The Dziani Dzaha Lake is a shallow tropical volcanic crater lake located on the Petite Terre Island of Mayotte (Comoros Archipelago, Indian Ocean, Figure 1). Its surface area is close to 2.36 × 105 m² and it is separated from the nearby seashore by a 220-m thick crater wall. The volcanic crater is approximately 1 km in diameter and 50–100 m high, resulting in a very restricted lake watershed. The average depth of the Dziani Dzaha water column is about 3 m with a narrow depression reaching 18 m depth, probably related to the phreatomagmatic eruption at the origin of this lake (i.e., between 9 and 4 ka, Zinke et al., 2003).

The Dziani Dzaha water column has a combination of physical, chemical and biological features that are atypical for a modern lacustrine system, which have been previously recently documented and are summarized here. Its waters probably originated from seawater and are now characterized by high salinity (i.e., ranging from 34 to 71 psu), strong alkalinity (i.e., ranging from 0.1 to 0.2 M), elevated pH value (i.e., comprised between 9 and 9.5), and very high primary productivity (i.e., 8 gC m⁻² day⁻¹ (Leboulanger et al., 2017; Sarazin et al., 2020). The lake ecosystem is dominated by microorganisms with a dense and perennial bloom of two photosynthetic species [i.e., Picocystis salinarum (Chlorophyta) and Arthrospira fusiformis (Cyanobacteria), Bernard et al., 2019]. The water column is permanently anoxic below ca. 1.5 m depth, with the anoxic water being periodically euxinic (i.e., presence of H₂S and HS⁻) (e.g., Sarazin et al., 2020). The SO₄ ²⁻ content is relatively low compared to seawater (<3 mM, Sarazin et al., 2020). Organic matter in the anoxic waters is remineralized through both sulfate reduction, as evidenced by the periodic total consumption of SO₄ ²⁻, when the lake becomes euxinic, and methanogenesis, as indicated by the high concentration of CH₄ dissolved in the water column that degasses into the atmosphere (Cadeau et al., 2020). Finally, the water column is nitrate-free while the reduced dissolved nitrogen species (i.e., NH₄ ⁺/NH₃) accumulate in the deeper part of the Dziani Dzaha lake (Bernard et al., 2019).

The physical and chemical structure of the Dziani Dzaha water column as well as the biological organism’s distribution within it (mostly dominated by cyanobacteria), are closely related to the seasonal variations (Hugoni et al., 2018; Bernard et al., 2019; Sarazin et al., 2020). The water column is periodically stratified with a halocline at about 2 m depth due a decrease in surface water salinity driven by rainy season precipitation (Figure 2). A permanent chemocline was present at ca. 14 m depth in the depression. During the stratified period (Figures 2, 3), salinity above the halocline was within 35–45 psu, alkalinity was close to 0.1 M, pH was close to 9.5, SO₄ ²⁻ was close to 3 mM, and only traces of NH₄ ⁺/NH₃ were observed (<0.03 mM). Below the halocline the salinity and alkalinity increased up to 70 psu and 0.2 M, respectively, pH decreased to a value close to 9, no sulfates were observed, and reduced species accumulated. It is worth noting here while that HS⁻/H₂S concentrations and CH₄ concentrations increased sharply at 2 m depth, i.e., across the halocline, up to 6 and 2 mM, respectively, and remained relatively constant down to the bottom, NH₄ ⁺/NH₃ concentration increased in two steps, one up to 0.4 mM at 3 m depth (i.e., 1 m below the halocline) and another one up to 6 mM below the deep chemocline. Above the 2 m-depth halocline, the diversity was dominated by photosynthetic microorganisms, while below it, it was dominated by a dense and diverse population of archaea and heterotrophic bacteria (Hugoni et al., 2018). During the non-stratified period most of the physical, chemical and biological parameters were constant with depth down to the deep chemocline at 14 m depth, except for the dissolved oxygen that was only present down to a maximum of about 1 m depth (Figures 2, 3). The salinity was close to 65 psu, the alkalinity close to 0.14 M, the pH close to 9.2, the SO₄ ²⁻ content close to 3 mM, NH₄ ⁺/NH₃ content was lower than 0.03 mM, and the planktonic biomass was dominated by photosynthetic microorganisms. The biological and physico-chemical parameters of the waters underlying the deep chemocline during non-stratified periods resembled strongly to stratified periods, albeit with even higher concentrations of reduced species (Sarazin et al., 2020).

