This is the accepted version of the article: Raja, Maria; Rosell-Melé, Antoni. «Appraisal of sedimentary alkenones for the quantitative 4 reconstruction of phytoplankton biomass». Proceedings of the National Academy of Sciences of the United States of America, Vol. 118, issue 2 (Jan. 2021), art. e2014787118. DOI 10.1073/pnas.2014787118 This version is available at https://ddd.uab.cat/record/243265 under the terms of the license 1 2 3 Main Manuscript for 4 Appraisal of sedimentary alkenones for the quantitative 5 reconstruction of phytoplankton biomass 6 7 Maria Raja*1 and Antoni Rosell-Melé1,2 8 1Institut de Ciència i Tecnologia Ambientals (ICTA-UAB), Universitat Autònoma de Barcelona, 9 Bellaterra, Catalonia, Spain 10 2 Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalonia, Spain 11 * Maria Raja 12 Email: Maria.Raja@uab.cat 13 14 https://orcid.org/0000-0003-0207-8189 15 16 Classification 17 Physical Sciences, Environmental Sciences. 18 Keywords 19 primary productivity; proxy; remote sensing; global export and burial; marine sediments 20 Author Contributions 21 Maria Raja: Methodology, Software, Validation, Formal Analysis, Investigation, Data Curation, 22 Writing-Original Draft, Writing-Review&Editing, Visualization. 23 Antoni Rosell-Melé: Conceptualization, Validation, Resources, Writing-Review&Editing, 24 Supervision, Funding Acquisition. 1 25 This PDF file includes: 26 Main Text 27 Fig.s 1 to 5 28 Table 1 29 30 Abstract 31 Marine primary productivity is the driving factor in the global marine carbon cycle. Its 32 reconstruction in past climates relies on biogeochemical proxies that are not considered to 33 provide an unequivocal signal. These are often based on the water column flux of biogenic 34 components to sediments (organic carbon, biogenic opal, biomarkers), although other factors 35 than productivity are posited to control the sedimentary contents of the components, and their flux 36 is related to the fraction of export production buried in sediments. Moreover, most flux proxies 37 have not been globally appraised. Here we assess for the first time a proxy to quantify past 38 phytoplankton biomass by correlating the concentration of C37 alkenones in a global suite of 39 core-top sediments with sea-surface chlorophyll-a (SSchla) estimates over the last 20 years. 40 SSchla is the central metric to calculate phytoplankton biomass and is directly related to primary 41 productivity. We show that the global spatial distribution of sedimentary alkenones is primarily 42 correlated to SSchla rather than diagenetic factors such as the oxygen concentration in bottom 43 waters, which challenges previous assumptions on the role of preservation on driving 44 concentrations of sedimentary organic compounds. Moreover, our results suggest that the rate of 45 global carbon export to sediments is not regionally constrained, and that alkenones producers 46 play a dominant role in the global export of carbon buried in the sea-floor. This study shows the 47 potential of using sedimentary alkenones to estimate past phytoplankton biomass, which in turn 48 can be used to infer past primary productivity (PP) in the global ocean. 49 50 Significance Statement 51 Biomarker proxies, namely alkenones, are commonly used to reconstruct past primary 52 productivity. However, their value has been questioned as they are posited to be controlled by 53 diagenetic processes and often limited to draw qualitative inferences on the carbon cycle. In fact, 54 most proxies have not been globally appraised with ocean biogeochemical data. Here we use a 55 combination of remote sensing and geochemical data to provide the first spatial-based global 56 calibration that show the use of alkenones to quantify past sea-surface chlorophyll-a 57 concentration, which ultimately can be used to infer quantitatively past primary productivity in 58 paleorecords. This calibration paves the way to clarify the relative role of the marine carbon cycle 59 in climate variability using field data, and test biogeochemical models. 60 61 62 Main Text 63 64 Introduction 65 66 Global carbon distribution between the ocean and the atmosphere regulates global climate on 67 Earth. This distribution is primarily controlled by marine primary productivity (PP) and 68 phytoplanktonic organisms, which transforms atmospheric CO 2 into organic matter. Only a 69 fraction of this produced organic matter is exported to the deep ocean. Global models estimate 70 that 48 PgC·yr-1 are produced in ocean surface waters (1) while 6 PgC·yr-1 (2) are exported out 71 from the photic zone, and 0.15 PgC·yr-1 are buried in sediments (3). Exported organic carbon is 72 out of contact with the atmosphere on decadal to millennial timescales or longer once is buried in 2 73 the seafloor, which exerts a major control on global climate by regulating the partial pressure of 74 atmospheric CO2 (4). Hence, estimating marine PP, export and burial productivity changes during 75 past key climatic periods (e.g. glacial-interglacial transitions) is essential to understand our 76 present climate and predict its evolution in the future. 77 To infer past PP a range of proxies based on the fluxes of biogenic components are available (5– 78 7). As flux proxies, they are related to changes in past export productivity, which are assumed to 79 be proportional to surface PP in paleoreconstructions. However, depositional factors such as 80 oxygen or ballasting effect are thought to be important in controlling organic matter export from 81 the upper water column to sediments (8–10) and thus, organic proxies sedimentary 82 concentration. The relative weights that control the spatial variability of organic matter 83 concentration in sediments, are still unconstrained, which leads to some uncertainty on the 84 applicability of organic matter proxies to infer PP (7). Consequently, available proxies are 85 sometimes interpreted to infer either changes in PP or depositional conditions (11). 86 One of the common approaches to reconstruct PP relies on the measurement of C 37 di- and tri- 87 unsaturated methyl ketones (i.e. C37 alkenones) concentrations or fluxes in sediments (12–18). 88 These organic molecules are biomarkers of the ubiquitous coccolithophore Emiliania huxleyi, 89 which is the principal source of alkenones and the most abundant coccolithophore in the modern 90 pelagic ocean (19–24). Geophyrocapsa oceanica and other coccolithophoral species from the 91 same genera are also considered important alkenones producers nowadays (20). 