Hydrobiologia 488: 27–41, 2002.D.M. Harper, R. Boar, M. Everard & P. Hickley (eds), Lake Naivasha, Kenya.© 2002 Kluwer Academic Publishers. Printed in the Netherlands.

27

Geochemical and physical characteristics of river and lakesediments at Naivasha, Kenya

Håkan Tarras-Wahlberg1,∗, Mark Everard2 & David M. Harper3

1Swedish Geological AB, Box 19090, 104 32 Stockholm, Sweden2The Natural Step UK, 9 Imperial Square, Cheltenham, Gloucestershire GL50 1QB, U.K.3Department of Biology, Leicester University, Leicester LE1 7RH, U.K.∗Author for correspondence. E-mail: hakan.tarras-wahlberg@hifab.se

Key words: Lake Naivasha, sediment stratigraphy, sediment geochemistry, metals

Abstract

Prior studies on Lake Naivasha relevant to understanding sediment dynamics include a bathymetric map, a pa-leolimnological study of fossil invertebrate assemblages in lake sediment, an overview of lake level fluctuationsthroughout the 20th century, and identification of a dynamic assemblage of macrophyte zones that has respondedboth to these changes in lake level and to more recent, alien species. Sediment samples collected from the riverssystems and the lake were examined physically and chemically. River sediment characteristics reflect geologyand geomorphological processes in the catchment, whereas lake sediment stratigraphy has responded to pastlake level changes. Such changes have caused significant changes in aquatic vegetation assemblages. Present daysediment dynamics in the lake are governed by the presence of river point sources in the north and wave-inducedre-suspension, such that sediments introduced by rivers are transported in easterly and southerly directions, andare eventually deposited in the eastern, central and southern parts of the lake. Sedimentary deposition is alsooccurring in northern areas that once were protected by papyrus swamp vegetation but now only have a narrowfringe, highlighting the important role of swamp vegetation in filtering out suspended particulates and therebycontrolling water quality in the lake. Geochemical analyses of river and lake sediments indicate that they representfairly undisturbed background conditions. Higher-than-expected concentrations of cadmium, iron, nickel and zincfound in both river and lake sediment are likely to derive from volcanic rocks and/or lateritic soils found in the lakecatchment.

Introduction

Lakes are strongly influenced by the natural and hu-man characteristics of their catchments. River sedi-ment in low order channels derive from a plethora ofhillslope and channel sources in the upstream catch-ment (Bowie & Mutchler, 1986). During downstreamtransport, sediment from different sources will mixand, therefore, investigations of the geochemical andphysical characteristics of river sediment may provideevidence of processes occurring in the entire catch-ment. As a consequence, stream sediment samplinghas long been used as a tool to prospect for metalli-ferous mineral deposits, and it is generally accepted

that such surveys provide more representative and costeffective results than surveys of soils or vegetation(Levinson, 1980).

Geochemical surveys of river sediment are increas-ingly being used also in environmental investigations,and applications include:

(i) assessments of environmental impacts, as sedi-ments’ contaminant concentrations downstreamof industrial, agricultural or other human activ-ities often are elevated (Darnley et al., 1995);

(ii) human and animal health studies as geologicallycontrolled deficiencies/excesses of certain nutri-ents/toxins in water and/or food products may

28

cause health problems (Fordyce et al., 1996); and(iii) to provide a geochemical baseline against

which any future pertubations may be appraised(Williams et al., 2000).

Similarly, investigations of lake sediments can alsobe used in environmental surveys. When rivers enter alake, the decrease of velocity results in sediment de-position. Investigations of geochemical changes withsediment depth may, thus, provide indications ofchanges in the characteristics of sediment input withtime. Such investigations may therefore be used todetermine both the character of contamination and, ifsedimentation rates are known, changes in contamin-ant loading with time (Nicholson, 1996).

Environmental applications of geochemistry of-ten investigate metal contamination, and it is gen-erally acknowledged that metals in their dissolvedionic state control direct toxic effects (Förstner, 1983;1988; Stumm & Morgan, 1996). However, metals inaquatic systems are usually predominantly associatedwith sediment (Förstner, 1983; Windom et al., 1991).Therefore, the concentration of metals in sediment isgenerally a more sensitive and accurate indicator ofcontamination of an aquatic system, compared to theconcentration of dissolved metals (Carelli et al., 1995).It is also recognised that sediment-bound metals havethe potential of changing to the ionic state (Willfordet al., 1987). Investigations of sediment geochemistryis, thus, directly relevant to assessing the possibilityof toxic effects of metal contamination. Furthermore,as metals are present in much higher concentrationsin sediment than in water, the analysis of sedimentis comparatively easy to perform and, therefore, lesscostly, enabling more samples to be analysed, and agreater geographical coverage.

The interpretation of sediment geochemistry datais, however, not without its problems. In rivers, dif-ferent morphological units often have quite differentgeochemical characteristics, resulting from physicalsorting processes within the stream bed (Ladd et al.,1998). Similarily, metals are often found preferentiallyconcentrated in the finer fraction of sediment (Först-ner, 1983). Consequently, care needs to be taken toensure that the same or at least similar morphologicalunits and size fractions are sampled in order for resultsfrom different rivers or even reaches of the same riverto be comparable (Ladd et al., 1998). Further, whereasenvironmental criteria for water have been establishedfor a varity of purposes, including the protection ofaquatic life, attempts to elaborate similar criteria for

sediment have been less succesful. It is difficult to de-termine to what extent potentially harmful elementsin sediment are, or may become, available for bio-logical uptake. Consequently, environmental criteriafor sediment are generally seen to be of limited valuedue to variations through time and space in the con-ditions that determine bioavailability (Darnley et al.,1995; Maher et al., 1999). Non-statutory systems forsediment quality guidelines have, nevertheless, beendeveloped in a number of countries, including Canadaand Sweden.

Studies of the input of nutrients and other materialfrom the Naivasha catchment to lake have been un-dertaken by Gaudet & Melack (1981), Harper et al.(1993), Harper et al. (1995), Everard et al. (2002), andKitaka et al. (2002). None have yet investigated lakeand river sediments. The aims of the study were: (i) toinvestigate the geochemical and physical characterist-ics of river and lake sediments, (ii) to investigate thesediment stratigraphy of the lake, and (iii) to assess theimplications of (i) and (ii) in informing of how to bestmanage Lake Naivasha and its catchment.

Study site description

Water level and vegetation changes

The water level and aquatic vegetation of Lake Na-ivasha changes continuously (Becht & Harper, 2002),and such fluctuations will have significant effects uponsediment characteristics. The lake was first describedin the writing of European explorers in the 1880s butits earliest record is that according to local tradition,there was a period immediately before the arrival ofthe first Europeans when the lake was almost com-pletely dry (Ase et al., 1986). Towards the end of the19th century, writings by European travellers indicatethat the lake’s level was continuously rising until 1895,when a maximum lake level of 1896 m above sea levelwas reached. This historic high water level is some 4–5m higher than any level reached since that time. Levelsthen declined around the turn of the century, so thatwhen continuous recordings commenced in 1908 thelevel was 1890 m (Ase, 1987). Thereafter, the recordsshow a progressive decline in lake levels during the1920s and 1930s, culminating in a historic (recorded)low of 1882 m in 1946. The lake level remained lowuntil about 1956, whereafter it rose to an intermediatelevel of 1887 m in 1980 (Ase et al., 1986). Lake leveldeclined progressively during the following decade,

29

rising rapidly by 1 m in May 1988, declining for a dec-ade then rising by 3 m in 1997. Both these latter eventswere following heavy rains associated with the ENSOclimatic phenomenon (see Becht & Harper (2002) fordetails of lake level changes).

