Depression wetland formation by redox-driven iron and silica cycling

doi:10.21203/rs.3.rs-2615893/v1

Download PDF

Research Article

Depression wetland formation by redox-driven iron and silica cycling

https://doi.org/10.21203/rs.3.rs-2615893/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 12 Jan, 2024

Read the published version in Wetlands Ecology and Management →

You are reading this latest preprint version

This paper investigates the origin and geomorphic evolution of Shadow Vlei, a depression wetland located approximately 30 km northwest of Port Elizabeth. The wetland is on sandstone on the African Erosion Surface. The depression wetland has a rim of high-lying ground surrounding a central depression. Fluctuations between wet and dry periods create both highly reducing conditions in soils during wet phases in the depression, and highly oxidising conditions in the depression margin. Under reducing conditions, iron(III) oxides that are insoluble, are reduced to soluble iron(II), which is transported to and trapped in the margins of the depression under oxidising conditions when the water table declines and iron(II) is oxidised to iron(III). There is also a moderate flux of silica from the surface in the centre of the depression to the margins, which is suggested to be associated with maximum alkalinity associated with iron(III) reduction, and decreased alkalinity in the margins associated with iron(II) oxidation. However, the main flux of silica is from the surface in the centre of the depression to depth of greater than 1.5m, again associated with variations in pH linked to iron-cycling. These reactions have thus caused a net volume loss in the centre of the depression, causing sagging, and a net volume gain at the margins of the depression, causing swelling. Wetland formation associated with alteration of landform morphology as a consequence of alternating redox conditions represents a novel mechanism that has not yet been described.

Depression wetlands are common globally and found across a wide range of landscapes and climates. These systems, which are most abundant in drylands where mean annual evaporation exceeds rainfall, are generally ephemeral, forming in localised topographic depressions where surface water accumulates due to precipitation and local runoff (Timms, 2006). Water seldom leaves depression wetlands via surface outflow but is lost via evaporation and groundwater recharge (Timms, 2006).

The term depression wetland is broadly applied to ephemeral and endorheic water bodies such as pans, playas, dune slacks, sumplands, damplands, or depression lakes, terms that are often used interchangeably. The term ‘depression wetland’ will be used in this study to describe endorheic depressions that are characteristically ephemeral and contain geomorphic, hydrologic, biotic and edaphic features that are typical of wetland systems. There are very limited descriptions and analyses of geomorphic processes in depression wetlands, but given their widespread occurrence in drylands there is general consensus that depression wetlands form as a result of aeolian processes (Goudie and Thomas, 1985; Timms, 2006). Central to the model of depression wetland formation presented by these authors is the process of aeolian deflation, which may be accelerated due to localized grazing pressure that reduces vegetation cover. As these depressions are endorheic, there is little fluvial infilling that takes place such that once the process of deflation has begun, the net sediment budget of the depression is negative. The model developed by Goudie and Thomas (1985) was refined by Thomas (2011), who identified structural controls (faulting, rock fracturing or crustal warping), ponding controls (interruption of surface flow by an obstruction such as an aeolian dune (Timms, 2006), trunk or tributary alluvial fans (McCarthy et al., 2011; Joubert and Ellery; 2013; Grenfell et al., 2010; Ellery et al., 2012) or erosional controls (aeolian deflation around animal watering points (Goudie and Thomas, (1985). Such controls may work individually or in combination to form depression wetlands (Thomas, 2011).

General interest in biogeochemical processes in wetlands in dryland environments, particularly of their contribution to landform modification through dissolution, weathering and chemical sedimentation, has grown substantially in the last three decades (cf. Tooth and McCarthy, 2007), including the formation of depression wetlands. Karst environments are associated with landforms related to dissolution of rocks rich in carbonate (Cabadas-Báez et al., 2010) and gypsum (Doğan and Özel, 2005) by slightly acidic meteoric waters infiltrating the rocks (cf. Gutiérrez et al., 2014). Dissolution leads to sagging or collapse and, in the latter case, there may be subsequent partial infilling that leads to formation of depression wetlands generally known as dolines and poljes. In contrast, calcium and magnesium introduced in solution into the Okavango Delta, Botswana, due to weathering of rock in the watershed, accumulate as magnesian calcite on island edges as a result of chemical precipitation due to solute enrichment caused by transpirational water loss (McCarthy and Metcalfe, 1990; McCarthy et al., 1993). Precipitation of magnesian calcite in island margins leads to volume expansion and the creation of water-filled depressions in the island centres. Long term chemical weathering of bedrock to considerable depths (10 to 15 m), and associated mineralogical simplification, leads to mass and volume losses that have been shown to lead to sagging and the creation of depression wetlands on the African Erosion Surface in the KwaZulu-Natal Midlands (Edwards et al., 2016) and the Eastern Cape (Alistoun, 2014), in South Africa. Export of material in solution in quartz-kaolinitic sedimentary rocks in the Taubaté basin in São Paulo, Brazil, has also been recognised as leading to sagging and the formation of depression wetlands (Filizola and Boulet, 1996), although the processes leading to dissolution were not described by these authors.

These geochemical processes associated with depression wetland formation have not been recognised until recently, although there is growing interest in these features, particularly in drylands where depression features that are temporarily flooded are abundant. This study investigates biogeochemical processes associated with redox transformation and margin-ward translocation of iron (and silica) as a possible mechanism leading to the formation of depression wetlands in a dryland environment.

Shadow Vlei (Fig. 1; 33°52'20"S and 25°24'38"E) is situated 210 m above sea level and located on municipal commonage approximately 30 km north west of the city Port Elizabeth. The climate is warm and temperate, receiving between 500 mm and 850 mm of rain fairly evenly distributed throughout the year (Mucina and Rutherford, 2006), but with peak rainfall periods in March and October/November (Jury and Levey, 1993). The mean annual evaporation varies from 1500 mm to 3000 mm in extreme years (Midgeley et al., 1994). The warmest months are January and February with a mean temperature of 20 ^oC, while the coldest month is July with an average temperature of 14 ^oC (O’Connor and Crow, 1999). The area is generally frost-free (Mucina and Rutherford, 2006).

