Development of a beef flavor lexicon and its application to compare the flavor profile and consumer acceptance of rib steaks from grass- or grain-fed cattle

Meat Science 90 (2012) 116–121

Contents lists available at ScienceDirect

Meat Science j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / m e a t s c i

Development of a beef flavor lexicon and its application to compare the flavor profile and consumer acceptance of rib steaks from grass- or grain-fed cattle Curtis Maughan, Rossarin Tansawat, Daren Cornforth, Robert Ward, Silvana Martini ⁎ Department of Nutrition, Dietetics, and Food Sciences, Utah State University, 8700 Old Main Hill, 750 North 1200 East, Logan, 84322-8700, UT, United States

a r t i c l e

i n f o

Article history: Received 23 November 2010 Received in revised form 7 February 2011 Accepted 3 June 2011 Keywords: Lexicon Beef Grass-fed Grain-fed Sensory evaluation Descriptive analysis

a b s t r a c t Ten panelists were selected from the local community to develop a meat lexicon composed of 18 terms that describe flavor attributes found in red meats. This flavor lexicon was used to compare the flavor profile of meat from beef cattle finished on grass or grain. Steaks from grass-fed animals were significantly (P b 0.05) higher in barny, bitter, gamey, and grassy flavor, and lower in juicy and umami notes. Gamey, barny, bitter and grassy were some of the attributes inversely correlated to the degree of liking of the meat and therefore can be classified as “negative” attributes. Brothy, umami, roast beef, juicy, browned, fatty and salty are some of the attributes positively correlated to the degree of liking of beef and therefore can be identified as attributes that drive consumers’ acceptance. Steaks from grass-fed cattle were rated by consumers as slightly liked (6.08 on a 9-point scale), while steaks from grain-fed animals were rated as moderately liked (7.05 on a 9-point scale). © 2011 Elsevier Ltd. All rights reserved.

1. Introduction Flavor is a combination of taste and aroma, and is one of the main factors that drive consumer acceptance of foods. Sensory evaluation is a powerful tool to evaluate the quality of a food product. In particular, sensory evaluation has been used during the last 20 years to identify meat flavors, both desirable and undesirable (Allen, Cornforth, Whittier, Vasavada, & Nummer, 2007; James & Calkins, 2008; Wadhwani, Murdia, & Cornforth, 2010). However, sensory studies usually differ in terminology, type of scale used, and type of panel (consumer vs. descriptive), and are usually focused on the negative attributes of beef. Variation among sensory panel methods hampers meaningful comparisons among studies. A standardized lexicon of terms applicable among sensory studies on fresh meats is strongly needed. Flavor lexicons have been used for decades in several high value products such as cheese, wine, whisky, coffee, and chocolate (Drake & Civille, 2003) where small changes in specific attributes can tremendously affect the acceptance of the product by the consumer. Johnson and Civille (1986) developed a flavor lexicon for warmed-over flavors (WOF) in meats. Their lexicon included terms such as: cooked beef lean, cooked beef fat, browned, serum/bloody, grainy/cowy, cardboardy, oxidized/rancid/painty, and fishy. They also included sweet, salty, bitter and sour in their lexicon. Their research

⁎ Corresponding author. Tel.: + 1 435 797 8136; fax: + 1 435 797 2379. E-mail addresses: [email protected] (C. Maughan), [email protected] (R. Tansawat), [email protected] (D. Cornforth), [email protected] (R. Ward), [email protected] (S. Martini). URL: http://www.MartiniResearch.com (S. Martini). 0309-1740/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.meatsci.2011.06.006

showed that WOF from reheated samples were associated with an increase of negative notes, such as cardboardy and oxidized, and a decrease in positive notes, such as cooked beef lean and cooked beef fat. Even though their research provided a lexicon for identifying and quantifying WOF, it did not provide a tool to evaluate the sensory profile of fresh cooked meats. Sensory evaluation techniques have also been used by Berry et al. (1980) to evaluate the flavor profile of loin steaks with different levels of maturity. These authors reported that beef from E maturity had higher aroma and flavor amplitudes and a greater predominance of “grassy” flavors. Stetzer, Cadwaller, Singh, McKeith, and Brewer (2008) also reported the effect of enhancement and ageing on flavor and volatile compounds in beef. They reported that several flavor-active compounds such as nonanal were increased by the enhancement and ageing of the beef, while pentanal, hexanal and hexanoic acid were decreased by these treatments. There are myriad studies using sensory evaluation to determine the flavor profile of the meat (Stelzleni & Johnson, 2008; Hamling, Jenschke, & Calkins, 2008; James & Calkins, 2008; Stetzer, Cadwaller, et al., 2008; Sitz, Calkins, Feuz, Umberger, & Eskridge, 2005; Stetzer, Tucker, McKeith, & Brewer, 2007; Stetzer, Tucker, McKeith, & Brewer, 2008; Rojas & Brewer, 2007); however, it is very difficult to compare results among these studies due to the lack of standardized terms. A standardized meat flavor lexicon will have immediate application to identify the effects of diet (grass or grain), maturity, and marbling on beef flavor profile. Other applications will include the identification of processing procedures (ageing, marinating) to minimize off-flavors (sour, grassy, gamey) and maintain desirable flavors (savory, umami, salty, brothy). The objectives of this research were: 1) to develop a standardized flavor lexicon for beef, 2) to use the new lexicon to identify and

