Geoecological evolution of New Zealand’s only inland salt lake

ABSTRACT

The Sutton Salt Lake is a rare occurrence of an inland evaporative saline lake on a small landmass with a temperate maritime climate. The lake lies on a bed of impermeable loess silt in a 2 ha depression between tor ridges of Otago Schist in inland east Otago. A rim of salt-tolerant plants (halophytes) occurs around parts of the lake margin. The vegetated lake margin has encroached laterally on to the lake bed by 5–30 m over the past 50 years. Salinity in the lake is dominated by Na and Cl (molar ratio ∼1) that has been concentrated by annual cycles of evaporation and replenishment by rain with marine aerosols. Drying lake margin sediments lose NaCl via rain-driven recycling back into the lake, with the formation of a salt crust that contains Ca, Mg and Na carbonate minerals, gypsum, and remnant halite. The lake and lake sediment chemistry are affected by carbonate dissolution and deposition, and more than 10-fold Ca–Mg carbonate supersaturation in lake waters contributes to elevated water pH (8.9–9.3). Geoecological evolution of the lake margin involves the encroachment of exotic weeds behind active halophytic habitats as dilute surface water lowers salinity.

KEYWORDS:

• Evaporation
• pH
• halite
• carbonate
• loess
• schist
• halophyte

Introduction

Inland salt lakes occur in rain shadows around the world where evaporation balances incoming precipitation and drainage inputs (Risacher et al. 2003; Lowenstein and Risacher 2009; Risacher and Fritz 2009; Zheng and Liu 2009). Many of these inland saline settings have unique endemic biota that are adapted to such saline environments (Zhao et al. 2005; Carling et al. 2013; Dunham et al. 2020; Martinez et al. 2021). Most of these inland saline lakes are in continental areas, far from the sea. In contrast, the South Island of New Zealand, as a small landmass in a temperate maritime setting, is different from these continental landmasses but does have one inland salt lake, Sutton Salt Lake, in the southeast (Figure 1A,B). This lake also has endemic salt-tolerant biota, including salt-tolerant plants (halophytes) whose habitats are the principal target of this geoecological study.

Figure 1. Location, with geological and topographic settings for the Sutton Salt Lake. A, Otago Schist forms basement and source of lake sediment in southern South Island. B, DEM of eastern Otago, showing the location of the lake between the Barewood plateau and the Strath Taieri basin. C, Vertical aerial photograph (from Google Earth, March 2021) of the lake (white; fully evaporated) in a depression between ridges of schist tors. D, Vertical aerial photograph (from Google Earth, December 2021) of the lake with partial evaporation. Whiter shore-parallel rings on west side are slightly higher ridges on lake bed with more advanced drying and salt accumulation.

Evaporative saline environments occur throughout inland Otago in the southern South Island (Figure 1A,B) within the rain shadow that has formed in the lee of the Southern Alps since the Pliocene (Allen and McIntosh 1997; Chamberlain et al. 1999; Craw et al. 2022a, 2023a). These saline sites are typically small (hectare scale) but they are mineralogically and geochemically anomalous compared to the surrounding areas which are dominated by the mineralogy of the underlying Otago Schist basement (Figure 1A,B; Craw et al. 2022a, 2022b). This anomalous saline mineralogy and geochemistry arises from a combination of NaCl-rich salts derived from marine aerosols in rain and minerals formed as a result of the interaction of saline shallow groundwater with schist-derived debris (Craw et al. 2022a, 2022b). These processes, combined with high evaporation rates in the semiarid climate, lead to the formation of bare surface substrates (salt pans) that are typically sodic and alkaline (Allen et al. 1997; Allen and McIntosh 1997; Craw et al. 2022a, 2022b, 2023a, 2023b). The geochemically extreme bare substrate environments host salt-tolerant plant species, some of which are rare and endangered, and several formal reserves have been established to protect the sites (Allen et al. 1997; Allen and McIntosh 1997; Allen 2000; Rogers et al. 2000; Rufaut et al. 2018).

Sutton Salt Lake (Figure 1A,B; Bayly 1967; Allen and McIntosh 1997; Craw and Beckett 2004) receives salts via marine aerosols in rain in a similar manner to the dry salt pans (above). However, the physical setting ensures that there is generally excess water accumulation in a depression (Figure 1C,D; Bayly 1967; Craw and Beckett 2004). The enhanced salinity of the lake has resulted from long-term annual evaporation cycles (Craw and Beckett 2004). Preliminary geological, topographic and geochemical descriptions of the site were compiled by Craw and Beckett (2004). The present study is intended to extend that work in the context of halophytic plant habitats around the lake margin, and the evaporative mineralogy associated with these habitats. We document the evaporite mineralogy, the marginal sedimentation, and the plants established on the marginal sediments, in the context of lake chemistry. The Sutton site has been formally reserved, partly for the preservation of the halophytes around the lake margin, and this paper provides geoecological context for on-going monitoring of the botanical values at the site. We investigate the geoecological aspects at the Sutton site with reference to geoecological features of other Otago saline sites farther inland and on the coast near Dunedin (Figure 1A,B; Rufaut et al. 2018; Craw et al. 2022b; Rufaut and Craw 2023). For these purposes, our focus is on surficial features and processes involving the ephemeral lake water and the immediately adjacent sedimentary shore area, some of which is exposed periodically and temporarily.

General setting

Sutton Salt Lake is the easternmost of the inland NaCl-rich evaporative sites in Otago, and lies ∼50 km from the east coast (Figure 1A,B). The nearest inland dry salt pans are in the Patearoa area of the Maniototo basin, 30 km to the northwest, and these host some endemic halophytic plants (Figure 1B; Allen and McIntosh 1997; Craw et al. 2022b). Substrates with elevated NaCl form on coastal clifftops and support halophytic coastal turf (Rogers 1999; Rogers and Wiser 2010), although surficial salt precipitates are rare because of frequent rain events (Rufaut and Craw 2023). Hence, the Sutton site lies within a regional transition between damp coastal settings and dryland environments farther inland. Evaporative precipitates form in mining areas at Macraes in a similar regional setting to the Sutton lake (Figure 1B), although the Macraes deposits are dominated by rock-derived carbonate and sulphate minerals with only minor marine aerosol components (Weightman et al. 2020; Rufaut et al. 2022).

Sutton Salt Lake has formed within a large area of exposed Otago Schist basement where the broad uplands (400–500 m above sea level) of the Barewood plateau slope northwards to the Strath Taieri basin (Figure 1B,C). The lake lies at 250 m above sea level in an erosional depression between prominent northwest-trending ridges of schist tors that are controlled by the underlying schist structure (Figure 1C,D). The lake is confined to this depression by schist outcrops and occupies ∼2 hectares when full (∼0.5 m deep). A thin (typically <1 m) layer of schist-derived fine sediment forms the lake bed (Craw and Beckett 2004). This sediment was derived from aeolian dust (loess) that has been washed into the depression by ephemeral surface streams (Craw et al. 2022a), and includes some particulate ferric oxyhydroxide washed from oxidising schist outcrops. The lake area is managed as a reserve and has been fenced to exclude grazing stock.

