LUCID’s Land Use Change Analysis as an Approach for Investigating Biodiversity Loss and Land Degradation Project

Multi-Scale Analysis of Land Use and Management Change

on the Eastern Slopes of Mt.

LUCID Project Wo rking Paper 20

By

Jennifer M. Olson with assistance of Bilal Butt, Fred Atieno, Joseph M. Maitima, Thomas A. Smucker, Eric Muchugu, George Murimi and Hong Xu

Department of Geography Michigan State University East Lansing, MI 48824 USA

March 2004

Address Correspondence to: LUCID Project International Livestock Research Institute P.O. Box 30709 , Kenya E-mail: [email protected] Tel. +254-20-630743 Fax. +254-20-631481/ 631499

Multi-Scale Analysis of Land Use and Management Change on the Eastern Slopes of Mt. Kenya

The Land Use Change, Impacts and Dynamics Project Working Paper Number 20

By

Jennifer M. Olson with assistance of Bilal Butt, Fred Atieno, Joseph M. Maitima, Thomas A. Smucker, Eric Muchugu, George Murimi and Hong Xu

Department of Geography Michigan State University East Lansing, MI 48824 USA

March 2004

Address Correspondence to: LUCID Project International Livestock Research Institute P.O. Box 30709 Nairobi, Kenya E-mail: [email protected] Tel. +254-20-630743 Fax. +254-20-631481/ 631499

Copyright © 2004 by the: Michigan State University Board of Trustees, International Livestock Research Institute, and United Nations Environment Programme/Division of Global Environment Facility Coordination. All rights reserved.

Reproduction of LUCID Working Papers for non-commercial purposes is encouraged. Working papers may be quoted or reproduced free of charge provided the source is acknowledged and cited.

Cite working paper as follows: Author. Year. Title. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper #. Nairobi, Kenya: International Livestock Research Institute.

Working papers are available on www.lucideastafrica.org or by emailing [email protected].

LUCID Working Paper 20 ii

TABLE OF CONTENTS

I. INTRODUCTION...... 1

II. METHODOLOGY...... 1 A. Site selection ...... 1 B. Land use change analysis ...... 3 1. Data acquisition and interpretation ...... 3 2. Land use and land cover classification ...... 5 C. Household and field level surveys ...... 5 D. Catchment field maps...... 6 E. Group and key informant interviews...... 6

III. LAND USE AND MANAGEMENT PATTERNS AND PROCESSES ...... 8 A. High elevation agriculture in Embu ...... 8 B. Low elevation, semi-arid Mbeere ...... 15 C. Income diversification and out-migration ...... 21 D. Effects of household composition and poverty on land use and management ...... 24 1. Widows ...... 24 2. The traditional group: husband present and working on own farm ...... 24 3. Female headed with husband working/ living away...... 25 4. Husbands present but work off-farm ...... 26 5. Summary of factors affecting soil maintenance and fertility ...... 27 E. Land and soil management variation across agroecological zones ...... 29 F. Regional land use and cover change ...... 33 1. Large scale private land holders ...... 34 2. The Government of Kenya...... 37 3. Small scale farmers and herders...... 38

IV. CONCLUSION...... 41

V. BIBLIOGRAPHY ...... 45

Annex 1. Land use and land cover classes for the Mt. Kenya region...... 47 Annex 2. Identification of land cover and land use classes...... 48 Annex 3. Population Table of Embu and Mbeere Districts 1969 to 1999 ...... 50

LUCID Working Paper 20 iii

MAPS

1. Study areas and transect location in Embu and Mbeere Districts, Kenya ...... 2 2. The eastern slopes of Mt. Kenya (2001 ETM+ draped over DEM) ...... 7 3. Land use change in northern Embu District from 1958 to 2001...... 10 4. Ndunduri and Kianjuki catchment maps and land use pies ...... 12 5. Land use change between 1961 and 2001 in the Kiritiri area of Mbeere District ...... 16 6. Kambita and Ivondo catchment maps and land use pies...... 19 7. Land cover in 1987 and 2001 of the eastern slopes of Mt. Kenya...... 35 8. Land use in 1987 and 2001 of the eastern slopes of Mt. Kenya ...... 36 9. Areas of significant land cover change on the eastern slopes of Mt. Kenya ...... 37

TABLES

1. Climate in the agroecological zones of Embu ...... 8 2. Land use change statistics, northern Embu site ...... 11 3. Climate in the agroecological zones of Mbeere...... 15 4. Land use change statistics, Kiritiri area, Mbeere District ...... 17 5. Changes in soil fertility and percent fields with inputs...... 31 6. Land cover change statistics, eastern slopes of Mt. Kenya...... 33 7. Population of Embu and Mbeere Districts from 1969 to 1999 ...... 40 8. Pattern of land use and economic change in lower, semi-arid zones...... 42 9. Major driving forces of land use change...... 43

FIGURES

1. Percent of husbands and adult sons conducting non- and off-farm work by farm size...... 22 2. Percent of husbands and adult sons conducting non- and off-farm work by agroecological zone ...... 23 3. Factors affecting soil maintenance and change in soil fertility...... 27 4. Household agricultural wealth affecting soil fertility and manure application...... 28 5. Percent of fields where inputs were applied the previous season ...... 30 6. Bar graph of land cover changes between 1987 and 2001...... 34 7. Population growth of Embu and Mbeere from 1969 to 1999 ...... 40

LUCID Working Paper 20 iv I. INTRODUCTION 1 Mt. Kenya, the second highest mountain in Africa, is the spiritual home of many Kenyans. It is located in the centre of the country, only a few hours from the capital city Nairobi, and has played a central role in Kenya’s history, economy and culture. It also provides a unique view of the ecology of Kenya and East Africa because it spans an altitudinal and land use gradient from the glaciers at the peak, to bamboo and tropical montane forests, to farms of tea, coffee and maize and finally to the semi-arid savannah. Population densities range from over 700 people/km2 in the higher elevation zones to 20 people/ km2 in the lower zones, only 100 kilometres apart. This diversity of ecologies and landscapes, combined with rapid economic and social change, has led to land use and cover change that vividly reflects evolving social and environmental patterns and processes.

The purpose of this study was to identify the spatial patterns and driving forces of changes in land use and land management practices since the 1950’s on the eastern slopes of Mt. Kenya, especially in the human managed landscapes. A multi-scale, historical approach was adopted to determine the impact of economic, political, social and demographic forces on the agricultural systems. A variety of research methods were employed including interpretation of aerial photographs and satellite images, household surveys, field level mapping, focus group interviews, and analyses of other environmental and socioeconomic data and literature. This information on changing land use and management is linked to field-level data collected on soil characteristics and plant species diversity presented in Land Use Change, Impacts and Dynamics (LUCID) project working papers 9 and 32 (Gachimbi 2002; Maitima et al. 2004).

The study was based on a political ecology conceptual framework, which proposes an approach to identify the broad societal context of changing environmental management and the effects of environmental change on land managers. Political ecology emphasizes the interaction between political, economic and social events and process at multiple scales, from the international to the household. These studies pay particular attention to societal differentiation, such as by wealth, power and gender, to understand how and why the system is changing, and how the changes affect people (Blaikie 1985; Blaikie and Brookfield 1987; Campbell and Olson 1991; Lambin 1997; Zimmerer 2000). Multiple methods and scales of analysis, as a combination of quantitative and qualitative information, are common.

In this study of the slopes of Mt. Kenya, the spatial analysis revealed that major land use conversions have occurred throughout the region in the past 50 years, changes that mirror the rapid evolution of the social and economic system since independence. The changes varied by zone and by time period. In some periods, particularly following a modification of government land policy, the changes occurred swiftly and were accompanied by major alterations of the economic basis of the farming systems. In other periods, the land use and cover evolved more gradually, reflecting economic and demographic process, but nonetheless covered a large area and had equally important social and environmental effects. Between the 1950’s and the present, the only land that had not experienced a land use/ cover conversion was part of the mountain’s protected forest and in large ranches. Otherwise, from the high potential humid to the semi-arid savannah zones, the land use and cover had been altered.

II. METHODOLOGY A. Site selection A goal of the LUCID project is to examine changing land use and management and their root causes in a human managed landscape, especially in agricultural and agro-pastoral areas.

1 Sections of this paper were previously presented by the lead author at the Association of American Geography annual meetings in 1997, 1998 and 2004, and in various papers (see bibliography).

LUCID Working Paper 20 1 StuStuddy ArArea Locatitionons EmEmbbuu aannd Mbeere Dissttricts

1

Tharaka-Nthi District

#Mbubori # # Ki ri ari Kianjokoma Gekou # 2 Kavutir i # Kai ruri # Runyenjes # # Manyata Ke v ot e # Ciangera # # # Ki ri gi Ishiara Nguviu # # Ma k e n gi

Ka t ha n ga ri # Mutunduri # # Ka t am a # Karingari Kieri ri #

4 Gi thegi 3 # Gi thimu #

# Ena Itabua Riandu # # EMB U #

5 District # Kiamuringa Kirinyaga # Si akago # Ndaguma # District Muchonoke # Kev oe Gac hok a # # # Itir a Ngenge # Mbi t a

6 Ngandure # Ki ri t ir i # 7 # Kar ii Gac hur ir i # # Iriamurai Karaba Chief's Camp

Makutano # # Ngegiri Kar aba # # # Machanga # Kamwendei Ma v u ri a Riachina # #

Ma r ik i #

# Riakanau District Study areas (catchment) Protected forests Infrastructure Ndunduri catchment 1 Mt. Kenya Forest Roads Kianjuki catcment 2 Kirimiri Forest # Central Places

Kambita catchment 3 Njukia-ini Forest Aerial photos interpreted Ivondo catchment 4 Maranga Forest Veg. and Soils Transect Protected areas (national parks & reserves) 5 Kiang'ombe Forest N Mt. Kenya national park 6 Kianjiru Hills 7 Mwea national reserve Kiambere Forest Kilometers 01020

Map 1. Study Areas and Transect Locations in Embu and Mbeere Districts, Kenya

LUCID Working Paper 20 2 Ecological gradients, such as found on mountain slopes, provide examples of a variety of farming and economic systems in a small area, and often are characterized by interactions between the gradient zones. Mt. Kenya encompasses such diversity. The temperature and rainfall differences are large, due to elevation and orographic rainfall. The differences range from the upper elevations being cool and very humid, supporting a tropical montane forest, to as one descends, changing to zones characterized by tea, then coffee, maize, sorghum and finally the hot, semi-arid savannah used for grazing animals that surrounds the mountain. This study is concerned with human-managed landscapes below the forest. See Vanleeuwe et al. 2003 for a study of changes within the forest.

Most of the study area is characterized by small holder, private farms which are the most common agricultural land use in Kenya. Detailed social and environmental analyses were conducted in the small holder farm area along a transect that ran in the Embu and Mbeere Districts from the forest edge to the dams along the Tana River (Map 1). Higher resolution land use change interpretation, household surveys, group interviews and soil and vegetation data collection were conducted following the approach outlined in the LUCID project’s methodological guide (Maitima and Olson 2001). In addition to small holder agriculture, large farms and ranches are found in the Timau area north of the mountain, and a parastatal tea plantation and a rice scheme are also present. Government of Kenya protected areas are also present, notably the Mt. Kenya Forest Park and Reserve, Imenti Forest Reserve, Meru Park, Mwea Reserve and smaller woodland and forest reserves. Land use change of the entire region of the eastern slopes of Mt. Kenya was also examined, and other project Working Papers present land use change and other research results from Meru and Tharaka Districts (Smucker 2002; Otuma 2004).

B. Land Use Change Analysis 1. Data Acquisition and Interpretation The study used a combination of aerial photographs for high resolution interpretation of small areas and to obtain older coverages, and satellite images that covered wider areas and different time periods.

The aerial photographs had been taken at irregular times and at varying resolutions because the photographs had been ordered for specific purposes; for example, to prepare topographic maps, for land registration, or to map the Mt. Kenya forest. The 1961 and 1982 photographs of the area in Mbeere (Map 5) were obtained from the Survey of Kenya with permission of the Kenya Department of Defence. The 1958 and 1985 prints were similarly obtained for the area in Upper Embu bordering Mt. Kenya Forest (Map 2). The prints were scanned and the files mosaiced and georeferenced to the digitized roads and rivers from 1:50,000 Survey of Kenya topographic maps.2 Visual interpretation of the photographs was placed onto plastic transparencies covering the photos. This interpretation was digitised on-screen over the mosaiced photos.3

The satellite imagery interpretation presented in this study was mostly derived from Landsat images of various dates. The eastern slopes region was interpreted at a 1:100,000 scale using a 25 February 1987 Landsat TM image and a 21 March 2001 Landsat ETM+ image. The 2001 image was also analysed at a higher resolution for the smaller Kiritiri and Mbuvori sites. A panchromatic image was captured by SPOT Image in November 1995. This was interpreted

2 The topographic maps at 1:50,000 were published by the Survey of Kenya, Government of Kenya. They were originally prepared in the 1950’s and 1960’s and have since been updated. 3 The aerial photographs were interpreted with Eric Muchugu and Georges Murimi, and the SPOT image with Jacques Imbernon, all at the time at the World Agroforestry Centre (ICRAF). Hong Xu, then at Ohio State University, assisted with re-interpreting the photos and analyzing the catchment data. The 1987 image was interpreted by Fred Atieno, and the 2001 image including Map 2 by Bilal Butt, both with ILRI. Thomas Smucker, then at Michigan State University, conducted group interviews in 2002.

LUCID Working Paper 20 3 visually on-screen for the Kiritiri and Mbuvori sites. The images and photographs were georeferenced using topographic maps and projected to UTM Zone 37 South. Interpretation of the satellite images was done using visual interpretation and vector digitizing on-screen at the 1:100,000 scale. The interpretations were corrected during ground truthing tours in 1996 and in 2002, by consulting experts of the area, and by referring to secondary sources including topographic and other maps. Details of the approach used to interpret the 2001 satellite image and to conduct the ground truthing, as well as a critique of how the study site characteristics affected the interpretation, are presented in Lucid Working Paper 16 (Butt and Olson 2002). A similar method was followed for interpreting other years’ images.

