Article pubs.acs.org/JAFC
Impact of Brassica and Lucerne Finishing Feeds and Intramuscular Fat on Lamb Eating Quality and Flavor. A Cross-Cultural Study Using Chinese and Non-Chinese Australian Consumers Damian Frank,*,† Peter Watkins,‡ Alex Ball,§ Raju Krishnamurthy,† Udayasika Piyasiri,† James Sewell,# Jordi Ortuño,† Janet Stark,⊥ and Robyn Warner⊗ †
Commonwealth Scientific Industrial Research Organisation (CSIRO), 11 Julius Avenue, North Ryde, NSW 2113, Australia Commonwealth Scientific Industrial Research Organisation (CSIRO), 671 Sneydes Road, Werribee, VIC 3030, Australia § Meat & Livestock Australia (MLA), Level 1, 40 Mount Street, North Sydney, NSW 2060, Australia # PGG Wrightson Seeds (Australia) Leigh Creek Research Station, 4 Blackswamp Road, Leigh Creek, VIC 3052, Australia ⊥ Commonwealth Scientific Industrial Research Organisation (CSIRO), 39 Kessels Road, Coopers Plains, QLD 4108, Australia ⊗ Faculty of Veterinary and Agricultural Science, The University of Melbourne, Royal Parade, Parkville, VIC 3010, Australia ‡
S Supporting Information *
ABSTRACT: Use of forage brassicas (Brassica napus) and lucerne (alfalfa; Medicago sativa) as ruminant feeds has been linked to unacceptable flavors in sheepmeat. Lambs from low and high intramuscular fat sires were allocated to one of four finishing feedsperennial ryegrass (Lolium perenne), lucerne, and two brassica foragesfor a 6 week period. Grilled loins (Longissimus thoracis et lumborum) were subjected to chemical and sensory analysis by a trained panel and also evaluated by non-Chinese and Chinese background Australian consumers. Consumer liking was similar for both groups, and liking was highest for the brassicaand lucerne-finished lamb, especially from high intramuscular fat sires. No evidence of a distinctive lucerne- or brassica-induced flavor taint was measured by the trained panel or gas chromatography−mass spectrometry−olfactometry. The diets influenced the composition of lipids and branched-chain fatty acids in the subcutaneous fat, and the concentration of total branched-chain fatty acids was positively correlated with flavor and overall liking. Significantly higher levels of key aroma volatiles were measured in the higher fat samples. KEYWORDS: lucerne, alfalfa, brassica, glucosinolates, olfactometry, branched-chain fatty acids
■
INTRODUCTION Many brassica cultivars (Brassica napus spp. biennis) and lucerne (alfalfa, Medicago sativa) are nutritious ruminant feeds; however, they are underutilized due to concerns that they can introduce undesirable flavors or taints into meat and other animal products.1−5 Although animal genetics is an important determinant of lamb eating quality and flavor, the diet can have direct and indirect effects on meat sensory properties.5−9 The energy and protein contents of feeds affect the rate of weight gain, muscle growth, fat deposition, and the underlying meat structure and consequently the textural sensory properties.10,11 Feeds and animal age may also affect the flavor through altering the composition and amount of lipids within the intramuscular fat (IMF); for example, the amount of branched-chain fatty acids (BCFAs) present in ovine fat is associated with “mutton” flavor.6,12,13 Consumers from some Asian backgrounds (e.g., Japanese, Singaporean) are known to be sensitive to BCFAs, which may be a barrier to acceptance of sheepmeat.14−17 China is a substantial export market for Australian lamb, and it would be useful to know whether Chinese consumers also have the same sensitivity to BCFAs in lamb. Other compounds derived from rumen metabolism, such as 4-methylphenol (p-cresol) and 3-methylindole, accumulate in fat reservoirs and are associated with distinct “pastoral” or “barnyard” flavor notes in sheepmeat, which some consumers find undesirable.18 © XXXX American Chemical Society
Lucerne (alfalfa) is a perennial high-quality summer feed and adapts to a wide range of environments, including dry conditions, and being a legume has the ability to fix atmospheric nitrogen. Despite these advantages, it remains underutilized in Australian mixed-farming systems.19,20 Past research indicated that lucerne can create undesirable flavors in Corriedale lambs.21,22 However, other studies have shown no negative effects.23 Forage brassicas including radish, turnip, swedes, and kale are used as animal feeds but generally not as a finishing diet. Forage rape is a useful feed during summer as it thrives in low water conditions and is high in protein and energy, with good digestibility.1,24 Two commercial forage brassicasTitan and Greenlandhave useful characteristics as ruminant fodders. Titan is a high-yielding forage rape of intermediate height with kale genetics and increased aphid tolerance. Greenland is a late-flowering forage rape, well suited to sowing during late summer until spring. Cruciferous or brassica plants are unique for their secondary metabolites known as glucosinolates (GSL). The type and concentration of GSLs vary widely,25 and their breakdown products, the Received: May 4, 2016 Revised: August 9, 2016 Accepted: August 13, 2016
