Termites can learn

It is generally believed that termites can ’ t learn and are not “ intelligent ” . This study aimed to test whether termites could have any form of memory. A Y-shaped test device with one release chamber and two identical test chambers was designed and constructed by 3D printing. A colony of damp wood termites was harvested from the wild. Worker termites were randomly selected for experiment. Repellent odors that could mimic the alarm pheromone for termites were first identified. Among all substances tested, a tea tree oil and lemon juice were found to contain repellent odors for the tested termites, as they significantly reduced the time that termites spent in the chamber treated with these substances. As control, a trail pheromone was found to be attractive. Subse-quently, a second cohort of termites were operant conditioned by punishment using both tea tree oil and lemon juice, and then tested for their ability to remember the path that could lead to the repellant odors. The test device was thoroughly cleaned between trials. It was found that conditioned termites displayed a reduced tendency to choose the path that led to expectant punishment as compared with naïve termites. Thus, it is concluded that damp wood termites are capable of learning and forming “ fear memory ” , indicative of “ intelligence ” in termites. This result challenges established presumption about termites ’ intelligence.


Introduction
As eusocial creatures, termites have long fascinated many naturalists, entomologists, biologists, and computer scientists, despite their reputation for destruction of wooden structures.It is generally believed that individual termites are not intelligent and lack the ability to learn (Srinivasan, 2018).Yet, the mounds that termites construct as a colony are among the largest and most sophisticated structures built by any nonhuman animal (Fagundes et al., 2021;King et al., 2015).Those marvelous creations, reaching as high as ten meters above ground, provide a microclimate of controlled heat, respiratory gas exchange and humidity for a colony of up to millions of termites dwelling beneath.Termite's remarkable proficiency in collaborative construction has captivated many scientists as a paradigm of "swarm intelligence" whereby complex and sophisticated behaviors emerge from the interactions of seemingly simple individuals (Krause et al., 2010;O'Bryan et al., 2020).Notable studies have been reported exploring termite-inspired analogous principles in artificial systems, particularly in the field of robotics and artificial intelligence (Carey et al., 2021;Werfel et al., 2014).
Despite the hype and hope, the fundamental presumption about individual termite's learning ability has not been tested experimentally.
Operant conditioning represents a learning mechanism whereby behavioral patterns undergo alteration through the linkage of stimuli with reinforcement or penalty (Brembs, 2003).This method finds frequent application in animal training endeavors.Termites, possessing an innate ability to detect alarm pheromones, exhibit a heightened sensitivity to potential threats within their ecological domain (Mitaka and Akino, 2021).Leveraging this trait, one may use aversive olfactory cues as punitive agents for conditioning termites and assessing their ability of associative learning.
In this study, we aim to empirically evaluate the aptitude of individual termites to acquire avoidance behaviors in response to anticipated punitive stimuli.Using a new-constructed Y-maze, we found that termites possessed the ability to recollect pathways linked with repellent olfactory cues.This study demonstrates individual termite's proficiency in learning and challenges the previously established assumption regarding individual termite's "intelligence".

Termite collection and rearing
Damp wood termites, Zootermopsis angusticollis, were harvested from the wild located at latitude 42.301170, longitude − 71.160096.The entire termite-living wood was collected for termite rearing and stored in a dark and humid plastic box.The temperature (approximately 25 • C) and humidity (around 60%) within the rearing box were consistently maintained and monitored at least weekly.The animals underwent a period of acclimatization to the new rearing environment, lasting for a minimum of one week before the commencement of experimentation.Only worker termites were used in this study.

Y-maze
Adapted from a previous study (Yanagawa et al., 2012), a Y-maze chamber was designed using a free 3D design software (www.tinkercad.com) and 3D-printed (PowerSpec 3D Pro Printer).The configuration of the Y-maze embodied a dual-choice paradigm, featuring three distinct compartments (Fig. 1).The lowermost compartment, denoted as "A," functioned as a release/departure point for termites.At the upper reaches of the Y-shaped structure, two compartments, labeled "B" and "C," were situated at the diverging arms of the letter "Y."When a two-compartment system was required, compartment A was blocked off using a plastic stick and only compartments "B" and "C" were accessible for termites.Each of the compartments harbored a circular chamber, interconnected by slender corridors, facilitating unidirectional movement of the test subjects.The base of the apparatus was furnished with a disposable, pristine paper surface, ensuring cleanliness and facilitating easy replacements between trials.A 15-cm petri dish was used to house the apparatus and testing animals.Video recording was performed to document and subsequent analysis of termite behaviors.

