Equatorward of 10° of latitude, as well as at high northern latitudes, where the chlorophyll concentration exceeds 0.05 mg/m3, the surface is anomalously warm and the subsurface anomalously cold when the chlorophyll concentration is interactive as compared to when it is kept at a (lower) constant value. More heat is trapped in surface and thus less heat penetrates into the ocean interior, as found in Lengaigne et al. (2006). The opposite effect takes place in the southern subtropics while the strong warming in the northern subtropics could be due to the specific timing of the phytoplankton bloom in this region

in IPSL-CM5A (Séférian et al., 2012). The middle and right panels in Fig. 6 show that this situation evolves after the first decade and the ocean globally becomes colder in CM5_piCtrl than in CM5_piCtrl_noBio. This suggests a delayed adjustment of the ocean overwhelming the direct MEK pathway 1-dimensional effect. This evolution is also seen in each basin taken individually, while the large-scale meridional transport is unchanged, as seen in Fig. 1 (bottom) for the Atlantic. A role selleck kinase inhibitor of the oceanic circulation and in particular the AMOC in this slow adjustment is thus excluded. As discussed in Gnanadesikan and Anderson (2009), the net effect detected in these kinds of twin experiments depends on the set-up of the control simulation without interactive biogeochemistry. We indeed found major differences in the chlorophyll

vertical distribution, in particular equatorward of 30° of latitude (Fig. 7) between our control run and the one used in Lengaigne et al. (2006), which was very close to CM4_piCtrl. More precisely, concentrations at the surface are similar, but CM5_piCtrl is much poorer than the previous model version between the surface and 150 m depth. This implies that the anomalous warming linked to the capture of light by the chlorophyll is weaker down to 150 m in CM5_piCtrl as

compared to Lengaigne et al. (2006). Consistently, photosynthetically available radiation (PAR) is weaker in CM5_piCtrl in upper layers (not shown). This might explain why eventually, in our experiments, subsurface cooling overwhelms surface warming. Differences in the interactive chlorophyll profiles are prominently driven by the vertical distribution of nutrients, the ocean circulation (mixed-layer oxyclozanide depth) and the incoming shortwave radiation, since these three parameters control the nutrient-to-light co-limitation of the phytoplankton growth. A quantitative skill assessment of the marine biogeochemistry has been performed with two control simulations of IPSL-CM4 and IPSL-CM5A in Séférian et al. (2012) and with the same forced configuration as F4 in Duteil et al. (2011). These two studies reveal in particular that errors in ocean circulation lead to an unrealistic distribution of nutrients, which in turn impacts the distribution of chlorophyll. These latters impact finally the penetration of the radiant heat, and thus the ocean circulation.