A
DOI: 10.1021/acs.jafc.6b02018 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
(equivalent to ∼1 ha or ∼150 m × 70 m) were sufficient to support more than 25 animals for a 6−8 week finishing period, although the animals on the ryegrass required some supplementation with barley and lupin in the final 2 weeks of the trial. The plots provided sufficient shade and water ad libitum for animals. Lambs were transferred to the experimental site, and the radio frequency identification tags were used to separate the animals into treatments to allow a balanced allocation of low and high IMF genotypes and liveweight, across the four finishing feed treatments. A feed budget was calculated for each treatment, and strip grazing was conducted, using an electric fence, within each plot to control feed intake throughout the trial. The lambs were finished on the respective trial plots conducted for 6 weeks in January and February 2013. Details of animal growth rates and feed utilization have been described elsewhere.28 Volatile and Odor Characterization of Brassica Feeds. Greenland and Titan stem and leaf material was collected at two time points in the field, January 22 (day 11) and February 13, 2013 (day 34), and frozen at −80 °C until later analysis. Replicate (n = 3) samples of 20 g (samples from leaves and stem) were macerated in icecold Milli-Q water (60 mL) and immediately transferred into gastight headspace sampling vessels along with the internal standard (IS) 4methyl-1-pentanol (40 ng g−1). After equilibration at 37 °C for ∼15 min, the headspace was purged with high-purity nitrogen for 30 min, and volatiles were concentrated onto Tenax traps and subjected to GC-MS and/or olfactometry under the same conditions as described in the following section. Three replicate samples for each brassica type were prepared and analyzed by GC-MS. Samples from February 13 were also subjected to qualitative olfactometry (three assessors). The odor quality (O) at the sniff port was recorded and matched to volatile compounds at the same elution time. Volatiles were tentatively identified on the basis of electron impact (EI) mass spectral data and matches with the NIST (version 2.0) mass spectral database and the existence of a corresponding M + H+ ion by methanol chemical ionization (CI) mass spectrometry. Semiquantitative estimation of volatile concentration was calculated against the IS (mg kg−1). The identity of some compounds was confirmed by the use of authentic reference standards (St). Animal Slaughter. Animals were slaughtered as part of commercial processing. Hot carcass weight (HCW) was recorded. After overnight chilling (2 °C) 24-h post-mortem, fat samples were removed from carcasses and GR-fat depth (mm) was measured over the 12th rib using a GR knife. Meat color was measured on the cut surface of the (M. longissimus thoracis et lumborum, LTL). Surface lightness (L*), redness (a*), and yellowness (b*) and ultimate pH (pHu) were measured according to published protocols.5 Loins (LTL) from both sides of the carcass were removed, and fat and other extraneous muscles were trimmed off; left and right striploins were separately labeled, vacuum packed, and wet aged in Cryovac bags for 5 days at 2 °C before freezing at −20 °C. Semifrozen striploins were cut into 15 mm steaks using a band saw, wrapped in cling film, vacuum packed in bags, and frozen until use (within 2 months for the sensory testing). Each steak was sufficient for two servings; for example, 5 steaks provided 10 servings. For larger animals, at least 15 × 15 mm steaks were obtained from each loin, sufficient to perform trained sensory analysis in triplicate. Right striploins were assessed by the trained sensory panel, and matching left striploins from the same animal were used in the cross-cultural consumer study. For each feed × genotype combination, eight replicate animals were subjected to trained panel and consumer sensory testing (n total = 64 for each). Remaining steaks and steaks from animals with lower HCW were used for sensory panel training and physicochemical analyses. Modified Warner−Bratzler Shear Force and Cook Loss. A modified Warner−Bratzler shear force test (WBS) was performed on cooked samples, allowing a smaller sample than normal to be utilized, according to previously reported protocols.29 Eight replicates were used for measurements on low and high IMF genotype samples for each of the four finishing feeds (total n = 64). Briefly, weighed samples in plastic bags were heated (70 °C, 60 min), cooled in an ice slurry (20 min), patted dry, and reweighed to determine the “cook loss”, the amount of water lost (% w/w) from meat samples during heating.