The lake sediment is dominated by silt and clay with minor sand (Craw and Beckett 2004), and forms a relatively impermeable lake bed that prevents accumulating rain water from discharging through underlying fractured schist. Salt accumulation has occurred via annual cycles of evaporation and refilling by aerosol-bearing rain for ∼20 ka (Jacobson et al. 2003; Craw and Beckett 2004; Craw et al. 2022a). From this perspective, the lake is unique in the area where numerous other similar depressions host pools that have sufficient water input and discharge to limit build-up of evaporative salts (Figure 1C,D).

The lake area receives ∼500 mm annual rainfall, with surface evaporation of at least 700 mm/year driven in part by frequent strong winds in this exposed low-relief landscape (Craw and Beckett 2004; NIWA 2021). Mean annual temperature is ∼12°C, with winter temperatures occasionally down to −10°C and summer temperatures rarely exceeding 30°C (Craw and Beckett 2004; NIWA 2021). The lake water is fully evaporated in dry summers to expose the whole lake bed (Figure 1C), while in some wetter years the lake merely partially contracts to expose large areas of damp mudflats and only minor dry shoreline (as depicted in Figure 1D). This latter situation is of most interest for studying surface features, and most observations for this paper were collected at such times (Figure 2A–E). Historical images to show changes with time are presented in Figure 3A,B and Supplementary Figs. 1–3.

Figure 2. Geoecological features of the Sutton lake shore area. A, EC in mS/cm. B, Zonation of salt grasses at the upper limit of lake filling, with drying saline lake bed sediments in foreground. C, View through shallow lake water of gas discharge features in bottom sediments. D, Biofilm (green) around evolving desiccation cracks on drying saline lake sediments. E, Close view of dry biofilm-covered lake sediment, with evaporative salt beneath the biofilm, in gaps in biofilm, and in some voids.

Figure 3. Evolution of the vegetated marginal flat on the western side of Sutton Salt Lake. A, Aerial view of the lake in 1945 (as in Supplementary Fig. 3a). B, Portion of the western margin of the lake (1945; dashed box in A), enlarged to the limit of resolution to show possible halophyte rim that forms a dark line at base of white schist outcrops on the lake edge. A small marginal flat has formed in the innermost part of the embayment. C, Same area as in B, viewed in 2016 (Google Earth) when lake was almost full. The positions of historic (1945) marginal flat and halophyte rim are indicated, in relation to the modern positions of the lake shore (white dry clay-rich edge), halophyte rim (dark green), and larger vegetated marginal flat of varying width. D, Summary sketch of a botanical transect across the marginal flat in c, measured in 2023. EC (mS/cm) and pH data for the substrates and lake water were measured in June 2023 when the lake was full. E, Photograph of part of a botanical quadrat (1 m steel outline; 50 cm yellow string internal divisions) at the lake-shore edge of the halophyte rim, in March 2023.

Methods

The lake site is highly variable because of the vagaries of weather each year (e.g. Figure 1C,D). For this study, we have compiled previously published geochemical and botanical observations and data, and augmented this information with several visits to the site in summer and autumn of 2023 when the lake filled approximately half the depression, and then as the lake filled in early winter. Additional data were collected in a similar manner to that presented in our previous publications on saline sites, to enable direct comparisons.

Lake water and water in marginal pools had pH and electrical conductivity (EC) measured directly in the field with access via wading in shallow water, at temperatures between 12 and 16°C. Measurements were made with a portable Oakton PC450 meter with separate pH and EC probes that has automatic temperature compensation and was calibrated in laboratory conditions (18–20°C) immediately before field work. Manufacturer-supplied standard pH 4 and 7 solutions were used for calibration along with a pH 10.01 (at 25°C) standard solution, consisting of sodium carbonate and sodium bicarbonate, from Rocker Scientific Co, Taiwan. Estimates of pH and EC of wet and dry lake surface sediments were obtained in the field (5–16°C) on slurries made with 20 ml of solids and 30 ml of added distilled water, in the same manner as at other Otago saline sites (Craw et al. 2022a, 2022b, 2023a, 2023b; Rufaut and Craw 2023). The slurries were stirred thoroughly over ∼5 min to ensure that all solid material had been disaggregated. Subsurface substrates beneath vegetated marginal flat areas were collected from shallow holes (<30 cm) dug to hard schist basement in June 2023 when the lake was full (<8°C). EC and pH data were collected on slurries made from samples at 5 cm depth intervals, and from waters that accumulated in the base of the sample holes.

Two different samples of lake water were collected in April 2023 for observed evaporation in laboratory conditions. Two litres of water were collected for each sample, and the samples were left to stand in sealed bottles on a laboratory bench for three days to settle sediments. Then exactly one litre of each sample was decanted into open one-litre containers with graduated volume scales. These were left to evaporate on the laboratory bench at 16–18°C for five months. The pH and EC of the waters were measured at the time of collection and at intervals as the water volumes decreased to the limits of practicality (<100 ml solution; EC beyond instrument range of 200 mS/cm). Thin precipitates coated container walls and were too difficult to extract, but most precipitates accumulated on the bottoms of the containers. Samples of these container-bottom evaporative precipitates were extracted for examination with a stereoscopic light microscope. These samples were taken while the precipitates were still damp with evaporative brine, to improve observational optics. Once the containers were fully dry, a precipitate sample was extracted to examine with scanning electron microscopy (SEM) by methods outlined below.

Samples of nearly dry lake sediment, with developing desiccation cracks and incipient salt encrustations were taken to the laboratory and left to dry fully to develop hard surface crusts. Centimetre-scale sub-samples from these crusts were mounted on aluminium stubs for examination with SEM. The mounted sub-samples were made electron-conductive with carbon tape and carbon paint, followed by standard carbon vapour coating under vacuum. SEM observations were made with a Zeiss Sigma VP (variable pressure) FEG-SEM fitted with an Oxford Instruments X-Max 20 mm silicon drift detector at the Otago Micro and Nanoscale Imaging (OMNI), University of Otago. Backscatter electron imaging (BSE) and qualitative energy dispersive spectrometry (EDS) analyses were carried out using an accelerating voltage of 15 kV and an aperture of 120 μm. Semiquantitative elemental compositions were obtained by spot analyses (∼3 µm) and we note that carbonate minerals had sufficiently large carbon and oxygen signals to distinguish their analyses from the carbon coating. Element maps were created of selected areas to document variations in fine-grained precipitates.