The quality and resolution of the photographs and images varied, as summarized below: 1. 1958 black and white aerial photographs at 1:25,000 (Mbuvori, Upper Embu). The quality of the prints was not very good, due to poor preservation, but the original photographs were of high quality with good contrast and details such as clusters of buildings clearly visible. 2. 1961 black and white aerial photographs at 1:50,000 (Kiritiri, Mbeere). The photos are of poor quality due to poor preservation of the negatives. The edges of the photos are darkened. Some detail, such as houses and individual trees, is visible under magnification. 3. 1982 black and white aerial photographs at 1:30,000 (Kiritiri, Mbeere). The photos are of very high quality with good detail and good contrast. Houses, trees, plough lines, and plot boundaries are visible. 4. 1985 black and white aerial photographs at 1:20,000 (Mbuvori, Upper Embu). The photos are of very good quality with good detail and good contrast, and without haze or clouds. Roads, farm and field boundaries, and even houses can been seen easily. 5. 25 February 1987 Landsat image at 30 meter resolution. The quality is good, with little cloud cover or haze except in one corner. The resolution and spectral differences are sufficient to identify boundaries between major land covers, such as between forest and cultivated areas, and to distinguish intact woodlands or forest from degraded woodlands or forest. The resolution however is not sufficient to visually distinguish areas outside of protected areas in the savannah that have scattered plots of cultivation versus those plots that only used for grazing. Automated classification produced “spotty” results that are difficult to interpret. 6. November 1995 panchromatic SPOT image at 10 meter resolution, 1:50,000. The image is of good quality with no haze and few clouds. Detail is not visible (houses, trees etc.) but farm boundaries in the lower zone are visible. Interpretation was conducted and digitized on-screen. Ground truthing was done in 1996. Later, the interpretation was refined in a few places where the class was unclear by using results of a supervised classification (three classes) based on 2 to 3 training sites per class, and by comparing the image with the 1985 mosaic. 7. 21 March 2001 Landsat EMT+ at 30 meter resolution (spectral bands) and 15 metre resolution (panchromatic band).Ground truthing was conducted in 2002. The quality of the image is good, with little haze and few clouds. The image is excellent for visual interpretation at the 1:100,000 scale to differentiate cultivated from natural vegetation, irrigated from rainfed agriculture, and between types of natural vegetation. In the small area of Kiritiri and Ndundori, visual interpretation was conducted at 1:30,000 using the panchromatic band. In semi-arid Kiritiri, farm boundaries are visible in the panchromatic band making it possible to differentiate farms from spectrally similar bush. Supervised and unsupervised classification schemes confused areas that are spectrally similar, such as grain crops, fallow, grassland and bush, whereas visual interpretation using texture and shapes was more useful. Using such a variety of sources of land use information presents difficulties when results are later compared. Cultivation mixed with bush in semi-arid areas provided the most difficulty in distinguishing boundaries between land covers and uses because of spectral and often textual similarity between crops and bush, and because small cropped fields are interspersed with bush. The percentage of cover under crops is a continuum that varies across the landscape (it was classes as “mixed crops and bush” in the lower resolution interpretations). Comparing high resolution interpretation of the semi-arid areas from the 1960’s to 2001, the approach changed from delimiting polygons of farms within the wider bush to delimiting bush

LUCID Working Paper 20 4 patches within the farmland as large extents of bush were converted. The actual amount of land being cultivated, however, is actually less than the amount classified as farmland, as seen in the highest resolution land use interpretation based on field surveys (see Map 6). After land adjudication in 1986 in the Kiritiri area, many farmers cleared the bush and fenced their new farms (e.g., with thorn or euphorbia triculi living hedges) to ensure their claim, to clarify the boundaries, and to prevent others from cropping or grazing on their land. Once cleared and fenced, however, many fields were used for grazing and/or trees were planted on them, so the land reverted to grassland or secondary bush. The high rate of out-migration by men and the low value of the crops compared to labour costs meant that farmers did not find it economical to plant areas much larger than needed for home consumption. The grassland or secondary growth within fields, therefore, is difficult to distinguish in remotely sensed information from other bush.

2. Land Use and Land Cover Classification A dual land use and land cover classification scheme was developed for the interpretation of the entire site (1987 and 2001 images). Since the Lucid project is interested in changes in the biophysical characteristics of the landscape, it was necessary to distinguish between degraded and intact forest, for example, which would be captured in a land cover classification (their land use would probably remain “forest reserve”). The project is also interested in identifying the socioeconomic root causes of land use change and projecting future changes, however, so it was necessary to distinguish between areas of different types of land ownership and management, a differentiation that could be captured in a land use classification. Both land cover and land use classifications, therefore, were used in this study.

A hierarchical system of nomenclature was adopted. The land cover scheme was adapted from the Food and Agricultural Organization of the United Nations (Latham, 2001) and the Biosphere-Atmosphere Transfer Scheme (BATS) classifications (Dickinson et al. 1996). The FAO land use classification scheme was refined to include land ownership and management variations. Polygons were also provided land use designations that reflect land ownership or management, such as protected area status or whether the agriculture was large or small scale. The land use class designations were assigned using information on ownership or management derived from topographic maps and other maps or GIS layers, ground truthing and expert knowledge. Where information from the maps or GIS layers differed (e.g., whether a forest was unprotected or a Forest Reserve), the topographic maps were used as the final arbitrator.

The land use and cover classification, and information on how the classes were identified on the photos and images, are provided in Annexes 1 and 2. Also see Butt and Olson 2002.

C. Household and Field Level Surveys4 Household and field level surveys were conducted in 1996 in a random cluster sampling framework in eight Embu and Mbeere communities located along the transect. Two communities are in the tea agroecological zone (UM1); two in the main coffee zone (UM2), one in the “marginal cotton” zone (LM4) and three are in the midland livestock-millet zone (LM5) as characterized by Jaetzolt and Schmidt (Jaetzold and Schmidt 1983). Four of those communities had been delimited by the Ministry of Agriculture’s Agricultural Extension Service as soil and water conservation catchments, where the Ministry was organizing a terracing programme. The other communities were chosen to be comparable to the catchments, and were within a few kilometres of the catchments.

4 The lead author was with the World Agroforestry Centre (formerly ICRAF) and the Kenya Agricultural Research Institute when the surveys were conducted. The surveys provided information on perceptions of land degradation, and land use and management.

LUCID Working Paper 20 5 Approximately 20 households were surveyed in each community, chosen randomly from comprehensive lists of households that had been prepared by the Chiefs or, in the lower zones where such lists were unavailable, by the headmasters of the local schools. In total, 80 households were surveyed in Embu District and 88 were surveyed in Mbeere District. The household-level questionnaires, translated into Embu and Mbeere languages, were designed to provide information on the household’s history in the area, out-migration, their land tenure status, and basic wealth, ethnicity, and labour resources information.

In addition to the household level questionnaire, three field-level questionnaires were completed for each household. A total of 238 fields were surveyed in Embu and 260 in Mbeere. The enumerator and household member walked to three fields managed by the household (one near to the homestead, one of a medium distance, and one far) and the enumerator collected information including current and prior land use, tenure status, agricultural inputs, and perceptions of soil characteristics. versions of the household and field questionnaires that were used are in the annex of Lucid Working Paper 15 (Maitima and Olson 2001). The analyses were conducted in SPSS for Windows version 11.0.1.

D. Catchment Field Maps In addition to the surveys, detailed field level maps of four catchments were prepared by ICRAF and Ministry of Agriculture staff in 1994 to 1996. A Ministry of Lands survey map was used as the base map and geo-corrected with GPS. Field assistants collected information on each field including type of crops grown, number and location of trees, soil conservation techniques, grass or shrub lines, and buildings.

E. Group and Key Informant Interviews Two types of group interviews were conducted. The first was basically information-gathering. In most of the communities where the surveys were conducted, groups of men and women met separately in 1996 and in 2002 to discuss in a semi-structured format the key land use issues of the study: the history of land use of the community, how and why land use and management was changing, how soil management and soil productivity differed between groups, and how they saw the future for themselves and their children. The second round of group interviews were feed-back seminars in 2004, during which the results of the research were presented and implications for the area’s agriculture, soils and livelihood system were discussed.

A variety of key informants were also interviewed in Embu and Mbeere. Semi-formal interviews were conducted of bee keepers and herbalists in 2001 and 2002 of their perceptions of changing native plant species distribution, and the impact if any of the change on their activities. A series of interviews of older people was conducted in 1997 concerning the land use history of their area. Other interviews of staff of the Ministry of Agriculture, District Headquarters and elsewhere were also conducted.

LUCID Working Paper 20 6 A

B

Map 2. The Eastern Slopes of Mt. Kenya (2001 ETM+ false colour composite draped over DEM). “A” is the land use study site in northern Embu District (Map 3), and “B” the land use study site in the Kiritiri area of Mbeere District (Map 5). Change of the entire area is shown in Maps 7-9.

LUCID Working Paper 20 7 III. LAND USE AND MANAGEMENT CHANGE PATTERNS AND PROCESSES Land use of the area was examined at three scales: 1. the regional scale of the eastern slopes of Mt. Kenya (change between 1987 to 2001), 2. the landscape scale in two areas, one in the highland coffee/tea zone in Embu District, and the other in the semi-arid lowlands in Mbeere District (change from 1950’s to 2001), and 3. the field level at four very small catchment sites along the agro-ecological gradient. Different changes and types of changes are visible at the various scales. Most land use change occurs at the field level as farmers decide, for example, to clear a field of bush for planting, to rotate crops, to plant trees, or to convert a cropped field to pasture or to a woodlot. Cumulatively these individual decisions have constituted most of land use change of the region, and these are described at the landscape and field levels.

First examined are the changes that have occurred in the high elevation, coffee and tea zone in Embu District since the 1950’s, and secondly changes in the low altitude, semi-arid livestock- millet zone in Mbeere District. This is followed by an examination of household and agroecological factors affecting land use and land management in both places. Finally, a wider view of changes in the Mt. Kenya region places the local landscape analyses into perspective. The impact of large scale land managers, particularly governmental and parastatal institutions, is evident at this scale.

A. High Elevation Agriculture in Embu Embu District covers the glaciated peak of Mt. Kenya, 5,200 meters high (17,780 feet), bamboo and tropical montane forest of the Mt. Kenya National Park and Forest Reserve, and the small, intensely cultivated tea, coffee and maize farms. The agro-ecological zones range from the tea-dairy zone to the marginal coffee zone (see Table 1). The eastern side of Mt. Kenya receives large amounts of orographic rainfall, and Embu has two rainy seasons per year, each sufficient to support maize production. The area is characterized by small hills with steep crests and narrow valleys. The soil is fertile to moderately fertile (Gachimbi 2002).

Table 1. Climate in the Agro-Ecological Zones of Embu Agro-Ecological Zone Altitude in m Annual mean temp Annual mean in °C rainfall in mm LH-1 Tea-Dairy 1,770 – 2,070 17.7 – 15.8 1,750 – 2,000 UM 1 Coffee-Tea 1,590 – 1,830 18.9 – 17.5 1,400 – 1,800 UM 2 Main Coffee 1,400 – 1,590 20.1 – 18.9 1,200 – 1,500 UM 3 Marginal Coffee 1,280 – 1,400 20.7 – 19.6 1,000 – 1,250 Source: Jaetzold and Schmidt 1983.

People who are the ancestors of the migrated into and settled in the area in the 15th century, and established a farming system based on sorghum, millet, and root crops (Fedders and Salvadori 1980). In the 1950’s, when the first aerial photographs were taken, the farming system was characterized by extensive grazing of grassland with small areas of shifting cultivation. The Embu people had developed a system of allocation of certain hills for cultivation and others for grazing while the riverbanks and the remaining hills were reserved as woodlands.5 Clan elders made the decisions on who was to cultivate what land. This system had served the community well because it separated the crops from grazing animals, ensured sufficient land for maintaining large goat herds, and provided sufficient woodfuel and other forest products. In the cultivated areas, soil productivity was maintained by long fallows.

The rural population density had, however, already begun to noticeably increase due to declining infant mortality rates. The system of shifting cultivation had endured, however. Old

5 Where no citation is provided, the narrative concerning the history of the area comes from interviews of older people or from group interviews of men and women.

LUCID Working Paper 20 8 farmers recall that difficulties were arising such as the need to walk several miles to find free land to crop. Production was primarily for subsistence and local trade even after the colonial economic structure was established (Kitching 1980), although some limited coffee production was permitted (Haugerud 1983).

The 1950's were also a time of civil unrest in Kenya with the Mau Mau revolt in full force. The Mau Mau revolt started as a protest against the British colonial government’s appropriation of 75 percent of the Kenyan Highlands for white settler farms (Kanogo 1993). The Embu people had not been directly affected by the appropriation of land because their area had been designated as a “native reserve,” but large estates had been carved out of Kikuyu land on the north and northwest side of Mt. Kenya and immediately southwest of Embu in and Machakos (Odingo 1971). The Mt. Kenya forest including the section of forest adjoining Embu was an important base of operations for the Mau Mau, and Embu people became closely involved in the revolt by joining the movement or by supplying the fighters with food and information. According to older Embu people, they aided the Mau Mau because otherwise the Mau Mau would forcefully retaliate and because the fighters were “our sons, our brothers, fighting for freedom:” “In the emergency, lots of men here were imprisoned in Gachoka, those accused of having taken the oath. We had a good harvest during the emergency; the main activity was to get food to the Mau Mau. Women walked up with the food to the forest. We were free to graze and cultivate but were forced to do communal work so that the colonial government could check if the men and women were around and didn’t join the others [Mau Mau].

The British responded by beating and imprisoning men suspected of aiding or having joined the Mau Mau, and, from around 1952 to 1959, forcing the entire Embu population to live grouped in “emergency villages.” These newly created villages were basically prison camps to control the population. The forced labour dug trench along the forest edge (to keep the elephants and Mau Mau from descending), and the emergency villages are clearly visible in the 1958 air photographs (villages are classed as “urban”, see Map 3).

When the emergency ended around 1959, Embu families were allowed to leave the emergency villages to return to their homes one by one, as a family was declared “clean.” A colonial administration land tenure ordinance had been implemented, however, that completely altered where each family could live and what land they could crop and graze. This ordinance, based on the Swynnerton Plan, was designed to increase land productivity and cash crop sales through privatization (Swynnerton 1954). In Embu and elsewhere around Mt. Kenya, it was also to develop a stable middle class to quell dissent and opposition (Haugerud 1983, 1993). It encompassed the entire Upper Embu region. No land was set aside for communal grazing or woodlands. Every hill was subdivided into permanent, private family farms. Legal boundaries were established separating the newly titled farms. In the areas surveyed, the sizes of plots ranged from 2 to 5 acres with 3 acres (1.2 hectares) being common (although clan elders often arranged to have somewhat larger land holdings).

The adjudication of land to individual families was, in a way, a grand experiment. Unlike most settlement schemes, the land being subdivided was not unclaimed or empty but already settled, by the same farmers who were now receiving their individual plots. This huge shift would probably not have been possible during peace time, according to older people. It was made possible by the forced abandonment of the land when families moved to the emergency villages, and by the farmers’ continued fear of the colonial government. By independence in 1963, the adjudication plan had been fully implemented and all families lived on and cultivated only their individual plots.

LUCID Working Paper 20 9 1985 1958

Land Cover Types Afro-Montane Forest Bush Shamba system Tea Plantation Tea/Coffee Urban 2001

1995 Projection:UTM Zone 37 South Datum: WGS 84

Map 3. Land use change in northern Embu District (Mbuvori) from 1958 to 2001.

LUCID Working Paper 20 10 1958 1985 1958-1985 1995 2001 1985-2001 Land use km2 % km2 % km2 % km2 % km2 % km2 % Forest 11.4 17.5 11.1 16.9 -0.4 -0.6 12.3 18.8 13.5 20.6 2.5 3.8 Bush 23.1 35.4 0.0 0.0 -23.1 -35.4 0.0 0.0 0.0 0.0 0.0 0.0 Shamba system 0.4 0.6 0.5 0.7 0.1 0.1 0.8 1.3 0.9 1.4 0.4 0.6 Tea Plantation 0.0 0.0 1.2 1.9 1.2 1.9 1.9 2.8 3.1 4.7 1.9 2.8 Tea/Coffee small farms 29.8 45.6 52.6 80.2 22.7 34.6 50.1 76.5 47.7 72.9 -4.8 -7.3 Urban 0.6 0.9 0.2 0.3 -0.4 -0.6 0.4 0.6 0.3 0.4 0.1 0.1 TOTAL 65.4 100 65.5 100 65.5 100 65.5 100 Table 2. Land use change statistics, northern Embu site (Mbuvori)

In the shift from clan controlled communal land to individual land holdings, an apparent shrinkage of available land occurred. Older people in Embu mentioned that they suddenly felt land constricted, felt especially in the loss of grazing land and an inability to fallow: “... No one could refuse… Before demarcation, we felt no land shortage. Only after demarcation did we feel it. It is like nyama choma [grilled meat] — when you see the whole goat it looks like allot. After you cut it in pieces it doesn’t look like much.”

In reality, the amount of land controlled by a clan had not changed, but individuals lost their ability to move spatially and temporally between ecological gradients and land covers. The perception of land shrinkage was sorely felt. The average three acre (1.2 ha) land holding was not then, and is certainly not today following sub-division of farms with inheritance, sufficient to support a family. Some older farmers still regret losing the old system of shifting cultivation because they feel it provided a fairer distribution of land. With the move to private land holdings and the new emphasis on land as wealth, the differences in wealth between families became more visible and seemed unalterably frozen. Haugerud (1983) found that despite the initial somewhat egalitarian distribution of land, informal land rights, power differentials and non-farm opportunities eventually led to wealth differences.