isothiocyanates (ITCs) and nitriles, may introduce undesirable taints into the meat. Greenland is known to have a higher total GSL content and a different GSL profile from that of Titan (unpublished data). A nonbrassica “washout” diet has been recommended to avoid the occurrence of brassica taints in lamb.1,2,4 Brassica plants also contain substantial amounts of Smethyl-L-cysteine sulfoxide, which generates potent odor-active sulfur volatiles such as methanethiol, dimethyl disulfide, and dimethyl trisulfide in the presence of plant or bacterial cysteine lyase enzymes.26,27 The main purpose of this study was to assess the potential for off-flavors or taints in the cooked meat of lamb finished on newer commercial varieties of brassica rape (Titan, Greenland) and lucerne, compared to lambs finished on conventional ryegrass (Lolium perenne). A secondary aim was to test whether lamb produced from sires with higher levels of IMF produce better eating quality and flavor, also when finished on brassica or lucerne feeds. To accomplish these goals, meat derived from lambs from low and high IMF sires, finished on four different feeds (ryegrass, lucerne, Titan, Greenland), was subjected to quantitative chemical analysis; the fatty acid composition of the lamb subcutaneous fat, including the BCFAs, and targeted analysis for 4-methylphenol and 3-methylindole. Furthermore, gas chromatography−mass spectrometry−olfactometry (GCMS-O) was used to characterize aroma volatiles and potential taint volatiles from brassica plant material and from freshly grilled lamb. Finally, grilled lamb was subjected to quantitative descriptive analysis by a trained sensory panel and was also evaluated by Chinese Australian and non-Chinese Australian consumers for liking and other sensory attributes.
■
MATERIALS AND METHODS
Chemicals. Authentic reference standard compounds were used to confirm the identity of most volatiles; 2-methylpropanal, 2-ethylfuran, 1-penten-3-one, 2,3-butanedione, methyl butanoate, ethyl butanoate, dimethyl disulfide, hexanal, 2-methylthiophene, 2-heptanone, heptanal, limonene, 2-pentylfuran, octanal, 1-octen-3-one, 1-octen-3-ol, nonanal, dimethyl trisulfide, (E)-2-nonenal, (E)-2-decenal, (E)-2-hexenal, 2acetyl-2-thiazoline, (E,E)-2,4-decadienal, 4-methoxyphenol, and 4methylphenol were supplied by Sigma-Aldrich (Castle Hill, Australia). 3-Methylbutanal, 2,3-pentanedione, 2-methylpyrazine, ethylpyrazine, 2,6-dimethylpyrazine, 2,3-dimethylpyrazine, 6-methyl-5-hepten-3-one, 2-ethyl-3,5-dimethylpyrazine, trimethylpyrazine, furfural, and 2-phenylethanal were supplied by Givaudan (ex-Quest) (Baulkham Hills, Australia). Methional, decanal, benzaldehyde, and 4-methylphenol were purchased from Fluka (Darmstadt, Germany). The fatty acid methyl ester standard mix (Supelco GLC-20), 4-methyl-1-pentanol internal standard, 3-methylindole, and 4-methyloctanoic, 4-ethyloctanoic, 4-methylnonanoic, and 2-butyloctanoic acids were also purchased from Sigma-Aldrich. Selection of Animals and Feeding Trial. Animal ethics approval in the state of Victoria was obtained (CSIRO-AEC 31592). Daily animal monitoring was carried out during the feeding trial. Ewe lambs (n = 125) were sourced from an existing experiment where extensive information on animal genetics was available. Sheep genetics are known to affect lean muscle growth and fat deposition.8,9 Data for dressing percent (%), lean meat yield (%), IMF (%), and shear force (N) were available for the sires of the lambs in the study (data not shown), and these data were used to select low and high IMF genotypes. The feeding trial was conducted at the PGG Wrightson Seeds Research Station, at Leigh Creek near Ballarat, Victoria, Australia (37°56′ S, 143°95′ E). Two brassica cultivars of commercial importance (cv. Titan and cv. Greenland) were selected for cultivation in separate plots. Lucerne and ryegrass (control) paddocks were also prepared adjacent to the brassica plots.28 The plots and feed available B