Geochemical modelling was carried out on various real and generalised water compositions to investigate relationships between waters and evaporative minerals, assuming inorganic chemical processes predominate. We used the software suite The Geochemists Workbench® (www.gwb.com), with the database thermo.hmw.dat (Harvie-Moller-Weare activity model) for the concentrated evaporation waters where appropriate. The program ACT2 was used to calculate model stability diagrams for minerals in relation to dissolved species. The program SpecE8 was used to calculate the chemical saturation of various minerals in solutions, resulting in values of saturation index, log (Q/K), where K is the equilibrium constant from the database, and Q is calculated from observed water compositions. The program REACT was used to model the precipitation of principal minerals from waters with progressive evaporation without re-dissolution, in contact with atmospheric CO₂(g). The modelling software assumes chemical equilibrium, but dolomite is notorious for not precipitating from saturated solutions, possibly for kinetic reasons (Arvidson and MacKenzie 1999), and we suppressed dolomite precipitation in our models. Some of our models also suppressed calcite and magnesite precipitation to achieve predicted levels of supersaturation observed in lake waters.

Results

Lake-bed sediment and vegetation

Mineralogy and geochemical compositions of the lake sediments are described by Craw and Beckett (2004) and are not repeated here, where we focus on surficial features and sediment distribution in relation to halophyte habitat. The eastern and southern lake margins are largely defined by rock outcrops (Figure 1D) with only minor exposed sediments when the lake is full. In contrast, the sediment floor is slightly thicker (∼20–30 cm) on the northern and western sides, so that the contracting lake is asymmetrically positioned in the depression (Figure 1D). Consequently, the northern and western shores are more frequently and more extensively exposed than the southeastern shore (Figure 1D). These northern and western sides are defined by sediments that form marginal flat strips 2–20 metres wide below schist outcrops (Figures 1D and 2A). The marginal flats are variably swampy, with ephemeral water pools (Figure 2A). The sediment surface gradients from the base of the rocky edges and across the marginal flats towards the deepest parts of the lake are very low (∼0.001). These slopes are almost imperceptible without the lake water edge as a reference, and are further disguised by the marginal vegetation cover (Figure 2A).

The marginal flats are well-vegetated, mainly with exotic grasses reflecting the surrounding agricultural land-use (Figure 2A) and the site’s pre-reserve use for agricultural grazing. However, salt grasses form a band around the shoreline, with distinct zonation: Puccinellia walkeri grows closest to the water, and Puccinellia fasciculata grows farther from the shore (Figures 2A,B and 3C–E). The native herb Oxybasis ambigua occurs locally on the flat strip, typically surrounded by exotic weeds, although scattered individuals of this species also occur locally with P. walkeri (Figure 3E). The oldest sediments, farthest from the lake shore (Figures 3A–C and 4A) have a well-defined organic soil layer (5–10 cm) developing on the surface, and abundant pasture grass roots extend almost to the base of the sediment profile. These sediments have a purple tinge indicative of weathering-related oxidation. In contrast, sediments farther out on the flat (Figure 4B,C) are grey with only minor organic material at the surface, although some roots extend to the base of these sediments as well. Deep root extension (to at least 30 cm) occurs beneath the halophytic grass farthest from the schist outcrops, on the active lake margin (Figures 3E and 4C).

Figure 4. Photographs of substrates below the vegetated western margin of Sutton Salt Lake (Figure 3C). Each profile was dug to hard schist bottom. Substrate EC (mS/cm) and pH were measured on slurries, and water EC and pH were measured on water that accumulated in the bottom of the holes after digging. Data collected when the lake was full in June 2023. A, Substrate profile in the oldest part of the margin (e.g. 1945, Figure 3A–C). B, Substrate profile through pasture grass, ∼2 m from schist outcrop, on the this extended marginal flat (formed since 1974; Supplementary Fig. 3b–d). C, Substrate profile beneath the active lakeward edge of the halophyte rim (Figure 3C–E).

The lake is home to flocks of water birds, and commonly has abundant algae in the water in summer (e.g. green tinge in Figure 1D), although this component appears to lessen towards autumn and winter. Algal biofilms form on the sediment in shallow water and on drying surfaces (Figure 2B–E). Lake sediments therefore have living and decaying organic components (2–3%; Craw and Beckett 2004) within the predominant silicate silt and mud. Organic material includes abundant ostracods (Diacypris sp.; Bayly 1967) that have calcareous body parts. Gas discharge features, including sediment disruption mounds and bubbles in overlying water, are widespread in shallow water (Figure 2C), and the whole site has a distinctly sulphurous odour from these gases on calm days.

Salt crusts

Sediments exposed when the lake contracts become progressively drier, with the development of polygonal desiccation cracks (Figures 2B,D, 3E and 5A). The cracks extend up to 10 cm below the drying surface, but the sediments remain moist or water-saturated below that level as the water table remains close to the surface over the whole site. Salts crystallise during the evaporative drying process, initially on subtly elevated surfaces and ridges (mm-cm scale) that form as a result of wind-driven sediment redistribution (Figures 1D, 2A,B and 5A). With further lake contraction, dry surfaces develop progressively over more extensive areas (tens of m²; Figure 1D) or even the whole lake bed (Figure 1C).

The salt crusts that develop during the drying process form hard brittle surface zones up to 3 cm thick (Figure 5B). The silt-dominated sediment has a matrix of clay, predominantly kaolinite, that binds the sediment structure on drying, and this combination of materials creates the white appearance that characterises the dry surface (Figures 1C,D and 2B). The crust is also partially cemented by evaporative salts, although these are not visibly prominent (Figure 5B). Salts form a thin (∼0.1 mm) white coating on dry exterior surfaces (Figure 5B–D) but these salts are visibly subtle unless the hosting sediments are only partially dry (Figure 5A). Salts are even more obscured by the presence of biofilms, as the encrustations develop beneath the biofilm layer (Figure 2E).

Figure 5. Photographs of salt crust on surface of the exposed lake bed. EC in mS/cm. A, Drying surface crust above damp sediment. White patches are initial salt deposits. B, Detailed view of a section through dry crust, showing internal structure, including abundant voids. C, Vertical close view of salt surface. White, shiny and translucent crystals are predominantly calcite and halite, and brown particles are clasts of Fe-stained quartz. D, Oblique view of broken edge (bottom) of salt crust, with exposed voids.

The hard salt crusts preserve fine (mm-scale) layering within the surface sediments (Figure 5B). Some of these layers contain voids with a wide variety of shapes, from rough and irregular (bottom of Figure 5B) to ellipsoidal or spheroidal shapes with smooth interiors (top of Figure 5B). The latter type can locally penetrate to the exterior surface (Figures 2E and 5D). At least some of these voids in the crusts are relics of voids and gas discharge mounds that are visible in surface sediments beneath shallow lake water (Figure 2C) and possibly beneath some biofilm (Figure 2D).

Salinity and pH of waters and surfaces

Salinity, as indicated by electrical conductivity (EC) measurements of Sutton lake waters ranges from ∼20–50 mS/cm and is highly variable depending on the extent of lake evaporation (Figure 6A,B; Table 1). These lake EC values are 1–2 orders of magnitude higher than waters and slurries from the base of wetlands immediately surrounding the lake, which have typical EC of 0.1–0.5 mS/cm (Figures 1D, 2A, 4A,B and 6A). The lake EC range (Figure 6A,B) is approximately equivalent to salinity of 25–35 parts per thousand (ppt), or a range from slightly dilute seawater composition to concentrated seawater compositions compared to typical New Zealand oceanic waters (30–40 mS/cm; 30–33 ppt; Tyler et al. 2017). This Sutton lake EC range is at the upper end of slurry measurements of dry substrate EC of inland salt pans and salt pan leachates (Figure 6A,B). Likewise, the Sutton lake waters are substantially more saline than typical substrate slurry measurements on saline coastal turf sites, although a few coastal data do overlap Sutton data (Figure 6C).

