applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern...

14
Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya Anna Tengberg 1,* , Jim Ellis-Jones 2 , Romano Kiome 3 , Michael Stocking 4 Kenya Agricultural Research Institute, Regional Research Centre, Embu, P.O. Box 27, Embu, Kenya Received 23 February 1998; accepted 30 June 1998 Abstract Agrodiversity – the diversity of cropping systems, crop species and farm management practices has received increasing attention in recent years as a way of spreading risk and supporting food security in resource-poor farming systems. This paper discusses the dynamic aspects of indigenous soil and water conservation (ISWC) practices in a semi-arid part of Kenya. The objective is to show the range of sources of variability and diversity that prevail in this environment, the responses of farmers to this variability, and the way farmers’ rationalise the heterogeneity of soil and water management practices. Methods used included participatory surveys and evaluations, on-farm monitoring, soil and rainfall data analyses, and questionnaire surveys. Sources of variability affecting cropping systems and land management practices included rainfall, soil fertility, farmer resource level and farm productivity. A decision-tree was developed to examine how biophysical and socio-economic variability affected farmers’ choice of ISWC. Different ISWC structures dominated on sandy and stony soils, respectively. Low resource farmers tended to choose cheaper and less labour demanding techniques, and constructed smaller ISWC structures than better endowed farmers. The largest diversity of ISWC practices was found on newly-opened land with mixed soils. Moreover, on-farm productivity levels indicated that costly investments in SWC are unfeasible, as this would further increase the risk for negative returns to farming. The wider implications of the results are that SWC interventions in marginal areas should build on the existing agrodiversity and an understanding of the complex interactions between environmental and socio-economic factors that give rise to differences in farming systems and land management practices. # 1998 Elsevier Science B.V. All rights reserved. Keywords: Agrodiversity; ISWC; Semi-arid; Land management; Kenya; Mbeere 1. Introduction There is a growing interest in diversity of cropping systems, crop species and farm management practices in low input agricultural systems. The belief that such systems are unsustainable is giving way to a recogni- tion that many small-holder farmers of the tropics utilise the diversity of their environments, manage a Agriculture, Ecosystems and Environment 70 (1998) 259–272 *Corresponding author. Tel.: +46-31-7734733; fax: +46-31- 7731986; e-mail: [email protected] 1 Present address: Go ¨teborg University, Earth Sciences Centre, Physical Geography, P.O. Box 460, SE-405 30, Go ¨teborg, Sweden. 2 Present address: Silsoe Research Institute (SRI), Wrest Park Silsoe, Bedford MK45 4HS, UK. 3 Kenya Agricultural Research Institute (KARI) HQ, P.O. Box 14733, Nairobi, Kenya. 4 Present address: School of Development Studies, University of East Anglia, Norwich NR4 7TJ, UK. 0167-8809/98/$ – see front matter # 1998 Elsevier Science B.V. All rights reserved. PII: S0167-8809(98)00153-4

Upload: anna-tengberg

Post on 16-Sep-2016

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

Applying the concept of agrodiversity to indigenous soil

and water conservation practices in eastern Kenya

Anna Tengberg1,*, Jim Ellis-Jones2, Romano Kiome3, Michael Stocking4

Kenya Agricultural Research Institute, Regional Research Centre, Embu, P.O. Box 27, Embu, Kenya

Received 23 February 1998; accepted 30 June 1998

Abstract

Agrodiversity ± the diversity of cropping systems, crop species and farm management practices has received increasing

attention in recent years as a way of spreading risk and supporting food security in resource-poor farming systems. This paper

discusses the dynamic aspects of indigenous soil and water conservation (ISWC) practices in a semi-arid part of Kenya. The

objective is to show the range of sources of variability and diversity that prevail in this environment, the responses of farmers

to this variability, and the way farmers' rationalise the heterogeneity of soil and water management practices. Methods used

included participatory surveys and evaluations, on-farm monitoring, soil and rainfall data analyses, and questionnaire surveys.

Sources of variability affecting cropping systems and land management practices included rainfall, soil fertility, farmer

resource level and farm productivity. A decision-tree was developed to examine how biophysical and socio-economic

variability affected farmers' choice of ISWC. Different ISWC structures dominated on sandy and stony soils, respectively.

Low resource farmers tended to choose cheaper and less labour demanding techniques, and constructed smaller ISWC

structures than better endowed farmers. The largest diversity of ISWC practices was found on newly-opened land with mixed

soils. Moreover, on-farm productivity levels indicated that costly investments in SWC are unfeasible, as this would further

increase the risk for negative returns to farming. The wider implications of the results are that SWC interventions in marginal

areas should build on the existing agrodiversity and an understanding of the complex interactions between environmental and

socio-economic factors that give rise to differences in farming systems and land management practices. # 1998 Elsevier

Science B.V. All rights reserved.

Keywords: Agrodiversity; ISWC; Semi-arid; Land management; Kenya; Mbeere

1. Introduction

There is a growing interest in diversity of cropping

systems, crop species and farm management practices

in low input agricultural systems. The belief that such

systems are unsustainable is giving way to a recogni-

tion that many small-holder farmers of the tropics

utilise the diversity of their environments, manage a

Agriculture, Ecosystems and Environment 70 (1998) 259±272

*Corresponding author. Tel.: +46-31-7734733; fax: +46-31-

7731986; e-mail: [email protected] address: GoÈteborg University, Earth Sciences Centre,

Physical Geography, P.O. Box 460, SE-405 30, GoÈteborg, Sweden.2Present address: Silsoe Research Institute (SRI), Wrest Park

Silsoe, Bedford MK45 4HS, UK.3Kenya Agricultural Research Institute (KARI) HQ, P.O. Box

14733, Nairobi, Kenya.4Present address: School of Development Studies, University of

East Anglia, Norwich NR4 7TJ, UK.

0167-8809/98/$ ± see front matter # 1998 Elsevier Science B.V. All rights reserved.

