Prediction of processed cheese instrumental texture and meltability by mid-infrared spectroscopy coupled with chemometric tools

The objective of this study was to determine the potential of mid-infrared spectroscopy coupled with multidimensional statistical analysis for the prediction of processed cheese instrumental texture and meltability attributes. Processed cheeses (n = 32) of varying composition were manufactured in a pilot plant. Following two and four weeks storage at 4 °C samples were analysed using texture profile analysis, two meltability tests (computer vision, Olson and Price) and mid-infrared spectroscopy (4000–640 cm⁻¹). Partial least squares regression was used to develop predictive models for all measured attributes. Five attributes were successfully modelled with varying degrees of accuracy. The computer vision meltability model allowed for discrimination between high and low melt values (R² = 0.64). The hardness and springiness models gave approximate quantitative results (R² = 0.77) and the cohesiveness (R² = 0.81) and Olson and Price meltability (R² = 0.88) models gave good prediction results.     

Effect of different stunning systems on meat quality of light lamb

The present study was proposed to compare the effect that different types of stunning (TS) had on the quality of refrigerated meat from light lambs of the Spanish Manchega breed at 24 h and 7 days post-mortem. Lambs were electrically stunned (ESL; n = 10), using CO₂ (GSL; n = 10) or slaughtered without previous stunning (USL; n = 10). Measurements on meat quality were carried out by evaluating pH, colour coordinates (L^*, a^*, b^*), water holding capacity (WHC), cooking loss (CL), shear force (SF) and drip loss (DL). At 24 h post-mortem, no significant differences were found in any of the variables studied. However, at 7 days post-mortem, meat quality was affected by the different TS: pH, CL and DL were lower (P < 0.001) in the USL group and GSL obtained the lowest a^* (redness) and b^* (yellowness) values (P < 0.01) than in the other groups. Ageing of meat affected SF in the ESL group (P < 0.01), although there were no significant differences due to treatments at any of the ageing times.     

The accuracy of predicting carcass composition of three different pig genetic lines by dual-energy X-ray absorptiometry

The accuracy of dual energy X-ray absorptiometry (DXA) for predicting carcass composition was evaluated by applying this technology to 95 half-carcasses of gilts from three genetic lines of widely varying composition. The DXA measurements of total weight, fat and lean masses and bone mineral content were used to predict the dissected lean, fat, bone and weight of the major primal cuts and the overall carcass. The study showed that it is not necessary to correct, or rather adjust, the prediction equations between genetic lines, given the small decrease in the prediction error (less than 1.43%). Except for the prediction for loin bone weight, the results showed that incorporating more than one of the predictive variables derived from DXA or from direct carcass measurements into the regression models was not justified in terms of prediction accuracy. The best predictions were those for half-carcass weight and primal cut weights (R² > 0.95). The prediction of dissected lean ranked second in accuracy (R² > 0.85), with the exception of the belly. The equations using DXA measurements provide good estimation (R² > 0.70) of the weight of fat in the carcass and primal cuts. However, DXA showed moderate accuracy (R² < 0.66) in predicting bone weight, although greater accuracy was attained for loin bone weight using a two-variable prediction model.     

Meat quality characteristics of springbok (Antidorcas marsupialis). 1: Physical meat attributes as influenced by age, gender and production region

Springbok is the most extensively cropped game species in South Africa. The effects of age (adult, sub-adult, lamb), gender and production region on the physical attributes (pH₂₄, cooking and drip loss, Warner Bratzler shear force and colour) were determined using samples of the M. longissimus dorsi (LD) muscles of 166 springbok. Stressed animals had a higher (P < 0.05) pH₂₄ (6.3 ± 0.07), as observed in the meat originating from the Caledon region. This meat had lower (P < 0.05) cooking loss (27.2 ± 0.62%) and drip loss (1.8 ± 0.08%) values in comparison to meat originating from the other regions. Inverse correlations were noted between pH₂₄ and drip loss (r = −0.26, P < 0.01) and cooking loss (r = −0.42, P < 0.001). Shear force values (kg/1.27 cm diameter) correlated positively (r = 0.25, P < 0.01) with pH₂₄. Age-related effects on tenderness were small in comparison with pH₂₄ effects. CIELab colorimetric values were typical of game meat and venison (L^* < 40, high a^* and low b^* values). It was noted that pH₂₄ correlated negatively (r = −0.51, P < 0.001) and positively (r = 0.33, P < 0.001) with the hue-angle and the chroma value of colour, respectively. Springbok originating from Caledon had a significantly (P < 0.05) higher a^* value, indicating meat to be more red with higher colour saturation.     

