Hydrology and erodibility of the soils and saprolite cover of the Swaziland Middleveld

Page 1

SOIL TECHNOLOGY

ELSEYIER Soil Technology ll(1997) 247-262

Hydrology and erodibility of the soils and saprolite cover of the Swaziland Middleveld

T. Scholten * Znstitutjir Bodenkunde and Bodenerhaltung, Justus Liebig Universitiit Giessen, Wiesenstrasse 3 - 5,

35390 Giessen, Germany

Abstract

The weathering mantle of the Middleveld of Swaziland consists of thick soil-saprolite complexes. The isovolumetric chemical weathering of the saprolites has led to mass losses of more than 50%. Compared with saprolites from quartz-diorite and granodiorite, those from diorite have higher portions of easily weatherable plagioclases and amphiboles and 20-30% greater total pore space. The macro pore space reaches a maximum (4.6-7.0%) in the central saprolite zone, corresponding to saturated hydraulic conductivities of 6.02-11.81 X low7 m/s. Similar to the saprolites, the overlying ferrallitic soils show total pore volumes ranging from 39 to 52%. Compared to the soils, the available water capacity of the underlying saprolites is two to four times higher and the saturated hydraulic conductivity is about two times higher due to the high portion of medium pores which amount to 70% of total pore space. In the areas affected by sheet erosion, most of the soil cover is denuded and the underlying saprolites essentially determine the site properties. The hydrological properties of the saprolites are therefore of great importance with respect to erosion during wet periods and plant growth during drought periods. The low structural stability of the saprolites, indicated by shear strength values < 5 kPa, results from a silty texture, absence of organic matter, and low contents of Fe- and Al-oxides. As a result, saprolites are highly susceptible to erosion and represent an essential precondition for the development and rapid expansion of deep incising erosion gullies in areas with magmatic rocks. In contrast, the clay-rich ferrallitic soils developed from saprolite are comparatively stable, indicated by shear strength values ranging from 7 to 12 kPa. The inherent stabilizing properties of the soil are altered by overgrazing and unwise land use leading to infiltration capacities below 65 cm/day and high overland flow potentials at low rainfall intensities. 0 1997 Elsevier Science B.V.

Keywords: Erodibility; Hydrology; Saprolite; Soil-saprolite complex; Structural stability; Swaziland

~- * Corresponding author

00933-3630/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved. PZZ SO933-3630(97)00011-l

Page 2

248 T. Scholteen/Soil Techology II (1997) 247-262

1. Introduction

The estimation of hydrological properties and erodibility is of great importance in understanding erosional processes and a key factor in many erosion risk assessments and soil management systems. Studies of hydrological properties and erodibility of soils have, for long, been objectives of pedological research and are available for soils from most parts of the world. In contrast, related investigations of saprolites (for definition, see Scholten et al., 1997) and, particularly, of the whole regolith, referred to as a soil-saprolite complex, are few.

The weathering mantle of the Middleveld of Swaziland consists of thick soil-sapro- Iite complexes, which have developed from magmatic and metamorphic rocks during weathering-intensive climatic periods. Compared to the solid rock, the saprolite is of low physical stability and represents the initial substratum for younger soils after erosion phases. Recently, soil erosion in Swaziland has been widespread and, together with the drought risk, represents the main problem facing agriculture (Mushala et al., 1993). The expanding areas of sheet erosion and deep erosion gullies have developed on deeply weathered magmatic rocks in the grass savanna of the densely populated Middleveld (WMS Associates, 1990; Scholten and Felix-Henningsen, 1993). Gully erosion is initiated mainly on cattle tracks on the lower slopes. As soon as the soil is breached, the gullies incise quickly into the underlying saprolite and migrate rapidly headwards. As the gully extends upslope, the rate of extension decelerates with decreasing catchment size. Within a few years, inactive gullies or inactive parts of a gully are characterized by gently sloping side walls stabilized by grass and shrub vegetation. Observations in Swaziland have shown that reclamation of saprolites by vegetation after erosion or drought periods is much faster than on the ferrallitic soils of the region.

Considering the characteristics of soil erosion as described above and a restricted water supply for vegetation and human use, the hydrological properties and erodibility of soil-saprolite complexes are of great importance as a basis for land evaluation and erosion risk assessment.

The objectives of this study were to characterize the hydrological properties based on pore volume, pore size distribution, available water capacity, saturated hydraulic conduc- tivity and infiltrability, and the erodibility determined by in situ shear strength measure- ments and calculation of a stability factor for typical soil-saprolite complexes from granodiorite, quartz-diorite, and diorite in the Middleveld of Swaziland. Moreover, the results have generic value for conditions in similar soil-saprolite regions.

2. Materials and methods

A detailed description of the geology, saprolites and soils of the Swaziland Mid- dleveld is given in Scholten et al. (1997). Three study sites (Kukanyeni, Mawelawela and Ntondozi area) in the center of the Middleveld were chosen to investigate the hydrology and erodibility of the soils and saprolite cover. The study sites represent typical landscape sections with respect to parent materials, relief energy and land uses (see Figs. 1 and 2 in Scholten et al., 1997).

Page 3

T. Scholten / Soil Technology 11 (1997.1247-262 249

At the study sites, 10 representative soil-saprolite profiles from granodiorite (study site l), diorite (study site 2) and quartz-diorite (study site 3) were described and sampled. The soil classification and designation of horizons are according to FAO (1989). The saprolites were divided into zones as described in Scholten et al. (1997). The rock classification was taken from Wilson (1982).

The infiltration capacity was determined in the field for different surface types by measuring the one-dimensional water flow into the soil per unit time with double-ring infiltrometers (300 mm inner ring diameter, from four to six replications) according to DIN 19682 (Hartge and Horn, 1989). The measurements were carried out at the end of the dry season in order to exclude the influence of different initial moisture contents as described by Turner and Sumner (1978). Soil moisture contents varied between 0.1 and 0.2 g/g oven-dry soil for all surface horizons. Shear strength was determined in the field for each horizon by measuring the soil cohesion at field capacity (5-10 replica- tions) with a torvane (Eijkelkamp Agrisearch Equipment).

