Salty Matters

The Blog is written by me, John Warren. Once every three or four weeks or so I will post an article or two on an evaporite topic that has piqued my interest. On the Saltwork Publications webpage (under "the Works") there is a growing library of pdfs and epubs based on these blogs. These articles on the website have much higher resolution extractable graphics in than in the blog. There is also a link to this set of pdfs and epubs on the home page (

Silica mobility and replaced evaporites: 1 - Alkaline lakes

John Warren - Saturday, July 02, 2016


In this series of blog articles, I plan to look at silica mobility, along with characteristic marine and nonmarine hydrogeochemistries over time, and how these parameters control chert and quartz precipitates and replacements in hypersaline settings. First, I will do this in modern surface and nearsurface settings in the marine and nonmarine realms, ending this first article with a focus on silica distribution and its precipitates in sulphate-depleted saline alkaline lacustrine sediments. Next we shall attempt a synthesis of controls on silica mobility and precipitation in the sulphate-enriched hypersaline marine surface, subsurface and burial realms, as well as defining relevant atmospheric and seawater chemistry changes across deep geologic time. And finally, we shall look at silica mobilisation in subsurface hydrothermal saline settings. The context for this discussion initially comes from the utility of recognising various silicified structures (including chert nodules) as indicative of typical marine (smooth-walled nodules), sulphate evaporite-enriched (cauliflower chert nodules) or sulphate-depleted alkaline lake (Magadi or crocodile skin chert) deposits (Figure 1).


Silica geochemistry

Modern river water typically contains less than 15 ppm dissolved silica, shallow meteoric groundwater typically has 10-50 ppm dissolved silica, while the modern ocean has between 0.5 and 10 ppm dissolved silica (Bridge and Demicco, 2008). Silica concentration is lowest in the ocean’s surface layer (less than 1 mg/l), and relatively constant at around 10 ppm below the thermocline. Deep saline sodium chloride and calcium chloride groundwaters contain comparatively little dissolved silica (30-80 ppm), relative to their total ionic content, although some saline subsurface waters in shale pore waters that also entrain dissolved organic acids can contain up to 330 ppm dissolved silica. Some of the highest values present in significant volumes of surface water are found in saline alkaline lakes with elevated pH levels. These waters can contain more than 1000 ppm silica in solution (Figure 3). One of the highest known natural water values, some 3970 mg/l of dissolved, is from a cold water spring known as Aqua de Rey, near the town of Mt. Shasta, California. The spring has a temperature around 54°C ana pH around 11.6.

At normal environmental pH, the dissolution-precipitation reaction of quartz,

Si02(s)quartz + H20(1) <-> H4SiO40(aq)

produces non-ionized silicic acid (H4SiO40). Because quartz is not very soluble at 25 °C, this reaction puts only ≈6 ppm silicic acid into solution. Therefore, most of the silicic acid in river water and groundwater is considered to come from the incongruent dissolution of silicate minerals, such as feldspars, during weathering. Non-crystalline amorphous silica gels are considerably more soluble, putting up to 120 ppm silicic acid into solution across the normal pH range (Figure 2a). The solubility of quartz is significantly affected by an increase in temperature, and at 300°C, approximately 600 ppm silicic acid is dissolved in groundwater with normal pH (Figure 2b; Verma, 2000; Fleming and Crerar, 1982). Siliceous sinter precipitates where such hot waters rise to the surface in hot springs and then cools, as at Mt Shasta. In the subsurface, cooling hydrothermal waters drive considerable silica mineral replacement and other cements associated with some types of epithermal and halokinetic ore-deposits (later blog). Rising pH also significantly affects the solubility of quartz, and this mechanism helps explain the elevated silica levels in the waters of many alkaline lakes (Figures 2a, 3a). At pH > 9 (at 25 °C) silicic acid dissociates:

H4Si04(aq) <-> H+(aq) + H3Si04(aq)

With the elevated pH of alkaline water, this reaction is driven to the right, and the solubility of both quartz and amorphous silica is greatly enhanced (Figure 1a). Saline alkaline lake waters with elevated pH, as in Lake Magadi in the African Rift Valley and the Alkali Valley playa brines of the south-west USA, consistently contain more than 1000 ppm dissolved silica as H3Si04(aq) (Figure 3a).


Modern siliceous sediments

Modern siliceous sediments accumulate as biogenic marine oozes, biogenic freshwater-lake deposits, chemical precipitates in alkaline lakes, chemical precipitates in soils (silcrete), and chemical precipitates around subaqueous and subaerial hot springs. Significant volumes of dissolved silica occur in the waters of saline alkaline lakes in the African Rift Valley (e.g. Lake Magadi) and a number of Basin and Range playa lakes in Oregon and California (e.g. Lake Abert, Oregon and Alkali Valley playa in California). Inflows in both regions are leaching highly labile volcanics (Figure 3).

Volumetrically, the most significant accumulations of modern siliceous sediments worldwide are constituted by seafloor deposits dominated by opal-A skeletons of planktonic diatoms. Today, diatoms (bacillariophytal algae with siliceous tests) scavenge virtually all of the silica in fresh to somewhat saline surface waters of the continental lakes and most significantly in open ocean waters of the marine realm (Figure 4). Diatoms arose in the Mesozoic, but have become particularly abundant over the past 30 million years, and their remains now dominate silica deposits of the ocean floor and the biogenic siliceous component of the sediment in many less-saline lakes and playa inflows (Knauth, 2003; Katz et al., 2004). Today, the oceans are everywhere undersaturated with amorphous silica, so diatoms build their shells despite thermodynamics of the dissolved content, so that some 90% of dead diatoms’ tests dissolve before they finally settle to the sea floor (Bridge and Demicco, 2008). However, appreciable thicknesses of diatom oozes can accumulate on the sea floor beneath regions with high productivity of diatoms, namely polar areas and areas of oceanic upwelling where there is a high flux of sinking diatoms, such as off the California coast. Other silica-secreting, single-celled eukaryotic plankton include the heterotrophic radiolarians and silicoflagellates. Radiolarian oozes are common beneath equatorial zones of oceanic upwelling.


Pore waters of marine siliceous oozes remain undersaturated with respect to amorphous silica for some depth below the sediment surface, and diagenetic reorganisation of silica is common in deep-sea siliceous sediments. This involves the slow conversion of opal-A to opal-CT, and eventually to microcrystalline quartz, via a complicated series of pathways involving quartz cementing and replacing the oozes. In turn, this leads to the diagenetic precipitation of significant volumes of smooth-walled chert nodules (Figure 1; Hesse, 1989). Worldwide the sampling of oceanic sediment by the Deep Sea Drilling Project (DSDP) has recovered well-developed chert in cemented layers within otherwise still unconsolidated Eocene siliceous oozes. The oldest sedimentary opal-A preserved in DSDP cores is Cretaceous in age, beyond that, cherts are made up of recrystallized quartz.

Dissolved silica and biogenic silica cycling in mesohaline lakes

Diatoms are a significant photosynthetic group driving the accumulation of amorphous silica in the bottom sediments of many freshwater temperate and saline lakes. When salinity and nutrient levels are appropriate, the diatom population in these lakes characteristically undergo explosive population growth. In the lit water column of temperate relatively-freshwater lakes, diatom blooms typically occur in the Spring and Fall of the year. This is when the water column in temperate lakes turns over, allowing nutrient-rich bottom waters to reach the near-surface lit zone at times when the surface water temperature is high enough to support diatom growth. In more saline lakes, diatom blooms occur whenever meromictic lake waters freshen to salinities appropriate for diatom growth. This leads to pulses of biogenic silica accumulating in the laminar distal (profundal) bottom sediments of these saline stratified lakes. As in the oceans, most lake columns in systems with periodic diatom blooms are undersaturated with amorphous silica, so most lacustrine diatom tests dissolve as they sink, and continue to dissolve in the bottom sediment. Closer to the lake shore, the greater volume of silica in the lake bottom sediments tends to come from detrital components washed into the lake as siliciclastic sand and mud.

Chert deposits through time preserve a record of secular change in the oceanic silica cycle. The evolutionary radiation of silica-secreting organisms resulted in a transition from abiological silica deposition, characteristic of the Archean and Proterozoic aeons, to the predominantly biologically-controlled silica deposition of the Phanerozoic (Maliva et al., 2005).

Biogenic lacustrine silica

Diatoms also flourish in the fresher water inflow areas of many salt lakes and playas. They are commonplace primary producers in mesohaline and lower penesaline environments with populations expanding across the lake when salinities are suitable. At less favourable time, healthy diatoms are restricted to refugia areas of fresher water springs and ponds. Their buoyant cells, often augmented by chitin threads or colonial adaptions, enable them to keep within the photic zone better than many other halotolerant algae. Some varieties of benthic diatoms live in lake brines with long-term salinities around 120 ‰, while the upper limit for diatom growth is around 180 ‰ (Clavero et al. 2000; Cook and Coleman 2007; Warren, 2016). The most halotolerant diatom taxa in the saltern ponds of Guerrero Negro are; Amphora subacutiuscula, Nitzschia fusiformis (both Amphora taxa), and Entomoneis sp.; all grow well in salinities ranging from 5 to 150 ‰. Three strains of the diatom Pleurosigma strigosum were unable to grow in salinities of less than 50 ‰ and so are true halophilic alga. A similar assemblage of Amphora sp., along with Cocconeis sp. and Nitzschia sp. dominate the high salinity (150 ‰) saltern ponds near Dry Creek, Adelaide, Australia (Cook and Coleman 2007).

