Effects of arena shape and exit location on the escapingbehavior of the Formosan subterranean termite, Coptotermesformosanus (Blattodea: Rhinotermitidae)

ZHANG Jian-Long, JIN Zheng-Ya, WEN Xiu-Jun, CHEN Xuan,CAI Jia-Cheng, WANG Cai,*

(1. College of Forestry and Landscape Architecture, South China Agricultural University, Guangzhou 510642, China;2. Department of Biology, Salisbury University, Salisbury, MD 21801, USA;3. Department of Mathematical Science, Salisbury University, Salisbury, MD 21801, USA)

Abstract: 【Aim】 Escaping from danger is a great challenge for group-living animals. Termites are eusocial insects that live in high densities. Therefore, they may have evolved some unique strategies to collectively escape from dangers. 【Methods】 The escaping behaviors of Coptotermes formosanus workers in arenas with different shapes (round- and square-shaped arenas without an exit) were compared under laboratory conditions, and the evacuation efficiencies of C. formosanus workers escaped from round-shaped arenas (with an exit) and square-shaped arenas (with an exit on the corner or middle of the sidewall) were investigated. 【Results】 The disturbed workers of C. formosanus rapidly moved to the edge of round- and square-shaped arenas (without an exit) and ran along the wall. However, this wall-following behavior of C. formosanus workers caused jamming on the corners of square-shaped arenas, where significantly higher density but lower moving speed of worker termites were observed compared with non-corner areas. When an exit was provided, no jamming was observed around exits because escaping C. formosanus workers were dispersed along the wall. Interestingly, the evacuation time of C. formosanus workers was similar when compared between round-shaped arenas with an exit and square-shaped arenas with an exit on the corner. However, C. formosanus workers spent significantly more time in evacuating from square-shaped arenas with an exit located in the middle of the sidewall. 【Conclusion】 These results suggest that both arena shape and exit location affect the escaping behavior and evacuation efficiency of C. formosanus workers. In addition, termites adopt unique escaping strategies to avoid the “faster-is-slower” effects usually exhibit in other group-living animals (e.g., humans, and mice). Because worker termites are blind, understanding the escaping behaviors of termites may bring insight to improving evacuation efficiency for humans under poor-visibility conditions.

Key words: Termites; Coptotermes formosanus; eusocial insects; collective behavior; escaping behavior

1 INTRODUCTION

Escaping from danger is one of the primitive instincts for animals, which have evolved diverse strategies (e.g., jumping, falling, running, and backward moving) to rapidly flee from predators and disturbances (Domenicietal., 2011a, 2011b; Card, 2012; Bulbertetal., 2015; Wilsonetal., 2020). For group-living animals, however, an effective evacuation needs the coordinated response and pedestrian of individuals. Any selfish behaviors (e.g., crowding, and backward stepping) of individuals would cause chaos and jamming, eventually decreasing the escaping efficiency at the group level (Shietal., 2018). This phenomenon, called the “faster-is-slower” effect, can be commonly observed in humans and other mammals under emergency conditions (Garcimartínetal., 2014; Zhaoetal., 2017; Shietal., 2018). However, this effect was not observed in eusocial insects such as ants, which exhibited few selfish behaviors and did not “produce a higher density zone near the exit door” (Parisietal., 2015).

Termites also are eusocial insects. Although termites usually aggregate in small and enclosed spaces (e.g., tunnels, chambers, mud tubes, and sheetings) with high densities (Graceetal., 1995), they can rapidly evacuate from the disturbed sites (Gautametal., 2012; Xiongetal., 2018a). Considering the blindness of worker termites that cannot depend on eyesight for orientation, it is assumed that they may have evolved some unique strategies to evacuate from dangers. From previous studies we observed the behavioral repertoires of escaping termites in collective level. In the field, the black-winged termite,Odontotermesformosanus, rapidly exhibited the following behavior (termites moved one by one from top to bottom) and repairing behavior (termites transported soil particles to restore the fractured tube) after the mud tube was broken (Xiongetal., 2018a). Interestingly, repairing termites move along the inner tube, away from the trunk where other termites escaped. For this reason, termites can carry out escaping and repairing simultaneously without mutual interference. Meanwhile, only a few termites exhibited selfish behaviors (e.g., suddenly stopping, backward stepping, and reverse moving) that may cause collisions and jamming. Similarly, the Formosan subterranean termite,Coptotermesformosanus, also showed a strong tendency to follow other individuals and move along the wall of Petri dishes after a disturbance was created under laboratory conditions (Wangetal., 2016).

