- Email: [email protected]
Freezing rate and frozen storage effects on the ultrastructure of samples of pork
Meat Science 53 (1999) 159±168
www.elsevier.com/locate/meatsci
Freezing rate and frozen storage eects on the ultrastructure of samples of pork T.M. Ngapo*, I.H. Babare, J. Reynolds, R.F. Mawson Food Science Australia, Sneydes Road, Werribee, Victoria 3030, Australia Received 14 August 1998; received in revised form 7 April 1999; accepted 13 April 1999
Abstract Cryo-scanning electron microscopy was used to study the ultrastructure of small samples (approximately 6 g) of pork. Combinations of six freezing rates, two storage times and three thawing rates were used. Cavities created after sublimation of the ice crystals were quantitatively analysed using an image analysis software package. The cross-sectional areas of cavities of meat samples in the frozen state were approximately ten times the areas of the cavities of the fresh and thawed samples. The large cavities in the frozen state grossly distorted the muscle cell structures. Upon thawing, the meat structure had almost completely recovered. No signi®cant freezing rate eects were observed, however, trends were evident. Signi®cant storage time eects were observed. In the frozen state, at the 90th percentile level, the hole area fraction was greater in stored samples for intermediate cavity areas. In thawed samples, hole area fractions of stored samples were greater than in samples without storage. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction The freezing of meat has been widely studied as a means to lowering the amount of proteinacious exudate (drip) lost on thawing. The loss of ¯uid generally reduces the eating quality, binding ability and the weight of meat, all factors contributing to its monetary value. The volume of drip produced on thawing has been extensively related to the rate of freezing, which in turn has been related to the size and location of ice crystals in frozen meat. The dierent quality obtained in frozen foods, in particular meat and muscle tissue, has been the subject of many reviews (such as Cassens, 1971; Fennema, Powrie, & Marth, 1973; Jeremiah, 1996) and different sizes of the crystals formed are generally considered to be one of the factors responsible for changes in the quality obtained. While there is a general agreement about the mechanism of freezing and location of ice crystals in frozen meat, there is a lot of con¯icting literature about the eects of freezing and thawing on meat. In contrast to the extensive amount of literature about the eects of freezing rate on changes in * Corresponding author at: Station de Recherches sur la Viande, INRA, Theix, 63122 St GeneÁs-Champanelle, France. Tel.: +33-473624-167; fax: +33-473-624-089. E-mail address: [email protected] (T.M. Ngapo)
physical, chemical and sensory properties of meat, there are few reports about the eects of thawing rates on meat characteristics and, in particular, the ultrastructure of thawed meat. Many of the reported histological studies examining the ultrastructure of fresh or thawed meat have used light and transmission electron microscopy. These have required sample preparation methods of freeze substitution (including Bevilacqua & Zaritzky, 1980) or ®xation (such as Grujic, Petrovic, Pikula, & Amidzic, 1993). An example of an investigation of the eects of frozen storage on the ultrastructure of muscle is that of Carroll, Cavanaaugh, and Rorer (1981) who used scanning electron microscopy (SEM) to observe the ultrastructure of samples frozen at ÿ18 C and stored for 5 days, 4, 12 or 26 weeks. The images obtained showed good preservation of structure with little apparent change from the untreated sample. These samples were cut (18 to 20 mm thick) perpendicular to the longitudinal axis, wrapped in ``freezer paper'' and frozen in a stationary air freezer at ÿ18 C (about 4 h to freeze). Thawing occurred at room temperature and took 3.5 h for the centre of the sample to reach 4 C. Tissue ®xation was used in the sample preparation for SEM. The cause of the ice structure produced in a meat sample during freezing is the process of nucleation that
0309-1740/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(99)00051-0
160
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
takes place at its border. Investigation of the conditions that regulate the number of crystals in this zone will aid in the prediction of the ice-®bre con®guration in frozen tissues (Menegalli & Calvelo, 1979). The aim of the present study was to investigate the eects of combinations of six freezing rates, three thawing rates and two frozen storage times on the ultrastructure of small meat samples using a cryo-SEM technique employed by Payne, Sandford, Harris, and Young (1994). This method allowed examination of the cavities created by sublimation of ice crystals without requiring chemical ®xation of the meat sample, and thus reducing possible artefacts. A small sample size was used to simulate the refrigerated border in a larger piece of meat. Preliminary results of these studies were presented in earlier papers (Ngapo, Babare, Murphy, Reynolds, & Mawson, 1997a,b). 2. Materials and methods 2.1. Materials Porcine M. biceps femoris were obtained from carcasses that had been stored for 24 h at 4 C after slaughter. Equal numbers of cross bred (Landrace, Large White, Duroc) entire male and female pigs, between 22 and 24 weeks of age, were slaughtered at a local abattoir. Meat was stored at 4 C for up to a further 48 h before use. Only muscles with an ultimate pH in the range 5.4 to 6.0 were used. 2.2. Measurement of pH Meat (2.0 g) was ground in a mortar and pestle with deionised water (10 ml). A standard pH meter equipped with a glass electrode was used. Measurements of pH were calculated from the average of four replicates. 2.3. Sample preparation Ð freezing Sample preparation was conducted at 4 C. Cylindrical samples (approximately 70 mm length10 mm diameter) were cut from the centre of the muscle with their axis parallel to ®bre direction. Care was taken to avoid obvious pieces of fat and connective tissue. The samples were placed in aluminium holders (80 mm length10 mm i.d., 2020 mm outer cross-sectional area, in two longitudinal pieces) and plugged with expanded polyurethane foam (5 mm length) at either end. The holders, ®rmly held together with rubber rings, were placed in latex sheaths and secured with cable ties. The samples were placed vertically in a B. Braun Frigomix S cooling bath with a B. Braun Thermomix UB attachment immediately prior to freezing. Ethylene glycol (90% w/w) was used as a coolant.
Freezing velocities were de®ned as the characteristic freezing times (tcf) to traverse the temperature range from ÿ1 to ÿ7 C. Times (tcf) of 12, 30, 60, 120, 240 and >900 min were used. The temperature of the cooling bath was changed manually every 1 C for samples with tcf of 12 min and every 0.5 C for samples with tcf from 30 to 240 min. The temperature of the cooling bath was changed every 0.2 C for samples with tcf of >900 min to ÿ3.6 C where the temperature was held for approximately 15 h, then changed every 0.2 C to ÿ7 C. After cooling at the speci®ed rates, the bath temperature was reduced from ÿ7 to ÿ20 C in 15 min. After 1 h at ÿ20 C, the samples were transferred to a freezer at ÿ18 C, removed from the holders, and wrapped in cryovac barrier bags. Four samples were frozen at each tcf. Two samples were examined by electron microscopy within 1 week of freezing; two samples were stored for 4 weeks at ÿ18 C and then examined by electron microscopy within 1 week. Each freezing rate was investigated using one muscle from one animal and the freezing process was replicated twice (i.e. two muscles, each from dierent animals, were frozen at the same rate but on dierent occasions). The experiment was accordingly a split-plot in a completely randomised design Ð the ``main plots'' were muscles, and each muscle received a freezing rate; the ``splitplots'' were pairs of samples from a muscle, and each pair received a storage time. 2.4. Sample preparation Ð thawing Sample preparation was conducted at either 4 or ÿ18 C depending on the state of the sample; fresh and thawed samples, or frozen samples, respectively. After freezing, by the described method, samples were either held in the cooling bath at ÿ20 C for up to 15 h (described as ``no storage'') or stored at ÿ18 C for 4 weeks. The stored samples were held at ÿ20 C for 1 h, then transferred to a freezer at ÿ18 C, removed from the holders, and wrapped in cryovac barrier bags. Immediately prior to thawing, the samples were replaced in the aluminium holders at ÿ18 C and transferred to the cooling bath at ÿ20 C. Thawing velocities were de®ned as the characteristic thawing times (tct) to traverse the temperature range from ÿ7 to ÿ1 C. Thawing times (tct) of 12, 60 and 180 min were used. The cooling bath temperature was raised from ÿ20 to ÿ7 C in 15 min. The bath temperature was then changed manually every 1 C for samples with tct of 12 min and every 0.5 C for samples with tct of 30 and 180 min to 2 C. Samples were held at this temperature for 10 min prior to use. Four samples were frozen and thawed at each freeze rate/thaw rate combination. Two were thawed with no storage and two were stored prior to thawing. Two further samples were analysed fresh (unfrozen). The fresh
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