Most mesohaline diatom species flourish at times of freshened surface lake waters or in and about perennial seepage and dissolution ponds (posas) about the edges of some salars, where they can be a major component in some lacustrine stromatolites (Figure 5). Several benthic diatom species are conspicuous in building diatomaceous stromatolites in these freshened (refugia) regions of the saline playa system, for example, Mastogloia sp., Nitzschia sp., Amphora sp., Diploneis sp. They function in a manner analogous to that of cyanobacteria in that they produce extracellular gel, are motile, phototropic, can trap and bind sediment, and create surface irregularities in the biolaminate mat (Winsborough et al., 1994).

Many Quaternary saline lakes have experienced significant fluctuations in water level and salinity across their millennial-scale sedimentary histories. For example, some 10,000 years ago Lake Magadi water depths were hundreds of metres above the present-day water levels and the diatomaceous High Magadi Beds (mostly laminites) were deposited. At that time most of the silica in the mesohaline stratified lake resided in sodium silicates deposited as laminites over most of the profundal lake floor. Diatoms flourished in the fresher waters inflow areas tied to deltaic sediments. Similar diatom rich-zones typify the fresher-water inflow areas of the Jordan River where it flows into the Northern end of the Dead Sea (Garber, 1980).

Silica in alkaline surface waters

It seems diatoms are efficient biogenic vectors for dissolved silica removal, not just in the oceans but also in many mesohaline lake settings. Significant diatom populations occur also in many saline lakes, even some hypersaline alkaline ones. They are important sediment contributors in Lake Manyara (Stoffers and Holdship, 1975), Lake Kivu (Degens et al., 1972), Abert Lake (Phillips and Van Denburgh, 1971), and perhaps also in the Great Salt Lake, Utah (Baxter et al., 2005). Silica levels in the water column of most of the lakesare typically very low (Hahl and Handy, 1969). So, in most mesohaline saline lakes the main biogenic form of silica is as amorphous in diatom skeletons, which tend to periodically accumulate in the bottom sediments with pore waters at the lower end of the hypersaline lacustrine salinity range (Warren, 2016, Chapter 9).

Coorong cherts

However, in some mesohaline to penesaline alkaline lakes, such as those of the Coorong region in South Australia, the ability of a diatom test to survive early burial is likely low. The siliceous frustules tend to dissolve and re-precipitate as amorphous inorganic silica. Abiogenic opal-CT precipitates are commonplace in evaporitic carbonate crusts of a number of Coorong Lakes about the edges of alkaline marginal-marine lagoons and lakes of the Coorong District of South Australia, especially those alkaline lakes containing magnesite or hydromagnesite (Peterson and von der Borch, 1965; Warren, 1988, 1990). Silica precipitation happens within the sediment, or just at the sediment surface, as opal gels (colloids) that can contribute up to 6% by weight of the high-magnesium carbonate sediment in the surface crusts of ephemeral saline lakes such as Milne Lake. Subsequent desiccation of the gel (locally known as yoghurt-textured mud) and consequent cracking creates hardened discs and plates of silica-impregnated mud, about 1 cm thick and 10 cm in diameter. About lake edges, these sediment discs and plates tend to erode into intraclast breccias that coat the uppermost massive unit as crust zones. In some lakes these crust breccia are located immediately edgeward of well-developed hemispherical stromatolites (Figure 6a and b; Warren, 1990).

Under seasonal high-pH conditions, the silica source in a Coorong ephemeral lake and its surrounds dissolves, it then re-precipitates as opal-CT as fresher subsurface groundwater with a lower pH seeps into the lake edges, and mixes with the surface lake brines. Measured pH in Milne Lake surface brines is ≈ 9.5 to 10.2. Silica precipitation tends to occur mostly in the periodically subaerial lake edges during times of incipient lake drying and shrinkage, prior to complete lake desiccation (that is abiogenic silica tends to precipitate in the late spring to early summer of the Southern Hemisphere). The initial silica phase impregnating the lake edge carbonate mud is opal-A. Peterson and von der Borch (1965) argued the likely source of the inorganically precipitated amorphous silica was the dissolution of detrital quartz (sand and silt), which is a common detrital component in the early stages of Holocene lake fill. These older Holocene units are now exposed about some lake edges (Warren, 1998, 1990).

However, diatoms do still thrive in many ephemeral Coorong lakes when surface waters have the appropriate levels of salinity and nutrients. They retreat to refugia about fresh water springs and seeps as the lake dries. Even tough common in plankton populations. intact diatom microfossils (siliceous frustules) have not been recognised in most cores from the same lakes. For example, diatom remains are not present in sapropelic muds in North Stromatolite Lake, a modern hydromagnesite-aragonite lake (Warren, 1990; McKirdy et al., 2010). But, within the organic geochemical constituents of the same cores, there are unusual T-shaped, C20 and C25 highly-branched isoprenoids, which are prominent among the aliphatic hydrocarbons in the extracted organic matter (Figure 7; Hayball et al., 1991). These unusual organic components were later recognised as bacillariophycean algal biomarkers (molecular fossils: McKirdy et al., 1995).

Diatoms are not organic-walled, and the silica of their frustules is highly susceptible to dissolution in present-day alkaline pore waters of this and other Coorong lakes. Hence, soon after burial, their cellular organic matter is destined to become part of the amorphous component of the kerogen (Barker, 1992). It is likely that the direct physical evidence for diatoms (viz. their siliceous frustules) is largely dissolved as waters become alkaline in many mesohaline Coorong ephemeral lakes, so that only biomarkers for a diatom source of the inorganically precipitated silica may remain.

Hence, the ultimate source of the inorganic siliceous carbonate breccia that defines the periodically subaerial edges of many ephemeral hypersaline Coorong lakes is likely from readily dissolved amorphous silica of diatom tests, not the much less soluble quartz, which was postulated as the likely silica source by Peterson and von der Borch (1995). Siliceous breccia zones in the edges of the ephemeral Coorong lakes are intimately tied to characteristic tepee expansion features known as extrusion tepees (Figure 6b). These expansion structures in cemented carbonate crusts are related to the desiccation/cementation of precursor gels washed into fractures beneath mobile sheets of colloid  muds ("yoghurts") that wash about the seasonally shrinking lake edge during the late spring to early summer (Kendall and Warren, 1987). No chert nodules are known to occur in the various Coorong Lakes, only siliceous carbonate mud layers and clasts associated with "extrusion" tepees.

Lake Magadi chert

Hypersaline chert is present as nodules, as well as siliceous breccia layers hosted in Pleistocene sediment that crops out landward of the current lake strandline. Precursors to these modern cherts are thought to initially deposit as late Pleistocene sodium silicates across significant portions of the Lake Magadi basin (Figure 8; Eugster, 1969). Today, these cherty precipitates comprise compact, well-indurated layers and nodules in what are otherwise unconsolidated lake sediments, rich in volcanic debris, sodium silicates and diatoms. The chert shows characteristic reticulate shrinkage cracks on the nodule surface giving it the name crocodile-skin or snake-skin chert or Magadi-style chert (Figure 1). The conversion to chert from its sodium silicate precursor is accompanied by many other enigmatic  textural and structural features such as large desiccation polygons, buckling, reticulation, extrusion, casts of mud-cracks and calcite cements.

Trenching in the regions landward of the current Lake Magadi strandline shows chert-rich zones laterally grade into sediments containing subsurface, still unconsolidated layers of sodium silicate gels dominated by the hydrous sodium silicate magadiite (NaSi7013(0H)3-3H20), with lesser amounts of kenyaite NaSi11O20.5(OH)3.H20 and makatite - NaSi2O3(OH)3.H2O (Figure 9a). Magadiite is a highly siliceous phase, running ≈78% SiO2 by weight (Table 1). When fresh, magadiite is white, soft, putty-like and readily deformable, but it dehydrates rapidly on exposure to air to harden irreversibly into fine-grained cherty aggregate. To date, magadiite has been found only in Quaternary alkaline lacustrine settings. In addition to Lake Magadi, Quaternary magadiite occurs in Lake Bogoria and Lake Kitagata, in Lake Chad in western central Africa, in the flats of Alkali Valley playa in Wyoming, and Trinity County in California (Sebag et al., 2001; Ma et al., 2011).


The solubility-equilibrium trends for silica and amorphous silica are similar, with a marked increase in solubility occurring in more alkaline conditions (pH>9; Figures 1a, 9b; Dietzel and Leftofsky-Papst, 2002). In contrast, SiO2 contents at equilibrium with magadiite show a minimum value at a pH around 8.5 and follow a different dissolution pattern to silica. At low pH the concentration of silica in solution increases, as it also does in alkaline solutions at the other end of the pH spectrum. Thus, the concentration of silica in a solution saturated with respect to magadiite, at constant Na content, is lowest in neutral to slightly alkaline solutions. Below pH 5.9, which is the intersection point of the magadiite and amorphous silica curves, magadiite exhibits a higher solubility than amorphous silica. Thus, at pH < 5.9 magadiite will dissolve, while amorphous silica precipitates (Figure 9; Dietzel and Leftofsky-Papst, 2002).