Despite these misgivings, individual land holdings have since spatially bounded the evolution of the farming system. Goat, sheep and cattle herds drastically shrank, and people became dependent almost entirely on their crop production. Long fallowing, the previous method to maintain soil fertility, became impossible to practice. After a lag period in the 1960’s during which soil fertility declined sharply, farmers said that they began to invest labour and money in their soil with organic and inorganic fertilizers, soil conservation measures, and crop rotation. The export crops of tea and coffee became the mainstay of the local economy and of household income. Since independence, the government has encouraged Embu farmers to plant coffee and, more recently, tea by providing extension services and supporting parastatals. By the 1970’s, most fields with these cash crops received manure and fertilizers, and terracing coffee fields was mandatory if not necessarily well maintained (Berlekom and Larsson 1984). Coffee and tea, that provide permanent vegetative cover and that are treated with chemical fertilizers, pesticides and fungicides, now cover approximately 50 percent of the land in the tea zone and 30 percent in the main coffee zone (see Ndunduri and Kianjuki catchments, Map 4). In the tea zone, particularly, farmers purchase almost all of their food and are extremely dependent on the marketing system of the parastatals and cooperatives for their income.

LUCID Working Paper 20 11 not cropped

coffee woodlot 13%

25% fallow, grass 10%

11% other food 25%

13% tea

maize

N

0.2 0 0.2 0.4 Kilometers

not cropp ed

13% coffee grass, woodlots

29% 10%

other food 7%

41%

N mai ze

100 0 100 200 Meters

Map 4. Ndunduri catchment (top, on forest edge in LH1 Tea-Dairy Zone) and Kianjuki catchment (bottom, in UM 2 Coffee Zone): land use maps and pies of percent of land under different uses. Yellow is maize, red is coffee, purple is tea, olive is other food crops, cyan or green is grass or woodlots, and grey is not cropped. Field boundaries are outlined.

LUCID Working Paper 20 12 Tree planting, especially of grevillea robusta, became widespread to provide woodfuel, construction materials, and fruits and nuts to sell— trees now cover approximately 8 percent of the land (recorded as points, not illustrated in the above maps). Fields except tea are usually planted with a mixture of various crops—over 30 main crops were surveyed. Field boundaries are also usually highly managed, with planted or maintained trees, vines (e.g., passion fruit, yams), forage grasses and bushes. The landscape is extremely heterogeneous and the field sizes tiny, an average of 0.2 acres in Kianjuki.

Government and non-governmental support programmes also led to a revival of cattle raising. A surprising high 75 per cent of households raise a cross-bred dairy cow fed with grass and tree leaves grown on-farm and with purchased concentrate meal. The milk is a commercial product and part of the income diversification strategy. Embu farmers, therefore, have developed a farming system specialized in high value commercial products while they purchase much of their food. Their commercial success is undoubtedly helped by their proximity to the capital city Nairobi, two hours away along a paved road. Today, the intensification of Embu agriculture appears successful in terms of high soil and commercial productivity and general well-being.

Population increases since the original adjudication of the land have led, however, to second and third generation sub-division of farms to sizes insufficient to support a family. Indeed, many parents are refusing to sub-divide their land even after their sons marry, so their sons control only small portions of the farm. Average land holdings per adult male, at 0.7 ha., are therefore lower than average total farm sizes of 1.3 ha., according to our survey. The farming system is not easily mechanized and requires high labour inputs, so it requires high population densities. It does not, however, support the increases in population numbers that have occurred since independence. In the highest zone where the Ndunduri catchment is located, population densities increased from 272 people/km2 in 1969 to 701 people/km2 in 1999. In the area of the Kianjuki catchment, densities increased from 339 to 662 during the same period (Kenya 1970, 2001).

In addition to the intensification of agriculture, therefore, people have been forced to develop alternative strategies. Those who can afford to, purchase or rent land “below” (in the lower elevation, hotter and drier zones) for themselves or their children. This led to the settlement and expansion of cultivation in the mid-zones of Mbeere. Another important response to the land shortage in Embu has been investment in children’s education so that future generations won’t need to depend on agriculture. With the development of the non-agricultural sector of the Kenyan economy, this investment has resulted in local non-farm employment and out-migration being common among young people. Much of the new employment is in the informal sector of the local and national economy, which has grown faster than the formal sector. Of those households with adult sons, 66 percent have at least one son who has migrated.6 Finally, practicing birth control has become common, with young women saying they will have one or two, three at the most, children because raising children is so expensive: education is of utmost importance while school fees have been expensive (primary school became free in 2004, but fees are still charged for other levels). Indeed, birth rates in Kenya have dropped faster than expected, according to census statistics. The total fertility rate declined from 7.8 births per woman in 1979, to 6.6 in 1989, and to 5.0 in 1999 (Mwai and Ndirangu 2003, UNFPA 2003).

6 Questions were asked concerning only adult sons, not daughters, because traditionally daughters married and left the parents’ farm whereas sons remained to work on their parent’s land which they would later inherit. Changes in sons’ behavior would therefore better reflect the future of the household’s farming system. Today, however, many daughters also remain on their parent’s land.

LUCID Working Paper 20 13 Embu agriculture has thus been profoundly affected by national level policies. The implementation of the Swynnerton Plan transformed the farming system and the landscape. Post- independence governmental programmes promoted export crops through establishment of parastatals. The growth of the national economy and the proximity of Embu to the national market of Nairobi have encouraged production of other commodities, as well, such as milk and macadamia nuts. The farming system is, however, deceptively vulnerable because it is so dependent on this national and the international economy, and the quality of governance. Embu farmers feel keenly the recent economic decline in Kenya and they closely follow news on weather affecting the Brazilian coffee crop.

The decline of international coffee prices since the early 1990’s, combined with the collapse of the coffee cooperative parastatal during the privatisation process associated with structural adjustment, led to a crisis in Embu agriculture as farmers received low or no remuneration for their coffee. At around the same time, the tea cooperative parastatal and tea prices also declined. After a few years, many farmers lost hope that the situation would soon improve. They first severely pruned and eventually removed many of their coffee and tea bushes, to replace them with comparatively crops such as cabbages, potatoes or maize that have a low comparative regional advantage.7 The government’s stricture against removing coffee bushes was relaxed during this time, perhaps because of its inability to guarantee a market for the berries. If the current government’s plans to rebuild the coffee sector succeed and highland coffee again generates a high gate price, Embu farmers would probably quickly rejuvenate existing and plant new coffee bushes.

The land use change maps reflect other changes in the Embu landscape, however, outside of the small-holder farmer sector. The national government has changed how the Mt. Kenya Forest land is used. Within the forest reserve in Thambana, first the colonial administration and then the Kenya government maintained a “Shamba system” tree plantation, in which trees were planted and maintained by villagers who could cultivate around the seedlings until they were a certain age. Once the trees were of sufficient size, they would be harvested and new seedlings planted in their wake. This system was conducted in an opening in the forest that continued to expand (north of the study area), and especially in the 1990’s new seedlings were not necessarily planted but the land was used mostly for grazing or cultivation of potatoes. The government changed its management of the plantation in 2000, and it no longer allows cultivation, only grazing and fuelwood cutting limited with purchased permits. It can be expected that the land will revert to secondary forest fairly quickly.

A second use of the forest reserve land by the Kenya government was the establishment and then expansion of a parastatal tea plantation, the Nyayo Tea Plantation, carved out of the forest within the reserve. The tea plantation is large, an irregularly shaped belt around much of the lower edge of the Mt. Kenya Forest Reserve. Despite the dissolution of many parastatals in the process of privatisation and a change in government, the Nyayo Tea Plantation will probably be maintained because it was one of the few that was economically self-sustaining.

The high elevation, intensely cultivated Embu landscape, therefore, has undergone major changes due to changes in governmental policies and programmes, changing agricultural commodity markets nationally and internationally and local population increases. Many of the changes have occurred at the field level, a scale not visible on imagery or air photographs because of the tiny size of fields.

7 The information on Map 4 was collected before farmers started removing their coffee and tea bushes. Small holder coffee farmers in Kenya have uprooted an estimated 16,000,000 acres of coffee, causing a 60 per cent drop in production in the past 14 years (Mathenge 2004). Production has recently increased.

LUCID Working Paper 20 14 2. Low elevation, semi-arid Mbeere Mbeere neighbours Embu to the southeast, and is directly below Embu as one descends the slopes of Mt. Kenya. Whereas cool and humid Embu produces tea and coffee, hot, semi-arid Mbeere was traditionally a millet growing and goat raising area. The Mbeere people are related to the Embu people , and share a similar language and many customs. The two groups have a history of economic and social ties, including the sale of food from Mbeere to Embu, and the sale of Mbeere labour especially in drought years. The land use and soil management history of Mbeere, however, is radically different from Embu because of the difference in agroecology, governmental programmes, and ties to the national market.

Table 3. Climate in the Agro-Ecological Zones of Mbeere Annual ave. temp Annual ave. rainfall Agro-Ecological Zone Altitude in m in °C in mm UM 4 Sunflower-Maize 1,280 – 1,400 20.7 – 20.0 960 – 1,100 LM 3 Cotton 1,070 – 1,280 22.0 – 20.7 900 – 1,100 LM 5 Lower Midland 830 – 1,130 23.5 – 21.7 700 – 900 Livestock-Millet IL 5 Lowland Livestock- 760 – 830 23.9 – 23.5 640 – 780 Millet Source: Jaetzold and Schmidt 1983.

In the 1950’s and 1960’s, Mbeere was mostly covered by bush or grassland that was used for raising large herds of goats and cattle. The vegetation was basically derived savannah, created by many years of grazing animals and use of fire (Brokensha and Riley 1977; Riley and Brokensha 1988). Indeed, the Mbeere people considered themselves herders. They regularly had tit-for-tat rustling of animals (and women) with the Machakos Kamba, and were victims of rustling by the Maasai. Shifting cultivation, or rotational bush fallow, of millet and sorghum was done on small fields in wetter areas and on better soils, as allocated by clan elders every year: “We just used a digging stick, a sharpened stick... so dug small areas of 2 to 3 plots. We looked for fertile areas, saw it was fertile from the bushes so it hadn’t been cultivated for along time. First you would slash the bushes, then burn them, and later come to plant when it was dry or after a rain. We used to dig afterwards only for weeding. We dug only a small area, but yields were very high...” People moved together in small groups, shifting their crops and homes, within the relatively small area controlled by their clans. Ties to urban centres and markets were weak, due to few schools and bad roads.

As population densities increased, the amount of land under crops gradually grew (Map 5). The establishment of primary schools and the growing of cash crops led to people settling and cultivating the same land repeatedly, although not continuously (Brokensha 1971). This relatively slow process of sedentarisation and expansion of cultivation suddenly changed, however, when in the 1970s and 1980s the Kenya government implemented a land adjudication programme similar to the Swynnerton Plan applied in Embu. Boundaries for individual farms were drawn onto aerial photographs, and all land was apportioned except protected areas and the steepest hillsides: “Demarcation was a government decision. We had absolutely no say. It was done on air photos, not the ground, so included the top of hills, rock. It wasn’t popular— people had to move who thought they had acquired land somewhere. Freedom was curtailed—before we could cultivate three acres for a few years and then leave it. I still prefer shifting cultivation; I see no advantage to demarcation.”

LUCID Working Paper 20 15

Map 5. Land Use Change between 1961 and 2001 in the Kiritiri area of Mbeere District

LUCID Working Paper 20 16 Table 4. Land Use Change Statistics, Kiritiri area, Mbeere District 1961-1982 1982-2001 1961 1982 1995 2001 Land use change change type Km2 % Km2 % Sq. Km % Km2 % Km2 % Km2 % Bush 52.9 65.1 36.8 45.3 -16.1 -19.8 4.8 6.0 5.9 7.2 -30.9 -38.0 Farm 19.4 23.8 32.0 39.4 12.6 15.6 69.0 84.9 70.6 86.8 38.6 47.4 Woodland 9.0 11.0 12.3 15.2 3.3 4.1 7.0 8.7 4.5 5.5 -7.8 -9.6 Urban 0.0 0.0 0.1 0.2 0.1 0.2 0.4 0.5 0.4 0.5 0.2 0.3 TOTAL 81.3 100.0 81.3 100.0 81.3 100.0 81.3 100.0

“It was done by the clan. Clans registered all members— male babies to old men— no ladies were registered even if unmarried. The Clan Committee then subdivided the land The Clan Committee members who were sharp got big pieces of land, and no Committee member got rocky land, all good land.”

Clan elders thus obtained larger farms, but they also lost their land allocation authority and much of their power, a continuation of a process that had started with the nomination of chiefs and others by the government (Njeru 1978). The adjudication programme was implemented clan-by-clan across the Mbeere landscape. The adjudication process is still occurring in neighbouring Tharaka with important land use and land management impacts (Smucker 2002). In Mbeere, each farm family was given one piece of land for their house and for crops in the clan’s “village” usually along a road, and another piece of land further away, for pasturing their animals. The resultant settlements were sometimes a formalization of existing villages, and sometimes newly created grouping of people of similar ages and from the same clan. An immediate result was that the land near the homes and often those further away were cleared, and barbed wire and/or fences put up establishing ownership.8 Almost all the sacred groves were on land that was distributed, and so were lost (Brokensha and Riley 1977). Trees, even native trees, changed from being communally to privately owned. Labour sharing, that had been based on kinship and other ties, became much more individualized and cash based (Brokensha and Riley 1980; Glazier 1985). Much of the bush in these settlements was cleared within the first years of adjudication, creating a dramatic land use/ cover conversion.

The pieces of land that farmers were allocated far from their homes were also usually fenced to claim ownership, and the fences acted as barriers to communal grazing. The far pieces of land are usually not cultivated because of problems of crop damage by wild and domestic animals, and theft. They are often given to sons as their inheritance, or sold especially by the poorer households. The result of the land transactions was a solidification of wealth disparities, that were later consolidated with differential investment in children’s education. Njeru (1978) found that landlessness increased soon after adjudication because of land sales.

Across the area, the land use conversion from bush to crops was rapid. The individual farms were not sufficiently large to support the former large goat herds and, much to the distress of the older Mbeere, raising large herds became impossible. The Mbeere people became primarily farmers. With adjudication, or demarcation, the economic and cultural system of the Mbeere was changed overnight: “Before, I was a pastoralist. Now, I am converted and am a cultivator. We were all pastoralists. We did farming to get bags of millet to barter for animals, but we were pastoralists.”

8 The fences were often made of barbed wire and thorn tree or euphorbia branches to form a living fence, and these are what are visible on remotely sensed data to indicate farmland.

LUCID Working Paper 20 17

The privatisation of land was also associated with increased land sales to people both within and outside the clan. Limited land sales had existed before adjudication, though they were rare because the clan members needed to agree to the sale. Some Embu and Kikuyu had settled in the area especially during the Emergency in the 1950’s (Brokensha 1971), and the mid-elevation zones had been receiving Embu and Kamba migrants for many years (Glazier 1985). According to the farmers, after adjudication in the Kiritiri area, Embu, Kikuyu from Central Province and others came to buy land from individuals who wanted or needed the cash. The difference in population growth rates between areas and over time (Table 7) would seem to confirm that migration occurred into this area especially in the 1980’s. Older men complained that young people tend to quickly sell their land even today. In-migrants have difficulties, however, in the harsh semi-arid environment: The strategy used by migrants, and by those who came here, well, they start by doing things the way they do it in their own home place. Like in the upper zones, they dig using jembes [hand hoes]. Now they come here and when they try to dig, they don’t make it. And most who do that, they are unable to cope with the situation and they go away.

When most outsiders come here, they will buy land during the rainy season. The situation is very good. But come the dry spell, the place looks worse! They start wondering why they had first found the place so good. Sometimes they just go away. The communal grazing land that we have is land that has been abandoned by such people, after they were unable to cope with the dynamics of this place.

Rather than adjudication leading to increased investment and intensification of agriculture, therefore, some of the driest areas had high land sales and reduced use.