C. Maughan et al. / Meat Science 90 (2012) 116–121

quantify differences between the flavor profile of beef from cattle fed different diets, and 3) to correlate these flavor differences with consumer acceptance. These objectives allow the identification of flavor notes that drive consumer acceptance in red meats. 2. Materials and methods 2.1. Meat samples Primal rib sections of three grass-fed steers were purchased from James Ranch, CO; while rib sections of two grain-fed steers and one heifer were obtained from USU's Animal Science Farm. The grain-fed animals were Black Angus bred, while the grass-fed animals were Red Angus sired with a mix of Hereford and Angus dams. The grass-fed animals were 24–27 months old and had a hanging weight between 318 and 360 kg. Their diets were supplemented with alfalfa during the winter, and they were finished for 120 days exclusively on grass. The grain-fed animals were 19– 20 months old, had a finish diet of 120 days consisting of 60% corn silage, 30% flaked barley, and 10% alfalfa, and were also 320–345 kg in hanging weight. The left and right rib sections were used from each animal. The characteristics of the beef samples used in this study are detailed in Table 1. The fat content of the samples was determined in uncooked rib steaks using the Soxhlet method using petroleum ether as the solvent (AOAC, 1990). Raw steak pH was measured on 10 g of sample that were finely chopped, diluted to 100 ml in distilled water, allowed to equilibrate at room temperature for 30 min and then filtered. Filtrate pH was measured, using a Fisher Accumet pH meter model 610 A (Fisher Scientific Inc, Salt Lake City, UT), equipped with a combination pH electrode calibrated immediately before use to pH 4.0 and 7.0. Primal ribs from each animal were vacuum packed after harvest, shipped to the Department of Nutrition, Dietetics, and Food Sciences at USU and immediately frozen at − 20 °C until use. The Longissimus dorsi muscles for each animal were used for the sensory tests. 2.2. Sample preparation for sensory tests Frozen ribeye steaks were cut to a thickness of 2.54 cm and thawed for 24 h before cooking. The ribeye steaks were then trimmed to leave only the Longissimus dorsi muscles for cooking, to avoid any variation in flavors between muscle types. Samples were prepared following the guidelines from the American Meat Science Association (1995). Steaks were cooked on electric griddles at 163 °C until reaching an internal temperature of 70 °C. Internal temperature was measured at the center of the steak using an AquaTuff 35200 digital thermometer (Atkins Technical Inc, Gainesvile, FL USA) equipped with a fast-responding microneedle probe. The probe was inserted horizontally from the side along the center line of the steaks during cooking. At least two readings were taken per steak to verify that steaks had reached the target internal temperature of 70 °C. They were then cut into 2.54 cm cubes and placed in covered aluminum dishes, and served to the panelists hot. Panelists tasted the samples in random order with 3-digit blinding codes under red colored lights to minimize bias.

117

Table 2 Taste concentrations used to achieve a specific taste intensity in aqueous phase during panel training. Attribute

Taste definition

Treatment

Levels (%)

