Transient hydrodynamic effects influence organic carbon signatures in marine sediments

Ocean dynamics served an important role during past dramatic climate changes via impacts on deep-ocean carbon storage. Such changes are recorded in sedimentary proxies of hydrographic change on continental margins, which lie at the ocean–atmosphere–earth interface. However, interpretations of these records are challenging, given complex interplays among processes delivering particulate material to and from ocean margins. Here we report radiocarbon (14C) signatures measured for organic carbon in differing grain-size sediment fractions and foraminifera in a sediment core retrieved from the southwest Iberian margin, spanning the last ~25,000 yr. Variable differences of 0–5000 yr in radiocarbon age are apparent between organic carbon in differing grain-sizes and foraminifera of the same sediment layer. The magnitude of 14C differences co-varies with key paleoceanographic indices (e.g., proximal bottom-current density gradients), which we interpret as evidence of Atlantic–Mediterranean seawater exchange influencing grain-size specific carbon accumulation and translocation. These findings underscore an important link between regional hydrodynamics and interpretations of down-core sedimentary proxies.

case. Figure 1: In panel B, the estimated depth of deglacial INL is marked. It would be useful to briefly explain what this reconstruction is based on. Figure 2: The blue shading between the lines indicating the ages of the different grain size fractions does not help much and should be removed. The age differences are better visible in Figure 3. Axis label "Sediment Fraction 14C Age" should use the unit "ky BP", and axis label "Calendar Age" should be in "ky" (an "age" cannot be "ago") Figure 3: In panels A, B, and F, the errors associated with the measurements should be displayed. In the caption, the reader should be made aware that axis in panel B is reversed. It is very hard to see the different colours used in Figure H, in particular the lavender line is hard to read.
Reviewer #2 (Remarks to the Author): The manuscript « Transient hydrodynamic effects influence organic carbon signatures in marine sediments » submitted by Magill et al. to Nature Communications is a novel and elegant work based on 126 radiocarbon ages on 21 paired discrete samples of clay, fine silt, coarse silt and sand compared to those obtained on bulk organic carbon and foraminifera from a southwestern Iberian margin core covering the last 25,000 years. It shows radiocarbon age offsets that are convincingly interpreted by the authors as the result of differential lateral transfer dynamics among grain-size classes during the last termination (~23,000-9,000 years ago). This differential lateral transfer would be mediated by changes in the intermediate nepheloid layer due to the dynamic of paleocurrent densities. The identification of this differential lateral transfer is very important for understanding climate dynamics as it can contribute to explain the observed leads and lags among proxies preserved in different grain-size classes. This work deserves publication in Nat. Comm. after the minor revisions I have listed below: -In lines 80-87, I have found several inconsistencies between the text, Figures 2 and 3 and Table  1 of the supplementary information. In the Table 1 and Figure 2, the maximum radiocarbon age of OC and the maximum calendar age of foraminifera are 21,409±84 and 21,728 years, respectively while in the text the values are 20,475±775 yr and 23,250 yr BP, respectively. In Figure 3, the oldest sample has also a calendar age of 23,250 yr BP. Similarly, the authors claim a consistent down-core difference between bulk sediment OC and foraminifera 14C ages of 1450±200 years but this difference is not observed at 65-66 cm where it is only of few centuries. Finally, the radiocarbon offset between organic carbon in paired fine or coarse-silts and foraminifera is between ~1000 and 3500 years ( Figure 3) and not between ~1000 and 3000 years as indicated in the text by the authors.
-Lines 97-107: In contrast with the author's statement, the relatively low radiocarbon offsets during the LGM do not transition into more moderate offsets during preliminary glacial termination. This first part of the glacial termination, ~19-17.5 kyr ago, lack of samples. This interval corresponds with a period of meltwater release into the Arctic/Nordic Seas when North Atlantic Deep Water (NADW) collapse (McManus et al., Nature, 2004;Standford et al., 2013, QSR, 2011 that may have strongly affected the northeast Atlantic margin and, consequently the 14C ages. I suggest to the authors that they rephrase this paragraph, and replace the solid lines between LGM and HS 1 with dashed lines in Figure 3A, 3B and 3F. -Line 128: Replace "…the MOW descended to at least 2600 mbsl during past Heinrich stadials as a consequence of decreased Atlantic-Mediterranean seawater exchange" with "…the MOW … as a consequence of increased…". -Lines 151-152: The authors suggest that periods of strong MOW flow would result in more fine grained terrigenous clastic transport to the studied site. I wonder whether the increased aridity associated with stronger MOW (e.g. Llave et al., 2006) could enhance erosion in the adjacent landmasses and contribute to the increase arrival of fine grained terrigenous clastic particles through the canyons of Cascais, Setubal and Lisbon (Jouanneau et al., 1998, Progress in Oceanography). The authors should discuss this issue.
-Lines 138 and 192: Replace RFS-y and RCS-y with "RFS-C and RCS-C" -Lines 235-241: the age offsets between the different SST proxies are not clear on Figure 3H. Also, the different blues of the SST curves are not clearly distinguishable. The authors should add a curve showing the variation of these offsets through time and chose distinct colors.
Reviewer #3 (Remarks to the Author): Evaluation of Transient hydrodynamics effects influence organic carbon signatures in marine sediments by Magill et al.
Although I support the overall conclusion that pre-depositional processes and lateral dynamics can differentially impact proxies with different grain-size associations, I have identified multiple issues that need to be resolved before publication. The data are of high quality and the writing is clear but sometimes the authors are little too confident about their interpretation. The conclusiveness of this paper needs additional work.
