Deriving indicators of biodiversity change from unstructured community‐contributed data

Opportunistic and unstructured observations of biodiversity crowdsourced from volunteers, community, and citizen scientists make up an increasingly large proportion of our global biodiversity knowledge. This incredible wealth of information exists in real time at both high resolutions and large extents of space, time, and taxonomy, thus holding huge potential to fill gaps in global biodiversity monitoring coverage in a cost‐effective way. Yet, the full potential of these data to provide essential indicators of biodiversity change for both research and management remains mostly unrealized, in large part due to the prevailing perception that the lack of standardization presents an unsurmountable barrier. In this paper, we provide an overview of the main challenges of working with unstructured community‐contributed data and synthesize the four fundamental approaches to overcome these challenges and extract useful inferences of biodiversity change, namely: 1) reverse‐engineering survey structure; 2) borrowing strength across taxa; 3) modeling the observation process, and; 4) integrating standardized data sources. To illustrate each of these approaches, we provide examples comparing community‐contributed observations crowdsourced via iNaturalist with long‐term standardized monitoring surveys for a subset of rocky intertidal organisms on the California coast from 2010 to 2019. We conclude by highlighting ways forward for the successful integration of unstructured community‐contributed observations within the global ecosystem of biodiversity change monitoring tools. Our ultimate goal is to update the prevailing perception among researchers and practitioners that unstructured community‐contributed observations of biodiversity are too noisy to use, and help establish this data stream as a key tool for research and management.