Fig. 2 Number of events that were detected correctly or with an error by the Bee Tracker software in the 15 videos that were checked visually
Discussion
The Bee Tracker software is a helpful tool to collect large amounts of data on the nesting and foraging behavior of solitary, cavity-nesting bees in an automated way. It identifies individual nesting females and assigns them to their nests. This permits to obtain robust data on per female reproductive success, if nesting progress within nests is additionally recorded. Moreover, the software counts the number of cavities a female probes until it finds its nest, collects information on the flight duration and allows to assess flight activity. Once the software is trained for the experimental setup in use, the method requires low labor input but can generate large data sets with a high measurement precision. Here we showed that a precision of 96% can be achieved with a relatively low training effort of about 30 working hours. Minor adaptations may further improve the performance of the software.
The software is designed to achieve a high precision at the expenses of the recall (fraction of events that was retrieved), which is of minor interest in this type of analysis as it only affects the sample sizes but not the extracted measurements themselves. The precision of theBee Tracker therefore exceeds precision values typically found in automated image analysis software (Eikelboom et al., 2019; Gallmann et al., 2020). The software may, however, only achieve the here reported precision of 96% in experiments with a similar setup, with respect to light conditions during video recording, hues and digits of bee IDs as well as the shape, size and location of the nest cavities in the nesting units. For variant setups, the training of the software may need to be repeated to achieve a comparable measurement precision of the software analysis. While errors by bee ID swapping cannot be entirely avoided due to the limitations of the centroid object tracking algorithm used by the software, errors caused by color misclassifications between green and yellow were probably caused by the convergence of spectra under different light conditions and could likely be reduced by choosing colors for ID tags with more distinct spectra. Thus, while an increased training effort may reduce the error rate, replacing either green or yellow by e.g. blue or red ID tags may completely eliminate color misclassifiactions, which would increase the precision to 98% in our data set.
Direct observations of the nesting activity of individually marked bees is very challenging and nearly impossible in most experimental setups, as several bees frequently aggregate in front of the nesting unit and the bee IDs are small for human vision while the bees move quickly. Researchers therefore used visual analysis of videos for the assessments of individual nesting and foraging behavior in solitary bees (McKinney & Park, 2012). In comparison to direct observations, a main advantage of the Bee Tracker is the large data sets that can be collected with relatively low time and labor input. Despite these advantages, the method also has some limitations. The main disadvantage of the software is its restriction to relatively large bee species that allow fixing ID tags on the bees’ thorax (e.g. tags produced for the marking of honey bee queens). Furthermore, the current version of the Bee Trackersoftware was trained on the model solitary bee species Osmia bicornis. Although bee recognition and the classification of movement (entering or leaving a cavity) seemed to work equally precise when tested on the closely related species O. cornuta (Knauer A., personal observation), further training may be required when working with other solitary bee species to obtain full precision of the software. Furthermore, the current version of the software can only analyze the above described 24 unique color-digit based bee IDs and identify cavities with a certain size and shape that are arranged in the nesting unit as described (Fig. 1). These limits can, however be adapted by training the software to additional bee IDs (with more digits or colors) and different nesting units. After such additional training, the software could be used in various experimental setups to study the behavior of solitary, cavity-nesting bees that can be established in standardized nesting units.