or thawed samples were each cut into six discs of approximately 10 mm length and the two end discs were discarded. The discs were plunged into liquid N2 until frozen. After removal from the liquid N2, the samples were stored in sealed plastic vials for up to 7 days at ÿ70 C, until 1 h prior to SEM, when the discs were transferred to a freezer at ÿ18 C. Each of the 18 freezing rate by thawing rate combinations was investigated using one muscle from one animal. These treatment combinations were not able to be replicated as only 18 muscles were used. 2.5. Monitoring of freezing and thawing rates The samples in the present paper were frozen with the samples used for drip loss analysis in the accompanying paper (Ngapo, Babare, Reynolds, & Mawson, 1999). Four samples used for drip loss analysis, had copper constantan thermocouples inserted into their centres, along the longitudinal axis of the cylinder. Two further thermocouples were placed in the coolant and one at room temperature. The thermocouples were connected as dierential inputs to a PCI-20303T Termination Panel which was part of an Intelligent Instrumentation Visual Designer PCI-20,000 system used to continuously monitor the freezing rate. 2.6. Scanning electron microscopy (SEM) The SEM was conducted using a Cambridge Instruments Stereoscan 90 scanning electron microscope ®tted with an Oxford Instruments CT-1000A cryostage. At least four images were captured from each stub using an Image Slave Software Package. Sample preparation was conducted at ÿ18 C. Cores (3 mm diameter) were removed from the meat samples parallel to ®bre direction and inserted into a hole (3 mm diameter) drilled in the centre of a cryomicroscope stub. The stubs were then held in a polystyrene cup over liquid N2 for up to 2 h before plunging into liquid N2. The meat sample was freeze fractured with a scalpel blade perpendicular to the ®bre direction, mounted on the cryostage and heated to ÿ60 C until the ice had sublimed. The tissue surface was examined with an accelerating voltage of 2.5 kV and at a constant working distance of 7 mm. 2.7. Data translation Quantitative data were obtained from the images using a Data Translation Global Lab Image Software Package. Two stubs of each sample were used for analysis. The numbers of crystals in the frozen meat were estimated by image analysis of the cavities observed in the meat after sublimation. At least 150 cavities were measured from each image and a minimum of three images were analysed from each stub. The cavities were grouped into categories
161
based on cross-sectional areas. The total area of meat viewed in each image was referred to as the region of interest (ROI). The fraction of the ROI occupied by the cavities of a category size was calculated and referred to as the hole area fraction for that category size. To categorise the cavities using image analysis, the closest approximation to zero that could be obtained, at the magni®cations of the images, was 0.2210ÿ3 mm2 for the frozen samples and 0.1210ÿ4 mm2 for the thawed samples. 2.8. Statistical analysis The 10th and 90th percentiles, lower and upper quartiles and medians were calculated from the hole area fractions for each freezing rate and storage time combination (in the case of the frozen samples) and for each freezing rate, thawing rate and storage time combination (in the case of the fresh and thawed samples). Percentiles of the hole area fractions were de®ned as the hole area fractions for which a percentage of the total cavity area was accounted for by cavities of that size or smaller. Separate analyses based on cavity size categories were undertaken on the frozen samples only. The analysis of variance (ANOVA) routine in the Genstat 5 statistical package (Genstat 5 Committee, 1993) was used to analyse the data. A conventional split-plot analysis of variance was used to analyse the data from the frozen samples. The analysis of the data from the fresh and thawed samples was facilitated by assuming that three-way (and higher order) interactions of the treatment eects were negligible and their associated observed mean squares represented residual or error variation. Consequently, the eect of storage and the two-way interactions of storage with freezing rate and thawing rate were able to be investigated, but not the main eects of freezing rate or thawing rate, and their interaction. For both sets of data, transformations to overcome heterogeneity of variance were explored. 3. Results 3.1. Freezing and thawing curves The temperature at the centre of selected samples used for drip loss analysis (Ngapo et al., 1999) was monitored and displayed continuously with the temperature of the cooling bath. No lag in sample temperatures was observed compared to the bath temperature. 3.2. Eect of freezing rate on cavity size of frozen samples The cross-sectional areas of the cavities of samples in the frozen state were approximately 10 times the areas of the cavities of the fresh samples and the samples after
162
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
thawing with and without storage. Figs. 1a±d illustrate the relative sizes of the cavities in fresh, frozen and thawed meat samples. The gross distortion of the meat structure in the frozen state compared to fresh and thawed samples is evident in these images. The percentiles calculated from hole area fractions are presented in Table 1. There was some evidence of heterogeneity of variance in the analyses of the 10th and 90th percentiles and a log-transformation corrected this problem, but as the conclusions from these analyses were essentially the same as the analyses on the original (untransformed data) all the results in Table 1 are based on analyses of the untransformed data. No signi®cant dierences were observed in the 10th percentiles, lower and upper quartiles and median values comparing samples with no storage and after 4 weeks of storage. A signi®cant dierence (p=0.034) was observed in the 90th percentiles. Without storage, 90% of the hole area fraction was occupied by cavities of cross-sectional area below 10.6710ÿ3 mm2. After 4 weeks storage, the cavities occupying 90% of the hole area fraction had cross-sectional areas of 8.9810ÿ3 mm2, and smaller (l.s.d.=1.4910ÿ3 mm2).
Figs. 2 and 3 illustrate the spread of the mean hole area fractions for each grouping of cavity size at each of the six freezing rates studied with and without storage, respectively. Note that the cavity size categories vary in area ranges: approximately 0.2210ÿ3 mm2 for the ®rst four categories up to a cavity size of 1.1510ÿ3 mm2 approximately 0.5710ÿ3 mm2 for the next six categories to 4.6010ÿ3 mm2 and 1.1510ÿ3 mm2 for the last eight categories to 13.7910ÿ3 mm2. Cavities of Table 1 Hole area fraction mean percentiles of frozen samples Percentile
Hole fraction area Hole fraction area l.s.d. (p=0.05) (10ÿ3 mm2) (10ÿ3 mm2) (10ÿ3 mm2) no storage 4 week storage
10th 0.67 25th (lower 1.61 quartile) 50th (median) 3.60 75th (upper 6.89 quartile) 90tha 10.67 a
0.73 1.59
0.08 0.22
3.32 6.04
0.57 1.10
8.98
1.49
Percentile values are signi®cantly dierent.
Fig. 1. A typical example of the eects of freezing and thawing on the ultrastructure of a small sample of pork; (a) fresh, (b) tcf of 60 min, not thawed, (c) tcf of 60 min, stored up to 15 h, tcf of 12 min, (d) tcf of 60 min, stored 4 weeks, tct of 12 min. Scale bar=200 m.
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
163
Fig. 2. The mean hole area fractions for each grouping of cavity size at each of the six freezing rates studied without storage.
Fig. 3. The mean hole area fractions for each grouping of cavity size at each of the six freezing rates studied after 4 weeks storage.
larger cross-sectional areas occurred infrequently and a sensible average could not be estimated. A maximum cavity size of 13.7910ÿ3 mm2 was therefore selected to illustrate the spread. Separate analyses for each cavity size category showed that the hole area fraction was signi®cantly greater in the samples stored for 4 weeks for the categories from 0.93 to 3.4510ÿ3 mm2 (p<0.05) than for the samples that had not been stored. The hole area fractions calculated from cavities in the area categories from 0.22 to 0.9310ÿ3 mm2 and from 3.45 to 13.7910ÿ3 mm2 were not signi®cantly dierent after 4 weeks storage. These dierences are apparent, but not signi®cant, in the quartiles and medians reported in Table 1 Ð the interquartile range is narrower (3.87 after 4 weeks storage compared to 4.59 with no storage), but not signi®cantly so (p=0.077), and has
shifted in the direction of smaller cavity areas after 4 weeks storage. Figs. 4 and 5 illustrate the hole area fractions for the six freezing rates with and without storage, respectively. Note that the cavity size categories vary in area ranges and are the same as those used to illustrate the spread of the cavity areas in Figs. 2 and 3. The pro®les suggest that the hole area fractions for samples with tcf of 12 and >900 min are approximately the same before and after 4 weeks storage. At tcf of 30 and 60 min the hole area fractions appear to be consistently greater after 4 weeks storage. At tcf of 120 and 240 min the hole area fraction is greater in the samples without storage for smaller area categories, but becomes similar as the cross sectional area increases. For larger cross-sectional areas the samples with tcf of 120 min tend to a greater hole area fraction after 4 weeks storage.
164
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
Fig. 4. Mean hole areas as fractions of the ROI at characteristic freezing times for samples with no storage. The cavity size categories are 10ÿ3 mm2.
Fig. 5. Mean hole areas as fractions of the ROI at characteristic freezing times for samples after 4 weeks storage. The cavity size categories are 10ÿ3 mm2.
