Residue Management Strategies for the Rainfed N-Deprived Maize-legume Cropping

Systems of Central Mozambique

Nascimento Salomão Nhantumbo

(BSc in Agronomy and MSc in Erosion, Soil and Water Conservation)

A thesis submitted for the degree of Doctor of Philosophy at

The University of Queensland in 2016

Queensland Alliance for Agriculture and Food Innovation Institute (QAAFI)

Page | i

ABSTRACT

Identifying best-fit residue management strategies for Central Mozambique rain fed N-

deprived maize-legume cropping (LEINA) systems managed under conservation agriculture

is critical to improve resource use efficiency and yields. Therefore, understanding under what

circumstances positive and negative responses from carbon rich residue mulches are likely to

occur and what drives farmers’ decision on the adoption of technologies is critical to design

relevant and feasible recommendations that fit farmers complex farming circumstances. Here

I used household survey data, field experimentation and cropping systems modelling to: 1)

characterize farm diversity and understand how it affects smallholder farmers resource use

attitudes, management decisions and the likelihood to engage in sustainable intensification

(SI) practices; 2) identify best-fit residue management strategies that are more likely to

improve resource productivity and yields; and 3) develop simple rules of thumb to match

residue management solutions to N-deprived smallholder farming systems.

Household characterization data showed that, rather than farm size and labour availability,

farmers resource use attitudes and household capacity to meet its annual food security and

income generation targets where key farm differentiation factors, therefore critical to map

farm typology feasible intensification options. Using a heuristic decision tree to group

smallholder farmer into a new set of mutually exclusive farm typologies that highlight

existing intragroup diversity among Mozambican small-scale farmers, three key typologies

where identified. The resource endowed and innovative farmers (Type B); resource endowed

but change resistant farmers (Type - A2); and resource constrained (Type - A1) farmers, who

can also be maize self-sufficiency (Type-A1a) and the food insecure (Type-A1b). For the

already maize self-sufficient Type-B and Type-A2 farms, the challenge is to improve

agronomy to generate enough surplus and income from maize. Mechanization and labour

qualification are critical for these groups. For resource constrained Type-A1 farmers, the

primary challenge towards intensification is to generate enough income to improve the

household maize self-sufficiency and income in the system. Considering its high market

value and further contribution to soil N-availability, legume production is a feasible

investment alternative. Nevertheless, two key intensification traps were identified: the

household maize self-sufficiency and soil fertility perception, and land availability perception

trap (extensification). The traps are perception based mind-sets that reflect farmers

understanding of their circumstances and the strategies they use to materialize their annual

food security and income generation targets.

Page | ii

Field trials and model-assisted analyses (64 years) of carbon rich residues effects on a cereal

(maize) and a legume (cowpea) crop, at different levels of N supply, and three soils of

contrasting water holding capacity (WHC) showed that for Central Mozambique rain fed N-

deprived systems, residues are more likely to (i) improve the yield, WUE and NUE of the

unfertilized legume crop across all tested environments; and (ii) that the application of carbon

rich residues reduces maize yields, NUE and WUE in a high WHC sandy clay loam (cSaCL)

soil, and on the wet lowland sandy loam textured soil (SaL). In high WHC soils, only a 5%

chance of positive yield response from carbon rich mulched soils was simulated at 0 (ND)

and 23 kg N/ha (LNA). At 92 kg N/ha (HNA) benefits from carbon rich residues are expected

only in 20% of the seasons in high WHC soils. However, such benefits are only attained

during the driest years. In the dry and low WHC fine sandy soil (fSa), positive responses to

residue application were simulated in 85-90% of the seasons in both cowpea and maize. In

terms of resource productivity, maize proved to be more responsive to N-application,

especially on the wet and moisture rich SaL soil. Here, maize response to N-application was

attributed to better nitrogen uptake (NUpE) and translocation efficiency (NUtE) due to high

in crop moisture regimes. In the drier fSa, poor soil water availability led to poor NUpE and

consequently lower maize NUE and yields. For cowpea, yields at HNA were hindered by

poor N-translocation into grain. Despite maize providing the best responses to N-fertilizer

application in mulch free systems, yield penalties from residue applications in continuous

maize cropping systems.

The findings from this study showed that, in the rain fed N-deprived systems of central

Mozambique, the overall performance of maize-legume cropping systems under residues is

governed by two critical interactions: 1) a crop and soil water induced residue response trade-

off; 2) a residue modified nitrogen driven WUE and NUE trade-off. Understanding these two

responses is key to validate locally feasible resource management strategies crucial to the

effective validation of CA systems. However, because resource access and management

differs across households, an interactive and personalized socio-technical intervention that

takes into account each group biophysical and socioeconomic circumstances is required to

successfully involve farmers in co-designing group specific technological interventions

centred on the improved use of locally available resources and those that farmers can afford

to mobilize and use.

Page | iii

Declaration by Author

This thesis is composed of my original work, and contains no material previously published

or written by another person except where due reference has been made in the text. I have

clearly stated the contribution by others to jointly-authored works that I have included in my

thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical

assistance, survey design, data analysis, significant technical procedures, professional

editorial advice, and any other original research work used or reported in my thesis. The

content of my thesis is the result of work I have carried out since the commencement of my

research higher degree candidature and does not include a substantial part of work that has

been submitted to qualify for the award of any other degree or diploma in any university or

other tertiary institution. I have clearly stated which parts of my thesis, if any, have been

submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University

Library and, subject to the policy and procedures of The University of Queensland, the thesis

be made available for research and study in accordance with the Copyright Act 1968 unless a

period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright

holder(s) of that material. Where appropriate I have obtained copyright permission from the

copyright holder to reproduce material in this thesis.

Page | iv

Publication during candidature

Peer-reviewed papers

Nhantumbo, N., Mortlock, M., Nyagumbo, I., Dimes, J and Rodriguez (nd) Residue and

fertilizer management strategies for the rain fed, N-deprived maize-legume cropping systems

in Central Mozambique, submitted to the Field Crop Research Journal

Conference abstracts

Nhantumbo, N. Dias, J., Mortlock, M., Nyagumbo, I., Dimes, J and Rodriguez, D (2014)

Best-fit residue allocation: A gate for legume Intensification in nitrogen constrained systems

of Central Mozambique. In “Proceedings from First African Congress on Conservation

Agriculture" (African Conservation Tillage Network, ed.), pp. 33-35. ACT, 18-21 March

Lusaka, Zambia.

Nhantumbo, N., Mortlock, M., Nyagumbo, I., Dimes, J and Rodriguez, D (2015) Improving

resource allocation in nitrogen constrained systems: acknowledging within and cross-farm

variability effect on yield and NUE. In “Proceedings from the 5th International Symposium

for Farming Systems Design: Multi-functional farming systems in a changing world”. Edited

by ESA and Agropolis International, pp 491-492, 7-10 September, Montpellier, France.

Nhantumbo, N., Mortlock, M., Nyagumbo, I., Dimes, J and Rodriguez, D (2015) Improving

farming systems design and management: the role of model assisted participatory crop season

planning in Sub-Saharan Africa. Presented at the TropAg2015 conference in Brisbane

Publications included in this thesis

None

Page | v

Contributions by others to the thesis

Key Mathews, a part-time statistical advisor for QAAFI RHD students, provided early advice

on experimental design and analysis techniques using GenStat during proposal

development and preparations for the first field trip.

For chapter 4, preliminary soil characterization and routine plant analysis were conducted at

the Cropnuts Soil and Plant Laboratory at ICRAF in Nairobi, with further analysis –

nitrate N (NO3-N) determined at the Soil and Plant Analysis Lab of the Faculty of

Agriculture of the Manica Higher Polytechnic Institute with Support from Eng.

Jacinto M. Andreque and Eng. Clemente Zivale.

Miranda Mortlock, have helped revise most of chapter 3, offering key advise on data

presentation and interpretation. Miranda also provided further statistical advice for

chapter 3 and 4 of this thesis.

John Dimes and Peter DeVoil provided technical backstop on APSIM model. This consisted

on direct training and further assistance with model parameterization for chapter 5.

Language editing: Dr John Schiller

Statement of parts of the thesis submitted to qualify for the award of another degree

None

Page | vi

Acknowledgements

I have learned a lot throughout this PhD journey. I am truly grateful to the SIMLESA Project

and also to the Australian Centre for International Agricultural Research (ACIAR) which,

through the John Allwright Fellowship Program, gave me this unique opportunity. My special

thanks go to my supervisor, Dr. Daniel Rodriguez, for keeping his word after our first casual

meeting in August 2010, during the SIMLESA launching meeting, when he gave me his

business card and immediately shared with me scholarship application information. I will be

always grateful for this gesture. I also acknowledge the SIMLESA Africa team, Dr Mullugeta

(SIMELSA Africa Coordinator), Dr Isaiah Nyagumbo (Malawi and Mozambique, Objective

2 Coordinator and my Co-supervisor) and Eng. Domingos Dias (SIMLESA Country

Coordinator) for bringing me into the SIMLESA family. I also want to thank Dr John Dimes

for the great discussions we had on residue management in Nairobi during the breaks of the

1st SIMLESA annual planning meeting back in March 2011. Thanks also for the endless

discussions and support relating to APSIM modelling throughout my PhD. My extended

thanks also go to the Instituto Superior Politecnico de Manica Director General, Dr Rafael

Massinga, for introducing me to the SIMLESA team. Without the direct or indirect

contribution of those people I wouldn’t have made it to UQ. For that reason, I will always be

grateful to you for all for this opportunity.

I am exceedingly grateful to my team of advisors for their support throughout my PhD

studies, and especially for keeping up with me throughout the process. Managing my field

work back in Mozambique, the re-allocation to Australia far from my family, and

administrating the tight schedule between my arrival in Australia after field work and the

milestone attainment deadlines was really challenging. I couldn’t have made it without the

support of my advisory team, especially Dr. Daniel Rodriguez, who had to directly manage

all these issues. To Miranda, for the valuable inputs in the survey design and analysis, and

precious time spent talking about it. Thanks also for bringing the human element into my

research and time I spent in Australia. All these experiences were rewarding for me and made

me grow both as a researcher and human being.

Page | vii

I would also like to send my special thanks to Ms Sharon Harvey in ACIAR for her support

throughout the scholarship process. Special thanks, also goes to my fellow PhD colleagues,

Abeya, Mesfin, Solomon Admassu and his family and Solomon Hassem, for the time we

spent in Australia together. To Caspar for the talks in Mozambique during our field work. I

wish you all the best, hoping that this is not the last time we do something together.

My special thanks to my beloved wife Ana, without whom I couldn’t have made it so far. Te

amo muito e ser-te-ei eternamente grata pelo amor, apoio incondicional, amizade,

companheirismo e principalmente por me aturares e teres sido o alicerce da família durante

este processo todo. To our kids, Makhida, Aione and Thulani, o pai pede imensas desculpas

por ter perdido estes 4 anos da vossa vida. Espero que um dia vos possa compensar e me

possam perdoar pela ausência. Espero também que este sacrifício tenha valido a pena e

possa ajudar a mostrar-vos que a gente pode chegar onde quiser desde que estude e corra

atras dos nossos objetivos.

Special thanks also to my family - mum, dad, brothers Betinho, Pique, Nina, and Salo, and

their families, for the endless support and words of encouragement throughout my studies.

Nothing that I do will ever payback the support you always gave me. Betinho obrigado por

olhares pelos meus matewes na minha ausência.

I would also like to thank the Instituto Superior Politecnico de Manica management board for

allowing me to pursue my studies. To my colleagues at the Faculty of Agriculture (DivAG),

without forgetting the Faculty of Economy, the Technical and Administrative Support Team

(CTA’s), students, my field team and friends in general, for their words of encouragement

throughout this journey. I hope my journey can help inspire you to fallow the same path so

that in the near future we can have an institution and country with people highly qualified and

actively conducting research that can help make a difference in the lives of those most in

need.

Page | viii

Keywords:

residue-fertilizer management, resource allocation, farm typologies, water use efficiency

(WUE), nitrogen use efficiency (NUE), maize-legume cropping systems, conservation

agriculture, APSIM, sustainable intensification, technology adoption

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 070105 Agricultural Systems Analysis and Simulation, 40 %

ANZSRC code: 070107 Farming Systems Research, 30 %

ANZSRC code: 070302 Agronomy, 30%

Fields of Research (FoR) Classification

FoR code: 0701 Agriculture, Land and Farm Management, 60 %

FoR code: 0703 Crop and Pasture Production, 30%

FoR code: 0799 Other Agricultural and Veterinary sciences, 10%

Page | ix

Table of Contents

ABSTRACT i

List of Tables xiii

List of Figures xv

Abbreviations xix

Thesis outline xx

CHAPTER 1 23

1. General Introduction 23

1.1 General Research Background 23

1.2. Research Questions, Objectives and Hypothesis 25

1.3 Study area description 27

1.3.1 Target Agroecological zone 28

1.4 General Methodology 29

1.4.2 Trial site characterization 34

CHAPER 2: 39

2. Literature Review 39

2.1 Fertilizer use and demand promotion in Sub-Saharan Africa 39

2.2 Tailoring fertilizer recommendations to SSA farming environments 40

2.2.1 Optimized fertilizer recommendations and Micro-dosing 41

2.2.2 Integrated Nutrient Management (INM) 42

2.2 Fertilizer, Residues and Legumes interaction in SSA Conservation Agriculture 45

2.2.1 Crop residue contribution to yield, water and N-availability 45

2.2.2 Yield penalties from residue retention in continuous cereal systems 46

2.2.3 Competing claims on residue use: from mixed crop-livestock systems to non-

cattle dominated systems 47

2.3 Legumes integration in SSA smallholder systems managened under CA 48

2.3.1 Legume performance in Africa 48

2.3.2 Potential contribution from legume crops in improving N-fixation 52

2.4 Using agricultural simulation models as decision support tools in SSA 54

2.4.1 Using the Agricultural Production Systems Simulation Model (APSIM) 55

2.4.2 Gaps in modelling crop residue decomposition and N-cycling with APSIM 57

References 59

CHAPTER 3: 74

Page | x

Characterising Farmers Diversity and Opportunities for Sustainable Intensification 74

Abstract 75

3.2 Materials and Methods 78

3.2.1 Geographic setting and context of our study 78

3.2.2 Sampling and data analysis 79

3.3 Results 83

3.3.1 Farm household characteristics - Household demographic and land use 83

3.3.2 Farming systems design and management 84

3.3.3 Resource use and allocation strategies 87

3.3.4 Farm typologies formulation 90

3.4.5 Farm typologies 94

3.5 Discussion 98

3.5.1 Improving agricultural advisory services through typology tailored intervention 98

3.5.2 Handling the key intensification traps across contrasting farm typologies 100

3.5.3 Building from current farming systems design 103

3.6 Conclusions 104

References 105

CHAPTER 4: 111

Best-fit Residue Management in Rain Fed N-Deprived Systems 111

Abstract 112

4.1 Introduction 113

4.2. Materials and Methods 114

4.2.1 Field Experimentation 114

4.2.2 Measurements 116

4.2.3 Statistical analysis of field experiment data sets 117

4.3. Results 117

4.3.1 Crop growing conditions – rainfall distribution 117

4.3.2 Maize and cowpea yield response to residue and fertilizer application across

sites 118

4.3.3 Resource Utilization 122

4.3.4. Short-term effects from residues and N-fertilizer allocation startegies 130

4.4. Discussion 134

4.6. Conclusions 138

Page | xi

Acknowledgements 139

References 139

CHAPTER 5: 146

Pathways for Sustainable Residue Management 146

Abstract 147

5.1 Introduction 148

5.1.1 Conceptual framework 149

5.2. Materials and Methods 151

5.2.1 Experiment for model parameterization 151

5.2.2 Long term simulations: residue induced yield, NUE and WUE responses 154

5.2.3 Model testing and Evaluation 155

5.3. Results 155

5.3.1 Model evaluation 155

5.3.2 Interactions between residue and fertilizer and their impact on yield 158

5.3 Resource productivity 161

5.4. Seasonal rainfall variability and their effects on residue responses 163

5.5 Discussion 166

5.6 Conclusion 170

References 170

CHAPTER 6 177

General Discussion 177

6.1 Summary of key findings 178

6.2 Typology tailored intervention: key entry points to downscale technologies 180

6.3 Optimizing responses through smart and pragmatic resource allocation strategies 182

6.3.1 Residue and fertilizer responses: linking empirical and model assisted analysis 182

6.3.2 Model assisted analysis and decision support framework 183

6.3.4 Demonstrating short term effects from alternative resource allocation strategies 185

6.4 Conclusions and way forward 186

6.4.1 Scaling out and scaling up innovations 187

References 191

Appendix 3.1 Generalized crop management timelines across regions and farmer

groups 199

Page | xii

Appendix 3.2 Crop distribution across farming environments and soils in both AEZ-

R4 and AEZ-R10. 200

Appendix 3.4: Farm household clusters by their fertilizer and manure allocation

preferences 201

Appendix 3.5: Households clustered by their reasoning for not using fertilizers or

manure 201

Appendix 3.6: Fertilizer and manure allocation preferences among residue users 202

Appendix 3.7: Correlation between household size and number of household farm

active people disaggregated by age and sex 204

Page | xiii

List of Tables

Table 1-1 Treatment key with indication of tested factors disaggregated by site ................... 33

Table 1-2 Soil starting condition in the first season (2013-14) of trials ................................. 35

Table 2-1 Main Lessons from Fertilizer Research and Promotion in SSA in Recent

Years .......................................................................................................................... 44

Table 2-2 Current findings on legume contributions to nitrogen fixation in African

cropping systems ....................................................................................................... 51

Table 2-3 The potential of APSIM in aiding agricultural research in SSA ............................ 56

Table 3-1 Household and farming systems characterization variables used for the

categorical principal components analysis (CATPCA) ............................................. 80

Table 3-2 Household characterization based on demographic data and land use ................... 84

Table 3-3 Loading values for the correlation matrix for 21 farm household

characteristics in Matsinho (Vanduzi) and Marera (Macate). ................................... 91

Table 3-4 Households distribution and characterization across farm types and sites ............. 95

Table 4-1 Significance levels for the main factors (Crop, R-levels, N-levels, Site)

and their interactions effects on yield, stover, crop water use (CWU), water

use efficiency (WUE), nitrogen use efficiency (NUE), nitrogen uptake

efficiency (NUpE), nitrogen utilization efficiency (NUtE) and N-

partitioning into shoot N (Nsh) and Grain N (Ng), at physiologic maturity ........... 121

Table 4-2 Significance levels for the residual effects from residue-fertilizer

allocation strategies on yield and resource productivity of cowpea-maize

and maize-maize sequences measured in the 2014-15 season ................................. 131

Table 4-3 Short term yield, N-uptake, NUpE, NUtE, NUE and WUE advantages

from residue and nitrogen fertilizer application into cowpea attained in a

subsequent bioassay maize crop. Results were obtained after one-year

comparison between unfertilized maize grown in plots previously planted to

cowpea and maize at different residue and N-application levels. ........................... 133

Page | xiv

Table 5-1 Initial soil organic carbon (SOC), nitrate-nitrogen (NO3-N) and soil water

description parameters from soil samples collected in the 2013-14 seasons

and were used for model parameterization .............................................................. 153

Table 5-2 Overall model evaluation for selected parameters ................................................ 156

Table 5-3 Inter-seasonal performance evaluated through variations in long term

simulated maize and cowpea water productivity (WUE) and nitrogen use

efficiency (NUE) response to residue application across N-levels for a low

N and low water holding capacity environment. ..................................................... 165

Page | xv

List of Figures

Figure 1-1 Trial layout. For each site a layout was generated through the GenStat 16

trial design tool. At site level, crop types was the whole plot, residue levels

where the sub-plots and N-level the sub-subplots ..................................................... 33

Figure 1-2 Topsoil volumetric soil water content (VWC) measured between

flowering and physiological maturity across the three tested soils in the

2013-14 and 2014-15 cropping seasons. ................................................................... 37

Figure 1-3 Post-harvest soil nitrate (NO3-N) distribution down the profile on maize

(ABC) and cowpea (DEF) treatments and sites. Shapes identification: ∆ -

Sandy Clay loam (cSaCL); O Fine sandy (fSa) and ◊ Sandy loam soil

(SaL). For each soil and N-level (ND: 0 kg N/ha; LNA: 23 kg N/ha and

HNA: 92 kg N/ha), R represents residue treatments, i.e., with (R+) and

without (R-) residue application ................................................................................ 38

Figure 3-1 Cropping systems description – crops and sequences for the AEZ-R4

gathered through the model assisted focus group discussions and resource

allocation maps (RAM’s) designed by local farmers ................................................ 86

Figure 3-2 Preferred fertilizer, manure and residue allocation across crops over the

last 10 years in sampled districts. Interpretation: 1-Mz is maize, 2-Leg is

legume and 3-Veg is vegetables and 4 is shared across crops, excluding

vegetables. ................................................................................................................. 89

Figure 3-3 Framework for farmer categorization and personalize agricultural

intervention. Key grouping factors were attitudes towards resource use

(fertilizer, manure) and the household maize self-sufficiency level .......................... 93

Figure 3-4 Hypothetical model for resource endowment based farm typologies.

Typology specific develop pathways are indicated by dashed arrows.

Continuous arrow indicates a stepping out situation (Dorward et al., 2009),

i.e., a level of income diversification where the farmer is able to generate

enough off the farm income to reduce dependence from farming. ............................ 94

Figure 4-1 Rainfall distribution, daily maximum, minimum temperature and

radiation from 1st September 2013 to 4th April 2014 in the 2013-14 cropping

Page | xvi

season (left) and 2014-15 cropping season (right). S is sowing date, (F)

flowering and (H) harvest ........................................................................................ 118

Figure 4-2 Crop response to residue application across different N-fertilization

levels on three contrasting soil types from Central Mozambique. ND:

Unfertilized N-deficient system (0 kg N/ha), LNA: low nitrogen application

system (23 kg N/ha), HNA: high N-application (92 kg N/ha). S – significant

differences between residue application levels. Different letters across N-

levels for same R-level indicated significant differences at p<0.05

significance. Caps lock – separate treatments without residues. Small letter

separate situations with residues .............................................................................. 120

Figure 4-3 Maize and cowpea grain yield as a function of crop water use from

emergence to maturity measured across sites. The continuous line

represents the attainable water use efficiency of 20 kg grain/ha.mm (a) and

10 kg grain/ha.mm (b), for maize and cowpea respectively. The dotted lines

represent the maximum actual WUE for each crop across the tested

environments. Theoretical WUE, based on APSIM model simulations for

this agroecological zone are of 17.9 kg grain/ha.mm for maize and 7.74 kg

grain/ha.mm for cowpea, respectively. .................................................................... 123

Figure 4-4 Relationship between crop WUEyield and N-fertilizer application across

sites (N-level x Site interaction). N-levels: ND: N-deficient system (0 kg

N/ha), LNA: low N-application system (23 kg N/ha) and HNA: Non-

limiting N (92 kg N/ha). Soil textures: fSa – fine sand soil, SaL – sandy

loam soil and cSaCL - coarse sandy clay loam soil. The vertical line in each

plot represents the LSD value for WUEyield comparison between N-levels

for each site at p<0.05. N-levels spaced higher than the LSD bar for the

same site are significantly different. ........................................................................ 125

Figure 4-5 Relationship between crops NUE and N-fertilizer application across

sites. N-levels: ND: N-deficient system (0 kg N/ha), LNA: low N-

application system (23 kg N/ha) and HNA: Non-limiting N (92 kg N/ha).

Soil textures: fSa – fine sand soil, SaL – sandy loam soil and cSaCL -

coarse sandy clay loam soil. Continuous line in each plot represent the LSD

value for NUE comparison between N-levels for each site at p<0.05. N-

Page | xvii

levels spaced higher than the LSD bar for the same site are significantly

different. ................................................................................................................... 127

Figure 4-6 Relationship between NUE and its components observed across sites of

contrasting moisture regime. In maize as NUpE was the primary driver of

NUE (a.1), while in cowpea (b.1, b.2) a, NUtE appeared to be major driver

of NUE changes. Shape identification: ∆ - Sandy Clay loam (cSaCL); O

Fine sandy (fSa) and ◊ Sandy loam soil (SaL). Empty and filled shape

indicate residue application levels ........................................................................... 128

Figure 4-7 Nitrogen driven trade-offs between NUE-WUE. NUE and WUE were

averaged across sites and residue application levels for maize (a) and

cowpea (b). Bellow: Maize (c) and cowpea (d) trade-offs as affected by

residue application levels. NUE and WUE for both crops were averaged

across sites. Residue application levels: R- and R+ indicate residue free and

residue applied systems. .......................................................................................... 129

Figure 4-8 Short term yield, NUpE and NUE return from a maize crop grown in

plots previously planted to cowpea (Cwp-Mz) and maize (Mz-Mz) under

contrasting residue and nitrogen application levels Treatment description:

ND, LNA and HNA represent unfertilized, low and high N-application

respectively. Previous residue status: with (+) and without (-) residue. (s)

represents significant differences between residual treatment effects at

p<0.05. ..................................................................................................................... 132

Figure 5-1 Above - Comparison between observed and APSIM model simulated

maize and cowpea yield distributed across three sites. Shapes identification:

∆ - Sandy Clay loam (cSaCL); O Fine sandy (fSa) and ◊ Sandy loam soil

(SaL). Filled shapes for each soil type represent residue treatments while

empty shapes did not receive any residue. Middle and bellow (A, B and C):

Overall comparison between observed and simulated maize and cowpea

yield (a), NUE (b) and WUE(c) distributed across tested sites for.

Continuous lines indicate a 1:1 fit and dotted lines indicate a best fit line

between observed and simulated. ............................................................................ 157

Figure 5-2 long term simulated maize and cowpea yield responses to residue

application and N-application levels (0, 23, 92 kg N/ha representing ND,

Page | xviii

LNA and HNA systems) in a high WHC (cSaCL) and prolonged wetness

(SaL) soils. No residue (R-); Residue applied (R+) ................................................ 159

Figure 5-3 Long term simulated maize and cowpea yield response to residue and N-

application levels (0, 23, 92 kg N/ha representing ND, LNA and HNA

systems) in low WHC (fSa) soil. No residues (R-), Residue applied (R+) ............. 161

Figure 5-4 Residue modified nitrogen driven NUE and WUE trade-off simulated

across maize and cowpea grown in contrasting farming environments of

central Mozambique. Overall means are presented for each level of N-

application namely: unfertilized (ND, 0 kg N/ha), Limited N-applications

(LNA, 23 kg N/ha) and high N-application (HNA, 92 kg N/ha). Residue

application levels are indicated as R- and R+ for no residue and residue

applied plots respectively. ....................................................................................... 162

Figure 5-5 Above, in the rainfall distribution graphs (A, B and C) for the period

1951-2014, seasons are categorized according to simulated maize yield

variations under high C/N residue application at three different soils. The

horizontal line indicates annual average rainfall over the simulated period.

Dark bars – indicate seasons were yield losses from residue application

were simulated; Red bars indicate seasons in which positive responses to

residue application are expected across N-application levels. In graphs A1

to C3, relative maize yield variations to residue application disaggregated

by soil and N-application levels over the simulated period. .................................... 164

Figure 5-6 Nitrogen and water stress levels in maize across residue application

levels for a high WHC environment in a normal and poor rainfall season. A

stress factor of 1 indicates full stress and 0 absence of stress. Stress indexes:

npho is the N-stress and wpho is water stress. ......................................................... 168

Page | xix

Abbreviations

APSIM Agricultural Production System Simulator

BNF Biological nitrogen fixation

CATPCA Categorical Principal Component Analysis

C/N Carbon to nitrogen ratio

CA Conservation agriculture

CMYP Continuous maize yield penalties

cSaCL Coarse sandy clay loam soil

CWU Crop water use

DAP Draft animal power

fSa Fine sandy soil

HNA High nitrogen application system

HTP Hired tractor power

INM Integrated Nutrient Management

ISFM Integrated Soil Fertility Management

LENI(S)A Low external nitrogen input sustainable agricultural systems

LENIA Low external nitrogen input agricultural systems

LNA Low nitrogen application system

LLR Land to labour productivity ratio

LSD Least significant difference

ND Unfertilized nitrogen deficient system

NNA Non-limiting nitrogen application

NO3-N Nitrate nitrogen

NUE Nitrogen use efficiency

NUpE Nitrogen uptake efficiency

NUtE Nitrogen utilization efficiency

PAWC Plant available water content

ppm Parts per million

R+ and R- With and without residue application

SaL Sandy loam soil

SIMLESA Sustainable Intensification of Maize-legume systems in Eastern and Southern Africa

SOC Soil organic carbon

WHC Water holding capacity

WUE Water use efficiency

YLR Yield to labour productivity ratio

Page | xx

Thesis outline

This study is about improving agricultural production and household incomes in nitrogen

constrained smallholders farming systems in Mozambique. This thesis document comprises

the following components (Figure 1):

Chapter I: Introduction

In this chapter, the general research background is set with a focus on resource use and

resource allocation for the rain fed N-deprived systems of Central Mozambique. The general

research background, the main question, objectives, hypothesis and general methodology are

presented. A short characterization of the environments where the trials were conducted, with

a focus on the biophysical aspects, is also presented.

Chapter 2: Literature Review

This chapter reviews the literature in reference to the key issues of: (1) fertilizer use in SSA;

(2) the effect of crop residues on yield, water and nutrient cycling in maize-legume cropping

systems; (3) legumes performance and potential contribution to improve N-availability

through biological N-fixation in Africa; and (4) how agricultural simulation models, such as

the Agricultural Production System Simulator (APSIM), can be used as analytical and

decision support tools to generate and disseminate relevant agricultural information in

Africa.

Chapter 3: Farmers diversity and opportunities to sustainable intensification

This chapter, reports on the findings of a household survey aimed at assessing farmers’

attitudes towards residues, fertilizers, manure and legume use, in their cropping systems.

Functional farmer’s typologies based on the combination of overall household livelihood

strategies, resource use behaviour and their perceived impact on local farming systems design

and the household ability to meet seasonal food security needs and income generation targets,

were built. These typologies were then used as the bases to formulate entry points for

sustainable intensification of maize-legume cropping systems. In the process, three main

technology adoption traps were identified, and their potential roles are also discussed.

Page | xxi

Chapter 4: Residue management in rain fed N-deprived systems

This chapter builds on the findings from chapter 3. The results from a multiple-site trial to

assess the short term responses from crop residue-fertilizer interactions and their impact on

yields and resource productivity (WUE, NUE, NUpE, NUtE) across crops and contrasting

farming environments are presented. Sites were selected based on data relating to plot

ownership collated during the survey. The focus of this chapter is in demonstration of

alternative allocation strategies of high C/N ratio residues and N-fertilizers that are more

likely to improve yields in the short term, and resource productivity for the rain fed N-

deprived systems.

Chapter 5: Pathways for sustainable residue management

The main focus of this chapter is on matching technological solutions to farmers’ needs in

terms of residue, legume and fertilizer management strategies in rain fed N-deprived

environments. A model based decision support framework, is presented and discussed as an

alternative toolkit to generate generalizable information that can be used to tailor residue

management strategies in the region. The framework presented in this chapter focus the

analysis of long term seasonal responses to residue application across crops, levels of N-

application and sites of contrasting soil water holding capacity. In this framework, the

understanding of seasonal variability effects on crop responses to residues is used as the key

to map the likelihood of positive or negative responses to residues for specific crops, sites and

levels of N-application.

Chapter 6: General discussion and conclusions

This chapter outlines and discusses the key findings from the study, the lessons learned, the

applicability of the results and the way forward in scaling out the results.

Page | xxii

Figure 1 Thesis outline layout, which includes chapters’ sequence and methodology used

Page | 23

CHAPTER 1

1. General Introduction

1.1 General Research Background

Over recent decades, the need to sustainably increase crop and land productivity in Sub-Saharan

Africa (SSA) has led to the widespread promotion of several agricultural technologies in Africa.

Conservation agriculture (CA) is one of the technologies that has received wide attention and has

been widely promoted throughout Africa under the assumption that it could help revert soil

degradation - erosion, mineral and organic matter decline, chemical in SSA agriculture (Giller et

al., 2011a; Ngwira et al., 2013; Thierfelder et al., 2013; Wall, 2007). However, the achievements

with CA in Africa are still far below those obtained in America, Asia, Europe and Australia

(Brouder and Gomez-Macpherson, 2014). In these regions, the adoption of CA has reached 70-

90% of the cropped land, and proved effective in slowing soil degradation while increasing crop

yields at the same time (D'Emden et al., 2006; Derpesch, 2008). However, attributing gains in

productivity solely to the adoption of CA principles is difficult, given the multiplicity of

associated investments made in those regions e.g. irrigation systems, mechanization, improved

seed and crop varieties, high use of inorganic fertilizers and chemical weed control.

It has been estimated that more than 50% of registered increases in agricultural production

achieved during the green revolution is attributable to high input uses, mechanization, irrigation

and enhanced crop response to inorganic fertilizers (Kelly, 2006; Roy et al., 2006). In SSA,

however, the practice of extensive rain fed agriculture (Alene and Coulibaly, 2009; Giller et al.,

2006), the recurrent poor access to inputs (Bationo et al., 2012a; Chianu et al., 2012), lower

mechanization levels and weak access to relevant agricultural information, are among the factors

that have hindered the realization of the continents agricultural potential. Despite these

constraints, in 2006 the Abuja declaration pledged to help increase fertilizer use in Africa from

8.3 to 50 kg/ha by 2015 (Kintché et al., 2015)1. However, to date there remain a number of

constraints to increased fertilizer use in Africa.

1 The declaration refers to gross nutrient amounts.

Page | 24

Despite the acknowledgement that, in the short term, “inorganic fertilizers are the best way to

feed Africa” (Gilbert, 2012), it is also true that smallholder farmers in Africa have been reluctant

to invest in inorganic fertilizers. Major constraints to fertilizer use in Africa are high fertilizer

costs mainly exacerbated by poor fertilizer markets, lack of institutional support and

inappropriate fertilizer packaging (Ade Freeman and Omiti, 2003; Chianu et al., 2012). Poor

perception about fertilizer benefits exacerbated by unreliable responses negatively affects

farmers decision to use inorganic fertilizers (Chianu and Tsujii, 2005). As a result, fertilizer

application rates in SSA are extremely low ca. 9-13 kg/ha (Tenkorang and Lowenberg-DeBoer,

2008). Therefore, within the context of poorly resourced smallholder farmers and the nitrogen

deprived farming systems of SSA, a more integrated and pragmatic resource use approach that

takes into account the socio-economic and biophysical diversity of smallholder farmers and the

resource pool at their disposal, appears to be a more feasible option to increase yields and

farmers’ food security prospects. In Africa, ‘integrated soil fertility management’ (ISFM) has

been presented as a viable and locally feasible alternative to improve nutrient availability

(Bationo et al., 2012b; Bationo and Waswa, 2011). The focus on a combined use of locally

available amendments and legume to increase nutrient availability links well with CA strategy in

Africa. Nevertheless, tailoring both ISFM and CA still poses a major challenge in Africa.

The CA strategy for Africa is built under the assumption that, minimum soil disturbance, crop

residue retention and legume integration in maize dominated systems offer a more affordable

option to sustainably increase soil fertility and yields in SSA. However, small amounts of plant

residues are left as ground cover due to low biomass production, high competition for the

residues mainly in livestock dominated areas (Erenstein, 2003; Erenstein et al., 2015;

Hauggaard-Nielsen et al., 2003; Valbuena et al., 2012). Poor legume performance in Africa

(Giller et al., 2009), is also a recurrent problem in Africa. Furthermore, the most widely available

and used residues in most smallholder farming systems are of high C/N ratio, and the

decomposition of such material is known to interact strongly with soil mineral N supply (Ros et

al., 2011). However, this relationship has been poorly studied, and integrated datasets reporting

on the combined effect of all CA principles are still scarce, particularly in SSA.

From 2011-2014 (SIMLESA Phase 1) to 2014-2019 (SIMLESA Phase 2), Mozambique has been

involved in an Australian Centre for International Agriculture Research (ACIAR) funded project

Page | 25

– The Sustainable Intensification of Maize-legume Cropping Systems in Eastern and Southern

Africa (SIMLESA). The SIMLESA project is being implemented across 5 countries, Ethiopia,

Kenya, Tanzania, Malawi and Mozambique. The intensification of maize-legume cropping

systems through the CA-ISFM platform is the focus of the SIMLESA intervention. However,

there is still a need to identify feasible options to sustainably intensify these maize-legume

cropping systems among resource poor smallholder farmers, considering that their farming

systems are diverse, complex and highly context bounded. Therefore, understanding the

interactions between farmers’ circumstances, their resource use behaviour and farming systems

design, is a critical step towards tailoring new technological packages to fit the local reality –

farmer groups and geographic context. In the meantime, is also true that little research has been

done on achieving an understanding of the SSA N-deprived smallholder farming systems those

dynamics in Central Mozambique.

This research, which is an integral part of the SIMLESA project in Mozambique is focused on:

1) identifying the best fit allocation strategies for limited resources such as high C/N ratio crop

residues and inorganic N-fertilizers, across crops and environments; 2) generating information

for a better understanding of the interactions between the widely used high C/N ratio residues

with N-availability in rain fed N-deprived systems; 3) understanding to what extent farmers

circumstances and their resource use behaviour affects local farming systems design and the

ability of those farmers to engage in new practices; and finally 4) because no change can be

achieved without a reliable and time effective decision support system, this study also explored

the capabilities of the Agricultural Production Systems sIMulator (APSIM) to simulate

alternative residue-fertilizer allocation strategies across crops and environments, in the hope that

the model can be used as part of a model assisted decision support framework on residue

management.

1.2. Research Questions, Objectives and Hypothesis

The overall research question of this study is: what are the benefits and trade-offs from

alternative crop residue management in the rain fed low external N-input maize-legume systems

in central Mozambique. The general hypothesis is that: ‘Best fit technological interventions need

to be found from an integrated systems approach that evaluates the benefits and trade-offs from

Page | 26

alternative resource management strategies and sustainable intensification pathways, in terms of

food production, risks and environmental outputs’.

General aim and specific objectives:

The aim of this research is to identify more productive and locally feasible sustainable residue

management practices capable of helping improve rainfall capture and nitrogen use efficiency,

and which are more likely to be adopted by resource poor smallholder farmers in rain fed low

external N-input agricultural (LENIA) systems. To achieve this goal, the following specific

research objectives were proposed:

Socio-economic and bio-physical characterization: To assess the performance of

existing practices of residue and nitrogen management, and to identify problems and

opportunities for improvements that are likely to be adopted by the poorly resourced

farmers of Central Mozambique;

Cropping systems research: To identify and test opportunities for better allocation of

crop residue and fertilizers that are more likely to provide short term benefits in terms of

improving crop yield, soil nitrogen, and water availability, in nitrogen constrained

environments;

Pathway to impact: To develop simple recommendations for better matching

technological solutions to smallholder farmers’ needs in terms of residue, legume and

fertilizer management strategies.

Hypothesis Tested

The objective of the research trials was to test the following specific hypotheses:

1) For the nutrient constrained smallholder farming systems of central Mozambique there

are household and site-specific management combinations of technologies e.g. residues,

and fertilizers that best fit farmers’ bio-physical and socioeconomic circumstances.

2) In wetter nitrogen constrained environments, investing the limited availability of high

C/N ratio stubble in improving the performance of the sole legume crop is more

beneficial to the system than using the residues for the maize crop.

Page | 27

3) In drier and variable climates, investing the limited availability of maize stubble in

improving the rainfall capture for the maize crop is likely to provide more benefits than

applying the residues to the legume crop.

4) For wetter soils, the beneficial effects of applying crop residues onto the legume crop are

likely to be observed on the following cereal crop.

5) The balance between the positive and negative effects of residue retention in maize-

legume systems (i.e. water balance, and the temporary nitrogen immobilization), can be

managed with nitrogen fertilisers.

1.3 Study area description

Central Mozambique is home for 58.6% of the country’s agricultural households distributed

across four provinces, Sofala, Manica, Tete and Zambezia. The region has one of the most

important agricultural breadbaskets, the Beira Agricultural Growth Corridor (BAGC) with ca.,

227.000 km2 covering Sofala, Manica and Tete provinces. The BAGC, has a total population of

4.8 million (19 persons/km2) and 10 million hectares of arable land, however, only 16.5% are

cultivated according to estimates from the last agricultural census (INE, 2011). The average

yields are around 1.2 t/ha for maize and less than 0.5 t/ha for most legume crops (FAOSTAT,

2014).

In 2011, with the approval of the countries Strategic Plan for the Development of the

Agricultural Sector (PEDSA)2 which points at the need to increase agricultural productivity

across the country’s agricultural breadbaskets as critical to improve the performance of the

agricultural sector, the BAGC have attracting several investments from the government and its

developmental partners, among them the World Bank, the Alliance for Green Revolution in

Africa (AGRA), the Food and Agriculture Organization (FAO), the International Fund for

Agricultural development (IFAD), the United States Agency for International Development

(USAID) and most recently the Australian Agency for International Development (AusAID)

through ACIAR agencies. This attention is in part due to: (1) the region’s rich agricultural

2 PEDSA – Plano Estratégico de Desenvolvimento do Sector Agrário. PESDA was approved in May 2011

and its main pillars are: (1) Increase agricultural productivity (ii) Facilitate access to market through the

improvement of infrastructure and development of value chain; (iii) Promote a sustainable use of natural

resource (iv) Strengthen local institutions – research and collaborators;

Page | 28

potential, has it hosts most of the highly productive agroecological regions of the country with an

expected production potential of more than 8.0 t/ha for maize and 3.0 t/ha for soybeans; (2) it

concentrates the highest extension of arable and cultivated land and (3) the existence of a

favourable support infra-structure mainly access to water, roads and transport to facilitate trade

within the regions and its surrounding provinces and neighbouring countries.

The vast mosaic of agroecological zones (AEZ) in central Mozambique, makes agriculture a

highly complex and diverse activity in the area. The agroecological diversity heavily influences

farming systems design and management strategies within and across regions. For example,

within the four provinces that constitute the central Mozambique agricultural development

region, at least 3 agro-ecologies can be identified in each central Mozambique province –

Manica (AEZ-R4, AEZ-R5, AEZ-R6, and AEZ-R10), Sofala (AEZ-R5, AEZ-R4, and AEZ-R6),

Tete (AEZ-R6, AEZ-R7, and AEZ-R10) and Zambezia (AEZ-R5, AEZ-R7, AEZ-R8, and AEZ-

R10) making farming highly context bounded. The intra and cross-agroecological variability that

characterizes the region can be seen in its climatic resource map as described in (Reddy, 1984).

1.3.1 Target Agroecological Zone (AEZ)

The study was conducted in Manica province in central eastern Mozambique in the medium

altitude agroecological zone (AEZ-R4). The AEZ-R4, falls within the plateau zone and are

characterized by two well-defined rainfall seasons: a wet season from November to late April

which receives 78-99% of the total annual rainfall, and a dry season that runs from May to

October. AEZ-R4 is characterized by a tropical dry climate and covers 80% of the Manica

province and part of the Sofala. The region is located in a medium planaltic zone with altitudes

between 200-800m. The annual rainfall is of approximately 834 mm, calculated over 63 years

between 1951 to 2014. The average minimum temperature in the region is 16.6°C, with an

average maximum of 32.5°C. The soils are predominately red ferralsols and luvisols (Gouveia

and Azevedo, 1954).

Page | 29

1.4 General Methodology

A combined methodology integrating socio-economic surveys, empirical research trials and bio-

physical modelling, were used to test the proposed hypotheses, as outlined below.

Hypothesis 1: For the nutrient constrained smallholder farming systems of central Mozambique

there are site and management specific combinations of technologies e.g. residues, legumes and

fertilizers, that best fit farmers’ bio-physical, market and socio-economic circumstances, which

will be tested using data generated from the following activities:

Socio-economic survey: The objective of this survey was to understand how farmers use

and manage crop residues, legumes and fertilizers, in their current cropping systems, and

how their attitudes towards resource use affects their ability to meet the household food

security and income generation targets. Survey data was also used to identify key

variables for the construction of new framework for farmer characterization that can help

aid the identification of scalable intensification pathways. To achieve this objective, a

random sample of 131 farming households were interviewed across two villages of

contrasting resource endowment – Matsinho in Vanduzi and Marera in Macate, both

located in the AEZ-R4. In addition, two focus group discussion were held in Marera with

29 farmers representing 7 farmer organizations from Macate, Zembe, Matsinho, Boavista

and Marera.

Participatory Crop Modelling (PCM) Workshops: These were used to complement

the household survey. The PCM workshops helped gather in-depth information on local

farming systems design, i.e., crops, sowing windows, cropping sequences, management

practices, yields, most common farming environments and their relative advantage. In

each meeting, resource allocation maps (RAM) were developed by farmers and used to

parameterize agent based typologies and also APSIM. APSIM was used to run technical

analysis of current cropping systems with an emphasis on plant densities and N-fertilizer

application. The simulations helped understand current cropping practices and identify

opportunities for improvements. The PCM’s also served as a platform to test APSIM

suitability to serve as a decision support tool for local research and agricultural extension

networks.

Page | 30

Hypotheses 2, 3 and 4: In wetter, nitrogen constrained environments, investing the limited

availability of high C/N residues on improving the performance of the sole legume crop is more

productive and beneficial to the system than applying those residues on the maize crop. In drier

and more variable climates, using the limited availability of maize stubble for improving rainfall

capture for the maize crop is likely to provide more benefits than applying these residues on the

legume crop. For wetter environments, the beneficial effects of applying crop residues on the

legume crop are likely to be observed on the following cereal crop. These hypotheses were tested

using the data generated from the following activities:

Cropping Systems Modelling: Modelling was used ex-ante to fine tune the treatments

for empirical research, and ex-post to: (1) explain the results from the empirical

experimentation; and (2) scale out the results from the study across seasonal conditions.