Scale value

Bitter

Taste elicited by caffeine

Caffeine

Salty

Taste elicited by salts

Sodium chloride

Sour

Taste elicited by acids

Citric acid

Sweet

Taste elicited by sugar

Sucrose

Umami

Taste elicited by monosodium glutamate

Monosodium glutamate

0.05 0.08 0.15 0.20 0.35 0.50 0.05 0.08 0.15 2 5 10 0.7 1.4 2.8

2 5 10 2.5 5 8.5 2 5 10 2 5 10 5 9 13

2.3. Descriptive sensory evaluation A sensory descriptive panel (n = 10) was recruited and selected from the local community to develop a flavor lexicon for meats. Potential panelists were recruited using local newspapers and flyers in the community, and were screened for the panel based on their ability to differentiate between basic tastes in both identification and intensity rankings, according to established guidelines (American Society for Testing and Materials, 1981). Panelists who passed basic screening were recruited for the panel and monitored over time for ability to identify and quantify meat attributes in order to be included in the final evaluation. Panelists ranged in age from 18 to 60, with 7 males and 3 females, though demographics are not expected to influence the ratings in a trained descriptive panel. Panelists were trained for a minimum of 50 h on beef flavors using the 15-point Spectrum intensity scale (Muñoz & Civille, 1998). A 15-point intensity scale was used in the development of this standardized lexicon to allow the use of this tool in different types of meats, from plain beef to more flavorful products such as jerky and salami. This scale is commonly used in the development of other flavor lexicons such as cheese (Drake & Civille, 2003). Panelists were first trained on the identification and rating of the five basic tastes: sweet, sour, bitter, salty and umami. Solutions of sucrose, citric acid, caffeine, sodium chloride, and monosodium glutamate were used, respectively. The concentrations used to achieve specific taste intensity are included in Tables 2 and 3. After training with the five basic tastes, panelists were introduced to different pieces of meat to develop the meat flavor lexicon. Meat references were created to train the panelists in the identification and quantification of the flavor intensity of the terms included in the flavor lexicon. These references were based on previous studies (Berry et al., 1980, Johnson & Civille, 1986; Stelzleni & Johnson, 2008; Stetzer, Cadwaller, et al., 2008), and supplemented with flavors identified by the panelists during the lexicon development. Each reference was presented at different levels of intensity to standardize the use of the 15-point intensity scale. The lexicon

Table 1 Carcass characteristics of grain and grass-fed animals. HW: hanging weight; REA: rib eye area; BFT: back fat thickness; MS: Marbling score; MA: moderately abundant; M: moderate; S: small; Sl: Slight. Samples

HW (kg)

REA (cm2)

BFT (mm)

MS

Quality grade

pH

Fat (%)

Grain #1 Grain #2 Grain #3 Grass #1 Grass #2 Grass #3

320 330 345 318 330 360

81.3 80.6 87.7 80.0 78.7 85.8

1.3 0.5 1.3 0.3 0.8 0.5

MA M S Sl Sl Sl

Prime (heifer) high Choice (steer) low Choice (steer) Select (steer) Select (steer) Select (steer)

5.13 ± 0.02 5.15 ± 0.01 5.06 ± 0.02 5.28 ± 0.02 5.27 ± 0.01 5.27 ± 0.01

13.86 ± 1.99 12.38 ± 1.45 11.05 ± 1.40 3.03 ± 0.20 3.51 ± 0.40 3.54 ± 0.23

118

C. Maughan et al. / Meat Science 90 (2012) 116–121

Table 3 Taste concentrations used to train descriptive panelists in the detection, identification and quantification of mixtures of the 5 basic tastes.

Sweet 2% sour 0.15% umami 0.2% Sweet 5% sour 0.2% Salty 0.35%, sour 0.2% Sweet 2%, sour 0.2% Bitter 0.05%, salty 0.5% Sweet 5%, salty 0.5%, bitter 0.15% Sweet 5% salty 0.5% Sweet 2%, sour 0.15%, umami 0.2%

Bitter

Salty

Sour

Sweet

Umami

0 0 0 0 2.5 4 0 0

0 0 3 0 3.5 2 3 0.5

1.5 4.5 5.5 5 0 0 0 3

1 3.5 0 2 0 3 4 1

3.5 0 0 0 0 0 0 2

development consisted in first familiarizing the panelists with different flavor attributes using the references available in the literature. The number of descriptors selected to be used in the flavor lexicon was determined by panel consensus. After the flavor lexicon was developed, training in the intensity of these attributes was performed using the references detailed in Table 4. In general, references were prepared with lean or full fat ground beef, depending on the sample, and spiked with a specific compound (or cooked in a particular manner) to develop the desired flavor (Table 4). For the reference on the grassy attribute, grass-fed steak was purchased from a local store. For gamey references meats from elk and deer were provided by the Meat Laboratory at Utah State University. As training progressed, panelists had more individual evaluations as well as group discussions. Meat and reference samples were presented to panelists in individual booths, and results were collected by computer software. Red lighting and random presentation with 3-digit blinding codes were used during training. Panelist performance was measured throughout the study using graphical panelist evaluation software to ensure consistency and reproducibility (Nofima & Denmark, 2008). Panelists were considered to be trained on an attribute based on their ability to evaluate identical samples the same over multiple repetitions, as well as rate the samples similarly to the panel as a

whole. Additional focus was given in training sessions to those areas where panelists showed they had difficulties in quantifying attributes. After the completion of training, meat samples from grain- and grass-fed animals were presented to the 10 panelists (eight males, two females). The panel evaluated ribs from the three animals from each dietary treatment, for a total of six steak samples. Ribs were chosen randomly from the left or right side of the animals. Meat was presented in 2.54 cm cubes with three-digit codes in a balanced and randomized design. Panelists tasted and evaluated each sample individually, using water and unsalted crackers to cleanse their palettes between samples. After tasting all samples, panelists waited for 15 min before tasting a replicate of the samples in a new randomized order.