Major comments: One, I have some reservations with their neglect of bioturbation. Bioturbation can also cause sizedependent displacement and thus age offsets as the authors do acknowledge. However, they then simply state that high accumulation rates (order of 15-20 cm/kyr) are enough to ignore bioturbation. This is questionable. A simple Peclet number analysis for a mixed layer of 10 cm (maximum), a sediment accumulation of 20 cm/kyr and a biodiffusion coefficient of 2 cm2/yr, reveals that mixing is more important than sediment accumulation (Pe<1). (Pe=w*L/D= 0.02*10/2= 0.1). A shallower mixing layer would even increase the relative importance of bioturbation mixing vs. accumulation. Moreover, the authors do present excess 210 Pb profiles in their supplementary table. They interpret these data with an accumulation model ignoring bioturbation. This may need revision; 210 Pb vs. depth profiles are governed by radioactive decay, sediment accumulation and bioturbation (ignoring compaction). Decay is known, but accumulation and bioturbation not. You can only separate these two having another, independent tracer, like for instance the 1963 or 1986 Cs peaks. Based on their model the upper 8 cm accumulated in about 160 years, in other words, about 50 cm/kyr. Compaction can account for part but not for all the difference. Most modern sediment biogeochemists would interpret these 210Pb profiles as evidence of bioturbation. Their mixed layer is about 7 cm, not unlike the global average bioturbated layer of about 10 cm (Boudreau paper). The age of their mixed layer is about 150-200 years based on 210 Pb, but order of 1000 years based on 14C (still 4 times higher if you correct for 14C reservoir effect). This being said, bioturbators go for food (organic matter) and OM input are related to grain size sorting and lateral processing.
Two, the text presented in lines 97-107 does not align with what figure 3 shows and the authors impose an interpretation rather than letting the reader infer her/himself the story. Line: 100-101: Overall lower radiocarbon offsets… during the LGM (this is based on a single data point!) transition into more moderate offsets of 2000 plus/minus 500 yr during preliminary glacial termination (not supported by a single data point!) and then rapidly peak to values of 1500-3000 yr during HS1 (the clays offsets are the lowest in HS1 rather than highest), and so on. For instance, line 104 after an interim of low radiocarbon offset throughout B/A interstadial, 14.7-12.8 kyr ago), the relative age differences… increase again. I really do see an increase during BA and decrease during YD. This whole section does not properly link to observation in Figure 3. Why is there no zero age in figure 3 and why are two zero offsets in Figure 2 for the clays and only one in Figure 3. Aren't these the same data.
Line 8: delivering particulate material to ocean shelves or from ocean shelves?
Line 90: large flocculates do indeed occur, but whether they survive settling to seafloor and diagenetic alteration in sediments remains to be seen. The reference cited here, as I recall, does not consider sediment aggregates, but water-column ones. Use other literature to support this claim.
Line 125: grain-size specific resuspension has also implication for how organisms mix material around.
Line 138:… with RFS-y I presume y should be replaced with C?
Line 140: … versus autochthonous clays.. Rewrite, this could be misread as new clay formation (the reverse weathering story).
Line 158: where is the evidence against bioturbation. I would say evidence for substantial bioturbation.
Line 172-174: the logic needs attention here. In principle OC should decrease if accumulation rates increase (simply by dilution), but usually OC and mineral delivery to sediments co-vary and OC (wt%) are rather invariant. Burial efficiencies increase of course with accumulation.
Line 185: why specifically oxic degradation? Is pre-depositional degradation not more correct?
Line 191-194: to make this more conclusive, OM quality/composition data would be needed.
Line 232-234: age offset of grain-size fractions go from zero to almost 3000, many of which 1500-2000, yet the lag (offset) in panel H, is only of the order of 200-300 years. This needs to be reconciled. Part of this may be explained by differences among sites, but how strong is your case then.

Reviewer #1 (Remarks to the Author):
The authors present a new data set on 14 C ages of bulk organic carbon (OC) in different grain size classes of sediments from the Iberian continental margin, which are compared with 14 C ages of planktic foraminifera taken to reflect depositional age. They observe large and variable offsets between the OC in different grain size fraction and the depositional age and suggest sedimentology to control these, which has further implications for expected leads and lags of proxy records associated with different grain size classes. This is a very interesting study taking a novel approach, which has great potential to improve the understanding of the effects of sedimentology on proxy records.
While I agree with the general hypothesis that differences in age of different grain size fractions likely reflect differential transport histories of the respective materials in response to their hydrodynamic properties, I am not convinced by the suggested mechanisms put forward by the authors to explain their observations. First of all, I do not see the "strong positive relationship" (line 122) described for down-core values of age offsets between fine silt and foraminifera with XRF derived ln(Zr/Al) values (as a side note: in the text often referred to as the linear ratio, while in the plots the logarithmic version is presented -this should be adjusted).
o There are obvious visual linear relationships apparent between calculated R F-FS/CS and XRF-derived ln(Zr/Al) (r 2 = 0.242 and 0.292, respectively [p-values < 0.05]); however, we agree that the correlation strength of this relationship is weak-to-moderate as opposed to strong. Furthermore, we also observed that the strength of this relationship is subject 'influential points' (SHAK06-5K sample depths of 12-23, 24-25, and 50-51 cm [equivalent to calibrated 14 C ages of ~670 to 3420 yr cal BP]). Thus, we have revised this section of text to more accurately present these relationships (L104-116). o We used the natural logarithm of Zr/Al to improve data normality (i.e., minimize residuals distribution), which is necessary for meaningful interpretations of Pearson/Spearman correlation coefficients in dynamic systems 1 . However, a strong positive correlation is apparent (r 2 = 0.995) between absolute and transformed data values since logarithmic adjustments show a stronger influence on more extreme values of the response variable. Even so, we revised the manuscript text to explain our approach and furthermore corrected the corresponding plots (c.f., Fig. 3

legend).
