the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 May 2025
Leveraging 20 Years of Remote Sensing to Characterize Surface Phytoplankton Seasonality and Long-Term Trends in Lake Tanganyika
Abstract. Lake Tanganyika, the world's second-largest freshwater lake by volume, is a vital resource for millions in East Africa, providing water, food, and economic opportunities while supporting exceptional biodiversity.
Chlorophyll-a concentration (Chl-a) is a key indicator of phytoplankton biomass and primary productivity, and thus a proxy for the health of aquatic ecosystems. In Lake Tanganyika, Chl-a is known to display strong spatiotemporal horizontal variability with an exceptionally low annual mean and wide ranges of concentrations compared to other tropical or temperate great lakes. This variability is influenced by the lake's hydrodynamic cycle driven by air temperature and wind seasonality. Phytoplankton biomass is suspected to be decreasing due to a strengthening of water column stratification induced by climate change. However, the particular spatiotemporal variability and trends in phytoplankton biomass have never been examined using a lake-wide, temporally continuous long-term record. This study bridges this gap by analyzing satellite remote sensing-derived Chl-a data from the ESA Climate Change Initiative Lakes dataset across the entire surface of Lake Tanganyika over a 20-year period. It offers insight into the Chl-a dynamics with an unprecedented timespan and spatial coverage.
The analysis reveals distinct seasonal patterns in Chl-a concentrations, with shallow regions (depth <170 m) maintaining high levels year-round, while deeper areas exhibit pronounced seasonality tightly linked to known wind patterns. To further explore these spatial differences in seasonal dynamics, the study identifies seven clusters of co-varying Chl-a concentrations, each displaying distinct seasonal behaviours that reflect the lake's hydrodynamic cycle. Long-term trends indicate a decline in Chl-a concentrations of -9 % per decade in deep regions, suggesting decreasing primary productivity. However, this overall decline is nuanced by monthly patterns. In deep regions, the low Chl-a concentrations, mainly observed between November and April, tend to decrease over time at rates between -5 to -15 % per decade when averaged over entire clusters. In contrast high Chl-a values recorded during the most productive months, from August to October, show increasing trends up to 25 %. Nearly all shallow areas, meanwhile, display year-round increases up to 35 % across the Chl-a distribution, with particularly sharp rises in extreme values.
The findings underscore the complexity of Lake Tanganyika's Chl-a dynamics. The observed trends may have significant consequences for the lake's trophic structure and the communities dependent on its resources. Further research is needed to elucidate the underlying drivers of these changes and to assess their broader ecological and socio-economic impacts.
Competing interests: Some authors (Marnik Vanclooster) are members of the editorial board of journal Hydrology and Earth System Sciences.
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 preprint. The responsibility to include appropriate place names lies with the authors.- Preprint
(1320 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 30 Jul 2025)
-
CC1: 'Comment on egusphere-2025-1326', Mariano Bresciani, 04 Jun 2025
reply
The manuscript is well written, easy to read, and makes an interesting point about the high spatial heterogeneity found in deep, very large lakes. Reading the abstract, introduction and methodology, it seems that the study is well structured. The methodology used appears appropriate for the study's objectives. However, the second part of the manuscript, including the results, discussion and conclusion, is less convincing.
The first remark concerns the use of the terms 'shallow coastal area', given that Lake Tanganyika is deep, even in the coastal zones. According to Moss et al. (2003), shallow lakes have a depth of less than 3 m. Secondly, the use of 'high values of Chl-a' is misleading. According to the OECD (1992), eutrophic lakes have a mean Chl-a concentration between 8 and 25 mg/m³. Therefore, the terminology must first be revised. In Lake Tanganyika, we can only speak of littoral/coastal versus pelagic areas, or of 'shallower areas' for comparison purposes, and of lower and higher concentrations of Chl-a, because Lake Tanganyika is an oligotrophic lake. See also later in the minor/specific comments.
Another point is the error associated with the concentration values. Maps showing the standard deviation, for example, are not displayed. There is a risk of speculating on data with a high degree of error because low Chl-a concentrations cast doubt on the reliability of the detection. This applies to both algorithms and spectrophotometric methods. This point is also valid for trend calculations.
