Palaeoenvironmental mosaic of Proconsul habitats: geochemical and sedimentalogical interpretation of Kisingiri fossil sites, Western Kenya

Palaeoenvironmental mosaic of Proconsul habitats: geochemical and sedimentalogical interpretation of Kisingiri fossil sites, Western Kenya

Journal of African Earth Sciences 39 (2004) 63–79 www.elsevier.com/locate/jafrearsci Palaeoenvironmental mosaic of Proconsul habitats: geochemical an...

577KB Sizes 2 Downloads 79 Views

Journal of African Earth Sciences 39 (2004) 63–79 www.elsevier.com/locate/jafrearsci

Palaeoenvironmental mosaic of Proconsul habitats: geochemical and sedimentalogical interpretation of Kisingiri fossil sites, Western Kenya Matthew S. Forbes a

a,1

, Erick A. Bestland

a,*

, Evelyn S. Krull b, Danielle G. Dicker

c

School of Chemistry, Physics and Earth Sciences, Flinders University, GPO Box 2100, Adelaide, Australia b CSIRO Land and Water, PMB 2, Glen Osmond, Australia c School of Biological Sciences, Flinders University, GPO Box 2100, Adelaide, Australia Received 27 August 2003; accepted 29 April 2004 Available online 20 July 2004

Abstract Geochemical differences in bulk rock geochemistry, mass balance weathering calculations and carbon isotope ratios of palaeosol organic matter between early Miocene palaeosols from the Kisingiri volcano are the result of both changes in the depositional regime and a shift in climatic conditions. Wayondo Formation palaeo-Alfisols located on Rusinga Island (Gumba Beds) and at nearby Karungu, which were deposited prior to the commencement of the main phase of volcanic activity (17.8 Ma), possess well developed clay horizons, moderate base cation contents and strong degrees of weathering. These palaeosols, when compared to their unweathered granitic parent exhibit collapse and loss of 80–90% of base cations and silica suggesting that their formation was during a period of landscape stability, when climatic conditions were humid to sub-humid. In comparison the overlying Kiahera Formation palaeosols, which were deposited during hydromaginatic activity at Kisingiri, exhibit a wide range of dilation and addition, together with individual elemental losses of up to 50%. Together, these weathering trends suggest that there was a climatic shift from wetter conditions to drier, semi-arid conditions, during the initial stages of the volcanoe’s growth. This is supported by d13 C values of soil organic matter from A-horizons of the Wayondo Formation palaeosols (Rusinga Island )25.9& and Karungu )25.0&) and the volcanically derived Hiwegi Formation ()23.2&) and Kiahera Formation ()24.2&) palaeosols. While all d13 C values are indicative of vegetation using the C3 synthetic pathway, d13 C values of the palaeosols that formed during the eruption (Hiwegi and Kiahera) show more 13 C-enriched values compared with the pre-volcanic palaeosols (Wayondo). The reason for the relatively 13 C-enriched values is suggested to be due to reduced fractionation due to seasonal or year-round water stress in an environment that had changed from humid (Wayondo) to sub-humid/semi-arid with the onset of volcanic activity. The greater abundance of calcareous material within these syn-eruptively formed fossil soils, compared with the pre-volcanic ones, also supports a change towards greater aridity. d13 C values of organic matter from the Kulu Lake Beds (in the Hiwegi Formation) have average d13 C value of )25.4&. However, it is not possible to determine whether these values reflect the d13 C values of the terrestrial vegetation of the watershed or whether they are dominated by lacustrine-produced organic matter. Our data indicate that the well-developed Gumba and Karungu fossil soils of the Wayondo Formation represent part of a sustained stable alluvial system and had formed under humid conditions. With the onset of volcanic activity and the creation of a rapidly aggrading pyroclastic system, on which the Kiahera, Kulu and Hiwegi Formations of the Rusinga Group formed, changes occurred in the weathering regime and in the d13 C values of organic matter, suggesting a change towards greater aridity in the pre-versus syn-volcanically formed palaeosols. Ó 2004 Elsevier Ltd. All rights reserved. Keywords: Early Miocene palaeosols; Kisingiri volcano; Mass balance geochemistry; Carbon isotopes

1. Introduction

*

Corresponding author. Fax: +1-208-236-4414. ,E-mail addresses: [email protected] erick.bestland@flinders.edu.au (E.A. Bestland). 1 Fax: +61-8201-2676.

(M.S.

Forbes),

0899-5362/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.jafrearsci.2004.04.003

During the Eocene and the Miocene, primates began to evolve, thriving in mainly tropical forest habitats (Andrews, 1992a,b). Previous research suggests that during this period these habitats gradually changed, becoming more open and grassy, possibly influencing

64

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

the development of bipedal species (Retallack et al., 1990; Cerling et al., 1991). A critical link in research between early arboreal primates and hominoids of more open country habitats are the early Miocene Proconsul fossil sites of Western Kenya (Shipman et al., 1981). Strata on Rusinga Island (Lake Victoria) and nearby sites such as Mfangano Island and Karungu contain one of the most prolific Proconsul bearing fossil assemblages discovered (Leakey, 1967; Pickford, 1986) (Fig. 1). The uncommon alkaline-rich nature of the volcanic debris from the Kisingiri Volcano greatly assisted in the preservation of animal and plant remains, making these deposits very fossiliferous (Van Couvering and Miller, 1969; Baker, 1987). Stratigraphic sections located on Rusinga Island and at nearby Karungu, both derived from Kisingiri, contain not only fossiliferous strata but also many palaeosols (Fig. 1). Palaeosols can be useful in deciphering ancient vegetation mosaics, because, unlike fossils, they are by definition in the place they are formed (Retallack et al., 1990). The identification and analysis of properties of palaeosol sequences such as horizon morphology and structure, clay structures and bulk rock chemistry provides estimates of degree of weathering and weathering flux in the geologic past (Bestland et al., 1996). It also allows for the palaeosol to be interpreted as a basic soil type (sensu strito, US Soil Taxonomy) with most soil orders having been recognised in the pre-Quaternary Stratigraphic record (Retallack, 1990; Retallack and Germ an-Heins, 1994; Driese et al., 1992; Bestland et al., 1996). The degree of accumulation of residual elements

Fig. 1. Location map of southwest Kenya showing the Kisingiri Volcano, and associated deposits, Rusinga Island and the Nyanza and Gregory Rifts.

such as iron, titanium and aluminium and the leaching of mobile elements such as sodium, potassium and calcium relates to the extent of weathering. Density measurements and geochemical analysis allow for mass balance calculation of pedogenic weathering using methods of Brimhall et al. (1988). Stable carbon isotope ratios of palaeosol organic matter have also been widely used as palaeoecological indicators (Cerling et al., 1989; Quade et al., 1995). Soil organic matter is a decomposition product of plant debris and as such records the carbon isotopic composition of vegetation that grew in the soil. The isotopic composition of modern plants using the C3 photosynthetic pathway have d13 C values ranging from )32& to )22&, with an average of )27&, whereas plants with the C4 pathway have average values of )13& and a range of )17 and )9& (Dienes, 1980; Cerling et al., 1991).

2. Geological setting The Kisingiri volcano has drawn considerable research attention due to its exposed volcanic core and central dome, as well as its volcaniclastic strata (LeBas, 1977; Pickford, 1982; LeBas, 1987; Bestland et al., 1995; Bestland and Krull, 1999). Kisingiri experienced two general phases of volcanic growth (Bestland et al., 1995). The first phase involved massive doming associated with intrusions that was followed by hydromagmatic and pyroclastic activity. The former resulting in alluvial material being shed from the low relief dome, and the latter resulted in pyroclastic and hydromagmatic material deposited from a low relief hydromagmal complex. The second phase comprised the building of a high relief central strato-cone resulting in the accumulation of coarse grained debris (Van Couvering and Miller, 1969; Bestland et al., 1995). The two phases are recorded in three distinct stratigraphic intervals on Rusinga Island and elsewhere, with two of these intervals combining to represent the first phase. Each consists of a lower section rich in Pre-Cambrian detritus and an upper section of volcaniclastic and pyroclastic deposits. Fossil sites of the Kisingiri strata are largely contained in the Rusinga Group (Fig. 2). However, underlying units from the initial doming event, such as the alluvial Gumba Bed on Rusinga Island and similar units at Karungu have also yielded fossils (Drake et al., 1988). K–Ar dating of biotite samples indicates that the first phase of volcanic development was much quicker and episodic than previously thought (Drake et al., 1988). This phreatic or hydromagmatic phase (cycles II and I) lasted only a few hundred thousand years, centred around 17.8 Ma, with the strato-volcanic phase following. At Kisingiri, intrusions of alkaline silicate magmas in the sub-surface resulted in the doming of the Pre-Cambrian basement prior to volcanic activity (LeBas, 1977).

