BEHAVIOR

Buccal Manipulation of Sand Particles During Tunnel Excavation ofthe Formosan Subterranean Termite (Isoptera: Rhinotermitidae)

HOU-FENG LI1 AND NAN-YAO SU

Department of Entomology and Nematology, Fort Lauderdale Research and Education Center, University of Florida,3205 College Avenue, Ft. Lauderdale, FL 33314

Ann. Entomol. Soc. Am. 102(2): 333Ð338 (2009)

ABSTRACT Serial tunneling behavior of Coptotermes formosanus Shiraki (Isoptera: Rhinotermiti-dae), the Formosan subterranean termite, includes excavation, loading, transportation, and depositionof sand particles. Functions of mouthparts, including mandibles, maxillae, labrum, and labium, for thetunneling behaviors were described. The four mouthparts form a buccal cavity for loading three tofour sand particles (ranging 0.300Ð0.355 mm in diameter) during sand displacement. The maxillae, notthe mandibles, are the major appendages for sand excavation and deposition. Previous studiesspeculated a “soil-compaction hypothesis” that subterranean termites might press the sand particleswith their head to build the tunnel. Examination of video recordings of tunnel excavation through amicroscope indicated sand was not compacted. When two photographs of tunnel tips taken before andafter excavation were superimposed, sand particles surrounding tunnel tips remained in place, whichdemonstrated that termites did not press sand to either side. Observations and experiments indicatethat particle displacement is the major mechanism for subterranean termites to build tunnels ratherthan particle compaction.

KEY WORDS Coptotermes formosanus, tunnel architecture, excavating behavior, mouthparts

Underground foraging galleries of Coptotermes spp.have been unearthed by some researchers who foundthat these gallery systems ranged in depth from 48 to304 cm and extended over a hundred meters from thenests at a depth from 5 to 300 cm (Oshima 1919,Ehrhorn 1934, Greaves 1962, Greaves and Florence1966, King and Spink 1969). Despite their extensiveunderground gallery systems, Coptotermes spp. andother subterranean termites are not known to movesoil to a great extent; thus, the speculation that sub-terranean termites compact soil to create gallery spacewas proposed by Ebeling and Pence (1957), and this“soil-compaction hypothesis,” was generally accepted(Greaves 1962, Greaves and Florence 1966, King andSpink 1969, Lee and Wood 1971).

When termites were placed in two-dimensionalarenas as described by Su et al. (2004), excavatedsand was often moved to the empty space within theintroduction chamber as the tunnel progressed.Transportation of excavated sand to an empty spacedoes not in itself invalidate the soil-compaction hy-pothesis, because termites may move and compactsoil simultaneously. The objective of this study is todescribe the functions of termite mouthparts fortunnel excavation, and to validate the soil-compac-tion hypothesis.

Materials and Methods

Individuals of three Þeld colonies of Formosan sub-terranean termites, Coptotermes formosanus Shiraki,were collected in Broward County, FL, by using un-derground bucket traps (Su and Scheffrahn 1986).Twenty workers (undifferentiated larvae third instaror older) from each colony were put in 80% ethanolfor measuring their head width, and dissecting theirmouthparts under a dissecting microscope (modelSZX12, Olympus Optical Co., Ltd., Tokyo, Japan). TheworkerÕs head was photographed with a stereomicro-scope (model MZ 12.5, Leica Microsystems GmbH,Weltzlar, Germany) and enhanced with Auto-Mon-tage program (version 5.02, Synoptics Ltd., Cam-bridge, United Kingdom). The photographs of eachmouthpart were taken with a compound microscopewith a digital camera (model BX51 and DP70, Olym-pus Optical Co., Ltd.).

Observations of workersÕ soil manipulation behav-iors were conducted by using a laboratory arena sim-ilar to that described by Li and Su (2008). The arenawas constructed of two sheets of transparent Plexiglas(15 by 15 by 0.6 cm) separated from each other by fourPlexiglas laminates (15 by 1.5 by 0.15 cm) placedbetween the outer margins to form a 0.15-cm gap of 12by 12 cm for loading sand and was held together withscrews (Fig. 1). To increase the color contrast be-tween sand particles for identifying each sand particle1 Corresponding author, e-mail: houfeng@ulf.edu.

