Assessing the variability and discriminatory power of elemental fingerprints in whiteleg shrimpLitopenaeus vannamei from major shrimp production countries

The origin of seafood is important to consumers and importers alike. Element profiling has been brought into play as a potential tool to improve seafood traceability. Results are presented here that demonstrate the potential of a traceability database for shrimp aquaculture products from predominant producing countries.

Most of the production of the whiteleg shrimp, Litopenaeus vannamei, is centered in a few countries in Latin America and Southeast Asia, including Ecuador, Thailand, Vietnam, India, Indonesia, and China. While China is the world’s leading producer of whiteleg shrimp, most of its production is for domestic consumption, while the rest of the countries listed above largely produce shrimp for international trade.

“The largest destination markets are Japan, USA, EU, Korea, and somewhat surprisingly because of its high domestic production, China (UN, 2020).”

Despite the importance of seafood to the human food supply chain, seafood is plagued by labeling fraud that is well documented in the United States and the European Union. Mislabeling is used to manipulate prices, improve marketing, replace farmed species with wild species, and meet quota requirements of processors.

In aquaculture shrimp products, transshipping has been an issue in the past, resulting in incorrectly labeled country of origin among retail products. Claims-making such as sustainability related claims in certifications also rely on traceability and accurately accounting for the origins of products. Several analytical methods have been explored as traceability tools including fatty acid analysis, stable isotope analysis, DNA profiling.

In addition, new technologies such as block chain and radio frequency identification have been implemented in seafood supply chains.

“Element profiling has been identified as another tool to increase traceability in seafood products, and has shown promise as a tool for discriminating geographic origins in shrimp.”

Ecuador has been an important producer of shrimp, since the 1970’s. To date, only one study evaluating elemental profiling as a traceability tool has included Ecuador (Smith & Watts, 2009), so the potential to differentiate shrimp from Ecuador and individual countries in Southeast Asia using this method is relatively unexplored.

Thus, the objective of the study presented here was to understand the potential of elemental profiling to differentiate between the major shrimp producing countries in Southeast Asia and Ecuador.

Methods

Whiteleg shrimp were collected from five countries: Ecuador, India, Indonesia, Vietnam, and Thailand. Utilizing the knowledge of the authors of the study who are experts in the aquaculture industries in their respective countries, collection sites were chosen in each country to be representative of the major production areas of shrimp (Figure 1).

All farms sampled utilized similar production practices (semi-intensive or intensive operations), and all shrimp were fed pelleted shrimp feeds. Samples were collected from farms in the same manner (Li et al., 2017).

Shrimp were prepared for storage until element analysis in the following manner. Shrimp were brought to a laboratory in the country of sampling and deheaded and peeled. The peeled tails were then dried in commercially food dehydrators at 50 ºC until the tissue reached constant mass, at least 12 h.

“Upon drying, the samples were stored frozen until shipment for analysis. In preparation for digestion, a sub sample of 3-5 shrimp from each sampling event were freeze dried overnight to remove any residual moisture.”

The digestion of the shrimp tissue was done following an adaptation of EPA, method 200.8 (US EPA 1994) for solid materials (Environmental Express, 2018). A NexION 350 d ICP-MS (PerkinElmer Inc., Waltham MA USA) was used to conduct the elemental analysis for this study.

Both standard and kinetic energy discrimination mode are used during analysis. Several steps were taken to ensure consistency between runs and within runs. Elements were removed from the statistical analysis if more than 20% of the samples were below detection limits.

The mean and standard deviation of element concentrations of samples are reported by country of origin. A one-way MANOVA type analysis was conducted utilizing a test statistic described in Friedrich and Pauly (2018), which is statistically robust to heteroscedasticity, does not rely on multivariate normality, and can be used with high dimensional data.

Results The MANOVA style test that was conducted found a statistical difference among the elemental profiles of the five countries in this study (MATS = 947.047, p < 0.001). A summary of the elemental concentrations in shrimp tail muscle tissue and the univariate statistical tests that followed the MANOVA with the MATS statistic are presented (Table 1).

Overall, significant differences were detected in 28 out of the 33 elements reported. In 15 of the 28 elements where statistical differences were noted, and in 17 elements overall, shrimp from Ecuador had the highest concentrations on average (e.g., Al, Fe, Li, Sr).

Out of the 28 elements with significant differences, Ecuador had unique group membership in seven of the elements (Al, As, Cs, Er, Fe, Gd, Sr). Vietnam and Thailand tended to belong to the same post-hoc groupings, being in the same group in 25 out of the 28 elements where significant differences were detected.

Overall, the random forest classification model obtained across validated accuracy of 91% (Table
2). The out-of-bag (OOB) accuracy for the same model was 89%. The most accurate results were for Ecuador (97%), while the least accurate results were for Thailand (80%), which had samples misclassified as Ecuador, Indonesia, and Vietnam.

While Thailand and Vietnam had lower accuracies than those for Ecuador, they were most frequently
misidentified as one another, with 5/48 samples from Thailand being identified as being from Vietnam and 5/53 samples being identified from Vietnam being Identified as Thailand. Indian shrimp were misidentified as Ecuadorian (2/61), but also there were samples misclassified as Thailand and Vietnam.