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

65

This report details the palaeosols of the Wayondo Formation (Gumba Beds) located on Rusinga Island, and at Karungu, both having formed prior to the commencement of the main phase of volcanic activity. Based on the stratigraphy and chronological determinations (Drake et al., 1988), the age of these deposits are somewhat older than 17.80.5 million years. For the purposes of comparing units within a chronology from the Kiahera and Hiwegi formations, 100 ka time intervals are used.

3. Objectives

Fig. 2. Miocene stratigraphy of Rusinga Island, adapted from Bestland (1991). Selected K–Ar age determinations of biotite, together with the mammalian fauna constrain the age of the Rusinga Group to early Miocene, some 17–18 Ma. The Wayondo Formation, which consists of both the Gumba Beds of Rusinga Island and the Karungu deposits reported in this study are pre-volcanic and are not part of the overlying volcanic derived Rusinga Group.

The resultant debris produced the Wayondo Formation, a conglomeratic unit dominated by granitic and gneissic clasts, but also consisting of inter bedded fine grain units of clay and silt. Following this initial phase of doming a sequence of repeated explosive eruptions produced pyroclastic material, composed primarily of tuffs and volcaniclastic sandstones of the Kiahera Formations (Fig. 2). Together, the Wayondo Formation and the Kiahera Formation constitute the first cycle of doming and eruption (Bestland et al., 1995). Interstratified within these units are moderately developed palaeosols (Bestland and Retallack, 1993; Bestland and Krull, 1999), which are sandwiched between hydromagmatic deposits and indicate periods of minimal volcanic activity. The second cycle consisted of doming and basement exposure, which again allowed coarse granitic debris to be deposited as the basal Rusinga Agglomerate (Boulder Breccia Member). A period of faulting soon after resulted in the block faulted lake basins and fan deltas of the Kulu Formation (Bestland, 1991). This was accompanied by the continuation of hydromagmatic pyroclastic flows and airfall, producing the sandy and ashy Hiwegi Formation (Pickford, 1982; Bestland et al., 1995). Phase two of Kisingiris formation refers to the development of a high relief (3000 m) strato-volcano (Shackleton, 1951). A more significant cycle of doming and basement exposure resulted in the widespread deposition of course granitic clast-rich debris around Kisingiri. The resulting Kiangata Agglomerate cut channels in existing deposits, filled and then overflowed from these channels to produce a coarse, debris-flow dominated apron facies (Pickford, 1986; Bestland et al., 1995).

The primary aim of this report is to further elucidate the palaeoclimatic conditions during the early stages of deposition associated with the Kisingiri volcano. In this study we present new data comprised of field observations, whole rock geochemistry and carbon isotope ratios on fossil soils (Gumba Beds) and lacustrine strata (Kulu Formation) located on Rusinga Island. Additionally detailed stratigraphic sections from Karungu are reported along with geochemistry and stable carbon isotope ratios from soil organic matter from palaeosols. The two sites are important in that both the Gumba and Karungu palaeosols analysed are from the Wayondo Formation, deposited as a result of the doming of Kisingiri prior to the onset of volcanic activity. With most of the previous stratigraphic work on the Kisingiriderived deposits having been completed on Rusinga and Mfangano Islands, investigations into the Karungu strata located on the opposite, or southern side of the volcano will allow for comparisons to be made between the two, thus providing a more detailed perspective of the nature of palaeoclimatic conditions. These data will help to arrive at a synthesis of a prevolcanic landscape mosaic, which can then be compared to previous palaeoenvirontnental interpretations from isotopic and geochemical data irom the younger volcanic deposits of the Kiahera, and Hiwegi Formations (Bestland and Krull, 1999) and the lacustrine sediments of the Kulu Formation. These interpretations will allow for an early Miocene palaeoenvironmental mosaic to be formed for the Kisingiri area, by identifying changes in climate through time and also variations in the environs in the Kisingiri area both prior to and after the onset of volcanic activity.

4. Field and laboratory methods 4.1. Field techniques The stratigraphic sections on Rusinga Island and at Karungu, described in this report, were measured and described at the horizon scale (5–10 cm) from fresh rock

66

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

excavated in trenches, while detailed fossil soil profiles were measured at the centimetre-scale. Standard geological techniques were used to estimate grain size and mineralogy, and field methods following Retallack (1988) were used to characterise pedological features. Macroscopic inspection of numerous samples in each horizon was undertaken in order to determine representative features. Semi-quantitative estimates of degree of fossil soil development, and the strength and abundance of palaeosol features (e.g. peds, cutans, root traces, and burrow structures) were made following a scale developed by Retallack (1988). 4.2. X-ray fluorescence (XRF) Samples were analysed at Adelaide University using a Phillips PW 1480 XRF Spectrometer. Major elements in their oxidised state such as SiO2 , A12 O3 , Fe2 O3 , MnO, MgO, CaO, Na2 O, K2 O, TiO2 , P2 O5 and SO3 (Table 1) were determined (Norrish and Chappell, 1977). Iron was analysed as a total (combining the ferrous and ferric forms); and expressed as Fe2 O3 . However, data for the Kiahera Formation from Bestland and Krull (1999) are reported as FeO. Samples for major elemental analyses were crushed, dried at 110 °C, and combusted at 960 °C overnight. Between 1.0 and 1.1 g of the combusted material was weighed and combined with four times this weight of flux (35.3% lithium tetraborate and 64.7% lithium metabotrate). Using Pt–Au crucibles the mixture was fused at approximately 1150 °C. Minor or trace element (Sr, Rb, Y, Zr, Nb, Pb, Th, U, Ba, Sc, Ga, V, Cr, Ce, Nd, La, Ni, Cu, Zn and Co) analysis was undertaken to obtain data in concentrations of one to several thousand parts per million (ppm).

sj;w ¼ sw el;w .p .w Cj;w Cj;p

Geochemical differences between palaeosol horizons and their parent materials can be used to determine the mass transfer of base cations and other pertinent elements (Brimhall and Dietrich, 1987; Brimhall et al., 1988; Chadwick et al., 1990; Bestland et al., 1996). These mass balance techniques are used to quantify the pedogenic strain (volume change), which can be directly related to the degree of weathering for palaeosols. During soil formation, chemical gains and losses depend on the nature and extent of pedogenic processes. The higher the intensity of weathering, the larger the percentage of original soil constituents that are leached from the profile, with resulting reduction of volume of the soil formed compared to a designated parent material. Using Eq. (1) with the assumption of a parent composition and density, numerical expressions can be derived to produce the pedogenic mass transport equation

ð1Þ

mass transport function of any element in weathered product strain of weathered product according to immobile element bulk density of parent material bulk density of weathered product conc. in wt% of chosen element in the weathered product conc. in wt% of chosen element in the parent material

Pedogenic mass transport ðsÞ is the gain or loss of an element in the soil when compared to the parent material normalised to an immobile element. Values for mass transport ðsÞ range from negative one (collapse) up to large positive values (dilation). When s is equal to zero the element is immobile, indicating that it is only affected by internal closed-chemical-system processes (Brimhall and Dietrich, 1987; Brimhall et al., 1988; Chadwick et al., 1990). To estimate the volume change we use an immobile element for which the mass transport is assumed to be zero. As such, Eq. (1) reduces to Eq. (2), the pedogenic strain el;w ¼ el;w .p .w Ci;p Ci;w