The landscape of central Mbeere District, Kenya has thus changed within a fifteen year period from one dominated by bush and woodland to one almost completely covered by agriculture. The economic and social system changed, as well, from one based on communal management of resources to private management, and from primarily pastoralism to cultivation with some animal raising. The former large herds (of around 10 to 20 goats, sheep and cattle) that were lost during the 1984 drought could not be replaced following demarcation and the loss of free grazing land. Herd sizes are smaller, around 3 to 6 animals, but because the number of households has increased, farmers estimate that the actual number of animals in an area has probably stayed the same. Over- grazing is seen as a problem: “With demarcation, we were disappointed because we had expected to be able to graze around but suddenly we found the area with good grass was fenced. Also, you were forced to cultivate whatever land you got and it could be overgrazed, eroded or naturally infertile. You can see an area where there is better soil but you can’t use it. Any advantages? Now, you plant high value crops, and perennials, like mangos. Now, you plant more trees. Anyway, it is too late to go back.”

.

LUCID Working Paper 20 18 not cropped

17%

maize, beans

39%

25% bush

12% N 8%

miraa other food 2000200400Meters

not cropped maize

19% 24%

N 15% other food 42% 200 0 200 400 Meters

bush

Map 6. Land use maps and pies of percent of land under land uses in Kambita (in LM4 cotton zone) and Ivondo catchments (LM 5 midland livestock-millet zone). Yellow is maize, olive is other food crops, red is Miraa, cyan is bush and grey is not cropped. Field boundaries are shown.

LUCID Working Paper 20 19 This sudden change in land use occurred not due to a large influx of population but to the implementation of land policy by the national government. The Mbeere people found themselves confined to family plots and necessarily to cropping. The average farm size, according to survey results, is 3.2 hectares (2.3 ha. per adult male). Population densities increased from 31 people/km2 in 1969 to 81 people/km2 in 1999 (Kenya 1970, 2001). Land tenure and land use changes that may have otherwise occurred gradually and unevenly with an increase in population densities and intensification of the agricultural system were mandated as part of the national government’s land adjudication programme. The result was that landscape was converted from 60 percent bush and woodland in 1982, to only 15 percent bush and woodland thirteen years later in 1995. Eight-four percent of the land was then covered by cropped fields.

With bush clearance and the continuous cultivation of the same fields, using ox ploughs rather than a digging stick or hand hoe became possible. The plough is commonly used today, especially by those who are rich enough to own or rent the plough and oxen (a few tractors are used but are owned by institutions rather than individuals). The ploughs, and having one’s fields near the home, made cultivating larger areas possible. Cereal crops have also changed from primarily millet and sorghum to primarily maize. In the 1960’s and 1970’s, cotton was an important cash crop organized by a parastatal. That ended when farmers weren’t receiving high prices for it, however, and today maize is increasing because of its growing market despite being vulnerable to drought. In some areas near the paved road, Miraa (Catha edulis Forsk), which is air freighted daily to Arabian and Somali markets, and horticultural crops such as melons and fruit are increasing.

The conversion of land from natural, albeit managed, shrub and grasslands to annual crops has many environmental implications. The soil is now tilled twice annually. Because of low rainfall, the ground is left bare for long stretches of the dry season open to wind and water erosion. Soil organic matter decomposes rapidly and nutrients are lost in the high temperatures, and soil fertility has declined (see Gachimbi 2002 and Section III.5. below). Farmers also complain about land being over-grazed due to the decline in available pasture. In the Kiritiri study area, the only woodland that hadn’t been cleared by 2001 was in the Kianjiru Hills reserve (the western edge of Map 5), other institutional land, and on some extremely rocky and steep hills. The trees in the remaining woodlands became less dense as people cut them for making charcoal to sell, and the woodland converted to bush.

An important landscape modification has been within the “bush” land cover class with the disappearance of native tree, bush, grass and other plant species. Bee keepers complain that they can no longer maintain bee colonies, for example, due to the lack of nectar from wild plants, and herbalists must travel to protected areas to find their medicinal plants. Farmers have significantly reduced the size of their animal herds as the little remaining grazing land is now privately owned and is usually fenced. This loss of grass and other natural vegetation, and fragmentation of the remaining bush and woodlands, has significantly affected wildlife in the entire eastern slopes of Mt. Kenya area (see LUCID Working Papers 8, 32, 36, 37 and 38).

Farmers soon found that their plots of land were small, the rainfall highly variable and marginal for crop production, the soil quickly degraded, and the harvested crops had a low value (e.g., millet, sorghum, pigeon peas, cow peas, green grams and increasingly maize and beans): “Now there’s just a bit of land and not enough for the next generation. With the population increase, farm sizes are smaller since the land is sub-divided. We’ll have a mix of crops but not enough food.”

LUCID Working Paper 20 20 The consequence is that household income from farming is limited in Mbeere, and more than half of the adult men have left the area to work in cities or as labourers on farms elsewhere (see Section III.3.).

This departure of men to search for off-farm income, combined with the departure of especially in- migrant families who left due to the harsh circumstances, has important land use implications (Map 6). The labour required even for low value crop production is high, while the opportunity costs for that labour appear to be greater elsewhere. The result is that after the initial fencing and clearance of bush following adjudication, much of the land in the drier zones is not being cultivated. Some has reverted to bush and is either not being used, is being used for grazing, or mango or other low maintenance trees have been planted on it. In Ivondo, the catchment in the driest area mapped, over 40 percent of the land has reverted to bush (cyan in Map 6), although field boundaries are still respected and fences are maintained. Some of this land is being used for communal grazing. The intensity of the land use system has therefore evolved—it gradually became more intense as population densities increased, it rapidly intensified following adjudication, but the more marginal areas appear to have reverted to less intensive uses.

The conversion of land use from mostly grasslands and woodlands to cropping is similar to processes occurring elsewhere in semi-arid regions of Kenya and other African countries. The rapidity and completeness of the conversion in Mbeere, however, is unusual. The changes in the landscape reflect equally momentous changes in the Mbeere culture, economy and social system.

C. Income diversification and out-migration An important component of income diversification in Embu and Mbeere has been searching for non- and off-farm income sources, especially by adult men.9 This is often portrayed in the literature as a coping strategy in response to low farm income and the expectation of higher urban income. High population pressure is often seen as an economic “push” factor leading to temporary and permanent out-migration, in conjunction with adaptations to the farming system (Zelinsky 1971; Grigg 1980; Boserup 1981; Binswanger 1983; Bilsborrow 1987; Bilsborrow and Okoth-Ogendo 1992; Turner et al. 1993; Tiffin et al. 1994; Reardon 1997). The amount of time invested by rural households on non-farm income activities is also an indication of the low relative returns to agriculture.

The migration patterns and household structure the rural areas of Embu and Mbeere is, however, complex. Instead of a simple dichotomy of men who have stayed on their farm and those have moved to settle elsewhere, a pattern of temporary migration and of maintaining two or more households is common. Many men live temporarily away from their farm while they are employed elsewhere. Others live permanently away but leave a wife and family on their home farm (and sometimes maintain a second family in the city). Still others cultivate more than one piece of land and live on one while a wife lives on the other. Respondents were asked, therefore, not only who was the head of the household, but also whether the husband lived or worked full, part-time or not at all on the farm. Household structure is not only complex but in constant flux. Migration patterns fluctuate as well, as employment opportunities change and men follow searching for jobs.

9 Non-farm employment is employment outside of agriculture and can include carpentry and other activities conducted while living on one’s own farm or elsewhere. Off-farm employment is salaried or otherwise remunerated activity conducted elsewhere, for example agricultural labour or business.

LUCID Working Paper 20 21 Figure 1. Percent of husbands (left) and adult sons (right) conducting non- and off-farm work according to farm size.

In Embu and Mbeere, rates of out-migration and conducting non-farm activities indeed appear to be related to restricted agricultural wealth. The percentage of households whose husband (male head of household or male spouse of head) is conducting non- or off-farm work is directly related to farm size and farm wealth. The fewer agricultural resources on the farm, particularly farm size but also number of animals and amount of agricultural labour, the greater the likelihood that the husband is not spending his time farming his own farm. Men from households with few resources are more likely to be either working on non- or off-farm activities nearby, or to have left the family on the farm to search for employment elsewhere (Figure 1). The most common type of employment of men from poor households is to work as artisans or agricultural labourers.

The absence of the husband has long-lasting impacts. According to statistical analyses, those families tend to stay poor, with farm sizes not increasing as the family ages. Investment in additional land, labour or animals remains very low (see discussion of the poverty/land management relationship below). The sons of these families also have a much higher rate of searching for work as agricultural labourers or other low-wage employment; less than 30% of adult sons stay to work on the family farm. The families left behind, therefore, do not benefit from the husband’s or adult sons’ agricultural labour or social role (important, for example, in purchasing land, or purchasing and caring for large animals), and the remittances are often small. The size of remittances in general are used for immediate consumption needs and not for investment in the farm. Hunt found that distress land sales to more wealthy families also reinforces the inability of these households to move from poverty, and increases wealth differentiation (Hunt 1996). This group is particularly vulnerable to the effects of drought of other crises, when sales of assets including land become more common. The loss of resources, especially land, are difficult to regain later, furthering the poverty of the poorest. Widows are the poorest and least able to produce enough food or maintain their farm compared to any other group. They are no longer only old women—younger women are increasingly widowed because of the effects of HIV-AIDS. This group in both Embu and Mbeere is moving towards or has reached being near landless.

An exception to this poverty/out-migration pattern, however, is the richest group of farmers. This group also has a high percentage of husbands who are earning non-farm income, but more of them stay to live on their own farm. Few of this group has out-migrated—they tend to stay on their farms and invest in it by buying additional land, labour, animals and other agricultural

LUCID Working Paper 20 22 inputs. They often have a professional salaried job or a business. Approximately half of their sons, too, tend to stay on the family farm although many have non-farm jobs.

The majority of the households, those of “medium” wealth, tend to still rely on their own farm for most of their income and continue to invest in their farm. As couples grow older, they invest in buying additional land, animals and in their soil. Whether this pattern will continue with the younger generation is uncertain because of the limited amount of land to purchase (and high land prices) and lower returns to agriculture relative to non-farm activities.

As seen above, the relationship between poverty and rates of non- or off-farm employment is strong and follows a similar pattern in Embu and Mbeere. The question of the role of population pressure affecting out-migration is, however, not as expected. Embu has high population densities and extremely small farm sizes, so high rates of out-migration would be expected. Mbeere has lower population densities and larger farms, so lower rates of out-migration would be expected. The rates of out-migration and of conducting local non-farm employment, however, appear to the opposite. There is a significant increase in the percent of husbands and adult sons in the lower population density and lower elevation zones who have left. The lowest zone surveyed (LM 5, labelled “cotton” in the figures but livestock-millet by Jaetzold and Schmidt 1983) has fewer than half of the husbands and less than a third of the adult sons working on their own farms. The majority are attempting to earn income elsewhere. Embu farms have higher per capita economic productivity despite their small size, and a higher proportion of young Embu men are able to make their livelihood near home.

This dramatic difference between zones is probably due to similar poverty/ out-migration factors as seen affecting households differently within zones. The lowest zone particularly has the lowest crop remuneration potential—it has low and variable rainfall, poor soils and no valuable cash crop. Links to a market for milk or another perishable product are weak. This area also has fewer non-farm employment possibilities because the villages are less developed and it is farther from Embu town. Young men especially have left, usually leaving their families behind. These households tend to have the smallest farms and the most problems producing enough food to feed themselves. The “push” factor of out-migration is poverty, therefore, not population pressure.

Figure 2. Percent of husbands (left) and sons (right) conducting non- and off-farm work by AEZ.

LUCID Working Paper 20 23 This higher level of out-migration currently occurring among Mbeere households is a switch from what it had been in the 1970s. Embu households migrated to Mbeere’s more humid zones (UM4 and LM3) in search for land in the 1970s and 1980s, and many Embu households have purchased relatively cheap Mbeere land as a second farm to produce maize for their household. Migration from Embu to Mbeere slowed in the 1990’s, however, has the land available was drier and more remote.

D. Effects of household composition and poverty on land use and management10 The “traditional” family, with a wife and husband who live and work on their own farm and whose sons expect to work on their parents land, is therefore no longer the experience of many households. Indeed, only 60% of the surveyed households in Embu and 50% in Mbeere meet this definition. In Mbeere especially, wives manage many farms because the husbands are absent. This frequent and long-term absence of husbands has affected the activities that women do on the farm, and their role in the society: “Before, jobs were divided. Men did the heavy job of digging. Now it is more equal, women see that they can do all jobs so they dig, and do other things. There is little difference in jobs. Women feed the animals, too.”

Households thus have varying compositions that include widows as heads, female headed with the husband away, and husband present but working elsewhere. The land management practices of each of these groups are compared below. Within each group, first Mbeere and then Embu households are examined.

1. Widows In the households surveyed in Mbeere, 6% (N=5) were headed by a widow. They are the poorest group in Mbeere, with the least amount of land (widows do not have clear rights to land following the death of their husband) and with few other agricultural resources. Few hire in extra labour, they do not practice fallowing, and they rarely use chemical fertilizers or apply manure to few fields. Indeed, 60% of these households do not produce enough food to feed their families. The younger widows are particularly poor—they have very few or no animals, little land, and their fields’ soil fertility is declining severely.

Thirteen percent (N=10) of the Embu households are headed by a widow, mostly older women whose small farms date from the 1960's adjudication. Many of these households have adult sons and their families living on the farm so the number of adults is fairly large. Many of the family members work as agricultural labourers, and little agricultural labour is hired in. These are among the poorest families and they do not produce sufficient food. Few of their fields receive manure, and this is the group with the highest percent of fields with declining fertility (47%) and with the worst erosion problems.

2. The “traditional” group: husband present and working on his own farm This group makes up 51% (N=45) of the surveyed Mbeere households. These households have the highest number of adults working on the farm; most households are composed of the husband, wife and at least one adult son. The husbands typically have only primary education and are older than the average head of household. Their farms are relatively large (3.6 ha) but, when divided up by adult males on the farm, they have around 2.4 ha/male (smaller than households where the

10 The information on soil management and soil characteristics is from the field level survey. Management questions concerned the past season, and perceptions of changing characteristics reflect the period since the respondent started cultivating the surveyed field.

LUCID Working Paper 20 24 husband works off-farm). Similarly, most (66%) own additional plots of land, but fewer than those households whose husband work off-farm. They apply manure more often, but chemical fertilizers less often, than average. Their soil fertility is changing at the about the average rate.

This group constitutes 61% (N=49) of the surveyed Embu households. As in Mbeere, they are the oldest group, and have little education above primary school. They have the most land per adult male, indicating that they’ve been able to maintain their original plots (perhaps by giving supplemental, purchased plots to their sons). Other indications that they are labour-rich and cash- poor are that many have family members who work as agricultural labourers, and that they apply less chemical fertilizer and have fewer animals than on average. They do apply quite a bit of manure and household residues, and fallow some fields. Their soil fertility is, in general, improving with around half of the fields experiencing improving fertility and around 39% with declining fertility.

3. Female headed with husband working/ living away This group makes up around one quarter (N=21) of the surveyed Mbeere households. The absent husbands work in Nairobi, Embu town and elsewhere at a variety of jobs, either salaried or as businessmen, artisans or agricultural labourers.

After the widows, this group is the poorest. It has the smallest land holdings, an average of 2.6 ha, with almost half of the households owning less than one hectare. They also have the fewest animals of any group, only 3.3. Their families are the smallest, with the wife as the sole adult on the farm. They apparently have a labour shortage despite their small farms—70% hire in labour while at the same time many (58%) also work as agricultural labourers themselves. After the widows, they have the most problems producing enough to feed their families (47% do not produce enough). It appears that despite the income their husbands may be earning while away, these families are suffering from a lack of resources.

Their soil management reflects their agricultural resource scarcity. Few (20%) of their fields receive chemical fertilizers and only 40% of their fields receive manure. This, again, is similar to widows’ farms. Their soil fertility, however, is deteriorating the worst of any group. They have highest percent of fields with declining soil fertility, 53%, and the lowest percent with improving fertility, only 36%.

