Insectes Sociaux

, Volume 60, Issue 4, pp 525–530

Colony-specific architecture of shelter tubes by termites


  • N. Mizumoto
    • Laboratory of Insect Ecology, Graduate School of AgricultureKyoto University
    • Laboratory of Insect Ecology, Graduate School of AgricultureKyoto University
Research Article

DOI: 10.1007/s00040-013-0319-1

Cite this article as:
Mizumoto, N. & Matsuura, K. Insect. Soc. (2013) 60: 525. doi:10.1007/s00040-013-0319-1


Social insects build sophisticated and complex architectures such as huge nests and underground galleries based on self-organizing rules. The structures of these architectures vary widely in size and shape within a species. Some studies have revealed that the current environmental and/or social factors can cause differences in the architectures that emerge from collective building. However, little is known about the effect of colony-level variations on the architecture. Here, we demonstrate that termite colonies build colony-specific architecture using shelter-tube construction as a model system. When we divided a colony into multiple groups of individuals, groups drawn from the same colony performed similar patterns of construction, whereas groups from different colonies exhibited different patterns. The colony variations in shelter-tube construction are generally thought to reflect differences in foraging strategy, and this difference can have important fitness consequences depending on the distribution of wood resources in the environment. This is the first demonstration of colony variation in the architecture that emerges from collective behavior. Colony-specific architectural variations provide new insights into our understanding of the self-organization systems, which were previously assumed to provide each species with a species-specific construction mechanism.


Collective behaviorSelf-organizationGallery architectureColony variationSocial insects


Social insects build various architectures such as huge nests and underground galleries, which can be very impressive given the differences between the individuals and their levels of cooperation in the absence of a supervisor or centralized control (Wilson, 1971; von Frisch, 1975; Theraulaz et al., 1998; Hansell, 2005). Construction of these structures is achieved by self-organization. In self-organizing systems, colony-level patterns emerge from local interactions among members that elicit positive and negative feedback responses. These interactions are often mediated by stigmergy (Deneubourg, 1977; Theraulaz and Bonabeau, 1995; Camazine et al., 2001), a form of indirect communication through the modification of the environment. Through careful observation and analysis, many studies on self-organizing mechanisms have demonstrated that complex and sophisticated structures emerge from simple interactions among members within systems in ants (Franks and Deneubourg, 1997; Theraulaz et al., 2002), bees (Camazine, 1991), wasps (Karsai and Penzes, 1993), and termites (Bonabeau et al., 1997). However, as the notion of each species having a species-specific construction mechanism has been frequently assumed, little is known about the colony variation of the structure that emerges from collective building.

The actual architectures built by social insects vary widely in size and shape within a species. Several studies have highlighted the effect of environmental conditions on collective building by social insects. Furthermore, variation in architecture within a species can be generated by exogenous environmental (Bollazzi et al., 2008; Toffin et al., 2010) or social (Franks et al., 1992; Tschinkel, 2004; Buhl et al., 2005; Toffin et al., 2009) factors. However, some recent studies have suggested that different constructions can emerge even in a homogeneous environment and within a species (Toffin et al., 2009, 2010); whether colonies display variations in the structure of architecture under identical exogenous conditions remains unclear. If colonies have colony-specificity in the building process and resulting architecture, each group drawn from the same colony would be expected to have the same characteristics and build the same structure. To test this idea, it is necessary to divide a colony into multiple groups of individuals and to examine the architecture produced by these fragmented groups. In most social insects, especially in bees or wasps, it seems impossible to conduct this experiment because colony fragmentation substantially disrupts group-level regulation and thus disturbs self-organizing building behavior. Moreover, although studies have conducted fragmentation experiments using ants or termites (e.g., Toffin et al., 2009, 2010; Cornelius and Osbrink, 2010), colony-specificity in building has not been tested or detected.

Like many other subterranean termites, Reticulitermes termites are classified as multiple-piece nesters based on their nesting and feeding habits, whereby nests of a single colony are interconnected by belowground tunnels and aboveground shelter tubes (Abe, 1987; Shellman-Reeve, 1997). Shelter tubes are made of wood pieces, soil and termite excretions, which provide shelter and protection to the termites from predators. In Reticulitermes termites, groups of individuals separated from an original colony perform collective activities, including shelter-tube construction. This characteristic makes them an ideal model system to study colony specificity in the patterns that emerge from collective activity.

In this paper, we used the shelter-tube construction system of Reticulitermes speratus to examine whether termites show intercolonial variation in the patterns of construction under the same external conditions and whether all subgroups from a colony generate the same pattern through collective building. We collected several field colonies and immediately divided each colony into fragment groups of individuals. Then, we compared the construction process and the consequent structure of shelter tubes among the groups to identify the colony-specificity of the morphology of the product and the collective behavior of the termites.


