sannethesis_moz

8/2/2019 sannethesis_moz

1/114

Exploring feasible yields for cassava production for food and fuel in the

context of smallholder farming systems in Alto Molcu,Northern Mozambique


2/114


3/114

Exploring feasible yields for cassava production for food and fuel in the context of smallholder

farming systems in Alto Molcu, Northern Mozambique

Sanne van den Dungen

MSc Thesis Plant SciencesPPS 80436

36 credits

June 2010

Supervisors: Ir. Sander de Vries

Dr. Ir. Gerrie van de Ven


4/114

Preface and acknowledgements

From the first contact I had with Ir. Sander de Vries, Dr. Ir, Gerrie van de Ven and Prof. Dr. Ken Giller, Iwas able to express my personal motivations and wishes concerning this thesis. I would especially like to

thank Sander and Gerrie for their patience and support during the length of this thesis. I was able to

shape the research in, what for me was the reason to come to Wageningen: explorative research on

farming systems in Sub-Saharan Africa. I am very happy to have been able to pursue this dream.

There are many people I would like to thank for making the four months of fieldwork research possible,

in Northern Mozambique 2009. First of all I would like to thank Sicco Kolijn for his contaminatingenthusiasm and motivation. I admire his network and am thankful for the help I received in finding a

suitable region for research, for hospitality and concerns. Because of his connections I was able to link

up to World Vision Quelimane under the supervision of Brian Hilton. I thank Brian for his support in

introducing me to the district of Alto Molcu and World Vision staff and thank World vision for all the

support and facilities I received. At World Vision Alto Molcu, I met Sansao Honwana, who has to been

much more than a supervisor to me and providing me with a place to stay and all the support from his

staff I could ask for. Thanks Raol for countless Saturdays picking me up from a weeks work in thevillages even though it was your free day. In Mugema, Nacuaca and Gafaria I thank all the farmers co-

operating in the research and being patient with me. Most of all I am grateful for the insights I was able

to get besides their farming systems: the everyday of life, the stories, the food, traditions and customs.

Antonio, Pedro and Gustodio have been amazingly patient with me during the length of fieldwork,

Trying to answer all my questions and providing me with solid friendship and support. The guest

families: the family of Antonio, Donna Celesta and the family of Xavier I show gratitude to their

hospitality for letting me camp at their premises and joining their meals. I had the honour to have BrunoGolden as my translator in the beginning of the fieldwork. There is absolutely no way of expressing my

appreciation to you. Many times during fieldwork I have thought about how to be able to show to you

what it was like having you to help me I trust you have felt it and I could not have wished for a better


5/114

SummaryCurrently the worlds energy supply system for the 21st century is under intensive debate. Renewable

energy sources, especially biomass, play an important role in the discussion where several scientists

have tried to identify high potential areas for the production of biomass for energy. African countries

such as Mozambique are currently under review, because of a favourable climate, ample land

availability and low population density (Batidzirai et al., 2006). Cassava, a potential crop for the

production of bio-ethanol, is currently being discussed as a feasible option for biofuel production by

smallholder farmers as well as maize and sorghum. However current (cassava) yields are low and

farmers have very limited resources for yield improvement.

This study was developed with the overall goal to explore feasible yields for cassava production for food

and fuel in the context of smallholder farming systems in Alto Molcu, Northern Mozambique with the

objectives of i) assessing the heterogeneity between farms by making a rapid farm characterisation, ii)

estimating current yields of cassava, sorghum and maize of selected smallholder farmers, iii) explaining

current yields of cassava, sorghum and maize, iv) making a yield gap analysis between actual yields from

selected crops from field estimations and feasible potential yields simulated using the FIELD model, v)

collecting parameter input used in the FIELD model.

Three villages (Mugema, Nacuaca and Gafaria) were selected in the district of Alto Molcu to represent

variability at district level. Experts, key-informants, back ground information, and first approach farm

characterisation interviews were used to describe and categorize bio-physical, socio-economic and farm

management practices variability found at village, farm and field level. A farm typology based on expert

knowledge and K-means clustering was developed to analyse variability found at farm level. A dynamic

simulation model FIELD was parameterized with field data, and simulation scenarios were developedfrom the rapid farm characterisation interviews and informal meetings with farmers. The model was run

to study and quantify the cassava yield response of different farm management practices. Due to

circumstances model simulations were only based on soil fertility properties of the sampled fields


6/114

However a large part of the same respondents would use fire for clearing of the field: 59%, 80% and 50%

in Mugema, Nacuaca and Gafaria. The effect of maize residue application on cassava fresh yield was

analysed with the help of the simulation model FIELD. Addition of maize residues had a positive effecton fresh cassava yield with an increase in yield of between 156 237kg/ ha in Nacuaca and 189-449

kg/ha in Gafaria over the three quartiles used per ton maize residue applied.

In general most farmers would only consider the use of manure if they would produce horticultures.

Generally manure was piled up and removed not to be of further use. Simulated yields comparing no

manure and manure application showed an increase of 402-482 kg and 433-1034 kg increase/ ton DM

manure applied for Nacuaca and Gafaria subsequently.

Average yield response per kg applied nutrient was highly variable comparing villages and comparing

quartiles. Increasing levels of potassium was most effective on soils in Nacuaca, while a higher response

(increase kg fresh cassava / kg applied N, P or K ha -1) for phosphorus was found in Gafaria. First and

second quartile classified fields had a lower response and remained at a low yield level compared to the

higher yielding fourth quartile fields. High increase in yield was obtained with NPK fertiliser application

(100:22:83 N:P:K) ranging from 37 to 65 kg yield increase per kg fertiliser added in Gafaria and between

49 and 67 kg yield increase per kg fertiliser added in Nacuaca. Improved fertiliser management was

developed to increase yield response per kg fertiliser applied with the help of available nutrients from

crop residues and manure. An increase in yield of 6, 21, and 35% was attained for 4 th quartile-2nd & 3 rd

quartile and 1st quartile fields for Nacuaca. For Gafaria yield increase due to improved management was

even higher: 22, 34 and 40% yield increase for 4th quartile-2nd & 3rd quartile and 1st quartile fields.

Ranked as the most important crop by 60% and 70% of the respondents in Nacuaca and Gafaria

respectively, cassava plays an important role for farmers in the research villages. Only one third of the

farmers in Gafaria (26%) sold cassava compared to 42% in Nacuaca and 48% in Mugema. Overall cassava

was mostly sold within the community (82% of respondents). When sold cassava was sold as less than


7/114

SUMMARY .................................................................................................................... 5

1.1 Introduction to Mozambique ................................................................................................................... 11

1.2 Agro climatic conditions and land use .................................................................................................... 11

1.3 Poverty and agriculture ........................................................................................................................... 12

1.4 Potential crops for bio fuel: cassava, maize and sorghum ...................................................................... 121.4.1 Reasoning in favour of biofuel production ............................................................................................. 131.4.2 Reasoning against biofuel production .................................................................................................... 14

1.5 Sustainability ........................................................................................................................................... 14

1.6 Introduction to research area ................................................................................................................. 151.6.1 Introduction to Zambzia province .................................................................................................... 151.6.2 Alto Molcu ........................................................................................................................................ 16

1.7 Objectives ................................................................................................................................................. 18

1.8 Theoretical background ........................................................................................................................... 19

1.9 Outline of the thesis ................................................................................................................................. 20

2. MATERIALS AND METHODS ................................................................................. 212.1.1 Introduction to work phases and working steps ..................................................................................... 212.1.2 Study area: Bio-physical and socio-economic description ...................................................................... 23


8/114

2.6 Statistical analysis .................................... ..................................... ...................................... ..................... 30

2.7 Dynamic simulation of crop performance and nutrient balances at field scale .................................... 312.7.1 Model for dynamic simulation at field scale: FIELD ............................................................................. 312.7.2 Developing scenarios .......................................................................................................................... 33

3 RESULTS ................................................................................................................. 35

3.1. Characteristics at village level ................................................................................................................. 35

3.1.1 Village biophysical variability ............................................................................................................... 353.1.2 Socio-economic variability at village level ............................................................................................ 393.1.3 Land use and management practices ...................................................................................................... 47

3.2 Between farm variability ......................................................................................................................... 523.2.1 Short characterisation per farm type ....................................................................................................... 52