The Agricultural Production Systems Modelling (APSIM) version 7.7 through its

nitrogen (SOILN), RESIDUE and water (SOILWAT) modules, were used as the main

modelling platform.

Field Trials: a researcher managed trials were conducted in Matsinho, central-eastern

Manica province (AEZ-R4) over two seasons to generate GxExM information to identify

best fit residue management strategies for the rain fed N-deprived systems managed

under conservation agriculture. Responses where tested across two crop types, i.e., sole

maize and sole cowpea. Three soils (environments) of contrasting texture and water

holding capacity (WHC) commonly used in Central Mozambique were used. A red

coarse sandy clay loam (cSaCL) textured Ferralsol which is a typical upland soil and

most dominant in the region; a dark sandy loam soil (SaL) Gleysol, typical lowland soil

with prolonged wetness periods; and a dry fine sandy textured (fSa) Arenosol of low poor

water holding capacity. Nitrogen was applied in three levels: 0 kg N/ha representing an

unfertilized N-deficient system (ND), 23 kg N/ha representing a low N-application (LNA

system and 92 kg N/ha representing a high N-application system (HNA). Finally, high

C/N crop residue mulches were applied in two levels, 0 and 6 t/ha. (Table 1.1)

The N-application levels were selected based on two criteria. The 23 kg N/ha assumed for the

LNA systems is an approximation of fertilizer package in FAO input voucher program in

Mozambique which provides approximately 29 kg N/ha through a 50 kg bag of NPK 12:24:12

Page | 31

and a 50 kg bag urea (46 N) to be applied as basal and top-dress fertilizer. To facilitate

recommendations and align the figure with regional fertilizer which range between 18.3 and 24.1

kg N/ha among resource endowed smallholder farmers in our study area an average N-amount of

23.8 kg N/ha averaged across this three values was kept to 50 kg bag of urea assuming only a

top-dress application. The HNA was based on reported N-application amount needed to satisfy

N-immobilization demands during high carbon residue decomposition in SSA LENIA systems

which ranges between 80-100 kg N/ha (Homann-Kee Tui et al., 2015; Vanlauwe et al., 2014).

The amount was kept to 92 kg N/ha, i.e., 4 bags of urea (46 N). The 6 t/ha residue application

rate, was selected to mimic an environment with a high N-immobilization risk due to continuous

incorporation of carbon rich maize stubble and grasses characteristic of central Mozambique

LENIA systems. This value is close to the average amount of maize stubble produced in

Southern Africa if we assume a 3.2 t/ha at a 0.45 harvest index (Kintché et al., 2015). Ex-ante

simulations for Central Mozambique showed an 80-75% chance of achieving a 3 and 4.5 t/ha

yield in ND and LNA systems without residue application meaning that 4±5.5 t/ha maize stubble

are likely to be attained in the region using a maize hybrid. These figures are in line with data

presented in (Baudron et al., 2015c) who estimated an approximately 2428 ± 5890 kg maize

biomass year-1 farm-1 in central Mozambique.

The experiment was conducted as completely randomized block design in a 3x3x2x2 factorial

arrangement, laid in a 4x3 layout (Figure 1-1). All the plots received the same recommended

amount of phosphorous (P) and potassium (K) as NPK 12:24:12 at sowing and nitrogen dozes

were applied in a 50:50 ratios as basal and top-dress. Cowpea, was inoculated at sowing. Plots

were 12m x 6m in the first season and were split into two 6x4m plots in the second season to

accommodate a bioassay maize crop. The bioassay was used to measure the short term residual

effect of the first year treatments while in the other half of the plot the first season trial was

continued to measure cross seasonal effects. No fertilizer neither residues were applied to the

bioassay maize crop in season two and residues from the previous season where retained in the

plot.

Page | 32

1.4.1 Trial data analysis

Before analysis, trial data was checked for homogeneity of variance using the Bartlett’s test.

Because variances where homogeneous, analysis of variance was performed using the following

statistical models for yield and resource productivity response to residue management strategies in year

one and year two bioassay maize crop:

Crop response to residue management strategies (Main Factors: Crop types, N-application levels,

and sites of contrasting soil water holding capacity). Three way interactions where tested as

model bellow shows:

i. (YEAR 1) Response variable = Constant + Crop + R_level + N_level + Site + Crop*R_level

+ Crop*N_level + R_level*N_level + Crop*Site + R_level*Site + N_level*Site

+ Crop*R_level*N_level + Crop*R_level*Site + Crop*N_level*Site +

R_level*N_level*Site + Crop*R_level*N_level*Site

ii. (YEAR 2 - Bioassay crop) Residual responses = Constant + P_Crop + R_level + N_level +

Site + P_Crop *R_level + P_Crop*N_level + R_level*N_level + P_Crop *Site

+ R_level*Site + N_level*Site + P_Crop*R_level*N_level +

P_Crop*R_level*Site + P_Crop *N_level*Site + R_level*N_level*Site +

P_Crop *R_level*N_level*Site

iii. Random model: blocks + blocks*wplots + blocks*wplots*subplots

iv. Number of units: 264

NB: For year 1 statistical model, Crop is maize and cowpea grown in this specific season while in year 2

model, P_Crop represents the previous crop sown, i.e., the sequences maize-maize or cowpea-maize

which were used to measure residual treatment effect.

For each individual crop, cross-site responses to residues across N-fertilizer application levels

were tested for the variables where sites proved to have a significant effect. The model below was

used:

i. Response variable: Constant + R_level + N_level + Site + R_level*N_level +

R_level*Site + N_level*Site + R_level*N_level*Site

ii. Random model: blocks + blocks*wplots + blocks*wplots*subplots

iii. Number of units:132

Page | 33

Table 1-1 Treatment key with indication of tested factors disaggregated by site

Factors Treatment3

Sites

(Whole plots)

Crop

(Subplot)

R-level

(Sub-subplot 1)

N-level

(Sub-subplot 2)

SaL fSa

0 t/ha

0 kg N/ha Cwp0R0N

Cowpea

23 kg N/ha Cwp0R23N

cSaCL

92 kg N/ha Cwp0R92N

6 t/ha

0 kg N/ha Cwp6R0N

23 kg N/ha Cwp6R23N

92 kg N/ha Cwp6R92N

Maize

0 t/ha

0 kg N/ha Mz0R0N

23 kg N/ha Mz0R23N

92 kg N/ha Mz0R92N

6 t/ha

0 kg N/ha Mz6R0N

23 kg N/ha Mz6R23N

92 kg N/ha Mz6R92N

3 Treatment key: Crop * R-level * N-level * Site

Figure 1-1 Trial layout. For each site a layout was generated through the GenStat 16 trial

design tool. At site level, crop types was the whole plot, residue levels where the sub-plots and

N-level the sub-subplots

Page | 34

Hypothesis 5: The balance between the positive and negative effects of residue retention on

maize yield, i.e. grain water productivity (WUE), and the temporary nitrogen immobilization,

respectively, can be managed with the use of nitrogen fertilisers. The experimental results were

used to test the performance of APSIM to simulate the observed crop responses to residue and N

inputs across the farming environments. The calibrated APSIM model and quality historical

climate records were used to extend the analysis across seasonal conditions as a means to

generate relevant and generalizable information.

1.4.2 Trial site characterization

Three commonly used soil types of Central Mozambique were used. These three soils are found

across two distinct farming environments, namely: 1) the upland farming environments, mostly

dominated by the red ferralsols (cSaCL) and Arenosols (fSa); and 2) the lowland wet farming

environments locally called Baixas, which are characterized by prolonged wet periods and are

dominated by Gleysol. Samples for initial soil characterization where collected at the start of the

2013-14 cropping season to 1.20 m depth (Table 1-2) at intervals of 0.15 m for the first two

layers and 0.30 m for the following five layers.

Sampling for starting soil water content and plant water use determination was done at the same

depth intervals, at sowing and physiological maturity. Additional soil moisture measurements

were conducted fortnightly across both seasons, 2013-14 and 2014-15 between maize flowering

and physiological maturity. Measurements were made at 0-15 cm depth in three points per

treatment with the help of a MPKit-406 Soil Moisture Instant Reading Kit from ICT

International. Results showed that the sandy loam soil (SaL), registered the highest topsoil (0-

15cm) volumetric soil water content (VWC) in all sampled dates. The fine sandy soil (fSa)

registering the lowest VWC (Figure 1-2). VWC measurements, as presented in figure 1-1 were

also critical to show the current site variability on in crop soil moisture regimes, an indication of

water holding capacity.

Page | 35

Permanent wilting point (CL15) and field capacity (DUL) were determined opportunistically by

combining the APSIM soil characterization protocol described by Dalgliesh and Foale (1998)

and also Pala et al. (2007). Here, a pond experiment was established to measure soil water

holding capacity and strengthen the texture based method.

Pre-sowing and post-harvest soil nitrate (NO3-N) content down the soil profile (0-1.20 m) was

measured across treatments in both seasons. The colorimetric method based on Merckoquant

nitrate strips - Merck KGaA from Germany was used (Cossani et al., 2012). Results from post-

harvest NO3-N measurements (0-1.20 m) at the end of the first season, showed higher residual

NO3-N in the cSaCL and fSa (Figure 1-3). These figures are in line with data on starting NO3-N

starting which was higher on both soils. Nevertheless, a slight reduction in the overall NO3-N

was measured compared to the starting condition in all soils. In the N-rich cSaCL, residue

application reduced the amount of profile residual NO3-N in ND and HNA systems meaning that

residue application favoured N-extraction especially at higher N-application (HNA). In the drier

fSa, the highest residual NO3-N was measured in residue free maize treatments, meaning that the

crop was unable to extract most of the applied and mineralized nitrogen. Contrarily to maize,

post-harvest NO3-N in cowpea was lower, meaning that the crop managed to extract most of the

N-applied at except at HNA under residue in the SaL.

Table 1-2 Soil starting condition in the first season (2013-14) of trials

Red ferralsol (Red sandy clay loam – cSaCL, local name: Kupswuka)

Soil layer

pH

P(O) K SOC NO3-N SON BD CLL15 PAWC

(ppm) (ppm) (%) (ppm) (%) g/cm3 (%) (mm)

0-15 6.50 25.3 296 1.93 7.79 0.19 1.50 12.1 0.36

15-30 6.37 19.6 206 1.67 6.92 0.19 1.45 13.9 4.47

30-60 6.62 49.2 190 0.90 4.12 0.15 1.52 10.3 19.4

60-90 6.45 27.4 121 0.62 2.29 0.17 1.39 17.9 9.94

90-120 6.47 22 78.5 0.59 1.36 0.16 1.39 17.2 7.81

Page | 36

Gleysol (Dark sandy loam – SaL, local name: N’dongo wa Kupfipa)

Soil layer

pH

P(O) K SOC NO3-N SON BD CLL15 PAWC

(ppm) (ppm) (%) (ppm) (%) g/cm3 (%) (mm)

0-15 4.93 19.8 14.5 2.04 5.10 0.17 1.50 11.4 4.32

15-30 4.89 18.2 13.5 2.37 0.10 0.17 1.52 10.9 8.20

30-60 4.89 18.7 15.2 1.14 0.10 0.19 1.52 10.9 19.1

60-90 4.89 19.3 10.7 0.51 0.01 0.09 1.45 14.0 32.5

90-120 5.01 21.3 13.4 0.31 0.01 0.06 1.43 15.2 27.1

Arenosol (Fine sandy soil – fSa, local name: Djetcha)

Soil layer

pH

P(O) K SOC NO3-N SON BD CLL15 PAWC

(ppm) (ppm) (%) (ppm) (%) g/cm3 (%) (mm)

0-15 672 35.4 97.7 0.59 5.72 0.15 1.64 7.34 0.24

15-30 6.55 42.9 105 0.56 3.27 0.15 1.65 7.14 0.26

30-60 6.61 28.3 100 0.13 0.86 0.12 1.65 7.13 2.74

60-90 6.41 14.3 87.9 0.07 0.87 0.12 1.52 11.4 3.76

90-120 6.36 51.7 60 0.02 0.67 0.09 1.52 11.4 4.67

Page | 37

Figure 1-2 Topsoil volumetric soil water content (VWC) measured between flowering and physiological maturity across the three

tested soils in the 2013-14 and 2014-15 cropping seasons.

Page | 38

Figure 1-3 Post-harvest soil nitrate (NO3-N) distribution down the profile on maize (ABC)

and cowpea (DEF) treatments and sites. Shapes identification: ∆ - Sandy Clay loam (cSaCL);

O Fine sandy (fSa) and ◊ Sandy loam soil (SaL). For each soil and N-level (ND: 0 kg N/ha;

LNA: 23 kg N/ha and HNA: 92 kg N/ha), R represents residue treatments, i.e., with (R+) and

without (R-) residue application

Page | 39

CHAPER 2:

2. Literature Review

2.1 Fertilizer use and demand promotion in Sub-Saharan Africa

In Africa, optimizing resource use and productivity among resource poor smallholder farmers is

paramount to feed the continent’s and growing world population. This is an important step

towards improving yields and food security prospects in the low external input smallholder

farming systems where extensive and poorly managed agriculture has prevailed for years.

Increased fertilizer use is regarded as being responsible for almost 50% of yields increases

during the Green Revolution (Kelly, 2006; Roy et al., 2006). However, despite being recognised

as critical to improve crop yields specially in Africa (Gilbert, 2012) fertilizer use is still the

lowest in the world. In the developing world, especially Sub-Saharan Africa, access to affordable

fertilizer is still a major problem (Darko et al., 2014; Poulton et al., 2006). For example, in the

period 1992-2002, Africa accounted for only 2.87% of the global fertilizer use (FAOSTAT,

2012). Over the same period, Asia, America and Europe, accounted for 56.2%, 21.1% and 17.4%

of global fertilizer consumption, respectively.

Despite a predicted 2.6% annual increase in global fertilizer consumption since the economic

crisis in 2008, Africa still accounts for less than 3.0% of global fertilizer consumption (FAO,

2010; Heffer and Prud'homme, 2011). This is a marginal contribution when compared to Asia

and America that are expected to account for 69% and 19% of the predicted annual growth

(FAO, 2010). The high transport cost in many landlocked countries and regions, poor conditions

of roads, and low population density (people/km2) in rural areas create barriers to the

establishment of viable fertilizer markets (FAO, 2008). The price of agricultural commodities,

farmers risk perception of fertilizer profitability, farmers’ liquidity and the lack of appropriate

fertilizer policies to regulate distribution, mainly among smallholder farmers in remote areas

(Kelly, 2006; Morris et al., 2007), are factors limiting fertilizer access and use. Moreover, lower

fertilizer returns especially in the rain fed smallholder farming systems of SSA where climate

variability has undermined fertilizer response is also a critical adoption factor (Darko et al.,

2014). As a result, only 9 kg/ha of fertilizer are used, a figure 82% less that the 50 kg inorganic

fertilizer use target set for 2015 in Arusha declaration in 2006 (Lambrecht et al., 2014).

Page | 40

In recent years, several attempts to increase inorganic fertilizer use in Africa have been put in

place. Fertilizer promotion initiatives, e.g., the Agro-dealers Development and Support Program

(ADP/ASP) are being implemented in Malawi, Kenya, Tanzania, Rwanda, Ghana and

Mozambique (IFDC, 2012; Minot and Benson, 2009; Poulton et al., 2006). In these countries,

subsidized input vouchers were distributed to smallholders farmers, allowing them to purchase

fertilizer at almost half the market price (IFDC, 2012; Minot and Benson, 2009). However, these

initiatives have come under criticism mainly because of its failure to stimulate independent

demand from smallholder farmers by creating dependence on government handouts (Jayne et al.,

2013). Nevertheless, this cannot be generalized since there is little data to back the claim. In

Mozambican the program has succeeded in helping establish a network of agro-dealers. It is true

however, that in many cases, the vouchers failed to reach the designated beneficiaries, i.e., the

poorest farmer – creating the “elite capture” which is when farmers who have the means to pay

are handed the vouchers in detriment of the one that really need them due to their poor purchase

power (Crawford et al., 2006; Jayne et al., 2013; Minot and Benson, 2009). Moreover, the

voucher programs are not accompanied by proper research and extension programs aimed at

educating farmers on associated practices that could help enhance fertilizer productivity

(Crawford et al., 2006).

2.1.1 Tailoring fertilizer recommendations to SSA farming environments

Despite the low fertilizer use in Africa it is well agreed that, in the short term, inorganic

fertilizers are the best way to improve yields in Africa (Gilbert, 2012). Therefore, extensive

research to optimize fertilizer use and response in SSA farming systems has been conducted in

recent years (Ade Freeman and Omiti, 2003; Kintché et al., 2015; Tabo et al., 2011). Here,

evaluating crop responses has been the main focus.

Several approaches to promote fertilizer use in SSA agro-environments have been hypothesized

in results of ongoing fertilizer use research. Among them, there is the development and

dissemination of broad and site specific fertilizer recommendations (Bationo et al., 2012a;

Bationo et al., 2012b), micro-dosing (Ncube et al., 2007; Tabo et al., 2007; Twomlow et al.,

2010) and most importantly integrated nutrient management (INM) strategies. In the INM

Page | 41

platform, integrated soil fertility management (ISFM) in systems managed under conservation

agriculture has been the major flagship in Africa (Bationo and Waswa, 2011; Mafongoya et al.,

2006).

2.1.2 Optimized fertilizer recommendations and Micro-dosing

The development and dissemination of broad and site specific fertilizer recommendations has

been pointed as a critical step towards improving fertilizer allocation and productivity in Africa.

The underlining principle behind this approach is that knowing African soils is key to improve

fertilizer use and allocation strategies (Bationo et al., 2012a). This goal has been pursued through

initiatives such as the African Soil Information Service (AfSIS) led by the Tropical Soil Biology

and Fertility Institute (TSBF) (Hengl et al., 2015; Towett et al., 2015) and the OFRA4 project led

by CABI International and the Nebraska Lincoln University and covers Kenya, Uganda,

Tanzania, Rwanda, Zambia, Malawi, Mozambique, Ghana, Mali, Burkina-Faso, Niger, Ethiopia

and Nigeria. Both interventions have invested considerable efforts in improving soil

characterization across SSA. Improved soil characterization is key for the development of better

fertilizer recommendation. Despite all efforts, current fertilizer recommendation in Africa are

still unattractive to smallholder farmers who cannot pay for fertilizers. This has been one of the

major criticism of optimal fertilizer recommendations in Africa, since fertilizer use levels among

African smallholder farmers are low. In response to SSA lower fertilizer use levels, alternative

solutions have been developed to overturn smallholder farmers’ inability to invest in the

recommended fertilizer amounts. Micro-dosing is one of those alternatives.

However, the successful implementation of micro-dosing is strongly linked to the existence of

reliable site specific fertilizer recommendations, which is not the case in Africa. Until now,

broad fertilizer recommendations based on the supply of optimal crop requirement ranges under

the assumption that farmers have the capacity to supply such amounts, are promoted. As for

micro-dosing, positive responses were reported across Africa. In Zimbabwe, 30-50% increases in

maize grain yield were measured from the application of 17 kg N/ha (Twomlow et al., 2010). A

42% and 55% grain yield increase in the Sahel was reported after the application of 0.3g N per

4 Optimizing Fertilizer Recommendations in Africa (OFRA)

Page | 42

plant hill in the form of NPK 16:16:16, i.e., 3 and 10 kg N/ha for sorghum and pearl millet after

three years (Aune and Bationo, 2008). Despite proved benefits results from these studies like

many nitrogen nutrition trials conducted to date in Africa and across the world, are highly

context bounded, therefore not generalizable (Sadras and Lemaire, 2014).

Despite the acknowledgement that site-specific fertilizer recommendations need to account for

site variability, little information is yet available on this aspect, as intra-agroecological zone

variability has been poorly studied. As presented in Table 2-1, fertilizer use and its relationship

with crop productivity in Africa has been widely studied. Factors hindering adoption and ways to

improve both fertilizer use and soil fertility management among SSA smallholder farmers, have

been widely theorized. However, there is still an inability to translate these findings into

pragmatic measures at farmer level.

Despite it being true that there has been considerable increase in the average use of fertilizers in

Africa it is also true that the rates are likely to vary across locations and farming environments,

within and across agro-ecologies. Therefore, optimizing inorganic fertilizer allocation in SSA

smallholder farming systems is as critical as access. This involves selecting the right fertilizer

allocation between crops and sites, choosing the right source of nutrients, the right timing of

application and the right amount – the 4R’s Principle to optimize fertilizer response (Ciampitti

and Vyn, 2014; Kaiser and Rubin, 2013; Ross R. Bender, 2013). Optimizing fertilizer allocations

and returns by advocating ‘best fit’ allocations of available fertilizer amounts in Africa is

important to improve nutrient productivity and yields in SSA diversified smallholders cropping

systems, where the main issue for farmers is finding the best ways to produce enough yield and

income with a reasonable amount of work and resources (Vanclay, 2004b) and resources.

2.1.3 Integrated Nutrient Management (INM)

In recent years, the concept of Integrated Nutrient Management (INM) practices have been

widely promoted in Africa (Bationo and Waswa, 2011; Mafongoya et al., 2006). Under INM,

locally tailored soil fertility management strategies focused in the combined use of inorganic

fertilizers with locally available amendments (manure, crop residues) and improved agronomical

Page | 43

practices such as, residue retention, legume integration and adequate soil and water conservation

measures. Integrated soil fertility management (ISFM) and conservation agriculture (CA) have

been the core of INM.

INM has been presented as a more farmer friendly technology since inorganic fertilizer use in the

practice can be kept at minimum and complemented by local amendments. However, like other

technologies promoted to date, INM is also knowledge intensive. Moreover, INM needs to be

tailored to meet contrasting agroecological and socioeconomic contexts. The fact that INM is

context bounded and knowledge intensive is among the factors that undermine its adoption.

Nevertheless, INM is an affordable and environmentally friendly practice with high potential to

improve soil fertility and raise crop yields in Africa where average maize yields are around 1.7

t/ha (1.2 t/ha in Mozambique), against 8.6 t/ha in developed countries (FAOSTAT, 2014).

Despite its potential to improve soil fertility and increase crop yield, INM benefits will not come

as easy as it is thought. INM needs to be part of an integrated farming systems designs where

optimizing resource allocation and responses for improved crop yield are the core of crop

husbandry. Here, alternative resource management strategies such as improved crop residue and

fertilizer allocation across crops and farming environments through GxExM analysis.

Nevertheless, a better understanding of residue-soil nutrient dynamics (Erenstein, 2003) within

existing cropping systems are yet to be tested.

Page | 44

Table 2-1 Main Lessons from Fertilizer Research and Promotion in SSA in Recent Years

Source Key message (Findings)

Ade Freeman and

Omiti (2003)

Understanding smallholder farmers (SHF) behaviour regarding fertilizer is key to fostering adoption

Bationo et al.,

(2012a)

Improving soil classification and availability of soil information can help improve fertilizer recommendations and nutrient

allocation

Bationo et al.

(2012b)

Tailoring fertilizer recommendations to the social and biophysical contexts of smallholder farmers are important if

fertilizer efficiency is to be achieved

Bationo and Waswa

(2011)

Crop productivity can be improved in Africa if widespread adoption of ISFM is achieved. However socio-economic and

biophysical factors still hinder ISFM adoption in Africa

Bekunda et al.

(2010)

Nutrient restoration in depleted SSA soils can be achieved through several cheap technologies, including ISFM and

biological nitrogen fixation (BNF). However, a proper policy environment is required

Chianu et al. (2012) Balanced fertilizer applications, the development of improved input-output markets, improved crop management and use

of nutrient budgets to demonstrate fertilizer profitability, is required to guarantee that SHF adopt fertilizers

Chianu and Tsujii

(2005)

Farmers decisions to adopt or not adopt the use of inorganic fertilizer is determined by their education level and knowledge

about fertilizers, and their perceptions of fertilizer impact on crop yield and profitability.

Enyong et al. (1999) Farmers’ implementation of soil fertility enhancing technologies (SFETs) is influenced by access to information and the

existence of a favourable policy environment.

Gilbert (2012) The key for tackling hunger in Africa is through enriching its soils. However, this cannot be achieved by just investing in

fertilizers. More locally tailored technologies, including the use of appropriate legumes for BNF are also important.

Mafongoya et al.

(2006)

ISFM through the combined use of inorganic fertilizers, legumes, animal manure and agroforestry, are the way to increase

and maintain soil fertility in SSA.

Page | 45

2.2 Fertilizer, Residues and Legumes interaction in SSA Conservation Agriculture

2.2.1 Crop residue contribution to yield, water and N-availability

In recent years, extensive research has been conducted to assess the complex interactions

between crop residues, fertilizers and tillage on yield, soil water availability and soil nutrient

turnover (Gangwar et al., 2006; Hoyle and Murphy, 2011; Kumar and Goh, 2002; Malhi and

Lemke, 2007; Mazzoncini et al., 2011; Sidhu and Beri, 1989). Nevertheless, contrasting results

have been found.

In India, results from a 9 year experiment designed to assess the impacts of incorporating 4 t/ha

of chopped wheat and rice residues on the yield of maize, wheat and rice crops grown at 0, 40,

80 and 120 kg N/ha (Sidhu and Beri, 1989), showed no significant effects on maize yield in the

first year. Nevertheless, significant yield changes were measured after the second year.

Significant increases in Total-N, 460 – 672 mg/kg were measured when compared to the initial

soil status. Data from the same trial also showed that, the incorporation of wheat residues

without fertilizer application in wheat plots, resulted in a significant yield decline in the first four

seasons compared to the fertilized plots. In this systems, no enough nitrogen was available to

satisfy the N-immobilization demands from the decomposition of incorporated residues.

Therefore, negative yields responses where measured. Negative returns from carbon rich

residues incorporation were also reported by Baudron et al. (2015a) with N-applications below

100 kg N/ha after 3 t/ha of maize stubble were applied to cotton.

The negative yield responses were also measured in unfertilized fields with high C/N residues

mulches (Cheshire et al., 1999; Turmel et al., 2015). These negative yield responses were

attributed to poor N-availability to plants due to N-immobilization. However, yield loss due to N

immobilization in residue applied fields can be minimized through increased high N-application.

Crop sequences and residue management were also found to significantly affect crop response to

residue application. Bakht et al. (2009), found that residue incorporation immediately after

harvest led to a 1.31 times increase in wheat yield. Legume-cereal sequencing (mung bean-wheat

rotations) led to the production of 2.09 and 2.16 times more wheat stover and grain yield when

compared to maize-wheat sequences. These results corroborate the findings of earlier studies

(Aggarwal et al., 1997; Gangwar et al., 2006; Samra et al., 2003; Sidhu and Beri, 1989; Wang et

Page | 46

al., 2010b), where positive impacts from high C/N residue application on yield, soil water

availability, soil biota and N-uptake were measured under adequate fertilizer-tillage-crop

sequencing management. It is important however, to take into consideration that most of these

studies were conducted in the humid tropics, and that the levels of N-fertilizer applied in the

residue-fertilizer interaction were rather high, i.e., between 80 to 300 kg N/ha in some studies.

This figures, differ from most of the environments where residue retention is promoted in Africa.

2.2.2 Yield penalties from residue retention in continuous cereal systems

Most positive responses to high C/N ratio residue use in conservation agriculture systems

reported to date comes from research dataset generated under considerable high inorganic N-

application (Kushwaha et al., 2000; Thierfelder et al., 2015; Vanlauwe et al., 2014; Wang et al.,

2011), which contrast SSA smallholder farming systems. Nevertheless, in high biomass

production systems like the corn-belt in the USA, a growing interest in understanding the trade-

offs between crop residue retention and removal have been reported in an attempt to minimize

yield penalties from residue incorporation in continuous maize systems (Luce et al., 2015;

Morrison, 2013).

An alternative to improve nutrient availability in low N systems managed under conservation

agriculture in Africa, is through the addition of N-rich legume residues (Giller et al., 2009). The

promotion of legume-cereal cropping sequences is another alternative (Luce et al., 2015).

Mazzoncini et al. (2011), found that the incorporation of N-rich legume residues grown as cover

crops, is able to contribute up to 430 kg/ha of extra soil organic carbon (SOC) and 20 kg N/ha a

year. Kumar and Goh (2002), measured significant improvements in yield and nitrogen uptake in

wheat grown after the incorporation of white clover and pea residues, compared to wheat grown

under wheat and ryegrass residues. These results strengthen the fact that responses from residue

application are highly dependent on residue type, amount, N-application levels, crop sequence

and the overall residue management strategy.

A recent simulation study based on data from the DayCent model, showed that up to 70% of

maize stover could be removed for biofuel production without significant losses in SOC (Liska et

al., 2014). Nevertheless, the 70% removal cup was contested (Robertson et al., 2014). More

recently, Wu et al. (2015) showed that 50% residue removal could be a reasonable cup for

Page | 47

maintaining SOC and soil fertility levels during the first years in these high biomass production

systems.

In the rain fed and N-deprived cropping systems of SSA managed under conservation

agriculture, smarter residue management strategies still need to be tested. Here, the emerging

question is on the best allocation of the widely available high C/N ratio residues, in both

livestock dominated and non-livestock dominated areas. Cross-site and crop-specific allocations

strategies which are key to optimize residue use have been little studied in Africa. Residue

allocation strategies and its interactive relation with crop type and inorganic fertilizer levels in

the SSA smallholder farming systems is an important area of attention, especially considering the

fact that crop residues can either help offset the nutrient cycling benefits in the system or

improve it if proper fertilizer compensation is made.

2.2.3 Competing claims on residue use: from mixed crop-livestock systems to non-cattle

dominated systems

Crop residues are an integral component of the conservation farming platform in Africa (Wall,

2007). If well managed, crop residue have the potential to improve soil microbial pool, thereby

influencing the nutrient cycling rate (Malhi and Lemke, 2007; Yadvinder et al., 2005). In the

semi-arid tropics where the soils are characteristically low in soil organic matter, crop residues

retention is crucial to compensate for physical biological degradation. Nevertheless, more has to

be done to identify under what circumstances positive and negative responses to residue

application are likely to occur.

In most mixed crop-livestock systems of SSA, residue use as feedstock is understood to limit

residue retention in the field (Erenstein, 2002). In this context, finding the best balance between

crop residue retention and removal for animal feedstock is critical to aid farmer’s decision

making process on best-fit residue use (Erenstein et al., 2015; Homann-Kee Tui et al., 2015).

Nevertheless, it also true that, in non-livestock based systems, residue retention is not an issue.

Here, analysing the potential trade-offs e.g., economic and environmental benefits associated

with the best-bet options for residue use – across contrasting crops, soil types, agro-environments

and rainfall season is crucial but stills lacking in Africa.

Page | 48

2.3 Legumes integration in SSA smallholder systems managed under CA

2.3.1 Legume performance in Africa

Almost 27 million hectares of land are planted to legumes in SSA every year (Tsedeke et al.,

2012), of which 11 million ha are concentrated in Western Africa, with 5 million being

distributed between Eastern and Southern Africa (Lupwayi, et al., 2011). Cowpea, Bambara

groundnut and common beans in particular, have received much attention due to their wide

adaptation in the diverse SSA farming environments (Dakora and Keya, 1997; Danso, 1984;

Lupwayi, 2011). Over recent decades, the need to improve productivity in Africa has led to

increased interest in the potential contribution of legume crops to the soil nitrogen pool (Asiwe,

2009; Jeranyama, 2007; Siame et al., 1998) through biological nitrogen fixation (BNF). Due to

this interest, an increasing range of grain legumes and green manure cover crops have attracted

the interest of SSA researchers and smallholder farmers. However, legume production potential

in African farming systems has not yet been fully realized (Sprent et al., 2010). This is in

contrast to the success of legume adoption in Latin America, Asia and Australia where legumes

are reported to perform well (Peoples et al., 2009) in terms of yield and nitrogen fixation.

Average yields between 0.45 and 0.6 t/ha for cowpea and common bean have been reported in

SSA (Chianu et al., 2011; Lupwayi, 2011; Tsedeke et al., 2012). Poor natural soil fertility,

characterized by low soil nitrogen and phosphorous content (Giller and Cadisch, 1995; Sanginga,

2003) in the dried semi-arid regions (Dakora and Keya, 1997) is undermining legume

performance. Drought in particular is of much importance. Drought and high temperature in the

semi-arid areas of SSA are threats to legume productivity, as they significantly reduce legumes

yields, biomass production and most of all, their N-fixation (Leport et al., 2006; Sangakkara et

al., 2002).

Despite the importance of legume-cereal rotations for the success of CA in SSA, most of the

legumes are grown as intercrops (Lupwayi, 2011), with a relatively small area being dedicated to

sole legume cropping. In West Africa. for example, approximately 90% of cowpea is grown as

an intercrop to sorghum and pearl millet. Despite the focus on intercropping, monoculture crops

have been proven to produce up to 38% higher yield and 35% more biomass compared to

intercrops (Dakora and Keya, 1997). Not less important is relay cropping that has been practiced

Page | 49

in the sub-humid and humid tropics of Africa (Awonaike et al., 1990). Here bimodal rainfall

patterns, e.g. in Guinean Savannah, offer the possibility of double cropping legumes. However,

little information is available on legume intensification systems in Africa.

Nonetheless, despite the on-going discussion over intercrops and sole cropping, increasing

research into the identification of adequate cropping sequences in legume-cereal cropping

systems has been gaining a momentum in Africa (Adjei-Nsiah, 2012; Bado et al., 2006; J. M.

Vesterager, 2007; Jeranyama, 2007; Sun et al., 2010). Maize yields of up to 7.0 t/ha was

measured in a pigeon pea-maize sequence, compared to 2.0 t/ha in a continuous maize systems in

Ghana (Adjei-Nsiah, 2012). Umeh and Mbah (2010) reported a significant yield increase in a

cassava crop harvested 8 months after soybean.

Legume-cereal and legume-tuber sequences are of much importance for SSA N-deprived

systems. In these systems the production and retention of N-rich residues provides a quick

alternative to improve N-availability. Therefore, considering the fact that most of the residual N

will be available for uptake by the second crop considerable investment needs to be made to

improve legume productivity (Derpesch, 2008). The current challenge in most SSA farming

environments is to establish which farming system design will provide farmers with better

legume yields and biomass production. Until now, intercrops and one cycle legume rotations

have not managed to produce enough N-rich biomass and/or fix the amounts of N achieved in

other parts of the world.

In West Africa, specifically, legume trees and shrubs have been alternatively used as N-sources

in legume-cereal and mixed crop-livestock systems. Alley cropping, planting hedgerows and

relay green cover crops using Acacia, Sesbania, Albizia, Leucaena and Gliricidia genus, have

been reported to fix between 36-581 kg N/ha year in trials conducted in Senegal, Nigeria and

Tanzania (Dakora and Keya, 1997; Gathumbi et al., 2002; Giller and Cadisch, 1995; Oyun,

2007). In these systems, leaf pruning from woody species and residues retained from relay green

cover crops have been successfully used as N-sources. Tree legumes have a double advantage,

especially in mixed crop-livestock systems where they can be used as N-sources and also as

stock feed, thereby reducing the pressure on field crop residues. Another important use of tree

legumes is for the control of erosion.

Page | 50

With regards to legume shrubs, they have been widely used as relay green manure cover crops,

which can later be incorporated as in-situ mulching. Crotalaria, Mucuna and Tephrosia have

been of great importance in these systems, due to their ability to fix nitrogen while providing

extra soil cover and weed suppressing effects (Odhiambo et al., 2010). However, N2-fixation

from green manure cover crops requires adequate in-field management, as their contribution to

N2-fixation is highly influenced by the cropping system and cultivation practices (Sanginga,

2003). Sanginga (2003) found that N2-fixation in Mucuna pruriens and Lablab purpureus

responded differently when used as live mulches when compared to in-situ mulching. Higher N-

fixation was achieved with in-situ mulching. In addition, Mucuna has been reported to fix

between 133 to 188 kg N/ha in N-fertilized and rhizobia inoculated plots.

Despite all the measured impacts from legume incorporation into SSA N-deprived farming

systems, a lot still need to be done, mainly in designing legume favourable cropping systems,

management and resource allocation strategies. This will help increase the potential contribution

from legumes (Table 2-2). However, independent of the achievements to date, it is agreed that,

(1) legumes in Africa are still an underestimated crop and have not yet achieved their full

production and nitrogen fixation potential. This is in part due to the continent’s focus on maize

which is the main dominant staple crop in Africa (Giller et al., 2011b); (2) improving legume

performance, especially in relation to biomass production in SSA smallholder cropping systems,

is of considerable importance if higher contributions to BNF are to be achieved; and (3)

improving management conditions and cropping systems design are key tools for improving

legume performance. Only good performing legumes will fix enough biological N and add

substantial N-rich residues into the systems.

Page | 51

Table 2-2 Current findings on legume contributions to nitrogen fixation in African cropping systems

Citation

Agro-ecological

Condition

Legume

used

N2-Fixation

(Kg/ha year)

Systems Comments

Adjei-Nsiah et al.

(2008)

Forest/savannah

transitional zone

(1271 mm)

Cowpea 32-44 Farmers evaluation of productivity

in cowpea-maize crop sequences

Indeterminate varieties produced 2.6

t/ha stover while determinate varieties

produced, on average, 1.2-1.3 t/ha;

Adjei-Nsiah

(2012)

Semi-deciduous

forest zone

(1433 mm)

Cowpea,

Pigeon pea

Groundnuts

64-160

N-return and yield in legume-

maize-cassava cropping sequences

Total N-returned to the system through

residues after harvest in one season in a

crop sequence that included maize and

cassava.

Yusuf et al. (2009) Guineas Savannah

(1038 mm)

Cowpea and

Soybean

61-83 Rotational effect of legumes and N-

fertilization on maize yield and

nitrogen fixation

Maize after legume yielded better than

continuous maize. High N-uptake in the

soybean-maize sequence was measured.

N agronomic efficiency (AEN) was

affected by N rate and rotation

Somado et al.

(2006)

Humid forest

(green house)

Crotalaria

micans

29-116 P-response effect on yield and N-

fixation

Legume N2-fixation increased fourfold

compared with the unfertilized legume.

Legumes produced less than 1- ton

biomass without P application

Siame et al. (1998) Kasama, Zambia

1286.5mm

Common

beans

- Intercropping and N-response At higher N-levels increased vegetative

growth of the legume decreased maize

yield by 30%, with some competition

noticed at lower N-levels

Jeranyama (2007) Sub-humid

(800-900mm)

Groundnuts 16-40 Rotation effect Maize yields increased by 0.7 t/ha

compared to continuous maize

Sanginga (2003) Northern Guinea

Savannah

Soybean 38-126 BNF in legume based cropping

systems

Soil accumulated N ranged between 8-

47 kg/ha, depending on the soybean

cultivar, with 10-24 kg/ha residual N in

the soybean-maize rotation

Page | 52

2.3.2 Potential contribution from legume crops in improving N-fixation

On a global scale, it has been estimated that chemical fertilizers only account for 25% of

total-N2 in the soil, with atmospheric biological nitrogen fixation (BNF) from legumes

contributing with almost 60% (Zahran, 1999). This represents a total annual input of

approximately 139 to 170 million tons of N2-biologically fixed, of which 25-30% (34-45

million tons of N) being in cropped land and 30% (45 million tons of N) being in permanent

grazing land (Peoples et al., 1995). Therefore, enhancing N-fixation is important, especially

for fertilizer constrained systems in the developing world, where soil quality depletion due to

continuous nutrient mining and poor crop management practices, are common. As in the

developing world smallholder farmers are less likely to invest in inorganic fertilizers for

staple food crop production (Serraj, 2005), cropping systems that advocate proper legume

incorporation would provide a long term alternative for improving soil fertility and yields.

However, for legumes to realize their nitrogen fixing potential, proper cropping

arrangements, sequences and varieties, need to be in place.

In recent decades, nitrogen fixation through symbiotic association between leguminous plants

and rhizobia has been widely studied (Di Ciocco et al., 2008; Giller and Cadisch, 1995;

Herridge et al., 2008; Peoples et al., 2009; Salvagiotti et al., 2008). Determining the

partitioning between plant parts of nitrogen fixed from the atmosphere (Ndfa) through 15N-

labelling and natural abundance methods (Peoples et al., 2009), the identification of the best

association between rhizobia species and legume crops (Awonaike et al., 1990), and the

performance of inoculants (Laditi et al., 2012), are just some of the key areas of progress in

legume research that can be used for the benefit of N-deficient systems. However, most of the

studies conducted to date have been in highly productive environments in the humid tropics

and temperate regions. Less information is available on BNF, especially in the SSA cropping

systems where underperforming legumes are grown in multi-nutrient depleted conditions,

e.g., N. P and K have yet to deliver the advantages shown in other regions, e.g., Latin

America, Asia and Australia.

Despite the success experienced in Latin America, Asia and Australia, BNF does not provide

straightforward benefits. Negative and positive N-balances have been reported in legume-

cereal cropping systems. N-balances have ranged from -132 to + 135 kg N/ha in soybean,

groundnut, chickpea, lentil, pea and lupin crops (Peoples et al., 1995). Several factors have

Page | 53

been cited as contributing to the negative N balance. The fact that most of the N2-biologically

fixed by legumes is stored in the seeds which are primary exports from the systems, is one of

the factors. Di Ciocco et al., (2002), found that the negative N-balances have been mainly

associated with high yielding legume varieties, e.g., soybean with high harvest index. With

such varieties, the N-harvested from seed is higher than the N2-fixed in vegetative parts,

leaving less N to be returned to the soil. In addition, the removal of crop residues from the

system (Peoples et al., 1995; Stern, 1993) is also a contributing factor for the negative N

balance. Adding to the discussion, Austin et al., (2006) argued that the highly fertile soils and

high input cropping systems where most of the legumes are grown in the humid tropics, were

acting against nitrogen fixation by legumes. This is because the high fertility of the soils and

the use of fertilisers, has a negative impact on the nodulation capacity of legumes (Austin et

al., 2006), thereby reducing N2-fixation.

The relationship between nitrogen fertilization and BNF is an important point of attention

from a management perspective, especially in SSA where most of the soils are known to be

N-deficient and require a certain level of nitrogen application to stimulate BNF. However,

rather than soil starting nitrogen levels, there are other potential constraints to nitrogen

fixation. In Africa, for instance, crop varieties and cropping systems are among the most

important factors affecting legume nodulation and N fixation capacity.

The selection of appropriate legume species, varieties and management options is also

important. Growing environments also play a major issue on BNF. In high rainfall and

bimodal environments for example, legume intensification systems and better resource

allocation strategies between cereals and legumes, can help improve N-fixation and

availability in the system. The same would be less likely to happen in unimodal regions

except if short cycled varieties are used to allow double sowing. Nevertheless, the complexity

of SSA smallholder cropping systems and the poor investment in agricultural research and

development makes the discussion about legume intensification systems in unimodal rainfall

regions such as central Mozambique rather difficult and mundane. Therefore, investing in

new approaches to test and convey relevant agricultural information to improve decision

making processes at farm and regional level is crucial in Africa. In that quest, agricultural

simulation models, can be an important part of the solution in SSA.

Page | 54

2.4 Using agricultural simulation models as decision support tools in SSA

In recent decades, with the advent of farming systems research, (Bouma and Jones, 2001;

Davidson, 1987; Jordan et al., 1997; Maxwell, 1986), the modelling of agronomic processes

has played an important role in helping researchers generate new virtual information that is

allowing them to integrate and understand the interactions between crops, soils, climate

processes and management practices. According to Mathews and Stephens (2002), the

integration of simulation modelling in research programs has helped to: (1) identify gaps in

existing knowledge; (2) generate and test hypotheses, and design of experiments; (3) select

the most influential parameters in a system (Seligman, 1990); and (4) bring researchers,

experiments and farmers together to solve problems (Carberry et al., 2009; Whitbread et al.,

2010).

In developed countries where simulation models are widely used, they have helped improve

the understanding of crop eco-physiological responses and allowed researchers to answer

basic questions, such as identifying the most suitable and remunerative cropping systems and

patterns for different climatic and socioeconomic conditions (Cox et al., 2010; Meinke et al.,

2001; Rodriguez et al., 2011). The use of crop and biophysical models capable of generating

ex-ante information through the simulation of real world responses to factors such as the

climate, soil and management (Bouma and Jones, 2001), has allowed researchers to provide

farmers with timely and relevant advice that has helped shape local farming practices. This

advice has significantly improved farmers’ ability to manage risk.

Despite the attention that agricultural simulation models experienced in developed countries,

they had been poorly integrated in SSA agricultural research. Several factors contribute to

this. The reliance on reductionist processes, such as the conventional agricultural research

model that views fields, plants and animals as separate and non-related factors within a farm

unit or the household (Ikerd, 1993); the lack of training and technical capacity to gather

quality data to test and validate crop models (Bouma and Jones, 2001); and above all, the lack

of faith in models’ predictive capacity (Mathews and Stephens, 2002), have hindered the use

of models in SSA. However, in SSA where expenditure on agricultural research and

development is low (Beintema and Stads, 2008), the use of simulation models that emphasize

process based research (Matthews and Stephens, 2002a) could potentially help increase the

efficiency of local research programs. Models would help define attainable research priorities

Page | 55

and suitable technologies to be tested with farmers, which would help make research more

effective. However, the selection and use of appropriate models is also challenging and

would depend on the questions needed to be answered.

For the specific case of SSA, models, need to be able to answer the practical problems faced

by farmers. In order to achieve this, models must be able to capture the complex routines of

the decision making process at farm level, while avoiding the mismatch between model

enquiry and end user needs (Matthews and Stephens, 2002a), a common burden with some

models. The collection and availability of quality data, e.g., soil and climate data, for initial

parameterization are key issues that need to be addressed in Africa. However, once these

issues are resolved, the time and resources invested in conventional research programs can be

improved, as models can be parameterized to a vast range of farming environments and

systems from where ex ante simulation can be developed in order to assist local research

initiatives.