The vegetation of Lake Naivasha is, in turn, influ-enced by changing lake levels. Beadle (1932) providedthe first scientific description of the lake’s vegetation.At that time, during a period of declining lake levels,papyrus (Cyperus papyrus) formed an extensive float-ing swamp in the whole northern part of the lake,north of an east–west horizontal line along the north-ern edge of Crescent Island (known colloquially as theNorth Swamp). Elsewhere varying widths of it fringedthe lake and formed a multitude of small, mobile is-lands. Various submerged macrophytes and the bluewater lily (Nymphaea nouchalii var. caerulea) grew inshallow lagoons inside these papyrus islands. To thelakeward side, as water deepened, the floating-leavedfollowed by dense beds of submerged macrophytesformed clear zonation. This pattern was consistentthroughout the water level changes up to the mid-1970s. Thereafter, it underwent dramatic and continu-ous changes. The driving forces behind these changeshave been various anthropogenic influences, includingthe introduction of alien species of animals and plants(Harper, 1984; Tarras-Wahlberg, 1986; Gouder et al.,1998), and the clearance of lakeside lands for intensiveagriculture, (Harper et al., 1990).

Lake bathymetry

Ase et al. (1986) provide a bathymetric map of thelake referring to a lake-surface of 1889 masl (repro-duced as Fig. 1). The map shows that the lake maybe divided into three semi-separate basins: the mainlake basin, the Crescent island crater in the east andLake Oloidien in the south. The main lake is charac-terized by a smooth flat relief with a maximum depthjust off Hippo Point in the southwest. The flat bottomtopography contrasts sharply with the hilly topographyof the surrounding land, indicating that the lake basinis filled with large amounts of sediments (Ase et al.,1986). A shallower area in the northern end of thelake is interpreted as a deltaic sediment accumulationderiving from the inflows of the Gilgil, Karati andMalewa rivers. At the time of the survey by Ase et al.(1986), the deltaic feature coincided closely with thepapyrus-dominated North Swamp. A feedback pro-cess is, thus, indicated in which the shallow delta areaprovides a preferred habitat for papyrus whilst at the

same time the papyrus swamp functions as a sedimenttrap and protects the delta from wave erosion and, inturn, promotes delta development.

Lake sedimentology

Verschuren (1996) conducted stratigraphic and geo-chronological analyses of three sediment cores takenfrom the main lake. He noted that sedimentationpatterns in the lake vary due to the effects of wind-induced re-suspension and re-deposition of sediments,and the presence of river point sources of sediment inthe north. As a result, two of the cores sampled did notrepresent longer periods of undisturbed sedimentation.In the central part of the lake, however, he sampledwhat was interpreted as a complete and uninterrup-ted sequence of sediments deposited during the 20thcentury, non-conformably overlying sediments datingback as far as the 16th–17th centuries. In this core, heidentified four main units: Unit 0, Unit I, Unit II andUnit III.

• Unit 0, found at the base of the core at 114–105 cm depths, was a peat horizon with abundantfragments of Cyperus papyrus and Ceratophyllum.Verschuren interpreted it as representing a periodwhen a low lake level transformed the lake intoa shallow, fragmented wetland overgrown by apapyrus swamp. A radiocarbon age of 290 ± 50years of sediments directly overlying this horizonindicates that Unit 0 was laid down, not during themid 19th century lowstand but during an earlierevent in the 16th–17th century.

• The clayey muds of Unit I, between 105 and82 cm, was interpreted as being composed of sed-iments laid down during low to intermediate waterlevels between the 16th–17th century and the endof the 19th century.

• Unit II sediments, between 82 and 44 cm, wereinterpreted as being laid down during high lakelevels in the late 19th and early 20th century. At48–58 cm there was a black layer with a significantadmixture of coarse partially decomposed papyrusmaterial. The unit’s black color was imparted bythe decomposed papyrus fragments, and the unitwas interpreted as representing sedimentation inshallow areas near papyrus stands, which grew onexposed mudflats, created as lake levels decreasedin the 1920s and 1930s.

• Lake levels declined in the 1920s, 1930s and 1940sand reached a lowstand of 1881.5 masl in 1946.

30

Figure 1. Bathymetric map of Lake Naivasha. Water depth contours are drawn at 2-m intervals. The lake contour represents a water level of

approximately 1887 m above sea level, (adapted from Ase et al., 1986).

The transition to Unit III sediments at 44 cm wereinterpreted as marking this return to lower lakelevels. Near the transition of Unit II and III, Ver-schuren noted the presence of a thin, light felthorizon, dated at 1941 ±7 years, composed ofnumerous needle shaped diatoms. The mechanismof formation is unclear, though Verschuren spec-ulates that its origin could be either a result ofmass sedimentation of diatoms triggered by unfa-vorable conditions during low lake levels, or thesedimentary redistribution and associated aggreg-ation of previously deposited diatoms. Above thislevel, Unit III consisted of a fairly homogenous or-ganic, clayey mud indicating that the depositionalenvironment has remained similar over the past40 years. The sedimentation rate in the main lakeduring this time was estimated at about 1 cm peryear.

Methods

Sampling was undertaken during the dry season inJuly/August 1997 and during October 2000.

Sediment samples from rivers were collected fromsixteen river sites during the dry season, between 19thJuly and 7th August 1997. Sample sites were selectedin low (first to third) order reaches to represent theoverall characteristics of river sediments in the entirelake catchment (Fig. 2). During this time, the Kar-ati river contained water in isolated pools only. Thesamples were taken with an Ekman grab from act-ive, oxygenated, mid-river sediments at sites wherethe channel is straight and the flow fairly rapid (withthe exception of the Karati samples, which were takenfrom standing pools of water and sample S10, takenfrom the Turasha dam in the Malewa catchment). Ateach site, about 1 kg of sediment was collected intwo to three grabs and these were subsequently mixed.A subsample of approximately 150 g (wet weight)

31

was retrieved from the master sample. This subsamplewas subsequently wet sieved through a 2-mm mesh(discarding the +2 mm fraction), and air dried atapproximately 40 ◦C.

The sample were then divided in two, and one sub-sample was investigated under the microscope (×10magnification) in order to determine grain size, angu-larity and sorting in accordance with tables providedin Pettijohn (1975). The other sub-sample was sentfor analysis of bulk and trace geochemistry by X–rayFluorescence spectrometry (XRF) at the Anglo Amer-ican research Laboratories in Johannesburg, SouthAfrica.

Care was taken to avoid cross sample contamina-tion and all equipment used were thoroughly cleaned,first in river water and then in de-ionised water. Theaccuracy and repeatability of the sampling procedureand the XRF analysis was checked by sampling andanalysing the sediment from one site in duplicate.

