When the VIBee project started, one of its central questions was deceptively simple: can we monitor honey bee colonies continuously, automatically and without disturbing their natural behaviour?
For decades, researchers have tried to count bees entering and leaving their hives. Such data can reveal daily flight activity, changes in foraging behaviour and potentially even daily losses of forager bees. However, automated bee counters are not just technical gadgets. If they are to be used in ecological research or pesticide risk assessment, their data must be reliable, validated and biologically meaningful.
VIBee helped to move this field forward. The project and its follow-up work contributed to several linked areas: automated bee counting, validation of monitoring devices, pollen resource assessment, colony-level indicators and, most recently, non-destructive in-hive monitoring of brood and pathogens.
Why counting bees is harder than it sounds
Automated bee counters have existed in different forms for almost a century. Modern systems use optical sensors, capacitive sensors, RFID technology or camera-based detection. All of them face the same basic challenge: bees do not behave like ideal objects passing through a laboratory device.
At the hive entrance, bees may hesitate, turn around, meet each other in the same channel, move slowly, sit in front of sensors, or pass in groups. These situations can lead to false counts or misclassification of incoming and outgoing bees. This is why VIBee placed strong emphasis on validation.
The BeeCheck system, based on capacitive sensors, was an important step in this development. It demonstrated that continuous recording of flight activity is possible under field conditions, but also showed where algorithms need to improve: especially when bees move back and forth, remain inside a passage, or meet other bees in the same entrance tube.
Validation is the key bottleneck
For research and regulatory use, it is not enough to know that a counter produces data. We need to know how accurate those data are.
A later VIBee-related study introduced a protocol for evaluating bee counters with respect to daily loss accuracy. The idea was to combine short-term observation with longer balance tests, such as the robbers test. Short video-based observations can reveal whether the system correctly detects individual events. Robbers tests, in contrast, can reveal whether small errors accumulate over a full day.
This distinction matters because daily loss is often much smaller than total daily traffic. A colony may generate thousands or tens of thousands of incoming and outgoing events per day, while the biologically relevant net loss may be much smaller. Even small systematic errors can therefore become important.
Why this matters for pesticide risk assessment
Traditional methods for assessing bee mortality, such as dead bee traps, are useful but incomplete. They mainly capture bees that die inside the hive and are carried out by undertaker bees. They do not capture foragers that die outside the hive or bees that fail to return because their orientation or flight behaviour was affected.
Automated bee counters could help close this gap. If validated properly, they could continuously measure flight activity, daily losses, homing success, foraging dynamics and changes in colony behaviour. This would be highly relevant for pesticide risk assessment, where both lethal and sublethal effects need to be considered under realistic field conditions.
However, this potential depends on one condition: the devices must be validated. Without robust error estimates, automated monitoring data cannot be interpreted transparently or compared across studies.
Looking beyond traffic: pollen as a window into the landscape
VIBee also contributed to automated pollen monitoring. Pollen is the main source of proteins and lipids for honey bees and reflects the floral resources available in the surrounding landscape. Traditionally, pollen diversity can be assessed by sorting pollen loads by colour or by laboratory-based palynological analysis.
The Pollenyzer app was developed to automate the assessment of pollen colour diversity from photographs of pollen trap samples. This can reduce subjective differences between observers and make pollen colour assessment faster and more reproducible.
At the same time, follow-up work showed an important limitation: pollen colour alone is not sufficient for reliable botanical classification. Natural colour variation within pollen types is large, and different plant taxa can produce similar colours. This means that colour-based tools are useful for monitoring colour diversity and resource patterns, but they should not be overinterpreted as precise species identification.
From the hive entrance into the colony
A more recent step extends automated monitoring from the hive entrance into the hive itself. In our study on in-hive flatbed scanners, a modified scanner was integrated into a brood frame to observe brood cells non-destructively over time.
The aim was to monitor processes that are usually difficult to follow without opening cells or disturbing the colony: egg laying, larval development, Varroa mite reproduction, chalkbrood development and hygienic behaviour. In a three-month pilot study, the system monitored hundreds of individual brood cells and generated thousands of images per cell.