4. DYNAMIC SIMULATION OF THE IMPACT OF MANAGEMENT ON CASSAVA

YIELDS ........................................................................................................................ 554.1 Introduction ........................................................................................................................................... 554.1.1 Initial simulated yields: Base run ........................................................................................................... 554.1.2 Nutrient yield response .......................................................................................................................... 59

4.2 Management simulations ......................................................................................................................... 604.2.1 Cassava residues .................................................................................................................................... 604.2.2 Maize residue ........................................................................................................................................ 62

4.2.3 Manure application ................................................................................................................................ 634.2.4 Fertiliser use .......................................................................................................................................... 654.2.5 Cost benefit analysis of fertiliser use ...................................................................................................... 654 2 5 Cost benefit analysis of fertiliser use 66


9/114

5. DISCUSSION .......................................................................................................... 73

5.1. Evaluating the effect of farm management on cassava yield .................................................................. 735.1.1 Incorporation of crop residues ............................................................................................................... 735.1.2 Manure application ................................................................................................................................ 745.1.3 Use of fertiliser ...................................................................................................................................... 745.1.4 Feasibility of yield increase scenarios .................................................................................................... 765.1.5 Model FIELD considerations ................................................................................................................. 77

5.2 Evaluation of feasibility .......................................................................................................................... 785.2. 1 Potential crops for biofuel; cassava, maize and sorghum .................................................................... 78

5.3.Methological considerations ........................................................................................................................... 805.3.1 The use of a farm typology .................................................................................................................... 805.3.2 Yield estimations ................................................................................................................................... 815.3.3 Limitations of research .......................................................................................................................... 81

6. CONCLUSIONS ................................................................................................. 83

REFERENCES ............................................................................................................ 85


10/114

APPENDIX I: ADDITIONAL BACKGROUND INFORMATION .................................... 88

APPENDIX II: SOIL FERTILITY ANALYSIS ............................................................... 89

APPENDIX III: SOCIO-ECONOMIC ANALYSIS ......................................................... 91

APPENDIX IV: FARMERS ESTIMATIONS AND QUEFTS MAIZE YIELD

ESTIMATIONS ............................................................................................................ 94

APPENDIX IV (CONT): CASSAVA AND SORGHUM YIELD ESTIMATIONS ............ 95

APPENDIX V: FARM TYPOLOGY .............................................................................. 96

APPENDIX VI: MODEL ASSUMPTIONS .................................................................... 98

APPENDIX VII: FIELD SIZE CASSAVA AND MAIZE ................................................. 99

APPENDIX VIII: CASSAVA AND MAIZE YIELD ESTIMATIONS .............................. 100

APPENDIX IX: FARM TYPOLOGY CONSIDERATIONS .......................................... 103

APPENDIX X: QUARTILE CONSIDERATIONS ........................................................ 104


11/114

1 Introduction

1.1 Introduction to MozambiqueMozambique is located in southern Africa and has around 80 Mha of land and a coastline of 2500 km, a

population density of 22/km2 and approximately 19 million inhabitants. Projections estimate an increase

to 28 million inhabitants by the year 2025 (Batidzirai et al., 2006). Mozambique has been reported as

one of the poorest countries despite the high economic growth

during the 1990s and has only recently been recovering from 16

years of devastating civil war and resulting famine (Bias and

Donovan, 2003 ; Maria and Yost, 2006). An estimated 80% ofMozambican labour is currently used in agriculture; productivity

however is low as agriculture only contributes up to of the GDP

(Bias and Donovan, 2003). According to Wils (2002) three macro-

socio-ecological zones can be defined in Mozambique: The Northern

region (Niassa, Cabo Delgado and Nampula), The Central region

(Zambzia, Tete, Manica and Sofala) and the Southern region

(Inhambane, Gaza and Maputo, see Figure 1). The Northern region

covers around 50% of the total land area and is inhabited by 33% of

the population. The Central region consists out of 29% of the total

land area with 41% of the Mozambican population. Finally the

Southern region forms 21% of the total land area and provides a

home for 26% of the population. These regions were derived based

on the existing administrative, demographic and hydrologic characteristics of the country.

1.2 Agro climatic conditions and land useThe county exhibits great diversity in geological and climatic diversity: from low lying coastal plains

( 200 ) h id i l i i ( 1000 ) d i f ll i f 00 i h

Figuur 1. Overview of Mozambique's

provinces

Figure 1. Overview of Mozambique's

provinces


12/114

1.3 Poverty and agriculture

In 2000, 70% of the Mozambican population lived below the poverty line (UNDP, 2000). Around three

million families are farming in the rural area and obtain 70% of their food from their own on-farm

production (MAM and MEM, 2008). Interestingly enough the poor and the less poor (classification used

by Bias and Donovan, 2003) have approximately the same amount of land per household (each plot of

land is called a machamba). The differences however lie in the amount and the type of input the poor

and less poor can use on their land. For example the amount of land irrigated is higher for the less poor.

Furthermore questionnaires (MINAG, 2007) show that only 4% of the farmers use fertiliser; generally,

the amount of inputs used in Mozambique is very low which is reflected in low crop yields. The low

usage of inputs such as improved seed, fertiliser and technology is mainly due to limited capital,

mechanization and poor access to financial support for the farmers. The Mozambique Biofuel

Assessment (MAM and MEM, 2008) concludes that the amount of land is not a limiting factor for poor

peasants, but rather the capacity to work the land. The average maize yield for Mozambique is

approximately 0.9 tons/ha, which is the lowest yield in Southern Africa: Malawi (1.7 tons/ha), South

Africa (2.8 tons/ha), Swaziland (1.5 tons/ha), Tanzania (1.2 tons/ha), Zambia (1.5 tons/ha) and

Zimbabwe with 1.5 tons/ha (FAO, 2005). Compared to the global average of 4.5 tons/ha these averages

are low, in European countries the production is overall higher and records exist of 15 tons/ha (MAM

and MEM, 2008).

Mozambican agriculture can be characterized by a large number of small-scale producers that have to

rely on rain fed production systems for their subsistence farming. Most important crops grown by small

holder farmers in Mozambique are (in order of importance): maize, cassava, cowpeas, groundnuts,

sorghum, rice, bambara beans/nuts, pigeonpeas , sweet potato, common beans, and millet (Bias andDonovan, 2003). Maize, cassava and sorghum are grown by 79%, 63% and 27% of the of small holder

farm households with a national production of 150, 755 and 34 metric tons subsequently (Bias and

2003 I G 200 ) i d h b h d


13/114

Sorghum provides the staple food of many of the worlds poorest and most food insecure people in the

semi-arid tropics of the world. Because of its genetics sorghum is able to grow in hot and dry agro-

ecological zones where it would be difficult to grow other food grain crops. Sorghum can produce grainsand stover in areas with frequent droughts (Borikar et al., 2007). Sweet sorghum (Sorghum bicolor L.

Moench) is similar to grain sorghum in appearance and agronomic performance. Because of its C4metabolism sorghum is photosynthetically efficient. Like sorghum, sweet sorghum can be cultivated

over a wide range of environments; the difference between the two is that sweet sorghum stores much

of its photosynthates as sugar in the stalks. However sweet sorghum also produces reasonable grain

yields. According to Reddy (2007) sweet sorghum could be grown on the 23.4 million hectares of dry

land in Africa (55% of global sorghum area) where it could produce more biomass and grain if yield-enhancing technologies would be stimulated by the bio fuel market.

Maize is produced almost throughout the country, however most of the production (70%) is

concentrated in the central region and in the north (MINAG, 2007). As sorghum, maize is a C4 crop

which under semi arid ecological conditions is favourable.

Government permission for using these three crops for biofuels is currently heavily debated in

Mozambique. However, in the scope of explorative research they will still be considered in this research,

but the main focus will be on cassava.