Conditions associated with the precipitation of magadiite from lake brines in Lake Chad, and probably most other soda lake occurrences, including Lake Magadi, require fluctuations in alkalinity or mixing interfaces between alkaline and less alkaline groundwaters (Figure 9b-d). Sebag et al. (2001) list the following conditions as typical of most modern magadiite occurrences; 1) Elevated alkalinity, typically in the lake dry season (pH >9) allow dissolution of silica, followed by lowering of alkalinity in the wet season driving precipitation of silica (Figure 9b, c), 2) High concentrations of dissolved silica (up to 2700 ppm), 3) Incorporation of sodium ions into the silica lattice that precipitates at the time of silica supersaturation (Figure 9d). Depending on the concentration of Na and Si in the brine at the time of precipitation, various sodium silicate minerals will precipitate (Figure 9).

Two general pathways have been proposed to explain the formation of magadiite in silica-rich sodium carbonate brines: a decrease in pH and evaporative concentration. Magadiite can precipitates when dilute inflow waters flow across a dense, sodium carbonate brine layer rich in dissolved silica interface mixing lowers the pH at the chemocline/halocline. In Lake Chad, and in some American examples in Califonia and Oregon, magadiite may have also precipitated by evaporative concentration or by capillary evaporation of saline, alkaline brines at a shallow subsurface water table. Other inferred mechanisms for sodium silicate precipitation include: 1) subsurface mixing of dilute and saline, alkaline groundwaters, 2) a reduction in pH of an alkaline brine resulting from an influx of biogenic or geothermally sourced CO2, and 3) precipitation from interstitial brines. Different sodium silicate minerals may form according to the concentrations of Na and SiO2 in the brine.

Magadiite (sodium silicate) layers and nodules in Lake Magadi weather into cherts and cherty breccia layers and so define Magadi-style cherts, with a characteristic reticulate, cracked or “crocodile-skin” surface created by shrinkage during the transformation from sodium silicate gel to chert nodule (Figure 1; Schubel and Simonson, 1990). In places, the Magadi chert layers preserve laminae of the original sodium silicate precursor. Conversion of magadiite to bedded and nodular chert is thought to take place close to the sediment surface and be related either to 1) the mobilisation and flushing of sodium by dilute waters in these shallow environments or, 2) to spontaneous conversion to chert in slightly deeper brine-saturated zones Intermediate diagenetic products, including kenyaite, amorphous silica and moganite, may form during the transformation to chert (see inset; Icole and Perinet, 1984; Sheppard and Gude, 1986). Both magadiite and the associated cherts have a distinctive trace element signature, unlike most other cherts (Kerrich et al., 2002).


Eugster et al., (1967) proposed that magadiite of the Lake Magadi High Beds was precipitated in the Late Pleistocene water column by diluting silica-rich, sodium carbonate lake brines with fresher waters at the lake chemocline or mid column interface. Mixing lowered the pH, and although the pH change may have been as little as 0.5, a decrease in pH from 10.3 to 9.8 lowers amorphous silica saturation by more than 500 ppm (Figure 2a). Silica solubility changes very little when pH varies below a maximum of 8. Highly alkaline sodium carbonate waters containing abundant SiO2 readily form in the Magadi rift valley via weathering and rapid subsurface hydrolysis of labile volcanic materials. Biogenically produced CO2 can also reduce the pH of the brine and drive magadiite precipitation (Eugster, 1969). Hay (1968) also suggested that simple evaporative concentration of the brine would lead to magadiite precipitation. Independent of freshwater flushing and pH changes, magadiite decomposes thermally in the lab into quartz and calcite at temperatures of 500-700°C (Lagaly et al., 1975).

In the perimeter sediments of Lake Magadi, Eugster (1969) described what he considered to be an impressive syndepositional result of this transformation of sodium silicates to chert, namely shrinkage megapolygons up to 50 m across in a bedded chert host, with bounding upturned chert ridges up to 2 m high. Historically, the megapolygons, extrusion tepees, convolute folds and other syntransformation features were interpreted as recording the shrinkage-induced flow and collapse of the sediments hosting the magadiite gels, as they lost sodium, dehydrated and shrank.

Subsequent work on the same megapolygonal structures by Behr and Röhricht (2000) concluded the megapolygons are not a response to mineralogical transformation, rather they are part of a suite of prelithification seismite structures in soft, siliceous lake sediment of the precursor Lake Magadi. That is, the chert megapolygons are a soft sediment response to intense deformation and local-scale diapirism, as are the numerous pillow-chert mounds, chert extrusives along dykes and fault ramps, horizontal liquefaction slides with breccias, slumps, petees, flows and shear-structures in the magadiite beds (now all preserved in chert at outcrop).

Collapse, liquefaction and extrusion of the pre-lithified siliceous matrix were caused by seismotectonic rift activity in the lake basin, and it activated fault scarplets and large-scale dyke systems. Seismic activity led to liquefaction and other earthquake-induced intrasediment deformation, especially along fault ramps and on tilted blocks. The textures all indicate the chert megapolygons are a form of seismite and do not mean volume changes in the transition from magadiite to chert. After liquefaction and extrusion, the exposed magadiite material solidified via spontaneous crystallisation to chert in an environment that was characterised by highly variable pH and salinity.

So, since the pioneering work of Eugster and others in the 1960s, three sets of diagenetic processes are now thought to be responsible for driving the conversion of magadiite to Magadi-style (crocodile-skin) chert in Lake Magadi and other soda lakes:

(1) Leaching of sodium by dilute surface runoff during weathering of the High Magadi beds, as evidenced by tracing unweathered beds into outcrop and summarised in the chemical transformation (Eugster, 1969; see inset);

NaSi7O13(OH)3.H2O + H+ —> 7SiO2 + Na+ + 5H2O

(2) Spontaneous release of sodium driving the conversion of magadiite to chert, whatever the nature of through-flushing solutions and environments (see inset). In this process sodium is expelled even in the presence of brines; it does not require the fresh water needed for process 1, and perhaps better explains the occurrence of calcite-filled trona casts in cherts and the presence of chert nodules in unweathered magadiite horizons in Lake Magadi (Hay, 1968, 1970).

(3) To this inorganic perspective on the transformation to form chert, Behr (2002) and Behr and Röhricht (2000) added a biological one. Based on field observations and microbiological studies of the cherts in Lake Magadi region, they argue that inorganic cherts are rare at the type locality of Magadi-style cherts. Rather, as inferred in many modern bacterial dolomites, the cherts at Lake Magadi may have been precipitated as amorphous silica via microbial processes and may not have had a sodium silicate precursor. They go on to note that most of the cherts in the Magadi depression are older than the High Magadi Beds and perhaps developed from flat-topped calcareous bioherms of Pleurocapsa, Gloecocapsa, and other coccoid cyanobacteria, along with thinly bedded filamentous microbial mats, stromatolites, bacterial slimes, diatoms, Dascladiacea colonies and other organic matter accumulations. Silicification occurred from a microbially mediated silicasol, via opal-A to opal-C, with final recrystallisation to a chert of quartzine composition. They conclude that metabolic processes of cyanobacteria controlled the pH of the brine and strongly influenced dissolution-precipitation mechanism that created the chert (Figure 1a). Today the debate as to inorganic versus organic origin of cherts in Lake Magadi continues, and is yet to be resolved.

Surdam et al. (1972) listed the following textures as indicators of Magadi-style cherts that have likely evolved from a sodium silicate gel: 1) Preservation of the soft-sediment deformation features of the putty-like magadiite, such as enterolithic folding, lobate nodular protrusions, casts of mudcracks and trona crystals, and extrusion forms; 2) Contraction features, especially the reticulate cracks and polygonal ridges on the surface of the chert, reflecting the loss of volume in the transition (crocodile-skin chert). If the arguments of Behr and Röhricht (2000) are accepted, then only criteria 2) the characteristic shrinkage-related reticulated surface texture (crocodile-skin) should be used to interpret ancient alkaline lake cherts, along with a lack of any indications of calcium sulphate in penecontemporaneous lake sediments.

Given the type-1 hydrogeochemistry needed for highly alkaline continental brines, evaporite minerals likely to be found in association with Magadi-style cherts are the sodium carbonate salts (trona, gaylussite or pirssonite; searlesite) or their pseudomorphs; gypsum and other forms of calcium sulphate are never present in type 1 (trona -precipitating) brines (Figure 10; Hardie and Eugster, 1970). This contrast with the silica replacement mechanisms that occur when gypsum or anhydrite nodules are silicified.

Across the Phanerozoic rock record, ancient examples of crocodile-skin cherts are not common, compared to documented examples of silicified anhydrite nodules (cauliflower chert). Documented examples include: Cambrian alkaline lacustrine sediments in South Australia (White and Youngs, 1980); Eocene Green River sediments in the USA (Eugster and Surdam, 1973), the Middle Devonian in the Orcadian Basin of Scotland (Parnell, 1988) and fluviolacustrine sediments of the Permian Balzano volcanic complex in Italy (Krainer and Spötl, 1998). For all such ancient occurrences of ancient crocodile-skin chert, it should be remembered that the precipitation mechanism in its type area of Lake Magadi is still contentious. That is, Magadi-type chert, is historically interpreted as being diagenetically derived from magadiite, a hydrous sodium silicate precursor deposited from strongly alkaline lake waters. More recent work in Lake Magadi concludes that the same chert, hosted in the same High Magadi Beds is due to chemical decomposition of pyroclastic deposits by alkaline groundwater, and that chert precipitation is strongly influenced by fluctuating levels of biogenic CO2. The numerous deformation features in the High Magadi Beds in this more recent interpretation are unrelated to the mineralogical transformation of magadiite to chert (Behr, 2002).