The fact that current models are based on a sustainable systems research platform rooted in a

holistic approach which views the different components of a farming system as complex

interrelated sub-organisms, with distinct biophysical and socioeconomic boundaries

(Bawden, 1991; Ikerd, 1993) is an advantage. The ability of models such as the Agricultural

Production Systems Simulator (APSIM) (Holzworth et al., 2014; Keating et al., 2003;

McCown et al., 1996) to predict potential environmental and economic impacts of

management decisions on the systems short and long-term responses is a major potential

benefit for agricultural research. Therefore, in the following section, APSIM capabilities as a

decision support tool will be discussed, with specific interest in its potential to simulate the

nitrogen and crop residue interactions in N-deficient cropping CA systems of SSA.

2.4.1 Using the Agricultural Production Systems Simulator (APSIM)

In recent years, APSIM has been used in studies covering economic, biophysical and

management interfaces of the model. In these studies, APSIM has been successfully used in

simulating a wide range of interactions. Among them, tillage responses, soil water balances,

nutrient and residue responses have received much attention (Table 2-3). With the growing

interest in conservation agriculture and integrated soil fertility management, which have been

largely promoted as the flagship for improving soil and crop productivity in Africa, several

technologies, including crop residues, conservation tillage, fertilizers, farmyard manure and

Page | 56

legumes, have been increasingly promoted to smallholder farmers (Chianu et al., 2011;

Giller et al., 2011a; Machido, 2011). However, information on the adaptability, best-fit

resource allocation and response of these practices, is still scarce and not yet tailored to the

diverse farming environments of Africa. Under these circumstances, modelling can be of

great use in: (1) synthesising current knowledge; (2) exploring alternative agronomic

practices; (3) measuring the effects of management options; and (4) bringing theory and

practice together (Stern, 1993). Therefore, calibrating and validating models such as APSIM,

have the potential to provide new information, especially in SSA.

APSIM’s major advantage is its ability to simulate biophysical processes in systems where

there is also interest in the economic and ecological impacts of management practices

(Keating et al., 2003). APSIM also includes most of the major tropical crops such as maize,

cotton and legumes e.g., cowpea, mucuna, pigeon pea, soybean and peanut (Keating et al.,

2003), which are of great importance in low input farming systems. This makes APSIM an

appealing analytical and decision support tool for application in SSA. In such places APSIM

could help generate timely information to support experimental design and discussions with

farmers and agribusinesses in participatory modelling research activities.

Table 2-3 The potential of APSIM in aiding agricultural research in SSA

Author Location Modelled parameters

Power et al. (2011) Australia Resource allocation in a farm business

Rodriguez et al. (2011) Australia Whole-Farm business design and analysis

Sadras and Rodriguez

(2010)

Australia Trade-off in Nitrogen and Water Use Efficiency

Balwinder et al. (2011) India Crop residue management and water use in wheat

crops

Mupangwa et al. (2011) Zimbabwe Tillage and residue effects on yield

MacCarthy et al. (2009) Ghana Nutrient and residue management

Lisson and Cotching

(2011)

Australia Fate of water and nitrogen in multiple cropping

systems

Ncube et al. (2009) Zimbabwe Residual effect of grain legumes,

Cox et al. (2010) Australia Profitability of cropping patterns,

Mupangwa and Jewitt

(2011)

Zimbabwe No-tillage effect on water fluxes and crop

productivity

Akponikpè et al. (2010) Niger Long term simulation of the effect of nitrogen

management

Page | 57

2.4.2 Gaps in modelling crop residue decomposition and N-cycling with APSIM

Residue decomposition and N-mineralization have been widely simulated with APSIM in

varying cropping environments, from drought affected semi-arid to humid tropics (Balwinder

et al., 2011; Gaydon et al., 2012; Mohanty et al., 2012; Mupangwa et al., 2007; Nascimento

et al., 2011). A mixture of results has been observed. APSIM simulated the decomposition of

Stizolobium aterrimum and Calopogonium mucunoides with a 37.6% and 48.2% error,

respectively, compared with the observed data in the absence of calibration, highlighting that

the reason for this error is the model’s inability to accurately capture initial residue

decomposition in the first 30 days. Separate studies have successfully predicted crop

responses to residue application (Balwinder et al., 2011; Gaydon et al., 2012) after assuming

a constant moisture factor of 0.5. The APSIM-residue module uses a combination of

algorithms where residue decomposition is a function of moisture, temperature and residue

C/N ratio (Probert et al., 1998), with adjustments made to the latter(Thorburn et al., 2001).

The algorithms have been basically developed to represent residue decomposition in large

residue mass per unit area, i.e., higher than 10 t/ha. A later modification was made for residue

amounts below 5-7 t/ha (Thorburn et al., 2001), which is a closer representation of SSA

smallholder farming systems where less residue is retained.

A positive aspect of the modified algorithms for residue decomposition is that they were

based on field experiments different from the mineralization studies. However, despite the

representativeness that the modification brings, further adjustments need to be made in the

algorithm of the RESIDUE module, to take account of the diversity of residues incorporated

through tillage in most conditions where the model is used. For some areas of Mozambique

and other semi-arid regions for instance, there is a need to make a distinction between termite

infested and non-infested soils, as the residue break-down and decomposition rate will differ

between these environments. However, while there is scope within the APSIM framework to

deal with termite loss, until now, this aspect has not yet been developed and tested.

As per N-dynamics, the process of N-mineralization and immobilization in the soil is the

main driver of nitrogen supply in cropping lands, thus the centre of research on N-dynamics

in agricultural land. In recent years, apart from model simulations such as in APSIM, N-

mineralization and immobilization have been studied through several incubation experiments

across the world (Coppens et al., 2006; Mary, 1996; Mohanty et al., 2010; Ros, 2012).

Page | 58

Initially, there was common agreement that the rate of N-mineralization and immobilisation

during crop residue decomposition depends largely on the type of residue applied and on the

existence of adequate soil moisture and favourable soil temperature to start the process. As a

result, several incubation studies have been conducted to analyse the decomposition rate of

different crop residue types under different management conditions and environments.

Despite several mineralization studies being conducted in soils with high amounts of mineral

N, it was possible to demonstrate that in N-constrained soils, the decomposition of soil

organic matter was not totally stopped but decreased; therefore, the mineralization intensity

was reduced and N-mineralization was delayed (Mary, 1996). The findings from incubation

trials for contrasting soil N-levels, (i.e., N-rich and N-poor soils), have proven that the

immobilization and mineralization process can actually be controlled through a feedback

effect by managing the soil N availability, plus the amount and type of residue applied,

thereby providing options for an improvement in the design of better crop residue

management strategies. However, there still remain questions to be answered regarding

mineralisation-immobilization turnovers in soils, especially regarding the role of the different

soil organic matter pools involved in N-cycling in the soil, and to what extent the laboratory

based mineralization results can be extrapolated to field conditions (Mary, 1996).

In a recent review that re-analysed 59 published datasets on the predictions of N-

mineralization, it was found that the amount of N mineralized from applied surface residues

was primarily linked to the size of the extractable organic matter (EOM) fraction, whereas

variables reflecting soil texture and organic matter quality were less important (Ros, 2012).

These findings add to the discussion that more mechanistic research is needed to explore the

biochemical basis of the relationship between total organic matter, EOM and N-

mineralization, rather than focusing on the C/N ratio of the material.

While using multivariate statistical modelling to analyse the role of organic matter and soil

proprieties in determining the variations of mineralizable N in 98 agricultural soils under

different management, it was found that the influence of variables reflecting soil texture or

organic matter quality (e.g., C/N ratio and extractable C and N), had little influence on the

amount of mineralizable N, compared to the size of the total organic matter pool. This means

that OM quality variables such as C:N ratio actually have less potential to improve the

accuracy of mineralizable N predictions (Ros et al., 2011). Another issue raised in this study

was the relevance of C-P and N-P ratios in explaining the magnitude of mineralizable N

Page | 59

variation, especially in N-constrained soils. In the experiment, these two ratios significantly

helped explain the variations in mineralizable N; however, the soil to which the study related

was not P-deficient, opening the question of whether these processes should be also studied

on P-deficient soils.

Based on these findings it can be concluded that key areas of research in terms of the

assumptions in the calculation of N and residue dynamics in the APSIM model are: (1)

modelling the contribution of the different organic matter and nitrogen pools to N-

mineralization and immobilization under different residue and nitrogen application levels,

especially in N-deficient soils; (2) to understand how management, mainly tillage, affects the

availability of the different organic matter and nitrogen pools, and its linkage with the

amounts of N-mineralization and immobilization under different residue-fertilizer

combinations in N-deficient soils; and (3) measuring N-mineralization and immobilization

during fallow is an area for exploration, as most of the residues from each season, is, in some

systems, incorporated right after crop harvest and left to decompose during the fallow period.

References

Ade Freeman, H., and Omiti, J. (2003). Fertilizer use in semi-arid areas of Kenya: Analysis of

smallholder farmers' adoption behaviour under liberalized markets. Nutrient Cycling

in Agroecosystems 66, 23-31.

Adjei-Nsiah, S. (2012). Evaluating sustainable cropping sequences with cassava and three

grain legume crops: Effects on soil fertility and maize yields in the semi-deciduous

forest zone of Ghana. Journal of Soil Science and Environmental Management 3, 49-

55.

Adjei-Nsiah, S., Kuyper, T., Leeuwis, C., Abekoe, M., Cobbinah, J., Sakyi-Dawson, O., and

Giller, K. (2008). Farmers’ agronomic and social evaluation of productivity, yield and

N-fixation in different cowpea varieties and their subsequent residual N effects on a

succeeding maize crop. Nutrient Cycling in Agroecosystems 80, 199-209.

Aggarwal, R. K., Praveen, K., and Power, J. F. (1997). Use of crop residue and manure to

conserve water and enhance nutrient availability and pearl millet yields in an arid

tropical region. Soil and Tillage Research 41, 43-51.

Akponikpè, P. B. I., Gérard, B., Michels, K., and Bielders, C. (2010). Use of the APSIM

model in long term simulation to support decision making regarding nitrogen

Page | 60

management for pearl millet in the Sahel. European Journal of Agronomy 32, 144-

154.

Alene, A. D., and Coulibaly, O. (2009). The impact of agricultural research on productivity

and poverty in sub-Saharan Africa. Food Policy 34, 198-209.

Asiwe, J. A. N. (2009). Needs assessment of cowpea production practices, constraints and

utilization in South Africa. African Journal of Biotechnology 8, 5383-5388

Aune, J. B., and Bationo, A. (2008). Agricultural intensification in the Sahel – The ladder

approach. Agricultural Systems 98, 119-125.

Austin, A. T., Piñeiro, G., and Gonzalez-Polo, M. (2006). More is less: agricultural impacts

on the N cycle in Argentina. Biogeochemistry 79, 45-60.

Awonaike, K. O., Kumarasinghe, K. S., and Danso, S. K. A. (1990). Nitrogen fixation and

yield of cowpea (Vigna unguiculata) as influenced by cultivar and Bradyrhizobium

strain. Field Crops Research 24, 163-171.

Bado, B., Bationo, A., and Cescas, M. (2006). Assessment of cowpea and groundnut

contributions to soil fertility and succeeding sorghum yields in the Guinean savannah

zone of Burkina Faso (West Africa). Biology and Fertility of Soils 43, 171-176.

Bakht, J., Shafi, M., Jan, M. T., and Shah, Z. (2009). Influence of crop residue management,

cropping system and N fertilizer on soil N and C dynamics and sustainable wheat

(Triticum aestivum L.) production. Soil and Tillage Research 104, 233-240.

Balwinder, S., Gaydon, D. S., Humphreys, E., and Eberbach, P. L. (2011). The effects of

mulch and irrigation management on wheat in Punjab, India—Evaluation of the

APSIM model. Field Crops Research 124, 1-13.

Bationo, A., Hartemink, A., Lungu, O., Naimi, M., Okoth, P., Smaling, E., Thiombiano, L.,

and Waswa, B. (2012a). Knowing the African Soils to Improve Fertilizer

Recommendations. In "Improving Soil Fertility Recommendations in Africa using the

Decision Support System for Agrotechnology Transfer (DSSAT)." (J. Kihara, D.

Fatondji, J. W. Jones, G. Hoogenboom, R. Tabo and A. Bationo, eds.), pp. 19-42.

Springer Science and Business Media, Netherlands.

Bationo, A., Kihara, J., and Adesina, A. (2012b). Beyond Biophysical Recommendations:

Towards a New Paradigm. In "Improving Soil Fertility Recommendations in Africa

using the Decision Support System for Agrotechnology Transfer (DSSAT)." (J.

Kihara, D. Fatondji, J. W. Jones, G. Hoogenboom, R. Tabo and A. Bationo, eds.), pp.

169-184. Springer Science and Business Media, Netherlands.

Page | 61

Bationo, A., and Waswa, B. S. (2011). New Challenges and Opportunities for Integrated Soil

Fertility Management in Africa. In "Innovations as Key to the Green Revolution in

Africa" (A. Bationo, B. Waswa, J. M. M. Okeyo, F. Maina and J. M. Kihara, eds.), pp.

3-17. Springer Science and Business Media, Netherlands.

Baudron, F., Delmotte, S., Corbeels, M., Herrera, J. M., and Tittonell, P. (2015). Multi-scale

trade-off analysis of cereal residue use for livestock feeding vs. soil mulching in the

Mid-Zambezi Valley, Zimbabwe. Agricultural Systems 134, 97-106.

Bawden, R. J. (1991). Systems Thinking and Practice in Agriculture. Journal of Dairy

Science 74, 2362-2373.

Beintema, N. M., and G. J. Stads. (2008). Agricultural R&D in Sub-Saharan Africa: An era of

stagnation. ASTI Background Paper. International Food Policy Research Institute

Washington, USA

Bekunda, M., Sanginga, N., and Woomer, P. L. (2010). Chapter Four - Restoring Soil

Fertility in Sub-Sahara Africa. In "Advances in Agronomy" (L. S. Donald, ed.), Vol.

Volume 108, pp. 183-236. Academic Press, USA.

Bouma, J., and Jones, J. W. (2001). An international collaborative network for agricultural

systems applications (ICASA). Agricultural Systems 70, 355-368.

Brouder, S. M., and Gomez-Macpherson, H. (2014). The impact of conservation agriculture

on smallholder agricultural yields: A scoping review of the evidence. Agriculture,

Ecosystems & Environment 187, 11-32.

Carberry, P. S., Hochman, Z., Hunt, J. R., Dalgliesh, N. P., McCown, R. L., Whish, J. P. M.,

Robertson, M. J., Foale, M. A., Poulton, P. L., and van Rees, H. (2009). Re-inventing

model-based decision support with Australian dryland farmers. 3. Relevance of

APSIM to commercial crops. Crop and Pasture Science 60, 1044-1056.

Cheshire, M. V., Bedrock, C. N., Williams, B. L., Chapman, S. J., Solntseva, I., and

Thomsen, I. (1999). The immobilization of nitrogen by straw decomposing in soil.

European Journal of Soil Science 50, 329-341.

Chianu, J., Chianu, J., and Mairura, F. (2012). Mineral fertilizers in the farming systems of

sub-Saharan Africa. A review. Agronomy for Sustainable Development 32, 545-566.

Chianu, J., Nkonya, E., Mairura, F., Chianu, J., and Akinnifesi, F. (2011). Biological nitrogen

fixation and socioeconomic factors for legume production in sub-Saharan Africa: a

review. Agronomy for Sustainable Development 31, 139-154.

Page | 62

Chianu, J. N., and Tsujii, H. (2005). Determinants of farmers' decision to adopt or not adopt

inorganic fertilizer in the savannas of northern Nigeria. Nutrient Cycling in

Agroecosystems 70, 293-301.

Ciampitti, I. A., and Vyn, T. J. (2014). Nutrient Sufficiency Concepts for Modern Corn

Hybrids: Impacts of Management Practices and Yield Levels. Crop Management 47,

38-44.

Coppens, F., Garnier, P., De Gryze, S., Merckx, R., and Recous, S. (2006). Soil moisture,

carbon and nitrogen dynamics following incorporation and surface application of

labelled crop residues in soil columns. European Journal of Soil Science 57, 894-905.

Cox, H. W., Kelly, R. M., and Strong, W. M. (2010). Pulse crops in rotation with cereals can

be a profitable alternative to nitrogen fertiliser in central Queensland. Crop & Pasture

Science 61, 752-762.

Crawford, E. W., Jayne, T. S., and Kelly, V. A. (2006). Alternative Approaches for

Promoting Fertilizer Use in Africa. pp. 88. World Bank, Washington.

D'Emden, F. H., Llewellyn, R. S., and Burton, M. P. (2006). Adoption of conservation tillage

in Australian cropping regions - An application of duration analysis. Technological

Forecasting and Social Change 73, 630-47.

Dakora, F. D., and Keya, S. O. (1997). Contribution of legume nitrogen fixation to

sustainable agriculture in Sub-Saharan Africa. Soil Biology and Biochemistry 29, 809-

817.

Danso, S. K. and Eskew D.L (1984). Enhancing biological nitrogen fixation. IAEA Bulletin

Vol 26 Nr. 2 p.1-6.

Dalgliesh, N., and Foale, M. (1998). "Soil matters: Monitoring soil water and nutrients in

dryland farming," 1/Ed. CSIRO, Australia, Toowoomba, Australia.

Davidson, A. P. (1987). Does farming systems research have a future? Agricultural

Administration and Extension 24, 69-77.

Derpesch, R. (2008). Critical Steps to No-Till Adoption. In "No-till Farming Systems" (T.

Goddard and M. A. Zoebisch, Gan, Y., Ellis, W., Watson, A. and Sombatpanit, S.,

eds.), World Association of Soil and Water Conservation. pp. 479 – 95.

Di Ciocco, C., Coviella, C., Penon, E., Diaz-Zorita, M., and Lopez, S. (2008). Short

communication. Biological fixation of nitrogen and N balance in soybean crops in the

pampas region. Spanish Journal of Agricultural Research 6, 114-119.

Page | 63

Enyong, L. A., Debrah, S. K., and Bationo, A. (1999). Farmers' perceptions and attitudes

towards introduced soil-fertility enhancing technologies in western Africa. Nutrient

Cycling in Agroecosystems 53, 177-187.

Erenstein, O. (2002). Crop residue mulching in tropical and semi-tropical countries: An

evaluation of residue availability and other technological implications. Soil and

Tillage Research 67, 115-133.

Erenstein, O. (2003a). Smallholder conservation farming in the tropics and sub-tropics: a

guide to the development and dissemination of mulching with crop residues and cover

crops. Agriculture, Ecosystems and Environment 100, 17-37.

Erenstein, O. (2011). Cropping systems and crop residue management in the Trans-Gangetic

Plains: Issues and challenges for conservation agriculture from village surveys.

Agricultural Systems 104, 54-62.

Erenstein, O., Gérard, B., and Tittonell, P. (2015). Biomass use trade-offs in cereal cropping

systems in the developing world: Overview. Agricultural Systems 134, 1-5.

FAO (2008). Current world fertilizer trends and outlook to 2011-12. pp. 57. FAO, Rome,

Italy.

FAO (2010). Current world fertilizer trend and outlook to 2014. The Fertilizer Organization

Working Group, FAO. Italy, Rome.

Fischer, R. A., Santiveri, F., and Vidal, I. R. (2002). Crop rotation, tillage and crop residue

management for wheat and maize in the sub-humid tropical highlands: II. Maize and

system performance. Field Crops Research 79, 123-137.

Gangwar, K. S., Singh, K. K., Sharma, S. K., and Tomar, O. K. (2006). Alternative tillage

and crop residue management in wheat after rice in sandy loam soils of Indo-Gangetic

plains. Soil and Tillage Research 88, 242-252.

Gathumbi, S. M., Cadisch, G., and Giller, K. E. (2002). 15N natural abundance as a tool for

assessing N2-fixation of herbaceous, shrub and tree legumes in improved fallows. Soil

Biology and Biochemistry 34, 1059-1071.

Gaydon, D. S., Probert, M. E., Buresh, R. J., Meinke, H., Suriadi, A., Dobermann, A.,

Bouman, B., and Timsina, J. (2012). Rice in cropping systems—Modelling transitions

between flooded and non-flooded soil environments. European Journal of Agronomy

39, 9-24.

Gilbert, N. (2012). Dirt Poor: The key to tackle hunger in Africa is enriching its Soils. Nature

483, 525-227.

Page | 64

Giller, K. E., and Cadisch, G. (1995). Future benefits from biological nitrogen fixation: An

ecological approach to agriculture. Plant and Soil 174, 255-277.

Giller, K. E., Corbeels, M., Nyamangara, J., Triomphe, B., Affholder, F., Scopel, E., and

Tittonell, P. (2011). A research agenda to explore the role of conservation agriculture

in African smallholder farming systems. Field Crops Research 124, 468-472.

Giller, K. E., Rowe, E. C., de Ridder, N., and van Keulen, H. (2006). Resource use dynamics

and interactions in the tropics: Scaling up in space and time. Agricultural Systems 88,

8-27.

Gregory, D. I., and Bumb, B. L. (2006). Factors Affecting Supply of Fertilizer in Sub-

Saharan Africa. In "Agriculture and Rural Development Discussion Papers", Vol. 24,

pp. 81. The World Bank, Washington, USA.

Hauggaard-Nielsen, H., Ambus, P., and Jensen, E. S. (2003). The comparison of nitrogen

uses and leaching in sole cropped versus intercropped pea and barley. Nutrient

Cycling in Agroecosystems 65, 289-300.

Heffer, P., and Prud'homme, M. (2011). "Fertilizer Outlook 2011-2015." IFA, Paris, France.

Herridge, D. F., Peoples, M. B., and Boddey, R. M. (2008). Global inputs of biological

nitrogen fixation in agricultural systems. Plant and Soil 311, 1-18.

Holzworth, D. P., Huth, N. I., deVoil, P. G., Zurcher, E. J., Herrmann, N. I., McLean, G.,

Chenu, K., van Oosterom, E. J., Snow, V., Murphy, C., Moore, A. D., Brown, H.,

Whish, J. P. M., Verrall, S., Fainges, J., Bell, L. W., Peake, A. S., Poulton, P. L.,

Hochman, Z., Thorburn, P. J., Gaydon, D. S., Dalgliesh, N. P., Rodriguez, D., Cox,

H., Chapman, S., Doherty, A., Teixeira, E., Sharp, J., Cichota, R., Vogeler, I., Li, F.

Y., Wang, E., Hammer, G. L., Robertson, M. J., Dimes, J. P., Whitbread, A. M., Hunt,

J., van Rees, H., McClelland, T., Carberry, P. S., Hargreaves, J. N. G., MacLeod, N.,

McDonald, C., Harsdorf, J., Wedgwood, S., and Keating, B. A. (2014). APSIM –

Evolution towards a new generation of agricultural systems simulation.

Environmental Modelling and Software 62, 327-350.

Homann-Kee Tui, S., Valbuena, D., Masikati, P., Descheemaeker, K., Nyamangara, J.,

Claessens, L., Erenstein, O., van Rooyen, A., and Nkomboni, D. (2015). Economic

trade-offs of biomass use in crop-livestock systems: Exploring more sustainable

options in semi-arid Zimbabwe. Agricultural Systems 134, 48-60.

Page | 65

Hoyle, F. C., and Murphy, D. V. (2011). Influence of organic residues and soil incorporation

on temporal measures of microbial biomass and plant available nitrogen. Plant and

Soil 347, 53-64.

IFDC (2012). "IFDC report: An update on the work and progress of IFDC." IFDC, Alabama.

USA

Ikerd, J. E. (1993). The need for a system approach to sustainable agriculture. Agriculture,

Ecosystems and Environment 46, 147-160.

Vesterager, J. M., Nielsen N. E. and Høgh-Jensen, H (2007). Nitrogen budgets in crop

sequences with or without phosphorus-fertilised cowpea in the maize-based cropping

systems of semi-arid eastern Africa. African Journal of Agricultural Research 2, 261-

268.

Jayne, T. S., Mather, D., Mason, N., and Ricker-Gilbert, J. (2013). How do fertilizer subsidy

programs affect total fertilizer use in sub-Saharan Africa? Crowding out, diversion,

and benefit/cost assessments. Agricultural Economics 44, 687-703.

Jeranyama, P., Waddington, S. R., Hesterman, O. B. and Harwood, R. R. (2007). Nitrogen

effects on maize yield following groundnut in rotation on smallholder farms in sub-

humid Zimbabwe. African Journal of Biotechnology 6, 1503-1508

Jordan, V. W. L., Hutcheon, J. A., Donaldson, G. V., and Farmer, D. P. (1997). Research into

and development of integrated farming systems for less-intensive arable crop

production: experimental progress (1989–1994) and commercial implementation.

Agriculture, Ecosystems and Environment 64, 141-148.

Kaiser, D. E., and Rubin, J. C. (2013). Corn Nutrient Uptake as Affected by In-Furrow Starter

Fertilizer for Three Soils. Agronomy Journal 105, 1199-1210.

Keating, B. A., P.S. Carberry, G.L. Hammer, M.E. Probert, M.J. Robertson, D. H., N.I. Huth,

J.N.G. Hargreaves, H. Meinke, Z. H., G. McLean, K. Verburg, V. Snow, J.P. Dimes,

M. S., E. Wang, S. Brown, K.L. Bristow, and S. Asseng, S. C., R.L. McCown, D.M.

Freebairn, C.J. Smith (2003). An overview of APSIM, a model designed for farming

systems simulation. European Journal of Agronomy, 267-288.

Kelly, V. A. (2006). Factors Affecting Demand for Fertilizer in Sub-Saharan Africa. In

"Agricultural and Rural Development Discussion Papers", pp. 89. The World Bank,

Washington, USA.

Kintché, K., Guibert, H., Sogbedji, J. M., Levêque, J., Bonfoh, B., and Tittonell, P. Long-

term mineral fertiliser use and maize residue incorporation do not compensate for

Page | 66

carbon and nutrient losses from a Ferralsol under continuous maize–cotton cropping.

Field Crops Research. 184, 192-200

Kumar, K., and Goh, K. M. (2002). Management practices of antecedent leguminous and

non-leguminous crop residues in relation to winter wheat yields, nitrogen uptake, soil

nitrogen mineralization and simple nitrogen balance. European Journal of Agronomy

16, 295-308.

Laditi, M. A., Nwoke, O. C., Jemo, M., Abaidoo, R. C., and Ogunjobi, A. A. (2012).

Evaluation of microbial inoculants as biofertilizers for the improvement of growth and

yield of soybean and maize crops in savanna soils. African Journal of Agricultural

Research 7, 405-413.

Lafolie, F., Bruckler, L., de Cockborne, A. M., and Laboucarié, C. (1997). Modeling the

water transport and nitrogen dynamics in irrigated salad crops. Irrigation Science 17,

95-104.

Leport, L., Turner, N. C., Davies, S. L., and Siddique, K. H. M. (2006). Variation in pod

production and abortion among chickpea cultivars under terminal drought. European

Journal of Agronomy 24, 236-246.

Liska, A. J., Yang, H., Milner, M., Goddard, S., Blanco-Canqui, H., Pelton, M. P., Fang, X.

X., Zhu, H., and Suyker, A. E. (2014). Biofuels from crop residue can reduce soil

carbon and increase CO2 emissions. Nature Climate Change 4, 398-401.

Lisson, S. N., and Cotching, W. E. (2011). Modelling the fate of water and nitrogen in the

mixed vegetable farming systems of northern Tasmania, Australia. Agricultural

Systems 104, 600-608.

Littleboy, M., D.M. Freebairn, Silburn, D. M., Woodruff, D. R., and Hammer, G. L. (1999).

PERFECT Version 3.0 - A computer simulation model of Productivity, Erosion,

Runoff Functions to Evaluate Conservation Techniques.

Lupwayi, N. Z., Kennedy, A. C and Chirwa, R. M. (2011). Grain legume impacts on soil

biological processes in sub-Saharan Africa. African Journal of Plant Science 51, 1-7.

MacCarthy, D. S., Sommer, R., and Vlek, P. L. G. (2009). Modeling the impacts of

contrasting nutrient and residue management practices on grain yield of sorghum

(Sorghum bicolor (L.) Moench) in a semi-arid region of Ghana using APSIM. Field

Crops Research 113, 105-115.

Page | 67

Machido, D. A., Olufajo, O. O, Yakubu, S. E., and Yusufu, S. S. (2011). Enhancing the

contribution of the legumes to the N-fertility of soils of the semi-arid zone of Nigeria.

African Journal of Biotechnology 10, 1848-1853.

Mafongoya, P., Bationo, A., Kihara, J., and Waswa, B. (2006). Appropriate technologies to

replenish soil fertility in southern Africa. Nutrient Cycling in Agroecosystems 76, 137-

151.

Malhi, S. S., and Lemke, R. (2007). Tillage, crop residue and N fertilizer effects on crop

yield, nutrient uptake, soil quality and nitrous oxide gas emissions in a second 4-yr

rotation cycle. Soil and Tillage Research 96, 269-283.

Mary, S. R., D. Darwis and D. Robin (1996). Interactions between decomposition of plant

residues and nitrogen cycling in soil. Plant and soil, 71-82.

Mathews, R. B., and Stephens, W. (2002). "Crop-soil simulation models applications in

developing countries," CABI- Publishing, UK.

Matthews, R. B., and Stephens, W. (2002). "CROP-SOIL SIMMULATION MODELS:

Aplications in Developing Countries," CABI Publishing, Wallingford, UK.

Maxwell, S. (1986). Farming systems research: Hitting a moving target. World Development

14, 65-77.

Mazzoncini, M., Sapkota, T. B., Barberi, P., Antichi, D., and Risaliti, R. (2011). Long-term

effect of tillage, nitrogen fertilization and cover crops on soil organic carbon and total

nitrogen content. Soil and Tillage Research 114, 165-174.

McCown, R. L., Hammer, G. L., Hargreaves, J. N. G., Holzworth, D. P., and Freebairn, D.

M. (1996). APSIM: a novel software system for model development, model testing

and simulation in agricultural systems research. Agricultural Systems 50, 255-271.

Meinke, H., Baethgen, W. E., Carberry, P. S., Donatelli, M., Hammer, G. L., Selvaraju, R.,

and Stöckle, C. O. (2001). Increasing profits and reducing risks in crop production

using participatory systems simulation approaches. Agricultural Systems 70, 493-513.

Minot, N., and Benson, T. (2009). Fertilizer Subsidies in Africa: Are Vouchers the Answer?

IFPRI, Issue Brief 60, Washington, USA

Mohanty, M., Probert, M. E., Reddy, K. S., Dalal, R. C., Mishra, A. K., Subba Rao, A.,

Singh, M., and Menzies, N. W. (2012). Simulating soybean–wheat cropping system:

APSIM model parameterization and validation. Agriculture, Ecosystems and

Environment 152, 68-78.

Page | 68

Mohanty, M., Probert, M. E., Reddy, K. S., Dalal, R. C., Rao, A. S., and Menzies, N. W.

(2010). Modelling N mineralization from high C: N rice and wheat crop residues.

Proceedings of the 19th World Congress of Soil Science: Soil solutions for a changing

world, Brisbane, Australia, 1-6 August 2010. Symposium 1.5.2 Modelling critical

processes in changing soil, 28-31.

Morris, M., Kelly, V. A., Kopicki, R. J., and Byerlee, D. (2007). Fertilizer Use in African

Agriculture: Lessons Learned and Good Practice Guidelines. pp. 162. World Bank,

Wanshigton, USA.

Mupangwa, W., Dimes, J., Walker, S., and Twomlow, S. (2011). Measuring and simulating

maize (Zea mays L.) yield responses to reduced tillage and mulching under semi-arid

conditions. Agricultural Sciences 2, 167-174.

Mupangwa, W., and Jewitt, G. P. W. (2011). Simulating the impact of no-till systems on field

water fluxes and maize productivity under semi-arid conditions. Physics and

Chemistry of the Earth, Parts A/B/C 36, 1004-1011.

Mupangwa, W., Twomlow, S., Walker, S., and Hove, L. (2007). Effect of minimum tillage

and mulching on maize (Zea mays L.) yield and water content of clayey and sandy

soils. Physics and Chemistry of the Earth, Parts A/B/C 32, 1127-1134.

Naab, J., Chimphango, S., and Dakora, F. (2009). N2 fixation in cowpea plants grown in

farmers’ fields in the Upper West Region of Ghana, measured using 15N natural

abundance. Symbiosis 48, 37-46.

Nascimento, A. F. d., Mendonça, E. d. S., Leite, L. F. C., and Neves, J. C. L. (2011).

Calibration of the century, apsim and ndicea models of decomposition and n

mineralization of plant residues in the humid tropics. Revista Brasileira de Ciência do

Solo 35, 917-928.

Ncube, B., Dimes, J., Twomlow, S., Mupangwa, W., and Giller, K. (2007). Raising the

productivity of smallholder farms under semi-arid conditions by use of small doses of

manure and nitrogen: a case of participatory research. Nutrient Cycling in

Agroecosystems 77, 53-67.

Ncube, B., Dimes, J. P., van Wijk, M. T., Twomlow, S. J., and Giller, K. E. (2009).

Productivity and residual benefits of grain legumes to sorghum under semi-arid

conditions in south-western Zimbabwe: Unravelling the effects of water and nitrogen

using a simulation model. Field Crops Research 110, 173-184.

Page | 69

Ndakidemi, P. A. (2006). Manipulating legume/cereal mixtures to optimize the above and

below ground interactions in the traditional African cropping systems. African

Journal of Biotechnology 5 2526-2533.

Ngwira, A. R., Thierfelder, C., and Lambert, D. M. (2013). Conservation agriculture systems

for Malawian smallholder farmers: long-term effects on crop productivity,

profitability and soil quality. Renewable Agriculture and Food Systems 28, 350-363.

Odhiambo, J. J. O., Ogola, J. B. O., and Madzivhandila, T. (2010). Effect of green manure

legume - maize rotation on maize grain yield and weed infestation levels. African

Journal of Agricultural Research 5, 618-625.

Oyun, M. B. (2007). Nodulation and Rhizobium Population in Root Nodules of Selected Tree

Legumes. Nigerian Journal of Soil Science 17, 65-70

Pala, M., Ryan, J., Zhang, H., Singh, M., and Harris, H. C. (2007). Water-use efficiency of

wheat-based rotation systems in a Mediterranean environment. Agricultural Water

Management 93, 136-144.

Peoples, M., Brockwell, J., Herridge, D., Rochester, I., Alves, B., Urquiaga, S., Boddey, R.,

Dakora, F., Bhattarai, S., Maskey, S., Sampet, C., Rerkasem, B., Khan, D.,

Hauggaard-Nielsen, H., and Jensen, E. (2009). The contributions of nitrogen-fixing

crop legumes to the productivity of agricultural systems. Symbiosis 48, 1-17.

Peoples, M. B., Herridge, D. F., and Ladha, J. K. (1995). Biological nitrogen fixation: An

efficient source of nitrogen for sustainable agricultural production. In "Management

of Biological Nitrogen Fixation for the Development of More Productive and

Sustainable Agricultural Systems." (J. K. Ladha and M. B. Peoples, eds.), Vol. 65, pp.

3-28. Springer Science and Business Media, Netherlands.

Poulton, C., Kydd, J., and Dorward, A. (2006). Increasing Fertilizer Use in Africa: What

Have We Learned? In "Agriculture and Rural Development Discussion Paper 25",

Vol. 25, pp. 52. The World Bank, Washington, USA.

Power, B., Rodriguez, D., deVoil, P., Harris, G., and Payero, J. (2011). A multi-field bio-

economic model of irrigated grain–cotton farming systems. Field Crops Research

124, 171-179.

Probert, M. E., Dimes, J. P., Keating, B. A., Dalal, R. C., and Strong, W. M. (1998). APSIM's

water and nitrogen modules and simulation of the dynamics of water and nitrogen in

fallow systems. Agricultural Systems 56, 1-28.

Page | 70

Robertson, G. P., Grace, P. R., Izaurralde, R. C., Parton, W. P., and Zhang, X. (2014). CO2

emissions from crop residue-derived biofuels. Nature Climate Change 4, 933-934.

Rodriguez, D., deVoil, P., Power, B., Cox, H., Crimp, S., and Meinke, H. (2011). The

intrinsic plasticity of farm businesses and their resilience to change. An Australian

example. Field Crops Research 124, 157-170.

Ros, G. H. (2012). Predicting soil N mineralization using organic matter fractions and soil

properties: A re-analysis of literature data. Soil Biology and Biochemistry 45, 132-

135.

Ros, G. H., Hanegraaf, M. C., Hoffland, E., and van Riemsdijk, W. H. (2011). Predicting soil

N mineralization: Relevance of organic matter fractions and soil properties. Soil

Biology and Biochemistry 43, 1714-1722.

Ros, G. H., Hoffland, E., van Kessel, C., and Temminghoff, E. J. M. (2009). Extractable and

dissolved soil organic nitrogen – A quantitative assessment. Soil Biology and

Biochemistry 41, 1029-1039.

Ross R. Bender, J. W. H., Matias L. Ruffo, and Fred E. Below (2013). Modern Corn Hybrids’

Nutrient Uptake Patterns. Better Crops 97, 7-10.

Roy, R. N., Finck, A., Blair, G. J., and Tandon, H. L. S. (2006). Plant nutrition for food

security - A guide for integrated nutrient management. (FAO, ed.), pp. 368, Rome.

Sadras, V. O., and Rodriguez, D. (2010). Modelling the nitrogen-driven trade-off between

nitrogen utilisation efficiency and water use efficiency of wheat in eastern Australia.

Field Crops Research 118, 297-305.

Salvagiotti, F., Cassman, K. G., Specht, J. E., Walters, D. T., Weiss, A., and Dobermann, A.

(2008). Nitrogen uptake, fixation and response to fertilizer N in soybeans: A review.

Field Crops Research 108, 1-13.

Samra, J. S., Bijay, S., and Kumar, K. (2003). "Managing crop residues in the rice-wheat

system of the Indo-Gangetic Plain.p.173-195. In "Ladha J.K. (Ed.) Improving the

productivity and sustainability of rice–wheat systems, Issues and impact." American

Society of Agronomy Special Publication, 65 Wisconsin, USA

Sangakkara, U. R., Ratnayake, A. D. A., and Kumarasinghe, K. S. (2002). "Impact of soil

moisture on nodulation, biological nitrogen fixation and yields of selected tropical

legumes," International Atomic Energy Agency (IAEA), Vienna, Austria.

Sanginga, N. (2003). Role of biological nitrogen fixation in legume based cropping systems;

a case study of West Africa farming systems. Plant and Soil 252, 25-39.

Page | 71

Shah, Z., Shah, S. H., Peoples, M. B., Schwenke, G. D., and Herridge, D. F. (2003). Crop

residue and fertiliser N effects on nitrogen fixation and yields of legume-cereal

rotations and soil organic fertility. Field Crops Research 83, 1-11.

Shivakumar, B. G., Mshhra, B. N., and Gautam, R. C. (2004). Nutrient economy through land

configuration and residue management in greengram (Phaseolus radiatus L.)-Wheat

(Triticum aestivum L.) cropping sequence with limited water supllies. Acta

Agronomica Hungarica 52, 369-379.

Siame, J., Willey, R. W., and Morse, S. (1998). The response of maize/Phaseolus

intercropping to applied nitrogen on Oxisols in northern Zambia. Field Crops

Research 55, 73-81.

Sidhu, B. S., and Beri, V. (1989). Effect of Crop Residue Management on the Yields of

Different Crops and on Soil Properties. Biological Wastes 27, 15-27.

Somado, E. A., Sahrawat, K. L., and Kuehne, R. F. (2006). Rock Phosphate-P enhances

biomass and nitrogen accumulation by legumes in upland crop production systems in

humid West Africa. Biology and Fertility of Soils 43, 124-130.

Sprent, J. I., Odee, D. W., and Dakora, F. D. (2010). African legumes: a vital but under-

utilized resource. Journal of Experimental Botany 61, 1257-65.

Stern, W. R. (1993). Nitrogen fixation and transfer in intercrop systems. Field Crops

Research 34, 335-356.

Tabo, R., Bationo, A., Gerard, B., Ndjeunga, J., Marchal, D., Amadou, B., Garba Annou, M.,

Sogodogo, D., Taonda, J.-B., Hassane, O., Diallo, M., and Koala, S. (2007).

Improving cereal productivity and farmers’ income using a strategic application of

fertilizers in West Africa Advances. In "in Integrated Soil Fertility Management in

sub-Saharan Africa: Challenges and Opportunities." (A. Bationo, B. Waswa, J.

Kihara and J. Kimetu, eds.), pp. 201-208. Springer Science and Business Media,

Netherlands.

Tenkorang, F., and Lowenberg-DeBoer, J. (2008). Forescasting Long-ter Global Fertilizet

Demand. FAO, Rome, Itally.

Thierfelder, C., Cheesman, S., and Rusinamhodzi, L. (2012). A comparative analysis of

conservation agriculture systems: Benefits and challenges of rotations and

intercropping in Zimbabwe. Field Crops Research 137, 237-250.

Thierfelder, C., Mombeyarara, T., Mango, N., and Rusinamhodzi, L. (2013). Integration of

conservation agriculture in smallholder farming systems of southern Africa:

Page | 72

identification of key entry points. International Journal of Agricultural Sustainability

11, 317-330.

Thorburn, P. J., Probert, M. E., and Robertson, F. A. (2001). Modelling decomposition of

sugar cane surface residues with APSIM-residue. Field Crops Research 70, 223-232.

Tsedeke, A., Alene, A. D., Bergvinson, D., Shiferaw, B., Silim, S., Orr, A., and Asfaw, S.

(2012). "Tropical Grain Legumes in Africa and South Asia: Knowledge and

Opportunities," International Crops Research Institute for the Semi-Arid Tropics,

Nairobi, Kenya.

Turmel, M.-S., Speratti, A., Baudron, F., Verhulst, N., and Govaerts, B. (2015). Crop residue

management and soil health: A systems analysis. Agricultural Systems 134, 6-16.

Twomlow, S., Rohrbach, D., Dimes, J., Rusike, J., Mupangwa, W., Ncube, B., Hove, L.,

Moyo, M., Mashingaidze, N., and Mahposa, P. (2010). Micro-dosing as a pathway to

Africa’s Green Revolution: evidence from broad-scale on-farm trials. Nutrient

Cycling in Agroecosystems 88, 3-15.

Umeh, S. I., and Mbah, B. N. (2010). Measuring the benefits of biological nitrogen fixation

of soybean (Glycine max (l.) Merrill) in cassava (Manihot esculenta crantz) and

soybean intercrop. African Journal of Agricultural Research 5, 3354-3359.

Unger, P. W., Stewart, B. A., Parr, J. F., and Singh, R. P. (1991). Crop residue management

and tillage methods for conserving soil and water in semi-arid regions. Soil and

Tillage Research 20, 219-240.

Valbuena, D., Erenstein, O., Homann-Kee Tui, S., Abdoulaye, T., Claessens, L., Duncan, A.

J., Gérard, B., Rufino, M. C., Teufel, N., van Rooyen, A., and van Wijk, M. T. (2012).

Conservation Agriculture in mixed crop–livestock systems: Scoping crop residue

trade-offs in Sub-Saharan Africa and South Asia. Field Crops Research 132, 175-184.

Valbuena, D., Tui, S. H.-K., Erenstein, O., Teufel, N., Duncan, A., Abdoulaye, T., Swain, B.,

Mekonnen, K., Germaine, I., and Gérard, B. (2015). Identifying determinants,

pressures and trade-offs of crop residue use in mixed smallholder farms in Sub-

Saharan Africa and South Asia. Agricultural Systems 134, 107-118.

Vanclay, F. (2004). Social principles of agricultural extension to assist in the promotion of

natural resource management. Australian Journal of Experimental Agriculture 44,

213–222.

Page | 73

Verburg, K., Bond, W. J., and Hunt, J. R. (2012). Fallow management in dryland agriculture:

Explaining soil water accumulation using a pulse paradigm. Field Crops Research

130, 68-79.

Wall, P. C. (2007). Tailoring Conservation Agriculture to the Needs of Small Farmers in

Developing Countries. Journal of Crop Improvement 19, 137-155.

Wang, X., Wu, H., Dai, K., Zhang, D., Feng, Z., Zhao, Q., Wu, X., Jin, K., Cai, D., Oenema,

O., and Hoogmoed, W. B. (2011). Tillage and crop residue effects on rain fed wheat

and maize production in northern China. Field Crops Research. 132, 106-116

Wang, Y., Xu, J., Shen, J., Luo, Y., Scheu, S., and Ke, X. (2010). Tillage, residue burning

and crop rotation alter soil fungal community and water-stable aggregation in arable

fields. Soil and Tillage Research 107, 71-79.

Whitbread, A. M., Robertson, M. J., Carberry, P. S., and Dimes, J. P. (2010). How farming

systems simulation can aid the development of more sustainable smallholder farming

systems in southern Africa. European Journal of Agronomy 32, 51-58.

Wu, Y., Liu, S., Young, C. J., Dahal, D., Sohl, T. L., and Davis, B. (2015). Projection of corn

production and stover-harvesting impacts on soil organic carbon dynamics in the U.S.

Temperate Prairies. Scientific Reports 5: 10830.

Yadvinder, S., Bijay, S., and Timsina, J. (2005). Crop Residue Management for Nutrient

Cycling and Improving Soil Productivity in Rice-Based Cropping Systems in the

Tropics. In "Advances in Agronomy", Vol. Volume 85, pp. 269-407. Academic Press.