0013-8746/09/0333Ð0338$04.00/0 � 2009 Entomological Society of America

under a microscope, 32 g of mixed colored sand (red,green, blue, purple, and gray) ranging from 0.300 to0.355 mm in diameter was used for tunneling substrate.An empty square space (2.8 by 2.8 by 0.15 cm) was leftat one corner for termites and a piece of moist wood(1.0 by 1.0 by 0.15 cm) was placed in the corner of thisspace. The top Plexiglas sheet had one access hole (0.6cm in diameter) at each of two corners for injecting�8 ml water to moisten sand and to introduce termites(Fig. 1).

The tunneling behaviors in the arena were observedand recorded through a video-equipped dissectingmicroscope. The speed of mouthpart movement wasquantiÞed by counting the video frames (60 framesper s). The number of sand particles moved over 5 cmper trip by 20 workers was counted separately. Theexperiment for testing the soil compaction hypothesiswas conducted by using 100 termites (90 workers and10 soldiers) and replicated three times with termitesfrom three different colonies. Observations weremade only of tunnels not touching the arena edges(Fig. 2A, tunnel b). Tunnels excavated along the arenaedge (e.g., Figure 2A, tunnels a and c) were not usedbecause termites were able to press sand particles inonly one direction. At the tip of the nonedge tunnel,a transparent Þlm with printed Xs (tunnel b, pointedby arrow tips, Fig. 2), at each corner of a 0.6-cm squarewas placed on the top Plexiglas sheet of the arena forphotographic orientation (Fig. 2). Back-lit digital im-ages were taken, before and after tunnel excavationbeneath the four marks, respectively (Fig. 2, A and B).At the midpoint of the marks, the width of the tunnel(Wt) was measured (Fig. 2B). Digital images werealso taken with an overhead light source through themicroscope, before and after termites excavated thetunnel beneath the four marks, respectively (Fig. 2, Cand D). By superimposing the two images (Fig. 2, Cand D), the sand particles that remained stationarywere identiÞed along with particles that were moved;

thus, the interface between the stationary and movedsand particles could be determined. The distance be-tween the two interfaces on two sides of the tunnel,termed as the width of the moved sand (Wm), wasmeasured (Fig. 2D). Wt and Wm were measured atthe same point. Five subsamples of Wt and Wm weremeasured along tunnel b in each replication. The sig-niÞcant difference between Wm and Wt was tested bypaired t-test at � � 0.05 (SAS Institute 1985). If ter-mites only removed sand particles, then the width ofa tunnel must be the same as the width of the movedsand or Wm � Wt. However, if termites removed andpushed sand particle simultaneously, then Wm � Wt.

Results and Discussion

Tunneling Behaviors. The worker buccal cavity isenclosed by the dorsal labrum, the lateral mandiblesand maxillae, and the ventral labium (Fig. 3A). Eachmouthpart is synchronized to perform a four-step tun-neling behavior that includes excavating sand, loadingsand in buccal cavities, holding sand for transporta-tion, and depositing sand. The video clips of Formosansubterranean termiteÕs tunneling behaviors can beviewed at http://ßrec.ifas.uß.edu/ent_nem/structural_entomology_su_videos.shtml.

Excavating sand is deÞned as moving sand particlesfrom its original location into the termitesÕ buccalcavity. The maxillae are the major appendages forexcavating sand (Fig. 3D). The basal plates, the stipes(S), support the laciniae (La) and the galeae (Ga)distally and maxillary palps (Mp) laterally. The laciniaterminates in a heavily sclerotized double tooth. Thegaleae cover the laciniae above and laterally. Beforeexcavating, the maxillary palps in front of the laciniaeand galea Þrst touched the sand particles. The exca-vating behavior was initiated by a lateral extension ofthe maxillae, and the distance between the two la-ciniae can be as wide as the width of the head. Themean (�SE) worker head width was 1.179 � 0.008 mm(n � 20). The extended maxillae moved forward andthe distal laciniae were inserted into gaps betweenparticles followed by grasping and moving sand par-ticles from original location to the buccal cavity. Thisprocess may take place as fast as 0.5 s. Termites oftenangled their heads so that the maxillae can easily graspand remove sand particles. When the pair of maxillaemoved backward while holding a sand particle, themandibles which bear heavily sclerotized teeth (Fig.3C)simultaneouslyopenedandreceived theparticles.

