Shortcut Navigation:

Memory Retention in Landscape Learning of Honeybees, Apis mellifera



Honeybees (Apis mellifera) play a vital role in U.S. agriculture as pollinators of several crops, including carrots, almonds and apples. Pollination is the result of honeybees' ability to remember foraging sites. This study investigated honeybee memory endurance using the relay landscape learning method. Honeybees (n=1,000, Day 0) were released 0.8 kilometers from the experimental hive in opposite directions, 500 for the relay and 500 for the control. On Days 1, 3, 6 and 9, 40 relay and 40 control honeybees were recaptured, repainted, and re-released 1.8 kilometers from the hive. The relay effect present on Day 1 showed strong memory retention (p=<0.001, G=39.95) while limited memory existed at Day 9 (p=0.22, G=1.5). After learning their necessary navigational cues, relay honeybees had an initially high homing rate (42.5% relay vs. 5% control on Day 1) but they lost their memory as time progressed (23.6% relay vs. 10% control on Days 6 to 9). Long-term memory loss was exhibited between Day 6 and 9. The exact demarcation of the long-term memory lapse is unknown and needs further investigation. Colony collapse disorder (CCD) also involves a gradual memory loss in honeybees respective to their hives; this study should be compared to CCD as a means to investigate causal factors for memory demise.


Apis mellifera, the female honeybee, has always been the model organism in the discovery of sensory abilities in insects (Menzel and Müller 1996). To assist with navigation, honeybees have a greatly evolved social structure coupled with unique behavioral techniques such as a cognitive map, which is the spatial representation of objects within the honeybees' internal environment (Wray et al. 2008). Honeybees—with minute brains approximately one millimeter in size and containing 960,000 neurons—have the ability to learn and recognize patterns of targets rapidly and effectively. This provides accurate navigation with a rapid homing rate (Srinivasan 2010).


A honeybee collecting nectar

Honeybees use landmarks, celestial cues and path integration to forage for pollen and nectar and natural resources like propolis and water (Srinivasan 2010). The act of foraging is necessary for the continuation and development of a hive (Greenleaf et al. 2007). After collecting food, honeybees can navigate back to their hives effectively using these sensory motor routines, which are learned to solve intricate navigational tasks.

Honeybees apply a navigational motor routine in which the bees are able to recognize a landmark and calibrate its distance from their hive. Path integration, or "dead reckoning," is the ability of a forager to remember all the turns it completed to calibrate a straight line to the hive. Honeybees form simple associations between landmarks to navigate along their flight path (Menzel et al. 2005). The honeybees also use the Sun and other celestial cues as a compass along familiar routes during the process of path integration. For these reasons, honeybees can fly back to their hives in complicated situations. In tasks, honeybees can employ these techniques to return to the hive rapidly (Menzel et al. 1998).

Memory is a vital characteristic for honeybees. Memory is developed through learning but is not established instantaneously; memory develops gradually over time (Menzel 1999). Honeybees employ two types of short-term memories: early short-term memory (eSTM) and late short-term memory (lSTM). The "working memory," or eSTM, involves keeping an active memory over a short span of time for the purpose of foraging cycles. Working memory is retrieved from a single learning task and is slowly consolidated into lSTM. lSTM leads to mid-term memory (MTM), which is stable with new learning experience. The transfer from eSTM to long-term memory (LTM) is time- and event-dependent. LTM develops only after three learning trials, and retention vanishes over several days if the actions are not repeated. Intertrial intervals (ITIs) are important in memory retention because they determine whether memory retention is susceptible to interference. If the intervals between the sequential learning trials are more than two days apart, impaired memory retention results (Menzel 1999). A single learning trial initiates a time-dependent process in which decay occurs because only multiple learning trials can establish LTM (Menzel and Muller 1996).


Figure 1. Relay honeybees are able to return home more rapidly than control bees (Huang, 2009; unpublished data).

A novel concept involving landmarks is called "relay landscape learning," which is a reconnaissance flight that tests honeybees' ability to remember their flight paths. In testing relay landscape learning, honeybees return to their hive after being released from a point near a previous release. Since they have previous experience near the release area, the honeybees are able to navigate more easily. Figure 1 shows the results of one such experiment. The data indicates that the honeybees that are familiar with a specific landscape can return to the hive more rapidly than inexperienced honeybees. Since only 50 percent of the honeybees returned home in the allotted time frame, the relay effect presented in Figure 1 is only somewhat successful. The first release allowed bees to learn more about the area in relation to their hive. During the second release, located in the same direction but a kilometer further away from the hive, bees used their previous memory and had more success in returning home compared to the control bees. This is called the "relay effect" because it allows bees to return home at a higher rate, as if this were a relay; bees released two kilometers away were helped by their prior experience at one kilometer away, if released in the same general direction. The control bees behaved differently because they were released in the opposite direction from the first release. A "relay breakdown" will occur when the second release is at a distance greater than three kilometers away from the hive.