While there seems to be an agreement for the broad HS1 interval (with some differences in timing, which might be explained by differences in the resolution of the two data sets), I do not observe such a relationship for the YD interval. Instead, there is a strong maximum in the age offset in the BA (or a local minimum defined by just one measurement around 14.5 kyr) followed by a decline during the YD toward the early Holocene. The authors should be more careful describing such apparent agreements, in particular since their sediment core has a very high sedimentation rate and, likely, a very good age control, as it is tuned to the very well dated record MD99-2334K (another side note: a better description of the age model used and its associated uncertainties should be provided, at least in the supplement; the figures would benefit by indicating age tie points for the records of this new core SHAK06-05K).
o No problem! We revised the Results section of this manuscript to more accurately describe apparent trends and down-core correlations in an effort to resolve this important issue (c.f., L104-116 and L151-159). o Likewise, multiple reviewers commented that further details are needed for our agedepth models, so we have added body (c.f., Methods section) and supplemental text in addition to marking pertinent tie-points in Supplementary Fig. 1 (c.f., L74 rebuttal below, as well). This additional figure and text describes our age models, uncertainty propagation, and the alignment techniques we used to compare down-core records.
Lateral transport of the silt fractions (coarse and fine silt) is suggested to occur over long timescales (and distances) in the intermediate nepheloid layers, and current flow strength controls the amount of laterally supplied silt. This suggestion includes the inherent assumption that material transported in INLs is pre-aged. This requires INL material to be resuspended after having been deposited elsewhere. Such re-suspended material is typically not transported very far, as these particles settle rather rapidly. I would like the authors to present evidence that pre-aged material is entrained in these INLs. Current velocities of the currents that are supposed to transport the silt fractions should also be considered in order to evaluate the likelihood of the proposed mechanism. o A very pragmatic approach to ascertaining 'pre-aged' bulk OM in nepheloid layers would be a comparison of the radiocarbon ages among grain-size sediment fractions in surface sediments and the concomitant SPM, though (unfortunately) this is well beyond the scope of this study. Even so, at a minimum, the occurrence of reworked Cretaceous nannofossils along the SW Iberian margin -and at SHAK06-5K (Fig. 3C) -suggests more ancient (pre-aged) material is transported from at least hundreds of kilometres away 2,3 because the nearest Cretaceous outcrops occur in southeast Spain 4 , although mud volcanoes of the southeast Cadiz diapiric ridge might likewise contribute some reworked Cretaceous material 5 . Concomitant Neogene nannofossils eroded from widespread Miocene outcrops along the Algarve margin insinuate more vicinal transport (10-100 km) from shoreline eroding processes, such as sea-level transgression 2 , although inconsistent trends observed between nannofossil distribution spectra 6 as compared to shoreline evolution 7 argue against decisive eustatic influences on reworking patterns since at least 25 kyr ago. o Rather, covariates of Iberian eustatic change, such as riverine suspended loads and discharge volume 8 , could drive apparent fluctuations in allochthonous sediment flux 9 . Such mechanisms square with the data shown in Fig. 3A,B in that there is a weak-tomoderate correlation apparent between radiocarbon offsets and local Iberian margin sea-level fluctuations (r 2 = <0.098-0.476) that features alongside much stronger correlations with inferred benthic oxygenation (Fig. 3D), MOW flow depth (Fig. 3E) and bottom-water circulation strength (Fig. 3G), each of which are subject to indirect, though co-variable, eustatic influences on water density gradients and INL formation 10-13 via nonlinear internal wave dynamics 14,15 . The mechanisms also square with studies of recent Iberian deep-sea water masses with low ratios of POC against total SPM (tSPM) that indicate a 're-suspended' origin of this organic carbon-rich suspended fraction 16 . High ratios of POC to corresponding pigment levels at depths of 1000 m and deeper moreover suggest that typical POC fallout does not incorporate much fresh material, but instead features more refractory ('pre-aged') OM that has undergone numerous (re)suspension cycles 16,17 . o Interestingly, potential for sediment (re)suspension is often calculated from the ratio of force exerted by shear stress, which acts to move appertaining grains across a substrate, as compared to the effective weight of the counteracting grains (c.f., Shields mobility parameter), which is dependent upon water density gradients at the sediment interface 14,18 . Upon resuspension, differences in settling particle velocities and their cohesiveness 19 will drive size-separation of coarse material from finer sediments, which will remain in suspension comparatively longer 14 . o With respect to current flow velocities, we added L254-L265. Further, we note available data indicate median grain-size of disaggregated particles within the nepheloid fall between about 10-20 µm at depths of 2000 m and below 16 along the Iberian margin. Associated benthic aggregates are in general <125 µm in diameter 20 . These data are also consistent with other studies suggesting progressive (re)suspension cycles within nepheloid layers results in episodic, although constant, horizontal fluxes (i.e., advection) of flocculated SPM together with rapid downslope influx of fresh phytodetritus 16,21 . With this in mind, typic organic matter aggregates show residence times of up to several months between (re)suspension events 20 , during which smaller aggregates (<125 µm) might travel distances of 200 km (N.B., displacement distances of 10-100 km) or more with episodic 22 rates of 1-10 km yr -1 (refs. 20,23,24).
The sediment at this core site accumulated at high but variable rates. The sedimentation rate estimates are presented in the supplement. Rough inspection reveals that the variability of the sedimentation rates resembles strongly the records of age offsets presented in Figure 3A, suggesting a coupling of sedimentation rate and the processes responsible for the age offset. Moreover, this agreement is likely better than that with ln(Zr/Al) ratios, which is discussed at length in the manuscript. The implications of this observation should be discussed. o Building upon our responses above, there is a notable correlation apparent between radiocarbon offsets and local Iberian margin sea-level fluctuations. However, there are some important differences in down-core evolution viz. slight decreases in sedimentation rate amid Heinrich Stadial 1 despite marked increases in radiocarbon offsets and the relative abundance of reworked nannofossils (c.f., Fig. 2 and Fig.  3A,C). Higher sedimentation rates amid the Last Glacial Maximum and latest Holocene also do not feature increases in reworked nannofossils. Considered together, we suggest that these data underscore a more complex or/and dynamic mechanism related to eustatic effects indirectly, such as continental hydroclimate (e.g., precipitation and fluvial discharge) in conjunction with hydrodynamic effects vis-à-vis current depth 25 (Fig. 3E), coastline evolution 7 (Fig. 3B), and bottom-water oxygen conditions 26 (Fig. 3D).