Trends in the patterns of hole area fractions with freezing rates are apparent (Figs. 4 and 5). A minimum is suggested at a tcf of 30 min, and maxima at 120 min. In the larger size categories, the minimum at 30 min disappears, and a minimum at 240 min becomes apparent. After 4 weeks storage, maxima at tcf of 120 and >900 min and a minimum at 240 min are more strongly suggested.
larger in the stored samples, particularly the long narrow cavities around the edge of the ®bre bundles (Fig. 1c and d). Storage time had signi®cant eects on the cavity areas of the thawed samples. Percentiles based on hole area fractions are presented in Table 2. While there was some evidence of heterogeneity of variance in the medians, upper quartiles and 90th percentiles, analyses of transformed data gave the same signi®cance test results as the untransformed data and so all means in Table 2 are based on analyses of the untransformed data. Signi®cant dierences were observed in the 10th percentiles (p=0.003) and lower quartiles (p=0.017) comparing cavity areas of frozen samples that had been stored for 4 weeks prior to thawing with cavity areas of samples frozen then thawed without storage. Only the frozen samples could be compared statistically, that is, the
3.3. Eect of thawing rate on cavity size of frozen samples It was observed that after thawing the ultrastructure of the meat sample almost wholly recovered from the grossly distorted structure observed in the frozen state (Figs. 1a±d). It was also noted that where dierences in the cavity areas were evident when comparing thawed samples with and without storage, the cavities appeared
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
cavity areas of the frozen/thawed samples, with or without storage, could not be compared to those of the fresh samples. A freezing rate by storage interaction was observed for the 10th percentile (p=0.039). The storage dierences were signi®cant at tcf of 12, 60 and >900 min (Table 3). Only the means of samples that had been frozen at the same rate could be compared statistically. 4. Discussion The temperature at the centre of the meat did not lag behind the temperature of the coolant and therefore there was no evidence of a temperature gradient in the meat sample during freezing. Menegalli and Calvelo (1979) suggested that the regulation of the size of the crystals by the speed of cooling is limited to samples with no temperature gradients (small pieces of meat) or to the peripheral region of large pieces of meat. Bevilacqua, Zaritzky, and Calvelo (1979) explained that when meat comes into contact with the refrigerating system a substantial supercooling is reached which generates a number of nuclei proportional to it. In the interphase of the crystals formed, the equilibrium temperature is soon reached due to a release of the latent heat of crystallisation. As crystal growth elevates the temperature, no further nucleation occurs. During this nucleation phase, crystal growth and nucleation occur simultaneously. A range of crystal sizes would therefore be expected at the termination of this phase. The continuation of crystal growth beyond this point suggests that there should be few of the initial small crystals (nuclei) remaining upon completion of the freezing regime, and that the time to reach the ®nal freezing temperature should in¯uence the sizes of crystals. Comparison of the meat samples frozen with tcf ranging from 12 to >900 min with meat samples frozen in liquid N2 shows the cavity cross-sectional areas were about 10-fold greater using the slower freezing rates; freezing in liquid N2 resulted in greater than 90% of cavity cross-sectional areas being less than 6.0010ÿ4
165
mm2 compared to less than 75% of cavity areas smaller than 6.0010ÿ3 mm2 and less than 10% smaller than 6.0010ÿ4 mm2 at the slower freezing rates. The large cavity areas in the samples frozen at tcf of 12 to >900 min created a gross distortion of the muscle cell structures compared to the structure observed in the samples frozen in liquid N2. To ensure that this was not an artefact created by the sample preparation for SEM, sublimation was observed as it occurred. No evidence of distortion as a consequence of the preparative procedure was found. The nuclei formed in the samples frozen in liquid N2 would have had very short growth phases, if any, and it is therefore reasonable to assume that these areas indicate values only slightly larger than that of the nuclei. The eective absence of these smaller cavities in the slower frozen samples is an indication that crystal growth has occurred. While the rate dierences between the cryo-SEM preparative freezing in liquid N2 and the experimental freezing (tcf from 12 to >900 min) in this study results in large dierences in cavity areas, it is suggested that the curves of nucleation rate and growth velocities versus supercooling for the experimental freezing rates are not so dierent. Therefore, more subtle dierences, if any, might be expected as a result of the smaller dierences in sizes of crystals formed at the dierent experimental freezing rates studied. The selection of freezing rates in the present study was based on the categories of ice crystals formed as suggested in the papers by Love (1958a,b) and Love and Haraldsson (1961). These studies were part of a comprehensive series of investigations examining the eects of freezing rate on cell disruption in cod ®llets. In the present study, a distinction could not be made between intra- and extracellular locations of ice crystals for comparison to this published work. However, while no signi®cant eects of freezing rate on cavity size were found, trends in the patterns of hole area fractions with freezing rates were apparent (Fig. 4) and comparisons could be made to this earlier work. At the very small cavity areas (less than 0.5010ÿ3 mm2) the trend was similar to that suggested by these workers. There
Table 2 Hole area fraction mean percentiles of fresh and thawed samples Percentile
Hole fraction area (10ÿ4 mm2) fresh sample
Hole fraction area (10ÿ4 mm2) no storage
Hole fraction area (10ÿ4 mm2) 4 wk storage
l.s.d. (p=0.05) (10ÿ4 mm2)
10th 25th (lower quartile) 50th (median) 75th (upper quartile) 90th
0.164 0.25 0.65 2.16 5.45
0.166a 0.26b 0.87 2.75 5.94
0.175a 0.29b 0.87 3.19 6.74
0.005 0.03 0.44 0.97 1.59
a b
Percentile values are signi®cantly dierent. Percentile values are signi®cantly dierent.
166
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
Table 3 Means of cavity size at dierent freezing rates (10th percentile level) Freeze
tcf (min)
Mean cavity size (10ÿ4 mm2) no storage
Mean cavity size (10ÿ4 mm2) 4 week storage
Numbers of samples analysed
No Yes Yes Yes Yes Yes Yes
0 (fresh) 12 30 60 120 240 >900
0.164 0.169a 0.167 0.169a 0.163 0.169 0.155a
0.193 0.158 0.184 0.168 0.174 0.171
36 12 12 12 12 12 12
a Means are signi®cantly dierent at the same freezing rates (l.s.d.=0.01310ÿ4 mm2 for comparison of samples at the same freezing rate).
appeared to be fewer of these small cavities at tcf of 30 min than at tcf of 12 min and a maximum at tcf of 120 min. Furthermore, there appeared to be more cavities up to about 2.5010ÿ3 mm2 at tcf of 120 min than at the other freezing rates. In contrast to the work of Love (1958a,b) and Love and Haraldsson (1961) there appeared to be large areas occupied by both very small cavities (less than 0.5010ÿ3 mm2) at a tcf of 240 min and cavities up to about 2.5010ÿ3 mm2 at tcf of >900 min. A more extensive study is required to expand these studies and for veri®cation of trends. Menegalli and Calvelo's (1979) theory of dendritic ice formation explains that the dierent zones of crystal sizes and locations observed in large pieces of meat where a temperature gradient exists, are a consequence of the sample size which leads to dierent thermal histories according to the zone considered. This could explain the dierences in results in the present work and the work of Love (1958a,b) and Love and Haraldsson (1961), who used a much larger sample size than the cylinders of meat used here. In larger samples of tissue, except for a small zone which corresponds to the refrigerated border of the sample (in which nucleation and thermal dendritic growth exist) nuclei grow through cells or cellular dendrites, according to whether the heat extraction is low or high, respectively. After 4 weeks frozen storage, a signi®cant dierence (p=0.034) was observed in the 90th percentiles of hole area fractions. With no storage, 90% of the hole area fraction was 10.6610ÿ3 mm2 and less. After four weeks storage, the cavities accounting for 90% of the hole area fraction were 8.9810ÿ3 mm2 and smaller. The lack of signi®cant dierences in 10, 25, 50 and 75 percentiles of hole area fractions with and without storage indicate that the relative proportions of cavities with cross-sectional areas up to about 6.510ÿ3 mm2 are similar regardless of the application of storage. However, the signi®cant dierence in the 90 percentile values suggests that in the samples with no storage there are proportionately fewer cavities within the area range from about