In social bee species, the number of adult bees, brood cells and the amount of food stores (honey and pollen) are used as indicators of colony strength and vitality (Dainat et al., 2020; Hernandez et al., 2020), while in solitary bees reproductive success measured by brood cell or offspring production is the most important proxy of fitness (Rundlöf et al., 2015; Stuligross & Williams, 2020; Zurbuchen et al., 2010). RFID technology has furthermore been used for the monitoring of foraging behavior in social species. RFID can automatically perform individual bee recognition and detect the inbound and outbound movements of tagged bees at the nest entrance where the antenna and reader are placed (Nunes-Silva et al., 2019). With this technology, flight activity, homing ability and flight duration of social bees can be studied (Henry et al., 2012; Schneider et al., 2012; Stanley et al., 2016; Tenczar et al., 2014). Such behavioral data can contribute to the understanding of behavior mediated impacts of environmental stressors on colony development (Henry et al., 2012). In addition, the measurements of behavior can be a powerful tool to assess the impact of specific stressors in (semi-)field experiments, especially as colony strength and development can be biased by various confounding factors (Oldroyd et al., 1992; Sandrock et al., 2014; Schmid‐Hempel & Schmid‐Hempel, 1998). Similarly, the Bee Tracker software can be used to collect large amounts of behavioral data to supplement and better understand measurements of reproductive success and fitness in solitary, cavity-nesting bees.
To our knowledge, the Bee Tracker software is the first automated tool that allows to efficiently collect large amounts of behavioral data on cavity-nesting solitary bee species. Foraging behavior can respond to various environmental stressors. Pesticide exposure for example, can impair orientation and memory in bees (Siviter et al., 2018) and increase flight duration or cause a reduction in homing or foraging activity (Artz & Pitts-Singer, 2015; Henry et al., 2012; Stanley et al., 2016). Flight duration may also be increased by habitat degradation or food competition, which can cause increased flight distances to food sources (Leonhardt et al., 2016; Thomson, 2004). Pathogens can reduce homing ability in honey bees (Li et al., 2013) or cause a premature onset of foraging and reduce the total activity span of foragers (Benaets et al., 2017). Overall, understanding bees’ foraging and flight activities can provide valuable information for evaluating the impact of a wide range of environmental stressors on bees. For example, behavioral data collected with RFID contributed to the detection of sublethal adverse effects of neonicotinoids which finally led to the ban of several compounds from this class of insecticides in the European Union (Gross, 2013).
The effect of different stressors can vary between species and depend on their functional traits such as body size, sociality or mode of nesting (Brittain & Potts, 2011; Sgolastra et al., 2019). A range of solitary bee species are therefore increasingly studied for the assessment and monitoring of stressors on pollinators (Boff et al., 2020; Ganser et al., 2020; Klaus et al., 2021; Stuligross & Williams, 2020; Zurbuchen et al., 2010). The Bee Tracker software can be a helpful tool to efficiently collect robust data on individual nesting and foraging behavior, of cavity-nesting solitary bees.
Data availability
The data associated with this manuscript and the software including the underlying Python code will be made available on dryad upon acceptance of the paper.
Author contribution
A.C.K. and M.A. conceived the study. J.G. developed the software. A.C.K. collected the data. A.C. conducted the statistical analysis. A.C.K. wrote the first version of the manuscript and all authors contributed to the writing of subsequent drafts.
Acknowledgements
We thank Jonas Winizki for technical support and advice. This project has received funding from from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 773921, PoshBee Project (www.poshbee.eu).
Competing interests
The authors declare no competing interests.
References
Artz, D. R., & Pitts-Singer, T. L. (2015). Effects of fungicide and adjuvant sprays on nesting behavior in two managed solitary bees, Osmia lignaria and Megachile rotundata. Plos One , 10 (8), e0135688. https://doi.org/10.1371/journal.pone.0135688
Benaets, K., Van Geystelen, A., Cardoen, D., De Smet, L., de Graaf, D. C., Schoofs, L., Larmuseau, M. H., Brettell, L. E., Martin, S. J., & Wenseleers, T. (2017). Covert deformed wing virus infections have long-term deleterious effects on honeybee foraging and survival.Proceedings of the Royal Society B: Biological Sciences ,284 (1848), 20162149. https://doi.org/10.1098/rspb.2016.2149
Boff, S., Raizer, J., & Lupi, D. (2020). Environmental Display Can Buffer the Effect of Pesticides on Solitary Bees. Insects ,11 (7), 417. https://doi.org/10.3390/insects11070417
Brittain, C., & Potts, S. G. (2011). The potential impacts of insecticides on the life-history traits of bees and the consequences for pollination. Basic and Applied Ecology , 12 (4), 321-331. https://doi.org/10.1016/j.baae.2010.12.004
Chollet, F. (2015). Keras .