Loading is deÞned as packing sand particles into thebuccal cavity. After placing the Þrst sand particle intobuccal cavity, the maxillae repeatedly moved forwardand backward to stuff the Þrst sand particle deeperinto the cavity, or continued excavating and loading asecond particle that simultaneously pushed the Þrstparticle deeper inward. Termites also rotated sandparticles in the buccal cavity with the maxillae. Duringloading, the labrum and labium were pushed upwardand downward, respectively, and some of the sandparticles may slip out of the cavity. The labrum, thedorsal appendage, is a transparent, convex, and semi-

Fig. 1. Experimental arena to observe workersÕ tunnelingbehaviors and to test the soil compaction hypothesis. (OnlineÞgure in color.)

334 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 2

circular ßap (Fig. 3B) covering the mandibles andprevents sand particles from slipping off the mandi-bles. The labrum can bend over a 60� angle when sandparticles projected above the dorsal surface of themandibles. As shown in Fig. 3E, the labium, the ventralappendage, consists of the postmentum (Po) and theprementum (Pm). The postmentum is a shield likeplate, and the prementum consists of a pair of labialpalps (Lp) and the ligula. The ligula is a four-lobedstructure with two inner lobes (the glossae [G]) andtwo outer lobes (the paraglossae [Pg]). The labiumcan angle downward to 50� or greater, and the fourlobes of the ligula spread laterally to support the sandparticles. During loading, termites sometimes partiallyextruded sand particles out of the buccal cavity byusing their hypopharynx (H) (Fig. 3A), to reposition

the particles. Termites may excavate and load sandalternately or simultaneously several times at the ex-cavation sites before moving sand to deposition sites.

During the step of transporting sand, termitesmoved it from the excavation site to the depositionsite. The mandibles, maxillae, labrum, and labium heldthe sand particles together until termites arrived at thedeposition site, and no movement of buccal append-ages was observed during the transportation. The av-erage number of sand particles carried by one workerper trip was 3.5 � 0.15 (n � 20).

We deÞned sand deposition as the releasing of sandparticles from the buccal cavity at deposition sites.Termites deposited sand in a given empty space, in-cluding the tunnel that they were digging, and theydid not always deposit all particles in one trip at the

Fig. 2. Measurement of width of tunnel (Wt) and width of moved sand (Wm). (A) Termites excavated three tunnels(a, b, and c) from the upper right corner of the arena. At the tip of tunnel b, a transparent Þlm with printed Xs at each cornerof a 6-mm square was placed on the top Plexiglas sheet. The arrows pointed at the four marks. (B) After termites haveexcavated the tunnel beneath four marks, the Wt was measured at the midpoint of the marks. (C and D) Microscopic imagesof the tunnel tip were taken before and after termites excavated beneath the four marks, respectively. By superimposingmicroscopic images C and D, the stationary and moved sand particles were identiÞed and the interfaces between moved andstationary sand particles on two sides of the tunnel were drawn in D shown as black lines. The distance between the twointerfaces, termed as the width of moved sand (Wm) was measured at the same point where Wt was measured. (Online Þgurein color.)

March 2009 LI AND SU: TUNNELING BEHAVIOR OF FORMOSAN SUBTERRANEAN TERMITE 335

samesite.Termitesusuallydepositedonesandparticleat time. Termites rotated their heads alternately fromside to side against the walls of deposition sites andpushed the sand particle from several directions withmaxillae to Þt the edge of the walls. Termites minutelyadjusted the position of each sand particle with the tipof maxillae (lacinia) rapidly at 0.25 s per round androunds repeated for up to 2Ð5 s. After depositing allsand particles, termites returned to the previous ex-cavation site and started another cycle of excavation,loading, transportation, and deposition.