Predicting the composition of lamb carcases using alternative fat and muscle depth measures

Hot carcase weight (HCW) and the GR tissue depth (total tissue depth over the 12th rib, 110 mm from the midline) were measured on 312 crossbred (Poll Dorset × Merino) lambs with a mean live weight of 47 kg. At a position 30 mm distal to the lumbar-sacral junction, the depth (RMD) of the rump muscles was measured along with subcutaneous fat depth overlaying the m. gluteus medius (RFD). The subcutaneous fat depth over the m. longissimus (loin) at the 12th rib was measured (Fat C), as was the depth and width of the loin cross section (EMD and EMW, respectively). The right side of each carcase was scanned by a dual energy X-ray bone densitometer and fat and lean weight determined and the percentage of lean and fat content in the carcases calculated. HCW alone accounted for 26% of the variation in percentage lean content (LC), with GR measurement being the single best predictor of LC (R² = 48.1; r.s.d. = 1.89). RFD provided a less accurate and precise estimate of LC (R² = 37.3; r.s.d. = 2.07) than either GR or Fat C (R² = 46.5; r.s.d. = 1.92). Inclusion of HCW with either GR or Fat C provided a small improvement in the amount of variation in LC which could be explained. The measurement of breeding animals by ultrasound includes the terms weight, Fat C and EMD. These predictors when combined explained 49% of the variation in LC whereas by comparison measures of weight, RFD and RMD only explained 42% of the variation in LC. The current method of measuring fat depth over the 12th/13th rib is practical and measurement over the rump does not present an advantage based on the results of this study.     

Drip loss of case-ready meat and of premium cuts and their associations with earlier measured sample drip loss, meat quality and carcass traits in pigs

Drip loss of 374 samples taken from porcine M. longissimus dorsi and M. semimembranosus was measured by using the “bag method” (BM), EZ-DripLoss (EZ-DL) from premium cuts (PC) and in retail tray (case-ready meat; CRM). This provided a comparison between these methods and their relationships to other meat quality and carcass traits. Samples were prepared at 24 h post-mortem (pm) and were measured 24 and 48 h after preparation (at 48 and 72 h pm) using the BM and after 48 h (at 72 h pm) with the EZ-DL and PC. Drip loss of meat kept in retail trays was measured after 7 days (CRM₇) and daily within a week (CRM_1–7). Average drip loss was 1.80% and 3.10% using the BM after 24 and 48 h, respectively. EZ-DL and CRM₇ showed higher drip losses of 4.71% and 4.00%. Daily loss of CRM_1–7 showed a concavely shaped curve and increased from 1.57% to 5.64% after 7 days. High correlations were obtained between drip loss of CRM₇ and BM (r = 0.88) or the EZ-DL (r = 0.91). The development of drip loss in case-ready meat fitted by linear-quadratic regression (y = 0.439 + 1.245x − 0.072x²) showed that high drip loss measured earlier by bag and EZ-DripLoss methods was highly associated with a high intercept (r = 0.63–0.72), a high linear increase (r = 0.77–0.81), but larger decrease in increments (r = −0.82 to −0.86) during weekly stored meat in retail trays as supplied at consumer level. Because the positive linear regression coefficient was substantially higher than the negative quadratic regression coefficient, the development of drip loss is mainly dependent on the initial drip loss. Therefore, animals with high drip loss within 72 h post-mortem also showed undesirable high drip loss curves over the entire retail period. Relationships between drip loss and other meat quality traits were similar for BM, EZ-DL and CRM₇. Of these the correlation between pH₂₄ and drip loss was highest with r = −0.54, −0.49 and −0.47 for BM, EZ-DL and CRMH₇, respectively. Interestingly, a correlation of r = −0.35 between blood pH value and CRML₇ was obtained. Carcass traits such as loin, ham, shoulder, belly weight or loin eye area showed only marginal correlations to drip loss. In conclusion, EZ-DL was the most appropriate method to predict drip loss of case-ready meat in retail trays and its development during a 7 day storage period.     

