ABSTRACT
This report indicates that the quality of fish meal can conveniently be expressed by two independent indices: a pre-processing spoilage index and a heat abuse index. It is shown that indices of spoilage such as total volatile basic nitrogen (TVBN) or free fatty acid (FFA) can be misleading. As an additional quality index the phospholipid (PL) content is suggested. This index measures the PL content of the residual lipids in the meal and allows the quality of the dried presscake to be assessed separately from the added stickwater. The polyunsaturated fatty acid (PUFA) content of the meal lipids is also an excellent guide for detecting heat abuse during drying of the meal. Using this measurement no evidence was found to conclude that steam (indirect) dryers cause more heat damage to meals than flame (direct) dryers.
INTRODUCTION
Fish meal is produced in South Africa from several species of fish unsuitable for human consumption. These fish include anchovy (Engraulis capensis), red-eye (Etrumeus whiteheadi), maasbanker (Trachurus trachurus), lantern fish (Lampanyctodes hectoris) and pilchard (Sardinops ocellata), the latter largely in the form of cannery waste. The meal is produced by a heating, pressing and drying process and normally consists of about 60–70% protein, 10–12% residual lipid and 8–10% moisture, with the rest consisting of bones and salt (ash). After heating, the fish is pressed, yielding a solid presscake and press liquor. The press liquor is centrifuged and separated into an aqueous phase (stickwater) and fish oil. The stickwater is then subjected to a three-stage evaporation process and added back to the presscake, which is dried in either flame (direct) dryers, or steam (indirect) dryers. Ethoxyquin (1,2-dihydro-6-ethoxy-2,2,4-trimethylquinoline, EQ) is normally added in quantities of about 400 mg/kg immediately after drying to prevent oxidation of the valuable polyunsaturated fatty acids (PUFA). The price of the meal in South Africa is solely determined by its protein content (% nitrogen×6.25) and no account is taken of protein quality. The quality of the meal depends on the freshness of the fish from which it was produced and on processing conditions. Normally, high quality meals are the result of extremely fresh fish and low temperature drying conditions, while spoiled fish inevitably yields poor quality meal.Citation[1] Enzymatic and bacterial spoilage of the fish leads to the formation of volatile nitrogenous bases (TVBN) from the proteinCitation[2] and free fatty acids (FFA) from the lipidsCitation[3] thus these quality indices are frequently used to express meal quality.
The TVBN content of a meal also depends upon drying conditions, i.e., temperature and duration, as some of the bases are volatilised during the drying process. Valid comparisons of TVBN content can therefore only be made on meals that were dried under identical conditions.Citation[2] The FFA content of a meal is also not a reliable index because FFA generated enzymatically increase to a maximum and then decrease.Citation[4] This is most likely due to long chain FFA being degraded into short chain volatile and water soluble FFA, which escape detection when titrating the meal lipids for FFA content. As an addition to these two quality indices the phospholipid (PL) level of a meal has been suggested.Citation[3] Fresh fish have a PL content of 0.5–0.8% (wet fish)Citation[5] which decreases rapidly with spoilage.Citation[3], Citation[6] In addition, PL remain almost exclusively in the meal and very little ends up in the press oil,Citation[3] so the meal's PL level should be a good index of the original freshness of the fish. This index also differs from the TVBN and the FFA contents in that it does not determine the level of spoilage products which are volatile, as in the case of TVBN, or vulnerable to further breakdown, as in the case of FFA, but rather the remaining quantity of a reasonably constant amount of starting material.Citation[3], Citation[6]
Another factor that affects the TVBN content of a meal is the variable addition of concentrated stickwater to the presscake prior to drying. This addition inevitably increases its TVBN level giving a false estimate of the underlying quality of the meal. This stickwater also contains some PL but its contribution to the PL level of a meal is only in the order of 1–6%, i.e., negligible.Citation[7] In short, the PL level of a meal should be included in expressing the quality of a meal as it accurately measures the quality of the dried presscake apart from the concentrated stickwater.