The determination of particle size distribution was carried out on H,O,-treated samples dispersed with sodium-pyrophosphate. Sand fractions were determined by sieving (63-2000 pm), and clay and silt fractions ( < 63 pm) by pipette analysis (Schlichting et al., 1995). In order to take the different stabilities of saprolite zones into consideration, all samples were crushed gently by hand and sieved under constant water flow and water pressure. This ensured a standardized mechanical pretreatment of the samples. Therefore, the particle size distribution of the saprolite reflects the dispersible fractions and not the primary particle size distribution. The saturated hydraulic conduc- tivity (K,) was measured on horizontally sampled undisturbed 100 cm3 volumetric cores (six replications) using a laboratory permeameter (Hartge and Horn, 1989). The bulk density was calculated from oven-dry (lOS°C) undisturbed 100 cm3 volumetric core samples (Hartge and Horn, 1989). The calculation of pore size distribution and available water capacity was based on the determination of water contents at different water retentions (6 kPa, 30 kPa, 1500 kPa; Hartge and Horn, 1989) using pressure plate extractors (Soil Moisture). According to Hartge and Horn (1989) the volumetric water content at 6 kPa is related to an equivalent pore diameter > 50 pm, 30 kPa to > 10 pm (referred to as coarse pores), and 1500 kPa to < 0.2 pm (referred to as fine pores), respectively. Organic carbon was determined on fine ground oven-dry (105°C) samples and analyzed by dry combustion (Carlo Erba ANA 1400). Dithionite-citrate-bi- carbonate and oxalate extractable iron, manganese, and aluminium were extracted according to Mehra and Jackson and Schwertmann (Schlichting et al., 1995) and analyzed by AAS. Exchangeable Al was measured by percolation with ammonium acetate at pH 7 (Schlichting et al., 1995). The Al was determined in the percolate using AAS equipment.

Wischmeier and Smith (1978) related the erodibility of a soil to the integrated effects of rainfall, surface runoff and infiltration. On the basis of experimental investigations, they developed a quantifiable soil erodibility factor K. The transferability of the USLE (Universal Soil Loss Equation; Wischmeier and Smith, 1978) to tropical environments is questionable (e.g., El-Swaify and Dangler, 1982) because the climatic, topographic, and soil factors differ strongly from those that were originally used to establish the USLE. Schieber (1989) also found little agreement between calculated and measured K-factors

Page 4

250 T. Scholten/Soil Techno1og.v I1 (1997) 247-262

for the subtropical soils of South Africa. As regards erodibility, the objective of this study was to evaluate relative differences of erodibility for soils and saprolite. Therefore, a stability factor was calculated using the relation of texture and organic carbon to erodibility as given in the modified equation for the K-factor in Schwertmann et al. (1987) for SI units. The term ‘stability factor’ was chosen to avoid confusion concerning the definition and validity of the K-factor.

3. Results

3.1. Porosity and pore size distribution

All sa$olites showed a decreasing pore volume with depth (Table 1). Owing to high silt and only low dispersible clay contents, medium pores (O-O.2 pm) predominate (Table 1). The small portion of fine pores ( < 0.2 pm) in the saprolites corresponds with the dispersible clay contents and provides independent confirmation of the particle size analysis.

With the transition from the saprolite upwards into the soil, the expected increase in total pore volume arising from the greatly increasing clay content (Table 1) is offset by the alignment of clay particles in the soil due to swelling and shrinking during structure formation. As a result, only slight differences were found between the soil horizons and the underlying saprolites (Table 1). Fine pores dominated in the soils, with 42-81% of the total pore volume (Table 1) and distinctly smaller amounts of narrow coarse pores (50-10 pm) and medium pores.

3.2. Available water capacity

For the saprolites, the available water capacity in the form of plant-available water in pores with equivalent diameters of 50 to 0.2 pm (pF 1.8-4.2) was two to four times higher than for the overlying soils (Table 1). Medium pores contribute 65-72% to the total pore volume independent of the amount of total pore volume. Thus, the available water capacity of the saprolites corresponds closely with the total pore volume.

The available water capacity increased with weathering intensity from bottom to top in all saprolites (Table 1). Flint and Childs (1984, cited in Jones and Graham, 1993) also measured an increase in the available water capacity with weathering intensity for saprolites from volcanic and metasedimentary rocks in Oregon, USA, with values of between 0.06 and 0.18 m3/m3. In the Middleveld of Swaziland, the highest values were reached in the diorite saprolite along with higher clay and silt contents (Table 1). The available water capacity increased only very slightly from the middle to the upper saprolite zone in the diorite saprolite as well as in the granodiorite saprolite (Table 1).

The soils showed a distinct lower available water capacity than the saprolites (Table 1). The lower amounts of medium pores in the Ah-horizons of study site 3 (Table 1) caused a decrease of the available water capacity in relation to the underlying B-hori- zons. In comparison, the available water capacity of the Ah-horizons of study sites 1 and 2 was twice as high (Table 1). A corresponding increase in pore volume and available

Page 5

Tabl

e 1

The

textur

e, po

re

size

dist

ribut

ion

and

tota

l po

re

volum

e (T

PV),

avail

able

water

ca

pacit

y (A

WC)

an

d sa

tura

ted

hydr

aulic

co

nduc

tivity

(K

,) of

diffe

rent

sa

proli

te

zone

s an

d ov

erlyi

ng

soil

horiz

ons

relat

ed

to

pare

nt

mat

erial

Horiz

on/zo

ne

Textu

re

(wt.%

> (

Mm)

Pore

siz

e di

strib

utio

n (%

) (p

m)

TPV

AWC

KS

Nam

e Sy

mbo

la <6

3 63

-200

0 >2

000

> 50

50

-10

10-0

.2

< 0.

2 (%

vo

l) (m

3/m3>

(lO

-7

m/s)

Dior

ite

(site

2)

N

Tops

oil

Ah

24

19

57

6 9

28

57

52

0.19

3.

23

:: $ Su

bsoi

l Bw

13

43

44

6

4 9

81

52

0.07

0.

39

2 Up

per

sapr

olite

Cw

l 19

65

16

5

8 59

28

54

0.

36

4.17

2 \

Midd

le sa

proli

te

cw2

30

63

7 9

10

59

23

51

0.35

6.

02

M c.

