Michael Barton School of Human Evolution & Social Change Center for Social Dynamics & Complexity Social Systems are Complex    Even simple social systems are complex Complexity has grown at increasing rate since beginnings of urbanism Measures of complexity  5,400 mammal species (350 primates)  28,800 occupational titles in USA (820 occupational groups) Socioecological Systems are More Complex   Social complexity amplified by interactions with the natural world Agriculture (last 10ka)  Human manipulation of ecosystems. Now on massive scale.  Society recursively impacted by consequences of environmental manipulation  Result  Coupled socioecological systems  Dynamics driven by combination of social practice and biophysical process  Dominate terrestrial landscapes today Socioecological Systems are More Complex  Coupled human-natural systems  Non-linear relationships between cause and effect in environmental change (e.g., threshhold and scale-dependent effects)  Emergent (often unexpected) ecological effects of human actions  Ecological consequences may take centuries to play out  Different social and biophysical time scales  Amplifying vs. dampening effects over time and space Socioecological Systems are More Complex Environmental “problems” are social problems (van der Leeuw)  Imperative to understand dynamics of socioecological systems   Current human-environmental interactions and landscapes  Informed environmental (and social) policy  Forecasting environmental change and risk Modeling Socieoecological Systems Coupled social-natural systems pose special challenges for modeling  Different components represented by different types of models, operating in different ways, with different timing  Outcomes often long-term. Make evaluating models difficult with modern data.  Modeling Socieoecological Systems  Human land use  Discrete entities, following decision rules, organized in nested hierarchies (individual, household, community, city, etc).  Agent-based models  Surface runoff  Landscape cells whose state is influenced by state of adjacent cells  Cellular automata  Stream flow  Competence varies continuously with velocity  Differential equations Modeling Socieoecological Systems  New cybertools  Permit explicit and quantitative representation of components of social and natural systems  Including dynamics across space and time  Important examples include  GIS and geospatial models  Agent based models  Interactive visualization of multivariate data  Often must combine to model coupled socioecological systems Modeling Socieoecological Systems  The past is an effective model testbed, especially for long-term consequences  Apply models to archaeological/historical record  Rigorous evaluation for models of socioecological dynamics  Record of sparse, discontinuous points, scattered in time and space  Models that can ‘pass through’ known points of past systems…  More robust  More confidently applicable in other times and places Mediterranean Landscape Dynamics  Example of research protocols for socioecological modeling  Large-scale landscape consequences of small-scale landuse decisions, and their feedbacks, over long time frames.  Hybrid modeling laboratory that integrates diverse model classes (ABM, CA, PDE, etc)  Uses archaeological data for model parameterization and evaluation. Characterize past system to create more robust general landuse/landscape models  Interdisciplinary collaboration  Archaeology, geosciences, life sciences, climatology, computer science, geospatial methods and statistics  NSF Biocomplexity program (BCS-0410269) Mediterranean Landscape Dynamics  Spans Mediterranean socioecosystems  Arid East, moister West  Range of social configurations  Earliest evidence for agriculture MedLand Modeling Laboratory  3 interlinked modeling environments  Potential landscape model  Archaeological/pale oecological record  Coupled agropastoral socioecology model MedLand Modeling Laboratory  Dynamically coupled models  Landuse ABM: ○ Agents (individuals, households, villages) ○ Behaviors based on decision rules and environmental information  Landscape cellular automata ○ Surface processes of erosion/deposition ○ Dynamics derived from differential equations MedLand Modeling Laboratory  Models for parameterization  Multiple regression ○ Climate: geospatial model  Values for each climate station  Estimated values between climate stations ○ Vegetation: geospatial model  Estimated values based on topography and climate  Currently simple algebraic derivation ○ Soils: geospatial model Model Components: Landuse  ABM in DEVSJAVA  Household is basic unit (agent)  Organized into villages  Landuse decisions  Potential productivity  Distance