|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 TRAFFIC FLOW AND TACTILE INTERACTIONS ORGANIZE THE LABOR OF SUBTERRANEAN TERMITES DURING TUNNEL EXCAVATION: AN ALTERNATIVE TO SCENT MEDIATED STIGMERY By PAUL MICHAEL BARDUNIAS A DISSERTATION PRESENTED TO THE GRADUATE SCHO OL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Paul Michael Bardunias
3 To my father, who taught me how to th ink, and my mother, who taught me why
4 ACKNOWLEDGMENTS I thank my family my wife Helaina, my daughter Arianna Rose, and my sons, Paul and Benjamin, for their enduring support throughout the long process of my education. I never gave up because you ne ver did. I received invaluable emotional support from my brothers, Peter and John, my beloved aunt Paulette Merchel and her husband Paul as well as Lloyd and Cynthia Greenberg, Sarah Greenberg, Andrew Greenberg and Marissa Lyons My parents, Peter and B onnie, and my grandparents, Florence Mauro and Helen, and Henry Poslu szny were my inspiration. I thank my advisor, Nan Yao Su the patron saint of second chances, for giving me the opportunity to study these fascinating insects. I also thank those who supported my research directly and my collaborators, Paul Ban, Ron Pepin, Stephanie Osorio, Aaron Mullins Sang Hee Lee, Thomas Chouvenc, Hou Feng Li and Rou Ling Yang for their assistance and scholarly discussions My gratitude goes to my committee, Robin Giblin Davis, Bill K ern, and Monica Elliott, for not being risk averse. I hope their bet pays off. Last ly I thank my partner Caroline Efstathion for showing me that entomology is more than theory and saving lives in the process
5 TABLE OF CONT ENTS ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW ................................ ..................... 11 Introduction ................................ ................................ ................................ ............. 11 Nest E xcavation ................................ ................................ ................................ ...... 12 More T han P he romone C ontrol I s R equired F or E xcavation ................................ .. 13 Tunnel O rientation ................................ ................................ ................................ .. 14 Choosing E xcavati on A nd B ranching S ites W ithout S cent L abeling ....................... 15 2 EXCAVATION SITE SELECTION ................................ ................................ .......... 19 Introduction ................................ ................................ ................................ ............. 19 Methods ................................ ................................ ................................ .................. 21 Results ................................ ................................ ................................ .................... 24 Tunneling ................................ ................................ ................................ .......... 24 Comparison O f T unnel W idths ................................ ................................ ......... 26 Dist ribution O f L ateral E xcavation D istances ................................ .................... 27 Discussion ................................ ................................ ................................ .............. 28 3 BRANCH FORMATION ................................ ................................ .......................... 42 Introduction ................................ ................................ ................................ ............. 42 Materials A nd Methods ................................ ................................ ........................... 44 Exp eriment 1: Proximity to T unnel T ip ................................ .............................. 46 Experiment 2: Size of D epression ................................ ................................ .... 46 Experimental R esults ................................ ................................ .............................. 47 Exp eriment 1: Proximity to T unnel T ip ................................ .............................. 47 Exp eriment 2: Size O f D epression ................................ ................................ ... 48 Discussion ................................ ................................ ................................ .............. 48 4 EXCAVATION THROUGH WOOD ................................ ................................ ......... 56 Introduction ................................ ................................ ................................ ............. 56 Material A nd Methods ................................ ................................ ............................. 58 Statistical Analysis ................................ ................................ ................................ .. 59 Results ................................ ................................ ................................ .................... 59 Discussion ................................ ................................ ................................ .............. 60 LIST OF REFERENCES ................................ ................................ ............................... 69
6 BIOGR APHICAL SKETCH ................................ ................................ ............................ 76
7 LIST OF TABLES Table page 2 1 A comparison of tunnel width (mm S.E.M.) at the digging face at the tip and at two distances from the t ip ................................ ................................ ............... 33 3 1 The mean number of excavation and deposition events ( SD) and the number of trials for which branches formed out of 15 replicates ........................ 51 3 2 The mean number of excavation and deposition events ( SD) ......................... 52
8 LIST OF FIGURES Figure page 2 1 Two dimensional are na design consist of two clear ................................ ........... 34 2 2 A typical queue at the tip of a tunnel. ................................ ................................ .. 35 2 3 Termites trave l ling down a tunnel ................................ ................................ ....... 36 2 4 Antennation of the digging face at the tunnel tip ................................ ................. 37 2 5 Tunnel prop agation at three time intervals ................................ ........................ 37 2 6 The distribution of the distances of excavation events ................................ ....... 38 2 7 The characteristic posture s of a termite digging towards the lateral edges of the tunnel tip. ................................ ................................ ................................ ...... 39 2 8 A model of termite excavation based on the b i directional traffic flow in tunnels and the responses of individual termites to tunnel features and traffic congestion. ................................ ................................ ................................ ......... 40 3 1 Experimental arenas ................................ ................................ .......................... 53 3 2 A 3 mm by 3 mm depression in the t unnel wall ................................ .................. 54 3 3 The distal end of a tunnel during excavation showing termites in queue ........... 55 4 1 T wo dimensional arena design. ................................ ................................ .......... 63 4 2 Two dimensional arena 72 h after exposure to termites. ................................ ... 64 4 3 The dry mass of paper before and after 72 h exposure to termites ................... 65 4 4 Paper parcels deposited along the inside of tunnels ................................ .......... 66 4 5 Paper parcels deposited within a t unnel, completely f illing it .............................. 67 4 6 Paper deposited within the ope ned introduction space ................................ ...... 68
9 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fu lfillment of the Requirements for the Degree of Doctor of Philosophy TRAFFIC FLOW AND TACTILE INTERACTIONS ORGANIZE THE LABOR OF SUBTERRANEAN TERMITES DURING TUNNEL EXCAVATION: AN ALTERNATIVE TO SCENT MEDIATED STIGMERY By Paul Michael Bardunias May 20 13 Chair: Nan Yao Su Major: Entomology and Nematology S tigmergy originated from the study of nest construction in termites and has been a model for self organized systems. All current models of termite construction rely on this type of indirect commun ication between termites in the form of the chemical marking of work product. We propose an excavation process that is governed solely by tactile interactions of termite excavators and patterns of traffic flow. Coptotermes formosanus Shiraki tunnels are bi directional and radiate away from an origin These vectors provide shared information that coordinates the labor of individuals. When excavation sites at the digging face of the tunnel tip were occupied by other termites, a queue of excavators awaitin g access f ormed. Termites in queue respond ed to this delay by excavating soil from the tunne l wall By examining excavation under conditions o f high flow and low flow we demonstrated that the rate of lateral excavation increased with queue length, a proc scales tunnel width to traffic flow.
10 The presence of a queue mediates branch formation. The likelihood that a depression became a branch was determined by competition between soil excavation enlarging it and dep osition filling it in. When 3 mm by 3 mm depressions were positioned within a zone of tunnel near the tip, branches invariably formed, but when these depressions were further away from the tunnel tip, branches were less likely to form. Increasing the siz e of depressions did not greatly increase the likelihood of their becoming branches. Coptotermes formosanus excavates through food sources in a process that results in the deposition of masticated food particles within tunnels in a manner previo usly shown for excavated soil. If soil Basidiomycota opportunistically invaded the microhabitat of fecal carton covered wood fragments along C. formosanus tunnels, then the basic mechanics of tunnel excavation through soil, when applied to a food source, serve as a preadaptation to evolution of the cultivation of fungi by termites. If, on the other hand, C. formosanus is exploiting fungi and fungal conditioned wood sequestered in tunnels, then a novel form of agriculture, perhaps the first to exist between fungus an d insect, may be described.
11 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW Introduction As the saying goes, there is no substitute for a good first impression Many people encounter their first termite with furrowed brow and opened wallet a s they examine the damage done to their houses by a hungry colony tunneling up from below But for others, the first termites they meet are the happy families of industrious little gnomes that construct soaring fairytale spires as portrayed in the academy award nominated movie From the titles of these movies, it is clear that the structures constructed by termites are the source of our fascination, and rightly so for they build domic iles that on their scale are three times higher than the tall est human sky scrapers (Hansell 2005), and incorporate architectural complexities that are instructing human builders (Holbrook et al. 2010). This dichotomy of insidious bu rrowers beneath and inspired architects above is pervasive among termitologists as (Wilson 1971), are primarily the subject of studies aimed at pest control, while st udies of behavior and computer science. If more primitive termites primarily excavate their living space (Wilson 1971), then the processes by which derived lineages build above ground may reflect a simple evolutionary tinkering with the mechanics of tunneling. Just as all construction is only as firm as its foundation, so too a true understanding of how termites evolved the process by which they build the above ground str uctures requires knowledge of the mechanics of how they construct below.