In this study, the tunneling material used wascolored sand of uniform size, which facilitated ob-

servations and quantiÞcations of the behavior andfunction of each mouthpart. However, it should benoted that soil properties such as moisture (Su andPuche 2003), density (Tucker et al. 2004), and par-ticle size (Cornelius 2005) can alter termite-tun-neling behavior. Moreover, the behaviors of exca-vating soil and cellulose involved different steps,because it probably takes less effort to separate soilparticles than to cut and tear off cellulose Þbers.Indrayani et al. (2007) reported that C. formosanusseparated wood fragments by cutting and pullingactivities. The maxillae grasped the wood and thenthe mandibles cut wood fragments. Termites pulled

Fig. 3. The mouthpart of the worker of C. formosanus. (A) Frontal view of the head. (B) Labrum. (C) Mandibles. (D)Maxillae. (E) Labium. B, bristle; G, glossa; Ga, galea; H, hypopharynx; L, labrum; La, lacinia; Lp, labial palp; M, mandible;Mp, maxillary palp; Pa, primary articulation point; Pg, paraglossa; Pm, prementum; Po, postmentum; S, stipes; Sa, secondaryarticulation point. (Online Þgure in color.)

336 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 2

wood fragments with both their maxillae and man-dibles while rocking their bodies. In this study, ter-mites could easily separate sand particles by usingtheir maxillae with no rocking motion.

The function of the termiteÕs mandibles for ex-cavating several substrates has been reported pre-viously. These functions include pushing sand par-ticles to create the tunnel (Reticulitermes hesperusBanks; Ebeling and Pence 1957), grasping and cut-ting kaolin clay to form a pill [Reticulitermes flavipes(Koller); Whitman and Forschler 2007], and cuttingand pulling wood fragments (Incisitermes minorHa-gen, C. formosanus, and Reticulitermes speratusKolbe; Indrayani et al. 2007). Because the maxillaeare covered beneath the mandibles, and their move-ment is rather fast, their function has not been welldescribed. In this study, by examining the videoframes recorded through the microscope, we foundthat the maxillae were the major appendage to ex-cavate and deposit sand, and the function of man-dibles was only for holding the sand particles duringthe serial tunneling behaviors.

The head capsule movement during excavation anddepositionof soil orothermaterialhasbeenpreviouslydescribed. Ebeling and Pence (1957) noted that R.hesperus pushed their heads forward through the sandand pushed sand to either side, which is similar to thehead alternating rotations of R. flavipes during exca-vation and deposition (Whitman and Forschler 2007)and to the side-to-side motion of the head of Zooter-mopsis spp. during placement of the fecal pellets onfecal cement (Stuart 1967). However, the interpreta-tions of the function of the head movement differedamong observers. Ebeling and Pence (1957) thoughtthat the head movement was used to press sand par-ticles to create tunnel space. Whitman and Forschler(2007) indicated that the head movement functionedto pull soil held by the mandibles during soil excava-tion, but they offered no explanation for the samebehavior during soil deposition. Stuart (1967) positedthat the termites moved heads to push the fecal pelletsheld by mandibles onto the wall with fecal cement.Our video recording showed that the rotating move-ment of the head capsule during sand depositionmainly allowed the maxillae to push sand particles intowall edges from several directions. C. formosanus didnot use the mandibles for pressing or pushing sandparticles.Rejecting theSoilCompactionHypothesis.Wm and

Wt were 3.16 � 0.18 and 2.90 � 0.15 mm, respectively.There is no signiÞcant difference between Wm andWt (n � 15, P � 0.12), which indicates that sand wasnot pushed to any signiÞcant degree. We concludethatC. formosanus removed the sand particles to maketunnels without pushing the sand particles to eitherside during excavation.

During sand deposition, the maxillae minutely ad-justed and pushed every sand particle to Þt the edgeof the wall one by one, which led to the proposal of themodiÞed soil-compaction hypothesis, in which ter-mites deposit sand in a greater density than before itwas removed to create space (Li and Su, 2008). How-

ever, the modiÞed hypothesis also has been rejected.Based on our behavioral observations and experi-ments, we speculate that soil displacement is the mainmechanism for subterranean termites to build under-ground galleries in sandy soil.