6.5 to 9.010ÿ3 mm2 and/or that the number of larger cavities is greater (for example, cavity areas >1310ÿ3 mm2) compared to the samples with storage. The latter suggestion requires that some of the largest cavities formed upon freezing become smaller with storage. Separate analyses for each cavity size category showed that the cavity area for each category as a fraction of the ROI was signi®cantly greater in the samples stored for 4 weeks for the categories from 0.93 to 3.4510ÿ3 mm2 (p<0.05) than for the samples that had not been stored. The hole area fractions calculated from cavities in the area categories from 0.22 to 0.9310ÿ3 mm2 and from 3.45 to 13.7910ÿ3 mm2 were not signi®cantly dierent after 4 weeks storage. The increase in this intermediate fraction of cavities could narrow the interquartile range. The cavities with areas >13.810ÿ3 mm2 were not included in these separate analyses. The impression gained is that storage for 4 weeks had narrowed the range of cavity areas and reduced the percentage of large cavities in samples left in the frozen state. Trends in the patterns of hole area fractions as a consequence of storage and freezing rate were observed (Figs. 4 and 5). The hole area fractions for samples with tcf of 12 and >900 min were approximately the same before and after 4 weeks storage. At tcf of 30 and 60 min the hole area fractions appeared to be consistently greater after 4 weeks storage. At tcf of 120 and 240 min the hole area fraction was greater in the samples without storage for smaller area categories (<0.510ÿ3 mm2). For larger cross-sectional areas (>2.510ÿ3 mm2) the samples with tcf of 120 min tend to a greater hole area fraction after 4 weeks storage. An increase in crystal sizes during storage can be explained by the process of recrystallisation. During storage, preferential growth in the solid state of large crystals occurs at the expense of smaller ones (Meryman, 1956). The data in the current study presents not only increases in crystal sizes, but a complex system of changes in crystal sizes at dierent rates of freezing, which is not obviously explained by recrystallisation. After thawing, the meat structure appeared to have almost completely recovered from the gross distortion observed in the frozen state. Wirth (1979) suggested that by using a small sample, a slow thawing rate results in reduced drip loss due to the partial absorption of exudate by the muscular ®brils. A consequence of this theory is that the ultrastructure of the sample partially recovers from the grossly distorted structure observed in the frozen state. Micrographs of typical images in the fresh, frozen and thawed states are given in Fig. 1a±c. The distortion of the meat ultrastructure in the frozen state and recovery from this state after thawing are evident in these images. Dierences in the cavity areas were evident in some samples when comparing thawed samples with and
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
without storage. The cavities appeared larger in the stored samples, particularly the long narrow cavities around the edge of the ®bre bundles (compare Fig. 1c and d). Signi®cant dierences were observed in the 10th percentiles (p=0.003) and lower quartiles (p=0.017) comparing hole area fractions of samples that had been frozen, stored for 4 weeks and then thawed, with hole area fractions of samples frozen then thawed with no storage. These percentile values were greater in the samples that had been stored (Table 2), indicating that there were proportionately fewer small cavities (<0.310ÿ4 mm2) and/or proportionately more intermediate cavities (about 0.3 to 0.910ÿ4 mm2) after thawing stored samples. Interestingly, the separate analyses on the cavity areas in the frozen state showed that the hole area fraction for each category was signi®cantly greater in the samples stored for 4 weeks for the categories from 0.93 to 3.4510ÿ3 mm2 (also an intermediate range of cavity areas) than for the samples that had not been stored. In the thawed samples, a freezing rate by storage interaction was also observed at the 10th percentile (p=0.039). The storage dierences were evident at tcf of 12, 60 and >900 min (Table 3) where the mean cavity areas were signi®cantly greater in the samples that had been stored. It is also of interest that signi®cant dierences in hole area fractions were observed after 4 weeks storage in both the frozen and thawed states, but that the dierences were only signi®cant at the 90% level in the frozen state and at the 10 and 25% levels in the thawed states. During thawing in the present study, a lag between the temperature at the centre of the meat and the coolant was not evident as a result of the small sample size used. There was, therefore, no evidence of a temperature gradient in the sample during the thawing process suggesting that recrystallisation due to a melting front was unlikely. However, recrystallisation while holding the sample at temperatures below the freezing point would have occurred. Earlier studies on drip loss in this system (Ngapo et al., 1999) showed that, at freezing rates with tcf of 12± 120 min, drip loss obtained was not signi®cantly dierent from that obtained from the fresh samples. At the slower freezing rates studied (tcf of 240 and >900 min) drip loss was signi®cantly dierent from drip loss obtained from the fresh samples. After 4 weeks of frozen storage, the drip loss was signi®cantly dierent at tcf of 12±120 min, compared to the samples without storage. However, there were no signi®cant dierences in drip loss of the stored samples comparing the freezing rates. Furthermore, the drip loss from stored samples was not signi®cantly dierent from that obtained at the two slowest freezing rates with no storage. This investigation illustrates that the combined eects of freezing rates, frozen storage and thawing rates on crystal formation and the ultrastructure of a small meat
167
sample present a complex system. Comparison of these results with those obtained from drip measurements adds to this complexity and any correlation of drip loss with crystal formation or meat ultrastructure is not apparent using a small meat sample. 5. Conclusions Large cavities observed in the frozen state caused gross distortion of the muscle cell structures, but after thawing, the meat structure appeared to have almost completely recovered. The combined eects of freezing rates, frozen storage and thawing rates on crystal formation and the ultrastructure of a small meat sample present a complex system. Comparison of ultrastructural data with drip measurements adds to this complexity and any correlation of drip loss with crystal formation or meat ultrastructure was not apparent using small meat samples. As the ice structure produced in a sample upon freezing depends upon the process of nucleation that takes place at the refrigerated border, study of the conditions that regulate the number of crystals in that zone is of great importance in predicting the ice ®bre con®guration in frozen tissues. The use of the cryo-SEM method on a larger sample size that has a thermal gradient across it during freezing, and thereby simulates industrial conditions of meat freezing, would provide valuable information to complement this work. Furthermore, observations were only made in cross-sections perpendicular to the ®bre direction and actual breakage of cells might be better seen using transmission electron microscopy. Acknowledgements The authors gratefully acknowledge the ®nancial support provided by the Australian Pig Research and Development Corporation and the Agriculture and Food Initiative of the Victorian Department of Natural Resources and Environment. References Bevilacqua, A. E., & Zaritzky, N. E. (1980). Ice morphology in frozen meat. Journal of Food Technology, 15, 589±597. Bevilacqua, A, Zaritzky, N. E., & Calvelo, A. (1979). Histological measurements of ice in frozen beef. Journal of Food Technology, 14, 237±251. Carroll, R. J., Cavanaugh, J. R., & Rorer, F. P. (1981). Eects of frozen storage on the ultrastructure of bovine muscle. Journal of Food Science, 46, 1091±1094, 1102. Cassens, R. G. (1971). Microscopic structure of animal tissue. In J. F. Price, & B. S. Schweigert (Eds), The Science of Meat and Meat Products (2nd ed.). San Fransisco: W. H. Freeman and Co. Fennema, O. R., Powrie, W. D., & Marth, E. H. (1973). Low temperature preservation of foods and living matter (pp. 150±239). New York: Marcel Dekker.
168
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
Genstat 5 Committee (1993). Genstat 5 release 3 reference manual. Oxford: Oxford University Press. Grujic, R., Petrovic, L., Pikula, B., & Amidzic, L. (1993). De®nition of the optimum freezing rate Ð I. Investigation of structure and ultrastructure of beef M. longissimus dorsi frozen at dierent freezing rates. Meat Science, 33, 319±331. Jeremiah, L. E. (1996). Freezing eects on food quality. New York: Marcel Dekker, Inc. Love, R. M. (1958a). The expressible ¯uid of ®sh ®llets. VIII. Cell damage in slow freezing. Journal of the Science of Food and Agriculture, 9, 257±262. Love, R. M. (1958b). The expressible ¯uid of ®sh ®llets. IX. Other types of cell damage caused by freezing. Journal of the Science of Food and Agriculture, 9, 262±268. Love, R. M., & Haraldsson, S. B. (1961). The expressible ¯uid of ®sh ®llets. XI.-Ice crystal formation and cell damage in cod muscle frozen before rigor mortis. Journal of the Science of Food and Agriculture, 12, 442±449. Menegalli, F. C., & Calvelo, A. (1979). Dendritic growth of ice crystals during the freezing of beef. Meat Science, 3, 179±198. Meryman, H. T. (1956). Mechanics of freezing in living cells and tissues. Science, 124, 515±521.