P I I : S 0 1 6 7 - 8 8 0 9 ( 9 8 ) 0 0 1 5 3 - 4

Page 2: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

large variety of crops and genotypes, and employ a

wealth of techniques both to exploit the diversity and

support rural livelihoods (Richards, 1985; Pretty,

1995; Reij et al., 1996). This paper is primarily about

one aspect of biological diversity at the farm level ±

the diversity of indigenous soil and water conservation

(ISWC) practices in semi-arid Kenya, an agro-ecolo-

gical area usually perceived for its paucity of practices

and its dif®culty in gaining a livelihood for farmers

living literally at the economic and environmental

margins.

The semi-arid tropics are not normally noted for

their diversity. Typically, they are stereotyped by

limited land use options, poor management practices

and poverty (Swift, 1996). Marginality, however, may

be hypothesised as a force impelling land users to

protect biological diversity on the grounds that spread-

ing of risk is a greater imperative than the maximisa-

tion of production (Ellis, 1993). Biological diversity,

then, is re¯ected in the diversi®cation of production

strategies and techniques, multiple and sequential

cropping, and a large variety of soil and water con-

servation practices. Appreciating this diversity and

identifying sources of variability in low input farming

systems is a ®rst step for addressing poverty and rural

livelihoods, and facilitating appropriate development

interventions (Brook®eld and Padoch, 1994). Instead

of limited land use options, there may be a wide range

of possibilities, each possibility needing to be targeted

to speci®c environments and households (Okali et al.,

1994).

This paper focuses on the dynamic aspects of

ISWC found in a semi-arid part of eastern Kenya.

Its objective is to show the range of sources of

variability and diversity that prevail in this dif®cult

environment, the responses of farmers to this varia-

bility, and the way farmers' perceive and rationalise

the surprising heterogeneity of soil and water manage-

ment practices.

2. Conceptualising `agrodiversity'

The term `agrodiversity' was used by Brook®eld

and Padoch (1994) to describe the variety of practices

and attributes of tropical small farm systems. Agro-

diversity has subsequently been seen as the way

resource-poor farmers spread risk and support their

own food security (Netting and Stone, 1996). Crop

biodiversity, that is, the number and diversity of

species used by farmers in their cultivation activities

and the number of wild and semi-domesticated species

used for food and other economic products (Old®eld

and Alcorn, 1987), is then only one aspect of agro-

diversity. Another aspect of agrodiversity which inter-

acts with cropping is the diversity of ISWC practices,

now being seen as a vital part in sustaining the

productivity of the soil (Reij et al., 1996).

Agrodiversity is a response of resource-poor farm-

ers to inherent environmental variability. Particularly

in the drylands, temporal and spatial variability domi-

nate. The ®rst is primarily a function of rainfall

variability, while the second is a re¯ection of land-

scape, relief and soil-type as well as spatial variability

in rainfall. Superimposed on this environmental varia-

bility is ethnic, cultural and economic diversity, char-

acterised by differences in wealth and access to

resources. Responding to such variability, farmers

choose different land management strategies accord-

ing to their asset holdings (Scoones, 1996), status

and livelihood requirements. Their choices are

revealed in an often-surprising heterogeneity of land

uses, crops and practices, even in apparently homo-

genous areas.

Farmers respond, therefore, to ecological, environ-

mental and socio-economic changes by ¯exible and

dynamic management strategies (Richards, 1985,

1986). This applies equally to ISWC practices. Most

are characterised by their multiple functions, spread of

labour demands and gender roles (Reij et al., 1996).

Some have been shown to be more economically

viable than introduced technologies (Kiome and

Stocking, 1993). An understanding of these functions

of ISWC is a pre-requisite to any external intervention

to promote agricultural sustainability and minimise

environmental impact of land use.

Agrodiversity, therefore, exists at a number of

spatial and sequential levels, conceptualised in a

simple model (Fig. 1). At the broadest level it exists

within a context of often-extreme environmental

variability which impels land users to adopt a

broad range of strategies for survival. Small-holder

farmers, for example, often change their crops and

practices as the nature of the individual growing

season unfolds. Different farmers, however, adopt

different strategies according to their socio-economic

260 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272

Page 3: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

circumstances and resource endowments. The

response is to reveal a surprising number of complex

farming systems and cropping practices. Crops, for

example, are often sited according to small pockets of

particular soil, or in created depressions which harvest

surrounding runoff. These micro-management prac-

tices have often been ignored. Overlooked also are the

complicated, farmer-speci®c, sets of ISWC practices.

Through the examination of one part of semi-arid

Kenya, this paper will show how ®rm linkages can

be drawn between the sources of variability ± envir-

onmental, social and economic, and the diversity of

farm practices that support livelihoods as re¯ected in

ISWC.

3. Study area and methods

This study was carried out in Mbeere district, a

semi-arid part of eastern Kenya, as part of a project

on indigenous soil and water conservation practices.

The study areas lie in the lower midland marginal

cotton (Gossypium hirsutum L.) and livestock-millet

(Panicum miliaceum) agro-ecological zones (Table 1).

Fig. 1. Conceptual framework for situating ISWC within overall agrodiversity. The arrows indicate that the system is dynamic and evolves

and changes over time.

A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272 261

Page 4: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

The rainfall pattern is bimodal and the rains normally

come in March to May and October to December.

Soils in the area range from nutrient-poor, sandy

Ferralsols to more fertile, stony Cambisols. The land-

scape is undulating with slopes between 3%±43%,

with stony soils on the steepest slopes. The farmers in

the area rely on a mix of rainfed farming, livestock

rearing and off-farm income for their livelihoods.

Maize (Zea mays L.), cowpeas (Vigna unguiculata),

beans (Phaseolus vulgaris L.) and sorghum (Sorghum

bicolor L. Moench) are grown in the April season and

millet, maize and sorghum are grown in the November

season (Gibbon and Pain, 1985; Riley and Brokensha,

1988).

