The role of airborne dust in the growth of tree islands in the Okavango Delta, Botswana

Geomorphology 206 (2014) 307–317

Contents lists available at ScienceDirect

Geomorphology journal homepage: www.elsevier.com/locate/geomorph

The role of airborne dust in the growth of tree islands in the Okavango Delta, Botswana M.S. Humphries a,⁎, T.S. McCarthy b, G.R.J. Cooper b, R.A. Stewart b,c, R.D. Stewart d a

Molecular Sciences Institute, School of Chemistry, University of the Witwatersrand, Private Bag 3, Wits 2050, Johannesburg, South Africa School of Geosciences, University of the Witwatersrand, Private Bag 3, Wits 2050, Johannesburg, South Africa Gold One, P Bag X17, Weltevreden Park 1715, Johannesburg, South Africa d Shango Solutions, Box 2591, Johannesburg, South Africa b c

a r t i c l e

i n f o

Article history: Received 4 June 2013 Received in revised form 12 September 2013 Accepted 16 September 2013 Available online 18 October 2013 Keywords: Tree islands Okavango Delta Dust accumulation Kalahari Makgadikgadi pans

a b s t r a c t The Okavango Delta, situated in the Kalahari Desert (Botswana), is host to extensive areas of seasonal swamp that are characterised by the presence of numerous tree-covered islands. Islands have been shown to play an important role in the landscape through the creation of habitat diversity, focusing of nutrients, and sequestration of salts. Islands are thought to grow through the subsurface precipitation of carbonate and silica, although recent work has suggested that other factors may be involved, notably airborne dust accumulation. In this study, we investigate the role of airborne dust in the maintenance and growth of tree islands in the seasonal swamps of the Okavango Delta. Chemical and grain size analyses indicate that whilst the channels and floodplains in the Okavango are dominated by well-sorted Kalahari sand that covers the entire region, material on the surface of islands is distinctly different. Island soils are enriched in Al2O3 and are characterised by higher proportions of poorly sorted, fine-grained material that we attribute to the addition of airborne dust. Large quantities of material that circulate in anticyclonic systems over southern Africa represent a potentially significant source of particulate sediment to the Okavango, whilst peat fires and the desiccation of the surrounding floodplains during the dry season are also considered to be important sources of local dust. The data suggest that varying proportions of Kalahari sand, dust, and chemical precipitate give rise to the range of compositions found on and within the islands of the Delta. Dust typically accounts for between 20 and 60% of material found on the surface of islands, whilst dust and chemical precipitate dominate the subsurface material. Islands thus appear to grow through a combination of ongoing surface and subsurface processes that result in considerable heterogeneity in soil composition. In both instances vegetation, especially trees, is the main driving force behind island development, not only causing the subsurface accumulation of CaCO3 and SiO2 in island soils, but also trapping airborne dust on the surface of islands. Our study suggests that dust fallout is an equally or possibly even more important contributor to the local topographic irregularities and thus habitat diversity in the Okavango Delta. Despite the potential importance of airborne material to the biogeochemistry and development of tree islands in wetland systems, our knowledge regarding these processes remains poor. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Tree islands form an important component of many wetlands because they increase habitat diversity in the landscape. They have been described from many parts of the world including the Okavango Delta in Botswana (Ellery et al., 1993; Ramberg and Wolski, 2008), the Everglades in Florida (Sklar and Van der Valk, 2002; Wetzel et al., 2011), the Pantanal of South America (Ponce and da Cunha, 1993), and the Nylsvlei wetland of South Africa (Tooth et al., 2002). Islands exhibit unique characteristics that differentiate them from the surrounding landscape and have been recognised as biogeochemical hot spots, serving as important sites for nutrient and chemical ⁎ Corresponding author. Tel.: +27 117176739. E-mail address: [email protected] (M.S. Humphries). 0169-555X/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.geomorph.2013.09.035

accumulation (McCarthy et al., 1993; Ellery et al., 1998; Troxler Gann et al., 2005). Although the mechanisms are still not fully understood, islands are thought to develop autogenously as a result of landscape processes and feedback between hydrological, biological, and climatic factors (Ellery et al., 1993; Sullivan et al., 2010; Wetzel et al., 2011). The origin of islands on the Okavango alluvial fan (Fig. 1) has been the focus of detailed study for a number of years (McCarthy et al., 1993; McCarthy, 2006; McCarthy et al., 2012). Morphological examination has revealed that some islands originate through fluvial activity on the fan surface and represent scroll bars or abandoned channels. However, the majority of islands appear to be unrelated to fluvial processes. These islands owe their origin to mound-building termite colonies. Once established, such elevated areas can support shrubs and trees, which are unable to survive on the seasonally flooded floodplains. Islands are

308

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

Fig. 1. The Okavango Delta in northern Botswana showing the location of the study area. The inset shows the extent of the Kalahari sands and mean annual rainfall distribution (in mm) over southern Africa.