4.3. Mass balance

qw Cj;w ðel;w þ 1Þ  1 qp Cj;p

qp Ci;p 1 qw Ci;w

ð2Þ

strain of weathered product according to immobile element bulk density of parent material bulk density of weathered product conc. in wt% of immobile element in the parent material conc. in wt% of immobile element in the weathered product

Positive strain indicates dilation. Negative strain indicates collapse, which is shown by an increase in the concentration of the immobile element. For many moderately to well-developed soils, pedogenic strain represents the fraction of parent material that has been weathered and removed from the system. 4.4. Carbon isotopes Palaeosols were crushed, sieved through a 0.210-mm screen, milled to a powder and then treated with 2 M HC1 to remove acid soluble minerals (CaCO3 ). Acidinsoluble residues were then washed until neutral and oven dried for at least 12 h. Carbon isotope results relating to the Gumba and Karungu sites were obtained from the Rafter Stable Isotope laboratory in Lower Hutt, New Zealand, using an Europa Geo 20–20 mass spectrometer. Isotope results presented (Table 2) are

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

67

Table 1 XRF bulk rock geochemical data for the Gumba and Karungu palaeosols of the Wayondo Formation and the Kiahera Formation of the Rusinga Group (from Bestland and Retallack, 1993) Al2 O3

Fe2 O3

MnO

Na2 O

K2 O

TiO2

P2 O5

LOI

Total

Karungu palaeosols Ku1-2 44.61 Ku1-3 44.16 Ku1-4 43.74 Ku1-5 44.00 Ku1-7 42.63 Ku1-9 43.64

12.20 12.05 11.96 11.89 11.11 11.48

13.32 13.41 13.48 13.48 13.45 12.83

0.09 0.11 0.11 0.10 0.09 0.09

3.03 3.02 3.02 2.97 2.70 2.72

0.92 0.85 0.93 0.95 2.77 3.23

1.12 1.18 1.24 1.34 1.94 1.71

5.96 5.88 5.83 5.77 5.30 5.42

2.67 2.66 2.59 2.56 2.43 2.59

0.52 0.47 0.52 0.52 0.69 0.69

14.9 15.8 16.11 15.81 16.0 15.0

99.4 99.6 99.6 99.4 99.2 99.4

Ku3-5 Ku3-7 Ku3-8

41.00 42.77 44.07

11.42 11.13 10.08

9.94 10.42 10.82

0.10 0.15 0.15

2.80 2.73 2.36

7.11 7.59 8.78

1.39 1.38 1.64

4.84 4.87 4.31

1.75 1.97 2.37

0.59 0.75 0.96

18.6 15.6 13.2

99.7 99.4 98.8

Ku4-2 Ku4-3 Ku4-4 Ku4-5

49.37 47.87 46.16 43.19

14.82 14.13 12.5 11.51

12.38 13.7 14.26 11.99

0.09 0.18 0.13 0.22

3.54 3.51 2.88 2.84

2.89 3.34 5.81 10.14

0.74 0.78 1.27 1.21

5.86 5.7 5.09 4.72

2.47 2.32 2.88 2.36

0.74 0.7 1.05 0.89

6.74 7.24 7.46 10.6

99.7 99.5 99.6 99.7

Gumba palaeosols GR(North) and GM(South) GR-1 40.82 15.47 8.92 0.31 GR-2 25.35 7.79 9.89 0.89 GR-3 40.91 14.41 9.41 0.46 GR-4 39.54 11.11 10.27 0.49 GR-5 39.83 10.72 8.76 0.39 GR-6 42.59 15.84 10.52 0.25 GR-7 45.76 17.3 11.54 0.17

2.95 3.61 3.08 2.04 2.09 2.4 2.48

11.22 25.02 10.78 14.63 15.53 8.89 5.2

1.21 0.88 1.08 1.15 1.14 0.87 0.99

2.29 1.28 2.64 2.11 2.15 2.42 2.53

1.37 1.3 1.27 1.97 1.9 1.67 1.81

1.13 1.54 1.38 0.81 0.74 0.45 0.47

13.14 21.31 12.72 14.65 15.41 13.17 11.18

98.9 98.9 98.2 98.9 98.9 99.1 99.5

GM-2 GM-3 GM-4 GM-5 GM-6 GM-7 GM-8

20.3 10.99 14.65 18.45 13.3 12.53 12.41

99.6 99.7 99.8 99.9 99.7 99.6 99.5

#

SiO2

34.33 47.33 41.83 35.66 45.5 45.93 45.91

9.13 12.65 11.61 11.32 15.66 16.28 16.55

Kiahera Formation palaeosols k1 31.61 7.6 k20 45.82 10.69 k2 37.62 8.62 k3 34.54 8.11 k4 55.4 12.71 k5 55.47 12.45 k6 55.5 12.45 k7 57.02 12.37 k8 51.59 10.65 k9 51.9 10.49 k10 52.05 11.12 k21 29.79 6.84 k22 45.88 10.57 k29 28.17 6.93 k1l 46.71 10.89 k12 43.1 10.1 k13 31.84 7.26 k14 35.97 8.14 k16 47.1 10.55 k17 41.95 9.5 k18 27.58 6.28 k19 46.8 10.64 k25 48.64 11.88 k26 47.07 11.22 k27 34.2 8.12 k28 44.46 10.67 k35 45.93 11.39 k36 46.87 10.66 k37 41.75 8.77

MgO

CaO

6.34 10.71 9.48 8.8 10.54 11.12 11.31

0.42 0.21 0.32 0.4 0.21 0.15 0.14

8.72 3.51 3.37 3.12 3.56 2.92 2.79

14.56 7.22 12.33 16.98 5.33 4.99 4.77

1.38 1.13 1.14 0.81 0.83 0.85 0.81

2.51 3.66 2.85 2.42 2.69 2.59 2.51

1.42 1.75 1.54 1.32 1.59 1.68 1.73

0.41 0.52 0.63 0.62 0.43 0.56 0.54

8.53 7.34 6.3 6.3 7.39 7.05 8.28 7.88 7.59 7.5 6.41 5 7.58 5.6 8.73 7.59 5.74 7.33 9.7 8.86 5.07 8.97 8.04 8.21 4.65 11.47 5.96 7.03 7.52

0.39 0.23 0.04 0.04 0.12 0.12 0.12 0.11 0.17 0.16 0.21 0.5 0.24 0.42 0.16 0.17 0.32 0.32 0.19 0.23 0.37 0.17 0.18 0.19 0.45 0.19 0.28 0.21 0.31

4.29 4.81 4.54 4.36 4.63 4.8 4.67 4.46 3.21 3.15 3.4 9.37 6.11 13.29 5.66 5.45 5.35 6.14 7.72 8.33 6.99 7.41 5.49 5.5 5.47 6.59 5.93 7.14 6.63

25.57 14.46 21.14 24.04 3.9 4.16 3.85 3.75 11.63 12.19 12.19 28.55 12.95 24.23 10.96 14.49 29.05 22.66 6.78 12.73 33.02 8.72 7.96 9.76 26.38 10.85 11.16 8.74 16.34

0.45 2.27 0.61 0.49 1.78 1.75 1.7 1.91 2.23 2.24 2.39 0.62 1.67 0.59 0.88 0.74 0.63 0.73 0.84 0.74 0.45 1.02 0.8 1.01 0.79 0.83 1.49 1.22 1.45

2.08 2.12 2.73 2.57 3.55 3.44 3.07 3.21 2.75 2.82 2.66 1.81 2.43 1.97 2.38 2.16 1.99 2.14 2.69 2.52 1.79 2.45 2.62 2.85 2.68 3.06 3.07 2.74 2.02

2.77 2.12 2.29 2.4 2.2 2.24 2.29 2.23 2.14 2.13 1.75 1.43 2.12 1.86 2.17 2.08 1.81 2.24 2.74 8.86 1.54 2.53 2.08 2.06 1.67 2.71 1.89 2.48 2.46

0.68 0.52 0.77 0.75 0.7 0.71 0.88 0.71 0.48 0.52 0.53 0.53 0.63 0.56 0.77 0.85 0.63 0.57 0.61 0.56 0.44 0.68 1.05 0.98 0.61 0.72 0.97 0.78 0.63