The quality of a meal may be also adversely affected by drying. This quality deterioration is usually referred to as heat abuse (burning or scorching) and is distinct from the deterioration due to spoilage of the fish prior to processing. Heat abuse destroys the PUFA of the meal lipids and can therefore be readily detected by analysing the meal for PUFA content. The lipids of a properly dried anchovy, pilchard and red-eye meal have PUFA contents of 40–45%, while a meal that has suffered heat damage has lower values.Citation[8]
Meals which have no added EQ inevitably show low PUFA levels that rapidly decrease with time.Citation[9] Such meals, which seldom occurred, have been deliberately excluded from this study. Steam (indirect) dryers are reported to cause more heat abuse than flame (direct) dryers.Citation[10] If so, then this is probably due to the fact that the residence times in steam dryers are much longer than in flame dryers (60–120 min vs. 10–20 min), respectively.Citation[10]
The present report represents a survey of 243 fish meals from seven different South African factories. These meals were evaluated for pre-processing fish spoilage by analysing for TVBN, FFA and PL levels. In addition, 215 meals made from anchovy, pilchard and red-eye (selected from the original 243 by excluding meals such as maasbanker and lantern fish) were evaluated for heat damage by determining PUFA levels. The survey extended over four fishing seasons which were conveniently divided into three: the combined 1996/1997, the 1998 and 1999 season. In addition, the survey critically evaluates the performance of the seven factories, and the validity of the different quality indices. A preliminary report of the results of the 1996/1997 fishing seasons has been issued.Citation[11]
MATERIALS AND METHODS
General
Samples of freshly produced fish meals were received at the Institute in air-tight polythene bags. They were stored at ambient temperature and analysed within two weeks of production, while the history of the meals was obtained from the factory concerned. The samples consisted mainly of anchovy, red-eye or pilchard meals, together with a smaller quantity of maasbanker or lantern fish meals. These latter two meal types were not included in the mean PUFA level calculations for possible heat abuse. All meals were qualitatively tested for the presence of the antioxidant EQ by means of fluorescence under ultra-violet light.Citation[12]
ANALYTICAL DETERMINATIONS
Moisture Content
This was determined by drying a weighed amount of meal (5–10 g) in an oven at 105°C for 4 h and noting the weight loss.Citation[13]
Protein Content
This was determined by the Kjeldahl method using cupric and potassium sulfate as catalyst.Citation[14]
Lipid Content
Twenty-five grams of fish meal were extracted three times with 75 mL chloroform: methanol (2:1, by vol) in a blender. The extracts were filtered, combined and washed once, according to Folch, Lees and Sloane-Stanley.Citation[15] The purified extract was dried over anhydrous sodium sulfate, filtered and the chloroform evaporated on a rotary evaporator. The lipids were taken up in a small volume of hexane, dried over sodium sulfate again, filtered, evaporated and dried on a freeze dryer. After weighing to obtain the fat content of the meal, the residual lipids were analysed for FFA and phosphorus (P) content.
FFA Content
This was determined by titrating the lipids with 0.02 M sodium hydroxide in hot ethanol:chloroform (4:1) using phenolphthalein as indicator, and expressed as oleic acid.Citation[16]
P and PL Content
The P content of the meal lipids was determined according to the method developed at our Institute, but replacing 2 M hydrochloric acid with 1 M sulphuric acid.Citation[17] The PL content of the meal was calculated by mul tiplying the P content of the lipids by a factor of 28.6 and the resulting number by the fat content of the meal.Citation[17] The results were expressed on a dry basis.