Ngapo, T. M., Babare, I. H., Murphy, K. L., Reynolds, J., & Mawson, R. H. (1997a). The eect of freezing rate on the ultractructure of a small sample of normal pork. In Proceedings of the 43rd International Congress of Meat Science and Technology (pp. 674± 675). Ngapo, T. M., Babare, I. H., Murphy, K. L., Reynolds, J., & Mawson, R. H. (1997b). The combined eects of freezing and thawing on the ultractructure of a small sample of normal pork. In Proceedings of the 43rd International Congress of Meat Science and Technology (pp. 676±677). Ngapo, T. M., Babare, I. H., Reynolds, J., & Mawson, R. H. (1999). Freezing and thawing rate eects on drip loss from samples of pork. Meat Science, 53, 149±158. Payne, S. R., Sandford, D., Harris, A., & Young, O. A. (1994). The eects of antifreeze proteins on chilled and frozen meat. Meat Science, 37, 429±438. Wirth, F. (1979). Chilling, freezing, storage and thawing of meat: present state of our knowledge. Fleischwirtschaft, 59, 1857(Cited by Ambrosiadis, I., Theodorakakos, N., Georgakis, S. and Lekas, S. (1994). In¯uence of thawing methods on the quality of frozen meat and the drip loss. Fleischwirtschaft, 74, 284±287).
www.elsevier.com/locate/meatsci
Freezing rate and frozen storage eects on the ultrastructure of samples of pork T.M. Ngapo*, I.H. Babare, J. Reynolds, R.F. Mawson Food Science Australia, Sneydes Road, Werribee, Victoria 3030, Australia Received 14 August 1998; received in revised form 7 April 1999; accepted 13 April 1999
Abstract Cryo-scanning electron microscopy was used to study the ultrastructure of small samples (approximately 6 g) of pork. Combinations of six freezing rates, two storage times and three thawing rates were used. Cavities created after sublimation of the ice crystals were quantitatively analysed using an image analysis software package. The cross-sectional areas of cavities of meat samples in the frozen state were approximately ten times the areas of the cavities of the fresh and thawed samples. The large cavities in the frozen state grossly distorted the muscle cell structures. Upon thawing, the meat structure had almost completely recovered. No signi®cant freezing rate eects were observed, however, trends were evident. Signi®cant storage time eects were observed. In the frozen state, at the 90th percentile level, the hole area fraction was greater in stored samples for intermediate cavity areas. In thawed samples, hole area fractions of stored samples were greater than in samples without storage. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction The freezing of meat has been widely studied as a means to lowering the amount of proteinacious exudate (drip) lost on thawing. The loss of ¯uid generally reduces the eating quality, binding ability and the weight of meat, all factors contributing to its monetary value. The volume of drip produced on thawing has been extensively related to the rate of freezing, which in turn has been related to the size and location of ice crystals in frozen meat. The dierent quality obtained in frozen foods, in particular meat and muscle tissue, has been the subject of many reviews (such as Cassens, 1971; Fennema, Powrie, & Marth, 1973; Jeremiah, 1996) and different sizes of the crystals formed are generally considered to be one of the factors responsible for changes in the quality obtained. While there is a general agreement about the mechanism of freezing and location of ice crystals in frozen meat, there is a lot of con¯icting literature about the eects of freezing and thawing on meat. In contrast to the extensive amount of literature about the eects of freezing rate on changes in * Corresponding author at: Station de Recherches sur la Viande, INRA, Theix, 63122 St GeneÁs-Champanelle, France. Tel.: +33-473624-167; fax: +33-473-624-089. E-mail address: [email protected] (T.M. Ngapo)
physical, chemical and sensory properties of meat, there are few reports about the eects of thawing rates on meat characteristics and, in particular, the ultrastructure of thawed meat. Many of the reported histological studies examining the ultrastructure of fresh or thawed meat have used light and transmission electron microscopy. These have required sample preparation methods of freeze substitution (including Bevilacqua & Zaritzky, 1980) or ®xation (such as Grujic, Petrovic, Pikula, & Amidzic, 1993). An example of an investigation of the eects of frozen storage on the ultrastructure of muscle is that of Carroll, Cavanaaugh, and Rorer (1981) who used scanning electron microscopy (SEM) to observe the ultrastructure of samples frozen at ÿ18 C and stored for 5 days, 4, 12 or 26 weeks. The images obtained showed good preservation of structure with little apparent change from the untreated sample. These samples were cut (18 to 20 mm thick) perpendicular to the longitudinal axis, wrapped in ``freezer paper'' and frozen in a stationary air freezer at ÿ18 C (about 4 h to freeze). Thawing occurred at room temperature and took 3.5 h for the centre of the sample to reach 4 C. Tissue ®xation was used in the sample preparation for SEM. The cause of the ice structure produced in a meat sample during freezing is the process of nucleation that
0309-1740/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(99)00051-0
160
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
takes place at its border. Investigation of the conditions that regulate the number of crystals in this zone will aid in the prediction of the ice-®bre con®guration in frozen tissues (Menegalli & Calvelo, 1979). The aim of the present study was to investigate the eects of combinations of six freezing rates, three thawing rates and two frozen storage times on the ultrastructure of small meat samples using a cryo-SEM technique employed by Payne, Sandford, Harris, and Young (1994). This method allowed examination of the cavities created by sublimation of ice crystals without requiring chemical ®xation of the meat sample, and thus reducing possible artefacts. A small sample size was used to simulate the refrigerated border in a larger piece of meat. Preliminary results of these studies were presented in earlier papers (Ngapo, Babare, Murphy, Reynolds, & Mawson, 1997a,b). 2. Materials and methods 2.1. Materials Porcine M. biceps femoris were obtained from carcasses that had been stored for 24 h at 4 C after slaughter. Equal numbers of cross bred (Landrace, Large White, Duroc) entire male and female pigs, between 22 and 24 weeks of age, were slaughtered at a local abattoir. Meat was stored at 4 C for up to a further 48 h before use. Only muscles with an ultimate pH in the range 5.4 to 6.0 were used. 2.2. Measurement of pH Meat (2.0 g) was ground in a mortar and pestle with deionised water (10 ml). A standard pH meter equipped with a glass electrode was used. Measurements of pH were calculated from the average of four replicates. 2.3. Sample preparation Ð freezing Sample preparation was conducted at 4 C. Cylindrical samples (approximately 70 mm length10 mm diameter) were cut from the centre of the muscle with their axis parallel to ®bre direction. Care was taken to avoid obvious pieces of fat and connective tissue. The samples were placed in aluminium holders (80 mm length10 mm i.d., 2020 mm outer cross-sectional area, in two longitudinal pieces) and plugged with expanded polyurethane foam (5 mm length) at either end. The holders, ®rmly held together with rubber rings, were placed in latex sheaths and secured with cable ties. The samples were placed vertically in a B. Braun Frigomix S cooling bath with a B. Braun Thermomix UB attachment immediately prior to freezing. Ethylene glycol (90% w/w) was used as a coolant.
Freezing velocities were de®ned as the characteristic freezing times (tcf) to traverse the temperature range from ÿ1 to ÿ7 C. Times (tcf) of 12, 30, 60, 120, 240 and >900 min were used. The temperature of the cooling bath was changed manually every 1 C for samples with tcf of 12 min and every 0.5 C for samples with tcf from 30 to 240 min. The temperature of the cooling bath was changed every 0.2 C for samples with tcf of >900 min to ÿ3.6 C where the temperature was held for approximately 15 h, then changed every 0.2 C to ÿ7 C. After cooling at the speci®ed rates, the bath temperature was reduced from ÿ7 to ÿ20 C in 15 min. After 1 h at ÿ20 C, the samples were transferred to a freezer at ÿ18 C, removed from the holders, and wrapped in cryovac barrier bags. Four samples were frozen at each tcf. Two samples were examined by electron microscopy within 1 week of freezing; two samples were stored for 4 weeks at ÿ18 C and then examined by electron microscopy within 1 week. Each freezing rate was investigated using one muscle from one animal and the freezing process was replicated twice (i.e. two muscles, each from dierent animals, were frozen at the same rate but on dierent occasions). The experiment was accordingly a split-plot in a completely randomised design Ð the ``main plots'' were muscles, and each muscle received a freezing rate; the ``splitplots'' were pairs of samples from a muscle, and each pair received a storage time. 2.4. Sample preparation Ð thawing Sample preparation was conducted at either 4 or ÿ18 C depending on the state of the sample; fresh and thawed samples, or frozen samples, respectively. After freezing, by the described method, samples were either held in the cooling bath at ÿ20 C for up to 15 h (described as ``no storage'') or stored at ÿ18 C for 4 weeks. The stored samples were held at ÿ20 C for 1 h, then transferred to a freezer at ÿ18 C, removed from the holders, and wrapped in cryovac barrier bags. Immediately prior to thawing, the samples were replaced in the aluminium holders at ÿ18 C and transferred to the cooling bath at ÿ20 C. Thawing velocities were de®ned as the characteristic thawing times (tct) to traverse the temperature range from ÿ7 to ÿ1 C. Thawing times (tct) of 12, 60 and 180 min were used. The cooling bath temperature was raised from ÿ20 to ÿ7 C in 15 min. The bath temperature was then changed manually every 1 C for samples with tct of 12 min and every 0.5 C for samples with tct of 30 and 180 min to 2 C. Samples were held at this temperature for 10 min prior to use. Four samples were frozen and thawed at each freeze rate/thaw rate combination. Two were thawed with no storage and two were stored prior to thawing. Two further samples were analysed fresh (unfrozen). The fresh