A variety of participatory survey and evaluation

methods were employed. Initially, two participatory

rural appraisals (PRAs) were conducted to enquire

of the type, and conditions of ISWC in two villages

(Altshul and Okoba, 1995; Okoba and Altshul, 1995).

A characterisation of farmers according to land man-

agement practices and access to resources was accom-

plished. Following the PRAs, 20 contact farmers,

representing different resource levels, were selected

and ISWC practices, cropping patterns and inputs

were monitored from the November 1995 cropping

season to the April 1997 season. Several workshops

were also held to evaluate and verify important ®nd-

ings with farmers (Okoba et al., 1998). Moreover, in a

household survey in May 1997 encompassing 48

randomly chosen households evenly distributed in

four of the villages in the study area, farmers were

interviewed to ascertain historical crop production,

labour inputs and soil and water conservation prac-

tices.

4. Sources of variability and diversity

As agrodiversity is the outcome of variability in

both the quality and access to natural resources and the

resources themselves, there are several sources of

variability in the study area that are likely to affect

cropping systems and management practices.

4.1. Rainfall variability

Large areas in central and eastern Kenya receive

less than 300 mm rainfall in six out of 10 growing

seasons (Downing et al., 1985), which is the minimum

amount considered essential for many dryland crops.

When looking at the rainfall at Ishiara, the rainfall

station with the longest record in the study area, it is

evident that the rainfall in both seasons is highly

variable. However, a rainfall cycle of about 10 years

seems to exist with maximum seasonal rainfall just

below 600 mm and minimum seasonal rainfall close to

300 mm for both seasons (Fig. 2). This indicates that

the likelihood of crop failure, particularly for maize

that requires about 600 mm in this environment

(ILACO, 1981), is very high. A statistical analysis

of rainfall in eastern Kenya showed that the rainfall is

subject to random variability rather than long-term

change (Downing et al., 1985).

Table 1

Biophysical characteristics of the study villages in eastern Kenya

Site Soil-type/FAO soil

classification

Agro-ecological

zonea

Altitude

(m)

Mean annual

rainfall (mm)

Temperature av.

max av. min

Mumburi Sandy/Ferralsols LM5 1095 830 29

17

Kathuri Sandy/Ferralsols LM5 1095 830 29

17

Karii Stony/Cambisols LM4 1158 849 No data

Mutuobare Stony loam/Cambisols and Luvisosls LM5 720 809 30

20

Kamwaa Sandy/Luvisols and Acrisols LM5 720 827b 32

19

a LM4: lower midland marginal cotton zone, LM5: lower midland livestock-millet zone (Jaetzhold and Schmidt, 1983).b Average rainfall at Ishiara.

262 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272

Page 5: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

4.2. Variability in soil fertility

In order to obtain a more detailed picture of soil

fertility in the study area, soil samples were taken in 20

representative farmers' ®elds in the study villages,

except Mutuobare. In each ®eld, three composite

samples of topsoil to a depth of 30 cm were taken.

Standard methods for tropical soils were employed for

the analysis of soil nutrients (Landon, 1984). Soil

nitrogen was not analysed due to technical problems

but can be inferred from organic C, the C:N ratio being

around 10:1 (Landon, 1984; Kiome and Stocking,

1993). The mean for each nutrient was calculated

for every ®eld.

Different sub-areas were classi®ed according to soil

fertility using discriminant analysis. Multigroup dis-

criminant analysis is a technique that uses classifying

functions to assign samples individually to different

groups, and is useful for determining whether several

groups are distinct (Davies, 1986). The objective is to

construct classifying functions that are linear combi-

nations of the original parameters so that the values of

the classifying functions for all groups are as different

as possible. The dataset contains organic C, Ca, K,

Mg, Mn, Na and P as original parameters. As pH and

nutrients are interrelated, this variable was not

included in the analysis. Three discriminant functions

were obtained, but only two functions were signi®cant

at the 5% level (Table 2). The ®rst two functions also

account for nearly all of the total variance (94.5%).

Discriminant function 1 was plotted against func-

tion 2 for all samples in the four groups (Fig. 3). For

function 1 there is an almost total overlap between

Mumburi and Kathuri, but a clear separation between

the other villages. Function 2 only discriminates

clearly between Karii and Kamwaa. Plotting these

two functions, three sub-locations can be inferred

Fig. 2. Variations in rainfall at Ishiara, the station with the longest rainfall record in the study area in Mbeere District. The graph depicts

annual rainfall and 5 year moving average for the April and November rains, respectively.

Table 2

Discriminant analysis of soil properties: group statistics associated

with the derived functions that are presented in Fig. 3

Function Percentage

of variance

Cumulative

percentage

Degrees of

freedom (df)

Significance

level

1 86.38 86.38 21 0.000

2 8.14 94.53 12 0.027

3 5.47 100.00 5 0.076

A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272 263

Page 6: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

based upon intrinsic soil conditions and related broad

fertility levels.

This multivariate analysis justi®es a spatial differ-

entiation of soil properties and a closer analysis of soil

fertility for the three sub-locations. The means for the

different monitored soil nutrients for the three identi-

®ed areas are presented in Table 3. Karii sub-location

was the most fertile area, being signi®cantly richer in

nutrient cations and organic C than the other areas.

Mumburi/Kathuri had the overall lowest ratings and

was the least fertile. Kamwaa was intermediate, but

had soils that were substantially more alkaline. The

spatial variability in soil conditions and potential

constraints to production were, therefore, signi®cant.

Karii had no appreciable current limitations and rea-

sonable nitrogen reserves as inferred by the organic

carbon; Kamwaa may suffer somewhat by high pH

and its interaction with nutrient availability; while

Mumburi and Kathuri were more generally de®cient

in most aspects and had a critically lower level of

Fig. 3. Plot of soil fertility discriminant functions for the study area in Mbeere District.

