3858 R. Steinfeldt et al.: Anthropogenic carbon in the Atlantic The Cant anomalies in the southwestern Atlantic are most pronounced over the first decade (1990–2000, Fig. 12a). They show negative values south of 40° S between 100 and 1000 m depth in the AAIW and positive anomalies directly south- and equatorward (south of 20° S) in a slightly shal- lower depth range within AAIW and the overlying SAMW. A similar structure has been inferred in Waugh et al. (2013) from transient tracer data for the southern parts of the At- lantic, Indian and Pacific oceans. These authors ascribe the changes in ventilation to a strengthening and southward movement of the westerly wind belt. This leads to enhanced upwelling of older water with low Cant south of the polar front and increased northward Ekman transport and forma- tion of mode waters (with high Cant) north of the front. A similar dipole in the upper 1000 m of the South Atlantic is also evident in the study of Gruber et al. (2019). Tanhua et al. (2017) also found large Cant storage in SAMW, at least between 1990 and 2005. The second area with extreme values of 1tCanomant in the tropical Atlantic is mainly restricted to the layer above 500 m (subtropical mode water). Positive and negative Cant accu- mulation anomalies alternate in latitudinal direction and be- tween the decades. They could be a consequence of variable mode water formation in the subtropics. Such changes in the subtropical cell with enhanced production and southward transport of Cant-rich mode water have been inferred from an inverse model in DeVries et al. (2017) with enhanced pro- duction and equatorward transport of Cant-rich mode water in the 1990s. Unfortunately, the study in DeVries et al. (2017) ended in 2010, and the decades in which the data are grouped are shifted by 5 years compared to our study, thus prohibiting a direct comparison of the decadal results. In contrast to our results, Gruber et al. (2019) find negative Cant anomalies in the whole tropical Atlantic over the upper 1000 m. The northernmost area with extreme values of 1tCanomant is located north of 40° N in the subpolar North Atlantic, includ- ing the Labrador Sea (Fig. 12). This structure reflects the ob- served variability of convective activity in the Labrador Sea and the associated changes in LSW formation. An unprece- dented deep-reaching convection formed a very dense mode of LSW from 1987 to 1994 (Yashayaev, 2007). During the following years, only lighter modes of LSW (Upper LSW, ULSW) have been formed (Stramma et al., 2004; Kieke et al., 2006; Yashayaev, 2007), whereas the pool of dense LSW (DLSW) has been exported from the formation region south- and eastward (Kieke et al., 2007; Rhein et al., 2015). These two processes are reflected in the positive1tCanomant for the 2000–2010 period around 1000 m (formation of ULSW modes) and the negative Cant anomalies between 1500 and 2000 m (export of DLSW) in Fig. 12c. This lack of Cant stor- age in the deeper part of the LSW between 2000 and 2010 is also visible in Fig. 11c. In 2008, convection in the Labrador Sea exceeded a depth of 1600 m for the first time in years (V?ge et al., 2009) but without a great impact on the Cant and oxygen trends (Rhein et al., 2017). In the study by Gruber et al. (2019), the Cant anomaly in the North Atlantic is negative down to a depth of ? 2500 m with the minimum in the upper ? 1000 m. Thus, a ULSW–DLSW dipole in Cant is not found there. Studies on the convection in the Labrador Sea indicate that at least the upper 500–1000 m of the water column has been con- vectively renewed every year since the 1990s (Yashayaev, 2007; Kieke and Yashayaev, 2015; Yashayaev and Loder, 2016), which makes a drastic decrease in the Cant storage in that depth range unlikely. Starting in 2014, deep-reaching convection in the Labrador Sea has re-emerged (Kieke and Yashayaev, 2015; Yashayaev and Loder, 2016). This is re- flected by the positive Cant anomaly in the LSW in the north- western Atlantic between 2010 and 2020. However, the den- sity of this recently formed LSW is still smaller than the den- sity of the LSW originating from the early 1990s (Yashayaev and Loder, 2016). Hence, the positive1tCanomant values do not reach the lower boundary of the LSW layer. The Irminger Sea has also been influenced by the enhanced deep convection in the subpolar northwestern Atlantic, and a large increase in Cant has been found there (Fröb et al., 2016). In the bottom waters north of 40° N (DSOW), there is an alternating pattern of negative and positive 1tCanomant values (Fig. 12a, c and e). From 1965–2000, the overflow waters experienced a freshening trend lasting more than 3 decades (Dickson et al., 2002). This long-term trend does not influ- ence the Cant uptake of ISOW and DSOW, as no such signal is evident in Fig. 12. Annual fluctuations in salinity (and also temperature) for the DSOW particularly coincide with the long-term freshening trend (Yashayaev, 2007). These differ- ent vintages of DSOW might be the reason for the alternating minima and maxima in 1tCanomant in the bottom waters north of 35° N. In the northeastern Atlantic, the Cant accumulation anomaly in the LSW for the 2000–2010 period is similar to that in the northwestern part, although less pronounced. LSW is formed in the western Atlantic, and the 1tCanomant of the newly formed LSW becomes diluted when the anomalies spread eastward. The recent positive 1tCanomant value in LSW did not become prominent in the eastern North Atlantic until 2020. Another small region with a Cant deficit is located within the deep and bottom waters (WSDW and AABW) of the southwestern Atlantic around 55° S between 1990 and 2000. This is the area where the newly formed deep water originat- ing from the Weddell Sea is advected eastward (see above). This recently ventilated WSDW is relatively high in Cant (Fig. 7a) but only shows a small decadal increase (Fig. 12a), lacking the expected growth from the atmospheric CO2. This result is in agreement with Huhn et al. (2013), who also found an aging and 1tCanomant deficit of AABW in the Wed- dell Sea. After 2000, this negative anomaly does not occur anymore, and the Cant increase in WSDW/AABW between 60 and 50° S is higher (Figs. 11c, e and 12c, e). Biogeosciences, 21, 3839–3867, 2024 https://doi.org/10.5194/bg-21-3839-2024