from village  Labor investment needed (e.g., clear land or simply cultivate)  Landuse activities      Clearing land Cultivating crops/herding Fallowing Harvesting crops/animal products Returns Model Components: Landscape  USPED (as implemented in GRASS GIS)  ED = d(Ep ⋅ qsx)/dx + d(Ep ⋅ qsy)/dy  ED is net potential erosion or deposition of sediment in any landscape cell  qsx and qsy are the sediment transport capacity coefficients in x and y directions (a function of slope, aspect, and flow accumulation) for a given surface process across the cell  Ep (potential erodability) is modified a RUSLE value that includes for each cell… ○ rainfall intensity ○ land cover Land use Rainfall Land cover Soil USPED ○ soil characteristics DEM ED map Model Components: Topography    Paleolandscape reconstruction and verification High-resolution DEM created from satellite imagery and stereo aerial photos Landform and geomorphic mapping with ground truthing in field residential flat to rolling surfaces active landslide active erosion hollow active drainage steep slopes terraces Model Components: Vegetation Potential landcover  Community models based on   climate  topography  soils  vegetation transects  Successional dynamics Model Components: Climate   Paleoclimate at 100 yr intervals for 40 ka Macrophysical climate models Retrodicted from modern weather station  Temperature and precipitation  Annual and monthly values Annual Precipitation 8000-2 Wadi Ziqlab Area Weather S 2500 2000 mm of precipitation  1500 1000 500 0 -9800 -9400 -9000 -8600 -8200 -7800 -7400 -7000 -66 -10000 -9600 -9200 -8800 -8400 -8000 -7600 -7200 -6800 Model Components: Climate Annual Precipitation 6000 BP 2500 2000 mm of precipitation Calculate multiple regression models for each century using topography/geograp hy  Apply regression coefficients to DEMs  Create paleoclimate “landscapes”  7000 BP 1500 1000 500 0 -10000 -9200 -8400 -7600 -6800 -6000 -5200 -4400 -9600 -8800 -8000 -7200 -6400 -5600 -4800 -4000 years BP 8000 BP Model Components: Soils     Initialize as constant thickness Calculate K-Factor from sand:silt:clay ratios or bedrock geology Dynamically model changing soil thickness and erodability Remantle paleosurfaces with Holocene soils. Coupled ABM & Landscape Model Vegetation modeling Agents (ABM) Climate model Terrain modeling Socioecosystem dynamics Experiments in Socioecological Dynamics: Spain early Holocenel andscape modern landscape Neolithic farming in the Penaguila Valley, central Mediterranean Spain Experiments in Socioecological Dynamics: Spain ABM land cover/landuse GIS landscape (erosion/deposition) 200 year coupled landuse-landscape simulation Experiments in Socioecological Dynamics: Jordan Tabaqat al-Buma PN hamlet intensive cultivation shifting cultivation Tell Rakkan PPNB village intensive cultivation shifting cultivation grazing catchments Experiments in Socioecological Dynamics: Jordan Modeled Wadi Ziqlab. 200 yrs shifting cultivation & grazing Wadi Ziqlab today Tell Rakkan: channel incision (shifting cultivation, grazing, low rainfall) Experiments in Socioecological Dynamics: Jordan Tell Rakkan: PPNB, high rainfall, vegetation change, 40 yrs vs 200 yrs (swidden agriculture, grazing) A Science of Social Dynamics Socioecological modeling a critical component of new science of social dynamics  Theory and concepts   Reconceptualize societies and environment as closely coupled complex system rather than distinct domains  Focus on dynamics of human-environmental interactions rather than their static relationships and results  Research protocols  Requires investment in “computational thinking” to represent human-environmental interaction and change as explicit, algorithmic models  Needs emerging cybertools to leverage traditional research protocols of observational data collection and confirmatory statistics  Benefits in new dimensions in social/natural science and added value for existing approaches and data Interdisciplinary Collaboration   ASU: School of Human Evolution and Social Change, Center for Social Dynamics & Complexity, School of Earth and Space Exploration, School of Computing and Informatics, Geographical Sciences, School of Sustainability Partners: Universitat de València, Universidad de Murcia, University of Jordan, North Carolina State University, University of Wisconsin, Hendrix College, Geoarchaeological Research Associates, GRASS GIS Development Team Open Agent Based Modeling Consortium  Building a community of researchers in social and ecological sciences, to:  Improve access to computational tools for complex systems modeling  Share experiences and strategies  Promote a science of social dynamics  http://www.openabm.org