The most important predictor of husbands working off-farm is their young age. Indeed, one-half of the husbands under age 35 have left their families to search for employment. The trend of young people leaving due to few opportunities in agriculture is expected to continue. Women said of their children: They can see that wealth comes from outside, not here because of the problem of land. Now here is the same problem of lack of land so children need to invest in business in Embu or in something else. While fathers noted: Now even those with education can’t find jobs, so we will probably send children to technical colleges to learn skills so they can be self-reliant and not depend on office jobs.

In Embu, the group whose husbands are not present makes up only 15% (N=12) of all households. Their husbands work a variety of jobs, mostly as artisans or in salaried jobs but some have remained in agriculture. Their educational levels are mostly high. They form two groups: one of young and poor households, the other of older and less poor households. The young have

LUCID Working Paper 20 25 very few animals whereas the older group has many; the young have very little land whereas a few of the older group has much. In general, around half of the young are near-landless (under 1 ha). On average, however, this group doesn’t appear to be very much poorer than the other groups, unlike Mbeere. They do have the same problem of a labour shortage: there are few adults on the farm, most (60%) hire in agricultural labour, and few (17%) have someone in the family who works as agricultural labourer. Their soil management is similar to other groups except that they use less manure on their fields and fewer fields have household residues applied. The soil’s fertility of this group is similar to the average with 37% of the fields having declining fertility and 51% with improving fertility.

4. Husbands present but work off-farm In Mbeere, the group of households whose husbands live at home but have an off-farm job is, at 19% (N=17), relatively large. The husbands work as either as artisans (carpenters, masons), have a salaried, professional jobs (teachers, agricultural extension agents), or are businessmen. The husbands are mostly middle-age. Very few have family members who work as agricultural labourers. This is the richest of all the Mbeere groups with the largest farms, especially in terms of land per adult male (3.1 hectares), and the most animals. Soil management reflects this wealth: this group uses the most manure and more fields receive chemical fertilizers. Their soil fertility reflects the inputs: this group has the highest percent of fields with improving fertility (60%) and the lowest percent of fields with declining fertility (35%).

The difference between the two Mbeere groups with husbands working off-farm is remarkable: the group whose husbands are absent is the poorest group whereas the group who husbands live on-farm is the richest. Indeed, almost half of the near-landless households of Mbeere have absent husbands.

A group of women interviewed about this observed: “The home depends on the woman. If the man works outside he can save some money to spend on labour or other things for the farm... but some men are drunkards and don’t give any money.”

While in a parallel group meeting, men said: “Usually, the home depends on the woman and if she is lazy or hard working. Men still participate when home on vacation or around.”

Whether the households with an absent husband will remain poor or, through remittances, slowly gain more land and animals remains to be seen. As seen above, though, these households are usually young, have the fewest resources to start out with, and are having a difficult time even producing enough food for themselves. In contrast, in the households with a husband present, the off-farm employment has led to accumulation of land and animals, and much investment in the soil. The wealth difference between the two groups will probably be accentuated with time, therefore.

In Embu, this is a small group (N=11%, 9) of households whose mostly middle-aged, educated husbands live on-farm and work as artisans, teachers or businessmen. They appear to be relatively large families with many adult males and, therefore, little land per male despite their average to large land holdings. Almost all (71%) also cultivate plots outside their farm. The labour abundance of these families is also reflected in that few hire in agricultural labourers. They have around the average number of animals and most of their fields receive manure. Similarly, many fields receive household residues, and all receive chemical fertilizers. Fallowing, however, is

LUCID Working Paper 20 26 Figure 3. Factors affecting a change in soil fertility. Data from household and field level surveys. Statistical correlation: * = significant at the 0.05 level, ** = significant at the 0.01 level.

rarely practiced. The soil has responded to the inputs and attention and is doing the best of any group with 76% of the fields with improving fertility. In general, then, this group appears to be slightly better off than similar households where the husband does not have a job: most families in this group own a second plot of land, they use more chemical fertilizer and their soil’s fertility is improving.

5. Summary of Household and Field Factors affecting Soil Maintenance and Fertility Despite the very different farming systems and crop profitability in Embu and Mbeere, similar factors have emerged as playing a significant role in affecting the household’s ability to maintain their soil, and changes in their soil’s fertility. Wealth, especially farm size, is seen by farmers to be playing an important role. One farmer in Mbeere explained: Rich people can leave a piece of land uncultivated if it is losing fertility, and then put animals there for their dung. Yes, there is a difference. The rich can buy manure and pay people to dig trenches or terraces, and employ people to dig five or more compost pits, compost stacks. But the common man can’t. So, the difference [in soil fertility] has to be there. The common man can dig maybe one pit, and can’t buy manure.

The relative importance of these factors in affecting soil maintenance and changing soil fertility were tested with correlation and a stepwise regression analyses. The variables are closely associated conceptually and statistically, e.g., farm size and number of animals are closely related to labour availability and use of manure, so agricultural wealth variables were condensed into an index. 11 Field characteristics were not included in the analysis because they tend to change along with change in fertility (e.g., degree of perceived erosion is the strongest predictor of changing fertility). Nevertheless, the following summary conclusions can be drawn:

11 The household agricultural wealth index is composed of farm size and a large animals index (controlled for agroecological zone), and the hiring-in of agricultural labour.

LUCID Working Paper 20 27 • The number of adults living in the household is closely associated with farm size and degree of soil maintenance. It is particularly those households with few adults and small farms that are also have few animals or other resources, place few inputs on their soil, and have declining soil fertility. In contrast, farms that hire-in agricultural labour have the highest levels of improved soil fertility. • Soil characteristics most closely associated with changing soil fertility are change in top soil depth, soil water holding capacity and degree of erosion. Soil maintenance variables most associated with changing fertility are application of manure and, to a much lesser extent, chemical inputs and whether soil conservation methods are present. • The impact of the absence of the husband from the farm on soil fertility is indirect via its effect on farm labour, farm size and other resources. The absence of the husband from the farm is closely associated with poverty. Husbands leave very small farms because the family needs non-farm income, but once the husband as left, farm sizes or other signs of agricultural wealth do not increase despite the household’s life cycle (age), number of adult sons, or availability of off-farm income. • Off-farm income from employment that has a higher return (e.g., professional salaried employment, business) appears to be invested into buying additional land and otherwise maintaining the farm. This is less important than the absence of the husband, indicating that outside income is less likely to be invested in the farm if the husband is absent. • A combination of household resources—animals, farm size, labour, and non-farm income—rather than any single resource is what affects the level of soil management and changing soil fertility (Figures 2 and 3).

LUCID Working Paper 20 28

In general, the impact of poverty on soil maintenance and fertility is significant (adjusted R2 of .238), if not as critical as it is in Rwanda, southwest Uganda or western Kenya (Olson 1994, 1998; Crowley and Carter 2000). Perhaps the high degree of income diversification and the support of the central government in agriculture has somewhat tempered the poverty/ degradation relationship in Embu and Mbeere. Nevertheless, the large difference in soil maintenance and soil degradation between the poorest and wealthiest farmers is striking.

E. Land and soil management variation between agroecological zones The variation in land and soil management between agroecological zones is as important as the variation between households. The extremely different farming systems—from almost entirely based on export crops in the high potential upper zones to low value mostly subsistence production in the lower—is associated with parallel differences in economic returns to investment in the soil, and to differences in governmental involvement in soil maintenance.

Embu, with its high population densities and high agricultural potential, has been the focus of governmental programmes to increase productivity since the 1930’s colonial administration. The colonial administration started a terracing programme that the chiefs enforced along with their enforcement of tax payment: “The job was open for Assistant Chief and people suggested me. The big job was to collect taxes— every person [man] over 18 years old paid 8, then 10, then 12 shillings. Only the head of households and unmarried men, not the young or elderly. I also made sure homes were clean and had pit latrines. Plus, if the Agricultural Officer told you to put in 2, 3 terraces and you didn’t, the Assistant Chief would take you to court and you could be jailed for 2, 3 months.”

Following independence, the enforced terracing programme ended and many people, very resentful of how they had been forced, removed them. The continuous cropping of the same fields, a necessary practice due to the small farms, combined with erosion led to rapidly declining soil fertility. Major changes occurred when the independent government permitted small-holders to grow the remunerative cash crops of first coffee and later tea. Parastatals gave credit for farmers to purchase chemical fertilizers, pesticides and fungicides for the coffee and tea. Farmers applied the fertilizers and the pesticides to their food crops, as well (see Table 5). Terracing coffee and tea fields became mandatory in the 1970’s, and in the 1980’s the Ministry of Agriculture and the Swedish International Development Agency organized terracing programmes for all fields within designated catchments (Berlekom and Larsson 1984; Admassie 1992). Indeed, Embu farmers applied chemical fertilizers to 90% of fields where they were growing food crops, and some form of soil conservation was present on 52% of their fields. Extension agents insisted that the coffee bushes be mulched, and the Ministry of Agriculture revived terracing programme for tea and coffee fields. This was later supplemented to cover all fields.

The level of soil and water conservation and application of inputs thus increased greatly in Embu, assisted not only by the government support or insistence but also because farmers appreciated the results. Farmers said that the fertility of many fields increased starting in the 1970’s. Manure was particularly appreciated because it improved the soil’s condition as well as increased yields (farmers complained that chemical fertilizers eventually ruin their soil). In the 1990’s before the coffee and tea prices collapsed, many of the wealthier Embu farmers were purchasing truck loads of manure brought in from Mbeere—ironic since the farms of Mbeere have much lower levels of fertility, and more of their fields show worsening fertility.

LUCID Working Paper 20 29 The level of soil maintenance in terms of applying inputs is, in general, much lower in Mbeere than in Embu (Figure 5). Mbeere has never had the same level of governmental support in terms of schools or agricultural development, and so has not had the same experience of receiving credit for fertilizers and pesticides. At the same time, the adjudication process occurred later and the conversion to dependence on cultivation has been fairly recent. Only since the late 1980’s have agricultural extension staff and non-governmental organizations focused on reducing soil erosion through terracing programmes (especially in drought years, terracing is a common food-for-work activity) or planting trees (Hunegnaw 1987). Mbeere farmers, however, developed an indigenous low-input system to reduce soil erosion through techniques such as trash lines (Tengberg et al. 1999).

Figure 5. Percent of fields with seasonal crops where soil inputs were applied by agro-ecological zone. Source: field level survey.

Mbeere farmers have noticed rapid declines in the fields’ soil fertility in the few years since adjudication and complain of the need to cultivate continuously (they can no longer practice fallowing, which had been their former means of increasing fertility) and their need to for manure. Indeed, fewer farmers apply manure to their fields than do Embu farmers. This could be that, ironically, a larger percentage of Mbeere households (30%) own no large animals, whereas only 20% of Embu households are without any. The poor in Mbeere are poorer and a larger group than the poor in Embu. Those households in Mbeere who own animals, however, have large herds. Mbeere farmers spoke, though, of their interest in applying inputs: Before, we would cultivate small plots, plus the yields were low since there was no soil conservation (only trash lines and they wash away in the rain), and no education for better techniques, especially manure (today you even can buy manure from others). There were no pesticides to keep insects from eating the crop, no fertilizer so it would give more, no pesticide for storage and no soil conservation so nutrients washed away...

They also said that their interest in investing in their soil increased with land privatisation:

LUCID Working Paper 20 30 It [demarcation] is good. Now, you can own stuff. Before, everything was communal, even the goats. Now you take care of the land and even infertile land is improving.

Embu Mbeere Change in soil worsened 37% 43% fertility no change 14% 8% improved 49% 48% Total 100% 100% Why soil fertility added 21% 29% changed manure added fertilizer + 15% 0% manure added chem. 6% 0% fertilizer added 7% 6% oganic inputs soil 6% 9% conservation fallowed 0% 8% lack 14% 21% manure, continuousf tili 12% 10% cropping erosion 10% 10% other 9% 7% Total 100% 100% Manure no 34% 48% yes 66% 52% Total 100% 100% Chemical fertilizer no 9% 85% yes 91% 15% Total Col % 100% 100% Have fallowed? yes 23% 27% no 77% 73% Total Col % 100% 100% N (fields) 238 260

Table 5. Changes in fertility since cultivating the field, and % of fields with applied inputs in Embu and Mbeere. Column percentages of cultivated fields. Source: field-level survey

LUCID Working Paper 20 31 Despite the lower levels of manure use, Mbeere farmers are not often using chemical fertilizers as a substitute (Figure 1 and Table 5). Very few fields in Mbeere have received fertilizer. The severe problems that Mbeere farmers experience with insect pests, however, has led them to purchase insecticide to apply to around 45% of their fields (especially on their maize, beans, and green grams, less so on millet and other crops). The returns to purchased inputs are probably relatively low, especially in the driest zone where rainfall is often unpredictable. Fallowing, the prior common method to maintain soil fertility, is now practiced on the fields where crops are grown at a similar, infrequent rate as in Embu. Farmers complained about their inability to fallow as long as they would like, and blamed their small farm sizes. Indeed, households with small farms were the least likely to have fallowed their fields—those who had fallowed the surveyed field had an average farm sizes of 4.0 ha, and those who hadn’t fallowed an average farm size of 3.0 ha (t-test significance at 0.00 level). A similar pattern exists in Embu (1.6 ha versus 1.2 ha farm size, statistically significant)

Soil management is therefore very intense in Embu—high levels of both organic and chemical inputs, and high labour use. This has been supported by the high value of their crop production. In Mbeere, the lower value of their crop production and higher levels of poverty is associated with lower levels of input application, despite poorer soil fertility.

F. Land Use and Cover Change at the Regional Level The local land use and cover change analyses discussed above reflected changes in small-scale, farming communities. The higher elevation Embu site had dramatic land cover conversions in the 1950’s and 1960’s, with the privatization of farmland and clearance of remaining bush. Modifications within the agricultural system have since occurred, such as a decline in coffee production. Similar conversions from bush to agriculture occurred in the lower elevation Mbeere site only in the 1980’s and 1990’s. A closer examination of Mbeere land use revealed that after the switch to individual, private farms, many fields in the driest zone were still not being cultivated but being used for grazing or for trees—around one-third of the farmland was actually under bush.

These local studies are representative of land use change in the small scale farming communities in central Kenya. The timing, rate and intensity of the conversion vary across the region depending on the drivers of change and contextual conditions. However, large scale land managers also control large amounts of land—the Kenya government, large farmers and ranchers, and institutions such as churches. A view of land use and cover changes in the wider region therefore provides a different perspective of what types of changes are occurring in the country, and why.

Satellite images of the eastern slope of Mt. Kenya from 1987 and 2001 were interpreted and then compared (Table 6, Figure 6). As explained in the methodology section above, both land cover and land uses classes were assigned to land areas so that, for example, degradation of a forest is recorded as a land cover conversion even if the land use remained the same, and a change in type of land ownership is recorded even if the land cover didn’t change. The land use classes provide essential ownership information for interpreting the causes of the conversions in the landscape.

The land use maps from 1987 and 2001 (Maps 7, 8 and 9) reveal that, indeed, much of the land is being managed by large scale land holders. In 2001, approximately 21% of the land was controlled by the government of Kenya (notably parks, reserves and university research land), 69% was managed by individual small scale farmers, 7% was managed communally by clans or

LUCID Working Paper 20 32 Woodland

Urban

Tundra/moorland

Tree_plant/Shamba

Tea

Coffee/Maize

Wheat/pasture

Bush/cultivation 2001 Lake 1987

Irrigated rice

Grassland

Degraded Forest

Dams

Bush

Bamboo forest

Montane forest

0 100 200 300 400 500 600 Area (Sq. Km)

other groups (non-protected forest, woodlands and bush), and 4% was managed by large scale land owners (ranchers, large farmers, churches, parastatals). Figure 6. . Bar graph of land cover changes between 1987 and 2001.

km2 % of all Land Cover type 1987 km2 2001 km2 % change change change Bush 89.0 64.7 -24.3 -27.3 -20.9 Afro-montane forest 87.6 76.5 -11.1 -12.7 -9.6 Woodland 36.2 29.7 -6.4 -17.7 -5.5 Irrigated rice, horticulture 16.9 11.1 -5.8 -34.2 -5.0 Degraded Forest 13.4 8.4 -5.0 -37.6 -4.3 Grassland 21.3 18.5 -2.8 -13.2 -2.4 Tree Plantation/Shamba system 8.6 7.1 -1.5 -17.7 -1.3 Coffee/Maize 220.3 219.3 -1.0 -0.5 -0.9 Urban 0.5 0.4 -0.1 -27.8 -0.1 Lake 0.1 0.0 0.0 -50.0 0.0 Bamboo forest 43.4 43.4 0.0 0.0 0.0 Dams 2.2 3.1 0.9 42.3 0.8 Tea 61.7 63.2 1.5 2.4 1.3 Wheat and pasture 15.6 18.1 2.5 16.0 2.2 Tundra/moorland 65.9 72.1 6.2 9.4 5.4 Mixed bush/cultivation 451.5 498.5 47.0 10.4 40.5 Total 1134.2 1134.1 Table 6. 1987 to 2001 land cover change statistics.