Acknowledgments

We thank Paul Ban and Ron Pepin (University of Florida)for technical assistance, and Rudolf H. Scheffrahn, RobinGiblin-Davis, and Brian Bahder (University of Florida) forreview of the manuscript. We also thank Lyle Buss for adviceand assistance with montage photography (University ofFlorida). This study was supported in part by a grant fromUSDAÐARS under the grant agreement 58-6435-2-0075.

References Cited

Cornelius, M. L. 2005. Effect of particles size of differentsands on the tunneling behavior of the Formosan sub-terranean termite (Isoptera: Rhinotermitidae). Sociobi-ology 45: 173Ð184.

Ebeling, W., and R. J. Pence. 1957. Relation of particle sizeto penetration of subterranean termites through barriersof sand or cinders. J. Econ. Entomol. 50: 690Ð692.

Ehrhorn, E. M. 1934. The termites of Hawaii, their eco-nomic signiÞcance and control, and the distribution oftermites by commerce, pp. 321Ð333. In C. A. Kofoid[eds.], Termites and termite control. University of Cal-ifornia Press, Berkeley, CA.

Greaves, T. 1962. Studies of foraging galleries and the inva-sion of living trees by Coptotermes acinaciformis and C.brunneus (Isoptera). Aust. J. Zool. 10: 630Ð651.

Greaves, T., andR.G. Florence. 1966. Incidence of termitesin blackbutt regrowth. Aust. For. 30: 153Ð161.

Indrayani, Y., T. Yoshimura, Y. Yanase, Y. Fujii,H.Matsuoka,and Y. Imamura. 2007. Observation of feeding behaviorof three termite (Isoptera) species: Incisitermes minor,Coptotermes formosanus, and Reticulitermes speratus. So-ciobiology 49: 121Ð134.

King,E.G.J., andW.T. Spink. 1969. Foraging galleries of theFormosan subterranean termite,Coptotermes formosanus,in Louisiana. Ann. Entomol. Soc. Am. 62: 536Ð542.

Lee, K. E., and T. G. Wood. 1971. Nests, mound and galler-ies, pp. 22Ð45. In K. E. Lee and T. G. Wood [eds.],Termites and soils. Academic Press, New York.

Li, H.-F., andN.-Y. Su. 2008. Sand displacement during tun-nel excavation by the Formosan subterranean termite(Isoptera: Rhinotermitidae). Ann. Entomol. Soc. Am. 101:456Ð462.

Oshima, M. 1919. Formosan termites and methods of pre-venting their damage. Philos J. Sci. 15: 319Ð383.

SAS Institute. 1985. SAS userÕs guide: statistics, version 5 ed.SAS Institute, Cary, NC.

Stuart, A.M. 1967. Alarm, defense, and construction behav-ior relationships in termites (Isoptera). Science (Wash.,D.C.) 156: 1123Ð1125.

Su, N.-Y., and R. H. Scheffrahn. 1986. A method to access,trap, and monitor Þeld populations of the Formosan sub-terranean termite (Isoptera: Rhinotermitidae) in the ur-ban environment. Sociobiology 12: 299Ð304.

Su, N.-Y., and H. Puche. 2003. Tunneling activity of subter-ranean termites (Isoptera: Rhinotermitidae) in sand withmoisture gradients. J. Econ. Entomol. 96: 88Ð93.

Su, N.-Y., B. M. Stith, H. Puche, and P. Bardunias. 2004.Characterization of tunneling geometry of subterranean

March 2009 LI AND SU: TUNNELING BEHAVIOR OF FORMOSAN SUBTERRANEAN TERMITE 337

termites (Isoptera: Rhinotermitidae) by computer simu-lation. Sociobiology 44: 471Ð483.

Tucker, C. L., P. G. Koehler, and F. M. Oi. 2004. Inßuenceof soil compaction on tunnel network construction by theeastern subterranean termite (Isoptera: Rhinotermiti-dae). J. Econ. Entomol. 97: 89Ð94.

Whitman, J. G., and B. T. Forschler. 2007. Observationalnotes on short-lived and infrequent behaviors displayedby Reticulitermes flavipes (Isoptera: Rhinotermitidae).Ann. Entomol. Soc. Am. 100: 763Ð771.

Received 24 February 2008; accepted 3 November 2008.

338 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 2