Sampling of lake sediment was performed by usinga sediment corer in 1997, and by using an Ekman grabin 2000.

Eighteen lake sites were sampled during 5–18 Au-gust 1997. Sediment cores were collected in 110-cmlong Perspex tubes, attached to a single-drive pistoncorer, which was deployed from a small boat. Thelocation of sample sites was fixed by using a hand-held Garmin GPS 12, Global Position System. Theexact sampling locations are shown in Figure 3. Thecores were extruded and described with regards totheir physical and biological character and appear-ance. Each stratigraphic unit identified was describedin detail.

Two cores were selected for geochemical analysis,one from the central lake (no. 12) and one in the North-ern swamp (no. 9), close to the outflow of the Malewariver. One sample of 50–100 g (wet weight) was col-lected from each of the stratigraphic units identifiedin the two cores, making a total of 10 samples. Carewas taken to take these samples from the centre of thecore, so as to avoid cross-sample contamination. Thesamples were kept cool in a refrigerator before beingsent for geochemical analysis at the Anglo-Americanresearch laboratories in Johannesburg, South Africa.The analysis was performed by plasma emission spec-troscopy (ICP-AES) following digestion in aqua regia.The accuracy and repeatability of the sampling andanalysis was checked by sampling and analysing onestratigraphic unit twice. Seven lake sediment sampleswere retrieved during the period 20–24 October 2000.At each site, about one kilo of sediment was collec-

ted in three grabs, which were subsequently mixed. Asubsample of approximately 150 grams (wet weight)was retrieved from the master sample. This subsamplewas air dried at approximately 40 ◦C, and thereafterkept cool in a refrigerator before being sent for ana-lysis of Loss on Ignition (LOI) and metal content atthe Leeds University School of Geography’s laborat-ories in the UK. The metal analysis was performedby plasma emission spectroscopy (ICP-AES) follow-ing digestion in aqua regia. Again, the accuracy andrepeatability of the sampling and analytical procedurewas checked by sampling and analysing one sampletwice.

The evaluation of sediment analysis was carriedout by comparison of the observed major and minorelement sediment concentrations with global aver-ages for shale (Turekian & Wedepohl, 1961; Mason& More, 1982), and Environment Canada’s provi-sional sediment quality guidelines for the protectionof aquatic life (Environment Canada, 1995). The lat-ter assume 100% bioavailablity for sediment-boundmetals which, in turn, make them possibly over pro-tective (Williams et al., 2000). Further, caution mustattend comparisons between river and lake sedimentsas the analytical procedures differ.

Results

River sediment description

Site descriptions are provided in Table 1, together witha sediment description and a linkage to RHS sitesexamined by Everard et al. (2002). Sediment char-acteristics differ markedly between the three riverssampled, reflecting the hydrological, topographic, andgeological differences in their catchments.

Sediment in the Karati is generally immature (i.e.,poorly sorted and angular in nature), reflecting the factthat the river’s gradient beneath the Kinangop Plateauis steep and that it flows only during the rainy season,providing little time for sorting and abrasion. As aconsequence of the episodic nature of the river’s flow,Karati sediment contain both fine particles (clay andsilt) and coarse material (gravel and coarse sand).

Sediment in the Gilgil system is different from thatof the Karati. Everard et al. (2002) demonstrate thatthe topography of the Gilgil system is more gentlethan the Karati, with few rapids. The flow of waterin the Gilgil system are also perennial. Consequently,the sediments found at Gilgil sampling sites are mod-

32

Tabl

e1.

Des

crip

tions

ofth

esa

mpl

ing

loca

tions

,the

phys

ical

char

acte

rist

ics

ofth

esa

mpl

edse

dim

enta

ndth

ean

alys

ispe

rfor

med

Sam

ple

Sam

plin

gL

ocat

ion

desc

ript

ion

and

land

use

Riv

erm

orph

olog

ySa

mpl

ede

scri

ptio

nX

RF

(RH

Sda

tean

alys

is

num

ber∗

)

s119

Jul9

7R

iver

pool

near

brid

geov

erth

eK

arat

i.N

ear

smal

lN

oco

ntin

uous

wat

erflo

w.R

elat

ivel

yst

eep

grad

ient

Sand

y,cl

ayey

silt

(957

6)vi

llage

(10

huts

).M

ixof

natu

ralv

eget

atio

nan

dsm

all

with

grav

elly

rive

rbed

.Cha

nnel

wid

thof

2–4

man

dsu

ba–

subr

scal

eag

ricu

lture

(mai

ze).

Pool

used

for

was

hing

dept

hof

0.3

m(i

npo

ol)

m-p

sort

ed

s219

Jul9

7Po

olat

uppe

rpa

rtof

ade

epgo

rge

near

brid

geov

erth

eN

oco

ntin

uous

wat

erflo

w,S

teep

grad

ient

with

Sand

y,cl

ayey

silt

(no

RH

S)K

arat

i.N

ear

smal

lvill

age

(30

huts

).Pa

stur

esan

dsm

all

grav

elly

rive

rbed

.Cha

nnel

wid

thof

2–3

m,a

ndde

pth

subs

–su

br

scal

efa

rmin

g(b

eans

).of

0.8

m(i

npo

ol)

m-p

sort

ed

s319

Jul9

7A

tbri

dge

over

the

uppe

rK

arat

ion

the

Nja

beni

–so

uth

No

cont

inuo

usw

ater

flow

.Ste

epgr

adie

ntw

ithG

rave

lly,c

laye

ysa

ndY

(957

5)K

inan

gop

road

.Nea

rtw

ofa

rmho

uses

.Pas

ture

s,sm

all-

grav

elly

rive

rbed

.Cha

nnel

wid

thof

2m

and

dept

hof

a

scal

efa

rmin

gof

0.4

m(i

npo

ol)

m-p

sort

ed

s419

Jul9

7U

pstr

eam

ofbr

idge

over

Kar

atio

nth

eN

aiva

sha–

Nak

uru

No

cont

inuo

usw

ater

flow

silty

sand

Y

(No

RH

S†)

high

way

.Pas

ture

san

dex

tens

ive

wat

erin

gof

cattl

eG

entle

grad

ient

and

smoo

thri

verb

ed.C

hann

elw

idth

subr

–su

ba

of3

m,a

ndde

pth

of0.

15m

(in

pool

)m

sort

ed

s520

Jul9

7D

owns

trea

mof

brid

geov

erth

eG

ilgil

onth

eno

rthe

rnSm

allr

apid

atbr

idge

,oth

erw

ise

gent

legr

adie

ntan

dsi

ltyfin

esa

nd

(957

9)L

ake

road

.Den

seac

acia

vege

tatio

n,20

–100

mw

ide,

smoo

thri

verb

ed.C

hann

elw

idth

of5

m,a

ndde

pth

ofsu

ba–

subr

past

ures

.0.

5m

w-m

sort

ed

s620

Jul9

7D

owns

trea

mof

brid

geov

erth

eM

alew

aon

the

Gen

tlegr

adie

nt.S

low

flow

ing

wat

er.C

hann

elw

idth

slig

htly

silty

sand

(No

RH

S†)

Nai

vash

a-N

akur

uhi

ghw

ay.N

ear

Nai

vash

ato

wn,

of10

–12

m,a

ndde

pth

of0.

