Detection and Control of Subterranean Termites Using In-Ground Termite Monitoring Stations and Chitin Synthesis Inhibitor Baiting for Singapore Urban Trees

Study Sites

In this study, a total of 145 trees from 14 species were assessed at 15 different study sites in Singapore from June 2023 to July 2024 (Table 1). Selected locations either had a recent history of or ongoing termite infestation based on visual assessment of signs and symptoms of termite infestation. These sites were broadly categorized into two main groups based on soil type: coastal areas (sandy and saline soil) and inland areas (loamy and nonsaline soil).

Inland Areas

Gardens by the Bay. At sites A, B, and C (Figure 1A), Phoenix sylvestris (L.) Roxb., Enterolobium cyclocarpum (Jacq.) Griseb., and Syzygium aromaticum (L.) Merr. & L. M. Perry trees were reported to have had termite infestations with aboveground symptoms in the past. Termite control by chemical treatment was applied about 6 months before the installation of IG termite monitoring stations. There were no aboveground signs and symptoms of termite infestations upon installation of the IG stations for sites A through C. At site D (Figure 1A), Araucaria columnaris (G. Forst.) Hook in this area had termite mud trails spotted during tree inspection in May 2023 with 2 trees showing extensive damage to central wood column. Due to proximity to a nearby freshwater lake, chemical application was not feasible, leading to removal of the 2 heavily damaged A. columnaris.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 1.

(A) Maps of different sites in which IG stations (depicted as orange circles) were installed. Sites A through D, Site E, and Site F are located in Gardens by the Bay, Hort Park, and Singapore Botanic Garden, respectively. Activated stations are depicted as black circles with an outer glow. For Site F, T1 and T2 stations were first activated by C. curvignathus before being taken over by Schedorhinotermes medioobscurus after the colonies were eliminated, while T3 station was activated by S. medioobscurus. (B) Maps of different sites in which IG stations (depicted as orange circles) were installed. Sites G, H, I, and J through L are located in Singapore Botanic Garden, Bidadari Park, East Coast Park Service Road, and East Coast Park, respectively. Activated stations are depicted as black circles with an outer glow. (C) Maps of different sites in which IG stations (depicted as orange circles) were installed. Sites M through N and O are located in Changi Beach Park and Changi Coast Road, respectively. Activated stations are depicted as black circles with an outer glow.

Hort Park. Site E contained a standalone Ceiba pentandra (L.) Gaertn. (Figure 1A). In 2022 termite infestation was detected with aboveground symptoms. Chemical application was carried out, and no termite activity or aboveground symptoms were observed after the chemical treatment.

Singapore Botanic Garden. Site F consisted of 8 trees: 5 Araucaria cunninghamii W. T. Aiton ex D. Don. individuals and 3 Agathis borneensis Warb. individuals. Chemical application via spraying was applied to all trees one month before the installation of IG termite monitoring stations (Figure 1A). Site G was located near a carpark with a total of 17 A. cunninghamii trees showing neither signs nor symptoms of termite infestation (Figure 1B). This site also had no past history of infestation.

Bidadari Park. This park (Figure 1B) was developed with the concept of rustic aesthetic design with several dead wood logs and trees as display pieces. Living trees at Site H showed active termite activities with several mud trails spotted during visual inspection. No chemical treatment had been applied for termite control in this location.

East Coast Park Service Road. Despite the proximity of this location to the sea, the soil was nonsaline and loamy, so the location was categorized as inland. The 5 Samanea saman (Jacq.) Merr were planted as a single row along the road while the 10 Caryota mitis Lour. individuals existed as a cluster within a forested patch opposite the S. saman trees (Figure 1B). No chemical treatment had been applied for termite control in this location.

Coastal Areas

East Coast Park. Sites J and K within this location (Figure 1B) were subjected to intermittent flooding due to sustained heavy rain or rising water tables leading to waterlogged conditions. All trees within the 3 sites showed visible signs and symptoms of termite infestation. No chemical treatment had been applied for termite control in this location.