Modeling transfer of Escherichia coli O157:H7 and Staphylococcus aureus during slicing of a cooked meat product

Cross contamination is one of the most important contributing factors in foodborne illnesses originating in household environments. The objective of this research was to determine the transfer coefficients between a contaminated domestic slicing machine and a cooked meat product, during slicing. The microorganisms tested were Staphylococcus aureus (Gram positive) and Escherichia coli O157:H7 (Gram negative). The results showed that both microorganisms were able to transfer to all slices examined (20 successively sliced) and at different inoculum levels on the blade (10⁸, 10⁶ and 10⁴ cfu/blade). The results also showed that the number of log cfu transferred per slice, during slicing, decreased logarithmically for both microorganisms at inoculum levels of 8 and 6 log cfu/blade. The type of microorganism significantly influenced transfer coefficients (p < 0.05) and there was an interaction between inoculum level and transfer coefficient for S. aureus (p < 0.05), but not E. coli O157:H7. Finally, to describe bacterial transfer during slicing, two models (log-linear and Weibull) were fitted to concentration on slice data for both microorganisms (at 6 and 8 log cfu/blade), obtaining a good fit to data (R²  0.73).     

Preliminary investigation of the application of Raman spectroscopy to the prediction of the sensory quality of beef silverside

The potential of Raman spectroscopy for the determination of meat quality attributes has been investigated using data from a set of 52 cooked beef samples, which were rated by trained taste panels. The Raman spectra, shear force and cooking loss were measured and PLS used to correlate the attributes with the Raman data. Good correlations and standard errors of prediction were found when the Raman data were used to predict the panels' rating of acceptability of texture (R²=0.71, Residual Mean Standard Error of Prediction (RMSEP)% of the mean (μ)=15%), degree of tenderness (R²=0.65, RMSEP% of μ=18%), degree of juiciness (R²=0.62, RMSEP% of μ=16%), and overall acceptability (R²=0.67, RMSEP% of μ=11%). In contrast, the mechanically determined shear force was poorly correlated with tenderness (R²=0.15). Tentative interpretation of the plots of the regression coefficients suggests that the -helix to β-sheet ratio of the proteins and the hydrophobicity of the myofibrillar environment are important factors contributing to the shear force, tenderness, texture and overall acceptability of the beef. In summary, this work demonstrates that Raman spectroscopy can be used to predict consumer-perceived beef quality. In part, this overall success is due to the fact that the Raman method predicts texture and tenderness, which are the predominant factors in determining overall acceptability in the Western world. Nonetheless, it is clear that Raman spectroscopy has considerable potential as a method for non-destructive and rapid determination of beef quality parameters.     

Thermal inactivation of the frozen thawed traditional meat starter culture, Pediococcus pentosaceus, in a meat model system

The equation, y_(t) = y₍₀₎e^kt, was fitted (R = 0.9281, 0.9220 and 0.9117, respectively) to thermal inactivation data (55, 60 and 65 °C) of the traditional meat starter culture Pediococcus pentosaceus (10⁷ cfu/ml) in a meat model system. The population reduction constant (‘k’) increased (about 2.5- and 3-fold) with an increase in the treatment temperature (from 55 to 60 °C and from 60 to 65 °C, respectively). The Q₁₀ (55–65 °C) for ‘k’ was 7.63. Thermal treatments of 19.1, 9.0 and 3.1 min (55, 60 and 65 °C, respectively) reduced the population of P. pentosaceus by 2.0 logs. The value of ‘k’ and the duration of the thermal treatment played an important role in the extent of the inactivation of the culture. The “zero inactivation” temperature (T₀) for P. pentosaceus was 49.9 °C. About 5 logs of the culture would be destroyed at 63 and 68 °C within about 15.5 and 6.5 min, respectively.     

Neural network prediction of physical property changes of dried carrot as a function of fractal dimension and moisture content

The relationship between microstructural and physical properties of dried foods is difficult to quantify. This study uses artificial neural network analysis (ANN) to predict shrinkage and rehydration of dried carrots, based on inputs of moisture content and normalized fractal dimension analysis (ΔD/D₀) of the cell wall structure. Several drying techniques were used including conventional hot air (HAD), low pressure superheated steam (LPSSD), and freeze drying (FD). Dried carrot sections were examined by light microscopy and the fractal dimension (D) determined using a box counting technique. Optimized ANN models were developed for HAD, LPSSD, HAD + LPSSD, and HAD + LPSSD + FD, based on 1–10 hidden layers and neurons per hidden layer. ANN models were then tested against an independent dataset. Measured values of shrinkage and rehydration were predicted with an R² > 0.95 in all cases.     

