Systems Science at the scale of impact reconciling bottom up participation with the production of widely applicable research outputs by Dr. Fergus Sinclair, Systems Science Leader, ICRAF
This document discusses the challenges of scaling up agroforestry systems to improve livelihoods across diverse landscapes and contexts in Africa. It argues that agronomic recommendations do not account for the complexity of farming systems within households and landscapes. To effectively scale up, research needs to characterize the fine-grained variation in factors like soil, climate, practices, markets and policy across regions. Participatory methods are then needed to develop a portfolio of intensification options tailored to different contexts. Monitoring the performance of these options in diverse environments can build understanding of what works where to improve livelihood systems at scale.
Similar to Systems Science at the scale of impact reconciling bottom up participation with the production of widely applicable research outputs by Dr. Fergus Sinclair, Systems Science Leader, ICRAF
Similar to Systems Science at the scale of impact reconciling bottom up participation with the production of widely applicable research outputs by Dr. Fergus Sinclair, Systems Science Leader, ICRAF (20)
Systems Science at the scale of impact reconciling bottom up participation with the production of widely applicable research outputs by Dr. Fergus Sinclair, Systems Science Leader, ICRAF
1. Systems science at the scale of
impact
Fergus Sinclair
World Agroforestry Centre
Bangor University, Wales, UK
CATIE, Costa Rica
2. Fertiliser
Crop
varieties
Agro-
chemicals
Higher yield
The cropping ‘system’ concept
Agronomy
Assumes that the crop is fairly independent of the
rest of the livelihood system – which (with due
regard to sustainability) is appropriate for a subset
of wealthier farmers in Africa
Package of measures to increase crop productivity by using inputs to make the
environment as suitable as possible for the crop – need to watch sustainability
esp. with respect to soil carbon and micronutrients
3. Crops within a livelihood system reality
Fertiliser
Crop
varieties
Agro-
chemic
als
Higher yield
Agrono
my
Thinnings and
residues
Manure
Fodder
at key
times
Livestock
Non agricultural
activity e.g. trade
Trees
Crops
Household
7. Systems research
• Farmers don’t follow agronomic
recommendations for maize
– plant at high population density, use thinings for
fodder end up with low densities
– intercrop – including with tree cover (globally
almost half agricultural land has >10% tree cover)
– apply fertiliser purposively (precision farming?)
– 30% increase in maize yield through participatory
varietal selection in Nepal (Tiwari et al., 2009)
Fertiliser
A
g
r
o
-
c
h
e
m
i
c
a
l
s
Higher y
A
g
r
o
n
o
m
y
Thinni
ngs
and
residu
es
Fo
dd
er
at
ke
y
ti
m
es
Livestock
Non
Trees
8. used by individual households collectively used
BARI
maize/millet intercropping
with fodder trees on crop
terrace risers
GRASSLAND F
O
R
E
S
T
KHET
paddy rice
LIVESTOCK
grazing
tree fodder
tree fodder/
crop residues
manure
crop residues
9. Systems interaction references
• Tiwari, T.P., Brook, R.M. and Sinclair, F.L. (2004) Implications
of hill farmers' agronomic practices in Nepal for crop
improvement in maize. Experimental Agriculture 40: 1-21
• Tiwari, T.P, Virk, D.S. and Sinclair, F.L. (2009). Rapid gains
in yield and adoption of new maize varieties for complex
hillside environments through farmer participation. I.
Improving options through participatory varietal selection
(PVS). Field Crops Research 111: 137–143
• Tiwari, T.P., Brook, R.M., Wagstaff, P. and Sinclair, F.L. (2012)
Effects of light environment on maize in hillside
agroforestry systems of Nepal. Food Security 4: 103-114.
10. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
Pruned trees
Free growing trees
Earthworm cast weight
Sample with no
earthworm casts
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 5 10 15
Separation distance (m)
Semivariance
Cross-semivariogram
Greater soil biological activity (earthworms) near trees but effect
greater for some tree species than others
Pauli et al 2010 Pedobiologia
11. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
12. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
13. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
14. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
15. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
16. The challenge
• Fine grained variation in:
– soil (biota)
– climate (altitude)