5m

subr

–r

exte

nsiv

ese

ttlem

ents

.Was

hing

and

bath

ing

wso

rted

dow

nstr

eam

s720

Jul9

7A

tbri

dge

over

Litt

leG

ilgil

atth

ePe

mbr

oke

scho

ol.

Smoo

than

dge

ntle

grad

ient

.Gra

ssed

bank

s.C

hann

elfin

esa

ndY

(957

1)E

upho

rbia

and

euca

lypt

usw

oodl

and.

Aca

cia

woo

dlan

dw

idth

of1

m,a

ndde

pth

of0.

1m

subr

dow

nstr

eam

.Sm

allv

illag

ene

arby

(50

hous

es).

mso

rted

s821

Jul9

7U

pstr

eam

ofbr

idge

over

Mal

ewa.

Den

seac

acia

Mai

nly

smoo

thw

ithge

ntle

grad

ient

.Sm

allr

apid

fine

sand

Y

(957

4)w

oodl

and.

Wat

erin

gpl

ace

for

cattl

e.be

low

sam

plin

gar

ea.C

hann

elw

idth

of20

m,a

ndsu

ba–

subr

dept

hof

0.5

mm

sort

ed

33

Tabl

e1.

Con

tinue

d

Sam

ple

Sam

plin

gL

ocat

ion

desc

ript

ion

and

land

use

Riv

erm

orph

olog

ySa

mpl

ede

scri

ptio

nX

RF

(RH

Sda

tean

alys

isnu

mbe

r∗)

s921

Jul9

7A

tpum

psat

ion

just

upst

ream

ofM

alew

a’s

confl

uenc

eFa

irly

smoo

thgr

adie

ntbu

twith

inte

rmitt

ent

smal

lFi

nesa

nd(9

573)

with

the

Tur

asha

.Sm

alls

cale

farm

ing

(mai

ze)

and

rapi

ds.C

hann

elw

idth

of16

m,a

ndde

pth

of0.

4m

suba

–su

brac

acia

woo

dlan

dm

sort

ed

s10

24Ju

l97

Ats

mal

lbri

dge

over

the

Tur

asha

dam

rese

rvoi

r.Sm

all

Res

ervo

irda

mw

ithsu

bmer

ged

mac

roph

ytes

and

Sand

ysi

ltY

(957

2)sc

ale

farm

ing

(mai

ze)

and

acac

iaw

oodl

and

float

ing

vege

tatio

nsu

br–

suba

wso

rted

s11

24Ju

l97

Atb

ridg

eov

erG

ilgil,

3km

east

ofG

ilgil

tow

nshi

p,Sm

ooth

and

gent

legr

adie

nt.G

rass

edba

nks.

Cha

nnel

Sand

ysi

ltY

(957

0)Pa

stur

esw

idth

of2–

4m

,and

dept

hof

0.4

msu

br–

suba

w-m

sort

ed

s12

24Ju

l97

Ats

mal

lbri

dge

dow

nstr

eam

ofbi

gger

brid

geov

erth

eSm

ooth

and

gent

legr

adie

nt.G

rass

edba

nks.

Cha

nnel

fine

sand

(956

8)M

alew

aon

the

OlK

alou

road

,eas

tof

OlK

alou

,w

idth

of5–

6m

,and

dept

hof

0.7

msu

ba–

subr

Past

ures

mso

rted

s14

25Ju

l97

On

the

Kar

atia

tthe

Del

amer

efa

rm,n

extt

ofe

nce

byN

oco

ntin

uous

wat

erflo

w.G

entle

grad

ient

with

silty

clay

Y(9

582)

the

nort

hern

swam

p.Pa

stur

esan

dne

arB

uffa

lo/c

attle

gras

sed

bank

s.C

hann

elw

idth

of1–

2m

,and

dept

hof

rw

ater

ing

plac

eof

0.1

m(i

npo

ol).

wso

rted

s15

25Ju

l97

Gilg

ilri

ver

onth

efa

rmof

Mr

A.S

imps

on.F

arm

land

Smoo

than

dge

ntle

grad

ient

.Cha

nnel

wid

thof

1–2

m,

clay

ey,s

andy

silt

Y(N

oR

HS)

<an

dpa

stur

es,n

ear

cattl

ew

ater

ing

loca

tion

and

dept

hof

0.1

m(i

npo

ol).

s–

mr

mso

rted

m1

7A

ug97

Eas

tern

frin

geof

Gilg

ilpa

pyru

ssw

amp,

onM

erul

aR

iver

frin

ged

bysh

allo

wpo

ols

ofw

ater

with

silty

clay

Y(9

577)

farm

,sw

amp/

wet

land

vege

tatio

nfo

r>

50m

each

side

abun

dant

papy

rus

and

salv

inia

.Cha

nnel

wid

thof

5m,

r–

sran

dde

pth

of1

mw

sort

ed

m3

7A

ug97

Sout

hern

part

ofM

erul

afa

rm,a

tcat

tlew

ater

ing

plac

eG

entle

grad

ient

,ste

epgr

asse

dba

nds.

silty

sand

Y(9

569)

near

smal

lfar

mvi

llage

(∼10

hous

es)

Cha

nnel

wid

thof

15m

,and

dept

hof

1.5

msu

brm

-wso

rted

Not

es:∗

RH

Ssi

tenu

mbe

rsan

dba

ckgr

ound

data

are

refe

rred

toin

Eve

rard

etal

.(20

02).

†N

eith

erof

thes

esi

tes

are

anR

HS

site

,but

both

are

clos

eto

Mal

ewa

RH

Ssi

te95

81(K

AR

Ifa

rm).

<T

his

isno

tan

RH

Ssi

te,b

utis

upst

ream

ofR

HS

site

9579

and

form

edth

eup

per

limit

ofbi

rdsu

rvey

stre

tch

inE

vera

rdet

al.(

2002

).Sa

mpl

ede

scri

ptio

n:a

–an

gula

r;r–

roun

ded;

w–

wel

l;m

–m

oder

atel

y;p

–po

orly

Not

e:X

RF–

X-r

ayflu

ores

cenc

e–

maj

oran

dtr

ace

geoc

hem

istr

y.

34

Figure 2. Sample site locations in the river sediment survey.

erately to well sorted, and the individual grains aresub-rounded to rounded. Since the flow of the Gilgilis relatively slower and more gentle, the sediment isgenerally fine (fine silt or clayey silt).

Sediments in the Malewa system represent an in-termediary stage between the poorly sorted and an-gular sediment of the Karati and the fine, well-sortedsediment of the Gilgil. This follows from the upperreaches of the Malewa being steep and similar to thoseof the Karati (although these upland reaches were notsampled in this study), whereas the lower reaches aresimilar to the Gilgil. Compared to the Karati and theGilgil, the Malewa is, furthermore, a more forcefulriver. Consequently, the sediment in the lower reachesis, though generally well-sorted, coarser-grained thanthose of the Gilgil.

