Thank you for your very interesting remarks, which will help improve this article.

First, regarding the use of the terms “shallow” and “deep”, they can indeed be misleading when comparing with typical lake depth classification. Instead of simply distinguishing between coastal and pelagic regions, the decision was made to classify zones based on depth rather than distance from the shore. This approach was adopted because certain nearshore areas in Lake Tanganyika are characterized by considerable depths and exhibit seasonal patterns and trends similar to those of offshore deep regions. Therefore, using the terms “coastal” and “pelagic” based solely on proximity to shore could be misleading. To reduce ambiguity, referring to the zones as “shallower” and “deeper” may offer a more accurate and consistent description. This change will be implemented in the whole manuscript, as in line 220 for example.

The use of “high” and “low” for describing the values of Chl-a can also be misleading in the context of Lake Tanganyika, which is oligotrophic. To better reflect the relative nature of chlorophyll-a (Chl-a) concentrations within the lake, the terms “high” and “low” will therefore be changed to “higher” and “lower” thourghout the text. The use of the terms “oligotrophic”, “oligotrophication” and “eutrophic”, as in lines 417-419 and 428-430, will be revised and changed to “lower levels”, “lowering levels” and “higher levels” of Chl-a.

Regarding the mention of air temperature and wind seasonality as drivers of Lake Tanganyika’s hydrodynamic cycle, they are first mentioned in the abstract in a paragraph summarizing established knowledge on Chl-a variability. The influence of these variables on the lake’s hydrodynamics is well documented in previous studies (Naithani et al., 2011; Naithani & Deleersnijder, 2004; Plisnier et al., 2023; Sterckx et al., 2023). In the discussion, the wind and air temperature data referenced are drawn from the literature, and the lack of direct data is explicitly acknowledged. While both variables are available from the ERA5-Land dataset, and while it would be highly relevant to study the influence of these variables on Chl-a using such data, this lies beyond the scope of the present study (this is a subject of another study). Consequently, current statements linking Chl-a variability to wind and air temperature are intended as interpretations without causal confirmation. The conclusion and perspectives of the revised paper will mention that the use of reanalysis products to investigate the influence of climatic variables on ecosystem dynamics - and more specifically phytoplankton concentrations, represents a valuable direction for future research.

A change will also be made in lines 423–425. The sentence will be revised to indicate that peak Chl-a concentrations occurs during periods of strong winds, rather than specifically following wind-driven mixing events, as the precise timing of such events cannot be determined without the use of appropriate wind datasets.

Lake Surface Water Temperature, indeed available in the Lakes_cci dataset alongside Chl-a data, represents an additional variable that could be exploited. However, it was not used in this study for two main reasons. First, lake temperature exhibits much lower spatiotemporal variability than Chl-a and its seasonal pattern closely follows the well documented wind seasonality. Secondly, the decision was made to focus solely on Chl-a in order to place emphasis on this variable. Another variablealso available from the Lakes_cci dataset is lake water level, which is identified in the introduction as a major threat for local populations. However, the relationship between lake level and Chl-a concentrations at the lake scale remains unclear and was not investigated within our study. It does not aim to study in detail the drivers of these dynamics, which is the subject of an additional forthcoming study. Further research would be needed to elucidate potential mechanisms underlying this interaction.

As mentioned in this review, transparency is indeed fundamental to phytoplankton blooms and to the remote sensing estimation of phytoplankton concentrations. Available Secchi disk depth will be included to support comment on water clarity in Line 104. Given that remote sensing data pertain to surface waters, the text consistently refers to surface Chl-a when presenting the results. However, a clarification will be added in the opening paragraphs of the Results section to emphasize that the findings reflect surface conditions and that the influence of phytoplankton concentrations on the Chl-a signal decreases with depth.

Regarding data availability, it does indeed influence the results, particularly regarding the statistical significance of observed trends. We will therefore include a comment in the discussion acknowledging the limitations of the dataset. Alongside this, the uncertainty inherent in Chl-a estimates—though not shown in the main figures to maintain clarity—is a critical factor in interpreting both absolute values and trend analyses. The Lakes_cci dataset provides uncertainty layers for Chl-a, with lines 142–143 indicating that 90% of the data have a fractional uncertainty between 39 and 60%. To enhance transparency and allow for a more nuanced assessment of the results’ robustness, we will explicitly address the impact of uncertainty on interpretation in the discussion and include supplementary maps depicting the monthly mean fractional uncertainty across Lake Tanganyika.