Honeybee memory is a primary source of honeybee prosperity. Therefore, determining the relationship between the duration of memory to relay landscape learning is the main purpose of the study. Flight direction is an important factor in determining long-memory retention because of the landmarks located in that specific region.



The map shows the release locations in Trial 1. Site A (0.8 km north of the hive) is the starting point for the relay group, and Site B (0.8 km south of the hive) is the starting point for the control group. Five hundred bees were released at each site on Day 0. On Days 1, 3, 6 and 9, honeybees were released at Site A and Site B, the relay release sites 1.8 km away from the hive.

The experiment locale was determined using Garmin® GPS coordinates: 42.4044°N, 84.2840°W, July to August 2010. Before performing the actual experiment, the relay landscape effect was conducted in order to identify the best experimental location (i.e. where honeybees had the most navigational success). The experimental release location was called Site A, and the control release location was called Site B. Site A was 0.8 km north and Site B was 0.8 km south of the hive (See map). For alternative trials, east and west directions, separated by 0.8 km, were used from the base hive to determine bee homing rate variations.

Honeybees were released on days with temperature ranges of 76° to 80°C without overcast. At Day 0, 1,000 bees were collected using a BioQuip insect vacuum. Forager bees were identified by their location outside the hive and their wing shape (for example, wings with jagged edges represent an older bee). All the forager bees had 21 days of experience in the local area. On Day 0 before the Site A and B releases, 500 bees were each painted blue or pink on the thorax. The blue bees and pink bees represented the relay and control groups, respectively. Paint colors also differentiated the honeybees used in multiple trials and the honeybees not involved in experimentation. The bees were placed in plastic cages, as relay and control sets, with an ample food supply.


The hive


Jill and a co-investigator prepare for the release.

Release of Honeybees: Days 0 to 9
Before the release, the hives were closed to prevent the relay and control bees from flying back to the hive. On Day 0, the blue bees were released 0.8 km north and the pink bees were released 0.8 km south of the hive. On Day 1, 40 blue bees and 40 pink bees were collected from the base hive. The bees were painted white on their abdomens to note the day they returned. These bees were then released 1.8 km north of the hive to determine the relay effect. After two hours, honeybees were no longer captured at the hive. The total number of bees retrieved between 0 and 120 minutes from the northern direction was recorded. This process was repeated for Days 3, 6 and 9. All of the experimental bees were kept in the hive and were retrieved each day to lower mortality.


Two trial bees return to the hive (note the orange and yellow on their thoraxes).

Repetition of Trials
The experiment was repeated two times with different hives. In the study field, the hives were placed approximately 0.26 km away. The hives used were relevant in this experiment because their locations were either close to landmarks or not. Landmarks aid honeybees in their flight home. In Trial 2, 1,000 bees were collected and then painted green or silver on the thorax. On Day 0, the bees were released at the relay site location 0.8 km west and a control site location 0.8 km east of the hive. From 1.8 km west, the relay effect was tested on Days 1, 3 and 6. Testing on Days 1 and 3 is pivotal because it shows how time is a factor of relay effect strength. In another trial, 1,000 bees were collected and then painted orange or yellow on the thorax. The honeybees were released at a relay site location 0.8 km west and at a control site 0.8 km east of the hive. The same procedure described above was also used in this trial, including the relay direction of 1.8 km. All statistical data, including P-values and G-values, was analyzed using StatView.



Figure 2. The average ratios of relay vs. control are shown from Days 1 to 9. As the relay vs. control ratio decreases, so does the relay effect. P=<0.001 ** P=<0.01*

The averaged ratios of the different days (Figures 2 and 3) illustrate the relay breakdown. When the ratio of relay bees to control bees was higher, the relay effect was stronger. P-values increased over time and became less significant. The p-values strongly support a decline in the number of bees returning to the hive. The honeybees experienced vanishing memory retention after Day 1. The honeybees were unable to use their sensory-motor routines to return from the second release sites (A and B).


Figure 3. The P-values and G-values are shown according to Figure 6. G tests the heterogeneity of the data. P= <0.05.

Long-term memory completely degraded somewhere between Day 6 and Day 9. Long-term memory was inhibited as the relay effect diminished. The G-values declined in this trial, showing a decreased likelihood of a strong relay effect. As the G-value declined, the observed and the expected values moved toward equilibrium.


Figure 4. Comparison of the different relay directions used in the trials. The site to the north produced a stronger relay effect than the site to the west.

The relay direction to the north shows a stronger relay effect than the relay effect to the west (Figure 4). The homing rate calculations provide evidence that, over time, the rate at which bees return to the hive significantly declines. In addition, the honeybees navigated more slowly as the days progressed. Their memory degraded during the trial, and there is an evident trend of deteriorating memory. The honeybees could no longer maintain a high homing rate for foraging and navigating (Figures 4 and 5). Therefore, it can be established that honeybee long-term memory vanishes if the learning trials are not repeated in ITIs of less than two days.


