Drifting behaviour in social bees using agent-based modelling to explore the effects of spatial layout on drifting patterns

Description

Social bees are commonly kept in artificial hives within apiaries to facilitate their management and resource harvesting. This practice favours forager drifting, i.e. the return to a nest that is not the forager's maternal nest, which can affect honey production, colony size and disease transmission. Drifting rates vary widely between studies, from 1% to 60%, and distance and hive configuration have been suggested as key drivers, but how drifting is shaped by apiary spatial layout remains unclear. We aimed to better understand how inter-hive distance and hive position shape drifting patterns, and whether drifting can emerge simply as a consequence of navigational mistakes. We hypothesised that a) larger distances between hives reduce drifting rates, and b) hive position affects drifting, with corner hives accumulating larger populations and more energy than hives positioned within a row. We developed an agent-based model that allowed us to simulate apiaries containing three or five hives, together with foraging bees and flowers. By varying inter-hive distance, hive position, and the radius at which bees recognise their hive, we explored how these spatial parameters shape drifting rates and colony-level outcomes such as population size and energy gain. Drifting decreased exponentially with increasing inter-hive distance, with even small increases in distance substantially reducing drifting rates. Hive position also affected colony dynamics: at close distances, corner hives accumulated higher populations and more energy than central and intermediate hives, although the proportion of drifters joining each hive position was similar regardless of position. Reducing the radius at which bees recognised hives reversed this pattern, with inner hives accumulating larger populations instead. Our findings support empirical evidence that apiary spatial layout, particularly inter-hive distance and hive position, is an important driver of drifting in social bees. Simulations suggest that relatively small adjustments to apiary layout could meaningfully reduce drifting, with implications for colony management, colony strength and disease transmission in managed bee populations.

Steps to reproduce

The dataset was produced using an agent-based model built in NetLogo 6.4, an open-source platform for agent-based simulation. The simulated environment was a two-dimensional square grid of 400 × 400 patches, where each patch represented 5 m and each simulation tick represented 1 second. Flowers (1000 food sources) and hives were placed within this grid, with hives positioned along the centre of the grid (y = 0) at varying distances depending on the configuration being tested. Individual bee agents followed a fixed set of behavioural rules at each tick: they moved at a flight speed of 1.4 patches/tick while searching for flowers, incurred a metabolic cost of 0.032 J/tick while flying, remained at a flower for 60 ticks per visit, and collected a fixed reward of 30 μl of nectar per trip. To return to the hive, agents used a directional "hive scent" cue, assessing scent strength within a return angle of 45° in front of them. An agent became a "drifter" if it entered the perception radius of a hive other than its hive of origin. Two model configurations were used to generate the core dataset. Model 1 simulated three hives, a central hive (the source of all initial foragers) and two corner hives, with the distance between the central hive and corner hive 2 varied from 1 to 100 patches (corner hive 1 fixed at 1 patch from the centre); one simulation was run for each distance value from 1 to 100. Model 2 simulated five hives in a row, each starting with 100 foragers, at inter-hive distances of 4, 6, 8, 10, 12, 14 and 16 patches, with 30 replicate simulations run per distance. Two further sets of simulations were generated to explore additional parameters: (1) variations in hive detection radius (1, 2 and 3 patches) at the closest and farthest distances tested (4 and 16 patches), and (2) a sensitivity analysis varying colony size (50, 100, 300), flight speed (1.2, 1.4, 1.6 patches/tick) and return angle (25°, 45°, 65°), also at 4 and 16 patches. For each simulation run, the proportions of agents were recorded throughout the run and averaged over the total foraging time. Total nest energy accumulated per hive was also recorded and divided by the number of bees to give energy collected per bee.

Institutions

• University of Bristol

  England, Bristol

Categories

Licence