Changi Beach Park. Casuarina equisetifolia L. was the main tree species planted in this park (Figure 1C). Casuarina equisetifolia trees in both Sites M and N showed visible signs and symptoms of termite infestations. No chemical treatment had been applied for termite control in this location.

Changi Coast Road. Casuarina equisetifolia and Xanthostemon chrysanthus (F.Muell.) Benth. trees were located as a single row along the main road (Figure 1C). They showed visible signs and symptoms of termite infestation. No chemical treatment had been applied for termite control in this location.

IG Termite Monitoring Station

In this study, we used IG stations and CSI baits produced by Exterminex (Agro Technic Ltd Pte, Singapore). A piece of preweighted lure wood (10.5 cm × 4.5 cm × 2.0 cm) made from pine was placed at the center of the station. The lure wood within the station served as a nontoxic food source for termite consumption when discovered by foraging termites and also served to evaluate the termite consumption rate and feeding behavior (Su and Scheffrahn 1986). For each tree, 3 stations were installed within intervals of 3 m at 3 directions (North, Southwest, and Southeast) as demonstrated in Figure 2. After installation, the stations were monitored weekly for termite activity. When the lure pine wood within the station was attacked by termites, stations were said to be activated. Lure wood pieces were then replaced with a new piece. The old wood piece was collected for weighing to obtain the baseline termite consumption rate for 2 to 4 weeks prior to set up of auxiliary monitoring stations and baiting. The described step aimed to ensure that the termites were continuously foraging within the same station and to ascertain that there was no feeding aversion to the food source given. After installation of the auxiliary station and addition of CSI bait to the auxiliary station, the consumption rate on the lure wood piece of the primary station was tracked weekly until 4 weeks after no more termite activity or presence were observed in both primary and auxiliary stations. To obtain lure wood consumption rate by detected termites, collected lure wood from activated primary stations were washed, cleaned, and oven-dried for 48 hours according to protocol of Umar and Ab Majid (2020b). After 48 hours, lure woods were weighed and compared to their original packaging weight.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 2.

In-ground (IG) station placement for individual trees. Top-down diagram of IG stations (Left). Side-view photo of actual IG stations installed for one tree (Right).

Auxiliary Monitoring Station and Bait Consumption

An auxiliary monitoring station was installed next to the primary monitoring station when termites were detected actively feeding within a primary monitoring station. The purpose of the auxiliary station was to deliver CSI to detected subterranean termites and provide observational data on termite presence. To quickly activate the newly installed auxiliary station, foraging termites from the primary station were transferred together into the auxiliary station. After successful establishment of termite activity, 100 g of cellulose-powdered chlorfluazuron bait was then moistened with 250 mL of distilled water to substitute the lure wood within the activated auxiliary station. The bait was monitored weekly and replenished when more than 50% of the original CSI bait had been consumed. The total weight of CSI bait consumption was determined when termite eradication, defined as the disappearance of termites from both primary and auxiliary stations, had been achieved.

Termite Species Identification

Activation of IG Stations and Detected Termite Species

Table 2 summarizes the termite species detected based on visual inspection and the actual termite species detected by IG stations as well as the associated activation time for all 145 trees in 15 sites (A through O) across all locations. For all sites in GB and HP, activation of IG stations by C. gestroi were achieved without apparent aboveground symptoms. Except for site D with trees displaying mud trails on trunks from active termite workers and soldiers found within the trails, the rest of the sites in these two locations did not have visible aboveground symptoms indicative of termite infestations. Activation time ranged from 1 week (site D) to 24 weeks (site G).

