No abstract

Ecosystem Services (ES) are an established conceptual framework for attributing value to the benefits that nature provides to humans. As the promise of robust ES-driven management is put to the test, shortcomings in our ability to accurately measure, map, and value ES have surfaced. On the research side, mainstream methods for ES assessment still fall short of addressing the complex, multi-scale biophysical and socioeconomic dynamics inherent in ES provision, flow, and use. On the practitioner side, application of methods remains onerous due to data and model parameterization requirements. Further, it is increasingly clear that the dominant “one model fits all” paradigm is often ill-suited to address the diversity of real-world management situations that exist across the broad spectrum of coupled human-natural systems. This article introduces an integrated ES modeling methodology, named ARIES (ARtificial Intelligence for Ecosystem Services), which aims to introduce improvements on these fronts. To improve conceptual detail and representation of ES dynamics, it adopts a uniform conceptualization of ES that gives equal emphasis to their production, flow and use by society, while keeping model complexity low enough to enable rapid and inexpensive assessment in many contexts and for multiple services. To improve fit to diverse application contexts, the methodology is assisted by model integration technologies that allow assembly of customized models from a growing model base. By using computer learning and reasoning, model structure may be specialized for each application context without requiring costly expertise. In this article we discuss the founding principles of ARIES - both its innovative aspects for ES science and as an example of a new strategy to support more accurate decision making in diverse application contexts.

No abstract

ABSTRACT. Ecosystem services mapping and modeling has focused more on supply than demand, until recently. Whereas the potential provision of economic benefits from ecosystems to people is often quantified through ecological production functions, the use of and demand for ecosystem services has received less attention, as have the spatial flows of services from ecosystems to people. However, new modeling approaches that map and quantify service-specific sources (ecosystem capacity to provide a service), sinks (biophysical or anthropogenic features that deplete or alter service flows), users (user locations and level of demand), and spatial flows can provide a more complete understanding of ecosystem services. Through a case study in Puget Sound, Washington State, USA, we quantify and differentiate between the theoretical or in situ provision of services, i.e., ecosystems' capacity to supply services, and their actual provision when accounting for the location of beneficiaries and the spatial connections that mediate service flows between people and ecosystems. Our analysis includes five ecosystem services: carbon sequestration and storage, riverine flood regulation, sediment regulation for reservoirs, open space proximity, and scenic viewsheds. Each ecosystem service is characterized by different beneficiary groups and means of service flow. Using the ARtificial Intelligence for Ecosystem Services (ARIES) methodology we map service supply, demand, and flow, extending on simpler approaches used by past studies to map service provision and use. With the exception of the carbon sequestration service, regions that actually provided services to people, i.e., connected to beneficiaries via flow paths, amounted to 16-66% of those theoretically capable of supplying services, i.e., all ecosystems across the landscape. These results offer a more complete understanding of the spatial dynamics of ecosystem services and their effects, and may provide a sounder basis for economic valuation and policy applications than studies that consider only theoretical service provision and/or use.

No abstract