Lowe

r sa

proli

te

cw3

37

51

6 4

8 64

24

43

0.

31

4.28

3

Quar

tz-dio

rife

(site

3)

ii-

To

psoi

l Ah

31

15

48

13

8

15

64

39

0.09

5.

67

0 r?.

Subs

oil

BW

25

17

58

9 I

20

64

40

0.11

0.

35

B Up

per

sapr

olite

Cw

l 40

53

7

15

13

56

16

41

0.28

7.

18

2 M

iddle

sapr

olite

cw

2 45

50

5

18

16

49

17

39

0.25

11

.23

2

Lowe

r sa

proli

te

CW3

5.5

41

3 9

11

59

21

36

0.25

5.

09

:

Gran

odior

ite

(site

1)

s

Tops

oil

Ah

31

18

51

8 10

34

48

42

0.

19

3.01

Su

bsoi

l Bw

s 16

16

68

k%

4

6 48

42

45

0.

24

0.58

Up

per

sapr

olite

Cw

l 36

53

11

10

11

59

20

44

0.

31

8.56

M

iddle

sapr

olite

cw

2 43

49

8

13

14

54

19

42

0.29

11

.81

Lowe

r sa

proli

te

cw3

52

42

6 8

8 61

23

38

0.

26

4.98

a FA

O,

1989

Page 6

252 T. Scholten/Soil Technology II (1997) 247-262

water capacity was reported by Morachan et al. (1972, cited in Bauer and Black (1992) for a silty clay, and was described by Hillel (1980) as the result of a stabilization of soil structure by organic matter.

3.3. Saturated hydraulic conductivity

Within all soil-saprolite complexes, the K, reached a maximum in the middle saprolite zone along with the highest amount of macro pores (Table 1). For all investigated samples, K, is related to the amount of pores > 50 pm with r = 0.84 (P = 0.01, n = 21). The bulk percentages by volume portion of coarse pores were 5.5% (granodiorite saprolite), 7.0% (quartz-diorite saprolite) and 4.6% (diorite saprolite) for the middle saprolite zone. These were higher than those for the lower saprolite zone with 3.0% (granodiorite saprolite), 3.2% (quartz-diorite saprolite) and 1.7% (diorite saprolite). Therefore, the decrease of K, in the lower saprolite zone (Table 1) was determined by a smaller portion of coarse pores due to the lower weathering intensity.

The soils showed the lowest K,-values at the highest clay contents within all soil-saprolite complexes (Table 1). In comparison with the B-horizons, the formation of secondary macropores due to the higher biological activity and rooting as well as the higher portion of sand in the Ah-horizons led to higher saturated hydraulic conductivi- ties (Table 1). The comparison of mean saturated hydraulic conductivities of the soil cover of all study sites (Table 1) indicates no distinct differences between the study sites. Single measured K,-values varied between 0.09 and 13.58 X lo-’ m/s corre- sponding to calculated values given in WMS Associates (1990) for different sites in Swaziland, ranging between 0.07 and 27.32 X lo-’ m/s.

3.4. Infiltration capacity

The objective of the field measurements was to determine the infiltration capacity according to different kinds of land use and different degrees of soil erosion. For this purpose, the soil surface categories ‘cultivated land’, ‘rangeland not overgrazed’, ‘rangeland overgrazed’ and ‘sheet erosion’ were chosen.

Final infiltration rates, measured as infiltration rates after reaching a quasi-constant value (for the given soils normally after 3-5 h of infiltration) were the lowest for sheet erosion surfaces with 24.5 cm/day and the highest for cultivated land with 212.5 cm/day (Fig. 1). Except for the ‘cultivated land’, the results confirm that an increasing organic carbon content is related to a higher infiltration capacity. The values for the ‘rangeland not overgrazed’ and ‘rangeland overgrazed’ are similar to those obtained in rainfall simulation experiments for rangeland with 90 and O-15% cover respectively (Morgan et al., 1997).

The rangeland areas with incomplete vegetation cover due to overgrazing showed only slightly higher infiltration rates and organic carbon contents than areas affected by sheet erosion (Fig. 1). Clogging during sheet wash and the compaction of the Ah-hori- zon by cattle have led to higher bulk densities (e.g., of 1.63 mg/m3 in the Ah-horizon as compared with 1.47 mg/m3 in the B-horizon at rnidslope position at site 3). Owing to the protective effect of the vegetation cover, clogging of pores and the formation of

Page 7

T. Scholten /Soil Technology 11 (1997) 247-262 253

250 Final infiltration rate [cm/day]

200

50

0

erosion overgrazed not overgrazed land

Fig. 1. The final infiltration rates and organic carbon content of selected soil surfaces (n is the number of samples).

surface crust were much reduced on rangeland that was not overgrazed. This confirms field observations at site 1 where infiltration capacities were higher.

The highest infiltration capacities were measured on cultivated land (Fig. 1). This is caused by ploughing, loosening the soil structure and forming large water-conducting pores (cf. Wendroth and Ehlers, 1989). Preferential water flow through these pores supersedes all other factors and, therefore, controls the infiltration capacity on cultivated land.

3.5. Estimation of overland flow

Smithen and Schulze (1982) gave the first detailed calculation of the spatial distribu- tion of rainfall erosivity in southern Africa. Because of the size of the region considered, the use of data from only one meteorological station for Swaziland and the high variability of rainfall intensities in Swaziland (Murdoch, 1968), accurate estimation of rainfall intensities for the Central Swaziland Middleveld is difficult. Kiggundu (1986) determined rainfall erosivities in Swaziland using daily means of rainfall intensities for four meteorological stations and daily rainfall records of 38 meteorological stations, and prepared erosivity contour maps for the whole country using the equation given in Wischmeier and Smith (1978).

