Introduction

Coastal lagoons are inland water bodies parallel to the coast, separated from the ocean by barriers, connected through narrow inlets, and characterised by shallow depths rarely exceeding a few metres1. They can be categorised into choked, restricted, and leaky systems based on the degree of water exchange with the adjacent coastal ocean2. The hydrographic conditions of the more isolated lagoons primarily depend on the local water balance from the effects of surface waves, atmospheric forcing, and tidal oscillations3,4,5. Such lagoons can sometimes exhibit different ecological characteristics to adjacent water bodies6. Coastal lagoons are complex habitats of global importance, offering multiple ecosystem services and supporting a diverse range of marine life due to their high productivity1,6. Ecological functions of different organisms such as zooplankton, macro-crustaceans, birds, fish or marine mammals in a lagoon can be beneficial in many ways. The ecosystem services provided can support primary and secondary production, nutrient cycling, sediment and soil formation, and habitat provision7,8,9,10,11,12. A healthy lagoon can also provide biological control of both marine and terrestrial pests via natural predation13,14,15. Seagrass meadows, which are commonly found in coastal lagoons, help mitigate climate change by sequestering carbon from the atmosphere16,17,18, and microbial communities and bivalves can enhance water purification19,20. However, close proximity to urban areas can introduce numerous anthropogenic stressors to coastal lagoons, including eutrophication, contaminant influx, non-indigenous species introduction, and overexploitation of fisheries, placing them among the world’s most endangered ecosystems, potentially exacerbating vulnerability to growing pressures from climate change11,13,21,22.

The Red Sea is a tropical marginal sea that hosts one of the longest coral reef systems in the world23. Its unique oceanographic setting, with rapidly increasing water temperatures from north to south24, makes it a natural laboratory to study the effects of climate change25. The strong temperature gradient has led to a high ratio of endemic species26, as well as corals that are more resistant to thermal stress27,28. The Red Sea also hosts several coastal lagoons, which are enclosed primarily by pristine coral reefs that are characterised by high biodiversity29,30.

Fig. 1
figure 1

(a) High-resolution bathymetry of the shallow areas surrounding Sharma Lagoon with a reference map pointing its location in the northern Red Sea. Data is taken from the Living Oceans Foundation, and the land basemap from ESRI. This dataset is limited to 20 m depth, therefore, areas deeper than that appear transparent. The extent of Sharma Lagoon, which is referred to in following figures, is highlighted by the white polygon. Maps were generated in ArcGIS Pro 3.0.3 (esri.com). (b) Bathymetry of the NEOM region covering the entire extent of the model. The dashed white box highlights the area used in (a), as well as in Figs. 2 and 3. Data is taken from the Delft3D model. Map was generated in Matlab 2023b (mathworks.com).

Full size image

Sharma is one such lagoon located in the northern Red Sea (NRS), and has experienced increasing anthropogenic pressure in the past few years31. Sharma can be classified as a restricted lagoon with a 20 km long elongated shape, and covers an area of \(\sim\)150 km2, including a small island (Fig. 1). Although, the lagoon is the endpoint of 3 dry riverbeds (wadi) of large catchment areas, the average annual rainfall is low (<40 mm yr-1) and usually occurs in short bursts that can cause flash floods32. The lagoon, which has an average depth of \(\sim\)20 m, is surrounded by coral reefs and is primarily connected to the NRS via one narrow entrance with a sill depth of \(\sim\)7 m (Fig. 1). Sharma Lagoon is characterised by relatively high chlorophyll-a (Chl-a) concentration and an opposite Chl-a seasonality compared to the open NRS waters31. From previous findings, the phytoplankton main growth period inside the lagoon occurs during summer, while the lowest concentrations are observed during winter31. In contrast, NRS waters outside the lagoon experience high Chl-a concentrations during winter and very low concentrations in summer due to thermal stratification limiting nutrient availability24,31,33. Diverse topographical features, hydrodynamic patterns, and climatic conditions can contribute to distinct seasonal variations in Chl-a within coastal lagoons. For instance, a similar paradoxical Chl-a seasonality was also observed in Al Wajh, a larger restricted lagoon in the NRS34,35. Its Chl-a peaks in September, and reaches its lowest concentration in March, which is opposite to the NRS Chl-a seasonality as described above24,31,33,36. This phenomenon was attributed to the role of water exchanges with the open sea, which is limited to surface waters by the semi-enclosed nature of the lagoon, and its effect on the stratification of the water column inside the lagoon37. Consequently, vertical mixing, which increases nutrient availability, peaks during different seasons inside and outside the lagoon35. Another example of a nutrient self-sustaining lagoon is Bahia-Magdalena at the west coast of the Peninsula of Baja-California. The Chl-a peak appears in June due to upwelling events, while the lowest concentrations occur in the winter, contradicting with the surrounding waters, where Chl-a peaks in spring due to upwelling events38. Lagoon hydrodynamics can also be largely influenced by tidal oscillations, controlling nutrient availability, such as in Setiu Lagoon, Malaysia39. Rainfall and river discharge has been shown to affect Chl-a seasonality in different ways in coastal lagoons. In Açu Lagoon, Brazil, rainfall stimulates phytoplankton growth40, while in Alvarado Lagoon, Mexico, Chl-a is at its lowest during the rainy season41. In other cases, intense weather events, such as tropical cyclones in Chilika Lake, a brackish coastal lagoon in India, can have a delayed effect on phytoplankton growth42. Comprehensive independent studies are necessary to understand the unique ecological characteristics of each system. Detailed studies of the lagoon have not been conducted yet, and the mechanism behind the phytoplankton seasonality paradox remains unknown.