Figure 6. General chemical comparisons between Sutton Salt Lake and other Otago saline sites. A, Electrical conductivity (EC) and pH of Sutton lake waters compared to slurry measurements from salt pans, laboratory leachates from salt pans, and rainwater runoff from salt pans of Central Otago (Alexandra area, Figure 1A; Patearoa area, Figure 1B; Craw et al. 2022a, 2022b, 2023a, 2023b). B, Range of EC and dissolved Na concentrations in Sutton lake waters, compared to laboratory leachates from Central Otago salt pans. Red curve is model NaCl-EC relationship (Sposito 2016). C, EC and pH of Sutton lake waters compared to slurry measurements from saline coastal turf near Dunedin (Figure 1B; Rufaut and Craw 2023). D, Surface water and slurry measurements taken at the Sutton lake site on a single day in May 2023 (12°C) while lake was low. E, Slurry data for subsurface substrates on vegetated Sutton lake marginal flat (Figure 2B) when lake was full in June 2023 (<8°C), and data for water that accumulated in sample holes. F, EC and pH evolution of two one-litre Sutton lake water samples that were evaporated in a laboratory (16–18°C) over 5 months.

Table 1. Representative water analyses for Sutton Salt Lake and its regional context.

Water source	pH	EC	K^+	Na^+	Ca^++	Mg^2+	Cl^-	SO4^2-	alk	Na/Cl
		mS/cm	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	Molar
Typical sea water^a	8.2	35	399	10,773	412	1294	19,344	2712	142	0.85
Sutton lake, dilute water 1966^b	7.7	20.7	230	5300	40	160	8980	na	430	0.90
Sutton lake, dilute water 1998^c	9.1	na	111	5070	64	176	7610	250	747	1.01
Sutton lake, concentrated water 2003^c	9.1	na	642	12,730	141	458	21,100	920	509	0.92
Sutton lake pore water 1998^c	8.1	na	113	3660	128	58	6280	250	100	0.89
Central Otago alkaline pan leachate^d	10.5	2.2	9	1020	2	1	209	15	2190	7.43
Central Otago alkaline pan leachate^d	8.1	6.8	44	3200	52	317	3800	3320	55	1.28
Dilute schist groundwater, Macraes^e	7.4	na	1	9	23	5	6	2	105	2.28
Evolved schist groundwater, Macraes^e	8.2	1.8	8	26	200	130	11	730	400	3.60

Notes: Alkalinity (alk) is all presented as HCO₃^–; na = not analysed.

Sources: ^aWilson (1975); Babel and Schreiber 2014; ^bBayly (1967); ^cCraw and Beckett (2004); ^dCraw et al. (2023b); ^eWeightman et al. (2020).

To investigate inherent variability in the lake, measurements were taken on a single day in May 2023 at different points while the lake was filling half the depression, and these show a range in EC from 34–46 mS/cm (Figure 6D). The highest of these EC values were measured in shallow (∼5 cm) waters near the western shore. In contrast, pools of water in the vegetated marginal strip (e.g. Figure 2A) generally have EC <10 mS/cm and some of these pools have EC < 1 mS/cm (Figures 4A,B and 6D,E). Experimental evaporation of two samples of lake water, with initial EC values of 22 and 32 mS/cm, showed progressive increase in EC that approximately paralleled the increases in EC observed in the lake at various stages of evaporation (Figure 6F).

The pH of the lake water was ∼9 during the present study and during the study by Craw and Beckett (2004), although Bayly (1967) reports pH as low as 7.7 (Figure 4A,D; Table 1). The lake water pH is in the middle of the range of pH values observed on inland salt pans which are highly variable on a regional scale (Figure 6A). The lake pH is also similar to that of saline coastal turf on calcareous substrates, but distinctly higher than pH of saline coastal turfs on basaltic substrates (Figure 6C). Marginal water pools at the Sutton lake site have lower pH (<8.5; Figure 6D,E). Experimental evaporation of the two Sutton lake waters, which had initial pH of 9.0 and 9.2, showed initial steep decline in pH and convergence of the two waters to a pH range of ∼8.2–8.6 (Figure 6F). These pH changes accompanied a steady increase in EC for both waters (Figure 6F).

Drying Sutton lake bed sediments had lower EC than lake waters (Figure 6D). Wet sediments had higher EC (10–25 mS/cm; Figures 2A,C and 6D) compared to slurries made from dry crusts that had EC (<10 mS/cm; Figures 2A,B and 5A–C). However, dry biofilm-coated sediment had locally elevated EC where salts had accumulated inhomogeneously (Figure 2E). The pH of wet sediments was 8.5–9.1, similar to the pH of lake waters (Figure 6D). In contrast, the dry sediments showed much wider variation, from pH 8.1–9.8 (Figures 5A–C and 6D). Samples from the surface of the dry crust and from ∼1 cm deeper showed some variation (Figure 5B), but values for both EC and pH fell into the same general range for dry sediments (Figure 6D). Subsurface samples from beneath vegetated marginal flat areas show a broad range of EC and pH data, from typical background soil values near to the lake depression edge to moderately saline values near the lake shore (Figures 4A–C and 6E).

Salt crust minerals

The on-site salt crusts are dominated by carbonate minerals, with intergrown halite and gypsum (Figure 7A–G). Calcite and Mg-bearing carbonate (possibly magnesite) are the most abundant minerals preserved in the crusts, and these form cements for the silicate minerals of the lake sediment (Figure 7A–C). Some Ca-rich carbonate is crystalline (e.g. Figure 7C), but most is fine grained (micron to sub-micron scale; Figure 7A,B). Halite also forms fine-grained cement (Figure 7D,E) with rarer coarse (50 µm scale) crystals coating some surfaces (Figure 7F). Sodium carbonate occurs as amorphous coatings and cement (Figure 7G), although detailed examination was difficult because of its instability under the SEM electron beam.

Figure 7. SEM backscatter electron images (BSE) and some supporting element maps for evaporative salts on and in the dry sedimentary bed of Sutton Salt Lake. A, A window through the silicate surface to carbonate-dominated amorphous evaporative cement. B, Exposed patch of amorphous evaporative cement incorporating silicate particles. C, Crystalline calcite with minor admixed Mg-carbonate. D, Amorphous halite cementing silicates, viewed in BSE and with a Cl map of the same area. E, Abundant amorphous halite cementing silicates. F, Crystalline halite, with minor gypsum coating, that has precipitated on the mounting medium during sample preparation. G, An amorphous particle of Na-carbonate accompanied by calcite and halite, viewed in BSE and with a C map of the same area.