12 The aesthetic form of nest mounds arises from the need to overcome the physiological challenges of their constructors posed by the environment (Aanen and Eggleton 2005 ). The mounds can rightly be viewed as an extension of their Dawkins 1999 Turner 2005). The termites cannot survive structure. In addition, the t ermites a re ecosystem engineers (Jones et al. 1994) specifically soil engineers whose actions modify the soil horizons in a manner that creates niches in the environment (Bignell 2006 ). If there is no centralized, top down, control of laborers, then how is the wo rkforce organized? Grasse (1959) put forth a simple scheme by which termites organize th eir tigmergy an indirect mode of communication between termites whereby the work product of a builder acts to guide the response of subs equent workers. This process relies on the application of various pheromones to mark sites and to identify the work product of construction to induce positive feedback loops, wherein building by one individual stimulates additional construction at the sce nt labeled site Nest E xcavation The most basal lineages of termites primarily excavate their dwellings in wood or soil rather than build complex structures (Wilson 1971) It may be accurate to say that product of the excavation of nests below ground and result from a need for space to place the tailings from the clearing of tunnels and chambers (Li and Su 2008 ). As with construction, termites work together to excavate soil. They do this by removing soil in a rot ation of individuals at the site of tunnel elongation, each of whom loads a parcel of soil by mouth ( Li and Su 2009 Bardunias
13 and Su 2010 ). Many cockroach species, related to the ancestor of termites, excavate soil, but they commonly do so by displacing and compactin g soil with the forelegs (Bell et al. 2007). This mode of excavation may be optimal fo r digging while moving through mouths, t ermites digging alone would nee d to make many, long trips from the tunnel tip to a remote site to deposit excavated soil if t he tunnel is to remain open and unobstructed. However, as group excavators, they are a ble to form digging groups that form et al. 20 02 ) which increases the efficiency of excavation by mouth It may be important that the closest relative to the termite clade, Cryptocercus spp ., are semi social and excavate both by forelimb and by mouth. More T han P heromone C ontrol I s R equired F or E xca vation Stigmergy forms the basis of all current models of nest construction, but when applied to the excavation of galleries within soil (Sudd 1975, Buhl et al. 2005, Kramer 2005), each of these failed to produce elements of natural tunnels or required add itional information and added complexity in the marking system. Stigmergic marking faces challenges as an organizing force in excavation. In the context of construction, termites are thought to apply a pheromone marker from their mouths as they place soi l parcels, and these deposited parcels then attract other termites with loads to drop. The use of stigmergy in excavation requires termites to label the void from which soil is removed in order to direct future excavators to the proper digging site. Furt hermore, the parcels of soil removed from a tunnel wall cannot be marked so as to attract other termites to excavate from wherever it is deposited or termites would be caught in a loop, picking up and dropping the same particles. To overcome this constrai nt, high dissipation rates for pheromone labels or the addition of more scent labels have been hypothesized (Buhl et
14 al 2005 Kramer 2005). This added complexity, based on no empirical data, violates the simplicity that has been suggested as support for t he stigmergic hypothesis The three main organizational forces hypothesized by stigmergic marking in excavation : tunnel orientation, the selection of sites at which individuals remove soil, and the production of branches from main tunnels. Focus on stigme rgy as an explanation of how termites build tunnels may have obfuscated the importance of other methods used by termites to orient and propagate their tunnels. Tunnel O rientation The subterranean termite, Reticulitermes flavipes (Kollar), has been shown to orient the initial excavation of tunnels around the rim of an introduction chamber in a nonrandom manner that may increase termite foraging efficiency by spreading exploratory tunnels away from one another and thus reducing overlap in a search area (Robso n et al. 1995). T unnels radiate from a central point, and this initial nonrandom element might be sufficient to ensure an evenly distributed search pattern in featureless soil Su et al (2004 ) simulated the patterns of tunnels radiating from a central sit e. From this study it became clear that termite tunnels do not loop backwards as we might he origin. Termites appear to b e orienting not just the origin of tunnels from a central int roduction site, but the propa gation of their tunnels in a nonrandom manner. Honey bees (Gould et al 1978) and ants (Camlitepe and Stradling 1995) use the as well (Becker and Gerish 1981 Maher 1998), but this remains to be studied. Although the use of physical irregularities (edge following) and olfactory gradients in the soil as guides for orientation has been demonstrated previously (Evans 2003), olfactory
15 gradients can only cause a change in tunnel heading over a close range to the source (Su 2005 ). It has been shown that termites only followed the edges of wooden guidelines (Swoboda and Miller 2004) which begs the question of whether the structure represents a guide to be followed or a goal in itself. One means of orienting foraging paths is dead reckoning, the integration of idiothetic or environmental cues to track position relative to a point of origination (Mittelstaedt 1983). Subterranean rodents use dead reckoning (Kimchi and Terkel 2003) to orient their tunnels in an efficient manner. Dead reckoning is common in insects (Collett and Collett 2000), and it has been implicated in termite orientation (Golberg 1975). Bardunia s and Su (2009 b ) demonstrated path integratio n in subterranean termites by forcing them to walk along an L shaped path between the tunnel origin and the tunnel end where they were free to dig. The termites did not excavate at random, but elongated their tunnels following a vector that was directly aw origin. They termed this the Global Away Vector (GAV). Choosing E xcavation A nd B ranching S ites W ithout S cent L abeling Studies to prove that pheromone labeling is not the guiding force in tunnel excavation are not easy to design Be cause the stigmergic scent labels used in excavation remain hypothetical, any direct evidence against their existence could be challenged by invoking further hypothetical pheromones in a more complex scheme. For my thesis I have chosen a different tact. Rather than disprove the current paradigm, I will pro pose another viable scheme and allow the scientific community to gauge the relative merit of each.
16 Stigmergic marking provides a straightforward means of directing future termites to excavate at specific sites. But direct labeling of specific sites may not be necessary if the movement of termites themselves is constrained within narrow tunnels causing termites to tend to loiter in the vicinity of sites where excavation is required. The bidirectional fl ow of termites through tunnels, the movement of prospective excavators towards the tip and termites loaded with soil back toward the mouth, can serve to focus labor at the tunnel ends where excavators must halt or turn around. Variation in the flux of ter mites at any point in the tunnel will change the tunnel environment, with an increase in worker worker interactions that limit movement options and access to some digging sites, while focusing labor on others. Traffic flow can constrain termite movement in a manner that focuses labor at specific sites along tunnels. We hypothesize that if subterranean termites in queue at the tip of a tunnel alter their behavior by switching tasks from attempting to move forward towards the end of the tunnel and the diggin g face to excavating from the tunnel wall at their current location (lateral excavation), then crowding and traffic congestion could focus excavation along tunnel walls near the tunnel tip. The objective of this study is to analyze the process of tunnel e xcavation in the Formosan subterranean termite, C. formosanus at the level of the individual termite to determine the mechanics of excavation. In this study we will observe the behavior of individuals preliminary to digging and determine the location of excavation sites within tunnels as a whole and within the area of the tunnels where queue formation occurs. By creating two distinct sizes of queue in tunnels, we examine d how queue size affected the manner in which tunnels were widen ed as they were elong ate d
17 The flow of termites moving past depressions may act as an alternate means of may be crucial to the correct functioning of the system. Bi directional traffic flow through tunnels results in termites that are seeking to excavate moving outward to dig at the tunnel tip and termites loaded with excavated soil moving back towards the beginning of the tunnel to drop their loads of soil. This may make excavation fro m depression near the tunnel tip more likely, while focusing the deposition of soil on depressions near the tunnels origin. While stigmergy can account for branch formation from tunnels, the movement of termites along tunnels and the focus of excavation n ear the tunnel tip may be the primary force in selecting where branch tunnels arise. We hypothesize that the orientation of potential termite excavators towards the tunnel tip and the formation of queues make excavation from depressions that are close to the tip more likely than Termites are known to create hollows through wood that appear similar to tunnels excavated through soil (D elaplane and Laf age way into a wood source, producing tunnel shaped artifacts, but they may excavate tunnels in wood following the same rules used in soil excavation. Tunnelling through a food source may pose additional challenges to a scent mediated system, because of the blending of stigm ergic and gustatory cues. If labor is organized by physical interactions between termites and their nestmates or the tunnel walls, then the physical nature of the substrate being excavated may not affect the process beyond the manner in which they contrib ute to or alleviate the formation of queues. Density of soil may cause delays due to the difficulty of loading dense substrate, clay for example (Tucker
18 et al. 2004). W e hypothesize that tunnels through a cellulose food source arise not from the substr previously outlined for tunnel creation in soil. If tunnelling through a food source follows the same mechanics as soil excavation, then we can expect tailings of wood will be dispersed t hroughout the tunnel complex. Wood fragment sequestered in this manner represent s a resource that can be exploited by the termites later, and create s a niche for microbes. S oil Basidiomycota may opportunistically invade this microhabitat of wood fragmen ts and produce a system which shares many elements of the type of fungal agriculture seen in higher termites (Aanen and Eggleton, 2005) Thus, the mechanics that evolved to excavate soil may, when applied to a wooden substrate, serve as a preadaptation t o the evolution of the cultivation of fungi by termites.