DOI: 10.1021/acs.jafc.6b02018 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry Samples were stored overnight at 4 °C before the initial yield (IY) and peak force (PF) were measured according to published protocols.29 Analysis of Lipids, Branched-Chain Fatty Acids, 4-Methylphenol, and 3-Methylindole. The percent fat in the LTL samples was determined using an oven moisture method.30 BCFAs are present in sheep fat in either the free form or bound within triacylglycerols.18 In the analytical method used in the present study the sum of free and bound BCFAs was measured as total BCFAs. Note that this method was in contrast to others reported where only the free BCFAs were measured.13 Both the free fatty acids and BCFAs were converted to fatty acid methyl esters (FAMEs) in the same step according to published procedures.31 Subcutaneous fat samples (ca. 5−10 g) were collected from the area over the gluteus medius, to minimize carcass damage, post slaughter and frozen at −20 °C until needed. Prior to analysis, the surface layer of the fat was removed and the remainder cut into small portions (ca. 0.5 cm3 squares). The fat was melted by heating the cut portions in a microwave oven for approximately 5 min. Four low and four high IMF samples were prepared for fat from each of the finishing feeds in duplicate (n total = 32 samples). An extract of the molten sheep fat (114 μL) was combined with 2-butyloctanoic acid (10 μL, 5 mg mL−1 as an internal standard), tetrahydrofuran (2 mL), and 5% (v/v) H2SO4/methanol (1 mL) and then heated at 50 °C for 16 h. After cooling, hexane (1 mL) and saturated NaCl solution (1 mL) were added and vortex mixed. The organic layer was washed with 5% NaHCO3 solution (1 mL), ready for GC analysis. The FAMEs (1 μL) were separated with isothermal heating (180 °C) using a Supelco SP-2560 column (l = 100 m, i.d. = 0.25 mm, film thickness = 0.2 μm) in an Agilent model 6890 gas chromatograph (GC) with a flame ionization detector (FID) using helium as the carrier gas (1.2 mL min−1). The injector was heated at 250 °C with split injection (50:1) used for the analysis. The FID was heated at 250 °C (the flow rates for H2, air, and nitrogen were 45, 450, and 45 mL min−1, respectively). FAME identification was made using a Supelco GLC-20 FAME mix standard and comparison with FAME solutions prepared from standard anhydrous milk fat. The results were reported as proportions of the total (identified) fatty acid content, with the response factor for each FAME set to 1. A 1 μL volume of BCFAFAMEs was injected at 250 °C in splitless mode with helium carrier gas (2.0 mL min−1) onto a GC-MS (Agilent 6890 GC, 5793 mass selective detector). FAMEs were separated on a DB-5 capillary column (J&W, 30 m, 0.32 mm, 0.32 μm). The oven was initially held at 80 °C (3 min), heated to 160 °C at a rate of 8 °C min−1, and held at 300 °C (2.5 min). The transfer line was held at 280 °C. The MS was operated in single ion monitoring (SIM) mode with the detector set to 400 V above the autotune value. The analyte response was quantified using a characteristic target ion using Chemstation software (Agilent) using the external standard technique. The same quantitative ion was used for the methyl esters of the BCFAs and 2-butyloctanoic acid, m/z 87, the McLafferty rearrangement ion. Additional qualifying ions were used to confirm the identification of 4-methyloctanoic (MOA), 4ethyloctanoic (EOA), 4-methylnonanoic (MNA), and 2-butyloctanoic acids, m/z 115, 113, 129, and 130, respectively. 4-Methylphenol and 3methylindole were analyzed according to published methods.5 Analysis of Grilled Lamb Volatiles by Gas Chromatography−Mass Spectrometry−Olfactometry. For volatile extraction, lamb meat samples were grilled in the same manner as for the sensory and consumer studies (see the following section). Three grilled lamb steaks from each animal were used to prepare each volatile extract. Five high and five low IMF samples were prepared for lamb samples from each finishing feed (n total = 40). The freshly grilled steaks were suspended in Milli-Q water (1:2 ratio, 37 °C) and homogenized to a fine slurry as described previously.5 A 60 g sample of suspension was placed in a headspace sampling vessel with added internal standard (4methyl-1-pentanol, 40 ng g−1) and a Teflon stir bar. The headspace was purged with high-purity nitrogen for 30 min at 37 °C, and volatiles were extracted onto Tenax-TA traps (60/80 mesh size, 100 mg). Samples were desorbed (260 °C, 5 min) by short-path thermal desorption (TD-5, Scientific Instrument Services, Ringoes, NJ, USA), transferred into the hot injector (240 °C, splitless), and analyzed using an Agilent (ex-Varian 4000 ion trap GC-MS) connected to a heated
