Monitoring the UK’s wild mammals: A new grammar for citizen science engagement and ecology

Anthropogenic activities have imperilled not just global ecosystems, but also the ecosystem services they provide which are crucial for human livelihoods. To understand these changes, there is a need for effective monitoring over large spatial and temporal scales. This thesis will build on two proposed solutions. First, citizen science – defined here as the involvement of non-professionals in scientific enquiry – allows the crowdsourcing of data collection and classification to expand monitoring in ways that are logistically infeasible for ecologists alone. Second, motion-sensing camera traps can reduce the labour needed for monitoring since they can be deployed for long periods and provide continuous, relatively unbiased observations. In this thesis, I describe MammalWeb, a citizen science project in north-east England where I enlisted the aid of the local community in wild mammal monitoring. Motivated by the current unevenness of survey effort and data for mammals in Great Britain, MammalWeb involves citizen scientists in both the collection and classification of camera trap images, a novel combination. This is a multidisciplinary project, and in the following chapters I will begin, in Chapter 2, with a detailed reflection on the organisation of the MammalWeb citizen science project and approaches to evaluating its performance. I observe that the majority of contributions came from a small subset of citizen scientists. In Chapter 3, I develop an economical approach to deriving consensus classifications from the aggregated input of multiple users, which is a crucial part of many citizen science projects. This is followed in Chapter 4 by a case study of a partnership I initiated between MammalWeb and the local Belmont Community School, where we empowered a group of secondary school students to not only aid in collecting data for MammalWeb, but also design and deliver ecological outreach to their community. This is now the template for a wider network of school partnerships we are pursuing. Chapter 5 will examine common concerns around estimating species occupancy from camera trap data, including post-hoc discretisation of observations and effects of missing data. I also develop a resampling method to account for uncertain detections, a common issue when crowdsourcing data classifications. I show that, through resampling, the estimated parameters from occupancy models are robust against high uncertainty in the underlying detections. Lastly, Chapter 6 will discuss how my work on MammalWeb has laid the foundation for a wider citizen science camera trapping network in the United Kingdom and avenues for future work. Importantly, I show that MammalWeb citizen scientists have been empowered to be more than “mobile sensors” and act as independent researchers who have initiated ecological studies elsewhere