1.4.1 Reasoning in favour of biofuel production

Currently the worlds energy supply system for the 21st century is under intensive debate. Renewable

energy sources, especially biomass, play an important role in the discussion where several scientists

have tried to identify high potential areas for the production of biomass for energy. African countries

such as Mozambique are currently under review (Batidzirai et al., 2006). Large scale biomass conversionto for example ethanol could offer a sustainable alternative for part of the current African fuel

consumption or provide export opportunities. Scenario studies estimate biomass production potential


14/114

1.4.2 Reasoning against biofuel production

In large parts of the country maize, cassava and sorghum are considered as staple food hence using

these crops for ethanol production could jeopardize food security. From surveys held in 96/97 it wasconcluded that a large amount of farmers growing the primary crops mentioned above were not able to

sell them: both in rural areas as for the urban poor with agricultural land. Cassava was grown

throughout the country but rarely sold (Bias and Donovan, 2003). Prices of those crops could rise and

availability could be limited creating problems for those without land or insufficient production. Local

processing mills that transfer maize into meal for household consumption using raw material could be

put into problems by the demand for maize for ethanol (MAM and MEM, 2008). However if production

would increase, maize could be considered for ethanol production, especially in certain parts of thenorth and the south where sorghum accounts for a larger part of peoples diet. It will be important to

include the impact of a diet crop being used for ethanol in that dietary region on household food

expenditure in the modelling since this is lacking in previous studies. For all crops investigated (maize,

cassava and sorghum) household consumption and surplus for the market therefore needs to be

identified. One of the main biophysical problems in the production of the crops mentioned above is soil

fertility.

1.5 Sustainability

Since its introduction in 1987 by the Brundtland Commission the concept of sustainability has been of

increasing importance in science, policy formulation and practice. In the Brundtland report sustainability

has been defined as meeting the needs of the present generation without compromising the ability of

future generations to meet their needs (Brundtland, 1987). However since the introduction of the

concept more than 300 different definitions have been developed. A shared characteristic amongst

these is the notion that sustainability is aimed at improving both the current and the future quality oflive. In the scope of scenario studies for biofuel production for the future, sustainability of farming

systems is a key factor. Sustainable agriculture is a form of agriculture concerned about the ability of an

t t i d ti i th l t S t i bl i lt h th j t


15/114

1.6 Introduction to research area

Zambzia province was selected as an interesting province for conducting the research. Below the

province and selected district will be further discussed, in section 2.1 the selection of the villages within

Alto Molcu district will be explained.

1.6.1 Introduction to Zambzia province

Zambzia province produces more than one quarter of the countries cassava, about 18% of the total

maize production and 15% of the countrys total sorghum production and this share remains stable over

the years (Bias and Donovan, 2003; MAM and MEM, 2008). In total Mozambiques area covers 79 mln

hectares of which 12% is arable land suitable for cropping (MAM and MEM, 2008). Zambzia provinceoccupies 105.000 km2 of land with 27.500 km2 suitable for arable production of which 23% is cropped

with cassava, 37% with maize and 5% with sorghum (Statoids, 2007). Other crops grown are pigeonpea

and rice (Table A 1 in the Appendix). Around 26% of the area of Zambzia province is suitable for arable

production, which is the highest percentage compared to other provinces such as Nampula (24%)

Maputo (17%) and Inhambane (14%) (MAM and MEM, 2008).

The highest yields per ha for cassava in Mozambique have been recorded in Zambzia and Nampula

province (Bias and Donovan, 2003). Average production ranges from 6.4 t/ha in Southern Province Gaza

to about 9.0 and 10 ton/ha in more Northern Provinces Zambzia and Nampula respectively (Table 1).

High yield losses ranging from 21-33% are mainly due to prevalence of diseases and plagues such as

cassava brown streak, cassava mosaic virus and termites. Sorghum yield data are unavailable but

estimated around 0.3-0.6 t/ha with a potential between 0.8 and 2 t/ha. For maize a range of 0.4-1.3 t/ha

and an estimated potential of 5-6 t/ha is estimated by the Ministry of Agriculture and Fisheries (MAP,

1997).

Table 1. Yields of fresh cassava (t/ha) per Province (IITA, 2003)


16/114

1.6.2 Alto Molcu

Alto Molcu district is one of the largest districts of Zambzia province and covers an area of 6.386 km2

which equals 6% (DRAM, 2007; Molcu, 2008). Located in the north of the province in between

Quelimane in the south (circa 400km) and Nampula in the north (219km) at national road number one,

latitude 1615 O and longitude 3715 W). The district borders in the north with the river Ligonha, which

separates Zambzia from Nampula province, in the south by the district of Ile, in the west by the district

of Guru and in the east by the district of Gile (Figure 2).


17/114

The district of Alto Molcu is characterized by its mountains (900m) in the west in the direction of Ile

and Guru, its plains (200-300m) in the south and centre and low riverbeds of the main river Molcu

that explain the name of the district (alto = high in Portuguese). Different crops can be grown accordingto the variation in elevation. Along the rivers: rice, coconut, sugarcane, tubers, fruits and vegetables and

grassland. The plains make up 2/3 of the districts surface undulating gradually in the direction of the

mountainous zones. Primary crops grown here: maize, sorghum, beans, sunflower, cotton, sesame,

sweet potato, cassava, fruits, cashew, tobacco and horticultures as well as grass for cattle. The

mountainous zone is cropped with: coffee, maize, sorghum, Irish potato, sweet potato, beans, sesame,

soy, cassava and fruits. Because of its differences in elevation two climatic zones can be identified:

tropical humid (90-100% humidity) and a tropical mountainous climate. Average yearly temperaturevaries between the plains and the mountainous area around 26C and 20C respectively. Two seasons

can be identified in the two climatic zones present in the district: the hot and rainy season from October

till April and the dry and fresh season from Mai until September. Yearly rainfall varies between 1000 mm

on the plains and 1300 mm in the mountains and is distributed in a uni modal pattern with peaks from

December to March (Figure 3)(DRAM, 2007).


18/114

1.7 Objectives

The overall goal of this work was to explore feasible yields for cassava production for food and fuel in

the context of smallholder farming systems in Alto Molcu, Northern Mozambique. This was done by

meeting with the following objectives in three phases.

Phase I objective was aimed at a concise literature study to provide with agronomic background

information on Mozambique and to explore the debate about potentials and drawbacks of biofuel

potential. The second phase was aimed at providing with information and insights from field work. The

third phase needed the input of the first and second phase for scenario development.

Phase I

- To make an exploration in literature to give agronomic background information in Mozambique

and to explore the biofuel debate in Mozambique

Phase II

- To asses heterogeneity between farms by making a rapid farm characterisation

- To estimate current yields of cassava and sorghum of selected smallholder farmers

- To explain current yields of cassava and sorghum of selected smallholder farmers in relation to

soil, landscape and management variability

- To collect parameter input to be used in model FIELD

Phase III

- To make a yield gap analysis between actual yields from field research and feasible yields

simulated using model FIELD


19/114

1.8 Theoretical background

The methodological approach followed in this thesis assumes that the sources of variability found atdifferent levels and scales are nested within each other: the smaller scale (e.g. field) variability

contributes to variability at larger scale (farm or village). Figure 4 illustrates how in a given region

comparisons can be made at village level: capturing average farm and field characteristics and

comparing between study villages, a step further downscaling farms can be compared using a developed

typology of farms within and between villages. Subsequently fields can be compared between

typologies and between villages.

Figure 4. Illustration of conceptual approach of variability at different levels in the research (for an n amount of

fields, farms and villages in a given region), arrows indicate comparisons between components

F t l


20/114

The positivist method followed as the second step in which the output of the rapid farm

characterisation formed the basis of designing a typology.

1.9 Outline of the thesisThe Sections 1.1 to 1.6 have provided the background information on Mozambique and the biofuel

debate. This is used as a starting point for the rest of the thesis. Chapter 2 introduces the methodology

used during the different phases and working steps and explains in several subchapters at which level

analysis are done. Chapter 3 is divided into three parts aimed at describing farm heterogeneity at

different scales of analysis. Part 1 (Section 3.1) analyses results found at village level: comparing

between villages in terms of soil fertility, socio-economic characteristics, production activities and

managerial aspects. Part 2 (Section 3.2) focuses on analyzing differences found comparing within

villages based on the designed typology. Subsequently Chapter 4 presents the results of FIELD model

simulations on cassava crop yields under different farm management scenarios. Chapter 5 is used for a

general discussion and Chapter 6 for concluding remarks and recommendations. Each chapter gives a

brief introduction about its contents and when indicated also a short summary at the end to make some

important concluding remarks to facilitate interpretation of the chapters. Many tables can be found in

the Appendices; within the main text explicit reference to the appendix is made.