19 CHAPTER 2 E XCAVATION SITE SELECTION Introduction The overall morphology of termite tunnel complexes has been quantified (Reinhard et al 1997, Pitts Singer and Forschler 2000, Campora and Grace 2001, Su and Puche 2003), but very little attention has been paid to the manner in which individual termites coordinate their digging efforts. Nest construction might be expected to share common principles with tunnel excavation and has been exa mined in a few highly derived termite species (Grasse 1959, Bruinisma 1979, and Jones 1979). Based on these few empirical studies, self organized nest building in termites has been extensively modeled (Courtois and Heymans 1991, Deneubourg and F ranks 1995 ool e et al 2003). Grasse (1959) proposed the concept of stigmergy, an indirect mode of communication between termites whereby the work product of a builder acts to guide the response of subsequent workers, to account for the app arent coordination of effort. A common feature of all these models of termite construction is the employment of various releaser pheromones to mark sites or pathways and to identify the work product of construction to induce stigmergic feed back loops. Termites move large particles by gripping them with their mandibles, but in digging parcels of soil the mandibles play a secondary r ole. The spread mandibles, along with the labrum and labium, form a cup for holdin g soil, while the maxillae move to load a nd pack soil into this cavity, with the maxillae alone having extensive contact with the hollow created by excavation (Li 2006). This process of removing parcels of soil
20 from the working face at the tunnel tip provides far less opportunity for pheromonal marking of the site of labor than the placing of parcels of soil, which were held for some time in the mouth and theoretically infused with exudate before being used in nest building. Even if we assume pheromone labeling occurs, the system in excavation m ust be more complex than in nest or gallery construction, for if a digging site is labeled to attract excavation, parcels of removed soil coated with this marker should not induce digging wherever they are dropped. Bonabeau et al. (1998 ) touched upon the n otion of an asymmetrical flux of uction. Jones (1979 Nasutitermes costalis (Holmgren). This behavior arose from direct int eraction between termites and involved termites that could not complete their tasks because of traffic conditions that stopped them for a given period and kept them from taking action. Recent models have taken into account halting times, as well as some e rgonomic factors such as crowding and the e et al. 1999, 2003). Contact between individuals has been shown to regulate traffic flow along foraging trails in ants in a self organizing manner (Couzin and Franks 2003). Our investigation of termite excavation breaks with previous theories of construction. Rather than begin with the assumption that tunneling is governed by the pheromone gradients and markers associated with stigmergic communication, we hypothesize that ph eromones play no direct role in the mechanics of excavation, relegating their role to trail marking and recruiting the labor force to tunnels. Along with possible tactile stigmergic signals generated by labor, we looked to the similar
21 dimensions of the bo dies of termite excavators, interactions between individuals, and flow is essentially bidirectional in tunnels that radiate from a common source, either a chamber or bran ch node. This provides common information to termites in the form of end points where termite movement is blocked or turning is required. Variation in the flux of termites at any point in the tunnel also changes the tunnel environment, with worker worker interactions increasing in frequency, limiting movement and access to tunnel walls. Herein we analyzed the process of tunnel excavation in the Formosan subterranean termite, C. formosanus at the level of the individual termite and described the behavior of termites at the digging face and in queues awaiting access. We demonstrated that a combination of bidirectional traffic flow in tunnels and local flux are sufficient to regulate the width of burgeoning tunnels. Methods Termites were collected from a C formosanus colony as described by Su and Scheffrahn (1986) in the summer of 2006. They were stored at 27 2C in plastic boxes with thin, moist wood chips for no more than 30 days prior to experimentation. The experimental arenas are similar to those used by Lee et al. (2008 a ). They consisted of three layers of clear acryl, top and bottom layers (13 13 cm 5 mm), and a 2 mm thick spacer in the middle to create a circular space (10 cm in diameter) in between (Fig. 2 1a). The enclosed space of the a rena was filled with sand (150 500 weight (Fig. 2 1b). At the outer edge of the top layer, a 1.5 cm diameter hole was drilled through it and the middle spacer. A lidded c ylinder (3 cm x 2 cm) was placed over this
22 hole to act as a termite release chamber (Fig. 2 1c). The chamber was provisioned with small slivers of pine wood and connected to the sand of the arena via a 1 cm x 4 mm tunnel formed by a cut out in the inner sp acer (Fig. 2 1d). The arenas were placed horizontally on a raised platform and tunneling activity was recorded at room temperature (26C) with a video camcorder ( Sony GR D22U, Tokyo, Japan ) mounted below the arena. Many factors potentially could influenc e the number of excavators entering tunnels, such as the motivational state and health of the termites and recruitment to the tunnel in response to soil environmental factors, but the most easily manipulated is the number of termites in the arenas. Two tr eatments were employed to create two distinct levels of flux in tunnels. To approximate low flux, 30 termites (27 workers and 3 soldiers) were introduced into the arenas, while 130 termites (117 workers and 13 soldiers) were used to produce high flux cond itions. We ran five trials of each treatment, where tunneling was recorded and data was transcribed for a 60 min period following an initial tunnel distance of 1 cm. From these recordings we measured variables including tunnel dimensions, the number of termites and the location of excavation (Fig. 2 2). Excavation could take place at the tunnel tip or occur when termites in a queue behind the individual at the tunnel tip take soil from the tunnel walls. The taking of soil from tunnel walls by termites By definition the width of the tunnel at its tip is not due to lateral excavation, whereas any widening that occurs at a distance from the tip results from lateral excavation. To determine the effect of lateral excavation on tunnel width, we compared
23 the width of tunnels every 10 min for 60 min at 0.0 (the tunnel tip), 0.5, and 1 cm distance from the tip to the nearest 0.25 mm. Because of heteroscedasticity and departures from normality in these data, we used the Kruskal Wallace test (Chi square approximation) to detect significant difference among means and then grouped by Dunn's post test at = 0.05. The mean tunnel widths for both high and low flux treatments at 1.0 cm, its expected max imum, was compared using a t test at = 0.05. For each 10 minute interval for both high and low flux treatments, we also measured the change in tunnel length during the interval. This elongation distance over time provided a rate of tunnel extension. We compared the rate of tunnel elongation between high and low flux treatments with Chi square analysis at = 0.05. In order to determine if the queue length is proportional to the total number of termite placed in arenas, we recorded the length of queues for both high and low flux treatments. Queue length was determined by counting the number of termites in mutual contact at the tunnel tip at 5 min intervals over 60 min. This resulted in 12 replicates for each treatment. The mean number of individuals at the tunnel tip, including both the foremost excavator and those in line behind, for both treatments was compared using a t test at = 0.05. Further variables associated with tunneling were needed to elucidate the mechanics of the process, but we did not seek to conduct a comparison between high and low flux treatments. These were derived solely from the videos of the high flux trea tment. We examined the average number of times that termites swept their antennae across the tunnel wall at the digging face prior to excavation, the average time (sec) spent at the digging face during the digging process, and the time (sec) spent in
24 the q ueue by those termites that excavate laterally prior to digging. We also determined the number of instances in which a termite excavated laterally prior to entering the queue at the tunnel tip. the queue limits the location of any la teral excavation it carries out to the area around its head. In the tight packed conditions at the tunnel tip, the size and shape of other termites may determine how close to the tunnel tip a potential excavator in a queue can progress. If there is a sys tem of scent marking for excavation sites, we would not expect the peaks of excavation to match those predicted by the limitation of access due to packing. We analyzed the first half hour of each of the five high flux recordings to determine the distribut ion of the distances of lateral excavation sites from tunnel tip. All digging events that took place within 1 cm from the digging face at the tunnel tip were measured to the nearest 1 mm. The number of excavation events occurring in each of these ten int ervals of 1 mm width was compared by ANOVA at = 0.05 (SAS Institute 1985) Results Tunneling Termites moved consistently from the tunnel mouth to the tunnel tip, antennating the tunnel walls as they passed (Fig. 2 3 A). In some rare instances (6.7 %, n = 113 excavation events), individuals stopped and excavated laterally prior to reaching the queue at the tunnel tip. This excavation appeared to be in response to a feature (Fig. 2 3 B, a) of the relief of the tunnel wall, a depression or a bulge (Lee et al. 2006). Upon arriving at the end of the tun nel (Fig. 2 3 C, a), if the digging face was unoccupied, they antennated the soil an average of 2.48 1.0 (mean SD, n = 25)
25 times before removing soil. Each antennation event consisted of sweeping the antennae inwards and outwards across the soil surfa ce. Rhienhard et al. (1997) be the primary sense in use. In contrast excavating termites pressed their moniliform (bead like) antennae into the soil until they deformed to match the relief of the tunnel surface (Fig. 2 4). There was high variance in the time it took for individuals to complete the digging process, which included antennating the tunnel walls prior to digging, loading their mouthparts with soil, and leavi ng the digging face (14.36 7.65 seconds, n = 50). The time increased if the termite did not fill its mouthparts with soil from one site, but moved between locations at the digging face, antennating and taking small portions of soil from each. A common s ource of delay was the impenetrability of the queue behind the foremost excavator. Often the queue had to unpack in order to give a loaded excavator a path to retreat through. When the digging face was occupied, termites formed a queue behind the forem ost excavator (Fig. 2 3 C, 2). While in queue, they pushed against the termites in front of them and jostled those beside them in an attempt to move forward. They often performed horizontal oscillatory movements (Stuart 1967), thought to have a communica tive function, and in rare cases would grip the abdomen or rear leg of termites ahead of them and pull them out of line. Individuals in queue could also turn to the tunnel wall at their current position and excavate laterally (Fig. 2 3 C, a). For those i ndividuals that excavated laterally, they did so after an average wait time of 7.005 4.207 ( n = 50) seconds.
26 After excavating a parcel of soil, termites walked backwards or turned around in the tunnel and moved back towards the tunnel mouth. They eit her unloaded beyond the mouth of the tunnel or pressed the parcels of soil against the tunnel wall, anywhere from the tip to the mouth (Fig. 2 3 D). The distance they traveled prior to unloading the excavated soil determined the length of their round trip and influenced the time between excavation events for each individual. Comparison O f T unnel W idths Any increase in the width of a tunnel beyond that seen at the tunnel tip must have been due to lateral excavation along the tunnel wall. The data from the high flux treatment showed a wedge shaped propagation of the tunnel because lateral excavation by those in queue widened the tunnel back from the tunnel tip (Fig. 2 5). There was a clear increase in tunnel width from the digging face at the tip to a dista nce of 1.0 cm at all time intervals at high flux (Table 2 1). The data at 0.5 cm shows the trend towards widening, but the values were intermediate and not significantly different from both those at the tip and at 1.0 cm. The low flux treatment data sh ows a more complex pattern, which changed over time (Table 2 1). At time 0, when the excavated tunnel is 1.0 cm in length, no significant difference in width at the various distances was recorded. As the tunnel elongated, there was a trend toward widenin g at the 0.5 cm distance, while at 1.0 cm the tunnel was narrower. The apparent temporal structure of the low flux treatment data on the tunnel wall where the 1.0 cm v alue was recorded is roughly the same place the 0.5 cm value was measured 2 3 time intervals earlier. The tunnel did not narrow
27 between the two distances. What occurred is that the whole tunnel gradually became wider, with the time delay explaining the d ifference in widths. The width of the digging faces at the tunnel tip for both high and low flux access to one excavator at a time. The mean widths of tunnels at 1.0 cm were significantly diff erent between high and low treatments ( t = 18.731, df: 128, n = 65, P < 0.0001). There was no significant difference ( P = 0.12, n = 35) in the rate of tunnel elongation (mm/hr) between high (15.3 1.4) and low flux (11.8 0.3) treatments. The similar r ates of tunnel elongation between treatments were due to this limitation of access to the digging face, for at the level of flux in both treatments there was an almost constant supply of workers at the tunnel tip. The average length of queues (number of individuals in line awaiting access to the digging face) differed significantly ( t = 6.2; df: 102; P < 0.0001, n = 35), between the high flux (7.2 0.2) and the low flux (5.1 0.3). Distribution O f L ateral E xcavation D istances The distribution of late ral excavation sites within 1 cm from the tunnel tip is shown in Fig. 2 6 A. From a minimum value at 1 mm, the occurrence of lateral excavation rose to a significant peak at a distance of 3 mm, and then tapered to a second low at 7 mm. A second, smaller peak occurred at a distance of 9 mm from the tunnel tip. Interestingly, the average number of excavation events was evenly divided between those that occurred at the tunnel tip, 540 15.378, and those that occurred along the tunnel wall within 1.0 cm from 2 = 1.040, d.f. = 1, P = 0.3078).
28 Discussion We created differential queue lengths by increasing the number of termites in the width was est ablished not at the digging face at the tunnel tip, but by the lateral excavations of termites in queue along the tunnel wall. The longer queues occurring in d to wider tunnels. The distribution of lateral excavation sites re sulted from interactions between individuals that limited the access of other excavators to specific areas of the tunnel rather than the detection of pheromonal marking. Termites are slightly wider than they are high, so that when they pack in tight or pa ss in a narrow tunnel they tend to orient themselves dorsal to dorsal orientation to minimize the cross sectional area of the pair (Fig. 2 3 B). In our arenas the vertical height was limited by the 2.0 mm vertical thickness of the inner spacer which led to tunnels with elliptical cross sect ion. This is not a biological artifact of arena design, because natural tunnels show a similar vertical compression and an oval cross section (Bardunias and Su 2005). There is some variability i n the postures that the foremosan termite could assume, or could be pushed into while excavating: hump backed, body down, or unbent. These positions and the resulting density of packing could help explain the initial peak of lateral excavation events (Fig. 2 6 A, B). Where the 5 mm long termites take soil from the tunnel tip limits the posture they can assume. When digging towards one side of the digging face, they tended to assume a hump backed stance as they lowered their mouth parts to the soil, their thorax humped up a nd their abdomen down (Fig. 2 7 A). The forward progress of the second termite in queue behind an excavator at the tunnel tip in this position was limited to about 3 mm from the tunnel tip.