Laboratory evaporation of Sutton lake water yielded precipitates greatly dominated by halite and sylvite (Figure 8A–E; Supplementary Fig. 4), in strong contrast to the on-site salt crusts (above) where chlorides are minor. In addition, the laboratory halite was well-crystallised and coarse grained, with cubes up to 2 mm (Figure 8A–E; Supplementary Fig. 4). Halite crystal size was bimodal, with coarse crystals surrounded by a mat of fine (<0.1 mm) crystals (Supplementary Fig. 4). Some of the halite crystals had well-defined hopper textures (Figure 8A,E; Supplementary Fig. 4). Larger halite crystals had irregular and fine-grained coatings of other evaporative minerals, especially carbonates (Figure 8B–E). Ca–Mg carbonate minerals are fine grained (sub-micron scale) and form botryoidal textures, commonly intergrown with halite crystals (Figure 8A,C). Most of the carbonate minerals were impregnated with brown ferric oxyhydroxide (Supplementary Fig. 4) that left a yellow-brown residual stain after HCl dissolution. Translucent gypsum needles coat some halite crystals as well, locally accompanied by epsomite (Figure 8A,B; Supplementary Fig. 4). Sodium carbonate is a minor component coating some halite crystals (Figure 8D) and rare K-Mg chloride, possibly carnallite, also makes thin coats on halite (Figure 8E).

Figure 8. SEM backscatter electron images of evaporative precipitate minerals derived from complete drying after the laboratory evaporation experiment (Evap 2; Figure 6F). A, Predominant halite crystals are cemented by sylvite and botryoidal Ca–Mg carbonate. B, Gypsum and epsomite coat halite crystals. C, Intergrown halite crystals and botryoidal Ca–Mg carbonate. D, Sodium carbonate coating on a halite crystal. E, Sylvite and K–Mg chloride, possibly carnallite, on a halite crystal.

Lake water compositions in relation to salt mineralogy

Sutton lake water is dominated by dissolved Na⁺ and Cl^- in accord with the abundant halite observed in the evaporative salts (Table 1). The molar Na/Cl ratios of lake waters are slightly lower than 1, similar to seawater and South Island rain water, and leachates from many inland saline sites (Figure 9A; Supplementary Fig. 5). This composition reflects the ultimate origin of the dissolved salts in marine aerosols (Figure 9A; Craw and Beckett 2004). Sutton lake waters have distinctly elevated alkalinity compared to seawater, although not as high as some inland salt pan leachates (Figure 9B). In detail, some of the Sutton lake waters have a combination of excess of both Na⁺ and alkalinity, compared to seawater, that is consistent with dissolution of Na-carbonate minerals (Figure 9C) as is observed in some inland salt pan leachates (Supplementary Figure 5). However, there is considerable scatter in these data, implying some more complex controls.

Figure 9. Sutton Salt Lake water chemistry shown in relation to typical seawater, seawater dilution or evaporation, and some mineral dissolution trends. Some leachate data from Central Otago salt pans (Craw et al. 2023b) are shown for comparison. Sutton data from Craw and Beckett (2004), summarised in Table 1. A, Na–Cl relationships for seawater and NaCl dissolution. B, Alkalinity and Cl relationship, showing elevated alkalinity in Sutton lake waters. C, Excess Na and alkalinity in Sutton lake waters compared to typical seawater, with trends for dissolution of Na-bearing carbonate minerals. D, Ca versus alkalinity, with trend for dissolution of calcite (molar 1:1 and 1:2 reactions; see text). E, Ca versus sulphate, with gypsum dissolution trend. F, Cl versus Mg, showing clear depletion of Mg relative to seawater.

In spite of the strong links between Sutton lake water and seawater that are implied by Na and Cl contents and ratios, the dissolved Ca²⁺, Mg²⁺ and sulphate concentrations in lake water are relatively low compared to seawater (Figure 9D–F). The Ca and sulphate contents of even the most concentrated lake waters are less than half the levels of seawater, although the lake waters have similar Ca/SO₄^2- ratio to seawater (Figure 9E). This Ca/SO₄^2- ratio in concentrated Sutton waters is not consistent with control by, for example, gypsum dissolution (Figure 9E) although gypsum is present in some evaporative salts at the site (Figures 7A,B and 8A,B), as predicted from modelling (Supplementary Fig. 6a,b).

Similarly, all the Sutton lake waters have low Mg²⁺ concentrations, with the most concentrated lake waters having only a third of the Mg²⁺ that concentrated seawater would have (Figure 9F). Evaporative precipitation, without re-dissolution, of Mg minerals such as Mg-carbonates (Supplementary Fig. 6c,d; Harrison et al. 2019) or epsomite (e.g. Figure 8B) is implied by the low Mg contents (Figure 9F). Freshwater Diacypris ostracods typically contain low levels of Mg in their calcareous body parts (Mg/Ca ∼0.01–0.03; De Deckler et al. 2011), and this may extract small amounts of Mg as well, although this is unlikely to have a major effect on the bulk lake water.

Discussion

Halophytes and salinity variations

The halophytes become established at or close to the edge of lake waters when the lake is almost full (Figures 2A,B and 3A–E). This narrow zone coincides with a steep salinity gradient, between the very high EC (>10 mS/cm) of the lake water and the very low EC (<0.5 mS/cm) of the marginal flats near the schist outcrops (Figures 2A–E, 3A–E, 4A–C, 5A–D and 6E). Observations from Central Otago salt pans and coastal settings suggest that high EC and abundant evaporative salts are not barriers for halophyte establishment, but are necessary to inhibit competition from taller plants (e.g. Figures 2A, 3D and 4A–C; Craw et al. 2023a, 2023b; Rufaut and Craw 2023). Hence, development and persistence of high-EC saline soil-free areas along the lake margin is critical for the longevity of the halophyte community.

Despite the high NaCl-dominated salinity of lake waters, the EC of dry salt crusts is relatively low (Figure 6D), and much lower than expected for halite-dominated evaporative salts (Figure 6B). Our SEM observations show that halite is present, but is subordinate to carbonates (Figure 7A–G). Since halite is the most soluble of the evaporative salts that form in Otago saline sites (Craw et al. 2023a, 2023b), it is likely that rain events remobilise at least some of the halite in dry crusts and transport it back into the lake, thereby enhancing NaCl concentrations in the lake water and depleting the EC of the marginal flats. This prediction is supported by our observations of laboratory evaporation of lake water, which yields precipitates dominated by halite (Figure 8A–E; Supplementary Fig. 4). Similar salt redistribution process occur on inland salt pans (Craw et al. 2023a, 2023b). The observed localised higher and inhomogeneous salinity in shallow marginal waters of the lake (Figure 6D) may reflect this recycling, in addition to on-going differential evaporation.