LUCID Working Paper 20 33 1987 2001

Meru Town

Embu Town

N

Land CoverTypes Rainfed - mixed bush/cultivation Afro-montane forest Rainfed - wheat and pasture Bamboo forest Rainfed Coffee/Maize Bush Rainfed Tea Dams Tree Plantations/Shamba systems Degraded Forest Tundra/moorland/glaciers/grasses Grassland Urban/built-up 0 10 Kilometers Irrigated - rice Woodland Lake

Map 7. Land cover in 1987 (left) and 2001 (right) of the eastern slopes of Mt. Kenya

LUCID Working Paper 20 34 1987 2001

Meru Town

Embu Town

N

Land Use Types Non-protected forest (degraded) Coffee/Maize (small scale) Non-protected forest (non-degraded) Dams Protected forest (degraded) Forest reserves Ranches Grazing land Tea (small scale) Irrigated rice - large scale Tree Plantations/Shamba systems Uni. of Nairobi land Lake 0 10 Kilometers Mixed Bush/crops Urban/built-up National Parks Wheat and grazing pasture

Map 8. Land use in 1987 (left) and 2001 (right) of the eastern slopes of Mt. Kenya

LUCID Working Paper 20 35 E B D 1987 to 2001 • Meru Town F Areas of Significant Land Cover Change peak Eastern Slopes of Mt. Kenya

E

•Embu Town E C

A From natural to cultivation Degraded natural Maize/coffee to bush/cultivation Changed natural Rice to maize/coffee Bush/cultivation to dam To irrigated horticulture N 10 0 10 20 Kilometers

Map 9. Areas of significant land cover change between 1987 and 2001

The most important land cover change that occurred during this thirteen year period was an increase in land covered by mixture of cultivation and bush, mostly at the expense of land that had been under bush. This alone accounted for almost half of the land use change. The second most important change was the degradation of natural areas—e.g., the thinning of forest or woodlands, the conversion from woodlands to bush, or bush to grassland. The reasons for these and for the other important changes are described below organized by the type of land manager.

1. Large Scale Private Land Holders Large scale private land holders and the government can convert large areas relatively quickly. In the large scale, private land holdings, significant land cover changes that occurred in the 14 year period included: • Development or expansion of irrigation systems (e.g., by the Don Bosco church school, by large scale farmers / ranchers), e.g., site “A” on Map 9; • Conversion of large areas from bush to managed pasture, pasture to crops, or crops to pasture (large scale farmers and ranches in both the northern and southern ends of the region), e.g., site “B”; • Conversion from one crop to another (e.g., rice to maize, by a parastatal). An example is site “C” which is in a large ranch that converted land from bush to grassland. These changes appear to be driven by responses to economic opportunities, such as the expansion of the national and international market for horticultural crops. The expansion of land under maize in the rice scheme may be due to decisions made when the scheme was having personnel and technical assistance management problems, and could be temporary.

LUCID Working Paper 20 36 2. The Government of Kenya The Kenya government also altered its land management practices during in this period, particularly of the Mt. Kenya Park and Reserves. Vanleeuwe et al. documented the impact of stricter governmental enforcement of laws and regulations—enforcement that included forced removal of squatters, destruction of illegal crops, a halt to logging hardwoods for the national and international market, using permits to control fuelwood collection and grazing, the designation of the forest as a National Reserve to be managed by the Kenya Wildlife Service, and removing administrators found to be engaged in illegal forest activities (Vanleewe et al. 2003). The enforcement was precipitated by the documentation of forest degradation by the governmental Kenya Wildlife Service with the help of the United Nations Environment Programme (Gathaara 1999), and internal and external pressure placed on the government. By 2003, the number of sites within the Mt. Kenya park and reserve that were being used for logging, cultivation, grazing and charcoal production declined and the forest re-generated in several sites (as in the north Embu Thambana area) (Vanleewe et al. 2003).

The regional land cover change analysis captured some of these changes, especially the degradation of Imenti Forest (site “D”). Imenti is important for wildlife conservation because it forms part of the elephant migration corridor between Mt. Kenya Forest and Meru Park. It also had been a sacred forest for the surrounding people (Maitima, pers.comm.). The lower part of this reserve was almost completed cleared prior to the 1997 election (a local official allocated plots to farmers from nearby areas). People were evicted in 2000, however, and by 2002 secondary forest covered the landscape. These dramatic changes are visible in images from years between 1987 and 2002, but appear less dramatic if only the 1987 and 2001 images are seen. This emphasizes the rapid nature that land cover change can occur, and that land use/ cover changes that occur between snapshots may be missed. It also illustrates the pressure that the forest is under—Mt. Kenya forest is bordered by some of the highest rural population densities in Africa, where the population is poor and experiences basically negative economic returns to having the forest maintained as a nature reserve (Emerton 1999).

The importance of governmental enforcement (or threat of enforcement) of restrictions to the clearance of Mt. Kenya park and reserve lands is witnessed by the sharp boundaries between the forest and cultivation on aerial photographs and satellite images. The boundaries around the Mt. Kenya Forest Reserve were not as sharp during the 1950’s (Map 2), with much woodland and grassland adjacent to the forest, but the distinction became clear afterwards as first farms were allocated up to the forest, and then the Nyayo Tea Plantation was carved out from the forest edge.

The “Shamba” forest plantation system above Imenti forest also underwent land cover changes—some large areas that had been under planted cypress or pine trees were cleared when the trees were harvested, and other large areas appeared to have secondary forest growth. The land use, however, remained as Shamba system (this may change in the future to be a protected forest, depending on decisions made by the Kenya government).

The Kenya government also completed a new hydroelectric dam along the Tana River, the Kiambere Reservoir, during this period.

3. Small scale farmers and herders: the impact of migration The group of land managers that controls most land is the small scale farmers and herders. They manage around 69% of land as individuals, and an additional 7% of the land communally by clans. The amount of land being managed communally is rapidly declining. Approximately one-third of the communally managed bush and woodlands were converted to individually managed farms and ranches during this thirteen year period.

The land under the management of small scale farmers and herders has seen major changes in the 1978 and 2001 period, although most of the changes were on small pieces of land.

LUCID Working Paper 20 37 Frequently, however, the accumulation of these many small changes led to land cover changes clearly visible at the 1:100,000 scale. During the thirteen year period, most of these changes occurred in the semi-arid area, much of the ring of land use change at the base of Mt. Kenya. The ring extends from where maize cultivation had extended downslope in 1987, to the new areas of cultivation that spread into the drier bush (in yellow and marked “E” on Map 9).

This conversion from bush to cultivation by farmers in the semi-arid zone was the largest land use and cover change in the Mt. Kenya region. The landscape changed from one of native plants in a savannah bush ecosystem to one of cleared bush and cultivated fields, though the fields were often scattered. The percent of land under cultivation is not possible to distinguish well at the 1:100,000 scale, but the results from the Mbeere catchment maps indicate that approximately 40% to 60% of the farmland in that area was being cropped (Map 6). Similarly, the Africover project estimated field densities at 40% to 60% and higher across much of the drier area, with smaller areas of densities of between 20% and 40%. Little land outside of the protected areas was classed as under 20% cultivated (Di Gregorio 2003). Environmental impacts include the loss and fragmentation of wildlife habitat, a decline in native plant species, and soil degradation (LUCID Working Papers that examine environmental changes in the Mt. Kenya site include numbers 8, 32, 36, 37 and 39).

With increasing rural population densities and sufficient economic returns to cultivation, the percentage of land under crops may increase as in-filling and intensification continues. The economic and environmental sustainability of such an increase, however, is not ensured especially in the marginal, driest zones where crop production is highly vulnerable to droughts. Unless drought resistant, high value crops are available, a high percentage of land may remain as pasture for grazing and future development dependent on irrigation. Especially if employment opportunities elsewhere improve, the relative economic returns to clearing and cultivating marginal lands will not be of interest. Particularly in the more remote semi-arid areas, with fewer commodity marketing or off-farm income opportunities, the pace of continued expansion and intensification of cropped agriculture may slow. Land speculation and use of land as security, even when not developed, however, remains important in this region. In summary, therefore, the pattern of continued individualization of land holdings even in the marginal, semi-arid areas will probably continue despite the high risk of failure and low economic returns to cropping. How intense the land is used for cropping will depend on relative returns to labour and animal raising.

A major change within the high potential farming system is a modification of land use around Nyambene Hills in the north of the image. It changed from a coffee/ maize system to one dominated by maize. This would be consistent with the removal of coffee bushes that was observed in the field after coffee prices and the coffee market plummeted.12

A significant change in the ring at the mountain base is the degradation of several un- protected woodlands and forests, usually found on hills (in pink on Map 9). Many are sacred groves or have sacred places within them. A few were completely cleared and replaced by cultivation, such as Mbita Hill. The hills near Siakago town were converted to grassland for grazing animals. Most, however, remained uncultivated but had a lower density of trees so changed from woodlands or forest to degraded woodlands or forest, or even bush. They had thinner tree densities, especially in their lower slopes, due to increased grazing and tree cutting for charcoal and fuelwood.

Adjacent to these hills and protected areas, however, cultivation expanded and produced new, sharp boundaries between cleared areas, and those not cleared. In semi-arid zones, the

12 It was not possible to delimit spatial boundaries between tea, coffee and coffee/maize based systems at this scale. The changes are on a continuum of percentage of the tiny fields under crops, and at best we could identify only a fuzzy boundary that was not comparable between years.

LUCID Working Paper 20 38 boundaries had formerly not been sharp—woodlands and bush covered much of the landscape around the reserves. A dramatic example of new cultivation expanding up to a protected area boundary is immediately south of Meru Park (site “F”). Otouma writes that this is primarily by a migrant community from the nearby higher-elevation, high population density zone attracted to the free land, security, and the potential for jobs in the conservation area. They were experiencing many human-wildlife problems, as wildlife destroyed their crops and threatened people (Otuma 2004).

This process of bush being cleared for cultivation in the lowest zones of the study area is a continuation of the expansion of cultivation that has occurred since the 1950’s. Short-distance migration from the upper to the lower zones to search for land has been the pattern in the area since the 1950’s. It was actually enforced by the colonial administration in a settlement scheme programme Tharaka. The purpose was to clear the land (prisoners were used), reduce tsetse fly problems, and increase food production. Families in Meru were forced to move. They were provided with houses, but remoteness and problems with wildlife were severe in the first few years (Maitima, pers. comm.). In the 1970’s, new migrants from the upper zones arrived voluntarily and the pattern continued. It led to the mid-elevation zones in Tharaka and Mbeere being now populated and intensely cultivated. Development of these mid-elevation zones was enhanced by the availability of high yielding maize (Katumani), a growing national market for maize, and infrastructure development (especially paved roads and schools). In the 1980’s, the pattern continued down the slope into the more marginal, semi-arid zones, but there were fewer migrants.

Population Density km2 Percent Change Growth Rate 1969- 1979- 1989- 1969- 1979- 1989- 1969 1979 1989 1999 1969 1979 1989 1999 79 89 99 79 89 99 EMBU 116,505 171,136 234,735 278,196 152 348 332 381 46.9 37.2 18.5 3.9 3.2 1.7 Manyatta 32,726 46,894 62,440 71,332 306 460 573 666 43.3 33.2 14.2 3.7 2.9 1.3 Runyenjes 28,809 41,579 56,777 64,111 182 287 406 432 44.3 36.6 12.9 3.7 3.2 1.2 Kyeni 21,427 30,945 41,374 48,385 219 326 427 461 44.4 33.7 16.9 3.7 2.9 1.6 Nembure 18,406 28,713 36,902 41,590 186 334 429 472 56.0 28.5 12.7 4.5 2.5 1.2 Central 15,087 23,005 37,168 52,446 260 359 581 743 52.5 61.6 41.1 4.3 4.9 3.5

MBEERE 62,407 92,037 135,403 170,953 30 46 65 82 47.5 47.1 26.3 4.0 3.9 2.4 Siakago 11,207 16,522 25,982 34,330 30 46 71 93 47.4 57.3 32.1 4.0 4.6 2.8 Gachoka 21,809 31,165 45,560 59,102 27 39 57 74 42.9 46.2 29.7 3.6 3.9 2.6 Mwea 12,915 22,642 32,911 40,680 25 53 65 79 75.3 45.4 23.6 5.8 3.8 2.1 Evurore 16,476 21,708 30,950 36,841 40 53 74 90 31.8 42.6 19.0 2.8 3.6 1.8

Table 7. Population of Embu and Mbeere Districts from 1969 to 1999. 1Note that District, sub-district and location boundaries changed over the four censuses (e.g., Central which includes Embu municipality). The 1999 boundaries are used as reference. See Annex 3. Source: (Kenya 1970, 1981, 1994, 2001).

The slowing of the trend of in-migration after the 1980’s could be due to the undesirability of the remaining land available—it was dry, often remote and marginal for crop production. Crops produce low returns and are vulnerable to drought. Indeed, as seen above, the driest zones of Mbeere have a high rate of temporary out-migration by husbands and sons who need to search for non-farm income.

The expansion of cropping into ever-drier areas by in-migrants is reflected in changes in population growth rates. While the population growth rate in Embu has been declining at a rapid rate since 1969, Mbeere’s annual growth rate continued to be extremely high at around 3.9% (Table 7 and Figure 7). This probably reflected high fertility rates, but also in-migration by families searching for land. Results from the latest census in conducted in 1999 shows a new, rapid decline in Mbeere’s population growth rate to 2.4, not much higher than what would be expected from high fertility rates. It appears, therefore, that the numbers of in- migrants to Mbeere have slowed to almost none.

LUCID Working Paper 20 39

Annual Population Growth Rate Population Numbers 4.5

300,000 4.0

250,000 3.5

200,000 3.0

Embu 2.5 150,000 EMBU Mbeere MBEERE 2.0 100,000 1.5 50,000 1.0 0 1969 1979 1989 1999 0.5

0.0 1969-79 1979-89 1989-99

Figure 7. Population Growth of Embu and Mbeere from 1969 to 1999. Source: Kenya (1970, 1981, 1994, 2001)

Embu’s low population growth rate of 1.7 reflects what families there are saying—that they cannot afford the school fees and other expenses of many children, do not have sufficient land to bequeath to them, and so are having only two to three children. Family planning methods are also readily available (UNFPA 2004). HIV-AIDS may also have a demonstrable impact on mortality rates (Mwai and Ndirangu 2003) . Due to momentum of population growth from prior years of high fertility, however, the population sizes and densities of both Embu and Mbeere will continue to rise in the near future. The rapidity of the decline in growth rates, however, is a dramatic illustration of how fast the societies and cultures are changing due to both economic and social factors.