According to Kiggundu (1986) the mean annual rainfall erosivity (EI,,) for the central Swaziland Middleveld is 450 kJ . mm/m2 . h with a mean rainfall energy of 18 kJ/m*. The annual rainfall erosivity for a 25year recurrence interval is 950 kJ * mm/m2 . h with a mean rainfall energy of 26 kJ/m*. The long-term mean maximum rainfall intensity for a 30-min interval (ZsO) is 25 mm/h and for a recurrence interval of 25 years, it is 36.5 mm/h. The high variability of the annual rainfall in Swaziland causes a relatively high range of values. Mushala and Kiggundu (1990) found for the rainy season 1989/1990 at three stations in the Highveld and one station in the Middleveld, a deviation from the long-term annual mean of EZ3,, between - 5.5 and +34.6%. In contrast to the rainfall intensities given by Hudson (1981) for Zimbabwe

Page 8

254 T. Scholten / Soil Technology I1 (I 997) 247-262

where up to 300 mm/h has been recorded, Kiggundu (1986) found that maximum rainfall intensities of 76 mm/h are exceeded only rarely in Swaziland (Mushala and Kiggundu, 1990).

The estimation of potential overland flow is based on these results. To evaluate the potential occurrence of surface runoff the percentage rate of runoff as a portion of total rainfall was calculated for I,, = 25 mm/h, 36.5 mm/h and 76 mm/h using the mean infiltration rate for different soil surfaces, based on a l-h rainfall event (Table 2). The cumulative infiltration after 60 min was used as a characteristic value for dry soil conditions at the beginning of the rainy season. The final infiltration rate was used to describe wet soil conditions. Losses due to interception and evapotranspiration as well as changes at the soil surface and of the pore space during a rainfall event as described by Boiffin and Monnier (1985), Boiffin (1986) and Roth (19921, were not taken into account.

The results show that overland flow on cultivated land is unlikely even at high rainfall intensities (Table 2). Only at Z30 > 88.5 mm/h for wet soil conditions and Zs,, > 124.5 mm/h for dry soil conditions are the infiltration capacities exceeded. The runoff potential increases in the order of ‘cultivated land’ < ‘rangeland not overgrazed’ < ‘rangeland overgrazed’ < ‘sheet erosion’. For the latter, rainfall intensities Zs,, > 13.9 mm/h for dry soil conditions and ZaO > 10.1 mm/h for wet soil conditions will generate runoff. At Zs,, = 25 mm/h the portion of surface runoff accounts for 44-60% of total rainfall (Table 2). The calculated runoff for the rangeland is somewhat higher than that obtained from rainfall simulation studies using a rainstorm of 75 mm/h for 15 min. The measured runoff ranged from 9 to 41% of the simulated rain (Morgan et al., 1997).

3.6. Stability of soil structure

The shear strength as a summarizing parameter of substratum cohesion was used to measure the stability of soil structure. With values of 2.3-4.4 kPa (Fig. 2) the shear strength of the saprolites was distinctly lower than that of the soil horizons and falls within the range of shear strength data that are typical of sandy to silty unconsolidated soils (cf. Morgan et al., 1991). The clay contents of all saprolites examined only amount to 4-16% depending on depth (Table 1). Together with the absence of organic matter and distinctly lower values for cementing pedogenic oxides (Fig. 21, this results in a

Table 2 The percentage of potential runoff on total rainfall based on a l-h storm related to different rainfall intensities and soil surfaces under dry (cumulative infiltration) and wet (final infiltration rate) surface conditions (I,, long-term mean maximum rainfall intensity within 30 min)

Soil surface & = 2.5 mm/h I,, = 36.5 mm/h Iso = 76 mm/h

Dry (%) Wet (I) Dry (o/o) Wet (%) Dry (%) Wet (%)

Sheet erosion 44 60 62 72 81 87 Rangeland overgrazed 0 0 1 27 52 64 Rangeland not overgrazed 0 0 0 0 23 48 Cultivated land 0 0 0 0 0 0

Page 9

T. SchoZten/Soil Technology ll(1997) 247-262 255

Study site 1 Study site 2 Study site 3 (Granodiorite) (Diorite) (Quartz-diorite)

Clay 1% WI] my 1% tij Clay 1% wt] 0 10 20 30 40 50 60 70

0 0 IO 20 30 40 50 60 70 0 10 20 30 40 50 60 70

0 0

200 200 200

400 5

400 400

2 0"

600 6W 600

600 +: Shear strength 800 800

* Organic m-bon

1000 1000 1000 0 0

Shea: slrenglh’gPa] 15 5 10 15 0 5 10 15

Shear strength [kPaJ Shear strength [kPa] F%CB [Wgl Fern, hWe1

.%cP.~ 10 [twig1 %c~IWgl

&CB~ 10 Imats Organic carbon x 0.5 [% WI, Organ1c carbon x 0 5 [% VA]

Ahx 10 lmpial Organic carbon x 0.5 [% wt]

Fig. 2. The shear strength and pedogenic oxides, clay and organic carbon content with depth at the different study sites (horizon designation according to FAO, 1989).

structure which is very unstable vis-A-vis erosion. The comparison of saprolites from different parent rocks showed only slight differences concerning the contents of clay and pedogenic oxides (Fig. 2). Therefore, a correlation of parent rock and structural stability

Table 3 The relationship between shear strength and different properties of soils and saprolites (r: correlation coefficient, S.E.: standard error, P: probability, n: number of samples)

Variable r SE. of r P(r=O) n

SOil Organic carbon 0.00 0.18 ns 32 Clay 0.29 0.17 ns 32

Fern6 -0.16 0.18 ns 32

ho - 0.27 0.18 ns 32

A1~o 0.07 0.18 ns 32 Al 0.66 0.24 b

CYCb’ 32

Saprolite Clay 0.74 0.17 d 18

Fewa 0.75 0.17 d 18

F~AO 0.64 0.19 c 18

40 0.74 0.17 d 18

ans: not significant. bP = 0.05. cP = 0.01. dP = 0.001.

Page 10

256 T. Scholten/Soil Technology II (1997) 247-262

Table 4 The calculated mean stability factors (SF) based on texture and organic carbon content as an index for erodibility of different soil and saprolite horizons (C.V.: coefficient of variation, n: number of samples)

Material SF C.V. (%) n

Topsoil (A-horizon) 0.13 33 15 Subsoil (B-horizon) 0.19 47 29 Transitional zone 0.57 13 10 Saprolite (all) 0.52 27 30 Diorite saprolite 0.66 14 9 Quartz-diorite saprolite 0.45 31 11 Granodiorite saprolite 0.48 21 10

of saprolites is not given. The slight increase of shear strength of the saprolite with depth (Fig. 2) could reflect the decreasing mass losses according to weathering intensity, weathering age and mineral composition (Scholten et al., 1997).