92 In this study, we evaluate for the first time the potential use of sedimentary C 37 alkenones 93 contents to infer past phytoplankton biomass at a global scale through the comparison of their 94 spatial variability in a global compilation of core-top sediments with sea-surface chlorophyll-a 95 (SSchla) (Fig. 1). This is the primary pigment of photosynthesis and is present in all 96 photosynthetic phytoplankton species. Its concentration in surface waters is commonly used as 97 an indicator of phytoplankton biomass and to infer primary productivity (25, 26). On a global 98 scale, its concentration in surface waters is estimated by remote sensing (27). We also assess 99 the effect of oxygen on the spatial accumulation of alkenones in sediments by comparing its 100 concentration in bottom waters with alkenones abundance on a global scale. 101 102 103 Results 104 105 Global comparison of SSchla with sedimentary alkenones 106 The comparison between sedimentary alkenones and SSchla concentration estimated over the 107 last 20 years is shown in Fig. 2. Log-transformation was needed, as the distribution of SSchla is 108 close to lognormal in our dataset (SI Appendix, Fig. S1). This is concordant with the natural 109 distribution of ocean chlorophyll (28). We applied standard major axis regressions (Type II 110 regressions), which assume that both variables are measured and include some error. For this 111 analysis we took into account that the remote sensing SSchla estimates using global standard 112 algorithms can have a wide range of error values and specific regional biases (29–35). To 113 minimize uncertainty in the correlation analysis, we focused our study on SSchla data located in 114 well-performing remote sensing regions. We applied the threshold stablished in (36), where 115 basins associated with lower 116 root-mean square logarithmic errors (RMS log errors) than the global mean (31%), are 117 considered well-performing regions. This value is within the range of 30-35% threshold values 118 proposed in the remote sensing literature, including the Global Climate Observing System 119 (GCOS) and the Ocean Colour Climate Change Initiative (OC-CCI), as the desired error 120 benchmark for open ocean waters, or optically clear waters (30, 31, 36–38). Consequently, we 121 did not included in our final analysis (see Fig. 2b) SSchla data from the North Atlantic (36, 39, 40) 122 and the Gulf of Alaska (41).Nor from the western South American margin, as the SSchla error 123 estimate of this region has not been thoroughly studied. Although data for the equatorial Atlantic 124 presents a RMS log error of 48%, it lowers to 23% when removing samples located offshore of 125 the north-eastern coast of South America (36). Thus, as our samples are not located in this 126 region, we included samples from the equatorial Atlantic in the correlation analysis. We obtained 3 127 a linear correlation between C37 alkenones and SSchla concentration with an RMS log error of 128 38% and a coefficient of determination (R2) of 0.60 (Table 1). Hence, our data show that changes 129 in the spatial variability of C37 alkenones are mainly related to changes in SSchla. 130 Several studies report the occurrence of regional variability in the vertical attenuation rate of 131 organic matter flux from the surface to the deep ocean (10, 42–44). This spatial variability could 132 lead to different regional correlations between sedimentary alkenones and SSchla concentration. 133 To investigate this issue in our data set, we divided the data by biogeochemical regions as 134 defined in (44), which are classified by factors influencing transfer efficiency, such as 135 phytoplankton community, nutrient concentration and temperature. The defined regions are as 136 follows: the tropics, the subtropics, the subarctic and the Southern Ocean. Note that the 137 correlation for the subarctic region was not calculated due to insufficient data. The data from the 138 subtropics show the lower RMS log error (31%) and is the region that can explain more SSchla 139 variability (59%). The tropics present the higher RMS log error (46%), with a coefficient of 140 determination of 0.49, and the Southern Ocean is the region that can explain less SSchla 141 variability (33%), with a log RMS error of 35% (Table 1). However, no significant differences were 142 found between sedimentary alkenones and SSchla correlations when comparing different oceanic 143 biogeochemical regions, as it is shown in Fig. 3. 144 145 Global comparison of bottom water oxygen with sedimentary alkenones 146 To study the influence of oxygen in the spatial accumulation of alkenones in sediment, we 147 correlated our global core-top sedimentary alkenones compilation with oxygen concentration in 148 bottom waters (Fig. 4). Data on oxygen concentration was extracted from the World Ocean Atlas 149 2018, which provide oxygen data with a spatial resolution of 1º. The relative low resolution of the 150 data may introduce some uncertainty in our analyses in some regions, as strong oxygen 151 gradients may occur that are smaller than the 1º grid of the dataset, such as those located in 152 oceanic ridges. However, our data do provide information about the influence of bottom oxygen 153 content on alkenones abundance at regional and global scales. 154 The global comparison between sedimentary alkenones and oxygen concentration in bottom 155 waters is shown in Fig. 4. Since our study comprises several biogeochemical regions (Fig. 1), it 156 covers a wide range of oxygen concentration conditions (from 18.72 to 268.11 µmol·kg -1). 157 Consequently, Fig. 4 shows different patterns that might reflect the influence of oxygen 158 concentration in different regions. Nevertheless, we do not observe a simple global overall trend 159 as in the case of SSchla, and we obtained a R2 of 0.02. Thus, oxygen concentration in bottom 160 waters is not the primary cause of alkenones abundance spatial variability on a global scale. 