Other comments:

Lines 34-46: Recent citations will be added to the first paragraphs of the introduction to strengthen it.

Regarding lines 115 to 135 and the suggestion to move them to the introduction or to support the discussion, we respectfully disagree. In the introduction, we provide a general context about Lake Tanganyika so that readers understand the significance of this ecosystem as well as the relevance of using remote sensing data to estimate phytoplankton concentrations. Section 2.1, "Study site," then presents a more detailed description of the lake’s physical and biological characteristics, explaining how these have changed over recent decades and their impact on the ecosystem.

It would be possible to merge these parts into a single introduction section. However, it does not seem coherent to separate lines 115–135 from the Study site section, as this would lead readers to first encounter detailed information on the lake’s evolving physical and biological features in the introduction, only to return to more general context in the Study site section. Such an arrangement would disrupt the logical flow and clarity of the presentation. We suggest to leave this section unchanged.

Line 144: This sentence will be changed to state that MODIS data (from 2012-2016) was not released for most lakes in the dataset, including Lake Tanganyika.

Line 155: The DINEOF methodology requires a minimum of available pixels of 5% for interpolation. Images with low data availability typically show patches of available data, not scattered pixels all around the lake.   

Naithani, J., & Deleersnijder, E. (2004). Are there internal Kelvin waves in Lake Tanganyika? Geophysical Research Letters, 31(6). https://doi.org/10.1029/2003gl019156

Naithani, J., Plisnier, P. D., & Deleersnijder, E. (2011). Possible effects of global climate change on the ecosystem of Lake Tanganyika. Hydrobiologia, 671(1), 147–163. https://doi.org/10.1007/s10750-011-0713-5

Plisnier, P., Cocquyt, C., Cornet, Y., Poncelet, N., Nshombo, M., Ntakimazi, G., Nahimana, D., Makasa, L., & MacIntyre, S. (2023). Phytoplankton blooms and fish kills in Lake Tanganyika related to upwelling and the limnological cycle. Journal of Great Lakes Research, 49(6). https://doi.org/10.1016/j.jglr.2023.102247

Sterckx, K., Delandmeter, P., Lambrechts, J., Deleersnijder, E., Verburg, P., & Thiery, W. (2023). The impact of seasonal variability and climate change on lake Tanganyika’s hydrodynamics. Environmental Fluid Mechanics, 23(1), 103–123. https://doi.org/10.1007/s10652-022-09908-8

Thank you for your helfpful feedback, which will contribute to improving this paper.

First, we acknowledge the relevance of the comments regarding the link between the hydrodynamics of the lake and surface Chl-a concentrations. Although a link was made in the manuscript between the hydrodynamic cycle described in Plisnier et al. (2023) and seasonal variation in surface Chl-a (lines 367–384), the discussion will indeed benefit from referencing recent studies that specifically unravel the lake’s complex hydrodynamics and their evolution under climate change, such as Delandmeter et al. (2018) and Sterckx et al. (2023). Modifications in the manuscript will be made accordingly.

As the reviewer further points out, phytoplankton blooms can result from a range of distinct processes including coastal phenomena, such as nutrient enrichment from localized upwellings, potentially triggered by internal waves or oscillations like Kelvin waves, as suggested by Antenucci (2005). Other mechanisms are the resuspension of periphyton growing in the littoral zone and being detached by wave-induced turbulence and drifting offshore and local nutrient inputs from rivers and shoreline settlements. This has been pointed out in our results in the area near the Malagarasi River mouth for example.

Satellite observations of Chl-a plumes drifting offshore over multiple days support the idea that some phytoplankton accumulations originate near the coasts and are advected by nearshore circulation. Disentangling these various processes using satellite-derived surface Chl-a alone, however, remains challenging. One influence is particularly evident though, and notably localized: that of rivers, as illustrated by the Chl-a hotspot observed near the mouth of the Malagarasi River, for example.

The manuscript will be revised to better convey the idea that pelagic and coastal regions are not isolated compartments but rather form a dynamic continuum shaped by physical processes such as currents, mixing, and wave action and that mechanisms can transport biomass and nutrients across zones.