21/114

2. Materials and methods

2.1.1 Introduction to work phases and working steps

Three work phases and four main working steps can be identified in this thesis to cope with the

variability found at different scales and to comply with the multiple objectives presented earlier

(Chapter 1.7). Figure 5 below gives a schematic overview of the work phases, working steps and the

interactions between the different components.

The first phase (Phase I) consisted of an elaborate literature review to provide a solid background for the

start of fieldwork. It provided necessary knowledge on the methods needed in the following steps of the

research such as the design of a questionnaire for the rapid farm characterisation and how to develop a

farm typology.

The second phase was aimed at field work and analysis of the data collected. Each step used a previous

step for input and provided information for the following one (Figure 5) For example previous to Step 1

experts were consulted and introduction visits to the area were made that provided the input needed to

start the rapid farm characterisation in the 3 villages. With the help of key informants, 25 farmers ineach village were chosen (74 in total, one missing). Socio-economic and managerial data was collected

as well as farmers yield estimations. The first approach questionnaire was used for this step (Appendix

A 23). The output of Step 1 served as a basis for Step 2: the typology. After development of this farm

typology selected farmers (a subset of 10 per village) was revisited for a second and third round of

questionnaires to cross-check and validate previous data and answers. During these second and third

visits soil samples were taken and area and harvest measurements for the crops of focus were done. As

a third step results were analysed at multiple scales for the villages, farms and fields (Figure 4).

Phase III continues at the end of step 3 from phase II: the analysis provided a framework for feasible

i f i d t b f fi ld d f l l M i ki t 4 i h III


22/114


23/114

2.1.2 Study area: Bio-physical and socio-economic description

The district of Alto Molcu was chosen before arrival in collaboration with the locally, nationally and

internationally active aid organization World Vision and agricultural experts of the region. Upon arrivalin the area in the end of April 2009, key informants (extension officers, researchers and NGOs) were

contacted in the district of Alto Molcu. After several meetings in which the objectives were explained

to those key informants three potential research villages were selected. Study sites with different soil

types, different importance of cassava in the farming system and different distance to the markets were

selected. The procedure was facilitated using a Global Position System (GPS) device to record GPS

coordinates and a GIS soil map (Appendix A24) of the district to cross check for soil type variability

between the sites. Main characteristics of the selected study sites are listed below in Table 2.

Mugema Nacuaca Gafaria

BiophysicalAltitude (masl) 610 711 710

Dominant soil types (FAO) Ferric Acrisol (sandy) Cambio Arenosol Ferric Acrisol (loamy)

Description soil type low levels of plant nutrients,

excess aluminium, and high

erodibility,

a fine-textured subsurface layer,

low water and nutrient holding

capacity

low levels of plant

nutrients, excess

aluminium high

erodibility, high

contents ofkaoliniticclay and iron and

aluminium oxides,

Landscape Plains slightly undulating rivers Plains slightly undulating Hilly surroundings

Table 2. Some biophysical and socio-economic characteristics of the selected villages (IIAM, 1996; Report, 2007; unknown,

2007)
http://www.britannica.com/EBchecked/topic/311686/kaolinitehttp://www.britannica.com/EBchecked/topic/311686/kaolinitehttp://www.britannica.com/EBchecked/topic/311686/kaolinite


24/114

Initially it was the intention to use the three selected villages Mugema, Nacuaca and Gafaria for all three

survey rounds, soil samples and harvest estimations. Unfortunately due to unforeseen circumstances

and limitations in time Mugema was only visited for the rapid farm characterisation. This made inclusionof Mugema for purpose of designing a farm typology possible, but further analysis on soil samples and

harvest estimations not possible. Result from the rapid farm characterisation in Mugema will therefore

only be used in the first part of Chapter 3: the analysis on village level (Section 3.1.2) and in the chapter

on farm typology (Section 3.2).


25/114

2.2 Farm sampling and characterisation

2.2.1 Rapid farm characterisationAfter the introduction and site selection local leaders in the villages were consulted about the possibility

and approval for carrying out the survey. In all three villages permission was obtained and a suitable key

informant per village was selected. The key informant then helped to select 25 households per village,

with contrasting household characteristics according to his knowledge. Special attention was paid to

avoid sampling bias by taken extra notice of the spatial distribution of selected farmers over the area of

study. Age distribution, ratio male: female respondents and NGO involvement of interviewees was taken

into account. Consequently surveys for the rapid farm characterisation were designed to capture bio-physical, socio-economic and managerial aspects of each selected farm and questions were checked

through triangulation. Emphasis was placed on quantifying yields, sales and the use of inputs such as

animal manure and hired labour and to explore the farming system in general. Overall the survey was

aimed at characterisation of typical smallholder farm in term of size, composition, production systems,

agricultural practices, labour and land with a focus on cassava, sorghum and maize. A mixture of

structured and un-structured questions was used and carried out by the author, a translator and the key

informant, which allowed for further questioning if answers given were unclear or not correspondingwith observations. The questionnaire is given in Appendix Table A 23.

2.2.2 Farm typology

After the first round of questionnaires (May- June 2009) obtained data was processed using statistical

software (SPSS 17.0) and Excel. Relationships between important variables on crop yield, crop

management, socio-economics, farming system and animal husbandry were explored using Pearson

bivariate correlations. Data from the rapid farm characterisation was analysed and farm household

typologies were constructed using the following multivariate statistical techniques as proxies for

categorization; Principal Component Analysis (PCA) and Cluster Analysis (CA) and were combined with

expert knowledge and preliminary household characterisation during fieldwork. PCA can be defined as a


26/114

preliminary characterisation with help of experts and exploration of field data was used to compare

different cluster outputs and to choose the most appropriate one.

2.2.3 Second and third round interviews

A subset of 2 farmers per typology (of the initially proposed typology) per village were chosen. The initial

first made typology was based on dividing the farmers according to farm income, farm involvement and

several wealth indicators suggested by experts. This approach will be discussed in Chapter 5.3.1 further

in this thesis. Later the used typology was revised, but the first typology made during fieldwork was used

for the selection of second and third round questionnaires. Besides asking for further details on crop

and soil fertility management, marketing and pest & diseases, the second round of questionnaires was

used to cross-check important information from the first round. For this second part of surveys fields of

cassava, sorghum, maize and fallow fields were visited together with the farmer. During the first round

of questionnaires and visits to farmers it was observed that what is called a machamba (field) by

farmers, is upon a closer look quite heterogeneous and could be divided in more homogenous sub-

machambas. A cassava field was considered as such if it would have the same characteristics for: i) time

of planting (in months), ii) ridges or heaps and iii) use of mixed cropping. For sorghum and maize only

the use of intercrop was used to identify a sub-field. Since farmers would fallow only a subplot of the

whole field, only the part used in the next cropping season was measured and recorded in the survey.

After defining the area was calculated using a differential GPS. Subsequently, the areas were plotted as

polygon in ArcGIS software in GIS soil layers.

2.3 Field biophysical characterisation

2.3.1 Soil sampling and analysis

For the cassava fields separate samples were collected from rows and from ridges/hills. Together withfarmers a distance of 20 cm from the stem was chosen for sampling in ridges/hills to avoid damaging the

tubers. To avoid oversampling the same number of soil samples in the ridges/hills as in between the


27/114

2.3.2 Field measurements

Bio-physical information per field was collected and included e.g. visible signs of erosion,

characterisation of surrounding fields (cultivated or fallow), and management: e.g. crop rotation, mixedcropping, history of cultivation, labour, weeding and handling of crop residues.

Distance in beds, distance between rows and height at 10 randomly chosen plants in each identified

cassava field were measured and recorded. A weed management score from 1 (very poor) to 5 (very

good) was used to quantify presence of weeds and subsequently weed management of the farmer.

Plant density (plants m-1) and plant population (PP, plants ha -1) were estimated by considering the

average over at least 5 distances between rows (BR, in meters) and between plants spacing (BP, in

meters) to calculate the expected plant population according to following equation:

Pl. density (plants m-2) = 1 / (BR x BP)

Because non productive stalks were replaced with productive stalks a correction factor for crop survival

was not included. This means that for estimation of plant population (PP, plants ha -1) the assumption

was made that all individual plants are productive and at least 5 measuring points between rows and

between plants are representative for plant density in each individual field.