As shown by the above site description, many physical, chemical and biological parameters have been published previously [pH, T, Salinity, O₂, NH₄ ⁺/NH₃ in Sarazin et al., 2020; SO₄ ²⁻, H₂S/HS⁻ in Sarazin et al., 2020; Chlorophyll a before 2012 in Leboulanger et al., 2017; carbon biomass associated to photosynthetic cell abundance in Bernard et al., 2019]. In this section we will thus describe only the sampling and the analytical methods used to acquire the new data presented here; chlorophyll a (for samples taken since 2012) and C/N and nitrogen isotope compositions.

Sampling were performed at the end of the rainy season (April) in 2012, 2014 and 2015, and at the end of the dry season (October-November) in 2011, 2014 and 2015. The “CLB” and “18 m” names refer to the water column stations where samples were collected, throughout 4.5 m depth and 18 m depth respectively, as shown on the Figure 1. Water samples were collected using a horizontal 1.2 L Niskin bottles along a vertical profile at the CLB and 18 m stations (Figure 1) every 0.25–0.50 m depth for the first 4.5 m of water column and either every 50 cm across the deep chemocline (between 13 and 15 m depth), or every 1–2 m between 5 and 13 m depth where no significant variation was observed on the profiles from the multiparameter probe. An aliquot of sampled water for each depth was filtered onto precombusted (450°C–4 h) 47 mm Whatman GF/F glass fibre filters (0.7 µm porosity) to recover suspended particulate matter for chlorophyll a, C/N and δ¹⁵N analyses. Several sediment cores serially numbered from the first survey were collected at different depths (Figure 1), freeze-dried, rinsed with deionized water, centrifuged three times, and finally freeze-dried and crushed down to <80 μm.

The concentration of chlorophyll a was analyzed after a two-steps extraction, the extract being then filtered and the filtrate analyzed spectrophotometrically at 400–750 nm (Leboulanger et al., 2017). The concentration of chlorophyll a was calculated according to Camacho et al. (2009). Nitrogen isotope and C/N measurements were performed on the same samples as the carbon isotope analysis reported in Cadeau et al. (2020) (i.e., suspended particles and sediments), but during a different run dedicated to the δ¹⁵N analyses and on bulk samples to avoid alterations of δ¹⁵N signal during the acid decarbonatation required for δ¹³C_org analyses (Brodie et al., 2011). Nitrogen isotope measurements of NH₄ ⁺/NH₃ were performed using the ammonium diffusion method (Sebilo et al., 2004), which is based on the conversion of NH₄ ⁺ to NH₃ by pH adjustments and subsequent trapping of the released NH₃ onto glass fibre filter. Nitrogen species retained on glass fibre filter (i.e., suspended particles, dissolved nitrogen trapped onto (NH₄)₂SO₄ form) and sediment samples were analyzed using a Flash EA1112 elemental analyser coupled to a Thermo Finnigan Delta^plus XP mass spectrometer via a Conflo IV interface (Thermo Fisher Scientific, Waltham, MA, United States). Nitrogen isotope ratios were calibrated against four internal standards of organic-rich soil or sediment included in the sample sequence and previously calibrated against the certified IAEA-N1 and IAEA-N2 international standards. The nitrogen isotopic signatures are expressed as ‰ relative to air with a reproducibility of ±0.3‰ (1σ). Routine replicate measurements on standards had internal deviations of ±0.15‰ (2σ) for δ¹⁵N values, and less than 4% of the measured value for the nitrogen content.

Two types of external N-sources can be envisaged: an influx of dissolved nitrogen-rich fluids and/or a detrital input of terrestrial organic matter. An influx of dissolved nitrogen-rich fluid (from seawater or hydrothermal fluids) is unlikely based on physical and chemical measurements in the water column, no on-going water infiltrations were identified and, from a water balance point of view, this lake seems to be endoreic (Sarazin et al., 2020). Moreover, seawater entrance would only marginally affect the dissolved nitrogen budget and its δ¹⁵N because the concentration of dissolved nitrate in the Indian Ocean (i.e., 5–15 μM up to 200 m depth, NOAA, 2017) are quite low compared to the concentration of dissolved reduced nitrogen species in euxinic waters of the Dziani Dzaha (i.e., mM range Figure 2) and the nitrate δ¹⁵N in the Indian Ocean (∼+6‰) is very close to that of the lake (∼+7 ‰). For an external dissolved nitrogen source to contribute significantly to the lake NH₄ ⁺/NH₃ budget, it would thus need to be sourced from NH₄ ⁺/NH₃-rich hydrothermal fluids. However, that would still not explain the fact that in the euxinic and NH₄ ⁺/NH₃-rich waters the δ¹⁵N value of suspended particles and sediments are more positive than the dissolved NH₄ ⁺/NH₃.