Insert Fig. 1 here

Based on a long-term analysis of rainfall from about 1880 to 1990, it is evident that droughts are a common feature of climate in the Eastern Cape Province, with only two extended periods of significantly above average annual rainfall over the periods 1880 to 1895 and 1971 to 1985 (Jury and Levey, 1993). However, over most of the rest of the period examined, “dry spells were the norm” (Jury and Levey, 1993: p 135). The region is currently experiencing a significant extended drought, with serious water shortages in all metropolitan areas, including Port Elizabeth.

Shadow Vlei lies upon the African Erosion Surface (AES), which is a gently undulating to flat surface and is the product of a long period of weathering and erosion from the Cretaceous (140 My BP) to the first Cenozoic uplift of the subcontinent of southern Africa about 30 My BP (Partridge and Maud, 1987). The erosion surface in the area where Shadow Vlei is situated is on a very gently sloping interfluve (about 1%) that is being dissected by small ephemeral streams (Fig. 2). As such the depression wetland is at a high point locally, surrounded by dissected valleys at lower elevations.

Insert Fig. 2 here

Shadow Vlei is located on the easternmost stretch of the Table Mountain Group sandstone (Thamm and Johnson, 2006). In the area south of Uitenhage, the approximately 2 500 m thick Table Mountain Group conformably overlies 500 m of grey quarzitic sandstone and shale (Thamm and Johnson, 2006). The sandstone upon which Shadow Vlei lies is composed primarily of mature quartz arenite within a feldspathic sandstone matrix – although the feldspars typically amount to less than 5% of the total mineralogy (Thamm and Johnson, 2006). The surficial lithology contains traces of iron and aluminium and therefore appears to be a brick red colour.

The predominant soil types in which Shadow Vlei is located are the hard-eluvic Wasbank soil form where an E horizon is present above the hard plinthic horizon, or a Dresden soil form where it is absent (Macvicar et al., 1991; Soil Classification Working Group, 2018). These soils are plinthosols (IUSS Working Group WRB 2015). Plinthic soils are characterized by enriched iron oxide concentrations in a hard plinthic B-horizon, often occuring in areas where a fluctuating water table gives rise to both aerobic and anaerobic conditions below ground (Fey, 2010).

The topography of Shadow Vlei was determined in order to establish variation in local topographic relief and relate this to variation in subsurface soil mineralogy and chemistry. Topographic variation was measured along eight survey transects that started about 100 m outside the depression, passed through the central point and extended about 100 m beyond the opposite margin of the depression. Topographic data was collected using a Topcon GTS 211D total station device and staff with reflective prism, with an accuracy of approximately 0.01 m in x, y and z dimensions. All points where soil cores were taken were included in the topographic survey. Contour lines with a contour interval of 0.5 m were manually interpolated to an estimated accuracy of 0.1m. Survey points were tied to a datum on a 1:50 000 topocadastral map, with an accuracy of approximately 2 m.

Soil samples for chemical and mineralogical analysis were extracted from nine soil cores taken along two perpendicular transects that intersected at a common soil core at the centre of the depression, such that three of the soil cores were located outside of the depression margin, two were in the depression margin and four were within the depression itself. Soil coring was undertaken using a 1996 Thor Drilling Rig equipped with a 100mm diameter percussion drill and a 1995 Ingersoll Rand Air Compressor XHP900. In each core, soil samples were collected at depth intervals of 0.10 m to a depth of 4 m, with an additional soil core to sample parent material from depths of 10 and 11 m.

Soil samples were dried at 60 ^oC until no further weight loss was measured before crushing for chemical and mineralogical analysis. About 50% of the sample was passed through a 1 mm sieve, with material retained by the sieve utilised for purposes of colour determination. The colour of each sieved sample was determined using a Munsell Soil Colour Chart (Munsell Color Co., Inc., 1992), which formed the basis of sample selection for chemical and mineralogical analysis. Samples within each soil core were selected from the mid-depth of each zone of approximately uniform colour such that 70 samples from the nine soil cores were selected for mineralogical analysis using X-Ray diffraction (XRD) and for soil chemistry using X-Ray fluorescence (XRF).

The 70 samples selected for chemical and mineralogical analysis comprised the bulk samples that had been oven dried but not sieved, and thus consisted of both fine material and larger clast particles. Large clasts in the bulk sample were broken up using a jaw vice and returned to the bulk sample for crushing to a fine powder of < 75 µm in a Lead Herzog Swing Mill with a. Each milled soil sample was then split with one half used for XRF analysis and the other for XRD analysis.

A qualitative XRD analysis was carried out at the iThemba Labs in Cape Town where a Bruker D8 Powder X-ray diffractometer system equipped with a theta-theta goniometer was used. Copper radiation (CuKα; λKα₁ = 1.5406 Å) and a 0.034 second time step was used for a continuous XRD scan. The analysis was carried out over a 2θ range of 5-80^o and each XRD pattern produced was analysed using the built in Bruker EVA software in combination with the ICDD PDF 1999 database.

The XRF analysis was carried out at Stellenbosch University using a PANalytical Axios Fast WDXRF spectrometer. Major element composition was determined on samples fused into a lithium borate glass disc (Norrish and Hutton, 1969). The spectrometer was equipped with a tungsten tube. The loss on ignition was established from the mass loss of samples heated to 1000°C.

Semi-quantitative mineralogical analysis were performed on a select number of samples based on the data from iThemba Labs using the Material Analysis Using Diffraction (MAUD) program. MAUD is a general diffraction/fluorescence/reflectivity program based primarily on the Rietveld method and can be used for quantitative phase analysis (Bish and Howard, 1988). Despite inherent difficulty in assigning quantitative values to bulk powder XRD samples, estimated mineral abundances were within approximately 15% of their actual abundances in the samples. The greatest source of potential error during the MAUD analysis was the difference in crystallinity that exists between phases present in each sample. Descriptive statistics from the package RStudio were used to summarise the mineral dataset. Regression and correlation statistics were computed to determine relationships between mineral distribution and to assess the distribution relationship between the mineral phases and the chemistry of soils.