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

thought to grow, in part, through the precipitation and subsurface accumulation of silica and calcite. The main process causing this precipitation is transpiration, particularly by trees, which results in an increase in groundwater solute concentration and ultimately to saturation in silica and calcite (Fig. 2A). Potential evaporation exceeds rainfall during all months of the year, and 96% of the water entering the Delta is lost to the atmosphere by evapotranspiration (Wilson and Dincer, 1976). Chemical precipitation thus constitutes a major sedimentary process on the fan, and ~360,000 t of dissolved chemicals (composed mainly of calcium, magnesium carbonate, and silica) is estimated to accumulate annually beneath islands. The high transpiration rate of island vegetation causes a local lowering of the water table, resulting in the continual flow of groundwater from the surrounding swamps to the islands. Ultimately, the salinity of groundwater beneath islands increases to the point where densitydriven subsidence occurs (Bauer-Gottwein et al., 2007). Islands, therefore, represent the final sink for groundwater solutes and play an important role in maintaining surface water of low salinity, particularly under an arid climate (McCarthy, 2006; Ramberg and Wolski, 2008). Vegetation on islands typically shows marked zonation, from evergreen tree species on the outer fringes, passing inward to deciduous tree species, and eventually to grassland interspersed with bare soil, often covered by a thin crust of sodium carbonate salts (Fig. 2B). This zonation is the result of pronounced gradients in groundwater salinity beneath the islands and reflects differences in the salt tolerance of island vegetation (Ellery et al., 1993). We recently calculated that ~30–40% of the total volume of islands could be attributed to chemical precipitation (McCarthy et al., 2012). The study suggested that continued island growth was thus a combination of processes possibly including the deposition of windborne

309

aerosols and local accumulation of dust, as well as ongoing termite activity. Airborne dust has been found to deposit preferentially on islands (Krah et al., 2004), although its contribution to total island volume is not known. The importance of airborne particulates to the functioning of ecosystems has been widely recognised. Because large quantities of airborne material can be transported over great distances, it is believed to play an important role in many biogeochemical processes, influencing soil characteristics, ocean productivity, and air chemistry (Prospero et al., 2002; Washington et al., 2003). Deserts and semiarid regions are known to be the main sources of global dust (Simonson, 1995; Prospero et al., 2002). Large quantities of material circulate in anticyclonic systems over southern Africa with an estimated 11.5 million t of material transported annually across the Okavango region (Tyson et al., 1996). Annual deposition of aerosols on the Delta from the atmosphere under anticyclonic conditions is estimated at only 250,000 t and represents more than half of the particulate sediment entering the Delta (Garstang et al., 1998). Airborne dust may thus constitute an important process contributing to island growth in the Okavango Delta. In order to explore this hypothesis, the topography, chemistry, and grain size characteristics of three islands in the seasonal swamps of the Delta were studied. Our aim was to provide insight into the relative contribution of airborne dust to tree island growth and local scale geomorphology. 2. Study area The Okavango Delta is located in the Kalahari Desert, Botswana, and forms part of the endorheic drainage of the Kalahari basin (Fig. 1). The Kalahari is characterised by aeolian sand that extends from the Equator

Fig. 2. (A) Schematic cross section of a typical island in the Okavango swamps illustrating the hydrogeochemical processes that result in the subsurface precipitation of minerals, mainly CaCO3 and SiO2. (B) Islands of various sizes and shapes in the permanent swamps showing the distinct zonation in vegetation that characterises islands.

310

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

in the north to the Orange River in the south, covering an area of 2.5×106 km2. The Delta is a large alluvial fan (40,000km2) and is hosted in a depression within the Kalahari basin that is an extension of the East African Rift system (Gumbricht et al., 2001). The wetlands are supplied by the Okavango River, whose headwaters arise in central Angola. In the upper reaches of the Delta, the Okavango River is confined in a narrow (b 12km) depression known as the Panhandle, but divides into a number of distributary channels farther downstream that disperse across a wide area to form the fan. The area around the fan is characterised by relict aeolian features consisting of linear dunes in the west and south (Fig. 3). Dune ridges are orientated in an ESE–WNW direction and reflect the dominant wind direction at the time of their formation. Although degraded, the dunes are up to 25 m high and some ridges extend for a distance of 200 km (Lancaster, 2000). The dune ridges terminate along the edge of the fan, and the floodplain itself is devoid of aeolian features, although soils are typically very sandy. The catchment is underlain almost entirely by aeolian Kalahari sand (Fig. 1, inset), and fluvial sediment entering the Okavango Delta (estimated to be 170,000t per annum) consists of fine sand, transported primarily as bedload (McCarthy et al., 1991a). Suspended sediment load in the Okavango River in the Panhandle is very low (~8–9 mg/L) and ~30,000 t of suspended sediment, consisting mainly of kaolinite, enters the fan annually (McCarthy et al., 1991a). Upper-fan and Panhandle channels are bound by thick stands of sedges (e.g., Cyperus papyrus) and emergent grasses (e.g., Phragmites spp.) rooted in peat. Water that leaks from the channels sustains large areas (ca. 12,000 km2) of permanent wetlands (Fig. 1). The dense vegetation filters suspended sediment and particulate material is thus almost completely deposited in the vicinity of channels. As a consequence, very little clay is introduced onto the distal seasonally flooded areas of the fan.