2.4 Yield estimations

Yield estimations of several crops produced by the farmers in the villages were estimated using the

following different approaches:

farmers estimates of household production (per crop produced in ton per household)

farmers estimates of crop production per individual field (for cassava, sorghum and maize fieldsin t/ha)

calculations from harvest estimations in the field (only for cassava in t/ha)


28/114

Farmers estimating crop yield per individual field

During the three rounds of farm visits farmers were asked to estimate cassava harvest (kg dried

cassava), maize and sorghum per field visited. Only for cassava a conversion from dried to kg freshweight tubers using 33% DM as a rule of thumb was used (Fermont et al., 2009).

Calculations estimating cassava yield from harvest measurements in the field

In consensus with the farmer, a number of plants were selected for harvest by the researcher. The

individual plants were specifically chosen to represent average plants present in each field. Farmers

helped in finding representative plants by clearing the soil around the cassava plant and judging its

representative characteristics by tuber size and amount of tubers. Selected plants were uprooted, soil

removed and weighed as a whole. Next tubers were removed and weighed in total, subsequently

marketable tubers were selected and weighed. Tubers were considered marketable if > 3cm in

diameter, damage and disease free. Name of cultivar was recorded together with the notation bitter or

sweet. Fresh tuber weight (IFW in kg plant-1) could then be calculated out of the average of measured 5

individual plant tuber weights corrected for months after planting (MAP) and set on a standard of 12

months. Total fresh tuber weight (TFW in t ha -1) was calculated using following equation:

TFW (t ha-1)=IFW (kg plant-1) x Plant population (PP, in plants ha-1)

Calculations estimating sorghum yield from harvest estimations in the storage facility

In the research area sorghum is harvested in between June and July. Initially the objective was to be part

of the harvest of sorghum in order to make harvest measurements. Because not all farmers were able to

estimate the amount of sorghum harvested (usually referred to as heap and store) an alternative had

to be found. Sorghum throughout the research area is stored in an identical way: the whole panicle is

piled up in a heap of panicles. The size of the storage of sorghum of farmers revisited was measured and

adjusted to geometrical shapes. For these shapes size in m3 was calculated. From a subset of farmers a

few panicles were collected. These were weighed as full panicle and volumetric (m3) size of the panicle

estimated using a beaker filled with water After drying the harvest index of the panicle (HI ) was


29/114

2.5 Income calculations

2.5.1 Household income calculationsA distinction in sources of income was made following the assumption that total estimated farm income

is the sum of income from crops, income from selling animals and animal products and off farm sources

of income as listed below:

Total estimated farm income = Income from crops + Income from animals + Off farm income (other

sources of income)

Income from crop productionGross value per crop or crop product (GVC) per household was estimated by multiplying fresh weight

(FW) of the product by the average selling price. Average selling prices and estimated fresh weights

were collected during the three farm visits. The sum of all GVC subsequently formed total income from

crop production in New Mozambican Meticals (MZN).

Income from animal production

Gross value from animal production (GVA) per household was estimated by multiplying the number of

animal type sold times average selling price. A summation of GVAs per animal type plus the value of

animal products sold formed the income from total animal production (GVAT).

Off farm income

During the interviews farmers were asked for involvement in off farm activities such as making blocks,

sieves or stoves, having a contract job (e.g. driver or teacher) and receiving pensions and remittances on

a yearly basis. The sum of all activities belonging to non-farm sources of income is off farm income.

2.5.2 Gross margin fertilizer analysis

Partial gross margins of NPK fertiliser for simulated yields were calculated for a total of 90 individual

fields in both Nacuaca and Gafaria Average wholesale market prices for 2009 were used for the full


30/114

2.6 Statistical analysis

Analysis of socio-economic diversity

Comparisons across sites and farm typologies in terms of socio-economic indicators, land use and

managerial variables were done through calculation of descriptive statistics and analysis of variance.

This analysis was done with the explanatory factors Village and Typology and their interaction, and

means were compared with the 5% LSD or 10% LSD if indicated. A principal component analysis (PCA)

was conducted using the socio-economic data (previous log or square root transformed and

standardized for comparable ranges) to identify important proxy indicators further used in the

development of the typology. Due to the experimental design of three villages and four farm types in a

non-equal distribution the ANOVAs were conducted under the unbalanced treatment structure of

offered in SPSS (17.0) under Univariate Generalized Mixed Models.

Soil fertility variability

Soil fertility status was assessed at village, farm and field level through calculation of descriptive

statistics and analysis of variance with Village or Quartile as explanatory factors. This analysis was done

based on soil properties pertaining to individual fields (N=90) expressed as results from chemical

analysis of soil organic C, total N (%), available P, exchangeable K, Ca and Mg, pH, CEC, sand (%), silt (%)

and clay (%) contents in a composite sample per field. The model FIELD was used to divide fields

according to their yields in four significantly different quartiles, to cope with variability within villages.

Explaining simulated yield variability

To be able to identify variables best explaining yield differences a multiple linear regression was carried

out. As a first step variables were checked for correlation using Spearmans and Pearsons correlation

analyses. For any pair of abiotic, biotic and management variables with inter-correlations (r) greaterthan 0.7 only one variable was retained and subsequently these were taken as independent variables.

The entire data set was used for the analysis using GenStats all-subset regression routine and the best

d l l t d ( i 13 1) I t f th t f h l t i bl


31/114

2.7 Dynamic simulation of crop performance and nutrient balances at field

scale

2.7.1 Model for dynamic simulation at field scale: FIELD

In this thesis two models for simulation of crop yields were used: FIELD for cassava yield simulations and

QUEFTS for maize yield simulations. In this section both models will be discussed, but the model FIELD

will be discussed more elaborately because of its importance in the modelling in this thesis.

To be able to explore short to medium term consequences of management options on cassava yield and

soil fertility indicators a model was used. Several models on cassava crop production exist but as stated

in research done by Fermont (2009) none of them has been calibrated for African conditions. A model

FIELD was developed as a crop and soil sub model of FARMSIM, which is a bio-economic model

developed to analyse tradeoffs between farming systems and environments focusing on strategic

decision-making and embracing the spatial and temporal variability of smallholder systems (FArm-scale

Resource Management SIMulator;www.africanuances.nl)(Tittonell, 2008). In overall FARMSIM consists

out of the components FIELD (crop-soil), LVSIM (livestock) and HEAPSIM (manure) modules that are

functionally integrated and allowing for feedbacks between these entities on farm scale (Figure 6). In

this thesis only the crop-soil component FIELD will be used.

FIELD focuses on long-term changes in soil fertility (C, N, P and K interactions between nutrients) which

determines crop production. It simulates crop responses to management interventions such as mineral

fertiliser, application of manure or organic amendments. Different fields within the farm are treated as

separate entities with their own set of soil properties.
http://www.africanuances.nl/http://www.africanuances.nl/http://www.africanuances.nl/http://www.africanuances.nl/


32/114

Crop biomass is calculated based on the intercepted amount of photosynthetically active radiation (PAR)

which is only a fraction of total PAR during the growing season. Light determined yield can

subsequently be estimated by using a light conversion efficiency coefficient and is affected bymanagement factors such as cultivar choice, planting density and planting time. Water-limited crop

production is calculated based on seasonal rainfall data and site and crop specific water use efficiency

coefficient and therefore depends on data availability of the study sites and has to be derived from

literature otherwise. Nutrient-limited crop production is calculated from nutrient availabilities and

nutrient use efficiency of the crop. Other nutrient sinks for example leaching and gaseous losses of N or

immobilization in the soil organic matter act as competition components to crop uptake an can be used

to derive the nutrient capture efficiency of the crop. To be able to calculate resource-limited cropproduction in FIELD the minimum of water-limited and nutrient limited crop production (determined by

the availability and use efficiency of N, P, K and their interactions ) has to be taken following Liebschers

Law of Optimum (Van Keulen, 1995). Reduction factors such as weed competition are used to calculate

actual yield and derived from multiplying actual biomass production with a harvest index coefficient

(Tittonell, 2008).

