1. Photo by Erwin Palacios (CI Colombia)© The Economist 1

2. Photo by Erwin Palacios (CI Colombia) To explicitly quantify the linkages between the natural ecosystem services that affect – and are affected by – food security and nutritional health for the rural poor at the forest-agricultural interface

3. Links between ecosystem services, benefits, food security and well-being

4. United Kingdom:  University of Southampton (PI Poppy)  University of Dundee (Dawson) United States of America:  Conservation International (Co-PI Honzak) Spain:  Basque Centre for Climate Change (Co-PI Villa) Colombia:  International Centre for Tropical Agriculture – CIAT (Co-PI Jarvis)  Research centres, universities and NGOs Malawi:  Chancellor College (Co-PI Chiotha)  LEAD Southern and Eastern Africa plus Ministry of Forestry, Forest Research Institute of Malawi  WorldFish Centre - Malawi  Rhodes University South Africa

5. High forest cover Low deforestation High forest cover High deforestation Low forest cover Low deforestation The study areas on the forest transition curve Colombia Peru Malawi Natural land use transitionForest replenishment period

6. 3 key interactions: Water, Biomass & People P EOPLE B IOMASS W ATER Monthly time-steps; globally available inputs

7. LPJ-Guess (Monthly) ProcessesOutputs Photosynthe sis Respiration Stomatal regulation Growth* Population dynamics* Disturbance* Land cover/use Vegetation biomass * Currently annual Water Balance Model (Monthly) ProcessesOutputs Surface water flow Soil water flow Surface water Soil water Globally Available GIS Inputs Temperature Precipitation Solar radiation Atmospheric CO 2 Soil type Cropland forcing Digital elevation model Ricker-type Agent Based Model (Monthly) ProcessesOutputs Pests Pest control Pollinators Game Culture Overall wildlife impact on agriculture Game availability Cultural ES Village Agent Based Model (Monthly) ProcessesOutputs Farming Irrigating Collecting water (rivers) Collecting water (boreholes) Harvesting (Crops) Harvesting (Wild veg.) Trading Casual Working Formal Working Fishing Hunting Sharing Borrowing Disturbance (agriculture) Disturbance (wild veg) Irrigation applied HH finances HH food security Game used Water used Household Agent Based Model (Monthly) ProcessesOutputs Food allocation Dietary preferences Individual health & energy Regional Agent Based Model (Monthly) SPATIAL ProcessesOutputs Trade Land use policy Land available for agriculture Market-based food availability & cost KEY Top priority linkage (given first preference) Low priority linkage (given what remains) Possible (with more complex animal agents) Include human- induced fire? Should these graze and grow? Urban agents? Pastoralist agents?

8. Water Balance Models Runoff – SCS curve number method Soil water – Thornthwaite-Mather soil water balance model Flow velocity (tbc) InputsOutputs Land coverRunoff Soil type InputsOutputs Land coverSoil water Soil typeGround water recharge

9. LPJ-Guess (Monthly) ProcessesOutputs Photosynthe sis Respiration Stomatal regulation Growth* Population dynamics* Disturbance* Land cover/use Vegetation biomass * Currently annual Water Balance Model (Monthly) ProcessesOutputs Surface water flow Soil water flow Surface water Soil water Globally Available GIS Inputs Temperature Precipitation Solar radiation Atmospheric CO 2 Soil type Cropland forcing Digital elevation model Ricker-type Agent Based Model (Monthly) ProcessesOutputs Pests Pest control Pollinators Game Culture Overall wildlife impact on agriculture Game availability Cultural ES Village Agent Based Model (Monthly) ProcessesOutputs Farming Irrigating Collecting water (rivers) Collecting water (boreholes) Harvesting (Crops) Harvesting (Wild veg.) Trading Casual Working Formal Working Fishing Hunting Sharing Borrowing Disturbance (agriculture) Disturbance (wild veg) Irrigation applied HH finances HH food security Game used Water used Household Agent Based Model (Monthly) ProcessesOutputs Food allocation Dietary preferences Individual health & energy Regional Agent Based Model (Monthly) SPATIAL ProcessesOutputs Trade Land use policy Land available for agriculture Market-based food availability & cost KEY Top priority linkage (given first preference) Low priority linkage (given what remains) Possible (with more complex animal agents) Include human- induced fire? Should these graze and grow? Urban agents? Pastoralist agents?

10. Biomass Model – LPJ-Guess Conceptual representation of LPJ-Guess Modes of LPJ-Guess Process-based carbon assimilation Smith (2001)

11. Globally Available GIS Inputs Temperature Precipitation Solar radiation Atmospheric CO 2 Soil type Cropland forcing Digital elevation model LPJ-Guess (Monthly) ProcessesOutputs Photosynthe sis Respiration Stomatal regulation Growth* Population dynamics* Disturbance* Land cover/use Vegetation biomass * Currently annual Water Balance Model (Monthly) ProcessesOutputs Surface water flow Soil water flow Surface water Soil water Ricker-type Agent Based Model (Monthly) ProcessesOutputs Pests Pest control Pollinators Game Culture Overall wildlife impact on agriculture Game availability Cultural ES Village Agent Based Model (Monthly) ProcessesOutputs Farming Irrigating Collecting water (rivers) Collecting water (boreholes) Harvesting (Crops) Harvesting (Wild veg.) Trading Casual Working Formal Working Fishing Hunting Sharing Borrowing Disturbance (agriculture) Disturbance (wild veg) Irrigation applied HH finances HH food security Game used Water used Household Agent Based Model (Monthly) ProcessesOutputs Food allocation Dietary preferences Individual health & energy Regional Agent Based Model (Monthly) SPATIAL ProcessesOutputs Trade Land use policy Land available for agriculture Market-based food availability & cost KEY Top priority linkage (given first preference) Low priority linkage (given what remains) Possible (with more complex animal agents) Include human- induced fire? Should these graze and grow? Urban agents? Pastoralist agents?

