FISHBOOST results: How can we select fish to improve fillet yield?
Marc van de Putte (INRA)
Selecting for fillet yield (percent fillet, as a proportion of body weight) is extremely important for species which are sold processed. Using real data from sea bass, sea bream and rainbow trout, as well as simulation approaches, INRA, Ifremer and Sysaaf showed that it is indeed possible to increase fillet yield by selection, to a modest but economically significant level, ranging from 0.4 to 1% per generation.
Why is it important to select for fillet yield ?
The objective of aquaculture is to produce edible fish flesh for human consumption. While some species are sold whole, processing is developing, as consumers more and more require easy to prepare fillets. When the commercial product is fillet, improving fillet yield (the ratio of fillet weight to body weight) improves the economic efficiency of production, as more valuable fillet can be sold from the same amount of fish produced at a farm. This also improves the environmental efficiency of farming, as effluents from fish farms are grossly proportional to the quantity of fish produced: thus, producing more fillet from the same amount of fish produced reduces the environmental impact per kg of fillet produced. In addition, less fish waste has to be disposed at the slaughterhouse. Fillet yield is thus a typical production efficiency trait.
Why is it difficult to select for fillet yield ?
There are two main reasons that render fillet yield a difficult trait to select for:
Which approach did we choose ?
We first identified traits that could describe the biological reality of fillet yield but would be less correlated than fillet weight and body weight. Our choice was to propose that a fish is the sum of two fillets and a “waste” component representing head, bones, viscera and fins.
The objective of aquaculture is to produce edible fish flesh for human consumption. While some species are sold whole, processing is developing, as consumers more and more require easy to prepare fillets. When the commercial product is fillet, improving fillet yield (the ratio of fillet weight to body weight) improves the economic efficiency of production, as more valuable fillet can be sold from the same amount of fish produced at a farm. This also improves the environmental efficiency of farming, as effluents from fish farms are grossly proportional to the quantity of fish produced: thus, producing more fillet from the same amount of fish produced reduces the environmental impact per kg of fillet produced. In addition, less fish waste has to be disposed at the slaughterhouse. Fillet yield is thus a typical production efficiency trait.
Why is it difficult to select for fillet yield ?
There are two main reasons that render fillet yield a difficult trait to select for:
- Measuring individual fillet yield requires the killing and processing of fish. Then, a fishcannot be used as a parent for the next generation once it has been measured. This can be solved by “sib selection” approaches, where fillet yield is measured on a number of fish per family, and the remaining live fish of the same families are selected according to the performance of their slaughtered sibs.
- Fillet yield is a ratio of two traits, fillet weight and body weight, which are very highly correlated (typically a correlation of 0.97-0.99, close to the theoretical maximum of 1). Based on this, many breeders consider that it is impossible to improve fillet weight independently of body weight, and thus that selecting for fillet yield will be inefficient. Additionally, ratio traits possess mathematical properties that complicate their use as selection criteria: a higher fillet yield can come from an increase in fillet weight for a given body weight, or a decrease in body weight for a given fillet weight, which obviously does not reflect the same biological variation. Different selection indices combining fillet weight and body weight have been proposed to try to solve this issue.
Which approach did we choose ?
We first identified traits that could describe the biological reality of fillet yield but would be less correlated than fillet weight and body weight. Our choice was to propose that a fish is the sum of two fillets and a “waste” component representing head, bones, viscera and fins.
The results show that the (phenotypic) correlation of fillet weight and waste weight was lower, in the range 0.82-0.93. Using family data, we also estimated the heritability and genetic correlations of fillet weight and waste weight in five populations of three species (sea bass, sea bream and rainbow trout). With these parameters, using genetics theory, we could simulate in silico populations of fish (using equations and probabilities in computers), and select them for improved fillet yield over ten generations, using nine different selection indexes that had been proposed to be used for improving fillet yield. Then, we compared the fillet yield of the different populations to evaluate the genetic improvement obtained using the different indexes in the different species.
What are the results?
We showed that choosing the fish with the best fillet/waste ratio, the best fillet yield, or the best residual fillet weight[1], was efficient at improving fillet yield by 0.4 to 1% per generation (0.66% on average) depending on the population and species. Selecting solely for fillet weight or body weight gave 55-65% less response on fillet yield than selection on the previous indices, while selecting against waste weight could even lead to diminished fillet yield. Thus, we showed that better gain was achievable by selecting for fillet yield or alternate measurements of it (such as residual fillet weight or fillet/waste ratio).
Where do we go from here ?
The simulations gave interesting results, but are tedious to perform, and we need to evaluate the possibility to predict genetic gains from simple prediction equations using the genetic parameters (heritability, correlations) of the selection indexes tested. This work is ongoing. Additionally, in several species in Fishboost and with industrial partners, indirect predictors of fillet yield using morphological 1,2,3D recordings are being developed in carp, sea bass, sea bream and rainbow trout. These may further facilitate selection for fillet yield by enabling measurements on live fish, and thus selection of individuals taking into account their own performance. This would make selection more effective and could reduce the need for sacrificing fish. Finally, an experimental selection for fillet yield in rainbow trout has been conducted with the industry partners Milin Nevez and Sysaaf. Results of selection response are expected by the end of 2018, and we hope it will confirm the practical feasibility of such an approach, and the real genetic gains to be expected in industry conditions.
[1] which represents the difference between the fillet weight of a fish and the average fillet weight of fish with similar body weight in the population