Under present conditions, channel systems are estimated to have a limited lifespan, forming and abandoning over a period of about 100 years (McCarthy et al., 1992). Following abandonment, peat desiccates in the arid climate and is destroyed by slow-burning peat fires (Ellery et al., 1989), leaving a residue of fine ash. Regular shifts in water and sediment across the fan have produced a surface with remarkably uniform gradient (1:3500; Gumbricht et al., 2001). The Okavango is subject to annual flooding that produces great hydrological variability. The flood wave results in an expansion of the inundated area of the Delta from around 5000 km2 to an annual maximum of 6000–12,000 km2 (Wolski et al., 2006). The local scale topography in the more distal portion of the Delta is gently undulating, with local relief in the region of 1.5 to 2 m. This higher ground forms islands during periods of flooding. Islands vary in size from small, irregular features b 1 m2 in area to large features covering many kilometres. Islands support trees, whereas the floodplains are vegetated mainly by short grasses and sedges. 3. Methods 3.1. Topographic surveying and sample collection Three islands (A, C, D) situated in the seasonal swamps of the Delta were selected for study (Fig. 4). Islands selected were in close proximity to one another and ranged in diameter from 300 to 800 m. Traverses were surveyed across the islands in a direction parallel to the prevailing wind (on a bearing of 290°) using a dumpy level and staff. One traverse was surveyed across island C, whilst two traverses were surveyed across the two larger islands (A and D). In addition to these five traverses, an additional ‘control’ traverse was surveyed at 90° to the prevailing

Panhandle

Linear dunes

N Kunyere Fault

Fig. 3. SRTM-derived DEM of the Okavango Delta showing the ESE–WNW aligned linear dune ridges that occur along the western boundary of the fan. The fan surface is characterised by a lace-like pattern of ridges and depressions. Inset: wind rose showing wind intensity and the prevailing direction (recorded at Maun).

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

311

Fig. 4. Google Earth image showing the islands studied and orientation of the traverses surveyed.

wind direction (on a bearing of 020°) on Island A. Surface (top ~ 5 cm) soil was collected along each traverse, sampling at 10-m intervals in the tree fringe area of the islands and at 20-m intervals on the floodplain and island centres.

3.2. Grain size and chemical analyses Samples were subjected to grain size analysis using a Malvern Mastersizer. Particle size analyses were also conducted on a selected number of dust samples collected using dust samplers by Krah et al. (2004). Mean grain size and sorting were calculated following Folk and Ward (1957). Major elements were analysed using a handheld Thermo Niton XRF analyser. As an internal check, selected samples were subject to full chemical analysis by XRF spectrometry, following fusion with Li2B4O7. Samples of dust from filters were also analysed by XRF spectrometry using fusion. Analytical errors in the case of Si are considered to be b3% (relative) and for Ca, Al, and Fe, b 10%. We hypothesized that islands are constructed from three primary components: aeolian Kalahari sand, airborne dust, and solutes precipitated in island soils during transpiration of groundwater. The chemical composition of each of these components is known and hence the proportion of each in samples of island material could be calculated using a least squares mixing procedure. Mixing calculations were carried out on the surface samples collected in this study and on subsurface material collected in previous studies (McCarthy and Metcalfe, 1990; McCarthy et al., 1991b). The end-member compositions used are given in Table 1.

4. Results 4.1. Island topography and grain size characteristics Topographic profiles of the islands investigated are shown in Fig. 5. Island elevation varies from 0.5 to 2.5 m relative to the floodplain. As documented in other studies from the Delta (e.g., McCarthy et al., 1993), islands often show elevated margins characterised by trees and palms (Figs. 4 and 5). The centres of islands are sparsely vegetated

Table 1 End-member percentage compositions used in the mixing calculation. End-member

SiO2

Al2O3

Fe2O3

CaO

Kalahari sand Dust Chemical precipitatea

100 85.8 56.0

0 7.6 0

0 2.6 0

0 1.1 24.6

a

Data from McCarthy et al. (2012).

Fig. 5. Island topographic profiles. The location of the tree fringe associated with each island is indicated by the broken line. The prevailing wind direction is indicated by the arrows.

