| 27 May 2026
Tidal modulation of nitrate supply and chlorophyll-a in the Amazon shelf–offshore continuum
Abstract. The Amazon shelf–offshore continuum is a dynamic biogeochemical hotspot of the tropical Atlantic, where river
discharge, ocean circulation, and strong tides interact to shape nutrient and phytoplankton distributions. The Amazon plume
and regional circulation have been widely studied and are known to strongly influence nutrient availability and biological
productivity in this region. However, shelf-break tides remain an overlooked pathway linking physical energy to offshore fertilization, and their contribution to seasonal and intraseasonal biogeochemical variability remains unclear. Here, we
quantify how tidal dynamics, including internal tides, modulate nitrate supply and chlorophyll distributions from the
Amazon shelf to offshore waters. We use a high-resolution coupled physical–biogeochemical model (1/36°), evaluated
against climatological, satellite, and in situ observations. The model reproduces the main observed patterns of surface nitrate
and chlorophyll, as well as key vertical features such as the nitracline and the deep chlorophyll maximum. We show that tides strongly enhance upward nitrate transport, increasing surface nitrate by more than 50% over the northern shelf, along
the shelf break, and within the main internal-tide pathway. This tidally supplied nitrate fuels offshore phytoplankton growth,
increasing chlorophyll by about 15–50%, while reducing surface chlorophyll near the Amazon mouth by 30–
40%.Seasonally, surface chlorophyll and nitrate are higher over the Amazon shelf during April–June but lower offshore,
revealing a marked cross-shelf contrast. When the tidal contribution is isolated, a similar but weaker spatial structure
30 emerges, with a cone-shaped chlorophyll anomaly extending from the shelf break toward the offshore internal-tide
propagation region. Remarkably, tides account for about 63% of the total seasonal variability in surface nitrate, meaning that
tidal forcing alone explains more than half of the seasonal nutrient signal. At intraseasonal timescales, tides generate a clear
spring–neap rhythm of about 15 days in both nitrate and chlorophyll. This spring–neap tidal pulse propagates from the shelf
break toward offshore waters and is especially pronounced near the deep chlorophyll maximum, where oscillations of the
upper nitracline periodically modulate nitrate availability and drive a corresponding chlorophyll response., where
chlorophyll variability is nearly doubled when tides are included. The concurrent increase in nitrate variability indicates that
this spring–neap phytoplankton response is sustained by tidally driven nutrient supply.
These findings identify internal tides as a key biogeochemical driver of the Amazon shelf–offshore continuum, linking tidal
energy to nutrient injection, offshore fertilization, seasonal redistribution, and rhythmic ecosystem variability in the western
tropical Atlantic.
The paper considers tidal effects on the distribution and growth of phytoplankton in response to transports, mixing, nitrate supply over the Brazil shelf near the Amazon outflow. It presents a very interesting picture of how internal tides affect biogeochemistry in the seasonally-variable presence of the Amazon plume. The work uses a NEMO-PISCEv4 framework with 3 km horizontal resolution, which is sufficient to resolve low-mode internal tidal waves.
The Introduction ends with a very clear description of the 3 key questions to be addressed.
The model is validated against climatological data for nitrate and chl (see general issues below). Most of the work focuses on the differences between the model with and without tides. Some of the strongest results are on the cross-shelf/shelf edge sections where internal tidal influences are very clear.
The analysis focuses on mean, seasonal and spring-neap changes. All of these are interesting, and there are both common (internal tides and mixing) and different (plume changes) physical drivers. The paper feels very descriptive. There are lots of interesting contrasts between tides/notides but the explanations of why these contrasts occur are largely asserted or linked to previous work. But this is an impressive modelling experiment that must also include model outputs that would allow thorough, objective explanations of the results.
I have made a lot of comments below, but it all summarises to two main suggestions: (1) provide more physical data (e.g. density) to help understand the nitrate and chlorophyll patterns, and (2) make better use of the model output to identify causal links (e.g. what does a map of nitrate flux to the DCM or diapycnal diffusivity look like – the tides/notides anomaly of this could be very informative and strengthen much of the explanations). Without (2) the paper feels very descriptive. If a more detailed analysis of model fields is used to identify causal links, than there may be more than one paper in all this (e.g. mean and seasonal fields, and spring-neap tidal behaviour). Or, keep it as one paper but simplify the results section and use the discussion section to demonstrate the causal relationships. Either way – this is good, novel material. High-resolution models that get internal tide mixing are rare and with a better focus on modelled processes and causal relationships there is potentially a very strong paper here.
General points to address:
The comparison of the model nitrate field with CARS is not convincing. Figure 3 should also show the CARS nitrate climatology, otherwise it is difficult to accept that the model does a reasonable job at simulating the broad distributions. The comparison (Notides-CARS) shows a coherent pattern of discrepancies that too me suggest the model results on the shelf are quite poor (e.g. high nitrate in coastal waters north of the Amazon, low nitrate over most of the rest of the shelf). I feel I still need to be convinced that the model is doing OK. Getting nitrate to the right order of magnitude does not seem like a difficult bar to get over. Later (line 297) it is stated that the model shows a moderate but statistically significant correlation with CARS – this seems to be a much more reasonable claim. However, the mean errors are still high and I feel are telling us something more interesting that the simple argument that CARS is relatively low resolution.
The comparison of the vertical profiles is overly optimistic, and confusing. Are the figure legends correct? The description of modelled and observed chl in the main text implies the observations are the green line and model the dashed line. Using the labels in the figures, the observations show a very clear subsurface chl maximum (45m depth), while the model has a very weak max at 25m and higher concentrations of surface chl (I would not agree that the model gets the subsurface max at the same depth as the observations). So, if the legends are correctly identifying which line is which, then I suspect too much nitrate is getting into surface waters in the model (e.g. at 40m there looks to be about x3 more nitrate in the model compared to the observations – again assuming the figure legend is correct). Might some of this discrepancy be caused by the wide range on environments covered by the observations, perhaps undermining a simple mean-profile approach? The observation locations cross some of the major gradients in model-obs discrepancies (Fig. 3). Or were the observed chl profiles based on fluorescence, so there might be some quenching in the surface? You could test the latter point by only selecting nighttime observations. Or, if the legends are incorrect (implied by the descriptions in the main text), then the observations have higher nitrate and chl in the surface – which would make me want to look at how the model vertical mixing works in stratification.
Was any tuning of PISCES attempted to improve the fit? I accept that the vertical nitrate gradients are OK, which is vital for subsequent calculations of nitrate fluxes, but I would want to be more convinced that the phytoplankton are also in the right places.
Figure 4: It would also be useful to see comparison of the vertical density structure as this is an important part of where nitrate and chl are found.
Figure 5: as with Fig. 3 it would be useful to also see the CCI observed chl field as the discrepancy plot possibly masks any consistency in model patterns. The precision of the bias and error in chl is too high – 1 decimal place is all that is justifiable.
Having said all that, the importance of the model results later mainly sits on the model-model comparisons with/without tides. So, getting the broad, approximate spatial patterns in nitrate and chl reasonable compared to the observations is fine –
Smaller points: