Internal-tide nutrient supply to the euphotic layer: vertical advection versus turbulent diffusion off the Amazon shelf from glider observations

M'hamdi, Amine; Koch-Larrouy, Ariane; Bosse, Anthony; Dadou, Isabelle; Vantrepotte, Vincent; Costa da Silva, Alex

doi:10.5194/egusphere-2026-4862

Preprints

https://doi.org/10.5194/egusphere-2026-4862

Preprints

19 Aug 2026

| 19 Aug 2026

Status: this preprint is open for discussion and under review for Ocean Science (OS).

Internal-tide nutrient supply to the euphotic layer: vertical advection versus turbulent diffusion off the Amazon shelf from glider observations

Amine M'hamdi, Ariane Koch-Larrouy, Anthony Bosse, Isabelle Dadou, Vincent Vantrepotte, and Alex Costa da Silva

Abstract. Tropical oligotrophic surface waters are typically nutrient-limited, and internal tides have been proposed as a key mechanism supplying nutrients to the euphotic zone through two complementary physical pathways: vertical advection and turbulent mixing. However, separating and quantifying these two pathways from in situ observations, and assessing their respective ecosystem implications, remains challenging. Unlike previous studies, which have generally investigated these mechanisms separately, this study directly compares their respective contributions within a common observational framework. Here, autonomous glider observations collected during the AMAZOMIX campaign off the Amazon shelf were used to reconstruct nutrient profiles using CANYON-B neural network, identify internal-tide-driven isopycnal and nutrient-associated vertical displacements, and independently estimate turbulent dissipation rates validated against microstructure profiler (VMP) measurements. Vertical nutrient profiles revealed contrasting nutrient limitation regimes, with a nitrate deficit in the upper layer and a silicate deficit at depth. Our results demonstrate that internal-tide-induced vertical advection dominates the mean nutrient enrichment at the base of the euphotic layer (F_ADV = 2.3 ± 0.5 mmol NO₃ m⁻² d⁻¹ and 0.4 ± 0.1 mmol Si m⁻² d⁻¹), whereas turbulent diffusion provides smaller mean fluxes (F_DIFF = 1.07 mmol NO₃ m⁻² d⁻¹ and 0.14 mmol Si m⁻² d⁻¹) but remains essential because it represents an irreversible transport pathway and can dominate during episodic high-energy mixing events. Together, these processes modify not only nutrient availability but also nutrient stoichiometry, highlighting internal tides as an important regulator of ecosystem functioning in the western tropical Atlantic. These results suggest that internal tides act both as a nutrient-supply mechanism and as a potential community-structuring process, with implications for nitrate–silicate-dependent phytoplankton groups at the base of the euphotic layer.

Received: 12 Aug 2026 – Discussion started: 19 Aug 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Amine M'hamdi, Ariane Koch-Larrouy, Anthony Bosse, Isabelle Dadou, Vincent Vantrepotte, and Alex Costa da Silva

Status: open (until 14 Oct 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-4862', Anonymous Referee #1, 18 Sep 2026 reply
General:
The paper investigates the relative contributions of vertical advection and diapycnal mixing, both caused by internal tidal waves, on the supply of deep nutrients to the euphotic zone. A key point is made, often missing from upwelling/advection arguments about nutrient supplies, that the irreversible nature of diapycnal mixing makes this source very important even though it appears to have a lower magnitude. The paper is reasonably well written, but there are gaps in some of the methods descriptions that need clarification.
The case for using CANYON-B is not strong. For instance, does the CANYON-B approach yield significantly better nutrient results compared to, say, a simple regression between nitrate and density? Is the neural net trained on the nutrient profiles first, and then let loose on the glider data? Fig 2: the crosses for the bottle data should be larger. It is unclear what the glider (St. 1) profile is based on – have glider nutrient profiles been calculated using CANYON-B? Overall, the figure does not persuade me that CANYON-B provides anything better than the original profiles – CANYON-B seems to give profiles with the same vertical resolution. Are there deeper nutrient measurements (the dotted lines for the in situ stations show increasing nutrients in deep water, but there are no bottle data shown to justify)? It would be good to see some clearer justification for using the CANYON-B approach rather than something simpler, and a clearer explanation of what the glider profiles are showing.

There are some confusing parts to the Methods section – section 2.2.2 mainly. I think I understood what had been done by the end, but it would help to have some clear statements at the start of the section on what is required so that the reader knows where the section is heading.

I am not sure I agree that the LEM reproduces the vertical epsilon structure. Below about 125m the comparison is good, but from 125 to about 90m the LEM tends to show opposite patterns to the VMP data. Some further work/insight would be helpful. Can you identify what parts of the LEM methods might have caused the departures? Can you make a reasonable argument that the LEM is still OK to get nutrient fluxes into the euphotic zone (accepting there might be poorer LEM-VMP comparisons above that, but at least this is a part of the water column where nutrient gradients are very weak and so fluxes are unimportant)?