Sharma Lagoon is part of the NEOM region, a gigaproject consisting of multiple development sites currently under construction in northwestern Saudi Arabia43,44. The project’s success largely depends on the sustainable development and conservation of the region’s natural resources that can attract tourism and investments. For NEOM, this primarily translates to its unique marine habitats, such as coral reefs and coastal lagoons, since most of its land is arid. Currently, a golf course with a small marina, which has experienced some dredging, is located in the east side outside of the lagoon. Two tourist resorts are located on the lagoon’s shoreline, and NEOM’s community area is located nearby (\(\sim\)3 km) since 2022. However, none of these establishments are likely to be a source of eutrophication in the lagoon. Sharma’s unique characteristics, alongside ongoing development activities in the NEOM region, highlight the importance of understanding and managing the region’s ecological dynamics.

Here we aim to identify the mechanisms underlying the difference in Chl-a seasonality between the interior and exterior of Sharma Lagoon, in the NEOM region. We hypothesise that the lagoon experiences enhanced vertical mixing during summer, that is increasing nutrient availability, similarly to Zhan et al.35. The synergy of topography and various environmental factors, such as water transport, is analysed with a multi-disciplinary approach. By utilising high-resolution satellite-derived ocean colour observations, high-resolution hydrodynamic model outputs and in situ measurements, we seek to provide a comprehensive understanding of why the phytoplankton peak occurs in summer within the lagoon, contrary to the winter peak observed at the surface level outside in the NRS open waters. Additionally, the potential impacts of human activities that can modify the lagoon’s water exchange rate, and thus its delicate Chl-a seasonal cycle, are discussed. By addressing these objectives, we aim to contribute to the broader discourse on sustainable coastal management and conservation practices in NEOM, which is critical for the preservation of these invaluable marine ecosystems that are facing increasing developmental pressures.

Methods and data

This study integrates multiple data sources to investigate the physical mechanisms driving phytoplankton variability in the Sharma Lagoon. We combine a high-resolution bathymetric dataset to accurately represent the lagoon’s complex topography, satellite-derived Chl-a estimates to characterise seasonal phytoplankton dynamics, a Delft3D numerical model to explore hydrodynamic processes, and in situ measurements to validate remote sensing observations and model outputs. Finally, a Lagrangian particle tracking experiment is conducted using OceanParcels to assess water exchange and residence times under different seasonal conditions.

Bathymetry data

The model bathymetry for the study region was constructed by merging three different datasets. Specifically, (i) the General Bathymetric Chart of the Oceans (GEBCO) at a 30 arc-second resolution45, which is one of the most commonly used bathymetry datasets; (ii) a very high-resolution satellite-derived bathymetry (\(\sim\)0.5 m), tuned with in situ measurements, from the Danish Hydrologic Institute Geographic Resource Analysis & Science (DHI GRAS), offering precise depth information for shallow areas with depths up to 30 m; (iii) digitized data sourced from the Saudi Port Authority hydrographic charts for offshore deep regions. The merging and interpolation of these datasets resulted in a comprehensive bathymetry dataset for the study area.

A separate high-resolution (2.4 m) shallow bathymetry dataset was obtained from the Living Oceans Foundation, and used to visualise the complex barrier structures surrounding the lagoon. It is based on QuickBird satellite imagery, and thus limited to 20 m depth. More details regarding the methodology that was used to create this dataset can be found in Bruckner et al.46.

Satellite remote sensing data

To characterise the seasonal variability of Chl-a, we utilised a dataset stemming from the Sentinel-3 Ocean and Land Colour Instrument (OLCI) at 300 m spatial resolution spanning the period from May 2016 to May 2020. This product was preferred over other well-established Chl-a products because its higher spatial resolution best suits to capture features inside the lagoon, while also providing enough data temporally (4 complete years) to observe patterns in the seasonal cycle. Daily data was composited into monthly and seasonal means to highlight the difference in Chl-a phenology inside and outside the lagoon.