In Singapore Botanic Garden, IG stations were activated by C. curvignathus, S. medioobscurus, Macrotermes gilvus, and Macrotermes carbonarius. It is important to note that the determination of C. curvignathus infestation in an A. cunninghamii tree at Site F was only obtained when one of the trees was felled, and actual workers and soldiers were collected from inside the felled tree trunk. Schedorhinotermes medioobscurus, M. gilvus, and M. carbonarius activities were not observable aboveground, with no indication of mud trails, mounds, or actual workers and soldiers foraging on the ground. For stations at trees T1 and T2 (Figure 1A, Site F), C. curvignathus was the original species activating the stations. After C. curvignathus had been eliminated through CSI baiting, S. medioobscurus was found feeding on the lure wood in the same stations after 2 to 3 weeks. Furthermore, IG stations activated by M. carbonarius and M. gilvus showed sporadic patterns: different stations were activated at different times, and termite activities within activated stations were not continuously observed for more than one week. In general, activation time ranged from 1 to 3 weeks for C. curvignathus, 2 to 3 weeks for S. medioobscurus, 5 to 13 weeks for M. gilvus, and 24 weeks for M. carbonarius.

Nasutitermes havilandi was the predominant species found foraging in Site H of Bidadari Park and Site K of East Coast Park with extensive mud trails on tree trunks and branches, but none of the installed stations were activated by this termite species. Interestingly, at Site K, one of the IG stations was activated by C. gestroi from an unknown location. However, due to the intermittent flooding of this site, C. gestroi was observed to abandon the station after the flood subsided. The same station was re-activated by C. gestroi after one year but was abandoned again after another flood.

Macrotermes carbonarius formed big mounds at the base of S. saman trees at Site I and activated IG stations installed in this area within 3 weeks after installation. Meanwhile, Termes rostratus were the predominant species found in the C. mitis palm cluster but did not activate any of the installed IG stations. Similarly, Microcerotermes crassus were the species commonly observed in Sites I, L, and O but did not activate most of the installed IG stations except for one station in Site O. This station was installed right next to the M. crassus mound found on the ground. After the initial activation, the lure wood was retrieved for lab measurement and replaced with a new piece. The new lure wood piece was never activated again by the M. crassus for the rest of the study.

Although C. gestroi showed consistent activation of IG stations for most of the inland sites, C. gestroi activities could be observed from mud trails on trunks of C. equisetifolia and X. chrysanthus but did not activate most of the IG stations in coastal sites (J through O), except for one station in Site K, which was described formerly, and another station in Site J. Site J activation only happened after 14 weeks from installation and approximately 1 to 2 weeks after 4 trees in this area were felled.

Lure Wood Consumption Rate and Feeding Pattern

As seen from Figure 3, M. carbonarius and M. crassus had very low average weekly wood consumption rates with only 0.4 g/week and 0.5 g/week, respectively. Macrotermes gilvus consumed 9.3 g/week. Coptotermes curvignathus, C. gestroi, and S. medioobscurus consumed the most with 22.1 g/week, 19.7 g/week, and 16.8 g/week, respectively.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 3.

Box-plot of average weekly wood consumption rates by different termite species.

Figure 4 shows feeding patterns by C. gestroi, C. curvignathus, M. carbonarius, M. gilvus, M. crassus, and S. medioobscurus in lure wood. Coptotermes gestroi and C. curvignathus fed by creating clean and sharp carvings with extensive galleries that cut through lure wood piece. Meanwhile, M. carbonarius and M. gilvus displayed scraping patterns of feeding. These species did not create galleries that cut through lure wood pieces. As compared to M. carbonarius, M. gilvus appeared to scrape more aggressively and left rough texture on lure wood surfaces. Microcerotermes crassus also displayed scraping feeding patterns similar to M. gilvus and M. carbonarius, but to a smaller extent. Schedorhinotermes medioobscurus displayed a mixed feeding pattern that included a combination of carving and scraping.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 4.

Feeding patterns of subterranean termites on lure wood pieces. (A) C. curvignathus. (B) C. gestroi. (C) M. crassus. (D) M. carbonarious. (E) M. gilvus. (F) S. medioobscurus.

