Introduction

Nests occur across a range of biological scales and serve various functions such as providing refuge, storage, transport, mating, and micro-climate regulation (1–10). Among the diversity of nests those constructed by social insects are highly complex in terms of function, architecture, and construction mechanisms. These nests are created through digging or modifying a substrate and can be arboreal or underground (11–18). Previous studies have demonstrated that these structures emerge from the interaction between individual workers and the environment in a decentralized manner, showcasing an exemplary instance of biological self-organization (19–28).

Previous studies on nest construction and dynamics often addressed mature colonies of fixed size and over short time scales (24, 29–31). They show that for ant groups of fixed size nest volume grows logistically to a saturation volume that scales linearly with population size (24, 29, 30, 32–34). This led to the hypothesis that workers can perceive the density of ants within the nest and adjust their digging activity accordingly. This mechanism may hold in relatively small populations and simple architectural designs. However, for more complex structures where ants congregate in specific chambers, workers are less likely to assess the overall nest density. In such cases, nest excavation is enhanced by recruiting additional workers from the neighborhood or facilitated by humidity or pheromones (22, 24, 27, 29, 35–38).

Ant colonies are composed of ants that vary in age. It is well known that the task allocation of individual workers can change throughout their lifetimes (age polyethism) (39–46). However, it is unknown which of the work groups in a developing colony participate in the digging effort. Additionally, ant colonies are comprised of the reproductive caste and one or more worker castes that can divide labor among themselves (23, 39, 44, 45, 47, 48). Thus, colony demographics play a significant role in the development of nest structure, and its role needs to be considered.

One way to view the social insect nest is from a colony-superorganism perspective (2, 19, 20, 22, 26). Within this analogy, the social insect nest can be likened to the superorganism’s skeleton. Inspired by the processes of skeletal growth and bone healing after fracture we set out to find how the volume of an ant nest is regulated during ordinary colony development and following a catastrophic collapse. Specifically, we were interested in the following questions:

1. How does a colony regulate the co-expansion of the population and the nest as the colony develops from a single queen to tens of workers?

2. How does a colony recover nest volume following a catastrophic collapse?

3. Which ants participate in nest excavation and does this answer differ between the cases (a) and (b)?

We performed two sets of experiments to study how a Camponotus fellah ant colony regulates the volume of their nest: The “Colony-Maturation” experiment monitored colonies as they developed from a single fertile queen to a mature colony. We used these experiments to track the coexpansion of the colony and its nest for an extended period (up to over six months). Whereas, in the “Fixed-Demographics” experiments, we monitored excavation by an ant group of determined size and age. We used these experiments to study the nest volume recovery following catastrophic events and the role of different age groups in the digging process.

Results

Nest design and experimental setup

All experiments were conducted within two-dimensional nests (Fig. 1, Materials and methods), nest sizes are reported as areas but these can be readily translated into volumes by multiplying by the thickness of the setup (0.8 cm). We performed two kinds of excavation experiments: colony maturation experiments that follow nest excavation alongside colony growth and fixed demographics experiments that track digging by already mature colonies.

Figure 1.

We initiated colony maturation experiments by introducing a single mated queen and several brood items (pupae) into the setup. This allowed us to capture nest excavation and colony development from their very first stages. Experiments were terminated either when the queen died or when she became irreversibly trapped after a structural collapse. Overall, colony maturation experiments (n = 22) spanned up to 6 months (see Table: Experimental details, Materials, and methods) during which the population size and nest area were recorded. The number of experiments decreased with time: for example, on day 130, we only had 13 ongoing excavation experiments (n = 9, collapses).

Fixed demographics experiments (n = 37) included colonies with known age demographics. In preparation for these experiments, we tracked the age of individual workers over the course of 10 months (Materials and Methods). We then extracted the queen alongside several workers of a given age and transferred them into the digging setup. Fixed demographics experiments spanned six conditions which varied by worker age (young vs. old workers) and group size (5, 10, &15 ants, including the queen). These experiments allowed us to track nest excavation as the ants were introduced into the setup and to further quantify the response to an induced catastrophic collapse.

Nest and colony development from the founding queen

Colony maturation experiments allowed us to track the concurrent growth of the colony and its nest.