2.4. Consumer test A consumer panel consisting of 120 panelists rated the grass and grain-fed beef steaks for liking on a typical 9-point hedonic scale. Each panelist was presented with two steaks, consisting of one grain-fed and one grass-fed sample. The steaks were randomly selected from one of the three animals from each diet treatment. For a constant treatment, each animal was presented 40 times. The steaks were prepared in the same manner as for the descriptive panel. In brief, each steak was cooked on an electric griddle to an internal temperature of 70 °C, cut into 2.54 cm cubes, and placed in covered aluminum dishes to keep the samples warm. The samples were presented in random order with three-digit blinding codes to minimize bias. Panelists were asked to rate their overall liking of each sample on a 9-point hedonic scale (1 = dislike extremely, 2 = dislike very much, 3 = dislike moderately, 4 = dislike slightly, 5 = neither like or dislike, 6 = like slightly, 7 = like moderately, 8 = like very much, 9 = like extremely), comment on each sample, and answer basic demographic questions such as gender, age, and frequency of steak consumption.

Table 4 Flavor lexicon references used in the training of the meat descriptive panel. Basic tastes were also included in the lexicon and the references used are reported in Table 2. Attribute

Taste definition in beef

Reference

Scale value

Astringent

Mouth-drying and harsh sensation

Barny

Aromatics associated with feces

Bloody Brothy

Taste associated with undercooked meat Flavors and aromatics associated with boiled meat or soup stock

Browned

Flavors associated with meat that is cooked more and charred on the outside Taste associated with wild game meat Aromatic found in grass-fed animals

Tannic acid 0.05% in water Alum 0.1% in water 0.5 μg skatole/g of beef 1 μg skatole/g of beef Steak cooked to 55 °C 5% of low-sodium beef broth in ground beef 10% of low-sodium beef broth in ground beef 20% of low-sodium beef broth in ground beef Steaks cooked to 71 °C, allowed to brown on each side Wild game meat such as deer and elk 1 drop hexanal in 300 g beef 1 drop hexanal in 100 g beef 3 drops hexanal in 100 g beef Different types of steaks with varying levels of juice/toughness 90% Lean ground beef 80% Lean ground beef 73% Lean ground beef 40% cow liver in ground beef 75% cow liver in ground beef 100% liver 0.36% Ferrous sulfate in ground beef 0.50% Ferrous sulfate in 100 g ground beef Ground beef cooked then refrigerated for at least 24 h before reheating Fresh ground beef Ground beef cooked, held in oven for 1 h Ground beef cooked, held in oven for 2 h

7–8 8–10 5 10 10–12 7–9 9–11 12–14 Depending on "browness" of sample, ranges from 7 to 12 Depends on animal 4–6 7–8 15 Depends on sample

Gamey Grassy

Juicy Fatty

Sensation caused by meats with higher levels of juices Sensation caused by various levels of fat in the beef

Livery

Taste found in animal organs

Metallic

Taste associated with various metal flavors found in meat Flavor of reheated meat

Oxidized/warmed over flavor Roast beef

Flavor developed in beef after holding at temperature for long periods of time

4–6 6–8 10–12 6–8 10–12 12–14 5–7 8–10 6–10 0 Roast beef 1–3, browned 1–3 Roast beef 3–6, browned 3–6