The supply of detrital organic matter from the surrounding catchment is also very unlikely to be responsible for the higher δ¹⁵N_SPM values of the euxinic waters and δ¹⁵N_sed values. The carbon and nitrogen contents and isotopic compositions of seven kinds of higher plants and soil sampled in the lake watershed (Table 1) exhibited δ¹⁵N values that were lower than the lake primary producers (<7‰), except for the rush sample with a δ¹⁵N value close to 9 ‰. This high value is best explained by the fact that rushes grow in the shallow waters of the lakeshore and are probably assimilating ¹⁵N-enriched nitrogen from the lake sediment pore waters. In addition, the measured plants δ¹³C and C/N values are significantly lower and higher, respectively, than those of suspended particulate matter and surface sediments (Figures 3, 4 and Table 1). A contribution of detrital organic matter to the suspended particulate matter or sediment would thus decrease their δ¹⁵N, rather than increase it. It would also decrease their δ¹³C_org and increase their C/N, as observed in some samples of the shallow cores (i.e., C2 and C9, Figure 4), which we accordingly explain by a contribution of detrital organic matter.

Isotopic alteration due to organic matter mineralisation in modern settings was relatively well investigated (e.g., Lehmann et al., 2002; Robinson et al., 2012; Tesdal et al., 2013; Katsev and Crowe, 2015). The extent of organic matter degradation as well as the early diagenesis processes, and subsequently the extent of isotopic alteration of the primary nitrogen isotope signature, appear to depend on the oxygen exposure time, water depth and organic matter availability (e.g., Lehmann et al., 2002; Thunell et al., 2004; Robinson et al., 2012). So far, increases in both δ¹⁵N_SPM and δ¹⁵N_sed of the order of a few per mil have only been reported in sinking particles in the deep ocean and have been interpreted as resulting from extensive organic matter remineralisation under oxic conditions (e.g., Lehmann et al., 2002; Gaye et al., 2009; Mobius et al., 2010; Mobius et al., 2011). It seems unlikely however that this process is applicable to the shallow and anoxic Dziani Dzaha for two reasons. First, ¹⁵N enrichment in sinking particles and in the surface sediments through organic matter mineralisation is usually associated with an increase in C/N ratio, as nitrogen is preferentially lost over carbon (Lehmann et al., 2002; Fenchel et al., 2012). In the Dziani Dzaha, as shown in Figure 4, the increase in δ¹⁵N_SPM is not associated with an increase in the C/N ratio but rather decreased with depth from approximately 7 to 5. Second, to our knowledge, such an isotopic increase with depth has never been reported so far in lacustrine systems (e.g., Lehmann et al., 2004; Menzel et al., 2013; McLauchlan et al., 2013) or in modern anoxic marine environments (e.g., Thunell et al., 2004; Fulton et al., 2012). In these systems, when δ¹⁵N-depths profiles were available, increases in the δ¹⁵N of dissolved N species or suspended particles have been reported but they are either spatially limited to the vicinity of the chemocline (e.g., Cariaco Basin, Thunell et al., 2004; Black Sea, Fry et al., 1991; Lugano Lake, Wenk et al., 2014) or temporally limited to water column mixing events (e.g., Kinneret Lake, Hadas et al., 2009). They have been interpreted as resulting from denitrification and/or anammox together with assimilation of the enriched NO₃ ⁻ and/or NH₄ ⁺ by chemoautotrophic organisms (Deutsch et al., 2001; Voss et al., 2001; Sigman et al., 2003; Thunell et al., 2004). These isotope signatures are however not transferred to the sinking particulate matter or the surface sediments, which both record the primary producer δ¹⁵N values (Thunell et al., 2004).