161 162 163 Discussion 164 165 Alkenones as a proxy for past phytoplankton biomass 166 Alkenones concentration in sediments is commonly used to qualitatively infer PP changes (14, 167 17, 18). However, as it is commonly stated in the literature, sedimentary alkenones concentration 168 may also be affected by depositional and burial processes, which could complicate the 169 interpretation of the proxy. Thus, alkenones concentration fluxes decrease with depth in the water 170 column (45, 46). In here, we do not appraise directly the link between PP and alkenones 171 sedimentary concentration, but indirectly through the use of SSchla which is a metric of 172 phytoplankton biomass (Fig. 2). The data span several oceanic biogeochemical regions (44), and 173 show that the spatial variability in sedimentary alkenones is primarily related to changes in 174 SSchla, following a linear relationship with a R2 of 0.60 and RMS log error of 38% (Table 2). Our 175 data permit to quantitatively reconstruct past SSchla concentration (equation 2). Therefore, our 176 results confirm previous assumptions on the use of sedimentary alkenones concentrations as a 177 proxy to qualitatively infer PP changes rather than redox conditions. 178 The correlations in Fig. 2 also allow the quantitative comparison of phytoplankton biomass data 179 from different sites and regions through time, which hitherto was considered hindered by 180 differences in depositional conditions between locations. Thus, one of the implications of 4 181 obtaining a global correlation between C37 alkenones and SSchla concentrations is that the 182 vertical rate of degradation of C37 alkenones from the sea-surface to sediments follows a similar 183 spatial pattern across different biogeochemical regions in the global ocean. We have obtained no 184 significantly different correlations between C37 alkenones and SSchla concentration for the 185 different oceanic biogeochemical regions defined in (44) (Fig. 3). This is consistent with studies 186 that have indicated that there are no regional differences in the vertical attenuation rate of organic 187 matter flux (47–49), even though other studies have found spatial variability in the export 188 efficiency of surface biomass to the deep ocean (10, 42–44). 189 Hence, our data support the results obtained in (50) and other earlier papers, which showed that 190 spatial variability in the vertical flux of particulate organic carbon decreases with increasing water 191 column depth. Our results are also in agreement with (51), who showed that the proportion of 192 primary productivity that reaches the deep sea does not vary greatly with latitude in the North 193 Atlantic. Besides, these results show that the global correlation is independent of the 194 biogeographic region, which implies that it is applicable globally despite any changes in 195 biogeochemical regions through time. In this sense, we assume the similar pattern of the vertical 196 degradation rate over different biogeochemical regions is also maintained after burial processes. 197 Future studies will evaluate to which extent subsurface degradation process might significantly 198 influence or bias the relationships described in Fig.s 2 and 3, and constrain their application in 199 paleo studies. 200 201 The relevant role of alkenones producers in the global carbon export 202 In contrast to SSchla, alkenones are mainly produced by coccolithophores (E. huxleyi and 203 Geophyrocapsa species) (20). Therefore, sedimentary alkenones concentration are often 204 considered an exclusive coccolithophore productivity proxy, which may not provide information 205 about total past phytoplanktonic biomass changes (15, 18, 52, 53). In contrast, C37 alkenones 206 concentration were also suggested to track PP not only from coccolithophores, but also from the 207 wider phytoplanktonic community in upwelling tropical areas (12). Subsequently, such 208 interpretation was corroborated at two more sites: ODP Site 982 in the North Atlantic and ODP 209 Site 846 in the eastern tropical Pacific (54). In spite of their findings, the authors suggested a 210 linear relationship would be not expected in oligotrophic areas, where alkenone-synthesising 211 coccolithophores constitute a small proportion of the coccolithophore population. 212 Contrary to previous assumptions, our global correlation shows that C 37 alkenones sedimentary 213 concentrations can be interpreted as a proxy for total phytoplankton biomass in different 214 biogeochemical regions (Fig. 2), and suggests that there are no differences between samples 215 located in oligotrophic (SSchla annual average < 0.1 mg·m-3) and non-oligotrophic regimes (SI 216 Appendix, Fig. S2). 217 Obtaining a global correlation between SSchla and alkenones indicates a proportionality between 218 total phytoplankton biomass and alkenones accumulation in sediments. In terms of total 219 phytoplankton biomass and PP, photosynthetic picoeukaryotes, which include E. huxleyi and G. 220 oceanica, have been reported to be the dominant contributors. For instance, photosynthetic 221 picoeukaryotes are responsible for more than the half the total PP in the North Atlantic (68%) 222 (55), southern California Bight (76%) (56), the eastern northatlantic subtropical gyre (54%) (57), 223 the southern Bay of Biscay (51%) (58) and the South East Pacific Ocean (>60%) (59). 224 However, there is not a consensus on which class of phytoplankton is the dominant carbon 225 exporter to the sea floor. Large phytoplankton, such as diatoms, were recognised to dominate 226 global carbon export to sediments (60, 61) and picoplankton (photosynthetic picoeukaryotes and 227 cyanobacteria) were claimed not to contribute significantly to carbon export because of their small 228 size (62). In contrast, more recent studies have suggested that picoplankton plays a major role in 229 the sedimentary carbon fluxes. For instance, massive picoplankton sedimentation in east of New 230 Zealand (63), and picoplankton export carbon fluxes dominance were found in the eastern 231 equatorial Pacific (64) and in the Arabian sea (65). Besides, subsequent studies showed that the 232 downward flux of organic carbon via small particles often constitutes the bulk of the total 5 233 particulate organic carbon flux in the North Atlantic (66), and that the export of small particles in 234 the Norwegian Sea contributed to long-term carbon sequestration (67). 235 We are not aware of any evidence that endorse the notion that alkenones producers are one of 236 the main contributors to global carbon sequestration. However, our results in Fig. 2 indicate that 237 alkenones producers export and burial contribution is proportional to total phytoplankton biomass 238 at global scale. Our results are in agreement to previous findings that show primary productivity to 239 be proportional to picoplankton biomass export from the euphotic zone at two different locations: 240 the equatorial Pacific and the Arabian Sea (68). We suggest that such proportionality is indicative 241 of the putative dominant role of photosynthetic picoeukaryotes to global carbon sequestration and 242 burial in the sea-floor. 