1. The ChitwanABM: Modeling Population- Environment Interactions and their Implications in the Chitwan Valley, Nepal Alex Zvoleff *1, Li An * 2011 AAAS Annual Meeting, Washington DC San Diego State University; San Diego, USA. 1 azvoleff@mail.sdsu.edu

2. Outline 1.Introduction 2.Study Site 3.Methods: The ChitwanABM 4.Scenario Comparison 5.Conclusions February 18, 20112AAAS Annual Meeting

3. Introduction Land-use and land-cover and human-decision making are intimately linked Micro-level decision-making can lead to broader scale (“emergent”) patterns on the landscape The link between these micro-scale decisions and the macro-scale landscape can be crucial for management February 18, 20113AAAS Annual Meeting

4. Research Question How does micro-level demographic decision making impact macro-level land use and land cover (LULC)? – Objective: Focus on feedbacks between marriage timing and LULC February 18, 20114AAAS Annual Meeting

5. Study Site: Western Chitwan Valley, Nepal February 18, 20115AAAS Annual Meeting

6. Study Site: Western Chitwan Valley, Nepal February 18, 20116AAAS Annual Meeting

7. Study Site: Western Chitwan Valley, Nepal February 18, 20117AAAS Annual Meeting

8. Chitwan Valley Family Study Longitudinal survey begun in 1996 Focusing on social context and family formation Human survey data – Three detailed interviews (1996, 2001, 2008) – Household registry (monthly since Feb. 1997) Environmental data – Flora count (1996, 2000, 2007) – Neighborhood mapping (1997, 2000, 2007) February 18, 20118AAAS Annual Meeting

9. Chitwan Valley Family Study: Household Registry February 18, 20119AAAS Annual Meeting

10. Study Site: Overview Population ≈ 250,000 National park: 932 Km 2 Buffer zone: 766 Km 2 Forest resources important – 93% use fuelwood – 76% gather fuelwood Ag. is dominant land-use – 80% of study area in 1996 February 18, 201110AAAS Annual Meeting

11. AGENT-BASED MODEL February 18, 201111AAAS Annual Meeting

12. Why an agent-based model? Agent-based models (ABM) represent individual “agents” and model their interactions ABM allows: – Representation of human-decision making – Consideration of feedbacks – Examination of system dynamics – Testing of alternative hypotheses February 18, 201112AAAS Annual Meeting

13. ChitwanABM Agent Hierarchy 151 neighborhood agents 1551 household agents 8414 individual agents February 18, 201113AAAS Annual Meeting

14. Parameterization Baseline probabilities calculated for: – Fertility First birth timing Desired number of children – Marriage – Migration – Mortality Feedbacks alter the baseline probabilities LULC determined by decisions of household agents February 18, 201114AAAS Annual Meeting

15. Parameterization: Mortality February 18, 201115AAAS Annual Meeting

16. Land-use/land-cover Classes Land-use/land-cover classes – Agricultural vegetation – Non-agricultural vegetation – Private buildings – Public infrastructure – Other (ponds, silted riverbanks, etc.) February 18, 201116AAAS Annual Meeting

17. Land-use/land-cover Change – 1996-2006 Class199620012006 Agricultural Vegetation 879.9 (80.0%)875.6 (79.4%)854.2 (77.6%) Non- agricultural Vegetation 50.2 (4.6%)35.3 (3.2%)54.4 (4.9%) Private Buildings 82.3 (7.5%)88.4 (8.0%)94.4 (8.6%) Public Buildings 59.2 (5.4%)64.3 (5.8%)66.9 (6.1%) Other28.4 (2.6%)39.5 (3.6%)31.2 (2.8%) Total1100.1 (100%)1103.1 (100%)1101 (100%) February 18, 201117AAAS Annual Meeting

18. Objective: Explore Two Scenarios Scenario 1: Baseline scenario Scenario 2: LULC-marriage timing feedback scenario February 18, 201118AAAS Annual Meeting

19. Scenario 1 February 18, 201119AAAS Annual Meeting

20. Scenario 2 Prior work (Yabiku 2007) has shown marriage timing – Highly correlated with age, sex, ethnicity – Positively related to agricultural land Higher percentage agricultural land leads to earlier marriage (get married younger) We explore two scenarios – Baseline scenario (no feedback) – Positive feedback of percent agricultural land on marriage February 18, 201120AAAS Annual Meeting

21. Scenario 2 CovariateRelationship Log % Agricultural Vegetation+ Sex (female)+ Age– Age (squared)+ Ethnicity Hill Tibeto-Burmese+ Lower-caste Hindu+ Newar– Terai Tibeto-Burmese+ Other+ February 18, 201121AAAS Annual Meeting

22. SCENARIO COMPARISON February 18, 201122AAAS Annual Meeting

23. Time Zero: Percent Vegetation February 18, 201123AAAS Annual Meeting

24. Scenario 1 (baseline): LULC February 18, 201124AAAS Annual Meeting

25. Scenario 1 (baseline) minus Scenario 2 (marriage) February 18, 201125AAAS Annual Meeting

26. DISCUSSION February 18, 201126AAAS Annual Meeting

27. Conclusions Feedback of LULC on marriage timing: – Shows little impact at the aggregate level – Shows large impact spatially Considering feedback – Agricultural areas grow quicker – Urbanized areas grow more slowly →Must consider spatial context of decision- making February 18, 201127AAAS Annual Meeting

28. Future Work 1.Incorporate additional feedbacks – LULC – first-birth timing – LULC – family size preferences – LULC – migration 2.Incorporate feedbacks between biomass harvesting ↔ ecosystem change February 18, 201128AAAS Annual Meeting

29. Acknowledgements With thanks to: Dr. William Axinn, UM Dr. Jianguo Liu, MSU Dr. Lisa Pearce, UNC-Chapel Hill Dr. Scott Yabiku, ASU Dr. Dirgha Ghimire, UM Dr. David López-Carr, UCSB Binoj Shrestha, ISER-N Krishna Shrestha, ISER-N With support from: February 18, 201129AAAS Annual Meeting

30. Thank you. Questions? Email: azvoleff@mail.sdsu.edu ChitwanABM is free and open-source: http://rohan.sdsu.edu/~zvoleff/ChitwanABM.php PyABM – an open-source ABM toolkit for Python (coming spring 2011): http://rohan.sdsu.edu/~zvoleff/PyABM.php February 18, 201130AAAS Annual Meeting