Although these observational studies described the escaping behaviors of disturbed termites (Wangetal., 2016; Xiongetal., 2018a), the factors that may influence escaping pattern and evacuation efficiency are still unclear. Our previous studies showed thatC.formosanususually constructs sheeting by partially or entirely covering the wood with a layer of soil, and termites can enter or leave the sheeting through one or multiple shelter tubes (Xiongetal., 2018b; Jinetal., 2020). Interestingly, when the wood is partially covered, the edge of the sheeting is usually curved. However, when the wood is entirely covered, the edge of sheeting may be angled according to the shape of the wood. We hypothesized that disturbing termites would show different pedestrian patterns when they ran along the wall with or without angled corners. In the present study, we first compare the pedestrian pattern (i.e., distribution and speed) ofC.formosanusworkers in the round- and square-shaped arenas after disturbing. Since jamming can be observed in corners of the square-shaped arenas (see results), we hypothesized that square-shaped arenas may negatively affect the escaping processes of termites. Therefore, the evacuation efficiency of termites from arenas with different shapes and exit locations were investigated. Because some previous studies reported the influences of arena shape and exit location on escaping ants and mice (Shiwakotietal., 2011, 2014; Zhangetal., 2019), our results would provide valuable information to compare the escaping strategies among termites and other group-living animals.

2 MATERIALS AND METHODS

2.1 Test termites

Five colony groups ofC.formosanuswere collected from different locations on the campus and arboretum of South China Agricultural University (SCAU), Guangzhou, China. The collection location for each colony group was >0.5 km apart from each other. Plastic termite monitors (upper diameter=15.0 cm, bottom diameter=13.5 cm, height=19.5 cm) containing 6 pinewood sticks [4.5 cm×3.0 cm×16.0 cm (width×height×length)] were buried in each location. The monitors were checked regularly, and wood sticks infested by termites were brought to the laboratory and placed in the plastic boxes (60 L). Before the experiment, wood sticks were gently knocked with each other to extract termites. These termites were placed in Petri dishes (188 mm in diameter) with moistened filter paper and woodblocks and allowed to acclimate in an incubator (25±1℃) under total darkness for >24 h. All termites were tested<1 week after extraction.

2.2 Escaping behaviors of termites in round- and square-shaped arenas

This experiment aims to investigate the escaping behaviors of termites in the round- and square-shaped arenas after a disturbance was created. Methods provided by Wangetal.(2016) were modified to set the experiments. The bioassay arenas were Petri dishes (bottom side: 85 mm in diameter) or acrylic boxes (bottom side: 85 mm× 85 mm) with a layer of 2% agar on the bottom, which provided a moist and smooth surface for termites to move. Fifty worker termites ofC.formosanuswere counted and released onto the center point of each arena at the height of 3 mm. A disturbance was immediately created by lifting the arena at the height of 30 mm (measured using rules each time) and allowing free-falling to knock on the desk. A lid was put to eliminate the impacts of airflow from the outer spaces of the arena. A 10-min video was then taken to record the escaping behaviors of termites. The test for each arena shape was repeated 30 times (6 replicates×5 colony groups). For this and following experiments, the termites were tested only once. In addition, only one video was taken each time, and the order of 60 replicates was randomly assigned. All experimental procedures were conducted in a quiet laboratory (no other experiment was set up simultaneously to avoid unexpected disturbances) at room temperature (25±2℃).

In all videos, termites immediately moved to the edge of arenas after disturbing (see results). To investigate the distribution of escaping termites, the bottom of round- and square-shaped arenas was divided into the edge and non-edge areas (Fig. 1). Since termites were heterogeneously distributed on the edge area of the square-shaped arenas (see results), we divided the edge area into the corner and non-corner areas (Fig. 1). A screenshot was taken at each minute of each video. The number of termites in each area was counted. If a termite was found on the boundaries of the areas, this termite was counted for the area where the termite’s head was located. The density of termites was calculated by dividing the number of termites by area. At 1 and 5 min, five termites in each area of each video were randomly selected, and 10 frames (total length=1/3 s) of the video film were checked to determine the moving speed of each termite using Tracker 5.1.5 (Brown and Cox, 2009). If there were<5 termites in the tested area, the speed of all termites in this area was measured.

2.3 Evacuation efficiency of termites in arenas with different shapes and exit locations

This experiment aims to investigate the evacuation efficiency of termites in arenas in different shapes and exit locations after disturbing. The experiment contained three treatments:

Treatment 1 (round-shaped arena with an exit): The bioassay arena was prepared by placing a small Petri dish (bottom side: 85 mm in diameter) in the center of a bigger Petri dish (bottom side: 188 mm in diameter). A round hole (diameter=5 mm) was made on the wall of the small Petri dish (the center point of the hole was 5.5 mm apart from the bottom of the small Petri dish) (Fig. 2: A). Agar (2%) was added to the bottom of two Petri dishes to the depth of 5.5 mm in the small Petri dish, and therefore the inner space of the two Petri dishes was connected through the semicircular hole (Fig. 2: D).