In all colonies, population size increased logistically (Fig. 2 A) and saturated at a value that was highly variable across colonies (typically between 5 and 20 ants, see SI. Fig 1 and Table 1). As the population expanded so did the nest (Fig. 2 B). The nest excavation was commenced by the founding queen which dug a mean area of 23.8(±1.7) cm² (n = 22; unless noted otherwise, means are presented as mean ± standard error). Once the first workers emerged from the pupae the queen ceased digging (personal observation). However, these first workers did not dig (see kink in Fig. 2 B-C) possibly because the initial area dug by the queen led to low ant densities (number of ants/area) at this stage. As more workers emerged the area of the nest increased and varied considerably across nests (SI. Fig 1).

Figure 2.

Table 1.

For instance, halfway through the experiment duration, around day 95, the excavated area ranged from 38.41 cm² to 173.28 cm² with a mean excavated area of 100.51(±11.54) cm² (n = 15). At day 190, excavated areas ranged from 149.87 cm² to 234.65 cm ² with a mean excavated area of 197.47(±13.8) cm² (n = 6). The number of active experiments decreased over time since experiments were terminated as specified above.

Previous works on already mature colonies revealed a linear relation between population and nest size (24, 29). We find that this relationship persists in cases where colonies and nests are allowed to co-expand (Fig. 2 C). A linear fit (R-square - 0.97), indicates that the colonies regulate an area of approximately 11cm² per each ant (SI Fig. 2 A, B). Deviations from this fit at early times result from the initial area excavated by the queen.

All the results stated so far deal with averages over different colonies. These do not allow us to address the causality of the regulation depicted in Fig 2 C. To test whether colony expansion triggered changes in nest-area or if it was the other way around, we tracked these concurrent processes one colony at a time. In all experiments, the initial population growth is followed, 15(±12) days later, by an increase in excavated area. Moreover, in some of the colonies these dynamics repeat themselves leading to a step-like pattern (SI Fig. 3). This staggered pattern suggests that the nest is excavated in separate digging episodes, possibly instigated by a population increase. Cross-correlation analysis between the digging rate and the population growth rate (Fig. 2 D) indeed reveals that an increase in the population size is followed by an increase in nest area on a time scale of one week.

Our results suggest that population size controls nest area and not vice-versa. As newly-born ants join the colony the size of the nest increases accordingly. It is, however, not clear which ants do the actual sensing or digging. One straightforward possibility would be that newborn ants are those that dig. Such a mechanism may rid ants of the requirement to determine overall density. Another possibility is that, following the emergence of new ants, every ant, irrespective of her age, senses the increase in density and commences to dig (24, 29, 30).

Fixed demographics experiments reveal age-dependent nest excavation

In “Colony-maturation” experiments ants emerge from their pupae inside the thin confines of the setup. This prevents us from marking them by age. To test how an ant’s age affects her propensity to dig we conducted shorter “Fixed-Demographics” excavation experiments with fixed population sizes (5,10, or 15 ants including a queen) where all workers are of similar age (either “young” (40 ± 16 days) or “old” 171 ± 20) (SI Fig. 4). These experiments (Materials, and Methods, Table 2) further allowed us to study excavation under conditions of extreme area shortage. Under our protocol, such conditions occur twice: as the ants are initially introduced to the setup and then again after we induce a collapse.

Table 2.

Following the introduction of the ants into the setup, the nest area grew logistically and stabilized after about three weeks independently of age or group size (Fig. 3 A, B). We found a significant difference in the maximum area excavated by same-sized cohorts as a function of worker age, with the younger ants excavating larger areas when compared to the older ants (Fig. 3 A, B-Unpaired Wilcoxon signed-rank test significant for group size: 10; p = 0.014, 15; p = 0.042). Moreover, the area excavated by the young cohorts was similar to that excavated by naturally maturing colonies at the point in which they reached the same population size (Tukey’s HSD; group size: 5; p = 0.61, group size: 10; p = 0.46, group size: 15; p = 0.20). This holds even after excluding the area excavated by the queen from the colony maturation experiments (SI. Fig 5).

Figure 3.