C. Maughan et al. / Meat Science 90 (2012) 116–121

2.5. Statistical analysis Statistical Analysis Software (SAS) version 9.1 (SAS Institute, Inc., Cary, NC) with the proc glm function was used for analysis of variance to identify statistically significant differences at the 95% confidence level. Comparison of the means was made based on p-values (α = 0.05) using the least significant different adjustment to obtain differences of least means squares. Principle component analysis (PCA) using proc corr was used to analyze lexicon terms and their relationship to the samples and consumer preferences. 3. Results and discussion 3.1. Meat characteristics Table 1 shows the characteristics of the grass- and grain- fed meat. No differences were found in the hanging weight and in the rib eye area between the different types of meat. As expected, slight differences were found in pH values, with slightly higher values observed for the grass-fed meat. Similarly, differences were found in fat content, back fat thickness, marbling scores and grade, with lower fat contents found for the grass-fed beef. These are expected differences due to the feeding regime, and also to the different ages of the animals used. Grass-fed animals usually gain weight more slowly than feedlot cattle. Thus, grass-fed cattle require longer time than grain-fed cattle to reach the target harvest weight range. 3.2. Meat flavor lexicon development The descriptive panel was used to develop a standardized meat flavor lexicon. The lexicon consisted of 18 terms that describe flavor characteristics commonly found in meat. These terms are detailed in Table 4 and include: astringent, barny, bitter, bloody, brothy, browned, gamey, grassy, juicy, fatty, livery, metallic, oxidized (warmed-over flavor), roast beef, salty, sour, sweet, and umami. Table 4 also provides a definition of each term and the concentrations used to train the panelists on the quantification of these attributes. 3.3. Descriptive analysis of meat from grain- and grass-fed cattle After intensive training in the identification and quantification of meat attributes included in the lexicon, meat samples from grain- and

Table 5 Flavor profile of beef obtained from grain and grass-fed cows. Ratings for each treatment are expressed as the mean values ± standard deviations of the three animals tested. Data was obtained by a 10-member descriptive panel using a 15-point category scale. Attribute

Grass

Astringent Barny Bitter Bloody Brothy Browned Fatty Gamey Grassy Juicy Livery Metallic Oxidized/WOF Roast beef Salty Sour Sweet Umami

1.69 ± 2.20 0.84 ± 1.16 0.48 ± 0.80 0.25 ± 0.58 1.57 ± 1.80 0.64 ± 1.12 1.89 ± 2.32 0.77 ± 1.49 1.17 ± 1.85 1.67 ± 1.87 0.51 ± 1.22 0.57 ± 1.01 0.24 ± 0.72 1.00 ± 1.22 1.23 ± 1.29 1.28 ± 1.52 0.44 ± 1.01 3.22 ± 1.60

a,b

Grain a a

a a b

b

1.49 ± 1.72 0.02 ± 0.13 0.23 ± 0.65 0.48 ± 1.05 1.92 ± 1.98 0.95 ± 1.29 2.30 ± 2.44 0.08 ± 0.32 0.13 ± 0.46 2.39 ± 2.23 0.20 ± 0.61 0.33 ± 0.77 0.08 ± 0.37 1.22 ± 1.60 1.35 ± 1.31 1.21 ± 1.54 0.73 ± 1.59 4.78 ± 2.18

P value b b

b b a

a

0.4245 0.0001 0.0039 0.0999 0.0673 0.0654 0.0778 0.0016 0.0003 0.014 0.0657 0.1322 0.0715 0.1788 0.2230 0.6158 0.0881 0.0001

Ratings with the same superscript are not significantly different (α = 0.05).

119

grass-fed cattle were analyzed using the descriptive panel. As described Section 2.1, three animals from each treatment (diet) were used in the experimental design. Table 5 shows the panel average rating for each sample and each attribute. For each attribute, samples identified with the same superscript letter are not significantly different (α = 0.05). Higher intensity values were observed in the meats from grass-fed animals for barny, bitter, gamey, and grassy and lower intensity values were observed for juicy and umami. It is interesting to note that variability was also observed in some of the grass-fed attributes (data not shown) suggesting that animals also contribute to flavor variability. Farmer and Patterson (1991) report that several disulphide compounds are related to beef flavor. Some of these compounds include 2-methyl-3furanthiol and bis(2-methyl-3-furyl) disulphide. These compounds are usually products from the Maillard reaction between sulphur-containing amino acids and the reducing sugars in the meat. Allen, Vercelloti, Dupuy, and Spanier (1988) have identified pentanal, 2,4-decadienal, hexanal, and 2,3-octanedione in meats, among other compounds. Stetzer, Cadwaller, et al. (2008) also identified hexanal, 3-hydroxy-2butanone, 1-octen-3-ol, butanoic acid, and nonanal in beef samples. These authors show that the livery off-flavor in the meat is positively correlated with pentanal, hexanal, 3-hydroxy-2-butanone, and hexanoic acid. While rancid off-flavor is correlated with pentanal and 2phenyl furan, it is not correlated with hexanal. Most of the research performed on grass-fed and grain-fed beef is based on the meat quality and very little data reporting flavor differences in these types of meat is available (Melton, Amiri, Davis, & Backus, 1982). In fact, most of the research is based on the improved nutritional quality of the grass-fed beef in terms of fatty acid profile and on the volatiles compounds released from the meat. In general, beef obtained from grass-fed cattle has a higher content of mono- and polyunsaturated fatty acids and common volatile compounds include 1-penten-1-ol, 2-penten-1-ol (Aurousseau et al., 2007a; Aurousseau et al., 2007b; Elmore, Mottram, Enser, & Wood, 2006; Gatellier, Mercier, Juin, & Renerre, 2005; Ponnampalam, Mann, & Sinclair, 2006; Yang, Lanari, Brewster, & Tume, 2002). The previous research discussed above suggests that the flavor profile of meats is strongly dependant on the volatile compounds of the meat, which in turn might depend on the chemical characteristics of the samples, such as fatty acid composition. A recent report from Brewer (2006) describes the association of specific beef flavors with volatile compounds. For example, the term “grassy” is associated with hexanal, while fatty is associated with nona-2(E)-enal and sweet with delta-nonalactone. We suggest that flavor differences reported in this research are a consequence of differences in the chemical composition of the meat such as fat content and fatty acid composition and that similar volatile compounds, as the ones described in previous research, can be associated with these flavor differences. The identification and quantification of these compounds exceed the scope of this research. Results from the descriptive panel suggest that the newly developed flavor lexicon can be used to detect and quantify flavor differences in meat. In addition, results reported in this paper indicate that diet and animal play an important role in the flavor profile of the meat. 3.4. Consumer tests One hundred and twenty consumers participated in the acceptance test. Fifty-five percent of the panelists were male, and 45% were female. Seventy-six (63.3%) panelists were between 18 and 25 years of age, 20 (16.7%) were 26–35 years old, 8 (6.7%) were between 36 and 45, 8 (6.7%) were in the range of 46–55, and 8 (6.7%) were 56 or older. Data recorded on the frequency of steak consumption shows that 52 panelists (43.0%) ate steak less than once a month, 49 (40.5%) ate steak at least once a month, 18 (14.9%) ate at least once a week, and 2 (1.7%) ate steak at least once a day. Table 6 shows the results obtained from the consumer acceptance test. All samples were well liked, with an average of a 7.05