CSI Baiting and Colony Eradication

In this study, only activation by C. gestroi, C. curvignathus, and S. medioobscurus in sites A through F showed consistent weekly feeding of the lure wood. Therefore, the installation of auxiliary stations in conjunction with the primary stations was attempted for activated stations in these sites. As seen from Figure 5, the lure wood consumption rate from primary stations all decreased following CSI bait addition. Time elapsed from CSI bait addition to auxiliary stations to disappearance of termites, which indicated colony elimination, ranged from 3 to 7 weeks for C. gestroi, 6 weeks for C. curvignathus, and 16 weeks for S. medioobscurus. The total CSI bait consumption to achieve colony eradication ranged from 206.5 g to 408.5 g for C. gestroi, 240.6 g for C. curvignathus, and 324.9 g for S. medioobscurus (Table 3).

• Download figure
• Open in new tab
• Download PowerPoint

Figure 5.

Weekly lure wood consumption rates by termites in primary IG stations for different sites and plant hosts. Addition of CSI bait into auxiliary stations was marked as week zero.

Monitoring Capacity of IG Stations for Subterranean Termite Detection

Based on the presented data, it was evident that IG stations were not a suitable tool for monitoring activities of subterranean termite species such as N. havilandi, M. crassus, and T. rostratus, with most of the installed stations remaining inactivated after one year. Meanwhile, the IG station was an effective monitoring tool for C. curvignathus, C. gestroi, M. carbonarius, M. gilvus, and S. medioobscurus, with installed IG stations activated by respective species in all inland sites. More importantly, for Sites F, G, and K, the presence of these subterranean termite species was not detected based on visual site inspection nor recorded in site inspection history.

However, the detection period showed great variation ranging from 1 week up to 24 weeks after installation. This observation aligns with that of field studies to delineate foraging territories and population sizes of C. gestroi that also found the inconsistency in activation of IG termite monitoring stations (Su et al. 2023). In our study, the activation inconsistency could be attributed to various factors. Firstly, when comparing the combined activation rates for termite species shown to be detectable by IG stations, we could see the stark contrast between inland sites (37.5%) and coastal sites (3.9%). Considering the subterranean behaviors of the detected termites, sandy and saline soil type in coastal sites could have contributed to reduced foraging behaviors of termites underground, leading to the low activation rate of IG stations. Probably, the loose sand particles in coastal sites did not provide sufficient structural support for underground tunnels as compared to loamy soil as seen in inland sites. The low belowground activity of Coptotermes in coastal area has also been reported by Li et al. (2017), who found that termite infestation was generally lower in regions with higher salinity level. Coptotermes has weak salinity tolerance whereby dehydration symptoms will be quickly displayed after being subjected to saline water (Chiu et al. 2021). Nevertheless, Coptotermes still manages to inhabit trees in coastal areas due to their ability to obtain freshwater from the host trees. This reduces the need for underground movement in order to prevent desiccation (Chiu et al. 2021). Furthermore, at Site K, frequent intermittent flooding caused termites to abandon IG stations shortly after activation, resulting in inconsistent detection. These sporadic activations may undermine the efficacy of CSI baiting by disrupting sustained bait consumption and reducing the quantity of toxicant transferred within the colony. Nonetheless, the successful detection of Coptotermes at Site K—despite its absence during visual assessment—underscores the value of IG stations in identifying infestations of this destructive species.

