Module 4

Nest Architecture

From self-organized excavation algorithms to thermoregulated mounds and floating fire ant rafts

4.1 Nest Excavation Algorithms

Ant nests are among the most impressive structures built by any animal relative to body size. A mature leafcutter ant (Atta) colony excavates a nest containing thousands of interconnected chambers reaching 6–8 metres underground, displacing up to 40 tonnes of soil. Yet no individual ant has a blueprint or overview of the nest. Instead, complex architecture emerges from simple local rulesfollowed by individual workers.

Local Excavation Rules

The core algorithm for nest digging can be described by two complementary rules:

Dig where pheromone concentration is high: ants preferentially excavate soil at sites marked with digging pheromone. Successful excavation sites are marked more heavily, creating a positive feedback loop.
Deposit soil pellets at low-concentration sites: ants carry excavated soil pellets away from the tunnel and drop them where pheromone concentration is low, creating organized refuse piles on the surface.

The excavation rate \(R\) at a given location can be modeled as a function of local ant density \(\rho_a\) and soil moisture \(w\):

\[\boxed{R(\rho_a, w) = R_{\max} \cdot \frac{\rho_a^2}{\rho_a^2 + K_a^2} \cdot \frac{w}{w + K_w}}\]

where \(R_{\max}\) is the maximum digging rate, \(K_a\) is the half-saturation density (crowding threshold), and \(K_w\) is the half-saturation moisture content. The \(\rho_a^2\) in the numerator reflects cooperative digging: excavation accelerates when multiple ants work at the same site.

The quadratic dependence on \(\rho_a\) (Hill coefficient \(n=2\)) means that the system is bistable: sites with slightly more ants attract even more, while sites with fewer ants are abandoned. This positive feedback produces the characteristic branching pattern of ant tunnels from an initially featureless soil volume.

Emergent Tunnel Geometry

Despite the simplicity of individual rules, the resulting nest architecture shows remarkable regularity. Tschinkel (2004) made aluminium casts of harvester ant (Pogonomyrmex badius) nests and found:

Chamber spacing increases with depth (roughly logarithmic)
Tunnel diameter is ~2 ant body widths (a traffic constraint)
Chamber area scales with colony size: \(A_{\text{chamber}} \propto N^{0.75}\), consistent with metabolic scaling
Total nest volume: \(V_{\text{nest}} \propto N^{1.0}\) (linear in colony size)

The logarithmic spacing of chambers likely arises from CO\(_2\) diffusion constraints: deeper chambers need more vertical separation to maintain adequate ventilation between levels.

Soil Mechanics

Ants prefer to dig in moist soil (wet sand grains stick together, preventing tunnel collapse). The cohesion from capillary bridges between soil grains follows\(F_{\text{cap}} \propto \gamma R_g \cos\theta_c\), where \(\gamma\)is surface tension, \(R_g\) is grain radius, and \(\theta_c\) is the contact angle. Too wet, and the soil is too heavy; too dry, and tunnels collapse. Optimal moisture is typically 15–25% by weight.

4.1b How an Ant Hill is Built: Structure & Construction

An ant hill is not a random pile of soil — it is a precisely engineered structure with functional zones, climate control, and structural integrity that would impress any architect. A mature wood ant (Formica rufa) mound can be 1.5 m tall and 3 m in diameter, containing over 100 chambers connected by a network of tunnels that extend 2–3 m below ground. The construction involves millions of individual decisions, yet produces a coherent building every time.

The Construction Process: From Soil to Superstructure

Phase 1: Site Selection

Scouts evaluate candidate sites based on soil moisture (15–25% optimal), sun exposure (south-facing slopes preferred in temperate regions), drainage, and proximity to food sources. Selection follows a quorum-sensing mechanism: scouts recruit others to the best site via tandem running, and construction begins when a critical density threshold\(\rho > \rho_{crit} \approx 5\,\text{ants/cm}^2\) is reached.