This approach opens a new perspective. Instead of looking only at adult bee traffic, we can begin to observe brood development and pathogen dynamics continuously inside the colony. This is especially relevant because simulation work suggests that brood-related indicators may provide early warning signals of colony stress before winter losses become visible.
Towards digital colony phenotyping
Together, these studies point in the same direction: automated bee monitoring should not be limited to one device or one endpoint.
A future monitoring framework could combine:
- bee counters to measure flight activity and daily losses,
- hive scales to assess nectar flow and colony-level resource dynamics,
- pollen analysis to understand nutritional landscape quality,
- brood monitoring to detect early colony stress,
- pathogen and pest monitoring to follow Varroa and brood diseases,
- and validated algorithms to make these data comparable and interpretable.
This would move bee research closer to digital colony phenotyping: a continuous, non-destructive and multi-level description of colony health.
What comes next?
The next challenge is standardisation. Automated systems are becoming more powerful, but their outputs must be validated under realistic field conditions. This includes sensor-based counters, optical systems, capacitive devices, RFID approaches and camera-based methods.
VIBee has contributed several building blocks: a review of automated bee counters, early validation work with BeeCheck, a protocol for daily loss accuracy, tools for automated pollen assessment, a critical evaluation of pollen colour classification, modelling work on early warning indicators, and now a non-destructive approach for in-hive brood monitoring.
The long-term goal is clear: automated monitoring should help researchers, beekeepers and regulators detect stress earlier, understand colony dynamics better and evaluate environmental risks more realistically.
Related publications
Groeneveld, Jürgen; Odemer, Richard; Requier, Fabrice
Brood indicators are an early warning signal of honey bee colony loss—a simulation-based study Journal Article
In: PLOS ONE, 2024.
@article{nokey,
title = {Brood indicators are an early warning signal of honey bee colony loss—a simulation-based study},
author = {Jürgen Groeneveld and Richard Odemer and Fabrice Requier},
url = {http://vibee-project.net/wp-content/uploads/2024/05/Groeneveld_et_al_2024.pdf},
doi = {10.1371/journal.pone.0302907},
year = {2024},
date = {2024-05-16},
journal = {PLOS ONE},
abstract = {Honey bees (Apis mellifera) are exposed to multiple stressors such as pesticides, lack of forage and diseases. It is therefore a long-standing aim to develop robust and meaningful indicators of bee vitality to support beekeeping. While established indicators often focus on expected colony winter mortality based on adult bee abundance and honey stores at the beginning of the winter, it would be useful to have early warning indicators that allow detection of stress effects earlier in the year to allow for adaptive management. We used the established honey bee simulation model BEEHAVE to explore the potential of different indicators such as population size, number of capped brood cells, flight activity, abundance of varroa mites, honey stores and a brood-bee ratio. We implemented two stressor types in our simulations: 1) parasite pressure, i.e. sub-optimal Varroa treatment by the beekeeper (hereafter referred as Biotic stress) and 2) temporal forage gaps in spring and autumn (hereafter referred as Environmental stress). Neither stressor type could be detected by bee abundance or honey stores at the end of the first year. However, all response variables used in this study (population size, number of capped brood cells, flight activity, abundance of Varroa mites, honey stores, brood-bee ratio) did reveal early warning signals during the course of the year. The most reliable and useful measures seem to be related to brood and the abundance of Varroa mites at the end of the year. However, while in the model we have full access to time series of variables from stressed and unstressed colonies, knowledge of these variables in the field is challenging. We discuss how our findings can nevertheless be used to develop practical early warning indicators. As a next step in the interactive development of such indicators we suggest empirical studies on the importance of the number of capped brood cells at certain times of the year on bee population vitality. },
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Honey bees (Apis mellifera) are exposed to multiple stressors such as pesticides, lack of forage and diseases. It is therefore a long-standing aim to develop robust and meaningful indicators of bee vitality to support beekeeping. While established indicators often focus on expected colony winter mortality based on adult bee abundance and honey stores at the beginning of the winter, it would be useful to have early warning indicators that allow detection of stress effects earlier in the year to allow for adaptive management. We used the established honey bee simulation model BEEHAVE to explore the potential of different indicators such as population size, number of capped brood cells, flight activity, abundance of varroa mites, honey stores and a brood-bee ratio. We implemented two stressor types in our simulations: 1) parasite pressure, i.e. sub-optimal Varroa treatment by the beekeeper (hereafter referred as Biotic stress) and 2) temporal forage gaps in spring and autumn (hereafter referred as Environmental stress). Neither stressor type could be detected by bee abundance or honey stores at the end of the first year. However, all response variables used in this study (population size, number of capped brood cells, flight activity, abundance of Varroa mites, honey stores, brood-bee ratio) did reveal early warning signals during the course of the year. The most reliable and useful measures seem to be related to brood and the abundance of Varroa mites at the end of the year. However, while in the model we have full access to time series of variables from stressed and unstressed colonies, knowledge of these variables in the field is challenging. We discuss how our findings can nevertheless be used to develop practical early warning indicators. As a next step in the interactive development of such indicators we suggest empirical studies on the importance of the number of capped brood cells at certain times of the year on bee population vitality.