Shelter-tube experiment

Eight colonies of R. speratus were collected during April–July 2012 from pine-oak forests at Uryuuyama (A and B), Takaragaike (C), Iwakura (D), and Yoshida (E–H), all northern suburbs of Kyoto, Japan. The eight colonies were at least 70 m apart. Termites were used for the experiment within 3 days of collection. We used workers that were large enough to conduct shelter-tube construction because larvae and small workers are unable to build shelter tubes (Crosland et al., 1998). Four hundred were randomly chosen from each colony and placed on a block of mixed sawdust food (36 mm diameter × 15 mm height) at the center of a short plastic container (221 × 141 × 37 mm; Fig. 1a, b). The sawdust block was prepared from brown rotten pinewood and cellulose powder (Nacalai Tesque, Kyoto, Japan) mixed at a ratio of 5 to 1 by volume. We made five replicate groups for each of the eight colonies. The plastic containers were maintained at 25 °C under a 16 h light:8 h dark photoperiod for 30 days. One group from colony E (group #E4) was excluded from analysis because its shelter tubes were accidentally broken at 24 days.
Fig. 1

Colony-specific architecture of shelter tubes. a A dendrogram of shelter-tube architecture resulting from a hierarchical cluster analysis. Ward’s distance is indicated on the scale above the dendrogram. Representative photographs of each of the four primary clusters are shown on the left of the dendrogram. An edging shelter tube (est) and a vagus shelter tube (vst) are indicated by arrows in the photograph. b Comparison of shelter tube construction among the fragment groups derived from the representative colonies B (above) and D (below) at 30 days

To record shelter-tube construction, we took vertical photographs of each plastic container every 24 h with a digital camera (D300, Nikon, Tokyo, Japan). The termites constructed shelter tubes and/or covered the bottom of a container with mats consisting of particles of mixed sawdust and termite feces (Fig. 1). Two distinct types of shelter tubes were constructed: an edging shelter tube (est) and a vagus shelter tube (vst). The former type was laid along the edge of the container, and the latter was formed in the middle of the bottom of the plate (Fig. 1).

To examine the colony characteristics in the construction process and the consequent structure of shelter tubes, we extracted the following data from the photographs taken for each 24-h interval: the starting date of construction, the total length of est, the total length of vst, the number of shelter tubes started from the sawdust block, the number of tube ends, and the number of intersections. Additionally, we measured the total area covered with shelter tubes and/or mats using planimetry in Adobe Photoshop software v.11.0.2 (Adobe Systems, Inc., San Jose, CA, USA).

Cluster analysis

To help identify colony characteristics of shelter-tube construction, we performed hierarchical cluster analysis (HCA). Six structural characteristics at 30 days (total length of est and vst, proportion of vst [vst/(est + vst)], the number of shelter tubes, the number of tube ends, the number of intersections, and covered area) were used in a HCA using Ward’s distances to produce a dendrogram. Prior to analysis, we used the Z-score transformation for all variables, which controls for the variance of the variables and fits well to variables in different units (Gotelli and Ellison, 2004).

Statistical analysis

To examine colony-specific characteristics in the construction process and structural forms, we tested colony differences in covered area, initial construction speed, and proportion of the length of vst using ANOVA with Tukey’s contrasts. Initial construction speed (mm/day) was determined for the first 5 days after the initiation of construction. Covered area and proportion of vst at 30 days were used in this analysis.

To compare overall construction and structural profiles among colonies, we performed a principal components analysis (PCA) on 13 variables: total length of est and vst and proportion of vst constructed during three time intervals (days 0–10, days 10–20, and days 20–30), the number of shelter tubes, the number of tube ends, the number of intersections, the number of days until the start of construction, initial construction speed, covered area, and laterality index (LI). LI was calculated as (R − L)/(R + L), where R and L represent the lengths of est at 5 days that were constructed clockwise and counterclockwise, respectively. The number of shelter tubes, number of tube ends, number of intersections, and covered area at 30 days were used in this PCA. For groups that did not form any shelter tubes (three of 40), vst, initial construction speed, and LI were set as 0, and the number of days until starting was set as 31. All variables were standardized by Z-transformation prior to the analysis (Gotelli and Ellison, 2004). The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy gave a value of 0.662, classified as “mediocre” after Kaiser for this data set (Kaiser, 1974). We reduced the variables to a single component representative of characteristics and then performed ANOVA on the component score. All analyses were performed with the software R (version 2.15.1).


Of the 40 groups (5 groups × 8 colonies), 37 groups formed shelter tubes and 3 did not, merely covering the container surface with particles. The dendrogram of the HCA showed four primary clusters of shelter-tube architecture (Fig. 1). We found significant differences among the original colonies in the following characteristics: covered area (ANOVA: F7,31 = 20.88, P < 0.0001), initial construction speed (ANOVA: F7,31 = 33.8, P < 0.0001; Fig. 2a), and proportion of vst (ANOVA: F7,31 = 12.04, P < 0.0001; Fig. 2b).
Fig. 2

A between-colony comparison of the construction process and the consequent architecture of shelter tubes. a Initial construction speed. b Proportion of vagus shelter tubes (vst). Significant differences among groups are indicated by different lower case letters (Tukey’s test, P < 0.05). Bars represent the mean ± SE