Repellent and attractive odors
Substances tested for repellency included tea tree oil (Handcraft Blends essential oil), fresh-made lemon juice and 5% acidity distilled white vinegar (Heinz).They were selected based on a search on Google using the term "substances that termites dislike".4,6-dimethyldodecanal, previously identified as a trail pheromone for damp wood termites (Mitaka and Akino, 2021), was synthesized by WuXi AppTec and prepared as 1 μg/μl in water.Before each trial, the testing apparatus was cleansed to minimize odor contamination.

Repellency testing
The testing termite was introduced into the release chamber A, which was sealed after the animal transitioned into compartment B or C. To quantify the allocation of time spent by the termites in compartments B and C, observations were conducted over a defined 3-minute interval after termites reached B or C. The time the termite spent in each of the two compartments was recorded after video analysis.10 μl of each testing substance was applied at the center of the chamber B. The paper bottom was replaced after each trial to avoid contamination of substances.10 randomly selected termites were tested for each substance and control.

Navigation choice and operant conditioning
To test the navigation choice, a termite was released into the chamber A and gently guided using a brush into the existing corridor.The percentage time that the termite went to B or C was recorded over 10 trials for each termite.More trials were not performed as the animal appeared to show "fatigue" when over-tested.20 randomly selected naive termites were tested as control.
To condition the termite, 10 μl of tea tree oil was applied at the center of the chamber B (conditioning chamber).When the terminal entered chamber B, a lemon juice-soaked cotton swab was used to expel it away from chamber B. When the termite was driven into the chamber C, one round of conditioning was considered complete.If the termite turned into chamber C directly from A, it was not considered a valid conditioning round.After each round, the termite was relocated into chamber A, and the conditioning started over again.10 conditioning rounds added up to one "conditioning set." 3 such sets were applied to each termite with a 5 minute "break" in a dark rest box after each conditioning set.After all 3 conditioning sets were done, the termite was immediately subjected to the navigation choice test as described above.20 randomly selected termites were conditioned and tested for navigation choice.

Statistics
Statistical analyses were conducted using the GraphPad Prism software suite.For comparisons involving three or more groups, a one-way ANOVA followed by Dunnett's post hoc test was employed.Two-group comparisons were performed using the t-test.The presented data are expressed as means accompanied by standard deviation of the mean (SEM).The difference was considered statistically significant if the P- Y.E.Ding and Z. Li value was less than 0.05.

Tea tree oil and lemon juice were repellent for termites
Th two symmetrical compartments B and C were used for testing repellency of selected substances with the compartment A blocked off (Fig. 2a).Emanation of tea tree oil, lemon juice or vinegar was introduced to the chamber B by adding 10 μl of the substance at its center.As shown in Fig. 2b, termites spent roughly equal time exploring both chambers B and C over a 3-min observation period, when nothing was added to the chamber B. The average percentage time termites spent in chamber B (i.e., conditioning chamber) was 51.1 ± 3.0% (N = 10).Significant reduction in percentage time in chamber B was observed when tea tree oil or lemon juice was present, indicating termite's aversion to these two substances (37.0 ± 3.3% with tea tree oil treatment and 31.9 ± 4.3% with lemon juice treatment; N = 10 each; P < 0.05).A video showing movement of a termite with tea tree oil treatment is attached as Supplementary Video 1.No discernible statistical distinction was observed between the control condition and vinegar treatment (45.1 ± 3.3%; N = 10; P = 0.551).As a positive control, the reported trail pheromone for damp wood termites, 4,6-dimethyldodecanal, was also tested.As expected, presence of this trail pheromone (1 μg/ μl, 10 μl) significantly increased the percentage time termites spent in the chamber B (81.6 ± 3.3%; N = 10; P < 0.0001).
Supplementary material related to this article can be found online at doi:10.1016/j.beproc.2024.105012.