As the core site is located at the continental margin, and the time period investigated includes a transition from low to high sea-level stand, the effect of shelf erosion induced by sea-level rise in re-distributing pre-aged sediment should be considered, too. Interestingly, the increase in age offsets (and the strongest maximum in age offsets between silt and clay) occurs approximately coevally with the onset of global sea-level rise at ~16.8 kyr as reconstructed by Lambeck et al., (2014). The authors should evaluate when exactly the (outer) shelf was flooded in the study area (using shelf bathymetry and the sea-level curve). o The Lambeck et al. 27 sea-level curve represents global eustatic effects, which might fail to capture more local developments associated with isostatic and tectonic influences 28 . Therefore, we adopt the more specific sea-level curve (c.f., Figure 3B) derived from esturine sediments along the Gulf of Cadiz coast 8 together with specific reconstructions of Iberian coastline evolution since the LGM 7 . o There was an estimated LGM sea-level low of ~130m in the Gulf of Cadiz 7,29 . Amid this low-stand, fluvial channels of the Guadalquivier river incised the shelf 30 , leading to substantial lateral erosion, especially just before the Bølling/Allerød interstadial. Thereafter, available reconstructions indicate outer shelf flooding occurred ~16 ky BP, followed by punctuated transgression events around 13-11 ky BP, during which the sea stood ~80 m below its current level 9 . With this in mind, the available sedimentological features and nanno/microfossil data suggest that fluctuations in regional (39-43 °N) lower-slope sediment flux were driven by a combination of bioproduction, sea-level transgression, and detrital input 7,31,32 , but deep-sea current (hydro)dynamics controlled local deposition since at least 30 kyr ago (ref. 31).
The data plots are presented without uncertainties, which may be intentional to ensure readability. However, at least for the relatively low resolution 14 C data, error bars should be displayed (Fig. 3A, F). o No problem! Age uncertainty bars have been added to corresponding panels of Figure 3, though we are interested in our peer's opinions about figure readability.
Overall, I think the study has a great potential to reveal the effect of sediment redistribution on proxy records, but more care must be taken in evaluating the possible mechanisms responsible for sediment transport, which each have a different implication. o Thanks for these encouraging words! We agree that this study has exciting potential, and have attempted to develop this potential by refining (c.f., L171-177) and better describing possible mechanisms associated with Iberian margin sediment transport (c.f., L171-177, L193-197 and L271-275). In particular, we attempted to describe more clearly potential transmission mechanisms associated with MOW flow dynamics (i.e., advection) as compared to down-slope or vertical settling processes (c.f., Fig. 3C-E and L151-159).
Below I list a number of more specific comments: Line 8: I think this should read "…delivering particulate material to continental slopes (or: ocean margins)…" rather than "ocean shelves", which are not considered in this study using a core retrieved from ~2.5 km depth o We agree and have revised this sentence to read "…delivering particulate material to ocean margins…".
Line 10: "radiocarbon …signatures measured for organic carbon in differing grain-size sediment fractions…" o Agreed -done! Lines 53 and following: Please provide more detail on the grain size separation methods: Did you dry sieve or wet sieve? How did you monitor recovery? Are weighted ages of OC in the grain size classes equivalent to bulk OC ages? If not, what are the potential loss mechanisms, could they be differential between compounds of different age/between different size classes? o We have added an explicit discussion about this outstanding question, which describes our assumptions with respect to mass/organic material losses and their isotopic effects as evidenced by previous studies vis-à-vis (wet) sieving procedures. o We also added Supplementary Fig. 2, which graphically depicts the relationship present between measured and 'mass-balanced' bulk organic signatures, which has a strong positive correlation (r 2 = 0.718).  Supplementary  Fig. 1) and furthermore reworded this sentence to read "Age-depth models were refined…".
Line 90: I suggest to move the word "here" to after "focus" o Agreed -this sentence is now re-worded.
Line 106: here and elsewhere (e.g., lines 122, 138): the increase in age offset, according to Figure 3, occurs prior to the YD in the BA warm period. o These inaccuracies are now resolved to accurately present the apparent decrease in age offset that falls at the coda of Heinrich Stadial 1 and initial Bølling/Allerød, followed by pulsed increase in age offsets at Bølling/Allerød-to-Younger Dryas transition.
Line 138: unclear what the terms R FS-y and R CS-y refer to. o We added the subscript 'C' to associated terms (i.e., R FS-C and R CS-C ) for clarification.  34,39 . Indeed, the existence of local INLs is inextricably linked to nepheloid layer formation within the Gulf of Cadiz and, further north, the shelf break 10 . As such, recent Iberian slope sedimentation is dominated by primary production in surface nepheloid layers ('local' input) and then rainout from intermediate nepheloid layers at depths of ~500-1500 m ('distal' input) 34 . o Other studies of the sediment dynamics along this margin demonstrate that local transmission of sediment from upper slope or shelf locations combined together with local bathymetric (morphological) features are of decisive import for sediment distribution on regional lower-slope sedimentary processes 31  Line 148: here the authors refer to a 230 Th-normalized flux record, which is not presented in the figure in spite of a reference to Figure 3G.  ), which are derived from mud diapirs and Guadalquivir river system, respectively 42 , occur within regional MOW flow cores as compared to coeval Atlantic water masses (e.g., surface waters and North Atlantic Deep Water).
Line 169 and 202: The term "pleniglacial" is uncommon in paleoceanography. It is more often used in terrestrial paleoenvironmental studies of the Eurasian continent, and it refers to the full glacial period prior to the LGM, which is not well resolved in the records presented here. I suggest to use more common terminology and suspect that indeed the authors meant "deglacial" o Indeed, we meant 'deglacial'! The incorrect term has been replaced in each instance.