The waters in the NEOM region, similarly to the wider Red Sea, are considered optically complex due to increased coloured dissolved organic matter absorption per unit Chl-a47. Consequently, standard ocean colour algorithms tend to overestimate Chl-a when compared with in situ data48,49. To overcome this issue, our dataset was tuned using the OC4ME_RG algorithm derived from the standard NASA OC4 algorithm50, the MERIS ocean-colour sensor wavelength, and Red Sea regional tuning, as shown in Table 3 of Brewin et al.51. This algorithm performs significantly better in the study region compared to standard NASA ocean colour algorithms and other semi-analytical algorithms51. Our Chl-a dataset was also validated with in situ Chl-a observations from the study area, both inside and outside the lagoon, with a correlation coefficient r = 0.75 (p = 3*10-6) from n = 29 matchups (Supplementary Figure S1b).

The Sharma Lagoon area includes many shallow parts (<20 m), usually covered by coral reefs and other complex habitats31, which can also introduce bias to the satellite observations. Underwater reflectance and scattering by sediments may increase the water-leaving radiance in the near-infrared wavelengths, affecting the atmospheric correction52. Therefore, an overestimation may be present in some of our Chl-a observations in the study area. To reduce observation bias, only the deepest parts of the lagoon (>20 m), which are the least affected, were used to describe the Chl-a seasonality. Observations in other areas in the Red Sea with similar topography have been validated with in situ data53,54. Despite the slight bias (-0.096 mg m-3), which mostly occurs in the lower Chl-a values in the data used, it is still possible to describe the general Chl-a seasonality inside the lagoon, which is the main goal of this study. In addition, the slope could be improved further with additional targeted in situ measurements covering the entire trophic range, due to the currently low number of matchups at high Chl-a concentrations, where the bias is smaller (Supplementary Figure S1b).

To describe seasonal processes that occur reliably each year, we focus our analysis on seasonal climatologies derived from multi-year satellite observations across multiple years (2016–2020). This approach allows us to isolate and characterise recurring seasonal patterns while minimising the influence of short-term anomalies or isolated events. Although interannual variability is an inherent part of environmental signals, the relatively low variability in key parameters such as Sea Surface Temperature (SST) and Chl-a in the NEOM region (Supplementary Figure S1c) supports the use of seasonal means for this purpose. The SST and Chl-a difference between the lagoon and the open waters exhibits a stable seasonal cycle as shown in Supplementary Figure S1c. This is particularly relevant for understanding mechanisms that are tightly linked to annual cycles within Sharma Lagoon.

Field measurements

The in situ data used in this study were collected from four oceanographic surveys conducted between August 2017 and April 2018, each corresponding to a different season (summer: August, autumn: November, winter: February, spring: April). Temperature and salinity measurements were taken at 1 m intervals with a CTD probe in multiple stations during August 2017 and February 2018 (Supplementary Figure S2a). These measurements were also used to calculate water density at the same intervals to compare seasonal differences in stratification inside and outside the lagoon.

Seawater samples were collected during November, February and April (locations shown in Supplementary Figure S2b) with 5 L NISKIN bottles from several depths, including the surface (1 m depth), allowing direct comparison with the satellite Chl-a data. The water samples were filtered on board using Whatman GF\(\backslash\)F filters, which were then kept frozen in a dark environment at approximately -20 oC. Chl-a concentration was measured in the Analytical Core Lab (ACL) of King Abdullah University of Science and Technology (KAUST) using a fluorometer according to an established protocol55. Nutrient analysis was also performed by filtering the water samples through a 0.45 \(\upmu\)m membrane Millipore filter and collected in 100 mL polyethylene bottles with a 10% HCl solution. These samples were also kept frozen and analyzed in the KAUST ACL by using a nutrient autoanalyser, following standard protocols such as described in Strickland and Parsons56. These measurements were used to validate the satellite-derived Chl-a observations in the region, and provide additional information about temperature and salinity along the lagoon’s water column.

Hydrodynamic model

The Delft3D numerical model was implemented to simulate the hydrodynamics of the Sharma Bay region. Delft3D-FLOW, the hydrodynamic module, was configured with a horizontal resolution of approximately 60 m and 10 vertical sigma levels. The model domain extended from \(27.5^{\circ }\)N to \(28.5^{\circ }\)N and from \(34.3^{\circ }\)E to \(35.5^{\circ }\)E, encompassing Sharma Bay and the Gayal lagoon. The model was configured with a curvilinear grid and used a Manning coefficient of 0.02 for bottom friction, along with horizontal eddy viscosity and diffusivity of 1 m2 s-1. The model was driven by high-resolution atmospheric forcing generated by downscaling the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis57 using a Weather Research & Forecasting model (WRF) configuration58. Details of the WRF configuration and validation are provided in Dasari et al.59. The resulting 600 m atmospheric fields were used to drive the hydrodynamic model, including wind speed and direction at 10 m height, temperature at 2 m height, relative humidity at 2 m height, cloud coverage, and accumulated rainfall made available at 30 min intervals. The tidal open boundary conditions were based on the TPXO 7.2 Indian Ocean (Red Sea) tidal model60 and integrated into a regional 1 km MIT General Circulation Model (MITgcm) of the Red Sea, which was driven by a 5 km atmospheric reanalysis59. Temperature, salinity, and water level data from MITgcm were used to establish boundary conditions for a higher-resolution Delft3D model through interpolation onto its grid, ensuring realistic, time-varying boundary conditions. The regional MITgcm outputs have been extensively validated in many studies of the general Red Sea circulation61,62,63,64,65.