Yusuf, A. A., Iwuafor, E. N. O., Abaidoo, R. C., Olufajo, O. O., and Sangin, N. (2009).

Effect of crop rotation and nitrogen fertilization on yield and nitrogen efficiency in

maize in the northern Guinea savanna of Nigeria. African Journal of Agricultural

Research 4, 913-921.

Zahran, H. H. (1999). Rhizobium-legume symbiosis and nitrogen fixation under severe

conditions and in an arid climate. Microbiology and Molecular Biology Reviews 63,

968-89

Page | 74

CHAPTER 3:

Characterising Farmers Diversity and Opportunities for Sustainable Intensification

No smallholder farmer chooses to remain a subsistence farmer. Smallholders are always

adapting and looking for a way out! Our responsibility is to provide farmers with the right

set of tools for them to make this transformation happen in a more productive, profitable, risk

minded, and planet friendly way, through a more conscious exploitation of the resources at

their disposal.

Page | 75

Abstract

Understanding farm diversity and its effects on farming system design and the household

ability to meet its food security and income generation targets is critical to identify scalable

intensification options and constraints. In this study, a new set of simple and mutually

exclusive farm typologies were generated and used to identify group best-fit intensification

options. A mixed dataset of qualitative and quantitative household survey data and resource

allocation maps collected across two villages in Central Mozambique was used to identify

key farm differentiation variables using Categorical Principal Component Analysis

(CATPCA). Selected variables were then used to define mutually exclusive agents using a

heuristic decision tree. In the classification tree Boolean statements based on the two key

household differentiation thresholds namely resource endowment (access and usage) and

household subsistence (food security and income generation capacity) level were developed.

Five different farm typologies were developed, resource endowed and semi-intensive farmers

(Type B); resource endowed but change resistant farmers (Type - A2); and finally resource

constrained farmers (Type - A1) who can be either maize self-sufficient (Type-A1a), or food

constrained (Type-A1b). For maize self-sufficient Type-B and Type-A2 farms, the challenge

is to improve agronomy to generate enough surplus and income. For that to happen,

mechanization and labour qualification are critical. For resource constrained Type-A1

farmers, the primary challenge towards intensification is to generate enough income to

improve the household maize self-sufficiency. Due to their high market value and further

contribution to soil N-availability, improving legume productivity appears to be a feasible

alternative for Type-A1 farmers. Two intensification constraints, were identified - the soil

fertility and maize self-sufficiency relation, and the land availability for extensive farming.

Based on current farm diversity, farm typology targeted agricultural interventions are

required to actively involve farmers in the design and implementation of locally feasible

intensification options centred on a smarter exploitation of available resources.

Keywords: Farm typologies, personalized agricultural extension, adoption traps

Page | 76

3.1 Introduction

In the last decades, the need to improve yields and food security prospects among Sub-

Saharan Africa (SSA) smallholder farmers, led to the promotion of several agricultural

technologies in a clear attempt to reproduce the 1960’s green revolution. Technological

packages such as conservation agriculture, (Giller et al., 2011a; Thierfelder et al., 2013; Wall,

2007), and more recently, sustainable intensification (Robinson et al., 2015; Tittonell and

Giller, 2013; Zimmerer et al., 2015) which has recently been presented as the “new paradigm

for Agriculture in Africa” in the 2013 Montpellier Panel Report under the proposition that

food production targets should be attained with the lowest possible environmental impact.

CA and SI are mere examples of the vast array of technologies and developmental models

that are being promoted across Africa. However, despite the efforts, technology adoption

rates in SSA are still low (Baudron et al., 2015c) compared to the rest of the world.

The low technology adoption rates among SSA farmers are in part due to the existing

inability to downscale complex and knowledge intensive technological packages into

practical onsite measures that fit farmer’s circumstances. This is continued due to the

prevalence of a supply driven agricultural intervention model, (Bembridge, 1987; Binns et al.,

1997), centred on single sized technological packages. Supply driven agricultural intervention

have failed to capture the complex and strong social dimension of technology adoption

(Vanclay, 2004a). Studying and incorporating these social dynamics when planning

agricultural intervention has been a major challenge in many developing countries. This is

exacerbated by the departmentalized interventions which lead to poor linkages between the

key players, i.e., researchers across disciplines, research-extension-farmers and policy makers

(Alene and Coulibaly, 2009). In SSA complex and diverse farming environments, bringing

farmers to the negotiation table is critical to understand their circumstances and how their

perception of those same circumstances shape their livelihood strategies, farming systems

design and the likelihood to engage in new practices (Giller et al., 2011b).

Clustering farmers in functional typologies have helped shed light into how local systems

work and how understanding farm households and regional dynamics can help trigger a

conscious change of practice. Nevertheless, despite farm typologies being pinpointed as a

critical tool to improve agricultural planning back in the 1970’s (Kostrowicki, 1976;

Page | 77

Kostrowicki, 1977), rigid typologies based on farm size and on the idea of large

homogeneous groups have dominated agricultural intervention in SSA. In recent years,

however, mapping intra and intergroup diversity became useful to understand farmers’

adoption profiles and livelihood strategies (Nainggolan et al., 2013; Valbuena et al., 2008).

Since then, several approaches to group farmers in functional typologies have been

hypothesized. Perception-based farmer typologies (Guillem et al., 2012); resource

endowment (Cortez-Arriola et al., 2015) and livelihood aspirations (Dorward et al., 2009)

have been used to group farmers across the globe. Daloğlu et al. (2014) proposed a

framework to link farmer typologies with an agent based model (ABM) to aid soil and water

assessment planning in the USA Corn Belt. In Malawi, Franke et al. (2014) grouped farmers

in typologies to map their likelihood to benefit from legume intensification systems. Despite

the late developments in farm categorization, agricultural interventions in SSA still fails to

incorporate the complex social dynamics that are characteristic of smallholder farm

enterprises into the design of locally feasible technological packages.

In this paper data from a farm household characterization survey conducted in Manica

province, central Mozambique is used to map farm diversity and identify the key farm

differentiation factors that can be used to develop an alternative and easy to use farm

categorization framework. The framework would be an entry point to explore the within

group diversity and help tailor sustainable agricultural intensification practices. This study

builds on two key prepositions:

1) Understanding smallholder farmers’ biophysical and socioeconomic

circumstances, is key to effectively involve them in the co-design of resource

efficient and profitable farming systems capable of helping them meet their food

and income generation goals at the minimum possible effort.

2) Smallholder farmers’ livelihood strategies shape their farming practices.

Therefore, understanding how individual farmers and groups pursue their seasonal

food security and income generation goals is critical towards mapping farmer’s

adaptability and likelihood to engage in new practices.

Page | 78

3.2 Materials and Methods

3.2.1 Geographic setting and context of the study

The study was conducted in Manica province in central Mozambique. Manica falls within the

plateau zone and is characterized by two well-defined seasons: a wet season with 78-99% of

the total annual rainfall, i.e., from November to late April; and a dry season that runs from

May to October. Data was collected across agroecological zone R4 (AEZ-R4). AEZ-R4 has a

tropical dry climate with average annual rainfall of 834.7 mm (1951-2014). The soils are

predominately red ferralsols and luvisols (Gouveia and Azevedo, 1954).

Two villages of contrasting were selected for the farm household characterization survey

namely Matsinho in Vanduzi and Marera in Macate districts. Matsinho is a resource limited

area dominated by maize, sorghum, sunflower and commercial cotton production. In contrast,

Marera is the most productive of the two districts with well-established commercial vegetable

and fruit production based systems.

In both communities, farmers are grouped in three major categories: small-scale (0-5 ha),

semi-commercial (5-20 ha) and large scale commercial farmers (>20ha)(INE, 2011a). These

rigid typologies emerged in late 1970’s right after independence and are based on farm size,

access to production means, farm labour and the share of the production destined for

household consumption and market sales.

According to the 2009-2010 Mozambican Agricultural Census (CAP-2009), small-scale

farmers represent 96.4% of the agricultural explorations in the country (Cunguara and

Darnhofer, 2011; INE, 2011a). The Mozambican farm categorization system like many in

SSA are somehow a rigid version of Dorward et al. (2009) classification where small-scale

family farmers would fall within the “hanging in” subsistence farming households. However,

several structural changes have occurred since the 1970’s e.g., the bankruptcy of most state-

owned agricultural enterprises in the 1980’s and the markets liberalisation in the 1990s,

forced the household farm enterprise to evolve in order to adapt to the new context. Farmers

became more diverse and the boundaries between groups became less distinct. Differences

Page | 79

have also emerged within groups, mainly based on farmer’s individual approach to their farm

enterprise management in order to meet the household food security and income generation

goals (Cunguara and Darnhofer, 2011). All these structural changes call for a renewed look

into how the “farm enterprise” operates in the increasingly dynamic and diverse farming

circumstances. Therefore, understanding what factors differentiate each farm from another is

critical to design interventions that are more likely to be adopted.

3.2.2 Sampling and data analysis

3.2.2.1 Farm household and farming systems characterization

Semi-structured interviews where used to collect data on 1) household composition –

household size and farm labour availability; 2) resource use behaviour - land, fertilizer,

manure and residue use; 3) farming systems design and management, and 4) farm

productivity. A mixed dataset composed of quantitative and categorical variables was

generated from the questionnaire (Table 3-1). These variables where then screened for

relevance to mapping farm diversity. In total a random sample of 131 households distributed

across two villages in the AEZ-R4, namely 65 of 5874 in Matsinho-Vanduzi and 66 of 3804

in Marera-Macate (INE, 2011b). In total 10 communities were sampled across both districts.

Additionally, 58 people met across two focus group discussions in Marera and Rotanda were

used to cross-validate the survey data.

Two model assisted focus group discussion were held to collect additional information that

would help glean a better light into local cropping systems and sequences, resource use and

allocation strategies with focus on farm inputs, risk management strategies, yields and future

aspirations. The focus group discussion also helped parameterize the pre-defined farm

typologies. Therefore, During the focus group discussion, resource allocation maps (RAMs)

were built to gather in depth information on the whole farm household enterprise functioning

and its effect on field level management decisions and farm productivity. Two model assisted

focus group discussion were held. The first one in Marera (Macate) and was attended by 29

farmers not involved in the survey and representing 11 farmer organisations from Matsinho,

Zembe, Boavista and Marera-Sede. Two agronomists and three extension officers from the

Gondola Agricultural Directorate (SDAE-Gondola) also attended the meeting. In Rotanda, 22

Page | 80

farmers representing 7 FO’s from Messambuzi, Tsetsera and Rotanda-Sede, attended the

meeting. Three extension officers, from the SDAE-Sussundenga, also attended the Rotanda

meeting. The survey data and PCM for the household characterization was explored using

simple descriptive statistics.

3.2.2.2. Farm level productivity

Maize (Zea mays L.) self-sufficiency index, was calculated as the ratio between total maize

production and annual household maize consumption requirements. Total household maize

consumption requirements were generated by correcting Mozambique maize daily intake of

128 g maize/person which supplies 20 % daily energy intake (Nuss and Tanumihardjo, 2011;

Ranum et al., 2014) to a fully maize based diet capable of supplying 80 % daily energy intake

required to meet 2280 Kcal daily caloric requirement. Adult equivalent scales (AE) were used

to narrow the food consumption gaps between members of the same household with distinct

energy requirements. The number AE was calculated using Deaton and Zaidi (2002)

guidelines to generate consumption aggregates. For this study, adult well all farm active

people older than 15. Household consumption needs were corrected to a 17% post-harvest

loss during storage. Land to labour ratio (LLR) and yield to labour ratio (YLR), i.e., share of

land management and yield attributed to each farm active member as determined in

(Leonardo et al., 2015; Lukanu et al., 2007) were also analysed. This analysis excluded hired

labour under the assumption that each household farm active member managed his share of

labour in his portion of managed land. Maize surplus was determined as the difference

between food requirements and total maize production. The income from maize selling was

the product between maize surplus and maize market prize.

Table 3- 1 Household and farming systems characterization variables used for the categorical

principal components analysis (CATPCA)

Input variable Unit Abbreviation

District dummy Dist

Household head age # Hh_age

Household head gender dummy Hh_gender1F2M

Household head literacy level dummy Hhlireacy_lvl

Page | 81

Household size # Hhld_size

Household farm active people # Hhld_farmactpeople

Household farm active elder people # Hhld_farmactEldery

Household farm active youth # Hhld_farmactYouth

Household farm active females # Hhld_farmactFemale

Household farm active males # Hhld_farmactMale

Total number of fields cultivated in the sampled season # TotlplotOwnd

Total farm size # Total_FarmSIZE

Fertilizer use dummy Fert_Use

Fertilizer application years # FertyearsApplied

Preferred fertilizer allocation (1-maize, 2-legumes, 3-

vegetables)

dummy FertCropApplied

Best fertilizer returns crops (1-maize, 2-legumes, 3-

vegetables)

dummy fertBestrespo

Reasons not to use fertilizers dummy fertReasNOT2use

Conditions to use fertilizers dummy fertCond2Use

Fertilizer investment source dummy Fert_InvstSOURCE

Manure Use dummy Manre_Use

Crop where manure is applied (1-maize, 2-legumes, 3-

vegetables)

dummy mnreAlloction

Best manure returns crops (1-maize, 2-legumes, 3-

vegetables)

dummy mBestresponse

Reasons not to use manure dummy mReasNOT2USE

Conditions to use manure dummy mCond2Use

Cattle ownership dummy Cattle ownership

Crop residue retention dummy rsd_Use

Highest recalled yield # TtplotRecalledyield

Maize self-sufficiency index # Maize_SFIndex

labour to yield productivity index # LP_index

Labour to land ratio # LLR

Income from maize grain selling # Income_Grainselling

Maize surplus # Maize_surplus

Page | 82

3.2.2.3. Farm categorization

Variable selection

Survey data was explored by categorical principal component analysis (CATPCA), using

SPSS 23. CATPCA is a non-linear Principal Component Analysis (PCA) method used in

social and behavioural science research where mixed datasets composed of binary and

quantitative variables are common and standard PCA or Multiple Correspondence Analysis

(MCA) cannot be used (Linting et al., 2007). In agricultural research and related fields,

CATPCA allows the analysis of farmer’s behavioural attitudes that are key to understanding

decision-making processes and tailor agricultural intervention (Rouabhi et al., 2014).

CATPCA has been successful used in agriculture and related fields (Browne et al., 2014;

Comoé and Siegrist, 2013; Dossa et al., 2011a; Keshavarz and Karami, 2014; Rouabhi et al.,

2014). In this study, CATPCA was used for two purposes: first as dimension reduction

method to reduce the original set of variables, into a smaller one and to understand how much

of the variability found in the data could be explained by each selected variable; second, to

select the most influential set of variables for farm categorization that could be used to aid the

design of sustainable agricultural intensification pathways that are reflective of farmers’

needs. Therefore, the variables should be reflective of farmer’s socioeconomic circumstances

(Valbuena et al., 2008) and highlight existing farm diversity based on the associations

between farmers’ resource use behaviour and farm level productivity. Proper variable

selection for farm categorization has been pointed as more important than the classification

technique itself (Daloğlu et al., 2014; Kostrowicki, 1977) especially in highly diverse and

complex farming environments such as SSA and central Mozambique in particular.

Therefore, CATPCA is used in this study to explore a mixed dataset in search for the most

relevant categorical and quantitative variables to aid the design of simple and mutually

exclusive farm typologies. Power scores in the general dataset and loadings across the first 4

components for each district were used to rank the individual contribution of tested variables.

At the end, only variables with loadings higher than ± 0.5 were retained for analysis (Dossa et

al., 2011b).

Page | 83

Typology formulation - agent definition and parameterization

Because a mixed dataset of qualitative and quantitative variables was used as base for farm

categorization, a “classification tree” was chosen as the most appropriate method for

typology formulation (Rounsevell et al., 2012; Valbuena et al., 2008). To effectively capture

farmer’s contrasting socioeconomic circumstances, attitudes towards resource use and their

impact of the household subsistence level and field level productivity, the classification tree

is based on Boolean statements that were key to set the household operational boundaries for

the selected variables. Two thresholds were fixed. For resource endowment level, the

threshold was fixed on the household capacity to mobilize enough funds to pay for an

unsubsidized FAO Group B input package with a market cost value of 7000 MT (205 USD at

34 MT/USD). For the household subsistence level, the threshold was fixed based on farm

productivity level measured on maize self-sufficiency basis. Once the typologies were

defined, they were parameterized based on additional variables, namely household

demographic composition (household size, number of farm active people youth and adults),

farming system design and management (crops, size of cultivated land, number of fields).

Analysis of variance (ANOVA) was used to define whether the difference between

typologies were significant based on selected quantitative variables. Grouping variables for

the analysis were selected based on the CATPCA results.

3.3 Results

3.3.1 Farm household characteristics - Household demographic and land use

The average household size was c.a., 8 people, of which half were farm active (Table 3-2). Of

the household farm active members, 43.1% were female and 47.1% were men, which shows a

gender balance in farm household labour. However, of the total sampled households, 33.6%

were female-headed and the remaining 66.4% were headed by men. In Marera (Macate), 33.7

% of the households were female-headed and 66.3% were male-headed. In Matsinho

(Vanduzi), a similar trend was observed, with male-headed households (66.7%), appearing in

a higher number than female-headed ones. In terms of literacy levels, 52.7% of the household

heads had primary education, 21.4% had secondary education and less than 1% had a

university degree. The overall illiteracy level was 25.2 %. Marera, had the lowest percentage

of illiterate household heads, with 24.2% compared with the 26.2% registered in Matsinho.

Household head literacy levels, where highly affected by gender. A 45.5% illiteracy level was

found in female-headed households but only 18.3% in male headed households.

Page | 84

Table 3-2 Household characterization based on demographic data and land use

* Significant at 0.05 probability

** Significant at 0.01 probability

3.3.2 Farming systems design and management

Both, the survey and focus group discussions highlighted that rain fed maize based systems

are the most dominant systems in both villages. Nevertheless, cropping systems and

sequences were found to be highly influenced by the farm position in the landscape catena -

upland and lowland farming environments (Figure 3-1). Soil type, water availability and in

crop soil moisture regimes were the key differentiating factors between farming

environments. To take advantage of such biophysical variability, 74.1% of the farmers in the

region cultivated more than two separate fields per season. Of those households, 62.8% had

two fields and a third and fourth field was owned by 29.9% and 7.21% households

respectively. Only 25.9% of the households cultivated one single field per season.

Most of the cultivated fields are located in the upland farming environment where more land

is available. Here, apart from maize, patches of early intercropped cowpea (Vigna

unguiculata (L.) Walp) are grown. Sunflower (Helianthus annuus) and sorghum (Sorghum

bicolor (L.) Moench) are also cultivated mostly in relay intercrops with maize. The lowland

landscapes (Baixas), are wet environments dominated by mixed vegetable-cereal-legume

systems. Production in this environment is intensive due to high water availability and incrop

Household demographic

Number of cultivated fields

1 (N=34) 2 (N=61) 3 (N=29) 4 (N=7) MEAN P-value

Household size 7.03 6.89 8.10 8.43 7.61

Farm active people 3.59 3.85 4.38 5.43 4.31

Farm active adults 1.72 1.67 1.48 2 1.66

Farm active youth 2.62 2.71 3.65 3.71 2.95 *

Total cultivated land (ha) 2.03 3.09 6.20 6.14 3.67 **

Field size (ha) 2.03 1.54 2.69 1.535 1.95

Page | 85

moisture regimes which allows for season-long commercial vegetable production and sowing

a second opportunistic maize crop (Figure 3-1). Farmers in these environment normally

invest in fertilizer, manure and improved seed.

Legumes are a significant component of local cropping systems. Nearly 72.5% of the

households grow legumes. Of planted legumes, cowpea (72.6%) and common beans

(Phaseolus vulgaris L.) (62.3%) are the most planted legumes. Additionally, groundnuts

(Arachis hypogaea L.) and Bambara groundnuts (Vigna subterranean (L) Verdc) are grown

by 24.7 and 23.2% of the farmers. Legumes are mostly planted in monoculture (52.3%) and

62.1% legume growers believe the best legume yields are achieved in monocultures. In

25.3% of the farms early sowing legumes (Nov-Dec window) was identified to be where the

best yield responses, were achieved mainly with determinate erect cowpea varieties.

However, the vast majority sow cowpea in the usual January to February window. Yield

(30.1%), local adaptability (15.0%) and marketability (12.4%) were the major factors for

legume selection. Seed availability (7.84%) is also a significant determinant for legume

selection.

Page | 86

End fallow Rain Season Cool season AEZ-R4

(Matsinho and Marera) Aug Sep Oct Nov Dec Jan Feb March April May June July

Fertilized commercial vegetable production for season long producers

Mixed

Vegetable-Cereal systems

(Lowland fields – Matoros)

Veg-Maize opportunistic sequence

Fertilized vegetable Regular Maize sowing

(Veg-Maize sequence)

Vegetable production

Maize harvest

Patches of early sown

legume with Maize

Fresh beans Legume harvest

Regular legume sowing

(Sole + relay Intercrop)

Dry beans

Harvest Maize Normal Maize sowing Maize harvest Opportunistic Maize Maize based systems

(low land fields – N’dongo) Fallow period Fallow period - Unmanaged

Fallow period Normal Maize sowing Maize harvest Fallow period

Mixed cropping systems

Upland fields

(Kupswuka/Tchica)

Patches of early sown

Legume with Maize

Regular legume sowing

(Sole + relay Intercrop)

Legume harvest

1st Sorghum sowing

(Sole/Intercrop-Mz)

Sorghum harvest

Sunflower and Sorghum

(Sole or relay with Maize)

Sunflower harvest

Figure 3-1 Cropping systems description – crops and sequences for the AEZ-R4 gathered through the model assisted focus group discussions

and resource allocation maps (RAM’s) designed by local farmers

Page | 87

3.3.3 Resource use and allocation strategies

Fertilizer allocation and perceived yield advantages

Only 15.3% of the surveyed households reportedly applied fertilizers in the last 10 years. Of

these fertilizer users, 45% are located in the Marera (Macate) and the remaining 55% in

Matsinho (Vanduzi). In Matsinho, an average amount of 270 kg/season was used over the last

4.27 years. This represents approximately 63.7 kg of fertilizer per year. Since most farmers

reportedly use the compost NPK 12-24-12 as basal and Urea (46% N) in top-dress, around

18.3 kg N/year were applied by each farmer, assuming a 50/50 share between fertilizer types.

Despite an apparently lower number of people reportedly using fertiliser in Marera, annual

fertiliser application rates were 86.4 kg/year, which represents 24.1 kg N/year per farmer.

In total, 65% of the farmers applied fertilizer into their commercial vegetable production

which is where the best economic returns to investments were expected. Only 20% have

applied fertiliser directly to maize, and just 15% to legume crops (Figure 3-2) - common

beans (Phaseolus vulgaris) and soybeans (Glycine max) in particular. As a result, 65% of the

farmers reportedly got the funds to purchase fertiliser from the income generated from selling

their vegetables. Money from household-owned businesses was also used to purchase

fertilizers in 30% of the cases. Among non-fertilizer users, 58.9% cited the high cost of

fertiliser as the main reason for not using fertilizers, while 16.7% identified knowledge as the

main factor. Furthermore, willingness to use fertiliser was identified in 27.1% of the farmers

under the condition that a local provider was available and they were instructed about how to

efficiently use fertilisers. However, 17.6% of sampled farmers did not see a need to invest

into fertilizers and pointed at decreasing yields or soil fertility has the main condition to use

fertilizers, since they believed their soils are fertile enough to keep up with their production

requirements.

Manure use and perceived yield advantages

Manure is currently used by 22.3% of the households in the area. The vast majority of

manure users, were located in Marera (58.6%). In Matsinho, only 16.7% of manure users got

it directly from their own kraal and the remaining users either purchased (50%) or got it free

Page | 88

(33.3%). In Macate, 29.1% used manure from their own kraal and only 17.6% purchased

from animal owners. Free access from neighbouring animal owner was high in Macate

(52.9%). In terms of allocation, 58.8% of the farmers applied manure on their commercial

vegetables, production gardens where the best economic returns are expected (Figure 3-2).

Only 23.5% and 14.7% of manure users reportedly applying it into maize and legumes,

respectively. Among non-users, the major determinants to start using manure were cattle

ownership, purchase capacity, knowledge and transport means - 30.6 %, 24.7%, 17.6% and

7.06% respectively. Furthermore, 16.5 % of the surveyed households sow no benefits from

manure use.

Residue use and perceived yield advantage

Residues are reportedly retained by 74.1% of the sampled farmers across both communities.

Macate is more residue friendly with 60.8% of the households reportedly retaining crop

residues in the field. In Vanduzi, only 39.2% are retaining crop residues. Residues are

incorporated through ploughing in 85.6% of the farms. In 5.15% of the farms residues were

piled off the field and used as feedstock. Post-harvest grazing was reported by 11.3% of the

households. Piling residues to be used as feedstock is a common practice among small scale

dairy farmers in both Matsinho and Macate. Despite the existence of such farmers in the

region, no-one is reportedly selling maize stubble. Of all farmers retaining residues, only,

26.3% of them use fertilizers. Returns from residue application were perceived to be high in

maize by 58.8% of the households. Application into legumes and vegetables was identified as

beneficial only by 6.18% and 3.09% of households, respectively.

Page | 89

Figure 3- 2 Preferred fertilizer, manure and residue allocation across crops over the last 10 years in sampled districts. Interpretation: 1-Mz is

maize, 2-Leg is legume and 3-Veg is vegetables and 4 is shared across crops, excluding vegetables.

Page | 90

3.3.4 Farm typologies formulation

3.3.4.1 Farm diversity – key farm differentiation factors

The CATPCA revealed that resource use related variables – manure and fertilizer use,

contributed the most for the differences between households in both villages. Among

resource use related variables, farmers reasoning for not using manure was the most

influential variable with a 0.69 and 0.82 contribution to the observed variation in the first

principal component (PC1). The first two principal components together captured 57.6% of

the variability in Matsinho (Vanduzi) and 59.1% in Marera (Macate). In table 3-3, loadings

for the correlation matrix for 21 household characteristics in Matsinho and Marera are

presented. The variable loading values suggest that among the categorical variables, resource

use related variables i.e., manure and fertilizer use, associated with farmers reasoning for not

using neither of them could be used as key variables to categorize farm households in the

region. Among the quantitative variables, the household food security and income generation

status measured through maize self-sufficiency index and income from maize surplus selling

proved relevant for farm categorization. The CATPCA also showed that for both villages,

household composition related variables, i.e., the household size, the age and literacy level of

the household head were less influential. Despite being a key determinant of farmer’s risk

management strategy and significantly affecting the overall size of the household cultivated

land, the number of cultivated fields had a poor contribution to the observed variability as

CATPCA results showed. Nevertheless, similarly to the gender of the household head,

despite not explaining much of the variability observed these two variables could be used as

important grouping factors for farm categorization. Based on CATPCA analysis results which

indicated farmer’s attitudes towards resource use (access capacity and usage level) and farm

level productivity (maize self-sufficiency and income) as the key farm differentiation factors,

a heuristic decision tree (Figure 3-3) is proposed to group farmers into simplistic as mutually

exclusive typologies that are representative of existing intragroup diversity.

Page | 91

Table 3-3 Loading values for the correlation matrix for 21 farm household characteristics in

Matsinho (Vanduzi) and Marera (Macate).

Variables Categorical Principal Components

Component 1 Component 2 Component 3 Component 4

(a) Matsinho

(Vanduzi)

Manre_Use 0.853 -0.362 -0.293 0.064

mReasNOT2use 0.697 -0.409 -0.164 -0.043

Fert_Use 0.687 0.276 0.620 -0.088

mCond2Use 0.573 -0.369 0.082 -0.033

fertReasNOT2use 0.321 0.208 0.450 0.663

fertCond2Use 0.006 0.244 0.551 0.604

Income ($) _Maize -0.214 -0.799 0.470 0.032

Maize_SFIndex -0.255 -0.737 0.469 -0.102

Maize_surplus -0.258 -0.822 0.449 -0.060

Total Prod -0.311 -0.776 0.419 -0.021

LP_index -0.326 -0.691 0.495 -0.009

LLR -0.491 -0.323 -0.079 0.335

fertAmountoverTiMe -0.522 -0.295 -0.614 0.149

FertyearsAppld -0.549 -0.329 -0.647 0.192

fertCropAppld -0.667 -0.262 -0.623 0.126

fertBestrespo -0.667 -0.262 -0.623 0.126

mnreAlloction -0.806 0.354 0.331 0.005

mBestresponse -0.806 0.354 0.331 0.005

manre_Type -0.809 0.258 0.226 -0.243

manre_Source -0.832 0.295 0.298 -0.042

manre_yrsofUse -0.848 0.362 0.249 -0.101

(b) Marera (Macate)

Manre_Use 0.908 -0.118 -0.350 -0.060

mReasNOT2use 0.824 0.048 -0.342 0.034

mCond2Use 0.598 -0.237 -0.387 0.175

Fert_Use 0.528 0.744 0.359 -0.141

fertReasNOT2use 0.291 0.507 -0.094 0.575

Page | 92

fertCond2Use 0.122 0.539 -0.099 0.576

LLR 0.055 0.232 0.088 0.541

fertAmountoverTiMe -0.415 -0.629 -0.441 0.088

LP_index -0.416 0.526 -0.616 -0.040

fertCropAppld -0.467 -0.774 -0.313 0.171

Maize_surplus -0.478 0.520 -0.656 -0.114

Income ($) Maize -0.481 0.538 -0.640 -0.110

FertyearsAppld -0.486 -0.760 -0.354 0.159

fertBestrespo -0.528 -0.744 -0.359 0.141

Total Prod -0.540 0.446 -0.676 -0.051

Maize_SFIndex -0.545 0.526 -0.626 -0.087

manre_Source -0.735 0.142 0.467 0.158

manre_Type -0.775 0.315 0.359 0.001

mBestresponse -0.802 0.144 0.435 0.043

mnreAlloction -0.841 0.120 0.416 0.030

manre_yrsofUse -0.851 -0.030 0.327 0.026

The variables are ranked in importance to the first principal component

Page | 93

How resources are

allocated?

Cropij ... CropN

Type B (a, b, ... N) Type A2 (a, b, …N)

(Type R ij * Cropij ... Type RN * CropN)

Type A1(a, b, ... N)

(Type R ij * Cropij ... Type RN * CropN)

RAM’s: Farm productivity and livelihood strategy: Are subsistence requirements met, i.e., maize self-sufficient and maize income? (YES/NO).

Secondary data: household demographics (size, farm active people), farming systems design and management (crops, and size of cultivated land)

Household Income and assets (Resource Endowment and usage)

(Can the household generate enough income to pay for 1 Group-B FAO input package: 50 kg NPK 12:24:12 + 50 kg Urea + 12.5 kg Maize seed)

(Yes) (No)

Uses? YES!

Resource Endowed and Uses either

fertilizer or Manure

Uses? NO!

Resource Endowed but Uses neither

Fertilizer nor Manure

Resource Constrained and Uses neither of them

Reasons not to Use

(Rij … RN)

Conditions to Use

(Condij … CondN)

Reasons not to Use

(Rij … RN)

Conditions to Use

(Condij … CondN)

Figure 3-3 Framework for farmer categorization and personalize agricultural intervention. Key grouping factors were attitudes towards resource

use (fertilizer, manure) and the household maize self-sufficiency level

Page | 94

3.4.5 Farm typologies

Three key structural typologies (Figure 3-4) were identified based on resource endowment

level and usage: (1) the Type-B farming households, which are resource-endowed innovative

farmers; (2) Type-A2 farmers who are resource endowed but technology averse farmers and

finally the (3) Type-A1 farmers who are resource constrained smallholder farmers. Type-A1

farmers can be either maize self-sufficient (Type-A1b) or not (Type-A1a). In both villages,

Type-B households had the largest household size. However, the number of farm active

people only differed across typologies in Marera where the number of young farm active

people also differed. The household subsistence level differed across typologies. Here, farm

productivity indexes, i.e., maize self-sufficiency, maize income, the labour to land ration

(LLR) and crop production per farm active member (YLR) differed significantly across

typologies in both villages. Maize self-sufficiency in particular, proved to be affected by the

gender of the household head specially in Matsinho.

Figure 3-4 Hypothetical model for resource endowment based farm typologies. Typology

specific develop pathways are indicated by dashed arrows. Continuous arrow indicates a

stepping out situation (Dorward et al., 2009), i.e., a level of income diversification where the

farmer is able to generate enough off the farm income to reduce dependence from farming.

Page | 95

Table 3-4 Households distribution and characterization across farm types and sites

Sites variables Hhold size Farm active people Youth Adults Cultivated land Maize Income LLR YLR_index SSFIndex

Farm types (FT) # p/Hh # p/Hh # p/Hh # p/Hh (ha) (USD) ha/p kg/p

Matsinho

(Vanduzi)

A1a 7.22 4.04 2.96 1.09 2.04 - 0.613 66 0.16

A1b 5.67 3.62 2.67 0.88 2.46 265.1 0.787 683 1.55

A2 4.5 2 1.5 1 5.12 652.2 2.562 2125 6.01

B 8.12 4.94 3.31 1.62 7.27 130.1 1.447 420 2.75

FT - - - 0.044 0.01 <.001 <.001 <.001 <.001

FT x #_fields - - - 0.036 - - - - -

FT x Hhhead gender - - - - - - - - 0.032

Marera

(Macate)

A1a 7.22 3.24 1.89 0.73 3.68 - 1.28 62 0.13

A1b 7.25 4 2.75 1.25 4.34 440 1.01 1089 0.95

A2 6.33 4.67 3.67 1 4.92 508.7 1.09 1105 3.77

B 9.93 5.27 3.13 1.2 3.23 100.7 0.99 267 3.07

FT 0.041 0.026 0.013 - - <.001

<.001 <.001

FT x #_fields - - - - - - - - -

FT x Hhhead gender - - - - - - - - -

Legend: p/Hh is people per household; ha/p is hectares per person; kg/p is kg of maize produced per farm active person

(a) Resource endowed innovative farmers – Type B (23.7%)

Farmers in this group normally use fertilisers and manure. Applied manure is from their own

kraal or acquired locally. Type – B farmers cultivate at least two fields per season, of which one

is the wet lowland farming environment where they run a commercial vegetable production farm.

This lowland commercial vegetable production unit is where most of the fertilizer and manure

are allocated. Type-B farmers normally sell part of their maize. However, despite cultivating

more land than other groups (Matsinho) and also being maize self-sufficient (Table 3-4), Type-B

had lower yield to labour productivity ratio (YLR) and income from maize selling. This is

explained by the fact that most of their income is being generated from their commercial

vegetable production and legume crops rather than maize grain selling. Poor maize agronomy is

also an issue. Type-B farmer’s lowland mixed vegetable-maize systems are semi-intensive and

specialized in a reduced number of crops per season. Normally marketability is the key crop

selection factor. Here, an opportunistic maize crop is normally grown right after the end of the

cool season’s in June-August and runs up until the start of the regular maize sowing season in

October-December (Figure 3-1). In this sequence, maize mostly benefits from the residual N

from vegetable production, which also includes N-fixing fresh and dry beans. To take advantage

of the residual N and prolonged wetness of these lowland fields, farmers normally purchase high

yielding hybrid varieties. In terms of labour, due to external cash availability and maize surplus,

Type B farmers are able to hire draft animal power (DAP), tractor (HTP) for land preparation.

Nevertheless, sowing is not mechanized. The household head is mostly well connected in the

community. These farmers are also market oriented legume growers.

(b) Non-resource users: Type - A (76.3%)

This group hosts approximately 76 % of surveyed households. The vast majority of the farmers

in this group run up to three or more fields per year. These are extensive and pro-diversification

farm enterprises. Within this group, two distinct categories of non-resource users can be

distinguished, based on their assets and reasons for not using either fertilizers or manure. The

resource endowed but resistant change farmer (Type-A2) and the resource constrained farmer

(Type-A1).

Page | 97

Resource endowed but change resistant farmers: (Type - A2; 6.11%)

These are multipurpose and labour intensive farms. These farmers are engaged in several on

farm and off-farm income generating activities that allow the household to hire DAP/HTP and

labour mainly for land preparation and weeding respectively. The majority of these farmers are

cattle owners, but do not use manure. Type -A2 farmers, generate enough income (Table 3-4) to

invest in fertilizers in both villages but don’t. The group’s maize self-sufficiency levels, 6.07 in

Matsinho and 3.77 in Marera contribute to the situation, since farmers manage to meet their food

security targets and generate enough surplus for income generation. Apart from assumed high

fertilizer price, Type – A2 farmer’s poor investment in fertilizers is associated with their maize

self-sufficient which gives the farmers the perception that their land is productive enough for

them to meet their production targets. However, their food security is attained through a labour

intensive farming approach. In Macate, labour competition between fruit crops – citrus and

banana with cereal and vegetables are common.

Two subgroups can be distinguished among type – A2 farmers. The traditional investment

resistant farmers. The household head is older (>45 years old) and points at decreasing soil

fertility and poor yields as pre-conditions to invest in fertilisers. However, this is less likely to

happen while the farmer has the means to run his normal farming cycle. Moreover, their

tightened budget and farm management approach is also a hindering factor, since most of the

farm income, is reinvested back into keeping the laborious farm cycle running. In Macate,

maintaining the citrus and banana orchards or the year-long commercial vegetable production

consumes most of the household income. There is also the resource endowed but input aware

and willing farmers. These are increasingly younger farmers, who like the previous subgroup,

have the asset base to invest in external inputs and practices that would likely reduce their work

load and improve the whole system performance. However, they are poorly connected to

information networks, therefore, do not have enough technical support to identify and test new

practices. Labour qualification and technical assistance is critical for this group.

Page | 98

Resource constrained farmers (Type - A1)

Most of the surveyed smallholder farmers fall within this group. They do not use fertilizers or

manure. Despite a vast majority being aware that fertilizers and manure would most likely

improve their yields, they are resource limited. Limited access to knowledge networks and a poor

asset base are major constraints to engage in new technologies especially among the food-

insecure ones, i.e., Type –A1a farmers (45.8%). Farmers in this group had the lowest YLR, 66

and 62 kg/Hh farm active household member in Matsinho and Marera respectively. Maize self-

sufficient was the lowest among Type-A1a farmers (Table 3-4). For that reason, most of the

farmers in this group engage in part-time on-farm and off-farm jobs to cover their expenses

while waiting for the next harvest. There is however, a growing number of resource constrained

but maize self-sufficient (Type –A1b; 24.4%) young male-led households where the household

head normally works several part-time jobs during the season or has an off-farm job. Therefore,

he is on the farm only during pick times, such as land preparation and weeding. Normally his

off-farm income is invested in small amounts of OPV seeds and is also used to pay for additional

labour for land preparation and weeding and cover household expenses during the low food

periods, i.e., December to March in bad years as they are self-sufficient most of the years. These

farmers are also typical traditional legume growers (L2-Type), planting a wide range of legumes

across the season from where they get most of their on-farm income. Fresh and dry groundnuts,

cowpeas sold as fresh pods and dry grain, Bambara groundnuts and also tuber and root crops are

key income sources. However, the amounts sold are considerable lower compared to other

groups.

3.5 Discussion

3.5.1 Improving agricultural advisory services through typology tailored intervention

Clustering farmers into functional typologies is crucial for central Mozambique and most SSA

countries, where farming across groups is a highly context bounded activity in both space and

time as contrasting household livelihood strategies, biophysical and socioeconomic

circumstances shape farming system design and management. Therefore, clustering farmers into

small homogeneous typologies that are more reflective of farmers biophysical and

socioeconomic circumstances is critical to glean a better light in farmers’ decision making

Page | 99

process. The advantage of clustering farmers as a mean to aid agricultural intervention has

already been proved valid in East Africa (Pannell and Vanclay, 2011; Tittonell et al., 2010)

where it helped understand how farmers’ livelihood strategies influence soil fertility

management practices. In Malawi, southern Africa, Franke et al. (2014), built typologies to

understand how different farmer groups would benefit from sustainable intensification. In both

situations, clustering farmers in functional typologies helped identify key points of intervention.

However, finding better ways to use the data from this type of exercises as baseline to

interactively design personalised agricultural support programs that suits each group

circumstances has been the main challenge in the pipeline to trigger effective change of practice

in SSA.

Typology targeted agricultural intervention, i.e., focusing on each group unique circumstances

and developmental needs, is the closest approach to 1:1 peer tutoring defined in Maertens and

Barrett (2013) as the most effective learning approach. Personalizing agricultural intervention

and allowing farmers to experiment and learn by doing in their own fields is a more effective

way to trigger adoption (Cameron, 1999; Munshi, 2004) compared to the conventional

intervention methods which treats farmers as equal. However, although, splitting farmers into

small functional typologies is a pivotal tool to make learning and interaction more effective, it

also true that typologies are dynamics and smallholder farmers are always evolving. Therefore,

exploring the similarities and transition factores between groups is as important as having the

contrasts clearly set.

To validate typology tailored agricultural intervention, it is important to design and validate new

and easy to use farm categorization frameworks that are likely to be used by field staff when

performing simple appraisals of agricultural intervention needs at community level. Based on

that, a new framework to map intragroup diversity (figure 3-3) is proposed in this study. The

framework is flexible in time and can accommodate the seasonal changes in farmer’s livelihood

strategies and socioeconomic circumstances. This is because the typologies generated through

the framework clearly explore the intra-group diversity based on the premise that the

differentiation is crucial to understand how each groups socioeconomic status and livelihood

strategy are critical to build the foundations to an effective design of group-specific agricultural

Page | 100

interventions (Daloğlu et al., 2014). The ability of the framework in exploring farmers reasoning

to use or not a certain resource and the preconditions to engage in its use is critical to the effort

of validating farm typology tailored interventions. This approach is important in the sense that,

rigid characterisation of farmer groups as it currently happens tells us little about each

household’s productive capabilities or their innovative approaches to farming which is critical to

re-negotiate and effectively co-design group specific interventions.

3.5.2 Handling the key intensification constraints across contrasting farm typologies

Soil fertility and maize self-sufficiency

Despite poor soil fertility being vastly identified as one of the major reason behind lower yields

in SSA agriculture (Bekunda et al., 2010; Mwangi, 1996), farmers identified decreasing yields

and soil fertility loss as pre-conditions for investing in fertilizers and manure. This attitude is

based on the premise that their soils are “fertile enough” to produce what they require to be food

safe - the soil fertility and maize self-sufficiency perception. This trend was observed across both

districts, mainly among maize self-sufficient Type-A2 and resource constrained Type-A1b

farmers, who are typical extensive farmers. However, this is a counterproductive approach since

the household food security targets are achieved through a laborious extensive farming approach

(Lotter, 2015). Nevertheless, misconceptions about soil fertility among smallholder farmers in

SSA are common especially among maize self-sufficient farmers (Scoones, 2001) who tend to

relate poor yield with low rainfall pattern rather than soil fertility.

The challenge in triggering a change of practice among Type-A2 farmers is that, the overall farm

performance is assessed through seasonal maize self-sufficiency and income levels rather than

technical efficiencies. This is in line with findings from Vanclay (2004a) who pointed that

“sustainability” for the farmer is about being in the farm and keep up with it.. An analogy can be

made to Type-A2 farmers in central Mozambique, to whom everything revolves around being

able to produce enough grain to keep up with the household’s basic needs – food and income,

throughout the year independently of how technically inefficient and laborious is the process: the

maize self-sufficiency trap. Therefore, introducing technological changes into these group needs

to take into account that they are already self-sufficient and that only interventions that can prove

Page | 101

capable to lift their economic and social status without an abrupt disruption of their own

reasoning to do the things the way they do are more likely to be put to the test. However, as in all

adoption processes which are basically ignited by a change of mind-set, it will not happen

immediately (Maertens and Barrett, 2013). However, continuous and personalized agricultural

support services aiming at demonstrating short term returns from investment and help farmers

visualize long term benefits of each agreed management change will likely challenge farmers to

move out of their comfort zone. These proposed changes need to be imbibed in what farmers

already do and fit into their own development plan. A similar approach was proposed by Vanclay

(2004a) when studying farmers’ attitudes and the entry points for successful extension programs

in Australia. The author pointed that for most farmers, farming is a socio-cultural practice rather

than just a technical activity, and most farmers do the things they do because it has become a

way of living for them. Therefore, understanding and taking into account these ways of living

when designing agricultural interventions is the first step towards stimulating a conscious change

of practice, since no single sized technological package is likely to fit all farms.

Land access and labour availability perception trap

Independently of the household size, cultivating multiple fields in a season is a common risk

management practice implemented by most households. This is done in an attempt to take

advantage of existing biophysical and climatic gradients – perceived soil fertility, rainfall, soil

moisture regimes and land productivity (Table 3-4). Despite being a traditional risk management

practice management capabilities differ across groups, consequently the viability of the practice

as well. Nevertheless, extensification rather than “(Sustainable) intensification” as a way of

improving production and on farm income among smallholder farmers in environments where

access to land is not a limiting factor was reported before (Cameron, 1999; Leonardo et al.,

2015). In Central Mozambique, access to additional man power, draft animal power (DAP) and

capacity to hire “tractorized” power (HTP) among resource endowed Type-B, Type-A2 and

some resource constrained self-sufficient farmers (Type-A1a) is a key factor in determining the

household likelihood to cultivate more than one fields. The number of households with access to

HTP and DAP has increased significantly from 7% in 1996 to 39.9 % in 2010 (INE, 2011a). This

figure is expected to have increased to approximately 51.7% by 2015 assuming a 2.35% annual

increase since 2010. In the last 5-10 years, facilities to HTP and DAP hiring have been created

Page | 102

through the local agricultural directorates (SDAE) and private providers from where farmers can,

access those services at rates ranging from 11 and 22 USD for an hour of tractor harrowing and

ploughing. Approximately 18 USD for animal draft power to cover an area of 0.25 ha. Most

farmers are willing to pay those prices mainly because cultivating multiple fields is also a

measure of social status (Place, 2009) and a mean to achieve surplus (Leonardo et al., 2015).