Rounded voids in dry evaporative crusts, such as those seen at the Sutton site (Figure 5B–D), are also seen in inland salt pans with halophytes, and these voids have been attributed to the dissolution of evaporative salts (Craw et al. 2023b). Similar salt dissolution may occur at Sutton lake site when halite-rich salts are recycled. However, gas bubbles and associated voids in the near-shore sediments at the Sutton site (Figure 2C,E) are also possible agents for the formation of voids in the dry crusts. The widespread sulphurous smell at the site, combined with observations of gas discharges (Figure 2C) are consistent with loss of sulphate via volatilisation, possibly facilitated by organic processes that produce hydrogen sulphide and/or dimethyl sulphide (Figure 12A–C; Gibson et al. 1991; Richards et al. 1994; Watts 2000; Reese and Anderson 2009). The relatively low dissolved sulphate concentrations in the lake waters compared to seawater, especially in the most concentrated lake waters (Figure 10E) is also consistent with a volatilisation process that is removing sulphate (Craw and Beckett 2004).

Figure 10. Model saturation indices (log Q/K) for selected minerals in Sutton Salt Lake waters (Craw and Beckett 2004), calculated using Geochemists Workbench® at 25°C (e.g. Table 2). A, Supersaturated minerals. B, Undersaturated minerals.

Carbonate supersaturation and halophyte habitat

As well as the high salinity, the lake waters have abundant other dissolved ions, especially in relation to carbonate minerals (Figure 9B–D). Consequently, Ca, Mg and Na carbonate minerals are the most abundant natural evaporative precipitates on bare substrates on the lake shore where halophytes become established (Figures 5A and 7A–C,G). Ordinary sea surface water is supersaturated with respect to Ca–Mg carbonates (Sulpis et al. 2021) and evaporating brines are commonly supersaturated before precipitation ensues (Babel and Schreiber 2014). Hence, supersaturation is expected to occur at the Sutton site as well. Model calculations of mineral saturation indices for Sutton Salt Lake waters suggest that Ca and Mg carbonate minerals are strongly supersaturated in those waters (Figure 10A; Table 2). Dolomite is the most strongly supersaturated mineral, but because of the widespread occurrences elsewhere of this phenomenon, this issue is ignored here (Arvidson and MacKenzie 1999; Weightman et al. 2020). It is possible that some nanocrystalline carbonate precipitates become suspended within the Sutton lake water column and pass through analytical filtration (0.45 µm), enhancing the level of apparent supersaturation in water analyses (Figure 10A). However, calcite and magnesite show extreme and consistent supersaturation (∼30 times higher concentrations) in all the lake waters (Figure 10A), so abundant fully dissolved species are suspected to contribute to these supersaturation models.

Table 2. Modelled mineral saturation indices (log Q/K) for concentrated Sutton Salt Lake water (as in Table 1).

Mineral	Formula	Log Q/K
Dolomite	CaMg(CO3)2	+4.17
Magnesite*	MgCO3	+1.76
Calcite*	CaCO3	+1.56
Aragonite	CaCO3	+1.38
Nesquehonite	MgCO3.3H2O	−0.93
Brucite	Mg(OH)2	−1.30
Gypsum*	CaSO4.2H2O	−1.54
Anhydrite	CaSO4	−1.74
Gaylussite	Na2Ca(CO3)2.5H2O	−2.03
Pirssonite	Na2Ca(CO3)2.2H2O	−2.19
Halite*	NaCl	−2.39
Nahcolite*	NaHCO3	−2.49
Mirabilite	Na₂SO₄.10H₂O	−2.70
Sylvite*	KCl	−3.28
Epsomite*	MgSO4.7H2O	−3.52
Thenardite	Na2SO4	−3.54
Hexahydrite	MgSO4.6H2O	−3.79
Natron*	(Na2CO3.10H2O)4.NaHCO3	−3.83
Na-carbonate*	Na2CO3.7H2O	−4.17
Glauberite	Na₂Ca(SO₄)₂	−4.69
Kalicinite	KHCO3	−4.74
Thermonatrite	Na2CO3.H2O	−5.06
Arcanite	K2SO4	−5.17
Kieserite	MgSO4.H2O	−5.22
Syngenite	K₂Ca(SO₄)₂.H₂O	−5.62
Trona	Na2CO3.2NaHCO3.3H2O	−6.43
Bloedite	Na2Mg(SO4)2.4H2O	−6.85

Notes: Model calculated with Geochemists Workbench® at 25°C; positive values are supersaturated and negative values are undersaturated (only values down to −7 are presented). Minerals in bold italics have been identified or inferred elsewhere in Otago evaporative contexts (see text for references).

*Minerals identified or inferred at Sutton Salt Lake via SEM (this study).

The lake waters have similar degrees of supersaturation of calcite irrespective of the amount of evaporative concentration of accompanying chloride (Figure 10A). This implies that some precipitation of calcite occurs during lake evaporation. Our SEM observations of natural and experimental evaporites support this modelling prediction for the fully evaporated salt crusts (Figures 7A–C and 8A,C). Since the amount of dissolved Ca²⁺ is lower than that of seawater even in the most concentrated lake waters, calcite supersaturation and calcite precipitation are apparently largely controlled by the elevated alkalinity. This assumes essentially all the alkalinity is bicarbonate ions (Figure 9D), as it is on the inland salt pans (Craw et al. 2022a, 2022b, 2023a, 2023b). Borate is the most likely non-carbonate contributor to alkalinity in a seawater-dominated setting, although this is likely to be minimal compared to the very observed high alkalinity at the Sutton site, because dissolved boron concentrations are low (1–6 mg/L) even near the sea in East Otago (Wangaloa site, Figure 1A; Craw et al. 2006). Further, even these low boron concentrations readily precipitate in evaporitic settings at theWangaloa site (Craw et al. 2006).

In strong contrast to lake water disequilibrium chemistry, pore waters collected from the lake sediments (Craw and Beckett 2004; Table 1) have compositions close to equilibrium saturation for calcite (Figure 10A). The relatively low pH of these sediments (∼8; Table 1; Figure 6E) is broadly similar to that reached during our experimental evaporation of lake water (Figure 6F), and may imply eventual equilibration with respect to calcite. This apparent equilibration suggests that some precipitation of calcite occurs within the sediments while they are wet. Modelling of equilibrium evaporation of seawater with atmospheric CO₂ predicts that precipitates will be dominated initially by small amounts of calcite (Supplementary Fig. 6A). The abundant ostracods in the lake sediments have calcareous body parts, and their life processes may be responsible for overcoming the disequilibrium supersaturation of calcite in lake waters to facilitate chemical equilibration in the lake bed pore waters (Supplementary Fig. 6b).

Modelling of saturation indices of Sutton lake waters suggests a wide variety of other minerals should precipitate with the carbonates during full evaporation (Figure 10B; Table 2). However, only small amounts (<<1 mg per litre of water) of most of these various minerals are predicted to form from these models. More of these minerals have been identified in the inland salt pans, in addition to the small subset of evaporite minerals that have been identified in the Sutton lake salt crusts via our SEM observations (Table 2). Brucite is predicted to occur in Sutton precipitates (Figure 10B; Table 2) but has not been detected, although it does occur in some inland sites (Craw et al. 2022a, 2023b). Also, ferric iron can dissolve at high pH (>9) but reprecipitates if pH is lowered (Langmuir 1997; Craw et al. 2022a; 2023b). This is the probable source of the ferric oxyhydroxide staining carbonates in our experimental evaporative precipitates (Supplementary Fig. 4).