To investigate seasonal variability, the model ran for 2 years (01/01/2013 to 31/12/2014) and the outputs were saved every hour. Density, temperature and salinity transect plots were created by interpolating the model outputs on a 2D grid for visualisation. The Brunt-Väisalä frequency (N2) was calculated using the standard formula to compare its seasonality with the satellite-derived Chl-a, and to compare it with the lagoon’s heat fluxes on an hourly scale. The mixed layer depth (MLD) was also calculated from the hourly outputs for each point inside the lagoon with a density threshold of 0.05 kg m-3, and then averaged between all points. We acknowledge the temporal mismatch between the simulation (2013–2014) and the observational datasets used in this study (2016–2020). While this discrepancy limits direct comparison, we justify the use of these model outputs by demonstrating the persistence of seasonal Chl-a and SST patterns in the region (Supplementary Fig. S1a, c), which supports the use of seasonal climatologies to investigate physical-biological mechanisms.

To assess model performance, the Delft3D model was validated against two Acoustic Doppler Current Profiler (ADCP) datasets collected between February and March 2018 inside and outside the lagoon, along with sea level observations from the OSU Tidal Inversion Software (OTIS). The modelled water level shows reasonable agreement with OTIS observations (Supplementary Figure S3a). The rose diagrams compare surface (2 m) current speed and direction from the model with ADCP measurements, showing that the model accurately represents the magnitude and dominant flow directions, albeit with a slight overestimation of strong currents in the south-southeastward direction (Supplementary Figure S3b). Model validation against in situ ADCP and temperature observations shows reasonable agreement. Surface current velocities have RMSE values of \(\sim\)18–21 cm/s and correlation coefficients of 0.61 (u) and 0.73 (v) at daily averages (Supplementary Table S1). However, directional agreement remained relatively low (R = 0.53). This reflects the dynamic nature of the lagoon exchanges, and is also affected by the model representation of the complex bathymetry at its entrance. SST and water level showed a stronger agreement, with correlation coefficients of 0.89 and 0.95, respectively, suggesting a very good representation of their variability in the lagoon. The agreement in directional distribution indicates the model’s ability to capture the lagoon’s water flows. Additionally, comparisons of simulated seawater temperature with in situ measurements demonstrate that the model effectively reproduces the thermal variations of the lagoon (Supplementary Figure S3c). The good agreement between model outputs and observations supports the use of this configuration for studying the lagoon’s hydrodynamic conditions.

Particle tracking

The OceanParcels package v3.0 for Python66 was used to release and track lagrangian particles. The field for the experiment was set for one week in late August (Chl-a peak within the lagoon) and one week in February (lowest Chl-a within the lagoon) during the same lunar phase, which was enough to observe differences in particle movement. The input data was 3-dimensional velocities (u, v, w) linearly interpolated on a high-resolution grid with 40 vertical layers (\(\sim\)10 m horizontal resolution and sigma vertical coordinates) from the Delft3D model outputs, extending between 27.89-28.12 \(^{\circ }\)N and 34.92-35.13 \(^{\circ }\)E. The input velocities have an hourly temporal resolution, while the experiment was tracking particle positions with a timestep of 15 min to enhance accuracy. Over 40,000 particles were released simultaneously throughout the water column of the lagoon in each run. To prevent particles from being stuck on land and the ocean floor, since the experiment was ran on an A-grid, a custom displacement field was created. The 3-dimensional location of each particle was written every hour in the output files, which were then used to estimate the lagoon’s flushing time.

Results

To investigate the underlying mechanisms behind the Chl-a seasonality paradox of Sharma Lagoon, both seasonal and diurnal analyses were performed. Firstly, we present the monthly Chl-a seasonality spatially for the entire study area, and compare the differences between summer and winter. We then examine the effects of mixing on Chl-a, and compare stratification strength between summer and winter using both in situ measurements and model outputs. Finally, we identify the primary mechanisms that contribute to the seasonal differences observed in the lagoon, namely the differences in water exchange dynamics and the diurnal variability of atmospheric forcing.

The chlorophyll-a seasonality paradox

Sharma is characterised by an opposite Chl-a seasonality, and consistently experiences higher surface Chl-a than the open NRS waters throughout the year (Fig. 2). Surface Chl-a in the open waters outside of the lagoon peaks during winter (January–February), while the lowest concentrations occur in summer (July–August). This seasonality is also amplified with increasing distance from the lagoon. The interior of the lagoon retains higher Chl-a throughout the year, but the seasonality is opposite to the exterior, peaking in late summer (August–September) and lower in winter (January-February). The observed patterns are spatially consistent within the lagoon. Available in situ measurements further indicate that the lowest Chl-a and silica nutrient concentrations occur during the winter (February) inside the lagoon (Supplementary Figure S4). These samples also reveal that Chl-a concentrations near the bottom of the lagoon are always higher compared to the surface, with the largest difference occurring in spring.