From the abstract, it seemed that meteorological and climatic data (e.g. air temperature and wind) were used in the analysis to establish a relationship with Chl-a variability. However, this data was not used in the study, which seems to have been based solely on an assessment of Chl-a evolution.
Lake Surface Water Temperature (LSWT, another Lakes_cci variable) was also mentioned in the Materials and Methods section, but was not presented in the Results section to complement the spatial analysis of Lake Tanganyika. Another point that is missing is the fact that transparency, or Secchi disk depth, was not considered in the work, even though it is fundamental to phytoplankton blooms (see also minor comments later). Remote sensing data relate to surface waters and the euphotic zone, neither of which were discussed in the study. Light penetration is essential not only for phytoplankton, but also for Chl-a measurements using remote sensing techniques.
Another variable which is also available from Lakes_cci dataset is lake water level, that can be related with LSWT and Chl-a evolution in the long-time series. The importance of the levels are mentioned in the discussion but the data are also available and should have been integrated.
As a consequence of this lack in the analysis, the conclusions drawn are not fully appropriate or supported by the findings of this work. See later in the specific comment section.
A significant part of the discussion relates to deep and shallow lakes, but this is not appropriate given the initial remark raised. We could speak of the coastal and pelagic zones for comparison. Furthermore, some statements relate to wind direction and intensity, as well as air temperature, but no data are reported (probably from the literature). While I agree that ground-based wind data are not available, reanalysis data such as wind components (as well as air temperature) are available from the ERA5-Land dataset. Given the size of Lake Tanganyika, the spatial resolution of this data is almost acceptable.
Finally, in the discussion it is useful to mention also the limits of the dataset which can influence the results (e.g. data gap from 2012 to 2016, less availability of data in the rainfall season due to cloud cover).
Other comments:
I suggest adding some more recent citations to the first paragraphs of the introduction, for example lines 34–46.
Line 104: “Lake Tanganyika’s water is remarkably clear and nutrient levels at the surface are generally low.” Some reference data for secchi disk depth would be appreciated.Line 110 and later: “mg C m-2” without point (mg C.m-2).
Line 115-135: Move to Intro or to support discussion?
Line 144: “Most lakes, including Tanganyika, show data gaps during the 2012 to 2016 period due to validation issues.” Incorrect, in the current available version (2.1), MODIS data for Tanganyika were not released. Delete this sentence or update accordingly.
Line 155: “95% missing values”. Is almost always and with a patchy distribution of the pixel? It seems a large “threshold”.
Line 220: “These coastal zones usually extend a few kilometres offshore and are relatively shallow, with depths generally not exceeding 250 m (Figure 1). They exhibit consistently high to very high median concentrations throughout the year, above 3 mg.m-3 in some areas.” I suggest to change the terminology for shallow areas and “high” median values, as shallow lakes are lakes with a depth < 3 m and eutrophic lakes have a mean Chl-a concentration in the range 8-25 mg m-3. For this reason it would be better say something like this: “These coastal zones usually extend a few kilometres offshore and are shallower (<250 m) compared to the pelagic areas, and they exhibit higher Chl-a concentration above 3 mg m-3 in some areas.”
Line 407: “Ground-based wind data are limited, and the reliability of historical wind speed records in the region remains a subject of debate (Eschenbach, 2004; O’Reilly et al., 2003).” Reanalysis data are available and suitable for this case study. I suggest including them in the work.
Lines:417-419: “This duality suggests that Lake Tanganyika exhibits characteristics of both a shallow lake becoming more eutrophic and a deep lake becoming more oligotrophic, adding to the complexity of its ecological functioning.” And Lines 428-430: “The contrasting trajectories suggest that Lake Tanganyika is simultaneously exhibiting characteristics of a deep, nutrient-limited system undergoing further oligotrophication and a shallow system experiencing intensified phytoplankton growth.
