Using Argo profiling floats, cruises and mooring data, we reconstructed the dissolved oxygen (O2) dynamics in the Gulf of Lion and the Ligurian Sea, with a focus on the intermediate waters. By applying the CANYON-MED neural network-based method on the large network of O2-equipped Argo floats we derived nutrients and carbonate system variables in the Gulf of Lion and the Ligurian Sea at different depths in the water column and derived trends over the 2012-2020 period. In these waters, the O2 minimum is strongly affected by the intermittent convection process, and the two areas show dissimilar responses to the mixing events. In the absence of deep convection events, the O2-depleted layer tends to spread vertically and intensify even more so in the Ligurian than in the Gulf of Lion. In both areas, over the 2012-2020 period, nutrients increase overall in deep layers, with a concomitant impact on nutrient molar ratios tending towards an increase in P-limitation. Acidification estimates derived in different layers of the water column show an overall increase in dissolved inorganic carbon and a concurrent pH decrease. These trends were strongly affected by convection events slowing down the overall acidification trend.

Key Points

In the absence of deep convection events, the O2-depleted layer spreads vertically and intensifies more in the Ligurian than Gulf of Lion.

Nutrients increase in deep and to a lesser extent in intermediate waters with a decoupling between nitrate and phosphate trends.

Dissolved inorganic carbon increases in intermediate and deep waters with a concurrent pH decrease over the period of study, 2012-2020.

Plain Language Summary

Using multiple observation platforms such as Argo profiling floats (autonomous, free-floating instruments equipped with sensors and profiling regularly in the first 2000 m of the water column), cruises on oceanographic vessels, and moorings (fixed position stations), we reconstructed the dissolved oxygen (O2) dynamics in the northwestern Mediterranean Sea. The intermediate waters are characterized by an O2 minimum strongly affected by the intermittent convection process (due to winter forcings and density changes, mixing of water masses occurs over the water column). We also derived nutrients and carbonate system variables (related to acidification) from a neural network methodology developed for the Mediterranean Sea, CANYON-MED, applied to Argo floats equipped with O2 sensors in different layers of the water column. Over the 2012-2020 period, the derived trends show an overall increase of nutrients in deep layers, with an impact on nutrient limitations. Acidification estimates show an overall acidification. These trends were strongly affected by convection events.