Marbling and sensory traits in beef: what controls juiciness and flavour

Why the marbling-tenderness link needed a closer look#

A long-standing assumption in both meat production and food service is that higher marbling makes beef more tender. The paper opens by testing that assumption against the published record, and finds the evidence underwhelming. Studies reviewed in the paper show the relationship between marbling and tenderness as low and variable, with some reporting a small positive association and others finding no significant trend at all.

Dikeman, cited in the paper, reviewed a number of earlier studies and put the share of variation in taste-panel tenderness scores attributable to marbling at only around 10 to 15%. That leaves most of the variation in tenderness explained by something other than fat within the muscle. Even so, the paper notes that marbling still functions as an assurance of tenderness and is used in that role in grading schemes, even where the direct relationship is weak.

The harder problem, according to the paper, is that sensory traits are tightly bound together. Correlations between tenderness, juiciness and flavour scores in the dataset examined ranged between 0.83 and 0.98, so any simple comparison between flavour and marbling partly reflects differences in tenderness as well. The paper's approach was to strip tenderness out of the picture statistically, using shear force as a covariate, so the fat-flavour and fat-juiciness relationships could be seen more clearly on their own terms.

Possible mechanisms behind marbling and sensory traits#

The paper sets out several mechanisms, discussed by Millar, through which marbling could plausibly affect eating quality. These mechanisms are also relevant to how fat develops within muscle in the first place, a question covered in our summary of the physiology of marbling in cattle. One is a change in bulk density: fat is lower density than lean, so a higher proportion of intramuscular fat produces a steak that needs less resistance to bite through. Another is a lubrication effect, where higher fat levels stimulate salivation and create a perception of juiciness while chewing, and sustain that sensation through the meal.

Flavour is linked to fat through a separate pathway. The paper notes that species-specific flavours are carried in fat, drawing on work by Hornstein and Wasserman, which helps explain why chemical fat percentage showed a measurable relationship with flavour scores once tenderness was accounted for. A third proposed mechanism concerns connective tissue: marbling fat is deposited in perivascular cells within the perimysium, and the theory holds that as marbling increases, connective tissue toughness decreases.

The paper is careful to say that working out how much each mechanism contributes on its own would be difficult in practice, given how the various structures of meat interact. It is likely, the paper suggests, that several of these mechanisms operate together to varying degrees, rather than one dominating.

Testing the overcooking insurance theory#

A separate idea popular in food service, attributed to Smith and Carpenter, is that marbling insures against the damage done by overcooking. Because fat conducts heat more slowly than lean, the theory goes, well-marbled meat should better protect muscle fibres from the shrinkage and denaturation that occurs at high cooking temperatures, making it both more tender and juicier when cooked well done.

This was tested directly in a CRC study by Rymill and colleagues, who cooked steaks spanning a spread of marbling scores to both rare and well done finishing points before running them through taste panel assessment. The result did not support the insurance theory: no interaction was found between marbling level and how well the steak was cooked, whether judged on tenderness or juiciness.

Instead, the authors of that study concluded degree of doneness mattered considerably more than intramuscular fat percentage in producing tender, juicy steaks. For anyone relying on marbling to compensate for a poorly managed grill, the paper's reading of this evidence is a caution rather than a reassurance.

How the CRC and MSA data were collected#

The sensory data came from the CRC straightbred and crossbred breeding programs, with design described by Upton and colleagues. Pedigreed weaner calves born from planned matings arrived at backgrounding properties in two annual intakes, then went onto either pasture or grain finishing before slaughter at one of three target carcass weight categories matched to domestic, Korean and Japanese market specifications.

For each animal, the front portion of the striploin was split into two blocks. One was frozen a day after slaughter and later analysed in the laboratory for toughness and fat content; the other was vacuum packed and sent to Meat Standards Australia, aged for a fortnight, then cut into steaks for consumer tasting. In total 3,613 samples carried a full set of live animal, carcass, laboratory and sensory records.

Sensory testing followed the MSA protocol described by Polkinghorne and colleagues. Consumers drawn from varied social and economic circumstances tasted samples only once; across a single sitting lasting a little over half an hour, each consumer worked through seven grilled half-steaks, rating tenderness, juiciness, flavour and overall liking on 100 mm scales running from very tough to very tender, very dry to very juicy, and extremely dislike to extremely like. Every reported value was an average built from ten separate consumer ratings.

One design quirk flagged by the paper: some kills where carcasses were not properly stimulated, producing cold-shortened, low-scoring, high-shear-force samples, were normally excluded from meat quality analyses but were kept in this one because they widened the range in tenderness and made the shear force adjustment more effective. The paper acknowledges this was not a randomly distributed inclusion and could bias some fixed-effect estimates.