243 244 The role of oxygen in the spatial accumulation of alkenones in sediment 245 Oxygen enhances organic matter degradation and its concentration in bottom waters is 246 considered a key factor in the preservation and accumulation of organic matter in sediments (69– 247 73). However, despite its strong influence in the vertical degradation of organic matter, its role in 248 the global spatial distribution of organic matter concentration remains unconstrained. 249 Some studies located in oxygen minimum zones, such as the Arabian Sea, reported empirical 250 relationships between oxygen concentration in bottom waters and sedimentary organic matter or 251 alkenones concentration (69, 74–76). Although these relationships only show a correlation when 252 oxygen is present at very low concentrations (≤ 50 µmol·kg-1), alkenones concentration is 253 sometimes used to reconstruct preservation conditions in other environments, such as the Pacific 254 and the Atlantic oceans (11, 77, 78). 255 Our results in Fig. 4 show that there is not a straightforward relationship between the spatial 256 concentration patterns of sedimentary alkenones and bottom water oxygen. In contrast to 257 previous studies, our data are located in different oceanic biogeochemical regions and include 258 low and high oxygen content environments (from 19 to 268 µmol·kg -1). Therefore, the wide range 259 of oxygen concentration presented in this study would be much more representative of the global 260 ocean conditions than those previously evaluated. These results can be interpreted as evidence 261 of the minor role of oxygen concentration in the spatial distribution of sedimentary alkenones for 262 the greater part of the ocean. Thus, our data do not support the global use of alkenones 263 concentration to reconstruct preservation conditions in the deep ocean. 264 265 Conclusions 266 The primary driver explaining the global spatial distribution of sedimentary alkenones is 267 phytoplankton biomass as reflected by SSchla concentration. Different biogeographic regions do 268 not show significant differences in their correlations between sedimentary alkenones and SSchla 269 concentration. Moreover, oxygen concentration in bottom waters does not strongly influence the 270 spatial concentration of alkenones in sediments on a global scale. Hence, our results do support 271 the use of alkenones sedimentary abundance to study primary productivity, and do not support 272 the use of alkenones as an indicator of preservation conditions in the deep ocean. Besides, our 273 data indicate that the global vertical attenuation rate of organic matter flux from sea-surface to 274 sediments is not regionally constrained, and suggest that the global carbon export and burial is 275 dominated by photosynthetic picoeukaryotes. 276 Thus, this study provides the first spatial-based global calibration that show the use of 277 sedimentary alkenones to quantify sea-surface chlorophyll-a, which ultimately can be used to 278 infer quantitatively PP in paleorecords. This calibration is independent of biogeographic regions, 279 which implies that besides relative PP changes through downcore, it can also be used for the 280 spatial comparison of PP between different locations at any site in the global ocean. Furthermore, 281 it is applicable despite any putative changes in biogeochemical regions through time. 282 283 284 285 Materials and Methods 6 286 287 Sedimentary alkenones 288 We compiled alkenone data from 226 locations widely distributed around the global ocean, and 289 across diverse biogeochemical regions as defined in (44) (Fig. 1). Most of the data are obtained 290 from previous published studies (95% of the compilation), while the rest were analysed in our 291 laboratory following the methodology explained hereafter. We assume that the interlaboratory 292 reproducibility of the different available methods to measure the absolute alkenones abundance 293 in sediments is 32% (79). The samples in the compilation generally correspond to the upper 1 cm 294 of a sediment core, and approximately 90% of them were retrieved using devices specially 295 designed for a minimum disturbance of the sediment surface, i.e. box corer or multicorer. The 296 compilation appears to be dominated by samples along continental margins, since they represent 297 the most intensively studied regions. However, sample sites span a wide range of alkenones 298 concentration distributed in the diverse biogeochemical regions defined in Weber et al. (2016) 299 around the global ocean (Fig. 1). In addition, over 80% of the compiled surface sediments are in 300 sites with a water column deeper than 1,000 meters, while over 90% of the samples were 301 retrieved from waters deeper than 500 meters (SI Appendix, Dataset S1). Consequently, the 302 compilation is mainly representative of the water column of the open ocean. We estimate that in 303 average, a sedimentary sample represents a time span of sedimentation of ca. 71 years. This has 304 been obtained from the sedimentation rate average of all sediments (14 cm·ka-1) using the map 305 published in (80). However, variability of sedimentation rates could play a role in driving the 306 concentration of alkenones in sediments. Consequently, to analyse changes in the flux of a 307 chemical or substance to the sea, the mass accumulation rates (MAR) must be used instead of 308 concentrations. Alkenones mass accumulation rates (MARalkenones) were estimated by 309 multiplying alkenones concentration by sedimentation rate and dry bulk density. In the absence of 310 these data for each sample site, we extracted sedimentation rates from two global maps 311 previously published (51, 80) and discussed in (81). However, sedimentation rates could only be 312 extracted for 54% of the sites. Dry bulk densities were assumed to be 0.9 g·cm -3, as it 313 corresponds to the mean dry bulk density for marine sediments in the global sediment core 314 database published in (81), which has been created by retrieving available data from online data 315 repositories. It is important to point out that these estimated accumulation rates are highly 316 uncertain, given that sedimentation rates can vary significantly over relatively short distances on 317 the seafloor due to winnowing and focusing, and the locations of cores are often biased towards 318 the highest accumulation rates in the search of retrieving records with high temporal resolution. 