Calpains from thaw rigor muscle

Pre-rigor beef M. Longissimus lumborum and diaphragma were frozen at −70 °C and thawed at different temperatures and the activities of extracted calpains and the toughness of heated meat compared with those in chilled muscle.

Fresh muscle contained about 14 μg of μ-calpain/g and was unaffected by freezing, but was reduced after thawing. Rapid thawing at 30 °C for 20 min reduced the μ-calpain to 14%. When cooked from the frozen state, extensive shortening occurred and tender meat was obtained.

By storing at −3 °C for 1 day, thaw-shortening was prevented, but tougher meat obtained. The μ-calpain decreased to 70% whilst the m-calpain was unaffected. Toughness decreased after further storage at −3 °C, as did the μ-calpain. The latter changes were similar to those during development of rigor mortis and ageing of non-shortened meat stored at 4 °C. Variation in calpain activity, rather than in sarcomere length, are likely to be the cause of toughness variation in thaw rigor muscle.     

Effects of rigor temperature and electrical stimulation on venison quality

The effects of rigor temperature and electrical stimulation on venison quality were assessed using venison longissimus dorsi muscle. In the first trial, effect of rigor temperature (0, 15, 25, 30, 35 and 42 °C) and time post-mortem (at rigor, 3, 7 and 14 days) on drip and cooking losses, % expressible water (water holding capacity, WHC), sarcomere length, protein solubility, meat tenderness and colour were investigated. In the second trial, the effects of rigor temperature (15 and 35 °C), electric stimulation (stimulated or not stimulated) and time (at rigor, 3 and 6 weeks post-mortem) on tenderness and colour were further investigated. Results of the first trial showed no clearly established trends of the effect of rigor temperature and time on the cooking and drip losses and protein solubility except venison muscles that went into rigor at 42 °C tended to have higher drip loss and lower protein solubilities compared to muscles that went into rigor at the other temperatures. Venison water holding capacity (WHC) decreased with the increase in rigor temperature (P < 0.001) and venison became more tender with time post-mortem. Venison colour improved with increasing rigor temperature. During display, samples that went into rigor at 15, 25 and 35 °C had the lowest and those at 0 and 42 °C had the highest rate of change of redness (a^*) value with time. In the second trial, tenderness was improved by stimulation (P = 0.01). Redness (a^*) values were affected by rigor temperature (P < 0.01) and post-mortem time (P < 0.001) but not by electrical stimulation. It is concluded that venison tenderness can be improved via the manipulation of rigor temperature to obtain acceptable level of tenderness early post-mortem with less damaging effect on colour stability.     

Analysis of potential lard adulteration in chocolate and chocolate products using Fourier transform infrared spectroscopy

Fourier transform infrared (FTIR) spectroscopy, in combination with attenuated total reflectance (ATR) and partial least square (PLS) regression, was used to detect the presence of lard in chocolate formulation. The spectral bands associated with lard, cocoa butter and their blends (ranging from 0% to 15% of lard in cocoa butter) were recorded, interpreted and identified. A semi-quantitative approach is proposed to measure the percent of lard in blends on the basis of spectral data at the frequency region 4000–650 cm⁻¹, using the equation y = 0.9225x + 0.5539. The coefficient of determination (R²) was 0.9872 with a standard error (SE) of 1.305. In this paper, the potential of FTIR spectroscopy as a rapid analytical tool for the quantitative determination of adulterants especially lard, in chocolate, is demonstrated.     

Effect of transport time and ageing on aspects of beef quality

Forty-eight bulls were transported by road in groups of eight for approximately 30 min, 3 or 6 h in two replicates. After slaughter, steaks from the longissimus dorsi thoracis of all transported animals were analysed in terms of pH, water-holding capacity, myoglobin concentration, texture at 7 and 14 days of ageing (compression and Warner-Bratzler analyses) and colour of the same surface of meat cut at 24 h post-mortem and allowed to bloom for 24 h and 7 d (L^*a^*b^*, chroma and hue). The myofibrillar resistance of the meat from animals transported 30 min was significantly greater at 20% compression (P<0.05). The maximum load and a^*b^* and chroma were all lower for meat aged 14 days.     