Line 172: "declining percent total OC" is a bit misleading; "low" or "decreased" might be clearer.
o We revised this sentence so it now reads, "…show decreased percent total OC…". Line 196 and 199: Again, the "strong parallels" between the two records are not evident from the figure; likely the authors refer mainly to the early phase of the HS1, where the age offset shows a pronounced maximum paralleled by a strong minimum in TOC%. In the YD, there is no clear correspondence; while the TOC% record has a local maximum in the BA and a minimum in the early YD which slowly increases toward the early Holocene, the age offsets are slightly higher in the BA or early YD and slowing increase through the Holocene. o As described in an earlier response, we revised the Results section of this manuscript to more accurately describe apparent trends and down-core correlations in an effort to resolve this issue (c.f., L104-116 and L151-159).  19,48 . Transport distances (and duration) are also subject to sediment (re)suspension dynamics, which in turn are subject to changes in (dis)aggregation, particle sphericity, grain cohesiveness, riverine discharge 37 , upcurrent turbidity (gravity) flow 35,36 , bottom-current detachment 34 , local flow velocities, and density gradients at deeper current boundaries 19 . Yet, despite such caveats, coarser particles (i.e., silt) fall from solution (i.e., settle out) before coeval finer material (i.e., clays) because, under similar conditions, coarser sediments are concentrated near the bottom of flows, whereas clays are distributed throughout flows, and have much higher settling velocities as compared to clays 19 .
Line 236: It would be helpful to specifically state which grain size class corresponds to which SST proxy. o We have added text for clarification, which links (i) alkenones with fine silt and (ii) GDGTs with coarse silt, though there is most likely partial between sediment-fraction overlap.
Line 237 and following: If transport causes the lags in the biomarker SST records relative to the foram-derived one, a different origin/source area for the biomarkers during the periods of stronger lateral transport would be implied, which would likely result in a shift in SST estimates (unless it is very local). o Again, it is difficult to ascertain exact hydrodynamic consequences of differing grainsize associations since several factors might influence apparent biomarker-sediment transport histories (c.f., L254-265 and L271-278), including provenance allocation. However, we still decided to connect down-core data with modern observations (c.f., L254-265) in an effort to resolve admittedly gross constraints on sedimentary provenance.
Line 240: do you mean "in spite of rather uniform nominal grain-size distributions"? o We agree with your suggestion, and have revised this sentence as "…despite rather uniform corresponding grain-size distributions…".
Line 255: The YD does not show higher age offsets than the BA period! Rather, the opposite is the case. o True! We have revised this sentence to omit the mention of the Younger Dryas and Bølling/Allerød because it doesn't add further strength to our comparison with the mid-Holocene, which was characterized by smaller radiocarbon offsets and higher organic carbon concentrations.   Figure 3. Axis label "Sediment Fraction 14C Age" should use the unit "ky BP", and axis label "Calendar Age" should be in "ky" (an "age" cannot be "ago") o No problem -we removed the shaded blue infill from this figure. Furthermore, we corrected the mislabelled figure axes as suggested to read 'k.y. BP'.
Figure 3: In panels A, B, and F, the errors associated with the measurements should be displayed. In the caption, the reader should be made aware that axis in panel B is reversed. It is very hard to see the different colours used in Figure H, in particular the lavender line is hard to read. o We have added propagated 1s uncertainty bars to panels A and F. We did not include uncertainty estimates for panel B because of limited replicates or else uncertainties were within symbol bounds. o We also mentioned the axis reversal in panel B to minimize confusion. o We replaced the difficult-to-read lavender colour with dark green and thickened this (and the other) lines to improve readability.

Reviewer #2 (Remarks to the Author):
The manuscript « Transient hydrodynamic effects influence organic carbon signatures in marine sediments » submitted by Magill et al. to Nature Communications is a novel and elegant work based on 126 radiocarbon ages on 21 paired discrete samples of clay, fine silt, coarse silt and sand compared to those obtained on bulk organic carbon and foraminifera from a southwestern Iberian margin core covering the last 25,000 years. It shows radiocarbon age offsets that are convincingly interpreted by the authors as the result of differential lateral transfer dynamics among grain-size classes during the last termination (~23,000-9,000 years ago). This differential lateral transfer would be mediated by changes in the intermediate nepheloid layer due to the dynamic of paleo-current densities. The identification of this differential lateral transfer is very important for understanding climate dynamics as it can contribute to explain the observed leads and lags among proxies preserved inmdifferent grain-size classes. This work deserves publication in Nat. Comm. after the minor revisions I have listed below: In lines 80-87, I have found several inconsistencies between the text, Figures 2 and 3 and Table 1 of the supplementary information. In the Table 1 and Figure 2, the maximum radiocarbon age of OC and the maximum calendar age of foraminifera are 21,409±84 and 21,728 years, respectively while in the text the values are 20,475±775 yr and 23,250 yr BP, respectively. In Figure 3, the oldest sample has also a calendar age of 23,250 yr BP. Similarly, the authors claim a consistent down-core difference between bulk sediment OC and foraminifera 14C ages of 1450±200 years but this difference is not observed at 65-66 cm where it is only of few centuries. Finally, the radiocarbon offset between organic carbon in paired fine or coarse-silts and foraminifera is between ~1000 and 3500 years ( Figure 3) and not between ~1000 and 3000 years as indicated in the text by the authors. o Thanks so much for catching this misstatement about down-core ages (N.B., the incorrect value was derived from an older age model) that have now been corrected (c.f., L86-88)! o We have also corrected the range of fraction-specific offsets to reflect the maximum age difference of ~3500 yr, and highlighted the exceptional (low) difference associated with our sample at 65-66 cm depth (c.f., L89-91). o These inaccuracies are now resolved (i.e., rephrased) to more accurately present the apparent trend in age offset that falls at the coda of Heinrich Stadial 1 and initial Bølling/Allerød (c.f., L104-116, L132-135). We have also added dashed lines connecting points that bridge between the LGM and Heinrich Stadial 1 (c.f., Fig.3 and legend).