A sensitivity analysis can be used to study the relative variation in model outputs in response to changes

in model input or parameters. The model was run on a number of fixed parameter settings presented in

Appendix Table A 12. The relative partial sensitivity of the model input can be calculated with the help of

following equation (Tittonell, 2003):

(dO/O) / (dI/I)

In this equation (dO/O) is the relative change in model output and (dI/I) is the relative change in the

value of parameter or input data. Sensitivity was calculated as the average sensitivity to changes in thevalue of parameters that were set according to information from the field (if available) or systematically

(e.g. increase and decrease of 10%). Calibration, validation and the sensitivity analysis have been done in


33/114

2.7.2 Developing scenarios

The modelling part of this thesis focuses on cassava, the major staple food crop in the farms under study

in Central to North Mozambique. In the simulation no distinction for different genotypes was made sothe outcomes are the result of an average performing cassava genotype.

Crop management

The model was run for six different regimes affecting short- and long term crop and soil productivity

listed in Table 4 : 1) no input farm practise, 2) increasing cassava residue returned to soil, 3) increasing

amendment of maize based organic residues, 4) increasing amendment of on farm manure, 5) increasing

amendment of NPK fertiliser, 6) increasing amendment of NPK fertiliser plus manure and organic

residue amendment. For each management regime the model was run for all fields sampled in both

villages (N=90) for a period of 10 years. A list of fixed model parameter input can be found in the

Appendix A 12.

ManureNutrient input from manure application was calculated by the production of manure (kg year -1 animal-1),

the dry matter content and the N, P and K fractions in the model. Animal manure production was

Scenario Name Quantity

I No input -

II Cassava residues 10% and 55% remaining in the field

III Maize residues 1-5 t/ha (steps of 1 t DM)IV Small stock farm manure 0.125, 0.25, 0.5, 0.75, 1.0 t DM ha

-1

V NPK fertiliser 25, 50, 100, 150, 200, 250, 350 kg ha-1

VI Improved NPK management NPK + 0.25t DM ha-

manure + 1t DM ha-

pigeonpea residues

Table 4. Overview of management scenarios applied


34/114

literature for the different crop residues used in the management regimes. For the purpose of the

different management regimes two types of residues were used: maize and pigeonpea. The amount of

nutrient in the residue was estimated using literature choosing a medium nutrient composition formaize, and an upper nutrient composition for pigeonpea. For maize 1%N, 0.2%P and 1%K was used

derived from work done by Dass (1979). For pigeonpea a high nutrient content of 4%N, 0.4%P and 4%K

was estimated to represent a hypothetical high nutrient rich legume. Lignin content was taken from

literature at rates of 5.4% and 10.7% and of total DM content for maize and pigeonpea (Bernard and

Jean-Marc, 1986) for maize and (Mapfumo and Mtambanengwe, 2004) for pigeonpea.

For maize farmers estimations on maize yield and QUEFTS simulations were used to calculate maize

biomass production and an HI of 40% was derived from work done by Hay and colleagues (2001). The

effect of cassava residue incorporation was simulated in two scenarios: 10% and 55% cassava residue

incorporation. 10% remaining residue complies with standard farm practice of removing all crop

residues and having some remains (in e.g. litter fall or some stalk remaining). The proposed 55% cassava

crop residues is in line with harvesting the tubers and stalks for planting but leaving the remaining stems

and leaves (instead of e.g. burning). Lignin content was assumed to be 8.4% of total DM and nutrient

content of crops residues set at 1.2% N, 0.2% P and 0.7%K , derived from Putthacharoen et al.(1998).

Table 5. Nutrient amendment in scenario simulations

Maize1

residues

Cassava

residues1

Pigeonpea1 Small stock

manure

HI 40 45 30 -

DM (%) 64 26 53 50Per ton DM applied

N (kg/ha) 9.0 12 36.0 20.0

P (kg/ha) 1.8 2 3.6 4.0K (kg/ha) 9.0 7 36.0 20.01Fraction incorporated was set at 0.9 (meaning 0.9 t of every ton added to the soil gets incorporated into the soil)

F tili


35/114

3 Results

3.1. Characteristics at village levelThe following section (Section 3.1.1) analyses results found at village level: comparing between the

villages Nacuaca and Gafaria for bio-physical variability.

3.1.1 Village biophysical variability

Average value of soil fertility indicators varied between villages in the district as predicted by the

variation in inherent soil types present. For most important soil fertility variables, box-plots are

presented in the Appendix Figure A 1, determination and calculation of the soil fertility variables used

can be found in the Appendix A 3. Comparing fields under cultivation between villages showed a

significant higher percentage of clay content in Gafaria compared to Nacuaca (25% compared to 12%,

P


36/114

Table 6. Percentage of fields (N=90) belonging to soil fertility class Very Low (VL), Low (L), Medium (M) and High (H)

according to soil property groupings of Instituto Nacional des Investigao Agronmica (Maria and Yost, 2006)

% of fields grouped per soil fertility class

Clay SOC N P K Al saturation

Nacuaca Class

VL 12 14

L 42 75 48 17 23 66

M 58 14 52 77 52 33

H 6 12 1

Gafaria Class

VL 3

L 5 74 5 21 67

M 87 24 92 79 45 31

H 8 3 55 2

Villages P< 0.001 ns 0.001 ns 0.007 ns

Fields of Gafaria were found significantly higher in exchangeable K (cmol(+) kg-1) compared to Nacuaca

with an average of 0.46 compared to 0.21 cmol(+) kg-1 respectively (P


37/114

Figure 8. Exchangeable K (cmol(+) kg-1

) plotted against clay content for all soil samples

combined (N=90) for Gafaria and Nacuaca


38/114

Sampled fields in both Nacuaca and Gafaria had a comparable pH, but the pH range of fields sampled

differed: a wider range was found in Gafaria (pH 5.3-6.8 for Nacuaca and pH 4.9-7.4 for Gafaria). None of

the fields sampled was classified as extremely acid (>4.5). 10% compared to 37% of the soils wereclassified as strongly acid (pH 4.5-5.5), 87% compared to 55% as slightly acid to acid (5.5-6.5) and only 2

samples from both villages were classified as neutral (6.5-7.3) in Nacuaca and Gafaria subsequently

(Table 7).

No differences were found for the Cation Exchange Capacity (CEC) of soils comparing both villages with

an average of 24.6 cmol kg-1 for Nacuaca and 26.0 for Gafaria. The Exchangeable Cation Exchange

Capacity (ECEC) was found to be an average of 10 cmol kg-1 for both villages. Base saturation (the sum of

bases Caexch, Mgexch, Naexch and Kexch/ECEC) did not differ significantly with a total average of 67% in both

villages.

Aluminium saturation (Alexch/CEC*100) can be an important indicator of aluminium toxicity on acid soils

and is considered low, medium and high at ratios of 0-15, 15-35 and >35% aluminium saturation. Of the

soils sampled 67% compared to 79% were classified as low and 33% and 21% as medium in Nacuaca and

Gafaria (Table 6).

Table 7. Percentage of fields grouped in the different pH ranges presented for Nacuaca and Gafaria (N=90),

classification used is derived from the Institute of Agricultural Research (Appendix Table A 2)

% of fields

pH range Nacuaca Gafaria

4.5-5.5 10 37

5.6-6.5 87 55

6.6-7.3 4 5


39/114

3.1.2 Socio-economic variability at village level

This section focuses on results found at village level: comparing between the villages Mugema, Nacuaca

and Gafaria for socio-economic characteristics, production activities and managerial aspects.

The farms sampled in this study showed characteristics in terms of family structure, income sources and

farm indicators that could consider them as representatives for the study area as compared to the

results of the baseline study conducted in the area (OVATA mid term, 2004). All results in this

subchapter are based on the total data set of farmers (N=74, three villages) unless indicated elsewhere.

Comparing villages no difference was found in the number of household heads being illiterate: anaverage of 11% of farmers per village did not go to school and had no knowledge how to read or write.

The remaining rest differed in level of education from 1st class until 8th class as highest education

(Appendix Table A 5). Men were more likely to have a higher education compared to women across

villages, females had a higher illiteracy rate compared to males across villages (not shown).

Sampled farms showed differences as well as shared characteristics across sites, some of these

characteristics are shown in Table 8 below. The average amount of fields per household was significantly

higher in Gafaria compared to farmers in Mugema (3.7 compared to 2.6 respectively). Household

members, the amount of people present at the farm, did not differ across sites with an average of 6-7

family members.