Function 1 � ÿ3:0Cÿ 0:04Ca� 25:9K� 3:6Mg� 2:6Mnÿ 48:6Na� 0:02Pÿ 2:0

Function 2 � 0:7C� 0:04Caÿ 3:5Kÿ 0:4Mg� 6:9Mn� 7:8Naÿ 9:6Pÿ 4:3

.

Table 3

Soil nutrient status in three sub-locations within the study area

Area Organic Ca% Cab cmol/kg Kb cmol/kg Mgb cmol/kg Mnc cmol/kg Nab cmol/kg Pd mg/kg pH-H2Oe

Mumburi and Kathuri 0.51 5.22 0.42 1.38 0.55 0.37 1.4 6.4

Karii 1.10 10.58 0.99 3.58 0.88 0.70 21.0 6.8

Kamwaa 0.72 11.10 0.52 2.00 0.44 0.42 16.4 7.5

aWalkley-Black method.bMeasured in unbuffered 1 M KCl.cHydroquine extraction.dOlsen method.e1:2.5 soil±water suspension.

264 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272

Page 7: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

organic matter, which would affect soil structure and

crusting.

4.3. Variability in farmer resource level

Cropping systems and farm management practices

are related to farmer resource level. Access to cash

income, farm size, livestock ownership, labour avail-

ability, input use and cultivation method, exposure to

external in¯uence, labour constraints, availability of

tools and sex of head of household are important

criteria in determining the resource level of house-

holds and ultimately their farming strategies and the

ISWC technologies they utilise (Table 4). In the PRA

exercises, three categories of farmers were identi®ed ±

high, medium and low resource farmers. In the house-

hold survey, farmers were categorised in a quick

appraisal based on use of draft animals and type of

housing and roo®ng (see Scoones, 1996).

A previous study of rural livelihood systems and

farm-non-farm linkages in Mbeere district between

1972/1974 and 1992/1993 revealed that almost every

Mbeere household is engaged in off-farm activities

and that 20% of adult males are absent for most of the

year (Hunt, 1995). However, many of these men remit

cash income to the household. External in¯uence in

the area also takes other forms. Non-governmental

organisations have been involved in numerous relief

and development activities, many women farmers are

members of women's groups and the Ministry of

Agriculture is present through the extension service.

An important factor differentiating high and low

resource farmers is the availability of draught animals.

High resource farmers also tend to have more

resources to spend on farm inputs, such as fertilisers

and pesticides. All farmer categories suffer from

labour constraints during peak periods. However,

the higher resource farmers normally have enough

cash to hire labour to offset the worst constraints.

4.4. Variability in farm productivity

Detailed monitoring of input use, including labour,

and outputs from different farm enterprises on each

contact farm enabled calculation of individual farmer

gross margins for four consecutive seasons (Novem-

ber 1995, April 1996, November 1996 and April

1997). An example from the November 1995 season

is shown in Table 5. This was based on market prices

for goods actually bought or sold and opportunity

costs for household supplied inputs or retained pro-

duce. Draft animal power and labour were valued at

local hire costs and retained produce at retail prices

Table 4

Farmer characterisation in Mbeere district (Source: PRAsa, survey and on-farm monitoring)

Resource level High Medium Low

Access to cash income Yes Some Very limited

Farm size Largest Smallest

Livestock nos Most (cattle and goats) Some (mostly goats) Nil

Tools Wide range (animal drawn

equipment, sprayers and

maintenance tools)

Good range (hand implements

and animal-drawn plough)

Only most basic (hoe)

Outside influence Most Some None

Labour constraints Some Some to severe Some to severe

Sex of head of household Mostly female (husband

remits income)

Mostly male Mostly female (widowed, aban-

doned, husband seeking work)

Input use Own draught animals Own or hired draught animals Little use of hired draught animals

Manure Some manure No manure

Crop chemicals Some crop chemicals No crop chemicals

Hired labour at peak periods Primarily family labour Family labour

Farming strategy Abandon land use in adverse

seasons

Labour trade-offs between

on-farm and off-farm

employment

Opportunistic: flexible labour re-

sponse

a PRA: participatory rural appraisal (see Altshul and Okoba, 1995 and Okoba and Altshul, 1995).

A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272 265

Page 8: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

obtained through regular market surveys (Okoba et al.,

1998). The analysis focused on the variability of

returns to land, labour and cash investment.

The November 1995 and April 1997 seasons had

rainfall close to average, whereas the April and

November 1996 seasons were well below average.

In November 1995 most farmers obtained positive

returns to land when costs for draft animals and labour

were excluded, the average for all farmers being 8360

KSh haÿ1. Similar ®gures for April 1996; November

1996; and April 1997 are 427, 145, and 7849

KSh haÿ1, respectively ± that is, most farmers experi-

enced a total crop failure in both seasons in 1996,

whereas in the April 1997 season, returns to land were

similar to the November 1995 ®gures.

The above data provided economic returns for two

seasons with close to normal rainfall and two seasons

with severe drought. However, it is also of interest to

establish the within season range between farmers. In

the two seasons with close to normal rainfall, returns

to land, excluding the costs of family supplied draft

animals and labour, ranged from 20 800 to ÿ1570

KSh haÿ1 and 26 660 to ÿ1180 KSh haÿ1, respec-

tively. Similar ®gures for the two low rainfall seasons

were 14 530 to ÿ1870 and 1130 to ÿ760 KSh haÿ1,

respectively. There are farmers that experience crop

failures even in years with normal rainfall. This can

either be due to labour constraints or an overall low

priority given to farming. The fact that some farmers

had reasonably good returns to land even in the April,

1996 season could possibly be due to factors such as

good timing of planting, choice of crop or fortunate

rainfall distribution. It is also apparent from the ana-

lysis that productivity levels do not allow costly

Table 5

Individual farm gross margins in the November, 1995 season according to on-farm monitoring in Mumburi, Kathuri, Karii and Kamwaa,