Learning by termites after operant conditioning
To investigate termite's potential capability of developing "fear  memory", a navigation choice test was employed (Fig. 3a).After being released into chamber A and gently guided into the existing corridor, the termite was afforded a "free" choice going to the chamber B or C. 10 such trials were performed for each termite and % probability of the termite going to the conditioning chamber B was calculated.For naïve, non-conditioned termites, the probability of them choosing the chamber B was found to be 50.5 ± 2.1% (N = 20) (Fig. 3b; Supplementary Video 2).
Supplementary material related to this article can be found online at doi:10.1016/j.beproc.2024.105012.
For operant conditioning, a punitive training regimen was executed for a second cohort of termites.Termites traversing from chamber A to B were subjected to dissuasive odor stimuli with combined treatment of tea tree oil and lemon juice until they redirected their path to chamber C (Supplementary Video 3).A series of 30 conditioning rounds were administered for each animal.Subsequently, the conditioned termite was subjected to the navigation choice test without presence of any odor stimuli.Impressively, a marked reduction was observed in the inclination of conditioned termites to opt for chamber B, where the punitive stimuli had been applied (Fig. 3b).Thus, conditioned termites exhibited a diminished propensity to select the route associated with anticipated adverse consequences.
Supplementary material related to this article can be found online at doi:10.1016/j.beproc.2024.105012.

Discussion
The biological mechanism underlying termite's "swarm intelligence" is still unknown.The current dominant theory explaining termite's collective intelligence is "stigmergy", a term combining the Greek stigma (mark) and ergon (action), which proposes that individual termites passively respond to traces and cues left by others, creating a positivefeedback loop (Oberst et al., 2020;Srinivasan, 2018).However, such theory has been inadequate in either explaining all termite's mastership or providing a framework for robots to mimic their biological models.The result from this study casts doubt on the long-established presumption regarding lack of learning abilities in individual termites.Adaptations and/or modifications of current framework for termite-inspired "swarm intelligence" may be warranted to incorporate the role of learning.
Our finding on learning in termites may be justifiable from the biology point of view.Acquisition of learning capabilities would hold practical implications for termites, enhancing their odds of survival by being sensitized to alarm cues.Any learning ability would also align with the communal nature of termite colonies, where synchronized efforts are requisite for accomplishing communal tasks.It is worth noticing that learning has been well documented in insects including flies, grasshoppers, parasitoid wasps, bees, and ants (Dawson et al., 2013;Dong et al., 2023;Dukas, 2008;Giurfa, 2015;Knaden, 2019).This study is among the first to demonstrate learning in termites.
The limitations of this study include: 1) The data collected from conditioned termites display some deviation, manifesting as instances where certain individuals avoid the conditioned chamber entirely, while others surpass the frequency of entries into the unconditioned chamber.Although statistical distinction is discernible between the conditioned and "naive" termites concerning their navigational preferences, future study may be better designed to examine differences of the same termites before and after conditioning.The seeming "fatigue" in termites after repeated testing had prevented us doing that in this study.It is also intriguing to know how dissociation of individual termites from others affect their responses to repellent stimuli, since termites are social insects.2) Only one single termite variant was examined.While this study serves as an initial step, extending similar investigations to encompass diverse termite types and species remains an avenue for future inquiry.
3) Neuroanatomy and neurophysiology of the termite nervous system were not studied.As learning is the acquisition of neuronal representation of new information, delving into the intricacies of termite nervous systems and elucidating the biological underpinnings of learning in these organisms represents another important direction for future research.Indeed, neural plasticity has been documented in termites during differentiation and aging (Ishibashi et al., 2023;Valadares et al., 2023).
In conclusion, this study has demonstrated learning in termites, in contrast to established presumption.Our finding has scientific meanings as it unveils new understanding in termites; it is also of practical significance in offering a potential avenue for better designing "swarm intelligence".

Funding
No funding was received for conducting this study.

Declaration of Competing Interest
The authors have no relevant financial or non-financial interests to disclose

Fig. 1 .
Fig. 1.Experimental steps and a Y-maze for testing termite behavior.

Fig. 2 .
Fig. 2. Repellency of selected substances for termites.a) Illustration of a two-compartment testing system.Asterisk indicates termite.b) Percentage time that termites spent in the chamber B (conditioning chamber) with nothing, tea tree oil, lemon juice, vinegar and a trail pheromone added.Termites spent significantly less time in the chamber with tea tree oil or lemon juice and more time in the chamber with trail pheromone.No statistically significant difference was observed between vinegar treatment and control.Mean ± SEM (N = 10 each), One-way ANOVA with Dunnett's post hoc test.*P < 0.05, **P < 0.01, ****P < 0.0001.

Fig. 3 .
Fig. 3. Learning by termites after punitive conditioning using repellent odors.a) Illustration of how navigation choice essay was performed.Asterisk indicates termite.b) Probability of termites going from chamber A to chamber B (conditioning chamber).During conditioning, termites were "punished" and driven out when entering from chamber A to B. While naive termites have similar probability of going to B or C, conditioned termites show decreased probability of going to the chamber where they were previously punished.Mean ± SEM (N = 20 each), t-test.**P < 0.01.