Fig. 2
figure 2

Monthly means of chlorophyll-a concentration in Sharma Lagoon and the surrounding area spanning from 2016 to 2020. Images are produced using data from Sentinel-3 OLCI at \(\sim\)300 m resolution using a regionally-tuned ocean colour algorithm. Different colour scales have been set for inside (dark green to yellow) and outside (white to dark green) the lagoon due to a large difference in the values. White areas, mainly populated by coral reefs, have no data due to the very shallow depths. Maps were generated in Matlab 2023b (mathworks.com).

Full size image

The largest Chl-a difference between the lagoon’s interior and exterior is observed in August and September, during the lagoon’s main growth period. In the winter, this Chl-a difference becomes smaller, especially in February, before increasing again abruptly in spring. The Chl-a difference between summer and winter is also visualised spatially in Fig. 3a. The relative difference is larger inside the lagoon, since Chl-a there is constantly higher than the oligotrophic NRS waters. Meanwhile, areas populated by coral reefs present lower seasonal variability with consistently higher Chl-a values than the open waters.

Fig. 3
figure 3

(a) Climatological chlorophyll-a difference map depicting summer (August and September) minus winter (January and February) at Sharma Lagoon. Images are produced using data from Sentinel-3 OLCI at \(\sim\)300 m resolution using a regionally-tuned ocean colour algorithm spanning from 2016 to 2020. (b) The spatial difference in mean buoyancy frequency (N2) from surface to 20m depth between summer and winter in the Sharma lagoon, northern Red Sea. Monthly mean data is taken from the regional Delft3D model outputs. Negative values (red) indicate stronger stratification during winter, while positive values (blue) indicate stronger stratification during summer. Note that the colourbar in (b) shows negative values in the top to allow better comparison with (a). Maps were generated in Matlab 2023b (mathworks.com).

Full size image

Seasonal stratification

A similar spatial pattern to the Chl-a seasonal difference is also observed in the lagoon’s stratification, represented by the seasonal difference of buoyancy frequency derived from the model outputs (Fig. 3b). The lagoon becomes more stratified in winter, in contrast to the exterior waters that are characterised by a deeper mixed layer during the same period. This is further supported by the in situ temperature and salinity measurements, three of which were selected as representative of the conditions both inside and outside of the lagoon, as well as at the deepest location of the channel (Fig. 4). While reduced stratification is typically associated with winter and cooler conditions, our analysis reveals a contrasting and counterintuitive result: stratification in this lagoon is actually weaker during summer. Specifically, although the overall density of lagoon waters increases in winter, the density difference between the surface and bottom waters is greater in winter (0.6 kg m-3) compared to summer (0.1 kg m-3), indicating stronger stratification during the colder months.. Additionally, there is a large difference between the density of the waters inside and outside the lagoon in winter (>1.5 kg m-3) compared to summer (<0.1 kg m-3 on average). In fact, the subsurface open sea waters (>15 m depth), which are not affected by the exchanges through the strait, have a higher density than the waters inside the lagoon during summer.

Fig. 4
figure 4

Vertical profiles of in situ temperature, salinity and density measurements from single stations during (a) February and (b) August. The locations of the stations are presented in the inset maps. The maximum sampling depth is slightly different between seasons for the interior and exterior stations. Each panel has different x-axis limits to highlight the variability between the stations during the same season. All panels were generated in Matlab 2023b (mathworks.com).

Full size image

The extreme water temperature range of the lagoon is the main driver of these differences. During winter, temperature declines with depth, specifically from 19.6 \(^{\circ }\)C near the surface to 18.2 \(^{\circ }\)C near the bottom, while salinity increases from 40.99 to 41.26. During summer, the variability inside the lagoon is smaller, with temperature decreasing from 31.6 to 31.2 \(^{\circ }\)C and salinity from 41.87 to 41.75. The lagoon exterior is characterised by the typical NRS seasonality, with a deep mixed layer during winter with a nearly uniform temperature of 22.6 \(^{\circ }\)C and density of 1028.2 kg m-3. On the other hand, strong stratification is observed during summer with temperature decreasing from 29 \(^{\circ }\)C at the surface to 27.4 \(^{\circ }\)C at the bottom and density decreasing by 0.35 kg m-3. The observations from the channel of the lagoon indicate mixing with waters from both inside and outside of the lagoon. The surface waters have a similar profile to the outside NRS waters, while the bottom waters have a profile similar to the lagoonal waters.