Across longer time frames than that preserved in the Pleistocene sediments of Lake Magadi, the chemical proportions of various ionic components in seawater are not constant (Figure 10). The varying proportions are intimately related to the evolution of the world’s atmosphere and rates of seafloor spreading (Warren, 2016; Chapter 2). Sulphate (rather than sulphide) only became a significant component in the world’s ocean around 2Ga. Before that, the world’s oceans and its atmosphere lacked significant oxygen, and entrained much higher proportions of CO2 and methane. In the Archean, the world’s oceans were Na-Cl-Ca-HCO3 waters, not the Na-Cl-Mg-SO4 systems of today and trona, along with halite were primary precipitates in marine hypersaline settings. Under that scenario, it is likely that some marine-associated hypersaline cherts were formed via replacement of sodium silicate precursors. In Phanerozoic strata, an ability to separate cauliflower cherts (after gypsum/anhydrite nodules) from crocodile-skin cherts (associated with silicate gels in trona/natron soda lakes) is considered significant in defining marine-fed versus continental saline hydrologies. Hydrochemistry and textures associated with silicification in CaSO4-rich environments is the topic of the next blog article.


Barker, P., 1992, Differential diatom dissolution in Late Quaternary sediments from Lake Manyara, Tanzania: an experimental approach: Journal of Paleolimnology, v. 7, p. 235-251.

Baxter, B., C. Litchfield, K. Sowers, J. Griffith, P. Dassarma, and S. Dassarma, 2005, Microbial Diversity of Great Salt Lake, in N. Gunde-Cimerman, A. Oren, and A. Plemenitaš, eds., Adaptation to Life at High Salt Concentrations in Archaea, Bacteria, and Eukarya: Dordrecht, Netherlands, Springer, p. 9-25.

Behr, H.-J., and C. Röhricht, 2000, Record of seismotectonic events in siliceous cyanobacterial sediments (Magadi cherts), Lake Magadi, Kenya: International Journal of Earth Sciences, v. 89, p. 268-283.

Behr, H. J., 2002, Magadiite and Magadi chert; a critical analysis of the silica sediments in the Lake Magadi Basin, Kenya, in R. W. Renaut, and G. M. Ashley, eds., Sedimentation in continental rifts, v. 73: Tulsa, Society for Sedimentary Geology (SEPM) Special Publication, p. 257-273.

Bridge, J. S., and R. V. Demicco, 2008, Earth Surface Processes, Landforms and Sediment Deposits, Cambridge University Press, 815 p.

Clavero, E., M. Hernandez-Marine, J. O. Grimalt, and F. Garcia-Pichel, 2000, Salinity tolerance of diatoms from Thalassic hypersaline environments: Journal of Phycology, v. 36, p. 1021-1034.

Cook, F. S., and P. S. J. Coleman, 2007, Benthic diatoms in the salinas of the Dry Creek saltfields, South Australia: Hydrobiologia, v. 576, p. 61-68.

Degens, E. T., R. P. Von Herzer, H.-K. Wong, W. E. Deu­ser, and H. W. Jannasch, 1972, Lake Kivu: Structure, chemistry and biology of an East African Rift lake: Geologische Rundschau, v. 62, p. 245-277.

Dietzel, M., and I. Letofsky-Papst, 2002, Stability of magadiite between 20 and 100°C: Clays and Clay Minerals, v. 50, p. 657-666.

Eugster, H. P., 1967, Hydrous sodium silicate from Lake Magadi, Kenya: precursors of bedded chert: Science, v. 157, p. 1177-1180.

Eugster, H. P., 1969, Inorganic bedded cherts from the Magadi area, Kenya: Contributions Mineralogy and Petrology, v. 22, p. 1-31.

Eugster, H. P., and R. Surdam, 1973, Depositional environment of the Green River Formation of Wyoming: A preliminary report: Bulletin Geological Society of America, v. 84, p. 1115-1120.

Fleming, B. A., and D. A. Crerar, 1982, Silicic acid ionization and calculation of silica solubility at elevated temperature and pH application to geothermal fluid processing and reinjection: Geothermics, v. 11, p. 15-29.

Garber, R. A., 1980, The sedimentology of the Dead Sea: PhD thesis, Rennsaler Polytechnic, 170 p.

Hahl, D. C., and A. H. Handy, 1969, Great Salt Lake, Utah: Chemical and physical variation of the brine, 1963-1966. : Utah Geo!. Mineral. Survey, Water Resour. Bull. 12. 33 pp.

Hardie, L. A., 1984, Evaporites: Marine or non-marine?: American Journal of Science, v. 284, p. 193-240.

Hardie, L. A., and H. P. Eugster, 1970, The evolution of closed-basin brines: Mineralogical Society America Special Paper, v. 3, p. 273-290.

Hay, R. L., 1968, Chert and its sodium-silicate precursors in sodium-carbonate lakes of east Africa: Contributions to Mineralogy and Petrology, v. 17, p. 255-274.

Hay, R. L., 1970, Silicate reactions in three lithofacies of a semi arid basin, Olduvai Gorge, Tanzania: Mineralogical Society of America Special Paper, v. 3, p. 237-255.

Hayball, A. J., D. M. MCKirdy, J. K. Warren, C. C. von der Borch, and D. Padley, 1991, Organic facies of Holocene lacustrine carbonates in North Stromatolite Lake, Coorong Region, South Australia. : 15th International Meet-ing on Organic Geochemistry, Manchester, U.K., Programme and Abstracts, p. 19-20.

Hesse, R., 1989, Silica diagenesis: origin of inorganic and replacement cherts.: Earth Science Reviews, v. 26, p. 253-284.

Icole, M., and G. Perinet, 1984, Les silicates sodiques et les milieux evaporitiques carbonates bicarbonates sodiques: une revue: Revue de Geologie Dynamique et de Geographie Physique, v. 25, p. 167-176.

Katz, M. E., Z. V. Finkel, D. Grzebyk, A. H. Knoll, and P. G. Falkowski, 2004, Evolutionary Trajectories and Biogeochemical Impacts of Marine Eukaryotic Phytoplankton: Annual Review of Ecology, Evolution, and Systematics, v. 35, p. 523-556.

Kendall, C. G. S. C., and J. K. Warren, 1987, A review of the origin and setting of tepees and their associated fabrics: Sedimentology, v. 34, p. 1007-1027.

Kerrich, R., R. W. Renaut, and T. Bonli, 2002, Trace-Element Composition of Cherts from Alkaline Lakes in the East African Rift, Sedimentation in Continental Rifts, v. 73, SEPM Society for Sedimentary Geology, p. 275-294.

Knauth, L. P., and D. R. Lowe, 2003, High Archean climatic temperature inferred from oxygen isotope geochemistry of cherts in the 3.5 Ga Swaziland Supergroup, South Africa: Geological Society of America Bulletin, v. 115, p. 566-580.

Krainer, K., and C. Spotl, 1998, Abiogenic silica layers within a fluviolacustrine succession, Balzano volcanic complex, Northern Italy - A Permian analogue for Magadi-type cherts: Sedimentology, v. 45, p. 489-505.

Krauskopf, K. B., 1967, Introduction to Geochemistry, McGraw-Hill, 721 p.

Lagaly, G., K. Beneke, and A. Weiss, 1975, Magadiite and H-magadiite 1. Sodium magadiite and some of its derivatives: American Mineralogist, v. 60.

Ma, L., T. K. Lowenstein, and J. M. Russell, 2011, A Brine Evolution Model and Mineralogy of Chemical Sediments in a Volcanic Crater, Lake Kitagata, Uganda: Aquatic Geochemistry, v. 17, p. 129-140.

Maliva, R. G., A. H. Knoll, and B. M. Simonson, 2005, Secular change in the Precambrian silica cycle: Insights from chert petrology: Geological Society of America Bulletin, v. 117, p. 835-845.

McKirdy, D. M., A. J. Hayball, J. K. Warren, E. D., and C. C. von der Borch, 2010, Organic facies of Holocene carbonates in North Stromatolite Lake, Coorong region, South Australia: Cadernos Laboratorio Xeolóxico de Laxe, v. 35, p. 127-146.

McKirdy, D. M., C. J. Hepplewhite, B. H. Michaelsen, A. G. Mazzoleni, and Y. Bone, 1995, Origin of sapropels in Holocene lakes of the Coorong region, South Australia, in J. O. Grimalt, and C. Dorronsoro, eds., Organic Geochemistry: Developments and Applications to Energy, Climate, Environment and Human History, AIGOA, Donostia-San Sebastián, p. 183-185.

Parnell, J., 1988, Significance of lacustrine cherts for the environment of source-rock deposition in the Orcadian Basin, Scotland: Geological Society, London, Special Publications, v. 40, p. 205-217.

Peterson, M. N. A., and C. C. Von der Borch, 1965, Chert: modern inorganic deposition in a carbonate precipitating localitty: Science, v. 149, p. 1501-1503.

Phillips, K. N., and A. S. Van Denburgh, 1971, Hydrology and geochemistry of Abert, Summer and Goose Lakes, and other closed-basin lakes in South-central Oregon: U.S. Geo!. Survey Prof. Paper 502B. 86 pp.

Schubel, K. A., and B. M. Simonson, 1990, Petrography and diagenesis of cherts of Lake Magadi, Kenya: Journal of Sedimentary Petrology, v. 60, p. 761-776.

Sebag, D., E. P. Verrecchia, S. J. Lee, and A. Durand, 2001, The natural hydrous sodium silicates from the northern bank of Lake Chad: occurrence, petrology and genesis: Sedimentary Geology, v. 139, p. 15-31.

