Atmospheric Fe supply has a negligible role in promoting marine productivity in 1 the Glacial North Pacific Ocean 2

13 Iron is a key element in the Earth climate system as it can enhance the marine primary productivity in the 14 High-Nutrient Low-Chlorophyll (HNLC) regions where, despite a high concentration of major nutrients, the 15 chlorophyll production is low due to iron limitation. One of the main Fe sources to the ocean is aeolian dust. 16 For this reason, ice cores provide a sensitive and continuous archive for reconstructing Fe fluxes over the last 17 millennia. Here we show the first Northern Hemisphere Fe record retrieved from the NEEM ice core, which 18 offers a unique opportunity to reconstruct the past Fe fluxes in the Arctic region over the last 108 kyr. 19 Holocene Fe fluxes to the Arctic were three times lower than the average recorded over the last glacial 20 period. They were greater during the Last Glacial Maximum (LGM) and the Marine Isotope Stage 4 (MIS 4). 21 Comparing our data with palaeoceanographic records retrieved from the HNLC North Pacific, we 22 demonstrated that during the coldest periods, characterized by the highest Fe fluxes, marine productivity in 23 the subarctic Pacific Ocean did not increase due to a greater sea-ice extent and the absence of upwelling 24 nutrient supply. This supports the hypothesis that Fe-fertilization was more effective in other regions, such as 25 https://doi.org/10.5194/cp-2020-77 Preprint. Discussion started: 10 July 2020 c © Author(s) 2020. CC BY 4.0 License.


Introduction 28
Greenland and Antarctic ice cores are unique archives that can provide information about how 29 temperature, atmospheric dust load and atmospheric gas composition have changed during the Holocene and 30 the late Pleistocene (Lambert et al., 2008;Schüpbach et al., 2018). Glacial periods were dustier and with a 31 lower CO 2 concentration (≈ 180 ppm) than interglacials (≈ 280 ppm). This dichotomy was explained through 32 different hypotheses: the increase in aridity and in newly exposed continental shelves (Fuhrer et al., 1999), 33 the enhancement of the atmospheric circulation (Delmas and Legrand, 1989), the increase in the aerosol 34 atmospheric life-time (Yung et al., 1996), the enhancement of glacier mobilization of highly bioavailable 35 iron from the bedrock (Shoenfelt et al., 2018) and, lastly, the enhancement of the polar circulation that might 36 have entrained additional dust from lower latitudes (Mayewski et al., 1994). The higher atmospheric dust 37 loading during glacial periods affected climate in both physical and biological ways. On the one hand, dust 38 particles absorbed and scattered the incoming solar radiation and the outgoing infrared radiation with a direct 39 effect on the Earth energy budget. On the other hand, once deposited on the ocean surface, the mineral dust 40 provided major and micronutrients (including iron) that could have stimulated the biological carbon pump 41 (Martin et al., 1990). This iron mediated mechanism has been demonstrated for the High-Nutrient Low-42 Chlorophyll (HNLC) regions, whose productivity is primarily Fe-limited, both through artificial (Smetacek 43  processes. It has been inferred that the observed decrease in the atmospheric CO 2 concentration during 45 glacial periods was linked to the Fe-modulated enhancement of the biological carbon pump in the Southern 46 Ocean due to the increase in Fe availability (Martin et al., 1990). However, according to both modeling 47 (Lambert et al., 2015) and observational (Gaspari et al., 2006;Röthlisberger, 2004;Vallelonga et al., 2013) 48 studies, the Fe-fertilization mechanism itself cannot completely explain the ≈100 ppmv glacial-interglacial 49 atmospheric CO 2 variability, but only around 8-20 ppmv of it. 50 Studies on leachable iron in ice cores were performed in Antarctica from Talos Dome (TD) (Spolaor et

Comparison with Fe fluxes from Antarctic ice cores 136
The NEEM iron ice core record allows the first comparison of leachable Fe concentrations and 137 fluxes between the Arctic and Antarctica (Figure 2) focusing on the Holocene, LGM and MIS 4 periods 138 (Table 3). 139 During the Holocene, average Fe fluxes in both NEEM (0.06 mg m -2 yr -1 , CV = 1.2) and Talos Dome 140 (0.09 mg m -2 yr -1 , CV = 1.2), were of the same order of magnitude suggesting a similar contribution from 141 aeolian mineral dust in both sites. They were significantly greater than the ones recorded in Law Dome (0.04 142 mg m -2 yr -1 , CV = 0.5) and Dome C (0.007 mg m -2 yr -1 , CV = 0.2). As previously mentioned, the greater Fe 143 fluxes in TD than in the other Antarctic sites, was related to the activation of a dust deflation area which 144 influenced Victoria Land more than the central Antarctic Plateau (Delmonte et al., 2013). were substantially lower suggesting a particularly enhanced deflation from Asian deserts (e.g. high wind 154 speeds, enhanced dustiness over the East Asian desert regions, increased aridity related to changes in the 155

Iron and marine productivity in the Northern Hemisphere 157
Considering the biological relevance of Fe and taking advantage from the Fe flux record retrieved from 158 the NEEM ice core, we questioned if the increase in its fluxes triggered the marine productivity in the HNLC 159 https://doi.org/10.5194/cp-2020-77 Preprint. Discussion started: 10 July 2020 c Author(s) 2020. CC BY 4.0 License. region of the North Pacific. Our interpretation is based on Fe fluxes retrieved from samples, which were 160 acidified with 2% HNO 3 for one month before the analysis. Therefore, they do not directly represent the 161 actual bioavailable Fe that can be dissolved into seawater at pH 8, but rather an upper limit of the aeolian Fe 162 potentially available for the phytoplankton (Edwards et al., 2006).  Accumulation Rate) C 37 alkenone (µg cm -2 kyr -1 ), was at its lowest level. A plausible explanation is that this 219 region was not characterized by stratified waters and thus it was not affected by the limitation of major 220 nutrients. Unfortunately, neither FB-δ 15 N nor information about water stratification are available for this 221 record. 222

From 108 kyr to the LGM 223
According to the available records, marine productivity changed heterogeneously in the Pacific 224 Ocean during the last glacial period (Figure 3). 225 It is challenging to state, with a high degree of confidence, whether Fe-fertilization triggered a 226 phytoplankton bloom or not in the HNLC subarctic North Pacific. This is due to the different responses that

Conclusions and future perspectives 248
In this study, we provided the first Fe record retrieved from the NEEM ice core. Through the comparison 249 with other available Fe records, we observed that during the Holocene, the Fe fluxes were similar between 250 TD and NEEM, while during the LGM all the investigated records showed Fe fluxes on the same order of 251 magnitude. The greatest difference arose during the MIS 4, when Fe fluxes in NEEM were 3 and 7 times 252 higher than in TD and DC respectively. 253 Merging our record with marine productivity data, we found that a direct link between Fe transport and

Competing interests 269
The authors declare that they have no conflict of interest. 270

Acknowledgments 271
We sincerely thank all the persons involved in the logistics, drilling operations, ice-core processing and   Pacific (Haug et al., 1995) and S-2, transition zone (Amo and Minagawa, 2003). Due to their limited 334 temporal extension, productivity records from SO202-07-6 and SO202-07-26 are not discussed in this figure, 335 but in Figure 4.