Compared to the saprolite, the clayey kaolinitic soil horizons (Scholten et al., 1997) have a greater specific surface and higher surface charge density. Together with organic matter and contents of Fe- and Al-oxides, which were up to 30 times higher, this led to a relatively higher stability of the soil structure (Fig. 2).

Although a close dependence of soil cohesion and contents of organic carbon, clay, pedogenic oxides and exchangeable Al can be traced for each specific soil-saprolite complex (Fig. 21, a general relation between these soil properties and measured shear strength of the soil cover could not be found (Table 3). In contrast, the cohesion of all saprolites is well defined by contents of clay and pedogenic oxides (Table 3).

3.7. Erodibility

As well as shear strength, the calculated stability factors show clear differences between the soil horizons and the saprolite (Table 4). The stability factor for the transitional zone (BwCw-horizon) and the saprolites is about three times that of the soil horizons. The Ah-horizon displays the lowest erodibility (Table 41, probably due to a higher organic matter content (Fig. 2). The texture is a critical determinant of the erodibility of the B-horizons and the saprolite, although a slight influence of organic matter on the stability factor can be traced down to the Cwl-horizons (Fig. 2). The various parent materials cause differences in erodibility between the study sites (Table 4). The individual saprolites were themselves very homogeneous due to the uniform particle size distribution of different saprolite horizons at each specific site (Table 1). Similar results for particle size distribution were reported by Stolt et al. (1992) from investigations in Virginia, USA.

4. Discussion

4.1. Hydrology of the soils and saprolite cover

Fine-grained rocks like the diorite at site 2 are more susceptible to chemical weathering than coarse-grained rocks because of the larger total area of particle surfaces

Page 11

T. Scholten / Soil Technology I1 (1997) 247-262 251

available. Mineral boundaries beside twin-planes (particularly for plagioclase) are the preferred targets of chemical weathering (Rodgers and Holland, 1979). According to investigations by Nixon (1979) orthoclases are clearly more stable as regards weather- ing than plagioclases. Also, Pye (1986) has shown for granitic rocks in southern Africa that weathering resistance increases with increasing amounts of orthoclases. Accord- ingly, the higher share of orthoclases in the granodiorite and quartz-diorite at study sites 1 and 3 (Scholten et al., 1997) leads to a sandier texture of the corresponding saprolites. These differences induced by parent material are weakened due to an increasing weathering rate of saprolites (cf. Wysocki et al., 1988). This is reflected in a slightly decreasing portion of the sand fraction in the sequence diorite saprolite-quartz-diorite saprolite-granodiorite-saprolite (Table 1). Accordingly, the diorite saprolite displayed about a 20-30% higher total pore volume, while quartz-diorite saprolite and granodiorite saprolite differed only little. Comparably high pore volumes were shown by Brimhall et al. (1991) for a saprolite from diabase in Mali (West Africa).

A relation between the pore size distribution of the soils and the underlying saprolites is not given. The differences in texture and size and amount of secondary pores between the soils due to structure and aggregate formation, biological activity and rooting patterns led to a relatively higher variability in the pore size distribution independent of parent material (Table 1).

Fine roots have diameters of > 10 pm (Baeumer, 1992) and so can only penetrate into coarse pores. Correspondingly, the rootable pore space of the saprolites is about 62-74% larger than that of the overlying soils. In addition to pore volume, the conditions of root growth and consequently water and nutrient uptake are determined by the spatial distribution of pores. The saprolites showed no secondary pores. Only the upper zone contained few relict root channels, witnessing a former deep-rooting forest vegetation. As a result, the primary coarse pores have an even and fine-meshed spatial distribution. In contrast, the soils showed subangular to angular blocky structured A-horizons and were partly columnar in the B-horizons. The mean aggregate sizes varied between 10 and 50 mm and the spatial distribution of pores was very coarse- meshed.

The soil cover at site 3 is obviously degraded by overgrazing and sheet erosion on most parts of the slope. Apart from the loss of organic matter through grazing, it can be assumed that the A-horizon is denuded by erosion. Both factors led to reduced organic carbon contents in the top soil (cf. Franzle, 1984). Although Bauer and Black (1992) could not find an increase in available water capacity with rising organic matter content for fine textured soils, it is probable that the distinctly higher organic carbon content of 2.7-2.9% at study sites 1 and 2 caused a formation of medium pores, which is not effective at site 3 with organic carbon contents varying between 1.0 and 1.4%. Compared to the soils, the available water capacity of the underlying saprolites is two to four times higher due to the higher portion of medium pores and this, therefore, promotes the water supply of the plants in dry periods.

In all saprolites, K, was lower in the upper saprolite zone along with decreasing amounts of coarse pores (Table 1). This is due to alteration of pore space in this zone through embedding of clay and humates which was shown by micromorphological investigation (Felix-Henningsen et al., 1995). On the contrary, both in the lower as well

Page 12

258 T. Scholten/Soil Technology I1 (1997) 247-262

as in the middle saprolite zone, K, depends only on water flux in primary pores since a pedogenetic structure formation is missing. Differences between granodiorite saprolite and quartz-diorite saprolite due to parent material are minor (Table 1). On the other hand, K, was distinctly lower for diorite saprolite (Table 1) due to the mainly higher share of easily weatherable plagioclases corresponding with a silty texture and, hence, a smaller portion of coarse pores.

At all sites, the B-horizons are characterized by illuvial clay accumulation, developed during an older soil formation phase (relic argillic B-horizons, Felix-Henningsen et al., 1995). This feature has also been described by Murdoch (19681, Renk (1977a,b), and Dardis (1990) for comparable sites in Swaziland and South Africa. The illuviation of clay reduced the size of pores, leading to a higher portion of fine pores and lower saturated hydraulic conductivities (Table 1). Hence, the B-horizons form a water-retain- ing layer which limits vertical water movement. This promotes the generation of surface runoff because of reduced infiltration and it lowers the water flux into the saprolite. The latter can distinctly impede further weathering of the saprolite (Gardner, 1992).

