants of sires with an average selection index (group 2) showed lower myoglobin content, cooking losses, WHC, maximum load, and compression test at 80% values (which would desirable for consumers), but they also presented higher compression at 20% values and the lowest overall appraisal scores.Finally, the animals in group 1, which were those with the lowest indexes and poor carcass quality, nevertheless had desirable values in terms of overall appraisal.From these results, it has to be concluded that there is as yet no variable that can be measured online on carcasses to predict meat quality.Consequently, more studies are necessary to further our knowledge on this subject.
In general, there is not much literature relating selection index and meat quality in beef, although a better conformation score is not related to a superior meat quality in Charolais (Maher et al., 2004).In pigs, however, most authors have reported that an improvement in carcass traits or lean meat produced a decrease in meat quality (Cameron et al., 1990;Knapp et al., 1997).
When p was adjusted by co-variate by slaughter weight, the significance of the effect remained mainly stable, except for 10 variables.For autopodus weight, the level of significance increased from p=0.159 to p=0.014, whereas in autopodus length, carcass width, hind limb length, hind limb width, hind limb perimeter, hind limb depth, pH at 7 days, WHC, and stress at 20%, the significance of the effect disappeared.Therefore, the effect of the selection index could be considered a consistent result.

Relationships between variables
Among carcass trait relationships, an especially important relationship was that between loin area and minimum diameter (r=0.83),demonstrating that variability in loin area is due to variability in dorso-ventral diameter rather than to any variability in medium-lateral diameter.It is also important to point out that muscle percentage was negatively related to both fatness and bone percentages, but that the relationship was greater with the former than with the latter.Therefore, it can be inferred that the heavier the carcass, the better conformation, more muscle, less fat, and less bone it has, as would be expected (Kempster et al., 1982;Barton and Pleasants, 1997).
Considering relationships between meat quality characteristics, the identified relationship between maximum load and compression stress at 80% of compression has already been defined in previous works, showing its similar biological significance (Campo et al., 2000;Panea, 2002).Myoglobin content was positively related to a* and negatively related to L*, indicating that a high myoglobin concentration was associated with redder, darker meat (Gil et al., 2001;Insausti et al., 2001;Serra et al., 2004).
With regard to sensory traits, there was a positive correlation between juiciness and tenderness, which is frequent in the sensory analysis.The more tender the meat, the more quickly the juices are released by chewing and the juicier the meat appears (Cross, 1988).Additionally, a negative correlation was found between cooking losses and tenderness, in accordance with the findings of other authors (Silva et al., 1999;Destefanis et al., 2000;Serra et al., 2004).
From the results of the present study, it can be concluded that there is great intrabreed variability for most of the traits studied and that this variability is greater in meat characteristics than in carcass traits.Thus, to achieve a standardised product, some meat quality characteristics should be included in selection programmes.Nevertheless, as indicated above, further studies seem necessary to identify a carcass measurement that can predict meat quality.
CENSYRA (Centre of Artificial Insemination and Animal Reproduction) of the Government of Aragón for their collaboration.This project has been financially supported by an agreement between University of Zaragoza, the National Confederation of Associations of Pyrenean Cattle, and the Government of Aragón.
pH 24 pH7 Fat L* Fat a* Fat b* Muscle L* Muscle a* Muscle b* Mb WHC Cooking Max C20

Table 1 .
Means, coefficient of variation (CV), and minimum and maximum values for carcass traits of Pirenaica young bulls (n=125)

Table 2 .
Means, coefficient of variation (CV), and minimum and maximum values for meat quality characteristics of muscle Longissimus thoracis (n=125)

Table 3 .
Cluster analysis to group nine different sires by their selection index (descendant weight at 210 days)

Table 5 .
Correlation coefficients (r) between carcass traits.Only significant correlations (at least p≤0.05) are shown may indicate that an additional tool should be taken into account to correctly classify carcasses, chiefly when very similar carcasses are compared.The use of a blockiness index for this purpose has already been suggested by Díez et al.

Table 6 .
Correlation coefficients (r) between meat quality characteristics.Only significant correlations (at least p<0.05) are shown

Table 7 .
Correlation coefficients (r) between carcass traits and meat quality characteristics