Variations of pH in halophyte habitats

Halophytic plants are remarkably tolerant of variations in pH on inland salt pans, where pH can range from 5 to 11 (Figure 6A; Craw et al. 2022a, 2022b, 2023a, 2023b). In this context, pH is unlikely to be a major factor affecting halophyte establishment and survival at the Sutton site, but pH may affect other encroaching plant species. Also, pH indirectly affects the carbonates discussed in the previous section. The variable and locally high pH at the Sutton site measured in this study (Figure 6D) are notably different from shallow schist-hosted groundwaters in nearby Macraes area that are dominated by Ca-carbonate interactions have consistent pH ∼8.2 (Table 1; Craw 2000; Weightman et al. 2020). Typical seawater also has pH ∼8.2 (Table 1).

At the Sutton site, there are clear differences in pH between the lake floor sediments (pH 7.5–8.2) and lake waters that have pH ∼8.9–9.3 (Craw and Beckett 2004; Figure 6D) and lake waters have been as low as pH 7.7 in the past (Table 1). It is possible that biological processes, such as diurnal variations in photosynthesis of lake biota, have driven the pH to higher levels locally at times (e.g. Wood et al. 2015). However, the lake pH stayed high (∼9) from summer through to early winter (May–June 2023; Figure 3D and 6D) as the lake filled, algae became much less abundant, and daylight hours became short. Hence, for this study we evaluate purely inorganic processes for pH variations via modelling of water compositions, as below, in comparison to inland salt pans.

Inorganic dissolution of calcite and/or magnesite can occur via a combination of chemical reactions that yield dissolved Ca and/or Mg and either 1:2 or 1:1 molar ratio with alkalinity (e.g. Figure 9D; Langmuir 1997). Both these carbonate dissolution reactions, involving mainly Ca-carbonate, occur in more limited evaporative environments in the nearby Macraes area (Figure 1B; Craw 2000; Weightman et al. 2020). These reactions are also probably responsible for the elevated alkalinity in Sutton lake waters (Figure 9B,D; Table 1) and lead to the substantial supersaturation of these two minerals in the waters (Figure 10A). For the saline waters of Sutton lake, modelling suggests that the resultant alkalinity would be a combination of numerous different dissolved carbonate species (Figure 11A). Formation of several of these species also consumes hydrogen ions to contribute to the pH rise, and at the typical pH ∼9 of concentrated lake waters, about half the alkalinity could be contributed by carbonate species other than HCO₃^- (Figure 11A). The observed high lake pH occurs at, and is possibly controlled by, the crossover of dominance of these different species types (Figure 11A). In comparison, sodic salt pans from inland Otago can have even higher pH, and HCO₃^- is subordinate (Figure 11B; Table 1).

Figure 11. Geochemical models (Geochemists Workbench®) of saline water compositions. A, Speciation of dissolved carbonate components in concentrated Sutton lake water, calculated at a range of pH. B, Speciation of dissolved carbonate components in leachate from a sodic salt pan leachate (Table 1; Craw et al. 2023b) plotted at same scale as a, for comparison. C, Rise in pH as a result of continued addition, to supersaturation, of dissolved calcite and magnesite to dilute seawater (models at 10°C and 25°C). D, Similar model to E, using dilute Sutton lake water at 25°C, starting with low alkalinity (100 mg/L).

This disequilibrium dissolution of excess calcite and magnesite to create supersaturated waters is one possible driver of the high pH of the lake waters, as depicted in models in Figure 11C,D. For a typical dilute seawater composition, adding calcite or magnesite to supersaturate the water can raise the pH as high as 9.6 at 25°C (Figure 11B). At lower temperatures, the resultant pH is raised even higher, by ∼0.2 pH units at 10°C (Figure 11C). Applying a similar model dissolution of calcite into dilute Sutton lake water (1998; Table 1), with initial alkalinity set at 100 mg/L, yields substantial increase in alkalinity to ∼750 mg/L, as is seen in some real waters (Figure 9B,D; Table 1). At the same time, the pH of the model waters rises to ∼9.2 at 25°C (Figure 11D), which is within the range of observed data (Figure 6D).

The pH of the laboratory evaporation experiments did not rise to the high levels seen in the lake, and that may be because the experimental evaporation was not affected by rain events that contribute to different precipitate mineralogy (Figure 7 versus Figure 8). Uninterrupted seawater evaporation leads to lowered pH (Supplementary Fig 6b; Babel and Schreiber 2014), which may have countered the effects of carbonate dissolution. This lowering of pH after evaporation of marine aerosols with only minor carbonate effects is observed at saline sites on the nearby Dunedin coast (Figure 1B; Rufaut and Craw 2023). Further, precipitation during evaporation may have assisted in limiting carbonate supersaturation.

In contrast to the inferred self-limiting pH of Ca and Mg carbonate chemistry in Sutton lake waters (Figure 11A,C,D), some inland sodic salt pans have abundant evaporative Na-carbonates that can lead to localised hyperalkaline pH values (as high as 11; Figure 6A; Craw et al. 2023a, 2023b). A parallel situation may arise at Sutton Salt Lake, where some dry crusts yield slurry pH measurements as high as 9.8 (Figure 6D), and some Na-carbonates have been observed in natural salts (Figure 7G) and experimental salts (Figure 8D). Slurry measurements are made on a time scale of minutes, which is too rapid for substantial dissolution of calcite and magnesite that typically dissolve over time scales of weeks (Craw et al. 2023a, 2023b; cf. Figure 11C,D). Na-carbonates have dissolution rates >4 orders of magnitude faster than calcite (Craw et al. 2023a, 2023b), and therefore probably do raise the short-term field measurements of pH.

Geoecological evolution of halophyte habitat on the lake margin

The most abundant halophyte species at the Sutton lake, salt grasses Puccinellia walkeri and P. fasciculata, occur at the shoreline of the lake when it is full (Figures 2A,B and 12A). P. walkeri is an endemic species that is closely related to native P. stricta, a species that is shared with Australia (Edgar 1996). P. fasciculata is a Northern Hemisphere species that is naturalised in New Zealand (Edgar 1996). These salt grass species are also common in saline areas on the Otago coast, and on the margins of many of the inland salt pans (Allen et al. 1997; Allen and McIntosh 1997; Rogers 1999; Rogers and Wiser 2010). The halophyte herb Oxybasis ambigua is a minor component of Sutton lakeshore vegetation (Figures 3D,E and 12A) and at inland saline sites, but is abundant in saline coastal areas (Rogers 1999; Rogers and Wiser 2010). Historically (in 1984), many wetlands in the area immediately surrounding the Sutton Salt Lake also hosted the native salt-tolerant herb Selliera radicans (Supplementary Fig. 1), which is also common on the coast (Rogers 1999; Rogers and Wiser 2010; Rufaut and Craw 2023), and is a minor component at some inland saline sites (e.g. Patearoa, Figure 1B). However, these Sutton wetlands have become progressively invaded by exotic weeds (Supplementary Figs 1, 3). The native halophytes have been fully displaced, and the substrates beneath the weeds now have low pH (5.5–7) and low EC (<0.5 mS/cm; Figure 6C). Our previous work, combined with other workers’ observations, indicates that substrate EC values >1 mS/cm are needed to exclude weeds from halophyte habitats (Figure 12B; Allen et al. 1997; Rufaut et al. 2018; Craw et al. 2023a, 2023b; Rufaut and Craw 2023).