Effects of age on composition and quality of muscle Longissimus thoracis of the Omani Arabian camel (Camelus dromedaries)

The aim of this study was to determine the effects of age on chemical composition and quality characteristics of the Arabian one-humped camel’s meat. Samples of longissimus thoracis (between the 10th and the 13th rib of the left side) were randomly collected from 21 Omani intact male camels of three different age groups: group 1 (1–3 years), group 2 (3–5 years) and group 3 (6–8 years). Samples were chilled (1–3 °C) for 48 h. Moisture, crude protein, fat and ash were determined on freeze dried ground muscle. Mineral contents were determined using an Inductively coupled plasma emission spectrometer (ICP). Meat quality including ultimate muscle pH, Warner–Bratzler shear force, sarcomere length, myofibrillar fragmentation index, expressed juice, cooking loss percent, and colour L^*, a^*, b^* were measured using standard methods.

The moisture, protein, fat and ash ranged from 64.4% to 76.7%; 18.6% to 25.0%, 1.1% to 10.5% and 1.0% to 1.4% on dry matter basis, respectively. The Ca, Mg, Na, K, P, Cad, Cr, Ni, Pb, Co, Mo, Be and V ranged from, 9.2 to 46.6, 24.7 to 57.3, 104.7 to 257.0, 471.4 to 1053.0, 249.9 to 584.0, 0.005 to 0.024, 0.020 to 0.410, 0.016 to 0.187, 0.010 to 0.299, 0.010 to 0.018, 0.050 to 0.470, 0.005 to 0.030 and 0.013 to 0.141 mg/100 g on dry matter basis, respectively. The percentage of protein decreased and that of fat increased with increasing camel age. The ultimate pH, shear force, sarcomere length, fragmentation index, expressed juice, cooking loss, lightness (L^*), redness (a^*) and yellowness (b^*) ranged from 5.46 to 6.64, 4.25 to 17.82, 0.96 to 2.50, 55.91 to 94.81,19.50 to 33.63, 13.18 to 29.88, 27.86 to 43.21, 10.46 to 22.81, and 4.63 to 10.11, respectively. Muscles of younger camels (group 1) had significantly (P < 0.05) lower shear force value, ultimate pH and higher sarcomere length, fragmentation index, expressed juice, cooking loss, and lightness color (L^*) by 48%, 3.4%, 43%, 25%, 28%, 14%, and 16% than those collected from older camels (group 3), respectively. Values of middle age camels (group 2) camels were in-between. This study confirmed that camel meat is healthy and nutritious as it contains low fat as well as being a good source of minerals. Age is an important factor in determining meat quality and composition.     

The relationship between post-mortem calcium concentration or pH and indicators of proteolysis in ovine muscle

The relationship between post-mortem calcium concentration in muscle and indicators of proteolysis was examined and compared with pH as a predictor of these changes. Muscle samples (m. longissimus et thoracis lumborum; LTL) were obtained from the left side of 24 lamb carcasses at pH 6.2 and 6.0 and then at 1 and 2 days post-mortem (n=96). Alternate carcasses were electrically stimulated (low voltage) within 15 min of death. The myofibrillar fragmentation index (MFI) was determined on samples taken at pH 6.2 and 1 and 2 days post-mortem (n=72). Protein solubility, pH and free calcium concentration were measured on all samples (n=96). The post-mortem degradation of myofibrillar proteins was followed using SDS electrophoresis on all samples (n=96) from which the relative amounts of the 30 kDa fragment and a protein designated M1 were quantified. Transformation of data values for each variable (MFI, protein solubility, 30 kDa fragment and protein M1) improved the normality of the residuals and increased the variance explained by either calcium concentration or pH. pH was a better predictor of MFI and the 30-kDa fragment than calcium concentration and this was reversed when protein solubility and the protein designated M1 was predicted. Of the variables protein solubility could be predicted with the greatest accuracy using calcium concentration (R²=0.64; R.S.D.=0.81) or pH (R²=0.60; R.S.D.=0.85). However overall calcium concentration was not superior to pH as a predictor of the changes in the different indicators of proteolysis examined in this study.     