TVBN Content
This was determined by distilling the volatile bases from a weighed sample of meal (1–3 g) to which magnesium oxide (about 2 g) was added into 30 mL of saturated boric acid. The solution was titrated with 0.02 M hydrochloric acid using screened methyl red as indicator. The TVBN was expressed as mg N per 100 g meal on a dry basis.Citation[18]
PUFA Content
The PUFA content of the meal lipids was determined by extracting 5 g of meal with chloroform:methanol (2:1, by vol) according to the Bligh and Dyer procedure.Citation[19] The extracted lipids were immediately converted into methyl esters and analysed by gas chromatography.Citation[20]
RESULTS AND DISCUSSION
Phospholipid Content
Lovern Citation[5] in reviewing the PL content of sixteen fish species comes to the conclusion that compared to the well known large variations in triglyceride storage lipids, fish flesh PL are reasonably constant. He quotes values for herring (Clupea harengus), a fatty fish comparable to anchovy, pilchard and red eye of between 0.7–0.8% (wet fish), while non-fatty fish have somewhat lower values of between 0.5–0.6% (wet fish).
By far the largest proportion of South African meals are made from anchovy, pilchard or red-eye. Of the 243 fish meals analysed in the 1996 to 1999 fishing seasons 215 (89%) were of these three species, while the remaining were lantern fish or maasbanker meals. PL contents of fresh whole samples as determined on various occasions (batches A–D) are shown in Table .
Table 1. Phospholipid Content of Some Whole South African Fish Used in the Fish Meal Process
These results clearly indicate the relative constancy of the PL content for the different species, but more importantly, in the case of anchovy, they show how enzymatic hydrolysis on spoilage rapidly reduces this PL content.Citation[6] For instance, anchovy of batch A, iced at sea and analysed within 12 h of catching, had a PL content of 1.26%, but this diminished to 1.00% on storage at 17°C for 24 h and to 0.74% on further storage for 48 h at 17°C. In his review, LovernCitation[5] does not mention the freshness of the fish nor whether whole fish or only the edible part was examined. This might explain the rather large variation (0.5–0.8% wet fish) in his data. Surprisingly, in spite of the large variation in his values he seems to have drawn the correct conclusion.
Fish Meal Composition
Table records the mean moisture, protein and fat contents of 243 fish meals during 1996–1999 produced by the seven factories included in this survey. The law requires that meals have a maximum moisture content of 10.0%, a maximum fat content of 12.0% and a minimum protein content of 60.0%.Citation[22] However, South African fish meal producers have agreed to raise the minimum protein level to 63.0% to satisfy their customers.
Table 2. Mean Moisture, Protein, and Fat Content (%) of 243 South African Fish Meals from Seven Factories in the 1996–1999 Fishing Seasons
The results show that on average only factory 2 completely fulfilled these requirements throughout the years, whereas the others either had too much moisture, too little protein or too much fat in their meals in one or other of the seasons. The worst offender seems to be factory 3, which only complied with the protein content in the 1996/1997 season. This factory was suspected of adding old fish oil back to fish meal, in particular in the 1998 season.
QUANTITATIVE QUALITY INDICES
General
The characteristics that determine the quality of the meals, i.e., their TVBN, FFA and PL levels, as well as their PUFA levels during the years 1996–1999, will be discussed for each individual factory. The results are recorded in Tables . Table compares data for TVBN, FFA and PL contents as indicators of the pre-processing spoilage of each factory during the years 1996–1999, while Table compares the performance of factory dryers as indicated by mean PUFA levels during those same years.
Table 3 Mean TVBN, FFA and PL Levels of 243 Fish Meals from South African Factories in the 1996–1999 Fishing Seasons
Table 4. Mean PUFA Levels of 215 Anchovy, Red-Eye, or Pilchard Meals from South African Factories in the 1996–1999 Fishing Seasons
Factory 1
Over the years, this factory showed a marked decline in quality, as evidenced by the three pre-processing spoilage indicators. The mean TVBN levels of their meals increased from 171±72 in 1996/1997 to 249±78 mg N/100 g in 1999. Similarly, over the same time-span their FFA levels increased from 11.35±3.33 to 17.46±6.38%, and their PL levels decreased from 3.77±1.02 to 3.00±0.71%. These are statistically significant changes (p<0.01 to 0.05). One of the reasons for the deterioration was the addition of excess low quality stickwater to some of the meals. This is a dangerous practice as these meals may be rejected by the buyers on the grounds of objectionable odour (e.g., a TVBN content of 417 mg N/100 g was recorded for one of the meals in 1999). On the other hand, the dryer of this factory performed very well, achieving the highest mean PUFA level of 42.42±1.86% in 1996/1997, and 42.78±2.99% in 1999. This clearly illustrates the independence of the heat abuse and the pre-processing spoilage index as quality indicators for fish meals.