29 When termites at the digging face excavated directly ahead, alon g the direction of tunnel propagation, they were able to assume a posture that allowed the second termite in queue to move closer to the tunnel tip ( Fig. 2 7 B, 2 3 B 5 mm long body, access is limited to that distance from that tip. Because the second individuals in queue have probably been waiting the longes t and are most likely to excavate laterally along the side wall, they cause the peak of lateral excavation over the range of distances determined by how close to the tunnel tip the position of the termite in front of them allowed them to approach. The 5 mm length of termites makes 4 5 mm a likely distance for the third termite in queue to excavate (Fig. 2 6 B). There is a zone between 6 and 7 mm where access is strictly limited by the bodies of other termites in a tight y packed queue, followed by an increase in excavation events further from the tip (Fig. 2 6 B). Our findings, while not directly disproving pheromonal labeling, allow us to propose an excavation process that is governed solely by tactile interactions (Fig. 2 8). Termite tunnel geo metry limited travel to two directions, out to the tip or back to the origin Individual termites travelled down tunnels (Fig. 2 3 A) until they were halted by 2 excavati on at the digging face and the surrounding tunnel wall with no need for specific pheromone labeling of digging sites. As termites travelled down the tunnels, they antennated the wall surface (Fig. 2 feature (Fi g. 2 3 B), an outcrop or a depression, if it was sufficiently stimulating to release digging behavior (Fig. 2 8
30 whole antennae when exploring the wall, rather than just the antennal tips, may indicate t hat tactile examination of the digging face plays an important role in the choice of digging site. Chemoreceptors are primarily found at the distal segments of C. formosanus antennae, while mechanoreceptors exist on all segments (Tarumingkeng et al. 1976) Because they had yet to dig and were unburdened (Fig. 2 8 they excavated laterally (Fig. 2 3 C, b) from the tunnel wall at their current position (Fig. 2 2 8 were carrying soil in their mouthparts (Fig. 2 8 load within the tunnel (Fig. 2 3 D) or beyond the tunnel mouth (Fig. 2 8 invert their heading to return to the tunnel tip for more soil (Fig. 2 If a feature failed to release digging behavior (Fig. 2 8 they continued on to the tunnel tip. If they reached the tunnel tip and the digging face was occupied, their forward progress blocked by one or more termites (Fig. 2 3 C), they then formed a queue of termites (Fig. 2 2 8 up soil (Fig. 2 8 excavated laterally (Fig. 2 sufficient to release digging behavior (Fig. 2 they reach ed the tunnel tip (Fig. 2 2 8 Many factors influence d the number of termites at the tunnel tip. The time spent by the foremost excavators at the digging face to choose a site and take on a load of soil defined the minimum delay of those awaiting access. The length of tunnel and the
31 walking speed of termites produced the round trip time for individual excavators. Longer trip times are expected to reduce pressure because termites will be spending less of their total excavation and travel time in queue at the tunnel tip. The geometry of the tunnel may cause traffic delays or congestion, while any features along tunnel walls could cause the termite to stop and antennate them (Lee et al. 2006). Lateral excavatio n from the tunnel wall was directly proportional to queue length. expands laterally in response to an increase in the number of excavators is analogous to the effect of in creasing pressure in a closed system. This expansion just behind the tunnel tip sets the width of expanding tunnels and leads to the characteristic wedge shaped advance of tunnel tips. Nowhere else will the digging pressure be as high as in a stretch of tunnel occupied by a queue, such as the tunnel tip or a zone of traffic congestion. In longer queues, individuals can be flanked on both sides by fellow excavators and denied access to the tunnel walls. Although the area of any stretch of tunnel may inc rease as it widens, the linear distance of tunnel wall from which soil may be taken remains constant. The length of a queue is far more important than the density of termites in determining digging pressure, for only those termites that have access to soi l can excavate and contribute to the widening. Lateral excavation could have resulted from interplay between conflicting forces: if there was an urge to reach the tunnel tip, it may have waned with time spent in the queue and been eclipsed by a rising ur ge to excavate. If the termites were responding to a specific geometry they associate with a tunnel tip (Lee et al. 2008 b ), they may have
32 responded to the constriction formed by the tunnel wall on one side and the body of a fellow excavator on the other a s a cue for excavation (Fig. 2 7, dashed line). Waiting time would be less important in this case than positioning in determining the likelihood of excavation. The delay in queue also increased the time of exposure of an excavator to a feature along the tunnel wall and would have increased the chances of generating a response, so there could be an interaction between queuing and tunnel geometry. There is another factor, only touched upon in this chapter which underlies the excavation process. Soil, onc e excavated, must be dropped at a remote location. The manner and distribution of the placement of excavated soil along tunnel walls and the environmental cues that release dropping behavior or excavation away from the queue at the tunnel tip will be addr essed in a separate chapter Digging pressure can produce tunnel features beyond initial width. The formation of queues due to the impeded flow of termites at any point in the tunnel, such as intersections or constrictions, will lead to widening. This w ill scale the tunnel width to traffic flow. The termite tunneling system relies on the existence of starting and ending points to coordinate labor. As a tunnel widens with increased pressure, the width of the digging face at the tunnel tip expands beyon d a width that allows one or two workers simultaneous access. When the digging face broadens, the location of the tip of the tunnel becomes unfocussed and fragments into multiple discrete digging sites. Multiple tunnel tip formation is the precursor to b ifurcation and branching, and the emergence of branching and tunnel branch patterns, from simple variations in flux and directed excavation.
33 Table 2 1. A comparison of tunnel width (mm S.E.M.) at the digging face at the tip and at two distances from th e tip Treatment Time (min) Distance from tunnel tip (cm) 0 0.5 1 0 1.9 0.1 a 2.9 0.1 ab 3.7 0.2 b 10 1.9 0.1 a 3.1 0.1 ab 3.7 0.2 b High flux 20 1.9 0.1 a 3.0 0.2 ab 3.5 0.1 b 30 2.0 0.1 a 3.0 0.2 ab 3.4 0.1 b 40 1.7 0.1 a 2.9 0.2 ab 3.2 0.1 b 50 1.7 0.1 a 2.9 0.1 ab 3.2 0.1 b 60 1.8 0.1 a 3.0 0.2 ab 3.4 0.1 b 0 1.6 0.1 a 1.5 0.1 a 2.0 0.1 a 10 1.2 0.1 a 1.6 0.1 ab 1.9 0.1 b 20 1.3 0.1 a 1.9 0.1 b 1.8 0.1 ab Lo w flux 30 1.6 0.1 a 2.4 0.1 b 1.8 0.2 ab 40 1.8 0.1 a 2.8 0.1 b 2.4 0.2 ab 50 1.8 0.1 a 2.9 0.1 b 2.4 0.1 ab 60 1.8 0.1 a 2.9 0.1 b 2.8 0.1 b V alues are the means of five replicates. Means followed by the same let ter in a row are not significantly different according to a Kruskal Wallis test and then grouped by Dunn's p
34 Figure 2 1 Two dimensional arena design A) These consist of two clear 5 mm thick acryl plates and a 2 mm thick central spacer between them B) A n opened central space filled with sand C) A cylindric al introductory chamber D) connected to the sand filled portion of the arena through a 1 cm 4 mm channel
35 Figure 2 2 A typic al queue at the tip of a tunnel E ach termite in queue is numbered successively (1 5), and arrows indicate the sites of excavation. Termite (1) is excavating at the digging face at the tip of the tunnel, indicated by the extent of the dotted line. The white dashed lines indicate the tunnel width at successive distances from the digging fa ce: 0.0, 0.5, and 1.0 cm.
36 Figure 2 3 Termites travel ling down a tunnel. A) A termite antennating the tunnel wall. B ) The back to back posture is assumed as they pass in a narrow tunnel; a feature, a, in this case a bulge in the tunnel w all that blocks passage, may stimulate excavation. C ) Termites at the tunnel tip. Termite 1 excavates from the digging face, a, while termite 2 excavates laterally, b. D ) A termite drops a load of soil by pressing it against the tunnel wall, a.
37 Fig u re 2 4 Antennation of the digging face at the tunnel tip A) Antennating the tunnel tip B) The moniliform (bead like) antennae deform extensively to match the relief of the wall profiles, indicated by the dotted lines. C) S ubsequent excavation Figure 2 5 Tunnel propagation at three time intervals Time intervals t 1, t 2, and t 3, traced from a high flux tunnel, showing the characteristic wedge shaped advance of the tunnel due to lateral excavation behind the digging face at the tunnel ti p.
38 F igure 2 6 The distribution of the distances of excavation events A) The mean dist ances ( S D ) from the tunnel tip within 1.0 cm compared by ANOVA (F = 25.2, df = 9, 40, P < 0.0001). Columns demarcated by the same letter are not significantly different. B ) Ind ividuals in queue at the tunnel tip and excavating laterally. The scale below is equivalent The dashed line indicates the relative position of the tunnel tip in both A and B. Each termite in queue is numbered, (1 6), and the arrows indicat e sites of excavation for direct comparison to the peaks of excavation in A.
39 Figure 2 7 The characteristic posture s of a termite digging towards the lateral edges of the tunnel tip. A) The dotted line shows the shape the body assumes, wh ile the dashed lines indicate a geometry that a termite might react to by excavating. B ) T ahead, along the direction of tunnel elongation.
40 Figure 2 8 A model of termite excavation based on the bi direction al traffic flow in tunnels and the responses of individual termites to tunnel features and traffic congestion Prospective excavators Move towards the tip of the tunnel. Along the way they Antennate the tunnel wall and may react to a Feature, depression or bulge, in the tunnel wall. If not Loaded, they will Excavate and Invert their movement vector. In most cases they continue to move down the
41 tunnel to the tip where they may form a Queue with other excavators awaiting access to the tunnel tip. If thei r Delay in queue reaches a critical level, or if they are able to reach the Tunnel end, and they are not Loaded, they will Excavate and Invert their movement vector. The termites head towards the mouth of the tunnel and, because they are Loaded, they Drop the soil and complete the cycle.