However, this come at a price mainly for resource constrained Type-A1 farmers who are unable

to mobilize enough man power to weed.

Mechanization of farm activities as a driver to optimize land usage and productivity is a valid

option to the already maize self-sufficient and non-labour constrained Type-B and Type-A2

farmers. In these farms, improving farm income from maize based system is the main driver for

mechanization. However, mechanization needs to accommodate the whole spectrum of farm

activities (Twomlow et al., 2002), i.e., land preparation, using improved varieties, fertilizer use

and most importantly weeding so that the benefits from sowing enough land to achieve surplus

are not offset by farmer’s inability to weed their fields as it is happening. The use of small

multipurpose machines like the two-wheel tractors (2WTs), (Baudron et al., 2015b), for sowing

and weeding would help improve crop arrangement, increase yields prospects and reduce time

invested in those activities. For resource constrained Type-A1 farmers, the primary challenge is

to generate enough income to supplement their low maize self-sufficiency and start investing in

external inputs - fertilizer, manure and improved seeds to boost crop land productivity.

Because Type-A1 farmers are capital constrained to invest in inorganic fertilisers (Liverpool-

Tasie and Takeshima, 2013), legume intensification systems are a feasible alternative to increase

farm income while improving soil N availability. However, more legume friendly resource

allocations strategies and cropping systems need to be considered. To validate these systems,

special attention needs to be given to legume selection in order to find a suitable balance between

grain yield and biomass production since high harvest index legumes do less for soil fertility

(Vanlauwe and Giller, 2006). For the laborious Type-A2 farms and labour constrained Type-A1b

farmers, improving labour quality, i.e., labour-land ratio (LLR) and yield to labour productivity

ratio (YLR) through the specialization of farm labour, is critical before considering field size

Page | 103

increments (Table 3-4). However, this needs to happen in parallel with investment in external

inputs and production means – fertilizers, improved seeds.

3.5.3 Building from current farming systems design

Resource access and smart allocation strategies

As already found in several other studies across SSA, capital constraints are a key limiting factor

for investing in inorganic fertilizers among resource poor Type-A1 farmers (Ade Freeman and

Omiti, 2003; Chianu and Tsujii, 2005; Lambrecht et al., 2014). Nevertheless, our results showed

that, despite lower use there is preferential resource allocation as well. These results are in line

with findings from Vanlauwe and Giller (2006) who also argued that fertilizers are used in SSA

often where they offer a favourable value-to-cost ratio. In central Mozambique, fertilizers and

manure are applied into vegetables were they are believed to provide better returns.

Nevertheless, more still needs to be done mainly regarding fertilizer promotion strategies so that

they are attractive to farmers. The fact that responses from low fertilizer rates are being reported

across Africa (Kgonyane et al., 2013; Vanlauwe et al., 2011) is a good starting point. However,

the low rate response data contradicts current blanket fertilizer recommendations (Chianu et al.,

2012; Rware et al., 2014), meaning that fertilizer recommendations still need to be tailored to fit

crops, farm variability and environments, purchase capacity and field sizes in some cases.

Farming environments proved also to play a pivotal role in shaping local cropping systems’

design (Figura 3-1). Understanding the biophysical differences across farming environments has

allowed smallholder farmers to innovate in order to take advantage of existing gradients. The

crops, sequences and resource allocation preferences across farming environments is a proof, of

how farmers are using the existing variability to their advantage. Better maize arrangements,

were observed among farmers who used improved OPV’s and hybrid maize seed varieties, once

germination of this material is not questioned (Biemond et al., 2013). As a result, yields and

management practices were better, especially when DAP was used to sow and weed in Type B

and Type A2 farms. In these systems, yields as high as 1.4 t/ha were measured with unfertilized

cross pollinated (OPV) maize in the upland maize systems in Macate. In the wet lowland mixed

maize-vegetable systems, a 3.78 t/ha yield was measured in Matsinho, with the hybrid PAN67,

Page | 104

planted at 3.3 pls/m2 density in a field previously under fertilized fresh beans. Taking local

management tactics into account when planning agricultural intervention is important as it will

help generate information that is likely to be used by farmers in making informed decisions

regarding best fit resource allocation strategies in their system. For that to happen, a whole farm

household analysis that looks at the overall livelihood strategies, resource allocation across

activities, crop management responses and income returns between management options across

typologies is required since there are clear trade-offs to be considered at household and field

level. Therefore, facilitating access to information is critical to help farmers smartly decide what

management option better suits his farm and will most likely help him attain his seasonal and

long term development goals more efficiently.

3.6 Conclusions

Overall, the findings from this study point that, understanding farmer’s socioeconomic

circumstances, their resource use behaviour and the reasoning behind their management

decisions, is critical to involve them in the co-design of locally feasible and sustainable farming

systems. In order to achieve that, a socio-technical approach that captures the dynamics within

the “whole farm enterprise” and builds from the existing set up, i.e., the whole household

livelihood strategies and own development needs is required. This approach is favoured over a

linear and technical intervention approach focused sole on field level agronomic management

since the farm enterprise is far more complex than the field. Our analysis also suggests that, at

field level, there are circumstantially induced trade-offs, i.e., plant densities, N-fertilizers,

varieties and residues in the system that need to be considered to see how each one fits into each

farm enterprise management model and possible development pathways. Adding to that, the

existing diversity of farms, livelihood strategies and management decisions, reveals that there are

farm-specific management combinations of technologies that best fit each farmers’ bio-physical

and socioeconomic circumstances. Therefore, no generalized single-sized solution is likely to fit

all farms. This because all modifications to be introduced on the way the “farm household

enterprise” is managed need to be rooted on farmers’ circumstances and elected paths to pursue

the household food and income generation targets. Assuring this key principle is critical to

actively involve farmers in the identification and implementation of locally feasible

intensification pathways.

Page | 105

Acknowledgement

The authors would like to thank the SIMLESA project that through the Australian Centre for

International Agriculture Research (ACIAR) awarded me the scholarship to pursue my studies.

The Soil Health Program of the Alliance for Green Revolution in Africa (AGRA) through the

2011SHP007 Project and 201-Inv-FNI for helping fund this work. Extended thanks also goes to

Eng.º Domingos Dias (IIAM), Clemente Zivale and Ricardo Maquina for the help with the

surveys. Thanks also to Eng.º Edwin Pinho, Eng.º Chiwiro Filipe, Tec., Rosario and Tec. David

from the Gondola and Sussundenga Agricultural Directorate for their help in organizing and

running the model assisted participatory focus group discussions in Marera and Rotanda.

References

Ade Freeman, H., and Omiti, J. (2003). Fertilizer use in semi-arid areas of Kenya: analysis of

smallholder farmers' adoption behavior under liberalized markets. Nutrient Cycling in

Agroecosystems 66, 23-31.

Alene, A. D., and Coulibaly, O. (2009). The impact of agricultural research on productivity and

poverty in sub-Saharan Africa. Food Policy 34, 198-209.

Baudron, F., Sims, B., Justice, S., Kahan, D., Rose, R., Mkomwa, S., Kaumbutho, P., Sariah, J.,

Nazare, R., Moges, G., and Gérard, B. (2015a). Re-examining appropriate mechanization

in Eastern and Southern Africa: two-wheel tractors, conservation agriculture, and private

sector involvement. Food Security 7, 889-904.

Baudron, F., Thierfelder, C., Nyagumbo, I., and Gérard, B. (2015b). Where to Target

Conservation Agriculture for African Smallholders? How to Overcome Challenges

Associated with its Implementation? Experience from Eastern and Southern Africa.

Environments 2, 338-357.

Bekunda, M., Sanginga, N., and Woomer, P. L. (2010). Chapter Four - Restoring Soil Fertility in

Sub-Sahara Africa. In "Advances in Agronomy" (L. S. Donald, ed.), Vol. Volume 108,

pp. 183-236. Academic Press, Netherlands.

Bembridge, T. J. (1987). Agricultural extension in the less developed areas of Southern Africa.

Agricultural Administration and Extension 27, 245-265.

Page | 106

Biemond, P. C., Oguntade, O., Stomph, T.-J., Kumar, P. L., Termorshuizen, A., and Struik, P.

(2013). Health of farmer-saved maize seed in north-east Nigeria. European Journal of

Plant Pathology 137, 563-572.

Binns, T., Hill, T., and Nel, E. (1997). Learning from the people: Participatory rural appraisal,

geography and rural development in the ‘new’ South Africa. Applied Geography 17, 1-9.

Browne, M., Ortmann, G. F., and Hendriks, S. L. (2014). Household food security monitoring

and evaluation using a resilience indicator: an application of categorical principal

component analysis and simple sum of assets in five African countries. Agricultural

Economics Research, Policy and Practice in Southern Africa 53, 25-46.

Cameron, L. A. (1999). The Importance of Learning in the Adoption of High-Yielding Variety

Seeds. American Journal of Agricultural Economics 81, 83-94.

Chianu, J., Chianu, J., and Mairura, F. (2012). Mineral fertilizers in the farming systems of sub-

Saharan Africa. A review. Agronomy for Sustainable Development 32, 545-566.

Chianu, J. N., and Tsujii, H. (2005). Determinants of farmers' decision to adopt or not adopt

inorganic fertilizer in the savannas of northern Nigeria. Nutrient Cycling in

Agroecosystems 70, 293-301.

Comoé, H., and Siegrist, M. (2013). Relevant drivers of farmers' decision behavior regarding

their adaptation to climate change: a case study of two regions in Côte d'Ivoire.

Mitigation and Adaptation Strategies for Global Change, 1-21.

Cortez-Arriola, J., Rossing, W. A. H., Massiotti, R. D. A., Scholberg, J. M. S., Groot, J. C. J.,

and Tittonell, P. (2015). Leverages for on-farm innovation from farm typologies? An

illustration for family-based dairy farms in north-west Michoacán, Mexico. Agricultural

Systems 135, 66-76.

Cunguara, B., and Darnhofer, I. (2011). Assessing the impact of improved agricultural

technologies on household income in rural Mozambique. Food Policy 36, 378-390.

Daloğlu, I., Nassauer, J. I., Riolo, R. L., and Scavia, D. (2014). Development of a farmer

typology of agricultural conservation behavior in the American Corn Belt. Agricultural

Systems 129, 93-102.

Deaton, A., and Zaidi, S. (2002). Guidelines for constructing consumption aggregates for welfare

analysis. (T. W. Bank, ed.). The World Bank, Washington, USA.

Page | 107

Dorward, A., Anderson, S., Bernal, Y. N., Vera, E. S., Rushton, J., Pattison, J., and Paz, R.

(2009). Hanging in, stepping up and stepping out: livelihood aspirations and strategies of

the poor. Development in Practice 19, 240-247.

Dossa, L. H., Abdulkadir, A., Amadou, H., Sangare, S., and Schlecht, E. (2011a). Exploring the

diversity of urban and peri-urban agricultural systems in Sudano-Sahelian West Africa:

An attempt towards a regional typology. Landscape and Urban Planning 102, 197-206.

Franke, A. C., van den Brand, G. J., and Giller, K. E. (2014). Which farmers benefit most from

sustainable intensification? An ex-ante impact assessment of expanding grain legume

production in Malawi. European Journal of Agronomy 58, 28-38.

Giller, K. E., Corbeels, M., Nyamangara, J., Triomphe, B., Affholder, F., Scopel, E., and

Tittonell, P. (2011a). A research agenda to explore the role of conservation agriculture in

African smallholder farming systems. Field Crops Research 124, 468-472.

Giller, K. E., Tittonell, P., Rufino, M. C., van Wijk, M. T., Zingore, S., Mapfumo, P., Adjei-

Nsiah, S., Herrero, M., Chikowo, R., Corbeels, M., Rowe, E. C., Baijukya, F., Mwijage,

A., Smith, J., Yeboah, E., van der Burg, W. J., Sanogo, O. M., Misiko, M., de Ridder, N.,

Karanja, S., Kaizzi, C., K’ungu, J., Mwale, M., Nwaga, D., Pacini, C., and Vanlauwe, B.

(2011b). Communicating complexity: Integrated assessment of trade-offs concerning soil

fertility management within African farming systems to support innovation and

development. Agricultural Systems 104, 191-203.

Gouveia, D. H. G., and Azevedo, A. L. (1954). The Provisional Soil Map of Mozambique. In

"Second Inter-African Soil Conference", Vol. 1459-1468. ISRIC, Leopoldville.

Guillem, E. E., Barnes, A. P., Rounsevell, M. D. A., and Renwick, A. (2012). Refining

perception-based farmer typologies with the analysis of past census data. Journal of

Environmental Management 110, 226-235.

INE (2011). "Censo Agro-Pecuario 2009-2010: Resultados Preliminares." Instituto Nacional de

Estatistica(INE), Ministerios da Agricultura (MINAG), Maputo, Mozambique.

Keshavarz, M., and Karami, E. (2014). Farmers' decision-making process under drought. Journal

of Arid Environments 108, 43-56.

Kgonyane, M. C., Mariga, I. K., and Dimes, J. P. (2013). Low rates of nitrogen and phosphorus

as fertilizer options for maize production by smallholding farmers in drier regions of

South Africa. Research on Crops 14, 444-454.

Page | 108

Kostrowicki, J. (1976). Agricultural typology as a tool in planning spatial organization of

agriculture—IGU Commission on Agricultural Typology. Geoforum 7, 241-250.

Kostrowicki, J. (1977). Agricultural typology concept and method. Agricultural Systems 2, 33-

45.

Lambrecht, I., Vanlauwe, B., Merckx, R., and Maertens, M. (2014). Understanding the Process

of Agricultural Technology Adoption: Mineral Fertilizer in Eastern DR Congo. World

Development 59, 132-146.

Leonardo, W., van de Ven, G. J., Udo, H., Kanellopoulos, A., Sitoe, A., and Giller, K. (2015).

Labour not land constrains agricultural production and food self-sufficiency in maize-

based smallholder farming systems in Mozambique. Food Security, 1-18.

Linting, M., Meulman, J. J., Groenen, P. J. F., and Van der Kooij, A. J. (2007). Nonlinear

Principal Components Analysis: Introduction and Application. Psychological Methods

12, 336-358.

Liverpool-Tasie, L. S. O., and Takeshima, H. (2013). Input promotion within a complex

subsector: fertilizer in Nigeria. Agricultural Economics 44, 581-594.

Lotter, D. (2015). Facing food insecurity in Africa: Why, after 30 years of work in organic

agriculture, I am promoting the use of synthetic fertilizers and herbicides in small-scale

staple crop production. Agriculture and Human Values 32, 111-118.

Lukanu, G., Green, M., and Worth, S. (2007). The influence of smallholder labour demand on

cultivation of cash crops in northern Mozambique. Development Southern Africa 24, 553-

573.

Maertens, A., and Barrett, C. B. (2013). Measuring Social Networks' Effects on Agricultural

Technology Adoption. American Journal of Agricultural Economics 95, 353-359.

Munshi, K. (2004). Social learning in a heterogeneous population: technology diffusion in the

Indian Green Revolution. Journal of Development Economics 73, 185-213.

Mwangi, W. M. (1996). Low use of fertilizers and low productivity in sub-Saharan Africa.

Nutrient Cycling in Agroecosystems 47, 135-147.

Nainggolan, D., Termansen, M., Reed, M., Cebollero, E., and Hubacek, K. (2013). Farmer

typology, future scenarios and the implications for ecosystem service provision: a case

study from south-eastern Spain. Regional Environmental Change 13, 601-614.

Page | 109

Nuss, E. T., and Tanumihardjo, S. A. (2011). Quality Protein Maize for Africa: Closing the

Protein Inadequacy Gap in Vulnerable Populations. Advances in Nutrition: An

International Review Journal 2, 217-224.

Pannell, D., and Vanclay, F. (2011). Changing Land Management: Adoption of New Practices

by Rural Landholders. CSIRO Publishing, Melbourne, Australia.

Place, F. (2009). Land Tenure and Agricultural Productivity in Africa: A Comparative Analysis

of the Economics Literature and Recent Policy Strategies and Reforms. World

Development 37, 1326-1336.

Ranum, P., Peña-Rosas, J. P., and Garcia-Casal, M. N. (2014). Global maize production,

utilization, and consumption. Annals of the New York Academy of Sciences 1312, 105-

112.

Robinson, L. W., Ericksen, P. J., Chesterman, S., and Worden, J. S. (2015). Sustainable

intensification in drylands: What resilience and vulnerability can tell us. Agricultural

Systems 135, 133-140.

Rouabhi, A., Hafsi, M., and Kebiche, M. (2014). Assessment of the farming transformation in a

rural region of Sétif province in Algeria. New Mediterranean 13, 38-46.

Rounsevell, M. D. A., Robinson, D. T., and Murray-Rust, D. (2012). From actors to agents in

socio-ecological systems models. Philosophical Transactions of the Royal Society of

Biological Sciences 367, 259-269.

Rware, H., Wairegi, L., Oduor, G., Macharia, M., Romney, D., Tarfa, B. D., de Maria, R., Ley,

G., Tetteh, F., Makumba, W., Dicko, M., Nouri, M., Leon, N. N., Gondwe, B., Kaizzi,

K., Negash, D., Kibunja, C., Alawy, A., Wartmann, C., and Rutsimba, E. (2014).

Assessing the Potential to Change Stakeholders Knowledge and Practices on Fertilizer

Recommendations in Africa. Agricultural Sciences 5, 1384-1391.

Scoones, I. (2001). "Dynamics and diversity: soil fertility and farming livelihoods in Africa: case

studies from Ethiopia, Mali, and Zimbabwe," Earthscan Publications Ltd, London.

Thierfelder, C., Mombeyarara, T., Mango, N., and Rusinamhodzi, L. (2013). Integration of

conservation agriculture in smallholder farming systems of southern Africa: identification

of key entry points. International Journal of Agricultural Sustainability 11, 317-330.

Page | 110

Tittonell, P., and Giller, K. E. (2013). When yield gaps are poverty traps: The paradigm of

ecological intensification in African smallholder agriculture. Field Crops Research 143,

76-90.

Tittonell, P., Muriuki, A., Shepherd, K. D., Mugendi, D., Kaizzi, K. C., Okeyo, J., Verchot, L.,

Coe, R., and Vanlauwe, B. (2010). The diversity of rural livelihoods and their influence

on soil fertility in agricultural systems of East Africa – A typology of smallholder farms.

Agricultural Systems 103, 83-97.

Twomlow, S., O'Neill, D., Sims, B., Ellis-Jones, J., and Jafry, T. (2002). RD—Rural

Development: An Engineering Perspective on Sustainable Smallholder Farming in

Developing Countries. Biosystems Engineering 81, 355-362.

Valbuena, D., Verburg, P. H., and Bregt, A. K. (2008). A method to define a typology for agent-

based analysis in regional land-use research. Agriculture, Ecosystems and Environment

128, 27-36.

Vanclay, F. (2004). Social principles for agricultural extension to assist in the promotion of

natural resource management. Australian Journal of Experimental Agriculture 44, 213–

222.

Vanlauwe, B., and Giller, K. E. (2006). Popular myths around soil fertility management in sub-

Saharan Africa. Agriculture, Ecosystems and Environment 116, 34-46.

Vanlauwe, B., Kihara, J., Chivenge, P., Pypers, P., Coe, R., and Six, J. (2011). Agronomic use

efficiency of N fertilizer in maize-based systems in sub-Saharan Africa within the context

of integrated soil fertility management. Plant and Soil 339, 35-50.

Wall, P. C. (2007). Tailoring Conservation Agriculture to the Needs of Small Farmers in

Developing Countries. Journal of Crop Improvement 19, 137-155.

Zimmerer, K. S., Carney, J. A., and Vanek, S. J. (2015). Sustainable smallholder intensification

in global change? Pivotal spatial interactions, gendered livelihoods, and agrobiodiversity.

Current Opinion in Environmental Sustainability 14, 49-60.

Page | 111

CHAPTER 4:

Best-fit Residue Management in Rain Fed N-Deprived Systems

This chapter was submitted for publication in the Field Crop Research Journal and the

original paper format was kept in the chapter:

Nhantumbo, N., Mortlock, M., Nyagumbo, I., Dimes, J and Rodriguez (nd) Residue and fertilizer

management strategies for the rain fed, N-deprived maize-legume cropping systems in

Central Mozambique.

Nhantumbo, N., Mortlock, M., Nyagumbo, I., Dimes, J and Rodriguez, D (2015) Improving

resource allocation in nitrogen constrained systems: acknowledging within and cross-

farm variability effect on yield and NUE. “In proceedings from the 5th International

Symposium for Farming Systems Design: Multi-functional farming systems in a

changing world”. Edited by ESA and Agropolis International, pp 491-492, 7-10

September, Montpellier, France

Page | 112

Residue and fertilizer management strategies for the rain fed, N-deprived maize-legume

cropping systems in Central Mozambique

Nascimento Nhantumbo1,2, Miranda Mortlock3, Isaiah Nyagumbo4, Daniel Rodriguez1

1 Queensland Alliance for Agriculture and Food Innovation (QAAFI), The University of Queensland PO

Box 102 Toowoomba Qld (4350) Australia; 2 Divisão de Agricultura (DivAG), Instituto Superior

Politecnico de Manica; Corresponding author: n.nhantumbo@uq.net.au; 3School of Agriculture and Food Science (SAFS), the University of Queensland; 4CIMMYT-Harare;

Abstract

Improving crop residues and nitrogen fertiliser management strategies, is critical to increase resource

productivity and yields in Sub-Saharan Africa’s N-deprived maize-legume systems. In this study we

quantified the immediate benefits and returns from the allocation of carbon rich residues and nitrogen

fertilisers between maize and cowpea crops, grown across soils of contrasting level of fertility and

moisture regime in Mozambique. Three contrasting soils: a fine sandy soil (fSa), a wetter sandy loam soil

(SaL), and a sandy clay loam soil (cSaCL) were used. The experiment was carried over two seasons. In

the first season maize and cowpea were sown under two levels of crop residue cover, i.e. 0 and 6t/ha;

three levels of N supply i.e. 0, 23 and 92 kg N/ha, representing unfertilized (ND), limited N-applications

(LNA), and high N-application (HNA) systems. Giant thatching grass was used as mulching. In the

second season a bio-assay maize crop was planted to evaluate the short term residual effects from

previous season treatments. Changes in yield, water use efficiency (WUE), nitrogen responses, nitrogen

use efficiency (NUE), and short term residual effects were analysed. Results showed that the application

of high C/N residues as mulches (i) improved cowpea yield and biomass production at all three sites; and

(ii) reduced the yield of unfertilized maize at all sites but effects were only significant on the fine sandy

soil, and sandy clay loam soil. Residue application in maize only proved beneficial under the HNA.

Residue application significantly increased WUE, NUE and nitrogen uptake efficiency (NUpE) across

residue levels in cowpea; though these efficiency indices were reduced on the maize crop. A higher soil

moisture regime in the wetter SaL soil led to higher NUE, while increased N supply reduced NUE and

increased WUE across all the sites. Results from the bioassay maize crop showed a 23.2%, 27.7% and

56.7% lower maize yield, NUpE and NUE, from plots previously under maize when residues were

applied. Results from this study suggest that there are crop and site specific responses and trade-offs that

need to be considered when promoting high C/N residues and fertilizers among resource poor smallholder

farmers of Mozambique.

Key words: Crop residues, Water use efficiency (WUE), Nitrogen use efficiency (NUE), maize-legume

systems, conservation agriculture

Page | 113

4.1 Introduction

Benefits from the adoption of conservation agriculture principles in Africa (Ngwira et al., 2013;

Thierfelder et al., 2012; Thierfelder et al., 2015) are contestable, reason current adoption levels

are low compared to the rest of the world (Baudron et al., 2015c; Giller et al., 2009; Govaerts et

al., 2009). In particular, the use of carbon rich mulches that are known to interact negatively with

plant N availability when used in N-deprived cereal cropping systems (Baudron et al., 2015c;

Giller et al., 2011a). Furthermore, data sets showing benefits from the use of C rich mulches are

usually from trials where considerable amounts of nitrogen fertilizer have been used (Ngwira et

al., 2013; Thierfelder et al., 2012; Vanlauwe et al., 2014); which does not compare to common

farmer practice. High prices, limited cash, and unreliable fertilizer responses resultant from poor

recommendations (Crawford et al., 2006; Enyong et al., 1999), have kept fertilizer application

rates in Sub Saharan Africa (SSA) extremely low, ca. 9-13 kg fertiliser/ha (Chianu et al., 2012).

This is much lower than the 50 kg fertiliser/ha target expected to be achieved by 2015 in the

Abuja declaration (Kintché et al., 2015).

An alternative to inorganic N-fertilizers in Africa, are rotations with legumes. However, legumes

are known to underperform in most SSA farming environments, producing lower yields, lower

shoot biomass and low N fixation (Giller et al., 2009). The availability of maize stubble is also

limited in regions were livestock freely roam open fields (Erenstein et al., 2015). In central

Mozambique the scarcity of residues is worsened by the presence of termites that feed on any

available source of carbon (Nyagumbo et al., 2015). As a result, incorporating locally available

grasses as an alternative to the in demand maize stubble maize is a common practice mainly in

cattle and termite dominated areas of Central Mozambique, thought poorly studied. African

grasses like Hyparrhenia rufa, Andropogon gayanus, Panicum maximum, Eragrostis ciliaris

(L.), and Penissetum glaucum are widely available in Central Mozambique and across southern

Africa where they colonize maize and sorghum fields (Vibrans et al., 2014) during undamaged

fallows.

The incorporation of alternative grasses and shrubs in CA based systems in Africa to build soil

organic matter (SOM) have been recommended in west African agroecosystems (Lahmar et al.,

Page | 114

2012). This is an important step towards tailoring CA in SSA since grasses and shrubs are

responsible for the increase in fire-mediated nitrogen losses in the tropics (Kugbe et al., 2015;

Rossiter-Rachor et al., 2008). Nevertheless, the incorporation of these carbon rich grasses in CA

based systems have detrimental effect on N-availability that still needs to be understood specially

in SSA N-deprived agroecosystems. Therefore, identifying optimum strategies to use and

allocate the available amounts of these C rich mulches, is of pivotal importance for the

sustainable intensification of agricultural production in smallholder farmers’ managed under

conservation agriculture (Tittonell et al., 2015).

The objective of this study was therefore, to quantify the short term returns and trade-offs from

the alternative allocation of carbon rich residues and the limited availability of mineral nitrogen

fertilisers, between maize and cowpea crops, grown on soils having contrasting level of moisture

regime. Responses were analysed in terms of water use (WU), nitrogen responses (NU) and

resource productivity – water use efficiency (WUE) and nitrogen use efficiency (NUE). This

study adds value to the current efforts to improve resource management in smallholder farming

systems in the SSA. Here, we hypothesize that, the best response from high C/N ratio mulches

and residues should be expected from the legumes phase of the rotation; and that limited

availabilities of nitrogen fertilisers should be primarily used to improve yields on the residue free

maize crop.

4.2. Materials and Methods

4.2.1 Field Experimentation

Field experiments were conducted at three soils of contrasting soil texture and moisture regime.

The soils were a red coarse sandy clay loam (cSaCL) Ferralsol i.e. a dry upland soil most

dominant in the region; a dark sandy loam soil (SaL) Gleysol, i.e. a wetter lowland soil, medium

to high fertility, located across the valleys and river streams; and a fine sandy soil (fSa)

Arenosol, mostly identified as dry and of low fertility soil. Plant available water content

(PAWCs) at sowing was 93 mm, 41.9 mm and 11.7 mm for the SaL, cSaCL and fSa respectively.

Initial nitrogen content was 23.6, 33.4 and 66.6 kg NO3-N/ha, in the SaL, fSa, and cSaCL soils,

Page | 115

to 1.20 m depth respectively (Table 1-1). Initial N (NO3-N and NH4-N) was determined using

NIR spectroscopy at ICRAF - Cropnuts Laboratory in Nairobi. Seasonal N-mineralization and

immobilization was monitored as extractable NO3-N at harvest of season 1 and 2 (Figure 1-3).

The colorimetric method based on Merckoquant nitrate strips - Merck KGaA from Germany was

used (Cossani et al., 2012). The SaL and fSa sites were established on-farm, while the cSaCL

was located at the ISPM research station. At each site, a fully randomized split plot factorial

(2x3x3) experiment was established in a 4x3 arrangement layout. Treatments included three

factors: crop type i.e., sole maize and sole cowpea; residue application at two levels (0 and 6

t/ha); and N-fertilization at three levels, an unfertilized N-deficient system (ND); a low N-

application rate (LNA, i.e. 23 Kg N/ha); and a high N application rate (HNA, 92 Kg N/ha). The

combination of these three factors resulted in 12 treatment replicated four times in each site. Plot

sizes were 12x4m in the first season (2013-14), and each plot was divided into two 6x4m in

2014-15 season to accommodate a bioassay maize crop planted without fertilizer application nor

residue to determine 1st season’s treatment effects. The other half was kept as in the first season.

Residues from the previous seasons where kept in the field and not incorporated in the bioassay.

The trial was dry sown between the 19th and 21st of November, in both seasons. Weeds were

brown killed using 6 l/ha glyphosate prior to sowing. Three in-crop weeding’s were conducted in

the maize, while one weeding, just before flowering, was required for the cowpea. All the plots

received 60 kg P/ha at sowing as single superphosphate, and the legume was inoculated prior to

sowing. Thatching grass straw having a 54:1 C/N ratio, was applied as mulch immediately after

sowing. The N-fertilizer was applied as Urea in a 50:50 rate at sowing and four weeks after

sowing. A short duration and self-pollinated erect cowpea variety IT16 was planted at 22.2 pl/m2

(0.45x0.20 m). A medium duration maize hybrid from PANNAR (PAN 67) was planted at 0.75

m between rows and at 0.25 m between plants, i.e. 5.3 pl/m2. An automatic weather station was

used to record daily values of rainfall, maximum and minimum temperatures, and solar radiation.

All trials were within a 2 km distance from the weather station.

Page | 116

4.2.2 Measurements

Soil water content (SWC) was determined gravimetrically in each plot at sowing and

physiological maturity at 0-15, 15-30, 30-60, 60-90 and 90-120 cm depth intervals using a soil

corer. Crop water use (WU) was estimated as the difference between soil water content at sowing

and physiological maturity, plus in crop rainfall (Aggarwal and Sinha, 1983; Cossani et al.,

2012). Losses by runoff were minimal, as the site was flat, though the method ignores water

losses by deep drainage. Plant available water content (PAWC) at each site was calculated by

correcting the SWC by a crop lower limit (CLL) (Dalgliesh and Foale, 1998; Pala et al., 2007).

Yield and dry matter production: dry matter production was determined at physiological maturity

by sampling two consecutive rows of 3 m length from the centre of each plot, i.e. 2.7 and 4.5 m2

in cowpea and maize, respectively. The same procedure was used to determine grain yield.

Resource productivity: the mass difference method was used to measure nitrogen use (NU)

(Montemurro et al., 2006). For the grain and straw, samples were collected at physiological

maturity, and total nitrogen content determined using Kjeldhal method. Nitrogen use efficiency

(NUE) was determined as the ratio between grain yield (Gw, kg/ha) and nitrogen supply (Ns, kg

N/ha), where Ns was the sum of applied inorganic N-fertilizer, aboveground plant N from the

control plots where no N fertilisers was added (Nt, kg N/ha), and the residual postharvest

mineral-N in the soil (Nsph, kg N/ha) (Huggins and Pan, 1993; Moll et al., 1982). Two NUE

components were also calculated. Nitrogen uptake efficiency (NUpE), which was determined as

Nt/Ns; and nitrogen utilization efficiency (NUtE), calculated as Gw/Nt. (Giambalvo et al., 2010).

Soil nitrogen availability (Nav) was estimated as the sum of aboveground N and postharvest

mineral N calculated as an approximation from soil nitrate levels, (Aggarwal et al., 1997; Bakht

et al., 2009). Aboveground nitrogen partitioning between stover and grain was calculated by

multiplying the percentage of nitrogen in the stover or grain, by the respective amount of

biomass of stover or grain produced. Water use efficiency (WUE, kg/ha mm-1), was calculated as

the ratio between grain yield and WU assuming 100 mm losses in soil evapotranspiration during

early crop establishment (Dalgliesh and Foale, 1998; French and Schultz, 1984).

Residual responses from residue and fertilizer treatments were calculated as the relative

difference between the yield of the maize crop in the bioassay maize crop grown in plots

previously under maize and the bioassay yield from plots previously under cowpea as described

Page | 117

in Gentry et al. (2013). The difference between those yields was defined as yield penalty from

growing a continuous maize (CMYP). Similar calculations were made for WUE, NUE, NUpE

and NUtE. Positive values indicate benefits from the cowpea-maize sequences, while negative

values indicate benefits from the continuous maize systems.

4.2.3 Statistical analysis of field experiment data sets

Analyse of variance (ANOVA) were performed using a generalized fixed linear model (RELM)

in GenStat 16 which was also used for trial design. ANOVA was conducted to test main factors

effect (residues, fertilizer, site) and their interactions on tested crops. Response variables where,

yield, water and N productivity (CWU, WUE, NUE, NUpE, NUtE). ANOVA was performed in

two stages. In the first season to test whether cowpea and maize responded differently to residue

and fertilizer application across tested sites. In the cases where different crop responses where

observed, further analysis where performed separately for each crop. Means were separated by

least significant differences (LSD) where significant differences where observed. Furthermore,

correlation analysis where conducted to explore the strength and significance of the relation

between the parameters of interest. In the second season, results from the bioassay maize crop

where analysed to measure treatment effects and identify the best fit resource residue and N-

fertilizer application across tested crops and sites. In the second season, because only 5 out of 12

treatments, i.e., 3 cowpea-maize and 2 maize-maize sequences were harvested in the drier fSa

soil residual treatment effects are only reported for the cSaCL and SaL sites where all treatments

were harvested.

4.3. Results

4.3.1 Crop growing conditions – rainfall distribution

Rainfall distribution throughout the growing season varied largely between months and was

better distributed in the first season, during the first season the trials experienced almost no

rainfall during the first 21 days after sowing (Figure 4-1). The amount of pre-sowing rainfall

Page | 118

differed between the two seasons though, during the 2013-14 season in-crop rainfall was 484 and

559 mm for the cowpea and maize, respectively. During the 2014-15 season in-crop rainfall on

the maize bioassay was 594 mm. During the first season crops experienced a short periods of

water stress at grain filling in cowpea and during early flowering of the maize. During the second

season rainfall distribution was much poorer, with 72.8% (432.9mm) falling between 9th

December to 9th January. During this season the sites experienced dry spells around flowering

and during grain filling of the maize

Figure 4-1 Rainfall distribution, daily maximum, minimum temperature and radiation from

1st September 2013 to 4th April 2014 in the 2013-14 cropping season (left) and 2014-15

cropping season (right). S is sowing date, (F) flowering and (H) harvest

Page | 119

4.3.2 Maize and cowpea yield response to residue and fertilizer application across sites

Maize and cowpea yield responses to the addition of the high C/N ratio residues differed across

sites and levels of N-applied. For cowpea, the application of high C/N ratio residues significantly

increased yields by 45.4% and 41.7% in the unfertilized (ND) cowpea of the high starting N

cSaCL and wetter SaL sites. On the drier fine sandy soil (fSa), positive responses to residue

application were also measured though the differences were not significant.

In maize, the response to residue application differed across levels of N-application (R-level x N-

level, p<0.05), and site (site, p<0.05, table 4-1). The application of high C/N ratio residues led to

significant yield losses in unfertilized (ND) maize plots, except on the wetter SaL site (Figure 4-

2). In the high starting N cSaCL site, residue application significantly reduced the yield of

unfertilized (ND) maize (33.8%), and significantly increased maize yield (44.7%) under HNA.

No significant yield increases were measured in the LNA plots (Figure 4-2). At the drier fSa site,

residue application significantly reduced the yield of the unfertilized (ND) maize (43.5%). In

contrast to the measured yield loss after the application of high C/N residues in unfertilized

maize, maize responded positively to the addition of the high C/N ratio residue for the at HNA

treatment, except at the drier sandy soil (fSa).

Cowpea and maize responded differently to nitrogen fertilisation (N-levels, p<0.05). Maize yield

responded positively to inorganic N-applications, while the yield of cowpea was reduced by N-

applications above 23 kg N/ha. The response to N-application also differed between sites (N-

level x Site, p<0.05). The largest responses were observed at the wetter SaL site. Individual

crops responses to N-applications (Figure 4-2), showed that maize’s responded to inorganic N-

applications at the high starting N but water limited cSaCL site, particularly when residues were

applied.

Page | 120

Figure 4-2 Crop response to residue application across different N-fertilization levels on

three contrasting soil types from Central Mozambique. ND: Unfertilized N-deficient system

(0 kg N/ha), LNA: low nitrogen application system (23 kg N/ha), HNA: high N-application

(92 kg N/ha). S – significant differences between residue application levels. Different letters

across N-levels for same R-level indicated significant differences at p<0.05 significance.

Caps lock – separate treatments without residues. Small letter separate situations with

residues

Table 4-1 Significance levels for the main factors (Crop, R-levels, N-levels, Site) and their interactions effects on yield, stover, crop

water use (CWU), water use efficiency (WUE), nitrogen use efficiency (NUE), nitrogen uptake efficiency (NUpE), nitrogen

utilization efficiency (NUtE) and N-partitioning into shoot N (Nsh) and Grain N (Ng), at physiologic maturity

Fixed term Yield Stover CWU WUEyld Shoot N Grain N Total N-uptake NUpE NUtE AEN

Crop ***

*** *** *** ** *** *** ***

R_level

**

**

N_level *** *** *** *** ** *** ***

Site *** *** *** *** *** *** *** *** ***

Crop x R_level

**

**

Crop x N_level ***

* ** * **

R_level x N_level *

*

*

Crop x Site *** *** *** *** *** *** *** ***

R_level x Site

*

N_level x Site ***

** *** *** * ***

Crop x R_level x N_level ** * **

** *

Crop x R_level x Site

*

*

Crop x N_level x Site ** **

**

** ***

R_level x N_level x Site

*

Crop x R_level x N_level x Site * *

*

* Significant at 0.05 probability level

** Significant at 0.01 probability level

*** Significant at 0.001 probability level

4.3.3 Resource Utilization

Crop water use (CWU)

Estimated crop water use (CWU) varied from a minimum of 384.8 mm to a maximum of 572.3

mm across crops and sites. Differences in WU across crops were due to different crop durations

and rainfall distribution, which affected the amount of in crop rainfall. However, site differences

effect on CWU were related to contrasting starting soil water contents (SWC) (R2 = 0.70,

p<0.05) and starting plant available water content (PAWC) (R2 = 0.59, p<0.05). In maize,

residue application significantly increased CWU.

Significant differences in CWU across sites were measured at both flowering and physiological

maturity. The cSaCL site, showed the highest value of CWU at flowering for both crops, and

differed significantly from the other sites. For both crops, CWU was different across levels of N-

application at all sites (N-level x Site, p<0.05). Figure 4.3 shows that the variation in CWU was

relative small compared to the variation in the yields of maize and cowpea. Figure 4-3 also

shows that in maize the maximum WUE was 13.8 kg grain/ha.mm, corresponding to the highest

levels of N supply under the use of residues at the SaL site. This value represented 77% of the

achievable WUE simulated by the APSIM model (17.9 kg grain/ha.mm) (Figure 4-3a). Figure 4b

shows that in cowpea the maximum WUE was 4.9 kg grain/ha.mm, corresponding to the SaL site

and low levels of N supply, with and without the use of residues. This value represented ca. 63%

of the achievable WUE calculated by the APSIM model (Figure 4-3b).

Figure 4-3 Maize and cowpea grain yield as a function of crop water use from emergence to maturity measured across sites. The

continuous line represents the attainable water use efficiency of 20 kg grain/ha.mm (a) and 10 kg grain/ha.mm (b), for maize and cowpea

respectively. The dotted lines represent the maximum actual WUE for each crop across the tested environments. Theoretical WUE, based

on APSIM model simulations for this agroecological zone are of 17.9 kg grain/ha.mm for maize and 7.74 kg grain/ha.mm for cowpea,

respectively.

Page | 124

Water use efficiency

Residue application significantly increased cowpea WUEyield, (Table 4-1). For maize, WUEyield

responses to residue application differed across levels of N-application (R-level * N-levels, p<

0.05). Nitrogen application significantly increased WUEyield (Table 10). For maize, the highest

WUEyield was measured under HNA and differed significantly from the other N-levels. For

cowpea, the highest WUEyield was achieved under LNA at all sites. Individually (Sites, p<0.05)

and through its interaction with N-levels (N-levels x sites, p<0.05), had the largest effect on

WUEyield. Crop WUEyield at cSaCL (9.16 kg grain/ha.mm) and SaL (8.67 kg grain/ha.mm) were

significantly higher than the values at the drier fSa site, which registered the lowest WUEyield

(Table 4-1). Analysis of individual treatment responses showed that the highest WUEyield for

maize was measured at the SaL site, in residue applied HNA (13.8 kg grain/ha.mm) plots.

However, this was not significantly different from the measured WUEyield in the cSaCL for the

same treatment. In cowpea the highest WUEyield, was observed on the sandy loam soil, i.e. 4.93

kg grain/ha.mm and differed significantly from the other two sites. Different effects from N-

application across sites were measured on WUEyield (N-level x Sites, p<0.05). In maize (Figure

4-4a), N-application increased WUEyield by an overall 76% across N-levels in the wetter SaL site.

However, no significant differences in WUEyield across N-application levels were observed for

the cSaCL and dry fSa sites. No intra-site effects were observed in cowpea (Figure 4-4b), but

clear inter-site differences in WUEyield were observed across N-levels.

Page | 125

Figure 4-4 Relationship between crop WUEyield and N-fertilizer application across sites (N-level x Site

interaction). N-levels: ND: N-deficient system (0 kg N/ha), LNA: low N-application system (23 kg N/ha)

and HNA: Non-limiting N (92 kg N/ha). Soil textures: fSa – fine sand soil, SaL – sandy loam soil and

cSaCL - coarse sandy clay loam soil. The vertical line in each plot represents the LSD value for WUEyield

comparison between N-levels for each site at p<0.05. N-levels spaced higher than the LSD bar for the

same site are significantly different.

Page | 126

Nitrogen Use Efficiency (NUE)

Residue application and N-levels differently affected NUE across crops and sites. No significant

differences were observed in maize. In cowpea, an overall 14.7% increase in NUE was

calculated for the treatments were high C/N residues were applied (Table 1). Across all sites, the

response of cowpea NUE to the use of residues was highest at the SaL site, where an overall

27.1% NUE increase was observed (R-level x Site, P<0.05). N and sites had the largest impact

on crop NUE. NUE was significantly reduced by N-applications above LNA, for both crops and

sites. For maize, 21.3 kg grain/kg N-supplied were produced under LNA; this was significantly

higher than the value under the HNA (16.2 kg grain/kg N-supplied). For cowpea the highest

NUE was 7.35 kg grain/kg N-supplied, which was observed at the LNA treatment. However,

NUE responses to N-applications differed across sites (N-level x Site, P<0.05). For maize, the

largest NUE loss from increased N-application was observed at the cSaCL site (Figure 4-5a). For

cowpea, variations in NUE as a function of the level of N-application were highest at the SaL

site. Moisture and starting N driven changes in NUE were observed. As we move from the drier

fSa to the high WHC cSaCL and wetter SaL sites NUE increased. Nevertheless, at high N-levels,

the starting high soil N-status negatively affected the values of NUE. As a result, no significant

differences between maize NUE were observed between cSaCL and SaL sites at HNA for maize

(Figure 4-5a). For cowpea (Figure 4-5b), the high starting N in cSaCL, led to poor NUE as well.

Page | 127

Relationship between NUE, NUpE and NUtE

NUE was highly related to its components in both crops. However, the magnitude the

contribution of each component to explain the variation in NUE differed between crops and sites.

In maize changes in NUE (Figure 4-6a) were primarily driven by changes in NUpE (R2 = 0.97, p

< 0.05) rather than NUtE (R2 = 0.25, p < 0.05). However, because NUE, NUpE and NUtE

differed significantly across sites (Table 4-1) the relationships between NUE and its components

also differed across sites. A split site analysis showed that for the cSaCL, NUE was sensitive to

variations in both NUpE (R2=0.72) and NUtE (R2=0.71), while at the wetter SaL, variations in

NUE were better explained by variations in NUtE, (R2=0.51). At the drier and N limited fSa site,

NUE was primarily driven by NUpE (R2=0.64), this compares with NUtE (R2=0.47). In cowpea,

overall NUE (Figure 4-6) shared a strong relation with both NUpE (R2 = 0.79, p < 0.05) and

Figure 4-5 Relationship between crops NUE and N-fertilizer application across sites. N-levels: ND: N-

deficient system (0 kg N/ha), LNA: low N-application system (23 kg N/ha) and HNA: Non-limiting N (92

kg N/ha). Soil textures: fSa – fine sand soil, SaL – sandy loam soil and cSaCL - coarse sandy clay loam

soil. Continuous line in each plot represent the LSD value for NUE comparison between N-levels for each

site at p<0.05. N-levels spaced higher than the LSD bar for the same site are significantly different.