Borlinghaus, Parzival; Tausch, Frederic; Odemer, Richard
Natural color dispersion of corbicular pollen limits color-based classification Journal Article
In: ISPRS Open Journal of Photogrammetry and Remote Sensing, 2024.
@article{nokey,
title = {Natural color dispersion of corbicular pollen limits color-based classification},
author = {Parzival Borlinghaus and Frederic Tausch and Richard Odemer},
url = {http://vibee-project.net/wp-content/uploads/2024/04/Borlinghaus_et_al_pollen_color_dispersion.pdf},
doi = {10.1016/j.ophoto.2024.100063},
year = {2024},
date = {2024-04-16},
urldate = {2024-04-16},
journal = {ISPRS Open Journal of Photogrammetry and Remote Sensing},
abstract = {Various methods have been developed to assign pollen to its botanical origin. They range from technically complex approaches to the less precise but sophisticated chromatic assessment, in which the pollen colors are used for identification. However, a common challenge lies in the similarity of colors of pollen from different plant species. The advent of camera-based bee monitoring systems has sparked renewed interest in classifying pollen based on color and offers potential advances for honey bee biomonitoring. Despite the promise of improved sensor accuracy, a critical examination of whether color diversity within a single species may be the primary limiting factor has been lacking. Our comprehensive analysis, which includes over 85,000 corbicular pollen from 30 major pollen species, shows that the average color variation within each species is distinguishable to a human observer, similar to the difference between two dissimilar colors. From today's perspective, the considerable color variation within a single pollen source makes the use of color alone to classify pollen impractical. When picking a single pollen color from the entire dataset, we report a correct pollen type classification rate of 67 %. The accuracy was highly dependent on the type and ranged from 0 % for rare types with common colors to 99 % for distinct colors. The large color dispersion within species highlights the need for complementary methods to improve the accuracy and reliability of color-based pollen identification in biomonitoring applications.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Various methods have been developed to assign pollen to its botanical origin. They range from technically complex approaches to the less precise but sophisticated chromatic assessment, in which the pollen colors are used for identification. However, a common challenge lies in the similarity of colors of pollen from different plant species. The advent of camera-based bee monitoring systems has sparked renewed interest in classifying pollen based on color and offers potential advances for honey bee biomonitoring. Despite the promise of improved sensor accuracy, a critical examination of whether color diversity within a single species may be the primary limiting factor has been lacking. Our comprehensive analysis, which includes over 85,000 corbicular pollen from 30 major pollen species, shows that the average color variation within each species is distinguishable to a human observer, similar to the difference between two dissimilar colors. From today's perspective, the considerable color variation within a single pollen source makes the use of color alone to classify pollen impractical. When picking a single pollen color from the entire dataset, we report a correct pollen type classification rate of 67 %. The accuracy was highly dependent on the type and ranged from 0 % for rare types with common colors to 99 % for distinct colors. The large color dispersion within species highlights the need for complementary methods to improve the accuracy and reliability of color-based pollen identification in biomonitoring applications.
