False positives and false negatives in citizen science monitoring data: should we be worried?
Date:
Awarded Student Talk: Second Place
Summary
The growth of citizen science, where volunteers participate in scientific data collection, represents a great opportunity for monitoring species population trends over large-scales. However, there is often concern over the quality of the data collected bythese projects,and the impacts this has on the outcome of such monitoring studies. We used a hierarchical modelling approach to test for the prevalence of two major types of observation error; imperfect detection (failing to detect a species when it is present), and species misidentification.
We estimated these rates using data from a 14 year volunteer monitoring study for amphibians in Switzerland. By comparing separate models that either account for orignorethese biases, we demonstrate the impacts that observer effect have on thecorrect assessment of population trends.
We demonstrate that amphibian detection probabilities are substantially lower than one, and that fora number of speciesmisidentification is a major issue. Ignoring these effects can be dangerous and can lead to inappropriate management. In our study region, in some cases, ignoring observer effects would have led us to conclude that populations that are in fact growing over time were locally extinct.
The observer effects we identified in our study are unlikely to be unique to volunteer-collected data. Withthe growth of citizen science programs, modelling techniques are developing to account for many of the issues we commonly encounter in species monitoring. Applying these new methods will allow us not only to initiate projects to collect data at a larger scale than previously possible, but is also essential if we are to avoid making inappropriate conservation management decisions