Page | 128

NUtE (R2 = 0.82, p < 0.05). However, in cowpea NUE was more sensitive to the N-translocation

into grain (NUtE), particularly at high levels of N supply (HNA). A cross site analysis showed

that, at the SaL site, NUE was better related to NUpE (R2 = 0.58) than NUtE (R2 = 0.23). While

at the cSaCL site, NUE, was better related to NUtE (R2 = 0.94), and had almost no relation with

NUpE. At the drier, fSa, NUE was strongly related to both NUpE (R2 = 0.67) and NUtE (R2 =

0.83).

Figure 4-6 Relationship between NUE and its components observed across sites of contrasting moisture

regime. In maize as NUpE was the primary driver of NUE (a.1), while in cowpea (b.1, b.2) a, NUtE

appeared to be major driver of NUE changes. Shape identification: ∆ - Sandy Clay loam (cSaCL); O Fine

sandy (fSa) and ◊ Sandy loam soil (SaL). Empty and filled shape indicate residue application levels

Page | 129

Resource allocation trade-offs: yield and resource productivity

A nitrogen driven trade-off between NUE and WUE was identified (Figure 4-7). Increases in N-

supply tended to increase WUE in detriment of NUE. However, the magnitude of this trade-off -

positive or negative - depended on the crops differential response to N-application (Figure 4-7a,

maize, Figure 4-7b cowpea) and residue treatments. As of residues, a crop driven residue

response modified trade-off between NUE and WUE was observed in both maize and cowpea.

Results also showed the existence of a cross-site interactive relationship between NUE and

WUE, which explained most of the variation in NUE. Here, improved soil water, i.e., a move

from the drier fSa to the relatively wetter SaL site, increased NUE.

Figure 4-7 Nitrogen driven trade-offs between NUE-WUE. NUE and WUE were averaged across sites

and residue application levels for maize (a) and cowpea (b). Bellow: Maize (c) and cowpea (d) trade-

offs as affected by residue application levels. NUE and WUE for both crops were averaged across sites.

Residue application levels: R- and R+ indicate residue free and residue applied systems.

Page | 130

4.3.4. Short-term effects from residues and N-fertilizer allocation strategies

Results from the bioassay crop, demonstrated that yield and resource productivity of the

subsequent maize crop were found to differ between plots previously under cowpea and maize

(Table 4-2). Of all tested factors, site differences had the largest effects on the subsequent crop

performance. In general, higher yields and resource productivity where measured from plots

previously planted to cowpea under both residue and fertilizer application. Resource allocation

into cowpea, led to an overall, 31.4%, 17.4% and 56.8 % higher yield, NUpE and NUE, in the

subsequent maize crop compared to the applications into maize. Moreover, an overall “penalty”

of 23.2%, 27.7% and 56.7% on yield, NUpE and NUE, was calculated from residue application

into maize (Table 4-3). Nevertheless, the largest yield, NUpE and NUE differences between

sequences was attained in the SaL site. Despite the highest N-uptake levels being observed at the

cSaCL site, the wetter SaL site, showed the largest overall NUE, particularly in plots previously

under cowpea (Table 4-3). However, differences were only significant at specific treatments

(Figure 4-8). Results also show that, at specific circumstances, residue allocation into the

continuous maize sequences are likely to bring benefits to the systems.

Page | 131

Table 4- 2 Significance levels for the residual effects from residue-fertilizer allocation strategies

on yield and resource productivity of cowpea-maize and maize-maize sequences measured in the

2014-15 season

Fixed term Yield Grain N N-uptake WUEyld NUE NUpE NUtE

Previous crop ** * *** *** **

R_level

N_level *

Site *** *** *** *** *** *** ***

Crop x R-level * *

Crop x N-level

R_level x N_level

Crop x Site

R_level x Site

N_level x Site

Crop x R-level x N-level

Crop x R-level x Site ***

Crop x N-level x Site

R-level x N-level x Site

Crop x R-level x N-level x Site

* Significant at 0.05 probability level,

** Significant at 0.01 probability level,

*** Significant at 0.001 probability level

Page | 132

Figure 4-8 Short term yield, NUpE and NUE return from a maize crop grown in plots previously

planted to cowpea (Cwp-Mz) and maize (Mz-Mz) under contrasting residue and nitrogen

application levels Treatment description: ND, LNA and HNA represent unfertilized, low and

high N-application respectively. Previous residue status: with (+) and without (-) residue. (s)

represents significant differences between residual treatment effects at p<0.05.

Page | 133

Table 4- 3 Short term yield, N-uptake, NUpE, NUtE, NUE and WUE advantages from residue

and nitrogen fertilizer application into cowpea attained in a subsequent bioassay maize crop.

Results were obtained after one-year comparison between unfertilized maize grown in plots

previously planted to cowpea and maize at different residue and N-application levels.

Fixed

term

Factor

level

∆yield

(%)

∆N-uptake

(%)

∆NUpE

(%)

∆NUtE

(%)

∆NUE

(%)

∆WUE

(%)

R-levels R- 39.7 17.1 7.06 12.9 56.9 25.6

R+ 23.2 10.7 27.7 11.0 56.7 25.8

N-level ND 23.6 16.6 36.4 7.61 41.9 6.43

LNA 37.9 13.9 0.94 14.4 71.9 34.0

HNA 32.8 11.1 14.8 13.8 56.5 36.8

Sites cSaCL 34.9 3.07 4.89 24.4 81.1 29.8

SaL 28.0 24.7 29.9 -0.54 32.4 21.6

cSaCL

ND - 24.0 11.8 42.4 13.6 49.4 22.1

LNA - 131 42.1 13.3 54.7 237.5 85.2

HLA - 13.1 -9.18 -12.5 13.3 44.2 1.85

ND + -7.48 -24.4 16.5 22.2 27.4 -21.9

LNA + 6.91 -13.2 -40.1 11.5 31.6 37.1

HNA + 41.7 11.3 9.77 31.1 96.7 54.7

SaL

ND - 15.3 29.0 0.26 -13.3 -12.5 12.1

LNA - 15.6 22.8 4.09 -2.93 -0.23 7.49

HNA - 39.2 6.20 -5.22 11.9 22.8 24.9

ND + 62.7 50.2 86.4 7.93 103.1 13.4

LNA + -1.98 3.97 26.5 -5.62 18.8 6.08

HNA + 37.1 36.2 67.3 -1.19 62.4 65.7

Interpretation: Residue levels: R- is without residues and R+ with residues applied. N-levels: ND

is 0 kg N/ha, LNA is 23 kg N/ha, HNA is 92 kg N/ha; Sites: Sandy clay loam soil (cSaCL) and

Sandy loam soil (SaL). For each specific circumstance, positive variations (∆), indicate gains

from resource allocation into cowpea in the previous season, while negative variations indicate

gains from resource allocation into maize

Page | 134

4.4. Discussion

Residue responses

Negative responses to high C/N ratio residues in unfertilized maize crops were reported before

and were mostly attributed to reduced nitrogen availability due to N immobilization stress in

mulched soils (Huang et al. 2008). This is in line with results from Andraski and Bundy (2008)

who found that mulching with high C/N ratio residues reduced NO3-N early in the season,

consequently reducing maize yield in unfertilized No-till plots. In a different study, Hoyle and

Murphy (2011), observed significant reductions in plant N concentration and yield in wheat after

the incorporation of high C/N ratio (40/1) oat residues. Similar results were also reported by

(Ambus and Jensen, 2001; Wells et al., 2013), who observed a reduction in soil inorganic N and

yield of cereals in mulched soils.

Despite the fact that the N-immobilized in mulched soils is retained in the soil and is expected to

become available to plants with time (Kumar and Goh, 2002), unless enough inorganic N-

fertilizer is supplied to satisfy N-immobilization demands. However, time is a limited

commodity for resource poor farmers, to whom the use of mineral fertilisers is unaffordable.

Therefore, demonstrating potential short-term technologies to improve soil N-availability and

yields in systems where high C/N residues are used as mulches, primarily to reduce erosion or

reduce soil evaporation losses (Bescansa et al., 2006; Ward et al., 2012), is of critical

importance. Based on our results, shifting residue application from cereals into legume crops

were significant yield performance improvement was measured, is a much better resource

allocation strategy for the N-deprived maize-legume cropping systems of Central Mozambique.

The high N turnover from the legume biomass (Nsh) to the system under cowpea, is more

beneficial to the system, especially if a cereal rotation is followed. In the USA corn belt,

interrupting continuous maize sequences with a legume cycle proved effective in helping reduce

maize yield penalties from the yearly incorporation of high C/N ratio maize residues in

continuous maize systems (Morrison, 2013).

Positive responses to high C/N ratio residue application observed in cowpea (Figure 4-2), for all

tested site. Rebafka et al. (1993), measured increased nitrogen fixation and dry matter production

in groundnut grown under 4 t/ha of millet straw in Niger per year. These response can be

Page | 135

explained by the legume ability to biologically fix their own N (Sprent et al., 2010), which

protects the crop from the severe N-immobilization induced plant N-stress. Positive responses to

residue application in legume crops were measured in other studies (Herrmann et al., 2014;

Krishna, 2001; Shah et al., 2003a) The positive responses to residue application in cowpea,

especially in unfertilized (ND) and low N-application (LNA) systems are an important result to

improve N-availability in the N-deprived maize-legume cropping systems managed under

conservation agriculture in Central Mozambique and SSA. This shift in resource allocation

already proved to provide better short term returns to subsequent cereal crops compared to

continuous maize systems (Homann-Kee Tui et al., 2015). In our trial, applying residues into

cowpea led to an overall 23.2% more yield in cowpea-maize sequences compared to continuous

maize under residue application.

Nitrogen fertilizer responses

Sites differences in starting mineral N and in-crop soil moisture regimes were detrimental to

observed differences N-fertilizer responses (N-level x Site, p<0.05). In the specific case of

maize, higher and more significant yield increments from increased N-application were

measured in the wetter SaL site, (Figure 4-2), compared to the high starting N cSaCL and drier

fSa were not enough soil water was available for the crop to use all available N. Water deficit

conditioned crop N responses as in our trial were also reported for several crops including maize

(Lemaire and Gastal, 2009; Pandey et al., 2000). Furthermore, cross-site variability effect on

resource allocation responses were also reported across SSA (Giller et al., 2006; Tittonell et al.,

2007b), strengthening the idea that there are site-specific responses that need to be considered to

improve resource allocation strategies in the N-deprived maize-legume systems of central

Mozambique.

The low N responses in the high starting N cSaCL for this AEZ-R4 is a critical result to raise

awareness on fertilizer recommendations in such unreliable rainfall environments (Dimes, 2011).

The measured poor response can be attributed to poor N-translocation into grain (NUtE)

resulting from lower water availability to match the nitrogen availability (Sadras, 2004). The

negative response to high N-application measured in the cSaCL and fSa especially in the absence

of residues, is also in line with findings from Cossani et al. (2010). They proposed that for rain

Page | 136

fed conditions N-availability needs to match water supply for the crop to maximize its grain

yield for a given environment. For our trail condition, a better match between N-supply and soil

water availability was only achieved in the wet SaL site which had the highest topsoil in-crop

moisture regime (Figure 1-2), consequently the highest NUpE and NUtE (Figure 4-6). In the

drier and loose fSa, poor N-responses can be partly explained by poor NUpE (Figure 4-6a). The

strong interactive relationship between water availability and NUE as reported in (Albrizio et al.,

2010; Teixeira et al., 2014), also explains how moisture regimes conditioned nitrogen responses

across our tested sites. Similar result was reported in Sun et al. (2012), while studying the effects

of different water and nitrogen management practices on yield and NUE of rice in China. They

found that, increased water availability improved NUE and rice yield compared to dry land rice,

were water deficit led to poor NUE. Despite being conducted under supplementary irrigation,

similarities can be drawn with central Mozambique were there are clear moisture driven cross-

site variability within and across agro-ecologies, which conditions resource responses and need

to be considered when recommending fertilizer management among resource poor smallholder

farmers of central Mozambique.

Resource productivity and nitrogen driven WUE - NUE trade-offs

Inorganic N-fertilizer applications above 23 kg N/ha significantly reduced NUE and its

components NUpE, NUtE. Decreases in NUE and its components with increased N-application

have been reported before (Giambalvo et al., 2010; Sadras and Rodriguez, 2010). NUE decreases

above such low levels of N-application clearly demonstrates how risky and unprofitable (Dimes,

2011) applying high amounts of inorganic N-fertilizer can be in unreliable rainfall environments.

Measured crop WUE was considerably lower than the levels proposed by Shah et al. (2003b) for

rain fed conditions i.e., 20 kg grain/ha.mm and 10 kg grain/ha.mm for maize and legume crops

grown respectively (Figure 4-3). It can be speculated that, poor rainfall distribution and the

occurrence of dry spells during early flowering and grain filling especially in the second season

negatively affected crop WUE. Rainfall distribution was reported to affect WUE in China (Sun

et al., 2010). Wang et al. (2010a) calculated lower maize WUE in dryland environments with

regular dry spells during tasselling. Despite, the observed low WUE, increased N-application

proved to increase WUE, as also reported by (Sun et al., 2012).

Page | 137

A nitrogen driven trade-off between NUE and WUE was also observed as in (Gong et al., 2011;

Sadras and Rodriguez, 2010). For both maize and cowpea, increased N-application reduced NUE

in detriment of WUE, whereas increased water availability reduced WUE in detriment of NUE

(Figure 4-7). Nevertheless, the trade-off was modified by residue application in both crops. The

direction and magnitude of these changes, depended on the individual crop and site response to

both nitrogen and residue application. Understanding the underlying factor behind the

modification of such important trade-offs and how these variations can be used for the benefit of

the system is fundamental to guide the design of improved and sustainable site and crop-specific

resource allocation recommendations for the N-deprived maize-legume systems of central

Mozambique.

Demonstrating immediate returns from alternative resource allocation strategies

Results from the bioassay crop, showed that for the N-deprived maize-legume cropping systems

of central Mozambique, the highest net returns from both high C/N ratio residues and N-fertilizer

application were achieved with applications into cowpea rather than maize. This is in line with

findings from Bakht et al. (2009), who reported a strong carry-over effect in legume-cereal

sequences as a key factor to improve N-availability and reduce N-losses in residue applied fields

(Ambus and Jensen, 2001). The high return measured in the cowpea-sequence, especially in ND

and LNA systems where losses are expected from residue applied maize, shows how allocation

the limited available resources into improving the performance of a sole legume crop is a better

and more pragmatic strategy (Kirkegaard et al., 2014), to improve N-deprived yields of the

following cereal crop in the N-deprived systems of Central Mozambique. This shift in resource

allocation is in part sustained on the premise that, with current low fertilizer use in SSA (Gilbert,

2012; Heffer and Prud'homme, 2011), resource poor farmers are less likely to afford the amounts

of N-fertilizers required to limit N-immobilization from yearly incorporation of high C/N

residues in continuous maize cropping systems (Morrison, 2013). In the USA corn belt, legumes

have also proved successful in reducing the economic optimum N-rates and improve yields in

wheat and maize systems under residue retention (Luce et al., 2015; Morrison, 2013). These

results strengthen our proposition that, in N-deprived maize-legume cropping system of central

Mozambique, investing the limited resources, i.e., residues and N-fertilizers to improve crop

Page | 138

performance during the legume phase of the rotation is likely to yield better returns to the

system.

The high N returns from previous crop and NUpE in the subsequent maize crop measured in the

cowpea-maize sequence, are in part resultant from the significant difference in N turnover from

biomass registered across sequences. In plots previously grown to cowpea, an average 122.7 kg

N/ha were returned to the soil, against 30.2 kg N/ha under maize. Because, in the continuous

maize sequence, N is still trapped into its high C/N ratio biomass, whose decomposition requires

further N to satisfy N-immobilization demands (Huang et al., 2008), yields penalties from

retained maize residue were measured in the subsequent season. Reversely, better residual

returns were attained in the cowpea-maize sequence where N from decomposing legume residues

were easily available through mineralized N (Kumar and Goh, 2002). For that reason, NUpE,

NUE and yields in the following season, were considerably higher in the cowpea-maize.

4.6. Conclusions

Results from this study, indicate that there are crop and site specific responses and trade-offs to

residue and N-fertilizer allocation that need to be considered in order to improve resource

productivity and yields in the rain fed N-deprived maize-legume cropping systems of central

Mozambique. Trial results clearly demonstrated that, N-fertilizers are better off, if used to

improve maize yields in residue free systems while legumes are a better match to residues. Better

returns from a combined application of available high C/N ratio residues and limited amounts of

inorganic N-fertilizers into the legume phase of the rotation were also measured. This shift in

practice proved beneficial in improving N-availability and yields of the following maize crop in

cowpea-maize sequences, contrary to the continuous maize systems yield losses from residue

application were registered. A nitrogen driven trade-off between NUE-WUE, modified by

residue and N-responses across crops and sites, was also demonstrated in our results. Therefore,

understanding how these trade-offs occurs in key to the design of feasible and locally tailored

resource management recommendations that are likely to help amplify desired responses or

alleviate unwanted ones in order to improve resource productivity and yields. To conclude, our

study results are of critical importance to SSA N-deprived maize-legume systems managed under

conservation agriculture, where demonstrating potential short term benefits of widely promoted

Page | 139

technologies such as residue retention, N-application and legume rotations through the

promotion of smart and pragmatic resource allocation strategies across crops and farming

environments is critical to easy a more conscious and effective adoption process. Nevertheless,

more research is required to create generalizable and robust information that can help understand

under what specific circumstances, i.e., farming environments, management and rain fall

patterns, benefits and penalties from residues are likely to be observed for a specific agro-

ecologies.

Acknowledgements

We would like to thank the Sustainable Intensification of Maize and Legume farming systems

for Eastern and Southern Africa (SIMLESA) program funded by the Australian Centre for

International Agriculture Research (ACIAR). ACIAR also awarded the PhD scholarship under

which this research was conducted. The AGRA Soil Health Program through the 2011SHP007

Project partially funded the fieldwork. We also thank Jacinto Andreque, Clemente Zivale,

Manuel Chirindzane, Cristovão D. Maquina, Eliseu D. Maquina and Ricardo Maquina for their

help with the fieldwork. We particularly acknowledge the support from SIMLESA’s

Mozambique team and country coordinator Mr Domingos Dias.

References

Aggarwal, P. K., and Sinha, S. K. (1983). Water stress and water-use efficiency in field grown

wheat: A comparison of its efficiency with that of C4 plants. Agricultural Meteorology

29, 159-167.

Aggarwal, R. K., Praveen, K., and Power, J. F. (1997). Use of crop residue and manure to

conserve water and enhance nutrient availability and pearl millet yields in an arid tropical

region. Soil and Tillage Research 41, 43-51.

Albrizio, R., Todorovic, M., Matic, T., and Stellacci, A. M. (2010). Comparing the interactive

effects of water and nitrogen on durum wheat and barley grown in a Mediterranean

environment. Field Crops Research 115, 179-190.

Page | 140

Ambus, P., and Jensen, E. S. (2001). Crop residue management strategies to reduce N-losses-

interaction with crop N-supply. Communications in Soil Science and Plant Analysis 32,

981-996.

Andraski, T. W., and Bundy, L. G. (2008). Corn Residue and Nitrogen Source Effects on

Nitrogen Availability in No-Till Corn All rights reserved. No part of this periodical may

be reproduced or transmitted in any form or by any means, electronic or mechanical,

including photocopying, recording, or any information storage and retrieval system,

without permission in writing from the publisher. Agronomy Journal 100, 1274-1279.

Bakht, J., Shafi, M., Jan, M. T., and Shah, Z. (2009). Influence of crop residue management,

cropping system and N fertilizer on soil N and C dynamics and sustainable wheat

(Triticum aestivum L.) production. Soil and Tillage Research 104, 233-240.

Baudron, F., Thierfelder, C., Nyagumbo, I., and Gérard, B. (2015). Where to Target

Conservation Agriculture for African Smallholders? How to Overcome Challenges

Associated with its Implementation? Experience from Eastern and Southern Africa.

Environments 2, 338.

Bescansa, P., Imaz, M. J., Virto, I., Enrique, A., and Hoogmoed, W. B. (2006). Soil water

retention as affected by tillage and residue management in semiarid Spain. Soil and

Tillage Research 87, 19-27.

Chianu, J., Chianu, J., and Mairura, F. (2012). Mineral fertilizers in the farming systems of sub-

Saharan Africa. A review. Agronomy for Sustainable Development 32, 545-566.

Cossani, C. M., Slafer, G. A., and Savin, R. (2010). Co-limitation of nitrogen and water, and

yield and resource-use efficiencies of wheat and barley. Crop and Pasture Science 61,

844-851.

Cossani, C. M., Slafer, G. A., and Savin, R. (2012). Nitrogen and water use efficiencies of wheat

and barley under a Mediterranean environment in Catalonia. Field Crops Research 128,

109-118.

Crawford, E. W., Jayne, T. S., and Kelly, V. A. (2006). Alternative Approaches for Promoting

Fertilizer Use in Africa. pp. 88. World Bank, Washington, USA.

Dalgliesh, N., and Foale, M. (1998). "Soil matters: Monitoring soil water and nutrients in

dryland farming," 1/Ed. CSIRO, Australia, Toowoomba, Australia.

Page | 141

Dimes, J. (2011). Formulating Crop Management Options for Africa’s Drought-Prone Regions:

Taking Account of Rainfall Risk Using Modeling. In “Innovations as Key to the Green

Revolution in Africa.” (A. Bationo, B. Waswa, J. M. M. Okeyo, F. Maina and J. M.

Kihara, eds.), pp. 785-794. Springer Science and Business Media, Netherlands.

Enyong, L. A., Debrah, S. K., and Bationo, A. (1999). Farmers' perceptions and attitudes towards

introduced soil-fertility enhancing technologies in western Africa. Nutrient Cycling in

Agroecosystems 53, 177-187.

Erenstein, O., Gérard, B., and Tittonell, P. (2015). Biomass use trade-offs in cereal cropping

systems in the developing world: Overview. Agricultural Systems 134, 1-5.

French, R., and Schultz, J. (1984). Water use efficiency of wheat in a Mediterranean-type

environment. II. some limitations to efficiency. Australian Journal of Agricultural

Research 35, 765-775.

Gentry, L. F., Ruffo, M. L., and Below, F. E. (2013). Identifying Factors Controlling the

Continuous Corn Yield Penalty. Agronomy Journal 105, 295-303.

Giambalvo, D., Ruisi, P., Di Miceli, G., Frenda, A. S., and Amato, G. (2010). Nitrogen Use

Efficiency and Nitrogen Fertilizer Recovery of Durum Wheat Genotypes as Affected by

Interspecific Competition All rights reserved. No part of this periodical may be

reproduced or transmitted in any form or by any means, electronic or mechanical,

including photocopying, recording, or any information storage and retrieval system,

without permission in writing from the publisher. Agronomy Journal 102, 707-715.

Gilbert, N. (2012). Dirt Poor: The key to tackle hunger in Africa is enriching its Soils. Nature

483, 525-227.

Giller, K. E., Corbeels, M., Nyamangara, J., Triomphe, B., Affholder, F., Scopel, E., and

Tittonell, P. (2011). A research agenda to explore the role of conservation agriculture in

African smallholder farming systems. Field Crops Research 124, 468-472.

Giller, K. E., Rowe, E. C., de Ridder, N., and van Keulen, H. (2006). Resource use dynamics and

interactions in the tropics: Scaling up in space and time. Agricultural Systems 88, 8-27.

Giller, K. E., Witter, E., Corbeels, M., and Tittonell, P. (2009). Conservation agriculture and

smallholder farming in Africa: The heretics’ view. Field Crops Research 114, 23-34.

Page | 142

Gong, X. Y., Chen, Q., Lin, S., and Brueck, H. (2011). Tradeoffs between nitrogen- and water-

use efficiency in dominant species of the semiarid steppe of Inner Mongolia. Plant and

soil 340, 227-238.

Govaerts, B., Verhulst, N., Castellanos-Navarrete, A., Sayre, K. D., Dixon, J., and Dendooven,

L. (2009). Conservation Agriculture and Soil Carbon Sequestration: Between Myth and

Farmer Reality. Critical Reviews in Plant Sciences 28, 97-122.

Heffer, P., and Prud'homme, M. (2011). "Fertilizer Outlook 2011-2015." IFA, Paris, France.

Herrmann, L., Chotte, J. L., Thuita, M., and Lesueur, D. (2014). Effects of cropping systems,

maize residues application and N fertilization on promiscuous soybean yields and

diversity of native rhizobia in Central Kenya. Pedobiologia 57, 75-85.

Homann-Kee Tui, S., Valbuena, D., Masikati, P., Descheemaeker, K., Nyamangara, J.,

Claessens, L., Erenstein, O., van Rooyen, A., and Nkomboni, D. (2015). Economic trade-

offs of biomass use in crop-livestock systems: Exploring more sustainable options in

semi-arid Zimbabwe. Agricultural Systems 134, 48-60.

Hoyle, F. C., and Murphy, D. V. (2011). Influence of organic residues and soil incorporation on

temporal measures of microbial biomass and plant available nitrogen. Plant and Soil 347,

53-64.

Huang, Z., Xu, Z., Blumfield, T., Chen, C., and Bubb, K. (2008). Soil nitrogen mineralization

and fate of (15NH4)2SO4 in field-incubated soil in a hardwood plantation of subtropical

Australia: the effect of mulching. Journal of Soils and Sediments 8, 389-397.

Huggins, D. R., and Pan, W. L. (1993). Nitrogen Efficiency Component Analysis: An Evaluation

of Cropping System Differences in Productivity. Agronomy Journal 85, 898-905.

Kintché, K., Guibert, H., Sogbedji, J. M., Levêque, J., Bonfoh, B., and Tittonell, P. (2015).

Long-term mineral fertiliser use and maize residue incorporation do not compensate for

carbon and nutrient losses from a Ferralsol under continuous maize–cotton cropping.

Field Crops Research 184, 192-200.

Kirkegaard, J. A., Conyers, M. K., Hunt, J. R., Kirkby, C. A., Watt, M., and Rebetzke, G. J.

(2014). Sense and nonsense in conservation agriculture: Principles, pragmatism and

productivity in Australian mixed farming systems. Agriculture, Ecosystems and

Environment 187, 133-145.

Page | 143

Krishna, N. R. (2001). Effects of Cereal and Legume Cover Crop Residues on Weeds, Yield, and

Net Return in Soybean (Glycine max). Weed Technology 15, 660-668.

Kugbe, J., Fosu, M., and Vlek, P. L. G. (2015). Impact of season, fuel load and vegetation cover

on fire mediated nutrient losses across savanna agro-ecosystems: the case of northern

Ghana. Nutrient Cycling in Agroecosystems 102, 113-136.

Kumar, K., and Goh, K. M. (2002). Management practices of antecedent leguminous and non-

leguminous crop residues in relation to winter wheat yields, nitrogen uptake, soil nitrogen

mineralization and simple nitrogen balance. European Journal of Agronomy 16, 295-308.

Lahmar, R., Bationo, B. A., Dan Lamso, N., Guéro, Y., and Tittonell, P. (2012). Tailoring

conservation agriculture technologies to West Africa semi-arid zones: Building on

traditional local practices for soil restoration. Field Crops Research 132, 158-167.

Lemaire, G., and Gastal, F. (2009). Chapter 8 - Quantifying Crop Responses to Nitrogen

Deficiency and Avenues to Improve Nitrogen Use Efficiency. In "Crop Physiology" (V.

Sadras and D. Calderini, eds.), pp. 171-211. Academic Press, San Diego, USA.

Moll, R. H., Kamprath, E. J., and Jackson, W. A. (1982). Analysis and Interpretation of Factors

Which Contribute to Efficiency of Nitrogen Utilization1. Agronomy Journal 74, 562-564.

Montemurro, F., Maiorana, M., Ferri, D., and Convertini, G. (2006). Nitrogen indicators, uptake

and utilization efficiency in a maize and barley rotation cropped at different levels and

sources of N fertilization. Field Crops Research 99, 114-124.

Morrison, L. (2013). Residue management can reduce continuous corn yield penalty. Corn and

Soybean Digest. Vol. 75 Issue 9, p 40

Ngwira, A. R., Thierfelder, C., and Lambert, D. M. (2013). Conservation agriculture systems for

Malawian smallholder farmers: long-term effects on crop productivity, profitability and

soil quality. Renewable Agriculture and Food Systems 28, 350-363.

Pala, M., Ryan, J., Zhang, H., Singh, M., and Harris, H. C. (2007). Water-use efficiency of

wheat-based rotation systems in a Mediterranean environment. Agricultural Water

Management 93, 136-144.

Pandey, R. K., Maranville, J. W., and Admou, A. (2000). Deficit irrigation and nitrogen effects

on maize in a Sahelian environment: I. Grain yield and yield components. Agricultural

Water Management 46, 1-13.

Page | 144

Rebafka, F. P., Ndunguru, B. J., and Marschner, H. (1993). Crop residue application increases

nitrogen fixation and dry matter production in groundnut (Arachis hypogaea L.) grown

on an acid sandy soil in Niger, West Africa. Plant and Soil 150, 213-222.

Rossiter-Rachor, N. A., Setterfield, S. A., Douglas, M. M., Hutley, L. B., and Cook, G. D.

(2008). Andropogon gayanus (Gamba Grass) Invasion Increases Fire-mediated Nitrogen

Losses in the Tropical Savannas of Northern Australia. Ecosystems 11, 77-88.

Sadras, V. O. (2004). Yield and water-use efficiency of water- and nitrogen-stressed wheat crops

increase with degree of co-limitation. European Journal of Agronomy 21, 455-464.

Sadras, V. O., and Angus, J. F. (2006). Benchmarking water-use efficiency of rain fed wheat in

dry environments. Australian Journal of Agricultural Research 57, 847-856.

Sadras, V. O., and Rodriguez, D. (2010). Modelling the nitrogen-driven trade-off between

nitrogen utilisation efficiency and water use efficiency of wheat in eastern Australia.

Field Crops Research 118, 297-305.

Shah, Z., Shah, S. H., Peoples, M. B., Schwenke, G. D., and Herridge, D. F. (2003). Crop residue

and fertiliser N effects on nitrogen fixation and yields of legume-cereal rotations and soil

organic fertility. Field Crops Research 83, 1-11.

Sprent, J. I., Odee, D. W., and Dakora, F. D. (2010). African legumes: a vital but under-utilized

resource. Journal of Experimental Botany 61, 1257-65.

St. Luce, M., Grant, C. A., Zebarth, B. J., Ziadi, N., O’Donovan, J. T., Blackshaw, R. E., Harker,

K. N., Johnson, E. N., Gan, Y., Lafond, G. P., May, W. E., Khakbazan, M., and Smith, E.

G. (2015). Legumes can reduce economic optimum nitrogen rates and increase yields in a

wheat–canola cropping sequence in western Canada. Field Crops Research 179, 12-25.

Sun, H., Shen, Y., Yu, Q., Flerchinger, G. N., Zhang, Y., Liu, C., and Zhang, X. (2010). Effect of

precipitation change on water balance and WUE of the winter wheat–summer maize

rotation in the North China Plain. Agricultural Water Management 97, 1139-1145.

Sun, Y., Ma, J., Sun, Y., Xu, H., Yang, Z., Liu, S., Jia, X., and Zheng, H. (2012). The effects of

different water and nitrogen managements on yield and nitrogen use efficiency in hybrid

rice of China. Field Crops Research 127, 85-98.

Teixeira, E. I., George, M., Herreman, T., Brown, H., Fletcher, A., Chakwizira, E., de Ruiter, J.,

Maley, S., and Noble, A. (2014). The impact of water and nitrogen limitation on maize

Page | 145

biomass and resource-use efficiencies for radiation, water and nitrogen. Field Crops

Research 168, 109-118.

Thierfelder, C., Cheesman, S., and Rusinamhodzi, L. (2012). A comparative analysis of

conservation agriculture systems: Benefits and challenges of rotations and intercropping

in Zimbabwe. Field Crops Research 137, 237-250.

Thierfelder, C., Matemba-Mutasa, R., and Rusinamhodzi, L. (2015). Yield response of maize

(Zea mays L.) to conservation agriculture cropping system in Southern Africa. Soil and

Tillage Research 146, Part B, 230-242.

Tittonell, P., Gérard, B., and Erenstein, O. (2015). Tradeoffs around crop residue biomass in

smallholder crop-livestock systems – What’s next? Agricultural Systems 134, 119-128.

Tittonell, P., Zingore, S., van Wijk, M. T., Corbeels, M., and Giller, K. E. (2007). Nutrient use

efficiencies and crop responses to N, P and manure applications in Zimbabwean soils:

Exploring management strategies across soil fertility gradients. Field Crops Research

100, 348-368.

Vanlauwe, B., Wendt, J., Giller, K. E., Corbeels, M., Gerard, B., and Nolte, C. (2014). A fourth

principle is required to define Conservation Agriculture in sub-Saharan Africa: The

appropriate use of fertilizer to enhance crop productivity. Field Crops Research 155, 10-

13.

Vibrans, H., García-Moya, E., Clayton, D., and Sánchez-Ken, J. G. (2014). Hyparrhenia

variabilis and Hyparrhenia cymbaria (Poaceae): New for the Americas, Successful in

Mexico. Invasive Plant Science and Management 7, 222-228.

Wang, X., Dai, K., Wang, Y., Zhang, X., Zhao, Q., Wu, X., Cai, D., Hoogmoed, W. B., and

Oenema, O. (2010). Nutrient management adaptation for dryland maize yields and water

use efficiency to long-term rainfall variability in China. Agricultural Water Management

97, 1344-1350.

Ward, P. R., Flower, K. C., Cordingley, N., Weeks, C., and Micin, S. F. (2012). Soil water

balance with cover crops and conservation agriculture in a Mediterranean climate. Field

Crops Research 132, 33-39.

Wells, M. S., Reberg-Horton, S. C., Smith, A. N., and Grossman, J. M. (2013). The Reduction of

Plant-Available Nitrogen by Cover Crop Mulches and Subsequent Effects on Soybean

Performance and Weed Interference. Agronomy Journal 105, 539-545.

Page | 146

CHAPTER 5:

Pathways for Sustainable Residue Management

Page | 147

Model assisted decision support frameworks for residue management in N-deprived maize-

legume cropping systems

Nhantumbo, Nascimento12., Mortlock, M 3, Nyagumbo, I4., and Rodriguez, D1 1 Queensland Alliance for Agriculture and Food Innovation (QAAFI), the University of Queensland PO

Box 102 Toowoomba Qld (4350) Australia; 2 Divisão de Agricultura, Instituto Superior Politecnico de

Manica; Corresponding author: n.nhantumbo@uqconnect.edu.au; 3School of Agriculture and Food Science (SAFS), the University of Queensland; 4CIMMYT-Harare

Abstract

Quality information is needed to support decision making on the use of limited resources in low

external N-input agricultural systems (LENIA) to improve yields and farmers’ profits,

particularly when high C/N residues are being promoted as mulches in conservation agriculture

(CA) systems. In this study, field trials and the Agricultural Production Systems Simulator

(APSIM) were used to research under what circumstances, i.e., crop type, farming environments,

and N-fertilizer regimes, the addition of high C/N ratio crop residues in CA systems is likely to

be beneficial in terms of crop yields and resource productivity. Results showed that the use of

high C/N ratio residues in LENIA systems managed under CA produced two responses 1) a crop

and soil type mediated residue effect on yield, water use efficiency (WUE) and nitrogen use

efficiency (NUE); 2) a residue driven nitrogen trade-off that was different between N fixing

crops (cowpea) and maize. In soils of high water holding capacity (WHC), gains in crop WUE

are likely to be outperformed by yield losses due to poor NUE caused by N-immobilization

during residue decomposition. Simulations showed a 90% chance of maize yield losses when

high C/N ratio residues are applied. However, cowpea responded positively to this type of

mulches. For dry and low WHC soils, moisture benefits from the application of high C/N residue

mulches were higher than any yield losses due to N-immobilization, and positive yield responses

from residue application were simulated. Results from this study shows the value of using

simulation modelling in combination with field experimentation to assess the outcome of

complex and dynamic processes to generate residue management recommendations.

Key words: APSIM, Conservation agriculture, residues, maize, Mozambique,

Page | 148

5.1 Introduction

Across most Sub-Saharan Africa (SSA) smallholder farming systems, improving the allocation

of locally available resources such as high C/N ratio residues, manure and limited amounts of

inorganic fertilizers, is critical to improve resource productivity and overall system performance.

However, in SSA low external N-input agricultural systems (LENIA) managed under

conservation agriculture, the role of high C/N ratio residue mulches has generated highly

contrasting views (Giller et al., 2009; Sommer et al., 2014; Vanlauwe et al., 2014). In part, most

of the discussion surrounding residues and fertilizer use in smallholder conservation agriculture

(CA) farming systems in Africa is related to the perceived need to produce generalizable

information easy to scale across regions, and highly contrasting farming situations.

To date, benefits from CA principles have been reported for specific situations (Baudron et al.,

2015c). Therefore, understanding the interactive relationships between the components of CA

system mainly crop residue mulches, fertilizers, crops and their management effect on yields and

resource productivity across farming environments (Giller et al., 2011a), is of extreme

importance to tailor CA principles to SSA diverse and complex farming circumstances.

Here we propose that understanding under what circumstances the use of high C/N ratio residues

as mulches is likely to increase or reduce yields, can help design more productive CA systems.

The use of dynamic cropping systems models can help quantify and untangle some of the

complexities in SSA smallholder CA system (Bouma and Jones, 2001; Jordan et al., 1997;

Maxwell, 1986). In Australia, for instance, the Agricultural Production Systems Simulator

(APSIM) (Holzworth et al., 2014) helped researchers identify the most suitable and remunerative

cropping systems for different climatic and socio-economic contexts (Cox et al., 2010; Meinke et

al., 2001; Rodriguez et al., 2011). Moreover, the integration of simulation modelling in research

programs across the world have helped to (1) identify gaps in existing knowledge; (2) generate

and test hypothesis to improve the design of experiments; (3) determine the most influential

parameters of a system (Matthews and Stephens, 2002b); and (4) bring researchers, experiments

and farmers together to discuss problems and identify solutions in participatory modelling

exercises (Whitbread et al., 2010).

Page | 149

Despite the attention that agricultural simulation models experienced in developed countries,

there has been little use in the developing world, were public expenditure in agricultural research

and development is the lowest (Beintema and Stads, 2008). Nevertheless, cost effective

approaches like agronomic models could help define attainable research outputs, and facilitate

the selection of suitable technologies to be tested in farmers’ fields. Among several models,

APSIM has proven valuable in Africa (Chikowo et al., 2008; Masikati et al., 2014; Mupangwa et

al., 2011; Ncube et al., 2009). The success of APSIM in modelling farming systems in Africa,

comes mainly as a result of integrating a systems approach (Whitbread et al., 2010) that allows it

to capture and analyse GxExM interactions (Holzworth et al., 2014), relevant to improving

research design and interpretation of African smallholder farming systems.

In this study, we used APSIM v7.7 to model the observed yield responses to crop residue mulch

management strategies and its linkages with water use efficiency (WUE) and nitrogen use

efficiencies (NUE) in maize-legume systems. The main objective of this study was to develop

simple rules of thumb to guide the allocation of high C/N ratio mulches between legumes and

cereals in LENIA systems managed under CA.

5.1.1 Conceptual framework

In the last decades, all alone in SSA, vast research have been conducted to demonstrate the

potential benefits of the conservation agriculture principles, of which residues retention are an

integral part (Ngwira et al., 2013; Thierfelder et al., 2012; Thierfelder et al., 2015; Thierfelder et

al., 2013). Fertilizer research has also deserved its share of attention (Gilbert, 2012; Ncube et al.,

2007; Twomlow et al., 2010). Nevertheless, most of these research was conducted under specific

trial conditions and focused in demonstrating the contribution of individual CA principals rather

than their interaction. Contrary to models, field trials by themselves have been so far unable to:

1) capture and explain the complex GxExM interactions and 2) explore the dynamics of WUE

and NUE which are of critical importance to generate relevant recommendation on the use of

high C/N residues and fertilizes that are relevant to mostly N-deprived SSA farming systems.

Therefore, results generated to date are of local interest with less generalizable value (Sadras and

Lemaire, 2014). To close this information gap, an APSIM based decision support framework for

residue management is proposed. The framework makes use of pre-established knowledge on

Page | 150

residue-fertilizer interactions to underpin under which circumstances benefits and losses from

alternative residues management strategies are likely to occur. To validate the framework, the

following operational principles were assumed:

1. In rain fed N-deprived systems, maize yield increases under residue application, are

likely to be attained at specific circumstances, i.e., crops, soils, N-availability levels, and

season types. Benefits from residues application are attained only when residue induced

crop WUE increments surpass yield losses from residue induced NUE decrease due to N-

immobilization.

2. There will always be a nitrogen driven trade-off between NUE and WUE (Sadras and

Rodriguez, 2010) whose magnitude can be alleviated or worsened by the amount and

type of mulch used for each crop type and fertilizer application level.

3. The balance between N-mineralization and immobilization during residue decomposition

is affected by the soils moisture regimes and N-supply to the system (McCown et al.,

1996; Probert et al., 1998). Therefore, wet and low N environments are expected to

provide a quick residue decomposition, and consequently higher chances of N-

immobilization induced plant N-stress than water limited environments.

Four propositions (rules of thumb) are used as background for the framework proposed in this

study:

1. In wet and high WHC soils, the use of high C/N ratio residues to improve the

performance of the sole legume crop is a better strategy than applying those residues on

the maize crop.

2. In dryer and low WHC soils, maize yields are likely to be increased by residue

application as a direct result of improved rainfall capture.

3. For wet and high WHC soils, the beneficial effects of applying crop residues on the

legume crop are likely to be observed on the following cereal crop.

4. The balance between the positive and negative effects of residue application on yield,

grain water productivity, and the temporary nitrogen immobilization in the rain fed

LENIA systems, can be managed with the use of nitrogen fertilisers and improved

legume-cereal sequences.

Page | 151

5.2. Materials and Methods

5.2.1 Experiment for model parameterization

The APSIM model was parameterised using data from field experiments conducted in central

Mozambique designed to evaluate the effects of different residue and fertilizer allocation

strategies in maize legume cropping systems. APSIM was parameterised for three contrasting

environments, namely: 1) a red ferralsol with sandy clay loam (cSaCL) texture representing an

upland farming environment with high starting soil N and high WHC; 2) a fine sandy (fSa)

textured Arenosol representing a dry and low WHC upland farming environment; 3) a dark

sandy loam (SaL) textured Gleysol representing a lowland farming environment with low

starting N and prolonged wetness, i.e., high in crop soil moisture regime. Soils were classified

based on the FAO-Unesco soil classification system.

In each environment twelve treatments were simulated, resulting from the combination of three

factors. Factors included crop type i.e., sole maize and sole cowpea, two levels of residue

application (0 and 6 t/ha); and three levels of N-fertilizer application - 0, 23 and 92 kg N/ha

representing an unfertilized N-deficient system (ND), a low N-application (LNA) and a high N

application system (HNA), respectively. The 6 t/ha residue application rate, was selected to

mimic an environment with a high N-immobilization risk due to continuous incorporation of

carbon rich maize stubble and grasses characteristic of central Mozambique LENIA systems.

This value is proposed on ex-ante model simulations for Central Mozambique which showed an

80-75% chance of achieving a 3 and 4.5 t/ha yield in ND and LNA systems without residue

application meaning that a 4±5.5 t/ha maize stubble amount is likely to be attained in the region

using a maize hybrid. Maize stubble amount is calculated based on Southern Africa average

maize yield of 3.2 t/ha assuming a 0.45 harvest index (Kintché et al., 2015). These figure are in

line with data presented in (Baudron et al., 2015c) who estimated an approximately 2428 ± 5890

kg maize biomass per year in central Mozambique farms.

The N-application levels were selected based on two criteria. The 23 kg N/ha assumed for the

LNA systems is an approximation of the fertilizer package in FAO input voucher in

Mozambique which provides approximately 29 kg N/ha through a 50 kg bag of NPK 12:24:12

and a 50 kg bag urea (46 N) to be applied as basal and top-dress fertilizer. To facilitate

Page | 152

recommendations and align the figure with sampled fertilizer applicant amounts in in our study

area (18.3 ± 24.1 kg N/ha) an average N-amount of 23.8 kg N/ha averaged across this three

values was kept to 50 kg bag of urea assuming only a top-dress application. The HNA was based

on reported N-application amount needed to satisfy N-immobilization demands during high

carbon residue decomposition in SSA LENIA systems which ranges between 80-100 kg N/ha

(Homann-Kee Tui et al., 2015; Vanlauwe et al., 2014). The amount was kept to 92 kg N/ha, i.e.,

4 bags of urea (46 N).

Soil parameterization in APSIM

Actual soil data collected at the field trials was used for the model parameterization. This

included bulk density, soil water, soil organic carbon (SOC) and mineral N (NO3-N) measured at

the start of the cropping (Table 5-1). In APSIM the nitrogen module was initialized with

measured soil nitrate N (NO3-N) in each site. Crop lower limit (CLL) and drained upper limit

(DUL) were determined opportunistically across the seasons (Pala et al., 2007). The lowest

measured soil water content at harvest in each soil was used as CLL. The DUL was averaged

between the highest measured soil water in each soil and the results from a pond experiment as

described in the soil matter procedural manual from Dalgliesh and Foale (1998). Soil profile (0-

1.20 m) plant available water content at sowing (PAWCs) of 91 mm, 41 mm and 11 mm for the

sandy loam, sandy clay loam and fine sand respectively were used to initialize the model. The

soil profile was considered to be filled from the top and the soil C/N ratio was set at 14.5 based

on soil analysis data.

Crop parameterization

For maize, a mid-duration hybrid PAN67 from PANNAR Seeds, was used. An existing cultivar

description in APSIM (SC501, a hybrid) was modified to represent PAN 67 yield potential (c.a.