The advection and diapycnal mixing mechanisms are largely described separately. But can they be tied together in the Discussion? Is there a link between increased advective supply and increased diapycnal mixing (caused by larger internal tidal waves)? In other words, do large-amplitude internal waves both raise nutrients and generate greater mixing?

Specific:
Line 82: is there a reference missing?

Line 82: some brief description of what CANYON-B is would be helpful here. Why is it necessary? What does it allow you to do that other methods fail at?

1: A scale bar for the purple colours, and a scale for the red vectors, are needed.

Lines 114-123: what was the typical dive profile time for the glider? i.e. what was thew temporal resolution of the glider profiles?

Line 136: should Zeu be more clearly a variable (e.g. z_eu)?

Line 167: “eta” – should that be Greek letter h?

Line 186-187. I cannot follow the normalisation argument and procedure here. I am not sure why it is necessary, and I do not understand how it was applied.

Line 196: equation 1 is invoked but not presented until much later. The methods here seem to be a little confused (or confusing).

Line 196-201: what is the difference between “directly estimating vertical velocities” and “velocities …inferred from” isopycnal displacement?

Lines 233-244: some of this gets confusing – possibly because some of the text is out of the correct order (e.g. “with w the vertical displacement….”)

Line 244: Figure 5 should be Figure 4?

4: what happens to the crest modelled nutrient profiles at depth (there is a sudden vertical switch in the profiles)? What do the blue, gey and red shading show?

5: the caption, and the main text, do not explain fully what the figure shows/means.

Line 258: why mention Jensen’s inequality? Is it useful?

Equation 3: why +/- 10m for the integral? It seems arbitrary. Why not integrate from Zeu to the surface to quantify the impact of nutrients available within the euphotic zone?

Line 276: Turbulent Kinetic Energy Dissipation?

Line 282: C-epsilon – the “epsilon” needs to be subscripted or it might be misunderstood as a variable in the equation (and inconveniently cancel the epsilon on the left of the equation).

7: I assume d(nitrate)/dz is based on the fit of glider data to the CANYON-B results? Why are there finite nitrate fluxes above about 70m event though the nitrate gradient is zero? Or is this just because fluxes are on a log scale and there are tiny gradients in the reconstructed nitrate profile?

Lines 335-338: The assumption of 50% of the diapycnal flux being attributable to internal tides is quite sweeping. Can you provide a better justification of why the Kouogang et al results are relevant? At least, note that the Kouogang results were for the same region?

Figs 8 and 9. My rough estimates of the indices based on Fig. 8 are as follows. Phosphate varies between about 0.1 and 0.6 mmol m-3 (Fig. 8), which therefore should lead to the phosphate availability index varying between 0.1 and 0.6 (using index = [Phosphate]/1 from Methods). Silicate index should be roughly 0.06 (surface) to 0.04 (silicate minimum) to 0.2 (silicate at 200m). Nitrate seems to me about 0.2 mmol m-3 at the surface and around 8 mmol m-2 at 200m. So, index looks to be 0.01 or so in the surface and 0.5 at 200m. But then there is the “normalisation” part of the method which I do not understand. The normalisation leads to the suggestion that phosphate is nowhere limiting, nitrate might be a little limiting, but remarkably depth-uniform, and silicate is most limiting at depth (where silicate is in the highest concentrations). I find it very difficult to make sense of these results!

Section 3.2: only nitrate and silicate are considered. This makes sense if phosphate is never limiting. But this should be explained.

Line 397-400: positive nutrient anomalies occur associated with spring tides? But Fig. 10a, b indicate the greatest negative anomalies at spring tides, and positive anomalies appear in the transitions between neaps and springs. There is no clear indication in Fig. 10c that anomalies are consistently greater at spring tides – this could be made clearer by reporting the mean values over neaps and spring, which then might help interpreting Fig. 11.

11: interquartile range across all triplets? I do not know what has been done here. I find it difficult to reconcile what I see in Fig. 11 with the patterns in Fig. 10a, b. For instance, spring tide nitrate anomalies below 150m seem to be biased strongly negative in Fig. 10a, but Fig. 11a suggests an even +/- distribution.

12: it would be helpful to make the euphotic zone boundary.

Line 449: Figure 13?

Figs 10, 11: The anomalies (delta-A) in Fig 10c are calculated from the wave-affected nutrient profiles (P) – to me showing Fig 10c after Fig 11 feels more logical.

Table 1: What is meant by “above Zeu”? Is that between Zeu and the surface? Also, “integrated flux range” – is that through the base of Zeu? “Mean flux” – where in the water column? Across Zeu?