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
or thawed samples were each cut into six discs of approximately 10 mm length and the two end discs were discarded. The discs were plunged into liquid N2 until frozen. After removal from the liquid N2, the samples were stored in sealed plastic vials for up to 7 days at ÿ70 C, until 1 h prior to SEM, when the discs were transferred to a freezer at ÿ18 C. Each of the 18 freezing rate by thawing rate combinations was investigated using one muscle from one animal. These treatment combinations were not able to be replicated as only 18 muscles were used. 2.5. Monitoring of freezing and thawing rates The samples in the present paper were frozen with the samples used for drip loss analysis in the accompanying paper (Ngapo, Babare, Reynolds, & Mawson, 1999). Four samples used for drip loss analysis, had copper constantan thermocouples inserted into their centres, along the longitudinal axis of the cylinder. Two further thermocouples were placed in the coolant and one at room temperature. The thermocouples were connected as dierential inputs to a PCI-20303T Termination Panel which was part of an Intelligent Instrumentation Visual Designer PCI-20,000 system used to continuously monitor the freezing rate. 2.6. Scanning electron microscopy (SEM) The SEM was conducted using a Cambridge Instruments Stereoscan 90 scanning electron microscope ®tted with an Oxford Instruments CT-1000A cryostage. At least four images were captured from each stub using an Image Slave Software Package. Sample preparation was conducted at ÿ18 C. Cores (3 mm diameter) were removed from the meat samples parallel to ®bre direction and inserted into a hole (3 mm diameter) drilled in the centre of a cryomicroscope stub. The stubs were then held in a polystyrene cup over liquid N2 for up to 2 h before plunging into liquid N2. The meat sample was freeze fractured with a scalpel blade perpendicular to the ®bre direction, mounted on the cryostage and heated to ÿ60 C until the ice had sublimed. The tissue surface was examined with an accelerating voltage of 2.5 kV and at a constant working distance of 7 mm. 2.7. Data translation Quantitative data were obtained from the images using a Data Translation Global Lab Image Software Package. Two stubs of each sample were used for analysis. The numbers of crystals in the frozen meat were estimated by image analysis of the cavities observed in the meat after sublimation. At least 150 cavities were measured from each image and a minimum of three images were analysed from each stub. The cavities were grouped into categories
161
based on cross-sectional areas. The total area of meat viewed in each image was referred to as the region of interest (ROI). The fraction of the ROI occupied by the cavities of a category size was calculated and referred to as the hole area fraction for that category size. To categorise the cavities using image analysis, the closest approximation to zero that could be obtained, at the magni®cations of the images, was 0.2210ÿ3 mm2 for the frozen samples and 0.1210ÿ4 mm2 for the thawed samples. 2.8. Statistical analysis The 10th and 90th percentiles, lower and upper quartiles and medians were calculated from the hole area fractions for each freezing rate and storage time combination (in the case of the frozen samples) and for each freezing rate, thawing rate and storage time combination (in the case of the fresh and thawed samples). Percentiles of the hole area fractions were de®ned as the hole area fractions for which a percentage of the total cavity area was accounted for by cavities of that size or smaller. Separate analyses based on cavity size categories were undertaken on the frozen samples only. The analysis of variance (ANOVA) routine in the Genstat 5 statistical package (Genstat 5 Committee, 1993) was used to analyse the data. A conventional split-plot analysis of variance was used to analyse the data from the frozen samples. The analysis of the data from the fresh and thawed samples was facilitated by assuming that three-way (and higher order) interactions of the treatment eects were negligible and their associated observed mean squares represented residual or error variation. Consequently, the eect of storage and the two-way interactions of storage with freezing rate and thawing rate were able to be investigated, but not the main eects of freezing rate or thawing rate, and their interaction. For both sets of data, transformations to overcome heterogeneity of variance were explored. 3. Results 3.1. Freezing and thawing curves The temperature at the centre of selected samples used for drip loss analysis (Ngapo et al., 1999) was monitored and displayed continuously with the temperature of the cooling bath. No lag in sample temperatures was observed compared to the bath temperature. 3.2. Eect of freezing rate on cavity size of frozen samples The cross-sectional areas of the cavities of samples in the frozen state were approximately 10 times the areas of the cavities of the fresh samples and the samples after
162
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
thawing with and without storage. Figs. 1a±d illustrate the relative sizes of the cavities in fresh, frozen and thawed meat samples. The gross distortion of the meat structure in the frozen state compared to fresh and thawed samples is evident in these images. The percentiles calculated from hole area fractions are presented in Table 1. There was some evidence of heterogeneity of variance in the analyses of the 10th and 90th percentiles and a log-transformation corrected this problem, but as the conclusions from these analyses were essentially the same as the analyses on the original (untransformed data) all the results in Table 1 are based on analyses of the untransformed data. No signi®cant dierences were observed in the 10th percentiles, lower and upper quartiles and median values comparing samples with no storage and after 4 weeks of storage. A signi®cant dierence (p=0.034) was observed in the 90th percentiles. Without storage, 90% of the hole area fraction was occupied by cavities of cross-sectional area below 10.6710ÿ3 mm2. After 4 weeks storage, the cavities occupying 90% of the hole area fraction had cross-sectional areas of 8.9810ÿ3 mm2, and smaller (l.s.d.=1.4910ÿ3 mm2).
Figs. 2 and 3 illustrate the spread of the mean hole area fractions for each grouping of cavity size at each of the six freezing rates studied with and without storage, respectively. Note that the cavity size categories vary in area ranges: approximately 0.2210ÿ3 mm2 for the ®rst four categories up to a cavity size of 1.1510ÿ3 mm2 approximately 0.5710ÿ3 mm2 for the next six categories to 4.6010ÿ3 mm2 and 1.1510ÿ3 mm2 for the last eight categories to 13.7910ÿ3 mm2. Cavities of Table 1 Hole area fraction mean percentiles of frozen samples Percentile
Hole fraction area Hole fraction area l.s.d. (p=0.05) (10ÿ3 mm2) (10ÿ3 mm2) (10ÿ3 mm2) no storage 4 week storage
10th 0.67 25th (lower 1.61 quartile) 50th (median) 3.60 75th (upper 6.89 quartile) 90tha 10.67 a
0.73 1.59
0.08 0.22
3.32 6.04
0.57 1.10
8.98
1.49
Percentile values are signi®cantly dierent.
Fig. 1. A typical example of the eects of freezing and thawing on the ultrastructure of a small sample of pork; (a) fresh, (b) tcf of 60 min, not thawed, (c) tcf of 60 min, stored up to 15 h, tcf of 12 min, (d) tcf of 60 min, stored 4 weeks, tct of 12 min. Scale bar=200 m.
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
163
Fig. 2. The mean hole area fractions for each grouping of cavity size at each of the six freezing rates studied without storage.
Fig. 3. The mean hole area fractions for each grouping of cavity size at each of the six freezing rates studied after 4 weeks storage.
larger cross-sectional areas occurred infrequently and a sensible average could not be estimated. A maximum cavity size of 13.7910ÿ3 mm2 was therefore selected to illustrate the spread. Separate analyses for each cavity size category showed that the hole area fraction was signi®cantly greater in the samples stored for 4 weeks for the categories from 0.93 to 3.4510ÿ3 mm2 (p<0.05) than for the samples that had not been stored. The hole area fractions calculated from cavities in the area categories from 0.22 to 0.9310ÿ3 mm2 and from 3.45 to 13.7910ÿ3 mm2 were not signi®cantly dierent after 4 weeks storage. These dierences are apparent, but not signi®cant, in the quartiles and medians reported in Table 1 Ð the interquartile range is narrower (3.87 after 4 weeks storage compared to 4.59 with no storage), but not signi®cantly so (p=0.077), and has
shifted in the direction of smaller cavity areas after 4 weeks storage. Figs. 4 and 5 illustrate the hole area fractions for the six freezing rates with and without storage, respectively. Note that the cavity size categories vary in area ranges and are the same as those used to illustrate the spread of the cavity areas in Figs. 2 and 3. The pro®les suggest that the hole area fractions for samples with tcf of 12 and >900 min are approximately the same before and after 4 weeks storage. At tcf of 30 and 60 min the hole area fractions appear to be consistently greater after 4 weeks storage. At tcf of 120 and 240 min the hole area fraction is greater in the samples without storage for smaller area categories, but becomes similar as the cross sectional area increases. For larger cross-sectional areas the samples with tcf of 120 min tend to a greater hole area fraction after 4 weeks storage.
164
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
Fig. 4. Mean hole areas as fractions of the ROI at characteristic freezing times for samples with no storage. The cavity size categories are 10ÿ3 mm2.
Fig. 5. Mean hole areas as fractions of the ROI at characteristic freezing times for samples after 4 weeks storage. The cavity size categories are 10ÿ3 mm2.
