Biogeochemical explanations for the world's most phosphate-rich lake, an origin-of-life analog

Evaporative lakes are promising origin-of-life settings because they can concentrate major biogenic elements and key reactive species (e.g., cyanide, sulfite/sulfide, phosphate). Soda lakes are alkaline, saline waters dominated by Na+, (bi)carbonate, and Cl−, and often feature elevated phosphate, potentially solving the “phosphate problem”: environmental phosphate is typically ~10^-6 M, whereas prebiotic phosphorylation of nucleosides requires ~0.1–1 M phosphate in experiments. Nucleoside phosphorylation is a condensation reaction likely driven by wet-dry cycles in evaporative lakes; high phosphate availability during dehydration is crucial to yield nucleotide concentrations sufficient for non-enzymatic RNA primer extension. In natural waters, phosphate is normally limited by precipitation as calcium phosphate (apatite) and biological uptake with burial. In soda lakes, it has been hypothesized that high carbonate causes early precipitation of CaCO3, lowering dissolved Ca2+ and preventing apatite saturation, allowing phosphate to remain in solution at ≥1 mM. However, direct demonstrations from sediments (identifying Ca phases) and field observations linking phosphate to evaporation are scarce, and the role of biological sinks (especially biological N2-fixation that could relieve N-limitation and stimulate P uptake) has not been well constrained in modern soda lakes. This study examines adjacent Last Chance Lake (LCL) and Goodenough Lake (GL) in British Columbia to test whether reduced geochemical and biological P sinks allow extreme phosphate accumulation and to quantify nitrogen cycling, particularly N2-fixation, that could modulate biological P demand.

Prior work proposed carbonate-rich soda lakes as solutions to the phosphate problem by suppressing apatite precipitation through low Ca2+ caused by carbonate mineral formation, supported by correlations between dissolved inorganic carbon and phosphate in soda lakes. Typical environmental phosphate sinks include calcium phosphate precipitation and biological uptake with burial. In lakes, N:P ratios are variable and biological N2-fixation (BNF) can, in principle, balance N relative to P over time, but empirical evidence is mixed. Soda lakes exhibit varied productivity limitations (light, nitrogen, phosphorus), and NH3 volatilization at high pH can deplete fixed N. BNF measurements in soda lakes are rare; water-column BNF was undetectable in Mono Lake despite elevated phosphate, though benthic mats showed BNF in Mono Lake and in hypersaline Russian soda lakes. No prior BNF rates were reported from lakes with ≥1 mM phosphate, leaving uncertainty about biological control on phosphate in such environments. Geological context indicates soda lakes commonly develop on basaltic terrains, with closed-basin hydrology favoring solute concentration. These backgrounds motivated direct sediment mineralogy tests, evaporation proxies (e.g., Cl−), and in situ biogeochemical measurements in LCL and GL.

Field campaigns at LCL and GL occurred in November 2021, June 2022, and September 2022, sampling spatially distributed stations across each lake. Water sampling employed Niskin bottles or clean tubing to avoid sediment disturbance; samples were filtered (0.2 µm) for dissolved analytes. Sediment cores (10-cm diameter PVC) were retrieved, sliced by depth, and porewaters extracted using rhizons or centrifugation followed by 0.2-µm filtration. Evaporites were collected from shores and lake surfaces, stored frozen. In situ measurements included dissolved O2, pH, and turbidity (YSI ProDSS). Total salinity was measured gravimetrically by drying filtered water; density by weighing a known volume at 22 °C. Dissolved ions were analyzed by ion chromatography (SO4^2−, Cl−) and ICP-MS (Ca, Fe, K, Mg, Mo, Na, P, S). Detection limits (dilution dependent) included: Ca 0.0014–0.84 mM; Fe 0.00016–0.097 mM; Mg 0.0045–2.7 mM; P 0.0015–0.87 mM; Mo 0.00019–0.11 mM. DIC was determined on a Shimadzu TOC/TN-V CSH by difference pre/post acidification and sparging. Phosphate was measured via the molybdate blue spectrophotometric method (≤1% precision). DIN species (NH4+ + NH3, NO2−, NO3−) were quantified by autoanalyzer with low-µM detection limits. Particulate analyses (POC, PON, POP, chlorophyll a) used vacuum filtration onto pre-combusted glass-fiber filters; Chl a was assessed fluorometrically after acetone extraction. Particulate N and δ15N were measured by EA-MS, standardized to air-N2 using international and in-house standards with blank and linearity corrections. Bulk sediments were freeze-dried, sieved (<2 mm), ground (≤125 µm), and processed for EA-MS or acid digestion (EPA-3050B) prior to ICP-MS total P analysis. Evaporite elemental composition was determined by dissolving salt in deionized water followed by ICP-MS. XRD: Clay-reduced sediment and evaporite samples were prepared and analyzed on a Bruker D8 with Cu anode and 2D detector; phases identified with Bruker EVA and ICCD PDF-4+; Rietveld refinements were performed in GSAS-2. Fertilization experiment: On 13 Sep 2022, lake waters were incubated in glass bottles (200 mL, 65 mL air headspace) under natural light in a water bath for ~48 h with three treatments (n=3 time points per lake): (1) +phosphate (Na3PO4 to +100 µM P), (2) +ammonium+nitrate ((NH4)2SO4 + NaNO3, 1:1 N molar, +100 µM N), (3) control. Time-series Chl a was measured at ~0, 1, and 2 days. 15N2-labeling experiments: Water and slurries (surface microbial mats from GL and surface sediments from LCL) were incubated with 15N-enriched N2 in June and September 2022 (two stations per lake; duplicates plus natural abundance references). 15N2-enriched water was prepared by He-bubbling, degassing, and adding pure 15N2 gas to achieve 1–10% 15N2; equilibration by 21 h shaking. Incubations ran ~48 h under natural light; 15N of dissolved N2 was measured from HgCl2-preserved headspace-equilibrated exetainers. Post-incubation, water samples were filtered; slurries were freeze-dried and homogenized. Uptake rates were computed from atom% 15N enrichment in particulates or slurries (Eq. 2), with detection limits set as three times the SD of reference measurements. Slurry volumetric rates were converted using measured wet densities, and GL mat rates integrated over the top 1 cm to estimate areal rates. Nutrient ratios were log-transformed for presentation.