319 Nonetheless, in the absence of more accurate data, we include the estimates to provide a first- 320 order information on the effect of sedimentation rates in our compilation. We compared 321 MARalkenones against alkenones concentration in SI Appendix, Fig. S3. The correlation between 322 the two variables is high as attested by their coefficient of determination of R 2=0.92 and R2=0.87, 323 depending on the sources of sedimentation rates (51, 80). Consequently, overall for our 324 compilation, the differences in using MARalkenones and alkenones concentration when 325 compared to values of SSchla concentration are not likely to lead to significantly different results. 326 To maximize the size of data points in our compilation we used the alkenones concentration 327 rather than its MAR. 328 329 In the samples analysed for the study, sediments (1-4 g) were extracted using 10 mL of a mixture 330 of dichloromethane and methanol (3:1) (GC grade, Suprasolv) in a MARS5 microwave 331 accelerated extraction system (CEM Corporation). Before extraction, 25 ng·µL-1 of the internal 332 standard 2-nonadecanone (Fluka, purity 0.97%) was added to the sediment placed in a Teflon 333 vessel. During extraction the mixture was stirred continuously with a magnetic bar, while 334 temperature was increased from ambient to 70ºC for 2.5 minutes and left at this temperature for a 335 further 5 minutes. After extraction, samples’ vessels were left to cool down at room temperature, 336 and the supernatants were decanted into glass tubes and centrifuged. Extracts were then dried 337 under vacuum and cleaned up using SiO2 columns eluted sequentially with hexane (first fraction) 338 and dichloromethane (second fraction). Extracts were dried again and 50 µL of isooctane was 339 added to the second fraction before their analysis with a gas chromatograph (GC) fitted with a 7 340 flame ionization detector at 320ºC (Agilent Technologies 7820A GC System). Samples were 341 injected in splitless mode in a GC column (HP-1 GC column, 60m length, 250µm internal 342 diameter, 0.25µm film thickness) with a flow rate of 1.5 mL·min-1. The GC method consists of 2 343 ramps, the first one from 80ºC to 120ºC with a temperature rate of 30 ºC·min -1, followed by a 344 second ramp that increases temperature at 6ºC min-1 until 320ºC, and held at this temperature 345 during 21 minutes. 346 347 348 Ocean surface chlorophyll 349 Satellite observation of ocean-colour, in the open ocean, is primarily dependent on phytoplankton 350 occurrence. These unicellular algae contain photosynthetic pigments, primarily chlorophyll-a, 351 which coexist together with other pigments, associated detrital, and coloured dissolved organic 352 matter (82). To infer SSchla from ocean-colour data, many algorithms have been constructed 353 relating characteristics of the water signal to the property of interest (83). There are large 354 uncertainties in the estimation of SSchla in some regions, particularly where the optical and 355 biological properties are complex (84, 85). For instance, some of the factors that affect the 356 properties of the surface waters are coloured dissolved organic matter, radiance-absorbing 357 aerosols, phytoplankton species diversity, suspended sediments and minerals, clouds, ice, sun 358 glint, and navigation/time space mismatches (86–89). Furthermore, desert dust (90) and bubbles 359 (91) make the water appear greener. Other sources of scatter are the remote sensing limitation of 360 detecting SSchla at different depths in the water column (92, 93). 361 Global SSchla algorithms, calibrated with global in situ datasets, show average root mean square 362 (RMS) log errors of 34% (29). The Global Climate Observing System uncertainty requirement for 363 SSchla is 30% (www.ncdc.noaa.gov/gosic/gcos-essential-climate-variable-ecv-data-access- 364 matrix/gcos-ocean-biogeochemistry-ecv-ocean-color). End-users of SSchla data commonly quote 365 uncertainty requirement of 35% (83). There are a number of challenges in the calibration of 366 remote sensing data, like the diversity of optical properties in various water types, comparability 367 of ocean in situ measurements using different analytical approaches, and the different spatial and 368 temporal scales of the ocean-in situ and satellite measurements (83). 369 The available ocean colour satellite data since 1997 have been merged in the GlobColour project 370 (http://globcolour.info), which combines measurements from the Sea-viewing Wide Field-of-view 371 Sensor (SeaWiFS; 1997-2010), the MEdium Resolution Imaging Spectrometer (MERIS; 2002- 372 2012), the MODerate-resolution Imaging Spectroradiometer (MODIS; 2002-2017), and the Visible 373 Infrared Imaging Radiometer Suite (VIIRS; 2012-2017) missions. The spatial resolution of the 374 available data for end-users is 1/24º (4.63 km at the equator). The algorithms used for obtaining 375 SSchla are OC4v5 for SeaWiFS, OC4Me for MERIS and OC3v5 for MODIS and VIIRS (94). 376 SSchla values for every sediment location were extracted from merged global-scale matrixes of 377 the Globcolour Project, which contain monthly data. For this paper, we extracted SSchla 378 concentration data from September 1997 to December 2017 for every core location. GlobColour 379 data used in this study have been developed, validated, and distributed by ACRI-ST, France. We 380 used the Matlab script provided by the Monterey Bary Aquarium Research Institute 381 (https://www.mbari.org/index-of-downloadable-files/) to calculate the standard major axis 382 regressions (type II regressions) between SSchla and sedimentary alkenones. 383 384 385 386 Acknowledgments 387 388 This research used samples provided by the OSU-MGR Collection and the Scripps Institution of 389 Oceanography and the Integrated Ocean Drilling Program (IODP). We are in debt with Niko 390 Lahajnar and Zanna Chase for providing samples. We also thank Eric D. Galbraith and Markus 391 Kienast for their insightful comments and discussion. GlobColour data (http://globcolour.info) 392 used in this study have been developed, validated, and distributed by ACRI-ST, France. Ocean 393 Data View is acknowledged for providing software. We acknowledge financial support from the 8 394 European Research Council (PALADYN-834934), the Spanish research Ministry (CTM2013- 395 43006-P) and the Spanish Ministry of Science, Innovation and Universities, through the “María de 396 Maeztu” program for Units of Excellence (MDM-2015-0552). 