312

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

and characterised by short grasses and shrubs, whilst grasses dominate the adjacent floodplain area. No obvious relationship between the prevailing wind direction and the morphology of islands is evident. Island surface soils are generally composed of medium to fine sand, although they display a wide range in mean grain size, varying from 78 to 318 μm (average = 213 μm). No systematic variation in mean grain size or sorting was observed across islands or between individual islands (Fig. 6). Samples tend to be poorly to extremely poorly sorted, with standard deviation ranging between 1.7 and 4.7 (Fig. 7). Surface samples from the floodplain consist predominately of coarser material, with a mean grain size of 266 μm, generally characteristic of Kalahari sands (Fig. 7). Material from the floodplains also tended to be better sorted (σ1 = 2.0–2.4) than surface material on islands. Despite little observable variation in grain size across islands, marked differences between island soils and those from the surrounding floodplain are noticeable. Compared with sands from the floodplain, island surface soils show distinctly higher proportions of finegrained (b 100 μm) particulates (Fig. 8). Grain size distributions on islands thus tend to be bimodal in nature, reflecting the addition of fine-grained material. Examination of dust samples shows that local airborne material is characterised by fine-grained (b100 μm), extremely poorly sorted (σ1 = 3.4–4.4) particles (Figs. 7 and 8). Sands from river channels are reasonably well-sorted (σ1 =1.4); and although they display a moderate range in mean grain size (328–565 μm), they generally resemble typical Kalahari sand in terms of their size distribution (Fig. 8). 4.2. Sediment chemistry Floodplain sands are dominated by well-rounded quartz grains and generally contain in excess of 95% silica. Average Al2O3 and CaO concentrations are very low, typically b2% and b0.5%, respectively. In contrast, surficial island sediments are enriched in Al2O3 (up to 4%) and CaO (up to 7%). Positive correlation between the b63-μm size fraction and Al2O3 and CaO abundances suggests that these components are associated with the fine fraction of sediments (Fig. 9). Sediments from the Okavango Delta can be distinguished from one another based on their chemistry. Fig. 10 shows that whilst there is a large range in the Al2O3 and Fe2O3 content of samples from different

areas of the Delta, most lie along a linear array with end members being zero and dust, which reflects the mixing of dust with an ironand aluminium-free component, viz. Kalahari sand. A plot of %SiO2 vs. %Al2O3 indicates that material from the Okavango falls within an envelope defined by three end-members (Fig. 11). The data suggest that varying proportions of Kalahari sand, dust, and chemical precipitate give rise to the range of compositions found on and within the islands of the Delta. Samples collected from specific environments tend to be dominated by particular end-members. Mixing calculations reveal distinct differences in the composition of surface and subsurface material on islands (Fig. 12). Surface material on islands is generally composed of Kalahari sand with varying proportions of dust. Dust generally contributes between 20 and 60% of the total mass in surface material. Chemical precipitation contributes very little and generally forms b10%, although a single sample taken from a termite mound showed 28%. In contrast, subsurface material from islands shows a high contribution from chemical precipitate (up to 80%) in a number of samples. Dust also appears to be an important contributor in subsurface material, ranging from 20 to 100%.

5. Discussion The channels and floodplains in the Okavango Delta are generally characterised by well-sorted sand with an average grain size of 223– 331 μm, typical of the Kalahari sand that blankets the entire region (Fig. 1, inset). This material is dominated by quartz with very low Al and Ca contents. However, analyses carried out in this study show that material on islands is distinctly different and has been modified by the addition of fine material that is enriched in Al2O3. Because of the low suspended sediment load of the Okavango River and the elevated nature of islands, such fine-grained material could not have been deposited by fluvial processes. Fines introduced by the seasonal flood water consist of clays and fine quartz, and much of this material is filtered by the dense vegetation that flanks the channels, and it therefore accumulates in the upper, permanent swamps of the Delta. Grain size analyses and low Al contents of floodplain soils confirm that very little suspended material is introduced onto the seasonally flooded floodplains. The data suggest

Fig. 6. Variation in mean grain size of surface material from islands.

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

313

Fig. 7. Relationship between mean grain size and sorting for different sediment types in the Okavango Delta. The range for Kalahari sand is shown for comparison (Lancaster, 1986; Livingstone et al., 1999).

that the bulk of the fine material can be attributed to the addition of windborne material. 5.1. Long-range aerosol fallout The Okavango Delta is situated in a semiarid region of southern Africa where annual rainfall is b500 mm. The potential for dust generation is thus significant. Large quantities of material circulate in anticyclonic systems over southern Africa with an estimated 11.5 million t of material transported annually across the Okavango region (Tyson et al., 1996). Garstang et al. (1998) estimated the annual aerosol-borne sedimentation over the Delta to be in the order of 250,000t. This equates to ~30% of the total material deposited annually on the Delta and thus represents a potentially significant source of particulate sediment. The aerosols

contained in these anticyclonic circulations may originate at some distance from the Okavango and cross the Delta approximately in a southeast to northwest direction. Analysis of total ozone mapping spectrometer (TOMS) images indicates two relatively small, but clearly developed dust sources in southern Africa (Prospero et al., 2002; Bryant et al., 2007). Although dust activity can usually be seen throughout the year, activity increases strongly in June–July with maximum activity over August–October (Prospero et al., 2002). Dust activity is centred over the Makgadikgadi pans, located southwest of the Okavango Delta (Fig. 1, inset). These pans are among the world's largest, covering ~22,000km2, and represent a complex of large ephemeral lakes that experience inundation on an intermittent basis. The Makgadikgadi pans have been identified as one of the principal dust sources in southern Africa (Prospero et al., 2002;