Closest fields for all study sites were 1.0-1.5 min walking away from the homestead. These fields would

be in the immediate vicinity of the family houses and storage sheds. Farmers of Gafaria all had their

fields close by the house (25 minutes) whereas on average 154 minutes had to be walked to reach the

furthest field in Nacuaca, Mugema was found intermediate with fields on 50 min walking distance.

The total amount of animals (including all animals) per household compared between study sites did not1


40/114

Table 8. Socio-economic indicators and production activity indicators per household for the three villages of study

(N=74)

Site

Indicator unit Mugema Nacuaca Gafaria P

Socio economic indicators

Household members present # 5.6 6.7 7.0 ns

Amount of machambas # 2.6a

3.2ab

3.7b

0.011Distance to closest field min 1.0 1.5 1.0 ns

Distance to furthest field min 50a 154b 24a 0.003

Months without food # .7a 1.3a 3.7b 0.001

Total income out of agriculture MZN1 5853ab 12286b 878a 0.038

that hire labour % 28 29 10 ns

that have off farm income % 44 21 41 ns

Production activity indicators

crops grown # 6.2a 7.2b 7.9b 0.009

cash crops grown # 1.0 1.0 1.0 ns

Fruit trees grown # 9.4a 11.6a 26.0b 0.001

Households that own animals % 80 92 92 ns

Households that own livestock % 48 67 64 ns

number of animals # 11.8 10.4 9.2 ns

number of goats # 1.0 9.7 5.0 ns

number of pigs # 5 2 4 9 6 4 ns


41/114

Food (in) security

Farmers were asked for the number of months per year with food insecurity/shortage. Respondents of

Mugema were able to feed themselves in average all year except for 1

month, respondents in Gafaria had a significant higher amount of months with food insecurity (3.7

p


42/114

Table 10. Start and end month of food insufficiency indicated by the respondents (% of households) in Mugema,

Nacuaca and Gafaria (N=74)

Mugema Nacuaca Gafaria

N 25 24 25

Start August 4 0 0

October 4 4 12

November 4 13 52

December 12 17 16

January 12 29 16

February 0 8 4

March 4 0 0

no hunger 60 29 0

End April 8 0 8

May 0 4 16

June 0 0 4November 4 0 0

December 12 8 0

January 4 0 4

February 4 46 20

March 8 13 48

no hunger 60 29 0


43/114

Labour

Labour was mainly derived from the household although some farmers from the survey also hired

labour from outside the farming household. Between 10 and 29% of the farmers hired labour on aregular and non-regular basis to relief pressure during the peak season of labour for preparing the fields

and harvesting of maize. Farmers themselves would work for their neighbour (ganho-ganho) at times of

hunger and get payment in food. Most commonly the whole family or the wife and husband contributed

to the largest part of the labour provided by the family (Table 11).

Table 11. Household members contributing the most to farm labour (percentage per household, N=74)

SiteTotalMugema Nacuaca Gafaria

N 25 24 25 74

Who works the fields the

most?

wife 8 8 5 7

wife and husband 40 75 33 50

whole family 40 8 48 31

husband 0 4 0 1

wife and children 12 4 10 9

wife and grandmother 0 0 5 1


44/114

Household income

Total household income was calculated summing the considered three main sources of income: income

from crop production, income from animal production and off farm income. The percentage ofhouseholds being involved with activities belonging to the different sources of income are listed below

in Table 12. Although no significant difference could be found in the amount of households being

involved in the different economic activities mentioned, a difference could be found in the percentage

off farm income contributed to total calculated household income comparing between villages. For 44%

of the farmers involved in the survey from Gafaria off farm income contributing for an average of 75% of

their total household income. For Mugema and Nacuaca 44 and 21% of the households had sources of

off farm income, contributing to 19% and 21% respectively of their total calculated household income.Off farm income activities mentioned were making crafts (45%: such as blocks, sieves or stoves), being a

middleman (21%: selling maize or beans) or working for others (14%), remittances or pensions were

received by the remaining 21% (N=29 out of 74,Table )

Table 12. Percentage of farms within the farm samples from the three sites that are involved in activities listed

below and their proportion of total income (N=74)

VillageMugema Nacuaca Gafaria

% of households selling animals or animal products 28 33 40

% of households selling crops 96 83 100

% of households having sources of off farm income 44 21 44

% of income from animals or animal products 9 12 17

% of income from crops 72 81 50

% of income from off farm 19 21 33


45/114

Income from crop production

Importance of on farm income from crops differed between villages: 72 and 81% of total income in

Mugema and Nacuaca was derived from crop sales compared to 50% in Gafaria (see Table 12 ).Households of Gafaria in average had the lowest income from agriculture compared to Mugema and

Nacuaca (878 compared to 5150 and 12285 MT hh-1; Table 8) and a higher percentage of income

attributed by selling of the alcoholic drink Ortega (37%, Figure 10). Ortega is an alcoholic beverage made

from cassava and sorghum flour as ingredients, fermented for about a week and a medium to high (

12-20%) alcohol percentage. A significantly higher percentage of farmers were selling Ortega in Gafaria

compared to farmers in Nacuaca and Mugema. Farmers in Mugema had a considerable higher

percentage of beans sales contributing to their from farm income (35% P


46/114

Table 13. Percentage of farms selling the most commonly grown crops and beverage Ortega listed below in the

three villages (only farmers growing the crop have been included see Table 12)

Table 14. Frequency of farmers (%) selling cassava and maize to listed buyers below (N=17)

Site

Mugema Nacuaca GafariaP 2). However to

be able to purchase NPK fertiliser farm households need capital for investment. Farm type 1 and 2farmers seem to have more capital available from crop and animal sellings (Figure 11). Farm type 3 has a

much higher percentage of income out of non-farming practises and doesnt seem to be as highly

dependent on agriculture, like the other farm types. As they do have capital to invest, they might lack

the involvement and expertise required make investments in agricultural inputs such as fertiliser. Farm

type 4 farmers are mostly single headed households that are not as involved in farming for income

generation, but mostly to support themselves and the remaining of the family present. Typologies found

in this research seem to be in line with characteristics of typologies found in previous research byTittonell (2009).


77/114

Improved management simulations included manure, crop residue and fertiliser in a complete

application package. The feasibility of this package requires availability of manure. Comparing farm

types, Type 1 and 2 farmers in general had the highest number of animals available on farm and were

more likely to own small stock (goat or pig).Figure 22 tries to illustrate capital availability on farm and animals present classified per farm type.

Farm types 1 and 2 seem to have as well a higher number of animals, as more capital for investment.

From literature the hypothesis is proposed that farms having on-farm income strategies (Farm type 1

and 2), are more focused on productivity and often more innovative. They can be characterized by an

earlier adoption and adaptation of technologies and could serve as an example for other farmers in the

community. Reij and Water-Bayer (Reij and Bayer, 2001) even state that this may even facilitate the

further outspread of technologies in the community.

5.1.5 Model FIELD considerations

Simulated yield response by the FIELD model used in this research was only influenced by the soil

fertility of the different fields used (N=90). Other variables such as weed management, pest and disease

pressure and harvest age have not been included in the simulation, while previous research showed

fertiliser response affected by soil fertility, rainfall and weed management (Fermont et al., 2009). It is

therefore important to look at the simulated result from the viewpoint of soil fertility alone, keeping in

mind several other variables that can influence actual yield.

Extra demand in labour, for the scenarios involving crop residue incorporation and manure applications

have not been taken into account in this study. Several attempts were made to quantify labour needs,

but did not succeed in sufficient and reliable data. Especially for cassava residues, which can be difficult

to handle in the field, extra labour could be required to cultivate the same field size. Even so, the size offield could suffer under the implication of cassava residue incorporation. At this stage of research, these

remain uncertainties.


78/114

5.2 Evaluation of feasibility

5.2. 1 Potential crops for biofuel; cassava, maize and sorghum

Firstly the discussion of cassava as a biofuel crop will be focussed upon: being the most important crop

of this study. Sorghum and maize will be used to illustrate and elaborate on the biofuel discussion. This

thesis can merely provide with insights from research done in three villages as case studies in Alto

Molcu district.