Lake sediment description

The lake stratigraphy is conveniently divided into sixsub-areas: central lake, delta front, northeast, northw-est, southeast, and southwest. These divisions are notabsolute, as differences between areas are gradational.The division reflects differences in water depth andlong term sediment dynamics, and serves to make bothanalysis and presentation of results more effective.

The Central Lake sub-area occurs in the centre ofLake Naivasha with depths of 5–6 m. The generalstratigraphy of the four cores corresponded to that sug-gested by Verschuren (1996), and is summarised inTable 2 . A thickness of 10–20 cm of sediments in thecentral lake sub-area, interpreted as being laid downduring the past 10–15 years, gives a current sediment-

35

Table 2. Idealised stratigraphic section of the central LakeNaivasha basin

Depth of Type of sediment and interpretationinterface

0 cm Brown, organic mudTrace macrophytes and papyrus, and trace diat-oms.Interpreted as sediment laid down during the last10–15 years during which time papyrus raftshave been more or less completely absent fromthis part of the lake.Distinct contact

10–20 cm Light brown, organic mudAbundant macrophytes and papyrus and tracediatoms.Interpreted as sediments laid down during pro-gressive increasing water levels from about 1950to 1980. A few cm thick layer of yellowish browndiatomaceous felt may in places be present at thebase of this unit. Together with the unit abovethey seem to correspond to Unit III of Verschuren(1996).Gradational contact

40–50 cm Black rooty and diatomaceous mudAbundant macrophytes, papyrus and abundantdiatoms.Interpreted as sediments laid down during a pro-gressive lowering of water levels in the early 20th

century when papyrus stands were continuouslygerminating on exposed mudflats. Probably cor-responding to Unit II of Verschuren (1996).Distinct non-conformity

70–75 cm Dark brown peaty mudAbundant macrophytes, papyrus and diatoms.Interpreted as sediment laid down during lowlake level stand of the mid 19th century. Probablytruncated by a period when the lake may havebeen completely dry. Probably corresponding toUnit I of Verschuren (1996).Distinct non-conformity

80–90 cm Black PeatAbundant papyrus and trace macrophytes.Hard and massive peat laid down when the wholelake must have been a shallow, fragmented wet-land overgrown with papyrus. Probably corres-ponding to Unit 0 of Verschuren (1996) and thuslaid down during 16th to 17th century.

ation rate of about 1 cm per year. The upper 40–50 cmappears to have been laid down conformably. Belowthis level, two distinct non-conformities, caused byerosion during times when lake levels were low, were

identified. A distinct horizon of yellowish brown, di-atomaceous felt was identified at the base of the lightbrown, organic mud in cores 10 and 15 only. The factthat this unit was not noted in all cores indicates thatthe unit is not continuous over the whole central basin.An explanation for the unit’s formation must there-fore be sought in a process that was not basin-wide.A system of smaller sub-basins, each with its specificsalinity and water quality conditions, situated withina larger wetland, may have created conditions for themass mortality and subsequent sedimentation of diat-oms in some basins but not others. A similar situation,albeit at a larger scale, occurs in the present day wherewater quality, and hence diatom assemblages, varysomewhat between the three sub-basins of the mainlake, that is the Main Lake, Oloidien Bay and CrescentIsland crater lake (Verschuren, 1996).

The Delta front sub-area occurs along the lakewardfringe of the North Swamp, at depths up to about 1 m.The deltafront sub-area is dynamic where sediment-ation patterns are sensitive to changes in water level,changes in location of river inflows and the extentof swamp vegetation. Consequently, sediments in thearea are an heterogeneous mix of massive brown clays,sandy mud, organic mud and peat horizons.

The Northeast sub-area, at depths of about 1–3m, contains sandy muds derived from the river inlets.These sediments are continuously being reworked andre-suspended by wave action. Below these active sedi-ments are stiff grey clays, interpreted as having formedfrom the previous aerial exposure and mineralizationof lacustrine sediments, and brown clays which areinterpreted as representing deltaic mudflat deposits.

The Northwest sub-area, depths of 2–3 m, containssediments which appears to correspond to the upperthree units of the central lake stratigraphy with twobrown organic mud units overlying a black rooted mudunit.

Core 11 was situated only about 3 km west of theMalewa outflow. Nevertheless, there was no evidenceof material directly derived from the river in this core.It is therefore indicated that river sediment input is,subsequent to being introduced to the lake, transportedin an easterly direction only.

The Southeast sub-area, together with one coretaken about 4 km further north, 2.5 km west of Cres-cent Island (core 5), but included in this group becauseof its stratigraphic similarity, appear to correspond tothe upper four units of the central lake stratigraphywith brown organic mud units above a black or darkbrown rooted mud unit over a dark brown peaty mud.

36

Figure 3. Sample site locations in the two lake sediment surveys. Sites in the 2000 survey are identified by the prescript L.

In the Southwest Bay sub-area, at depths of 3–4 m,a far greater rate of deposition was evident. The strati-graphies was somewhat different from the central lakestratigraphy. The two cores consisted of organic mudunits, interpreted as corresponding to the upper twounits of the central lake stratigraphy. However, thoughboth cores were nearly 100 cm in length, neither in-tersected the organic rich, black mud units laid downin other parts of the lake in early to middle part ofthis century. Consequently, the sedimentation rate inthis relatively sheltered part of the lake appears to bemore rapid than in the central part of the lake. A 10-cmthick brown organic mud unit with abundant Salviniaand papyrus fragments at a depth of about 40 cm isinterpreted as a unit laid from the 10–15 years beforethe 1987 low level. At this time, both papyrus raftsand Salvinia mats were common in this part of thelake (Tarras-Wahlberg, 1986). Such an interpretationleads to a sedimentation rate of about 3 cm per year inthis part of the lake, which is considerably more thanestimated sedimentation rates in other parts of the lake.

Present-day sediment dynamics

Comparisons of the sediment composition at thesediment-water interface of the 18 cores enables ananalysis of present day sedimentation patterns.

Near the outflow of the Malewa the surface sed-iments are either massive brown clay or sandy mudunits. The clay is interpreted as a deltaic deposit (amud flat) which is now actively being eroded by waveaction, whereas the sandy mud is material directlyderived from the Malewa and Gilgil Rivers. Con-sequently, the sedimentary regime of the pro-delta areais dominated by erosion and sedimentary transport,rather than sedimentation. Presumably, sedimentationand associated deltaic growth occurs in the reducedareas now protected by swamp vegetation.

The surface sediments in the bay in the north-eastern part of the lake and areas directly west andnorth west of Crescent Island are, like the pro-deltaarea, dominated by sandy muds and clay units. Con-sequently, the sedimentary regime here is also dom-inated by transport and erosion. However, the stiffclays here are not only brown clays (interpreted asdeltaic deposits) but also stiff clays that are grey ordark grey in colour. These stiff clays are interpreted

37

as having formed from lacustrine sediments that wereexposed to air, such that they became mineralised andoxidised. This may be a consequence of historic lowlake levels, and is consistent with the inferences aboutdelta-forming processes based on RHS and other geo-morphological observations by Everard et al. (2002).Organic muds dominate the surface sediments col-lected elsewhere across the lake in this study and,hence, these areas represent depositional sedimentaryenvironments.