The data from the XRF analysis were analysed using Microsoft Excel and RStudio. Most of the data analysis focused on establishing relationships between the quantities of SiO₂, Al₂O₃ and Fe₂O₃ because jointly they generally accounted for > 90% of the chemical composition of samples. The XRF results were used to construct cross-sections showing the distribution of SiO₂, Al₂O₃ and Fe₂O₃ in soil samples. Regression and correlation statistics were computed in RStudio to generate scatterplots that were used to determine relationships between these oxides.

The concentrations and variability of all oxides were presented as a series of boxplots as calculated in RStudio, showing the mean and quartiles from maximum to minimum values. A hierarchical agglomerative cluster analysis was used to group samples with similar chemical compositions, which allowed for the identification of patterns in the distribution of different oxides below ground. Descriptive statistics (mean and standard deviation) were then used to analyse differences in the concentrations of SiO₂, Al₂O₃ and Fe₂O₃ within and between clusters. One-way ANOVA and a Tukey’s HSD post-hoc test were conducted to determine the differences between oxide concentrations of clusters.

4.1 Morphology of Shadow Vlei and surficial ferricrete distribution

Shadow Vlei comprises a central depression surrounded by elevated margins that are approximately 3 m higher than the deepest point. The elevated margin gives way laterally to a relatively flat plain about 0.5 m lower than the margin (Fig. 3). The entire wetland is approximately 150 m long and 100 m wide. A crescent-shaped mound exists on the north side of the depression where there are scattered ferricrete (iron cemented concretions) outcrops. A ferricrete outcrop occurs extensively along the eastern margin of Shadow Vlei with smaller ferricrete outcrops scattered around many of the remaining marginal areas of the wetland. These outcrops vary in their extent, but the average thickness of the exposed ferricrete is between 1 and 2 m.

Insert Fig. 3 here

4.2 Mineralogy and Geochemistry of Shadow Vlei

4.2.1 Variation in mineralogical and chemical composition

The two bedrock samples analysed for their bulk mineralogy were predominantly composed of quartz (approx. 90% modal), with subordinate K-feldspar and sericite making up the remainder. Although no XRF was conducted on these samples, the above mineralogy suggests a simple bulk geochemical composition overwhelmed by the abundance of SiO₂ (quartz), with much lower contents of Al₂O₃ and K₂O contained in the K-feldspar and sericite mineral fractions. Trace Fe occurs mainly as very fine interstitial particles of Fe oxides and hydroxides.

Compared to bedrock compositions, soils at Shadow Vlei also have a correspondingly simple mineralogy, with microcrystalline quartz, sericite and clay minerals being the dominant species. XRD results point to mixtures of illite and less so kaolinite as the chief components of the clay mineral fraction, accounting for the bulk of the Al and K soil concentrations (along with sericite), although minor K-feldspar cannot be discounted as an additional host. The abundance of clays beneath the depression mirrors the correspondingly high concentration of aluminium in that part (R² = 0.7646). With respect to Fe oxides, goethite (FeO(OH)) is a dominant albeit minor phase with an average concentration for all samples of 7.8%. The presence of goethite, particularly at the marginal areas of the depression, accounts almost solely for the presence of iron in the soil chemistry (R² = 0.7708).

It follows that Shadow Vlei soils have a relatively simple bulk geochemical composition. The oxides of Ca (CaO), K (K₂O), Mg (MgO), Mn (MnO), Na (Na₂O), P (P₂O₅) and Ti (TiO₂) are present at concentrations generally less than 1% each and collectively no more than 10% of the total soil mass (Fig. 4A). Between 2% and 7% of the soil material also comprises organic matter and water bound to clay minerals and Fe hydroxides; both components are measured and herein reported as loss on ignition (LOI). The soil is thus predominantly composed of the oxides of silicon (SiO₂), aluminium (Al₂O₃) and iron (Fe₂O₃). The SiO₂ makes up by far the largest proportion of the soil mass with an average concentration of 83.14%, while Al₂O₃ and Fe₂O₃ are present with average concentrations of 5.16% and 5.62% respectively (Fig. 4B). Minimum and maximum values respectively are SiO₂ = 55.64% and 97.77%, Fe₂O₃ = 0.71% and 22.71% and Al₂O₃ = 0.87% and 14.60%.

Insert Fig. 4 here

The dendrogram in Fig. 5 shows the clusters into which each sample was classified based on chemical composition. The divisions identified at a cluster height of 9, were considered to represent sample heterogeneity in an effective way. There were a total of 5 clusters that each contained between 3 and 25 samples. Clusters 1 and 2 contained about 67% of the samples while Clusters 3, 4 and 5 contained approximately 20%, 9% and 4% respectively.

Insert Fig. 5 here

The clusters were split based mainly on variation in SiO₂, Al₂O₃ and Fe₂O₃. Cluster 1 contains samples that are rich in SiO₂ and have very low Al₂O₃ and Fe₂O₃ concentrations. The mean weight percentage for this cluster is 94.12% for SiO₂, 2.26% for Al₂O₃ and 1.60% for Fe₂O₃ (Table 1), with values significantly different from the means in all other clusters (Table 2). Cluster 2 contains samples that have slightly lower SiO₂ concentrations (x̅ = 85.23%) and slightly higher Al₂O₃ (x̅ = 4.84%) and Fe₂O₃ (x̅ = 4.16%) concentrations than Cluster 1, and once again values in Cluster 2 are significantly different to all other clusters except for Al₂O₃, where the mean in Cluster 2 is not significantly different to Cluster 5 (Table 2). The third cluster contains intermediate concentrations of SiO₂, Al₂O₃ and Fe₂O₃ (x̅ = 73.54%, 8.76% and 6.80% respectively) compared to other clusters, and is the most variable cluster (Table 1). The means for Cluster 3 are significantly different to those in all other clusters except in the case of Al₂O₃, where the difference between the means of Cluster 3 and Cluster 5 are not significant (Table 2). The highest mean weight percentage for Al₂O₃ is in Cluster 4 (x̅ = 12.91%), which has mean values of 61.45% and 10.48% for SiO₂ and Fe₂O₃ respectively. The means in Cluster 4 are significantly different to all other clusters except in the case of SiO₂, where the means in Clusters 4 and 5 do not differ significantly. The highest mean weight percentage for Fe₂O₃ is contained within Cluster 5 (x̅ = 21.83%) while the lowest mean concentration of SiO₂ is also contained in this cluster (x̅ = 61.16%).