Line 128: Replace "…the MOW descended to at least 2600 mbsl during past Heinrich stadials as a consequence of decreased Atlantic-Mediterranean seawater exchange" with "…the MOW … as a consequence of increased…". o We decided to remove the entire clause ("…as a consequences of decreased Atlantic-Mediterranean seawater exchange…") because it imposes a mechanistic interpretation into an otherwise descriptive sentence without due cause. o It is difficult to disentangle the sources or/and the mechanisms responsible for increased terrigenous input without discrete molecular isotopic analyses of biomarkers in corresponding grain-size sediment fractions (N.B., Ausin et al. in preparation will focus on alkenone-specific radiocarbon analyses). Even so, down-core sediment bulk C/N ratios at SHAK06-5K feature values indicate marine-dominated OM (Supplmentary Table 1). Further, studies at U1385 highlight that terrigenous material from rivers do not (directly) reach the Principe de Avis plateau 37,50 , upon which SHAK06-5K lies. Respective studies indicate relative input from siliclastics as compared to carbonate at U1385 is influenced by glacioeustatic effects on centennial and longer timescales 50 , though these effects do not appear to have a marked impact on apparent terrigenous input to southwest Portuguese margin slopes 51 . Rather, regional increases in rainfall lead to (i) increased fluvial discharge of freshwater (which influences MOW formation and buoyancy) 52,53 and discharge of terrigenous clays 51 ) and (ii) increases in chemical weathering on vegetated landscapes (which influences supplies and discharge of terrigenous clays) 51 ), with both mechanisms acting in-phase 37 .
Lines 138 and 192: Replace RFS-y and RCS-y with "RFS-C and RCS-C" o Both instances are now revised accordingly.
Lines 235-241: the age offsets between the different SST proxies are not clear on Figure 3H. Also, the different blues of the SST curves are not clearly distinguishable. The authors should add a curve showing the variation of these offsets through time and chose distinct colors. o It is difficult to show centennial 14 C differences among proxy records on a 25 kyr age axis. Moreover, we used Analyseries in our analyses of lead/lag phase relationships, which computes cross-covariance of down-core records and then interpolates each to create evenly spaced data at intervals commensurate with the lowest time resolution averaged throughout the entire record (N.B., bin intervals of Fig. 3H are equal to about 250 yr). Thus, discrete age offsets are sometimes difficult to observe since we show original data as opposed to interpolated data, which we used for establishing phase relationships. o Regardless, we revised the colour scheme of Fig. 3H to include more contrast (e.g., green replaced lavender) and thicker lines to improve readability. o We also added Supplementary Fig. 3 featuring pertinent SST reconstructions derived from at MD95-2042. The records are all derived from identical down-core sample intervals, thus removing potential biases associated with absolute age differences among stratigraphically aligned (paleo)oceanographic records 1 .

Reviewer #3 (Remarks to the Author):
Evaluation of Transient hydrodynamics effects influence organic carbon signatures in marine sediments by Magill et al.
Although I support the overall conclusion that pre-depositional processes and lateral dynamics can differentially impact proxies with different grain-size associations, I have identified multiple issues that need to be resolved before publication. The data are of high quality and the writing is clear but sometimes the authors are little too confident about their interpretation. The conclusiveness of this paper needs additional work. o We have made a major effort to temper the conclusiveness of our interpretations, and instead have attempted to more clearly describe the cantilevered logic we used to connect data -which is straightforward to discuss -with our mechanistic interpretations, which are much more subjective.

Major comments
One, I have some reservations with their neglect of bioturbation. Bioturbation can also cause size-dependent displacement and thus age offsets as the authors do acknowledge. However, they then simply state that high accumulation rates (order of 15-20 cm/kyr) are enough to ignore bioturbation. This is questionable. A simple Peclet number analysis for a mixed layer of 10 cm (maximum), a sediment accumulation of 20 cm/kyr and a biodiffusion coefficient of 2 cm2/yr, reveals that mixing is more important than sediment accumulation (Pe<1). (Pe=w*L/D= 0.02*10/2= 0.1). A shallower mixing layer would even increase the relative importance of bioturbation mixing vs. accumulation. Moreover, the authors do present excess 210 Pb profiles in their supplementary table. They interpret these data with an accumulation model ignoring bioturbation. This may need revision; 210 Pb vs. depth profiles are governed by radioactive decay, sediment accumulation and bioturbation (ignoring compaction). Decay is known, but accumulation and bioturbation not. You can only separate these two having another, independent tracer, like for instance the 1963 or 1986 Cs peaks. Based on their model the upper 8 cm accumulated in about 160 years, in other words, about 50 cm/kyr. Compaction can account for part but not for all the difference. Most modern sediment biogeochemists would interpret these 210Pb profiles as evidence of bioturbation. Their mixed layer is about 7 cm, not unlike the global average bioturbated layer of about 10 cm (Boudreau paper). The age of their mixed layer is about 150-200 years based on 210 Pb, but order of 1000 years based on 14C (still 4 times higher if you correct for 14C reservoir effect). This being said, bioturbators go for food (organic matter) and OM input are related to grain size sorting and lateral processing. o We addressed this issue indirectly in our revisions with the addition of ichnofabric evidence (c.f., L127) and down-core reworking percentages (c.f., Fig. 3B). Below, we also include results of biodiffusion models, which can also be added as a supplementary figure if need be.
Rebuttal Fig. 1: Heat map of optimal parameter solutions for the biodiffusive model. In this heat map, warmer colors (dark red) indicate the optimal solution pairs for sedimentation rate (S, cm 1000yr -1 ) and the biodiffusive coefficient (D b , cm 2 yr -1 ) given 210 Pb activity at depth zero (A 0 = 44.5 dpm g -1 ). The parameters were optimized by minimizing the sum of the negative log-likelihood between observed and modeled 210 Pb activity (A x ) for each depth (x) using the formula 57 : " = % ℮ "(()*( + ,-⋋/ 0 )(2/ 0 ) 34 Two, the text presented in lines 97-107 does not align with what figure 3 shows and the authors impose an interpretation rather than letting the reader infer her/himself the story. o Aspects of this issue were also raised by other reviewers (c.f., L97-107 rebuttal); therefore, we added text that describes -rather than interprets -the corresponding patterns shared between radiocarbon offsets and complementary proxies of (paleo)oceanographic variability. This revision is accompanied by poignant text revisions in the same paragraph to resolve inconsistencies as compared to Fig. 3.