Odemer, Richard; Jakoby, Oliver; Barth, Markus; Knäbe, Silvio; Pistorius, Jens; Schmidt, Katharina
Making way for the implementation of automated bee counters in regulatory risk assessment Journal Article
In: Journal of Applied Entomology, 2024.
@article{nokey,
title = {Making way for the implementation of automated bee counters in regulatory risk assessment},
author = {Richard Odemer and Oliver Jakoby and Markus Barth and Silvio Knäbe and Jens Pistorius and Katharina Schmidt},
url = {http://vibee-project.net/wp-content/uploads/2024/04/VIBEE_Odemer_Perspective_BeeCounter_RiskAssessment_2024.pdf},
doi = {10.1111/jen.13256},
year = {2024},
date = {2024-04-03},
urldate = {2024-04-03},
journal = {Journal of Applied Entomology},
abstract = {Measuring adverse effects on honey bees and their colonies requires a suitable methodology. For example, due to the large number of bees in a hive and the foraging activity, measuring the mortality of individuals is a difficult task that has not yet been adequately addressed. Knowing the natural daily mortality rate of a bee colony would be of great benefit in assessing whether and to what extent external influences and stress factors affect mortality. More precise mortality data could in turn help refining specific protection goals for regulatory purpose. The European Food Safety Authority (EFSA) recently published a document that estimated such mortality rates based on a systematic literature review, but none of these rates were assessed from continuous monitoring of colonies. Currently, bee mortality is routinely evaluated with various types of dead bee traps that prevent deceased bees from being removed from the colony. Both the literature review and the dead bee traps are relevant to regulatory risk assessment, but in our opinion are not describing the total mortality. Bee counters capable of precisely determining daily loss rates meet the above points and combine them with generating automated and continuous monitoring data. Lately, the field has gained a lot of importance in research and technological advances offer new possibilities in regulatory risk assessment. We will highlight these possibilities and discuss their future application in practice.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Measuring adverse effects on honey bees and their colonies requires a suitable methodology. For example, due to the large number of bees in a hive and the foraging activity, measuring the mortality of individuals is a difficult task that has not yet been adequately addressed. Knowing the natural daily mortality rate of a bee colony would be of great benefit in assessing whether and to what extent external influences and stress factors affect mortality. More precise mortality data could in turn help refining specific protection goals for regulatory purpose. The European Food Safety Authority (EFSA) recently published a document that estimated such mortality rates based on a systematic literature review, but none of these rates were assessed from continuous monitoring of colonies. Currently, bee mortality is routinely evaluated with various types of dead bee traps that prevent deceased bees from being removed from the colony. Both the literature review and the dead bee traps are relevant to regulatory risk assessment, but in our opinion are not describing the total mortality. Bee counters capable of precisely determining daily loss rates meet the above points and combine them with generating automated and continuous monitoring data. Lately, the field has gained a lot of importance in research and technological advances offer new possibilities in regulatory risk assessment. We will highlight these possibilities and discuss their future application in practice.
Borlinghaus, Parzival; Jung, Jakob; Odemer, Richard
Introducing Pollenyzer: An App for Automatic Determination of Colour Diversity for Corbicular Pollen Loads Journal Article
In: Smart Agricultural Technology, 2023.
@article{nokey,
title = {Introducing Pollenyzer: An App for Automatic Determination of Colour Diversity for Corbicular Pollen Loads},
author = {Parzival Borlinghaus and Jakob Jung and Richard Odemer },
url = {http://vibee-project.net/wp-content/uploads/2023/06/Borlinghaus_et_al_Pollenyzer.pdf},
doi = {10.1016/j.atech.2023.100263},
year = {2023},
date = {2023-06-01},
urldate = {2023-06-01},
journal = {Smart Agricultural Technology},
abstract = {Pollen is known to be the only source of proteins and fats for honey bees. Therefore, it is an important component of nutrition, essential for brood care and a good indicator for the availability of resources in a landscape. It is also known that a diverse diet is beneficial for bee health, also in relation to winter losses. In this work, an app is presented that allows to quantify the pollen from a pollen trap and to determine its colour diversity in an automatic way. The colour diversity is closely related to the actual plant diversity. This correlation allows conclusions to be drawn on the apicultural importance of a landscape and on biodiversity in general. In this way, the app provides beekeepers with important information about the nutritional condition of their colonies, while scientists can benefit from aggregated information about local biodiversity. The app is free of use and available as a web app on all devices.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Pollen is known to be the only source of proteins and fats for honey bees. Therefore, it is an important component of nutrition, essential for brood care and a good indicator for the availability of resources in a landscape. It is also known that a diverse diet is beneficial for bee health, also in relation to winter losses. In this work, an app is presented that allows to quantify the pollen from a pollen trap and to determine its colour diversity in an automatic way. The colour diversity is closely related to the actual plant diversity. This correlation allows conclusions to be drawn on the apicultural importance of a landscape and on biodiversity in general. In this way, the app provides beekeepers with important information about the nutritional condition of their colonies, while scientists can benefit from aggregated information about local biodiversity. The app is free of use and available as a web app on all devices.