Sheppard, A. P., and A. J. Gude, 1986, Magadi-type chert - A distinctive diagenetic variety from lacustrine deposits, in F. A. Mumpton, ed., Studies in Diagenesis, v. 1578, US Geological Survey Bulletin, p. 335-343.

Surdam, R. C., H. P. Eugster, and R. H. Mariner, 1972, Magadi-type chert in Jurassic and Eocene to Pleistocene rocks Wyoming: Geol. Soc. Am. Bull., v. 83, p. 1739-1752.

Verma, M. P., 2000, Revised quartz solubility temperature dependence equation along the water–vapor saturation curve: Proceedings of the 2000 World Geothermal Congress. Kyushu, Tohoku, Japan, p. 1927-1932.

Warren, J. K., 1988, Sedimentology of Coorong dolomite in the Salt Creek region, South Australia: Carbonates and Evaporites, v. 3, p. 175-199.

Warren, J. K., 1990, Sedimentology and mineralogy of dolomitic Coorong lakes, South Australia: Journal of Sedimentary Petrology, v. 60, p. 843-858.

Warren, J. K., 2016, Evaporites: A compendium (ISBN 978-3-319-13511-3) Published Feb. 22, 2016: Berlin, Springer, 1854 p.

White, A. H., and B. C. Youngs, 1980, Cambrian alkali playa-lacustrine sequence in the northeastern Officer Basin, South Australia.: Journal of Sedimentary Petrology, v. 50, p. 1279 - 1286.

Winsborough, B. M., J. S. Seeler, S. Golubic, R. L. Folk, and B. Maguire, Jr., 1994, Recent fresh-water lacustrine stromatolites, stromatolitic mats and oncoids from northeastern Mexico, in J. Bertrand-Sarfati, and C. Monty, eds., Phanerozoic Stromatolites II: Amsterdam, Kluwer Academic Publishers, p. 71-100.


Lake Nakuru flamingos– Life's response to feast and famine in schizohaline lacustrine hydrologies

John Warren - Monday, May 23, 2016

The Flamingo Connection

An aviator once described Lake Nakuru as “a crucible of pink and crimson fire,” with a million flamingos painting an astonishing band of colour that burst into pieces as the birds took flight (Figure 1).

Flamingo population levels in Lake Nakuru and any mass “die-offs” are popularly considered as indicators of the environmental health of Nakuru and other lakes in the African rift valley with significant flamingo populations. In 2006, more than 30,000 of the birds were found dead at Nakuru, leaving enough pink carcases to spur an international newspaper to describe the lake as a “flamingo death camp.” Two years prior, 43,800 of the birds had perished at Tanzania’s Lake Manyara, the first major die-off documented at that alkaline, soda-rich lake. Previous mass die-offs occurred at Lake Nakuru and two other Kenyan lakes in 1993, 1995, and 1997, as well as at two lakes in Tanzania in 2002. At the same time, birds were gathering in places they have never been documented before. Since 2006 there have been additional population crashes at Nakuru and Elmentia (Table 1).


Flamingos numbers in Lake Nakuru are perhaps one of the most visually impressive responses to episodic but very high levels of organic productivity, driven by a well-adapted species feeding in a layered saline water body subject to periodic salinity stress (Warren, 2011). The fluctuating richness of the lake’s flamingo population was dubbed the “flamingo connection” in a benchmark paper by Kirkland and Evans (1981) that considered mesohaline evaporitic carbonates as hydrocarbon source rocks.

Flamingo biology

Flamingos (Aves, Phoenicopteridae) are an ancient lineage of long-legged, microphagous, colonial wading birds. Although popularly misperceived as tropical species, flamingo distribution is more closely tied to the great deserts of the world and to hypersaline lake sites, than it is to equatorial regions (Bildstein, 1993). Flamingos are filter feeders that thrive on halotolerant cyanobacterial blooms in mesohaline shallows of saline lakes around the world. This creates the context between flamingos, mesohaline planktonic blooms, and saline lakes, well documented in Lake Nakuru by Vareschi (1982) and first noted in the geological literature in that benchmark Kirkland and Evans paper. Sedimentary textures and structures associated with flamingo lifestyles, where these birds dominate the macrofauna in some modern saline lakes, are described by Scott et al. (2009, 2012). Ancient avian counterparts can leave a characteristic set of trackways and trace fossils (including nest mounds) that can be used to refine mid-late Tertiary lacustrine depositional models (Melchor et al., 2012). Flamingo-like ancestors, which would give rise to modern ducks, even left traces in the shallow saline mudflats of Eocene soda lake sediments that define the saline portions of the trona/nahcolite-bearing sediments of the Green River formation in Utah (Figure 2).

Two species of flamingo gather in huge numbers in Lake Nakuru and a few other East African rift lakes, namely, the greater and the lesser flamingo (Phoenicoptus ruber roseus and P. minor respectively), with the lesser flamingo having characteristic and spectacular pink-red colouration in their feathers. These bright pink waders feed and breed in mesohaline rift lake waters where cyanobacterial blooms can be so dense that a Secchi disc disappears within a few centimetres of the lake’s water surface (Warren, 1986, 2011). Lake Natron (Koeppen climate Aw), where trona is the dominant evaporite, is a major breeding ground for flamingos in East Africa and is the only regular breeding site for the lesser flamingo in Africa (Simmons 1995). Lesser flamingos build nesting platforms on the trona pavement in the more central parts of the lake. These spectacular birds not only feed in saline waters, they can choose to build nest mounds on evaporite pavements!

Worldwide, only six sites are used for breeding by the lesser flamingos: Lake Natron (Tanzania), Etosha Pan (Namibia), Makgadikgadi-Pan (Botswana), Kamfers Dam (South Africa), as well as two pans in the “Little Rann of Kachchh” (India). Recent estimates of lesser flamingos at the main distribution areas are as follows: 1.5–2.5 million in eastern Africa; 390,000 in northwestern India; 55,000-65,000 in southwestern Africa; and 15,000–25,000 in western Africa. The highest population densities occur in Kenya (1.5 million) and Tanzania (600,000) (Childress et al. 2008). Lesser flamingos are well adapted to the harsh conditions associated with living and breeding in hypersaline alkaline conditions. Worldwide, they follow an itinerant lifestyle, ranging across their distribution areas in search of saline water bodies with appropriate cyanobacterial blooms. In the east African rift the flocks can travel up to 200 km a day between feeding and breeding sites, which are generally at geographically separate locations (Figure 3).

Lake Natron has the highest concentration of breeding flamingos of any lake in East Africa. Both the greater and the lesser flamingo are found there, with the lesser flamingo outnumbering the greater by a hundred to one. Lesser flamingos bred at Lake Natron in 9 out of 14 years from 1954 to 1967. But while the trona-rich nearby Lake Natron is an essential breeding site, it is not a focal feeding site for flamingos. Major feeding sites in the Africa rift valley are Lakes Nakuru and Bogoria (formerly Lake Hannington) in Kenya and entrain waters that are mesohaline with an abundance of halotolerant cyanobacteria, dominated by Arthrospira (Schagerl et al., 2015). Lake Nakuru is a mesohaline soda lake with a pH ≈ 10.5 and a typical annual salinity range of 15-45‰, nearby lake Bogoria is somewhat larger, also an alkaline soda lake, with somewhat higher salinities. The Lake Nakuru depression measures some 6.5 km by 10 km, with a water covered area of some 5-45 km,2 experiencing an annual pan evaporation rate of 1500 mm beneath a Cfb Köppen climate (Figure 3b; Vareschi,1982; Krienitz and Kotut, 2010). It contains a eutrophic bottom water mass in the lake centre, with thermally stratified water column that can be up to 4.5 metres deep. The Lake Bogoria’ water mass is up to 4 km wide, some 17 km long with thermally stratified eutrophic waters up to 10 m deep. It lies beneath a Cfb climate with a pan evaporation of 2600 mm. It is fed by a combination of rainfall (≈760mm/year) and numerous (>200) hydrothermal springs about the lake edge (Figure 3c).

Lake Bogoria hydrochemistry, it seems, is more stable compared to other endorheic lakes in Kenya, because of its greater depth (10 m), steep shores and larger water volume preventing it from drying up. In contrast, Lake Nakuru resembles more a flat pan; it is much shallower (1 m) and is more subject to changes in size related to changes in water levels, at times Lake Nakuru can dry up completely. Water depths in both lakes vary from year to year, in large part depending on the vagaries of annual runoff/ inflow. The higher level of hydrothermal inflow in Bogoria’s hydrology, along with its somewhat steeper hydrographic profile, means the lake area salinity and water depth vary less from year to year compared to Nakuru (Figure 4).

Lakes Nakuru and Bogoria, which at peak breeding times can support lesser flamingo populations in excess of 1.5 million birds, have surface areas that are less than half those of lakes Natron and Magadi, which in turn have small areas compared to most ancient lacustrine evaporites. Yet Nakuru and Bogoria are two of the most organically productive ecosystems in the world (Warren, 2011). What makes both lakes so productive are dense populations of halotolerant cyanobacteria, especially Arthrospira sp, which flourish and periodically reach peak growth in their waters yet at other times organic productivity can crash. The morphological and hydrochemical contrasts likely accounts for the more frequent Arthrospira crashes and biotal community changes in Nakuru when compared to Bogoria and other lakes in the region (Table 1).