The most important element in the classification model was Cs, followed by As and Se (Figure 2). The least important elements were Cr and Pb.

The canonical discriminant analysis (CDA) reduced the dataset to four canonical variables. In the first two dimensions, which account for ~78% of the total variation, Ecuador and India separate from the three countries in Southeast Asia (see Figure 3a).

Indian shrimp separate from shrimp from the other countries along the second canonical variable, while Ecuador separates along the first canonical variable. Elements with strong factor loadings in the first canonical variable include Al, As, Sr, while variables highly correlated with canonical variable 2 are Al, Co, and V (Table 3).

While Thailand, Vietnam, and Indonesia overlap in the first two dimensions of the CDA, Indonesia separates from Thailand and Vietnam in the first and third dimension, which is highly associated with the elements Ca, Cs, and V (Figure 3b).

Discussion

It was conducted a discriminant analysis with a random forest model with farmed whiteleg shrimp from major exporting countries with element concentrations from shrimp muscle tissue. The basis for elemental profiling is that patterns of heterogeneity exists in samples from different predefined groupings (e.g., species, production method, geographic origins).

In this dataset, 28 out of 33 elements reported showed statistical difference among groupings. In general, samples from Ecuador were more mineralized than samples from other countries. This may be because of the freshwater runoff that comes from the Andean mountains in Ecuador being higher in minerals than the lowland regions of Southeast Asia.

Overall, the accuracy of the discrimination procedure with 328 samples was 91%. This compares favorably to other studies that have been done with regards to identifying geographic location in cultured shrimp. Li et al. (2017) covered a subset of the geographical areas in this study, and had an overall accuracy of 97%, but with less samples and a more limited scope.

Gopi, Mazumder, Sammut, Saintilan, et al. (2019) was able to identify black tiger shrimp Penaeus monodon to regions in Australia and Southeast Asia with 98% accuracy, however the results of that study may be confounded with the mixture of wild capture and cultured shrimp in their sample, as culture vs. wild capture has been successfully delineated by elemental profiling in other cases (Anderson et al., 2010; Varra et al., 2019).

Seafood traceability is a growing concern for producers, retailers, and consumers. Consumers are becoming more aware of the impact of their purchasing power and are therefore willing to buy products that are perceived as “sustainable,” even if they have a higher price tag (Roheim et al., 2011).

In many cases, retailers use certification schemes as a proxy for sustainability and claims making, as it removes them from the process of validating the claims about the product but allows them to project environmental consciousness.

In aquaculture, both the Aquaculture Stewardship Council and the Global Aquaculture Alliance Best
Aquaculture Practice (BAP) shrimp standards have chain of custody and traceability requirements. Findings add evidence towards the possibility of using elemental profiling as a traceability tool.

Here, shrimp samples from five countries that are the world’s leaders in shrimp exports were successfully discriminated based on country of origin. This is also a considerably larger sample than other similar studies and is not confounded by any species or cultured vs. wild artifacts in the data.

Moreover, study shows that shrimp sampled from Ecuador likely have unique element profiles compared to shrimp in Southeast Asia, which has been the focus of previous efforts in elemental profiling in shrimp.

Conclusion

Overall, this work shows that with more robust sampling (sampling in more locations and more samples within each farm), a database could be generated to discriminate geographic origins of shrimp from the world’s leading shrimp exporting countries.

The authors sought to classify shrimp to country of origin to five of the largest shrimp exporters in the world. The potential of elemental profiling as a traceability tool in practical applications is still relatively unexplored.

With an overall accuracy of 91% using a random forest classification model, the results show that farm-raised shrimp can serve as a database for potential traceability applications, whether for private retailers or government agencies.

This is a summarized version developed by the editorial team of Aquaculture Magazine based on the review article titled “ASSESSING THE VARIABILITY AND DISCRIMINATORY POWER OF ELEMENTAL FINGERPRINTS IN WHITELEG SHRIMP LITOPENAEUS VANNAMEI FROM MAJOR SHRIMP PRODUCTION COUNTRIES)” developed by: ROBERT P. DAVIS AND CLAUDE E. BOYD – Auburn University; RAVIBABU GODUMALA and AVANIGADDA B. CH MOHAN- Seafood Solutions; ARTURO GONZALEZ World Wildlife Fund, Guayaquil, Ecuador; NGUYEN PHUONG DUY – World Wildlife Fund for Nature, Hanoi, Viet Nam; PANDE GDE SASMITA J – Udayana University; NUR AHYANI – World Wildlife Fund, Jakarta, Indonesia; OLGA SHATOVA and JOSHUA WAKEFIELD – Oritain Global Limited; BLAKE HARRIS and AARON A. MCNEVIN – World Wildlife Fund; D. ALLEN DAVIS – Auburn University.
The original article was published on OCTOBER, 2021, through FOOD CONTROL.
The full version, including tables and figures, can be accessed online through this link: https://doi.org/10.1016/j.foodcont.2021.108589.