Trends in the patterns of hole area fractions with freezing rates are apparent (Figs. 4 and 5). A minimum is suggested at a tcf of 30 min, and maxima at 120 min. In the larger size categories, the minimum at 30 min disappears, and a minimum at 240 min becomes apparent. After 4 weeks storage, maxima at tcf of 120 and >900 min and a minimum at 240 min are more strongly suggested.
larger in the stored samples, particularly the long narrow cavities around the edge of the ®bre bundles (Fig. 1c and d). Storage time had signi®cant eects on the cavity areas of the thawed samples. Percentiles based on hole area fractions are presented in Table 2. While there was some evidence of heterogeneity of variance in the medians, upper quartiles and 90th percentiles, analyses of transformed data gave the same signi®cance test results as the untransformed data and so all means in Table 2 are based on analyses of the untransformed data. Signi®cant dierences were observed in the 10th percentiles (p=0.003) and lower quartiles (p=0.017) comparing cavity areas of frozen samples that had been stored for 4 weeks prior to thawing with cavity areas of samples frozen then thawed without storage. Only the frozen samples could be compared statistically, that is, the
3.3. Eect of thawing rate on cavity size of frozen samples It was observed that after thawing the ultrastructure of the meat sample almost wholly recovered from the grossly distorted structure observed in the frozen state (Figs. 1a±d). It was also noted that where dierences in the cavity areas were evident when comparing thawed samples with and without storage, the cavities appeared
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
cavity areas of the frozen/thawed samples, with or without storage, could not be compared to those of the fresh samples. A freezing rate by storage interaction was observed for the 10th percentile (p=0.039). The storage dierences were signi®cant at tcf of 12, 60 and >900 min (Table 3). Only the means of samples that had been frozen at the same rate could be compared statistically. 4. Discussion The temperature at the centre of the meat did not lag behind the temperature of the coolant and therefore there was no evidence of a temperature gradient in the meat sample during freezing. Menegalli and Calvelo (1979) suggested that the regulation of the size of the crystals by the speed of cooling is limited to samples with no temperature gradients (small pieces of meat) or to the peripheral region of large pieces of meat. Bevilacqua, Zaritzky, and Calvelo (1979) explained that when meat comes into contact with the refrigerating system a substantial supercooling is reached which generates a number of nuclei proportional to it. In the interphase of the crystals formed, the equilibrium temperature is soon reached due to a release of the latent heat of crystallisation. As crystal growth elevates the temperature, no further nucleation occurs. During this nucleation phase, crystal growth and nucleation occur simultaneously. A range of crystal sizes would therefore be expected at the termination of this phase. The continuation of crystal growth beyond this point suggests that there should be few of the initial small crystals (nuclei) remaining upon completion of the freezing regime, and that the time to reach the ®nal freezing temperature should in¯uence the sizes of crystals. Comparison of the meat samples frozen with tcf ranging from 12 to >900 min with meat samples frozen in liquid N2 shows the cavity cross-sectional areas were about 10-fold greater using the slower freezing rates; freezing in liquid N2 resulted in greater than 90% of cavity cross-sectional areas being less than 6.0010ÿ4
165
mm2 compared to less than 75% of cavity areas smaller than 6.0010ÿ3 mm2 and less than 10% smaller than 6.0010ÿ4 mm2 at the slower freezing rates. The large cavity areas in the samples frozen at tcf of 12 to >900 min created a gross distortion of the muscle cell structures compared to the structure observed in the samples frozen in liquid N2. To ensure that this was not an artefact created by the sample preparation for SEM, sublimation was observed as it occurred. No evidence of distortion as a consequence of the preparative procedure was found. The nuclei formed in the samples frozen in liquid N2 would have had very short growth phases, if any, and it is therefore reasonable to assume that these areas indicate values only slightly larger than that of the nuclei. The eective absence of these smaller cavities in the slower frozen samples is an indication that crystal growth has occurred. While the rate dierences between the cryo-SEM preparative freezing in liquid N2 and the experimental freezing (tcf from 12 to >900 min) in this study results in large dierences in cavity areas, it is suggested that the curves of nucleation rate and growth velocities versus supercooling for the experimental freezing rates are not so dierent. Therefore, more subtle dierences, if any, might be expected as a result of the smaller dierences in sizes of crystals formed at the dierent experimental freezing rates studied. The selection of freezing rates in the present study was based on the categories of ice crystals formed as suggested in the papers by Love (1958a,b) and Love and Haraldsson (1961). These studies were part of a comprehensive series of investigations examining the eects of freezing rate on cell disruption in cod ®llets. In the present study, a distinction could not be made between intra- and extracellular locations of ice crystals for comparison to this published work. However, while no signi®cant eects of freezing rate on cavity size were found, trends in the patterns of hole area fractions with freezing rates were apparent (Fig. 4) and comparisons could be made to this earlier work. At the very small cavity areas (less than 0.5010ÿ3 mm2) the trend was similar to that suggested by these workers. There
Table 2 Hole area fraction mean percentiles of fresh and thawed samples Percentile
Hole fraction area (10ÿ4 mm2) fresh sample
Hole fraction area (10ÿ4 mm2) no storage
Hole fraction area (10ÿ4 mm2) 4 wk storage
l.s.d. (p=0.05) (10ÿ4 mm2)
10th 25th (lower quartile) 50th (median) 75th (upper quartile) 90th
0.164 0.25 0.65 2.16 5.45
0.166a 0.26b 0.87 2.75 5.94
0.175a 0.29b 0.87 3.19 6.74
0.005 0.03 0.44 0.97 1.59
a b
Percentile values are signi®cantly dierent. Percentile values are signi®cantly dierent.
166
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
Table 3 Means of cavity size at dierent freezing rates (10th percentile level) Freeze
tcf (min)
Mean cavity size (10ÿ4 mm2) no storage
Mean cavity size (10ÿ4 mm2) 4 week storage
Numbers of samples analysed
No Yes Yes Yes Yes Yes Yes
0 (fresh) 12 30 60 120 240 >900
0.164 0.169a 0.167 0.169a 0.163 0.169 0.155a
0.193 0.158 0.184 0.168 0.174 0.171
36 12 12 12 12 12 12
a Means are signi®cantly dierent at the same freezing rates (l.s.d.=0.01310ÿ4 mm2 for comparison of samples at the same freezing rate).
appeared to be fewer of these small cavities at tcf of 30 min than at tcf of 12 min and a maximum at tcf of 120 min. Furthermore, there appeared to be more cavities up to about 2.5010ÿ3 mm2 at tcf of 120 min than at the other freezing rates. In contrast to the work of Love (1958a,b) and Love and Haraldsson (1961) there appeared to be large areas occupied by both very small cavities (less than 0.5010ÿ3 mm2) at a tcf of 240 min and cavities up to about 2.5010ÿ3 mm2 at tcf of >900 min. A more extensive study is required to expand these studies and for veri®cation of trends. Menegalli and Calvelo's (1979) theory of dendritic ice formation explains that the dierent zones of crystal sizes and locations observed in large pieces of meat where a temperature gradient exists, are a consequence of the sample size which leads to dierent thermal histories according to the zone considered. This could explain the dierences in results in the present work and the work of Love (1958a,b) and Love and Haraldsson (1961), who used a much larger sample size than the cylinders of meat used here. In larger samples of tissue, except for a small zone which corresponds to the refrigerated border of the sample (in which nucleation and thermal dendritic growth exist) nuclei grow through cells or cellular dendrites, according to whether the heat extraction is low or high, respectively. After 4 weeks frozen storage, a signi®cant dierence (p=0.034) was observed in the 90th percentiles of hole area fractions. With no storage, 90% of the hole area fraction was 10.6610ÿ3 mm2 and less. After four weeks storage, the cavities accounting for 90% of the hole area fraction were 8.9810ÿ3 mm2 and smaller. The lack of signi®cant dierences in 10, 25, 50 and 75 percentiles of hole area fractions with and without storage indicate that the relative proportions of cavities with cross-sectional areas up to about 6.510ÿ3 mm2 are similar regardless of the application of storage. However, the signi®cant dierence in the 90 percentile values suggests that in the samples with no storage there are proportionately fewer cavities within the area range from about