Geochemistry

Tables 3 and 4 show the results of the geochemicalanalyses. Lake and river sediments are geochemicallysimilar, and the contents of major and minor elementsare mostly comparable to global averages (Table 3).The only noticeable differences between river and lakesediments are that river sediment contain somewhatmore P2O5 and Pb, somewhat less Cu and Ni, andsignificantly less SO3, than lake sediment. These dif-ferences are, with the exception of SO3, less than oneorder of magnitude and may arise from the differentanalytical methods used. The higher content of SO3in lake sediment compared to river sediment is, how-ever, to be expected due to the tendency of animalsand plants, more common in the lake compared to therivers, to concentrate sulphur. This hypothesis is cor-roborated by the fact that the only two lake sedimentsamples to have low SO3 concentrations are 12 A and12 D, both of which represent inorganic clay units.

River sample S10, taken from the Turasha dam inthe Malewa catchment, differs from the other samplesin that it contains more organic and volatile materi-als (as indicated by the low total percentage). Thisreflects, the low energy environment of the dam wherefine organic material can settle out to a greater extentthan in-stream.

All sediment samples contain elevated levels ofFe2O3, and the concentrations of 6–13% are consider-ably higher than the 4% expected in an average shale.The high iron content could either be caused by thebed rock (i.e., the volcanic rocks of the area) beingnaturally enriched in iron, or caused by the erosion ofiron rich lateritic soils, which are common throughoutthe area.

The content of minor and potentially toxic metalsand metalloids are, in comparison to global averages,mostly unremarkable. Exceptions are elevated con-centrations of Zn in all samples, Ni in lake sediment

and in the Turasha dam sediment, and Cd in two lakesediment samples.

Zinc concentrations are similar in all samples,including samples taken from remote parts of the cath-ment (e.g., S12 and S3), which strongly indicatesthat relatively high concentrations are natural in ori-gin. Zinc concentrations vary from 88 to 190 mgkg−1, which can be compared to the ‘threshold effectlevel’ of the sediment quality criteria, set at 123 mgkg−1. The latter assumes however, 100% bioavailabil-ity, which is unlikely to be the case, so the Zn levelsare not likely to be of concern.

Nickel levels in sediment are elevated in com-parison to the sediment quality guidelines, althoughconsiderably lower than the global average shale. Fur-ther, the Ni concentrations are considered as only‘somewhat elevated levels’, as defined by the SwedishEnvironmental Protetion Agency (SEPA, 1998). Thus,it may be the case that the Canadian criteria are overlyconservative. Nickel is, nevertheless, enriched in ba-sic rocks such as the basalts of the Nyandarua range,which may explain the levels encountered. Further,Ni concentrations are somewhat higher in lake sedi-ment and in the Turasha dam sediment compared toriver sediments, indicating that this metal, for reas-ons unknown, is preferentially enriched in still watersediment compared to active river sediment. Pos-sible anthropogenic sources of Ni include batteries,electroplating and paints. None of these sources ap-pear overly likely in the Lake Naivasha area, whichharbours little or no such industry.

The lake sediment samples 9C and 12B contain7 and 6 mg kg−1 Cd, respectively, which is signi-ficantly above global average levels and also abovethe ‘probable effect level’ of the Canadian sedimentquality criteria. There are a number of explanations forthese seemingly high concentrations. Firstly, the con-centrations are fairly close to the low limit of detectionof the analytical method used (2 mg kg−1) and there-fore analytical error cannot be ruled out. Secondly,cadmium is a common contaminant of rock phos-phorus, the major ingredient of chemical fertilisers(Driver et al., 1999). Excessive inputs of Cd into thelake may therefore result from the use and release ofchemical fertilisers. Both units represent sediments atdepth below the sediment–water interface (39–49 and8–44 cm, respectively). The high Cd concentrationsmay thus be a result of Cd being mobilised from theupper oxidising sediments and re-precipitating out atdepths where conditions are reducing. Thus a horizonenriched in Cd could be generated from the concentra-

38

Tabl

e3.

Con

cent

ratio

nof

sele

cted

maj

oran

dm

inor

elem

ents

inri

ver

sedi

men

tsam

ples

from

Lak

eN

aiva

sha’

sca

tchm

ent.

Ana

lyse

sw

ere

perf

orm

edby

X-R

ayFl

uore

scen

ce(X

RF)

.Low

erto

tals

are

caus

edby

orga

nic

mat

ter

and

othe

rvo

latil

em

ater

ial

Sam

ple

Tota

lSi

O2

Al 2

O3

Fe2O

3K

2O

Na 2

OM

gOP 2

O5

SO3

CaO

MnO

As

Co

Cr

Cu

Ni

PbZ

n

%%

%%

%%

%%

%%

%m

g/kg

mg/

kgm

g/kg

mg/

kgm

g/kg

mg/

kgm

g/kg

S392

5712

133.

22.

20.

530.

310.

040.

910.

22<

10<

6<

100

22

1415

0

S491

5713

114.

22.

70.

730.

290.

041.

20.

41<

10<

6<

100

37

1715

0

S791

5713

123.

62.

00.

810.

200.

031.

70.

38<

10<

6<

100

410

1915

0

S890

5514

9.9

3.6

2.3

0.85

0.26

0.05

1.8

0.23

<10

8<

100

811

1614

0

S10

8349

1411

2.0

1.1

0.84

0.54

0.08

1.6

0.31

<10

14<

100

2220

1217

0

S11

9058

1310

3.1

1.7

0.72

0.41

0.03

1.0

0.38

<10

<6

<10

05

1116

150

S14

9361

149.

32.

61.

41.

00.

310.

031.

60.

23<

10<

6<

100

2016

1217

0

S15

9061

138.

03.

72.

20.

660.

240.

041.

00.

20<

10<

6<

100

66

1414

0

M1

8958

159.

42.

00.

900.

880.

490.

040.

990.

12<

10<

6<

100

1915

1119

0

M1-

repe

at87

5615

9.3

2.0

0.90

0.86

0.28

0.04

1.0

0.14

<10

8<

100

1714

1119

0

M3

9155

1410

3.6

2.4

0.89

0.37

0.03

2.1

0.17

<10

11<

100

710

1313

0

Ave

rage

shal

e158

.115

.44.

023.

241.

302.

440.

170.

643.

10.

10–

1990

4568

2095

Thr

esho

ldef

fect

leve

l:5.

9–

3736

1835

123

Sedi

men

tqua

lity

crite

ria

for

prot

ectio

nof

aqua

ticlif

e2Pr

obab

leef

fect

leve

l:17

–90

197

3691

315

1M

ajor

elem

ents

take

nfr

omM

ason

&M

ore

(198

2)an

dm

inor

elem

ents

from

Tur

ekia

n&

Wed

epoh

l(19

61).

2E

nvir

onm

entC

anad

a(1

996)

39

Tabl

e4.

Con

cent

ratio

nof

sele

cted

maj

oran

dm

inor

elem

ents

inla

kese

dim

ent

sam

ples

from

Lak

eN

aiva

sha

and

itsca

tchm

ent.