Fig. 13: negative fluxes can only occur if the nutrient gradient switches direction. For Si there is some evidence of this in the upper water column. I find it difficult to believe positive nitrate gradients. The existence of negative fluxes below about 100m for both nutrients probably reflects noise in the nutrient calculations from glider data rather than anything real?

Reply
Citation: https://doi.org/10.5194/egusphere-2026-4862-RC1
- AC1:
  'Reply on RC1', Amine M'hamdi, 25 Sep 2026 reply
  
  We are very grateful to the Reviewer for hisconstructive and relevant comments, which have helped us improve the quality and clarity of this work.
  Please find attached
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-4862-AC1
  - AC2: 'Reply on AC1', Amine M'hamdi, 25 Sep 2026 reply
    
    Dear Editorial Team, Dear Reviewer
    I would like to remove the reply I posted under the RC1 comment on my manuscript, as it was submitted by mistake (it was for another publication).
    Could you please delete this reply from the discussion?
    Thank you very much for your help.
    Best regards,
    
    Amine M’Hamdi
    
    Reply
    
    Citation: https://doi.org/10.5194/egusphere-2026-4862-AC2
RC2: 'Comment on egusphere-2026-4862', Anonymous Referee #2, 26 Sep 2026 reply

The manuscript is generally well written (although some obvious typos/omissions running through the manuscript which I’ve mostly highlighted below). The authors give a good overview and understanding of the topic, highlighting the importance of viewing advection and turbulent diffusion fluxes separately and discussing the implications to nitrate and silicate supply to the euphoric zone.
My main concerns are as follows:
I can understand why the nutrients have been derived to get higher resolution and estimates of stoichiometric variability, but I’d like to see more evidence to why Canyon-B has been used instead of for instance ESPER-NN. Further to this, the variability needs to be shown further in the water column in Figure 2 (to the next points which the lines go to at the very least). It looks like the offset gets bigger for silicate in the nutricline >200m, this would need to be discussed in the context of Si:N variability through the manuscript.
From my understanding the glider profiles went deeper than 200m and I think the paper would be improved with an increase in the depth shown in figures to at least 250m, perhaps 300m. This would be particularly relevant with Figure 2. Also in figure 8 it would be good to see how silicate and nitrate change as the concentration increases in the nutricline.
The euphotic depth being fixed - can you elaborate further on why this was considered representative? Why was PAR only measured at one of the three stations?
Minor suggestions:
Title - internal tide driven nutrient supply?
Formatting - In general a lot of instances where there are spaces required or where there are double spaces, have an extra check through on revision.
21 - using the Canyon B
Line 24 - add space after full stop
83 - need to add a reference here!
Figure 1 - the depth contour labels that are there are difficult to read
104 - Data and methods
236 - capital W for start of sentence
Figure 8 - its hard to see the variability in nutriclines, could you add to the bottom of the figure the variability between each, e.g Nitracline minus phosphacline and nitracline minus silicaline - would help with your paragraph at line 365
Figure 9 remove s’s from the legend
Figure 13 is split over two pages
Figures 10 and 13 should be more consistent with each other - same red-white-blue colorbar and both a+b laid equally
531 remove present
543 For nitrate the
550 For silicate the
557-559 - how does the vertical nutrient gradient differ? How much is gradient-driven and how much internal wave driven?
596 need a comma after indices
605-610 this section is really interesting, I’d like to see more papers referenced here, even if not focussed on internal tides, but on the pulses of higher Si vs more consistent N fluxes

Reply

Citation: https://doi.org/10.5194/egusphere-2026-4862-RC2

Amine M'hamdi, Ariane Koch-Larrouy, Anthony Bosse, Isabelle Dadou, Vincent Vantrepotte, and Alex Costa da Silva

Viewed

Total article views: 233 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
107	73	53	233	52	43

HTML: 107
PDF: 73
XML: 53
Total: 233
BibTeX: 52
EndNote: 43

Views and downloads (calculated since 19 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	16	8	2	26
Sep 2026	91	65	51	207

Cumulative views and downloads (calculated since 19 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	16	8	2	26
Sep 2026	91	65	51	207

Viewed (geographical distribution)

Total article views: 228 (including HTML, PDF, and XML) Thereof 228 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 30 Sep 2026

Short summary

Internal tides are underwater waves that can transport nutrients from deep waters toward the sunlit ocean. Using glider observations off the Amazon shelf, we compared vertical advection and turbulent mixing. Vertical advection supplied more nitrate and silicate on average, while turbulence became important during energetic events. Together, these processes change nutrient supply and balance, potentially influencing phytoplankton communities in the tropical Atlantic.


Total:	0
HTML:	0
PDF:	0
XML:	0