Flamingos as filter feeders

Flamingos pass water through their bill filters in two ways (as documented by Penelope M. Jenkin in her classic article of 1957): either by swinging their heads back and forth just below the water surface, so permitting the water to flow passively through the filters of their beak, or by more efficient and more usual system of an active beak pumping. The latter is maintained by a large and powerful tongue that fills a large channel in the lower beak. As it moves rapidly back and forth, up to four times a second, it drawing water and plankton through the beak filters on the backwards pull and expelling it on the forward drive. The tongue’s surface also sports numerous denticles that scrape the collected food from the filters.

Flamingo beaks have evolved into highly efficient plankton-extraction apparati that exploit the dense cyanobacterial populations periodically found in mesohaline lakes worldwide (Figure 5; Gould, 1987; Bildstein, 1993). The beak is unlike that possessed by any other bird group on earth; the affinity is more to the baleen of whales used to filter planktonic krill from the lit upper waters of world's oceans. A flamingo beak houses a high volume water-filtering system made up of a piston-like tongue and hair-like structures called lamellae made up of rows of fringed platelets that line the inside of the mandible. In the lesser flamingo, the lamellae fibres have the appropriated spacing for capturing coiled filaments of Arthrospira. Lamellar spacings are wider in the beaks of the greater flamingo than those in the beaks of the lesser flamingo so these larger-sized birds are more generalist feeders of lake zooplankton. Thus, in any mesohaline lake where the two bird species feed and co-exist, they do not compete for the same food source. By swinging their upside-down heads from side to side just below the water surface and using the piston-like tongue to swish water through their lamella-lined beaks, flamingos can syphon the lake plankton into their gullets at phenomenal rates. Lesser flamingos can pump and filter as many as 4 beakfuls of plankton-rich water a second. This means some individuals filter upwards of 20,000 litres of water per day.

How the birds manage to cycle so much brackish to mesohaline waters, while maintaining their osmotic integrity, remains a mystery. When the rift lakes are typified by dense cyanobacterial blooms, each adult bird ingests around 72 g dry weight of Arthrospira per day (Vareschi 1978). This means the Nakuru lesser flamingo flock is able to ingest 50–94% of the daily primary production in the lake (about 60-80 tons). Rates of planktonic renewal in these rift lakes is obviously extremely high and the required rates of biomass production by Arthrospira can be spectacular. Vonshak (1997) reported doubling times of 11–20 hours of Arthrospira in a culture growing under mesohaline conditions at 35°C.

Flamingos (flamingoes) are mostly nocturnal feeders and will feed for up to 12-13 hours in a 24 hour period. The preferred planktonic food of the lesser flamingo is the cyanobacterium Arthrospira platensis (formerly known as Spirulina platensis), which for much of an average year is the widespread phytoplankton component in Lake Nakuru and Lake Bogoria waters (Figure 6). Unfortunately, dense populations of Arthrospira function best at a preferred range of temperature and salinity, meaning acme populations tend to collapse irregularly and unpredictably in both Lake Nakuru and Bogoria, leading to highly-stressed malnourished flocks of flamingos subject to mass dieback (Figure 6; Vareschi 1978; Kreinitz and Kotut, 2010).

Arthrospira has high levels of the red pigment phycoerythrin and so when ingested in large volumes it accumulates in flamingo feathers to give the birds their world famous colouration, hence the “flamingo connection.” Once digested, the carotenoid pigment dissolves in fats, which are then deposited in the growing feathers. The same effect is seen when shrimp change colour during cooking due to carotenoid alteration. The amount of pigment laid down in the feathers depends on the quantity of pigment in the flamingo’s diet. Lesser flamingos, with beak design maximised to feed on Arthrospira have a more intense pink colour in their feathers than greater flamingos. The latter species sits higher in the Lake Nakuru food chain and so gets the slight pink tinge in its feather colour, mostly second-hand from the lake zooplankton, which also feed on Arthrospira.

As well as possessing very high levels of phycoerythrin in its cytoplasm, Arthrospira is also unusual among the cyanobacteria in its unusually high protein content (some ten times that of soya). This, and the high growth rate of this species, explains why Nakuru and Bogoria acme populations can support such spectacular population levels of flamingos. A lack of cellulose in the Arthrospira cell wall means it is a source of plant protein readily absorbed by the gut, making it a potentially harvestable human food source in saline water bodies in regions of desertification. In Lake Chad, and in some saline lakes in Mexico, Arthrospira accumulates as a lake edge scum that has been harvested for millennia by the local people (including the ancient Aztecs) and used to make nutritious biscuits.

When Arthrospira stocks are low in the rift lakes, the lesser flamingo will consume benthic diatoms. However, net primary productivity of benthic diatoms in East African soda lakes is one to two orders of magnitude less than that of Arthrospira, and the carrying capacity of the habitat with diatoms is lower by the same order (Tuite 2000). This lower productivity is seen in the peak 25,000 bird population, which are diatom feeders, in Laguna de Pozuelos in Argentina (area ≈100-130 km2, more than 3 times that of Nakuru, yet the peak flamingo numbers are two order of magnitude lower). In Lake Nakuru, Arthrospira and other lake plankton are also consumed by one species of introduced tilapid fish and one species of copepod and a crustacean. Rotifers, waterboatmen, and midge larvae also flourish in the mesohaline waters of Lake Nakuru. The mouth-breeding tilapid Sarotherodon alcalicum grahami was introduced to the lake in the 1950s to control the mosquito problem and fish-feeding birds (such as pelicans) have flourished ever since (Vareschi, 1978). During times of non-optimum water conditions, when either freshening or somewhat elevated hypersalinity lessens the number of Arthrospira, the tilapids can displace the flamingos as the primary consumers of planktonic algae.

In 1972 Lake Nakuru waters held a surface biomass of 270 g/m3 and an average biomass of 194g/m3 but, as in most hypersaline ecosystems, Nakuru’s organic production rate varies drastically from year to year as water conditions fluctuate (Figure 7; Vareschi, 1978). Arthrospira was in a long-lasting bloom in 1971-1973, and accounted for 80-100% of the copious phytoplankton biomass in those years. In 1974, however, Arthrospira almost disappeared from the lake and was replaced by much smaller-diameter planktonic species, such as coccoid cyanobacteria that dealt better with elevated salinities. This transfer in primary producer make-up in the lake waters also made the lesser flamingos less efficient feeders. When the relatively large filaments of Arthrospira dominate the lake plankton, the flamingo’s beak filters between 64 and 86% of the plankton held in each mouthful of lake water (Vareschi, 1978). When the much smaller coccoids come to dominate, the filters are a much less efficient feeding mechanism. The change in plankton species was also tied to a severe reduction in algal biomass (and protein availability), which in 1974 was down to 71 g/m3 in surface waters and averaged 137 g/m3 in the total water mass. As a result, the flamingo numbers feeding in the lake declined from 1 million to several thousand, driving a significant die-back as the salinity-stressed flamingo population moved to other lakes, like Bogoria, where Arthrospira were flourishing (Vareschi, 1978).

The lower salinity limit for an Arthrospira bloom is ≈ 5‰, but it does better when salinity is more than 20‰. The species dominance of Arthrospira and its higher biomass in somewhat more saline lake waters in the African Rift lakes is clearly seen in the near unispecific year-round biomass of nearby Lake Bogoria (salinity 40-50‰). Its surface salinity is higher year round than Lake Nakuru (salinity ≈ 30‰) and the somewhat fresher waters of Lake Elmenteita (salinity ≈ 20‰; Figure 3a). Because of this, Lake Bogoria is a more reliable food source for feeding flamingos compared to either Nakuru or Elmenteita. Birds tend to migrate there to feed when conditions for Arthrospira growth are not ideal in other nearby alkaline lakes (too fresh or too saline). This was the case in 1999 when high rainfall and dilution of lake waters caused the Arthrospira levels to fall in both Nakuru and Elmenteita. It was also true in late 2012 when Nakuru water levels were at near-historic highs and the waters too fresh to support a healthy Arthrospira population.

Flamingo biomass controls

A driving mechanism for the abrupt change in biomass in Lake Nakuru in the period 1972 -1974 was not clearly defined. It was thought to be related to increased salinity and lowering of lake levels, driving the growth of coccoid species other than Arthrospira sp. that are better adapted to higher salinity, but offering less protein to the feeding birds (Figure 7a; Vareschi, 1978). There is also the simple fact that in a lake with no surface outflow, ever more saline waters cover ever-smaller areas on the lake floor. There have been times in the last 70 years when most of Lake Nakuru has dried up and pools of saline water only a few tens of centimetres deep remained. This was the case in 1962 and most recently the case in 2008 (Figure 4).

After the lake level lows of the 1960s, during the mid to late 1970s and in the 1980s the Nakuru hydrology returned to more typical inherent oscillations in water level and salinity (schizohalinity). The flamingo populations in Nakuru returned to impressive numbers but followed the vagaries of Arthrospira blooms. Since the early 1990s more reliable long-term datasets on physical lake condition and flamingo numbers have been compiled (Figures 5, 7b). In that time frame, in 1993, 1995 1998, 2008 and 2012, the flamingo populations feeding in Lake Nakuru were once again at very low levels and the remaining bird populations were stressed and subject to mass dieback.