Regarding the findings of Descy et al. (2006), we agree that their field data did not systematically show higher Chl-a concentrations in coastal compared to offshore waters. Our results are consistent with this as we do not observe a systematic difference either for all coastal regions. However, we were able to distinguish between different regions with different Chl-a seasonality and trend patterns. Some coastal and shallower regions exhibit persistently higher Chl-a concentrations and the strongest 20-year increasing trends. In contrast, offshore and deeper coastal regions show more pronounced seasonal patterns in Chl-a concentrations and generally negative 20-years trends. These spatial differences do not contradict Descy et al. (2006), but rather complement their site-based observations.

Regarding the validation satellite-derived Chl-a concentrations, we agree that it is a key point. However, it is important to note that the Chl-a product used in this study, ESA’s Lakes_cci dataset, includes extensive algorithm development and validation steps across a large range of lakes and trophic states from around the world. Moreover, the spatial and temporal patterns we observe in Lake Tanganyika are consistent with previous studies based on both in situ and other satellite products which supports the reliability of the Chl-a estimates. While we acknowledge that direct in situ validation for this lake would strengthen the argument, such data are not available for the full time period covered in this study. We therefore rely on the consistency with prior studies and on the quality of the Lakes_cci product itself.

Other comments:

L10, 42, 105: We agree that chlorophyll-a concentration is a proxy for phytoplankton biomass, but not directly for primary productivity or ecosystem health as a whole. We will revise the manuscript accordingly to remove references that mistakenly associate Chl-a with primary production or overall ecosystem health. Instead, we will consistently refer to Chl-a as an indicator of phytoplankton biomass and, where appropriate, trophic status.

L50: While the references cited here provide general groundworks for the use of optical remote sensing to estimate phytoplankton biomass, the study by Horion et al. (2010) will be added as citation as well as it specifically applied those techniques on Lake Tanganyika by developing an algorithm tailored to its optical properties.

L28: We acknowledge the reviewer’s point regarding the potentially misleading use of the terms “shallow” or “deep” in this study. In the revised manuscript, we propose to distinguish between “shallower coastal” and “deeper pelagic and coastal’ regions rather than relying solely on distance-to-shore distinctions. This terminology helps delineate regions that showed similar patterns in both the seasonality and trends in surface Chl-a.

L64, 96: We agree with the comment regarding line 64 and will replace “upwelling” by “availability”. Line 96 will also be reformulated to state that upwellings bring nutrients from deep waters to the euphotic zone.

L104: We agree that a reference is needed and suggest citing the reference book on Lake Tanganyika by Coulter (1991).

L336-337: Regarding the reviewer statement about lines 336-337, we agree to change the phrasing to “Most previous studies relied on sampling methodologies to characterize phytoplankton concentrations”.

L: 347: Our results did indicate that some coastal areas, particularly those that are relatively shallow, do show distinct patterns of surface phytoplankton concentrations. These areas show year-round higher concentrations and less pronounced seasonal variations compared to deeper regions of the lake. We acknowledge that the dataset used does not allow to draw conclusions regarding phytoplankton community composition which our study did not attempt to characterize.

L 417: We agree with the reviewer’s comment on line 365. The use of the terms “oligotrophic” and “eutrophic” should be revised as Chl-a levels in Lake Tanganyika, even in extreme events do not necessarily make it eutrophic.

L407: We agree with the relevance of this point and will include a sentence in the revised manuscript to highlight the importance of modelling climate-induced hydrodynamic changes, as suggested by Sterckx et al. (2023).

L431: We agree with the reviewer’s suggestion and will revise the sentence accordingly to reflect a more cautious interpretation.

Antenucci, J. P. (2005). Comment on “Are there internal Kelvin waves in Lake Tanganyika” by Jaya Naithani and Eric Deleersnijder. Geophysical Research Letters, 32(22), 1–2. https://doi.org/10.1029/2005GL024403

Coulter, G. W. (1991). Lake Tanganyika and its life. Oxford University Press.

Delandmeter, P., Lambrechts, J., Legat, V., Vallaeys, V., Naithani, J., Thiery, W., & Deleersnijder, E. (2018). A fully consistent and conservative vertically adaptive coordinate system for SLIM 3D v0.4 with an application to the thermocline oscillations of Lake Tanganyika. Geoscientific Model Development, 11(3), 1161–1179. https://doi.org/10.5194/gmd-11-1161-2018

Descy, J.-P., André, L., Vyverman, W., & Deleersnijder, E. (2006). Climate variability as recorded in Lake Tanganyika (CLIMLAKE). http://www.belspo.be