The decrease in C/N ratio within euxinic waters coincides with a strong change in biological diversity (Hugoni et al., 2018). In terms of abundance, the biological communities are dominated by two primary producers (the picoeukaryote Picocystis salinarum and the cyanobacterium Arthrospira fusiformis) above the halocline at 2 m depth during stratified periods, or above the deep chemocline at 14 m depth during non-stratified periods, and by bacteria and archaea in euxinic waters below the halocline or deep chemocline (Leboulanger et al., 2017; Hugoni et al., 2018). The C/N value close to 7 in surface waters is consistent with the predominantly cyanobacterial biomass (Godfrey and Glass, 2011). Indeed, previous work on continuous cultures of Arthrospira fusiformis from Lake Chitu (Ethiopia) reported that the C/N ratio was stable in this species, around 6.25 (Kebede and Ahlgren 1996). The generally lower C/N values in the euxinic waters, sometimes approaching 5, are consistent with a predominantly bacterial or archaeal biomass (Muller, 1977; They et al., 2017). From there, it is tempting to imagine that both the decrease in C/N and the increase in δ¹⁵N_SPM observed in the euxinic waters could be linked to a higher proportion of bacterial and archaea biomass, which would be characterized by a more positive δ¹⁵N_SPM values than the cyanobacterial biomass. However, even if in euxinic waters bacteria and archaea are predominant in terms of number of individuals compared to cyanobacteria (Hugoni et al., 2018), biomass estimations from cells abundance measurements and biovolumes show that cyanobacteria remain by far the main constituent of the biomass (Figure 3F) and are responsible of most of the primary production in the lake (Leboulanger et al., 2017). Therefore, in spite of their coincidence with a strong biodiversity change, the combined C/N decrease and δ¹⁵N_SPM increase in euxinic waters are not directly related to it.

In this section, we investigate the possibility that the combined C/N decrease and δ¹⁵N_SPM increase in the euxinic waters reflect NH₄ ⁺ assimilation and storage by cyanobacteria (Figure 3). As previously shown in modern alkaline anoxic settings (e.g., Lonar Lake, Menzel et al., 2013), pH controls the dissolved reduced nitrogen speciation through the reaction of NH₄ ⁺ dissociation to NH₃ (i.e., Eq. 1 below, Li et al., 2012): NH 4 + = NH 3 + H + (1)

At isotope equilibrium the isotope exchange between NH₄ ⁺ and NH₃ is associated with a strong isotopic fractionation (i.e., of 45 ‰ at 23°C, Li et al., 2012), forming ¹⁵N-depleted NH₃ and ¹⁵N-enriched NH₄ ⁺. Considering an average temperature of 30°C in the NH₄ ⁺-rich part of the Dziani Dzaha water column (Sarazin et al., 2020; Figure 2), and based on the equations proposed in Li et al. (2012), the pKa of acid-base couple (Eq. 1) is of 9.09 in the Dziani Dzaha conditions. Considering this and the minimum and maximum pH values observed in the NH₄ ⁺-rich and euxinic waters of the Dziani Dzaha (i.e., close to 9 in April 2012 and 9.5 in April 2014), between 45 and 72% of the NH₄ ⁺ is dissociated in NH₃, enriching NH₄ ⁺ in ¹⁵N by about 20‰ and 32‰ , respectively at pH 9 and 9.5, compared to the total dissolved inorganic nitrogen. These values represent a qualitative estimation of the dissolved NH₄ ⁺ isotopic signatures range in equilibrium with ammonia, which strongly depends on variations of both pH values and total dissolved inorganic nitrogen δ¹⁵N.

Arthrospira fusiformis, the main cyanobacteria species that constitute most of the biomass in the Dziani Dzaha, is known to harbour aerotopes (gas vacuoles) that enhance cell buoyancy, and cyanophycin-rich granules increasing nitrogen intracellular storage capacity (Cellamare et al., 2018). These characteristics are consistent with a vertical migration process where cells increase in density during photosynthesis in lit layer, which makes them sink in the NH₄ ⁺/NH₃-rich water where they store nitrogen as cyanophycin or in phycobilisomes. When they reach aphotic waters, photosynthesis stops and respiration reduces their cell density, allowing them to move upward (Carey et al., 2012). However, when the lake is stratified and its deep waters are euxinic, this process could eventually lead to a death trap due the combination of the sulfide toxicity. The fact that the ¹⁵N-enrichment and C/N decrease of suspended particles is limited to the euxinic waters at times of stratification could thus reflect the assimilation of this ¹⁵N-enriched NH₄ ⁺ by cyanobacteria, and their entrapment below the halocline. This would also be compatible with the delayed increase in NH₃/NH₄ ⁺ concentration in the first meter below the halocline (compared to CH₄ and H₂S⁺/H S⁻) where cyanobacteria might still be active enough to assimilate NH₄ ⁺ efficiently. In addition, δ¹⁵N_SPM in the euxinic and NH₃/NH₄ ⁺-rich waters shows an average value close to 12 ‰ (Figure 3), which represents an isotopic enrichment of only about 5 ‰ compared to the δ¹⁵N_SPM observed in surface water despite the estimated δ¹⁵N values of dissolved NH₄ ⁺ in the NH₃/NH₄ ⁺-rich waters comprised between 20‰ and 32 ‰. This is most likely related to the fact that NH₄ ⁺ assimilation favours ¹⁴N uptake (Pennock et al., 1996; Vo et al., 2013) and that only up to approximately 29% of the total nitrogen in the ¹⁵N-enriched cyanobacterial biomass originated from assimilation of ¹⁵N-enriched NH₄ ⁺ according to the change in C/N ratio. Finally, δ¹⁵N_SPM values in the euxinic waters vary by only 1 ‰between 2012 and 2014 in spite of significant the variations of pH and NH₄ ⁺/NH₃ dissolved concentrations which impact of the δ¹⁵N of NH₄ ⁺ may have balanced each other. Unfortunately, this cannot be assessed further here as in the absence of measurement of the total dissolved inorganic nitrogen δ¹⁵N in 2012 and 2014.