Crall, J. D., Switzer, C. M., Oppenheimer, R. L., Versypt, A. N. F., Dey, B., Brown, A., Eyster, M., Guérin, C., Pierce, N. E., & Combes, S. A. (2018). Neonicotinoid exposure disrupts bumblebee nest behavior, social networks, and thermoregulation. Science , 362 (6415), 683-686. https://doi.org/10.1126/science.aat1598
Dainat, B., Dietemann, V., Imdorf, A., & Charrière, J.-D. (2020). A scientific note on the ‘Liebefeld Method’to estimate honey bee colony strength: its history, use, and translation. Apidologie , 1-6. https://doi.org/10.5281/zenodo.3341580
EFSA, G. O. (2013). EFSA Guidance Document on the risk assessment of plant protection products on bees (Apis mellifera, Bombus spp. and solitary bees). EFSA Journal , 11 (7), 3295. https://doi.org/10.2903/j.efsa.2013.3295
Eikelboom, J. A., Wind, J., van de Ven, E., Kenana, L. M., Schroder, B., de Knegt, H. J., van Langevelde, F., & Prins, H. H. (2019). Improving the precision and accuracy of animal population estimates with aerial image object detection. Methods in Ecology and Evolution ,10 (11), 1875-1887. https://doi.org/10.1111/2041-210X.13277
Franke, L., Elston, C., Jütte, T., Klein, O., Knäbe, S., Lückmann, J., Roessink, I., Persigehl, M., Cornement, M., & Exeler, N. (2021). Results of 2‐Year Ring Testing of a Semifield Study Design to Investigate Potential Impacts of Plant Protection Products on the Solitary Bees Osmia Bicornis and Osmia Cornuta and a Proposal for a Suitable Test Design. Environmental toxicology and chemistry ,40 (1), 236-250. https://doi.org/10.1002/etc.4874
Gallmann, J., Schüpbach, B., Jacot, K., Albrecht, M., Winizki, J., & Aasen, H. (2020). Flower Mapping in Grasslands with Drones and Deep Learning. Authorea . https://doi.org/10.22541/au.158740047.73964506
Ganser, D., Albrecht, M., & Knop, E. (2020). Wildflower strips enhance wild bee reproductive success. Journal of Applied Ecology , 486-495. https://doi.org/10.1111/1365-2664.13778
Goulson, D., Nicholls, E., Botías, C., & Rotheray, E. L. (2015). Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science , 347 (6229). https://doi.org/10.1126/science.1255957
Gross, M. (2013). EU ban puts spotlight on complex effects of neonicotinoids. 23 (11), R462-R464. https://doi.org/10.1016/j.cub.2013.05.030
Henry, M., Beguin, M., Requier, F., Rollin, O., Odoux, J.-F., Aupinel, P., Aptel, J., Tchamitchian, S., & Decourtye, A. (2012). A common pesticide decreases foraging success and survival in honey bees.Science , 336 (6079), 348-350. https://doi.org/10.1126/science.1215039
Hernandez, J., Maisonnasse, A., Cousin, M., Beri, C., Le Quintrec, C., Bouetard, A., Castex, D., Decante, D., Servel, E., & Buchwalder, G. (2020). ColEval: Honeybee COLony Structure EVALuation for Field Surveys.Insects , 11 (1), 41. https://doi.org/10.3390/insects11010041
Klaus, F., Tscharntke, T., Bischoff, G., & Grass, I. (2021). Floral resource diversification promotes solitary bee reproduction and may offset insecticide effects–evidence from a semi‐field experiment.Ecology Letters , 24 (4), 668-675. https://doi.org/10.1111/ele.13683
Klein, A.-M., Vaissiere, B. E., Cane, J. H., Steffan-Dewenter, I., Cunningham, S. A., Kremen, C., & Tscharntke, T. (2007). Importance of pollinators in changing landscapes for world crops. Proceedings of the Royal Society B: Biological Sciences , 274 (1608), 303-313.