Mbeere District

SWCa Cropb Categoryc Value of

produce

Total costs

of inputs

Total costs

of labour

Total costs

of DAPd

Total costs

of SWC

Returns

to land

Returns

to lande

Returns

to cash

Returns

to labourNovember

1995 KSh/haf KSh/ha KSh/ha KSh/ha KSh/ha KSh/ha KSh/ha KSh/Sh KSh/day

FJ ma L2 15 000 309 2250 365 872 11 569 14 691 48 263

LSB ma H3 19 998 269 4678 536 398 14 118 19 729 74 242

SSB�TL gg�ma L1 13 475 179 5940 0 975 6381 13 296 75 117

SSB�TL mi L2 1764 535 7199 0 975 8355 16 529 32 112

SSB cp�mi L3 10 730 284 6075 0 405 3966 10 446 38 10

LL�SB mi L3 21 336 535 17 830 0 825 2146 20 801 40 66

FJ mi�ma H1 9000 2120 4260 884 872 1748 6880 4 65

FJ mi L1 10 664 530 5260 130 872 4002 10 134 20 61

LL mi L2 19 336 875 13 700 365 450 4312 18 461 22 58

None ma�be H3 10 039 1362 6615 1040 0 2062 8677 7 56

FJ mi L3 6480 646 6640 520 872 ÿ1677 5835 10 54

None gg�ma L2 1723 299 2160 106 0 ÿ843 1423 6 49

FJ ma�be L2 4463 812 4420 0 872 ÿ1641 3651 6 47

FJ ma�be L2 2925 929 3260 0 872 ÿ2136 1996 3 39

SSB ma�gg H2 6090 1489 11 569 0 405 ÿ7373 4601 4 31

SB�TL mi L3 8272 572 21 685 0 945 ÿ14 930 7701 14 22

LTL mi�ma H2 1800 705 5070 650 1320 ÿ5295 1095 3 12

FJ ma�be L1 1040 1304 9775 520 872 ÿ10 911 ÿ264 0 0

FJ ma�gg L1 0 549 5800 78 872 ÿ7221 ÿ549 0 0

SSB�TL mi�cp L2 0 1569 9530 0 975 ÿ12 074 ÿ1569 0 0

a Soil and water conservation (SWC) ± FJ: Fanya Juu; SB: medium size stone bunds; LSB: large stone bunds; SSB: small stone bunds; TL:

medium size trash lines; LTL: large trash lines; LL: log lines.b Crops ± ma: maize; gg: green grams; mi: millet; cp: cowpeas; be: beans.c Category ± L: low soil fertility; H: high soil fertility, 1, 2 and 3� high, medium and low resource farmers.d DAP: draft animal power.e Costs for draft animal power (DAP) and labour are excluded.f 1US$ �63.65 KSh in November 1997.

266 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272

Page 9: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

investments in SWC. The risk of negative returns

could be increased if the cost of SWC is high.

5. Agrodiversity

The environmental and socio-economic variability

described in the previous section results in highly

diverse farming systems and practices in Mbeere.

With these sources of variability noted, agrodiversity

was now analysed to identify the main discriminating

factors behind different cropping systems and man-

agement practices. First, cropping patterns and crop

output were examined followed by farming systems.

To take account of as many aspects as possible of

the variations in cropping and management patterns, a

number of different approaches were employed.

Detailed farm maps were constructed for three con-

secutive seasons for the 20 contact farmers in four of

the ®ve villages in Table 1. The maps indicated the

area planted with different crops and combinations of

crops as well as soil and water conservation practices.

The household survey, in contrast, focused on total

farm output of different crops and dominant soil and

water conservation practice on each.

5.1. Crops planted

In terms of crops and combinations of crops

planted, the November season had the highest degree

of diversi®cation with a total of 25 different cropping

systems (Table 6). When looking at both seasons,

medium resource farmers had the largest number of

crop combinations and the high resource farmers the

lowest. Other interesting contrasts between farmer

categories were found in the choice of crop. In the

November season, high resource farmers grew mainly

millet intercropped with a legume, mostly beans.

Medium resource farmers concentrated on intercrop-

ping maize and a legume, usually cowpeas and/or

green grams (Vigna radiata; Gibbon and Pain,

1985). Likewise, low resource farmers also preferred

intercropping maize with a legume. In the April

season, the dominant cropping system for high

resource farmers shifted to maize plus a legume or

a sole legume. Cowpeas was the preferred legume. For

medium and low resource farmers intercropping of

maize with a legume also dominated in this season, but

medium resource farmers favoured green grams and

low resource farmers cowpeas and beans. Moreover,

in this season, other crops, such as pigeon peas

(Cajanus cajan), tobacco (Nicotiana tabacum) and

cotton were also important (Gibbon and Pain,

1985), especially for the medium resource farmers.

As a strategy to reduce the risk of crop failure, crop

diversity was most marked in the medium resource

group. Risk averse behaviour was also seen with low

resource farmers, but their choices were more con-

strained by lack of cash and consumption of stored

crops and seed supplies, particularly following a

drought.

Table 6

Farm area cultivated with different crops for 20 farmers in Mumburi, Kathuri, Karii and Kamwaa, Mbeere district (Source: on-farm

monitoring)

Crops November season April season Both seasons

Farmer Category:a H M L All H M L All H M L All

Average area (ha) per farm 1.4 0.8 0.5 0.9 1.5 0.8 0.5 0.9

Percentage of total:

Maize/cowpeas 0 26 11 11 32 0 47 24 16 13 29 17

Maize/beans 6 24 5 13 2 6 27 8 4 15 16 10

Maize/green gram 16 7 28 15 2 46 7 18 9 27 18 17

Millett/sorghum/cowpeas 13 2 24 11 28 10 13 19 20 6 18 15

Millet/beans 42 2 1 18 0 0 0 0 21 1 1 9

Cowpeas 0 1 2 1 33 15 4 22 16 8 3 11

Millet/ sorghum/maize 18 33 24 25 0 3 0 2 9 18 12 13

Other 4 5 6 7 3 19 2 8 4 12 4 8

Total 100 100 100 100 100 100 100 100 100 100 100 100

Crop combinations 10 16 17 25 7 11 8 17

a H: high resource farmers, M: medium resource farmers, L: low resource farmers.