An important factor influencing the form and stability of soil structure and corre- spondingly the infiltration capacity of a soil is organic matter (Fiedler, 1958; Katschin- ski, 1958; Bachmann, 1989; Zhang and Hartge, 1990; Bohne, 1991). The topsoils of sheet erosion areas had the lowest organic carbon contents (Fig. 1). The discharge of humic topsoil horizons by erosion and the corresponding nutrient loss degrade these areas and impede the establishment of new vegetation and soil fauna, essential for formation of stable aggregates and large animal burrows which increase the infiltration capacity.

Apart from organic matter, soil crusting has been identified as one of the most important factors influencing infiltration capacity (Morin et al., 1981; Le Bissonnais, 1990; Roth, 1992). Soil crusts have higher bulk densities, finer pores and lower hydraulic conductivities than the underlying soil (Smith et al., 1990). As a consequence, soil crusts that have formed on all sheet eroded areas and most vegetation-free spots of rangeland areas reduce infiltration capacity.

4.2. Structural stability and erodibility of the soils and saprolite cover

The structural stability of the soils is influenced by several different soil properties. Clay acts as a binding agent, improving the aggregation of soil colloids (Skidmore and Layton, 1992). Organic matter causes a distinct increase in aggregate stability (Bartoli et al., 1992; Roth, 1992) and thus in the stability of soil structure. Also Fe- and Al-oxides have a stabilizing effect on the soil structure (Giovannini and Sequi, 1976). Moreover, the increasing release of Al 3f-ions below pH 5 leads to higher structural stabilities due to a connection of negatively charged clay particles (Emerson and Bakker, 1973) and formation of complexes with organic matter (Bartoli and Philippy, 1990). As a matter of coincidence, the positive charge of pedogenetic oxides increases with decreasing pH. Kostopoulou and Panayiotopoulos (1993) also showed that the formation and stabiliza- tion of microaggregates in soils is a result of organic matter, Fe- and Al-oxides. For the Ah-horizons with the highest organic carbon contents, it can be assumed that the formation of clay-organic complexes is the dominant stabilizing process (cf. Bartoli et al., 1992). Further, the progressively increasing Al-saturation led to a higher shear

Page 13

T. Scholten / Soil Technology I1 (1997) 247-262 259

strength (Table 3) due to microaggregation by coagulation of clay particles. In the B-horizons with distinctly lower contents of organic matter, the cementing effect of Fe- and Al-oxides due to their positive charge under acid conditions and also the coagulation of soil particles by exchangeable Al-ions (Table 3) are most pronounced (cf. Edwards and Bremner, 1967; Bartoli and Philippy, 1990).

One reason for a lack of correlation between shear strength of the soils and the above described soil properties could be the varying quality of organic matter and microbiolog- ical activity and differences in soil biota (cf. Griffiths and Burns, 1972; Tisdall and Oades, 1982; Brussaard and Kooistra, 1993; Hartge and Stewart, 1995) between the sites and horizons. In addition, recent research has shown that the decomposition of organic matter takes place preferentially on aggregate surfaces (Augustin, 1992), leading to a reduced cementing effect of organic matter between single aggregates.

The stability factor shows that the saprolite and the transitional zone between the soil and the saprolite are very unstable as regards soil erosion. Within the saprolite, no organic carbon was present at all depths except a small amount (< 0.2%) in the Cwl-horizons (Fig. 2). Hence, the higher values of the stability factor result exclusively from decreasing contents of silt and very fine sand. In the field, these results are confirmed by the rapid retreat of the profile walls exposed in erosion gullies as soon as erosion undercuts the soil and reaches the saprolite. Overhangs of the solum fall in huge blocks and indicate that undercutting due to the distinctly lower structural stability of the saprolite favors the fast extension of erosion gullies.

5. Conclusions

In areas where the soil cover is completely denuded or only some tens of centimetres thick as a consequence of continuous sheet erosion, underlying saprolites form major parts of the rooting zone and, therefore, largely control the available water capacity. Thus, the conditions for revegetation after erosion or drought periods are better in places where the regolith cover is composed of soil-saprolite complexes because of the much higher available water capacity of the saprolites as well as the homogeneous distribution and larger amount of coarse pores in the saprolites that favor the potential water and nutrient uptake per unit volume by plants. This is of great importance especially in subtropical regions like Swaziland with a marked dry period.

Due to their thickness and high pore volume, saprolites are an important groundwater reservoir, especially in remote areas without access to a nearby river. Furthermore, the saprolites in the Swaziland Middleveld serve as a reservoir during high-intensity rainfall events lowering the risk of mass movement.

In landscapes outside the present humid tropics, saprolites predominantly represent a relic formation (cf. Felix-Henningsen et al., 1993). In areas where saprolite is still preserved and where it forms as part of a regolith, destabilization of the weathering mantle occurs, which becomes apparent in areas affected by soil erosion. High mass losses due to isovolumetric weathering and low amounts of cementing agents lead to low structural stability and high erodibility of saprolite, especially that derived from different magmatic rocks. All saprolite zones down to fresh bedrock are extremely sensitive to erosion. Therefore, the genetic and geochemical differences between the

Page 14

260 T. Scholten/Soil Technology II (1997) 247-262

saprolite zones have only a minor influence on erodibility. In contrast, the susceptibility to erosion of ferrallitic soils, originally developed from saprolite, is relatively low due to high soil structural stability. This inherent soil strength is superseded in importance by the alterations of the soil surface due to overgrazing and unwise land use, which lead to low infiltration capacities. The occurrence of runoff at low rainfall intensities especially on sheet erosion-affected and overgrazed rangeland areas underlines the high erosion risk.

If the saprolite is exposed by sheet-like or linear erosion of the solum, the erosion rates and land loss increase dramatically. Therefore, the saprolite is one essential precondition for the rapid development of deeply incised gullies in areas with magmatic rocks. This is in contrast to the good hydrological properties of the saprolites which result in faster regrowth of vegetation on eroded areas. The results show the importance of understanding the spatial distribution of soil-saprolite complexes as a basis for land evaluation and erosion-risk assessment.

Acknowledgements

The research was funded by the European Union STD 3 project on Soil Erosion and Sedimentation in Swaziland (Contract No. TS * CT90-0324) carried out in collaboration by the University of Swaziland, the Westfdische Wilhelms-Universitat Miinster and Cranfield University.