Leonhardt, S. D., Kaluza, B. F., Wallace, H., & Heard, T. A. (2016). Resources or landmarks: which factors drive homing success in Tetragonula carbonaria foraging in natural and disturbed landscapes?Journal of Comparative Physiology A , 202 (9), 701-708. https://doi.org/10.1007/s00359-016-1100-5
Li, Z., Chen, Y., Zhang, S., Chen, S., Li, W., Yan, L., Shi, L., Wu, L., Sohr, A., & Su, S. (2013). Viral infection affects sucrose responsiveness and homing ability of forager honey bees, Apis mellifera L. Plos One , 8 (10), e77354. https://doi.org/10.1371/journal.pone.0077354
McKinney, M. I., & Park, Y.-L. (2012). Nesting activity and behavior of Osmia cornifrons (Hymenoptera: Megachilidae) elucidated using videography. Psyche , 2012 .
Nunes-Silva, P., Hrncir, M., Guimarães, J., Arruda, H., Costa, L., Pessin, G., Siqueira, J., De Souza, P., & Imperatriz-Fonseca, V. (2019). Applications of RFID technology on the study of bees.Insectes Sociaux , 66 (1), 15-24. https://doi.org/10.1007/s00040-018-0660-5
Oldroyd, B. P., Rinderer, T. E., Harbo, J. R., & Buco, S. M. (1992). Effects of intracolonial genetic diversity on honey bee (Hymenoptera: Apidae) colony performance. Annals of the Entomological Society of America , 85 (3), 335-343. https://doi.org/10.1093/aesa/85.3.335
Ollerton, J., Winfree, R., & Tarrant, S. (2011). How many flowering plants are pollinated by animals? Oikos , 120 (3), 321-326. https://doi.org/10.1111/j.1600-0706.2010.18644.x
Pennington, Z. T., Dong, Z., Feng, Y., Vetere, L. M., Page-Harley, L., Shuman, T., & Cai, D. J. (2019). ezTrack: An open-source video analysis pipeline for the investigation of animal behavior. Scientific Reports , 9 (1), 19979. https://doi.org/10.1038/s41598-019-56408-9
Potts, S. G., Biesmeijer, J. C., Kremen, C., Neumann, P., Schweiger, O., & Kunin, W. E. (2010). Global pollinator declines: trends, impacts and drivers. Trends in Ecology & Evolution , 25 (6), 345-353. https://doi.org/10.1016/j.tree.2010.01.007
Potts, S. G., Imperatriz Fonseca, V., Ngo, H. T., Biesmeijer, J. C., Breeze, T. D., Dicks, L., Garibaldi, L. A., Hill, R., Settele, J., & Vanbergen, A. J. (2016). Summary for policymakers of the assessment report of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services on pollinators, pollination and food production.
R Core Team. (2021). R: A language and environment for statistical computing.
Redmon, J., & Farhadi, A. (2018). Yolov3: An incremental improvement.arXiv preprint arXiv:1804.02767 .