Phase 2: Initial Excavation

Workers begin digging from the surface, carrying soil pellets one at a time in their mandibles. Each pellet weighs ~1 mg (a 2 mg ant carries 50% of its body weight). The digging rate is\(\sim 1\,\text{pellet per 2 minutes}\) per ant. With 10,000 workers, this produces ~5 kg of excavated soil per day. Tunnels initially follow moisture gradients downward.

Phase 3: Chamber Formation

Chambers form at tunnel junctions where multiple digging fronts meet. Chamber size is self-regulated: ants stop excavating when they can no longer detect pheromones from the walls (chamber radius \(r \approx L_{diffusion} = \sqrt{D\tau}\) where\(D\) is pheromone diffusion and \(\tau\) is detection threshold time). This automatically produces chambers proportional to colony size.

Phase 4: Mound Construction

Excavated soil is deposited above ground in a dome shape. Workers add pine needles, resin, and plant material (wood ants). The mound is not random: it is asymmetric — steeper on the north side, gentler slope facing south for solar heating. Material is cemented by a mixture of soil, saliva, and honeydew that hardens to form a weatherproof crust.

Construction Rate Model

The total mound volume grows following a logistic model constrained by colony size:

\[ \frac{dV_{mound}}{dt} = r_V \cdot N_{workers} \cdot V_{pellet} \cdot \left(1 - \frac{V_{mound}}{V_{max}(N)}\right) \]

where \(r_V \approx 0.5\,\text{pellets/min/ant}\), \(V_{pellet} \approx 1\,\text{mm}^3\), and \(V_{max}(N) = k \cdot N^{1.0}\) (linear scaling with colony size,\(k \approx 0.5\,\text{cm}^3/\text{ant}\)). A colony of 100,000 workers builds a mound of \(\sim 50,000\,\text{cm}^3 = 50\,\text{L}\).

Structural Zones of an Ant Hill

A mature ant hill has distinct functional zones, each optimised for a specific purpose. From top to bottom:

Surface

Weatherproof Crust (0–2 cm)

Outer layer of cemented soil, resin, and plant material. Sheds rainwater (contact angle ~110°). Thermal insulation: k ≈ 0.3 W/(m·K). Repaired within hours of damage.

Shallow

Solar Heating Chamber (2–10 cm)

Shallow chambers just below the south-facing surface. Dark material absorbs solar radiation. Temperature can reach 35–40°C on sunny days. Brood is moved here for accelerated development.

Mid-level

Brood Nurseries (10–40 cm)

Largest chambers in the nest. Eggs, larvae, and pupae sorted by developmental stage. Temperature maintained at 25–30°C. Humidity 60–80%. Workers relocate brood vertically throughout the day to track optimal temperature.

Deep

Queen Chamber (30–80 cm)

Central, well-protected chamber. Slightly larger than brood chambers. Queen surrounded by attendant workers. In polygynous species (multiple queens), several queen chambers may exist.

Variable

Food Storage (variable depth)

Dedicated chambers for stored food: seeds (harvester ants), honeydew crops, fungus gardens (leaf-cutters). Temperature and humidity optimised for preservation. Some ants store liquid food in repletes (living storage vessels).

Deepest

Waste Chambers (deepest)

Refuse, dead ants, and depleted fungus substrate deposited in the deepest chambers. Pathogen containment: workers handling waste do not enter brood chambers (behavioural hygiene). Some species have dedicated waste workers.

Throughout

Ventilation Shafts

Vertical channels connecting deep chambers to the surface. Diameter ~2–5 mm. Create passive airflow via stack effect (ΔT-driven convection). Some species actively fan to regulate CO₂ and humidity.

