How training citizen scientists affects the accuracy and precision of phenological data
Abstract
Monitoring plant and animal phenology is a critical step to anticipating and predicting changes in species interactions and biodiversity. Because phenology necessarily involves frequent and repeated observations over time, citizen scientists have become a vital part of collecting phenological data. However, there is still concern over the accuracy and precision of citizen science data. It is possible that training citizen scientists can improve data quality though there are few comparisons of trained and untrained citizen scientists in the ability of each to accurately and precisely measure phenology. We assessed how three types of observers-experts, trained citizen scientists that make repeated observations, and untrained citizen scientists making once-per-year observations-differ in quantifying temporal change in flower and fruit abundance of American mountain ash trees (Sorbus americana Marsh.) and arthropods in Acadia National Park, Maine, USA. We found that trained more so than untrained citizen science observers over- or under-estimated abundances leading to precise but inaccurate characterizations of phenological patterns. Our results suggest a new type of bias induced by repeated observations: A type of learning takes place that reduces the independence of observations taken on different trees or different dates. Thus, in this and many other cases, having individuals make one-off observations of marked plants may produce data as good if not better than individuals making repeated observations. For citizen science programs related to phenology, our results underscore the importance of (a) attracting the most number of observers possible even if they only make one observation, (b) producing easy-to-use and informative data sheets, and.
(c) carefully planning effective training programs that are, perhaps, repeated at different points during the data collection period.