Ammonium assimilation and storage by cyanobacteria, beyond accounting for the δ¹⁵N_SPM increase in the euxinic waters, also explains the fact that the sediment δ¹⁵N is biased towards the δ¹⁵N of ¹⁵N-enriched cyanobacteria, even at shallow depths where the water column remains oxic. The high buoyancy of cyanobacteria that allows them to move vertically in the water column when they are alive and active, also prevents them from sinking quickly to the surface sediment when they are dead. This is consistent with the fact that although the Dziani Dzaha is a highly productive and shallow lacustrine system exhibiting highly favourable conditions for the preservation of organic matter (e.g., Leboulanger et al., 2017), the proportion of the net primary production preserved in the sediments is only of 2.9% (Cadeau et al., 2020), e.g., only slightly higher than the maximum value estimated in the stratified and euxinic Black Sea basin (i.e., 0.5–1.8%, Karl and Knauer., 1991).

As shown by the chlorophyll a and biomass data (Figures 3F,G), when the water column is sulfide-free (i.e., up to 2 or 14 m depth according to the stratified or non-stratified period, respectively), most cyanobacteria are active and the amount of biomass is stable with depth. Their high buoyancy, together with the wind agitation of the surface waters that propagates down to the lake bottom when it is non-stratified (as shown by the similarity between surface and bottom water temperatures and their high amplitude temporal changes, Sarazin et al., 2020), probably limits the export of organic matter to the sediment. Preferential organic matter export would thus probably occur from the bottom waters when the lake is density stratified and the mixing dynamics of bottom waters strongly reduced (as shown by the very dissimilar surface and bottom water temperatures with constant temperatures in bottom waters unaffected by the high frequency changes in surface temperatures, Sarazin et al., 2020), allowing the dead or weakened cyanobacteria to sink in the water column and accumulate in the sediment.

In the shallow parts of the lake (i.e., up to 2 m depth), cyanobacteria remain active all year long, even when they settle down onto the water/sediment interface as suggested by observation of a green surface layer in the flocs resting on the top of the shallow sediment cores. Since the water column is never euxinic and NH₄ ⁺-rich at shallow depth, we would not expect shallow surface sediments to be ¹⁵N-enriched, and yet they are. It is most probable that the interstitial waters of the colloidal floc and surface sediments are probably anoxic and NH₃/NH₄ ⁺-rich and that ¹⁵N-enriched NH₄ ⁺ is assimilated by cyanobacteria when they reach the floc, explaining the increase observed in the sedimentary δ¹⁵N values even at shallow sites.

Finally, NH₄ ⁺ incorporation into clay minerals could also contribute to modifying the sediment δ¹⁵N towards more positive values (Muller, 1977). Indeed, the shallowest sediment cores (e.g., C10) contain half the amount of organic matter than those collected at 4 and 18 m depths (i.e., C4 and C6 respectively) (Milesi et al., 2019), and a higher clay mineral (e.g., saponite) content (Milesi et al., 2019). Pore waters pH being basic within the first 30 cm studied here (Milesi et al., 2019; Milesi et al., 2020), NH₄ ⁺ is also ¹⁵N-enriched by isotope equilibrium with NH₃. An incorporation of ¹⁵N-enriched ammonium into clay minerals by substituting for K⁺ could make the sediments isotopically heavier in all sediment cores (Muller, 1977; Stueken et al., 2019), especially into the core C10 presenting a higher clay mineral content (Milesi et al., 2019). Still, although this process cannot be ignored, it is probably not significant compared to the nitrogen input from primary producers (at the surface or deep in the water column), because organic matter content in all sediment cores remains significant and the proportion of ammonium potentially incorporated into the clay mineral fraction would represent only a small fraction of the overall clay mineral content.

