The here presented data comprises of benthic foraminifera stable carbon (δ¹³C) and oxygen (δ¹⁸O) isotope, bulk sediment neodymium (εNd) isotope and XRF scanning data from IODP Sites U1302/03. The age model was developed by tuning the benthic δ¹⁸O record from Sites U1302/03 to the global benthic δ¹⁸O stack LR04 (Lisiecki and Raymo, 2005, doi: 10.1029/2004PA001071). For the stable carbon (δ¹³C) and oxygen (δ¹⁸O) isotope records, we selected 3-8 specimens of either epifaunal living benthic foraminifera species Cibicidiodes wuellerstorfi and Hoeglundia elegans, or shallow infaunal living species Uvigerina peregrina from the >106 μm grain-size fraction. The alternating use of these three species was necessary as none of the species were consistently present throughout the studied sediment interval. Prior to measurement, the selected foraminifera tests were carefully crushed, ultrasonicated in methanol to physically remove contaminations (e.g., clays and nannofossils) and subsequently dried at 50°C. The stable isotope measurements were carried out on a MicroMass Isoprime at the GEOTOP-UQUAM. The precision of the measurements is ±0.05 ‰ for δ¹³C and δ¹⁸O, respectively. The results were calibrated using the international standard NBS-19, and two in-house standards. Isotopic values are reported in standard delta notation (δ) relative to the Vienna Pee Dee Belemnite (VPDB). Interspecies correction between the selected foraminifera species follows Shackleton and Hall (1984, doi:10.2973/dsdp.proc.81.116.1984) by applying a constant fractionation factor of +0.64 ‰ for δ¹⁸O of C. wuellerstorfi and +0.9 ‰ for the δ¹³C values of U. peregrina. H. elegans was corrected to equilibrium by applying a constant fractionation factor of -0.4 ‰ and -1.5 ‰ for δ¹³C and δ¹⁸O, respectively. The high-resolution (~1 cm interval) elemental composition of the sediments at Sites U1302/03 were obtained by non-destructive XRF core scanning at Cambridge University using the Avaatech XRF core scanner. The instrumental settings for the XRF measurements as well as the Ca/Sr and Si/Sr ratios are already outlined in Channell et al. (2012, doi:10.1016/j.epsl.2011.11.029). To eliminate non-linear matrix effects and constant-sum constraints we used log ratios following Weltje et al. (2015, doi:10.1007/978-94-017-9849-5_21). We extracted the εNd isotope composition from a total of 84 samples by weak acid-reductive bulk sediment leaching following the method described by Blaser et al., (2016, doi:10.1016/j.chemgeo.2016.06.024). Briefly, 250-300 mg of dried and ground sediment were washed for 30 min with ultra-pure deionised water. After centrifugation the water was decanted and discarded. The authigenic fraction and carbonates were then leached with 10 ml of an acid-reductive aqueous solution containing 3mMNa-EDTA, 5mMhydroxylamine-hydrochloride, and 1.5% trace metal clean acetic acid buffered to a pH of 4 using trace metal clean 25% ammonia solution. After ∼60 min of agitation the samples were centrifuged and pipetted for further analysis. However, recent εNd results by Blaser et al. (2020, doi:10.1016/j.epsl.2020.116299) have shown that detrital carbonate can have an effect on the leaching efficiency on sediments from the Labrador Sea which can be enriched in detrital carbonate. Hence, for comparison a total of 48 samples across the studied time frame were decarbonated before subjection them to the same leaching procedure as stated above. For carbonate removal 10 ml of an acetic acid/sodium acetate buffer (NeAc/CaOH (ada)) was applied to the dried and grounded sediments for 30 minutes. Afterwards the sample was washed and rinsed three times with ultra-pure deionised water before proceeding with the leaching protocol following Blaser et al. (2016; doi:10.1016/j.chemgeo.2016.06.024). The dual approach of analysing decarbonate and non-decarbonated samples side-by-side allows for a direct comparison and assessment of potential detrital carbonate influences on the resulting εNd signals through time. Subsequent to the leaching, the remaining sample solution was purified with a two-step column chemistry protocol for the measurements of Nd iso-topes on two Nu Instruments "Plasma II" MC-ICP-MS at GEOTOP at Université du Québec à Montréal and at the GEOMAR Helmholtz Centre for Ocean Research, Kiel, and a "Neptune Plus" MC-ICP-MS at Heidelberg University. Nd isotope data were corrected to 146Nd/144Nd of 0.7219 with an exponential law and normalised to JNdi-1 standards with 143Nd/144Nd = 0.512115 (Tanaka et al., 2000, doi:10.1016/S0009-2541(00)00198-4). Reproducibility was assessed by repeated measurements of respective in-house standard solutions and varied in Nd from 0.12 to 0.63 (two standard deviations) between measurement sequences. The resulting uncertainty for εNd values is ~ 0.2 epsilon units (95% confidence).