8t/ha) and duration to physiological maturity (120 days). The main adjustments for this

calibration were; maximum grain number = 550 (from 520), grain growth rate was kept at 9

mg/grain/day, and thermal time from flowering to maturity = 980 oC days (from 730). The

genetic coefficients for maize were estimated using a vet fit method (Balwinder et al., 2011).

Here, variation on the genetic coefficients in a pre-existing variety were made in order to

produce a close match between measured and simulated phenology and grain yield. For cowpea,

Page | 153

an existing short duration variety, Banjo was adjusted to represent the IT16. Adjustments were

made on thermal degrees’ day from emergence to floral initiation (552 to 565), cumulative

vernal day from 100 to 85, estimated days from emergence to floral initiation from 20 to 25 days

to reproduce the yield of the unfertilized and no residue applied cowpea treatment. These

adjustments were based on averaged data from two cowpea crops, sown in November (early

sowing), and February of the 2012-13 and 2013-14.

Table 5-1 Initial soil organic carbon (SOC), nitrate-nitrogen (NO3-N) and soil water description

parameters from soil samples collected in the 2013-14 seasons and were used for model parameterization

Farming Environment Parameter

Soil layer (cm)

0-15 15-30 30-60 60-90 90-120

Arenosol

Low N and low WHC (fSa)

SOC (%) 0.594 0.564 0.133 0.070 0.020

NO3-N (ppm) 5.720 3.270 0.860 0.870 0.670

Air dry (mm/mm) 0.030 0.050 0.087 0.127 0.137

CLL (mm/mm) 0.074 0.074 0.087 0.127 0.134

DUL (mm/mm) 0.140 0.140 0.140 0.140 0.140

SAT (mm/mm) 0.331 0.327 0.327 0.376 0.376

Bulk density (g/cm3) 1.640 1.647 1.648 1.521 1.521

Gleysol

Low N and high WHC (SaL)

SOC (%) 2.040 2.370 1.140 0.520 0.370

NO3-N (ppm) 5.100 0.100 0.100 0.010 0.010

Air dry (mm/mm) 0.040 0.060 0.180 0.239 0.252

CLL (mm/mm) 0.100 0.110 0.180 0.239 0.252

DUL (mm/mm) 0.200 0.200 0.283 0.290 0.300

SAT (mm/mm) 0.384 0.376 0.376 0.403 0.410

Bulk density (g/cm3) 1.503 1.518 1.522 1.455 1.432

Ferralsol

High N and high WHC (cSaCL)

SOC (%) 1.930 1.670 0.900 0.620 0.590

NO3-N (ppm) 7.790 6.920 4.120 2.290 1.360

Air dry (mm/mm) 0.050 0.070 0.183 0.193 0.199

CLL (mm/mm) 0.121 0.121 0.183 0.193 0.199

DUL (mm/mm) 0.258 0.286 0.296 0.298 0.299

SAT (mm/mm) 0.384 0.403 0.420 0.425 0.422

Bulk density (g/cm3) 1.500 1.450 1.520 1.390 1.390

Page | 154

5.2.2 Long term simulations: residue induced yield, NUE and WUE responses

APSIM v7.7, was used to evaluate the expected variation in maize and cowpea response to a

range of fertiliser x mulch inputs in Chimoio. As cowpea is a major legume for farmers in this

region, the analysis was extended to assess whether farmers would obtain a higher return from

high C/N maize residues if applied as mulch to cowpea crops. Once the model was

parameterized, it was used to:

a) Test whether the model was able to reproduce the observed treatment and site responses;

b) Identify under which circumstances, i.e., crops, sites, N-levels and seasons, responses to

high C/N residues would be likely. Responses were evaluated in term of yield (kg/ha),

water use efficiency (WUE, kg grain/ha.mm), and NUE (kg grain/kg N-supplied).

Multi-year simulations (64 years) were conducted to assess the potential response of maize and

cowpea to residues. The C/N ratio of applied residues was kept at 54:1. Daily climate records for

rainfall, temperature and radiation at Chimoio were available for the period January 1951 to

April 2015, were used. Simulations were initiated on 01 September (end of dry season) and

ended on 15th April. Weed competition was simulated using a summer grass cultivar with a

population of 3 pl/m2, based on weed counts taken along the maize crop. The soil module was

initialised at 27 and 64% plant available water content evenly distributed in the soil profile to

simulate two contrasting environments. Initial NO3-N and NH4-N in the soil profile was set at 10

and 5 kg N/ha to 1.20 m, respectively.

Calculations

Water use efficiency (WUE) was calculated for both maize and cowpea as the ratio between

yield and evapotranspiration. Evapotranspiration was determined as the difference between in-

crop rainfall and soil water losses by the system which a sum of surface soil runoff and water

losses through deep drainage outside the root zone. Nitrogen use efficiency (NUE) was

calculated based on (Moll et al., 1982; Sadras and Lemaire, 2014).

Page | 155

5.2.3 Model testing and Evaluation

Simulated model outputs were compared with observed values for crop yield, WUE and NUE for

all treatments and sites for the 2013-14 cropping season. The fit between observed and simulated

results was determined using the root mean squared error (RMSE), model efficiency (ME) and

the coefficient of determination (r2). The coefficient of determination r2 was calculated from the

regression of observed against simulated data. RMSE and ME were calculated through formulas

(1) and (2) respectively. A ME and r2 vary between -1 and +1. A ME as closes as possible to 1

indicates a good fit, while a negative ME suggests that the observed data are a better

representation of the situation in cause. ME close to zero represents as poor fit.

RMSE = (1)

And

ME = 1 - (2)

5.3. Results

5.3.1 Model evaluation

The model was able to reproduce the experimental results i.e. overall model efficiency (ME) of

0.64 and RMSE of 957.3 kg/ha; with a r2 of 0.75 between observed and predicted yields (Table

5-2). A separate analysis of model performance for each individual crop showed an overall yield

RMSE of 1131.8 kg/ha and 0.60 ME for maize. However, the model over predicted maize yield

for specific treatments (Figure 5-1), especially HNA in fSa and cSaCL. For cowpea model

performance was less impressive, i.e. with an overall ME of 0.37 and RMSE of 232.2 kg/ha.

However, over predictions of cowpea yield were observed for the cSaCL site. The model

represented better crop NUE than WUE (Table 5-2).

Page | 156

Table 5- 2 Overall model evaluation for selected parameters

Model attribute Number of data

points

r2 RMSE ME

Maize yield (kg/ha) 18 0.63 1131.8 0.60

Cowpea yield (kg/ha) 18 0.42 233.2 0.37

Overall Grain yield (kg/ha) 36 0.78 831.5 0.73

Water use efficiency (WUE, kg grain/ha.mm) 36 0.45 - -

Nitrogen Use Efficiency (NUE, kg grain/kg N-

applied)

36 0.81 4.912 0.80

Page | 157

Figure 5-1 Above - Comparison between observed and APSIM model simulated maize and cowpea

yield distributed across three sites. Shapes identification: ∆ - Sandy Clay loam (cSaCL); O Fine sandy

(fSa) and ◊ Sandy loam soil (SaL). Filled shapes for each soil type represent residue treatments while

empty shapes did not receive any residue. Middle and bellow (A, B and C): Overall comparison

between observed and simulated maize and cowpea yield (a), NUE (b) and WUE(c) distributed across

tested sites for. Continuous lines indicate a 1:1 fit and dotted lines indicate a best fit line between

observed and simulated.

Page | 158

5.3.2 Interactions between residue and fertilizer and their impact on yield

A crop and soil driven yield response to a high C/N ratio mulch and inorganic N fertilizer

application was simulated. In general, cowpea responded positively to residue application in all

environments, except in the drier and low water holding capacity sandy soil (fSa), were mixed

responses to residues were simulated across levels of N-application. For maize, responses to

residue application were affected by the starting soil N status and N-fertilizer application levels.

Soil moisture regimes also led to contrasting responses to residue and N-application.

Residue responses in high WHC and wetter soils

Maize yield losses were simulated in almost 85% to 80% of the seasons in ND and LNA systems

under residue application in the of the N-rich cSaCL (Figure 5-2). In HNA systems, yield losses

from residue application were simulated in 75% of the seasons with positive responses to

residues being expected in approximately 20 % of the seasons. Nevertheless, yield losses due to

high C/N residue application were reduced with increased N-application in this soil. Overall

maize yield losses of 11.2% were calculated for HNA systems, against 31.2% and 38.4% in LNA

and ND systems after high C/N residue application. Simulations results also showed that a 3 t/ha

yield is only likely to be attained in 10% and 40 % of the seasons in ND and LNA systems under

residue application but in almost 80% of the seasons in residue free ND and LNA systems. In

contrast, a 95% chance of achieving yields higher than 3 t/ha was simulated in HNA systems

under residue application. This was 5% higher than residue free systems under HNA. For the wet

sandy loam soil (SaL), were enough soil moisture to trigger a faster residue decomposition and

consequently N-immobilization, high C/N residue application led to an overall yield loss of

46.3%, 40.7% and 18.6% in ND, LNA and HNA maize systems over the simulation period. In

this soil, benefits from residue application were only attained in 5 % of the seasons in ND and

LNA systems. Nevertheless, maize yields in those seasons were lower than 1 t/ha. In HNA

systems, positive yield responses were simulated in 20% of the seasons under high C/N residue

application. Simulation results also showed that the likelihood of producing yields higher than 3

t/ha, was of approximately 80 % at all levels of N-application in residue free systems compared

to less than 1% and 20% in ND and LNA systems under residue application. In HNA systems, an

85% chance to produce yields higher than 3 t/ha yield plateau was simulated.

Page | 159

In both soils, positive responses to high C/N residue application were simulated in more than

90% of the seasons in ND, LNA and HNA systems for cowpea. Nevertheless, increased N-

applications have reduced cowpea yields. Cowpea yields higher than 1t/ha were attained in

almost 85% of the season in ND and LNA systems but in less than 80% in HNA systems in the

N-rich cSaCL. In the wet SaL, increased N-application have reduced cowpea yields to less than 1

t/ha in 85% in ND and LNA systems. In HNA systems, yields higher than 1 t/ha were only

attained in almost 75% of the seasons under residue application.

Figure 5-2 long term simulated maize and cowpea yield responses to residue application and N-

application levels (0, 23, 92 kg N/ha representing ND, LNA and HNA systems) in a high WHC

(cSaCL) and prolonged wetness (SaL) soils. No residue (R-); Residue applied (R+)

Page | 160

Residue responses in dry and low WHC soils

In dry and low WHC (fSa), yield increments from the application of high C/N ratio residues

were simulated for maize in almost 90% of the seasons at all N-application levels. Maize yields

were lowest at the fSa site (Figure 5-3). Nevertheless, residue application in this environment,

increased the chances of achieving yields higher than 1 t/ha, from less than 40%, 30% and 25%

of the seasons in residue free systems under ND, LNA and HNA, to more than 75% of the

seasons in all systems once residues were added. Overall maize yield increments of 43.3%,

22.2% and 16% in ND, LNA and HNA were attained under residue application in the fSa.

Despite the measured positive response to residue application, a 3 t/ha yield is likely to be

surpassed in less than 30% and 35% of the seasons in ND and LNA systems. A 50% chance of

surpassing the 3 t/ha mark was simulated for HNA systems under residue application in the drier

fSa.

For cowpea, simulated results showed almost 90 and 80% probability of surpassing 0.5t/ha

(SSA, average) and 1t/ha, respectively, were simulated under residue application in all sites.

Despite an overall positive response to residue application in cowpea, mixed responses to residue

were simulated in this drier and low water holding capacity soil. In unfertilized ND cowpea

systems, advantages from residue application, were only observed in 50% of the simulated

seasons. However, advantages from residues application were increased with levels of N-

application. A pre-assessment of crop performance based on residue application, would suggest

that for this site, when fertilizers are not available, investing residues to improve moisture

retention in the maize crop, is a more viable option that applying then into the legume crop

whose response to residues proved to increase with levels of N-application.

Page | 161

5.3 Resource productivity

Residue modified N-driven trade-off between NUE and WUE

A nitrogen driven trade-off between NUE and WUE was simulated (Figure 5-4). For both crops,

increased N-applications above 23 kg N/ha, reduced NUE and improved WUE. Increased water

availability across environments increased crop NUE especially for maize. This phenomenon can

be seen in the fSa-SaL and fSa-cSaCL transitions. The magnitude of the nitrogen driven NUE

and WUE trade-offs was dependent on individual sites, i.e. soil moisture and initial soil N-status.

Results also show that in LENIA systems, the N-driven NUE and WUE trade-offs is affected by

site and crop individual responses to the application of high C/N residues. Across all sites,

increased N-fertilizer applications reduced NUE but improved WUE (Figure 5-4).

Figure 5-3 Long term simulated maize and cowpea yield response to residue and N-application

levels (0, 23, 92 kg N/ha representing ND, LNA and HNA systems) in low WHC (fSa) soil. No

residues (R-), Residue applied (R+)

Page | 162

Figure 5-4 Residue modified nitrogen driven NUE and WUE trade-off simulated across maize and cowpea grown in contrasting farming

environments of central Mozambique. Overall means are presented for each level of N-application namely: unfertilized (ND, 0 kg N/ha),

Limited N-applications (LNA, 23 kg N/ha) and high N-application (HNA, 92 kg N/ha). Residue application levels are indicated as R- and R+

for no residue and residue applied plots respectively.

Page | 163

5.4. Seasonal rainfall variability and their effects on residue responses

Responses to residue application across crops and farming environments were found to be

largely affected by in-crop rainfall distribution patterns (Figure 5-5). For each environment,

positive and negative yield variations across N-application levels were simulated at specific

seasons. In the low WHC and drier fSa, positive contributions from residue application into

maize yield were simulated in most of the seasons, except in the wettest years, i.e., 1952, 1956,

1960, 1977, 1989 and 2001 which registered annual rainfall higher than 1000 mm. In the high

WHC cSaCL and wet SaL, were soil moisture regimes are considerable good enough to sustain

crop growth over the season, benefits from residue application were only simulated in the driest

(e.g.,1973, 1968, 1992 and 2003) and poorly distributed rainfall seasons (e.g., 1954, 1983 and

2010).

Page | 164

Figure 5-5 Above, in the rainfall distribution graphs (A, B and C) for the period 1951-2014, seasons are

categorized according to simulated maize yield variations under high C/N residue application at three

different soils. The horizontal line indicates annual average rainfall over the simulated period. Dark bars –

indicate seasons were yield losses from residue application were simulated; Red bars indicate seasons in

which positive responses to residue application are expected across N-application levels. In graphs A1 to

C3, relative maize yield variations to residue application disaggregated by soil and N-application levels

over the simulated period.

Page | 165

Table 5-3 Inter-seasonal performance evaluated through variations in long term simulated maize and cowpea water productivity (WUE) and

nitrogen use efficiency (NUE) response to residue application across N-levels for a low N and low water holding capacity environment.

Farming Environment N-level Systems

Performance Maize WUE Cowpea WUE Maize NUE Cowpea NUE

∆WUE (%) Seasons ∆WUE (%) Seasons ∆NUE (%) Seasons ∆NUE (%) Seasons

Arenosol

Low N and low WHC

(fSa)

0N Benefits 135 0.79 32.5 0.95 77.4 0.25 22.0 0.41

Losses -34.6 0.21 -12.0 0.05 -29.8 0.75 -3.38 0.59

23N Benefits 72.1 0.78 27.4 0.98 52.9 0.24 24.3 0.40

Losses -27.4 0.22 -19.9 0.02 -29.9 0.76 -3.30 0.60

92N Benefits 54.7 0.83 31.8 0.95 54.8 0.40 25.5 0.65

Losses -29.8 0.17 -9.49 0.05 -17.8 0.60 -2.68 0.35

Gleysol

Low N and high WHC

(SaL)

0N Benefits 70.4 0.18 64.6 1 51.9 0.37 33.2 0.94

Losses -37.8 0.82 - - -17.6 0.63 -2.94 0.06

23N Benefits 71.3 0.20 51.2 0.98 92.1 0.20 35.8 0.94

Losses -36.1 0.80 -0.95 0.02 -25.4 0.80 -4.01 0.06

92N Benefits 88.2 0.30 44.4 1 87.0 0.30 42.0 0.97

Losses -14.6 0.70 - - -16.6 0.70 -5.66 0.03

Ferralsol

High N and high WHC

(cSaCL)

0N Benefits 64.9 0.22 33.1 0.56 49.0 0.49 61.0 0.95

Losses -32.9 0.78 -19.1 0.44 -6.28 0.51 -2.38 0.05

23N Benefits 52.4 0.19 34.0 0.44 83.3 0.30 40.2 0.84

Losses -33.4 0.81 -13.6 0.56 -20.2 0.70 -0.90 0.16

92N Benefits 35.4 0.33 29.1 0.59 35.7 0.76 45.0 0.87

Losses -12.3 0.67 -11.1 0.41 -21.8 0.24 -2.26 0.13

Legend: Systems performance: Mean is the overall relative variation (%) of yield, WUE and NUE across residue application levels; Benefits and

losses, are averaged positive and negative variations in yield, WUE and NUE over the simulated period. Seasonal responses: presented as a ratio

between the number of seasons with calculated benefits over the all spectrum of simulated seasons (N/64), Benefits and losses are averaged across

seasons were positive and negative variations in simulated yield, WUE and NUE were calculated for each specific environment

Page | 166

5.5 Discussion

APSIM suitability to simulate residue-fertilizer interactions in N-deprived systems

The overall performance of APSIM in reproducing observed maize and cowpea yield

responses to residue mulch and fertilizer application across farming environments was good.

However, in certain circumstances, the model either over predicted or under predicted

individual treatments responses. Example, for the specific case of the cSaCL, a response to

N-application was simulated in the absence of residues in both years contradicting the poor

response to N-application attained in the field during the first season. However, long-term

simulations showed that in low and poorly distributed rainfall seasons a similar result is likely

to occur in this environment. In the wet but low N environment (SaL), a good agreement

between model and observed data was achieved. In this environment, positive responses to

N-application in both situations, with and without residues were also simulated. Nevertheless,

a lower response was simulated in the transition between ND and LNA systems in both

residue levels.

In the low N and drier environment (fSa), the model has done well at lower N-levels but did

not quite well mimicked trial results at higher N-level. Poor APSIM performance in water

limited environments was reported before (Balwinder et al., 2011). For this specific case,

there is a need to mention that, a single crop soil parameterization, i.e., maize was used,

letting all alone the model to capture cowpea responses for the same conditions. This

somehow would likely affect model predictions for cowpea. However, it was not the case,

since a better model agreement was simulated in cowpea. Despite the discrepancy between

observed and simulated yield in some treatments, there was an overall good agreement

between observed and simulated responses in all tested environments, which makes APSIM a

candidate tool for ex-ante and post-ante analysis that are residue management and maize-

legume cropping systems design in Central Mozambique and SSA as shown before

(Akponikpè et al., 2010; Mohanty et al., 2010; Ncube et al., 2009). Nevertheless, model

accuracy can still be improved, in order to make the model an integral part of local research

systems. However, improved crop phenology descriptors for locally grown cereals and

especially legume varieties need to be developed.

Page | 167

Understanding residue induced responses through seasonal yield, WUE, NUE variations

Simulation results showed that there is almost 85 % chance that residue application will

reduce maize yield in rain fed LENIA systems of central Mozambique, especially in ND and

LNA maize crops cultivated in considerably wet and high WHC soils. In the AEZ-R4, which

has an average annual rainfall of 834 mm (1951-2014), appears to provide enough soil

moisture to trigger residue decomposition and consequently a N-immobilization induced

plant N-stress which leads to maize yield losses especially in wet and high WHC soils

(Figure 5-5).

In a normal rainfall season no significant water stress is likely to occur throughout the crop

cycle, especially in the predominantly cultivated soil - red cSaCL ferralsol in AEZ-R4.

Therefore, maize yield losses are derived from poor NUE resulting from N-immobilization

induced plant N-stress under residue application (Table 5-3). However, in poor rainfall

seasons, i.e., less than 20% over the simulated period, both water and N-stress are likely to

occur. Nevertheless, residue application is likely to improve soil water availability and

consequently yields only in poor rainfall seasons – low rainfall and badly distributed in-crop

rainfall years.

Poor responses from C rich residues are likely to occur across most central Mozambique

agro-ecologies, namely the medium altitude – AEZ-R5, AEZ-R7 and AEZ-R8 and in the

central Mozambique highlands (AEZ-R10) which have rainfall regimes between 800-1000

mm/year and above 1200 mm/year respectively (Reddy, 1984). Here, more pragmatic residue

application strategies need to be considered to take advantage of residues in the system. In the

semi-arid regions of northern Manica province and south Tete provinces where lays the R6

agro-ecology with annual rainfall between 400-600 mm, benefits from residue application are

likely to occur as simulated for the drier and low WHC fSa soils. In drier and low WHC soils,

where there is not enough soil moisture to trigger a faster residue decomposition, residues are

likely to help increase maize WUE at all levels of N-application, and consequently increase

yields (Proposition 2).

Page | 168

Figure 5-6 Nitrogen and water stress levels in maize across residue application levels for a

high WHC environment in a normal and poor rainfall season. A stress factor of 1 indicates

full stress and 0 absence of stress. Stress indexes: npho is the N-stress and wpho is water

stress.

Maize yield losses, under residue application in considerably wet environments, are

associated to the high moisture induced residue decomposition, which leads to N-

immobilization. In high WHC and clay rich soils, gross N-immobilization lasting between

14-545 days were reported (Gentile et al., 2008; Recous et al., 1995; Sakala et al., 2000).

High moisture and microbial activity in those soils trigger a fast residue decomposition

leading to N-immobilization at low N-levels (Guntiñas et al., 2012; Tian et al., 1992). Chen et

al. (2012), reported high N-mineralization and N-immobilization during residue

decomposition in soils at 75% WHC compared to soils at 35% WHC.

Contrarily to high WHC soils, poor microbial activity in low WHC soils, i.e., dry

environments like AEZ-R6, delays residue decomposition allowing them to stay longer in the

soil providing further surface cover and soil moisture benefits to the crops. As a result,

positive yield responses mainly due to increased WUE are expected in such circumstances as

the simulation results also showed for the fSa. Nevertheless, a risk of N-immobilization is

also present and residue application are likely to reduce maize NUE. However, losses from

NUE do not surpass the gains from WUE in this environment. Benefits from residue

application, in drier and coarse textured low WHC soil such as the drier fSa were also

Page | 169

reported by Gentile et al. (2008). Improved maize yields under residue application were also

reported under drought conditions in the Central Mexico Highlands (Romero-Perezgrovas et

al., 2014), strengthening our finding that high C/N ratio residues can be used to improve

water capture and increase yields in considerably dry soils and years.

Experimental data has also showed that yield losses from N-immobilization during residue

decomposition in N-deprived systems might be alleviated by improved residue management

strategies. Increased inorganic N-application especially in wet and high WHC soils, was

reported to reduce N-immobilization and improve N-release to the crops (Esther et al., 2014;

Recous et al., 1995; Ribeiro et al., 2002). Nevertheless, results also showed that the

magnitude of the negative effects from high C/N ratio residues application on maize yields,

WUE and NUE was largely dependent on the initial soil N status and could be alleviated by

increased levels of N-application in wet environments (Proposition 4). In most field trials,

however positive, yield and resource productivity attained under high C/N ratio residue

application where reported at considerably high N-application levels, i.e., applications above

80 kg N/ha or more in N-rich environments where inorganic N-supply levels are high enough

to partially or fully satisfy N-immobilization demands during residue decomposition (Bakht

et al., 2009; Baudron et al., 2015a; Tao et al., 2015; Verhulst et al., 2011; Xiukang et al.,

2015). However, this is not the case of most resource poor farmers in central Mozambique

and SSA.

An alternative to inorganic fertilizer whose use is considerable low among resource poor

smallholder farmers of Central Mozambique, is shifting high C/N ratio residue mulch

application into legumes crops. This presents as more affordable option to improve soil N-

availability in the system once positive yield, NUE and WUE responses to high C/N ratio

residues were simulated in almost 50-90 % of the seasons with an early sown cowpea crop

across tested soils in the AEZ-R4. For example, a 13% grain yield benefit in mungbean under

wheat residue application were reported by Shah et al. (2003b). In the same trial, the

percentage of mungbean N-derived from N2-fixation was increased by 55.6% and a +64 kg

N/ha average N-balance was achieved under residue application. Positive yield responses and

high N-turnover from N-rich residues returned to the system from legume crops grown under

high C/N ratio residue applied plots, were reported in separate studies (Luce et al., 2015;

Monzon et al., 2014; Neugschwandtner and Kaul, 2015). These results, prove that using

Page | 170

residues to improve legume performance in maize dominated systems, might be a viable

option to improve N-availability in smallholder farming systems. This is a critical result for

central Mozambique maize dominated systems, where legume incorporation is key to help

reduce yield penalties from continuous maize systems under conservation agriculture.

5.6 Conclusion

As field trials can only capture a limited part of the vast and complex interactions between

crops, farming environments and their response to several management conditions. Model

assisted research and decision support is critical especially in countries where research is

scantly funded. Therefore, owning to its simplicity, time and cost effectiveness, the model

assisted decision support framework proposed here can be useful to capture not only crop and

site-specific responses to residue and fertilizer allocation strategies, but also the trade-offs

characteristic of central Mozambique complex farming environment mosaic. Nevertheless,

despite APSIM’s acceptable performance in simulating the complex residue and fertilizer

interaction and their effects on crop yields, WUE and NUE, model accuracy can still be

improved to make it a reliable and integral component of local research and extension

systems.

Acknowledgements

Would like to thank the SIMLESA project and the Australian Centre for International

Agriculture Research (ACIAR) for the scholarship under which this research was conducted.

The Soil Health Program of the Alliance for Green Revolution in Africa (AGRA) through the

2011SHP007 Project for helping fund the fieldwork. Thanks also to Peter De Voil (APSRU-

DAF-QAAFI) for support with APSIM modelling.

References

Akponikpè, P. B. I., Gérard, B., Michels, K., and Bielders, C. (2010). Use of the APSIM

model in long term simulation to support decision making regarding nitrogen

management for pearl millet in the Sahel. European Journal of Agronomy 32, 144-

154.

Page | 171

Bakht, J., Shafi, M., Jan, M. T., and Shah, Z. (2009). Influence of crop residue management,

cropping system and N fertilizer on soil N and C dynamics and sustainable wheat

(Triticum aestivum L.) production. Soil and Tillage Research 104, 233-240.

Balwinder, S., Gaydon, D. S., Humphreys, E., and Eberbach, P. L. (2011). The effects of

mulch and irrigation management on wheat in Punjab, India—Evaluation of the

APSIM model. Field Crops Research 124, 1-13.

Baudron, F., Delmotte, S., Corbeels, M., Herrera, J. M., and Tittonell, P. (2015a). Multi-scale

trade-off analysis of cereal residue use for livestock feeding vs. soil mulching in the

Mid-Zambezi Valley, Zimbabwe. Agricultural Systems 134, 97-106.

Baudron, F., Thierfelder, C., Nyagumbo, I., and Gérard, B. (2015b). Where to Target

Conservation Agriculture for African Smallholders? How to Overcome Challenges

Associated with its Implementation? Experience from Eastern and Southern Africa.

Environments 2, 338.

Bouma, J., and Jones, J. W. (2001). An international collaborative network for agricultural

systems applications (ICASA). Agricultural Systems 70, 355-368.

Chen, Q.-H., Feng, Y., Zhang, Y.-P., Zhang, Q.-C., Shamsi, I. H., Zhang, Y.-S., and Lin, X.-

Y. (2012). Short-Term Responses of Nitrogen Mineralization and Microbial

Community to Moisture Regimes in Greenhouse Vegetable Soils. Pedosphere 22,

263-272.

Chikowo, R., Corbeels, M., Tittonell, P., Vanlauwe, B., Whitbread, A., and Giller, K. E.

(2008). Aggregating field-scale knowledge into farm-scale models of African

smallholder systems: Summary functions to simulate crop production using APSIM.

Agricultural Systems 97, 151-166.

Cox, H. W., Kelly, R. M., and Strong, W. M. (2010). Pulse crops in rotation with cereals can

be a profitable alternative to nitrogen fertiliser in central Queensland. Crop and

Pasture Science 61, 752-762.

Dalgliesh, N., and Foale, M. (1998). "Soil matters: Monitoring soil water and nutrients in

dryland farming," 1/Ed. CSIRO, Australia, Toowoomba, Australia.

Esther, O. J., Guo, C.-h., Tian, X.-h., Li, H.-y., and Zhou, Y.-x. (2014). The Effects of Three

Mineral Nitrogen Sources and Zinc on Maize and Wheat Straw Decomposition and

Soil Organic Carbon. Journal of Integrative Agriculture 13, 2768-2777.

Page | 172

Gentile, R., Vanlauwe, B., Chivenge, P., and Six, J. (2008). Interactive effects from

combining fertilizer and organic residue inputs on nitrogen transformations. Soil

Biology and Biochemistry 40, 2375-2384.

Gilbert, N. (2012). Dirt Poor: The key to tackle hunger in Africa is enriching its Soils. Nature

483, 525-227.

Giller, K. E., Corbeels, M., Nyamangara, J., Triomphe, B., Affholder, F., Scopel, E., and

Tittonell, P. (2011). A research agenda to explore the role of conservation agriculture

in African smallholder farming systems. Field Crops Research 124, 468-472.

Giller, K. E., Witter, E., Corbeels, M., and Tittonell, P. (2009). Conservation agriculture and

smallholder farming in Africa: The heretics’ view. Field Crops Research 114, 23-34.

Guntiñas, M. E., Leirós, M. C., Trasar-Cepeda, C., and Gil-Sotres, F. (2012). Effects of

moisture and temperature on net soil nitrogen mineralization: A laboratory study.

European Journal of Soil Biology 48, 73-80.

Holzworth, D. P., Huth, N. I., deVoil, P. G., Zurcher, E. J., Herrmann, N. I., McLean, G.,

Chenu, K., van Oosterom, E. J., Snow, V., Murphy, C., Moore, A. D., Brown, H.,

Whish, J. P. M., Verrall, S., Fainges, J., Bell, L. W., Peake, A. S., Poulton, P. L.,

Hochman, Z., Thorburn, P. J., Gaydon, D. S., Dalgliesh, N. P., Rodriguez, D., Cox,

H., Chapman, S., Doherty, A., Teixeira, E., Sharp, J., Cichota, R., Vogeler, I., Li, F.

Y., Wang, E., Hammer, G. L., Robertson, M. J., Dimes, J. P., Whitbread, A. M., Hunt,

J., van Rees, H., McClelland, T., Carberry, P. S., Hargreaves, J. N. G., MacLeod, N.,

McDonald, C., Harsdorf, J., Wedgwood, S., and Keating, B. A. (2014). APSIM –

Evolution towards a new generation of agricultural systems simulation.

Environmental Modelling and Software 62, 327-350.

Homann-Kee Tui, S., Valbuena, D., Masikati, P., Descheemaeker, K., Nyamangara, J.,

Claessens, L., Erenstein, O., van Rooyen, A., and Nkomboni, D. (2015). Economic

trade-offs of biomass use in crop-livestock systems: Exploring more sustainable

options in semi-arid Zimbabwe. Agricultural Systems 134, 48-60.

Jordan, V. W. L., Hutcheon, J. A., Donaldson, G. V., and Farmer, D. P. (1997). Research into

and development of integrated farming systems for less-intensive arable crop

production: experimental progress (1989–1994) and commercial implementation.

Agriculture, Ecosystems and Environment 64, 141-148.

Kintché, K., Guibert, H., Sogbedji, J. M., Levêque, J., Bonfoh, B., and Tittonell, P. (2015).

Long-term mineral fertiliser use and maize residue incorporation do not compensate

Page | 173

for carbon and nutrient losses from a Ferralsol under continuous maize–cotton

cropping. Field Crops Research 184, 192-200.

Luce, M., Grant, C. A., Zebarth, B. J., Ziadi, N., O’Donovan, J. T., Blackshaw, R. E., Harker,

K. N., Johnson, E. N., Gan, Y., Lafond, G. P., May, W. E., Khakbazan, M., and

Smith, E. G. (2015). Legumes can reduce economic optimum nitrogen rates and

increase yields in a wheat–canola cropping sequence in western Canada. Field Crops

Research 179, 12-25.

Masikati, P., Manschadi, A., van Rooyen, A., and Hargreaves, J. (2014). Maize–mucuna

rotation: An alternative technology to improve water productivity in smallholder

farming systems. Agricultural Systems 123, 62-70.

Matthews, R. B., and Stephens, W. P. D. (2002). "Crop-soil simulation models: applications

in developing countries," CABI Pub, Wallingford, England.

Maxwell, S. (1986). Farming systems research: Hitting a moving target. World Development

14, 65-77.

McCown, R. L., Hammer, G. L., Hargreaves, J. N. G., Holzworth, D. P., and Freebairn, D.

M. (1996). APSIM: a novel software system for model development, model testing

and simulation in agricultural systems research. Agricultural Systems 50, 255-271.

Meinke, H., Baethgen, W. E., Carberry, P. S., Donatelli, M., Hammer, G. L., Selvaraju, R.,

and Stöckle, C. O. (2001). Increasing profits and reducing risks in crop production

using participatory systems simulation approaches. Agricultural Systems 70, 493-513.

Mohanty, M., Probert, M. E., Reddy, K. S., Dalal, R. C., Rao, A. S., and Menzies, N. W.

(2010). Modelling N mineralization from high C/N rice and wheat crop residues.

Proceedings of the 19th World Congress of Soil Science: Soil solutions for a changing

world, Brisbane, Australia, 1-6 August 2010. Symposium 1.5.2 Modelling critical

processes in changing soil, 28-31.

Moll, R. H., Kamprath, E. J., and Jackson, W. A. (1982). Analysis and Interpretation of

Factors Which Contribute to Efficiency of Nitrogen Utilization1. Agron. J. 74, 562-

564.

Monzon, J. P., Mercau, J. L., Andrade, J. F., Caviglia, O. P., Cerrudo, A. G., Cirilo, A. G.,

Vega, C. R. C., Andrade, F. H., and Calviño, P. A. (2014). Maize-soybean

intensification alternatives for the Pampas. Field Crops Research 162, 48-59.

Page | 174

Mupangwa, W., Dimes, J., Walker, S., and Twomlow, S. (2011). Measuring and simulating

maize (Zea mays L.) yield responses to reduced tillage and mulching under semi-arid

conditions. Agricultural Sciences 2, 167-174.

Ncube, B., Dimes, J., Twomlow, S., Mupangwa, W., and Giller, K. (2007). Raising the

productivity of smallholder farms under semi-arid conditions by use of small doses of

manure and nitrogen: a case of participatory research. Nutrient Cycling in

Agroecosystems 77, 53-67.

Ncube, B., Dimes, J. P., van Wijk, M. T., Twomlow, S. J., and Giller, K. E. (2009).

Productivity and residual benefits of grain legumes to sorghum under semi-arid

conditions in south-western Zimbabwe: Unravelling the effects of water and nitrogen

using a simulation model. Field Crops Research 110, 173-184.

Neugschwandtner, R. W., and Kaul, H.-P. (2015). Nitrogen uptake, use and utilization

efficiency by oat–pea intercrops. Field Crops Research 179, 113-119.

Ngwira, A. R., Thierfelder, C., and Lambert, D. M. (2013). Conservation agriculture systems

for Malawian smallholder farmers: long-term effects on crop productivity,

profitability and soil quality. Renewable Agriculture and Food Systems 28, 350-363.

Pala, M., Ryan, J., Zhang, H., Singh, M., and Harris, H. C. (2007). Water-use efficiency of

wheat-based rotation systems in a Mediterranean environment. Agricultural Water

Management 93, 136-144.

Probert, M. E., Dimes, J. P., Keating, B. A., Dalal, R. C., and Strong, W. M. (1998). APSIM's

water and nitrogen modules and simulation of the dynamics of water and nitrogen in

fallow systems. Agricultural Systems 56, 1-28.

Recous, S., Robin, D., Darwis, D., and Mary, B. (1995). Soil inorganic N availability: Effect

on maize residue decomposition. Soil Biology and Biochemistry 27, 1529-1538.

Reddy, S. J. (1984). Clima Geral de Mocambique In "Serie Terra e Água" (INIA, ed.), pp. 41.

Instituto de Investigação Agronómica de Moçambique (INIA), Departamento de Terra

e Agua (DTA), Maputo, Mozambique.

Ribeiro, C., Madeira, M., and Araújo, M. C. (2002). Decomposition and nutrient release from

leaf litter of Eucalyptus globulus grown under different water and nutrient regimes.

Forest Ecology and Management 171, 31-41.

Rodriguez, D., deVoil, P., Power, B., Cox, H., Crimp, S., and Meinke, H. (2011). The

intrinsic plasticity of farm businesses and their resilience to change. An Australian

example. Field Crops Research 124, 157-170.

Page | 175

Romero-Perezgrovas, R., Verhulst, N., De La Rosa, D., HernÁNdez, V., Maertens, M.,

Deckers, J., and Govaerts, B. (2014). Effects of Tillage and Crop Residue

Management on Maize Yields and Net Returns in the Central Mexican Highlands

Under Drought Conditions. Pedosphere 24, 476-486.

Sadras, V. O., and Lemaire, G. (2014). Quantifying crop nitrogen status for comparisons of

agronomic practices and genotypes. Field Crops Research 164, 54-64.

Sadras, V. O., and Rodriguez, D. (2010). Modelling the nitrogen-driven trade-off between

nitrogen utilisation efficiency and water use efficiency of wheat in eastern Australia.

Field Crops Research 118, 297-305.

Sakala, W. D., Cadisch, G., and Giller, K. E. (2000). Interactions between residues of maize

and pigeon pea and mineral N fertilizers during decomposition and N mineralization.

Soil Biology and Biochemistry 32, 679-688.

Shah, Z., Shah, S. H., Peoples, M. B., Schwenke, G. D., and Herridge, D. F. (2003). Crop

residue and fertiliser N effects on nitrogen fixation and yields of legume–cereal

rotations and soil organic fertility. Field Crops Research 83, 1-11.

Sommer, R., Thierfelder, C., Tittonell, P., Hove, L., Mureithi, J., and Mkomwa, S. (2014).

Fertilizer use should not be a fourth principle to define conservation agriculture:

Response to the opinion paper of Vanlauwe et al. (2014) ‘A fourth principle is

required to define conservation agriculture in sub-Saharan Africa: The appropriate use

of fertilizer to enhance crop productivity’. Field Crops Research 169, 145-148.

Tao, Z., Li, C., Li, J., Ding, Z., Xu, J., Sun, X., Zhou, P., and Zhao, M. (2015). Tillage and

straw mulching impacts on grain yield and water use efficiency of spring maize in

Northern Huang–Huai–Hai Valley. The Crop Journal 3, 445-450.

Thierfelder, C., Cheesman, S., and Rusinamhodzi, L. (2012). A comparative analysis of

conservation agriculture systems: Benefits and challenges of rotations and

intercropping in Zimbabwe. Field Crops Research 137, 237-250.

Thierfelder, C., Matemba-Mutasa, R., and Rusinamhodzi, L. (2015). Yield response of maize

(Zea mays L.) to conservation agriculture cropping system in Southern Africa. Soil

and Tillage Research 146, Part B, 230-242.

Thierfelder, C., Mombeyarara, T., Mango, N., and Rusinamhodzi, L. (2013). Integration of

conservation agriculture in smallholder farming systems of southern Africa:

identification of key entry points. International Journal of Agricultural Sustainability

11, 317-330.

Page | 176

Tian, G., Kang, B. T., and Brussaard, L. (1992). Biological effects of plant residues with

contrasting chemical compositions under humid tropical conditions—Decomposition

and nutrient release. Soil Biology and Biochemistry 24, 1051-1060.

Twomlow, S., Rohrbach, D., Dimes, J., Rusike, J., Mupangwa, W., Ncube, B., Hove, L.,

Moyo, M., Mashingaidze, N., and Mahposa, P. (2010). Micro-dosing as a pathway to

Africa’s Green Revolution: evidence from broad-scale on-farm trials. Nutrient

Cycling in Agroecosystems 88, 3-15.

Vanlauwe, B., Wendt, J., Giller, K. E., Corbeels, M., Gerard, B., and Nolte, C. (2014). A

fourth principle is required to define Conservation Agriculture in sub-Saharan Africa:

The appropriate use of fertilizer to enhance crop productivity. Field Crops Research

155, 10-13.

Verhulst, N., Govaerts, B., Nelissen, V., Sayre, K. D., Crossa, J., Raes, D., and Deckers, J.

(2011). The effect of tillage, crop rotation and residue management on maize and

wheat growth and development evaluated with an optical sensor. Field Crops

Research 120, 58-67.

Whitbread, A. M., Robertson, M. J., Carberry, P. S., and Dimes, J. P. (2010). How farming

systems simulation can aid the development of more sustainable smallholder farming

systems in southern Africa. European Journal of Agronomy 32, 51-58.

Xiukang, W., Zhanbin, L., and Yingying, X. (2015). Effects of mulching and nitrogen on soil

temperature, water content, Nitrate-N content and maize yield in the Loess Plateau of

China. Agricultural Water Management 161, 53-64.

Page | 177

CHAPTER 6

General Discussion

The most important thing in triggering any kind of change is by first finding people’s own

passion to grow. Once we find that passion our responsibility is to provide them with the

right kind of knowledge and technical support to materialize their passion! For that to

happen we need to listen and not force our will into people. (Ernesto Sirolli, in Ted Talks

2015)

(Agricultural intervention is not an exception to that rule)

Page | 178

6.1 Summary of key findings

Over the last decades, residue retention and fertilizer application have been widely promoted

across Africa. Crop residues are used as part of a wider conservation agriculture (CA)

platform that also includes rotation with legumes and minimum soil disturbance (Wall, 2007).

In the short term fertilizers appear to be the best way to improve yields in Africa (Chianu et

al., 2012; Gilbert, 2012); though there are many situations where the degree of soil

degradation is such that fertilisers alone will not be enough to lift crop productivity(Tittonell

et al., 2007a). CA appears like an optimum solution then in the view of researchers (Ngwira

et al., 2013; Thierfelder et al., 2012; Thierfelder et al., 2015) while adoption of these

practices continues to be a low priority for resource poor smallholder farmers in Africa. So,

compared to the rest of the world, adoption of CA principles in Sub Saharan Africa (Baudron

et al., 2015c), as that of fertilizers (Heffer and Prud'homme, 2011) remains negligible. This

raises questions over the suitability of these technologies and the way they are promoted for

the complex and diverse African situations.

Single-sized technological packages have been widely promoted (Giller et al., 2010), based

on results from researcher-managed trials. Nevertheless, these trials have failed to

demonstrate the how, where, when and what technology would fit what situation. This

because field trials alone have made it harder to scrutinize the real source of reported CA

benefits making most results reported to date not generalizable since they fail to capture the

complex of the system and situation where CA is promoted. The complexity of the situation

involves not only the need to understand numerous components of the system i.e., farmer’s

endowment levels, access to inputs and markets, availability and quality of mulches, crop and

sites responses across seasons; but also risk perceptions, values, and aspirations of poorly

resourced farmers. In addition to that, the use of inorganic N-fertilizers a fundamental input

to validate CA and increase yields is SSA is prohibitive to cash constrained farmers (Chianu

et al., 2012; Rware et al., 2014), and is making fertilizer use less attractive to most farmers.

Despite resource access being essential, recent research has demonstrated that rather than

access itself, optimizing resource management, i.e., allocation across farms, crops and

locations (Bucagu et al., 2014; Giller et al., 2006; Rusinamhodzi et al., 2016) may play a key

role in improving the performance of the highly diverse and complex SSA farming systems

(Tittonell et al., 2010; Tittonell et al., 2007a; Vanlauwe et al., 2006). Therefore, this study

Page | 179

investigated: 1) how farmer’s circumstances and perception of their biophysical and

socioeconomic circumstances affect their resource use decisions and farming systems design

and management; 2) how can carbon rich residues and available limited amounts of N-

fertilizers be optimally used in rain fed N-deprived maize-legume cropping systems of central

Mozambique. This study was conducted around four key propositions:

1) There are household and site-specific managements e.g. fertilizers, manure, residues

and legumes, that best fit farmers’ bio-physical and socioeconomic circumstances.

2) The balance between the positive and negative effects from the use of carbon rich

residues in maize-legume systems can only be managed with nitrogen fertilisers.

3) In wetter nitrogen constrained environments, investing the limited availability of high

C/N ratio residues on improving the performance of the sole legume crop is more

beneficial than applying those residues on the maize crop. The benefits from this

resource allocation strategy are likely to be observed in the following cereal crop.

4) In dryer nitrogen constrained agricultural systems, investing the limited availability of

maize stubble on improving rainfall capture for the maize crop is likely to provide

more benefits than applying those residues on the legume crop.

Each of the results chapters, i.e., 3, 4 and 5, address one or a set of the propositions presented

above under the assumption that they are the key to optimize resource allocation strategies

and their responses. The main hypothesis in the thesis is that the likelihood to sustainably

improving resource productivity, yields, and the household food and income situation in the

N-deprived systems of central Mozambique, can be increased through a better allocation of

available resources. However, this can only be done through interactive co-design of new

practices with active farmer participation from the onset of the project. Furthermore, it is

hoped that the results from this research will be widely applicable to a range of agro-

ecologies of central Mozambique and across SSA since there is little generalizable

information that can help design sustainable intensification intervention that are more likely

to be adopted by poorly resourced farmers.