sniffing port (ODO-II, SGE/Trajan Analytical Science, Ringwood, Australia). Volatiles were separated using a polar capillary column (Zebron-Wax, 60 m, 0.32 mm, 0.5 μm), using a temperature program from an initial temperature of 30 °C (held 10 min), increased at 8 °C min−1 to 250 °C (held 2 min). The effluent from the sniff port was simultaneously subjected to olfactory assessment (GC-MS-O) by trained sniffers using direct time intensity.31 Five assessors evaluated each eight samples, representing a low and high IMF sample from each of the four finishing feeds (n = 40 total). The integrated odor intensity was calculated as the average area under the curve for replicate response from assessors.5 Volatiles eluting at the same retention time were tentatively identified on the basis of EI and CI mass spectral data and odor quality (O). The identity of most compounds was confirmed by authentic reference standards (St). Semiquantitative estimation of volatile concentration was calculated against the IS (mg kg−1). Panel Training and Quantitative Descriptive Analysis. Ethics approval was granted for the trained panel study (LRRP15/2012-E). The 10-member sensory panel was composed of 9 females and 1 male; average age = 51 ± 6 years. Panelists were remunerated for their time. All assessors had been screened for sensory acuity and had extensive prior experience in participating in descriptive sensory analysis of meat and other products. Eight low IMF and eight high IMF samples for each finishing feed type were used in the trained panel and consumer testing (n total = 64). After thawing overnight (4 °C), lamb steaks were grilled in a clamshell grill (Silex, Marrickville, Australia) at 220 °C for 180 s. according to a standardized grilling protocol.32 After grilling, samples were covered loosely with aluminum foil and allowed to rest (3 min), before cutting into four equal pieces. Two pieces of warm lamb were placed into a coded standard wine glass and covered before serving to panelists or consumers according to a randomized presentation order. Five 2 h training sessions were held to generate and define the sensory vocabulary that best described the differences in odor, flavor, taste, texture, aftertaste, and after-feel attributes. During training, assessors were equally exposed to samples representing experimental design variables, that is, grilled meat from each of the four feeds and low and high IMF genotypes. Quantitative descriptive analysis was performed in 2 h sessions over a 2 week period. Attributes were rated on 100 mm unstructured line scales anchored at 5 and 95% for each descriptive attribute. Nine grilled lamb odor attributes were evaluated orthonasally (e.g., by sniffing the headspace): overall impact, lamb/ mutton, grilled, bloody, caramel, barnyard, hay/grainy, livery, and oxidized fat odor. Panelists then placed a warm lamb sample in their mouth (using a toothpick) and chewed three times before assessing eight flavor attributes (retronasally): overall impact, lamb/mutton, dairy, grass, vegetal, fatty, livery and metallic. Taste attributessweet, salty, sour/acidic, and umamiwere then assessed, and the sample was swallowed. Five aftertaste sensations were measured after swallowing: acidic, astringent, oily mouthcoating, metallic, and lingering. The second piece of warm lamb was chewed three times, and tenderness and juiciness were rated. The number of chews was counted up to the point of swallow; the number of chews to swallow and the amount of connective tissue in the mouth were assessed at this point. Panelists were required to cleanse their palate with cucumber slices, dry crackers, and water between samples. Data were recorded and stored using the Compusense sensory data acquisition software (version 5.2, 2004; Compusense Inc., Guelph, ON, Canada). Cross-Cultural Consumer Testing. Ethics approval was obtained for the consumer study (LRRP-17/2013). Metropolitan Sydney has a large population of recent Chinese immigrants; cultural aversions to lamb flavor (if they exist) are likely to still be present. A panel (n = 60) of Australian consumers from a Chinese background (CB) and Australian consumers from a non-Chinese background (NCB) (n = 60) were recruited by an external market research organization. Consumers were remunerated for their participation. Inclusion criteria for CB consumers were as follows: of Chinese ethnicity but living in Australia for