Lastly, we noticed that once successfully infesting a tree host, subterranean termite species such as C. gestroi and C. curvignathus had the tendency to forage within or near to the host trees. This behavior could be inferred from the fact that IG stations installed for trees in Sites C and E took 10 and 20 weeks, respectively, to be activated by C. gestroi, much higher than the median activation period of 3 weeks. Meanwhile, when trees infested with C. gestroi or C. curvignathus were felled, the removal of the primary food source repeatedly appeared to force the termites to actively expand their foraging range to find new food sources, leading to activation of nearby IG stations. Site D, with the highest activation rate (3 out of 4 trees) and fastest activation time (1 week after IG station installation) had 2 heavily infested A. columnaris removed before the study started. In Site J, all IG stations remained inactive for 13 weeks, but one of the stations became activated 1 week after a cluster of 4 heavily infested C. equisetifolia were felled. Similarly, activations of all stations at Site F happened 1 week after heavily infested trees were felled, although the same set of stations remained inactive for 2 weeks prior to tree removal.

Efficacy of Chlorfluazuron in CSI Bait as a Subterranean Termite Control

Previous studies confirmed the elimination of termite colonies through CSI baiting via destructive excavation of termite mounds (Lenz et al. 1996; Lee et al. 2014) and underground nests (Yamauchi et al. 1997). In this study, we demonstrate the efficacy of CSI baiting through a 4-week baseline assessment of termite consumption rates (Figure 5). Initially, termites consistently consumed newly replenished wood blocks. However, following CSI baiting (designated as Week 0 in Figure 5), a sustained cessation of feeding was observed (3 to 7 weeks), strongly indicating colony elimination. Additionally, an increased soldier-to-worker ratio was observed toward the end of baiting, consistent with the known larvicidal mode of action of CSI. This effect directly reduces worker populations over time, further supporting colony collapse (Lee et al. 2014).

In terms of termite control, the reported data showed that CSI bait containing chlorfluazuron as the active ingredient was effective in eliminating C. curvignathus and C. gestroi, moderately effective against S. medioobscurus, and not effective in managing M. carbonarius and M. gilvus. According to Lee et al. (1999) and Sajap et al. (2000), Coptotermes has a higher feeding rate under the Malaysian tropical conditions. Tropical termite species tend to have smaller population sizes and smaller foraging territories when compared to temperate species (Lee 2002). In Malaysia, a colony of Coptotermes requires a minimum of one month for suppression of the colony (Lee et al. 2007). In comparison with our results, the time taken to achieve complete termite colony elimination is similar, with the shortest time required for elimination being 6 weeks and the longest time required being 10 weeks. For S. medioobscurus, a minimum of 6 months was required for eliminating the entire Schedorhinotermes colony. This result aligns with those reported by Lee et al. (2007). Based on our observations of Schedorhinotermes, this species has shown feeding preference for wood over bait. This feeding behavior led to a slower horizontal transfer of active ingredients in the bait, resulting in a longer period required for colony elimination. Nonetheless, as discussed previously, S. medioobscurus has not been attributed as the primary cause of any tree failure in Singapore.

In this study, no baiting attempts were made for M. carbonarius and M. gilvus, as these species are not considered the target species for urban tree termite management. Furthermore, higher termites like M. carbonarius and M. gilvus exhibit distinct biological traits compared to lower termites, particularly the nonmoulting nature of higher termite workers and their sporadic foraging behavior. These characteristics rendered low efficiency of CSI baiting (Lee et al. 2014; Chiu and Li 2024). Therefore, from a tree health management perspective, CSI baiting was an effective control method for controlling the two destructive species of subterranean termites in Singapore, which are C. curvignathus and C. gestroi.