– farming practices
– household characteristics
– market opportunities
– social capital
– policy and its implementation
17. What to scale up?
• Participation (PAR) largely replaced systems
methods (farmer or community integrates)
• Options refined through PAR at a few sites
don’t scale because context varies, BUT
• scaling only innovation processes (rather than
options to improve livelihood systems) is not
cost effective. Options are:
Technology
(components and
their management)
Effective delivery
mechanisms / markets
Appropriate enabling
policy and institutional
environment+ +Ingredients that can be combined in different ways across scales
18. Characterize variation in
context across scaling domain
Influence development
projects so that sufficient
intensification options are
offered to farmers across
sufficient range of variation
in drivers of adoption
Initial matrix of
intensification and
resilience options and
the contexts in which
they work (soils, climate,
farming system, planting
niche, resource
availability, institutions)
Participatory monitoring and
evaluation system for the
performance of options
Scaling up
Simple to use tools to
match options to sites
and circumstances across
the scaling domain
Generate understanding of
suitability of options in
relation to context – and the
cost effectiveness of
different combinations
refined
characterization
refined
options
Scaling out
Application of
understanding about cost
effective options for
different contexts beyond
the current scaling domain
Global comparative
understanding of how to
improve livelihood systems,
emergent from the place-
based research complex.
Coe, R., Sinclair, F. and Barrios,
E.(2014). Scaling up agroforestry
requires research ‘in’ rather than ‘for’
development. Current Opinion in
Environmental Sustainability, 6: 73–77.
20. Genotype x Environment (GxE)
interaction
Drought stress
Crop yield
A
B
C
Used by breeders
21. GxE → OxC
Tree species
Management
package
Training approach
Organisational
model
Genotype
Climate
Soil
Farm resource
endowment
Market integration
Gender, HH type
Ethnic group
Environment Production
Risk
Profitability
Acceptability
Env impact
X =
X
Option Context PerformanceX =
=
Anystyleofparticipation
Quantitative-Qualitative-Mixed
22. Local effects - trees increase crop yields from
meta analysis of >90 trials across sub Saharan Africa
• Mean yield of maize after coppiced
and non-coppiced tree fallows (various
species) is > 1 t ha-1 doubling default
practice of many farmers in many
years (no nutrient inputs).
• Very large standard error around the
mean – indicates performance varies
with circumstances – we need to know
where particular trees will increase
yields by a large enough amount to
merit farmer input in the technology
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
N
aturalfallow
H
G
M
Ls
N
on-coppicing
C
oppicingFullfertilizer
Yielddifference(tha-1
)
Yield difference = Treatment-control yield
Control = maize without nutrient input
HGMLs = herbaceous green manure legumes
Sileshi G, Akinnifesi FK, Ajayi OC and Place F (2008) Meta-
analysis of maize yield response to planted fallow and
green manure legumes in sub-Saharan Africa. Plant and
Soil 307: 1-19.
23. 0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
-1.0 0.0 1.0 2.0 3.0 4.0 5.0
Increase in maize yield over control after sesbania fallow (t ha-1)
Cumulative
probability
From: Sileshi et al., 2010. Field Crops Research. Based on meta analyses of over 90 trials across sub-saharan Africa
50% probability of no increase in yield or
worse on Nitosols (saturated fertility?)
Risk – what is the probability that a farmer will get a threshold increase in yield on different soils?
60% probability of > 1 t ha-1 increase in
yield on Luvisols
25. Images and analyses courtesy of Thomas Gumbricht
Matching technologies to fine scale variation in food resilience
RATA – Resilience indicator
26. Roots of recoveryA tale of two villages – Africa Rising
Ministries go for cocoa options in Peru Rediscovering our trees - DRC
27. I am trying things out here first and if
they work well, I will expand to other
areas of my farm over there
Learning from experience - teasing out
ingredients of success and the contexts they
are appropriate for:
scaling domains for ingredients and their
combination
28. Key steps
• Optimise the system not one component
– total factor productivity
• Vertical and horizontal integration
– across scales
– food, water, energy
• Options by context not silver bullets
– systematic large N trials of range of options over range of context
– nested scale options (T+M/E+P/I) and planned comparisons
– measure performance with appropriate indicators including of
resilience
– refine recommendations through action of feedback loops to create
easy to use tools (co-learning) at right scales and resolution
– partner with development organisations and the private sector
– bridge knowledge systems (local, global science, policy makers)
29. It’s more than…
• … classic ‘participatory research’
– aiming for scale
– integrating testing of options, delivery, institutional
arrangements
• …project M and E
– planned to generate needed information efficiently
– contribute to global K-base, not just project
requirements.
• … action research
– learning principles, contributing to global K-base
– not just optimising locally