Our October Editor’s Choice looks at the value citizen science brings to monitoring programmes and how to ensure that value doesn’t go to waste. Associate Editor, Yolanda F. Wiersma, discusses the selected article, Balancing sampling intensity against spatial coverage for a community science monitoring programme.

Citizen science (also termed ‘community science’), the involvement of non-credentialed scientists (‘ordinary citizens’) in a scientific research project, has a long history. Innovations in the internet and mobile technology have opened up new possibilities for widely distributed data collection, processing, and analysis connected to citizen science and there are many categories of participants and types of participation that can be involved.

The potential for data collection by a large number of people over large extents of space and time is one of the things that makes citizen science attractive to wildlife and natural resource managers. Much of this distributed monitoring falls in the category of ‘surveillance monitoring‘, rather than as part of a hypothesis-driven scientific research project. The thinking behind citizen/community monitoring programmes is that, by distributing citizens to collect data across a broad area and over long periods, it will quickly become possible to detect changes in populations of species or attributes of the environment (e.g. contaminant levels in air or water).

However, for monitoring to be effective, it needs to be designed in such a way that real changes and trends are detected (that is, to avoid a type I error), while at the same time avoiding seeing changes in trends that might be simply an artefact of data collection (or in other words, commit a type II error). Thus, careful consideration of how much survey effort is needed, and where, is necessary.

This Editor’s Choice article by Weiser et al. examines the trade-off between spatial coverage and sampling intensity for a monarch butterfly monitoring programme. Based on census work on their winter range, monarchs are in decline. Because of their migratory nature and distribution across North America, citizen science appears to be the most tractable solution to monitor their population trends outside of the winter range. The Integrated Monarch Monitoring Program (IMPP) is developing a sampling strategy to collect community science data, towards the goal of continental scale monitoring outside the overwinter period. However, Weiser et al. point out possible biases in volunteer monitoring. Such biases have been documented in other citizen science programmes as well and include proximity to human populations centres and observer bias to high quality habitats (citizens are motivated to help because they want to see the natural feature that the project is focused on, so will preferentially survey in more pristine conditions). To ensure that citizen science efforts yield useable data for trend monitoring, Weiser et al. set out to assess what would be an appropriate sampling strategy for the IMPP to carry out effective citizen-based monitoring.

Weiser et al. conducted a simulation study to determine how to optimise sampling breadth (number of sites distributed across the continent) against sampling intensity (number of visits to a particular site). They modelled different sampling scenarios to determine the statistical power of these designs to detect a 4% annual decline in monarch populations the (equivalent to the decline estimated based only on overwinter surveys). They tested how statistical power varied in three habitat types and for samples of adult butterflies, eggs and milkweed. Simulated protocols followed that of the IMPP and used pilot data from various sources. The analysis tested how variation in number of sites and number of years (sampling breadth) and variation in number of visits per site per year, and number of subplots (sampling intensity) affected the statistical power to detect a 4% decline. The authors found that sampling breadth had stronger effects than sampling intensity. When sampling milkweed or adult monarchs, statistical power did not change when sampling intensity was greater than three times per year or with more than 50 subplots. Asking citizens to sample more intensively than this could be wasted effort. The required sampling intensity for milkweed varied slightly by land use type, but all recommended sampling were considered moderate levels of effort for a continental scale community-science project. Thus, based on this prpject, it appears that citizen monitoring could effectively and efficiently detect a 4% decline.

The work of Weiser et al. provides valuable insight for the IMPP and thus for monarch butterfly conservation. Beyond work on monarch butterflies, it is a useful article for any practitioner involved in a citizen science initiative. It illustrates how to harness existing data in simulation modelling to test for statistical power and to evaluate the trade offs inherent in any sampling design. More citizen science-based projects should invest energy in this kind of up-front power analysis to ensure that the efforts of citizens do not go to waste, and more importantly, that they yield scientifically defensible monitoring data.

The full Editor’s Choice article, Balancing sampling intensity against spatial coverage for a community science monitoring programme, is free to read for a limited time in issue 56:10 of Journal of Applied Ecology.