Treatment 2 (square-shaped arena with a corner exit): The bioassay arena was prepared by placing an acrylic box (bottom side: 85 mm×85 mm) in the center of a Petri dish (188 mm in diameter). A round hole (diameter=5 mm) was made on the wall near the corner of the box (the center point of the hole was 5.5 mm apart from the bottom and edge of the box)(Fig. 2: B). Agar (2%) was added to reach the depth of 5.5 mm in the box, and the inner space of the box and Petri dish was connected through the semicircular hole (Fig. 2: E).

Treatment 3 (square-shaped arena with a middle exit): The bioassay arena was prepared as described in treatment 2, but the hole was made in the middle of one of the four sidewalls of the box (the center point of the hole was 5.5 mm apart from the bottom) (Fig. 2: C and F).

Before the experiment, 50 worker termites ofC.formosanuswere counted and released into the center point of the inner container (round-shaped Petri dish or square-shaped box). A disturbance was created as described above. A lid was put on each arena, and a 60-min video was taken. Each treatment was repeated 15 times (5 replicates×3 colony groups). Only one video was taken each time, and the order of 45 replicates was randomly assigned. All experiments were conducted in a quiet laboratory under constant environmental conditions (25±2℃). A screenshot of the video was analyzed at each minute, and the number of termites in the inner container was counted. The duration for 50%, 70% and 90% of termites evacuated from the inner container was recorded. Because a few termites escaped from the arena may return (see results), we only recorded the first time that the percentage of evacuated termites reached 50%, 70% and 90%. If evacuated termites did not reach 50%, 70% and 90% at the end of the experiment, the evacuation time was given the value of 60 min. To investigate the distribution of termites, the arena was divided into the edge and non-edge areas as described above. Also, a square area (11 mm× 11 mm) near the exit was defined as the exit area (Fig. 2).

2.4 Data analyses

The normality of data was examined using Shapiro-Wilk test (SAS 9.4, SAS Institute, Cary, NC). Since data obtained from all experiments are not normally distributed, ln-transformation was carried out to meet the assumption of normality. For the experiment to compare the escaping behaviors of termites in round- and square-shaped arenas, the density and moving speed of termites distributed in each area were compared using three-way repeated measures analysis of variance (ANOVA) with time as the within-subject effect, and termite colony (treated as the random effect) and location (treated as the fixed effect) as the between-subject effects (SAS 9.4). Statistical results showed that the effects of time and location and their interactive effects on density and moving speed of termites were significant (Table 1). Thus, the densities and moving speed of termite in different areas of the round- and square-shaped arenas were compared at each time point. For the experiment to compare the evacuation efficiency of termites in arenas with different shapes and exit locations, the time for 50%, 70% and 90% of termites to evacuate was compared using the two-way ANOVA with colonies as the random effect and treatment as the fixed effect. Tukey’s honest significant difference (HSD) test was used forpost-hoccomparisons after each ANOVA. In all tests, the significance levels were determined at α=0.05.

Table 1 Statistical results of the experiment to compare escaping behaviors of Coptotermesformosanus workers in round- and square-shaped arenas

The density and moving speed of termites distributed in each location were compared using three-way repeated measures analysis of variance (ANOVA) with time as the within-subject effect, and termite colony (treated as the random effect) and location (treated as the fixed effect) as the between-subject effect.

3 RESULTS

3.1 Escaping behavior of termites in round- and square-shaped arenas

C.formosanusworkers immediately ran along the wall of round-shaped and square-shaped arenas after disturbing (Figs. 3 and 4). In the round-shaped arenas, escaping flows (i.e., some termites followed one by one and moved toward the same direction) rapidly formed along the wall (Fig. 3). In general,C.formosanusworkers had significantly higher density on the edge area of the round-shaped arena than non-edge areas throughout 10-min videos (Fig. 5: A). Also,C.formosanusworkers on the edge area moved significantly faster than termites on non-edge areas (Fig. 6: A and B).

In square-shaped arenas,C.formosanusworkers also had significantly higher density on the edge areas (except corners) than non-edge areas at each time point (Fig. 5: B). However, jamming was observed on the corners (Fig. 4), where worker termites had significantly higher density than other locations (Fig. 5: B). In addition,C.formosanusworkers moved significantly faster on the edge areas (except corners) than on the other locations (Fig. 6: C and D). The moving speed ofC.formosanusworkers were significantly faster on the non-edge areas than on corners at 1 min after disturbing (Fig. 6: C), but they were not significantly different at 5 min (Fig. 6: D).