A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272 267

Page 10: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

5.2. Total farm crop output

In the household survey, farmers were interviewed

about crop output from their farms, use of draft

animals and labour input into different farming opera-

tions as well as soil and water conservation. Most

farmers experienced an almost total crop failure in the

November 1996 season and, therefore, when asked

about their latest harvest, reported yields from the

April 1996 season. Most farmers could recall their

total yields as far back as the November 1995 season

and some even further back.

The total farm crop output (Table 7) was analysed

according to area and farmer category using the non-

parametric Mann±Whitney U-test (Ebdon, 1985) ± for

details see Okoba et al. (1998). There were no statis-

tically signi®cant differences in yields between the

different areas for the two seasons with most data,

November 1995 and April 1996. However, when the

grouping variable was transformed to represent two

levels of soil fertility (high for Karii and Mutuobare

and low for Mumburi/Kathuri), the difference was

signi®cant at the 10% level for both seasons, with

higher crop output in the high fertility areas. Likewise,

there was a signi®cant difference in total crop output

between the different farmer categories at the 10%

level for the April 1996 season with low resource

farmers having the highest output. This interesting

observation, resource-poor farmers reporting the

highest yields, is also valid for the November

1995 season, although the differences were not sta-

tistically signi®cant. However, this can be explained

by the fact that a large proportion of the low resource

farmers live in a recently settled area with high soil

fertility.

5.3. Farming systems

A conclusion from the above analyses is that soil

fertility and farmer resource level are signi®cant

factors in determining cropping systems, farm output

and crop diversi®cation. To investigate this further, the

farming systems were grouped according to differ-

ences in farmer category and soil fertility.

Intercropping of maize and a leguminous crop

predominated in the April season for all farmer cate-

gories and both levels of soil fertility. In the November

season there appeared to be a difference in terms of

cropping strategies between areas with high and low

soil fertility, a feature, which was not revealed by the

contact farmers, but by the larger survey. In high

fertility areas, intercropping of maize and a legumi-

nous crop was still predominant, whereas intercrop-

ping of millet and a leguminous crop became the main

cropping system in areas with low soil fertility. Total

farm crop output varied both according to soil fertility

level and farmer category (Table 7), as well as

between years with high and low rainfall. Also the

proportional output of different crops varied according

to the two grouping variables, that is, farmer category

and soil fertility. In areas with high soil fertility, maize

accounted for the largest portion of all crops in both

seasons. Moreover, in areas with low soil fertility high

resource farmers grew more leguminous crops than

low resource farmers.

6. Indigenous soil and water conservationpractices

The large-scale environmental and socio-economic

variability revealed in this study and the resulting

Table 7

Total farm crop output in Mbeere district

Soil Fertility Farmer resource level Na April season November Season Yieldb (kg/ha) Yieldc (kg/ha)

High High 12 Maize�legume Maize�legume 350 1450

High Medium 5 Maize�legume Maize�legume 150 1250

High Low 11 Maize�legume Maize�legume 200 950

Low High 9 Maize�legume Millet�legume 50 600

Low Medium 8 Maize�legume Millet�legume 100 550

Low Low 2 Maize�legume Millet�legume 50 400

a Sample size.b Yield in low rainfall year.c Yield in normal rainfall year (data from high rainfall years is not available).

268 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272

Page 11: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

diversi®cation of farming systems were also re¯ected

in diversi®cation in other management practices, par-

ticularly soil and water conservation.

The most common techniques are trash lines, stone

bunds, Fanya Juus and log lines. Trash lines are

formed from crop residues that are placed in surface

strips that follows the contour. They are temporary

structures, laid seasonally, and are sometimes moved

from year to year in order to exploit trapped fertility

gains. Stone bunds are permanent structures of stones

also aligned along the contour that are common on

stony soils. They form a semi-permeable type of

barrier but with time an impermeable barrier develops.

Fanya Juu is a Swahili term meaning `make it up'. A

type of back slope trench is dug and soil from the

trench is thrown up slope to form a riser bank. The

trench by design is meant to be level on the contour,

trapping rainwater which is retained to percolate into

the ®eld below the ditch. Log lines are only found on

recently cleared land and are usually formed by logs

that are not suitable for charcoal production. When

new land is opened up, trees are felled two seasons in

advance of actual cultivation to allow drying out of the

logs. The ®eld is set on ®re during the season of

cultivation. The logs left unburned are then used for

making the log lines. For detailed technical descrip-

tions of the different structures see Altshul et al.

(1996) and Okoba et al. (1998).

When trash lines are combined with other struc-

tures, they are often superimposed on the other mea-

sure, but can also be placed in between structures. The

trash used is composed of crop residues from millet or

sorghum, but residues from maize can also be used if

they are not required for feeding livestock. Moreover,

there is a large range in dimensions and spacing of

structures, which depend on the availability of mate-

rial and tools as well as ®eld slope. The spacing of

structures ranges from 2.3 to 35.4 m and there seems

to be a negative exponential relationship between

spacing and ®eld slope (Altshul et al., 1996). Stone

bunds have the closest spacing, which is due to the fact

that stony soils are mainly found in the steepest parts

of the study area. This wide variation of structures and

systems is symptomatic of farmers' ability to experi-

ment and adapt technologies to their own environ-

mental circumstances and resource availability. For

example, in the case of Fanya Juus, ditches can be

used for compost making or planting of bananas

(Musa spp.; Williams et al., 1980), and ridges can

be planted with makarakari grass (Panicum color-

atum; Wrigley, 1981), which is used as livestock feed.