120

C. Maughan et al. / Meat Science 90 (2012) 116–121

Table 6 Consumer acceptance (degree of liking) of beef obtained from grain- and grass-fed animals. Three animals of each treatment were evaluated by a 120-member consumer panel. Attribute

Grass 1

Liking

5.93 ± 1.86

Grass 2 c

6.10 ± 1.43

Grass 3 c

6.20 ± 1.63

Grain 1 bc

Grain 2

7.15 ± 1.56

a

Grain 3

6.75 ± 1.69

ab

7.25 ± 1.45

P value a

0.0001

Ratings with the same superscript are not significant different (α = 0.05).

browned, roast beef, juicy, fatty, sweet, salty, and bloody, respectively. These attributes can therefore be identified as positive attributes. Similarly, meats obtained from the grass-fed animals are strongly correlated to attributes such as oxidized, grassy, barny, bitter, gamey, astringent, sour, livery, and metallic which can be associated to negative terms due to their negative correlation to the degree of liking (Fig. 1). Correlation values for these attributes were −0.90, −0.90, −0.84, −0.83, −0.65, −0.51, −0.33, −0.24, −0.16, and for grassy, barny, oxidized, bitter, gamey, livery, metallic, astringent, and sour, respectively. It is interesting to observe that even though significant differences were observed in the degree of liking from consumers, these differences were small (see Section 3.3). In addition, these results suggest that the slightly lower consumer acceptance for meats obtained from the grass-fed animals can be attributed to the increased intensity of negative attributes such as barny, bitter, grassy, gamey, and decreased intensity of positive attributes such as juicy and umami. It is important to note here that “positive” and “negative” attributes are defined according to the degree of liking obtained from the consumer tests performed in our facilities. Therefore this classification is relative to the population tested. Consumers with different eating habits might rate some of the “negative” attributes as “positive” and vice versa. Given the higher price of grass-fed beef, one may conclude that demand is driven by its perception of high nutritional value, with less regard for its sensory properties (Umberger, Boxall, & Lacy, 2009). And as consumers of grass-fed beef often point out, cooking methodologies and recipes have been developed to enhance eating qualities of grass-fed beef.