References

Augustin, S., 1992. Mikrobielle Stofftransfonnation in Bodenaggregaten. Berichte des Forschungszentrums Waldokosysteme, Reihe A, 85. Forschungszentrum Waldiikosysteme, GSttingen.

Bachmann, J.J., 1989. Auswirkungen der organischen Substanz verschiedenen Zersetzungsgrades auf die physikalischen Bodeneigenschaften von Sandboden. Mitteilgn. Dtsch. Bodenkundl. Gesellsch. 59 (I), 53-62.

Baeumer, K., 1992. Allgemeiner Pflanzenbau. Ulmer, Stuttgart. Bartoli, F., Burtin, G., Gdrif, .I., 1992. Influence of organic matter on aggregation in Oxisols rich in gibbsite

or in goethite: II. Clay dispersion, aggregate strength and water stability. Geoderma 54, 259-274. Bartoli, F., Philippy, R., 1990. Al-organic matter associations as cementing substances of ochreous brown soil

aggregates: preliminary examination. Soil Sci. 150, 745-751. Bauer, A., Black, A.L., 1992. Organic carbon effects on available water capacity of three soil textural groups.

Soil Sci. Sot. Am. J. 56, 248-254. Bohne, H., 1991. Stabilitat des BodengeRiges unter Einflug der Bodennutzung-Vorraussetzungen, An-

forderungen, Moglichkeiten. Berichte iiber Landwirtschaft, Sonderheft 204: Bodemmtzung und Boden- fruchtbarkeit, Bd. 2: 43-54. Parey. Hamburg, Berlin.

Boiffin, J., Momrier, G., 1985. Infiltration rate as affected by soil surface crusting caused by rainfall. International Symposium on the Assessment of Soil Surface Sealing and Crusting. Ghent, Belgium, pp. 210-217.

Boiffin, J., 1986. Stages and time dependency of soil crusting in situ. In: Callebaut, F., Gabriels, D., De Boodt, M. @Is.), Assessment of Soil Surface Sealing and Crusting. Flanders Centre of Soil Erosion and Conservation. Ghent, Belgium.

Page 15

T. Scholten / Soil Technology 11 (1997) 247-262 261

Brimhall, G.H., Lewis, C.J., Ford, C., Bratt, J., Taylor, G., Warin, O., 1991. Quantitative geochemical approach to pedogenesis: importance of parent material reduction, volumetric expansion and eolian influx in lateritization. Geoderma 51, 51-91.

Brussaard, L., Kooistra, M.J., (Ed%), 1993. Soil Structure/Soil Biota Interrelationships. Elsevier, Amsterdam. Dardis, G.F., 1990. Late Holocene erosion and colluvium deposition in Swaziland. Geology 18, 934-937. Edwards, A.P., Bremner, J.M., 1967. Microaggregates in soils. J. Soil Sci. 18, 64-73. El-Swaify, S.A., Dangler, E.W., 1982. Rainfall erosion in the tropics. A state of the art. Soil erosion and

conservation in the tropics. Soil Cons. Sot. Am. Spec. Publ. 43, l-25. Emerson, W.W., Bakker, AC., 1973. The comparative effects of exchangeable calcium, magnesium, and

sodium on some physical properties of red-brown earth subsoils: II. The spontaneous dispersion of aggregates in water. Aust. J. Soil Res. 11, 151-157.

FAO, 1989. Soil map of the world-revised legend. Reprint of World Soil Resources Report 60, FAO, Rome, 1988. ISRIC, Technical Paper 20. Wageningen.

Felix-Henningsen, P., Schotte, M., Scholten, T., 1993. Mineralogische Eigenschaften von Boden-Saprolit- Komplexen auf Kristallingesteinen in Swaziland (siidliches Afrika). Mitteilgn. Dtsch. Bodenkundl. Gesellsch. 72, 1293-1296.

Felix-Henningsen, P., Schotte, M., Scholten, T., 1995. Mikromorphogenese von Boden-Saprolit-Komplexen auf Kristallingesteinen in Swaziland (sildliches Afrika). Mitteilgn. Dtsch. Bodenkundl. Gesellsch. 76, 1423-1426.

Fiedler, H.J., 1958. Die Bedeutung synthetischer organischer Linearkolloide als Modellsubstanzen natiirlicher, and er Aggregatbildung und -stabilisierung beteiligter Verbindungen. Deutsche Akademie der Land- wirtschaften, Tagungsberichte 13: 245-260. Eigenverlag, Berlin.

Fr%nzle. O., 1984. Bodenkunde-Sidafrika (Mocambique, Swaziland, Republik Siidafrika). Afrika-Karten- werk, Beiheft S 4. Berlin.

Gardner, L.R., 1992. Long-term isovolumetric leaching of aluminium from rocks during weathering: implica- tions for the genesis of saprolite. Catena 19, 521-537.

Giovannini, G., Sequi, P., 1976. Iron and aluminium as cementing substances of soil aggregates: II. Changes in stability of soil aggregates following extraction of iron and aluminium by acetylacetone in a non-polar solvent. J. Soil Sci. 27, 148-153.

Griffiths, E., Bums, R.G., 1972. Interactions between phenolic substances and microbial polysaccharides in soil aggregation. Plant and Soil 36, 599-612.

Hartge, K.H., Horn, R., 1989. Die Physikalische Untersuchung von BSden: 2. Auflage. Ferdinand Enke Verlag, Stuttgart.

Hartge, K.H., Stewart, B.A., 1995. Soil structure-its Development and Function. CRC Press, London. Hillel, D., 1980. Applications of soil physics. Academic Press, Toronto, New York. Hudson, N.W., 1981. Soil conservation. Batsford, London. Jones, D.P., Graham, R.C., 1993. Water-holding characteristics of weathered granitic rock in chaparral and

forest ecosystems. Soil Sci. Sot. Am. J. 57, 256-261. Katschinski, N.A., 1958. Die Natur der mechanischen Stabilitat und Wasserstabilitit der Bodenstruktur.

