To make the HSI operational, we collected data on the health-related components of the colony. In short, we used a 3-tiered process that spanned 3 bee seasons and that gradually expanded from a restricted local level to a pan-European level. A BEEP digital sensor system (BEEP base) that measures the hive weight as well as the temperature and sound in the hive, was installed in all bee colonies studied (N = 383), whilst the digital BEEP logbook allowed to keep track of all manual inspections and automatically acquired data from sensors. Worker bee samples were taken from all bee colonies three times a year (in spring, summer and fall) for testing for pests and pathogens, but also for genotyping (variants related to Varroa resistance and subspecies determination) and to study viral diversity. Two supervised machine learning algorithms were employed to extract meaningful patterns from the automated data streams coming from the BEEP bases and to accurately predict colony survival outcomes. We also developed separately a high-performance unsupervised learning algorithm capable of identifying colonies whose size, honey productivity and HSI lie well beyond those of the other colonies (their collective success or failure) of the apiary. Hive weight emerged as the most accurate indicator for predicting honey bee colony survival, suggesting its pivotal role in HSI components.
To refine the HSI, several innovative tools were developed. The use of accelerometer systems for hive vibrational data recordings allowed the automated detection and categorisation of honeybee vibrations. This resulted in the discovery of a new-to-science pulsed honeybee vibration which we called the ‘purring bee’ signal, and which apparently correlates with queenlessness. We also installed an electromagnetic shaker required for assessing the frame content and determining the colony’s reaction to short vibrational pulses. It permitted the assessment, non-invasively, of colony overall mobility, the presence of the queen in the active season, the clustering of the colony, and its ‘restfulness’. Gas sensors deployed in hives demonstrated both the daily and long-term trends in the carbon dioxide and humidity. A spatially resolved sensor system with 48 temperature sensors in each of the ten brood frames of a bee colony recorded data with a living colony. These measurements provide experimental data for the calibration of the ApisRAM model of the European Food Safety Authority for a bee colony. The bee counter discovered an average daily loss of 2% of bees in 18 colonies (from 3 countries). To allow beekeepers and scientists to test different matrices for the presence of pesticides, we developed quick tests (dual-Lateral Flow Device, LFD) for on-site screening of neonicotinoids and sulfoxaflor.
We also developed floral resource models that integrated data on the composition of ‘bee-friendly’ plants within the main landscape elements/habitat units important for bees across Europe with data on floral resource production, flowering abundance and flowering phenology of individual plants. These floral resource models were then integrated into the ALMaSS landscape representation, that typically operates within a window of ten by ten kilometres, with a spatial resolution of 1 m². To link the ApisRAM model with the flower resource model, a foraging model was developed and incorporated into ApisRAM. We then assessed the impact of environmental temperature changes on honey bee foraging activity. Seven simulation scenarios were conducted to evaluate the impacts of two landscape management practices (flower strips and set-aside) on the foraged nectar and sugar. We created a Landscape Suitability Map for Honeybees across Europe, which led to a comprehensive assessment of the suitability of different habitats and landscapes for honeybees across Europe.
We assessed the socio-economics of healthy and sustainable beekeeping and identified viable and sustainable business models for European beekeeping. In the second Reporting Period (RP) a total of 504 stakeholders from 10 European countries completed the quantitative survey. The data were now further analyzed to better understand stakeholder views on i) policy priorities for fostering healthy and sustainable beekeeping, ii) the perceived and experienced impacts of climate change and iii) honeybee health. A total of 844 beekeepers from 18 European countries completed the quantitative survey in RP2 and yielded further insight into beekeepers’ attitudes and motivations for beekeeping, management practices, honeybee colony health, production efficiency and the quality of the natural environment for honey bees. Five distinct European beekeeper segments have been identified and profiled. During RP3 the data were further analyzed i) to assess the extent to which stakeholders involved in the European beekeeping sector and European beekeepers perceive and experience the impacts of climate change on their operations, and whether they had to adapt their practices accordingly and ii) to characterize European beekeepers who are implementing digital monitoring in their beekeeping operation versus those who are not.
Dissemination highlights of this period include further Beekeepers’ stories and scientific posters, a legacy booklet summarising all B-GOOD outcomes, a final beekeeper event in Wageningen/NL, and a closing conference organised as a satellite event to the COLOSS conference in Bled/SI.
Our Multi-actor Forum (MAF) has been continuously enlarged and guaranteed the exchange of best practices and uptake of our results by the beekeeping community. Continuous MAF feedback enabled us to refine our activities to ensure their relevance and exploitability.
In terms of exploitation, B-GOOD achieved a number of new technologies, tool kits, standardisation efforts, models and networks. Most notable achievements are related to the BEEP platform and making the BEEP measurement system future-proof.
The follow-up project Better-B started in June 2023.