84.93 92.12 85.65 85.42 93.75 93.64 94.12 94.96 93.96 93.45 93.86 85.84 91.76 85.24 90.51 87.93 85.66 87.55 90.55 89.43 84.51 91.08 89.92 89.92 86.03 88.99 89.91 89.79 89.69

68

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

Table 2 d13 C carbon isotope results of organic matter from the Rusinga Island and Karungu palaeosols of the Wayondo Formation d13 C

% orgC

Hiwegi Fm KN-M2 KN-M2(d) KN-M3 KN-M3(d) KN-M3-L KN-M4 KN-M4(d) KN-M4-L KN-M4-L(d) KN-M5 KN-M5-L KN-M6 KN-M6-L KN-M13 KN-M13(d) KN-M14 KN-M15 WT-7 WT-7(d) WT-8

)23.2 )23.0 )22.8 )22.6 )24.0 )22.5 )23.9 )23.1 )23.7 )23.2 )23.6 )23.1 )23.3 )22.4 )22.9 )24.5 )23.1 )23.8 )23.9 )24.3

0.06 0.05 0.03 0.03 0.03 0.02 0.02 0.02 0.03 0.02 0.02 0.06 0.02 0.04 0.04 0.03 0.02 0.08 0.07 0.09

Average

)23.4

0.04

Wayondo Fm Gl G2 G3 G4 G5

)25.9 )25.2 )24.7 )24.0 )23.6

0.02 0.03 0.02 0.02 0.03

Average

)24.7

0.02

Karungu Fm KU1-1 KU1-2 KU1-3 KU1-4 KU1-5 KU1-6 KU1-7 KU1-8 KU1-9

)25.2 )25.2 )23.5 )22.7 )29.9 )23.9 )25.1 )23.6 )22.4

0.06 0.06 0.05 0.05 0.06 0.05 0.04 0.02 0.05

Average

)24.0

0.05

KU3-3 KU3-4 KU3-5 KU3-6 KU3-7 KU3-8

)27.6 )22.6 )22.2 )23.2 )22.3 )25.1

0.08 0.08 0.10 0.09 0.09 0.10

Average

)22.4

0.09

Kiahera Fm KN-C1 KN-Cl(d1) KN-Cl(d2) KN-C2 KN-C2(d1) KN-C3 KN-C3(d1) KN-C4 KN-C5

)24.2 )24.0 )24.1 )24.3 )24.2 )24.8 )24.1 )24.5 )24.6

0.07 0.04 0.05 0.06 0.06 0.09 0.07 0.17 0.13

C/N

(continued on next page)

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

69

Table 2 (continued) d13 C

% orgC

KN-C6 KN-C6(d1) KN-C6(d2) KN-C7 KN-C7(d1) KN-C7(d2) KN-C9 KHG-llA-Ll KHG-11A-L2 KHG-11 KHG-11-L1 KHG-11-L2 KHG-11-L3 KHG-12-L1 KHG-12-L2 KHG-13-L1 KHG-13-L2 KHG-15-L1 KHG-l5-Ll(d)

)24.7 )24.1 )24.7 )24.9 )23.8 )24.8 )24.9 )25.0 )24.9 )24.6 )24.4 )22.8 )22.4 )22.9 )22.3 )23.4 )24.2 )24.6 )24.6

0.03 0.03 0.03 0.03 0.02 0.02 0.18 0.14 0.12 0.12 0.20 0.04 0.04 0.04 0.03 0.05 0.03 0.03 0.03

Average

)24.2

0.07

Kulu Fm NG-1 NG-2C NG-3A NG-3C NG-5 NG-6A NG-8B NG-9 NG-10

)22.8 )27.8 )24.7 )25.7 )25.8 )25.1 )24.4 )26.0 )26.1

0.05 0.05 0.04 0.03 0.04 0.04 0.03 0.02 0.03

Average

)25.4

0.04

C/N

5.2 2.8 2.8 4.5 5.4 4.7 3.6 5.6 4.9

Also data from the lacustrine Kulu Formation of the Rusinga Group, compared to previous data from Bestland and Krull (1999) regarding the Kiahera and Hiwegi Formations.

reported in the conventional d-notation as per mil deviation from the PDB standard (Peterson and Fry, 1987).

5. Karungu site results and discussion 5.1. Karungu stratigraphy The Karungu section consists of approximately 25 m of strata (Fig. 3), which are exposed laterally over several kilometres. Saprolitised and deeply weathered PreCambrian basement rocks underlie the section which predominantly is a basal laterite conglomerate-breccia that grades up-section into conglomerates with mixed limestone and silicate detritus. Between 5 and 10 m from the base of the section are algal ball conglomerates, which contain rounded and coated micrite clasts. There are a number of intervals with palaeosols, which are distinctive in their rhythmic interbedding of mudstones and fine sandstones. Eleven distinct groups of fining upward sequences of bedded mudstones–fine sandstones with overlying palaeosols have been recognised (Fig. 3).

Layers of conglomerate or coarse sandstone define the base of each of the fining upward sequences. We interpret these sequences as channel and levee splay deposits of a meandering river system, where distance from channel is related to sedimentation rate and grain size, following general models of alluvial aggradation (Coleman, 1969; Farrell, 1987; Bown and Kraus, 1987). The observed pattern of maturing-upward palaeosols directly overlain by channel–levee deposits argues against a systematic, regular channel migration across a floodplain. This model would produce matureupward and then weakening-upward palaeosols as the channel migrates away and then toward the site of deposition (Bown and Kraus, 1987). Therefore the thickness and lateral extent of the fining-upward sequences, which are on the order of a few metres and a few tens of metres, respectively, are interpreted as the products of a laterally migrating medium sized river system. However, the distinct lack of major channel cuts within the stratigraphic sequence suggests that the stream system was only small to moderate in size. It would be expected that larger channel systems would

70

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

Palaeosol features are distinct and include root traces, clay-filled pedotubes, clay veins and ped structures. These features are all indicators of moderate soil development (Retallack, 1990). A gradation exists between massive to poorly layered mudstones and the horizons with clear pedogenic modification. The degree of palaeosol development, which is assessed by observing the pedologic characteristics generally, increases up-section within the fining-upward sequence, with sandstones and siltstones of the next sequence directly overlying the most mature palaeosol horizon. Three palaeosols within the fining-upward sequences that had the strongest developed pedologic characteristics and which best represented the majority of fossil soils in the strata were investigated in detail. These were palaeosol KU-1, KU-3, and KU-4 located 9, 11 and 14 m from the bottom of the section, respectively. 5.3. KU-1 profile

Fig. 3. Composite stratigraphic section of the Karungu site, which is representative of the Wayondo Formation. Evident is the 11 repeating patterns of fining upward sequences. The three detailed palaeosols KU-1, KU-3 and KU-4 analysed in this project are identified in the diagram.

produce substantial incision of at least several metres into underlying deposits. Another possibility of a forcing mechanism for the fining-upward sequences involves volcanic eruption and pyroclastic deposition proximal to the volcanic centre. Erosion of this material with transport and reworking on the floodplain could cause a depositional sequence to form. The presence of fresh volcanic minerals such as feldspars, pyroxene and biotite in only the very top of the section discounts this volcanically related depositional scenario. 5.2. Palaeosols The Karungu palaeosols are predominantly located in the upper half of each sequence and are formed on layered siltstones, mudstones and sandstones (Fig. 3).