Figure 5. Homing rate calculations of a strong relay effect (the ratio of relay vs. control is significant)

As the trial proceeded from Days 1 to 9, the honeybees were progressively unable to compute the location of the flight path back to the hive. Honeybee long-term memory degraded during the trial. Because many bees did not return to the hive, it is most likely that the bees experienced impaired retention. Bees were no longer able to retrieve their previously learned memory of the novel route. The honeybees could no longer maintain a significant homing rate for foraging and navigating.



Figure 6. Homing rate calculations of a weak relay effect (the ratio of relay vs. control is insignificant)

The honeybees that were released at the beginning of each trial (Day 1 and Day 3) were more successful in navigating than the bees released on later days of each trial (Day 6 and Day 9). In order for the bees to return efficiently to the hive, they used path integration and/or landmarks. However, the bees that experienced a long-term degradation of memory were unable to remember how to navigate back to the hive. The number of bees returning each day declined, and they returned at a slower rate. Long-term memory retention is a time-consuming process because intertrial intervals play a significant role. The homing rate calculations (Figures 5 and 6) demonstrate the changing homing rate with time as the variable. A high homing rate occurs when honeybees are able to apply their previous acquired knowledge to navigate rapidly. Landmarks also help the honeybees return home more successfully (Collett and Collett 2000). A weak homing rate exists when honeybees cannot relate existing landmarks to the distance home. To forage long distances, honeybees have to have a strong memory. Based on the results of this project, their long-term memory decays rapidly over a period of nine days.


Relay breakdown in honeybees should be tested every couple of years. If relay breakdown becomes more prominent, it could be related to increased pesticide application in agriculture. The use of pesticides could be contributing to decay in memory over time in the hive. It is believed that the combination of a fungus and a virus may be the culprit in colony collapse disorder (CCD). The application of pesticides could be making the bees more vulnerable to this disease (Johnson 2010). The effects of learning, homing skills and memory should all be tested in correlation to CCD. By doing this, researchers would have a greater insight into CCD. This is important because in most regions around the world, honeybees are the main pollinators. If honeybees did not forage, most of the crops worldwide would not exist (Johnson 2010). Fifteen to 30 percent of the U.S. food supply comes from animal pollination (Greenleaf et al. 2007). The economic significance of honeybees to the U.S. economy is approximately $15 billion to $20 billion a year. Knowing more about honeybee navigational skills is vital to agriculture.

The relay breakdown effect should also be tested using different hives. Different hives in different locations may produce stronger or weaker relay effects. The placement of landmarks is an important factor in honeybee navigation. It is pertinent to retest this experiment in different areas to ensure accurate results. The time of these trials should also be extended to establish more of a trend in the data. The precise timing of long-term memory disappearance is unknown, so the exact day of the breakdown (somewhere between Day 6 and Day 9) should be tested.


I would like to thank Dr. Zachary Huang for providing me with the proper guidance and equipment in order to conduct my research. I also would like to thank my science research teacher, Serena McCalla for her encouragement and guidance on this project.


Collett, T.S., and M. Collett. "Path Integration in Insects." Current Opinion in Neurobiology 10 (1 December 2000), 757-762. Retrieved from

Greenleaf, S.S., N.M. Williams, R. Winfree, and C. Kremen. "Bee Foraging Ranges and Their Relationship to Body Size." Plant Animal Interaction 153 (2007): 589-596.

Huang, Z.Y. Unpublished data. 2009.

Johnson, R. "Honeybee Colony Collapse Disorder." Congressional Research Service. 7 January 2010. Retrieved from

Menzel, R., et al. "Bees Travel Novel Homeward Routes by Integrating Separately Acquired Vector Memories." Animal Behavior 55 (1998): 139-152.

Menzel, R. "Searching for the Memory Trace in a Mini-Brain: The Honeybee." Learning & Memory 8 (2001): 53-62.

Menzel, R., et al. "Honeybees Navigate According to a Map-like Spatial Memory." Proceedings of the National Academy of Sciences 102 (February 2005): 3040-3045.

Menzel, R., and U. Müller. "Learning and Memory in Honeybees: From Behavior to Neural Substrates."Annual Review of Neuroscience 19 (1996): 379-404.

Srinivasan, M.V. "Honeybees as a Model for Vision, Perception, and Cognition." Annual Review of Entomology 55 (2010): 267-84.

Wray, M.K., B.A. Klein, H.R. Mattila, and T.D. Seely. "Honeybees Do Not Reject Dances for 'Implausible' Locations: Reconsidering the Evidence for Cognitive Maps in Insects." Animal Behaviour76 (2008): 261-269.

American Museum of Natural History

Central Park West at 79th Street
New York, NY 10024-5192
Phone: 212-769-5100

Open daily from 10 am - 5:45 pm
except on Thanksgiving and Christmas
Maps and Directions