Table 1

The mean weight percentages, standard deviations and sample number (Total N = 68) for aluminium, iron and silicon oxides across the five clusters. Values in bold are the maximum values for each column.
	Al₂O₃		Fe₂O₃		SiO₂
Cluster	Mean (x̅)	Std. Dev.	Mean (x̅)	Std. Dev.	Mean (x̅)	Std. Dev.	N
Cluster 1	2.26	0.76	1.60	0.60	94.12	1.64	25
Cluster 2	4.84	1.70	4.16	2.17	85.23	2.56	21
Cluster 3	8.76	2.47	6.80	2.95	73.54	3.64	13
Cluster 4	12.91	1.15	10.48	1.20	61.45	3.57	6
Cluster 5	7.02	1.72	21.83	0.69	61.16	2.79	3

Table 2

Post-hoc comparisons using Tukey’s HSD test (N = 68). Differences between weight percentage means are shown where * indicates difference is significant at the 0.05 level, ** indicates difference is significant at the 0.01 level, *** indicates difference is significant at the 0.001 level.
Al₂O₃
	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Cluster 5
Cluster 1	0	2.58***	6.50***	10.64***	4.76***
Cluster 2		0	3.92***	8.07***	2.18
Cluster 3			0	4.15***	1.74
Cluster 4				0	5.89***
Cluster 5					0
Fe₂O₃
	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Cluster 5
Cluster 1	0	2.57*	5.20***	8.88***	20.23***
Cluster 2		0	2.64*	6.32***	17.66***
Cluster 3			0	3.68**	15.02***
Cluster 4				0	11.34***
Cluster 5					0
SiO₂
	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Cluster 5
Cluster 1	0	8.89***	20.58***	32.67***	32.96***
Cluster 2		0	11.69***	23.78***	24.07***
Cluster 3			0	12.09***	12.38***
Cluster 4				0	0.29
Cluster 5					0

Insert Tables 1 and 2 here

SiO₂, and Fe₂O₃ show an inverse relationship as low concentrations of SiO₂ are associated with high concentrations of Fe₂O₃, but as SiO₂ concentration increases, Fe₂O₃ concentration decreases (R² = 0.6497; Table 3). A similar inverse relationship exists between Al₂O₃ and SiO₂ (R² = 0.8003). The relationship between Al₂O₃ and Fe₂O₃ is weak (R² = 0.2962), but which is nevertheless statistically significant.

Table 3

Pearson correlation coefficients and significance values for geochemical results (N = 68)
	Al₂O₃	Fe₂O₃
SiO₂	R² = 0.8003	R² = 0.6497
	p < 0.001	p < 0.001
Al₂O₃		R² = 0.2962
		p < 0.001

Insert Table 3 here

4.2.2 Variation in soil chemistry in relation to depression morphology

There is a clear link between the morphology of Shadow Vlei and the distribution of SiO₂, Al₂O₃ and Fe₂O₃. Al₂O₃ and Fe₂O₃ are concentrated beneath the mounds that form the margins of the depression whereas SiO₂ is depleted in these regions (Figs. 6 and 7). Fe₂O₃ is concentrated close to the surface in the margins of the depression with concentrations often reaching values greater than 12%. It is moderately concentrated near the soil surface beneath the centre of the depression with concentrations that range between 4% and 8%. Fe₂O₃ is depleted with increasing depth and increasing distance from the margins of the depression, with average concentrations declining to less than 1%.

Al₂O₃ has a similar distribution to Fe₂O₃ but is present at more uniform concentrations beneath the centre of Shadow Vlei. Al₂O₃ is also concentrated beneath the margins of the wetland, reaching concentrations greater than 12%. However, at depth Al₂O₃ is at relatively higher concentrations than Fe₂O₃. The concentration of SiO₂ is broadly inverse to those of Al₂O₃ and Fe₂O₃. Beneath the margins and at the centre of the depression, it occurs at concentrations that range from 55–70%. However, in the surrounding landscape and with increasing depth, SiO₂ generally occurs at bulk concentrations that are greater than 85%. In general, there is a high degree of agreement between the concentrations of SiO₂, Al₂O₃ and Fe₂O₃ and the clusters presented in the cluster analysis.

Insert Figs. 6 and 7 here

The wetland formation model that was presented by Goudie and Thomas (1985) cannot be applied to Shadow Vlei. Firstly, Shadow Vlei is distinctly atoll shaped, suggesting that there has been both aggradation and degradation. In a wetland where sediment is removed solely by aeolian processes, the creation of a large topographic feature downwind of the depression is likely to be observed (Timms, 2006), but it is unlikely that positive relief would be created evenly around the entire wetland as in Shadow Vlei. Secondly, the resistant nature of the sandstone lithology upon which Shadow Vlei is located means that removal of sediment by wind or biotic processes is also unlikely. Erosion by wind might account for a small amount of sediment removal – especially fine-grained silts or clays – but it is unlikely to explain the relief that has been created all the way around Shadow Vlei. Thirdly, the wetland is in an area that receives sufficient rainfall to support permanent ground covering vegetation, particularly as the area receives rainfall throughout the year (Mucina and Rutherford, 2006). Finally, the presence of ferricrete in the depression margin suggests that chemical rather than physical processes are likely to have contributed to the formation of this wetland landform. It is for these reasons that an alternative model of wetland formation is called for.