Horion, S., Bergamino, N., Stenuite, S., Descy, J. P., Plisnier, P. D., Loiselle, S. A., & Cornet, Y. (2010). Optimized extraction of daily bio-optical time series derived from MODIS/Aqua imagery for Lake Tanganyika, Africa. Remote Sensing of Environment, 114(4), 781–791. https://doi.org/10.1016/j.rse.2009.11.012

Plisnier, P., Cocquyt, C., Cornet, Y., Poncelet, N., Nshombo, M., Ntakimazi, G., Nahimana, D., Makasa, L., & MacIntyre, S. (2023). Phytoplankton blooms and fish kills in Lake Tanganyika related to upwelling and the limnological cycle. Journal of Great Lakes Research, 49(6). https://doi.org/10.1016/j.jglr.2023.102247

Sterckx, K., Delandmeter, P., Lambrechts, J., Deleersnijder, E., Verburg, P., & Thiery, W. (2023). The impact of seasonal variability and climate change on lake Tanganyika’s hydrodynamics. Environmental Fluid Mechanics, 23(1), 103–123. https://doi.org/10.1007/s10652-022-09908-8

Thank you for your comments.

Abstract: Regarding the mention of wind and water temperature in the abstract, this was intended to highlight their role as drivers of Lake Tanganyika’s hydrodynamic cycle, and therefore of phytoplankton biomass. Further mentions in the discussion are intended as interpretation without conclusive causal link.

Dataset / validation with in-situ data: We agree that linking satellite-derived Chl-a estimates with in situ measurements is important. In the revised manuscript, we will add a section in which the satellite-derived seasonality is compared to the seasonality described in previous works based on in situ measurements. This will help place our results in the context of existing field-based studies.

Implication for other studies: Our study provides a comprehensive view of the seasonality of surface phytoplankton biomass and its changes across Lake Tanganyika over a 20-year period. By establishing a robust picture of surface phytoplankton dynamics, it offers a relevant basis for future studies on the ecology of Lake Tanganyika and fisheries-related research, including those examining the links between phytoplankton productivity, food-we processes, and fisheries yields.

Results section numbering: We thank the reviewer for noticing the duplicated section numbers. In the final manuscript, these sections will be reorganized and renumbered accordingly.

Language: We acknowledge the language issues raised and will revise the manuscript for clarity and style, including correcting the examples highlighted by the reviewer.

We thank the reviewer for the detailed review.

We acknowledge the reviewer’s valuable comment on the current scale of the bathymetry map. Defining representative east-west cross sections is indeed challenging, as the extent and slope of the shallower zones vary considerably across different regions of Lake Tanganyika. However, we agree that the bathymetry map (Figure 1) could be made clearer. We will therefore adjust the map scale to better illustrate depth variations within the key intervals mentioned by the reviewer (<170m, 250m and 0-40m).

We thank the reviewer for this detailed suggestion regarding the methodological structure. However, our approach intentionally differs from the proposed framework. Rather than first partitioning the lake by depth, we chose to evaluate the temporal dynamics of Chl-a (seasonality and long-term trends) at the pixel level across the entire lake, independently of water depth. This allows a good assessment of temporal behaviour before any spatial aggregation.

Subsequently, pixels are grouped into clusters based on the temporal covariance of Chl-a, which enables us to identify coherent regions with similar variability patterns. Water depth is indeed not included in this clustering approach. These clusters are then used to describe Chl-a seasonality and analyse quantile-based trends. This approach focuses on the intrinsic temporal coherence of Chl-a dynamics rather than on predefined bathymetry zones, which we believe provides a more data-driven characterization of spatial variability.

Regarding the reviewer’s comment on the availability of in-situ datasets covering large areas of Lake Tanganyika over multiple years, they are unfortunately not available for wind speed, surface water temperature or water turbidity. For Chl-a, however, Horion et al. (2010) validated a MODIS-based algorithm for estimating Chl-a concentrations in Lake Tanganyika using in-situ measurements collected between 2002 and 2006. The algorithm applied in this study is similar to the one used in the lakes_CCI processing chain, ensuring consistency with that validated approach.

We confirm the statement that “90% of the dataset had an associated fractional uncertainty between 39 and 60%”, and we acknowledge that an uncertainty analysis of the computed trends is necessary. We propose to propagate the pixel-level uncertainties though the trend estimation process using a Monte Carlo approach, thereby providing an uncertainty estimate for each computed slope.  