Different ISWC practices are used in different

cropping systems. For example, high resource farmers

in high fertility areas tend to grow more maize and

legumes and are also likely to have cattle. Such

farmers are unlikely to combine trash lines with other

practices because of the need to feed crop residues to

livestock. Instead, they would more likely look to crop

combinations of maize and legumes for both produc-

tion and conservation, or use stones or grass strips.

To further explore interactions between ISWC and

different biophysical and socio-economic farm char-

acteristics, an hierarchically structured decision-tree

was developed to combine the most important criteria

determining farmers' choice of ISWC. Each criteria

was only given two classes or levels, such as `yes' or

`no' and `high' or `low' (Fig. 4). This is, of course, a

simpli®cation as there for most criteria existed a

continuum of cases. For example, farmer resource

level can be everything between high and low and

soils can have varying proportions of sand and stones.

The decision-tree was veri®ed in four parts with the

assistance of the agricultural extension service in

Mbeere district. Four meetings were organised with

farmers living in areas with different natural settings ±

that is, old land and newly-opened land with stony and

sandy soils, respectively. The outcome of the decision-

making processes in Fig. 4, according to the different

groups of farmers, are presented in Table 8, but note

that ISWC option 25±32 are on mixed soils and not on

sandy soils as indicated in the simpli®ed decision-tree.

Not surprisingly, stone bunds dominated in stony areas

and Fanya Juus were only found in sandy areas. There

were also interesting differences between high and

low resource farmers in choice of ISWC practices.

Low resource farmers tended to choose cheaper and

less labour demanding techniques such as trash lines

and log lines and, in stony areas, construct smaller

stone bunds than better endowed farmers. The largest

diversity of ISWC practices was found on newly-

opened land with mixed soils.

7. Conclusions

The above analysis revealed highly complex and

diverse farming systems in semi-arid Kenya, where

A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272 269

Page 12: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

Fig. 4. Decision-tree on ISWC for Mbeere district. For ISWC options see Table 8.

Table 8

Soil and water conservation techniques resulting from decision-making processes are presented in Fig. 4

Old land and secondary bush land Newly-opened land (primary bush land)

Stony soil: Stony soil:

1. Large SB 17. Large SB�TL

2. Large SB 18. Large SB�TL

3. Large SB�large TL 19. Large SB�TL

4. Large SB�large TL 20. Large SB�TL

5. Large SB�small TL 21. Small SB�TL

6. Small SB�small TL 22. Small SB�TL

7. Large SB�large TL 23. Small SB�TL

8. Large TL 24. Shifting cultivation

Sandy soil: Sandy-stony-clay soil:

9. FJ ± planted with makarikari grass 25. FJ�TL, LL�TL and small SB�TL

10. FJ ± planted with makarikari grass 26. Small SB�TL, TL and FJ

11. FJ�TL 27. Small SB�TL and small FJ

12. FJ�TL 28. LL ± planted with makarikari grass, FJ and incorporation of mulch

13. FJ ± planted with makarikari grass 29. Incorporation of mulch, TL, LL and FJ

14. TL 30. Small SB, LL and fallow strips

15. FJ 31. TL, LL, small SB, check dams and fallow strips

16. TL 32. TL, LL, small SB and TL

Note that option 25±32 are on mixed soils and not on sandy soils as indicated in Fig. 4.

SB: stone bund, TL: trash line, FJ: Fanya Juu terrace, LL: log line, SB�TL: TL superimposed on SB, FJ�TL: TL superimposed on FJ,

LL�TL: TL superimposed on LL.

270 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272

Page 13: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

environmental as well as socio-economic variability

give rise to a wide range of land management strate-

gies, choice of crops and soil and water conservation

practices. The term `agrodiversity' captures this diver-

sity at a number of temporal and spatial scales.

The study has demonstrated that the identi®cation

of improved ISWC lies at the interface between land

management and cropping strategies. There is a need

for continuing research and experimentation on the

integration of cropping and SCW practices. Further-

more, the people of Mbeere are famous for their use of

indigenous plants, which has already been subject to

detailed investigation (see Riley and Brokensha, 1988;

Kidundo, 1997; Roothaert et al., 1997). However,

these studies focus mainly on the use of trees. A

closer look at crop biodiversity in the study area is,

therefore, indicated. How many varieties of different

crops are farmers growing and on what criteria do they

base their choice? How does choice of crop varieties

relate to overall land management, such as soil and

water conservation and fertility and pest management?

These are the sort of research topics in agrodiversity

that have components other than ISWC as the focus,

and which could prove valuable to the development

community.

Moreover, interventions in the area of soil and water

conservation must build on the existing agrodiversity

and an understanding of the complex interaction

between environmental and socio-economic factors

in giving rise to different farming systems and man-

agement practices. The present study could form the

basis for such an approach to soil and water conserva-

tion, whereby development of improved SWC builds

on indigenous technologies, and ¯exibility and diver-

sity are acknowledged as the most important proper-

ties. Thus, developing a range of options of land

management strategies that build on local practices,

that are ¯exible and cheap seems to be the way

forward. Indeed, this approach is not new and has,

for example, previously been advocated for promotion

of cropping systems in marginal areas in Africa (Okali

et al., 1994).

Acknowledgements

The research was carried out at Kenya Agricultural

Research Institute (KARI), Regional Research Centre

in Embu and was funded by the Department for

International Development (DFID). The input from

Barrack Okoba and Helen Altshul is gratefully

acknowledged. The ®rst author had a post-doctoral

grant from Wenner±Gren Foundations, Stockholm,

during the course of the study.

References

Altshul, H., Okoba, B., 1995. Report of a participatory rural

appraisal in Siakago division, Embu district, Kenya, 26 March±

1 April 1995 with emphasis on indigenous soil and water

conservation. ODA/SRI/KARI, p. 67.