Seasonal water exchange variability

Tidal oscillations are the primary drivers of water exchange between Sharma and the exterior NRS waters. This is shown indirectly by the exchange of particles and the variations in sea surface height (SSH), which follow the tidal cycle (Fig. 5). Particles escape mostly during ebb tides, when water is flushed from the lagoon, especially in summer when the mean SSH is lower, indicating that water exchange is mainly tide-driven. There is a significant seasonal difference in the lagoon’s flushing time, as well as in the average SSH. The results from the particle release reveal a faster flushing time during winter (Fig. 5b), indicating increased water exchange with the NRS open waters. In winter, over 40% of the particles that were released inside the lagoon escaped within 6 days, while only \(\sim\)14% escaped in summer.

Fig. 5
figure 5

(a) Timeseries of Sea Surface Height inside Sharma lagoon, in NEOM, during a week in summer (red) and in winter (blue). (b) Sharma lagoon flushing time during summer and winter, showing the percentage of particles that have escaped the lagoon boundaries. The same lunar phase (full moon spring tide) was chosen for both seasons. Minimum and maximum values for each season are highlighted in the y-axis with the corresponding colours. Particles mostly escape during the ebb tide phases, when more water is exiting the lagoon.

Full size image

The total heat fluxes (qtot)in combination with tidal oscillations also play an important role in the lagoon’s N2 and MLD diurnal cycles (Fig. 6). The cross-correlation between N2 and (qtot) showed maximum correlation r = 0.75 at 3 hours lag in summer, and r = 0.38 at 2 hours lag in winter (Supplementary Figure S6). Similarly, the cross-correlation between N2 and the MLD reaches maximum r = 0.87 at 4 hours lag in summer, and r = 0.58 at 3 hours lag in winter. The seasonal difference can be attributed to flood tides, when the incoming NRS waters significantly strengthen stratification inside the lagoon in winter by creating a “capping” effect with the incoming lower density waters (Fig. 7). This does not appear to be the case in summer since the open sea waters have similar or higher density (Fig. 4), thus enhancing mixing.

Fig. 6
figure 6

Diurnal cycle of mean buoyancy frequency (N2), total heat flux (qtot) and mixed layer depth (MLD) daily mean values for both winter (blue) and summer (red) within Sharma Lagoon. Higher N2 values indicate stronger stratification in the lagoon. The cross-correlation between N2 and qtot is strong during summer (maximum r = 0.75 at lag = 3 hours), and weaker during winter (maximum r = 0.38 at lag = 2 h) due to the tidal oscillations. Similarly, the cross-correlation between the MLD and qtot is strong during summer (maximum r = 0.87 at lag = 4 h), and weaker during winter (maximum r = 0.58 at lag = 3 h) qtot peaks around 9am, and reaches the lowest values around 1 am. Data is taken from the regional Delft3D model outputs. The MLD is based on a density threshold of 0.05 kg m-3, and N2 was averaged over the entire water column.

Full size image
Fig. 7
figure 7

Mean density latitudinal transects at the center of Sharma Lagoon, as shown on the map, during (a) winter day, (b) winter night, (c) summer day, and (d) summer night. Panels from the same season share the same colourscale limits. The red contour lines represent isopycnals at the depicted density values. Data is taken from regional Delft3D model outputs. All panels were generated in Matlab 2023b (mathworks.com).

Full size image

Discussion

This study investigates the Chl-a seasonality paradox and its driving forces in Sharma, a restricted coastal lagoon in the NRS, using regionally-tuned satellite-derived Chl-a data, in situ measurements, and outputs from a validated regional hydrodynamic model. Typically, surface Chl-a is negatively correlated with SST in the NRS33,67, which is indicative of vertical mixing and supply of nutrients by deeper layers36. The phytoplankton main growth period in the NRS occurs during winter, while the lowest Chl-a is observed during summer24,31. On the contrary, Chl-a is positively correlated with SST inside Sharma, peaking during late summer, and reaching the lowest concentrations during winter throughout the lagoon (Fig. 2). The results show that, due to the lagoon’s enclosed nature, Chl-a phenology is controlled by the seasonal and diurnal variability of the hydrodynamic and atmospheric conditions, which enhance nutrients availability through increased vertical mixing from the lagoon’s seabed during summer compared to winter.

Seasonal drivers of the chlorophyll-a phenology paradox

A key factor influencing Chl-a variability in Sharma is tidal oscillations, which drive water exchange by reversing direction roughly every 6 h (Fig. 5a), which is a common phenomenon in coastal lagoons68,69,70,71,72. These oscillations not only control lagoon flushing but also enhance mixing and elevate Chl-a and nutrient levels, as observed in other coastal lagoons where inorganic sediment and organic matter can accumulate73,74,75. This could explain why even at its lowest concentration in winter, Chl-a is higher inside the lagoon than outside during the NRS main growth period, as shown in Fig. 2. However, increased water exchange in Sharma during winter, as shown by the particle release (Fig. 5b), does not coincide with increased Chl-a. Instead, it is negatively correlated with Chl-a, and thus with nutrient availability.