Once the fixed-demographics nests reached a steady-state area, we artificially collapsed 25-30 % of the steady-state area (star - Fig 3A, & circle - Fig 3B). We found that irrespective of age, ants dug after the collapse. Moreover, colonies ceased digging when they recovered (93 ± 3)% of the area lost by the manual collapse (crossed circles - Fig 3A, & 3B). Saturation areas (Fig. 3 C, SI. Fig 7 A) increase linearly with the colony size (for the excavated area, General Linear Model (GLM, F(_35,37>); p < 0.0001)). Normalizing saturation areas and digging rates by the number of ants allowed us to obtain the single ant digging rate as well as the target area per ant (this is the reciprocal of ant density which, for short, we refer to as the target area from here on). We find that while the target area of an ant depends on her age (Fig. 4 A), digging rates remain stable (, SI Fig. 6 A, B), over the first 200 days of an ant’s life (SI. Fig 7 B).

Figure 4.

Age-dependent model predicts minimal digging by older ants

The fixed-demographics experiments revealed that while all ants dig at the same rate, younger ants aim to approach a higher area target. Building on this, we assume that the age-dependent target area identified in the fixed demographic experiments persists during colony maturation experiments. This assumption facilitates a re-analysis of the colony maturation experiment, allowing us to determine which ants are involved in nest excavation during normal nest growth and investigate its implications.

To do this we present a modified version of a previous model (24, 29, 30) where the digging rate of each ant depends on how similar the current area per ant (a(t) which is one over the density) is to her age-dependent target;

Where a_age is the target area of a specific age (Fig. 4 A) and the digging rate is (as above). The instances where the current area per ant (‘a’) is higher than the steady-state area per ant a_(age) resulted in a negative digging rate. While there were very few instances where ants filled up the excavated regions they were extremely rare. Therefore, in the model, negative digging rates were taken to be zero.

The total excavation rate ^dA for a colony that includes ^′N^′ ants of different ages can be obtained by summing over the contributions of all individual ants:

Since we recorded the population sizes in all colony maturation experiments, and mortality events were rare - we could readily deduce the entire age demographics of the colony for each experiment (SI section: Age quantification from colony maturation experiments, Table 3). Using this data, we could integrate Equation 2, over each experiment to obtain the total excavated area.

Table 3.

The models’ results and their comparison to colony maturation experiments are shown in Fig. 4 B. In both model and colony maturation experiments, ‘day 0’ corresponds to day in which the first workers emerge. The area at this day is the area dug by the queen before this event. We find that the age-dependent model fits the excavation dynamics observed in the colony maturation experiments. Furthermore, we simulated an age-independent density model (SI. Fig 8) with a constant target area for each ant (11.6cm²ant⁻¹, SI. Fig 2) as obtained from the colony maturation experiments, rather than an area target that decreases with age (Fig. 4 A). We find that the age-independent model also provides an ample fit for colony maturation experiments, however, it is inconsistent in the light of fixed demographics experiments. An interesting observation is that the target area we fit for the age in the age-independent model closely approximates the empirically measured age-dependent target extrapolated to very young ants (Fig. 4 A). This provides further evidence that, in the colony maturation experiments, it is the youngest ants that do most of the digging.

Since the target area is age-dependent and ants in the colony vary in age, not all ants perform the same amount of digging. By integrating equation 2, we can track the specific terms in the sum and quantify the area excavated by each ant. When setting an age threshold at 56 days (mean + standard deviation of the age of the younger cohort) in the age-dependent model, we find that, after the initial excavation by the queen, younger ants performed the vast majority of the digging (Fig. 4 C). The area excavated by older ants (age of 57 days and above) was nearly negligible. Thus, the age-dependent model provides indirect evidence indicating that younger members of the colony predominantly carry out nest excavation. Furthermore, we quantified the area dug by each ant in the normal colony growth experiment as estimated from the age-dependent model and found that all ants excavated more or less the same amount, with a mean area of 7.84(±3.44)cm² (where the error bar represents the standard deviation) (Fig. 4 D; Kolmogorov-Smirnov test, p = 0.56).

These findings suggest that while all ants can dig and will do so following catastrophic events (see Fig. 3), the regular nest expansion is primarily carried out by young ants.