Effect of high power ultrasound and ageing on the physical properties of bovine Semitendinosus and Longissimus muscles

Tenderness is an important meat quality parameters and the use of high power ultrasound to disrupt muscle structure may prove effective for reducing both myofibrillar and collagenous toughness. The experiment was carried out with Longissimus lumborum et thoracis and Semitendinosus muscles from 3 to 4 year old steers. Uncooked beef samples (60 × 40 × 20 mm) were treated with high power ultrasound (24 kHz, 12 W/cm²) for up to 240 s, and aged for up to 8.5 days before evaluation of pH, drip loss, cook losses Warner–Bratzler shear (WBS), compression hardness, and colour. Ultrasound treatment significantly reduced WBS force and hardness, but significantly increased pH. Ageing significantly reduced hardness and WBS force, but there was no significant interaction between ultrasound treatment and ageing time. Ultrasound treatment did not affect any of the colour parameters (L^*a^*b^*, chroma and hue) but the ageing time significantly increased the lightness, chroma and hue. There was no significant effect of ultrasound treatment on drip loss, but it did significantly reduce the cook and total loss. During ageing, cook loss and total losses significantly increased. The results suggest that high power ultrasound is capable of reducing objective texture measurements of beef without compromising the other quality parameters investigated.     

Sodium chloride levels in comminuted chicken muscle in relation to processing characteristics and Fresnel reflectance detected with a polarimetric probe

Increasing NaCl from 0 to 2% in comminuted samples of chicken breast meat increased the resistance to compression of cooked samples (r=0.98, p < 0.001), increased water-holding capacity (r=0.95, p < 0.001), decreased cooking losses (r=−0.97, p < 0.001) and decreased paleness (L^*) before and after cooking (r = −0.96 and r=−0.99, respectively, p < 0.001). The decrease in paleness caused by NaCl was associated with an increase in the relative amount of reflected light maintaining its original plane of polarisation, so that NaCl% was correlated with an extinction coefficient for polarised light, r=0.91, p < 0.001. A possible explanation is that myofibrillar proteins may have been dissolved at high levels of NaCl, thus reducing light scattering by myofibrils and allowing more Fresnel reflectance to be detected.     

The influence of early post-mortem enhancement and accelerated chilling on pork quality

To improve pork quality, the effectiveness of early post-mortem enhancement and accelerated chilling were investigated. The four treatments evaluated were: Enhancement with Accelerated Chilling (ENAC), Accelerated Chilling Only (ACO), Enhancement with Conventional Chilling (ENCC), and Conventional Chilling Only (CCO). ENAC had a higher (P < 0.05) pH than all other treatments. CCO resulted in the highest (lightest; P < 0.05) L^*, while ENAC had the lowest L^* value (darkest; P < 0.05). Subjective color and striping did not differ (P > 0.05) between ENAC and ENCC, although ENAC was numerically higher for both parameters. Sensory analysis for juiciness and tenderness were not different (P > 0.05) between ENAC and ENCC, but both were higher (P < 0.05) than ACO and CCO. Enhancement early post-mortem coupled with accelerated chilling may be used to improve instrumental color and pH over conventional processing methods.     

Aluminium contents in baked meats wrapped in aluminium foil

In this investigation, the effect of cooking treatments (60 min at 150 °C, 40 min at 200 °C, and 20 min at 250 °C) on aluminium contents of meats (beef, water buffalo, mutton, chicken and turkey) baked in aluminium foil were evaluated. Cooking increased the aluminium concentration of both the white and red meats. The increase was 89–378% in red meats and 76–215% in poultry. The least increase (76–115%) was observed in the samples baked for 60 min at 150 °C, while the highest increase (153–378%) was in samples baked for 20 min at 250 °C. It was determined that the fat content of meat in addition to the cooking process affected the migration of aluminium (r² = 0.83; P < 0.01). It was also found that raw chicken and turkey breast meat contained higher amounts of aluminium than the raw chicken and turkey leg meat, respectively. Regarding the suggested provisional tolerable daily intake of 1 mg Al/kg body weight per day of the FAO/WHO Expert Committee on Food Additives, there are no evident risks to the health of the consumer from using aluminium foil to cook meats. However, eating meals prepared in aluminium foil may carry a risk to the health by adding to other aluminium sources.