What the adjusted analysis showed#

Mixed models examined juiciness and flavour scores against chemical fat percentage after adjusting for shear force, with terms built in for breed, market category, finishing system and sex of the animal, alongside random effects for cohort and kill group. The data themselves were varied: sensory scores ranged across roughly 80 units on a 100 point scale, shear force spanned a nine-fold range up to 18 kg, and chemical fat percentage averaged 3.3% but reached as high as 15% in some samples.

Once shear force was accounted for, chemical fat percentage still showed strong effects on both juiciness and flavour scores, and the relationship was curvilinear rather than a straight line: scores rose with fat but at a decreasing rate. Points of inflexion for the curves sat at 19% chemical fat for juiciness and 14% for flavour, at or beyond the upper range of the data, and the practical plateau for both traits was identified at 15 to 20% chemical fat, adjusted to a shear force of 5.0 kg.

The paper states that it could not find prior published work setting out this particular curved shape between juiciness, flavour and chemical fat once a tenderness measure had been allowed for, which is one reason the finding is presented as a distinct contribution rather than a confirmation of prior work.

Breed, finish and market differences in marbling and sensory traits#

Even once shear force and chemical fat percentage were held constant across comparisons, breed, finish and market still produced significant differences in flavour and juiciness. Tropically adapted breeds and crosses scored 5 to 10 units lower on both traits than Bos taurus breeds and crosses at matching fat and toughness levels, and this breed gap held regardless of animal age.

Finish and location told a different story. Cattle grown out on pasture in the north scored up to five sensory units lower for juiciness than stock finished on southern pasture or in feedlots, with a broadly similar pattern for flavour. Carcasses destined for export weights, Korean or Japanese specifications, scored around three units lower in flavour and juiciness than those slaughtered at domestic weights. Once animal age was added to the models, these finish and market effects dropped out of significance, indicating the northern pasture cattle were less juicy and flavoursome largely because they were older, not because of where or how they were finished.

Feedlot finishing's reputed flavour advantage was examined directly, and the same question of how nutrition and management shape marbling is explored at greater length in our summary on marbling in feedlot cattle and nutrition management. In the southern comparison, where pasture and feedlot cattle were slaughtered within 50 days of age of each other, feedlot carcasses scored three flavour units higher before adjusting for fat, though the juiciness advantage of two units did not reach significance. Once the two groups were set side by side at matching toughness and fat levels, the flavour difference disappeared, pointing to higher chemical fat in feedlot cattle, rather than grain feeding itself, as the real driver. This lines up with a review by Muir and colleagues finding little flavour difference between pasture and grain feeding when fatness or carcass weight was matched.

Age, flavour and the limits of this dataset#

Flavour showed a persistent negative relationship with age even after holding both toughness and fat percentage constant, meaning older animals scored less desirable flavour at matching tenderness and fat level. When chemical fat was removed from the flavour model altogether, the age coefficient stayed negative, though it dropped to half its previous size, indicating fat's positive flavour contribution does not fully offset the ageing penalty. Juiciness, by contrast, was not significantly affected by age in this analysis, consistent with other studies cited in the paper.

The paper itself flags a methodological limitation worth weighing carefully: flavour and juiciness were measured after a fortnight of ageing, but adjusted against a toughness measurement taken only a day after slaughter, so the resulting relationship between sensory scores and chemical fat is confounded with whatever sensory changes occurred during that ageing period. Combined with the non-random inclusion of cold-shortened kills, this means the plateau figures should be read as a solid pointer from a large, well-structured dataset rather than a fully isolated cause-and-effect result. We think that caveat matters in practice: a producer or processor using marbling as a flavour lever should treat 15 to 20% chemical fat as a useful guide, not a precise threshold.

Practically, the paper translates its headline figure into grading terms: a 15% level of chemical fat is equivalent to an AUSMeat marble score 4. For an industry moving to control tenderness more systematically, the paper's conclusion is that marbling's role may shift from insuring against toughness to genuinely driving flavour and juiciness, once myofibrillar tenderness is otherwise managed. This workshop paper sits within the broader Marbling Symposium collection, alongside other contributions gathered in the Livestock Library's conference proceedings collection.

Sources and further reading#

• Trove library search: find a library that holds the paper
• Meat & Livestock Australia: red meat industry research and marketing body
• Australian Meat Processor Corporation: research for red meat processors

Questions#

— The Livestock Library team