Structural classification of nests

The colony maturation and fixed demographics experimental conditions differed not only in the total volume of the nest but also in its architecture. We compared nest architectures by segmenting the nest into chambers and tunnels, where chambers are flat horizontal structures, whereas tunnels are more narrow and vertical (SI. Fig. 9, SI section: Nest skeletonization, segmentation and orientation, (13, 30, 49, 50)).

We found that the colony maturation colonies exhibited a “top-heavy” design (13, 51, 52): large chambers (marked in Yellow in Fig. 5 A) comprise most of the nest area at the top part of the nest (52). Tunnels generally branch out from the large chambers and mostly lead to the bottom parts of the nest. The wide tunnels are primarily seen originating from the tunnels and are oriented parallel to the chambers (Fig. 5 A). In the fixed demographics experiments, the nest structure mostly lacked large chambers, unlike the colony maturation experiments (Fig. 5 B, C, D, E, and F). This could be attributed to the absence of a young queen prone to dig and/or to the absence of brood. We analyzed the orientation of all tunnels and chambers relative to the direction of gravity. We found that in the colony maturation and fixed-demographics young experiments, the bulk of tunnels and wide tunnels are oriented between (30^° - 60^°). Surprisingly, in the fixed-demographics old colonies, the orientation of tunnels ranged from (70^° - 90^°) (Fig. 5 G, H; SI Fig. 10). Thus, the tunnels and wide tunnels in the colony maturation experiments are similar to the fixed demographics of young colonies but not the old ones.

Figure 5.

Maturing colonies are composed of ants of different ages. Nevertheless, colony maturation experiments displayed tunnel orientation similar to those in the nest of young fixed demographics cohorts. This supports our prediction of the age-dependent model that younger ants primarily undertake digging tasks within maturing colonies.

Discussion

Group size is a key determinant in the behavior of ant colonies. Groups of different sizes differ in their foraging strategies (53–57), division of labor (2, 42, 45–48, 58, 59), and communication networks (60–63). It is less clear what happens in the expansion period that follows colony foundation, when group size is constantly on the rise. A growing colony may exhibit collective patterns that differ from a fixed group of the same size. This is especially relevant if the system preserves long-term memories such as the stigmergic alteration of the environment evident in nest construction (22, 30, 64–67).

To date, the overwhelming majority of studies on nest excavation in ants were short-term experiments performed on a mature group of fixed size (24–26, 29–31, 35, 52, 67). As we have shown, these short term studies likely address construction in emergency scenarios where there is a huge deficit in available space. Studying animal behavior in ethologically relevant conditions is most valuable for being able to correctly capture the nature of these animals and is crucial for deciphering the underlying meaning of their behavior. Furthermore, testing animals in the wrong context/conditions could lead to misleading conclusions (68–73).

The current study follows ant colonies over extended periods as they mature from a single reproductive individual. We find significant quantitative and qualitative differences between nests constructed within this natural context and nests constructed in the context of an emergency. These include the total volume of the nest, the way this volume is divided into tunnels and chambers, the orientation of the tunnels, and the identity of the ants that do the actual digging.

A main advantage of social insects is their ability to allocate different tasks to different individuals. A well-known example includes age-polyethism wherein ants perform age-specific tasks (2, 42, 45–48, 58, 59). Response thresholds that vary with age have been suggested as a mechanism for age polyethism (23, 74–78). Previous studies of nest excavation indeed rely on a threshold mechanism that prompts ants to dig extra space when the nest becomes too crowded (24, 29, 30, 35). Our results reveal that this area target decreases linearly with age such that young ants are more sensitive to shortages in space.

Age polyethism plays an important role in controlling the volume of the nest. During normal growth conditions, nest volume is maintained proportional to the number of ants, and densities never become too high. This means that density thresholds for older ants are never crossed and excavation is performed almost exclusively by younger ants. This fact has implications for the volume of the nest. A natural growing nest includes ants that have eclosed inside the nest and thus have the highest possible volume per ant target. This is in line with the fact that the volume of such a naturally formed nest is higher than the volume of a nest excavated by an ant group of equal size. Under these conditions, effectively, each new ant joins the colony and adds a small volume to the nest. Following a catastrophic collapse or displacement volume shortages become critical and ants of all ages join the excavation effort.