Regarding the reviewer’s comment on how missing Chl-a data were handled, we clarify that the DINEOF (DATA Interpolating Empirical Orthogonal Functions) method was used to reconstruct missing values. The interpolation was restricted to areas that were observed at least once during the study period, ensuring that no extrapolation was performed beyond data-supported regions. Moreover, DINEOF was applied only to dates for which the available data covered at least 5% of the lake’s surface.

We thank the reviewer for the helpful suggestion of merging sections 2.4, 2.5 and 2.6. We will merge them into a single section titled, as suggested, “Analysis Methods for Seasonal and Interannual Variations of Chlorophyll-a”.

We appreciate the reviewer’s suggestion to change the title of section 2.7. It will be renamed “Partition and Shifts in Chl-a Distribution”.

We agree with the reviewer’s detailed comments regarding Figure2. We will enlarge the April and October/November panels to better highlight differences in Chl-a seasonal trends between coastal shallow zones and deeper regions, as well as north-south contrasts. Specific areas mentioned by the reviewer such as the Gulf of Burton, will be indicated on the first bathymetry map of Lake Tanganyika, along with the main surrounding cities, as already shown.

We thank the reviewer for the comments regarding Section 3.2. Regarding the clustering approach, we note that the spatial aggregation was designed to groups areas exhibiting similar temporal variations in Chl-a. We acknowledge that some intra-cluster variability among pixels remains, even if differences are small.

Concerning Figure 4, the legend depicts the generic structure of the graph, with shaded areas representing the 25-75% and 10-90% intervals. The 25-75% interval is colored according to the cluster colors to facilitate interpretation, ensuring color consistency across Figures 3, 4 and 7.

With respect to the observed seasonal patterns, the article already mentions the northern upwellings in September, followed by secondary upwellings in the south, along with dampened thermocline oscillations. We agree that these dynamics can lead to multiple phytoplankton peaks, as observed in Clusters 1 and 2, and we will add further discussion to explicitly link these physical processes to the observed seasonal Chl-a patterns in the south.

We thank the reviewer for this comment. We note that Figure 7 already identifies areas shallower than 170 m, which allows readers to visualize those two zones (shallower or deeper than 170m). We will revise the text to explicitly refer to this figure when discussing trends and their relationship with depth, guiding readers to the appropriate reference.

We thank the reviewer for the suggestion of adding absolute monthly trend magnitude maps, along with the associated p-values. We will provide these maps in the Supplementary Materials to avoid overcrowding the main text with additional figures.

About the reviewer’s comment on Quantile-Based Trend Analysis, the line graph in Section 4 was designed to summarize multiple annual quantile maps while limiting the number of figures presented in the manuscript. While it is possible to provide separate maps for each annual quantile, the line graph effectively conveys the median trends per cluster, distinguishing between pelagic (deep water) and shallower zones. We acknowledge that some intra-zone pixel variability remains, but the figure provides a clear overview of the general patterns across depth zones. 

Thank you for the comment on the spatial context of areas with increasing Chl-a values. Regions with high Chl-a values and increasing trends are usually located near the coasts with relatively shallow waters. What we meant in the text is not that there are exceptional high-Chl-a areas in deep waters, but rather that among the shallow regions, some areas showing high Chl-a are neither close to major river estuaries nor near urban centers. At this stage, our analysis is descriptive, and the underlying causes of the positive trends in Chl-a levels remain to be further investigated.

We appreciate the reviewer’s comment regarding the proposed three-season framework for Chl-a dynamics. However, we respectfully disagree with the suggestion that the data are inconsistent with a three-season pattern. While Cluster7 does not exhibit clear seasonal variations, the remaining six clusters do show three distinct seasons: the first spanning from April to September, the second until December and third Chl-a season from December until April. This pattern is consistent with the three-season framework emerging from the remote sensing observations.

We thank the reviewer for the comment regarding the decline in primary productivity. The current version of the text incorrectly refers to Chl-a as a direct indicator of primary productivity. In fact, Chl-a is only a partial indicator, as it reflects phytoplankton biomass rather than production rates. This will be corrected in the next version of the manuscript.

Primary productivity and fish yields are influenced by multiple factors including nutrient availability, itself driven by upwelling dynamics and light availability influenced by water turbidity, and solar radiation. The seasonality and long-term trends we have computed should therefore be interpreted as an indicator of phytoplankton biomass seasonality and its evolution over the 20-year period, rather than a direct measure of primary productivity or fisheries yield.