Ana

lyse

sof

lake

sam

ples

wer

epe

rfor

med

byIC

P-A

ES

afte

rdi

gest

ion

inaq

uare

gia

LO

IA

l 2O

3Fe

2O

3M

gOP 2

O5

SO3

CaO

MnO

As

Cd

Co

Cr

Cu

Ni

PbZ

n

%%

%%

%%

%%

mg/

kgm

g/kg

mg/

kgm

g/kg

mg/

kgm

g/kg

mg/

kgm

g/kg

Cor

esa

mpl

es(1

997)

9A(0

–11c

m)

–17

110.

940.

180.

630.

940.

212

<2

99

2328

<5

160

9B(1

1–39

cm)

–14

9.0

0.65

0.16

0.70

1.1

0.22

3<

28

1829

41<

513

0

9C(3

9–49

cm)

–13

7.3

0.56

0.16

0.58

2.0

0.15

57

76

918

<5

110

9D(4

9–73

cm)

–9.

99.

50.

480.

160.

681.

30.

28<

2<

211

3617

38<

588

9E(7

3–85

cm)

–12

7.9

0.55

0.16

0.60

0.95

0.15

2<

29

3013

35<

514

0

9F(8

5–10

1cm

)–

137.

90.

550.

090.

500.

970.

122

<2

622

1322

<5

150

12A

(0–8

cm)

–11

7.3

0.48

0.11

0.05

0.67

0.15

<2

<2

1015

1019

517

0

12A

-rep

eat

–11

7.5

0.50

0.09

0.05

0.67

0.15

<2

<2

1111

1018

519

0

12B

(8–4

4cm

)–

106.

00.

510.

090.

301.

10.

224

610

316

216

160

12C

(44–

60cm

)–

167.

20.

630.

070.

250.

830.

084

<2

107

1824

718

0

12D

(60–

73cm

)–

198.

60.

700.

070.

080.

600.

065

<2

1114

1722

719

0

Gra

bsa

mpl

es(2

000)

L-1

14.

8–

––

––

––

––

––

––

––

L-1

224

––

––

––

–0.

98<

0.17

––

1517

<0.

316

0

L-1

331

––

––

––

–0.

84<

0.17

––

1113

<0.

314

0

L-1

420

––

––

––

–0.

55<

0.17

––

7.4

6.4

3.7

100

L-1

4-re

peat

20–

––

––

––

0.52

<0.

17–

–7.

65.

73.

710

0

L-1

626

––

––

––

–0.

99<

0.17

––

1715

<0.

318

0

L-1

733

––

––

––

––

––

––

––

–

L-1

839

––

––

––

–0.

87<

0.17

––

1319

3.6

112

Ave

rage

shal

e115

.44.

023.

240.

170.

643.

110.

10–

0.3

1990

4568

2095

Thr

esho

ldef

fect

leve

l:5.

90.

6–

3736

1835

123

Sedi

men

tcri

teri

afo

rpr

otec

tion

ofaq

uatic

life2

Prob

able

effe

ctle

vel:

173.

5–

9019

736

9131

5

1M

ajor

elem

ents

take

nfr

omM

ason

&M

ore

(198

2)an

dm

inor

elem

ents

from

Tur

ekia

n&

Wed

epoh

l(19

61).

2E

nvir

onm

entC

anad

a(1

996)

.

40

tion of existing natural Cd or, alternatively, through thesimilar concentration of contaminant Cd. The analysisof surface sediments performed in 2000 does not in-dicate any significant supply of contaminant Cd fromfertilisers. This follows since Cd levels are equally lowat sites situated near the big flower farms on the east-ern shore of the lake (L-13 and L-14), as at sites nearthe much less developed western shore (L-16 and L-18). Hence, the relatively high Cd levels encounteredare unlikely to be caused by contamination, but ratherdue to geochemical redistribution of naturally existingCd. The content of other minor and potentially toxicmetals and metalloids is low.

Conclusions

The description and analyses of Lake Naivasha sedi-ment have provided extensive data on both present andpast sedimentation patterns. Importantly, the findingsshow that natural lake level changes have caused dra-matic changes to the lake ecosystem in the past, andthat the instability of the lake’s ecosystem is in partnatural.

At present, sediment dynamics in the lake are gov-erned by the presence of river point sources in thenorth, the deposition of silt and sand sized materialnear the river outlets; and wave-induced re-suspensionof fine and organic rich sediment in the north, whichare transported in easterly and southerly directions anddeposited in the lake’s central, southern and easternparts. Deposition of fine, organic rich sediment isalso occurring in northern areas that are protected bypapyrus swamp vegetation.

The sedimentation rate in the central lake is foundto be about 1 cm per year, which is in agreements witha previous estimate (Verschuren, 1996). However, thesedimentation rate in Elsamere Bay in the southwestappears to be higher, at up to 3 cm per year.

In the north-western part of the lake, historic lowsin lake levels has lead to the formation of a possiblycontinuous layer of stiff grey, mineralised clay. Thislayer will act as an effective water barrier and its exist-ence indicates that ground water recharge in this partof the lake, suggested by among others Ase (1987)may be impossible.

The geochemical analysis of a selected number ofsediment samples from the lake and its catchment in-dicate that the concentrations of Fe, Zn, Cd and Nisediment are elevated compared to global averagesand/or sediment quality criteria. The reasons for the

relatively high concentrations are likely to be found inthe sediment source materials (volcanic rocks or later-ite soils), which are naturally rich in these elements.The concentrations of other potentially toxic elementsare low.

Lake Naivasha area is experiencing a period ofrapid development with ever-increasing use of landand irrigation water for flower farming and the ex-pansion of the Olkaria hydrothermal plants (Becht &Harper, 2002). In addition to these intensive uses,Everard et al. (2002) note an increasing intensity ofsubsistence agriculture in the catchment, includingwidespread poor land management practices. Con-sequently, it is not unlikely that the lake’s catchmentmay soon be severely affected by impacts related tothese developments as well as the pressures of a risingpopulation. In addition to these immediate pressures,the planned construction of a water supply reservoirfor the nearby town of Nakuru, that will export a sub-stantial proportion of the river Malewa’s water, willhave an immediate and large scale affect both on theMalewa River and the lake itself (Harper et al., 1995).

The present study indicates that the river and lakesediment of the Lake Naivasha catchment are stillrather pristine, and that anthropogenic impacts are notyet significant. The data obtained may therefore beused as reference values for assessing the possible en-vironmental impacts of future developments aroundthe lake.

Acknowledgements

This study formed past of the work of the Universityof Leicester research project at Lake Naivasha, whichsince 1987 has been authorised by the Office of thePresident, Government of Kenya under research per-mit to Dr D. M. Harper no. OP 13/001/12C 46. Theproject was funded by the Earthwatch Institute, Bo-ston, USA, and Oxford, England. The data collectionwould not have been possible without the assistance ofnumerous Earthwatch volunteers. Our sincere thanksgo to the numerous colleagues in Kenya in particularJill & Angus Simpson and Velia Carn.

References

Ase, L. E., K. Sernbo & P. Syren, 1986. Studies of Lake Naivasha,Kenya and its Drainage Area. Forkningsrapporter 63, Stock-holms Universitet Naturgeografiska Institutionen, Stockholm: 75pp.