Previous work in Al Wajh lagoon, located in the NRS, has demonstrated that the semi-enclosed nature of these regions is crucial to their functioning and distinctive ecological behaviour35. Subject to seasonally fluctuating strong surface heat fluxes, their shallow and confined waters experience significant temperature and density extremes that diverge from the more stable temperature patterns observed in the open waters. During winter, the density of the inflowing waters is lower than the cooler ambient waters, meaning that the water entering the Al Wajh lagoon remains at the surface and strengthens stratification. In contrast, during summer, denser open sea waters flow in, weakening stratification and inducing mixing throughout the water column. This increases nutrient availability and shifts phytoplankton growth to an opposite timing compared to the outside waters. Therefore, the density difference of the two water masses interacting plays an important role in the intensity of the mixing35,76. This effect can also be observed, but to a lesser extent in Sharma. The incoming open sea waters have equal or lower density compared to the lagoonal waters during summer (Fig. 4), thus dispersing nutrients from the bottom of the lagoon, which has a higher concentration of Chl-a and nutrients (Supplementary Figure S4). This is especially true during summer daytime, when the lagoon’s surface water temperature increases rapidly (Supplementary Figure S5c). On the other hand, the density difference between the two water masses is much larger in the winter, leading to surface retention of incoming waters and further strengthening stratification. The amount of water transported between Sharma and the NRS further influences nutrient availability within the lagoon. Variations in sea surface height (SSH) within the lagoon (Fig. 5a) are directly linked to water exchanges between Sharma and the surrounding NRS waters. The observed sinusoidal signal with a daily period closely resembles the dominant tidal oscillation of the region. The seasonal differences in SSH, with lower values in winter compared to summer, are consistent with patterns observed throughout the Red Sea77. This variability is primarily driven by seasonal mean sea level changes, which are influenced by the Red Sea’s connection to the open ocean. This further enhances the effects of density difference described earlier. In winter, this translates into stronger stratification as lower density open sea waters flow inside the lagoon during flood tides. In summer, however, limited water exchange allows atmospheric forcing to play a larger role in controlling stratification strength.

Diurnal variability and its role in chlorophyll-a seasonality

In addition to seasonal variations, Sharma exhibits strong diurnal fluctuations that further shape its Chl-a dynamics. The lagoon’s shallow topography makes it highly responsive to atmospheric heat fluxes, leading to pronounced day-night temperature variations (Supplementary Figure S5). During summer nights, the weakly-stratified lagoon’s waters rapidly cool down due to the large negative qtot, mixing the entire water column of the lagoon (Figs. 6, 7), which can redistribute nutrients from the sediments. Even though air temperature is significantly lower in winter, the highest nighttime heat loss (minimum qtot) from the lagoon’s surface is usually observed from late summer to early autumn. This can be attributed to increased evaporative heat fluxes, which is the main cooling process in coastal lagoons78, during summer nights due to the dry land breeze originating from the surrounding desert79,80. Moreover, the increased water exchange dampens the effects of atmospheric heat fluxes during winter, when the warmer NRS waters (Fig. 4) enter the lagoon’s surface strengthening stratification. These findings are also supported by the cross-correlation analysis between qtot and both the MLD and N2 (Supplementary Figure S6). This complex balance between the effects of hydrodynamic and atmospheric forcings in combination with Sharma’s topography results in the observed Chl-a seasonality paradox.

Limitations of the study and future directions

One limitation of this study is the temporal mismatch between the hydrodynamic model outputs (2013–2014) and the observational datasets (2016–2020). In addition, due to the remote location of the study area, in situ measurements are only available over one calendar year from August 2017 to April 2018. While this prevents direct validation, the use of model climatologies is supported by the low interannual variability in SST and chlorophyll-a and the stable seasonal patterns of physical drivers in the region (Supplementary Fig. S1a, c). Nonetheless, this temporal offset and the lack of multiyear field observations add uncertainty to the alignment between modeled processes and observed biological signals.

Although we were able to describe the mechanisms that control the Chl-a seasonality in Sharma, our findings are subject to limitations due to the scarcity of in situ observations and ecological studies from within the lagoon. While there is an indication that Chl-a is elevated near coral reefs (Fig. 2), their effect on nutrient fluxes and phytoplankton composition in the lagoon remains unknown. Regardless of the indirect relationship between Chl-a and temperature through nutrient fluxes, temperature is not directly correlated with Chl-a in the Red Sea81. However, it is correlated with increased phytoplankton metabolic rates and shifts towards heterotrophic communities82. From the available data, we are unable to conclude how the high SST range of the lagoon (Fig. 4) affects phytoplankton communities and growth rates seasonally. Consequently, it becomes difficult to predict how the ecosystem services provided by the lagoon can be impacted in different scenarios.