Variable digging behaviors impact not just the nest’s volume but also its structure. During the initial phases of colony establishment, only the queen participates in excavation activities. Our findings reveal that, during this period, the queen typically creates several large chambers near the surface. Aligned with the density-target threshold model, these chambers prolong the emergence of the first worker ants, who eventually take over excavation tasks. Consequently, the queen’s initial digging alters the nest’s architecture to become more top-heavy. Moreover, we found that an ant’s age influences not only its likelihood to engage in a task but also how it performs that task.

We believe that the nest reflects the cumulative effects of past events and conditions experienced by the colony. For example, in line with our experimental findings, if the colony expanded during a period with many young ants, more slanted tunnels are expected, indicating accommodation for a growing population. Conversely, nest construction initiated by mature ants or a catastrophe leading to nest area reduction would result in newly constructed tunnels aligning more with gravity, running straight down. Thus, the nest structure could be treated as a historical record of the colony.

We wish to stress that tunnel orientation and its dependence on the age of the diggers constitute a new kind of age-polyethism. Previously reported examples of age-polyethism regarded it as a means of allocating a given task to different subgroups in the colony (79–85). Our results demonstrate a new kind of behavioral plasticity wherein the task itself, and not just the propensity to do it, drastically changes with the age of the ant. This new kind of age polyethism adds previously unknown regulatory capabilities to the ant colony and is distinct from task performance based on morphological differences seen in polymorphic ants (27, 31).

To conclude, ant colony demographics play a pivotal role in nest excavation. They affect the identity of the ants that do the digging, the dynamics of nest growth, the volume of the nest per population size, and the chamber and tunnel architecture of the nest. We find that age-polyethism provides a major means by which the colony exerts control over its nest in different scenarios.

Materials and methods Study species

We chose to work with Camponotus fellah, a common species in Israel of the ground-dwelling carpenter ants (7.8–17.2 mm body length, Ionescu-Hirsch, 2009). These ants are omnivorous, monogynous, and singly mated (86), with colony sizes ranging from tens to thousands of workers. We collected newly-mated queens in the Weizmann Institute of Science campus after nuptial flights in March 2018. For the colony maturation experiments, we used only queens with pupae, that had already nurtured larvae to increase survival rates. We used colonies of group sizes (5,10, and 15 ants) for the fixed demographics experiments. We housed the single queen and fixed colonies in plastered nest containers and environmental conditions (climate-controlled chamber under controlled humidity (65%), temperature (27^◦ C), and a light-dark cycle of 12 hours) before the experiments. We supplied the colonies with a food mixture (tuna, honey, eggs, vitamin mix, and water), water, and a 20 % sucrose solution.

Nest structure

The artificial nest structure consists of a 2D Plexiglas frame 60 cm wide, 80 cm high, and 8 mm deep made from 5 mm thick cast acrylic sheets. Six support blocks (2 cm) held the front and back frames together. We equipped the nests with a 5 cm wide water column connected by a series of 0.4 mm holes in the bottom that allow water transfer but not sand grains (Fig. 1 A). The top part of the frame was coated with Fluon (Sorpol) to prevent the escape of ants. A removable plexiglass lid perforated with numerous small (1 mm) holes for ventilation covers the top of the nest.

Nest preparation

Each nest was filled 70 cm high with homogenous fine-grained sand (AccuSand 30 − 40, Unimin Co, USA) with grain sizes of 425-600 µm. Before the experiments began, water infiltrated gradually from the water column to the sand, slowly saturating the entire sand column from the bottom. Apart from wetting the sand, this action rearranged the sand grains in even more compact packaging and lowered the chance of spaces opening up in the column. On top of the sand layer, we placed a 1-2 cm layer of soil on the sand column, thus significantly reducing evaporation and the subsequent collapses. The soil was collected from the Weizmann campus orchard and sieved through coarse (2 mm) and fine (0.5 mm) sieves (Haver & Boecker, Germany), and was sterilized. We manually made a 30 mm long and 8 mm wide hole in the middle of the solid layer (seeding point) to induce a digging behavior.