13. LPJ-Guess (Monthly) ProcessesOutputs Photosynthe sis Respiration Stomatal regulation Growth* Population dynamics* Disturbance* Land cover/use Vegetation biomass * Currently annual Water Balance Model (Monthly) ProcessesOutputs Surface water flow Soil water flow Surface water Soil water Globally Available GIS Inputs Temperature Precipitation Solar radiation Atmospheric CO 2 Soil type Cropland forcing Digital elevation model Ricker-type Agent Based Model (Monthly) ProcessesOutputs Pests Pest control Pollinators Game Culture Overall wildlife impact on agriculture Game availability Cultural ES Village Agent Based Model (Monthly) ProcessesOutputs Farming Irrigating Collecting water (rivers) Collecting water (boreholes) Harvesting (Crops) Harvesting (Wild veg.) Trading Casual Working Formal Working Fishing Hunting Sharing Borrowing Disturbance (agriculture) Disturbance (wild veg) Irrigation applied HH finances HH food security Game used Water used Household Agent Based Model (Monthly) ProcessesOutputs Food allocation Dietary preferences Individual health & energy Regional Agent Based Model (Monthly) SPATIAL ProcessesOutputs Trade Land use policy Land available for agriculture Market-based food availability & cost KEY Top priority linkage (given first preference) Low priority linkage (given what remains) Possible (with more complex animal agents) Include human- induced fire? Should these graze and grow? Urban agents? Pastoralist agents?

14. 1.PRA used to build conceptual framework

17. 17

18. 1.PRA used to build conceptual framework 2.National census can be used to build agent typologies

19. i) Selection of Variables  Variables associated with household land, labour, livestock assets and gender were extracted from National Census Data  Data was split into a training set (80%) and validation set (20%) ii) Principal Components Analysis (PCA)  To reduce the dimensionality of data iii) Hierarchical Clustering of PC  Objects are grouped into clusters based on a distance (Euclidean) and an agglomeration criterion (Ward) to produce a hierarchical tree.  K-means Clustering is then used to partition the tree, defining the clusters iv) Validation  Testing the stability of clusters by a) one-way ANOVA; b) changing the order of variables; c) using the validation set v) Results  Three distinct clusters were recognised and related to household livelihood strategies: 1) on-farm workers 2) Off-farm agricultural workers and 3) Off-farm non-agricultural workers. This can be carried out using the FactoMineR package in R http://factominer.free.fr/index.html

20. 1.PRA used to build conceptual framework 2.National census can be used to build agent typologies (see next slide) 3.HH survey to be used build decision rules based on Bayesian networks

21. Measuring poverty and food security  Two 6-month waves per study site.  Samples in Malawi and Peru, census in Colombia.  Standard demographic, expenditure and income estimates BUT… o Includes use of natural resources for basic wellbeing dimensions: health, housing, food, agricultural and business inputs, etc. o Subjective assessments of well-being o Anthropometric measures for children < 5.

22. 22 This presentation was produced by ASSETS (NE-J002267-1), funded with support from the Ecosystem Services for Poverty Alleviation Programme (ESPA). The ESPA programme is funded by the Department for International Development (DFID), the Economic and Social Research Council (ESRC) and the Natural Environment Research Council (NERC), as part of the UK’s Living with Environmental Change Programme (LWEC). The views expressed here are those of the authors and do not necessarily represent those of the funders, the ESPA Programme, the ESPA Directorate, or LWEC.

23. ARIES: summary  A rapid spatial assessment tool for ecosystem services and their values; not a single model but an artificial intelligence assisted system that customizes models to user goals.  Demonstrates a mapping process for ecosystem service provision, use, sink and flow while most ES assessments only look at provision.  Probabilistic, Bayesian models inform decision-makers about the likelihood of possible scenarios; users can explore effects of policy changes and external events on estimates of uncertainty.

24. Natural sinks Removal by people Forest ES Crop ES Water ES Game ES

25. Uses Sources Sinks

28. Controlling frequency and size of meals Social support networks Collection of wild vegetables (limited) Selling assets Begging and prostitution Fishing Hunting Collection of forest products Social support networks Irregular rains Increasing population Deforestation (reserves) Increasing population (settlers) Deforestation Over exploitation (commerce) Small (rain-fed) farmland Most have minor livestock Regular hungry seasons Casual labour (illegal or risk-prone) Most cover basic needs Most families have farmland Hunting, fishing are common Limited financial and physical capitals Very little forest cover Decreasing water availability Limited land availability Diminishing soil fertility Land still available but more limited No changes in forest cover Decreasing variety of fish Diminishing variety of wildlife

29. Working towards influencing policy-making  The project is working in collaboration with key governmental institutions, such as the Ministry of Forestry and the Ministry for Local Government and Rural Development in Malawi.  The project has set up National Steering Committees in Colombia and Malawi with high-profile scientific researchers.  Local partners have access to relevant policy-making instances and public- agenda setters: Malawi TV documentary discussed the potential contribution of the ASSETS project to current environmental issues affecting Malawi. Caqueta’s Regional Government Development Plan (2012-2015) states that the project’s “scientific contributions are expected to serve as input for the formulation of better policies for intervention, prioritization of actions and management of regional and local authorities”.