Fig. 8. Particle size distributions of representative material from (A) the surface of islands, (B) the floodplain surface, (C) dust collectors, and (D) channels. A typical Kalahari sand (broken line) is shown for comparison.

314

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

Fig. 9. Relationship between the b63-μm size fraction and %CaO and %Al2O3 in island sands.

Washington et al., 2003; Bryant et al., 2007) and are thus likely to be an important point source of atmospheric dust to the Okavango Delta. Etosha pan, located in the extreme northwest of the Kalahari basin, is a second potential source of dust. During some years, TOMS shows dust activity across the entire region extending from the Makgadikgadi pans to the Etosha pan (Prospero et al., 2002).

5.2. Locally generated dust Rainfall over the Okavango Delta is strongly seasonal with effectively no rain falling in the winter months from May to September. During this time, vegetation on the floodplain dies and is often burnt off by fires. Seasonal floodplains remain barren until the flood arrives in late July to August. This creates ideal conditions for dust generation and the entrainment of significant quantities of fine-grained material. Measurements made by Krah et al. (2004) show evidence for the significant movement of dust over both island and floodplains in the seasonal swamps. Dust loads were highest over floodplains and lowest in the sparsely vegetated island interiors. This airborne material consists of fine-grained quartz and amorphous silica (including phytoliths) with assorted clay minerals, especially kaolinite, and is enriched in Al2O3, Fe2O3, and CaO (Table 1). Fine particulates introduced fluvially into the Delta may also ultimately contribute to the dust load. Peat accumulates in the upper, permanent swamps and traps fine particulates, including clay minerals carried into the Delta from the catchment as suspended sediment. Although highly variable, analysis shows that the inorganic fraction of

Fig. 11. Relationship between %SiO2 and %Al2O3 in different materials from the Okavango Delta, showing the three end-members.

peat consists mainly of kaolinite (40%) and quartz (20%) and is Al2O3rich (McCarthy et al., 1989). Peat fires burn constantly in the Delta and the combustion of dry peat leaves behind a very fine-grained inorganic residue. Peat fires create thermals which disperse dust into the atmosphere, where it becomes entrained. Krah et al. (2004) observed significant vertical variations in the dust load, with most of the dust (84%) carried below a height of 3 m above the ground (Fig. 13). This suggests that much of the dust carried in the lower portion of the air column is probably of local origin, being transported and redistributed by gusts of wind. The movement of dust on the floodplains was observed to take place as a result of the development of ‘dust devils’ or strong thermals that cross islands in random directions, especially in spring prior to the onset of the summer rain. Dust loads were found to be highest on the floodplains and lowest over the interiors of islands. Between the floodplains and islands, Krah et al. (2004) measured a 30% reduction in the dust load at a height of 2 m above the ground. Whilst the floodplains are broad, open areas where winds can attain and maintain fairly high velocities, trees in the riparian fringes around islands substantially reduce near-ground wind velocities. This appears to result in the deposition of suspended dust and the accumulation of fine material on islands. These observations are supported by grain size analyses, which showed higher quantities of fine-grained material on islands. Similar interactions between vegetation patches and wind-driven processes have been documented by Field et al. (2012) who identified vegetation height as a key factor determining the ability of vegetation patch types to capture wind-blown material. During the hot summer months, the sparsely vegetated interiors of larger islands may also become sufficiently heated to produce thermals that transfer fine material from island centres to the tree-covered margins, thus further enhancing the topographic difference between island centres and margins (Fig. 5). No obvious relationship between the prevailing wind direction and the morphology of islands is evident. It might be expected that windborne material would accumulate preferentially in the tree fringe on the windward side of islands, but this is not observed (Fig. 5). It thus appears that local eddies and dust devils play a more significant role in redistributing airborne material in the Delta than sustained, groundlevel wind. This may be a consequence of the random distribution of tree islands, which break the flow of air near ground level.

5.3. Island composition and growth

Fig. 10. Relationship between %Al2O3 and %Fe2O3 in different materials from the Okavango Delta.