41

Ase, L. E., 1987. A note on the waterbudget of Lake Naivasha,Kenya. Geografiska Annaler, 69A: 415–429.

Beadle, L. C., 1932. Scientific results of the Cambridge Expeditionto East Africa. Linn. Soc. J. Zool. 38: 137–141.

Becht, R. & D. M. Harper, 2002. Towards an understanding ofhuman impact upon the hydrology of Lake Naivasha, Kenya.Hydrobiologia 488 (Dev. Hydrobiol. 168): 1–11.

Bowie, A. J. & C. K. Mutchler, 1986. Sediment sources and yieldsfrom complex watersheds. In: Wang, S. Y., Shen, H. W. & Ding,L. Z. (eds), Proceedings of the Third International Symposiumon River Sedimentation. pp 1224–1232. University of Miss.,Jackson.

Carelli, G., N. Sannolo, G. De Lorenzo & N. Castellino 1995. Eco-systems. Chapter 3 In Inorganic Lead Exposure – Metabolismand Intoxication, Lewis Publishers, Boca Raton, FL: 15–51.

Darnley, A. G., A. Björklund, B. Bolviken, N. Gustavsson & P.Koval, 1995. A Global Geochemical Database: Recommenda-tions for International Geochemical Mapping. Final Report ofIGCP Project 259. UNESCO, Paris.

Driver, J., P. Lijmbach and I. Steen, 1999. Why recover phosphorusfor recycling, and how? Environ. Tech. 20: 651–662.

Environment Canada, 1995. Interim Sediment Quality Guidelines,Draft. Prepared by Environment Canada, Ecosystem Con-servation Directorate, Evaluation and Interpretation Branch,Guidelines Division, Ottawa.

Everard, M., J. A. Vale, D. M. Harper & H. Tarras-Wahlberg, 2002.The physical attributes of the Lake Naivasha catchment rivers.Hydrobiologia 488 (Dev. Hydrobiol. 168): 13–25.

Fordyce, F. M., D. Masara & J. D. Appleton, 1996. Stream sedi-ment, soil and forage chemistry as indicators of cattle mineralstatus in Northeast Zimbawe. In Appleton J. D., R. Fuge & G. J.H. McCall (eds), Environmental Geochemistry and Health, withSpecial Reference to Developing Countries. Geological SocietySpecial Publication No. 113, The Geological Society, London:23–37.

Förstner, U., 1983. Assessment of metal pollution in rivers andestuaries. In Thornton, I. (ed.), Applied Environmental Geo-chemistry. Academic Press, London.

Förstner, U. & G. T. W. Wittman, 1983. Metal Pollution and theAquatic Environment. Springer Verlag, Amsterdam.

Gaudet, J. J. & J. M. Melack, 1981. Major ion chemistry in a tropicalAfrican lake basin. Freshwat. Biol. 11: 309–333.

Gouder de Beauregard, A.-C., D. M. Harper, F. Malaisse & J.-J. Symoens, 1998. Dynamique récente et cartographie de lavégétation aquatique (1960–1996) du lac Naivasha (Rift Val-ley, Kenya). Bull. Séanc. Acad. r. Sci. Outre-Mer 44 (1998–3):373–389.

Harper, D. M., 1984. Recent changes in the ecology of LakeNaivasha. Verh. int. Ver. theor. angew. Limnol. 22: 1192–1197.

Harper, D. M, K. M. Mavuti & S. M. Muchiri, 1990. Ecology andmanagement of Lake Naivasha, Kenya, in relation to climatechange, alien species’ introductions, and agricultural develop-ment. Environ. Conserv. 17: 328–336.

Harper, D. M., G. Phillips, A. Chilvers, N. Kitaka & K. M. Mavuti,

1993. Eutrophication prognosis for Lake Naivasha, Kenya. Verh.int. Ver. theor. Angew. Limnol. 25: 861–865.

Harper, D. M., C. Adams & K. M. Mavuti, 1995. The aquaticplant communities of the Lake Naivasha wetland, Kenya: pattern,dynamics and conservation. Wetlands Ecology and Manage. 3:111–123.

Kitaka, N., D. M. Harper, N. Pacini & M. Kenneth Mavuti,2002. Chemical characteristics, with particular reference tophosphorus, of the rivers draining into Lake Naivasha, Kenya.Hydrobiologia 448 (Dev. Hydrobiol. 168): 55–70.

Ladd, S. C., W. A. Marcus & S. Cherry, 1998. Differences intrace metal concentrations among fluvial morphological unitsand implications for sampling. Environ. Geol. 3: 259–270.

Levinson, A. A., 1980. Introduction to Exploration Geochemistry.Applied Publishing Ltd., Calgary.

Mason, B. & C. B. More, 1982. Principles of Geochemistry, 4thedn. John Wiley and Sons, New York.

Maher, W., G. E. Batley & I. Lawrence, 1999. Assessing thehealth of sediment ecosystems: use of chemical measurements.Freshwat. Biol. 41: 361–372.

Nicholson, K., 1996. Lacustrine sediment geochemistry as a toolin retrospective environmental impact assessment of mining andurban development in tropical environments: examples fromPapua New Guinea. In Appleton, J. D., R. Fuge & G. J. H. Mc-Call (eds), Environmental Geochemistry and Health, with specialreference to developing countries. Geological Society SpecialPublication No. 113. The Geological Society, London: 195–200.

Pettijohn, F. J., 1975. Sedimentary Rocks. 3rd edn. Harper and Row,New York.

Stumm, W. & J. J. Morgan, 1996. Aquatic Chemistry. Chem-ical Equilibria and Rates in Natural Waters, 3rd edn. Wiley-Interscience, New York.

Swedish Environmental Protection Agency, 1998. Bakgrundsdok-ument till bedömningsgrunder för sjöar och vattendrag, SNV,Stockholm (in Swedish).

Tarras-Wahlberg, N., 1986. Observations on Salvinia and its envir-onment at Lake Naivasha (Kenya). J. East Africa nat. Hist. Soc.nat. Mus. 76: 189.

Turekian, K. K. & K. H. Wedepohl, 1961. Distribution of the ele-ments in some major units of the earth’s crust. Bull. geol. Soc.Am. 72: 175–192.

Verschuren, D., 1996. Recent and late-Holocene palaeolimnologyof Lake Naivasha and Sonachi, Kenya. PhD thesis, University ofMinnesota.

Windom, H. L., J. T. Byrd, R. G. Smith & F. Huan, 1991. In-adequacy of NASQAN data for assessing metal trends in thenation’s rivers. Environ. Sci. Technol. 25: 1137–1142.

Willford, W. A., M. J. Mac & R. J. Hesselberg, 1987. Assessingthe bioaccumulation of contaminants from sediments by fish andother aquatic organisms. Hydrobiologia 149: 107–111.

Williams, T. M., P. N. Dunkley, E. Cruz, V. Acitimbay, A. Gaibor, E.Lopez, N. Baez & J. A. Aspden, 2000. Regional geochemical re-connaissance of the Cordillera Occidental of Ecuador: economicand environmental applications. Appl. Geochem. 15: 531–550.