• Extreme phosphate in LCL and evaporative control: Across seasons and stations, phosphate in LCL scaled linearly with chloride (R^2 ≈ 0.93), a conservative salinity proxy, indicating evaporative concentration. Peak concentrations in September 2022 concentrated brines were tdP = 37 ± 9.0 mM, PO4^3− = 37 ± 0.37 mM, Cl− = 2.5 ± 0.0041 M. Spring inputs to LCL were low (tdP 21 ± 15 µM; PO4^3− 5.7 ± 0.057 µM; Cl− 4.5 ± 3.2 mM). Mean PO4^3−:Cl− ratios: LCL water 0.02 ± 0.006; LCL sediment porewater 0.03 ± 0.009; GL water 0.002 ± 0.001; GL sediment 0.006 ± 0.005, showing LCL’s high phosphate is not solely due to greater evaporation than GL.
• Calcium removal via carbonate, not apatite: Cl− correlated with Na+ and K+ but not with Ca2+ or Mg2+, implying early precipitation of divalent cations. Bulk sediment Ca greatly exceeded dissolved Ca, particularly in LCL, indicating early Ca removal to sediments. XRD showed LCL sediments dominated by plagioclase (anorthite 38.5 ± 13.4%, albite 31.5 ± 20.3%) and dolomites (22.3 ± 8.21%) as the major non-clay authigenic Ca phase. Evaporites lacked authigenic Ca phases; LCL evaporites were dominated by sodium (bi)carbonates, chiefly thermonatrite (66.1 ± 20.5%). P- or Fe-bearing evaporite minerals were absent by XRD; ICP-MS of LCL evaporites confirmed low P (0.13 ± 0.08%), Fe (0.11 ± 0.19%), and Ca (0.10 ± 0.17%). These lines of evidence support suppression of apatite precipitation by early carbonate formation and Ca depletion, enabling phosphate accumulation.
• Nitrogen limitation and weak biological P sink in LCL: DIN was orders of magnitude lower than phosphate in LCL; DIN:PO4^3− ratios were far below Redfield, whereas GL’s ratios approximated Redfield. Particulate N:P ratios were near-Redfield in GL but much lower in LCL. Fertilization experiments showed N addition doubled Chl a in LCL within 48 h, while N addition did not stimulate growth in GL, indicating N-limitation in LCL.
• N2-fixation rates: Water-column BNF rates were low-to-moderate and similar between lakes: GL June 24.0 ± 11.7 nmol N L−1 d−1; GL Sep 0.2 nmol N L−1 d−1 (n=1); LCL June 24.4 ± 10.2 nmol N L−1 d−1; LCL Sep 9.7 ± 1.7 nmol N L−1 d−1. Slurry incubations revealed hotspots in GL mats: GL microbial mat surface slurries fixed 1674.3 ± 263.9 nmol N g_dry−slurry−1 d−1 (≈115.3 µmol N L−1 d−1; ≈48.1 µmol N m−2 h−1 for top 1 cm). LCL sediment surface slurries fixed 2.7 ± 1.4 nmol N g_dry−slurry−1 d−1 (≈0.8 µmol N L−1 d−1). δ15N_PON values supported these patterns: LCL biomass was 15N-enriched (6.9–16.4‰) consistent with NH3 volatilization at high pH; GL surface mats approached low δ15N indicative of active BNF. Estimated GL mat BNF could significantly draw down P by creating biological P demand comparable in magnitude to the lake’s P inventory (~98 µmol P cm−2), whereas LCL’s very low BNF is negligible relative to its mM phosphate pool.
• Mechanism for low BNF in LCL: Trace metals (Fe, Mo) were not lower in LCL than GL, arguing against trace metal limitation. High salinity likely suppresses nitrogenase synthesis and activity, reducing BNF and thus biological P demand.
• Environmental context: Both lakes are alkaline (pH ≥ 9.9) and oxic, with LCL showing higher, more variable salinity (up to 462 g/L) and turbidity. LCL lacks extensive cyanobacterial mats (only thin, nearshore mats of photoautotrophic eukaryotes), whereas GL has thick perennial photo-diazotrophic cyanobacterial mats (~10 cm).