Although our results are based on measurements from a relatively small number of islands from one region of the Delta, our data suggest that airborne material is important in the construction of tree islands

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

315

Fig. 12. Ternary plots showing the relative contribution of Kalahari sand, dust, and chemical precipitate in (A) island surface material and (B) island subsurface material.

in the Okavango. The growth of islands in the Okavango Delta thus appears to occur through a combination of processes. Once formed, islands are able to support a variety of tree species that are unable to survive on the floodplains. Through transpiration, these trees cause the subsurface precipitation of CaCO3 and SiO2 in island soils. Accumulation of chemical precipitate beneath trees typically results in the development of elevated fringes, which characterise the morphology of many of the islands. In some cases, chemical precipitation can account for up to 80% in subsurface soils. Trees also appear to play an important role in trapping dust on islands. When compared with the surrounding floodplain, surface soils on islands are enriched in Al2O3 and Fe2O3 and contain higher quantities of fine-grained material, reflecting the effects of dust accumulation. A lack of systematic variation of grain size parameters across islands suggests that fine-grained material on islands is subject to constant reworking and redistribution.

We recognise that our results are somewhat limited and site-specific and that variations in the relative importance of airborne material to the growth of islands across the Delta may occur. Islands tend to increase in size and become more numerous toward the distal reaches of the Delta (Gumbricht et al., 2004). This may reflect higher dust loads that develop over the seasonally inundated floodplains. Many islands in the Delta are believed to have originated as termite mounds (McCarthy et al., 1998, 2012), and islands may therefore continue to grow as a result of ongoing termite activity. Chemical analyses of surface material collected from termite mounds show higher Fe2O3/ Al2O3 ratios, suggesting the local addition of Fe to these soils. A similar observation was made by McCarthy et al. (1998) who suggested that this represented Fe precipitated by plants growing on the mounds. Soils in the Okavango lack natural cohesion and termites make use of finegrained material for the construction of their termitaria. McCarthy et al. (1998) identified two types of mortar: one rich in iron oxide and clay, and the other calcite-rich. Termites, therefore, appear to bring calcite and Fe-coated clay to the surface for the construction of their mounds. This accounts for the observed Ca and Fe enrichments seen in surface material from termite mounds. Termites thus probably play a role in mixing surface and subsurface material on islands. In addition, termites probably also continue to transport material from the floodplain to islands as they excavate foraging tunnels beneath the floodplain during the dry season. This material is likely to be dominated by Kalahari sand. 5.4. Implications for other wetland systems

Fig. 13. Variation in total dust collected as a function of height (Krah et al., 2004).

Our study indicates the importance of windborne material to the construction of tree islands in the Okavango Delta, and we expect that other tree island dominated wetland systems are likely to be characterised by similar processes. Soil dust is a major constituent of airborne particles in the global atmosphere and can be transported over thousands of kilometres (Middleton and Goudie, 2001). The Sahara, in particular, has long been recognised as the major source of aeolian soil dust in the world (Swap et al., 1992; Middleton and Goudie, 2001; Washington et al., 2003) and is responsible for producing almost half of the world's mineral dust. Strong convective disturbances that develop over West Africa carry entrained material and dust plumes westward across the North Atlantic, influencing areas as far afield as the Amazon (Swap et al., 1992) and the southeast USA (Prospero, 1999). In particular, African dust is known to comprise an important part of the

316

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317

ambient aerosol in South Florida (Prospero, 1999) and is thus likely to be an important contributor to the soils on islands in the Everglades. Although the potential influence of windborne material on tree islands in the Everglades has been recognised, it has not been studied (Wetzel et al., 2011). Knowledge regarding the influence of airborne material on the biogeochemistry and development of tree islands in the Everglades and other wetland systems in the world represents a major gap in our understanding. 6. Conclusions We now have a fairly comprehensive understanding of how islands in the Okavango Delta form and develop over time. Islands appear to grow through a combination of ongoing surface and subsurface processes that result in the considerable heterogeneity in soil composition observed in this and previous studies. In both instances, vegetation is the main driving force behind island development. Trees not only cause the subsurface accumulation of CaCO3 and SiO2 in island soils, but also play an important role in trapping airborne dust on the surface of islands. Dust typically accounts for between 20 and 60% of material found on the surface of islands, whilst dust and chemical precipitate dominate the subsurface material. Overall, island soils thus owe their character to the addition of airborne dust and accumulation of chemical precipitate. The relative importance of these processes may change as islands evolve in character over time. McCarthy et al. (2012) showed that islands in the Okavango are the product of long-term aggradation processes, likely to develop over extended timescales (10,000–100,000 years). Although initially colonised by trees, the buildup of salinity in the soils eventually results in trees being replaced by more salt-tolerant grasses and shrubs, ultimately giving way to barren, salt-encrusted soil. Transpiration and the subsurface precipitation of minerals as well as dust entrapment thus probably decline and island growth slows or even ceases. However, the distribution of water across the seasonal swamps is constantly changing because of channel failure; and as a consequence, flooding in certain areas may periodically cease completely. During such periods, rain flushes accumulated salts from island soils, thereby regenerating their fertility. Accumulated calcite and silica precipitates are unaffected and thus represent a permanent increment of aggradation on the fan surface. Whilst chemical precipitation was previously considered to be the primary cause of island growth, it is evident from this study that the nonuniform distribution of dust fallout is an equally or possibly even more important contributor to the local topographic irregularities and thus habitat diversity in the Okavango Delta. The capture of windblown sediment and nutrients by tree islands may also have an important biological feedback that further contributes to the development of islands. Whilst airborne dust may have important geomorphological and biogeochemical influences on tree islands in other wetland systems, knowledge regarding such processes remains poor. Acknowledgements We thank the South African National Research Foundation and the University of the Witwatersrand for funding. We also thank Jaqui Areias for assistance with some of the analyses; Lisa Ramsay for use of the Malvern Mastersizer; and I. Mosie, M. Krah, and E. Rice for assistance in the field. The Okavango Research Centre provided logistical support. Constructive reviews from three anonymous reviewers are gratefully acknowledged. References Bauer-Gottwein, P., Langer, T., Prommer, H., Wolski, P., Kinzelbach, W., 2007. Okavango Delta islands: interaction between density-driven flow and geochemical reactions under evapo-concentration. J. Hydrol. 335, 389–405. Bryant, R.G., Bigg, G.R., Mahowald, N.M., Eckardt, F.D., Ross, S.G., 2007. Dust emission response to climate in southern Africa. J. Geophys. Res. 112, D09207.