IV. CONCLUSION In summary, the largest land use and cover change that occurred in the Mt. Kenya region since the 1960’s has been the expansion of cultivation from the higher elevation areas into the mid- elevation, then low elevation and most recently to the semi-arid bush. The result is a “ring” of recent land cover change at the base of Mt. Kenya. The landscape of the semi-arid zones of Mbeere and Tharaka has changed from one dominated by bush savannah and woodland to one almost completely covered by farms, if not all cropped. The only large areas remaining of these natural vegetation types are in protected areas or on rocky, steep hills, but even there many of the woodlands have been cleared or thinned. The large scale conversion from bush to farmland was precipitated in many places by the implementation of a land adjudication programme of the national government, and sustained by rapid population growth due to high fertility rates and in-migration. Land use and tenure changes that may have otherwise occurred gradually and unevenly with an increase in population densities and intensification of the agricultural system were mandated as part of the government’s land programme. The Mbeere and Tharaka people, previously agropastoralists raising herds of goats and cattle on communal clan land, have found themselves confined to family plots and dependent on crops. Much of the farmland in the drier zones is not intensely cultivated, however, due to low returns to labour. The crops are of low value, and risks of drought and pests are high. Indeed, these areas are the poorest despite low population densities, and many husbands and adult children have left the family farm to search for other sources of income. The semi-arid zones appear to be in danger of worsening soil degradation due to this expansion of cropping into environmentally fragile and climatically risky areas, and the land cover change has had a large impact on native vegetation and wildlife.

On the other hand, the well-watered, densely populated Embu and Meru districts on the upper slopes appear to represent a success story, with the development of a highly productive, profitable

LUCID Working Paper 20 40 agricultural system that includes intensive soil management. Governmental and parastatals programmes were instrumental in the development of the tea and coffee sectors to such an extent that farmers here bought most of their food, and very high population densities have been supported primarily by agriculture. Here, too, however, farmers have been forced to respond to changing economic circumstances by modifying their farming system. Coffee and tea prices declined precipitously and many farmers eventually removed their coffee bushes to plant other, if less remunerative, crops. Land shortage is seen as a major problem and wealthy farmers since the 1970’s have been renting or purchasing land ever lower into semi-arid zones. It is also being addressed by measures to reduce dependency on the land: farm families have focussed their investments on their children’s education since the 1970’s so that they learn a trade other than farming, many young people seek non-farm employment, and birth rates and population growth rates have declined rapidly.

1950’s: Grasslands and woodland, patches of shifting cultivation. Economy based on goat herding. 1970’s Expansion of cultivation into pasture, use of plough 1980’s Initial intensification near settlements: from shifting to fixed cultivation 1980’s Adjudication from clan to family, private landholdings, rapid bush clearance. fewer animals: economy changes 1990’s Further intensification of the core area: from animal grazing to crops permanent cultivation or short fallows of seasonal crops limited tree planting, some soil management: manure, erosion control Crops bring limited wealth, much poverty. Out-Migration and income diversification by fathers and children to cities, other farms Table 8. Pattern of land use and economic change in semi-arid zones

In the small-scale farming and herding systems around Mt. Kenya, therefore, the pattern of land use and cover change appears to have followed a non-linear path of intensification and extensification, with variations in rate and with reversals. The generalized pattern is summarized in Table 8 and includes an initial intensification (sedentarisation), followed by extensification (search for land, associated with much of the visible land use change) and continued intensification (increased use of inputs and labour). Across the region, economic conditions and land tenure programmes have greatly influenced how and when this has happened: • the upper zones were able to continue to economically benefit from increased intensification of their system until a drop in the international price of coffee and tea, and decline in governmental support of the coffee sector. Farmers then reversed course, changed to less intense if marketable crops, and adopted demographic and other responses to reduce land pressure and their dependence on the land. • In the mid and low elevation zones, the extensification of cropping has been dramatic. Where marketing and environmental conditions are conducive, this has been followed by an intensification of the system and development of a productive, market-oriented system. An example of this type of intensification later reversing was the arrival then departure of cotton production within a 15 year period. Maize is the current basis of the farm economy. • In the semi-arid areas where marketing and environmental conditions are marginal, however, an initial clearance of bush has been followed by low and in some cases

LUCID Working Paper 20 41 declining intensity of use. Land adjudication initially led to bush clearance, land sales and in-migration, but after it became clear that crop production and sales were undependable, out-migration especially by members of the poorest families has reduced labour availability. Some fields have reverted to bush or were planted with low-maintenance trees. Decisions on whether to stay or leave, or whether to work on one’s own farm or elsewhere, are linked to evolving local and national economies and can change quickly.

Political: Kenya government: land polices (adjudication, conservation), changing enforcement of regulations, elections/ stability, infrastructure investment, coffee & tea promotion, health of cooperatives/ parastatals International: environmental organisations, structural adjustment, aid, commodity prices Economic: Changing relative labour and economic returns to crops vs. animals, to farm vs. non-farm Changing markets for maize, coffee, charcoal, timber , miraa, horticulture Availability of off-farm employment Demographic: Local pop. growth and low land availabilityÆin-filling and intensification Migration: to look for land (upper to lower zones)Æextensification, to look for employment Falling birth rates, HIV-AIDS Social/ cultural: power from clans to families, changing gender roles, wealth differences, high value placed on education, on sacred forests Locational context: remoteness, marginal environment, strength of national ties Table 9. Major driving forces of land use change

Agricultural intensification and extensification processes, therefore, have been closely tied to national and international economic and political forces, and to non-agricultural sectors (Table 9). Rather than being a uniform success story, the impacts on poverty, soil management and fertility particularly in the semi-arid zones are mixed. This is somewhat different from in neighbouring Machakos, which benefited from high capital and technological investments over a sustained period of time (Tiffin et al. 1994).

Rather than population pressure being the main driver of extensification and intensification processes, population dynamics themselves have varied in response to changing circumstances. First short-distance migration to clear new land, then longer-distance moves to search for land in remote or less productive, drier areas, along with high rates of temporary labour migration, were critical responses to changing economic constraints and opportunities—a familiar pattern in East Africa (Gould 1995). Migration patterns were primarily from Embu and surrounding highland areas into Mbeere and were instrumental in affecting the clearance of new land. First the mid- elevation areas, then the semi-arid lower elevation zones, were destinations for migrants especially after the privatisation of land holdings permitted easier land sales. Some migrants have found the semi-arid areas risky and untenable for agriculture and have left. Meanwhile, temporary labour migration by members of the poorest households, and particularly from the semi-arid areas, is a common response to diversify the family’s income. Poverty rather than population pressure has been associated with out-migration. The poorest households are investing little of their outside income in the farm. Their poverty is being perpetuated by distress land sales and low investment

LUCID Working Paper 20 42 in children’s education. The pattern of labour migration, however, is bi-modal with the poorest and the richest families most engaged. A second demographic change has been a decline in population growth rates due to a decline in fertility rates, especially in Embu. The decline was expected, but the rapidity of the decline was not (UNFPA 2004). It was due to increased availability of family planning and a desire to have fewer children. Currently, therefore, the extensification of agriculture in the semi-arid zones is slowing, as both in-migration and local population rates decline. Future increases in the percent of land being cropped will be highly tied to relative returns of semi-arid agriculture vis a vis other activities. Development opportunities such as irrigation or introduction of a high value crop would change both the extensification and intensification dynamics.

The other major actor affecting land use and cover in the eastern Mt. Kenya region is the Kenyan government. It directly manages around 21% of the land surface—under parks, reserves, research land, and reservoirs. It, too, has been a very active agent affecting land cover. It reversed years of how it managed the Mt. Kenya forests, and started strictly enforcing regulations restricting logging, grazing, cultivation and other uses of the forest. During the thirteen year interval between the satellite images that were interpreted, one large section of forest was cleared, cultivated, and then allowed to revert to secondary forest growth. The reasons for these turns in governmental action include: 1) as a response to international and national pressure by environmental groups and activists; 2) power reversals at the local to national level (of individuals, agencies, political groups), and 3) attempts by a politician to please voters prior to an election. These types of reasons, and thus future land cover changes, are difficult to predict. International governmental forces affecting the agricultural and other sectors, such as structural adjustment and aid, are similarly difficult to predict.

Acknowledgements The authors would like to sincerely thank the following institutions for their generous support of this research: the United Nations Environment Programme, the Global Environment Facility, Michigan State University, the Rockefeller Foundation, the World Agroforestry Centre (formerly the International Centre for Research in Agroforestry), the Kenya Agricultural Research Institute, the Kenya Ministry of Agriculture, the Ohio State University, the International Livestock Research Institute, and the National Science Foundation under grant number 0308420.

LUCID Working Paper 20 43 V. BIBLIOGRAPHY Admassie, Yeraswarq. 1992. The Catchment Approach to Soil Conservation in Kenya. Regional Soil Conservation Unit Paper 6. Nairobi, Kenya: Regional Soil Conservation Unit Swedish International Development Authority. Berlekom, Maria, and Rolf Larsson. 1984. Soil Conservation in Kenya Technical and Socio- economic Aspects in Runyenjes Division, Embu District, Kenya. Nairobi, Kenya: Ministry of Agriculture, Soil and Water Conservation Branch. Bilsborrow, Richard E. 1987. Population Pressure and Agricultural Development in Developing Countries: A Conceptual Framework and Recent Evidence. World Development 15 (2):183-203. Bilsborrow, Richard E., and H.W.O. Okoth-Ogendo. 1992. Population-driven Changes in Land Use in Developing Countries. Ambio 21 (1):37-45. Binswanger, H. P. 1983. Agricultural Growth and Rural Non-Farm Activities. Finance and Development June:33-40. Blaikie, Piers. 1985. The Political Economy of Soil Erosion in Developing Countries. Essex: Longman Group Limited. Blaikie, Piers, and Harold Brookfield. 1987. Land Degradation and Society. London: Methuen & Co., Ltd. Boserup, Ester. 1981. Population and Technological Change: A Study of Long-Term Trends. Chicago: The University of Chicago Press. Brokensha, David. 1971. Mbere Clans and Land Adjudication. Institute for Development Studies Staff Paper 96. Nairobi, Kenya: Institute for Development Studies, University of Nairobi. Brokensha, David, and Bernard Riley. 1977. Vegetation Changes in Mbere Division, Embu. Institute for Development Studies Staff Paper 319. Nairobi, Kenya: Institute for Development Studies, University of Nairobi. Brokensha, David, and Bernard W. Riley. 1980. Introduction of Cash Crops in a Marginal Area of Kenya. In Agricultural Development in Africa, edited by R.H. Bates and M.F. Lofchie. New York: Praeger Publishers. Butt, Bilal, and Jennifer M. Olson. 2002. An Approach to Dual Land Use and Land Cover Interpretation of 2001 Satellite Imagery of the Eastern Slopes of Mt. Kenya. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper 16. Nairobi, Kenya: International Livestock Research Institute. Campbell, David J., and Jennifer M. Olson. 1991. Framework for Environment and Development: The Kite. CASID Occasional Paper 10. East Lansing: Center for Advanced Studies in International Development, Michigan State University. Crowley, Eve., and Simon E. Carter. 2000. Agrarian Change and the Changing Relationships Between Toil and Soil in Margoli, Western Kenya (1900-1994). Human Ecology 28 (3):383-414. Di Gregorio, Antonio. 2003. Africover Project Overview On Methodological Approach. Rome, Italy: Food and Agriculture Organisation. http://africover.org/. Emerton, Lucy. 1999. : The Economics of Community Conservation. Community Conservation Research in Africa, Principles and Comparative Practice 6. Institute for Development Policy and Management, University of Manchester. Fedders, Andrew, and Cynthia Salvadori. 1980. Peoples and . Nairobi: Transafrica. Gachimbi, Louis N. 2002. Technical Report on Soil Survey and Sampling: Embu - Mbeere Districts. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper 9. Nairobi, Kenya: International Livestock Research Institute. Gathaara, Gideon N. 1999. Aerial Survey of the Destruction of Mt. Kenya, Imenti and Ngare Ndare Forest Reserves, February - June 1999. Nairobi: Kenya Wildlife Service. www.kenya-wildlife-service.org. Glazier, Jack. 1985. Land and the Uses of Tradition among the Mbeere of Kenya. Landham: University Press of America.

LUCID Working Paper 20 44 Gould, W. T. S. 1995. Migration and Recent Economic and Environmental Change in East Africa. In The Migration Experience in Africa, edited by J. Baker and T. Akin Aina. Gotab, Sweden: Nordiska Afrika Institutet. Grigg, David. 1980. Population Growth and Agrarian Change: An Historical Perspective. Cambridge: Cambridge University Press. Haugerud, Angelique. 1983. The Consequences of Land Tenure Reform among Smallholders in the Kenya Highlands. Rural Africana 15-16 (Winter-Spring):65-89. ———. 1993. The Culture of Politics in Modern Kenya, Cambridge University Press. Hunegnaw, Tamiru. 1987. Technical Evaluation of Soil Conservation Measures in Embu District in Kenya. Nairobi, Kenya: Ministry of Agriculture Soil and Water Conservation Branch. Hunt, Diana. 1996. The impacts of individual land titling in Mbeere, Eastern Kenya. Discussion Papers in Economics 01/96. University of Sussex. Jaetzold, Ralph, and Helmut Schmidt. 1983. Farm Management Handbook of Kenya. Vol. II/C. Nairobi, Kenya: Ministry of Agriculture. Kanogo, Tabitha. 1993. Squatters and the Roots of Mau Mau. Nairobi: East Africa Educational Publishers. Kenya. 1970. Kenya population census, 1969. Nairobi, Kenya: Central Bureau of Statistics. ———. 1981. Kenya population census, 1979. Nairobi, Kenya: Central Bureau of Statistics. ———. 1994. Kenya population census, 1989. Nairobi, Kenya: Central Bureau of Statistics. ———. 2001. Kenya population census, 1999. Nairobi, Kenya: Central Bureau of Statistics. Kitching, Galvin. 1980. Class and Economic Change in Kenya: The Making of an African Petite- Bourgeoisie. New Haven: Yale University Press. Lambin, Eric F. 1997. Modelling and Monitoring Land-Cover Change Processes in Tropical Regions. Progress in Physical Geography 21 (3):375-393. Maitima, Joseph M., Simon M. Mbugua, Fred Atieno, and Stephen Mathai. 2004. Impacts of Land Use on Vegetation Species Composition, Distribution, Structure and Diversity: The Case of Embu and Mbeere Districts, Kenya. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper 32. Nairobi, Kenya: International Livestock Research Institute. Maitima, Joseph M., and Jennifer M. Olson. 2001. Guide to Field Methods for Comparative Site Analysis for the Land Use Change, Impacts and Dynamics Project. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper 15. Nairobi, Kenya: International Livestock Research Institute. Mathenge, Gakuu. 2004. The World Wants Kenya's Coffee? We Won't Sell. The East African, April 5, 2004. Mwai, Muthui, and Mwangi Ndirangu. 2003. Life expectancy on the decline; More people dying in Kenya today than was the case before 1990s. , December 5, 2003. Njeru, Enos H. N. 1978. Land Adjudication and its Implications for the Social Organization of the Mbere. Land Tenure Center Report 73. Madison, WI USA: Land Tenure Center, University of Wisconsin-Madison. Odingo, Richard S. 1971. The Kenya Highlands: Land Use and Rural Development. Nairobi, Kenya: East African Publishing House. Olson, Jennifer M. 1994. Farmer Responses to Land Degradation in Gikongoro, Rwanda. Ph.D., Geography Department, Michigan State University, East Lansing. ———. 1998. A Conceptual Framework of Land Use Change in the East African Highlands. Paper read at Earth's Changing Land: Joint Global Change and Terrestrial Ecosystems and Land Use and Land Cover Change Open Science Conference on Global Change, at Barcelona, Spain. Otuma, John. 2004. The Effects of Wildlife-Livestock-Human Interactions on Habitat in the Meru Conservation Area, Kenya. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper 39. Nairobi, Kenya: International Livestock Research Institute. Reardon, Thomas. 1997. Using Evidence of Household Income Diversification to Inform Study of the Rural Non-farm Labor Market in Africa. World Development 25 (5):735-747. Riley, Bernard, and David Brokensha. 1988. The Mbeere in Kenya. Vol. 1: Changing rural ecology. Lanham: University Press of America.