In 1998, unlike 1974, the stress on the flamingo population was related to lake freshening and rising water levels driving the decrease in Arthrospira biomass, not increased salinity and desiccation. In the preceding bountiful years, the Arthrospira-dominated biomass had bloomed at times when salinities were favourable and died back at times of elevated salinities and lake desiccation, as in 1974. By 2000 formerly low salinities had once again increased making surface waters suitable for another widespread Arthrospira bloom and the associated return of high numbers of feeding flamingos, which continued until 2007 (Figure 7c). In 2013 there was another freshening event, with associated rising water levels and the lakes flamingo population moved to Lake Bogonia to feed.

Freshening favours a cyanobacterial assemblage dominated by picoplanktic chlorophytes (Picocyctis salinarum) and the nostocalean Anabaenopsis; the latter creates slimy masses that clog the flamingo’s feeding apparatus. The combination can drive much of the flamingo population to starvation or migration to other lakes with suitable salinities (Krienitz and Kotut, 2010). With freshening, comes also the possibility of the growth of strains that produce toxins (such as Anabaenopsis or Microcystis), possibly not in the feeding areas, which tend to remain too saline for Microcyctis, but in the spring waters where the flamingos fly in order to bathe and cleanse their feathers after a night spent feeding (Kotut and Krienitz, 2011).

It seems that breeding flamingos come to Lake Nakuru to feed in large numbers when there is water in the lake with appropriate salinity and nutrient levels to facilitate an Arthrospira bloom. In some years when heavy rains occur, lake levels rise significantly and the lake waters, although perennial, stay in the lower salinity tolerance range for Arthrospira platensis, keeping cyanobacterial numbers and protein levels at the lower end of the spectrum, as in the El Niño period between October 1997 to April 1998 and again in 2013. Once lake levels start to fall, salinities and rates of salinity change return to higher levels, then water conditions once again become appropriate for an Arthrospira bloom. But the environmental stress on the flamingos also comes with further-elevated salinities and desiccation moving lake hydrochemistry into salinities at the upper end of Arthrospira tolerance.

One of the reasons Lake Nakuru is suitable for phenomenal cyanobacterial and algal growth at times of Arthrospira bloom is the maintenance of suitable temperatures and oxygenation in the upper water mass, where the photosynthetic Arthrospira thrive. Nakuru develops a daily thermocline in the top 1.5 metres of the water column that dissipates each day in the late afternoon via wind mixing. Overturn recycles nutrients (derived from the decomposition of bird and other droppings, including those of resident hippopotami) back to the oxygenated lit surface waters to facilitate an ongoing bloom the next day (Figure 7c).

Numbers of flamingos feeding in Lake Nakuru and Lake Bogoria are used in the popular press as indicators of the environmental health of the lakes. Thousands of birds died in Lake Nakuru in 1995 and more than 30,000 birds may have died in Lake Bogoria in the first half of 1999. The most dramatic die-offs in the last two decades were at Lake Nakuru in August 2006, when some 30,000 died, and Lake Bogoria in July 2008, when 30,000 birds died. Some environmentalists have argued in the popular press that mass die-offs and their perception of lowered numbers of flamingos in Lake Nakuru and Lake Bogoria across the 1990s and 2000s were indicators of uncontrolled forest clearance, an uncontrolled increase in sewerage encouraging eutrophication, and increase in heavy metals from increasing industrial pollutants in the lake, along with general stress on the bird population from tourists and the drastic increase in local human population centred on the town of Nakuru (third largest in Kenya). Numbers of people in the town, which is the main city in the rift valley, have grown by an average of 10% every decade for the past 30 years.

But like much environmental doomsday argument, it is more based on opinionated prediction than on scientific fact. When numbers of feeding flamingos in Lake Nakuru are plotted across last few decades, it is evident that flamingo numbers oscillate widely, but it is also apparent that the peak numbers in 2000s are equivalent to the peak numbers in 1990s. A notion of longterm fall in numbers rather than wide natural fluctuations in numbers of feeding flamingos in the lake is not based on scientific reality.

Likewise, when studies were done on the cause of the mass die-off in Lake Bogoria in 1999, it was found to have a natural, not an anthropogenic cause (Krienitz at al., 2003). The flamingos had ingested the remains of toxic cyanobacteria that constitute part of the population of the microbial mats that had bloomed to form a floor to the fresh water thermal spring areas about the lake edge. There the mats are dominated by thermally tolerant species; Phormidium terebriformis, Oscillatoria willei, Arthrospira subsalsa and Synechococcus bigranulatus. The influence of cyanotoxins in the deaths of the birds is reflected in autopsies which revealed: (a), the presence of hot spring cyanobacterial cells and cell fragments (especially Oscillatoria willei), and high concentrations of the cyanobacterial hepato- and neurotoxins in flamingo stomach contents and faecal pellets; (b), observations of neurological signs of bird poisoning - birds died with classic indications of neurotoxin poisoning - the ophistotonus behaviour (neck snapped back like a snake) of the flamingos in the dying phase, and the convulsed position of the extremities and neck at the time of death. Cyanobacterial toxins in stomach contents, intestine and faecal pellets were 0.196 g g-1 fresh weight (FW) for the microcystins and 4.34 g g-1 FW for anatoxin-a. Intoxication with cyanobacterial toxins probably occurred via uptake of detached cyanobacterial cells when birds come daily to the springs to drink and wash their feathers after an overnight feeding session in the saline waters of the lake proper.

When heavy metal studies were undertaken in Nakuru lake sediments, the amount of heavy metals (Cd, Cr, Cu, Hg, Ni, Pb, Zn) were found to be in the typical range of metals in sediments in lakes worldwide. The exception is Cd, which is elevated and can perhaps be ascribed to anthropogenic activity (Svengren, 2002). All other metals are present at low levels, especially if one considers that Lake Nakuru lies within a labile catchment where the bedrock is an active volcanogenic-magmatic terrane.

Nearby Lake Magadi is also characterised by seasonal freshening, high productivity levels and bright red waters. In this case, the colour comes from haloalkaliphilic archaea, not cyanobacteria. Archaeal species belonging to the genera Natronococcus, Natronobacterium, Natrialba, Halorubrum, Natronorubrum and Natronomonas, all occur in soda brines of Lake Magadi. Lake centre brines where this biota flourishes is at trona/halite saturation with a pH up to 12. Stratified moat waters around the trona platform edge are less chemically extreme and moat bottom sediments preserve elevated levels of organics (≈6-8%). Lake Magadi also harbours a varied anaerobic bacterial community in the moat waters, including cellulolytic, proteolytic, saccharolytic, and homoacetogenic bacteria (Shiba and Horikoshi, 1988; Zhilina and Zavarzin, 1994; Zhilina et al., 1996). When the homoacetogen Natroniella acetigena was isolated from this environment, its pH growth optimum was found to be 9.8–10.0, and it continued to grow in waters with pH up to 10.7 (Zhilina et al., 1996).


Flamingo numbers track feast and famine

Rise and fall of lake levels, drastic changes in salinity, a periodically stressed biota, and a lack of predictability in water character are endemic to life in saline ecosystems (cycles of “feast or famine”). Natural variations in hydrochemistry control the number of feeding flamingos in Nakuru and Bogoria. In general, sufficient base-line scientific data in these schizohaline ecosystems is not yet available and so accurate determinations of the relative import of increased human activity versus natural environmental stresses on longterm bird numbers are not possible.


Regionally, salinities in the east African rift valley lakes range from around 30-50‰ total salts (w/v) in the more northerly lakes in the rift (Figure 3a; Bogoria, Nakuru, Elmenteita, and Sonachi) to trona and halite saturation (>200‰) in lakes to the south (Lakes Magadi and Natron). Yet across this salinity range a combination of high ambient temperature, high light intensity and a continuous resupply of CO2, makes some of these soda lakes amongst the highest in the world in terms of their seasonal planktonic biomass (Grant et al., 1999) and also places them among the world’s most productive ecosystems (Figure 8; Melack and Kilham, 1974). Organic production is periodic, and pulses of organic product periodically swamp the ability of the decomposers and so accumulate as laminites in the perennial water-covered areas of some lake centres.

Less-alkaline lakes in the rift valley are dominated by periodic blooms of cyanobacteria, while the hypersaline lakes, such as Magadi, can on occasion support blooms of cyanobacteria, archaea and alkaliphilic phototrophic bacteria (Jones et al., 1998). Halotolerant and halophilic biota living in the variably saline and layered water columns constitute small-scale “feast or famine” ecosystems, which at times of “feast” are far more productive than either tropical seagrass meadows or zones of marine upwelling (Figure 8).

The “flamingo connection” across the African Rift Lakes supports a general observation that short periods of enhanced organic productivity are followed by episodes of lessened productivity in various schizohaline saline lake and seaway waters worldwide. both past and present (Warren, 2011). It reflects the general principle that increased environmental stress favours the survival of a few well-adapted and specialised halotolerant species. This biota is well adapted to the feast and famine life-cycle that exists in most saline depressions and means their numbers are subject to wide fluctuations tied to wide fluctuations in a saline lakes hydrochemistry (schizohaline waters). This same general principle of schizohaline ecosystem adaption is clearly seen in the periodic decrease in invertebrate species (grazers and predators) numbers with increasing salinity in the carbonate lakes of the Coorong of Southern Australia, where the only metazoan to remain alive in waters with salinities more than 200‰ are the southern hemisphere brine shrimp (Paratemia zietziania). It is seen in population fluctuations of the motile alga Dunaliella sp. in the Dead Sea, and in the fluctuating purple bacterial communities of Lake Mahoney in Canada (Warren, 2011). All these examples underline a general principle of “life will expand into the available niche” a paradigm that in the cases we have discussed is driven by fluctuating salinities inherent to saline-tolerant and saline-adapted ecosystems.