Factory 2
This factory showed a loss in quality of their meals in 1998 followed by a recovery in 1999. This is clearly shown in a significant (p<0.1) decrease in the mean PL level of their meals from 3.38±0.38% in 1996/1997 to 2.75±1.01% in 1998. In 1999 the mean PL level improved again to 3.22±0.80%, but this was not statistically significant (p>0.1). This drop in quality in 1998 was not shown in either the mean TVBN or FFA levels, which had such large standard deviations that no significant annual differences between them could be established. However, as the TVBN and FFA pre-processing spoilage indicators can be manipulated, the PL level seems a better guide.
This factory showed consistently good drying performance with almost constant mean PUFA levels of about 41.0% over the years. This again demonstrates that the quality of meals should be expressed by two independent indices: PUFA levels for heat abuse and preferably PL levels for pre-processing spoilage.
Factory 3
The quality of their meals as expressed in mean PL contents remained remarkably constant, varying from 3.88±0.61% in 1996/1997 to 3.74±0.88% in 1999. However, both mean TVBN and particularly high FFA levels indicated a highly significant (p<0.01) drop in quality in 1998. This is probably due to the addition of extremely low quality fish oil and stickwater to the meals in 1998. In that year the factory produced individual meals with FFA contents of 32.01%, 42.80% and 33.31%, while the respective fat contents of these meals were 13.47%, 14.42% and 17.08%. This seems sufficient evidence to maintain that fish oil with high FFA levels was added to these meals, and this means that, without knowledge of the history of the meal, FFA and TVBN values cannot be trusted as reliable fish spoilage indices. This seems to be confirmed by feeding trials on Atlantic salmon (Salmo salar L) where it was found that TVBN levels of fish meals were not a good indicator of meal quality.Citation23-24 This factory also showed an apparent increase in mean PUFA levels from 40.31±2.52% in 1996/1997 to 42.16±3.56% in 1999. However this increase was not significant (p>0.1).
Factory 4
This was the most consistent factory, with almost constant TVBN, FFA and PL levels. It is remarkable that the mean FFA levels of meals from this factory never exceeded the 10.0% level, which is the maximum level allowed for import into Taiwan.Citation[25] The mean PUFA levels of meals from this factory were also virtually constant throughout the years having a value of 40.39±2.50% in 1999.
Factory 5
The quality of the meals produced by this factory as expressed by their mean PL level significantly (p<0.1) decreased from 3.32±0.45% in 1996/1997 to 2.84±0.80% in 1999. In contrast, the TVBN and FFA values remained almost constant at the 1996/1997 levels of 157±47 mg N/100 g and 11.36±4.49% respectively. The performance of their dryers also remained virtually constant, with PUFA levels in 1996/1997 of 40.03±2.21% and 40.04±2.09% in 1999.
Factory 6
This factory maintained almost constant TVBN and FFA contents at the 1996/1997 levels of 201±64 mg N/100 g and 10.20±2.28% respectively. Its mean PL levels, however, improved significantly (p<0.1) from 3.03±0.60% in 1996/1997 to 3.43±0.46% in 1999, indicating less pre-processing spoilage, a fact not indicated by its TVBN and FFA levels.
In addition, this factory improved its dryer performance from a mean PUFA value of 37.44±2.77% in 1998 to 40.18±2.73% in 1999. This significant (p<0.05) improvement was undoubtedly due to the instalment of a new steam dryer at the beginning of 1999, after the factory management had scrutinised reports from our Institute of PUFA values of meals from all the factories.