5.1 Coupled Fe- and Si-cycling and mass balance implications

The process of sagging and volume loss that is observed in the centre of Shadow Vlei, and the swelling in the depression margin, is likely to be a consequence of processes related to coupled redoximorphic reactions involving Fe and Si in association with variation in pH related to Fe reduction and oxidation. In small wetlands, non-flooded and flooded soils exist in close proximity, with flooding leading to anaerobic conditions dominated by reduction reactions, while adjacent non-flooded soils are dominated by oxidation reactions (Mitsch and Gosselink, 2015). Such conditions are likely to exist in and adjacent to Shadow Vlei in response to alternating high and low water levels. A process of iron cycling, whereby iron is constantly being transformed from its ferric (Fe³⁺) to its ferrous state (Fe²⁺) under periodically changing redox conditions, is a process that has been well documented (Schwertmann, 1991; Cornell and Schwertmann, 2003). The reduction of ferric iron generally occurs in waterlogged, anaerobic soils where ferric iron (Fe³⁺) is reduced to ferrous iron (Fe²⁺), with concurrent oxidation of organic matter. In turn, oxidation of ferrous iron occurs in aerobic soils where Fe²⁺ is exposed to oxygen and can act as a primary electron donor.

In a system dominated initially by a uniform Si-dominated lithology, the observed variation in soil chemistry and associated topography can effectively be explained by iron and silica chemistry, on the assumption that aluminum has remained largely immobile. We adopt a simplistic approach by assuming that, during weathering, all iron is converted to stable Fe-(hydr)oxides, while all Al, K, Mg and Na are hosted in the clay fraction (mainly illite). Based on stoichiometric information for typical illite, we can derive the corresponding amount of SiO₂ in the clay mineral structure, calculate the modal abundance of illite, and therefore determine the excess SiO₂ which must be accounted for as various mineral forms of free silica such as micro-crystalline quartz and/or chalcedony (conveniently grouped here under the term “quartz”).

On the basis of Al immobility, we infer that the antithetic relationship between modal illite and quartz (Fig. 8) must be due to the effect of preferential silica mobility during periods of iron reduction. Si mobility would be associated with flooding and microbial decomposition of organic matter by iron oxyhydroxide at the depression center, driven by a resultant increase in alkalinity (Brinkman 1970; Hobson & Dahlgren 1998). Given that silica is relatively soluble under periodically high pH, we propose that flooding of soils is likely to be associated with preferential dissolution of iron (as Fe²⁺) as well as silica (as H₄SiO₄) and their mobilisation away from the centre of the depression to the margins thereof.

Insert Fig. 8 here

The relationship for the plot of Fe-oxide against modal illite is broadly positive but slightly more complex, revealing distinct clusters of data distribution with variable slopes, recording correspondingly variable modal illite/Fe ratios (Fig. 9A). Those samples with a relatively high illite/Fe ratio, which reflect Fe depletion, occur near the depression surface at a depth of less than 1.5 m (Fig. 9B), and correspond largely with Clusters 2 and 3 in the cluster analysis. However, for samples with a low illite/Fe ratio, which represent samples enriched in Fe, a further distinction can be made into two sub-populations of data using the conventional value of 5 wt%: those with less than 5 wt% FeO, labelled here as low illite/low Fe, and those with greater than 5 wt% FeO, labelled as low illite/high Fe. Samples with low illite/low Fe correspond to Cluster 1 of the cluster analysis and are distributed at depth in the centre and margins of the depression and at a range of depths beyond the centre and margins of the depression. Samples with low illite/high Fe occur in the depression margins near the soil surface (less than 2 m depth).

Insert Fig. 9 here

During the oxidizing phase of iron cycling, Fe²⁺ is oxidised to Fe³⁺ to form insoluble ferric (oxy)hydroxide such as goethite FeOOH or ferrihydrite Fe(OH)₃ (Cornell and Schwertmann, 2003). The reaction that occurs during oxidation of Fe²⁺ produces acidity and thus lowers the pH in the soil profile during dry periods when oxygen diffuses into the soil (Mann, 1983). The resultant decline in pH produced during this reaction would be responsible for lowering solubility and causing at least partial re-precipitation of the silica that was rendered mobile during the iron-reductive stage at the depression centre.

In order to examine the dynamics of chemical mass balance in more detail, we have graphically linked changes in soil geochemistry to corresponding mass changes during the weathering process, by employing the classic isocon diagram (Grant, 1986). In the case of Shadow Vlei, which occupies the crest of a dissected interfluve on the African Erosion Surface such that the water table slopes away from the depression (Melly et al., 2017; Schael et al., 2015), we have assumed that the meteoric fluids involved are unlikely to have caused major metasomatic introduction of new soluble material from sources exotic to the system. We therefore consider the Shadow Vlei sub-system as effectively semi-closed, characterised largely by internal chemical redistribution and only partial solute loss away from the system during periods of increased mineral solubility.

For Shadow Vlei, our application of the isocon diagram links the mean values for all samples – which we assume closely approximate the primary bulk soil composition – with the bulk compositional averages of the three different sample populations identified in Fig. 9A. The results are shown in Fig. 10. Aluminum immobility is based on the bulk clay mineral fraction (mainly illite) being the predominantly stable mineral species in the Shadow Vlei system compared to both free SiO₂ mineral forms and iron oxides. This renders Al as the conservative element of choice during weathering and depression development. This is supported by the statistically significant relationship between the generally insoluble TiO₂ and Al₂O₃ (R² = 0.727; P < 0.0001) for all samples excluding those at the near-surface samples (< 0.2 m depth), where bioturbation may redistribute material in a manner unrelated to weathering.

In the depression at a depth of less than 1.5 m, soils have a high illite/Fe ratio such that against Al₂O₃, SiO₂ appears to show substantial relative depletion (Fig. 10 IA). The same is true for Fe₂O₃, which is also depleted relative to Al₂O₃ (Fig. 10 IB). In the depression at a depth below where the illite/Fe ratio is high, the illite/Fe ratio is low and the Fe₂O₃ concentration is also low (low illite/low Fe sample group). For these samples, and relative to Al₂O₃, SiO₂ appears to have been enriched (Fig. 10 IIA), but there has been a net loss of Fe (Fig. 10 IIB). A corresponding decrease in LOI reflects this relative dearth of Fe oxyhydroxide and clay minerals as these are the key carriers of volatiles in the soils.