Ranked as the most important crop by 60% and 70% of the respondents in Nacuaca and Gafaria

respectively, cassava plays an important role for farmers in the research villages. As discussed before

one of the main advantages of cassava is the storing of roots in the soil. While other crops such as maizeand sorghum have a more defined time frame of harvest, cassava can be kept undisturbed for a later

period of harvesting. Only one third of the farmers in Gafaria (26%) sold cassava compared to 42% in

Nacuaca and 48% in Mugema. When sold cassava was sold as less than half of the produced quantity:

47, 49 and 43% respectively for Mugema, Nacuaca and Gafaria. From the interviews it was understood

that mostly cassava is sold within the community (82%) but some was also sold to people coming from

far (18%). If farmers would be involved in a emerging cassava for biofuel market, with current

production levels, the surplus being sold to neighbours with insufficient cassava production at themoment, could be sold for biofuel purposes. This could be an important consideration to be investigated

in further detail. The resilience of a community for food insecurity can be outbalanced if cassava is sold

outside the community. Currently low priced (contributing to an average of 12% of total income)

cassava could increase in value because of extra demand of the market from the biofuel industry.

Therefore making it more expensive as a food crop purchase for cassava consumers.

As an illustration the crop maize can be used. As well as an important food crop, maize also serves as a

commonly used crop to generate cash. Recent programmes such as People for People (PforP) try to

increase food security in Mozambique and surrounding countries by buying farmers surplus of maize,


79/114

bottleneck in the adoption of sweet sorghum cultivation. However, since farmers are familiar with

sorghum growing practises, sweet sorghum could be introduced not as a crop to replace sorghum but as

an extra cash crop. The same way, sesame has been introduced a couple of years ago to generate cash

for the farmers while hardly being consumed.

The biggest question remains if current cassava yields can be increased. This thesis has discussed ways

to increase yields starting from low capital required residue and manure applications to higher capital

investments needed if fertiliser would be applied. Previous research has shown that only 4% of the

farmers have access to credit, although rural banks are emerging and more micro-credit incentives are

being developed (MINAG, 2007). Outgrower schemes for e.g. the production of tobacco are excisting

and providing farmers with fertiliser under contract production. These existing outgrower schemes

could provide with examples and frameworks on how commercial firms can arrange input purchases or

input credit for farmers under production contracts.


80/114

5.3.Methological considerations

5.3.1 The use of a farm typologyInitially after the first rapid farm characterisation (May-June 2009) a farm typology was designed to

characterize farm types and have a better understanding on within and between village farm

heterogeneity (and homogeneity). This first typology was based on dividing farmers on farm income,

their involvement in farming and several wealth indicators as suggested by experts (amount of animals,

hired labour, maize and bean production). A total of 5 farm types were identified for all three villages

(Mugema, Nacuaca and Gafaria) and 2 farmers per village were selected for re-visit (excluding Mugema).

After returning from fieldwork and the start of analysis questions arose about the correctness of the

chosen typology. Ideally, a chosen farm typology should capture the most important characteristics that

would typify farm households according to their similarities within a typology and their dissimilarities

between typologies. A re-evaluation was made using the proposed methodology presented in Chapter

2.2.2. After designing the new typology, more profoundly based on the results of the rapid farm

characterisation, and making divisions in farmers based on several variables and less strongly on farm

income it was found that on revisiting the selected farmers a farm type group appeared under sampled.

Farm type 3 farmers characterized by high off farm involvement seemed to have been not revisited in

sufficient amount in Gafaria and even more in Nacuaca where farm type 3 was left out totally (Appendix

Table A 18). The first farm typology design was mainly based on variables presented in Appendix A

19.where a separate typology was made for farmers hiring labour. As shown farm management

practices such as burning of crop residues were taken into dividing farmers over typologies. Pros and

cons for both types of typologies can be thought of but keeping in mind the function of a typology in

this research: a tool for understanding variability between farmers within a village, the revised typology

seems to have a more profound basis, even though sampling has been done with a less favourable

division.


81/114

5.3.2 Yield estimations

Yield estimations in this study were done following several methods as presented in Chapter 2.4 and

listed in the Appendix Table A 8, Table A 9 and Table A 10. For cassava three methods were used to

estimate production/ha. Because of limitation in field work, only a limited amount of plants could be

harvested per plot, calculations from harvest measurements remain uncertain. Previous research done

by Hilton (2000) showed farmers underestimating their yields by 25-50%. This would imply that actual

yields should be considered somehow in between the range of (lowest to highest): estimated by farmers


82/114


83/114

6. ConclusionsInitially the aim of this study was to explore smallholder farming systems producing cassava, maize and

sorghum in the scope of the current raging discussions on crops for biofuel production. Although

Mozambique has been identified as a promising country for biofuel production little is known about

current farming systems.

An important finding of this research was the high heterogeneity found comparing villages and within

villages. Analysis at different levels: village, farm and field provided helpful tools and the design of a

typology helped to analyse within village variability. An interdisciplinary approach in which bio-physical,

socio-economic and farm management practices was used to explain variability can be considered key to

this sort of research. As an example the case study of Gafaria can be mentioned: based on bio-physical

analysis alone Gafaria would be thought to be more fertile and higher in production compared to

Nacuaca. From socio-economic analysis the opposite was found: farmers in Gafaria seemed to be less

well off compared to farmers in Nacuaca.

Cassava yield estimations, currently lacking in many research projects, have proven to be very difficult in

this research. Initially the aim was to quantify yields of cassava, maize and sorghum but due to severalunforeseen circumstances this could not be accomplished. Therefore explaining current yields in relation

to soil, landscape and management variability was not possible. This is to be regretted because it could

have provided with important insights on most important factors, besides bio-physical properties,

limiting yields.

As a result of this, simulated cassava simulated yields by the model FIELD, are based on soil fertility

alone and not on interactions with crop management such as weeds and pests. It is a major limitation

that has to be taken into account. Simulated yields of FIELD were higher than estimated yields by

farmers but lower than calculated yields from field work. Harvest measurements are lacking to assess


84/114

The development of cassava for ethanol could strongly increase the demand for cassava and could be an

incentive for farmers to adopt technology packages to improve productivity and profitability of cassava

production. However which farmers would be more suitable to involve in production increasing

activities (e.g. the adoption levels of farmers to improved techniques) could be assessed with the help of

farm typologies. Considered solely as a tool they can provide with insight on important farm

characteristics and could be used in allocating promising pilot projects. Production orientation, resource

possibilities and farm involvement could be important indicators derived from a Farm typology. Four

different farm types over the three villages were identified in this study. From this research Farm Type I

and II would be indentified as feasible candidates for pilot projects.

Within farms it was found that fields could be classified into quartiles corresponding to FIELD simulatedyields. 4th quartile fields seemed to be more responsive to any nutrient amendment (in-organic and

organic) and had overall higher yields compared to 1st and 2nd & 3rd quartile fields. Allocation of available

nutrients to these higher producing could induce soil fertility gradients over farms and should be taken

into consideration.

Recommendations

This study has provided insights in farming systems from three villages in Alto Molcu district, Northern

Mozambique and can underline the importance of studies focused on exploring the farming system.

Biofuel projections are based on assumptions and estimations on national level, with 80% of the

population of Mozambique being involved in agriculture one can imagine the heterogeneity in

possibilities for production.

On farm improved farm management trials could provide with more insights on factors affectingyields such as weeds, pests and diseases. The model FIELD could be improved and further

calibrated to have a better fit; it could provide as a helpful tool in simulation of different


85/114

ReferencesBatidzirai, B., Faaij, A. P. C., and Smeets, E. (2006). Biomass and bioenergy supply from Mozambique.

Energy for Sustainable Development X, 27.

Bernard, C., and Jean-Marc, B. (1986). Yield and composition of cell wall residues isolated from various

feedstuffs used for non-ruminant farm animals. Journal of the Science of Food and Agriculture

37, 341-351.

Bias, C., and Donovan, C. (2003). "Gaps and opportunities for agricultural sector development in

Mozambique," Rep. No. 54E.

Bidogeza, J. C., Berentsena, P. B. M., De Graaff, J., and Lansink, A. G. J. M. O. (2007). Multivariate

Typology of Farm Households Based on Socio-Economic Characteristics Explaining Adoption of

New Technology in Rwanda. In "Second International Conference, August 20-22". African

Association of Agricultural Econo

sannethesis_moz

Documents