Factory 7
This factory showed significant (p<0.05) improvement in relation to fish spoilage by increasing its PL content from its 1998 level of 2.73±0.79% to 3.60±0.67% in 1999. This was not apparent from either its TVBN or its FFA levels, which were respectively 265±113 mg N/100 g and 15.86±3.81% in 1999, and 183±49 mg N/100 g and 14.82±7.88% in 1998.
This factory showed the most significant (p<0.001) improvement in dryer performance in 1998 by increasing its mean PUFA level of 37.52±1.74% in 1996/1997 to 41.09±1.41% in 1998. This improvement was due to the overhaul of their dryer at the beginning of the 1998 fishing season as a result of a report from our Institute. The performance of the dryer dropped slightly in 1999 to a mean PUFA level of 40.26±1.13%, but this level is still very satisfactory.
CONCLUSIONS
It is clearly shown in this report that the pre-processing spoilage quality indices frequently used for fish meals may sometimes be misleading. Both the TVBN and FFA content are dependent on the variable addition of concentrated stickwater and sometimes even low quality fish oil. These indices can therefore be misleading and without knowledge of the history of the meal are not reliable. Ideally, the PL content of a meal should be included to establish whether the TVBN and FFA contents accurately reflect the quality of the meal. This is demonstrated in the survey by the sometimes contradictory results for a PL index and a TVBN or FFA index. In instances like this the PL index must be included to obtain a correct quality assessment. It is also shown that the PUFA level of meals can be used as an excellent guide for detecting heat abuse during fish meal drying. Evidence for this is presented by factories improving their dryer performance significantly by either renewing or overhauling their dryer after receiving a report from our Institute of PUFA levels of all the factories. No evidence was obtained to categorically state that steam (indirect) dryers cause more heat damage to meals than flame (direct) dryers, despite the fact that the residence time in steam dryers is longer than in flame dryers. Overheating of meals seems therefore due to faulty dryers, whether flame or steam. It is therefore important to monitor PUFA levels of meals regularly to ensure that these fatty acids are protected as they are beneficial to animals. They provide them with better resistance to disease, better bone formation and improved fertility.Citation[26]
ACKNOWLEDGMENT
The author wishes to thank Mr. Colin Simmonds for his assistance with the statistical calculations.
REFERENCES
- McCallum , I.M. and Higgs , D.A. 1989 . An Assessment of Processing Effects on the Nutritive Value of Marine Protein Sources for Juvenile Chinook Salmon (Oncorhynchus tshawytscha) . Aquaculture , 92 : 181 – 200 .
- Sandfeld , P. 1983 . Quality Criteria for Fish Meal 221 Copenhagen, , Denmark : Association of Fish Meal and Fish Oil Manufacturers in Denmark .
- de Koning , A.J. , Mol , T.H. , Przybylak , P.F. and Thornton , S.J. 1990 . The Free Fatty Acid Content of Fish Oil. Part II. The Effect of Anchovy Quality on the Free Fatty Acid Content of the Resulting Anchovy Oil and Meal . Fat Science and Technology , 92 : 193 – 197 .
- de Koning , A.J. , Milkovitch , M. and Wessels , J.P.H. 1985 . The Free Fatty Acid Content of Fish Oil. FFA Formation During Pre-Processing Storage of Anchovy at Different Temperatures . Fishing Industry Research Institute Progress Report , 244 : 1 – 5 .
- Lovern , J.A. 1962 . Fish in Nutrition Edited by: Heen , E. and Kreuzer , R. 86 – 110 . London : Fishing News Books .
- de Koning , A.J. 1995 . The Free Fatty Acid Content of Fish Oil. Part IV: Rates of Free Fatty Acid Formation from Phospholipids and Neutral Lipids in Anchovy (Engraulis capensis) Stored at Various Temperatures . Fat Science Technology , 97 : 341 – 346 .