The KU-1 palaeosol is below a sandstone unit that exhibits cross bedding and sand tilled holes, burrows and cracks (Fig. 4). The palaeosol profile is more than 60 cm thick and displays an olive brown (2.5Y 4/2) colour throughout, with some distinct red mottled root traces towards the top of the profile. A coarser-grained horizon occurs in the upper 10 cm of the profile, and is interpreted as a sandy A-horizon. At 35 cm the profile displays the strongest clay structures and is interpreted as a Bt horizon. Decreasing Al2 O3 below this point supports the observed decrease in clay content. The presence of abundant clay with defined pedogenic clay structures suggests the KU-1 palaeosol is an Argillisol (following the classification of Mack et al., 1993). Analysis of base cations (Ca, Mg, K and Na) is greater than ten percent for all samples in the profile. Nettleton et al. (2000) has used this method to distinguish cation rich Argillisols (Alfisols) from cation depleted Argillisols (Ultisols). Alfisols form in less intensely weathered environments than Ultisols and retain a moderate abundance of the mobile elements that otherwise would leach from the soil profile. Generally Alfisols form in environments that are sub-humid to humid (Hall, 1983), however they are known to develop in drier environs. 5.4. KU-3 profile The KU-3 palaeosol is located below several layers of yellow to olive-yellow (2.5Y 6/6) sandstones (Fig. 5), which contain small root traces and pedotubes, indicators of minor pedogenesis. The main fossil soil horizon of KU-3 consists of a distinct olive (5Y 4/3) to olive grey (5Y 4/2) clayey horizon between 20 and 35 cm depth. This horizon exhibits good clay structure in the form of columnar peds with clay slickensides and is interpreted as a Bt horizon. Towards the bottom of the profile sand

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

71

Fig. 4. Geochemical profile of atomic abundances, soil horizon interpretations and carbon isotopic data for the Karungu palaeosol KU-1.

Fig. 5. Geochemical profile of atomic abundances, soil horizon interpretations and carbon isotopic data for the Karungu palaeosol KU-3 (symbols as in legend in Fig. 4).

content increases and the clay content decreases until at 50 cm it grades into conglomerate. Base cations total over 15% for the entire palaeosol and combined with pedogenic features support the interpretation of this profile as a palaeo-Alfisol similar to KU-1.

5.5. KU-4 profile The KU-4 profile consists of two distinct horizons: a grey-green (G/2 5/5) horizon that overlies a red-brown (7.5YR 5/4) horizon (Fig. 6). The amount of iron is

72

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

Fig. 6. Geochemical profile of atomic abundances and soil horizon interpretations for the Karungu palaeosol KU-4 (symbols as in legend in Fig. 4).

greater in the lower, red horizon than in the grey-green horizon. Evidence of clay translocation (slickensides and cutans) occurs throughout the two horizons. Concentration of these clay-structures was significantly higher within the grey-green horizon than in the red-brown horizon, this is reflected by both the Al2 O3 and SiO2 concentrations, which are lower in the red-brown horizon than in the green horizon. Clay structures and features seen in this profile are more predominant than in the other two Karungu palaeosols, allowing for this fossil soil to be nominated as the strongest developed in the overall sequence. The grey-green horizon has a higher calcareous component that the red-brown horizon. Noticeable within the grey-green horizon are two calcareous veins, which lie at an angle to the rest of the profile. These are likely to be dipping carbonate horizons, a modern contribution, to the system, deposited after the initial formation of the soil profile. In comparison calcareous nodules evident in the red-brown horizon suggests that the carbonate content here most likely developed during formation of the palaeosol. The red-brown clay horizon is a Bk horizon while the grey-green clay horizon is a Bt horizon. The occurrence of feldspars, mica and quartz within the red-brown palaeosol provides support for a parent material composition consisting of weathered detritus from the PreCambrian granitic basement. The abrupt colour change between the two horizons suggests that the grey-green palaeosol probably formed in an environment saturated with water and/or there was abundant organic matter. As such the grey-green soil would have experienced reduction of iron (Fe2þ ), while the iron in the red-brown horizon would be oxidised (Fe3þ ). The presence of strong clay features in this palaeosol initially suggests an Arg-

illsol classification. However the presence of carbonate nodules, suggests that this cation-depleted palaeo-Alfisol does not represent a humid environment, but rather a sub-humid to semi-arid environment. The repeating regularity of the fining-upward sequences of conglomerate, sandstone and palaeo-Alfisols evident in the Karungu sequence argues for a cyclic forcing mechanism. General soil principles worked out from soil chronosequences (Birkeland et al., 1991), infers the time of formation required for channel and levee deposits would be a few tens to hundreds of years. For the palaeosols in this section, a time frame of a thousand years to at most a few thousand years for Bw horizons and a few thousand years to 10,000 years for Bt horizons is a reasonable estimate. Given these very approximate estimates, each sequence could represent one of the wet and dry climate cycling following Milankovitch forcing that would cause changes in sediment discharge and deliver during flood peaks. Based on the estimates of time of soil formation of the fossil soils, Milankovitch climatic cycles spanning 19, 23 or possibly 41 ka could be responsible for these sequences. 5.6. Organic carbon isotopes Carbon isotopes of fossil soil organic matter from palaeosols KU-1 and KU-3 were analysed with the aim of further defining the palaeoenvironmental conditions. d13 C values in KU-1 became more 13 C enriched from the A to the sandy A/B horizon (0–10 cm depth) and then reached the most 13 C depleted value ()25.1&) in the Bt horizon at 35 cm depth (Fig. 5). The overlying sediment of KU-1 was characterised by similar d13 C values as the A-horizon.

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

In profile KU-3 (Fig. 6) the A-horizon had been eroded and only the clayey Bt and B/C horizons were preserved, which is evident in the relatively consistent d13 C values down profile. Values in the Bt horizon of KU-3 vary from )22.2& to )23.2&, but become more 13 C depleted in the B/C horizon ()25.0&). The overlying sediment of KU-3 was characterised by a much more 13 C-depleted value ()27.6&) compared with the palaeosol organic matter. R2 values for KU-1 (0.1601) and KU-3 (0.0926) indicate no correlation of total organic carbon (TOC) content and isotopic composition, implying that changes in d13 C values were not a function of TOC, and therefore probably not a result of diagenetic processes. To infer palaeoenvironmental conditions from d13 C values of palaeosol organic matter, it is important to differentiate between isotopic values that are likely to represent the source vegetation and isotopic changes that occur subsequently during soil development. A number of authors have established that d13 C values tend to become more 13 C-enriched with age and depth in the soil profile (e.g. Nadelhoffer and Fry, 1988; BeckerHeidmann and Scharpenseel, 1992; Balesdent and Mariotti, 1996; Ehleringer et al., 2000). This 13 Cenrichment is most likely due to decompositional processes and the addition of 13 C-enriched microbial biomass. The degree of 13 C-enrichment in temperate and semi-arid soils is usually between 1 and 4& (Stout et al., 1981; Nadelhoffer and Fry, 1988; Balesdent et al., 1993; Balesdent and Mariotti, 1996); however, 13 C-enrichment can be as high as 6& in rainforest soils (Garten et al., 2000; Krull et al., 2002). Therefore, d13 C values from the A-horizon or, alternatively, from sediment overlying palaeosols (as it is likely to contain plant material), are more likely to reflect the d13 C values of the vegetation. By comparison, d13 C values from B-horizons could provide information with respect to soil organic matter decompositional changes and should be used with caution for palaeoenvironmental reconstruction. Following this line of reasoning, the d13 C values of the A-horizon of KU-1 as well as organic matter from the overlying sediment of KU-1 and KU-3 can be used as a proxy for d13 C values of the existing vegetation. These data are comparable with d13 C values from modern humid, temperate or sub-tropical environments and do not indicate evidence for water-stress (Balesdent et al., 1993; Guehl et al., 1998; West et al., 2001).