42 CHAPTER 3 BRANCH FORMATION Introduction Many species excavate extensive tunnel complexes for habitation and in the course of foraging for resources. The mechanics of tunnel excavation by groups of social insects is not well understood, but heterogeniety in the subterranean environment influences the tunneling behavior of both termites and ants. Variation in soil featu res, such as temperature (Arab and Costa Leonardo 2005 ), soil density (Tucker et al. 2004, Nobre et al. 2007), gaps or objects in the s oil (Evans 2003), moisture (Su and Puche 2003, Mikheyev and Tschinkel 2004), and chemosensory gradients (Tschinkel 2004, Su 2005) have been demonstrated to influence the tunneling behavio r of either termites, ants, or both. One feature common to ant and termite excavation systems is that digging behaviour has been shown to be stimulated by the presence of small depressions in the tunn el wall (Sudd 1970, Bardunias and Su 2009a), also termed (Lee et al. 2008b ) depressions occur along the walls of a tunnel, excavation at the sites leads to branch formation. The reaction of C formosanus to depressions in the tunnel wall of a range of sizes was examined by Lee et al. (2008b). In that study, termites were presented with an artificial tunnel, a pre excavated channel in the sand of the arena, stretching between two empty chambers. Only a specific size range of depressions, placed at the mid point of the tunnel, stimulated excavation. For example, 3 mm wide by 3 mm deep depressions in the tunnel wall of a 3 mm wide tunnel were not excavated into branches, while 5 mm by 3 mm depression s were (Lee et al. 2008b). This suggested that the determining factor in whether a depression induced excavation behavio r was the e ase
43 with which termites could turn into the opening of the depressions based on the dimensions of both depression and tunnel. Sudd (1970) showed that single workers of Lasius niger (L) and Formica lemani Bondroit, when isolated f rom fellow colony members, excavate d at the location of depressions in tunnel walls. In his study, roughly half of all excavation events for both species occurred along the lateral wall of a tunnel, presumably at the site of a pe rceived depression, instead of at the tunnel tip, though in F. lemani excavation events were less tip. Sudd (1971) added a second ant to form pairs of excavating ant s, but this did not make digging at the tunnel tip more likely. Bardunias and Su (2010) described potential excavators moving down tunnels to along the tunnel wall i n the manner described by Sudd (1970, 1971). This may have been the result of the type of group interactions shown to organize lane formation from traffic flow in ants (Couzin and Franks 2003, Dussutour et al. 2004) with physical encounters between termit es reducing the likelihood of individuals stopping along the way. If this is true then single individuals, and perhaps pairs, excavating in isolation from other colony mates wi ll not display the same behavio r as individuals excavating in groups. Lee et al (2008 b), by placing the depression midway along a tunnel between two chambers, was unable to include two processes that might be crucial to tunneling in no clear end of the tunnel and no formation of a queue of excavators near the tunnel tip.
44 (Buhl et al. 2005 ) where high intersection rates and convergence effectiv ely eliminated many tunn el tips. Secondly, because there was no tip, there was no movement of termites carrying excavated soil past the site of the depression. When termites excavate soil from the tip of tunnels, it must be deposited in a manner that does not interfere with the growth of the tunn el (Li and Su 2008). Bardunias and Su (2009b) showed that the soil that was excavated could be deposited back along the length of the tunnel within depressions in the tunnel wall, effectively closing them over. We hypothesize that the orientation of potential termite excavators towards the tunnel tip and the formation of queues make excavation from depressions that are close examined depressions along the wall of a stretch of tunnel that is subject to the bi directional traffic flow of multiple termites and the movement of soil excavated at the tunnel tip back along the length of the tunnel. In a series of experiments we investigated the role o f the size and location of depressions on the probability of branch formation. Materials A nd Methods Individuals of C. formosanus were collected from three field colonies in Broward County, FL, by using the methodology of Su and Scheffrahn (1986). They we re stored at 27 2C in plastic boxes with thin, moist wood chips for no more than 30 d prior to experimentation. Horizontal arenas for the following experiments consisted of two 9 by 7.5 cm 2 sheets of transparent Plexiglas with a 0.2 cm inner Plexiglas spacer (Fig. 3 1 a) surrounding an elongated hexagonal (6.5 cm width, 7.5 cm height) sand filled space
45 (Fig. 3 1 b ). The sifted sand (150 moistened with deionized water at al (Nalge Nunc In ternational, Rochester, NY) were used as an introduction chamber (Fig. 3 1c, 3 cm diameter x 4 cm height). This was connected to the arena through a 1 cm length of op aque polyethylene tubing (Fig. 3 1 d, 4 mm diameter, Watts, Andover, MA). The arena was bisected by a 3 mm wide channel made by removing sand during arena construction to simulate a tunnel excavated by termites (Fig. 3 1e). Along this simulated tunnel, at distances that varied by experiment, we carved a depression of varying dimensions into the tunnel wall in imitation of a natural depression that could lead to a newly forming branch (Fig. 3 1 f). A short length of op aque polyethylene tubing (Fig. 3 1 g ) 4 mm inner diameter, Watts, Andover, MA) was joined to the arena at the te rminus of the simulated tunnel opposite the introduction chamber on one side and to a 10 cm length o f clear vinyl tubing (Fig. 3 1 h, 5 mm inner diameter, Watts, Andover, MA) on the other. Both of these tubes were filled completely with blue sand (Activa P roducts, Inc., Marshall, TX) to facilitate the observation of deposition. The introduction chamber was loaded with a (1 by 1 by 2 cm 3 ) piece of wood as a food source and 40 rd instar, and 4 soldiers), and experimental trials were immediately conducted. Upon entering the arenas, excavators were allowed to travel through the 3 mm wide channel excavated in the s oil of the arena that simulated a tunnel (Fig. 3 1 e), enc ount ering the depression (Fig. 3 1 f) whose dimensions and placement along the tunnel wall varied in different experiments. Upon reaching the tip of the simulated tunnel, termites were able to dig fr om the blue coloured sand in the adjoining tube (Fig.
46 3 1 g) and bring the blue sand back down along the tunnel. The arenas were placed horizontally on a raised platform and tunneling activity was recorded at room temperature (26 C) for 60 min with a vide o camcorder ( Sony GR D22U, Tokyo, Japan ) mounted below the arena Recording was done under the illumination of the overhead fluorescent lighting of the laboratory. Most excavation within tunnels occurs at the digging face at the ti p of the tunnel (Barduni as and Su 2010), but in this study we recorded only the number of excavation and deposition events occurring at the site of the experimental depression in the tunnel wall over a period of 60 min. Fifteen replicates were conducted for each experiment, 5 fr om each of the 3 colonies. The numbers of soil parcels removed from or deposited into each depression were compared via a Mann Whitney U test due to departures from normality in the data. Experiment 1: Proximity to T unnel T ip In this experiment, we soug ht to determine the distance from the end of the main tunnel, where excavators form queues awaiting access to the digging face at the excavation and deposition within a depres sion. A depression in the tunnel wall was created at one of three distances, 0.5 cm, 1.0 cm, or 6.0 cm, from the digging face at the tip of the tunnel provided by the tube filled with blue sand. The depression was 3 mm wide, matching the width of the mai n tunnel, and 3 mm deep. In a previous study (Lee 2008b) depressions of these dimensions in the absence of a clear tunnel tip did not result in branch formation. Experiment 2: Si ze O f D epression The formation of a branch from an initial depression is gov erned by competition between soil being removed and incoming soil bein g deposited (Bardunias and Su
47 2009a). In this context, we sought to determine if a larger initial depression size, requiring more incoming sand to fill, would give an advantage to the e xcavators and make branch formation more likely. This experiment was conducted as above at distances of either 0.5 cm or 1.0 cm, but the width of the depression was increased to 5 mm, wider than the main tunnel, while the depth rema ined 3 mm. In a previo us study (Lee 2008b) depressions of these dimensions resulted in branch formation. The depression at a distance of 6.0 cm in experiment 1 did not induce either enough excavation to form branches or sufficient deposition to fill them in the previous exper iment, thus we did not seek to use depressions at that distance for a comparison of the effect of depression size on branching. Experimental R esults Depressions (Fig. 3 2 A) were either filled over completely b y the deposition of soil (Fig. 3 2 B), or succ e ssfully formed branches (Fig. 3 2 C). The mean numbers of excavation and deposition events for each treatment are presented in Table 3 1 along with a notation on whether the depressions were excavated to form branches by the end of the 1 h recording period A branch was noted as forming if the termites had extended the depression beyond its initial dimensions. Experiment 1: Proximity to T unnel T ip Depressions (3 mm by 3 mm) located at 0.5 cm from the distal end of the tunnel invariably formed branches (Tab le 3 1). Branch formation was reflected by the manner in which the number of excavation events significantly (P < 0.001) outweighed deposition at the same site. When the distance of the depression from the digging face at the tip of the main tunnel was d oubled to 1 cm depressions were completely filled in all but one trial (Table 3 1). The mean number of excavation events when the same
48 sized depressions were placed at a distance of 0.5 cm and 1 cm were significantly different (P < 0.001), as were the mea n number of deposition events (Table 3 2). Depressions placed at greater distance, 6 cm, from the distal end of the main tunnel did not form branches within the 60 min time frame of the experiment, and received a reduced level of deposition, insufficient to completely fill the opened space (Table 3 1) Experiment 2: Size O f D epression At the 0.5 cm distance, both the 3 mm by 3 mm and 5 mm by 3mm depressions invariably formed branches (Tab le 3 1). At a distance of 1.0 cm one of the 3 mm by 3 mm depressio n s formed branches, while the wider, 5 mm x 3 mm depressions successfully elongated into branches in 4 of 15 instances. The larger sized depressions attracted more excavation and deposition, but required more sand to be deposited to completely fill them i n. Discussion The likelihood of a depression being excavated into a branch was not a simple function of depression size. Depressions of a size that had not stimulated excavation in previous studies (Lee 2008b) invariably did so when they were placed in cl ose proximity, 0.5 cm, to the tunnel tip. This distance is roughly the b ody length of a termite (Fig. 3 3 ), so that when termites formed a queue behind an individual digging at the tunnel tip (Bardunias and Su 2010), some of them were adjacent to this dep ression. Our findings suggest that the presence of termites waiting next to the depression made their excavation at the site more likely. Because these depressions were within the stretch of tunnel generally occupied by a queue of excavators awaiting acc ess to the tunnel tip, they may form through the same process as a bifurcation of the tip itself.
