Network analysis to analyse complex agroecosystems

Network analysis to analyse
complex agro­ecosystems
Mariana Rufino, Pytrik Reidsma, Santiago López­Ridaura, Pablo Tittonell, Ken Giller
Plant Production System Group

The context
� Rural­poor farm households
� Food security (primary goal)
� Risk (income and/or consumption)
� (Scarce) resources

Population (People per Km 2 )
0 ­ 2
3 ­ 10
11 ­ 20
21 ­ 50
51 ­ 100
101 ­ 200
201 ­ 500
501 ­ 1000
>1000
Tigray ­
Ethiopia
Kakamega ­
Kenya
Murewa ­
Zimbabwe
Source FAO

The Highlands of Tigray ­ Northern
Ethiopia

Communal farming ­ Murewa NE Zimbabwe

The Highlands of Western Kenya

The objectives
� To test:
� Whether ecological network analysis was applicable to agro­
ecosystems
� Whether their application provide insights into current
performance and development across environments
� To reflect
� Whether the results of the applications were informative
enough to contribute to discussion on sustainable
livelihoods

The system definition – the network
� The unit of analysis ­> the farm household
� Network compartments ­> the farming activities
� Difference with ecosystems ­> human agency
� Resource flows ­> purposively managed
� Currency ­> Nitrogen
� Timescale ­> year

Using existing information
� Farm typology + household data
� Resource flow mapping
� Observations and farmers interviews
� Field measurements
� Existing information from:
� Tigray – Northern Ethiopia (n=50)
� Kakamega – Western Kenya (n=15)
� Murewa – NE Zimbabwe (n=50)

Conceptualising the network
For each
farm
type
Pasture
Maize
Fodder crops
Fertiliser + seeds
Manure
storage
Maize
Compost
Feed
Waste
Excreta
Maize
Household
Grain
Food
Vegetables
Animal products
Chicken
Feed
Cattle
Ground
nuts
Food crops
Excreta
Products
Goats
Food
Livestock
Excreta
Feed

The network analysis
� Ecological network analysis
� Leontief (1951) input­output analysis in economics
� Hannon (1973) application in Ecology
� J.T. Finn (1976, 1980)
� R.E. Ulanowicz (1980, 1986, 1997,…)
� B.C. Patten, B.D. Fath, S.E. Jørgensen and others

The indices of the network analysis
� System size (activity)
� System structure and organisation
� Efficiency measures
� A measure of growth and development
� Ascendency

Agro­ecosystems as networks
Roles (#)
12
10
8
6
4
2
0
0 2 4 6 8 10 12 14
Connectivity (flows node ­1 )
A window of vitality
Random networks
Natural ecosystems
Agroecosystems
Effective # of nodes
14
12
10
8
6
4
2
0
Natural ecosystems
Agroecosystems
0 5 10 15 20 25 30 35 40 45
Effective # of flows
After Zorach and Ulanowicz (2003) Complexity 8, 68­76

Size and organisation
Measure of growth & development (A)
2000
1500
1000
500
0
Kakamega
Murewa
Tigray
0 500 1000 1500
2000
1500
1000
500
0
0.5 1 1.5 2
Measure of size (T) Measure of organisation (AMI)
Ascendency = T x AMI

Size and organisation
Measure of growth & development (A)
2000
1500
1000
500
0
Kakamega
Murewa
Tigray
livestock
0 500 1000 1500
2000
1500
1000
500
0
0.5 1 1.5 2
Measure of size (T) Measure of organisation (AMI)
Ascendency = T x AMI

Capacity
Inputs (Φ I )
Exports (ΦE)
Dissipations (Φs)
Pathway redundancy (Φo)
Ascendency (A)
5000
4000
3000
2000
1000
0
farm types ­­>
P M W P M M W P M W
Tigray­
Ethiopia
Ascendency = size x organisation
Overhead
Murewa­
Zimbabwe
Currency: kg N fhh ­1 y ­1 bits
Kakamega­
Kenya

Relationship to food self­sufficiency
Food selfsufficiency
4.0
3.0
2.0
1.0
0.0
other
Ethiopia
0 500 1000 1500 2000
4.0
3.0
2.0
1.0
0.0
0 10 20 30 40 50 60 70
Ascendency Network organisation
Food self­sufficiency = Food produced/food needed

Relationship with productivity
Biomass (t ha ­1 )
12.0
8.0
4.0
0.0
0 400 800 1200 1600
Biomass (t ha ­1 )
12.0
Ascendency (kg N fhh ­1 y ­1 bits) Network organisation
8.0
4.0
0.0
0 10 20 30 40 50 60 70

Relationship with productivity
Biomass (t ha ­1 )
12.0
8.0
4.0
0.0
0 400 800 1200 1600
Biomass (t ha ­1 )
12.0
8.0
4.0
others
Kakamega­Kenya
0.0
0 10 20 30 40 50 60 70
Ascendency (kg N fhh ­1 y ­1 bits) Network organisation

N cycling (% of total)
Elements of stability and resilience
12.0
9.0
6.0
3.0
0.0
Nutrient cycling Dependency
0 300 600 900 1200 1500
1.0
0.8
Fraction of N
imported
0.5
0.3
0.0
Ascendency (size x organisation)
0 300 600 900 1200 1500

Elements of stability and resilience
N cycling (% of total)
12.0
8.0
4.0
0.0
Nutrient cycling Dependency
0 10 20 30 40 50 60 70
Fraction of N
imported
1.0
0.8
0.6
0.4
0.2
0.0
Network organisation
0 10 20 30 40 50 60 70

Ascendency / overhead
5000
4000
3000
2000
1000
0
farm types ­>
P M W P M M W P M W
5000
Current systems 4000
Without livestock
3000
2000
1000
0
P M W P M M W P M W
Tigray Murewa Kakamega Tigray Murewa Kakamega

Ascendency / overhead
5000
4000
3000
2000
1000
0
30%
10%
5000
Current systems 4000
Without livestock
44%
3000
2000
~10%
1000
P M W P M M W P M W
0
Self­
insurance
P M W P M M W P M W
Tigray Murewa Kakamega Tigray Murewa Kakamega

From the analysis:
� The household and the livestock are the most
important sinks for N (key nodes)
� Livestock represent a ‘non­productive cost’
…but also self­insurance
…and a driver for nutrient cycling
� Agro­ecosystems are refined networks

Conclusions
� NA is useful to analyse agro­ecosystems properties:
e.g. allowed to identify elements of productivity, stability and
resilience
� Results can contribute to discussions on livelihoods
because NA includes:
� The rural family and their primary goals
� The natural resources (constraints and opportunities)
� And shows the effects of human agency (adaptive management)

Thanks
and the contribution of:
Huib Hengsdijk, Jan Verhagen, Shamie Zingore and Luke Latham
© Wageningen UR