OPS 1-4
Assessing accuracy in volunteer-based plant phenology monitoring
Rigorous evaluations of patterns and changes in phenology necessitate repeated observations that are spatially and temporally comprehensive. In the United States, thousands of non-scientist volunteers are engaged in tracking plant and animal phenology through a variety of citizen science programs in an attempt to meet these data needs. However, the quality of these volunteer-based phenology observations and their suitability for scientific analysis remains largely untested. In this study we aimed to evaluate the accuracy of plant phenology observations collected by non-scientist volunteers following protocols designed by the USA National Phenology Network (USA-NPN). Phenology observations made by non-scientist volunteer observers were compared to those collected independently by a professional ecologist. Approximately 11,000 observations were recorded by 28 volunteers over the course of one field season. Our questions were: (1) How accurately were trained volunteer observers able to assess plant leaf, flower, and fruit phenophases overall? (2) Did volunteers’ accuracy in identifying phenophase status vary by species, plant functional group or phenophase? (3) Did volunteers’ ability to accurately identify phenophase status decrease during periods of phenophase transition? (4) Did volunteers that submitted more observations over the period of study exhibit a lower error rate than those that submitted fewer total observations?
Results/Conclusions
Volunteer participants consistently identified phenophases correctly (over 91% accurate overall) for the 19 plant species observed. However, when plants were transitioning between stages, the ability of volunteers to correctly identify phenophase status decreased to an average of 77% correct for phenophase onset and phenophase end. Accuracy also varied by species, phenophase and functional group. Volunteers demonstrated greatest accuracy identifying unfolded leaves, open flowers and ripe fruits. Volunteers that submitted fewer observations over the period of study did not did not exhibit a significantly higher error rate than those that submitted more total observations. These results suggest that even at the outset of participating, volunteers can provide reliable observations of plant phenology when following explicit, standardized protocols. These findings demonstrate the legitimacy and accuracy of phenology observations collected by volunteers, an important finding for analysts relying on data collected by non-scientists.