Ren, S., He, K., Girshick, R., & Sun, J. (2017). Faster r-cnn: Towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 39 (6), 1137-1149. https://doi.org/10.1109/TPAMI.2016.2577031
Rundlöf, M., Andersson, G. K., Bommarco, R., Fries, I., Hederström, V., Herbertsson, L., Jonsson, O., Klatt, B. K., Pedersen, T. R., & Yourstone, J. (2015). Seed coating with a neonicotinoid insecticide negatively affects wild bees. Nature , 521 (7550), 77-80. https://doi.org/10.1038/nature14420
Sandrock, C., Tanadini, M., Tanadini, L. G., Fauser-Misslin, A., Potts, S. G., & Neumann, P. (2014). Impact of Chronic Neonicotinoid Exposure on Honeybee Colony Performance and Queen Supersedure. Plos One ,9 (8). https://doi.org/e10359210.1371/journal.pone.0103592
Schmid‐Hempel, R., & Schmid‐Hempel, P. (1998). Colony performance and immunocompetence of a social insect, Bombus terrestris, in poor and variable environments. Functional Ecology , 12 (1), 22-30. https://doi.org/10.1046/j.1365-2435.1998.00153.x
Schneider, C. W., Tautz, J., Grunewald, B., & Fuchs, S. (2012). RFID Tracking of Sublethal Effects of Two Neonicotinoid Insecticides on the Foraging Behavior of Apis mellifera. Plos One , 7 (1). https://doi.org/10.1371/journal.pone.0030023
Schönfelder, M. L., & Bogner, F. X. (2017). Individual perception of bees: Between perceived danger and willingness to protect. Plos One , 12 (6), e0180168. https://doi.org/10.1371/journal.pone.0180168
Sgolastra, F., Hinarejos, S., Pitts-Singer, T. L., Boyle, N. K., Joseph, T., Lūckmann, J., Raine, N. E., Singh, R., Williams, N. M., & Bosch, J. (2019). Pesticide exposure assessment paradigm for solitary bees.Environmental Entomology , 48 (1), 22-35. https://doi.org/10.1093/ee/nvy105
Siviter, H., Koricheva, J., Brown, M. J., & Leadbeater, E. (2018). Quantifying the impact of pesticides on learning and memory in bees.Journal of Applied Ecology , 55 (6), 2812-2821. https://doi.org/10.1111/1365-2664.13193
Siviter, H., Richman, S. K., & Muth, F. (2021). Field‐realistic neonicotinoid exposure has sub‐lethal effects on non‐Apis bees: A meta‐analysis. Ecology Letters . https://doi.org/10.1111/ele.13873
Stanley, D. A., Russell, A. L., Morrison, S. J., Rogers, C., & Raine, N. E. (2016). Investigating the impacts of field‐realistic exposure to a neonicotinoid pesticide on bumblebee foraging, homing ability and colony growth. Journal of Applied Ecology , 53 (5), 1440-1449. https://doi.org/10.1111/1365-2664.12689
Stuligross, C., & Williams, N. M. (2020). Pesticide and resource stressors additively impair wild bee reproduction. Proceedings of the Royal Society B , 287 (1935), 20201390. https://doi.org/10.1098/rspb.2020.1390
Tenczar, P., Lutz, C. C., Rao, V. D., Goldenfeld, N., & Robinson, G. E. (2014). Automated monitoring reveals extreme interindividual variation and plasticity in honeybee foraging activity levels. Animal Behaviour , 95 , 41-48. https://doi.org/10.1016/j.anbehav.2014.06.006
Thomson, D. (2004). Competitive interactions between the invasive European honey bee and native bumble bees. Ecology , 85 (2), 458-470. https://doi.org/10.1890/02-0626
Woodard, S. H., Federman, S., James, R. R., Danforth, B. N., Griswold, T. L., Inouye, D., McFrederick, Q. S., Morandin, L., Paul, D. L., & Sellers, E. (2020). Towards a US national program for monitoring native bees. Biological Conservation , 252 , 108821. https://doi.org/10.1016/j.biocon.2020.108821
Zurbuchen, A., Cheesman, S., Klaiber, J., Müller, A., Hein, S., & Dorn, S. (2010). Long foraging distances impose high costs on offspring production in solitary bees. Journal of Animal Ecology ,79 (3), 674-681. https://doi.org/10.1111/j.1365-2656.2010.01675.x