Summary

The study provides novel data on how the biological gravity pump transports carbon from the surface to the deep ocean in the equatorial Atlantic upwelling region, using a biogeochemical (BGC) Argo float equipped with an Underwater Vision Profiler 6 (UVP6). Satellite data are used for sea surface chlorophyll-a, sea surface temperature, and Lagrangian diagnostics. Particle abundances are used to calculate carbon flux. Export events are identified based on carbon flux anomalies and a regime shift detection algorithm. Morphological properties of detritus are categorized and assigned using Ecotaxa and a principal component analysis. Flux attenuation and biological carbon pump efficiency are calculated using the Martin function.

The authors identify two distinct periods with cold surface waters and chlorophyll-a peaks and suggest that these periods of increased productivity are linked to Tropical Instability Waves. A correlation between surface chlorophyll-a and abundance of both smaller and larger particles is described, and two export events are identified. Marine snow is grouped into five different morphotypes, based on size and packaging. Both denser and more loosely packed smaller particles are most abundant at the beginning of the export events, suggesting that they are the precursors of larger particles. Low export efficiencies are attributed to the continuous presence of grazers in the equatorial region, as opposed to higher latitudes. An increase of denser particles was observed at 300-600m water depth and attributed to zooplankton feeding.

General comment

The study provides a comprehensive data set, observing and describing two export events in the equatorial Atlantic and key factors that may drive the observed carbon flux. They present details on particle properties such as their size and denseness. I think that is a valuable case study of how recent developments in underwater imaging technology can be used to gain detailed insights into carbon flux mechanisms. The data analysis and method description are clear and thorough.

Here are some specific comments to the different sections:

L44 – 56: Mineral ballasting could be mentioned as an additional factor

L315: “best transfer efficiency” – rephrase to “highest transfer efficiency” to avoid a misleading qualitative assessment

Section 3.4.1. : I found this section difficult to read as five acronyms were introduced at once. In many points the authors make, there is a pattern by either size or by packaging. I would suggest spelling this out by describing the categories as either small or large and either densely packed, loosely packed, or fiber. Also, I think that the word “porous” is misleading, I would suggest “dense” and “loose” or “dark” and “light”, based on the image property.

Example: “They shared similar temporal dynamics primarily in the surface layer: all types decreased exponentially between 0 and 150 m. While fibers and loose/light particles decreased slowly throughout the water column in the mesopelagic layers, dense/dark particles increased gradually between 400-600 m…”

The morphotypes identified here are different than the clusters identified by Trudnowska et al (which are cited here). Discussing why different categories were identified in the context of this ecosystem would add value to the paper.

L434-435: In my experience, denser particles can also be more intact, larger pieces of organic detritus, or aggregated phytoplankton after a longer period of time. I think that it can’t automatically be concluded that the main source of dense particles are fecal pellets.

L646 ff: This is an interesting observation!

L482: How much of the organic matter reached those depths?

Figure 6: Would it be possible to reflect the patterns you see in particle size in this figure?

Code availability

I think it would be helpful for the scientific community and enhance the transparency of the data set to share the code in the supplement or on a platform such as GitHub rather than making the reader rely on direct communications with the authors.

The authors investigate two bloom events and associated particle/carbon export dynamics in the equatorial Atlantic using BGC Argo data and remote sensing products. Identification of bloom characteristics, potential drivers and mechnisms determining export of biogenic particles during and outside these two events are presented in a largely convincing manner. The manuscript presents an interesting case study that nicely highlights the value of UVP data for more in-depth assessments of the composition and fate of organic matter associated with phytoplankton blooms. The manuscript is largely well written but may benefit from minor language editing. I have a few general and minor comments. Once these have been addressed I would recommend the manuscript for publication.



General remarks

The criterion used to constrain the particle plumes, i.e. sloped lines corresponding to a particle sinking speed of 30 m day-1, does not seem to adequately capture the second plume (Fig. 4d). However, the actual sinking speed of particles in this plume seem to be higher than the literature value of 30 m day-1 (which could be explained by the higher concentration of large particles). Wouldn't it make sense to adjust the criterion to get better fitting constraints for the plume and subsequent analysis?

The carbon flux units mentioned throughout the manuscript and figures need to be revised. Units are given in mgC m^-3 d^-1 in the text (e.g. lines 279, 281) and in mgC m^-3 in the figure labels (Figs. 3 and 4). However, fluxes should be given in mass per area per time, i.e. mgC m^-2 d^-1.

In the discussion on diel vertical migration as an explanation for the increase in small dense particles at depth, I suggest mentioning the local time of day at which the float profiled (which would also be helpful information to include in the methods). According to the Argo fleet monitoring website, profiles were all taken near local midnight, which would support the hypothesis that the increased abundance in particles at depth are a diel migratory signal.

Specific comments

Line 100: Why were different spatial resolutions used for the two satellite products? VIIRS should be available in 4km resolution as well, right?

Line 137: What wavelength was BBP measured at? 470 or 700 nm?

Line 152: Is 60m the maximum mixed layer depth that was reached?

Line 252-254: Another factor that would contribute to the difference in variability: Float and satellite chl are derived in fundamentally different ways. Float chl is derived from fluorescence which is affected by physiology (primarily by light and nutrient availability). Satellite chl is derived from remote sensing reflectance, which is largely unaffected by physiology. Float-based chl estimates are a lot more variable for this reason. See Long et al. (2024), https://www.nature.com/articles/s43247-024-01762-4

Fig. 3: Are the data shown here log10 or natural log transformed?

Line 340: I assume this is a typo: "SDP were most abundant ..." (instead of SPP).

Line 417: Correct me if I'm wrong, but this seems to contradict what's stated in the results. Here: "Both MiP and MaP in the top 100 n are correlated to in situ chl-a biomass". In line 269: "No significant correlation was found between surface chl-a and MaP abundance."

Fig. S2: Consider adding SD shades behind the climatology.

Fig. S7: Colours for FD and BPP are not colourblind-friendly.

Fig. S9: The labels are somewhat cryptic without an explanation of what the variables mean.