Indice
Farmed Salmon
Nowadays, aquaculture represents one of the most important farming activities in the world with a significant contribution to the global economy and the aquatic foods supply. The control over fish production processes in aquaculture is more efficient than in capture fisheries and for this reason, aquaculture products could be integrated into the fish food supply chains in a better way.
The latest report on the State of World Fisheries and Aquaculture (FAO, 2018) indicates that during 2016, global aquaculture production reached a total amount of more than 110 million tons, including food fish (finfish, mollusks, crustaceans and other aquatic animals) and aquatic plants valued in around USD 243.5 billion. Among the farmed food fish, the production of finfish species reached 54.1 million tons that represented 57 per cent of the total value (USD 138.5 billion). Finfish farming is, in fact, an essential source of income and livelihood for many people around the world, and is considered a significant driver of economic development in different countries.
Finfish farming is the most diverse aquaculture subsector and the number of the different finfish species that are farmed represented over 90 per cent of the total production in 2016, making fish farming the most significant form of aquaculture all around the world. Atlantic salmon (Salmo salar) is the most heavily traded species in the world, and one of the main finfish farmed species. Even though Salmo salar represents only 4 per cent of the finfish species produced in world aquaculture and its economic contribution is significantly higher than the rest of the farmed fish.
Quality management systems in the seafood industry, including aquaculture, is critical to cope with the need to satisfy consumer requirements and safety as essential components of any food-producing sector. One of the main advantages of aquaculture is the degree of control over the production process, which would have a direct impact on the quality prediction models accuracy. Considering this aspect, if aquaculture practices are optimized to guarantee a quality stable product, the performance of these tools will improve significantly.
Freshness and spoilage profile
Fish spoilage is the consequence of many autolytic reactions and bacterial metabolites production. They are easily detectable through the assessment of the appearance and odour perception, moreover could provide enough and complete information about freshness degree of fish with high accuracy primarily when ice has used to extend the shelf life [1]. The best way together to establish the quality of fresh fish continues to be sensory analysis.
In this study, a trained sensory assessors panel with experience to evaluate seafood was used to carry out three different sensory methods: Torry [2], UE Scheme [3] and Quality Index Method [4]. The aim was to establish in an objective way, and with high precision, the freshness degree for each sensory attribute assessed in the fish. Also, analytical assays (physico-chemical and microbiological count) were carried out, simultaneously, to compare with sensory results.
Fig.1 shows the evolution over time of sensory attributes key for quality and freshness for farmed salmon stored in ice. As could be seen, gills and eyes were the most critical aspect due to their notable changes over time. Spoilage profiles for salmon based on freshness (%) are shown in Fig.2.
All of them agree that gills were the most vulnerable aspect because it was the worst valued characteristic in the salmon. Eyes colour and shape (convex) had an important role too to indicate a marginal quality (marked with dotted red lines in the graphics). Conversely, shiny and clear mucus on surfaces of the skin in the fish was the better-assessed attribute for a long time. From 10 days, a considerable loss in quality was observed; however, at 12 days, a fall in overall freshness degree indexes was detected.
Development of predective tools for quality aspect
Chemical and microbiological methods for assessing freshness and quality in fish are useful for research or product development. Still, they are not practical for routine use due to its costs and the time required to obtain results. Moreover, most of the physical analyses, generally considered as rapid methods, cannot be used with those fish subjected to different treatments (for example thawed) or bruised from mechanical damage due to unreliable results.
Based on the above, the development of models to estimate quality fish (freshness degree or number of days postharvest) through the combination of results obtained from simple routinely analysis techniques with a multifactorial statistic could represent an opportunity to provide an integral criterion based on methods of different natures that considering the most relevant aspects of fish quality.
Implementation of predictive models contributes to the improved control of food safety and spoilage, e.g., by quantifying the effect of storage and distribution on microbial proliferation via the HACCP system. Predictive microbiology has been accepted as a tool to define the safety of food products in the food industry [5]. In the short term, the demand for such tools will increase, and significant jump would happen in the field of predictive modelling in foods, primarily predictive modelling in the food chain, quality and safety management, modelling of food processes, sampling and experimental designs/plans [6].
In this work, two predictive models were developed:
1) Ice Storage Time (IST) which denotes the elapsed time (in hours) from the slaughter and placement on the ice of the fish to a defined sampling time over its shelf life.
2) Reference microbial load (MVC) from Mesophilic Viable Count (log CFU/g).
Models were designed considering physico-chemical, microbiological and sensory parameters that previously demonstrated to have a good correlation among them, and they were able to use as reference values for future estimations. The experimental guideline established to design these models were: a) Origin: Farmed salmon, b) Size and weight: 20 +/- 0,30 kg with 6-7 as commercial size, c) Storage temperature: 0 ºC, d) Variation rank for Storage temperature: +/-1.5 ºC, e) Internal Temperature in the Fish: ≤ -1ºC, f) Pre-treatment applied: Gutted and g) Ice replacement frequency: daily.
A Partial Least Square Regression (PLS-R) was used obtaining a general Plot as could be seen in Fig.3. This graphic demonstrated that all sensory methods were very well correlated among them as shows a weighted coefficient regression -BW- (Fig.4). Variables inside the first ellipse (Hotelling test) have less than 50% of weight to describe the variance of this model; therefore, they are considered of little importance. PLS regression provided general and complex models with limited practical application.
Due to the above, single models with high capability to prediction has to be accomplished because the explanation of a response variable through few predictors (variables used for estimation) is an excellent indicator to obtain simple models that would exhibit high accuracy for future evaluations.
IST = -27.79 TM + 9,83 MVC + 384.33 (1)
MVC = -0.073 SFI (UE) – 0.015 TM + 7.81 (2)
Where IST is Ice Storage Time (in hours), TM is a value for electrical conductivity (Standard scale) provided by torrymeter® equipment (Distell™), MVC is Mesophilic Viable Count (log CFU/g) determinates by standard ISO 4833-1:2013 [7] and SFI (UE) is Sensory Freshness Index using European Official Method. Regression Coefficients Graphic (Fig.5). Eq.1 showed that TM and MVC were variable with major weight and minor standard error. After cross-validation Eq.1 had a slope=0.95 with a R-square=0,95; and Eq.2 showed a Slope=0.95 and R-square:0.96.
Due to the relevance of microbial count as a requirement for food safety, a Response Surface graphic (Fig.6) for Eq.2 was generated. The landscape obtained allow to indicate confidence intervals for future estimations in any salmon samples. This interval comprised between 1.4 and 5.3 log UFC/g. According to Regulation (EC) 2073/2005 [8] a bacterial count of 6 log UFC in fresh meat or fish could be considered as a critical limit to considerate food as “suitable to consumption”. The orange zone in the Fig.6 represents the marginal quality for fresh salmon
Pratical validation and prospects
Both Predictive tools designed (Eq.1 & Eq.2) were tested with fresh salmons from market. Theory values estimated using these equations were compared with results from experimental analyses did in those fish. The Fig.7 shows practical validation test for models assayed. Eq.1 achieved an accuracy of 92% with a regression coefficient of 0.90. If a transformation from hours to days (dividing in 24 h) is applied to estimated IST, so Accuracy Factor (Af) improves to achieve a 95%. In the case of Eq.2, it demonstrated an Af= 87% with r= 0.92 especially sensitive in a logarithmical phase of microbial growth (> 2 log UFC/g).
In the case of MVC, a complementary external validation was applied through comparison of results using a predictive tool designed with a data provided by recognized software frequently used for shelf life study (Food Spoilage and Safety Predictor –FFSP- Version 4.0 developed by DTU aqua -2019-).
The software was configured as follow Microbial Spoilage Model, H2S producing Shewanella, specifically Fresh Seafood stored in air. The initial count was 1.48 log UFC/g and temperature in average 0.07ºC (32.12ºF). According to the above FSSP predict a shelf life of 14 days while tool designed (Eq.1) set a time of 12 days. However, bacterial growth curve estimated by FSSP (Fig.8A) demonstrate that at day 12 count was more than the recommended limit -6 log UFC- [8, 9].
A comparative test was made for FSSP count and MVC values (Eq.2). An accuracy of more than 85% was achieved between them. However, after 5 log UFC/g, predicted counts were infra valued by Eq.2 concerning FSSP results (Fig.8B). The above fact could be due to that many factors (physico-chemical, sensory and microbial) were taken into account to design prediction tool, so it resulted with a more rigorous criterion which reduces the possibility of offering dangerous and defective foods to the consumers.
Predictive models designed in this work represent a novel and powerful tool for the fishing industry, whose competitive advantage is based on the combination of sensory evaluation with routine analytical methods. They resulted in economic option, easy to apply and able to achieve fast and highly reliable estimations that reduce times required for decision-making in the fishing industry company.
1) Olafsdottir, G., Martinsdottir, E., Oehlenschlager, J., Dalgaard, P., Jensen, B., Undeland, I., Mackie, I. Henehan, G. Nielsen, J. & Nilsen, H. 1997. Methods to evaluate fish freshness in research and industry (Review). Trends in Food Science & Technology. 8, 258-256.
2) SeaFish. 2010. Sensory assessment scoresheets for fish and shellfish – Torry & QIM. Research & Development Department. UK. www.seafish.org.
3) Council Regulation (EC) No 2406/96 of 26 November 1996 laying down common marketing standards for certain fishery products OJ L 334, 23.12.1996, p. 01–15.
4) AZTI. 2008. Frescura del pescado: guía visual para su evaluación. AZTI-tecnalia. Alimentatec. Bilbao, Spain.
5) Van Impe, J. Vercammen, D. & Van Derlinden, E. 2013. Toward a next generation of predictive models: A systems biology primer. Food Control. 29 336-342.
6) ICMPF. (2011). International Committee of Predictive Modelling in Food. Home. http://www.icpmf.org
7) ISO 4833-1:2013. Microbiology of the food chain — Horizontal method for the enumeration of microorganisms — Part 1: Colony count at 30 degrees C by the pour plate technique. Genève, Switzerland.
8) Commission Regulation (EC) 2073/2005. 15 November 2005 on microbiological criteria for foodstuffs (Text with EEA relevance). Official Journal of the European Union.
9) IFST. (1999). Development and use of microbiological criteria in foods. London: Institute of Food Science & Technology (UK).
Dr. José Antonio Beltrán Gracia
R&D project manager. PhD UNIZAR (1988), He has a wide experience in R&D for more than 30 years he has managed R&D projects in the agrifood sector at International, European and National level. He is Full Professor of food science and technology ay University of Zaragoza. Currently performs duties as R&D Project Manager at the AgriFood Institute of Aragon (IA2).
Dr. Juan B. Calanche Morales
Science and technology Researcher in IA2 at University of Zaragoza, PhD. (UNIZAR, 2015) and MSc. (UDO, 2009). He has a wide knowledge in R&D projects and more than 20 years of experience teaching at Universities of America and Europe in the field of Fish and seafood technology as well in sensory analysis and consumers’ behaviour.
Dr. Adrián Jesús Hernández Arias
Adrián Hernández received his Ph.D. in Aquatic Biosciences from the Tokyo University of Marine Science and Technology in 2005. Currently working as Associate Professor at the Catholic University of Temuco, Chile. He has a rich experience in aquaculture and in the course of his research activities he had dealt on different aspects of fish nutrition.