As in the previous comment, I don't think these statements are entirely accurate, because Tanganyika doesn't exhibit more "eutrophic" waters (at least not a higher concentration of 3 mg/m³ compared to open waters).
Lines 423-425: “The analysis confirms that shallow areas exhibit persistently high Chl-a concentrations year-round, whereas pelagic regions show strong seasonal variability, with peak productivity following wind-driven mixing events.” This is not really an outcome of this study, as wind was not included in the analysis. It is more a speculative sentence most likely based on literature research.
Citation: https://doi.org/10.5194/egusphere-2025-1326-CC1 -
CC2: 'Comment on egusphere-2025-1326', J-P. Descy, 17 Jun 2025
reply
Review of egusphere-2025-1326
The work uses teledetection estimation of chla concentration in Lake Tanganyika to examine log-term (~20 y) and seasonal changes of phytoplankton biomass in Lake Tanganyika. Like in previous work based on a similar approach, it shows the spatial heterogeneity of phytoplankton biomass over this large lake, further insisting on differences between the pelagic zone and some coastal areas or the southern and northern ends of the lake.
I agree with the authors that, given this great spatial heterogeneity, field-based chlorophyll measurements at fixed sites for long-term monitoring cannot capture all changes that occurred and will occur in such a large lake, as a result of climate change and other anthropogenic influences. This aligns very well with conclusions drawn from Descy et al (2005):
“Interannual variability in weather pattern and lake hydrodynamics can be high, and only monitoring of phytoplankton composition, biomass and production over long periods of time, and at several sites to account for spatial variation, allow proper assessment of relative effects of short- and long term environmental variability on this community”.
In this context, teledetection is a great tool to examine variation of lake surface Chla, lake surface temperature (LST) and even light penetration at large spatio-temporal scale, and this paper does the job quite well for chla. Particularly interesting is the information presented in figure 6, which shows the very different seasonal patterns depending on the lake region, with contrasting behaviour between the northern part of the lake (strongly stratified during the rainy season) and the more dynamic southern part (due to lower temperature gradient and more frequent mixing).
All this is fine, however we miss a few important things here, in particular the link with the complex hydrodynamics in lake Tanganyika, on which important work has been done in the past few years (e.g. Delandmeter et al., 2018; Sterckx et al., 2023). Establishing this link might help to point out the cause for some observations, as the “blooms” (I believe the term should be used with more caution in the ms.) that occur the coastal areas. They might be the results of coastal nutrient input from upwelling caused by oscillations such as Kelvin waves, the existence of which being likely, according to Antenucci (2005).
They might also consists of periphyton growing on substrates in the littoral zones, brought into suspension by turbulence from wave action and transported offshore. And of course the littoral zone may be subjected to local nutrient inputs, from rivers and from villages on the shore. Actually, several processes can explain that chlorophyll a is often higher in the coastal zone than in the pelagic zone, although there is no compelling evidence from field data that inshore waters have sytematically higher chla than offshore waters : data can be found in Descy et al. (2006). Of course these data came from sampling fixed pelagic and coastal sites, which may not reflect the situation at larger spatio-temporal scale.
What I also miss is a proper validation of the chla concentrations determined from satellite data. Despite a good general agreement with the published chla concentration in lake Tanganyika, both from field measurements and estimates from satellite data, this needs to be done to prove that their derived satellite data are realistic and in the right range. Here the authors seem to have missed a calibration (to my knowledge the first one for this lake) between whole-lake field concentrations (from north-south cruises) and the estimates from the application of the existing algorithms at that time (Horion et al. , 2010, in RSE). Using these published data, the authors could have assessed the agreement with their own data.
Other comments
L10 “Chlorophyll-a concentration (Chl-a) is a key indicator of phytoplankton biomass and primary productivity, and thus a proxy for the health of aquatic ecosystems”. Chlorophyll a is a proxy of phytoplankton biomass and not of productivity; and it is not a proxy of ecosystem health. Ecosystem health is a broader concept, which encompasses not only trophic status, but other criteria and indicators, including foodweb structure, resilience, biodiversity ...