6. Rusinga Island results and discussion

73

lands overlooking Lake Victoria (Fig. 1), situated approximately 200 m apart and are overlain by grey conglomeratic beds. The conglomeratic beds are interpreted as channel deposits of a fluvial system deposited from the initial doming stage of Kisingiri and are therefore stratigraphically equivalent to the base of the Rusinga Island section. These units are interpreted to be analogous to the Karungu site stratigraphy in that the deposition of both are presumed to have resulted from erosion of the Pre-Cambrian basement due to pre-volcanic doming (McCall, 1958; Bestland et al., 1995). 6.2. South Gumba The South Gumba palaeosol profile is approximately 160 cm thick and contains three distinct dark reddish brown (2.5YR 3/4) to dark red (2.5YR 3/6) clay horizons (Fig. 7). At depths of 40 and 80 cm are two red (2.5YR 4/6) sandy calcareous layers which separate the clay units. They have abundant iron nodules, mica and have high CaO content (15–25 wt%), hence they are interpreted as Ck horizons. There is well-developed clay structure throughout the profile indicated by features such as clay slickensides, mangans and an obvious, welldefined ped structure. An increase in clay content below the second C-horizon (80 cm) is indicated by the presence of an increasingly distinct ped structure. The top two clay horizons that are overlying the C-horizons are defined as Bk due to their high CaCO3 content. The better-developed clay horizon situated lower in the profile has lower CaCO3 content and as such is defined as a Bt horizon. Increase of both SiO2 and Al2 O3 down the profile correlate well with clay contents in the Bt and Bk horizons. An exception to the well-developed clayey horizon lower in the profile is a small laminated horizon at approximate 90 cm which displays no ped structure at all. It appears to have escaped any major pedogenic alteration since its deposition at the base of the coarser gram material. The thin, laminated horizon is interpreted as a boundary between underlying over bank flood plain deposits and coarser overlying channel deposits. The South Gumba palaeosol with base cations percentages of greater than 10% is interpreted as an Argillisol, (palaeo-Alfisol), suggesting the possibility of sub-humid to humid conditions existing during its formation. However, as with the palaeosols investigated at Karungu, the North Gumba profile exhibits Ck horizons, suggests that this profile experienced possibly slightly drier conditions than what is optimum for palaeo-Alfisol development.

6.1. Gumba palaeosol profiles 6.3. North Gumba Located at the southwestern end of Rusinga Island are the two palaeosol profiles investigated (Gumba North and Gumba South). Both are exposed in bad-

The North Gumba profile (Fig. 8) is approximately 320 cm thick and possesses similar soil characteristics to

74

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

Fig. 7. Geochemical profile of atomic abundances, soil horizon interpretations and carbon isotopic data for the Rusinga Island palaeosol South Gumba (symbols as in legend in Fig. 4).

Fig. 8. Geochemical profile of atomic abundances, soil horizon interpretations and carbon isotopic data for the Rusinga Island palaeosol North Gumba (symbols as in legend in Fig. 4).

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

the South Gumba profile. An obvious colour change is present between the greyish brown conglomerate (10YR 5/2) and the strong brown (7.5YR 5/4) palaeosol. However, the boundary is much more gradational in comparison to the sharper boundary seen at South Gumba. Variation of grain size within this profile is much less than at South Gumba. Two coarser horizons are observed at 20 and 120 cm, respectively, however the North Gumba coarse-grained units are predominantly silt in comparison to the higher sand component seen in South Gumba. Again there is mica in the coarser-grained horizons. Pedogenic carbonate horizons are obvious in the top of the profile with the best-developed horizon occurring in the second coarse horizon (120 cm). Carbonate nodules are present below this horizon, although their concentration is minimal. Thus the coarse horizon at 20 cm is defined as a Bk horizon due to its carbonate content, while the second coarse unit is a Ck horizon. Of all the profiles investigated, the North Gumba shows the best-developed chemical depth trends for

75

clayey horizons, Clay slickensides and mangans as well as other ped structures are present in the clayey horizons located at 60–80 and 200–240 cm. The degree of development of these clay properties increases slightly at the base of the profile, with clay development throughout the rest of the clay horizons being quite uniform. Increases in SiO2 and Al2 O3 concentrations support field interpretations of these clay layers as Bt horizons. The depletion of base cation content in the Bt horizons in comparison to the C-horizons indicates a strong weathering signature. A significant decrease in MgO content down profile could be an indication of a decrease in mafic minerals, which correlates with the observation of coarser horizons within the top half of the profile. Coarse-grained detritus contain minerals such as hornblende and pyroxene, which are common in these deposits. Total base cations (Naþ , Ca2þ , Kþ and Mg2þ ) is greater than ten percent down the profile, inferring a palaeo-Alfisol classification. The significant presence of calcareous material in the profile suggests

Fig. 9. Mass balance weathering trends for collective Karungu and Gumba beds of the Wayondo Formation and the Kiahera palaeosols of the Rusinga Group. Evident is a 90% loss and collapse of SiO2 , Na2 O and K2 O for the Gumba and Karungu palaeosols compared to a granitic parent material, representing multi-stage weathering in an alluvial system. In comparison, the Kiahera Formation displays dilation of up to 90% based on a nephelinite parent material. This infers an input of pyroclastic material and in situ weathering combined with minimal alluvial reworking. Bulk rock chemical data for Kisingiri’s granitic material from unpublished data (M. LeBas).

76

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

Fig. 10. Comparison of d13 C values from A horizons and sedimentary deposits from the pre-volcanic Wayondo Formation (closed circles) and from the syn-eruptively formed Kiahera and Hiwegi palaeosols (open circles) with average isotopic distribution among C3 and C4 plants (modified from Cerling and Quade, 1993). The y-axis represents the number of species. The d13 C values of the two different palaeosol groups shows that the Wayondo palaeosols fall into the d13 C range typical for average C3 plants, whereas the Hiwegi and Kiahera palaeosols fall into the higher end, usually represented by waterstressed,vegetation. It is not believed that these values represent a mix of C3 and C4 species, since the occurrence of C4 plants has not been verified for the early to mid Miocene.

that the environment that this palaeosol formed in was at least sub-humid to humid. Carbon isotopic values from the A horizon of the Gumba profile (reported in a previous paper by Bestland and Krull, 1999) are similar to those of the laterally equivalent Karungu palaeosols ()25.9&) and are therefore consistent with the interpretation of a generally humid environment during formation of the pre-volcanic Wayondo Formation (Fig. 10). By comparison, d13 C values of the syn-eruptively formed palaeosols of the Hiwegi and Kiahera formations show much more 13 C-ennched values (as high as )23.0&), are characterised by less organic matter (Table 2), and greater abundance of carbonate compared to the Wayondo palaeosols. Such 13 C-enriched values could be the result of input from grasses, using the C4 photosynthetic pathway, which have average d13 C values of )13& compared with an average value of )27& for C3 plants (Dienes, 1980; Fig. 10). However, there is currently no reliable evidence for the presence of C4 grasses in East Africa during the early to mid Miocene (Cerling et al., 1991). Therefore, we interpret the relatively 13 C-enriched values of the syn-eruptively formed palaeosols as an indicator for water-stress under a more arid climate. 6.4. Kulu Lake Beds The Kulu Formation which overlies the Rusinga Agglomerate and interfingers with the Hiwegi Formation (Fig. 2) is lacustrine, containing fish in varved

shales, and was deposited during a period of relative volcanic hiatus (Shackleton, 1951; Bestland, 1991). This formation is part of a syns-depositionally-faulted succession resulting from volcano-tectonic activity of the Kisingiri volcano and Nyanza Rift (Bestland, 1991). Faultscarps are indicated by the stepped, drop off appearance of the top surface of the Rusinga Agglomerate and are perpendicular to the slope of Kisingiri volcano. These faultscarps are situated on either side of a 5 km long and 2–3 km wide LugongoLunene Horst (Bestland, 1991). The result of this faulting event was damming and the formation of local lakes. Organic matter buried in the Kulu Formation sediments were subjected to carbon and nitrogen isotopic analyses with the aim of producing a record of the general type of organic matter delivered and accumulated in these small temporary alkaline lakes.