Figure 12. Summary of principal physical, chemical and biological features, and inferred geochemical processes, of the lake bed and margin at Sutton Salt Lake, compared to an alkaline sodic salt pan in Central Otago. A, Block diagram (vertical scale exaggerated) of the Sutton lake margin. B, and C, Sketch section through the surface crust in a, with biofilm and salt coatings, respectively. D, Schematic diagram depicting controls on alkaline pH at the Sutton lake site. E, Photograph of a high-pH sodic Central Otago salt pan showing principal features.

The demise of the native halophytes in the Sutton lake’s surrounding wetlands raises the issue of potential displacement of halophytes by exotic weeds in the Sutton lake depression itself. This issue is most prominent on the western side of the asymmetrical lake depression, where the lake is shallowest and a well-defined weedy vegetation strip has already developed, with its widest parts at the northwestern end (Figures 2A and 3A; Supplementary Figs. 1–3). The reason for the asymmetry of sediment distribution is not clear, but it may be related to the original shape of the depression and/or sediment mobilisation by southerly winds and associated waves in the shallow lake. Irrespective of the cause, the end result has been the formation of a wider flatter western lake shore zone than on the eastern side (Figures 1D and 2A; Supplementary Figs. 1–3). This wide western shoreline is most susceptible to weed incursion, especially at the northwestern tip of the lake (Figures 2A and 3A; Supplementary Fig. 3).

Salt grasses, especially P. walkeri, can grow in damp or wet sediment on the lake margin (Figures 2A,B and 12A) and the lakeward extent of this plant colonisation reflects the maximum extent of lake water when the lake is full. The combination of underlying root mass and subaerial grass (Figure 12A) inevitably traps suspended sediments when the lake is full (Langlois et al. 2003), or even when the lake is at lower levels and dust is blown from dry surfaces. Hence, the establishment of grasses can raise the substrate slightly (centimetres) above lake level (Figure 2A,B). Resultant elevated and vegetated lake margins can then trap incoming rainwater, both directly and from rainwater seeping from the margins of the depression, leading to the formation of low-salinity pools and associated saturated low-salinity ground (Figures 2A and 6D). It is in these low-salinity areas on the marginal flats that weeds can become established and displace the halophytes (Figures 2A, 3A–E and 4A–C). Rain-driven leaching of salts from marginal bare dry sediments (Figure 12A–D) can lead to lowered surface EC (Figure 6D), and may also permit localised incursion of vegetation, native or exotic, on slightly elevated dry ridges (e.g. Figures 1D and 2B).

The above-described processes are likely to result in progressive lakeward migration of the vegetated margin on the western side of the lake, as has clearly happened at the northwestern tip (Figure 2A; Supplementary Fig. 3). There has been lateral encroachment of vegetation on to the lake bed on the scale of 5–30 m since 1945 and probably since 1975 (Figure 3A–E; Supplementary Fig. 3). While a narrow zone of salt grasses may follow this retreating lake edge (Figures 2B and 3A–E), it is likely that areas of exotic weeds will become more extensive on the flat areas behind the salt grasses (Figures 2A, 3A–E, 4A–C, and 12A). If the shore area becomes elevated by more than a few centimetres above lake level, or if the lake remains low over several dry years, weeds may totally displace the salt grasses from areas of lowered surface EC, and extend on to the dry lake bed.

One possible remediation approach is to re-introduce managed stock grazing, especially at a time when the lake is full. Trampling of the flat saturated marginal substrates by stock may induce sufficient disturbance to allow saline lake waters to reoccupy the areas that currently host low-salinity water pools (Figure 2A) and underlying sediments (Figures 4A–C and 6E). Such extension of saline surface waters into the weedy areas could permit re-establishment of salt grasses farther towards the edge of the lake depression.

Conclusions

The Sutton Salt Lake lies approximately half-way between coastal saline environments and inland salt pans and has features in common with both settings. The Sutton lake salt, like the coastal sites and inland pans, is derived from marine aerosols in rain, followed by evaporative concentration. Like saline coastal turfs, the Sutton lake chemistry is dominated by Na and Cl with only minor evidence for water-rock interaction that is a more significant component of many inland salt pan sites. Evaporative precipitates on the lake margin are dominated by carbonates because soluble halite and sylvite are recycled back into the lake by rain events.

The Sutton Salt Lake site hosts some halophytes, mostly salt grasses. The halophyte communities are being invaded by exotic weeds where substrates have lowered EC (<1 mS/cm) in a similar manner to coastal turfs and inland salt pans. Some wetlands surrounding the Sutton lake have been completely taken over by weeds on what are now low-pH, low-EC substrates, with no remaining native halophytes. Similar processes are occurring on the western margin of the main Sutton salt lake where the lake is shallowest and a broad flat shore area is exposed for much of the year. Vegetation encroachment and associated weed incursion have occurred on the scale of 5–30 metres laterally over the past 50 years. This encroachment may possibly be reversed by managed stock grazing, preferably when the lake is full. Stock trampling of the vegetated areas may permit flooding of the areas by saline lake waters, to the detriment of exotic species and potential advantage of halophytes.

Pore waters in Sutton lake sediments have pH ∼8 that is controlled principally by Ca-carbonate dissolution and precipitation, and this is similar to some coastal turf and some inland salt pans. However, there is clear chemical disequilibrium between Sutton sediment pore waters and the overlying lake waters, which have anomalously high pH (8.9–9.3). This high pH of Sutton lake is at least partially a result of disequilibrium supersaturation of the waters with respect to Ca and Mg carbonate, and this high pH and supersaturation is stable irrespective of the amount of lake filling and dilution or evaporative concentration. In contrast, some evaporative dry crusts on the Sutton lake bed have short-term (minutes to days) locally elevated pH when wet as a result of the dissolution of minor Na-carbonates, similar to some highly alkaline sodic inland salt pans.

Acknowledgements

We thank Scott Jarvie of ORC for his interest and support of our research project on Otago saline areas. We appreciate the enthusiasm of Ellery Mayhence and Clement Lagrue (Department of Conservation) who provided useful discussions. SEM observations were made at Otago Micro and Nanoscale Imaging (OMNI), University of Otago. Constructive comments from two reviewers and journal editor substantially improved the presentation of the ms.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data relevant to this study are included within this paper and cited references, with additional figures and tables in a Supplementary file available at: figshare.com/articles/journal_contribution/Sutton_Salt_Lake_Supplementary_Figures_and_Tables/24899853.

Additional information

Funding

This research was funded by University of Otago and Otago Regional Council.