Comparative Field Study of Subterranean Termite Biology

There were 6 termite species found in this study to be foraging on the lure wood: C. gestroi, C. curvignathus, M. carbonarius, M. gilvus, M. crassus, and S. medioobscurus. Besides C. gestroi and C. curvignathus, which have been well studied due to their destructive feeding behaviors on structures and trees, the biology and feeding behaviors of M. carbonarius, M. gilvus, M. crassus, and S. medioobscurus have not been well documented. According to the literature, M. carbonarius and M. gilvus are mainly foraging species that feed on dead plant matters such as dead wood, leaf litter, and tree barks (Foo et al. 2014); M. crassus feed on decayed or weathered wood and dead trees (Foo et al. 2014); and S. medioobscurus feeds on dead trees or tree stumps (Jasmi and Ahmad 2011; Foo et al. 2014). To our knowledge, there has been no published study that provided data giving comparative insight of potential damage of C. gestroi, C. curvignathus, M. carbonarius, M. gilvus, M. crassus, and S. medioobscurus in trees. Based on the reported weekly consumption rate of lure wood, M. carbonarius and M. crassus showed negligible lure wood consumption. For M. carbonarius, they have a preference for leaf litter over lure wood, which could be clearly indicated by the lesser amount of leaf litter around the termite mound. As for M. crassus, their diet is mainly on decayed tree bark and wood litter. This species showed no preference towards fresh wood in our study. Coupled with superficial scraping marks observed on lure wood, our data evidently showed these two species did not pose any risk to tree structural integrity.

Additionally, although belonging to the genus Macrotermes, M. gilvus displayed a much more aggressive feeding behavior on lure wood, with weekly consumption rates approximately 23 times higher than that of M. carbonarius. Despite the relatively higher consumption rate, M. gilvus is rather sensitive to its feeding resources. This species tends to abandon occupied stations easily due to slight disturbances which were likely a consequence of routine checks to the station. This finding aligns with the results reported by Iqbal et al. (2017). Furthermore, the feeding pattern of M. gilvus was mostly surface scraping with no tunneling, implying that this species would be limited to tree bark and exposed dead wood.

Interestingly, S. medioobscurus lure wood consumption was almost comparable to the notoriously destructive C. gestroi and C. curvignathus. More importantly, S. medioobscurus displayed a nearly similar feeding pattern on lure wood as compared to C. gestroi and C. curvignathus, as this species also created sharp tunnels through the wood as it fed, but to a much lesser extent. Such findings imply potential high damage on trees by S. medioobscurus. However, based on our field observations and tree inspection records, all cases of tree failure due to termite infestation in Singapore have been caused by C. gestroi or C. curvignathus. These results are similar to those described by Cheng et al. (2008), where C. curvignathus was the primary species that attacked both living and dead oil palms while Schedorhinotermes was associated with plant residues. Therefore, despite the tunneling feeding behavior with a high consumption rate, S. medioobscurus appears to be a potent species that feeds on dead wood while being mostly nondestructive to living trees.

Besides providing a common criterion for comparison in terms of wood damage, we found that IG stations also provided insight into the subterranean termite interspecies dynamics within the site. For instance, data collected from the network of IG stations at Site G suggested that M. gilvus was the dominating species foraging the forest floor and M. carbonarius was the less dominant species. Meanwhile, data from the Site F stations revealed that C. curvignathus played a dominant and antagonistic role to S. medioobscurus. The latter species could only expand their foraging range and feeding behavior after C. curvignathus had been eliminated from the site.

Ultimately, effective termite management should begin with recognizing their ecological significance through the understanding of termite biology. Termites are not simply pests; many species serve as vital decomposers, contributing to global carbon cycling, nutrient redistribution, and soil health (Umar and Ab Majid 2020a; Ahmad et al. 2021). Concurrently, host susceptibility is another factor that should be considered, because different ornamental trees are known to have varying susceptibility to termite attack (Lee 2014). Generally, softwoods, such as pine, are among the more vulnerable to termites. In contrast, tropical species, such as teak, exhibit greater resistance towards attack by genus Coptotermes (Judd 2018). However, not all hardwoods are termiteresistant, as there are multiple factors that could affect the preference of termites, including secondary metabolites, wood density, wood composition, and nutritional value (Judd 2018).

The central challenge, then, is to move beyond broad elimination toward targeted, ecologically grounded strategies—prioritizing surveillance around susceptible tree hosts, focusing control efforts on the most destructive termite species, and preserving nondestructive decomposer species that play critical roles in sustaining urban ecosystem health.