Line: 100-101: Overall lower radiocarbon offsets… during the LGM (this is based on a single data point!) transition into more moderate offsets of 2000 plus/minus 500 yr during preliminary glacial termination (not supported by a single data point!) and then rapidly peak to values of 1500-3000 yr during HS1 (the clays offsets are the lowest in HS1 rather than highest), and so on. For instance, line 104 after an interim of low radiocarbon offset throughout B/A interstadial, 14.7-12.8 kyr ago), the relative age differences… increase again. I really do see an increase during BA and decrease during YD. This whole section does not properly link to observation in Figure 3. Why is there no zero age in figure 3 and why are two zero offsets in Figure 2 for the clays and only one in Figure 3. Aren't these the same data. o Our original text 'inflated' the interpretive significance of limited LGM data, so we revised/tempered the sentence to read, 'A lower radiocarbon offset…", but we are also open to reviewer suggestions about how we can further improve this sentence without distracting from its descriptive clarity. o The above revision comes in addition to more extensive changes in our manuscript text to square descriptions of data with their associated figure(s). o With respect to data offset between sediment fractions viz. clays and foraminifera, we omitted the zero value of Fig. 3A for aesthetic reasons, which we still favour. However, we accept 14 C differences of <500 years are difficult to see in Fig. 2 because of the visual disparities with respect to age scales. Suggestions are welcome, and for now, we added text to Fig. 2 legend that highlights expanded data axes of Fig.  3 and the contents of Supplementary Table 1.

Minor points of attention
Line 8: delivering particulate material to ocean shelves or from ocean shelves? o Both! We revised this sentence to read, "…interplays among processes delivering particulate material to and from ocean margins…".
Line 90: large flocculates do indeed occur, but whether they survive settling to seafloor and diagenetic alteration in sediments remains to be seen. The reference cited here, as I recall, does not consider sediment aggregates, but water-column ones. Use other literature to support this claim. o This is an important distinction, and therefore we have augmented the references to include additional literature sources specific about benthic aggregates and boundarylayer sediment dynamics.
Line 125: grain-size specific resuspension has also implication for how organisms mix material around. o We agree with this statement, but -in an effort to maintain manuscript focus -have decided to reword the sentence so our assertion about bottom-current influences on size-specific re-suspension is relegated to a subordinate clause as opposed to a main subject.
Line 138: …with RFS-y I presume y should be replaced with C? o Indeed, the 'y' terms have been replaced with 'C' both here and throughout the manuscript to improve readability.
Line 140: … versus autochthonous clays.. Rewrite, this could be misread as new clay formation (the reverse weathering story). o Agreed. We decided to remove corresponding parenthetical information outright to avoid future confusion.
Line 158: where is the evidence against bioturbation. I would say evidence for substantial bioturbation.
Line 172-174: the logic needs attention here. In principle OC should decrease if accumulation rates increase (simply by dilution), but usually OC and mineral delivery to sediments co-vary and OC (wt%) are rather invariant. Burial efficiencies increase of course with accumulation. o This is a very good observation, which we attempted to rectify by removing the contradictory perspectives inherent to our argument. That is, we removed the words 'Yet', 'despite' and 'counterintuitive', which together make our sequence of logic more…well, logical.
Line 185: why specifically oxic degradation? Is pre-depositional degradation not more correct? o You make a convincing point! Therefore, we replaced 'oxic' with "…progressive pre-depositional degradation." We also added a more applicable reference.
Line 191-194: to make this more conclusive, OM quality/composition data would be needed. o We agree that these data would benefit our manuscript, but -alas -we are yet to complete such analyses outside of bulk C/N ratios (c.f., Supplementary Table 1). Therefore, we added a clause about the speculative nature of our conclusions in lieu of biomarker molecular and isotopic measurements among grain-size sediment fractions.
Line 232-234: age offset of grain-size fractions go from zero to almost 3000, many of which 1500-2000, yet the lag (offset) in panel H, is only of the order of 200-300 years. This needs to be reconciled. Part of this may be explained by differences among sites, but how strong is your case then. o The sentences in question discuss relative age offsets apparent between concurrent fine silts versus coarse silts, as opposed to 'absolute' age offset between sediment fractions and foraminifera. This distinction is important because the offset (lag) discussed in this paragraph (c.f., L266-278) is between silt fractions only. The offset between either silt fraction and foraminifera is somewhat larger (500-1000 yr), though still far less than the differences shown in Fig. 3A. We suggest that these discrepancies are consequent to differences in mixing proportions among proxies (i.e., relative abundance of 'fresh' vs. older material in corresponding proxy pools) in combination with uncertainties or natural fluctuations in sedimentary particle associations. That is, we assume specific and invariable molecular sediment-fraction associations (i.e., alkenones with fine silt; GDGTs with coarse silt) that might be over simplified for our system. Indeed, large discrepancies can arise when measured 14 C differences among grain-size fractions are compared to apparent lead/lag phase relationships among proxies signals derived from bulk sediments. o Discrepancies in age offsets as compared to signal lead/lag phase relationships might likewise arise if two or more source areas share similar oceanographic conditions, such as sea-surface temperature, since apparent lead/lag patterns are related to offsets in signal timing (phase) as opposed to age sensu stricto. o We hesitate to include a discussion of this issue at the moment because of its complexity, but likewise welcome reviewer opinions about the matter! Figure 3: why is the bulk TOC content scale presented upside down (low is up). o Bulk TOC % (panel B) is shown upside down to facilitate comparison amongst downcore records (i.e., most readers find it easier to visualize positive, as opposed to inverse, relationships). This confusing point was also noted by other reviewers; as such, we added text to Fig. 3 (legend) to underscore our reasoning. General note: The line numbers given in the rebuttal do not agree with the line numbers in the revised manuscript, which makes it harder to assess whether appropriate changes were made.