Bildstein, K. L., C. B. Golden, B. J. McCraith, and B. W. Bohmke, 1993, Feeding Behavior, Aggression, and the Conservation Biology of e: Integrating Studies of Captive and Free-ranging Birds: American Zoologist, v. 33, p. 117-125.

Childress, B., S. Nagy, and B. Hughes, 2008, International Single Species Action Plan for Conservation of the Lesser Flamingo (Phoenicopterus minor): CMS Technical Series No. 18, AEWA Technical Series No. 34, Bonn, Germany, 59 pp.

Gould, S. J., 1987, The Flamingo's Smile: Reflections in NAtural History: New York, W. W. Norton and Company.

Grant, S., W. D. Grant, B. E. Jones, C. Kato, and L. Li, 1999, Novel archaeal phylotypes from an East African alkaline saltern: Extremophiles, v. 3, p. 139-145.

Jenkin, P. M., 1957, The Filter-Feeding and Food of Flamingos (Phoenicopteri): Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, v. 240, p. 401-493.

Jones, B. E., W. D. Grant, A. W. Duckworth, and G. G. Owenson, 1998, Microbial diversity of soda lakes: Extremophiles, v. 2, p. 191-200.

Kirkland, D. W., and R. Evans, 1981, Source-rock potential of evaporitic environment: Bulletin American Association of Petroleum Geologists, v. 65, p. 181-190.

Kotut, K., and L. Krienitz, 2011, Does the potentially toxic cyanobacterium Microcystis exist in the soda lakes of East Africa?: Hydrobiologia, v. 664, p. 219-225.

Krienitz, L., A. Ballot, K. Kotut, C. Wiegand, S. Pütz, J. S. Metcalf, G. A. Codd, and S. Pflugmache, 2003, Contribution of hot spring cyanobacteria to the mysterious deaths of Lesser Flamingos at Lake Bogoria, Kenya: FEMS Microbiology Ecology, v. 43, p. 141-148.

Krienitz, L., and K. Kotut, 2010, Fluctuating algal food populations and the occurrence of Lesser Flamingos (Phoeniconaias minor) in three Kenyan rift valley lakes: Journal of Phycology, v. 46, p. 1088-1096.

Melack, J. M., and P. Kilham, 1974, Photosynthetic rates of phytoplankton in East-African alkaline saline lakes: Limnology and Oceanography, v. 19, p. 743-755.

Melchor, R. N., M. C. Cardonatto, and G. Visconti, 2012, Palaeonvironmental and palaeoecological significance of flamingo-like footprints in shallow-lacustrine rocks: An example from the Oligocene-Miocene Vinchina Formation, Argentina: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 315-316, p. 181-198.

Schagerl, M., A. Burian, M. Gruber-Dorninger, S. O. Oduor, and M. N. Kaggwa, 2015, Algal communities of Kenyan soda lakes with a special focus on Arthrospira fusiformis: Fottea, v. 15, p. 245-257.

Scott, J. J., R. W. Renaut, L. A. Buatois, and R. B. Owen, 2009, Biogenic structures in exhumed surfaces around saline lakes: An example from Lake Bogoria, Kenya Rift Valley: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 272, p. 176-198.

Scott, J. J., R. W. Renaut, and R. B. Owen, 2012, Impacts of flamingos on saline lake margin and shallow lacustrine sediments in the Kenya Rift Valley: Sedimentary Geology, v. 277-278, p. 32-51.

Shiba, H., and K. Horikoshi, 1988, Isolation and characterization of novel anaerobic, halophilic eubacteria from hypersaline environments of western America and Kenya: In: Proceedings of the FEMS symposium—The microbiology of extreme environments and its biotechnological potential, Portugal, p. 371-373.

Simmons, R. E., 1995, Population declines, viable breeding areas and management options for flamingos in southern Africa: Conservation Biology, v. 10, p. 504-514.

Svengren, H., 2002, A study of the environmental conditions in Lake Nakuru, Kenya, using isotope dating and heavy metal analysis of sediments: Masters thesis, Dept. Structural Chemistry, University of Stockholm, Swededn.

Tuite, C. H., 2000, The Distribution and Density of Lesser Flamingos in East Africa in Relation to Food Availability and Productivity: Waterbirds: The International Journal of Waterbird Biology, v. 23, Special Publication 1: Conservation Biology of Flamingos, p. 52-63.

Vareschi, E., 1978, The ecology of Lake Nakuru (Kenya). I. Abundance and feeding of the lesser flamingo: Oecologia, v. 32, p. 11-35.

Vareschi, E., 1982, The ecology of Lake Nakuru (Kenya). III. Abiotic factors and primary production: Oecologia, v. 55, p. 81-101.

Vonshak, A., 1997, Spirulina: Growth, physiology and biochemistry, in A. Vonshak, ed., Spirulina platensis (Arthrospina): Physiology, cell-biology and biotechnology: London, Taylor and Francis, p. 43-65.

Warren, J. K., 1986, Shallow water evaporitic environments and their source rock potential: Journal Sedimentary Petrology, v. 56, p. 442-454.

Warren, J. K., 2011, Evaporitic source rocks: mesohaline responses to cycles of “famine or feast” in layered brines, Doug Shearman Memorial Volume, (Wiley-Blackwell) IAS Special Publication Number 43, p. 315-392.

Zhilina, T. N., and G. A. Zavarzin, 1994, Alkaliphilic anaerobic community at pH 10: Curr. Microbiol., v. 29, p. 109-112.

Zhilina, T. N., G. A. Zavarzin, F. Rainey, V. V. Kevbrin, N. A. Kostrikina, and A. M. Lysenko, 1996, Spirochaeta alkalica sp nov, Spirochaeta africana sp nov, and Spirochaeta asiatica sp nov, alkaliphilic anaerobes from the Continental soda lakes in Central Asia and the East African Rift: International Journal of Systematic Bacteriology, v. 46, p. 305-312.



Recent Posts


mirabilite Sulphate of potash recurring slope lines (RSL) trona Deep Evaporite-source rock association Jefferson Island salt mine carbon oxygen isotope cross plots Sumo extrasalt crocodile skin chert Belle Isle salt mine mummifiction dihedral angle 18O carnallitite Lake Peigneur deep meteoric potash carbon cycle wireline log interpretation sinkhole jadarite Turkmenistan Zabuye Lake gas outburst collapse doline hydrothermal karst CaCl2 brine H2S allo-suture capillary zone organic matter gem nitrogen Stebnyk potash Atlantis II Deep evaporite-metal association venice hectorite potash ore sedimentary copper mass die-back source rock nacholite kainitite CO2 Realmonte potash Dallol saltpan antarcticite bischofite sulfate hydrological indicator potash ore price intrasalt stevensite High Magadi beds halotolerant Mesoproterozoic Zaragoza Lake Magadi Platform evaporite flowing salt sinjarite silica solubility salt leakage, dihedral angle, halite, halokinesis, salt flow, well logs in evaporites cryogenic salt HYC Pb-Zn SOP sodium silicate North Pole Corocoro copper lapis lazuli nuclear waste storage well log interpretation Musley potash water on Mars basinwide evaporite Dead Sea caves waste storage in salt cavity auto-suture McArthur River Pb-Zn meta-evaporite potash gas in salt 13C enrichment salt tectonics salt karst salt mine hydrohalite vadose zone anomalous salt zones intersalt vanished evaporite Great Salt Lake Paleoproterozoic Oxygenation Event lithium brine Crescent potash sepiolite DHAL oil gusher snake-skin chert salt suture Neutron Log subsidence basin mine stability ancient climate Neoproterozoic salt periphery Badenian Kara bogaz gol Pangaea cauliflower chert Hyperarid Ingebright Lake salt trade evaporite-hydrocarbon association seal capacity Ethiopia brine evolution Salar de Atacama Pilbara lithium battery CO2: albedo Ripon Archean Five Island salt dome trend stable isotope alkaline lake blowout Kalush Potash rockburst hydrothermal potash natural geohazard halite black salt NPHI log Deep seafloor hypersaline anoxic lake sulfur NaSO4 salts water in modern-day Mars GR log Koppen climate African rift valley lakes doline magadiite SedEx circum-Atlantic Salt Basins Stebnik Potash Clayton Valley playa: Proterozoic sulphate palygorskite Hell Kettle Hadley cell: perchlorate sulphur lithium carbonate solikamsk 2 gypsum dune Muriate of potash anthropogenically enhanced salt dissolution Precambrian evaporites hydrogen evaporite RHOB well blowout Gamma log dark salt marine brine supercontinent Catalayud salt ablation breccia Koeppen Climate Magdalen's Road Messinian Karabogazgol 13C bedded potash eolian transport Boulby Mine Density log Neoproterozoic Oxygenation Event dissolution collapse doline evaporite karst Quaternary climate MVT deposit chert lunette Lomagundi Event salt seal Danakhil Depression, Afar knistersalz lot's wife saline clay methane zeolite York (Whitehall) Mine astrakanite halokinetic causes of glaciation Warrawoona Group halite-hosted cave 18O enrichment geohazard lazurite Weeks Island salt mine freefight lake Ure Terrace climate control on salt Mega-monsoon authigenic silica base metal epsomite Red Sea gassy salt evaporite dissolution tachyhydrite Dead Sea karst collapse namakier MOP silicified anhydrite nodules K2O from Gamma Log