COS 81-1 - Using big data to understand cultural ecosystem services at landscape scales

Thursday, August 11, 2016: 1:30 PM
124/125, Ft Lauderdale Convention Center
Rebecca L. Hale, Biological Sciences, Idaho State University, Pocatello, ID, Elizabeth M. Cook, Instituto de Ciencias Ambientales y Evolutivas, Universidad Austral de Chile, Valdivia, Chile and Bray J Beltrán, Heart of the Rockies Initiative, Missoula, MT
Background/Question/Methods

Ecosystem service research has accelerated in the past decades, with advances in quantifying relationships between ecosystem structure and function and human well-being. However, research on cultural ecosystem services such as aesthetics, spiritual and religious values, education, and recreation has lagged. One reason for this gap is the difficulty in quantifying these values without using labor-intensive methods such as interviews. A second reason for this gap is the much less clear relationship between ecosystem structure and function and cultural services. Flickr, an image hosting website, provides a unique opportunity to quantify how and where users enjoy the environment. Images can be geotagged, and users can provide titles, descriptions, and tags for their images, which provide information about important user-defined image content. We asked: How can Flickr metadata be used to understand patterns of ecosystem service use? What are the relationships between ecosystem structure and cultural services? Do ecosystem service users cluster into different groups based on the types of services or types of ecosystems they enjoy? Here we used the 100M Flickr dataset to infer cultural service provisioning based on where and what about the environment and ecosystem structure people take pictures of in the state of Idaho.

Results/Conclusions

The majority of images had some ecologically-relevant tag in all land covers except developed areas. People most commonly tagged pictures for rivers, sagebrush, lakes, and parks, suggesting a preference for these features. Specific tags about the content of images were not evenly distributed across Idaho. Tags matched major ecosystem transitions (e.g., from sagebrush steppe to forest), but the locations of specific tags were dependent on ecosystem as well as access, such as parks and roads. The types of tags and proportion of pictures with ecosystem tags also varied by land cover. Approximately 1/3 of images on pasture lands were of views and plants, 1/3 of images in evergreen forest were of recreational activities, and 1/3 of images in grasslands and shrublands were of plants. Flickr users clustered into distinct groups based on the tags that they use. Three distinct specialist groups emerged that posted images primarily of views and scenery, animals, and recreation. Two additional generalist groups posted images with a mixture of tags. Flickr metadata offers a unique opportunity to understand how people use and perceive landscapes and a semi-quantitative method to characterize cultural ecosystem services and help prioritize conservation planning based on cultural services.