(moderately liked) and 6.08 (slightly liked) rating for the meat obtained from the grain- and grass-fed animals on a 9-point hedonic scale. These ratings were significantly different showing a slightly lower degree of liking for the meat obtained from the grass-fed cattle. No significant differences were observed between the animals from each treatment. However, the acceptance rating for one of the grainfed animals was not significantly different from the acceptance rating of one of the grass-fed animals (Table 6). As explained in Section 3.2, this demonstrates that some of the variability found in meat flavor profile is due to animal variation. No significant differences were observed when the consumer data was evaluated for differences between age groups and sex. It is important to note here that consumer data obtained is limited to the demographics used in this research. To extrapolate these results nationwide, consumer tests should be performed in different parts of the country with significantly more consumers. However, such broad study exceeds the scope of this research. 3.5. Relating meat flavor profile with consumer acceptance Principal component analysis (PCA) was performed with the flavor profile and consumer acceptance data. The objective of this step was to find relationships between the flavor profiles of the samples and their acceptance by consumers and identify the attributes in red meats that drive consumer acceptance. Fig. 1 shows the PCA plot of these data. It can be seen that principal component 1 contributes to 61.4% of the data variability while principal component 2 contributes to the 16.8% of the variability. These two principal components explain 78.2% of the variability. It is clear from Fig. 1 that meat obtained from grain-fed animals was strongly correlated with attributes such as salty, juicy, fatty, sweet, bloody, umami, brothy, roast beef and browned. These attributes are also highly correlated with the degree of liking of the samples. Correlation values were 0.97, 0.88, 0.88, 0.85, 0.82, 0.82, 0.74, 0.71, and 0.68 for brothy, umami,

4. Conclusions A standardized lexicon to describe flavor properties of meats was developed by our research team. This flavor lexicon includes 18 terms that describe positive and negative flavor attributes found in meats. The lexicon was used to describe the flavor profile of meat samples

2 Grass 1 1.5

PC2 (16.8%)

Livery Bitter Barny

Metallic

Gamey Sour

1

0.5

0 -2

-1.5 Grass 3

-1 -0.5 Grassy Oxidized Grass 2

0 Astringent -0.5

Salty

Juicy

Grain 3 Fatty Sweet Bloody Umami Brothy

Roast Beef 0.5 Grain 1

1 Liking Browned

1.5

2

Grain 2

-1

-1.5

-2

PC1 (61.4%) Fig. 1. PCA relating the flavor profile of beef obtained from grain and grass-fed cows to consumer acceptance.

C. Maughan et al. / Meat Science 90 (2012) 116–121

obtained from animals that were fed different diets. The newly developed flavor lexicon was used to demonstrate that animal diets can affect the flavor profile of meat. Relating the lexicon to consumer acceptance allows for classification of these attributes as positive or negative attributes for a specific consumer market. Terms associated with “positive” attributes are: brothy, umami, roast beef, juicy, browned, fatty, and salty. Terms associated with “negative” attributes include: oxidized, bitter, barny, gamey, grassy, livery, metallic, and astringent. The flavor lexicon can be used in the future, along with other consumer acceptance tests, to better understand consumers’ preferences of one type of meat over another. Acknowledgements The authors thank the panel members for their commitment to the sensory panel. We would also like to thank the Institutional Review Board for approving the project and The Beef Checkoff for financial support. This paper was approved by the Utah Agricultural Experiment Station as paper # 8237. References Allen, K., Cornforth, D., Whittier, D., Vasavada, M., & Nummer, B. (2007). Evaluation of high humidity and wet marinade methods for pasteurization of jerky. Journal of Food Science, 72, C351–C355. Allen, J. A., Vercelloti, J. R., Dupuy, H. P., & Spanier, A. M. (1988). Assessment of beef flavor quality: A multidisciplinary approach. Food Technology, 133–138. American Meat Science Association (1995). Research guidelines for cookery, sensory evaluation, and instrumental tenderness measurements of fresh meat. American Meat Science Association in cooperation with National Live Stock and Meat Board. Chicago, IL: Meat Science Association. American Society for Testing and Materials (1981). Guidelines for the selection and training of sensory panel members. American Society for Testing and Materials. Philadelphia, PA: American Society for Testing and Materials. AOAC, Assn. Official Analytical Chemists, Inc. (1990). Official methods of analysis. (15th ed). Procedure 960.39. Arlington, Va, Vol. 2. Aurousseau, B., Bauchart, D., Galot, A. L., Prache, S., Micol, D., & Priolo, A. (2007a). Indoor fattening of lambs raised on pasture: 1. Influence of stall finishing duration on lipid classes and fatty acids in the longissimus thoracis muscle. Meat Science, 76, 241–252. Aurousseau, B., Bauchart, D., Galot, A. L., Prache, S., Micol, D., & Priolo, A. (2007b). Indoor fattening of lambs raised on pasture: 2. Influence of stall finishing duration on triglyceride and phospholipid fatty acids in the longissimus thoracis muscle. Meat Science, 76, 417–427. Berry, B. W., Maga, J. A., Clakins, C. R., Wells, L. H., Carpenter, Z. L., & Cross, H. R. (1980). Flavor profile analyses of cooked beef loin steaks. Journal of Food Science, 45, 1113–1121. Brewer, S. M. (2006). The chemistry of beef flavor. National Cattlemen's Beef Association, 1–13.