Experimental overview

The study included 22 colony maturation and 37 fixed demographics experiments (Fixed demographics young, group size 5, n = 4, group size 10, n = 6, group size 15, n = 6); Fixed demographics old (group size 5, n = 6, group size 10, n = 7, group size 15, n = 8) of ant nest excavation. We placed the founding queen on top of a 60 cm wide, 70 cm high, and 8 mm thick soil-sand column in each nest. We photographed the nest development every 1-3 days. terminated the investigations once the queen died or became trapped after a collapse.

Fixed demographics colonies

Queens used for the fixed demographics experiments were collected during the nuptial flight of May 2020. Forty-five newly mated queens were reared for the fixed demographics experiments and 37 colonies that attained the required colony sizes needed for the fixed demographics were chosen for the experiments. The colonies were maintained with a 12-hour light/dark cycle, room temperature of 25 ± 2°C, and 60% humidity. Each freshly born ant was marked using a unique three-color combination of acrylic paint on the head, thorax, and abdomen to determine the age. We checked each colony at least once in 5 days to check whether new workers emerged. The marked worker’s age was recorded and updated in due course. In the fixed demographics experiments, we separated colonies into two different demographics: Young and Old. The young and old colonies had workers with an average age of 40 ± 16 days and 171.56 ± 20 days, respectively. Error represents the standard deviation of the mean.

Experimental conditions

We maintained a stable environmental condition in the experimental room by maintaining a 12-hour light/dark cycle and room temperature of 25 ± 2°C. Daily variations were 2-3 degrees, usually 24 − 27°C. More significant deviations ranging at most 20 − 30°C did occur sporadically due to technical problems, but they did not last more than a few hours at a time. Humidity was more challenging to maintain, and it fluctuated in the range of 40 − 70%, averaging around 60%.

Nest maintanence

We administered food for ants ad libitum via three tubes of water, sucrose water (20%), and protein food. The protein consisted of egg powder, tuna, prawns, honey, agar, and vitamins. We filled the water and protein food in 5 ml polystyrene tubes and placed a cotton stopper at the end to prevent overflow. We connected the tubes at a slight angle to ensure the flow of liquids using a custom-made plexiglass adapter. We replenished three tubes (water, sucrose water, and protein food) regularly and performed nest maintenance once every two weeks.

Nest photography - Data collection

A photography sequence of different exposures recorded the nest development in a measurement. The nests were placed one meter away from the camera against a diffuser-covered LED backlight. Images were initially taken with Canon EOS 800D DSLR cameras with a 24 MP resolution and fitted with EF-S 18-55mm f/4-5.6 IS STM lenses. We controlled the cameras remotely using a USB cable through the ‘DigiCamControl’ software (Istvan, 2019) with the photography sequence programmed using MATLAB CameraController (Serge, 2019). The photography sequence consisted of four photos taken at 2-second intervals.

Image processing and Excavated area

The images were post-processed using the MATLAB Image Processing Toolbox. The first step was image registration, where we aligned nest images with the first image taken before introducing ants. The algorithm registers the images through a projective geometric transformation using the six support blocks in the frame as control points. After registration and cropping, we binarized the images to show only the excavated areas of the sand column. The first binarization step involved subtracting the initial image from the processed image to reduce glare and noise. We then converted the RGB image into an intensity image by choosing the maximum value between the green or blue channels. Next, we used adaptive thresholding to produce BW (black and white) images of the nest architecture in which white pixels represented the excavated areas. Lastly, we made manual corrections to mark known excavated zones, which were not evident in the images, or remove other apparent noises. The excavated area refers to each Black and White image’s white pixel area.

Acknowledgements

We thank Dr. Yuri Burnishev, Guy Han, Ezra Eliyahu, the physics workshop, and the physics electronics team for the technical support in making the nest frame, and camera system. We also thank Dr. Aviram Gelblum, Dr. Lior Baltiansky, Dr. Atanu Chatterjee, Tabea Heckanthler, Dr. Yael Heyman, Michal Roitman, Yohai Nurenburg, Krishna Sindhu, Nina Weinberger for the instructive comments during the project and the manuscript. OF thanks the Israeli Science Foundation (ISF grant 1727/20) for their support.