6.5 to 9.010ÿ3 mm2 and/or that the number of larger cavities is greater (for example, cavity areas >1310ÿ3 mm2) compared to the samples with storage. The latter suggestion requires that some of the largest cavities formed upon freezing become smaller with storage. Separate analyses for each cavity size category showed that the cavity area for each category as a fraction of the ROI was signi®cantly greater in the samples stored for 4 weeks for the categories from 0.93 to 3.4510ÿ3 mm2 (p<0.05) than for the samples that had not been stored. The hole area fractions calculated from cavities in the area categories from 0.22 to 0.9310ÿ3 mm2 and from 3.45 to 13.7910ÿ3 mm2 were not signi®cantly dierent after 4 weeks storage. The increase in this intermediate fraction of cavities could narrow the interquartile range. The cavities with areas >13.810ÿ3 mm2 were not included in these separate analyses. The impression gained is that storage for 4 weeks had narrowed the range of cavity areas and reduced the percentage of large cavities in samples left in the frozen state. Trends in the patterns of hole area fractions as a consequence of storage and freezing rate were observed (Figs. 4 and 5). The hole area fractions for samples with tcf of 12 and >900 min were approximately the same before and after 4 weeks storage. At tcf of 30 and 60 min the hole area fractions appeared to be consistently greater after 4 weeks storage. At tcf of 120 and 240 min the hole area fraction was greater in the samples without storage for smaller area categories (<0.510ÿ3 mm2). For larger cross-sectional areas (>2.510ÿ3 mm2) the samples with tcf of 120 min tend to a greater hole area fraction after 4 weeks storage. An increase in crystal sizes during storage can be explained by the process of recrystallisation. During storage, preferential growth in the solid state of large crystals occurs at the expense of smaller ones (Meryman, 1956). The data in the current study presents not only increases in crystal sizes, but a complex system of changes in crystal sizes at dierent rates of freezing, which is not obviously explained by recrystallisation. After thawing, the meat structure appeared to have almost completely recovered from the gross distortion observed in the frozen state. Wirth (1979) suggested that by using a small sample, a slow thawing rate results in reduced drip loss due to the partial absorption of exudate by the muscular ®brils. A consequence of this theory is that the ultrastructure of the sample partially recovers from the grossly distorted structure observed in the frozen state. Micrographs of typical images in the fresh, frozen and thawed states are given in Fig. 1a±c. The distortion of the meat ultrastructure in the frozen state and recovery from this state after thawing are evident in these images. Dierences in the cavity areas were evident in some samples when comparing thawed samples with and
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
without storage. The cavities appeared larger in the stored samples, particularly the long narrow cavities around the edge of the ®bre bundles (compare Fig. 1c and d). Signi®cant dierences were observed in the 10th percentiles (p=0.003) and lower quartiles (p=0.017) comparing hole area fractions of samples that had been frozen, stored for 4 weeks and then thawed, with hole area fractions of samples frozen then thawed with no storage. These percentile values were greater in the samples that had been stored (Table 2), indicating that there were proportionately fewer small cavities (<0.310ÿ4 mm2) and/or proportionately more intermediate cavities (about 0.3 to 0.910ÿ4 mm2) after thawing stored samples. Interestingly, the separate analyses on the cavity areas in the frozen state showed that the hole area fraction for each category was signi®cantly greater in the samples stored for 4 weeks for the categories from 0.93 to 3.4510ÿ3 mm2 (also an intermediate range of cavity areas) than for the samples that had not been stored. In the thawed samples, a freezing rate by storage interaction was also observed at the 10th percentile (p=0.039). The storage dierences were evident at tcf of 12, 60 and >900 min (Table 3) where the mean cavity areas were signi®cantly greater in the samples that had been stored. It is also of interest that signi®cant dierences in hole area fractions were observed after 4 weeks storage in both the frozen and thawed states, but that the dierences were only signi®cant at the 90% level in the frozen state and at the 10 and 25% levels in the thawed states. During thawing in the present study, a lag between the temperature at the centre of the meat and the coolant was not evident as a result of the small sample size used. There was, therefore, no evidence of a temperature gradient in the sample during the thawing process suggesting that recrystallisation due to a melting front was unlikely. However, recrystallisation while holding the sample at temperatures below the freezing point would have occurred. Earlier studies on drip loss in this system (Ngapo et al., 1999) showed that, at freezing rates with tcf of 12± 120 min, drip loss obtained was not signi®cantly dierent from that obtained from the fresh samples. At the slower freezing rates studied (tcf of 240 and >900 min) drip loss was signi®cantly dierent from drip loss obtained from the fresh samples. After 4 weeks of frozen storage, the drip loss was signi®cantly dierent at tcf of 12±120 min, compared to the samples without storage. However, there were no signi®cant dierences in drip loss of the stored samples comparing the freezing rates. Furthermore, the drip loss from stored samples was not signi®cantly dierent from that obtained at the two slowest freezing rates with no storage. This investigation illustrates that the combined eects of freezing rates, frozen storage and thawing rates on crystal formation and the ultrastructure of a small meat
167
sample present a complex system. Comparison of these results with those obtained from drip measurements adds to this complexity and any correlation of drip loss with crystal formation or meat ultrastructure is not apparent using a small meat sample. 5. Conclusions Large cavities observed in the frozen state caused gross distortion of the muscle cell structures, but after thawing, the meat structure appeared to have almost completely recovered. The combined eects of freezing rates, frozen storage and thawing rates on crystal formation and the ultrastructure of a small meat sample present a complex system. Comparison of ultrastructural data with drip measurements adds to this complexity and any correlation of drip loss with crystal formation or meat ultrastructure was not apparent using small meat samples. As the ice structure produced in a sample upon freezing depends upon the process of nucleation that takes place at the refrigerated border, study of the conditions that regulate the number of crystals in that zone is of great importance in predicting the ice ®bre con®guration in frozen tissues. The use of the cryo-SEM method on a larger sample size that has a thermal gradient across it during freezing, and thereby simulates industrial conditions of meat freezing, would provide valuable information to complement this work. Furthermore, observations were only made in cross-sections perpendicular to the ®bre direction and actual breakage of cells might be better seen using transmission electron microscopy. Acknowledgements The authors gratefully acknowledge the ®nancial support provided by the Australian Pig Research and Development Corporation and the Agriculture and Food Initiative of the Victorian Department of Natural Resources and Environment. References Bevilacqua, A. E., & Zaritzky, N. E. (1980). Ice morphology in frozen meat. Journal of Food Technology, 15, 589±597. Bevilacqua, A, Zaritzky, N. E., & Calvelo, A. (1979). Histological measurements of ice in frozen beef. Journal of Food Technology, 14, 237±251. Carroll, R. J., Cavanaugh, J. R., & Rorer, F. P. (1981). Eects of frozen storage on the ultrastructure of bovine muscle. Journal of Food Science, 46, 1091±1094, 1102. Cassens, R. G. (1971). Microscopic structure of animal tissue. In J. F. Price, & B. S. Schweigert (Eds), The Science of Meat and Meat Products (2nd ed.). San Fransisco: W. H. Freeman and Co. Fennema, O. R., Powrie, W. D., & Marth, E. H. (1973). Low temperature preservation of foods and living matter (pp. 150±239). New York: Marcel Dekker.
168
T.M. Ngapo et al. / Meat Science 53 (1999) 159±168
Genstat 5 Committee (1993). Genstat 5 release 3 reference manual. Oxford: Oxford University Press. Grujic, R., Petrovic, L., Pikula, B., & Amidzic, L. (1993). De®nition of the optimum freezing rate Ð I. Investigation of structure and ultrastructure of beef M. longissimus dorsi frozen at dierent freezing rates. Meat Science, 33, 319±331. Jeremiah, L. E. (1996). Freezing eects on food quality. New York: Marcel Dekker, Inc. Love, R. M. (1958a). The expressible ¯uid of ®sh ®llets. VIII. Cell damage in slow freezing. Journal of the Science of Food and Agriculture, 9, 257±262. Love, R. M. (1958b). The expressible ¯uid of ®sh ®llets. IX. Other types of cell damage caused by freezing. Journal of the Science of Food and Agriculture, 9, 262±268. Love, R. M., & Haraldsson, S. B. (1961). The expressible ¯uid of ®sh ®llets. XI.-Ice crystal formation and cell damage in cod muscle frozen before rigor mortis. Journal of the Science of Food and Agriculture, 12, 442±449. Menegalli, F. C., & Calvelo, A. (1979). Dendritic growth of ice crystals during the freezing of beef. Meat Science, 3, 179±198. Meryman, H. T. (1956). Mechanics of freezing in living cells and tissues. Science, 124, 515±521.
Ngapo, T. M., Babare, I. H., Murphy, K. L., Reynolds, J., & Mawson, R. H. (1997a). The eect of freezing rate on the ultractructure of a small sample of normal pork. In Proceedings of the 43rd International Congress of Meat Science and Technology (pp. 674± 675). Ngapo, T. M., Babare, I. H., Murphy, K. L., Reynolds, J., & Mawson, R. H. (1997b). The combined eects of freezing and thawing on the ultractructure of a small sample of normal pork. In Proceedings of the 43rd International Congress of Meat Science and Technology (pp. 676±677). Ngapo, T. M., Babare, I. H., Reynolds, J., & Mawson, R. H. (1999). Freezing and thawing rate eects on drip loss from samples of pork. Meat Science, 53, 149±158. Payne, S. R., Sandford, D., Harris, A., & Young, O. A. (1994). The eects of antifreeze proteins on chilled and frozen meat. Meat Science, 37, 429±438. Wirth, F. (1979). Chilling, freezing, storage and thawing of meat: present state of our knowledge. Fleischwirtschaft, 59, 1857(Cited by Ambrosiadis, I., Theodorakakos, N., Georgakis, S. and Lekas, S. (1994). In¯uence of thawing methods on the quality of frozen meat and the drip loss. Fleischwirtschaft, 74, 284±287).