7. Carbon isotopes There is no apparent correlation ðR2 ¼ 0:0006Þ between the TOC content and d13 C values, indicating that carbon content did not determine the d13 C values. Lake organic matter is mostly a mixture of terrestrial-derived organic matter (soil and litter) and organic material produced in situ by lacustrine algae (Meyers and Ishiwatari, 1993; Meyers and Lallier-verges, 1999). In the absence of C4 plants, lake-derived organic matter produced by phytoplankton is, usually isotopically indistinguishable from organic matter produced by C3 plants in the surrounding watershed and varies between )25& and )30& (Meyers, 1997; Meyers and Lallier-verges, 1999). However, d13 C values of algal organic matter can become as 13 C-enriched as )9& in alkaline lakes and microbial re-working of organic matter can result in up to 2& more 13 C-depleted values (Meyers and Lallierverges, 1999). Thus, the d13 C isotopic values (average )25.4&; Table 2) from lacustrine sediments of the Kulu Lake Beds cannot be used as a similar proxy as palaeosol organic matter to infer palaeoenvironmental conditions. However, the average C/N value of 4.4 is supportive of a predominantly algal source and comparatively very little input from litter or soil material (Meyers and Lallier-verges, 1999). Given that the d13 C values fall into the more 13 C-enriched end of the isotopic values typical for lacustrine algae (generally the decay of carbon in lacustrine sediments is preferential to 13 C (Meyers and Ishiwatari, 1993)), suggests that the increased aridity, as indicated by palaeosol d13 C values, was not severe enough to cause a significant increase in salinity and decrease in productivity, which would result in depleted d13 C values in lacustrine algae (Meyers and Lallier-verges, 1999).

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

8. Mass balance geochemistry Mass balance geochemistry (Fig. 9) was undertaken on all sedimentary units discussed in this paper except the Kulu Formation lake beds. For easier evaluation and discussion of weathering trends calculated using Eqs. (1) and (2), bulk rock chemistry from all three Karungu palaeosols and the two Gumba palaeosols were grouped. Both the Gumba and Karungu palaeosols have similar and much more pronounced weathering signatures than that of the overlying Kiahera Formation (values from Bestland and Retallack, 1993). The palaeosols from the basal part of the section exhibit collapse and loss of silica, sodium, and potassium of approximately 90%, indicating strong weathering when compared to the granitic composition of parent material. Iron displays additions of up to 50% in an overall collapsed system, again supporting a history of strong weathering. The Kiahera Formation palaeosol horizons display dilation values up to 90% and base cation trends display losses (weathering) of between 60 and 90%. Silica spans both the loss and addition field with higher degrees of dilation corresponding to higher levels of addition. Iron displays trends similar to silica, suggesting that the Kiahera Formation palaeosols are only moderately to weakly weathered compared to their nephelintic parent. More importantly, the strong dilatational trends are in agreement with the active pyroclastic setting that existed during deposition of the Kiahera Formation. Both the Gumba and Karungu palaeosols existed in a more humid climate than what was experienced by the Kiahera Formation. However, the fact that the Kiahera Formation has experienced various amounts of dilation for base cations infers that this unit experienced inputs of additional material during soil formation. This suggests that the difference in weathering signatures between the Kiahera and the Wayondo Formations is not only a reflection of climate but also a change in the depositional regime that coincided with the onset of Kisingiri’s volcanic activity.

9. Conclusions Investigations of the early Miocene Wayondo Formation palaeosols located on Rusinga Island (Gumba Beds) and at Karungu provide further insight into the environmental conditions around Kisingiri volcano approximately 17.8 million years ago. Both are alluvial in nature, products of the initial doming stage of the Kisingiri volcano, with these deposits being a precursor to the fossil-rich deposits of the Kiahera, Rusinga, Kulu and Hiwegi Formations, all of which formed during later stages of Kisingiri’s volcanic activity. The Karungu and Gumba profiles were identified as palaeo-Alfisols

77

using a combination of field techniques, XRF bulk rock geochemistry and mass balance calculations. These palaeo-Argillisols possess well-developed clay horizons, moderate base cation contents and pronounced weathering signatures, all characteristics of this type of palaeosol. Both sets of palaeosols display collapse and loss of silica, sodium and potassium of 80–90%. These weathering trends and the above palaeo-Alfisol classification suggest the existence of a humid to sub-humid environment on both the northern and southern sides of the Kisingiri Volcano prior to the commencement of volcanic activity. Furthermore, the close correlation in weathering signatures seen for the palaeosols, suggests a similar environment existed on either side of Kisingiri during this period. Mass balance calculations indicate clearly that the Karungu and Gumba palaeo-Alfisols are more weathered and more developed than the palaeoInceptisols identified by Bestland and Retallack (1993) in the Kiahera Formation (dilation 0–100% addition and losses of 0–50%). This decrease in weathering could be interpreted as a shift from wetter warmer conditions of the Wayondo Formation (Gumba and Karungu palaeosols) to drier conditions associated with the deposition of the Kiahera and Hiwegi Formations. However, the fact that the Kiahera Formation experienced large inputs of material, as indicated by dilation signatures, suggests that at least part of the difference in weathering can be attributed to the commencement of volcanic activity and increased pyroclastic inputs to the depositional environment. Evidence from stable carbon isotopes suggest that a change in climate indeed took place as shown by the more 13 C-enriched values in the syn-eruptively formed palaeosols, suggesting a more water-stressed environment. The combination of mass balance calculations and d13 C data from palaeosol surface horizon and sedimentary organic matter shows that conditions changed from a stable landscape of a humid to sub-humid environment to a landscape of decreased stability and greater aridity. However, given that the lacustrine d13 C values do not support long-term aridity, it is likely that changes in precipitation were relatively small and were probably seasonal, resulting in different effects in different parts of the landscape. Especially in a relatively destabilised system, relatively minor precipitation changes would be expected to have a greater impact on soils than on lakes.

Acknowledgements This work was funded by a grant from the National Geographic Society (grant number 6644-99) to Erick A Bestland and Evelyn S Krull. We thank J. Stanley at Adelaide University for analytical support. A special mention must go to the Okomo Ocheng. family in

78

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79

particular Kedrick and Joel for their assistance while undertaking field work. The field research was done with the permission of the Government of Kenya, under the auspices of the Department of Palaeontology, National Museums of Kenya, and its head, M. Leakey. References Andrews, P., 1992a. Community evolution in forest habitats. Journal of Human Evolution 22, 423–438. Andrews, P., 1992b. Evolution and environment in the Hominoidea. Nature 359, 641–646. Baker, B.H., 1987. Outline of the petrology of the Kenya Rift alkaline province. In: Fitton, J.G., Upton, B.G.J. (Eds.), Alkaline Igneous Rocks: Geological Society of London Special Publication 3, pp. 293–311. Balesdent, J., Girardin, C., Mariotti, A., 1993. Site-related 13 C of tree leaves and soil organic matter in a temperate forget. Ecology 74, 1713–1721. Balesdent, J., Mariotti, A., 1996. Measurement of soil organic matter turnover using 13 C natural abundance. In: Boutton, T.W., Yamaski, S. (Eds.), Mass Spectrometry of Soils. Marcel-Dekker, New York, pp. 83–111. Becker-Heidmann, P., Scharpenseel, H.W., 1992. Studies of soil organic matter dynamics using natural carbon isotopes. Sciences of the Total Environment 117/118, 305–312. Bestland, E.A., 1991. A Miocene Gilbert-type fan-delta from a volcanically influenced lacustrine basin, Rusinga Island, Lake Victoria, Kenya. Journal of the Geological Society, London 148, 1067–1078. Bestland, E.A., Krull, E.S., 1999. Palaeoenvironments of early Miocene Kisingiri volcano Proconsul sites: evidence from carbon isotopes, palaeosols and hydromagmatic deposits. Journal of the Geological Society, London 156, 965–976. Bestland, E.A., Retallack, G.J., 1993. Volcanically influenced calcareous palaeosols from the Miocene Kiahera Formation, Rusinga Island, Kenya. Journal of the Geological Society, London 150, 293–310. Bestland, E.A., Thackray, G.D., Retallack, G.J., 1995. Cycles of doming and eruption of the Miocene Kisingiri Volcano, Southwest Kenya. Journal of Geology 103, 598–607. Bestland, E.A., Retallack, G.J., Rice, A.E., Mindszenty, A., 1996. Late Eocene detrital laterites in central Oregon: mass balance geochemistry, depositional setting, and landscape evolution. Geological Society of America Bulletin 108 (3), 285–302. Birkeland, P.W., Machette, M.N., Haller, K.M., 1991. Soils as a tool for applied Quaternary geology: Utah Geological and Mineral Survey Miscellaneous Publication 91-3, p. 67. Bown, T.M., Kraus, M.J., 1987. Integration of channel and floodplain suites in aggrading fluvial systems, 1. Developmental sequence and lateral relations of lower Eocene alluvial palaeosols, Willwood Formation, Bighorn Basin, Wyoming. Journal of Sedimentary Petrology 57, 587–601. Brimhall, G.H., Dietrich, W.E., 1987. Constitutive mass balance relations between chemical composition, volume, density, porosity, and strain in metasomatic hydrochemical systems: results on weathering and pedogenesis. Geochimica et Cosmochimica Acta 51, 567–587. Brimhall, G.H., Chadwick, O.A., Lewis, C.J., Compston, W., Williams, I.S., Danti, K.J., Dietrich, W.E., Power, M.E., Hendricks, D., Bratt, J., 1988. Deformational mass transport and invasive processes in soil evolution. Science 255, 696–702. Cerling, T.E., Quade, J., Wang, Y., Bowman, J.R., 1989. Carbon isotopes in soils and palaeosols as ecology and palaeoecology indicators. Nature 341, 138–139.

