Calibration and validation of honey bee foraging flight duration as an estimate of distance flown

Abstract

Organic beekeeping standards require that beekeepers provide for their honey bees (Apis mellifera, L.) with a radius of 6.4 km of foraging land that is free from (3 km) or with reduced use of (additional 3.4 km) synthetic pesticides. This requirement is based on the honey bee's large maximum foraging range of 10-12 km, even though most foraging trips occur at c. 2 km from the hive. Because of this mismatch between organic foraging range requirements and more recent foraging range data, large scale honey bee foraging patterns on organic farms need to be investigated. But, current methods of estimating honey bee foraging ranges, including our current gold-standard of waggle dance decoding, are unsuitable for use in a large-scale, organic farm context. One potential new method of estimating distance flown is by using a bee's flight duration, or the time she spends outside the hive. However, what is not known is how, or how well, durations translate to distance. Here, we test the viability of the new method first with a Calibration Experiment, where we observed the flight durations of individually marked bees foraging at known distances. Then we performed a Validation Experiment, where we used a model fitted to the Calibration Experiment's data to predict flight distance from flight duration for freely flying bees, which we then compare to a distance estimate generated from the same bee's decoded waggle dance. We found in our Calibration Experiment that flight duration had a significant, positive relationship with distance flown when bees foraged upon artificial feeders. We then created two Calibration Models that both produced accurate, but imprecise, predictions of distance for bees traveling to feeders. These predictions were significantly correlated with the true, measured distances, as tested both with our analysis bees and a hold-out sample, which were not part of the main analysis (analysis population: p < 0.001; holdout sample: p < 0.001). However, the Validation Experiment revealed that a freely foraging bee's flight duration is not significantly related to her waggle run duration which has a known, linear relationship with distance. Therefore, flight duration did not produce accurate estimates of distance flown for freely foraging bees that, importantly, recruited nestmates. This may be because the use of dancing (recruiting) bees nonrandomized the sample of foragers. Additionally, we observed that pollen foragers had significantly longer flight durations (24.5 ± 11.8 minutes) than nectar foragers (16.7 ± 8.8 minutes; p < 0.001), even though pollen foragers had significantly shorter waggle run durations (1.11 ± 0.5 s) compared to nectar foragers (1.60 ± 0.42 s; p < 0.001). Overall, these results reveal that we can use flight duration as a proxy for distance, but cautiously, as predictions will be noisy and often reflect the upper boundary of flight distance for a given duration. Although we cannot yet explain why the validation does not support our calibration, we suspect that the sub-selection from freely foraging bees to those that are also at the stage of making a recruitment dance, interacting with different dance thresholds for differing resources, may produce the observed inverse relationship. Future work should track the flight duration ontogeny of bees foraging at known distances until they produce a waggle dance, which will allow us to determine the potential impact of distance, resource reliability, and optimized flight durations on waggle dance production.