49 Such bifurcations are much more common than branches that arise from tunnel walls (Bardunias and Su 2009a). The relative paucity of excavation at depressions at the further distances from the tunnel tip (1.0 cm and 6.0 cm) indicates that termites tended to walk past the same size depression (3 mm by 3 mm) without stopping to excavate when it was not near the tip. This focus of excavation near the tunnel tip i s counter to what was reported by Sudd (1970), where isolated, individual ants in his study readily excavated along tunnel walls. Reduced excavation at depressions remote from the tunnel tip gave an advantage to deposition, which smoothed tunnel walls, th us eliminating cues that could have led to inefficient over branching. Li and Su (2008) suggested that the relatively small amount of space needed for the deposition of the excavated soil in a tunnel system could be accounted for if soil were placed in t he voids created by consuming wood. The deposition of soil along tunnel walls is another means of conserving space while sequestering excess soil. An increased number of excavators near the tunnel tip led to the widening of tunnels by termites excavating from tunnel walls (Bardunias and Su 2010). If traffic flow later decreases, then the tunnel will be wider than necessary and sand may be deposited along the walls without hindering the movement of termites. Variations in the flow of excavators at the ti p as the tunnel elongates may result in tunnels that widen and narrow periodically, rather than extend at a uniform width. Bardunias and Su ( 2009b ) suggested that small stretches of wide tunnel caused by pulses of excavation from the tunnel walls by term ites waiting in longer queues, flanked by narrower sections, appear as depressions of the type used in this study. The effect of sand deposition in
50 depressions or at widened sections of a tunnel, such as at intersections, is to smooth tunnel walls and sca le tunnel width to traffic flow. A result of the competition between excavation and deposition is that tunnels with little traffic flow may be completely filled in (Su and Le e 2009). In this way the tunnel pattern may be optimized by reducing the number o f tunnels that may be dead ends or simply longer paths to a destination for which shorter routes are available. Our findings act as a reminder to those who would model excavation that group behavi o rs by social insect colonies cannot be understood by viewin g digging individuals in isolation from interaction with other colony members, and as a caution against oversimplifying the agents used. In this system, the same stimulus, depressions, elicited either of two responses, based on the behavioural state of th e termite and the context in which it encountered the cue. We further suggest that simply modelling excavation upon the pooled probability of behaviours derived from haphazard sampling obscures variations in likelihood that are tied to localized heterogen eity in the environm ental cues that release behavio rs. An accurate representation of an heading within tunnels must be a feature of future models.
51 Table 3 1. The mean num ber of excavation and deposition events ( SD) and the number of trials for which branches formed out of 15 replicates Experiment n Dist. 1 Size 2 Excavation Deposition Branches 1: Proximity 15 0.5 cm 3 x 3 mm 35.66 5.23a 21.53 1.84b 15 15 1.0 cm 3 x 3 mm 3.66 6.58a 34.4 2.16b 1 15 6.0 cm 3 x 3 mm 1.25 1.28a 11.08 0.9b 0 2: Size 15 0.5 cm 5 x 3 mm 52.4 8.91a 33.66 3.75b 15 15 1.0 cm 5 x 3 mm 20.53 18.41a 65.73 21.22b 4 M eans followed by different letters within a row are significan tly differ Whitney U test ). 1 Distance was measured from the tip of the main tunnel. 2 Depression size was measured as width x depth
52 Table 3 2 The mean number of excavation and deposition events ( SD) Distance 1 Size 2 n 0.5 cm 1.0 cm 3 x 3 mm Excavation 15 35.66 5.23a 3.66 6.58 b Deposition 15 21.53 1.84 a 34.4 2.16b 5 x 3 mm Excavation 15 52.4 8.91a 20.53 18.41 b Deposition 15 33.66 3.75 a 65.73 21.22b M eans followed by different letters wit hin a row are significantly differ Whitney U test ). 1 Distance was measured from the tip of the main tunnel. 2 Depression size was measured as width x depth
53 Figure 3 1 Experimental arenas A) T wo sheets of transparent Plexiglas with an inner Plexiglas spacer B) A n elongated hexagonal sand fille d space C) An introduction chamber D) A 1 cm length of polyethylene tubing connected to the arena via a partially obscured by its insertion into the chamber and the chamber lid. E) Th e arena was bisected by a 3 mm wide simulated tunnel, a channel dug in t he sand of the arena F) A pre cut depression that varied in geometr y and placement by treatment G) A short l ength of polyethylene tubing filled completely with blue sand joined to the arena at the terminu s of the simulated tunnel opposite the intro duction chamber on one s ide H) A length of clear vinyl tubing filled completely with blue sand.
54 Fi gure 3 2 A 3 mm by 3 mm depression in the tunnel wall. A) P laced 1 cm from the digging face at the end of the ma in tunnel. B ) The same depression after experimentation showing that it has been completely filled by the deposition of blue sand that was excavated at the tunnel tip. C ) Shows a depression of t he same dimensions but at a distance of 0.5 cm from the tunnel tip, showing a fully formed branch.
55 Figure 3 3 The distal end of a tunnel during excavation showing termites in queue Termites labeled 1 5 and demarcations (dotted lines) a distance of 0.5 cm and 1 cm from the digging face at the tunnel tip w here depressions would have been placed.
56 CHAPTER 4 EXCAVATION THROUGH WOOD Introduction Termites in the subfamily Macrotermitinae (Termitidae) have evolved a mutualistic relationship with the fungi of the genus Termitomyces (Basidiomycota: Agaricales) ( Bignell 2006 complex habitat, and a variety of microbial s ymbioses have been described ( Visser et al. 2011), but the evolutionary origin of the garden system is poor ly understood. Eggleton ( 2006) sug gested that the evolution of the fungal garden system seen in the Macrotermitinae today originated in a Coptotermes like (Rhinotermitidae) ancestor when the fecal carton of the central mound was opportunistically invaded by soil fungi. A major challenge t o direct fungal infestation of fecal comb in the ancestor of the Macrotermitinae is the need to overcome antifungal properties of fecal carton demonstrated for Coptotermes and other Rhinotermitidae ( Hamilton et al. 2011). Where the fungus appears to have successfully invaded brood rearing portions of the nest of these termites, it was not as gradual invaders of fecal carton, but as egg mimics ( Matsuura et al. 2009 ). If Coptotermes is a valid model for the ancestral condition of the Macrotermitinae, then p erhaps an examination of their behavior may shed light on the process by which opportunistic soil fungi transitioned to living within fecal carton materials Subterranean termites encounter fungi commonly when foraging through decaying wood, and have been shown to benefit from consuming wood that h as been conditioned by fungi ( Smythe et al. 1971 ). Delaplane and LaFage (1990 ) showed that C. formosanus Shiraki
57 increas e the surface area available to feeding termites, and presumably opportunis tic fungi as well. Thus, fungi growing on the insides of tunnels within wood may be the earliest site of infestation, but there was an element overlooked in the study. The authors describe d a process by which wood b and LaFage 1990 ) after exposure to termites. Did the termites simply eat their way into the wood source, or could some of this debris have been parcels of wood excavated from th e main source? Bardunias and Su ( 2010) have described the basic mechanics of tunnel excavation through soil in the subterranean termite C. formosanus The process involves a queue of excavators that remove parcels of soil from the burgeoning ends of tunnel s. These excavators then deposit their loads remote from the site of excavation. The tailings of soil may be placed along tunnel walls or into any available voids, such as unused tunnels, and subsequently coated w ith a fecal carton envelope ( Bardunias and Su 2009 a ). If termites dig through wood according to the same mechanics described for soil excavation, then tunnels in soil near wood sources would become lined, and lengths of unused tunnels packed, with excavated wood fragments. The infestation of fung i into these pockets of masticated wood fragments deposited along tunnel walls and sealed behind an envelope of fecal carton may have served as the route by which soil fungi evolved to exploit fecal carton. In this scenario, termites did not alter their f oraging behavior to begin consuming fungi within the structural carton of their central nest site, where the reproductives are located and where brood rearing occurs. Instead, they exploited pockets of wood fragments in
58 proximity to food sources that have undergone conditioning by fungus in a manner proposed to benefit fungal gardeners ( Bignell 2006 ). The shift in behavior would be the relatively modest transition from feeding on fallen logs, in which fungus may be present, to additionally feeding on seque stered wood fragments within tunnels adjacent to logs. Once a symbiosis was established between termite and fungus, selection may have then economized the gardening process by driving the gardens into the larger, dedicated growing chambers seen in the Mac rotermitinae and closer to the central sections of the colony. In this study, we presented C. formosanus with a cellulose source in order to determine how they create tunnels within it, either by eating the material to create voids, or by excavating accord ing to the mechanics previously outlined for tunnel creation in soil. Material A nd Methods Termites were collected from three field colonies of the Formosan subterranean termite C. formosanus in Fort Lauderdale, FL using the method described by ( Su and Sc heffr ahn 1986 ) and processed via methods outlined by (Tamashiro et al., 1973) Arenas (Fig. 4 1 ) consisted of 2 sheets of PlexiglasTM (12 x 12 x 0.2 cm) with 4 0.2 cm laminates sandwiched between them in a manner that created a central void. Preliminary observation of termites presented with thin slats of wood ( Picea sp.) or paper as a cellulose source showed no apparent difference in the manner of excavation. Thus, for ease of handling, a sterile absorbent pad (45mm diam.) was placed to serve as a cellul ose source after the dry weight of the pad was obtained. The arenas were then filled with red colored sand (Activa Products, Inc., Marshall, TX),