In the depression margins close to the surface (to a depth of 2 m), SiO₂ is slightly enriched (weathered / primary composition = 0.908) relative to Al₂O₃ (weathered / primary composition = 0.907; Fig. 10 IIIA). However, Fe₂O₃ has been considerably enriched relative to Al₂O₃ (weathered/primary composition = 2.58; Fig. 10 IIIB). These samples are referred to as low illite/high Fe. Comparison of parent and weathered material shows a 7.97 wt% enrichment of Fe₂O₃ and a (-)7.58 wt% depletion of bulk SiO₂ in the depression margins, suggesting that net Fe₂O₃ enrichment largely accounts for the observed SiO₂ dilution effect. These data collectively suggest that enrichment of Fe₂O₃ in the depression margin, together with some SiO₂ addition, contribute most to the volume gains.

Insert Fig. 10 here

These analyses shed light on the underlying processes contributing to variation in soil chemistry and the observed topography. In the near-surface environment in the centre of the depression where organic matter is present, iron reduction takes place that increases the pH of meteoric water to a point where mineralogical forms of free silica are relatively soluble, leading to dissolution and loss of SiO₂ to the fluid medium that amounts to 7.22 wt% of the parent rock mass. This loss of SiO₂ is approximated by an increase in Al₂O₃ (-2.98 wt%) and LOI (-3.37 wt%), both hosted primarily in the bulk clay mineral fraction. However, below the depth where organic matter is present in the central depression, it seems that iron reduction and associated pH changes are limited by the lack of organic matter, such that dissolution of SiO₂ is curtailed. Therefore, observed relative declines in bulk Al₂O₃ (-3.18 wt%), Fe₂O₃ (-3.74 wt%) and LOI (-3.57 wt%) appear to be counter-balanced by the relative excess in bulk SiO₂ by approximately 11.74 wt%.

5.2 Iron-cycling And The Origin And Development Of Shadow Vlei

Processes of iron cycling under alternating oxidising and reducing conditions have the potential to explain the occurrence of ferricrete in the marginal areas of Shadow Vlei as the occurrence of ferricrete is commonly associated with a fluctuating water table (Pain and Ollier, 1992; Phillips et al., 1997; Hobson and Dahlgren, 1998). Ferricrete duricrusts are often located a short vertical distance above the average elevation of the water table as iron is extremely sensitive to oxidising conditions and will precipitate out of solution rapidly upon exposure to oxygen. The occurrence of ferricrete in the marginal areas of depression wetlands has been widely documented in both Australia and the United States and is said to be a direct result of lateral transport of iron in ground water and exposure of dissolved iron to oxidising conditions (Pain and Ollier, 1992; Hobson and Dahlgren 1998). Ferricrete duricrusts associated with depressions are rarely thicker than 2 m but can cover extensive areas depending on the extent of the iron supply.

The creation of positive relief has been attributed to the formation of ferricrete in seasonally wet conditions in Australia (Conacher et al., 1991; Pain and Ollier, 1992), Mali and Burkina Faso (Butt and Bristow, 2013), and Central Sudan (Schwarz, 1994). Ferricrete creates positive relief both as a consequence of the accumulation and consolidation of ferruginous material such as goethite, haematite and limonite in a single location (in this case in the depression margin), leading to swelling, and also as a consequence of the lowering of the adjacent surface due to volume losses associated with weathering.

During its early formation, the depth of Shadow Vlei might have been measured in decimeters, and the diameter in meters to tens of meters. Weathering would likely have been limited to a few meters below the soil surface. In the presence of organic matter in the near-surface soil, sufficiently prolonged inundation of such a shallow and small depression would promote the presence of anaerobic conditions, leading to iron reduction and increased alkalinity in the near-surface weathering soil environment. Under these conditions, silica solubility may be favoured and, provided localised recharge of groundwater was taking place from the depression, SiO₂ would be transported away from the centre of the depression where the soil is flooded for longest, to be precipitated in the depression margins due to the lowering of pH associated with iron oxidation (Fig. 11A). As such, during the early stages of depression development, the near-surface soils at the centre of the shallow depression would become depleted in both iron and SiO₂, leading to a net volume loss. Lateral movement of dissolved silica (and iron) from the more permanently flooded soils beneath the centre of the depression and their accumulation in the margins in the form of silica-indurated ferricrete, would have led to the creation of minor topographic relief. Over long periods, the depth of weathering would have increased such that in the subsoils, where the amount of organic matter was too low to promote sufficiently alkaline conditions, silica solubility would have been impeded and possibly accompanied by top-down re-precipitation beneath the depression Fig. 11B). By comparison, the depression margins became relatively more iron-rich, leading to ferricrete formation and aggradation.

Insert Fig. 11 here

Iron cycling occurs primarily in the marginal areas of the wetland for several reasons. The process requires alternating oxidising and reducing conditions produced by a fluctuating water table as well as a constant supply of iron. The movement of iron from the centre towards the margin of the depression due to groundwater recharge means that iron is readily available in the depression margin. Furthermore, iron cycling is dependent on microbial activity associated with oxidation of organic substrates. The central areas of Shadow Vlei are always vegetated and preferred sites of organic matter accumulation in the soil due to prolonged flooding, such that there is likely to be a sufficient supply of organic material for soil microbes to use as electron sources. The combination of chemical weathering in the centre, and ferricrete formation in the margins of Shadow Vlei, has led to a combined sagging/aggradation effect that has created sufficient relief to form a depression wetland.