Including 13 construction and structural characteristics, the PCA identified four components with eigenvalues larger than 1 (Table 1). The first component (Comp1) explained 47.85 % of the total variance. The highest loadings for this component were related to construction speed (the number of days until the start of construction and the proportion of vst constructed during days 20–30). The second component (Comp2) explained 17.99 % of the total variance and had the highest loadings for parameters related to shelter-tube architecture (covered area and proportion of vst constructed during days 0–10). Both Comp1 and Comp2 were significantly different among colonies (ANOVA, Comp1: F7,31 = 49.89, P < 0.0001; Comp2: F7,31 = 63.55, P < 0.0001) and clearly differentiated the original colonies (Fig. 3).
Table 1

Results of principle component analysis of the construction process and the consequent architecture of shelter tubes


Comp. 1

Comp. 2

Comp. 3

Comp. 4






% Variance explained





Total length of est and vst

 Days 0–10





 Days 10–20





 Days 20–30





Proportion of vst

 Days 0–10





 Days 10–20





 Days 20–30





No. of shelter tubes





No. of tube ends





No. of intersections





Days until starting construction





Covered area





Laterality index





Initial construction speed




Fig. 3

The results of a principal components analysis including the characteristics of the construction process and the consequent architecture. The first component (Comp1) explained 47.85 % of the total variance. The second component (Comp2) explained 17.99 % of the total variance. Bars represent the mean ± SE


The termites showed distinct colony-specificity in their shelter-tube construction, that is, individual groups of workers derived from the same colony showed similar patterns of shelter-tube construction, whereas groups derived from different colonies showed distinctly different construction patterns (Fig. 3). The shelter-tube construction can be influenced by physical properties of the building materials in the subterranean termites (Cornelius and Osbrink, 2010). In this study, however, pattern variations among colonies were attributable not to exogenous factors but to colony characteristics because all termite groups constructed the shelter tubes under identical environmental conditions. Therefore, non-exogenous factors involved in colony characteristics, such as genetic differences (Robinson et al., 1997; Ben-Shahar et al., 2002; Ingram et al., 2005), learning (i.e., past environmental conditions experienced by the colony; Leadbeater and Chittka, 2007) or physiological differences (e.g., nutritional conditions), need to be taken into account to understand the observed between-colony variation in shelter-tube construction.

The tunneling behavior and the shelter-tube construction were known to be influenced by worker size, which varied according to age (Crosland and Traniello, 1997; Crosland et al., 1998; Yang et al., 2009; Haifig et al., 2011). In Reticulitermes fukienesis, larvae and small workers are unable to build shelter tubes, and the shelter-tube construction is mainly conducted by large workers (Crosland et al., 1998). In the present study, we excluded small workers (firstst and second stage workers) and used only large workers (fourth and the later stage workers) in each colony, which were considered able to engage in shelter-tube construction. Therefore, the observed variation in shelter-tube patterns cannot be attributed to the division of labor involving worker size or age within the groups of termites.

In social insects, natural selection primarily operates on differences among colonies (Korb and Heinze, 2004), and several studies have focused on intercolonial variation from other perspectives than the architecture they build. For example, the collective personality, that is, the consistent differences in the collective behavior, just like solitary animals, can have important consequences for colony fitness (Wray et al., 2011; Scharf et al., 2012). In this study, some groups formed no shelter tubes, but simply covered the bottom of the container with mats (Fig. 1). Other groups laid many shelter tubes from the nest, and still others constructed fewer but longer shelter tubes (Fig. 1). The differences in this shelter-tube structure would involve decisions among the following strategies: searching woods near the nest area by walking without shelter tubes, searching carefully near the nest area by constructing a fine web of shelter tubes, searching roughly over a wide area by laying a small number of long shelter tubes, and other intermediate strategies. The colony difference in the strategies could have important fitness consequences, depending on the distribution of wood resources in the environment (Araújo et al., 2011).

In general, one primary goal in studies of self-organization is to understand the mechanisms that construct adaptive structures and organize vast numbers of individuals (Camazine et al., 2001). Therefore, only a few studies have focused on the colony variation of the structure that emerges from collective building. On the other hand, although researchers are focusing on colony-level variation in collective behaviors (Gordon et al., 2011), investigating whether differences in the collective behaviors result in different structures has been technically difficult. Producing exactly the same pattern is theoretically possible even when the behaviors of constituent members and/or collective dynamics are different. In this study, we showed that termite colonies displayed remarkable colony-specificity in construction. This is the first demonstration on colony variation in the architecture that emerges from collective behavior. Differences in the architecture among colonies lead to variations that introduce natural selection to the colonies. Examining the adaptive significance of colony variation in architecture and how differences emerge from individual differences could provide an understanding of the selective pressures operating on self-organizing systems and their subsequent evolution.


We thank Dr. Kazuya Kobayashi and Dr. Jin Yoshimura for helpful comments. This work was supported by the Japan Society for the Promotion of Science (No. 09001407) and by the Sumitomo Foundation to K. M.

Copyright information

© International Union for the Study of Social Insects (IUSSI) 2013