Cerling, T.E., Quade, J., Ambrose, S.H., Sikes, N.E., 1991. Fossil soils, grasses, and carbon isotopes from Fort Ternan, Kenya: grassland or woodland? Journal of Human Evolution 21, 295–306. Chadwick, O.A., Brimhall, G.H., Hendricks, D.M., 1990. From a black box to a gray box––a mass balance interpretation of pedogenesis. Geomorphology 3, 369–390. Coleman, J.M., 1969. Brahmaputra river: channel processes and sedimentation. Sedimentary Geology 3, 129–239. Dienes, P., 1980. The isotopic composition of reduced organic carbon. In: Fritz, P., Fontes, J.Ch. (Ed.), Handbook of Environmental Geochemistry 1: The Terrestrial Environment, pp. 329–406. Drake, R.E., VanCouvering, J.A., Pickford, M.H., Curtis, G.H., 1988. New chronology for the early Miocene mammalian faunas of Kisingiri, Western Kenya. Journal of the Geological SocietyLondon 145, 51–68. Driese, S.G., Mora, C.I., Cotter, E., Foreman, J.L., 1992. Palaeopedology and stable isotope chemistry of Late Silurian vertic palaeosols, Bloomsburg Formation, Central Pennsylvania. Journal of Sedimentary Petrology 62, 825–841. Ehleringer, J.R., Buchmann, N., Flanagan, L.B., 2000. Carbon isotope ratios in below ground carbon carbon cycle processes. Ecological Application 10, 412–422. Farrell, K.M., 1987. Sedimentology and facies architecture of overbank deposits of the Mississippi River, False River Region, Louisiana. In: Etridge, F.G., Flores, R.M., Harvey, M.D. (Ed). Recent developments in fluvial sedimentology. Soc. Econ Paleontol. Mineral Spec Pub. 39. pp. 111–120. Garten, C.T., Cooper, L.W., Post, W.M., Hanson, P.J., 2000. Climate controls on forest soil C isotope ratios in the Southern Appalachian Mountains. Ecology 81, 1108–1119. Guehl, J.M., Domenach, A.M., Bereau, M., Barigah, T.S., Casabianca, H., Ferhi, A., Garbaye, J., 1998. Functional diversity in an Amazonian rainforest of French Guyana: a dual isotope approach (delta 15N and delta13C). Oecologia 116, 316–330. Hall, G.F., 1983. Pedology and geomorphology. In: Wilding, L.P., Smeck, N.E., Hall, G.F. (Eds.), Pedogenesis and Soil Taxonomy. I Concepts and Interactions. Elsevier publishers, Amsterdam Netherlands, pp. 117–140. Krull, E.S., Bestland, E.B., Gates, W.P., 2002. Soil organic matter decomposition and turnover in a tropical Ultisol: evidence from d13 C, d15 N and geochemistry. Radiocarbon 44, 93–112. LeBas, M.J., 1977. Carbonatite––Nephelinite Volcanism. Wiley, New York, USA. 347. LeBas, M.J., 1987. Nephelinites and carbonatites. In: Fitton, J.G., Upton, B.G.J. (Eds.), Alkaline Igneous Rocks: Geological Society of London Special Publication 30, pp. 53–83. Leakey, L.S.B., 1967. An early Miocene member of Hominidae. Nature 213, 155–163. Mack, G., James, W.C., Monger, H.C., 1993. Classification of palaeosols. Geological Society of America Bulletin 105, 129–139. McCall, G.J.H., 1958. Geology of the Gwasi Area. Geological Survey of Kenya Report 45. Meyers, P.A., 1997. Organic geochemical proxies of palaeoceanographic, palaeolimnologic, and palaeoclimatic processes. Organic Geochemistry 27 (5–6), 213–250. Meyers, P.A., Ishiwatari, R., 1993. Lacustrine organic geochemistry–– an overview of indicators of organic matter sources and diagenesis in lake sediments. Organic Geochemistry 20, 867–900. Meyers, P.A., Lallier-verges, E., 1999. Lacustrine sedimentary organic matter records of late Quaternary palaeoclimates. Journal of Palaeolimnology 21 (3), 345–372. Nadelhoffer, K.J., Fry, B., 1988. Controls on natural nitrogen-15 and carbon-13 abundances in forest soil organic matter. Soil Science Society of America Journal 52, 1633–1640. Nettleton, W.D., Olson, C.G., Wysocki, D.A., 2000. Palaeosol classification: problems and solutions. Catena 41, 61–92.

M.S. Forbes et al. / Journal of African Earth Sciences 39 (2004) 63–79 Norrish, K., Chappell, B.W., 1977. X-ray fluorescence spectrometry. In: Zussman, J. (Ed.), Physical Methods in Determinative Mineralogy, second ed. Academic Press, New York, pp. 201– 272. Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Annual Review of Ecological Systems 18, 293–320. Pickford, M.H., 1982. The tectonics, volcanics and sediments of the Nyanza Rift Valley, Kenya: supplement band. Zeitschrift Fur Geomorphology 42, 1–33. Pickford, M.H., 1986. Sedimentation and fossil preservation in the Nyanza Rift system, Kenya. In: Frostick, L.E., et al. (Ed.), Sedimentation in the African Rifts. Geological Society, London, Special Publications 25, pp. 345–362. Quade, J., Crater, J.M.I, Ojha, T.P., Adam, J., Harrison, T.M., 1995. Late Miocene environmental change in Nepal and the northern Indian sub-continent: Stable isotopic evidence from palaeosols. Geological Society of America Bulletin 107, 1381–1387. Retallack, G.J., 1988. Field recognition of palaeosols. Geological Society of America. Special Paper 216, pp. 1–20. Retallack, G.J., 1990. Soils of the Past: An Introduction to Palaeopedology. Unwm-Hyman, London, UK.

79

Retallack, G.J., German-Heins, J., 1994. Evidence from palaeosols for the geological antiquity of rain forest. Science 265, 499–502. Retallack, G.J., Dugas, D.P., Bestland, E.A., 1990. Fossil soils and grasses of a middle Miocene East African grassland. Science 247, 464–475. Shackleton, R.M., 1951. A contribution to the geology of the Kavirondo Rift Valley. Quarterly Journal of the Geological Society, London 106, 345–392. Shipman, P., Walker, J.A., Van Couvenng, P.J., Hooker, J.A., Miller, J., 1981. The Fort Ternan hominoid site, Kenya: geology, age, taphonomy and paleoecology. Journal of Human Evolution 10, 49–72. Stout, J.D., Goh, K.M., Rafter, T.A., 1981. Chemistry and turnover of naturally occurring resistant organic compounds in soil. In: Paul, E.A., Ladd, J.N. (Eds.), Soil Biochemistry. Marcel-Dekker, New York, pp. 19–24. Van Couvering, J.A., Miller, J.A., 1969. Miocene stratigraphy and age determinations, Rusinga Island, Kenya. Nature 221, 628– 632. West, A.G., Midgley, J.J., Bond, W.J., 2001. The evaluation of d13 C isotopes of trees to determine past regeneration environments. Forest Ecology and Management 147, 139–149.