Altshul, H.J., Okoba, B.O., Willcocks, T.J., 1996. On-farm studies

of water and soil conservation in Mbeere district, Kenya. In:

Willckocks, T.J., Gichuki, F.N. (Eds.), Conserve water to Save

Soil and the Environment. Proceedings of an East African

Workshop, Nyeri, May 1996. SRI (Silsoe Research Institute)

Report No IDG/96/15, pp. 82±93.

Brookfield, H., Padoch, C., 1994. Appreciating agrodiversity: A

look at the dynamism and diversity of indigenous farming

practices. Environment 36(5), 7±45.

Davies, J.C., 1986. Statistics and data analysis in geology. Wiley,

New York, p. 646.

Downing, T.E., Mungai, D.N., Muturi, H.R., 1985. Drought

climatology of central and eastern Kenya. Climatic variability

and agricultural production in central and eastern Kenya.

Ministry of Environment and Natural Resources, Nairobi,

Kenya, pp. 1±37.

Ebdon, D., 1985. Statistics in Geography, 2nd ed. Basil Blackwell,

Oxford, UK, p. 232.

Ellis, F., 1993. Peasant Economics: Farm Households and Agrarian

Development. 2nd ed. Cambridge University Press, Cambridge,

UK, p. 309.

Gibbon, D., Pain, A., 1985. Crops of the Drier Regions of the

Tropics. Longman Scientific and Technical, Essex, UK, p. 157.

Hunt, D., 1995. Rural livelihood systems and farm non-farm

linkages in Lower Embu, Eastern Kenya: 1972/1974 to 1992/

1993. Preliminary Survey Findings, With Special Reference to

Agriculture. Manuscript, Sussex University, Sussex, UK.

International Land Development Consultants B.V., Arnhem,

Netherlands (ILACO), 1981. Agricultural Compendium for

Rural Development in the Tropics. Elsevier, Amsterdam,

Netherlands, p. 739.

Jaetzhold, R., Schmidt, H., 1983. Farm Management Handbook for

Kenya, Vol IIC, Eastern Kenya. Ministry of Agriculture and

GTZ, Nairobi, Kenya, p. 207.

Kidundo, M., 1997. Participatory Technology Development.

Nursery Propagation of Melia Volkensii (GuÈrke): A Potential

Agroforestry Tree Species for Semi-arid Mbeere. M. Phil.

Thesis. Centre for Arid Zone Studies. School of Agriculture

and Forest Science, University of Wales, Bangor, UK, p. 194.

Kiome, R., Stocking, M., 1993. Soil and Water Conservation in

Semi-arid Kenya. Natural Resource Institute Bulletin 61,

Chatham, UK, p. 59.

A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272 271

Page 14: Applying the concept of agrodiversity to indigenous soil and water conservation practices in eastern Kenya

Landon, J.R., 1984. Booker Tropical Soil Manual. A Handbook for

Soil Survey and Agricultural Land Evaluation in the Tropics

and Subtropics. Longman, Essex, UK, p. 474.

Netting, R.McC, Stone, M.P., 1996. Agrodiversity on a farming

frontier: Kofyar smallholders on the Benue plains of central

Nigeria. Africa 66, 52±70.

Okali, C., Sumberg, J.E., Reddy, K.C., 1994. Unpacking a technical

package: Flexible messages for dynamic situations. Expl.

Agric. 30, 299±310.

Okoba, B., Altshul, H., 1995. Report of a participatory rural

appraisal in Gachoka division, Embu district, Kenya, 12±18

March 1995 with emphasis on indigenous soil and water

conservation. ODA/SRI/KARI, p. 81.

Okoba, B., Altshul, H., Tengberg, A., Twomlow, S., Ellis-Jones, J.,

1998. Conserve water to save soil and the environment:

An Evaluation of Indigenous Soil and Water Management

Systems in Mbeere District. SRI Report No. IDG/98/16,

p. 80.

Oldfield, M.L., Alcorn, J.B., 1987. Conservation of traditional

agroecosystems. Bio Science 37, 199±208.

Pretty, J.N., 1995. Regenerating agriculture: policies and practice

for sustainability and self-reliance. Earthscan (London), p. 320.

Reij, C., Scoones, I., Toulmin, C., 1996. Sustaining the Soil:

Indigenous Soil and Water Conservation in Africa. Earthscan

(London), p. 260.

Richards, P., 1985. Indigenous Agricultural Revolution: Ecology

and Food Production in West Africa. Hutchnison, London, UK,

p. 192.

Richards, P., 1986. Coping with Hunger: Hazard and Experiment in

an African Farming System. Allen and Unwin, London, UK,

p. 176.

Riley, B.W., Brokensha, D., 1988. The Mbeere in Kenya. vol. 1,

Changing Rural Ecology. University Press of America,

Lanham, p. 366.

Roothaert, R.L., Arimi, H.K., Kamau, E.N., 1997. Improving

indigenous fodder trees and shrubs in Kenya: Starting with

farmers' knowledge. AFRENA (Agroforestry Research Net-

works for Africa) Report, No 111, p. 27.

Scoones, I., 1996. Hazards and Opportunities. Farming Livelihoods

in Dryland Africa: Lessons from Zimbabwe. Zed Books,

London, UK, p. 267.

Swift, J., 1996. Desertification Narratives, Winners and Losers. In:

Leach, M., Mearns, R. (Eds.), The Lie of the Land: challenging

received wisdom on the African environment. The International

African Institute. James Currey, Oxford, UK, pp. 73±90.

Williams, C.N., Chew, W.Y., Rajartnam, 1980. Tree and Field

Crops of the Wetter Regions of the Tropics. Longman Scientific

and Technical, Essex, UK, p. 262.

Wrigley, G., 1981. Tropical Agriculture: The Development of

Production. Longman, Essex, UK, p. 496.

272 A. Tengberg et al. / Agriculture, Ecosystems and Environment 70 (1998) 259±272