59 moistened with deionized water 7% by sand weight, to maximize the contrast to the white cellulose pad. Along one wall, a 2 cm void was left within the arena to allow for the introduction of 60 termites (54 workers of at least 3rd instar and 6 soldiers). A small wafer (~1 x 1 cm) of wood ( Picea sp.) was placed in the arena as a supplementary food source. In 15 tr ials, termites were allowed to excavate in the arenas for 72 h. Arenas were subsequently opened and all fragments of paper separated from sand. Dry weights were determined for the remaining mass of each pad and all fragments of paper recovered from the a rena. These weights were then compared to the initial dry weight of each pad in order to determine the mass removed from the pad that could be accounted for by deposition. Any difference in the dry weight of the pad prior to the experiment and the sum of the remaining pad and paper particles recovered after the experiment might have been ingested. Statistical Analysis Dry weights of cellulose pads in each arena prior to exposure to termites were compared to the combined dry weight of the pad and pad frag ments found in the arena after exposure to termites via a paired t ( SAS 2002 ) Results The termites tunneled into the c ellulose pads extensively (Fig 4 2 ). While there was a significant difference between the dry mass of paper before an d after 72 h exposure to termites ( t (14 ) = 3.05, p= 0.0086), the mass unaccounted for by d eposition was very small (Fig 4 3 ). The cellulose pads had an average initial dry weight of 0.896 0.02 g (mean, std dev). After 72 h, an average of 0.208 0.014 g was removed from
60 the cellulose pads and deposited elsewhere in the arena. When the remnants of the pad after the experiment and the paper recovered from arenas were added together, only an average of 0.027 0.029 g of paper remained unaccounted for by deposition. Thus, the amount of paper deposited within the arena was roughly 10 fold greater than the maximum amount that might have been ingested by termites. Paper particles were deposite d throughout the tunnel complex Depositi o n occurred along tunnels (Fig 4 4 ), and often deposition was so heavy that tunnels were completely obstructed and filled in (Fig 4 5 ). Most of the paper was deposited within the void left in the arena u pon initial construction (Fig 4 6 ). The creation of a lining of feces over pap er particles was initiated with in the sho rt duration of this study Discussion Coptotermes formosanus excavates through food sources in a process that results in the deposition of masticated food particles within tunnels in a manner previ ously shown fo r e xcavated soil (Bardunias and Su 2010) Our findings should wood by comparing before and after dry mass of sound wood. Clearly, much of the mass of removed wood may not be eaten by termites but remains in the surrounding environment as deposited wood fragments within the tunnel complex In sequestering wood fragments along and within tunnels, Coptotermes creates a microbial microniche (Bignell 2006). Fecal carton infested by fungi and/ or bacteria that help to condition cellulose and lignin expelled in feces has been suggested to represent where digestive function has been exported outside of the body into the tunnel system that repre
61 phenotype. The obligate symbiosis between the Macrotermitinae and Termitomyces represents a derived condition, with no obvious outgroups representing the evolutionary stages that led to it (Aanen and Eggleton 2005 ). If soil Basidiomycota opportunistically invaded the microhabitat of fecal carton covered wood fragments deposited along C. formosanus tunnels, then the basic mechanics of tunnel excavation through soil, when applied to a food source, serve as a preadaptation to evolution of the cultivation of fungi by termites. If, on the other hand, C. formosanus is exploiting fungus and fungal conditioned wood sequestered in tunnels, then a novel form of agriculture, perhaps the first to exist between fungus and insect, ma y be described. Sanchez Pena (2005) suggested that fungiculture in Attine ants represented the acquisition of fungal strains previously domesticated by Scolytine and Platypodine beetles. Given the early derivation of Coptotermes and similar subterr anean termites relative to the other insect taxa shown to rear fungal gardens, it is possible that a facultative relationship between termite and fungus acted to select for fungal traits that facilitated the invasion of both ant and beetle galleries. Although i t is unclear exactly how far the zone of directly deposited excavated wood fragments extends, cellulose from a food source may be deposited at greater distance in the form of feces. Arquette and Rodriguez (2011) describe feces ticles that when dried become a fine powder. This is not directly deposited from excavation, but results from incomplete digestion in the gut of wooden fragments. This paste forms the substrate from which tunnel linings are formed. There is some evidence for the reingestion of this type of fecal carton lining tunnel walls in soil feeding termites, though it remains to be proved if micro bial
62 conditioning is occurring ( Bignell 2006 ) It would be interesting to compare this ca to manure their fungal gardens (Aanen and Eggleton 2005 ). Excavated wood fragments cached within the tunnel complex are sealed behind a fecal paste envelope. Deposited wood becomes sandwiched between fecal carton on one side and soil containing opportunistic fungal spores on the other. This results in a gradient of any antifungal properties of the fecal carton diffusing through the sequestered wood to the soil. Perhaps this facilitated soil basidiom ycetes in invading termite colonies because the fungi could evolve to overcome diluted defenses along this gradient, eventually becoming competent to survive in whole fecal carton.
63 Figure 4 1 Two dimensional arena design. T wo clear 5 m m thick acryl plates and a 2 mm thic k central spacer between them. A ) R ed sand B ) A n opened space for t he introduction of termites C) A piece of wood D) A steri le absorbent pad (45mm diam.) E) A preformed channel through the sand connecting the in troduction area with the pad
64 Figure 4 2 Two dimensional arena 72 h after exposure to termites. Paper particles have been deposited throughout the tunnel complex and the introduction space at the top of the arena has been filled with excavated sand and paper.
65 Figure 4 3 The dry mass of paper before and after 72 h exposure to termites The hows both the dry mass of the paper pad after the experiment in red, and the aggregate of all paper particles recovered from within the arena.
66 Figure 4 4 Paper parcels deposited along the inside of tunnels
67 Figure 4 5 Paper parcels deposited within a tunnel, completely filling it
68 Figure 4 6 Paper deposited within t he opened introduction space Note the initiation showing the initiation of a lining of feces over the pape r particles
69 LIST OF REFERENCES Aanen, D.K. and P. Eggleton. 2005. Fungus growing termites originated in African rain forest. Curr Biol 15: 851 855. Anderson,C., J.J. Boomsma, and J.J.Bartholdi III 2002. Task partitioning in insect societies: buc ket brigades Insectes S oc. 49 : 1 10. Arab, A., and A. M. Costa Leonardo. 2005. Effect of biotic and abiotic factors on the tunneling behavior of Coptotermes gestroi and Heterotermes tenuis (Isoptera: Rhinotermit idae). Behav Process 70: 32 40 Arquett e, T. J., and J. M. Rodriguez. 2011 Examination of methods for Formosan subterranean termite (Isoptera: Rhinotermitidae) feces recovery Fla. Entomol 94: 109 111 Bardunias, P. M., a nd N. Y. Su. 2005 Comparison of Tunnel Geometry of Subterranean Termi Arenas. Sociobiology 45: 679 685 Bardunias, P. M., and N. Y. Su. 2009a. Opposing headings of excavating and depositing termites facilitate branch formation in the Formosan subt err anean termite. Anim Behav 78: 755 759 Bardunias, P. M., and N. Y. Su. 2009b. Dead Reckoning in Tunnel Propagation of the Formosan subterranean termite (Isoptera: Rhinotermitida e). Ann. Entomol. Soc. Am. 102 : 158 165 Bardunias, P. M., and N. Y Su. 2010. Queue size determines the width of tunnels in the Formosan subterranean termite (Isoptera: Rhinotermitidae). J. Insect Behav. 23: 189 204 Becker, G., and W. Gerish. 1981. Geomagnetic rhythms of termite feeding activity and related rhythms. J. Inter discipl. Cycle Res. 12: 247 256. Bell, W.J., L.M. Roth, and C.A. Nalepa. 2007. Cockroaches: Ecology, Behavior, and Natural History JHU Press Bignell, D.E., Y. Roisin, and N. Lo. 2000. Termites as soil engineers. Biology of Termites: A Modern Synthesis (Bignell DE Roisin Y & Lo N, eds), 193 210. Springer Verlag, Berlin. Bignell, D.E. 2006. Termites as soil e ngineers and soil processors. 8: 183 220. Bonabeau, E., G. Theraulaz, J. L Deneubourg, N.R Franks, O Rafelsberger, J.L. Joly, and S. Blanco. 1998. A Model for the emergence of pillars, walls and royal chambers in termite nests. Phil Trans R Soc B 353 : 1561 1576
70 Bruinsma, O. H. 1979. An analysis of building behaviour of the termite Macrotermes subhyalinus (Rambur). Ph. D. dissertation Wageningen University, The Netherlands Buhl, J., J. L. Deneubourg, A. Gimal, and G.Theraulaz. 2005. Self organized digging activity in ant colonies. Behav Ecol. Sociobiol. 58: 9 17 Camlitepe, Y., and D. J. Stradling. 1995. Woodants orient to magnetic fields. Philos. Trans. R. Soc. B 261: 37 41. C ampora, C. E., and J. K. Grace. 2001. Tunnel orientat ion and search pattern sequence of the Formosan subterranean termite (Isoptera: Rhinotermitidae). J Econ Ent 94 : 1193 1199. Chowdh ury, D., A. Schadschne ider and K. N ishinari 2005. Physics of transport and traffic phenomena in biology: from molecular motors and cells to organi sms. Phys. Life Rev. 2: 318 352. Collet, M., and T.S. Collett. 2000. How do insects use path integration for navigation? Biol. Cy bernet. 83: 245 259 C ourtois, P. J. and F. Heymans. 1991. A simulation of the construction process of a termite nest. J Theor Biol 153: 469 475 Couzin, I. D., and N.R. Franks. 2003. Self organized lane formation and optimized traffic flo w in army a nts. Proc. R. Soc. B. 270: 139 146 Dawkins R. 1999. The extended phenotype: the long reach of the gene. Oxford University Press, Oxford, UK Delaplane, K.S ., and J.P LaFage. 1990. Wood excavations of three species of subterranean termites Entomol. E xp. Appl. 54: 81 87 Dene uborg, J. L., and N. R. Franks. 1995. Collective control with out explicit coding: the case of communal nest excavation. J Insect Behav 8 : 417 432 Dussutour, A., V. Fourcassi, D. Helbing, and J. L. Deneubourg. 2004. Optimal t raffic organisation in ants u nder crowded conditions. Nature 428: 70 73. Etienne, A. S., R. Maurer, and V. Seguinot. 1996. Path integration in mammals and its interaction with visual landmarks. J. Exp. Biol. 199: 201 209. Evans, T. A. 2003. The influenc e of soil heterogeneity on exploratory tunneling by the subterranean termite Coptotermes frenchi (Isoptera: Rhinotermitidae). Bull. En t. Res. 93: 413 423