Structural Engineering Principles

Arch & Dome Construction

Tunnels and chambers use catenary arch geometry — the same shape used in Roman aqueducts. Ants do not “know” the catenary equation, but the arch shape emerges naturally: soil grains that would fall under gravity are removed, while those in compression (supporting the arch) remain. The result is a self-supporting structure that distributes load evenly:

\[ y(x) = a \cosh\!\left(\frac{x}{a}\right), \qquad \sigma_{compression} = \frac{\rho g y}{t_{wall}} \]

where \(a\) is the catenary parameter (related to tunnel width),\(\rho\) is soil density (~1600 kg/m\(^3\)), and \(t_{wall}\) is wall thickness (~2–5 mm for small ant tunnels). Maximum span for ant tunnels: ~20 mm (limited by soil cohesion, not ant engineering).

Pillar & Column Support

In large chambers, ants construct soil pillars to prevent ceiling collapse. The critical column height before buckling (Euler buckling) is:

\[ h_{crit} = \frac{\pi}{2}\sqrt{\frac{EI}{P}}, \qquad I = \frac{\pi r^4}{4} \quad \text{(circular cross-section)} \]

For a soil pillar with \(r = 3\,\text{mm}\), \(E_{soil} \approx 5\,\text{MPa}\), supporting its own weight: \(h_{crit} \approx 40\,\text{mm}\). Ant chambers wider than ~30 mm typically contain 1–3 support columns (observed in Tschinkel aluminium casts).

Waterproofing & Drainage

The mound surface is hydrophobic: ants mix saliva (containing proteinaceous glue) and resin with soil particles, creating a composite with water contact angle \(\theta \approx 110°\). Tunnels near the surface are angled slightly downward to drain rainwater. Deep chambers have drainage channels that redirect groundwater around the nest. Some species (Atta) line chamber walls with a clay layer that acts as a moisture barrier.

Ant Hill Cross-Section: Complete Architecture

Simulation: Ant Hill Construction & Structure

Modelling mound growth dynamics, chamber depth distribution, temperature profiles, and the emergent tunnel network structure.

Python

script.py175 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from matplotlib.gridspec import GridSpec

fig = plt.figure(figsize=(16, 16))
gs = GridSpec(2, 2, figure=fig, hspace=0.35, wspace=0.35)
fig.patch.set_facecolor('#030712')
fig.suptitle("Ant Hill Construction & Architecture", fontsize=16, color='white', fontweight='bold', y=0.98)

# ============================================================
# Panel 1: Mound volume growth over time (logistic model)
# ============================================================
ax1 = fig.add_subplot(gs[0, 0])
ax1.set_facecolor('#1a0505')

days = np.linspace(0, 365 * 3, 500)  # 3 years
colony_sizes = [5000, 20000, 50000, 100000]
colors = ['#60a5fa', '#34d399', '#fbbf24', '#ef4444']
r_v = 0.5  # pellets per min per ant
V_pellet = 1e-3  # cm3 per pellet
k_vol = 0.5  # cm3 per ant

for N, col in zip(colony_sizes, colors):
    V_max = k_vol * N / 1000  # liters
    # Logistic growth: V(t) = V_max / (1 + exp(-r*t))
    r_growth = 0.005  # per day
    V = V_max / (1 + np.exp(-r_growth * (days - 365)))
    ax1.plot(days / 365, V, color=col, linewidth=2.5, label=f'N = {N:,} workers')
    ax1.axhline(V_max, color=col, linestyle=':', linewidth=0.8, alpha=0.4)

ax1.set_xlabel('Time (years)', color='white', fontsize=12)
ax1.set_ylabel('Mound Volume (liters)', color='white', fontsize=12)
ax1.set_title('Mound Growth: Logistic Model\nV_max scales linearly with colony size', color='#fca5a5', fontsize=13, fontweight='bold')
ax1.legend(fontsize=10, facecolor='#1a0505', edgecolor='#ef4444', labelcolor='white')
ax1.tick_params(colors='white')
ax1.grid(True, alpha=0.15, color='#ef4444')
for sp in ax1.spines.values(): sp.set_color('#ef4444')

# ============================================================
# Panel 2: Chamber depth distribution (Tschinkel-style)
# ============================================================
ax2 = fig.add_subplot(gs[0, 1])
ax2.set_facecolor('#1a0505')