LUCID Working Paper 20 45 Smucker, Thomas A. 2002. Land Tenure Reform and Changes in Land Use and Land Management in Semi-arid Tharaka, Kenya. Land Use Change Impacts and Dynamics (LUCID) Project Working Paper 11. Nairobi, Kenya: International Livestock Research Institute. Swynnerton, R.J.M. 1954. A Plan to Intensify the Development of African Agriculture in Kenya. Nairobi: Government Printer. Tengberg, Anna, Linus Murithi, and Barrack Okoba. 1999. Land Management on Semi-Arid Hillsides in Eastern Kenya: Learning from Farmers' Diverse Practices. Mountain Research and Development 19 (4):354-363. Tiffin, Mary, Michael Mortimore, and Francis Gichugi. 1994. More People, Less Erosion: Environmental Recovery in Kenya. Chichester: John Wiley & Sons. Turner, B.L II, Goran Hyden, and Robert W. Kates, eds. 1993. Population Growth and Agricultural Change in Africa. Gainesville: University Press of Florida. UNFPA. 2004. Country Profiles for Population and Reproductive Health: Policy Developments and Indicators 2003. New York: United Nations Population Fund. Vanleewe, H., B. Woodley, C. Lambrechts, and M. Gachanja. 2003. Change in the State of Mt. Kenya Forest: 1999-2002. Nairobi, Kenya: Kenya Wildlife Service and the United Nations Environment Programme.http://www.unep.org/dewa/mountain/mtkenya- forest2.asp. Zelinsky, W. 1971. The Hypothesis of the Mobility Transition. Geographical Review 61:219-49. Zimmerer, K. S. 2000. The Reworking of Conservation Geographies: Nonequilibrium Landscapes and Nature-Society Hybrids. Annals of the Association of American Geographers 90 (2):356-369.

LUCID Working Paper 20 46 Annex 1 Land use and land cover classes for the Mt. Kenya region

Land Use Code General Land Use Type Specific Land Use Type Sub-Specific Land Use Type

1000 Agriculture - Small Scale 1100 Rainfed Cropping 1110 Tea 1120 Maize Dominant 1130 Mixed Bush/Crops 1140 Coffee 1200 Irrigated Cropping (Horticulture Dominant) 1300 Grazing Land (Bush and Grassland) 2000 Agriculture - Large Scale 2100 Rainfed Cropping (Wheat Dominant) 2200 Irrigated Cropping (Horticulture Dominant) 2300 Shamba System (Mix of Crops and Tree Plantations) 2400 Ranches 2500 Wheat and Grazing Pasture) 3000 Protected Areas 3100 National Parks 3200 National Reserves 3300 Forest Reserves 4000 Institutional Land Uses 4100 University Research Plot 4200 KenGen Land 4300 Don Bosco Farms 5000 Tree Plantations 6000 Urban and/or Built-up Areas 7000 Water Bodies 7100 Dams 7200 Lakes 8000 Non-Protected Forest Areas 8100 Non-Degraded 8200 Degraded

Land Cover Code General Land Cover Type Specific Land Cover Type Sub-Specific Land Cover Type

1000 Tundra/Mooorland/Glaciers/Grasses 2000 Forest 2100 Bamboo Forest 2200 Afro-Montane Forest 2300 Woodland (Open Canopy, mostly Dryland Forests) 2400 Tree Plantations 2500 Shamba System 2600 Degraded Forest 2700 Degraded Woodland 3000 Bush 4000 Cultivated Land 4100 Rainfed Cultivation 4110 Tea 4120 Maize Dominant 4130 Mixed Bush/Cultivation (Grains Dominant) 4140 Wheat and Pasture 4150 Coffee 4200 Irrigated Crops 4210 Rice 4220 Horticulture 5000 Urban and/or Built-up Area 6000 Water Bodies 6100 Dams 6200 Lakes 7000 Grassland

LUCID Working Paper 20 47 Annex 2. Identification of land cover and land use classes.

1. Agriculture - Small Scale a. Rainfed Cropping. Farm land including cultivated fields, fallow fields, grassland/ pasture, trees (often in lines), and houses. The farms are under 5 acres. Bright color especially where the soil is bare, and has a smooth texture (higher resolution) or is checkered (lower resolution). Houses and plot boundaries present. Zones of dominant crops include coffee/tea, maize, and other grains distinguished by spectral reflectance (from brightest to darkest, respectively), expert knowledge and ground truthing. b. The mixed bush/ crops class, found in the semi-arid zone, is identified by the presence of farm boundaries (light colored rectangles) within open bush land. c. Grazing Land (Bush and Grassland): open to dense grass intermixed with widely spaced shrubs or small trees 0.5-5.0 meters tall with individuals or clumps not touching. Medium to light color, smooth texture with irregularly spaced spots, absence of plot boundaries or houses.

2. Agriculture -Large Scale a. Rainfed cropping: large fields of cultivated land and maintained pasture. The fields are over 10 hectares each, with most around 100 hectares. Bright color, smooth texture, large uniform areas with linear boundaries. Zones of dominant crops include maize in one area, and wheat and grazing elsewhere distinguished by ground truthing and expert knowledge. b. Irrigated cropping: highly organized fields with vigorous new vegetation in otherwise dry areas. Some have standing water and one has a characteristic round spigot irrigation pattern. Bright vegetation in the infrared band. Types of crops grown identified during ground truthing (rice or horticulture). c. Shamba System: Very large rectangular fields (around 250 hectares each) within a Forest Reserve that are under crops or planted trees. The fields are planted mostly with potatoes and after two years are also planted with cedar or other tree seedlings that grow to maturity, to be later harvested. Distinguished by the fields’ large size, rectangular shape, road network, and location within a Reserve. Confirmed during ground truthing, expert knowledge and consultation with data from the KWS-UNEP study (Gathaara 1999). Bright reflectance in the infrared band, and smooth texture especially compared to the natural forest. d. Ranch. Large area of grassland under single ownership or management used for grazing. Distinguished by a large area of uniform, smooth coverage of grass within an area otherwise covered by small farms, and confirmed by consulting topographic maps.

3. Forest types and protected area status. a. Bamboo forest. Distinguished by location (above the tree line on Mt. Kenya), colour (bright in infrared) and texture (mostly smooth). Confirmed by the vegetation coverage in Gaathara 1999. b. Afro-montane forest: natural forest with complete canopy coverage and crowns interlocking. Distinguished by location (high altitude starting from 1600 to 2900 metres), color (mottled but bright in infrared) and contiguous surface with a rough texture. Confirmed by the vegetation coverage in Gaathara 1999. c. Woodland: Open stands of natural trees often over 5m tall with crowns not usually touching. Distinguished by location (in warmer and drier areas than the Forest, below approximately 1600 metres), colour (darker than the surrounding cropped areas in panchromatic, lighter in infrared) and texture (smooth with some mottling, no farm boundaries). Degree of degradation (or conversion to grassland) for most lowland hills confirmed during ground truthing.

LUCID Working Paper 20 48 d. Tree plantation: planted trees distinguished from forest by location, smoothness of texture, and presence of rectangular boundaries and roads. Their presence was confirmed during ground truthing and by consulting the coverage of Gaathara 1999. When plantations were within the Shamba System zone, they were included in the Shamba System designation. e. Degraded forest/ woodland: areas still mostly covered by forest or woodland, but whose canopy is not as complete as non-degraded, and with visible bare soil in woodlands. Distinguished in the forest zone by texture (e.g., in 2001 in Imenti smoother than the rough non-degraded forest because of secondary growth of uniform age, and in the woodlands smooth where trees had been removed) and by color (in woodlands, degraded areas are darker than non-degraded in infrared band, darker in panchromatic). The status was confirmed during ground truthing. f. Protected area status for the land use classification: the management status of the forests and woodlands (National Park, National Reserve, Forest Reserve or not government protected) was obtained from topographic maps. 4. Bush. Open to dense grassland intermixed with shrubs or small trees 0.5-5.0 meters tall with individuals or clumps not touching. Medium to light color, smooth texture with some spots, absence of plot boundaries or houses. 5. Grassland: Open grass distinguished by uniform, smooth area and confirmed during ground truthing and by consulting topographic maps. 6. Institutional land uses: large expanses of land managed by an institution. Distinguished by their uniform color and texture with clear boundaries often within heterogeneous cropped areas, and confirmed by ground truthing, consulting topographic maps and expert knowledge. 7. Urban areas: Distinguished by very light color in panchromatic, and by location. 8. Water bodies: distinguished by dark color and smooth texture.

LUCID Working Paper 20 49 Annex 3. Population Table of Embu and Mbeere Districts 1969 to 1999 1969 1979 1989 1999 % Change Ann. Growth Rate Admin Area 1969- 1979- 1989- 1969- 1979- 1989- Pop. Km2 Density Pop. Km2 Density Pop. Km2 Density Pop. Km2 Density 79 89 99 79 89 99 EMBU (includes Embu Municip., 116,505 769 152 171,136 492 348 234,735 708 332 278,196 792.4 351 46.9 37.2 18.5 3.92 3.21 1.71 Mbeti North & Mt. Kenya Forest) Central (Mbeti No. + Embu M.) 15,087 58 260 23,005 64 359 37,168 64 581 52,446 70.6 743 52.5 61.6 41.1 4.31 4.91 3.50 Mbeti North 11,159 55 203 7,019 40 175 10,554 40 264 15,526 47.1 330 -37 50.4 47.1 -4.53 4.16 3.94 Embu Municipality 3,928 3 1,314 15,986 24 648 26,614 24 1,109 36,920 23.5 1,571 307 66.5 38.7 15.07 5.23 3.33 Kyeni 21,427 98 219 30,945 95 326 41,374 97 427 48,385 104.9 461 44.4 33.7 16.9 3.74 2.95 1.58 Kyeni North 9,450 28 338 15,549 35 444 17,100 26 658 20,533 29.3 701 64.5 10.0 20.1 5.11 0.96 1.85 Karurumo 5,675 34 167 8,632 33 262 12,060 34 355 11,935 44.5 268 52.1 39.7 -1.0 4.28 3.40 -0.10 Kyeni South 6,302 36 175 6,764 27 251 12,214 37 330 15,917 31.1 512 7.3 80.6 30.3 0.71 6.09 2.68 Runyenjes 28,809 158 182 41,579 145 287 56,777 140 406 64,111 148.5 432 44.3 36.6 12.9 3.74 3.16 1.22 Kagaari North 9,000 43 209 14,567 42 347 20,786 42 495 24,608 43.0 572 61.9 42.7 18.4 4.93 3.62 1.70 Kagaari South 6,572 81 81 10,396 72 144 15,554 67 232 16,691 73.9 226 58.2 49.6 7.3 4.69 4.11 0.71 Runyenjes Town 13,237 34 389 16,616 31 536 20,437 31 659 22,812 31.6 722 25.5 23.0 11.6 2.30 2.09 1.11 Manyatta 32,726 107 306 46,894 102 460 62,440 109 573 71,332 107.1 666 43.3 33.2 14.2 3.66 2.90 1.34 Gaturi North 6,098 18 339 8,272 17 487 10,557 19 556 12,252 18.5 662 35.7 27.6 16.1 3.10 2.47 1.50 Ngandori 10,552 31 340 14,367 30 479 17,801 30 593 22,258 32.6 683 36.2 23.9 25.0 3.13 2.17 2.26 Nginda 11,442 42 272 17,239 35 493 24,150 40 604 27,763 39.6 701 50.7 40.1 15.0 4.18 3.43 1.40 Ruguru 4,634 16 290 7,016 20 351 9,932 20 497 9,059 16.4 552 51.4 41.6 -8.8 4.23 3.54 -0.92 Nembure 18,406 99 186 28,713 86 334 36,902 86 429 41,590 88.1 472 56.0 28.5 12.7 4.55 2.54 1.20 Gaturi South 5,038 31 160 10,227 26 390 12,550 25 502 12,857 22.0 584 103.0 22.7 2.4 7.34 2.07 0.24 Kithimu 6,900 47 148 9,861 42 236 13,258 42 316 17,251 47.1 366 42.9 34.4 30.1 3.64 3.00 2.67 Makengi 6,468 21 308 8,625 18 479 11,094 19 584 11,482 19.0 604 33.3 28.6 3.5 2.92 2.55 0.34 Mount Kenya Forest 50 249 0 74 212 0 332 210.2 2 348.6 16.20

LUCID Working Paper 20 50

1969 1979 1989 1999 % Change Ann. Growth Rate Admin Area 1969- 1979- 1989- 1969- 1979- 1989- Pop. Km2 Density Pop. Km2 Density Pop. Km2 Density Pop. Km2 Density 79 89 99 79 89 99 MBEERE (does not include 62,407 2,102 30 92,037 1,995 46 135,403 2,097 65 170,953 2,092.5 82 47.5 47.1 26.3 3.96 3.94 2.36 Embu M. or Mbeti North) Siakago 11,207 368 30 16,522 361 46 25,982 364 71 34,330 367.3 93 47.4 57.3 32.1 3.96 4.63 2.83 Gitiburi 1,982 49 40 3,197 43 74 4,955 49 101 6,815 49.2 139 61.3 55.0 37.5 4.90 4.48 3.24 Muminji 2,606 92 28 3,858 108 36 5,602 104 54 7,142 103.7 69 48.0 45.2 27.5 4.00 3.80 2.46 Mutitu 1,891 146 13 2,401 128 19 4,625 127 36 5,993 127.6 47 27.0 92.6 29.6 2.42 6.78 2.63 Nthawa 4,728 81 58 7,066 82 86 10,800 84 129 14,380 86.8 166 49.5 52.8 33.1 4.10 4.33 2.90 Evurore 16,476 411 40 21,708 412 53 30,950 419 74 36,841 410.0 90 31.8 42.6 19.0 2.80 3.61 1.76 Ishiara 5,332 130 41 6,683 127 53 10,109 135 75 11,556 135.0 86 25.3 51.3 14.3 2.28 4.23 1.35 Kanyambora 4,117 56 74 6,173 55 112 8,234 47 175 9,924 46.5 213 49.9 33.4 20.5 4.13 2.92 1.88 Kiangombe 2,882 72 40 3,756 69 54 5,831 64 91 7,061 63.4 111 30.3 55.2 21.1 2.68 4.50 1.93 Ndurumori 4,145 153 27 5,096 161 32 6,776 173 39 8,300 165.1 50 22.9 33.0 22.5 2.09 2.89 2.05 Gachoka (does not include Embu M.) 21,809 807 27 31,165 797 39 45,560 806 57 59,102 800.3 74 42.9 46.2 29.7 3.63 3.87 2.64 Mbeti South 4,068 171 24 6,724 178 38 10,556 177 60 13,695 169.4 81 65.3 57.0 29.7 5.15 4.61 2.64 Kiambere 3,471 282 12 4,648 271 17 7,429 282 26 10,810 280.9 38 33.9 59.8 45.5 2.96 4.80 3.82 Kianjiru 7,042 122 58 9,766 138 71 13,002 123 106 17,457 138.9 123 38.7 33.1 34.3 3.32 2.90 2.99 Mavuria 7,228 232 31 10,027 210 48 14,573 224 65 17,140 211.1 81 38.7 45.3 17.6 3.33 3.81 1.64 Mwea - 1999 12,915 516 25 22,642 425 53 32,911 508 65 40,680 514.9 79 75.3 45.4 23.6 5.77 3.81 2.14 Makima 6,630 121 55 10,334 337 31 12,822 342.7 37 55.9 24.1 4.54 2.18 Karaba 6,018 88 68 8,444 90 94 12,463 82 152 15,786 83.5 189 40.3 47.6 26.7 3.45 3.97 2.39 Riakanau 6,897 428 16 7,568 214 35 10,114 89 114 12,072 88.7 136 9.7 33.6 19.4 0.93 2.94 1.79 Notes: District, division and location boundaries changed over the four censuses. The 1999 census boundaries were used as reference and previous years’ figures adjusted accordingly as well as could be estimated (figures in italics have been adjusted to account for the changes). The 1999 population density was re-calculated for Embu. Prepared by Marian Mitchell, April 7, 2004. Source: Kenya (1970, 1981, 1994, 2001)

LUCID Working Paper 20 51