Author contributions

Conceptualization and methodology, all authors; software, validation, formal analysis, investigation, and visualization, all authors; resources, all authors; writing - original draft and review & editing, all authors; supervision and funding acquisition O.F.

Declaration of interests

The authors declare no competing interests.

Data availability

All data presented in this manuscript will be made available in a data repository upon acceptance.

Supplementary information

Individual experiments - Colony maturation

SI Figure 1.

Area per ant from colony maturation experiments

SI Figure 2.

Individual colony maturation experiments reveal a step-like pattern

SI Figure 3.

Age distribution in fixed demographics experiments

SI Figure 4.

Fixed demographics young colonies are similar to colony maturation experiments

SI Figure 5.

Maximum digging rate from fixed demographics experiments

SI Figure 6.

Excavated area, and digging rate from fixed demographics experiments

SI Figure 7.

SI Figure 8.

Nest skeletonization, segmentation and orientation

To estimate the total length of the nests, we created skeleton images of the binary images (SI. Fig 9 A). Initially, images were preprocessed with an averaging filter to smooth the shape edges. The skeleton was derived by MATLAB’s, ‘bwmorph’ function using the ‘thinning’ morphological operator set to infinity. We further performed a ‘Medial Axis Transform’ (MAT) on the nest skeleton, where every skeleton pixel is assigned a distance value to the closest boundary of the excavated area. Many natural nests can be clearly divided into chambers and tunnels, while in other species the division is unclear (52). Moreover, previous studies defines chambers as flat horizontal structures compared to vertical narrow tunnels. With no available data on the architecture of natural nests of Camponotus fellah ants to compare with, we segmented the nest into tunnels, wide tunnels, and chambers using the decision algorithm as described in (SI. Fig 9 B). On the inverse of the binary image, we assigned each pixel in the excavated area to its shortest distance to the edge. Those assigned with a distance of more than 10 pixels (≈ 0.65cm) were designated as seeds for large chambers. Next, a geodesic distance transform assigned pixels the distances to these seeds and outlined the chambers. This created separate blobs, groups of connected pixels in a binary image, for the two classes with a width threshold of 1.30 cm between tunnels and potential chambers. From the width distribution, we choose 3.15 cm and 1.6 cm as high and low thresholds. Blobs above the high threshold were assigned as chambers. Blobs under the low threshold and relatively flat (absolute value of orientation under 30°) and overall regular in shape are added to the first class of tunnels. The residual blobs were defined as wide tunnels. The shape regularity was estimated by blobs with less than 30 % deviation from the mean of the MAT pixels. The resultant segmented images from the colony maturation, fixed demographics young, and old colonies are provided in (Fig. 5 A, B, and C). The orientation analysis was performed to quantify the angular orientation of all tunnels, wide tunnels, and chambers from the initial seeding point concerning the direction of gravity.

SI Figure 9.

Nest architecture

SI Figure 10.

Colony maturation, and fixed demographics experiments

Population change in the colony maturation experiments

We quantified the population growth in colony maturation experiments. Like the nest area, the population growth among the nest varied highly. In some experiments, only one or two workers eclosed (n = 5), while in others, the colony grew more than 15 workers (n = 12). The colony growth increased gradually in all nests until it saturated at a certain level. On average, the population grew logistically, reaching a plateau at roughly 12 ants after around 100 days. We fitted a classical population growth model to describe the population kinetics:

On solving the equation, we get,

We used K = 19.0, the maximal number of ants, and r = −0.032, to describe the population growth. We find that the logistic model can describe the population growth relatively well (R² = 0.98).

Age quantification from colony maturation experiments

During the nest photography session, we meticulously recorded the population of ants inhabiting each nest. By analyzing these population records, we were able to construct a comprehensive age demographics profile for the colony. For instance, consider the population change in the following nest:

On day 10, the population remained constant at 1 ant, with the ant being 10 days old. However, on day 11, the population increased by one, indicating the birth of a new ant. Therefore, on day 11, the newly born ant was 1 day old, while the existing ants were 11 days old. Subsequently, we analyzed the rest of the time points, allowing us to estimate the age distribution of ants in the entire nest.