Page | 180

6.2 Typology tailored intervention: key entry points to downscale technologies

The results in chapter 3, show that, understanding the drivers for smallholder farmers’

decisions in the quest to meet seasonal food security and income generation targets is

paramount to learn how farmers can be actively involved in the co-design of more relevant

and actionable management options. As African smallholder farms are highly diverse in

space and time, grouping farmers in small homogeneous functional groups or typologies can

help target interventions so that adoption is more likely to take place (Daloğlu et al., 2014;

Dorward et al., 2009; Guillem et al., 2012). For the specific case of Mozambique, exploring

the intra-group diversity among small scale farmers in the existing rigid categorization (INE,

2011b) is key to develop recommendations that are more representative of farmers

circumstances. Here we proposed a framework to categorize farmers in small mutually

exclusive functional typologies.

The typologies proposed in this study are flexible in time and are in line with findings

showing how contrasting households access knowledge (Kraaijenbrink and Wijnhoven,

2008); and understanding of their levels of resources endowment to achieve food self-

sufficiency (Bucagu et al., 2014; Guillem et al., 2012) affected the way farmers designed and

managed their farms. Nevertheless, in central Mozambique, the relation between farmer’s

perception of their biophysical and socioeconomic circumstances and the way it affects local

farming systems design and management (Zingore et al., 2007) translated into technology

adoption traps. The adoption traps, are perception based mind-sets that reflects farmers

understanding of their circumstances and the farm management strategies they choose to

implement and believed are more likely to help them achieve their seasonal food security and

income generation targets considered their current circumstances. These management choices

are made in detriment of existing technological recommendation – fertilizers. Three key

adoption traps were identified across sampled typologies in central Mozambique, i.e.,

farmers’ perception about the level of soil fertility in their farm, the degree of maize self-

sufficiency, and availability of land i.e. farm size.

The results from chapter 3 also validate the proposition 1, that, there might be household

(typology) and site-specific management combinations of technologies e.g. fertilizers,

manure, residues, legumes and resources usage that best fit farmers’ bio-physical and

socioeconomic circumstances. For example, farmers’ investment on improved hybrid seed

Page | 181

planted mainly in the lowland mixed vegetable-maize farms that are normally fertilized since

farmers believe to have better returns from investments, is in line with the findings from

(Vanlauwe and Giller, 2006), who showed that farmers invest when they believe there is a

clear value-to-cost ratio. In the meanwhile, resource poor farmers, choose to diversify their

crops and cultivate more plots a season or engage in on and off the farm jobs in order to

improve their food and income prospects (Cunguara et al., 2011; Sichoongwe et al., 2014).

Therefore, understanding and taking these combinations of factors and management

circumstances into account is key to actively involve farmers in scaling out knowledge

intensive technologies such as the conservation agriculture principles and sustainable

intensification into locally feasible practices that are more likely to be tested and adopted by

farmers. However, it is also clear that given the diversity in households’ sources of

livelihoods there is no single-sized technology that will fit all households. Recent studies

revealed that factors such as the adopter characteristics (Meijer et al., 2015; Muzari et al.,

2012) are critical determinants of adoption.

Results from the household survey also showed that despite labour constraints (Leonardo et

al., 2015), farmers opt for extensive farming as a mean to minimize risk and achieve a

surpluses. Nevertheless, the average cultivated land area was usually not larger than 3ha/year

in our sampled population. Plot size is an issue to be further studied across typologies. Here,

identifying economic plot sizes for each typology is a critical step towards making farms

manageable and profitable (Ali et al., 2015). For example, a rough estimate indicates that

each household needs to produce a surplus of at least 7.8 ton/maize/year to generate enough

income to pay for 12 months of basic salary in the agricultural sector, i.e., ca. 960 USD/year

at a rate of 80 USD/month to the household head. Unfortunately, this cannot be achieved

from 0.5-1.0 ha of a maize monoculture with current varieties and levels of resource use.

Diversification of crops, income sources and cultivated land are the only ways this can be

achieved. The likelihood of producing surplus or generate the income from other farm and

off-farm activities will differ across groups. Therefore, an exhaustive analysis of household

dynamics and the trade-offs associated to different management strategy at both plot and

household level, needs to be considered in order to design resource efficient and profitable

systems capable of providing farmers with what they need.

Page | 182

6.3 Optimizing responses through smart and pragmatic resource allocation strategies

6.3.1 Residue and fertilizer responses: linking empirical and model assisted analysis

In chapter 4, the results clearly demonstrate that in the rain fed N-deprived systems of Central

Mozambique where there is enough soil moisture to trigger residue decomposition, a more

pragmatic and cautious approach needs to be taken when recommending residue application.

Field trial results showed that high C/N residues are better used to improve the performance

of the legume crop during the rotation phase rather than applied directly into an unfertilized

or poorly fertilized maize crop, where yield losses due to residue application were registered.

Nevertheless, positive responses to crop residue application into maize were attained with

increased N-applications. HNA, i.e., 92 kg N/ha, an amount far from the reach of most

resource poor farmers proved effective in offsetting the N-immobilization demand, allowing

enough N to be available to plant growth in tested seasons.

Previous research has also demonstrated that yield losses in CA cropping systems can be

alleviated by using inorganic N-application, which reduces N-immobilization demands and

improves N-release to the crop (Esther et al., 2014; Recous et al., 1995; Ribeiro et al., 2002).

Inorganic N-applications of around 80-100 kg N/ha in soils with considerable soil N content

were considered good enough to partially or fully off-set N-immobilization (Bakht et al.,

2009; Baudron et al., 2015a; Tao et al., 2015; Verhulst et al., 2011; Xiukang et al., 2015).

However, this is not the case for most resource poor households from central Mozambican.

Therefore, applying high C residues to legumes and using the limited amounts available of N

(ca. 23 kg N/ha) on the maize crop, appears to be the best resource allocation option for the

N-deprived maize-legume systems managed under conservation agriculture. This is because

better returns from residue and N-fertilizer application were measured in the legume-maize

sequence after just one season. This shift in practice is supported by the cowpeas positive

response to residues, as also reported in other studies before (Herrmann et al., 2014; Krishna,

2001; Shah et al., 2003a). In maize, fertilizer application at lower levels should only be

recommended in residue free systems, where better responses are attained.

Maize proven to be more responsive to increased N-application than cowpea. Cowpea yield

was actually reduced by N-applications above 23 kg N/ha. Despite, positive responses to N-

application in maize, responses differed across sites, with the wet sandy loam soil (SaL)

Page | 183

showing the highest overall agronomic response. The high returns from N-fertilizer

application in the wetter environments (SaL) rather than water limited fSa and cSaCL sites, is

in line with results from other studies (Cossani et al., 2010, 2012; Shangguan et al., 2000;

Sun et al., 2012; Teixeira et al., 2014). In these studies, strong interactive and co-limitation

effects between nitrogen use and soil water availability were observed. This is an important

finding, considering the high intra and cross-agroecological zone soil diversity in Central

Mozambique that implies that smarter fertilizer allocation strategies are required. Farmers

preferential fertilizer allocation into the wet lowland farming environment (Chapter 3), where

fertilizers are believed to offer better economic returns is in line with the interactive relation

between N-responses and water availability.

Fertilizer responses and productivity analysis demonstrated that the highest return from each

kg of N applied (NUE) was achieved under limited N-application (LNA), i.e., 23 kg N/ha

compared to 92 kg N/ha, in all the tested environments. This result is important in regards to

raising awareness about the importance of understanding the value-to-cost ratio of existing

blanket fertilizer recommendations in such unreliable rainfall environments (Dimes, 2011).

Nitrogen driven trade-offs between NUE and WUE were also observed (Sadras and

Rodriguez, 2010). In maize, increased N-application increased WUE and reduced WUE,

while increased water availability increased NUE to the detriment of NUE. A strong

relationship between NUE and its components was also observed with site moisture regime

playing a detrimental role in NUpE and its contribution to NUE. NUE in maize was highly

affected by variation in NUpE, while for cowpea NUtE had the highest impact on NUE,

especially at higher levels of N-application. Understanding all this interactions, resource

productivity trade-offs and the responses to several management conditions (e.g., residue

application and sites variability) is key to generate relevant and generalizable information

(Sadras and Lemaire, 2014) that could be used to optimize resource allocation.

6.3.2 Model assisted analysis and decision support framework

The results from long-term simulations with APSIM presented in Chapter 5, reproduced the

field trial results of chapter 4, and allowed further extrapolation. To facilitate the analysis in

an area with a vast array of soils, soil textures (soil WHC), was used to group residue

responses across soil types and agro-ecologies. The results demonstrated that there are site

Page | 184

and crop-specific responses to residues and inorganic N-fertilizer application that need to be

considered to design effective management recommendations in the N-deprived systems of

Central Mozambique. A 90% chance that a high C/N ratio residue application as mulch will

likely reduce maize yield, was simulated in considerably wet and high WHC soils – the

lowland sandy loam (SaL) and the upland sandy clay loam soils (cSaCL). For example,

simulated results showed an 80% chance to achieve 3 t/ha in residue free maize at 23 kg N/ha

in both cSaCL and SaL. This contrasted with only 40% at the cSaCL, and 10 % at the SaL

sites. In these soils, high soil moisture content leads to high microbial activity which triggers

faster residue decomposition resulting in N-immobilization, particularly in soils where not

enough N is available to satisfy N-immobilization demands and increase N-availability to

crops (Guntiñas et al., 2012; Probert et al., 1998; Tian et al., 1992). High N-immobilization

during residue decomposition in soils at 75% WHC compared to soils at 35% WHC was also

reported by Chen et al. (2012). In contrast, positive responses to residue application into

cowpea were simulated in almost 95% of the seasons in the same environments. This

validates our third proposition that in wetter nitrogen constrained environments, investing the

limited availability of high C/N ratio stubble on improving the performance of the sole

legume crop is more beneficial to the system than applying those residues on the maize crop.

The returns from this resource allocation strategy are likely to be observed in the following

cereal crop.

Results also showed that, the negative responses to high C/N ratio residue application on

maize yields, WUE and NUE will largely depend on the levels of N-application and type of

season. In both SaL and cSaCL soils, benefits from residue application were only observed

in the driest seasons, i.e., 5-10% seasons over the simulated period. Similar results were

reported in the Central Mexico Highlands where improved maize yields under residue

application were attained in dry years (Romero-Perezgrovas et al., 2014). In contrast to the

wet and high WHC soils, in drier and low WHC coarse textured soils (fSa), residue

application increased maize yields through improved water capture (WU) at all levels of N-

application. In these dry and low WHC soils, the poor microbial activity delays

decomposition (Chen et al., 2012), allowing residues to remain longer and provide further

soil cover. Benefits from residue application in drier and coarse textured soils were also

reported by Gentile et al. (2008). These are the only soils were residue applications into

maize proved to be more beneficial to the system compared to applications into legumes.

Page | 185

6.3.4 Demonstrating short term effects from alternative resource allocation strategies

Demonstrating short term effects from investments in residues and fertilizer application, in

the N-deprived smallholder farming systems is a critical challenge defined in the CA research

agenda of most research and development institutions across SSA (Giller et al., 2011a).

Rising awareness about short-term effects can help promote a more conscious and informed

adoption process. In this study higher returns from residue and limited N-fertilizer application

were obtained for the legume phase of the cowpea-maize sequence. High returns from

investment in legume-cereal sequences have been reported in several other studies before

(Luce et al., 2015; Monzon et al., 2014; Neugschwandtner and Kaul, 2015), mainly resulting

from high N-turnover and carry-over effect from the retained N-rich legume biomass. In the

continuous maize systems, penalties to yield from residue application and retention (NUpE

and NUE), were observed in the subsequent bio-assay maize crop. Penalties from yearly

residue incorporation in continuous maize systems have also been reported in the United

States corn belt (Gentry et al., 2013; Sindelar et al., 2013), though those are high residue and

N-input systems.

For Central Mozambique and SSA N-deprived smallholder farming systems managed under

conservation agriculture, where residue retention has been widely recommended as a mean to

reduce soil erosion, yield penalties from residue retention in continuous maize systems need

to be further studied in order to find effective ways to alleviate them. In the meanwhile,

promoting regular legume rotations (Morrison, 2013; Stute and Posner, 1995) appears to be a

cost-effective option to minimize such penalties among resource poor farmers who are unable

to invest in fertilizers. In these systems, effective legume phases can also help reduce the

economic optimum of N-fertilizer application (Luce et al., 2015). In our experiments

(Chapter 4), differences between the yield of fertilized continuous maize systems and the

unfertilized maize yield in the cowpea-maize sequence were only found for in specific

treatments. This means that more pragmatic and smarter resource management strategies

need to be considered for the N-deprived smallholder farming systems. However, this needs

further research. For the specific case of the N-deprived systems of Central Mozambique,

bold measures need to be considered in order to increase soil N-availability and reduce

penalties from yearly incorporation of high C/N ratio residues in continuous maize systems.

Climatic analyses for Central Mozambique indicate the existence of regions where double

growing periods are possible (Reddy, 1984). This means that an opportunistic legume crop,

either early or as a relay crop, could be grown. In the course of this research, an early cowpea

Page | 186

crop was successfully grown (Nhantumbo et al., 2014), leaving enough time to grow a

second crop starting in February. This proves that two legume crops can be grown in the

region, even if only one is for grain and the other one as a green manure cover crop.

6.4 Conclusions and way forward

From this study it can be concluded that, apart from the protracted problem of limited access

to production means, optimizing resource allocation across activities within the household is

critical to improve the performance of the whole farm enterprise and consequently their food

and income generation prospects. However, because the farm enterprise, its design and

management approach differs in space, time and across typologies, there is no single-sized

technological package capable of solving all farmers’ problems at once. Therefore, an

integrated socio-technical approach that looks at the whole household “farm enterprise”

dynamics (i.e., from the homestead to the plot level management) is required. This is because

the overall household livelihood strategy and the farmers’ ability to meet their annual food

and income generation targets, affects their resource use behaviour – investment and

relocation across activities and plot level management. Personalized typology specific

agricultural interventions are required to effectively involve farmers in the negotiation and

co-design of more farmer suitable and sustainable farming systems and management

practices.

At the filed level, the results have demonstrated that there is crop, site and management

specific responses and trade-offs that need to be understood and explored in a smarter and

more pragmatic way. This, in order to improve resource productivity and yields in N-deprives

smallholders farming systems managed under conservation agriculture in Central

Mozambique. However, because field trials can only capture a limited part of the vast and

complex GxExM interactions and their effect on resource productivity and yields, model

assisted research and decision support systems are critical to improve decision making

processes, especially in contexts where research for development is poorly funded, like

Mozambique. Nevertheless, model integration into local research and extension platforms

needs to be part of a wider intervention platform that not only looks at plot level responses,

but which can also capture and exploit the inter linkages between components of the system –

a whole household farm enterprise. It’s critical, however, to combine agronomic simulation

Page | 187

models such as APSIM with agent based models (ABM) (e.g., multiple goal programming or

the trade-off analysis model (TOA-MD Model)) to set clear optimization goals and co-design

development roadmaps to each farm enterprise and region. This approach is critical in the

sense that it would help generate information that can help farmers visualize all available

alternatives on what they can achieve for their farms and what approach better suits their

circumstances.

6.4.1 Scaling out and scaling up innovations

Central Mozambique, presents a vast mosaic of agro-ecologies with highly diverse edaphic

and climatic characteristics. For example, in the four provinces that constitute the Central

Mozambique agricultural growth corridor, at least 3 agro-ecologies can be identified in each

one of them – Manica (AEZ-R4, AEZ-R6, and AEZ-R10), Sofala (AEZ-R5, AEZ-R4, and

AEZ-R6), Tete (AEZ-R6, AEZ-R7, and AEZ-R10) and Zambezia (AEZ-R7, AEZ-R5, AEZ-

R8, and AEZ-R10). Moreover, clear intra and cross-agroecological zone differences with

regards to soil type, soil fertility and soil texture induced in crop moisture regimes are also

characteristic of the region, and dictate how farmers manage their systems within the context

of existing and perceived variability. Farmer diversity is also an issue that characterizes the

agricultural sector in the region, as our results demonstrated. All these factors make farming a

highly complex activity in the region. Because no single-sized technological package will

help solve all farmers’ problems at once, understanding how each group of farmers perceive

their circumstances and manage their farm enterprise is critical to actively involve them in the

co-generation of relevant agricultural information that can help make farming a smarter and

more effective activity, is critical. However, finding timely and cost-effective methods to co-

generate relevant and pragmatic agricultural information that can provide feedback to farmers

within a time frame that can allow them to reflect and use it to improve their farming systems

design and management still constitutes a major challenge.

In this thesis, three new approaches that need to be further explored at both national and local

level by research institutions and also local agricultural authorities, since they have the

potential to help minimize the existing information gap and its relevance burden by making

agricultural interventions more reflective of farmers’ circumstances and the complexity of

their farming systems design, management and their agro-environment are proposed:

Page | 188

1) Personalized agricultural interventions: a framework to cluster and characterize

farmers in new simple and mutually exclusive functional typologies that reflect

the diversity among smallholder farmers, based on their resource use behaviour

and openness to engage in new practices. Contrary to the current farmer

classification that assumes a wide homogeneity among farmers, the approach

proposed in this thesis is based on the proposition that farmers are different, and

their livelihood strategy and perception of their own circumstances shapes how

they manage their whole household farm enterprise.

2) Creating more generalizable information by conducting research that explores the

interactions within the farm household enterprise and finding ways to amplify or

minimize management associated trade-offs: Model assisted decision support

frameworks for residue and fertilizer management in N-deprived systems. The

frameworks will allow researchers to extend the scope of their analysis in order to

produce information that has the potential for more general application, this being

a problem to date. This framework used a simplistic approach that allows the

grouping of several soils into the three textural classes – heavy, medium and light

textured, as a key to further explore the responses and resource productivity trade-

offs associated with residues and N-fertilizer allocation strategies across farms and

agro-ecologies.

3) Bringing researchers and farmers closer: The use of model assisted participatory

crop season planning has also been tested and proved to be a valid option for

integration into local research and extension platforms. This is an interactive

method that allows a closer farmer-researcher interaction.

Taking into account the fact that the farm enterprise is managed differently across typologies,

depending on each household biophysical and socioeconomic circumstances, which

influences current farming system design and management, agricultural interventions needs

to match household’s circumstances. Therefore, the following management challenges were

identified as key to sustainably improve current farming systems performance and boost each

farm typology and the household ability to meet their annual food and income generation

targets:

Page | 189

1) For resource endowed Type-B and Type-A2 farms who are already maize self-

sufficient and have the necessary investment capacity to purchase the basic inputs

required, the challenge is to improve resource productivity (land, labour and

inputs) and yields in order to generate enough surplus and on farm income. This

can be achieved through smarter resource mechanization of farm activities and

relocation strategies.

2) For resource constrained Type-A1 farmers, the primary challenge is to generate

enough income either on-farm or off-farm to supplement their low maize self-

sufficiency. For these farms, improving legume productivity appears to be the best

option considered legumes high market price and further contribution to soil N-

availability. Diversification of farm activities is also a better option for this group

since it will reduce the risk.

3) Maximizing land use and productivity through better crop arrangement and

improved plot level management is critical to improve returns from land and

labour. Here, mechanizing farm activities such as sowing and weeding is

fundamental for all groups especially considering the fact that labour availability

and quality are limiting and each farm will need to smartly and efficiently

cultivate all owned land to produce enough surplus to generate on farm income to

cover household daily expenses. This is critical in Central Mozambique, were land

is not yet a limiting factor, but surplus will hardly be achieved in 0.5 ha plots of

land in a region with less than 50 days of rainfall a season.

At field level improving resource allocation and productivity is a key area to be addressed

across all farm types. Here understanding field level responses and resource allocation trade-

offs is critical to take advantage of existing cross-site variability and potential short term

benefits. Therefore, the following rules of thumb are proposed in order to guide resource

allocation – residues and inorganic N fertilizers, across farms, in the region:

1) Although residues from previous maize crops and weeds germinated during

unmanaged fallows, are incorporated across all farm types, residues application

into maize is only viable if enough fertilizer (≥ 92 kg N/ha), can be mobilized to

Page | 190

satisfy N-immobilization demands which is not the case of most resource poor

farmers. Therefore, shifting residue application into legumes is the best short term

option.

2) For the ones using inorganic N-fertilizers, investing the limited amounts of N-

fertilizers into improving the performance of the legume crop (biomass and yield)

or increasing the yield of residue free maize is the best N-fertilizer allocation

strategy compared to applying this N into residue applied maize;

3) Cattle owners and those who can afford to purchase enough manure should also

consider incorporating manure into maize fields rather than just using it in

vegetable production fields were it is likely to help improve soil biophysical

proprieties.

Because central Mozambique presents a vast mosaic of soils and farming environment, it is

critical that resource allocation strategies match existing cross-site variability. To fine tune

residue and fertilizer allocation and field level, a set of rules based on existing soil WHC

gradients in the region are proposed:

1) In high water holding capacity (WHC) soils where moisture is not a limiting

factor, high C/N ratio residues should be used to improve the performance of the

sole legume crop during the legume phase of the rotation rather than applied into

an unfertilized cereal crop where the soil moisture advantage provided by residues

(WUE increments) do not compensate for the yield losses due to N-

immobilization;

2) High C/N ratio residues only offer a significant yield advantage to the cereal crops

in dryer and low WHC soils. In these soils, residues significantly improve incrop

rainfall capture generating a moisture benefits that surpasses the negative impacts

from N-immobilization on maize yield;

3) Best responses from inorganic N-fertilizer (NUE, NUpE, NUtE and NHI) are

likely to be attained in wet and high WHC environments where there is enough

moisture for the crop to efficiently use the supplied inorganic N fertilizer. This

because in dry and low WHC soils, poor soil moisture regimes reduce crop

nitrogen uptake leading to poor responses to inorganic N-application;

4) For wet and high WHC soils, the beneficial effects of applying crop residues on

the legume crop are likely to be observed on the following cereal crop.

Page | 191

References

Ali, D., Daniel Ayalew, A., and Klaus, D. (2015). Is There a Farm Size-Productivity

Relationship in African Agriculture? Evidence from Rwanda. Land economics 91,

479-500

Bakht, J., Shafi, M., Jan, M. T., and Shah, Z. (2009). Influence of crop residue management,

cropping system and N fertilizer on soil N and C dynamics and sustainable wheat

(Triticum aestivum L.) production. Soil and Tillage Research 104, 233-240.

Baudron, F., Delmotte, S., Corbeels, M., Herrera, J. M., and Tittonell, P. (2015a). Multi-scale

trade-off analysis of cereal residue use for livestock feeding vs. soil mulching in the

Mid-Zambezi Valley, Zimbabwe. Agricultural Systems 134, 97-106.

Baudron, F., Thierfelder, C., Nyagumbo, I., and Gérard, B. (2015b). Where to Target

Conservation Agriculture for African Smallholders? How to Overcome Challenges

Associated with its Implementation? Experience from Eastern and Southern Africa.

Environments 2, 338-357.

Bucagu, C., Vanlauwe, B., Van Wijk, M., and Giller, K. (2014). Resource use and food self-

sufficiency at farm scale within two agro-ecological zones of Rwanda. Food Security

6, 609-628.

Chen, Q.-H., Feng, Y., Zhang, Y.-P., Zhang, Q.-C., Shamsi, I. H., Zhang, Y.-S., and Lin, X.-

Y. (2012). Short-Term Responses of Nitrogen Mineralization and Microbial

Community to Moisture Regimes in Greenhouse Vegetable Soils. Pedosphere 22,

263-272.

Chianu, J., Chianu, J., and Mairura, F. (2012). Mineral fertilizers in the farming systems of

sub-Saharan Africa. A review. Agronomy for Sustainable Development 32, 545-566.

Cossani, C. M., Slafer, G. A., and Savin, R. (2010). Co-limitation of nitrogen and water, and

yield and resource-use efficiencies of wheat and barley. Crop and Pasture Science 61,

844-851.

Cossani, C. M., Slafer, G. A., and Savin, R. (2012). Nitrogen and water use efficiencies of

wheat and barley under a Mediterranean environment in Catalonia. Field Crops

Research 128, 109-118.

Page | 192

Cunguara, B., Langyintuo, A., and Darnhofer, I. (2011). The role of nonfarm income in

coping with the effects of drought in southern Mozambique. Agricultural Economics

42, 701-713.

Daloğlu, I., Nassauer, J. I., Riolo, R. L., and Scavia, D. (2014). Development of a farmer

typology of agricultural conservation behavior in the American Corn Belt.

Agricultural Systems 129, 93-102.

Dimes, J. (2011). Formulating Crop Management Options for Africa’s Drought-Prone

Regions: Taking Account of Rainfall Risk Using Modeling. In “Innovations as Key to

the Green Revolution in Africa.” (A. Bationo, B. Waswa, J. M. M. Okeyo, F. Maina

and J. M. Kihara, eds.), pp. 785-794. Springer Science and Business Media,

Netherlands.

Dorward, A., Anderson, S., Bernal, Y. N., Vera, E. S., Rushton, J., Pattison, J., and Paz, R.

(2009). Hanging in, stepping up and stepping out: livelihood aspirations and strategies

of the poor. Development in Practice 19, 240-247.

Esther, O. J., Guo, C.-h., Tian, X.-h., Li, H.-y., and Zhou, Y.-x. (2014). The Effects of Three

Mineral Nitrogen Sources and Zinc on Maize and Wheat Straw Decomposition and

Soil Organic Carbon. Journal of Integrative Agriculture 13, 2768-2777.

Gentile, R., Vanlauwe, B., Chivenge, P., and Six, J. (2008). Interactive effects from

combining fertilizer and organic residue inputs on nitrogen transformations. Soil

Biology and Biochemistry 40, 2375-2384.

Gentry, L. F., Ruffo, M. L., and Below, F. E. (2013). Identifying Factors Controlling the

Continuous Corn Yield Penalty. Agronomy Journal 105, 295-303.

Gilbert, N. (2012). Dirt Poor: The key to tackle hunger in Africa is enriching its Soils. Nature

483, 525-227.

Giller, K. E., Corbeels, M., Nyamangara, J., Triomphe, B., Affholder, F., Scopel, E., and

Tittonell, P. (2011). A research agenda to explore the role of conservation agriculture

in African smallholder farming systems. Field Crops Research 124, 468-472.

Giller, K. E., Rowe, E. C., de Ridder, N., and van Keulen, H. (2006). Resource use dynamics

and interactions in the tropics: Scaling up in space and time. Agricultural Systems 88,

8-27.

Page | 193

Guillem, E. E., Barnes, A. P., Rounsevell, M. D. A., and Renwick, A. (2012). Refining

perception-based farmer typologies with the analysis of past census data. Journal of

Environmental Management 110, 226-235.

Guntiñas, M. E., Leirós, M. C., Trasar-Cepeda, C., and Gil-Sotres, F. (2012). Effects of

moisture and temperature on net soil nitrogen mineralization: A laboratory study.

European Journal of Soil Biology 48, 73-80.

Heffer, P., and Prud'homme, M. (2011). "Fertilizer Outlook 2011-2015." IFA, Paris, France.

Herrmann, L., Chotte, J. L., Thuita, M., and Lesueur, D. (2014). Effects of cropping systems,

maize residues application and N fertilization on promiscuous soybean yields and

diversity of native rhizobia in Central Kenya. Pedobiologia 57, 75-85.

INE (2011). Censo Agropecuário 2009-2010: Resultados Preliminares. pp. 105. Instituto

Nacional de Estatistica(INE), Ministerios da Agricultura (MINAG), Mapuo,

Mozambique.

Kraaijenbrink, J., and Wijnhoven, F. (2008). Managing heterogeneous knowledge: a theory

of external knowledge integration. Knowledge Management Research and Practice 6,

274-286.

Krishna, N. R. (2001). Effects of Cereal and Legume Cover Crop Residues on Weeds, Yield,

and Net Return in Soybean (Glycine max). Weed Technology 15, 660-668.

Leonardo, W., van de Ven, G. J., Udo, H., Kanellopoulos, A., Sitoe, A., and Giller, K.

(2015). Labour not land constrains agricultural production and food self-sufficiency in

maize-based smallholder farming systems in Mozambique. Food Security, 1-18.

Meijer, S. S., Catacutan, D., Ajayi, O. C., Sileshi, G. W., and Nieuwenhuis, M. (2015). The

role of knowledge, attitudes and perceptions in the uptake of agricultural and

agroforestry innovations among smallholder farmers in sub-Saharan Africa.

International Journal of Agricultural Sustainability 13, 40-54.

Monzon, J. P., Mercau, J. L., Andrade, J. F., Caviglia, O. P., Cerrudo, A. G., Cirilo, A. G.,

Vega, C. R. C., Andrade, F. H., and Calviño, P. A. (2014). Maize-soybean

intensification alternatives for the Pampas. Field Crops Research 162, 48-59.

Morrison, L. (2013). Residue management can reduce continuous corn yield penalty. Corn

and Soybean Digest. Vol. 75 Issue 9, p 40

Page | 194

Muzari, W., Gatsi, W., and Muvhunzi, S. (2012). The Impacts of Technology Adoption on

Smallholder Agricultural Productivity in Sub-Saharan Africa: A Review. Journal of

Sustainable Development 5, 69-77.

Neugschwandtner, R. W., and Kaul, H.-P. (2015). Nitrogen uptake, use and utilization

efficiency by oat–pea intercrops. Field Crops Research 179, 113-119.

Ngwira, A. R., Thierfelder, C., and Lambert, D. M. (2013). Conservation agriculture systems

for Malawian smallholder farmers: long-term effects on crop productivity,

profitability and soil quality. Renewable Agriculture and Food Systems 28, 350-363.

Nhantumbo, N., Dias, J., Mortlock, M., Nyagumbo, I., Dimes, J., and Rodriguez, D. (2014).

Best-fit residue allocation: A gate for legume intensification in nitrogen constrained

cropping systems of Central Mozambique. In "First African Congress on

Conservation Agriculture" (A. C. T. Network, ed.), pp. 33-35. ACT, 18-21 March

Lusaka, Zambia.

Probert, M. E., Dimes, J. P., Keating, B. A., Dalal, R. C., and Strong, W. M. (1998). APSIM's

water and nitrogen modules and simulation of the dynamics of water and nitrogen in

fallow systems. Agricultural Systems 56, 1-28.

Recous, S., Robin, D., Darwis, D., and Mary, B. (1995). Soil inorganic N availability: Effect

on maize residue decomposition. Soil Biology and Biochemistry 27, 1529-1538.

Reddy, S. J. (1984). Clima Geral de Mocambique in "Serie Terra e Água" (INIA, ed.), pp. 41.

Instituto de Investigação Agronómica de Moçambique (INIA), Departamento de Terra

e Agua (DTA), Maputo, Mozambique.

Ribeiro, C., Madeira, M., and Araújo, M. C. (2002). Decomposition and nutrient release from

leaf litter of Eucalyptus globulus grown under different water and nutrient regimes.

Forest Ecology and Management 171, 31-41.

Romero-Perezgrovas, R., Verhulst, N., De La Rosa, D., Hernandez, V., Maertens, M.,

Deckers, J., and Govaerts, B. (2014). Effects of Tillage and Crop Residue

Management on Maize Yields and Net Returns in the Central Mexican Highlands

Under Drought Conditions. Pedosphere 24, 476-486.

Rusinamhodzi, L., Dahlin, S., and Corbeels, M. (2016). Living within their means:

Reallocation of farm resources can help smallholder farmers improve crop yields and

soil fertility. Agriculture, Ecosystems and Environment 216, 125-136.

Page | 195

Rware, H., Wairegi, L., Oduor, G., Macharia, M., Romney, D., Tarfa, B. D., de Maria, R.,

Ley, G., Tetteh, F., Makumba, W., Dicko, M., Nouri, M., Leon, N. N., Gondwe, B.,

Kaizzi, K., Negash, D., Kibunja, C., Alawy, A., Wartmann, C., and Rutsimba, E.

(2014). Assessing the Potential to Change Stakeholders Knowledge and Practices on

Fertilizer Recommendations in Africa. Agricultural Sciences 5, 1384-1391.

Sadras, V. O., and Lemaire, G. (2014). Quantifying crop nitrogen status for comparisons of

agronomic practices and genotypes. Field Crops Research 164, 54-64.

Sadras, V. O., and Rodriguez, D. (2010). Modelling the nitrogen-driven trade-off between

nitrogen utilisation efficiency and water use efficiency of wheat in eastern Australia.

Field Crops Research 118, 297-305.

Shah, Z., Shah, S. H., Peoples, M. B., Schwenke, G. D., and Herridge, D. F. (2003). Crop

residue and fertiliser N effects on nitrogen fixation and yields of legume-cereal

rotations and soil organic fertility. Field Crops Research 83, 1-11.

Shangguan, Z. P., Shao, M. A., and Dyckmans, J. (2000). Nitrogen nutrition and water stress

effects on leaf photosynthetic gas exchange and water use efficiency in winter wheat.

Environmental and Experimental Botany 44, 141-149.

Sichoongwe, K., Mapemba, L., Tembo, G., and Ng'ong'ola, D. (2014). The Determinants and

Extent of Crop Diversification Among Smallholder Farmers: A Case Study of

Southern Province Zambia. Journal of Agricultural Science 6, 150-159.

Sindelar, A. J., Coulter, J. A., Lamb, J. A., and Vetsch, J. A. (2013). Agronomic Responses of

Continuous Corn to Stover, Tillage, and Nitrogen Management. Agronomy Journal

105, 1498-1506.

St. Luce, M., Grant, C. A., Zebarth, B. J., Ziadi, N., O’Donovan, J. T., Blackshaw, R. E.,

Harker, K. N., Johnson, E. N., Gan, Y., Lafond, G. P., May, W. E., Khakbazan, M.,

and Smith, E. G. (2015). Legumes can reduce economic optimum nitrogen rates and

increase yields in a wheat–canola cropping sequence in western Canada. Field Crops

Research 179, 12-25.

Stute, J. K., and Posner, J. L. (1995). Legume Cover Crops as a Nitrogen Source for Corn in

an Oat-Corn Rotation. Journal of Production Agriculture 8, 385-390.

Page | 196

Sun, Y., Ma, J., Sun, Y., Xu, H., Yang, Z., Liu, S., Jia, X., and Zheng, H. (2012). The effects

of different water and nitrogen managements on yield and nitrogen use efficiency in

hybrid rice of China. Field Crops Research 127, 85-98.

Tao, Z., Li, C., Li, J., Ding, Z., Xu, J., Sun, X., Zhou, P., and Zhao, M. (2015). Tillage and

straw mulching impacts on grain yield and water use efficiency of spring maize in

Northern Huang–Huai–Hai Valley. The Crop Journal 3, 445-450.

Teixeira, E. I., George, M., Herreman, T., Brown, H., Fletcher, A., Chakwizira, E., de Ruiter,

J., Maley, S., and Noble, A. (2014). The impact of water and nitrogen limitation on

maize biomass and resource-use efficiencies for radiation, water and nitrogen. Field

Crops Research 168, 109-118.

Thierfelder, C., Cheesman, S., and Rusinamhodzi, L. (2012). A comparative analysis of

conservation agriculture systems: Benefits and challenges of rotations and

intercropping in Zimbabwe. Field Crops Research 137, 237-250.

Thierfelder, C., Matemba-Mutasa, R., and Rusinamhodzi, L. (2015). Yield response of maize

(Zea mays L.) to conservation agriculture cropping system in Southern Africa. Soil

and Tillage Research 146, Part B, 230-242.

Tian, G., Kang, B. T., and Brussaard, L. (1992). Biological effects of plant residues with

contrasting chemical compositions under humid tropical conditions—Decomposition

and nutrient release. Soil Biology and Biochemistry 24, 1051-1060.

Tittonell, P., Muriuki, A., Shepherd, K. D., Mugendi, D., Kaizzi, K. C., Okeyo, J., Verchot,

L., Coe, R., and Vanlauwe, B. (2010). The diversity of rural livelihoods and their

influence on soil fertility in agricultural systems of East Africa – A typology of

smallholder farms. Agricultural Systems 103, 83-97.

Tittonell, P., Vanlauwe, B., de Ridder, N., and Giller, K. E. (2007). Heterogeneity of crop

productivity and resource use efficiency within smallholder Kenyan farms: Soil

fertility gradients or management intensity gradients? Agricultural Systems 94, 376-

390.

Vanlauwe, B., and Giller, K. E. (2006). Popular myths around soil fertility management in

sub-Saharan Africa. Agriculture, Ecosystems and Environment 116, 34-46.

Page | 197

Vanlauwe, B., Tittonell, P., and Mukalama, J. (2006). Within-farm soil fertility gradients

affect response of maize to fertiliser application in western Kenya. Nutrient Cycling in

Agroecosystems 76, 171-182.

Verhulst, N., Govaerts, B., Nelissen, V., Sayre, K. D., Crossa, J., Raes, D., and Deckers, J.

(2011). The effect of tillage, crop rotation and residue management on maize and

wheat growth and development evaluated with an optical sensor. Field Crops

Research 120, 58-67.

Wall, P. C. (2007). Tailoring Conservation Agriculture to the Needs of Small Farmers in

Developing Countries. Journal of Crop Improvement 19, 137-155.

Xiukang, W., Zhanbin, L., and Yingying, X. (2015). Effects of mulching and nitrogen on soil

temperature, water content, Nitrate-N content and maize yield in the Loess Plateau of

China. Agricultural Water Management 161, 53-64.

Zingore, S., Murwira, H. K., Delve, R. J., and Giller, K. E. (2007). Soil type, management

history and current resource allocation: Three dimensions regulating variability in

crop productivity on African smallholder farms. Field Crops Research 101, 296-305.

Page | 198

Appendix 1.1 Mozambique agroecological regions with their main farming systems

Altitude Description Agro-

ecological

zone (AEZ)

Farming Systems

Low altitude

Semi-arid and arid

regions of Southern

Mozambique

In lad semi-arid zone in Maputo province R1 Dry land maize and pastoralism

Semi-arid littoral in southern Mozambique,

Inhambane and Maputo Province

R2 Dry land maize-legume based systems. Coconut

based systems in the cost intercropped with

groundnuts and cassava

Arid and semi-arid parts of in land Gaza, north

Maputo and Inhambane province

R3 Dry land maize, lowland irrigated rice and maize

systems (Chókwè) with Commercial vegetables;

Cotton based systems; crop-livestock systems

Medium altitude

Covers almost 80% of Manica province and part of

Sofala in central Mozambique

R4 Maize-legume with Banana based systems in the

highlands of Manica

Coastal area of central Mozambique covering

Sofala and Zambezia province

R5 Maize-legume, rice based systems and

commercial sugar cane

Semi-arid region of Southern Tete and northern

Manica and Zambezia

R6 Livestock dominated systems with sorghum and

dryland maize

The largest agroecological region, covering 5

provinces in Central and Northern Mozambique,

namely Zambezia, Nampula, Tete, Niassa and

Cabo Delgado

R7 Maize-legumes and cassava-legume systems;

Cotton and tobacco based systems. Groundnuts,

beans, pegeonpea are important legume crops

High altitude

Northern Mozambique littoral covers Zambezia,

Nampula and Cabo Delgado

R8 Coconuts based systems

Maize/cassava-legume systems

Highlands of Cabo Delgado – planalto de Mueda R9 -

Manica, Tete, Zambezia and Niassa highlands R10 Maize-legume systems

Small patches of wheat in Rotanda

Commercial vegetable gardens

Compiled from: (Kajisa and Payongayong, 2011; Lukanu et al., 2009; Reddy, 1984)

Page | 199

Appendix 3.1 Generalized crop management timelines across regions and farmer groups

Activity Timing

Land preparation Land preparation mostly takes place from September-December before maize sowing.

In Rotanda, ploughing right after harvest in March/April and August/September is a

common practice among cattle owners, to incorporate residues and give them time to

decompose.

Maize sowing

window

Normally split across 15 October to 15 December: Early sowing in mid-October to

late November after the first rainfall, depending on the region. This window is mostly

used by resourceful farmers who prepare their land on time.

Normal sowing in early December using early to late maturity varieties. In irrigated

areas: mid-September and irrigated until the start of the rainy season.

Legumes sowing

window

Small patches of cowpea at the same time as maize for leaf and fresh pods.

The main window is in January/February after the rains, extending to March in the

wetlands. Harvest is normally in April/May. Sole legumes–common beans are sown

in March in irrigated areas and in the wetlands.

Varieties used Maize: local varieties, Matuba (OPV) and local hybrids PAN 67 and PAN 53 from

Pannar Seeds; Beans: Diacol Calima (IIAM).

Cowpea: IT16 and IT18 (IIAM) but mostly local varieties with indeterminate growth.

Spacing Random sowing with 3-6 pls/station for farmers without animal draft power. A better

in-line sowing pattern is seen with farmers with animals. Plant spacing recommended

by extension: 90x50 cm for Matuba, 90x40 cm for PAN67 and 90x60 cm for

Rasposta, with 2 pls/station, as recommended in Farmer Field Schools.

Fertilizer

application

(when available)5

Basal: at sowing with 3.4g/hill of NPK 12:24:12

Top dressing: 30-40 DAS with 3.4 g/hill Urea (46% N). This application coincides

with the second weeding.

Weeding 1st weeding at 21-30 DAS and 2nd weeding in January around tasselling. The third

weeding takes place in early February after the first rains and is also intended to

prepare land for beans that will planted as a relay crop.

5 8 farmers of the 20 that attended the PCM workshop in Rotanda, reported using fertilizer in the last season. The

amounts reported were, 75 kg NPK 12:24:12 (1.5-50 kg bags in a 1ha), 100 kg Urea (2-50 kg bag in 1ha), 100 kg of

Urea (2-50kg bags in 0.5ha), 100 kg of urea (2-50 kg bags in 0.5ha), 100 kg of urea (2-50 kg bags in 1ha), 100 kg of

urea (2-50 kg bags in 0.5ha), 50 kg of urea (1-50 kg bags in 0.5ha) and 100 kg of urea (2-50 kg bags in 1ha)

Page | 200

Harvest Harvesting from late March to early May for maize. Beans are harvested around May-

June.

Crop residues Left in the field in Macate. In Rotanda, the residues are normally incorporated after

harvest for those with access to draft animal power, and sometimes burned during

land preparation by those without access to draft animals. Residues are not sold and

animal graze freely after harvest.

Manure All present animal owners reportedly use manure but most of it is applied into

commercial vegetables production.

Appendix 3.2 Crop distribution across farming environments and soils in both AEZ-R4

and AEZ-R10.

Soil Type (local name) Fertility Yield6 Most common crop allocation

Red sandy clay loam

(Kupswuka)

Average Average Maize, sunflower, sorghum, legumes, cotton and

sesame

Sandy soil (Djetcha) Poor Low Maize-sorghum-sunflower and legumes

Dark sandy loam

(N’dongo & Matoro)

High Higher Vegetables, maize, legumes, tubers and root crops

6 Yield capacity is defined on a maize base: In an average season, without applying any fertilizer or

manure, farmers in Rotanda (R10) reportedly harvest between 27 bags of 50kg/bag maize in the lowland

wet environments and only 16 bags in upland environments using recycled local seed. During the focus

group discussion, a farmer reported having harvested 32 bags of 50 kg/bag in the lowland environment in

the 2012-13 season after planting 0.5 ha with PAN67, a high yielding hybrid with a yield potential of 6-8

t/ha

Page | 201

Appendix 3.4: Farm household clusters by their fertilizer and manure allocation

preferences

Figure A-0-1 Farm household clustered by fertilizer and manure allocation across

crops. Codding: X-Y-Z(v): number X is gender of the household head (1-M 2-F), Y-Z

household size and farm active people; (v) and (M), where (v) is Matsinho in Vanduzi

district and (M) is Marera in Macate district. Resource use: fert+manure: household uses

fertilisers and also apply manure; Nofert_Usemanure: household doesn’t use fertilizer but

applies manure; Allocation preference presented according to the resource in use: VEG -

resource applied into vegetable, Mz - applied into maize, Leg – applied into a legume,

shared – resource shared across all crops.

Page | 202

Appendix 3.5: Households clustered by their reasoning for not using fertilizers or manure

Codding: Household characteristics and location indicated through X-Y-Z(v): number X is gender of

the household head (1-Male, 2-Female); Y-Z are household size and farm active people respectively;

(v) and (M) stand for sampled communities, where (v) is Matsinho in Vanduzi district and (M) is

Marera in Macate district. Resource use: fert+manure: household uses fertilisers and also apply

manure; Nofert_Usemanure: doesn’t use fertilizer but applies manure; Nofert_NoManure: doesn’t use

both. Reasons for not using fertilizer and manure: f(X-Y) reasons for not using fertilizers. 1- No

interest, 2 - See no benefit, 3 - high cost, 4 – No local supplier, 5 – don’t know how to use them

properly; m(X-Y) reasons for not using manure: 1- No interest, 2 – Don’t see a benefit, 3 - high cost,

4 – No local supplier, 5 – don’t know how to use them properly, 6 – other

Page | 203

Appendix 3.6: Fertilizer and manure allocation preferences among residue users

Codding: Household characteristics and location indicated through X-Y-Z(v): number X is gender of

the household head (1-Male, 2-Female), Y-Z are household size and farm active people, respectively;

(v) and (M) stand for sampled communities where (v) is Matsinho in Vanduzi district and (M) is

Marera in Macate district. Resource use: fert+manure: household uses fertilisers and also applies

manure; Nofert_Usemanure: household doesn’t use fertilizer but applies manure; Allocation

preference presented according to the resource in use: VEG - resource applied into vegetable, Mz -

applied into maize, Leg – applied into a legume, shared – resource shared across crops in the farm

enterprise

Page | 204

Appendix 3.7: Correlation between household size and number of household farm active

people disaggregated by age and sex

Legend: 1-household size, 2-household farm active people (age>16), 3-household farm active

older people (age>45), 4-household farm active youth (age: 16-35), 5-household farm active

female members, 6-household farm active male members.