Ellery, W.N., McCarthy, T.S., Cairncross, B., Oelofse, R., 1989. A peat fire in the Okavango Delta, Botswana, and its importance as an ecosystem process. Afr. J. Ecol. 27, 7–21. Ellery, W.N., Ellery, K., McCarthy, T.S., 1993. Plant distribution in islands of the Okavango Delta, Botswana: determinants and feedback interactions. Afr. J. Ecol. 31, 118–134. Ellery, W.N., McCarthy, T.S., Dangerfield, J.M., 1998. Biotic factors in Mima mound development: evidence from the floodplains of the Okavango Delta, Botswana. Int. J. Ecol. Environ. Sci. 24, 293–313. Field, J.P., Breshears, D.D., Whicker, J.J., Zou, C.B., 2012. Sediment capture by vegetation patches: implications for desertification and increased resource redistribution. J. Geophys. Res. 117, G01033. Folk, R.L., Ward, W.C., 1957. Brazos River bar: a study in the significance of grain size parameters. J. Sediment. Petrol. 27, 3–26. Garstang, M., Ellery, W.N., McCarthy, T.S., Scholes, M.C., Swap, R.J., Tyson, P.D., 1998. The contribution of aerosol- and water-borne nutrients to the functioning of the Okavango Delta ecosystem, Botswana. S. Afr. J. Sci. 94, 223–229. Gumbricht, T., McCarthy, T.S., Merry, C.L., 2001. The topography of the Okavango Delta, Botswana, and its tectonic and sedimentological implications. S. Afr. J. Geol. 104, 243–264. Gumbricht, T., McCarthy, J., McCarthy, T.S., 2004. Channels, wetlands and islands in the Okavango Delta, Botswana, and their relation to hydrological and sedimentological processes. Earth Surf. Process. Landf. 29, 15–29. Krah, M., McCarthy, T.S., Annegarn, H., Ramberg, L., 2004. Airborne dust deposition in the Okavango Delta, Botswana, and its impact on landforms. Earth Surf. Process. Landf. 29, 565–577. Lancaster, N., 1986. Grain-size characteristics of linear dune in the southwestern Kalahari. J. Sediment. Petrol. 56, 395–400. Lancaster, N., 2000. Eolian deposits. In: Partridge, T., Maud, R. (Eds.), The Cenozoic of Southern Africa. Oxford University Press, Oxford, UK. Livingstone, I., Bullard, J.E., Wiggs, G.F.S., Thomas, D.S.G., 1999. Grain-size variation on dunes in the southwest Kalahari, southern Africa. J. Sediment. Res. 69, 546–552. McCarthy, T.S., 2006. Groundwater in the wetlands of the Okavango Delta and its contribution to the structure and function of the ecosystem. J. Hydrol. 320, 264–282. McCarthy, T.S., Metcalfe, J., 1990. Chemical sedimentation in the Okavango Delta, Botswana. Chem. Geol. 89, 157–178. McCarthy, T.S., McIver, J.R., Cairncross, B., Ellery, W.N., Ellery, K., 1989. The inorganic chemistry of peat from the Maunachira channel-swamp system, Okavango Delta, Botswana. Geochim. Cosmochim. Acta 53, 1077–1089. McCarthy, T.S., Stanistreet, I.G., Cairncross, B., 1991a. The sedimentary dynamics of active fluvial channels on the Okavango fan, Botswana. Sedimentology 38, 471–487. McCarthy, T.S., McIver, J.R., Verhagen, B.Th., 1991b. Groundwater evolution, chemical sedimentation and carbonate brine formation on an island in the Okavango Delta swamp, Botswana. Appl. Geochem. 6, 577–595. McCarthy, T.S., Ellery, W.N., Stanistreet, I.G., 1992. Avulsion mechanisms on the Okavango fan, Botswana: the control of a fluvial system by vegetation. Sedimentology 39, 779–795. McCarthy, T.S., Ellery, W.N., Ellery, K., 1993. Vegetation-induced, subsurface precipitation of carbonate as an aggradational process in the permanent swamps of the Okavango (Delta) fan, Botswana. Chem. Geol. 107, 111–131. McCarthy, T.S., Ellery, W.N., Dangerfield, J.M., 1998. The role of biota in the initiation and growth of islands on the floodplain of the Okavango alluvial fan, Botswana. Earth Surf. Process. Landf. 