Toshifumi Kimura

A new approach for the simultaneous tracking of multiple honeybees for analysis of hive behavior

Social activities are among the most striking of animal behaviors, and the clarification of their mechanisms is a major subject in ethology. Honeybees are a good model for revealing these mechanisms because they display various social behaviors, such as division of labor, in their colonies. Image processing is a precise and convenient tool for obtaining animal behavior data, but even recent methods are inadequate for the identification or description of honeybee behavior. This is because of the difficulty distinguishing between the large number of individuals in a small hive and their multiple movements. The present study developed a new computer-aided system, using a vector quantization method, for the identification and behavioral tracking of individual honeybees. The vector quantization method enabled separation of honeybee bodies in photographs recorded as a movie. This system succeeded in analyzing a huge number of frames quickly and can thus save both time and labor. Moreover, the system identified more than 72% of the bees in a hive and found and determined the active areas in the hive by extracting the trajectories of walking bees. In addition, useful behavioral data on the honeybee waggle dance were obtained using the present system.

Waggle dance effect: dancing in autumn reduces the mass loss of a honeybee colony

SUMMARY A honeybee informs her nestmates about the location of a profitable food source that she has visited by means of a waggle dance: a round dance and a figure-of-eight dance for a short- and long-distance food source, respectively. Consequently, the colony achieves an effective collection of food. However, it is still not fully understood how much effect the dance behavior has on the food collection, because most of the relevant experiments have been performed only in limited locations under limited experimental conditions. Here, we examined the efficacy of the waggle dances by physically preventing bees from dancing and then analyzing the changes in daily mass of the hive as an index of daily food collection. To eliminate place- and year-specific effects, the experiments were performed under fully natural conditions in three different cities in Japan from mid September to early October in three different years. Because the experiments were performed in autumn, all six of the tested colonies lost mass on most of the experimental days. When the dance was prevented, the daily reduction in mass change was greater than when the dance was allowed, i.e. the dance inhibited the reduction of the hive mass. This indicates that dance is effective for food collection. Furthermore, clear inhibition was observed on the first two days of the experiments; after that, inhibition was no longer evident. This result suggests that the bee colony adapted to the new environment.

Development of a new method to track multiple honey bees with complex behaviors on a flat laboratory arena.

A computer program that tracks animal behavior, thereby revealing various features and mechanisms of social animals, is a powerful tool in ethological research. Because honeybee colonies are populated by thousands of bees, individuals co-exist in high physical densities and are difficult to track unless specifically tagged, which can affect behavior. In addition, honeybees react to light and recordings must be made under special red-light conditions, which the eyes of bees perceive as darkness. The resulting video images are scarcely distinguishable. We have developed a new algorithm, K-Track, for tracking numerous bees in a flat laboratory arena. Our program implements three main processes: (A) The object (bees) region is detected by simple threshold processing on gray scale images, (B) Individuals are identified by size, shape and spatiotemporal positional changes, and (C) Centers of mass of identified individuals are connected through all movie frames to yield individual behavioral trajectories. The tracking performance of our software was evaluated on movies of mobile multi-artificial agents and of 16 bees walking around a circular arena. K-Track accurately traced the trajectories of both artificial agents and bees. In the latter case, K-track outperformed Ctrax, well-known software for tracking multiple animals. To investigate interaction events in detail, we manually identified five interaction categories; ‘crossing’, ‘touching’, ‘passing’, ‘overlapping’ and ‘waiting’, and examined the extent to which the models accurately identified these categories from bees interactions. All 7 identified failures occurred near a wall at the outer edge of the arena. Finally, K-Track and Ctrax successfully tracked 77 and 60 of 84 recorded interactive events, respectively. K-Track identified multiple bees on a flat surface and tracked their speed changes and encounters with other bees, with good performance.

Observation system for the control of the hive environment by the honeybee (Apis mellifera)

The honeybee can control its hive environment to survive drastic changes in the field environment. To study the control of multiple environmental factors by honeybees, in this experiment, we developed a continual and simultaneous monitoring system for the temperature, moisture, and carbon dioxide (CO2) concentration in a honeybee hive. Changes in hive weight, CO2 production rate, and honeybee behavior were also monitored to estimate energy costs and behavioral activity for the environmental regulation. Measurements were conducted in August 2008. We found that the honeybee hive has a microclimate different from the ambient climate, and that the difference was partly accompanied by changes in honeybee activity. Our results also suggest that hive temperature, humidity, and CO2 concentrations are controlled by different mechanisms. Additional monitoring of the hive environment and honeybee behavior for longer periods would enable us to understand the mechanisms of environmental control by honeybees, which is one of the behaviors that define honeybees as social insects.

Error in the Honeybee Waggle Dance Improves Foraging Flexibility

The honeybee waggle dance communicates the location of profitable food sources, usually with a certain degree of error in the directional information ranging from 10–15° at the lower margin. We simulated one-day colonial foraging to address the biological significance of information error in the waggle dance. When the error was 30° or larger, the waggle dance was not beneficial. If the error was 15°, the waggle dance was beneficial when the food sources were scarce. When the error was 10° or smaller, the waggle dance was beneficial under all the conditions tested. Our simulation also showed that precise information (0–5° error) yielded great success in finding feeders, but also caused failures at finding new feeders, i.e., a high-risk high-return strategy. The observation that actual bees perform the waggle dance with an error of 10–15° might reflect, at least in part, the maintenance of a successful yet risky foraging trade-off.

Automated analysis of two-dimensional positions and body lengths of earthworms (Oligochaeta); MimizuTrack.

Earthworms are important soil macrofauna inhabiting almost all ecosystems. Their biomass is large and their burrowing and ingestion of soils alters soil physicochemical properties. Because of their large biomass, earthworms are regarded as an indicator of “soil heath”. However, primarily because the difficulties in quantifying their behavior, the extent of their impact on soil material flow dynamics and soil health is poorly understood. Image data, with the aid of image processing tools, are a powerful tool in quantifying the movements of objects. Image data sets are often very large and time-consuming to analyze, especially when continuously recorded and manually processed. We aimed to develop a system to quantify earthworm movement from video recordings. Our newly developed program successfully tracked the two-dimensional positions of three separate parts of the earthworm and simultaneously output the change in its body length. From the output data, we calculated the velocity of the earthworms movement. Our program processed the image data three times faster than the manual tracking system. To date, there are no existing systems to quantify earthworm activity from continuously recorded image data. The system developed in this study will reduce input time by a factor of three compared with manual data entry and will reduce errors involved in quantifying large data sets. Furthermore, it will provide more reliable measured values, although the program is still a prototype that needs further testing and improvement. Combined with other techniques, such as measuring metabolic gas emissions from earthworm bodies, this program could provide continuous observations of earthworm behavior in response to environmental variables under laboratory conditions. In the future, this standardized method will be applied to other animals, and the quantified earthworm movement will be incorporated into models of soil material flow dynamics or behavior in response to chemical substances present in the soil.