# Based on Tschinkel 2004 Pogonomyrmex data
depths_cm = np.array([5, 15, 25, 40, 60, 85, 115, 150, 190, 240])
chamber_areas = np.array([25, 35, 30, 45, 50, 40, 35, 30, 20, 15])
chamber_types = ['Solar', 'Brood', 'Brood', 'Brood', 'Queen', 'Food', 'Food', 'Storage', 'Waste', 'Waste']
type_colors = {
    'Solar': '#fbbf24', 'Brood': '#ef4444', 'Queen': '#f472b6',
    'Food': '#fb923c', 'Storage': '#a78bfa', 'Waste': '#6b7280'
}

for d, a, t in zip(depths_cm, chamber_areas, chamber_types):
    ax2.barh(-d, a, height=8, color=type_colors[t], edgecolor='white', linewidth=0.5, alpha=0.8)
    ax2.text(a + 2, -d, f'{t} ({a} cm\u00B2)', color=type_colors[t], fontsize=9, va='center')

ax2.set_xlabel('Chamber Area (cm\u00B2)', color='white', fontsize=12)
ax2.set_ylabel('Depth (cm)', color='white', fontsize=12)
ax2.set_title('Chamber Distribution with Depth\n(based on Tschinkel 2004 casts)', color='#fca5a5', fontsize=13, fontweight='bold')
ax2.tick_params(colors='white')
ax2.grid(True, alpha=0.15, color='#ef4444', axis='x')
for sp in ax2.spines.values(): sp.set_color('#ef4444')

# Add depth markers
ax2.axhline(-5, color='#fbbf24', linewidth=0.5, linestyle='--', alpha=0.3)
ax2.axhline(-60, color='#f472b6', linewidth=0.5, linestyle='--', alpha=0.3)
ax2.axhline(-150, color='#6b7280', linewidth=0.5, linestyle='--', alpha=0.3)

# ============================================================
# Panel 3: Temperature profile through the mound
# ============================================================
ax3 = fig.add_subplot(gs[1, 0])
ax3.set_facecolor('#1a0505')

depth = np.linspace(-50, 250, 300)  # cm, negative = above ground (mound)
T_ambient = 20  # C
T_soil_deep = 12  # C at 2.5m

# Summer midday profile
T_summer = np.where(depth < 0,
    T_ambient + 15 * np.exp(depth / 20),  # mound heated by sun
    T_ambient + (T_soil_deep - T_ambient) * (1 - np.exp(-depth / 100)) +
    5 * np.exp(-depth / 30))  # metabolic heating near surface

# Winter profile
T_winter = np.where(depth < 0,
    5 + 3 * np.exp(depth / 15),
    5 + (T_soil_deep - 5) * (1 - np.exp(-depth / 80)) +
    8 * np.exp(-((depth - 80)**2) / (2 * 40**2)))  # cluster heating at queen depth

# Optimal brood zone
ax3.axhspan(10, 40, alpha=0.1, color='#ef4444')
ax3.text(32, 25, 'Brood\nzone', color='#ef4444', fontsize=10, fontweight='bold')

ax3.plot(T_summer, -depth, color='#ef4444', linewidth=2.5, label='Summer midday')
ax3.plot(T_winter, -depth, color='#60a5fa', linewidth=2.5, label='Winter')
ax3.axhline(0, color='#86efac', linewidth=1.5, linestyle='-', alpha=0.5)
ax3.text(33, 2, 'Ground surface', color='#86efac', fontsize=10)

ax3.axvline(25, color='#fbbf24', linewidth=1, linestyle=':', alpha=0.5)
ax3.axvline(30, color='#fbbf24', linewidth=1, linestyle=':', alpha=0.5)
ax3.text(27.5, -45, '25-30C\noptimal', color='#fbbf24', fontsize=9, ha='center')

ax3.set_xlabel('Temperature (C)', color='white', fontsize=12)
ax3.set_ylabel('Position (cm, negative = above ground)', color='white', fontsize=12)
ax3.set_title('Temperature Profile Through Ant Hill\nSummer vs Winter', color='#fca5a5', fontsize=13, fontweight='bold')
ax3.legend(fontsize=10, facecolor='#1a0505', edgecolor='#ef4444', labelcolor='white')
ax3.tick_params(colors='white')
ax3.grid(True, alpha=0.15, color='#ef4444')
for sp in ax3.spines.values(): sp.set_color('#ef4444')
ax3.invert_yaxis()

