Comparing time‐lapse photography and traditional human observation in collecting crop pollinator visitation data

Agricultural pollinator management requires knowledge on pollinator‐crop interactions, including information on pollinator visits to crop flowers. Time‐lapse trail cameras may provide an efficient and simple tool to collect such information. However, research evaluating their performance against traditional direct observation of crop pollinators is scarce. Here, we compare trail camera and human observer performance for collecting flower visitation data in two crops, apple and caraway, in boreal farmland. Five commercially available trail cameras recorded time‐lapse images of crop flowers throughout one flowering season. Near each camera, a human observer monitored flower visits five times per crop. In apple, cameras recorded more diverse pollinator assemblages than the human observer. This was due to better detection of less conspicuous flower visitors, as well as to a larger time window for the camera monitoring. Honeybees dominated flower visits in apple but with a lower proportion in the camera monitoring. In caraway, hoverflies dominated flower visits in both methods. Their proportion was similar between the methods during similar observation periods, but lower in the camera monitoring when using full camera data with larger temporal coverage. Detecting pollinators from images was harder for caraway than for apple. Time‐lapse photography with manual image analysis was more time‐ and cost‐efficient than human observation. In conclusion, trail camera photography appears as a competitive method for monitoring pollinator flower visits in apple but not in caraway. Automated pollinator detection from images can improve the method's efficiency in the future.