Deutsche Akademie der Landwirtschaften, Tagungsberichte 13: 139-150. Eigenverlag, Berlin. Kiggundu, L., 1986. Distribution of rainfall erosivity in Swaziland. Research Paper 22. University of

Swaziland, Kwaluseni Campus, Swaziland. Kostopoulou, S.K., Panayiotopoulos, K.P., 1993. The influence of organic matter, cation bridges and Fe- and

Al-oxides and hydroxides on the aggregation of three Altisols. 2. Pflanzenem’ahr. Bodenk. 156, 435-440. Le Bissonnais, Y., 1990. Experimental study and modelling of soil surface crusting processes. Catena Suppl.

17, 13-28. Morgan, R.P.C., McIntyre, K., Vickers, A.W., Quinton, J.N., Rickson, R.J., 1997. A rainfall simulation study

of soil erosion on rangeland in Swaziland. Soil Technol. Morgan, R.P.C., Quinton, J.N., Rickson, R.J., 1991. EUROSEM-a User Guide. Silsoe College, Bedford. Morin, J., Benyamini, Y., Michaeli, A., 1981. The dynamics of soil crusting by rainfall impact and the water

movement in the soil profile. J. Hydrol. 52, 321-335. Mushala, H.M., Kiggundu, L., 1990. Rainfall characteristics for soil erosion monitoring in Swaziland. SADCC

Land and Water Management Research Programme. First Scientific Conference October 1990. Gaborone, Botswana.

Page 16

262 T Scholten / Soil Technology I I (1997) 247-262

Mushala, H.M., Felix-Henningsen, P., Scholten, T., 1993. Some preliminary investigations on soil erosion and saprolites in Swaziland. UNISWA J. Agric. 2, 20-28.

Murdoch, G., 1968. Soils and land capability in Swaziland. Swaziland Ministry of Agriculture. Mbabane, Swaziland.

Nixon, R.A., 1979. Differences in incongruent weathering of plagioclase and microcline-cation leaching versus precipitates. Geology 7, 221-224.

Pye, K., 1986. Mineralogical and textural controls on the weathering of granitoid rocks. Catena 13, 47-57. Renk, B., 1977. Die Morphogenese des swaziltidischen und transvaalischen Lowvelds und angrenzender

Fhachensysteme. Dissertation des Fachbereichs Mathematik-Naturwissenschaften der Christian-Albrechts- Universit%t Kiel. Unverdffentlicht.

Renk, W., 1977. Die raumliche Struktur und Genese der Bodendecke im Bereich der Grogen Randstufe Transvaals und Swazilands. Dissertation des Fachbereichs Mathematik-Naturwissenschaften der Christian- Albrechts-Universifat Kiel. Unverijffentlicht.

Rodgers, G.P., Holland, H.D., 1979. Weathering products within microcracks in feldspars. Geology 7, 278-280.

Roth, C., 1992. Die Bedeutung der Oberfhachenverschfammung fti die Ausldsung von AbfluS und Abtrag. BodenSkologie und Bodengenese, 6. Technische Universitit Berlin. Eigenverlag.

Schieber, M., 1989. Zwei modifizierte K-Nomogramme zur Bestimmung der Erodierbarkeit subtropischer BGden in Stidafrika. Z. Geomorph. N.F., Suppl.-Bd. 74, 147-158.

Schlichting, E., Blume, H.-P., Stahr, K., 1995. Bodenkundliches Praktikum. Blackwell, Hamburg. Scholten, T., Felix-Henningsen, P., 1993. Gully-Erosion in Boden-Saprolit-Komplexen auf Kristallingesteinen

in Swaziland (siidliches Afrika). Mitteilgn. Dtsch. Bodenkundl. Gesellsch. 72, 1247-1250. Scholten, T., Felix-Henningsen, P., Schotte, M., 1997. Geology, soils and saprolites of the Swaziland

Middleveld, Soil Technol. Schwertmann, U., Vogl, W., Kainz, M., 1987. Bodenerosion durch Wasser. Ulmer, Stuttgart. Skidmore, E.L., Layton, J.B., 1992. Dry-soil aggregate stability as influenced by selected soil properties. Soil

Sci. Sot. Am. J. 56, 557-561. Smith, H.J.C., Levy, G.J., Shainberg, I., 1990. Water-droplet energy and soil admendments: effects on

infiltration and erosion. Soil Sci. Sot. Am. J. 54, 1084-1087. Smithen, A.A., Schulze, R.E., 1982. The spatial distribution in southern Africa of rainfall erosivity for use in

the Universal Soil Loss Equation. Water SA, 8: 74-78, 165-167. Stolt, M.H., Baker, J.C., Simpson, T.W., 1992. Characterization and genesis of saprolite derived from gneissic

rocks of Virginia. Soil Sci. Sot. Am. J. 56, 531-539. Tisdall, J., Oades, J., 1982. Organic matter and water-stable aggregates in soils. J. Soil Sci. 33, 141-163. Turner, D.P., Sumner, M., 1978. The influence of initial soil water content on field measured infiltration rates.

Water SA 4, 18-24. Wendroth, 0.. Ehlers, W., 1989. Wassertransportparameter als Kenngr%en des Bodengeftiges. Mitteilgn.

Dtsch. Bodenkundl. Gesellsch. 59 (I), 261-266. WMS Associates, 1990. Investigation of the causes and hydrological implications of gully erosion in

Swaziland. WMS Associates, Fredericton, New Brunswick Canada, Ministery of Agriculture and Coopera- tives and Ministry of Natural Resources, Land Utilization and Energy. Mbabane, Swaziland.

Wilson, A.C., 1982. Geological Map of Swaziland Scale 1:250,000-Geological Summary. Mbabane, Swaziland.

Wischmeier, W.H., Smith, D.D., 1978. Predicting rainfall erosion losses. Agricultural Handbook No. 537. United States Department of Agriculture, Washington, DC.

Wysocki, D.A., Lietzke, D.A., Zelazny, L.W., 1988. Effects of parent material weathering on chemical and mineralogical properties of selected Hapludults in the Virginia Piedmont. Soil Sci. Sot. Am. J. 52, 196-203.

Zhang, H.Q., Hartge, K.H., 1990. Die Kohlsion ungesattigter Sandbdden und deren Beeinflussung durch organische Substanz. NZ: Soil Technol. 3, 31 l-326.