# ============================================================
# Panel 4: Tunnel network — simulated branching
# ============================================================
ax4 = fig.add_subplot(gs[1, 1])
ax4.set_facecolor('#1a0505')

np.random.seed(42)

def grow_tunnel(x0, y0, angle, depth, max_depth=8):
    if depth > max_depth:
        return
    length = 15 + np.random.uniform(-5, 5)
    x1 = x0 + length * np.cos(angle)
    y1 = y0 + length * np.sin(angle)
    lw = max(0.5, 3.5 - depth * 0.35)
    alpha = max(0.3, 1.0 - depth * 0.08)
    color = '#d97706' if depth < 3 else '#ef4444' if depth < 5 else '#94a3b8'
    ax4.plot([x0, x1], [y0, y1], color=color, linewidth=lw, alpha=alpha)

# Chamber at junction
    if depth > 0 and np.random.random() > 0.4:
        r = 3 + np.random.uniform(0, 4)
        c_color = '#ef444430' if depth < 4 else '#fb923c20'
        ax4.add_patch(plt.Circle((x1, y1), r, color=c_color, ec=color, lw=0.5))

# Branch
    if np.random.random() > 0.3:
        grow_tunnel(x1, y1, angle + np.random.uniform(0.3, 0.8), depth + 1, max_depth)
    if np.random.random() > 0.4:
        grow_tunnel(x1, y1, angle - np.random.uniform(0.3, 0.8), depth + 1, max_depth)
    if np.random.random() > 0.6:
        grow_tunnel(x1, y1, angle + np.random.uniform(-0.2, 0.2), depth + 1, max_depth)

# Grow from surface
ax4.axhline(0, color='#86efac', linewidth=2, alpha=0.5)
for x_start in [-10, 0, 15]:
    grow_tunnel(x_start, 0, np.pi/2 + np.random.uniform(-0.3, 0.3), 0)

ax4.set_xlim(-80, 80)
ax4.set_ylim(-5, 120)
ax4.invert_yaxis()
ax4.set_xlabel('Horizontal Extent (cm)', color='white', fontsize=12)
ax4.set_ylabel('Depth (cm)', color='white', fontsize=12)
ax4.set_title('Emergent Tunnel Network\n(recursive branching simulation)', color='#fca5a5', fontsize=13, fontweight='bold')
ax4.tick_params(colors='white')
ax4.grid(True, alpha=0.1, color='#ef4444')
for sp in ax4.spines.values(): sp.set_color('#ef4444')

# Surface marker
ax4.fill_between([-80, 80], [-5, -5], [0, 0], color='#1a0f05', alpha=0.5)
ax4.text(60, -3, 'Surface', color='#86efac', fontsize=11)

plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#030712')
print("=== Ant Hill Construction Simulation ===")
print(f"Colony 100k workers -> mound volume: {k_vol * 100000 / 1000:.0f} liters")
print(f"Construction rate: {0.5 * 100000 * 1e-3 * 60 * 8 / 1000:.1f} liters/day (100k colony)")
print(f"Summer mound surface temp: {T_ambient + 15:.0f} C")
print(f"Winter queen chamber temp: {max(T_winter):.1f} C (metabolic heating)")
print(f"Optimal brood zone: 10-40 cm depth, 25-30 C")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

4.2 Thermoregulation

Many ant species actively regulate the temperature of their nests, especially for brood development. The most impressive thermoregulators are the wood ants (Formica), which build large thatched mounds that function as solar-heated incubators, maintaining brood chambers at 25–30 \(^\circ\)C even when external temperatures fluctuate wildly.