397 398 The authors declare no competing interest. 399 400 401 References 402 403 1. J. L. Sarmiento, N. Gruber, Ocean Biogeochemical Dynamics (Princeton Univ. Press, 404 2006) https:/doi.org/10.1063/1.2754608. 405 2. D. A. Siegel, et al., Global assessment of ocean carbon export by combining satellite 406 observations and food-web models. Global Biogeochem. Cycles 28, 181–196 (2014). 407 3. F. E. Muller-Karger, et al., The importance of continental margins in the global carbon 408 cycle. Geophys. Res. Lett. 32, 1–4 (2005). 409 4. P. W. Boyd, H. Claustre, M. Levy, D. A. Siegel, T. Weber, Multi-faceted particle pumps 410 drive carbon sequestration in the ocean. Nature 568, 327–335 (2019). 411 5. K. E. Kohfeld, C. Le Quéré, S. P. Harrison, R. F. Anderson, Role of marine biology in 412 glacial-interglacial CO2cycles. Science (80-. ). 308, 74–78 (2005). 413 6. M. Zhao, J. L. Mercer, G. Eglinton, M. J. Higginson, C. Y. Huang, Comparative molecular 414 biomarker assessment of phytoplankton paleoproductivity for the last 160 kyr off Cap 415 Blanc, NW Africa. Org. Geochem. 37, 72–97 (2006). 416 7. S. D. Schoepfer, et al., Total organic carbon, organic phosphorus, and biogenic barium 417 fluxes as proxies for paleomarine productivity. Earth-Science Rev. 149, 23–52 (2015). 418 8. T. DeVries, T. Weber, The export and fate of organic matter in the ocean: New constraints 419 from combining satellite and oceanographic tracer observations. Global Biogeochem. 420 Cycles 31, 535–555 (2017). 421 9. R. Francois, S. Honjo, R. Krishfield, S. Manganini, Factors controlling the flux of organic 422 carbon to the bathypelagic zone of the ocean. Glob. Biogeochem. Cycles 16, 1087 (2002). 423 10. S. A. Henson, R. Sanders, E. Madsen, Global patterns in efficiency of particulate organic 424 carbon export and transfer to the deep ocean. Global Biogeochem. Cycles 26, 1–14 425 (2012). 426 11. R. F. Anderson, et al., Deep-Sea Oxygen Depletion and Ocean Carbon Sequestration 427 During the Last Ice Age. Global Biogeochem. Cycles 33, 301–317 (2019). 428 12. C. T. Bolton, et al., Glacial-interglacial productivity changes recorded by alkenones and 429 microfossils in late Pliocene eastern equatorial Pacific and Atlantic upwelling zones. Earth 430 Planet. Sci. Lett. 295, 401–411 (2010). 431 13. K. T. Lawrence, Evolution of the Eastern Tropical Pacific Through Plio-Pleistocene 432 Glaciation. Science (80-. ). 312, 79–83 (2006). 433 14. A. Moreno, I. Cacho, M. Canals, J. O. Grimalt, A. Sanchez-Vidal, Millennial-scale 434 variability in the productivity signal from the Alboran Sea record, Western Mediterranean 435 Sea. Palaeogeogr. Palaeoclimatol. Palaeoecol. 211, 205–219 (2004). 436 15. B. Petrick, et al., Oceanographic and climatic evolution of the southeastern subtropical 437 Atlantic over the last 3.5 Ma. Earth Planet. Sci. Lett. 492, 12–21 (2018). 438 16. F. G. Prahl, R. B. Collier, J. Dymond, M. Lyle, M. A. Sparrow, A biomarker perspective on 439 prymnesiophyte productivity in the northeast pacific ocean. Deep. Res. Part I 40, 2061– 440 2076 (1993). 441 17. F. Rostek, E. Bard, L. Beaufort, C. Sonzogni, G. Ganssen, Sea surface temperature and 442 productivity records for the past 240kyr in the Arabian Sea. Deep. Res. 44, 1461–1480 443 (1997). 444 18. O. Seki, et al., Reconstruction of paleoproductivity in the Sea of Okhotsk over the last 30 445 kyr. Paleoceanography 19 (2004). 9 446 19. S. Schmidt, et al., Particle export during a bloom of Emiliania huxleyi in the North-West 447 European continental margin. J. Mar. Syst. 109–110, S182–S190 (2013). 448 20. J. K. Volkman, S. M. Barrett, S. I. Blackburn, E. L. Sikes, Alkenones in Gephyrocapsa- 449 Oceanica - Implications for Studies of Paleoclimate. Geochim. Cosmochim. Acta 59, 513– 450 520 (1995). 451 21. J. K. Volkman, G. Eglinton, E. D. S. Corner, T. E. V. Forsberg, Long-chain alkenes and 452 alkenones in the marine coccolithophorid Emiliania huxleyi. Phytochemistry 19, 2619– 453 2622 (1980). 454 22. I. T. Marlowe, et al., Long chain (n-c37-c39) alkenones in the prymnesiophyceae. 455 distribution of alkenones and other lipids and their taxonomic significance. Br. Phycol. J. 456 19, 203–216 (1984). 457 23. L. L. Rhodes, B. M. Peake, A. L. MacKenzie, M. Simon, A. R. Edwards, Coccolithophores 458 Gephyrocapsa oceanica and Emiliania huxleyi (Prymnesiophyceae = Haptophyceae) in 459 New Zealand’s coastal waters: Characteristics of blooms and growth in laboratory culture. 460 New Zeal. J. Mar. Freshw. Res. 29, 345–357 (1995). 461 24. M. H. Conte, J. K. Volkman, G. Eglinton, Lipid biomarkers of the Prymnesiophyceae. The 462 Haptophyte Algae 51, 351–377 (1994). 463 25. M. J. Behrenfeld, E. Boss, D. A. Siegel, D. M. Shea, Carbon-based ocean productivity and 464 phytoplankton physiology from space. Global Biogeochem. Cycles 19, 1–14 (2005). 465 26. Y. Huot, et al., Relationship between photosynthetic parameters and different proxies of 466 phytoplankton biomass in the subtropical ocean. Biogeosciences 4, 853–868 (2007). 467 27. Z. Lee, J. Marra, M. J. Perry, M. Kahru, Estimating oceanic primary productivity from 468 ocean color remote sensing: A strategic assessment. J. Mar. Syst. 149, 50–59 (2015). 469 28. J. W. Campbell, The lognormal distribution as a model for bio-optical variability in the sea. 470 J. Geophys. Res. 100, 13237–13254 (1995). 471 29. R. J. W. Brewin, et al., Uncertainty in Ocean-Color Estimates of Chlorophyll for 472 Phytoplankton Groups. Front. Mar. Sci. 4 (2017). 473 30. S. Dutkiewicz, A. E. Hickman, O. Jahn, Modelling ocean-colour-derived chlorophyll a. 474 Biogeosciences 15, 613–630 (2018). 475 31. Global Climate Observing System, “Systematic Observation Requirements For Satellite- 476 Based Data Products for Climate - 2011 Update” (2011). 477 32. IOCCG, Partition of the Ocean into Ecological Provinces : Role of Ocean-Colour 478 Radiometry, M. Dowell, T. Platt, Eds. (Reports ofthe International Ocean-Colour 479 Coordinating Group, No. 9, IOCCG, 2009). 480 33. T. Jackson, S. Sathyendranath, F. Mélin, “An improved optical classification scheme for 481 the Ocean Colour Essential Climate Variable and its applications” (Elsevier Inc., 2017). 482 34. T. Moore, J. Campbell, M. Dowell, A class-based approach to characterizing and mapping 483 the uncertainty of the MODIS ocean chlorophyll product. Remote Sens. Environ. 113, 484 2424–2430 (2009). 485 35. I. Shulman, et al., Bio-Optical Data Assimilation With Observational Error Covariance 486 Derived From an Ensemble of Satellite Images. J. Geophys. Res. Ocean. 