Plain Language Summary: Social media and deep learning reveal specific cultural preferences for biodiversity.

By Ilan Havinga, Diego Marcos, Patrick Bogaart, Dario Massimino, Lars Hein, and Devis Tuia.

This Plain Language Summary is published ahead of the article discussed; check back soon for a link to the full paper.

The experience of nature generates many cultural benefits, whether it’s the aesthetic enjoyment associated with a scenic walk or the sense of cultural identity gained from the presence of particular species. These contributions of nature to people are difficult to quantify but no less important than other, more tangible benefits such as food and medicine. Nevertheless, measuring these contributions is important to integrate this information in national-level assessments, such as the National Ecosystem Assessment conducted in the UK.

We looked at how we could draw on AI and social media to measure and understand the cultural contributions of biodiversity in Great Britain. Social media is a rich source of data due to the vast number of interactions being shared by people. In particular, the images people share contain a detailed insight into their preferences for natural elements in their environment. However, due to the size of the datasets, special AI techniques are needed to process this data.

In our study, we trained and applied a set of AI models on nearly 10 million images from the social media platform Flickr to identify people’s interactions with individual species in Great Britain. We found that applying these techniques allowed us to map people’s interactions with specific species, such as birds, insects and mammals. At the same time, the preferences reflected in these interactions were found to be unique to Flickr in comparison to those on iNaturalist, a citizen science platform.

We also found, however, that the preferences reflected in people’s images did not match ecological measure of bird biodiversity based on a selection of 36 bird species. This suggests that using these data sources and techniques does not reveal the complete cultural value of biodiversity in Great Britain, considering the importance placed on it through conservation policies. Nevertheless, the detail and scale in which these novel techniques can reveal individual preferences for biodiversity means more, and much-needed, information can be integrated into national-level ecosystem assessments.