Heat Equation for a Hemispherical Mound

Consider a simplified model of the nest mound as a hemispherical dome of radius \(R\)with uniform internal temperature \(T\). The thermal energy balance is:

\[\boxed{C \frac{dT}{dt} = Q_{\text{metabolic}} + Q_{\text{solar}} - h \cdot A \cdot (T - T_{\text{amb}})}\]

where:

\(C\) = thermal heat capacity of the mound (J/K)
\(Q_{\text{metabolic}}\) = metabolic heat production of all ants (W)
\(Q_{\text{solar}}\) = solar radiation absorbed by the dome surface (W)
\(h\) = heat transfer coefficient (W/m\(^2\)/K)
\(A = 2\pi R^2\) = surface area of the hemisphere
\(T_{\text{amb}}\) = ambient temperature

At steady state (\(dT/dt = 0\)):

\[T_{\text{ss}} = T_{\text{amb}} + \frac{Q_{\text{metabolic}} + Q_{\text{solar}}}{h \cdot A}\]

For a large Formica rufa mound (\(R \approx 0.75\) m, ~500,000 ants),\(Q_{\text{metabolic}} \approx 2\)–5 W (ants produce ~5–10 \(\mu\)W each) and \(Q_{\text{solar}}\) can reach 10–50 W on a sunny day. The dark thatch material (resin-treated plant debris) absorbs solar radiation efficiently (\(\alpha \approx 0.85\)).

Active Thermoregulation Behaviours

Wood ants supplement passive solar heating with active behaviours:

Solar basking: in the morning, worker ants emerge to bask on the sunny side of the mound, absorbing solar radiation. They then re-enter and release their body heat in the brood chambers — acting as living heat-transfer units.
Ventilation control: ants open or close surface openings to regulate convective heat loss.
Brood relocation: ants move brood to warmer or cooler chambers depending on temperature — effectively choosing the optimal microclimate.

Each individual ant acts as a thermostat with a simple rule: if the local temperature exceeds the set-point, move brood away or open vents; if too cold, bring warm bodies in or close vents. The colony-level regulation emerges from many such local decisions.

Leafcutter Fungus Garden Temperature

Leafcutter ants (Atta and Acromyrmex) cultivate the fungus Leucoagaricus gongylophorus as their sole food source. This fungus requires a remarkably narrow temperature range: \(25 \pm 1\,^\circ\)C. The ants achieve this precision deep underground where temperature fluctuations are damped by thermal inertia of the soil:

\[T(z, t) = \bar{T} + \Delta T \cdot e^{-z/\delta} \cos\left(\omega t - \frac{z}{\delta}\right)\]

where \(z\) is depth, \(\delta = \sqrt{2\kappa/\omega}\) is the thermal penetration depth, \(\kappa\) is soil thermal diffusivity, and\(\omega = 2\pi/\text{year}\) for seasonal variations. At typical fungus garden depths (2–4 m), seasonal temperature oscillations are reduced to less than 1\(^\circ\)C.

4.3 Ventilation & Gas Exchange

Underground ant nests face a critical challenge: CO\(_2\) accumulation and O\(_2\)depletion. A colony of 500,000 ants produces CO\(_2\) at roughly the same rate as a small mammal. Without ventilation, CO\(_2\) levels in deep chambers can reach 3–6% (vs. 0.04% in ambient air), which inhibits fungus growth and is physiologically stressful for the ants themselves.

