Abstract
A common debate on the value of citizen science projects is the accuracy of data collected and the validity of conclusions drawn. Sow Wild! was a hypothesis-driven citizen science project that investigated the benefits of sowing a 4 m2 mini-meadow in private gardens and allotments to attract beneficial insects. The use of researcher-verified specimen-based methods (pan traps, yellow sticky traps) and observational insect watches allowed investigation of potential bias in identification skills and sampling methods conducted by citizen scientists. For bumblebees and honeybees, identification of pan trap insect specimens was similar between researchers and citizen scientists, but solitary bees were possibly misidentified as social wasps or hoverflies. Key results of the Sow Wild! project differed between specimen-based and observation-only data sets, probably due to unconscious bias, such that incorrect conclusions may have been drawn if we had relied solely on observations made by citizen scientists without detailed training. Comparing the efficiency of sampling methods, insect watches produced the most insect observations overall. Yellow sticky traps collected more solitary wasps, social wasps, hoverflies and honeybees than pan traps. There was also variation in the abundance of insects caught according to the four pan trap colours. While all of these sampling methods can be successfully incorporated into citizen science projects to monitor a range of flying insects in urban landscapes, we recommend that verification of data by taxonomic experts is a valuable component of hypothesis-led citizen science projects, and increased training is required if target taxa include less conspicuous insect groups.