The study demonstrates that in carbonate-rich soda lakes, early precipitation of Ca as carbonates (e.g., dolomite) depletes dissolved Ca2+, preventing apatite saturation and geochemical P removal. In Last Chance Lake, this geochemical context, combined with severe N-limitation and suppressed N2-fixation at high salinity, minimizes both abiotic (apatite) and biotic (BNF-driven productivity) P sinks, allowing evaporative processes to concentrate phosphate linearly with chloride to extreme levels (tens of mM). In contrast, Goodenough Lake’s robust benthic photo-diazotrophic mats support high areal BNF rates that generate biological P demand sufficient to maintain lower phosphate and decouple phosphate from conservative evaporation signals. These findings address the long-standing question of how natural environments could accumulate sufficient phosphate for prebiotic chemistry. The carbonate control on Ca2+ and the suppression of biological P sinks in highly saline, alkaline settings provide mechanistic support for soda lakes as plausible Hadean origin-of-life environments. Under prebiotic, potentially more anoxic conditions with lower ferric iron availability (reducing P adsorption to Fe oxides) and absent BNF, phosphate concentrations may have been even higher. Enhanced P inputs on early Earth from more CO2-driven weathering of igneous apatite and meteoritic schreibersite, with photochemical oxidation to phosphate, would further augment phosphate availability. The results therefore strengthen the case that basalt-hosted, evaporative soda lakes on early Earth could have sustained phosphate levels adequate for nucleotide synthesis and other prebiotic pathways.

By combining geochemical, mineralogical, and biogeochemical measurements, the study provides direct field evidence that carbonate-driven Ca depletion suppresses apatite precipitation, enabling extreme phosphate accumulation in a natural soda lake (LCL; up to ~37 mM). Simultaneously, low N2-fixation rates—likely repressed by high salinity—limit biological productivity and phosphate uptake, distinguishing LCL from nearby GL where benthic cyanobacterial mats drive high BNF and greater biological P demand. These mechanisms together explain why LCL attains the world’s highest natural dissolved phosphate and position soda lakes as strong analogs for prebiotic environments with abundant phosphate. Future work should quantify long-term organic P burial fluxes, sediment–water ammonium exchanges, and seasonal/spatial variability of BNF and productivity; further trace-metal clean studies and expanded anoxic simulations could refine controls on BNF and P cycling. Investigations of prebiotic-relevant inputs (e.g., schreibersite-derived reduced P) and their fates in calcium-poor, carbonate-rich brines will help constrain phosphate availability on early Earth and other worlds.

• Trace-metal protocols: Metal concentration data were not obtained under strictly trace-metal–free conditions, complicating definitive assessment of Fe or Mo limitation.
• Temporal and replication limits: BNF rates were measured during two seasons (June and September) with limited replication in some cases (e.g., GL September water-column n=1), potentially underrepresenting variability.
• Experimental scope: Incubations lasted ~48 h under natural light in bottles; in situ heterogeneity (e.g., microenvironments in mats, diel cycles) may not be fully captured.
• Hydroclimatic variability: September LCL sampling included rainwater–brine mixtures and highly concentrated brines beneath salt crusts, which may introduce short-term variability.
• Biological activity present: Some modern biological production (e.g., nearshore biofilms) persists in LCL, which can impose a residual biological P sink; extrapolation to fully prebiotic conditions requires caution.
• Unconstrained sediment fluxes: Ammonium fluxes from sediments and long-term organic P burial efficiencies were not directly quantified; these could modulate steady-state phosphate over longer timescales.