- de Koning , A.J. and Hearshaw , K.D. 1990 . Investigation of the Oxidation and Hydrolysis of Anchovy Lipids in the Manufacture of Anchovy Meal at a Number of Different Factories . Fishing Industry Research Institute Technical Report , 5 : 99 – 103 .
- de Koning , A.J. 1996 . The Quality of South African Fish Meal. Part I. An Investigation of the Quality of Fish Meals Produced by Different Factories in the 1996 Fishing Season . Fishing Industry Research Institute Progress Report , 375 : 1 – 17 .
- de Koning , A.J. and Mol , T.H. 1989 . Lipid Determination in Fish Meal: An Investigation of Three Standard Methods Applied to Stabilised and Non-Stabilised Anchovy Meals at Increasing Stages of Maturity . Journal of the Science of Food and Agriculture , 46 : 259 – 266 .
- Dichmont , M.P. and Stewart , C.P. 1992 . Residence Time of Fish Meal in Direct and Indirect Dryers and the Effect on Quality . Fishing Industry Research Institute Technical Report , 7 : 180 – 185 .
- de Koning , A.J. 1999 . Quantitative Quality Tests for South African Fish Meal. An Investigation into the Validity of a Number of Quality Indices . International Journal of Food Properties , 2 : 79 – 92 .
- Spark , A.A. 1982 . Ethoxyquin in Fish Meal . Journal of the American Oil Chemists' Society , 59 : 185 – 188 .
- 1987 . Standard Methods for the Analysis of Oils, Fats and Derivatives , 7 Oxford, , United Kingdom : International Union of Pure and Applied Chemistry . Method 1.121.
- Bradstreet , R.B. 1965 . The Kjeldahl Method for Organic Nitrogen 239 New York and London : Academic Press .
- Folch , J. , Lees , M. and Sloane Stanley , G.H. 1957 . A Simple Method for the Isolation and Purification of Total Lipids from Animal Tissues . Journal of Biological Chemistry , 226 : 497 – 509 .
- 1988 . Official Methods and Recommended Practices of the American Oil Chemists Society Illinois : Champaign . Method Ca 5a-40
- de Koning , A.J. and Mol , T.H. 1989 . A Convenient Method for the Accurate Determination of Phosphorus in Fish Oils . Fat Science Technology , 91 : 36 – 38 .
- 1975 . Official Methods of Analysis of the Association of Official Analytical Chemists , 12 Washington Method 2.055.
- Bligh , E.G. and Dyer , W.J. 1959 . A Rapid Method for Total Lipid Extraction and Purification . Canadian Journal of Biochemistry and Physiology , 37 : 911 – 917 .
- Christie , W.W. 1982 . Lipid analysis , 2 207 Oxford : Pergamon Press .
- de Koning , A.J. and Evans , A.A. 1991 . Phospholipids of Marine Origin. The Lantern Fish (Lampanyctodes hectoris) . Journal of the Science of Food and Agriculture , 56 : 503 – 510 .
- " Government Gazette ." Republic of South Africa 180 (7105) , pp. 17 . 1980
- Anderson , J.S. , Lall , S.P. , Anderson , D.M. and McNiven , M.A. 1993 . Evaluation of Protein Quality in Fish Meals by Chemical and Biological Assays . Aquaculture , 115 : 305 – 325 .
- Anderson , J.S. , Higgs , D.A. , Bearnes , R.M. and Rowshandeli , M. 1997 . Fish Meal Quality Assessment for Atlantic Salmon (Salmo salar L) Reared in Sea Water . Aquaculture Nutrition , 3 : 25 – 38 .
- Stadler, J. South African Fish Meal Marketing Association [Pty] Ltd, Cape Town, Republic of South Africa, Personal Communication.
- Pike, I.M. Evaluation of the Use of Fish Oil in Aquatic and Land Animal Feeds. Annual Conference, International Fishmeal and Oil Manufacturers Association, Cape Town, South Africa, 18–21 November 1996; Record of Proceedings; 1–23.