23, 291–316. McCarthy, T.S., Humphries, M.S., Mahomed, I., Le Roux, P., Verhagen, B.Th., 2012. Island forming processes in the Okavango Delta, Botswana. Geomorphology 179, 249–257. Middleton, N.J., Goudie, A.S., 2001. Saharan dust: sources and trajectories. Trans. Inst. Br. Geogr. 26, 165–181. Ponce, V.M., da Cunha, C.N., 1993. Vegetated earthmounds in tropical savannas of central Brazil: a synthesis. J. Biogeogr. 20, 219–225. Prospero, J.M., 1999. Long-range transport of mineral dust in the global atmosphere: impact of African dust on the environment of the southeastern United States. Proc. Natl. Acad. Sci. 96, 3396–3403. Prospero, J.M., Ginoux, P., Torres, O., Nicholson, S.E., Gill, T.E., 2002. Environmental characterization of global sources of atmospheric soil dust identified with the NIMUS 7 total ozone mapping spectrometer (TOMS) absorbing aerosol product. Rev. Geophys. 40, 1002. Ramberg, L., Wolski, P., 2008. Growing islands and sinking solutes: processes maintaining the endorheic Okavango Delta as a freshwater system. Plant Ecol. 196, 215–231. Simonson, R.W., 1995. Airborne dust and its significance to soils. Geoderma 65, 1–43. Sklar, F.H., Van der Valk, A.G., 2002. Tree Islands of the Everglades. Kluwer Academic, Dordrecht, The Netherlands. Sullivan, P.L., Price, R.M., Ross, M.S., Scinto, L.J., Stoffella, S.L., Cline, E., Dreschel, T.W., Sklar, F.H., 2010. Hydrologic processes on tree islands in the Everglades (Florida, USA): tracking the effects of tree establishment and growth. Hydrogeol. J. 19, 367–378. Swap, R., Garstang, M., Greco, S., 1992. Saharan dust in the Amazon basin. Tellus B44, 133–149. Tooth, S., McCarthy, T.S., Hancox, P.J., Brandt, D., Buckley, K., Nortje, K., McQuade, S., 2002. The geomorphology of the Nyl River and floodplain in the semi-arid Northern Province, South Africa. S. Afr. Geogr. J. 84, 226–237. Troxler Gann, T.G., Childers, D.L., Rondeau, D.N., 2005. Ecosystem structure, nutrient dynamics, and hydrological relationships in tree islands of the southern Everglades, Florida, USA. For. Ecol. Manag. 214, 11–27. Tyson, P.D., Garstang, M., Swap, R., Kallberg, P., Edwards, M., 1996. An air transport climatology for subtropical southern Africa. Int. J. Climatol. 16, 265–291.

M.S. Humphries et al. / Geomorphology 206 (2014) 307–317 Washington, R., Todd, M., Middleton, N.J., Goudie, A.S., 2003. Dust-storm source areas determined by the total ozone monitoring spectrometer and surface observations. Ann. Assoc. Am. Geogr. 93, 297–313. Wetzel, P.R., Sklar, F.H., Coronado, C.A., Troxler, T.G., Krupa, S.L., Sullivan, P.L., Ewe, S., Price, R.M., Newman, S., Orem, W.H., 2011. Biogeochemical processes on tree islands in the Greater Everglades: initiating a new paradigm. Crit. Rev. Environ. Sci. Technol. 41, 670–701.

317

Wilson, B.H., Dincer, T., 1976. An introduction to the hydrology and hydrography of the Okavango Delta. Proceedings of the Symposium on the Okavango Delta and Its Future Utilisation. Botswana Society, Gaborone, pp. 33–48. Wolski, P., Savenije, H.H.G., Murray-Hudson, M., Gumbricht, T., 2006. Modelling of the flooding in the Okavango Delta, Botswana, using a hybrid reservoir-GIS model. J. Hydrol. 331, 58–72.