The origin of depression wetlands is generally poorly understood and there are clearly a range of processes that lead to the formation of features that are morphologically similar. The aeolian deflation model of Goudie and Thomas (1985) gives rise to a depression with a leeward mound, but models that involve biogeochemical processes generally give rise to depressions with no mound in the depression margin (Dartmoor Vlei described by Edwards et al., 2016), or with a mound surrounding the depression (Okavango island depressions described by McCarthy et al., (1993), Grahamstown wetland described by Alistoun (2014) and Shadow Vlei described in this study). Of these models, the Okavango island depressions involve transpiration-driven precipitation of allochthonous solutes that enter the ecosystem from weathering in the catchment. The other models involve autochthonously derived solute transfer from the depression centre, either to be exported (Dartmoor Vlei) or redistributed locally (Grahamstown wetland and Shadow Vlei). In the case of Dartmoor Vlei, export of weathered solutes probably occurs due to the high rainfall and semi-permanently flooded conditions that prevail. The Grahamstown wetland and Shadow Vlei occur in somewhat drier conditions, and the prevailing circumstances that give rise to these features are related to redox-driven solute dissolution in the depression and transpiration-driven precipitation of solutes in the margin (Grahamstown wetland) or redox- and associated pH-driven dissolution and precipitation of solutes in the depression and margin respectively (Shadow Vlei). The dominant processes that prevail in these cases seem to be related to the mineralogy and chemical composition of bedrock and associated soils. Generalised models of depression wetland formation will require more widespread examination of these features given our relatively poor understanding at present.

Acknowledgements

This research was supported financially by the National Research Foundation (NRF grant number: 93590) and by Rhodes University (RU grant number: 37407). Thanks to Dr Phumelele Gama, Dr Denise Schael, Dr Brigitte Melly and Mr Japie Buckle for introducing us to the study site. Thanks to Mr Julián Arenzon, Mr JC Bonnici, Mr Simon Bremner, Ms Gina Frangs and Mr Geoffrey Smuts who assisted with field work. Thanks to Ms Caitlin Rybko and Mr Abe Ngoepe for their assistance in the labs.