123, 1801–1813 487 (2018). 488 36. W. W. Gregg, N. W. Casey, Global and regional evaluation of the SeaWiFS chlorophyll 489 data set. Remote Sens. Environ. 93, 463–479 (2004). 490 37. C. Hu, L. Feng, Z. Lee, Uncertainties of SeaWiFS and MODIS remote sensing reflectance: 491 Implications from clear water measurements. Remote Sens. Environ. 133, 168–182 492 (2013). 493 38. C. R. McClain, A Decade of Satellite Ocean Color Observations. Ann. Rev. Mar. Sci. 1, 494 19–42 (2009). 495 39. M. Stramska, D. Stramski, H. Ryszard, S. Kaczmarek, S. Joanna, Bio-optical relationships 496 and ocean color algorithms for the north polar region of the Atlantic. J. Geophys. Res. 497 108, 3143 (2003). 498 40. A. Mendonça, et al., Evaluation of ocean color and sea surface temperature sensors 499 algorithms using in situ data: a case study of temporal and spatial variability on two 10 500 northeast Atlantic seamounts. J. Appl. Remote Sens. 4, 043506 (2010). 501 41. J. N. Waite, F. J. Mueter, Spatial and temporal variability of chlorophyll-a concentrations in 502 the coastal Gulf of Alaska, 1998-2011, using cloud-free reconstructions of SeaWiFS and 503 MODIS-Aqua data. Prog. Oceanogr. 116, 179–192 (2013). 504 42. P. J. Lam, S. C. Doney, J. K. B. Bishop, The dynamic ocean biological pump: Insights 505 from a global compilation of particulate organic carbon, CaCO3, and opal concentration 506 profiles from the mesopelagic. Global Biogeochem. Cycles 25, 1–14 (2011). 507 43. C. M. Marsay, et al., Attenuation of sinking particulate organic carbon flux through the 508 mesopelagic ocean. Proc. Natl. Acad. Sci. 112, 1089–1094 (2015). 509 44. T. Weber, J. A. Cram, S. W. Leung, T. DeVries, C. Deutsch, Deep ocean nutrients imply 510 large latitudinal variation in particle transfer efficiency. Proc. Natl. Acad. Sci. 113, 8606– 511 8611 (2016). 512 45. M. Yamamoto, et al., Seasonal and depth variations in molecular and isotopic alkenone 513 composition of sinking particles from the western North Pacific. Deep. Res. Part I 514 Oceanogr. Res. Pap. 54, 1571–1592 (2007). 515 46. P. J. Müller, G. Fischer, A 4-year sediment trap record of alkenones from the filamentous 516 upwelling region off Cape Blanc, NW Africa and a comparison with distributions in 517 underlying sediments. Deep. Res. Part I Oceanogr. Res. Pap. 48, 1877–1903 (2001). 518 47. E. Suess, Particulate organic carbon flux in the oceans—surface productivity and oxygen 519 utilization. Nature 288, 260–263 (1980). 520 48. J. H. Martin, G. A. Knauer, D. M. Karl, W. W. Broenkow, VERTEX: carbon cycling in the 521 northast Pacific. 34, 267–285 (1987). 522 49. R. S. Lampitt, A. N. Antia, Particle flux in deep seas: Regional characteristics and 523 temporal variability. Deep. Res. Part I Oceanogr. Res. Pap. 44, 1377–1403 (1997). 524 50. M. Lutz, R. Dunbar, K. Caldeira, Regional variability in the vertical flux of particulate 525 organic carbon in the ocean interior. Global Biogeochem. Cycles 16, 11-1-11–18 (2002). 526 51. R. A. Jahnke, The global ocean flux of particulate organic carbon : Areal distribution and 527 magnitude. 10, 71–88 (1996). 528 52. C. Pelejero, J. O. Grimalt, M. Sarnthein, L. Wang, J. A. Flores, Molecular biomarker record 529 of sea surface temperature and climatic change in the South China Sea during the last 530 140,000 years. Mar. Geol. 156, 109–121 (1999). 531 53. M. J. Higginson, M. A. Altabet, Initial test of the silicic acid leakage hypothesis using 532 sedimentary biomarkers. Geophys. Res. Lett. 31, 4–7 (2004). 533 54. C. T. Bolton, K. T. Lawrence, S. J. Gibbs, P. A. Wilson, T. D. Herbert, Biotic and 534 geochemical evidence for a global latitudinal shift in ocean biogeochemistry and export 535 productivity during the late Pliocene. Earth Planet. Sci. Lett. 308, 200–210 (2011). 536 55. W. K. W. Li, Composition of ultraphytoplankton in the central North Atlantic. 122, 1–8 537 (1995). 538 56. A. Z. Worden, J. K. Nolan, B. Palenik, Assessing the dynamic and ecology of marine 539 picoplankton : the importance of eukaryotic component. Limnol. Ocean. 49, 168–79. 540 (2004). 541 57. E. Teira, et al., Variability of chlorophyll and primary production in the Eastern North 542 Atlantic Subtropical Gyre: Potential factors affecting phytoplankton activity. Deep. Res. 543 Part I Oceanogr. Res. Pap. 52, 569–588 (2005). 544 58. X. A. G. Morán, Annual cycle of picophytoplankton photosynthesis and growth rates in a 545 temperate coastal ecosystem: A major contribution to carbon fluxes. Aquat. Microb. Ecol. 546 49, 267–279 (2007). 547 59. Y. M. Rii, et al., Diversity and productivity of photosynthetic picoeukaryotes in 548 biogeochemically distinct regions of the South East Pacific Ocean. Limnol. Oceanogr. 61, 549 806–824 (2016). 550 60. V. S. Smetacek, Role of sinking in diatom life-history cycles: ecological, evolutionary and 551 geological significance. Mar. Biol. 84, 239–251 (1985). 552 61. C. Sancetta, T. Villareal, P. Falkowski, Massive fluxes of rhizosolenid diatoms : A common 553 occurrence ? Notes, 1452–1457 (1989). 11 554 62. A. F. Michaels, M. W. Silver, Primary production, sinking fluxes and the microbial food 555 web. Deep. Res. 35, 473–490 (1988). 556 63. A. M. Waite, K. A. Safi, J. A. Hall, S. D. Nodder, Mass sedimentation of picoplankton 557 embedded in organic aggregates. Limnol. Oceanogr. 45, 87–97 (2000). 558 64. T. L. Richardson, G. A. Jackson, H. W. Ducklow, M. R. Roman, Carbon fluxes through 559 food webs of the eastern equatorial Pacific: An inverse approach. Deep. Res. Part I 560 Oceanogr. Res. Pap. 51, 1245–1274 (2004). 561 65. T. L. Richardson, G. A. Jackson, H. W. Ducklow, M. R. Roman, Spatial and seasonal 562 patterns of carbon cycling through planktonic food webs of the Arabian Sea determined by 563 inverse analysis. Deep. Res. Part II Top. Stud. Oceanogr. 53, 555–575 (2006). 564 66. S. L. C. Giering, et al., High export via small particles before the onset of the North Atlantic 565 spring bloom. J. Geophys. Res. Ocean. 121, 2268–2285 (2016). 566 67. G. Dall’Olmo, K. A. Mork, Carbon export by small particles in the Norwegian Sea. 567 Geophys. Res. Lett., 1–6 (2014). 568 68. T. L. Richardson, G. A. Jackson, Small Phytoplankton and Carbon Export from the 569 Surface Ocean. Science (80-. )., 838–840 (2007). 570 69. K. G. J. Nierop, G. J. Reichart, H. Veld, J. S. Sinninghe Damsté, The influence of oxygen 571 exposure time on the composition of macromolecular organic matter as revealed by 572 surface sediments on the Murray Ridge (Arabian Sea). Geochim. Cosmochim. Acta 206, 573 40–56 (2017). 