Nitrogen isotopic signatures recorded in the Dziani Dzaha surface sediments would then result from a combination of 1) the cyanobacterial initial biomass with its original δ¹⁵N acquired from surface waters, 2) their nitrogen storage compartments with more positive δ¹⁵N values acquired through ¹⁵N-enriched ammonium assimilation in the euxinic part of the water column, in the floc or in the sediments interstitial waters, and possibly in a lesser extent 3) some abiotic incorporation of ¹⁵N-enriched ammonium into authigenic silicates.

As previously suggested, albeit for a different reason, these results confirm that basic pH conditions have a strong control on the sedimentary nitrogen isotopic signature, which hence represents an useful tool to track such geological settings over times (e.g., Stueken et al., 2019). However to date, the mechanisms envisioned to explain the overall ¹⁵N enrichment of sediment in modern or past basic lakes are NH₄ ⁺ isotope equilibrium with NH₃ coupled to NH₃ volatilization into the atmosphere and/or denitrification or anammox (e.g., Menzel et al., 2013; Stueken et al., 2015; Stueken et al., 2019; Stueken et al., 2020). Our results demonstrate that additional processes to those commonly used to explain these high values can contribute to increase the sedimentary δ¹⁵N. Indeed, in the Dziani Dzaha Lake a significant isotopic ¹⁵N enrichment is observed at depths within the water column and/or directly within surface sediments. We propose here that it is mainly due to ¹⁵N-enriched NH₄ ⁺ assimilation by cyanobacteria either in the deep-water column or in surface sediments. The most important aspect of this result is that the primary producer δ¹⁵N values observed in the surface water of the Dziani Dzaha Lake are not well recorded in surface sediments. This represents a strong contradiction with one of the main assumptions underlying the use of nitrogen isotope compositions as a tool for the reconstruction of the past nitrogen cycle: i.e., that sediment δ¹⁵N values record those acquired by primary producers, themselves recording the δ¹⁵N value of the nitrogen species assimilated in the photic zone (e.g., Ader et al., 2016). This hypothesis is well established for all previously studied modern continental platforms or anoxic basins, which are usually characterised by high export production, low oxygen content and high organic matter preservation (e.g., Pride et al., 1999; Altabet et al., 1999; Emmer and Thunell, 2000; Thunell and Kepple, 2004; Thunell et al., 2004). But our results show that it may not apply to ammonium replete environments or paleo-environments of the Precambrian, which are thought to have been inhabited by dominantly microbial ecosystems (e.g., Butterfield., 2015), in which cyanobacteria were responsible for the bulk of the primary productivity.

Depending on the water pH, the ability of cyanobacteria to assimilate ammonium when sinking in the water column may have various impacts on their initial δ¹⁵N. Although under basic settings, the large isotopic fractionation associated with NH₄ ⁺ dissociation into NH₃ leads to ¹⁵N enrichment through the assimilation of ¹⁵N-enriched NH₄ ⁺, NH₄ ⁺ assimilation is associated with an isotopic fractionation that favours ¹⁴N and is proportional to the NH₄ ⁺ concentration (Pennock et al., 1996; Vo et al., 2013). Hence, under the neutral condition (i.e., without NH₄ ⁺ dissociation into NH₃), NH₄ ⁺ assimilation could modify the cyanobacteria δ¹⁵N towards more negative isotopic values, leading, if ignored, to an underestimation of the assimilated nitrogen δ¹⁵N in such settings. For instance, in the Dziani Dzaha Lake the total dissolved nitrogen δ¹⁵N in the water column (i.e., NH₄ ⁺ and NH₃) exhibits an isotopic value close to 7‰ . If the pH values in the Dziani Dzaha were neutral, NH₄ ⁺ assimilation could have modified the δ¹⁵N of primary producers towards negative values. These findings require further consideration about the reliability of paleoenvironmental evidences provided by the δ¹⁵N in such stratified and NH₄ ⁺-rich settings.