To improve these limitations and further advance our understanding of the lagoon’s dynamics, additional studies are necessary to address the gaps in our data and explore the complex interactions within the ecosystem. There are several directions to continue research in Sharma Lagoon. For instance, comparison of planktonic communities between seasons could provide more insight into this paradoxical seasonality and its importance for the region. Even though occasional spring Chl-a spikes can coincide with the spring dust storms that are common in the region between March and May83,84, they have no significant impact on Chl-a in the Red Sea84. However, the effects of other extreme weather events, such as rainstorms, on phytoplankton growth inside the lagoon remains unknown, and would require targeted in situ sampling for a comprehensive analysis39. Sharma also appears to be the endpoint of three dry riverbeds (wadi) that in such events may provide some fresh water to the lagoon altering further the stratification mechanisms32. Finally, future studies could model the effects of even more extreme SST ranges, due to climate change, on phytoplankton dynamics, or the impacts of altering the lagoon’s water exchange rate.

Recommendations to NEOM policy makers

With the area’s topography being a permanent limiting factor for water exchange, any alterations, such as increasing the channel’s width or the sill depth, can change the lagoon’s dynamics drastically, as it has been reported for lagoons elsewhere85,86. The current water exchange rate is vital for nutrient suspension from the seabed, which maintains the lagoon’s high productivity during summer, when Chl-a is especially low at the surface of the NRS. The restricted nature of Sharma Lagoon also prevents the gradual homogenisation of its waters with the oligotrophic NRS waters87 and maintains its unique Chl-a seasonality and phytoplankton biomass. Any alterations to the water exchange levels can disrupt the lagoon’s unique Chl-a seasonality, potentially impacting the area’s biodiversity and the ecosystem services it currently provides6. For instance, disturbances to phytoplankton phenology can resonate through marine ecosystems, as the survival and fitness of organisms at higher levels of the food web is dependent on food availability88,89.

Due to its grand-scale development plans in the area, it is imperative to address potential impacts of the NEOM project to this unique environment, and ways to prevent them. Even though information regarding the ecological importance of Sharma Lagoon is limited, several recommendations to NEOM stakeholders can be derived from the results of this study:

  • Alterations to the lagoon’s entrance width or depth should be avoided. The rate of water exchange in the lagoon, and its flushing time, are controlling the abundance and growth timing of phytoplankton (food source for higher trophic levels). Many organisms, such as marine mammals, turtles, seabirds and fish, may rely on the highly productive waters of Sharma during summer, when the NRS surface waters reach their lowest levels of productivity. The lagoon also has the potential to be an important nursing or migratory ground for endemic or endangered species in the region.

  • On-site studies and monitoring projects should be reinforced to emphasize the lagoon’s biodiversity and ecological services, and how they extend to the broader northern part of the Red Sea basin. Due to the uneven spatial and temporal distribution of animal assemblages within coastal lagoons, implementing ecosystem-level management plans becomes crucial to safeguard habitat heterogeneity and its biodiversity, thereby ensuring the delivery of services from all organisms in the coastal zone6.

  • Any anthropogenic inflow of nutrients or pollutants into the lagoon should be avoided. Excessive concentrations of nutrients can cause eutrophication in the lagoon due to its restricted and self-sustaining nature. This could cause not only the loss of water quality, but also the disappearance of important habitats, such as macrophyte meadows90. Recovery from eutrophication is typically considered difficult, with no direct reversal even with nutrient reduction efforts12. On the other hand, brine from desalination could further increase salinity, and consequently the density of the lagoon’s waters, indirectly intensifying stratification (reducing nutrient availability), but also directly impact benthic organisms.

In conclusion, the distinctive attributes of Sharma Lagoon make a compelling case for its designation as a Nature Reserve, underscoring its significance within the northern Red Sea ecosystems and contributing to the success of the NEOM project.

Conclusion

This study reveals the driving forces behind the paradox of the Chl-a seasonality in Sharma Lagoon (summer peaking environment) compared to the Red Sea open waters (typically winter peaking), and highlights its ecological importance to the region. Tidal oscillations are the primary drivers of the lagoon’s water exchange, and have a different effect depending on the density difference between the northern Red Sea waters and the lagoonal waters. This difference is large in winter with the open waters being much warmer, and very small in summer. Therefore, water inflow via flood tides in winter strengthens stratification, while in summer it can induce mixing, especially during daytime. In addition, water exchange rate is significantly higher during winter because of the sea surface height seasonality of the northern Red Sea, further enhancing the effects of flood tides. Diurnal heat fluxes also affect vertical mixing due to the lagoon’s enclosed and shallow topography, which limits water exchange, with a clear correlation between the stratification index (N2) and total heat flux. Heat loss is higher during summer nights compared to winter due to increased evaporation from dry land breezes, which enhances vertical mixing and increases nutrient availability in the lagoon. Consequently, the current water exchange rate is vital for maintaining the Chl-a seasonality of the lagoon by the described processes. Given these findings, it is crucial to implement conservation measures that preserve the lagoon’s natural water exchange dynamics, ensuring its ecological functions remain intact alongside NEOM’s development plans.