Climate-driven feedbacks
Counter-intuitively, the increased NPP and terrigenous inputs do not translate into an increased export flux (Fig. 4). In line with the usual paradigm that changes in the lower trophic level cascade up the food chain35, we find that more NPP is generally accompanied by increased zooplankton grazing both in terms of carbon and nitrogen (C: +143%, N: +13%, Fig. 2). This leads to greater losses of organic detritus from plankton to the surrounding seawater (C: +59%, N: +34% for phytoplankton; C: +113%, N: +15% for zooplankton), which adds to the additional organic detritus directly discharged from land. Climate change does not only stimulate NPP and zooplankton grazing, but also instigates significant alterations in other crucial biogeochemical processes. Among those processes, remineralization is one of the most strongly affected (Fig. 2, 5A). As a metabolic process, remineralization is primarily governed by temperature36 (see methods) and consequently exhibits a high sensitivity to warming37,38. Here, we quantify that carbon and nitrogen remineralization increase by 55% and 44%, respectively, due to climate-driven warming (Fig. 2). This increase in remineralization is the fastest in the central AO, in the Eurasian basin in particular (Fig. S1).
We find that changes in carbon and nitrogen export fluxes differ in magnitude and direction (C: +8.5%, N: -0.5%; Fig. 2, 5BC). Both fluxes, however, experienced a change-point in the 2010s (see methods, Fig. S2), suggesting that most of the increase already occurred, while they are expected to stagnate or even decrease in the future compared to present levels. This can be explained by the fact that remineralization increases faster than the ability of the AO to export organic detritus, as the remineralization efficiency (defined as the ratio between the remineralization rate and the export flux) increases by up to 40% (Fig. 6A). This positive feedback loop (Fig. 4B) directly reduces the export efficiency (C: -38%, N: -30%, Fig. 6BC). Such results are confirmed by the analysis of other IPCC-like models (Fig. S3).
Only a small share of the exported organic detritus is sequestrated, as the transfer efficiencies remain low (C: 13.9%, N: 13.3%, Fig. 2), and those numbers are decreasing with climate change (C: -1.8%, N: -4.8%, Fig. 2). Below the surface layer, plankton respiration and remineralization (mainly driven by bacteria39) dominate over photosynthetic processes (Fig. S4). Notably, we observed a mesopelagic end-of-century warming of the AO that surpassed ~1.8 times that of the global ocean (Fig. 1F, SSP3-7.0). Mesopelagic oxygen concentrations decrease with increasing mesopelagic temperatures (Extended Fig. 2B). The decrease is attributed to biological processes since changes in physical processes act to increase oxygen concentrations through a slight increase in vertical mixing (Fig.1E) and a decrease in stratification (Extended Fig. 2A). Our findings thus reveal heightened mesopelagic respiration processes (i.e. remineralization) in the AO due to accelerated warming that extends well beyond the surface layers.
Terrigenous inputs driven feedbacks
Remineralization is not only sensitive to temperature, but also to the size (and quality) of the organic matter reservoir 40–42. Our results show that an increase in terrigenous inputs substantially stimulates NPP and net nitrogen assimilation (+21.4% and +33%; Fig. 2, 3AB). Similar to the climate-driven response discussed above, terrigenous inputs cause escalating losses of detritus emanating from both phytoplankton (C and N: +18%, Fig. 2) and zooplankton (C: +25%, N: +26%), expanding the size of the organic pool. Consequently, water-column remineralization provides the largest nitrogen mass increase among biogeochemical processes with +6.5 TgN (+18%, Fig. 5A, Extended Table 2), followed by the benthic remineralization with +5.5 TgN (+44%). Terrigenous inputs also represent a +11% and +31% increase in pelagic and benthic carbon remineralization (Fig. 2), which relative to the large size of the flux, represents the major biogeochemical source of inorganic matter.
Surprisingly, terrigenous inputs seem to increase remineralization faster than organic matter production and export, also fueling the positive feedback loop (Fig. 4B). The increase in export fluxes remains limited (C: 9%, N: 7%; Fig. 2, 5) compared with NPP (+21.4%) and remineralization (+18%). This may be explained by the fact that remineralization is influenced by the size of the organic pool and temperature, whereas NPP is limited by nutrients. As a result, the remineralization efficiency increased by about +10% with terrigenous inputs, resulting in a decrease of the export efficiency (C: -10%, N: -24.4%; Fig. 5B). This terrigenous-stimulated change is substantial, as it had the same order of magnitude as the changes induced by about one century of climate change. Through the same mechanisms, terrigenous inputs decreased the transfer efficiency by about -1.5% (Fig. 2). The decline in both export and transfer efficiencies reflect a decline in the BCP efficiency.
Terrigenous carbon inputs can also affect the AO carbon sink through the release of dissolved inorganic or organic carbon. The organic carbon is eventually remineralized and adds up to the dissolved inorganic carbon, whose concentration then increases at the surface. The partial pressure of CO2 in the surface ocean increases, which, along with sea ice loss, affects air-sea CO2 exchanges. In fact, our results indicate terrigenous inputs are responsible for intense CO2 outgassing at the pan-Arctic scale, with about 32.7 ± 3.3 TgC yr-1 for the 2090s (Extended Fig. 5), more than 50% of Spain's CO2 emissions in 2021 (equivalent to 120 Gt CO2 yr-1)43. Most of the outgassing occurs on continental shelves, supporting recent regional findings44,45. Terrigenous inputs therefore reduce the AO’s carbon sink by releasing large quantities of CO2 to the atmosphere The total amount of CO2 released to the atmosphere due to terrigenous inputs fluctuated around 27-47 TgC yr-1 in the Terr simulation. Given the AO carbon sink increases with climate change, the relative reduction of the AO carbon sink due to terrigenous inputs decreased from about 23.2% in the 1970s to about 10.3% in the 2090s.
The avoidable effects of climate change
Shifting from a high to a low emission scenario leads to positive impacts on the efficiency of the BCP in the AO. We define the so-called “avoidable effects of climate change” the high (SSP3-7.0) minus the low (SSP1-2.6) emissions scenarios in the 2090s (both with terrigenous inputs, see methods). In a high emission scenario, the BCP efficiency is damped by the alteration of biogeochemical processes, compared with a low emission scenario, through the same mechanisms as for climate change (i.e. warming and subsequent effects). Sticking to a high emission scenario would decrease by 18 and 2.7% the carbon export and transfer efficiencies (respectively 19 and 3.1% for nitrogen, Fig. 2AB). In general, the “avoidable effects of climate change” are similar in magnitude to those of adding terrigenous inputs, but smaller than climate change. Those results mean that although our projections are sensitive to the choice of CO2 emissions scenario (and terrigenous inputs), climate change is consistently driving a substantial decrease in the efficiency of the BCP in all emission scenarios. Additionally, following the trajectory of a low emission scenario would also avoid a shift towards a severe nutrient limitation in the AO (Fig. 3C) because it would reduce NPP and net nitrogen assimilation by 21% and 17% (Fig. 2).