Productivity can be assessed from estimates of primary and secondary production over a sufficiently long period of time, e.g. a year or several years. For example a typical unit of primary productivity is g C.m-2.y-1. Chla is just a concentration, which can be expressed as g C.m-3 or g C.m-2 if integrated over the water column, using a C:Chla conversion factor.
Elsewhere in the ms., there is a confusion between chla concentration (phytoplankton biomass) and primary production. The authors should be aware that primary production depends not only on chlorophyll a concentration but also on light penetration and on photosynthetic capacity of the phytoplankton assemblage. They should also know that phytoplankton biomass at a given time/site results from the balance between production and losses (incl respiration, grazing, sedimentation...). So for example, it is quite possible that chla in the pelagic zone is lower than in the coastal zone because losses by grazing and sedimentation are lower in the coastal zone, assuming a same production rate in both zones.
LO 28 and elsewhere : avoid to use the term “shallow” which is used for lakes of only a few m deep. Lake Tanganyika is a deep lake, with a littoral zone (defined by light penetration down to a depth which allow the development of macrophytes and periphyton) of limited extension, sometimes only a few dozen of m. Using “coastal” or “inshore”, as opposed to pelagic and offshore may be more appropriate.
L42 and following : see comment above. Chla is an indicator of trophic status, among other indicators such as Secchi depth and macronutrient concentration. Lake ecosystem health is something different as mentioned above.
L50 Remote sensing provides a measurement of surface chlorophyll a (with no indication of distribution of phytoplankton biomass in a stratified water column); among the references cited, Horion et al. 2010, who made a first estimate of chla in Lake Tanganyika from remote sensing, is missing
L64 I suggest to replace “upwelling” by “availability” or some equivalent term, as increase of stratification does not necessarily reduce upwelling.
L 96 upwelling does not increase nutrient cycling (which depends on biological processes) but brings nutrients from deep water to the euphotic zone
L 104 “Lake Tanganyika’s water is remarkably clear and nutrient levels at the surface are generally low” References missing here.
L 105 “Phytoplankton concentration show great spatiotemporal variability depending on the nutrient availability in the euphotic zone”; As mentioned above, phytoplankton biomass depends on several processes, and it is phytoplankton growth (not biomass) which depends on nutrient availability.
L 336-337 “Most previous studies did not use remote sensing data and focused on phytoplankton concentrations near the coasts” : this is incorrect
L 347 “that a key factor differencing these two types of areas is depth, rather than simply their distance from the shore. Lake
Tanganyika could be seen as having two distinct ecosystem areas, each with its own characteristic phytoplankton
concentration levels and seasonality”. Yes, coastal waters have different food webs and vastly different biodiversity, but when it comes to phytoplankton biomass and composition, pelagic and coastal zones are very similar.
L 365 “Lake Tanganyika appears to exhibit characteristics of both lake types, depending on the depth” : if this means that the coastal zone is shallow and eutrophic whereas the pelagic zone is oligotrophic, the statement is incorrect. Lake Tanganyika is oligotrophic and the range of chla in the coastal zones remains in the oligotrophic range.
L390 I would suggest to use instead a C:Chla ratio determined for this lake, from POC – Chla regressions.
L407 Modelling climate impact on the lake hydrodynamics may help for the prediction of the outcome of different scenarios, as suggested by Sterckx et al. (2023)
L 417 “This duality suggests that Lake Tanganyika exhibits characteristics of both a shallow lake becoming more eutrophic and a deep lake becoming more oligotrophic” . See comment on L 365
“These shifts could have cascading effects throughout the lake’s trophic network, potentially altering food web dynamics and ecosystem stability” . This would rather be a bottom-up effect, as cascading suggests a top-down effect. Plus they don’t know much about the food web and its dynamics ... I suggest to be more careful, as this is an ancient lake, which has seen many changes over time, perhaps of greater importance than the current anthropogenic alterations.
Citation: https://doi.org/10.5194/egusphere-2025-1326-CC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 120 | 19 | 9 | 148 | 5 | 9 |
- HTML: 120
- PDF: 19
- XML: 9
- Total: 148
- BibTeX: 5
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1