3.2 Evacuation efficiency of termites in arenas with different shapes and exit locations

Similar to the observations mentioned above, the disturbed workers ofC.formosanusrapidly moved to the edge of the round- and square-shaped arenas and ran along the wall (Fig. 7). Also, jamming was observed on the corners of square-shaped arenas (Fig. 7). However, no jamming was observed in exit area of the three types of arenas (Fig. 7). WhenC.formosanusworkers moved to the exit of round-shaped arenas, some termites passed the exit and evacuated, while some termites still moved along the wall (Fig. 8: A). Likewise, effective evacuation behaviors can be observed near the corner exits of the square-shaped arenas, where groups ofC.formosanusworkers moved along the wall and directly passed the exits (Fig. 8: B). However, escaping worker termites tended to ignore the middle exit when they moved along the sidewall of the square-shaped arenas, and were more likely to pass the exit when they were around the exit and moved angled to the sidewall (Fig. 8: C). In general,C.formosanusworkers showed high evacuation speed at the initial stage of escaping (Fig. 9). However, the escaping speed slowed down after most worker termites evacuated from the arena, regardless of arena shapes and exit locations (Fig. 9). Similar time was required for 50%, 70% and 90% of worker termites to evacuate from round-shaped arenas and square-shaped arenas with the corner exit (Fig. 10). However, significantly more time was required for worker termites to escape from square-shaped arenas with the middle exit (Fig. 10).

4 DISCUSSION

Our study showed that: (1)C.formosanusworkers run along with physical guidelines (e.g., wall of arenas) during escaping and evacuating, but jams can be observed in corners of square-shaped arenas (Figs. 3-6); (2)C.formosanusworkers evacuate fast from the square- and round-shaped arenas with an exit, and few jams were observed around the exits (Figs. 7 and 8); and (3) exits located on the corner of square-shaped arenas significantly increased the evacuation efficiency than exits located on the middle of the sidewall (Figs. 9 and 10).

The wall-following behaviors have been widely observed in ants and termites under normal conditions. For example, the black garden ants,Lasiusniger, usually prefer to move along the arena wall instead of walking on the shortest path directly connecting the nest and food source (Dussutouretal., 2005). Also, the subterranean termitesReticulitermesflavipes,R.virginicus, andR.chinensisshowed a strong tendency to move along the inner wall of arenas or enter pre-formed tunnels (Pitts-Singer and Forschler, 2000; Hassanetal., 2021). Interestingly, Hassanetal.(2021) reported that different casts ofR.chinensisspend significantly more time in moving along the wall of Petri dishes than inner zones, but knockdown of phosphofructokinase gene significantly decreased the duration of termites moved in the wall zone. Our previous and present studies showed that worker and soldierC.formosanusalso tended to escape along the wall under panicked conditions (Wangetal., 2016). Such behavior may have some adaptive value for escaping termites. First, worker termites are blind, and the wall of arenas can provide physical guidelines for termites, thus reducing random movements during evacuation. Also, walls may prevent termites from running outside the biostructures (e.g., shelter tubes, sheetings, chambers and tunnels) and exposing to the unexpected environments and predators.

In addition, the present study showed that escapingC.formosanusworkers dispersed on the edge area when ran along the wall, and therefore prevent jamming and clogging near the exit. We believe that this can explain the high evacuation efficiency of termites. Interestingly, previous studies showed that ants adopt different strategies from termites to avoid jamming near the exit. For example, Parisietal. (2015) reported that groups of panicked carpenter ants,Camponotusmus, distributed uniformly in the whole arena, and therefore the density of ants remained low near the door (exit). Chung and Lin (2017) reported that when escaping from a heated arena, the black spiny weaver ants,Polyrhachisdives, actively clustered near the exit to avoid gathering and jamming. Compared with termites and ants, other group-living animals (especially mammals) usually exhibited low evacuation efficiency due to their selfish behaviors resulting in blocks near exits. For example, many previous studies showed large crowds of mice, sheep, and humans near the exit, which caused jamming and clogging and greatly decreased the evacuation efficiency (Garcimartínetal., 2014; Linetal., 2016; Zurigueletal., 2016; Wangetal., 2018; Zhangetal., 2018, 2019).