Borlinghaus, Parzival; Odemer, Richard; Tausch, Frederic; Schmidt, Katharina; Grothe, Oliver
Honey bee counter evaluation – Introducing a novel protocol for measuring daily loss accuracy Journal Article
In: Computers and Electronics in Agriculture, vol. 197, 2022.
@article{nokey,
title = {Honey bee counter evaluation – Introducing a novel protocol for measuring daily loss accuracy},
author = {Parzival Borlinghaus and Richard Odemer and Frederic Tausch and Katharina Schmidt and Oliver Grothe},
url = {http://vibee-project.net/wp-content/uploads/2022/04/Borlinghaus_et_al_2022.pdf},
doi = {10.1016/j.compag.2022.106957},
year = {2022},
date = {2022-04-26},
urldate = {2022-04-26},
journal = {Computers and Electronics in Agriculture},
volume = {197},
abstract = {Automated bee counters advanced over the last hundred years and became increasingly diverse. However, to date, there is no method for standardized validation of counting accuracy and thus no reliable data on daily bee losses, or background mortality in colonies. However, such data are in urgent need by regulators to establish future guidelines for pesticide risk assessment. In this work, existing approaches were combined to form a novel protocol for validating bee counters. In a case study, we demonstrated that the protocol is sufficiently feasible to determine the measurement accuracy of a commercial counting system. Measurement accuracy was modeled by the difficulty of specific measurement conditions. Daily loss, i.e., the difference between incoming and outgoing bees, can be used to assess colony health, environmental impacts, and infer the effect of pesticides on bee colonies. The developed protocol makes innovations in this field measurable and creates a foundation for the benchmarking of different types of bee counting systems. We discuss how it can be utilized in an effort to move the sector forward in the future.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Automated bee counters advanced over the last hundred years and became increasingly diverse. However, to date, there is no method for standardized validation of counting accuracy and thus no reliable data on daily bee losses, or background mortality in colonies. However, such data are in urgent need by regulators to establish future guidelines for pesticide risk assessment. In this work, existing approaches were combined to form a novel protocol for validating bee counters. In a case study, we demonstrated that the protocol is sufficiently feasible to determine the measurement accuracy of a commercial counting system. Measurement accuracy was modeled by the difficulty of specific measurement conditions. Daily loss, i.e., the difference between incoming and outgoing bees, can be used to assess colony health, environmental impacts, and infer the effect of pesticides on bee colonies. The developed protocol makes innovations in this field measurable and creates a foundation for the benchmarking of different types of bee counting systems. We discuss how it can be utilized in an effort to move the sector forward in the future.
Odemer, Richard
Approaches, challenges and recent advances in automated bee counting devices: A review Journal Article
In: Annals of Aplied Biology, 2021.