Overall, the authors have responded in detail to each of the points raised in my review of the originally submitted manuscript. However, the revised manuscript still stresses the importance of intermediate nepheloid layers (as opposed to bottom nepheloid layers, which could also be created during sea-level rise, or enhanced supply of riverine material) for lateral particle transport. I think this is not substantiated by the data, as the paper does not provide evidence that pre-aged organic matter is indeed entrained in these INLs. The text should be more thoroughly revised to make clear that transport in INLs is one of several possible scenarios.
Replies to my comments on the original manuscript: -The authors argue that the putative "strong positive relationship between downcore values of age offsets with XRF-derived ln(Zr/Al) values is apparent visually and from linear regression; it would be good if these regressions were shown in the SI. The r2 values of 0.242 and 0.292 do not convince this reviewer even of a moderate positive relationship -age model: a new paragraph was added describing how the age model was constructed. I am wondering why the authors used the calibration software Calib in the slightly outdated version 6.0?
-description of the temporal evolution of age offsets (re. Fig. 3f). I am still not satisfied with the way the age offsets are described. For instance, the authors state: "low-to-intermediate average radiocarbon offsets …persist through the mid-Holocene", while no mention is made of the rather large age offsets of up to 2000 yr observed in the late Holocene. These offsets are in fact much higher than the putative "peak" during the "B/A-YD transition". While I agree that local maxima may be apparent, in particular during the early HS1, the data from the early BA on to my eyes only show a gradual increase; there are two contradicting local maxima for the two FB and CS grain size fractions.
-In my previous comment I suggested that lateral transport of pre-aged material might not happen in the intermediate nepheloid layer but instead by other ways. The reply presented by the author does not address this concern but instead lists numerous lines of evidence for the occurrence of lateral transport, which is not doubted. I would like to refer to a study on intermediate and bottom nepheloid layer dynamics, which shows that most of the material in INTERMEDIATE nepheloid layers is actually very fresh (Karakas et al, 2006, JGR). The references cited in the rebuttal refer to BOTTOM nepheloid layers, which have inherently different erosion and re-deposition dynamics. I would like to suggest that the author make a clear distinction between intermediate and bottom nepheloid layers and refrain from directly attributing transport to INLs (e.g., line 168). I also note that in the rebuttal, several months are cited as average residence times in re-suspension events of organic matter aggregates. This estimated timescale for lateral transport is in conflict with the data presented here, where organic matter in the fine fraction is up to 3500 yr older than the depositional age.
-I am not entirely satisfied with the reply to my comment suggesting a potential tie between sedimentation rate and age offset. The authors should present a plot of sedimentation rate along with the other data. In line 195, they refer to figure 3 a,b in context of a correlation between age offsets and sedimentation rates, but on the figure, plots of age offsets (a) and sea-level and bulk TOC% (b) are shown. Likewise, I acknowledge the inclusion of the local sea-level reconstruction. However, the first derivative of this record, i.e., the rate of sea-level change would be more appropriate. From visual inspection of the local sea-level curve it appears that the strongest increase in sea-level happened in the second half of the younger Dryas right after 12 kyr BP, probably in line with the global sealevel rise. When inspecting the references cited for this record, I could not find the data to support the displayed curve (ref 79 provides a plot of a previously published sea-level reconstruction with some relatively strong oscillations, while ref 80 covers only the time period from about 12 kyr to present). Moreover, it seems from the plot displayed in Figure 3b that for the intervals between ~23 kyr BP and 17.5 kyr BP and between 17.5 kyr BP and ~13 kyr BP, no local data exist and the record displayed is a linear interpolation between existing data points. If this is the case (displaying fixed data points might have helped in evaluating this), the global reconstruction might be more suitable for assessing this question.
A one-to-one relationship between sedimentation rate and age offsets would be rather surprising, as both depend on a variety of different factors, as rightly stated in the reply letter. However, the multiple potential influences on the age offsets and the different mechanism of sediment remobilization should be discussed in more detail in the manuscript. As is, the discussion remains centered around the assumed intermediate nepheloid layer transport, which I doubt is the proper mechanism.
- Figure 3, label of x-axis: Still the misleading unit (k.y. ago) is displayed. Please change to ky BP Further notes: Line 108-109: Here, it sounds like the decrease in age offsets is due to an unexpectedly low TOC age, while from figure 2 it looks like the decrease is due to a older than expected foram age.
Line 175: It is not clear to me what is meant here. Do the authors imply that redox conditions impact the age of OM in sediment and SPM? How would this occur?
Line 195 and following: Here a reference is made to Fig 3a,b when discussion down-core sedimentation rate. However, sedimentation rate is not displayed in this figure. Please add a plot of sedimentation rate (see comment above). Further, the sentence starting in 196 seems grammatically incorrect. "Aspiration" seems an odd term in context of ocean circulation. I recommend that they include the rebuttal Fig. 1 (about accumulation vs. bioturbation) in the SI. Why? I do not consider the ichnofabric argument against bioturbation as strong evidence. And this rebuttal Fig. 1 shows precisely why bioturbation cannot be ignored (as also evidenced by the lack of Cs peak (xls file0, but smearing all over, given the resolution; and a Pe of 1 based on their data, implying mixing and sediment accretion are similarly important). Also, a lack of a mixed layer (with 210Pb excess, L147) does not mean there is no bioturbation. A mixed layer is usually only visible for longer-lived radionuclides. Is the Zonneveld paper the best reference for this? L131: is it wise to use the word 'coda' for an international audience dominated by non-native speakers nowadays? L174: .. that, in turn, impact… (this logic requires a reference).