For ants, the wall-following behavior may depend on visual, chemical and/or mechanical cues (Dussutouretal., 2015). Because worker termites are blind, visual cues may not play a role in the wall-following. Also, in the present study,C.formosanusworkers rapidly moved to the edge area after disturbing (Figs. 3 and 4), and therefore they may not have chances to mark the edge area with trail pheromone. Thus, the chemical cues may not be a reason for the wall-following behavior of termites at the initial stage of escaping. Over some time, however, termites moving along the wall may drop enough trail pheromone on the path, which may strengthen and amplify the following behavior. It would be valuable to verify the effect of chemical cues in termite escaping. We believe that mechanical cues play a significant part in termite escaping. When collisions (mechanical cues) occur between termites and walls, running termites can adjust their moving directions parallel or tangent to the wall. In addition, our previous study showed that when termites moving from different directions crashed, they showed a strong tendency to change their original pedestrian pattern and follow other termites (Cai and Wang, unpublished data). The tangoreceptor located on the antenna and/or body of termites may provide the cues during colliding or crashing so that some termites can adjust their moving directions and follow other ones. Once the unidirectional escaping flow (latter termites escaped following the former termites) formed, few termites exhibited step backing, reverse moving, or stopping behaviors that may cause collisions within the flow (Wangetal., 2016; Xiongetal., 2018a). Therefore, the escaping flow can be observed in this and our previous studies (Wangetal., 2016; Xiongetal., 2018a). However, full-frontal crashes between running termites and vertical wall occurred when termites moved to the corners of square-shaped arenas. The crashed termites further collided with following termites, and therefore jamming can be formed.

Although square-shaped arenas can negatively affect the collective movement ofC.formosanusworkers, the exit located on corners effectively decreased the density of jamming termites and therefore caused high evacuation efficiency(Figs. 7 and 8). In contrast, termites had the lowest evacuation efficiency when escaping through the middle exits on the sidewall. In this case, the moving direction of termites is entirely parallel to the sidewall. Therefore, they may ignore the exit due to the lack of mechanical cues (collision between antennae and/or body of termites and the sidewall). Interestingly, we observed that some escaped termites returned to the inner arena. This result corroborates Gautametal.(2012) that escaped termites would not abandon their habitat where a disturbance occurred.

Studies on the evacuation process of other animals may reveal different results from termites. For example, Zhangetal.(2019) reported that under low-stress conditions, mice showed a higher evacuation rate when the exit (width=1, 1.5 and 2 cm) was located on the middle of the sidewall than the corner exit (Zhangetal., 2019). However, the evacuation rates of mice were not significantly different between two exit locations under high-stress conditions (Zhangetal., 2019). In addition, the evacuation rate of mice usually remained relatively constant during the whole evacuating process (Shiwakotietal., 2014; Zhangetal., 2019). However, our study showed thatC.formosanusworkers evacuated faster at the initial stage of escaping, and the evacuation rate slowed down after the majority of individuals successfully escaped (Fig. 9). The different results between termites and mice may be due to their different escaping strategies. When mice crowed near the exit, the number of escaped individuals may remain steady throughout the process. However, when termites running along the wall successfully evacuated, there may be small portions of termites still staying on non-edge areas, which moved slowly and may not easily find their way out.

In our study, 60 and 45 replicates were tested in randomly assigned orders for the two experiments. In addition, multiple colony groups of termites were tested and treated as the random effect. Therefore, we believe the sample size is sufficient to draw a valid conclusion for each experiment. However, there might be some differences in escaping behavior of termites between laboratory and natural conditions. Firstly, termites usually stay in dark and enclosed spaces inside tunnels and chambers, so it is difficult for us to observe their natural behaviors. In our study,C.formosanusworkers were allowed to escape in transparent arenas under the light. It would be valuable to investigate whether the light exposure affects the escaping behaviors of subterranean termites. Secondly, a matureC.formosanuscolony may contain millions of individuals (Graceetal., 1995; Henderson, 2008; Rust and Su, 2012), and therefore termites usually live in extremely high densities. In this study, only one density of termites (50 termites/arena) was tested. We plan to compare the escaping behaviors and evacuation efficiency of termites with different densities in future studies.

In conclusion, our study showed that disturbed workers ofC.formosanustended to move along the wall of arenas, which provides guidelines for escaping termites and reducing random movement. The wall-following behavior also dispersed termites and avoided a high density of termites accumulating near the exit that may cause jamming. In addition, the location of exits exerted a significant effect on termite evacuation. Because worker termites are blind, studying termites may provide some valuable strategies to improve evacuation efficiency of humans under poor-visibility conditions. For example, people can move along the wall or other physical cues and try to follow other people if they escape from rooms filled with smoke or under dark conditions. Also, opening the door near the corner may reduce jams and increase the evacuation speed.