71 Golberg, J. 1975. C R Acad Sci Hebd Seance s Acad Sci D 281: 667 670 Gould, J. L., J. L. Kirschvink, and K. S. Deffeyes. 1978. Bees have magnetic remanence. Science (Wash., D.C.) 201: 1026. Grasse, P. P. 1959. La reconstruction du nid et les coordinations inter individuelles chez Bellicosite rmes nataliensis et Cubitermes sp La theore de la stigmergie: essai Insectes Soc 6 : 41 81 Hamilton C F Lay and M.S. Bulmer 2011. Subterranean termite prophylactic secretions and extern al antifungal defenses. J Insect Physiol 57: 1259 1266. Hansell M. H. 2005. Animal architecture. Oxford University Press, Oxford, UK. Holbrook, C.T., R.M. Clark, D. Moore, R.P. Overson, C.A. Penick, and A. A. Smith. 2010. Social insects inspire human d esign. Biol. Lett. 6: 431 433. John, A., A. Scha dschneider, D. Chowdhury, a nd K. Nishinari 2004. Collective effects in traffic on bi dire ctional ant trails. J. Theor. Biol. 231: 279 285. Jones, R. J. 1979. Expansion of the nest of Nasutitermes coasta lis Insecte s Soc 26: 322 Jones, C.G., H.J., Lawton and M. Shachak. 1994. Organisms as E cosystem Engineers. Oikos 69 : 373 386. Kimchi, T., and J. Terkel. 2003. Detours by the blind molerat follow assessment of location and physical properties of under ground obstacles. Anim. Behav. 66: 885 891. Kimchi, T., A. S. Etienne, and J. Terkel. 2004. A subterranean mammal uses the magnetic compass for path integration. Proc. Natl. Acad. Sci. U.S.A. 101: 1105 1109. Kramer, R.S. S. 2005 Three dimensional ant ne st excavation using stigmergic rules. M. S. thesis, The University of Sussex, Brighton. Ladely, D. and S. Bullock. 2005 The role of logistic constraints in termite construction of chambers and tunnels. J Theor Biol 234: 551 564 Lee S H P.M. Bar dunias and N Y Su 2006. Food encounter rates of simulated termite tunnels with variable food size/distribution pattern and tunnel branch length. J. Theor. Bio. 243 : 493 500
72 Lee, S. H., P.M. Barduni as, and N. Y. Su. 2007. Optimal length distribution o f termite tunnel branches for efficient food search and resource transportation. Biosys. 90: 802 807. Lee, S. H., P.M. Bardunias, and N. Y. Su 2008a. Rounding a corner of a bent termite tunnel and tunnel traffic e fficiency. Behav. Process. 77: 135 138 Lee, S. H., P.M. Bardunias, N. Y. Su, and R. L. Yang 2008b. Behavioral response of termites to tunnel surface irregularity. Behav. Process. 78: 397 400 Li, H. F. 2006. Soil displacement during tunnel excavation by the Formosan subterranean termite, Coptotermes formosanus Shiraki (Isoptera: Rhinotermitidae). M.S. thesis, Univer sity of Florida, Gainesville Li, H. F., and N. Y. Su 2008. Sand displacement during tunnel excavation by the Formosan subterranean termite (Isoptera: Rhinotermitidae). Ann. Entomol. Soc. Am. 101: 456 462. Li, H. F., and N. Y. Su. 2009. Buccal manipulation of sand particles during tunnel excavation of the Formosan subterranean termite (Isoptera: Rhinotermitidae). Ann. Entomol. Soc. Am. 102: 333 338. Maher, B. A. 1998. Magne tite biomineralization in termites. Philo s. Trans. R. Soc. B 265: 733 737 Matsuura, K., T. Yashiro, K. Shimizu, S. Tatsumi, and T. Tamura. 2009. Cuckoo fungus mimics termite eggs by producing the cellulose digestin g lucosidase. Curr Biol 19: 30 36. Mikheye v, A. S., and W.R. Tschinkel. 2004. Nest architecture of the ant Formica pallidefulva : structure, costs, and rules of e xcavation. Insectes Soc. 51: 30 36. Mittelstaedt, H. 1983. The role of multimodal c onvergence in homing by path integration. Fortschr. Zool. 21: 46 85. Muller, M., and R. Wehner. 1988. Path Integration in Desert Ants, Cataglyphis fortis Proc. Natl. Acad. Sci. 85: 5287 5290 Nishinari, K., K. Sugawara, T. Kazama, A. Sch adschneider, an d R. Chowdhury. 2006. Modelling of self driven particles: foraging a nts and pedestrians. Physica A. 372: 132 141. Nobre, T ., L. Nunes, and D.E. Bignell. 2007. Tunnel geometry of the subterranean termite Reticulitermes grassei (Isoptera: Rhinotermitidae) in response to sand bulk density and the presence of food. Insect S ci. 14: 511 518
73 Nobre T C Rouland Lefvre and D.K. Aanen 2011 Comparative biology of fungus cultivation in termites and ants Biology of Termites: A Modern Synthesis ( Bignell DE R oisin Y & Lo N eds), 193 210 Springer Verlag, Berlin binson and M. R. Myerscough. 1999. Self organized criticality in termite architecture: a role of crowding in ensuring ordered nest expansion. J Theor Biol 198: 305 327 binson and M. R. Myerscough. 2003. Self organized criticality and emergent oscillations in models of termite architecture with crowding. J Theor Biol 221: 15 27 Pitts Sing er, T. L., and B. T. Forschler. 2000. Influence of gui delines and passageways on tunnelling behavior of Reticulitermes flavipes (Kollar) and R. virginicus (Banks) J. Insect Behav 13 : 273 290 Reinhar d, J., H. Hertel, and M. Kaib. 1997. Systematic Search for food in the subterraneantermite Reticulitermes santonensis De Feytaud (Isoptera, Rhinotermitidae). Insectes Soc 44 :147 158 Robson, S. K., M. G. Lesniak, R. V. Kothandopani, J. F. A. Traniello B. L. Thorne, and V. Fourcassie. 1995. Nonrandom search geometry in subterranean termites. Naturwissens chaften 82: 526 528 Snchez Pea S.R 2005. New view on origin of attine ant fungus mutualism: exploitation of a pre existing insect fungus symbiosis (Hymenoptera: Formicidae). Ann. Entomol. Soc. Am. 98: 151 164. SAS Institute. 2002 uide, version 9.1. SAS Institute, Cary, NC. R.V. Smythe, F.L. Carter, and C.C. Baxter. 1971. Influence of wood decay on feeding and survival of the eastern subterranean termite, Reticulitermes flavipes (Isoptera: Rhinoter mitidae). Ann. E nt. Soc. Am. 64: 5 9 62 Stuart A M 1967. Alarm, defense and construction behaviour relationships in termites (Isoptera) Science 156: 1123 1125 Su, N. Y., and R. H. Scheffrahn. 1986. A method to access, trap, and monitor field populations of the Formosan subterranean termite (Isoptera: Rhinotermitidae) in the urban environment. Sociobiology 12: 299 304 Su, N. Y. and H. Puche. 2003. Tunneling activity of subterranean termites (Isoptera: Rhinotermitidae) in sand with moisture gradients. J Econ Ent 96 : 88 93 Su, N. Y., B. M. Stith, H. Puche, and P. M. Bardunias. 2004. Characterization of
74 tunneling geometry of Subterranean Termites (Isoptera: Rhinotermitidae) by Computer Simulation. Sociobiology 44: 471 483 Su, N. Y. 2005. Directional Change in Tunneling of Subt erranean Termites (Isoptera: Rhinotermitidae) in Response to Decaye d Wood Attractants. J. Econ. Entomol. 98: 471 475 Su, N. Y., and S. H. Lee 2009. Tunnel Volume Regulation and Group Size of Subterranean Termites (Isoptera: Rhinotermitidae). Ann. Ent. Soc. Am. 102: 1158 1164 Sudd, J. H. 1970 The response of isolated digging worker ants [ Formica lemani Bondroit and Lassius niger (L.)] to tunnels. Insectes Soc. 4: 261 272 Sudd, J. H. 1971. The effect of tunnel depth and of working in pairs on the spe ed of excavation in ants ( Formica lemani Bondroit). Anim. Behav. 19: 677 686 Sudd, J. H. 1975. A model of digging behaviour and tunnel production in ants. Insectes Soc 22: 225 235 Swoboda, L.E. and D.M Miller. 2004. Laboratory assays evaluate t he influence of physical guidelines on subterranean termite (Isoptera: Rhinotermitidae) tunneling, bait dis covery, and consumption. J. Econ Entomol. 97: 1404 1412. Tamashiro M J.K. Fujii and P O Lai 1973. A simple method to observe, trap, and prep are large numbers of subterranean termites for laboratory and field experiments. Environ Entomol 4: 721 722. Tarumingkeng R C H .C. Coppel and F. Matsumura 1976. Morphology and ultrastructure of the antennal chemoreceptors and mechanoreceptors of wo rker Coptotermes formosanus Shiraki. Cell Tis. Res. 173: 173 178 T ucker, C. L., P. G. Koehler, and F.M. Oi. 2004. Influence of soil compaction on tunnel network construction by the eastern subterranean termite (Iso ptera: Rhinotermitidae). J. Econ. Entomo l. 97: 89 94 Turner, J S. 2005. Extended physiology of an insect built structure. Am Entomol 51 : 36 38 Tschinkel, W. R. 2004. The nest architecture of the Florida harvestor ant, Pogonomyrmex badius J. Insect Sci. 4, 21, available online. http://www.insectscience.org/4.21 Visser, A. A., T. Nobre, C.R. Currie, D.R. Aanen, and M. Poulsen. 2011. Exploring the Potential for Actinobacteria as Defensive Symbionts in Fu ngus Growing Termites. Microb. Eco l. 6 3: 975 985.
75 Wilson E. O. 1971 The insect societies. Harvard University Press, Cambridge, MA.
76 BIOGRAPHICAL SKETCH Paul Michael Bardunias was born to Bonnie and Peter Bardunias in Carmel, NY, a suburb of New York City. He grew up with his older brother Peter, three years his senior and his little brother John, four years his junior. He graduated from Lakeland High School in 1987, then went on to earn his Bachelor of Science degree at the University of Miami, in Coral Gables, FL, in 1992. Paul then wen t on to attend graduate school at the University of Kansas, where he began work on a Doctor of Philosophy degree in Entomology. Unfortunately, the illness and loss of his beloved parents derailed his course of study, but after working as a researcher in t he lab of Nan Yao Su for years, he was able to receive a Master of Arts d egree from the University of Kansas and enroll in the Entomology program at the University of Florida in 2010, with Nan Yao Su as his advisor. His thesis work on self organization in termites has redefined our understanding of excavation by social insect groups. He received the 2012 Lafage student research award, from the Entomological Foundation, for his discovery of a preadaptation to fungal agriculture of the type seen in Macroter mes termites and Leaf Cutter ants that arose from the simple mechanics of tunnel excavation. After graduation, Paul will be working as a Post Doctoral associate for J. Scott Turner, in apt the building mechanics and features of termite construction to robot swarms and human architecture.