the project

Fisheries-independent abundance indicators for bluefin and tropical tuna

The general objective of BLUEMED is the development of new methods for the derivation of fisheries-independent abundance indicators for bluefin and tropical tuna.

Tropical and temperate tuna, whose global annual catches correspond to around 5 million tons, constitute a major commercial group of species worldwide. Since the 2000s, in order to face the risks of overfishing, catch limits have been introduced for the Atlantic bluefin tuna. More recently, certain tropical tuna species, such as yellowfin tuna in the Indian Ocean, have also been subject to catch restrictions because of their fragilization induced by the strong fishing pressure. In order to guarantee a sustainable management of these stocks, indicators providing information on their abundance are necessary. However, since tuna occupy very large habitats, obtaining these indicators is a major operational and scientific challenge. Indeed, the limitations of catches and the rapid technological development of industrial fisheries make catch data difficult to use to provide information on the abundance of tuna. The BLUEMED project aims to contribute to the development of new methods for deriving abundance indicators for bluefin and tropical tunas. These indicators, based on fisheries-independent data, aim to provide a more reliable estimate of tuna populations and thus contribute to their sustainable management.


The BLUEMED project aims at contributing to the derivation of novel fisheries-independent abundance indices for bluefin and tropical tunas. Despite their different habitats, ecology and behavior, the project relies on a common scientific approach that is applicable to these two groups of species. Such an approach combines observations at the scale of the schools/aggregations with behavioral analyses conducted at the scale of the individuals, to produce novel abundance indices.

For several years, aerial surveys have been carried out by scientists in the North-Western Mediterranean Sea, in order to count the number of schools of bluefin tuna feeding on the surface. At the same time, scientists can now dispose of new acoustic data, collected by echo sounder buoys, providing information on the biomass of tropical tunas associated with fish aggregation devices (FADs) deployed by industrial purse seiners. These observations (bluefin tuna present on the surface and tropical tunas associated with FADs) can be used for the construction of alternative abundance indicators. However, they are strongly linked to the behavior (surface or association) of tuna. The methods developed within the BLUEMED project aim to take these behavioral factors into account, by characterizing the states where tuna are detectable (on the surface or associated with FADs) via electronic tagging techniques. The key idea is to couple this information collected at the individual level through tagging with observations obtained on the schools/aggregations via modeling approaches, to link the components of the population observed (by aerial or acoustic monitoring) to the total population of tuna present in the study area.

Main results

Bluefin tuna
A total of 24 tags were deployed between 2015 and 2016 on late juvenile/young adult ABFT (Bauer et al 2019). The data analysis of ABFT vertical movements demonstrated that the proportion of time that ABFT spent at the near surface layer (0-20m) during daytime shows a seasonal pattern, increasing during spring, peaking in summer and decreasing with the decline of thermal stratification of the water column (Bauer et al 2019). The duration of the residence times of ABFT in the visible surface (0-1m) during daytime was found to last less than 2 minutes, independently on the season. However, the rate of individual surfacing events was correlated with the proportion of time that ABFT spent in the near surface layer (0-20m). Our results demonstrated that computing an index of ABFT abundance on aerial survey is meaningful in the Gulf of Lion from August to October, but these estimates should account for the seasonality of the time that ABFT spent at the near surface. The empirical model coupled to aerial surveys data demonstrated a high variability of the observed number of ABFT schools (Perez et al. 2017). This variability could not be fully explained through the seasonal changes in the vertical behavior of ABFT, indicating that further components (social or environmental) should be accounted for. Also, the comparison between the observed schools and the model outputs revealed spatial aggregations of the observed ABFT schools (Pérez et al. 2017).

Tropical tuna
The processing of the M3I echosounder buoys data through the random forest algorithm showed an accuracy of 75 and 85 % for the recognition of presence/absence of tuna aggregations under DFADs in the Atlantic and Indian ocean, respectively (Baidai et al 2018, Baidai et al 2019). Similarly, accuracies of 85% could also be found for the M3I+ echosounder buoys in the Indian ocean (Diallo et al 2019). The assessment of the size of the tuna aggregations showed lower accuracies for both oceans and buoy models. Using such presence/absence classification model, we were able to characterize for the first time the dynamcis of a FAD aggregation in the Atlantic and the Indian Ocean respectively (Baidai et al 2019), a key input for the field-based models estimating tuna abundance. Analysis of the resulting time series evidenced that the FAD aggregations have different dynamics depending on the ocean (Baidai et al 2019). The analysis of CRT and CAT from acoustic tagging data collected on both drifting and anchored FADs revealed a species-dependent associative behavior (Rodriguez et al 2017, Tolotti et al 2019), with skipjack tuna that generally spends less time associated to the FADs with respect to bigeye and yellowfin tuna. Survival curves of CRTs and CATs were best fitted by exponential models, implying a memory-less process of association (Rodriguez et al 2017, Tolotti et al 2019). Finally, the role of the inter-FAD distances on the CRTs and CATs could be evaluated, with smaller absence times for FAD arrays presenting higher FAD densities (Pérez et al 2019). The parameters of the correlated-random walk model, namely the sinuosity of the trajectory and the orientation radius (i.e. the distance at which tuna can detect the FADs) were calibrated using the field data collected on different arrays of anchored FADs. The model was then run on different FAD-array densities and the relation between the inter-FAD distances and the theoretical relation between the time spent by tuna out of the FADs were derived (Perez et al 2019). Finally, the empirical model to obtain an abundance index for tropical tuna based on its associative behavior (Capello et al 2016) was applied on the case study of acoustically-tagged yellowfin tuna (Thunnus albacares) in Hawaii. For the first time the ratio between the associated and the total population could be derived, demonstrating that in the FAD array of Hawaii, during the tagging period, around 70% of the local population of yellowfin tuna was associated with the FADs (Capello et al 2016).


The project is funded by the French National Research Agency ANR (call @RAction -2014