Alistoun J. 2014. The origin of endorheic pans on the African Erosion Surface north of Grahamstown, South Africa. Rhodes University: Grahamstown. (MSc Dissertation) [PDF].
Bish DL, Howard SA. 1988. Quantitative phase analysis using the Rietveld method. Journal of Applied Crystallography 21(2) : 86–91. DOI: 10.1107/S0021889887009415
Brinkman R. 1970. Ferrolysis, a hydromorphic soil forming process. Geoderma, 3(3):199–206. DOI: 10.1016/0016-7061(70)90019-4
Butt CRM, Bristow APJ. 2013. Relief inversion in the geomorphological evolution of sub-Saharan West Africa. Geomorphology 185(1) : 16–26. DOI: 10.1016/j.geomorph.2012.11.024
Cabadas-Báez H, Solleiro-Rebolledo E, Sedov S, Pi-Puig T, Gama-Castro J. 2010. Pedosediments of karstic sinkholes in the eolianites of NE Yucatán: A record of the Late Quaternary soil development, geomorphic processes and landscape stability. Geomorphology 122(2-3) : 323-337. DOI: 10.1016/j.geomorph.2010.03.002
Conacher AJ, Ollier CD, Galloway RW. 1991. Lateritic duricrust and relief inversion in Australia. Catena 18(1) : 585–588. DOI: 10.1016/0341-8162(91)90041-U
Cornell RM, Schwertmann U. 2003. Th iron oxides: structure, properties, reactions, occurrences and uses (2^nd Edition). WILEY-VCH Verlag GmbH & Co. Weinheim.
Doğan U, Özel S. 2005. Gypsum karst and its evolution east of Hafik (Sivas, Turkey). Goemorphology 71 (3-4) : 373-388. DOI: 10.1016/j.geomorph.2005.04.009
Edwards RJ, Ellery WN, Dunlevey J. 2016. The role of the in situ weathering of dolerite on the formation of a peatland: The origin and evolution of Dartmoor Vlei in the KwaZulu-Natal Midlands, South Africa. Catena 143(2):232–243. DOI: 10.1016/j.catena.2016.04.016
Ellery WN, Grenfell SE, Grenfell MC, Humphries MS, Barnes K, Dahlberg A and Kindness A. Peat formation in the context of the development of the Mkuze floodplain on the coastal plain of Maputaland, South Africa. Geomorphology 141-142:11-20. doi:10.1016/j.geomorph.2011.11.009
Fey M. 2010. Soils of South Africa. Cambridge University Press: Cape Town.
Filizola HF, Boulet R. 1996. Evolution and opening of closed depressions developed in a quartz-kaolinitic sedimentary substratum at Taubaté basin. Geomorphology 417:1-10.
Goudie AS, Thomas DSG. 1985. Pans in southern Africa with particular reference to South Africa and Zimbabwe. Zeitschrift fur Geomophologie 29(1):1–19.
Grant, J.A. The isocon diagram-a simple solution to Gresens' equation for metasomatic alteration. Economic Geology, 81(8):1976-1982. DOI: 10.2113/gsecongeo.81.8.1976
Grenfell SE, Ellery WN, Grenfell MC, Ramsay LF and Flugel TJ. 2010. Sedimentary facies and geomorphic evolution of a blocked-valley lake: Lake Futululu, northern KwaZulu-Natal, South Africa. Sedimentology 57:1159–1174.
Gutiérrez F, Parise M, De Waele J, Jourde H. 2014. A review on natural and human-induced geohazards and impacts in karst. Earth Science Reviews 138:61-88. DOI: 10.1016/j.earscirev.2014.08.002
Hobson WA, Dahlgren RA. 1998. Soil forming processes in vernal pools of northern California, Chico area. In: CW Witham, ET Bauder, D Belk, WR Ferren and R Ornduff (eds.). Ecology, conservation and management of vernal pool ecosystems. Proceedings from a 1996 Conference. California Native Plant Society, Sacramento, CA. pp 24-37.
IUSS Working Group WRB. 2015. World Reference Base for Soil Resources 2014, update 2015. International soil classification system for naming soils and creating legends for soil maps. World Soil Resources Reports No. 106. FAO, Rome.
Joubert R, Ellery WN. 2013. Controls on the formation of Wakkerstroom Vlei, Mpumalanga province, South Africa. African Journal of Aquatic Science 38(2):135–151. DOI: 10.2989/16085914.2012.762897
Jury MR, Levey K. 1993. The climotology and characteristics of drought in the eastern Cape of South Africa. International journal of climatology 13(6) 629-641. DOI: 10.1002/joc.3370130604
Macvicar CN, Soil Classification Working Group. 1991. Soil classification: A taxonomic system for South Africa (2e). Department of Agricultural Development: Pretoria
Mann AW. 1983 Hydrogeochemistry and weathering on the Yilgarn Block, Western Australia-ferrolysis and heavy metals in continental brines. Geochimica et Cosmochimica Acta 47(2):181–190. DOI: 10.1016/0016-7037(83)90131-X
McCarthy TS, Ellery WN, Ellery K. 1993. Vegetation-induced, subsurface precipitation of carbonate as an aggradational process in the permanent swamps of the Okavango (delta) fan, Botswana. Chemical Geology 107(1–2):111–131. DOI: 10.1016/0009-2541(93)90105-R
McCarthy TS and Metcalfe J. 1990. Chemical sedimentation in the semi-arid environment of the Okavango Delta, Botswana. Chemical Geology 89:157-178. https://doi.org/10.1016/0009-2541(90)90065-F
McCarthy TS, Rowberry MD, Brandt D, Hancox PJ, Marren PM, Tooth S, Jacobs Z, Thompson M, Woodborne S, Ellery WN. 2011. The origin and development of the Nyl river floodplain wetland, limpopo province, South Africa: Trunk-tributary river interactions in a dryland setting. South African Geographical Journal 93(2):172–190. DOI: 10.1080/03736245.2011.619324
Melly BL, Schael DM, Gama PT. 2017. Perched wetlands: An explanation to wetland formation in semi-arid areas. Journal of Arid Environments 141:34–39. DOI: 10.1016/j.jaridenv.2017.02.004
Midgeley DC, Pitman WV, Middleton BJ. 1994. Surface Water Resources of South Africa. Water Research Commission: Pretoria.
Mitsch WJ, Gosselink JG. 2015. Wetlands (5^th Edition). John Wiley & Sons: New York.
Mucina L, Rutherford MC. 2006. The vegetation of South Africa, Lesotho and Swaziland. South African National Biodiversity Institute: Pretoria.
Munsell Color Co., Inc., 1992, Munsell soil color charts (revised ed.): Baltimore.
Norrish K, Hutton JT. 1969. An accurate X-ray spectrographic method for the analysis of a wide range of geological samples. Geochemica et cosmochimica acta 33(4):431-453. DOI: 10.1016/0016-7037(69)90126-4
O’Connor TG, Crow VRT. 1999. Rate and pattern of bush encroachment in Eastern Cape savanna and grassland. African Journal of Range and Forage Science 16(1):26–31. DOI: 10.2989/10220119909485715
Pain CF, Ollier CD. 1992. Ferricrete in Cape York Peninsula, north Queensland. BMR Journal of Australian Geology & Geophysics 13(3):207–212.
Partridge TC, Maud RR. 1987. Geomorphic evolution of southern Africa since the Mesozoic. South African Journal of Geology 90(2):179–208.
Phillips JD, Lampe M, King RT, Cedillo M, Beachley R, Grantham C. 1997. Ferricrete formation in the North Carolina coastal plain. Zeitschrift fur Geomophologie 41(1):67–80.
Schael DM, Gama PT, Melly BL. 2015. Ephemeral Wetlands of the Nelson Mandela Bay Metropolitan Area : Classification , Biodiversity and Management Implications. [Online] Water Research Comission: Pretoria. DOI: 10.13140/RG.2.1.2029.9122 [27/12/2017]
Schwarz T. 1994. Ferricrete formation and relief inversion: an example from Central Sudan. Catena 21(2–3):257–268. DOI: 10.1016/0341-8162(94)90016-7
Schwertmann U. 1991. Solubility and dissolution of iron oxides. Plant and Soil 130:1-25. https://0-doi.org.wam.seals.ac.za/10.1007/BF00011851
Soil Classification Working Group. 2018. Soil Classification: A natural and anthropogenic system for South Africa. ARC-Institute for Soil, Climate and Water. Government Printer, Pretoria.
Thamm AG, Johnson MR. 2006. The cape Supergroup. In The Geology of South Africa, Johnson MR, Anhaeuser CR, Thomas RJ. (eds). Geological Society of South Africa: Johannesburg. 443-460.
Thomas DSG. 2011. Arid environments: their nature and extent. In Arid zone geomorphology: process, form and change in drylands (3e.), Thomas DSG. (eds). John Wiley & Sons: New York, 1–16.
Timms BV. 2006. The geomorphology and hydrology of saline lakes of the middle Paroo, arid-zone Australia. Proceedings of the Linnean Society of New South Wales 127:157-174.
Tooth S, McCarthy TS. 2007. Wetlands in drylands: geomorphological and sedimentologicalcharacteristics, with emphasis onexamples from southern Africa. Progress in Physical Geography31(1):3–41.

No competing interests reported.

Download PDF

Journal Publication

published 12 Jan, 2024

Read the published version in Wetlands Ecology and Management →

Editorial decision: Accepted
23 Nov, 2023
Reviewers agreed at journal
28 Mar, 2023
Reviewers invited by journal
26 Feb, 2023
Editor assigned by journal
22 Feb, 2023
Submission checks completed at journal
22 Feb, 2023
First submitted to journal
22 Feb, 2023

You are reading this latest preprint version

Depression wetland formation by redox-driven iron and silica cycling

Status:

Journal Publication

Version 1

Abstract

Figures

1 Introduction

2 Study Area

3 Methods

4 Results

5 Discussion

5.1 Coupled Fe- and Si-cycling and mass balance implications

5.2 Iron-cycling And The Origin And Development Of Shadow Vlei

6 Conclusion

Declarations

Acknowledgements

References

Additional Declarations

Status:

Journal Publication

Version 1