121

Drake, M. A., & Civille, G. V. (2003). Flavor lexicons. Comprehensive Reviews in Food Science and Food Safety, 2, 33–40. Elmore, J. S., Mottram, D. S., Enser, M., & Wood, J. D. (2006). Chapter 3: The effects of diet, breed, and age of animal at slaughter on the volatile compounds of grilled beef. In F. Shahidi, & H. Weenan (Eds.), Food lipids: Chemistry, flavor and texture (ACS Symposium Series) (pp. 35–48). : American Chemical Society. Farmer, L. J., & Patterson, R. L. S. (1991). Compounds contributing to meat flavor. Food Chemistry, 40, 201–205. Gatellier, P., Mercier, Y., Juin, H., & Renerre, M. (2005). Effect of finishing mode (parture- or mixed-diet) on lipid composition, colour stability and lipid oxidation in meat from charolais cattle. Meat Science, 69, 175–186. Hamling, A. E., Jenschke, B. E., & Calkins, C. R. (2008). Effects of aging on beef chuck and loin muscles enhanced with ammonium hydroxide and salt. Journal of Animal Science, 86, 1200–1204. James, J. M., & Calkins, C. R. (2008). The influence of cooking rate and holding time on beef chuck and round flavor. Meat Science, 78, 429–437. Johnson, P. B., & Civille, G. V. (1986). A standardized lexicon of meat WOF descriptors. Journal of Sensory Studies, 1, 99–104. Melton, S. L., Amiri, M., Davis, G. W., & Backus, W. R. (1982). Flavor and chemical characteristics of ground beef from grass-, forage-grain- and grain-finished steers. Journal of Animal Studies, 55, 77–87. Muñoz, A. M., & Civille, G. V. (1998). Universal, product and attribute specific scaling and the development of common lexicons in descriptive analysis. Journal of Sensory Studies, 13, 57–75. Nofima, Marin, & Denmark, Technical University of (2008). (Version 1.3.2) [Software]. Available from http://www.panelcheck.com/ Ponnampalam, E. N., Mann, N. J., & Sinclair, A. J. (2006). Effect of feeding systems on omega-3 fatty acids, conjugated linoleic acid and trans fatty acids in Australian beef cuts: Potential impact on human health. Asia Pacific Journal of Clinical Nutrition, 15(1), 21–29. Rojas, M. C., & Brewer, M. S. (2007). Effect of natural antioxidants on oxidative stability of cooked, refrigerated beef and pork. Journal of Food Science, 72, S282–S288. Sitz, B. M., Calkins, C. R., Feuz, D. M., Umberger, W. J., & Eskridge, K. M. (2005). Consumer sensory acceptance and value of domestic, Canadian, and Australian grass-fed beef steaks. Journal of Animal Science, 83, 2863–2868. Stelzleni, A. M., & Johnson, D. D. (2008). Effect of days on concentrate feed on sensory off-flavor score, off-flavor descriptor and fatty acid profiles for selected muscles from cull beef cows. Meat Science, 79, 382–393. Stetzer, A. J., Cadwaller, K., Singh, T. K., McKeith, F. K., & Brewer, M. S. (2008). Effect of enhancement and ageing on flavor and volatile compounds in various beef muscles. Meat Science, 79, 13–19. Stetzer, A. J., Tucker, E., McKeith, F. K., & Brewer, M. S. (2007). Quality changes in beef Gluteus Medius, Infraspinatus, Psoas Major, Rectis Femoris, and Teres Major enhanced prior to aging. Journal of Food Science, 72, S242–S246. Stetzer, A. J., Tucker, E., McKeith, F. K., & Brewer, M. S. (2008). Quality changes in beef Complexus, Serratus, Ventralis, Vastus Lateralis, Vastus Medialis, and Longissimus Dorsi muscles enhanced prior aging. Journal of Food Science, 73, S6–S10. Umberger, W. J., Boxall, P. C., & Lacy, R. C. (2009). Role of credence and health information in determining US consumers’ willingness-to-pay for grass-finished beef. The Australian Journal of Agricultural and Resource Economics, 53, 603–623. Wadhwani, R., Murdia, L. K., & Cornforth, D. P. (2010). Effect of muscle type and cooking temperature on liver-like off-flavour of five beef chuck muscles. International Journal of Food Science & Technology, 45, 1277–1283. Yang, A., Lanari, M. C., Brewster, M., & Tume, R. K. (2002). Lipid stability and meat colour of beef from pasture- and grain-fed cattle with or without vitamin E supplement. Meat Science, 60, 41–50.