574 70. E. Kristensen, M. Holmer, Decomposition of plant materials in marine sediment exposed 575 to different electron acceptors (O2, NO3−, and SO42−), with emphasis on substrate origin, 576 degradation kinetics, and the role of bioturbation. Geochim. Cosmochim. Acta 65, 419– 577 433 (2001). 578 71. B. Dauwe, J. J. Middelburg, P. M. J. Herman, Effect of oxygen on the degradability of 579 organic matter in subtidal and intertidal sediments of the North Sea area. Mar. Ecol. Prog. 580 Ser. 215, 13–22 (2001). 581 72. D. E. Canfield, Factors influencing organic carbon preservation in marine sediments. 582 Chem. Geol. 114, 315–329 (1994). 583 73. H. E. Hartnett, R. G. Keil, J. I. Hedges, A. H. Devol, Influence of oxygen exposure time on 584 organic carbon preservation in continental margin sediments. Nature 391, 572–574 585 (1998). 586 74. K. A. Koho, et al., Microbial bioavailability regulates organic matter preservation in marine 587 sediments. Biogeosciences 10, 1131–1141 (2013). 588 75. M. M. M. Kuypers, R. D. Pancost, I. A. Nijenhuis, J. S. Sinninghe Damsté, Enhanced 589 productivity led to increased organic carbon burial in the euxinic North Atlantic basin 590 during the late Cenomanian oceanic anoxic event. Paleoceanography 17, 3-1-3–13 591 (2002). 592 76. M. Rodrigo-Gámiz, S. W. Rampen, S. Schouten, J. S. Sinninghe Damsté, The impact of 593 oxic degradation on long chain alkyl diol distributions in Arabian Sea surface sediments. 594 Org. Geochem. 100, 1–9 (2016). 595 77. A. W. Jacobel, et al., Deep Pacific storage of respired carbon during the last ice age: 596 Perspectives from bottom water oxygen reconstructions. Quat. Sci. Rev. 230 (2020). 597 78. W. Lu, et al., I/Ca in epifaunal benthic foraminifera: A semi-quantitative proxy for bottom 598 water oxygen in a multi-proxy compilation for glacial ocean deoxygenation. Earth Planet. 599 Sci. Lett. 533, 116055 (2020). 600 79. A. Rosell-Melé, et al., Precision of the current methods to measure the alkenone proxy 601 U37K′ and absolute alkenone abundance in sediments: Results of an interlaboratory 602 comparison study. Geochemistry, Geophys. Geosystems 2 (2001). 603 80. J. P. Dunne, B. Hales, J. R. Toggweiler, Global calcite cycling constrained by sediment 604 preservation controls. Global Biogeochem. Cycles 26, 1–14 (2012). 605 81. O. Cartapanis, D. Bianchi, S. L. Jaccard, E. D. Galbraith, Global pulses of organic carbon 606 burial in deep-sea sediments during glacial maxima. Nat. Commun. 7, 10796 (2016). 607 82. IOCCG, “Remote Sensing of Ocean Colour in Coastal , and Other Optically-Complex , 12 608 Waters” (2000). 609 83. S. B. Groom, et al., Satellite ocean colour: Current status and future perspective. Front. 610 Mar. Sci. 6 (2019). 611 84. C. W. Brown, J. A. Yoder, Cocolithophorid blooms in the global ocean. J. Geophys. Res. 612 99, 7467–7482 (1994). 613 85. D. Blondeau-Patissier, J. F. R. Gower, A. G. Dekker, S. R. Phinn, V. E. Brando, A review 614 of ocean color remote sensing methods and statistical techniques for the detection, 615 mapping and analysis of phytoplankton blooms in coastal and open oceans. Prog. 616 Oceanogr. 123, 123–144 (2014). 617 86. H. M. Dierssen, R. C. Smith, Bio-optical properties and remote sensing ocean color 618 algorithms for Antarctic Peninsula waters. 105, 26301–26312 (2000). 619 87. S. Sathyendranath, G. Cota, V. Stuart, H. Maass, T. Platt, Remote sensing of 620 phytoplankton pigments: A comparison of empirical and theoretical approaches. Int. J. 621 Remote Sens. 22, 249–273 (2001). 622 88. H. Claustre, S. Maritonera, The Many Shades of Ocean Blue. Ocean Sci. 302, 1514–1515 623 (2003). 624 89. G. Volpe, et al., The colour of the Mediterranean Sea: Global versus regional bio-optical 625 algorithms evaluation and implication for satellite chlorophyll estimates. Remote Sens. 626 Environ. 107, 625–638 (2007). 627 90. H. Claustre, et al., Is desert dust making oligotrophic waters greener? Geophys. Res. Lett. 628 29, 107-1-107–4 (2002). 629 91. D. Stramski, J. Tegowski, Effects of intermittent entrainment of air bubbles by breaking 630 wind waves on ocean reflectance and underwater light field. J. Geophys. Res. 106, 631 31345–31360 (2001). 632 92. J. J. Cullen, The deep chlorophyll maximum: Comparing vertical profiles of chlorophyll a. 633 Can. J. Fish. Aquat. Sci. 39, 791–803 (1982). 634 93. J. Huisman, N. N. Pham Thi, D. M. Karl, B. Sommeijer, Reduced mixing generates 635 oscillations and chaos in the oceanic deep chlorophyll maximum. Nature 439, 322–325 636 (2006). 637 94. J. E. O’Reilly, et al., SeaWiFS Postlaunch Calibration and Validation Analyses, part 3. 638 NASA Tech. Memo. 11, 1–49 (2000). 639 13 640 641 642 Figures and Tables 643 644 645 646 Fig. 1. Global core-top sediments distribution. Lines delineate distinct biogeochemical regions 647 defined on the basis of temperature and nutrient concentration (44) Abbreviations: subarctic (SA), 648 subtropics (ST), tropics (T) and Southern Ocean (SO). 649 14 650 651 652 Fig. 2. Global correlations between sedimentary C37 alkenones concentration and the sum of 653 sea-surface chlorophyll-a (SSchla) concentration from 1997 to 2017. a. includes the whole 654 compilation, and b. includes regions presenting RMS log errors lower than 31% between remote 655 sensing and in situ SSchla. Samples are classified by biogeochemical regions as defined in Fig. 656 1. 657 658 15 659 660 661 Fig. 3. Regional correlations between sedimentary C37 alkenones concentration and the sum of 662 sea-surface chlorophyll-a (SSchla) concentration from 1997 to 2017. Different colours for lines 663 and text indicate the evaluated biogeochemical regions, which are defined in Fig. 1. 664 665 16 666 667 668 Fig. 4. Global comparison of sedimentary C37 alkenones concentration and oxygen concentration 669 in bottom waters. Different colours for lines and text indicate the evaluated biogeochemical 670 regions, which are defined in Fig. 1. 671 672 673 17 674 675 Table 1. Equation coefficients and errors from correlations in Fig. 2 and 3. Log[C37 676 alkenones]=a·log∑[SSchla]+b. Abbreviations: sea-surface chlorophyll-a (SSchla), coefficient of 677 correlation (r), coefficient of determination (R2), root-mean square logarithmic error (RMS log error), 678 number of samples (n). 679 2 Equation Region a b r R RMS log error (%) n 1 Global 1.55 -3.42 0.67 0.45 53 226 2 Global 1.64 -3.46 0.77 0.60 38 194 3 Subtropics 1.31 -2.87 0.77 0.59 31 75 4 Tropics 1.77 -3.77 0.70 0.49 46 88 5 Southern Ocean 2.29 -4.64 0.57 0.33 35 26 680 681 682 683 18