@article{nokey,
title = {Approaches, challenges and recent advances in automated bee counting devices: A review},
author = {Richard Odemer},
url = {http://vibee-project.net/wp-content/uploads/2021/10/Odemer_2021_BeeCounter_Review.pdf},
doi = {10.1111/aab.12727},
year = {2021},
date = {2021-08-08},
urldate = {2021-08-08},
journal = {Annals of Aplied Biology},
abstract = {For nearly 100 years, electronic bee counters have been developed using various technologies to track the foraging activity of mostly honey bee colonies. These counters should enable remote monitoring of the hives without disturbing natural flight behavior while generating precise scientific data. Today, however, there are not many counters on the market, that are able to fulfill this task. One main challenge is the lack of standardized methods to validate a counter’s precision, but validation is crucial to categorize and judge the data produced by the counter, especially for scientific purposes. Another challenge is the interpretation of flight data to measure the effects of environmental or anthropogenic sources. Nevertheless, recent developments in the field are promising. This review describes the historic development of automated bee flight measurement and critically compares validation methods to encourage their improvement. Lastly, to increase the comparability of future analyses with bee counters, current advances in data interpretation are also presented.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
For nearly 100 years, electronic bee counters have been developed using various technologies to track the foraging activity of mostly honey bee colonies. These counters should enable remote monitoring of the hives without disturbing natural flight behavior while generating precise scientific data. Today, however, there are not many counters on the market, that are able to fulfill this task. One main challenge is the lack of standardized methods to validate a counter’s precision, but validation is crucial to categorize and judge the data produced by the counter, especially for scientific purposes. Another challenge is the interpretation of flight data to measure the effects of environmental or anthropogenic sources. Nevertheless, recent developments in the field are promising. This review describes the historic development of automated bee flight measurement and critically compares validation methods to encourage their improvement. Lastly, to increase the comparability of future analyses with bee counters, current advances in data interpretation are also presented.
Bermig, Sven; Odemer, Richard; Gombert, Alina J.; Frommberger, Malte; Rosenquist, Ralf; Pistorius, Jens
Experimental validation of an electronic counting device to determine flight activity of honey bees (Apis mellifera L.) Journal Article
In: Journal fur Kulturpflanzen, vol. 72, pp. 132-140, 2020, ISSN: 1867-0938.
@article{Bermig2020,
title = {Experimental validation of an electronic counting device to determine flight activity of honey bees (Apis mellifera L.)},
author = {Sven Bermig and Richard Odemer and Alina J. Gombert and Malte Frommberger and Ralf Rosenquist and Jens Pistorius},
url = {http://vibee-project.net/wp-content/uploads/2021/10/JKPF_BeeCheck_Artikel_final.pdf},
doi = {10.5073/JfK.2020.05.03},
issn = {1867-0938},
year = {2020},
date = {2020-05-01},
urldate = {2020-05-01},
journal = {Journal fur Kulturpflanzen},
volume = {72},
pages = {132-140},
publisher = {Eugen Ulmer KG},
abstract = {In this work, a functional prototype of the BeeCheck counting device was evaluated for its accuracy to validate its suitability for scientific purposes. Two different approaches were applied: (i) we manually compared electronic data of the counting device by video recordings of entry and exit events, and (ii) by using the so-called "robber's test" in a tunnel tent. The results showed an expected temperature dependency of the general flight activity. Difficulties occurred with certain activities at the hive entrance. The various running speeds of individuals, approaching or stuck bees, and bees moving back and forth in the tube were a challenge for sensor technology and the mathematical algorithm. To minimize such mistakes and to increase the counting accuracy, it is necessary to correct the algorithm accordingly. This will be addressed in the "V-I-Bee" follow-up project and future perspectives of using an improved counting device are discussed.
This article was not funded by the project, but can be considered as predecessor which is why it is presented here.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In this work, a functional prototype of the BeeCheck counting device was evaluated for its accuracy to validate its suitability for scientific purposes. Two different approaches were applied: (i) we manually compared electronic data of the counting device by video recordings of entry and exit events, and (ii) by using the so-called "robber's test" in a tunnel tent. The results showed an expected temperature dependency of the general flight activity. Difficulties occurred with certain activities at the hive entrance. The various running speeds of individuals, approaching or stuck bees, and bees moving back and forth in the tube were a challenge for sensor technology and the mathematical algorithm. To minimize such mistakes and to increase the counting accuracy, it is necessary to correct the algorithm accordingly. This will be addressed in the "V-I-Bee" follow-up project and future perspectives of using an improved counting device are discussed.
This article was not funded by the project, but can be considered as predecessor which is why it is presented here.