Stack Ventilation (Chimney Effect)

Harvester ant nests (Pogonomyrmex) and many other species exploit the stack effect (also called chimney effect or Bernoulli-driven ventilation) to passively ventilate their nests. The basic mechanism is analogous to how a chimney works:

Warm air inside the nest is less dense than cool external air, creating a pressure difference that drives upward convection through the tunnel system. The driving pressure difference is:

\[\boxed{\Delta P = \rho_{\text{ext}} \cdot g \cdot h \cdot \frac{T_{\text{in}} - T_{\text{out}}}{T_{\text{out}}}}\]

where \(\rho_{\text{ext}}\) is external air density,\(g\) is gravitational acceleration, \(h\) is the effective chimney height (vertical extent of the tunnel system), and temperatures are in Kelvin.

This can be derived from Bernoulli's equation and the ideal gas law. The density difference between inside and outside air is:

\[\Delta\rho = \rho_{\text{ext}} - \rho_{\text{int}} = \rho_{\text{ext}} \left(1 - \frac{T_{\text{out}}}{T_{\text{in}}}\right) \approx \rho_{\text{ext}} \frac{T_{\text{in}} - T_{\text{out}}}{T_{\text{out}}}\]

The hydrostatic pressure difference over height \(h\) is then\(\Delta P = \Delta\rho \cdot g \cdot h\), which gives the boxed expression above.

Wind-Driven Ventilation

In many species, surface wind also drives ventilation. Nests with multiple openings at different heights experience differential wind pressure (Bernoulli effect):

\[\Delta P_{\text{wind}} = \frac{1}{2}\rho v_{\text{wind}}^2 (C_{p,\text{upper}} - C_{p,\text{lower}})\]

where \(C_p\) are pressure coefficients that depend on opening orientation relative to wind direction. Leafcutter ant nests in tropical forests show a distinctive pattern: central chimneys (turrets) surrounded by lower peripheral openings, optimizing wind-driven flow.

The volumetric flow rate through the nest follows the Hagen-Poiseuille law for each tunnel segment:

\[Q = \frac{\pi r^4}{8\mu} \frac{\Delta P}{L}\]

where \(r\) is the tunnel radius, \(\mu\) is air viscosity, and\(L\) is tunnel length. This imposes a strong constraint on tunnel geometry: flow scales as \(r^4\), so slightly wider tunnels dramatically improve ventilation.

CO\(_2\)-Driven Active Ventilation

When passive ventilation is insufficient, some species use active fanning. Workers station themselves in tunnels and fan their wings (vestigial in workers, but still functional for air movement). The CO\(_2\) threshold for initiating fanning behaviour is approximately 2–3%, suggesting individual ants can sense CO\(_2\)concentrations and respond accordingly.

4.4 Fire Ant Rafts

When Solenopsis invicta (red imported fire ant) colonies are flooded — a common event in their native South American floodplains — the entire colony self-assembles into a floating raft within minutes. This raft can persist for weeks, carrying the queen, brood, and thousands of workers across flood waters.

Superhydrophobicity and Buoyancy

Individual fire ants are coated with a waxy cuticle that makes them mildly hydrophobic. When they link together, the spaces between their bodies trap a layer of air. This air plastron serves two purposes: (1) it provides buoyancy, and (2) it allows submerged ants to breathe through their spiracles.

The buoyancy condition for the raft to float is:

\[\boxed{\rho_{\text{water}} \cdot g \cdot V_{\text{submerged}} = N \cdot m_{\text{ant}} \cdot g}\]

The effective density of the raft is much less than the density of individual ant tissue (~1.1 g/cm\(^3\)) because of the trapped air:

\[\rho_{\text{raft}} = \frac{N \cdot m_{\text{ant}}}{V_{\text{total}}} = \rho_{\text{ant}} \cdot (1 - f_{\text{air}})\]

where \(f_{\text{air}}\) is the air volume fraction. Measurements by Mlot et al. (2011) found \(f_{\text{air}} \approx 0.25\)–0.45, giving an effective raft density of 600–800 kg/m\(^3\) — well below the density of water.

Viscoelastic Material Properties

The fire ant raft is a remarkable material: it behaves as both a solid and a liquid, depending on the timescale. Hu and colleagues (2016) performed rheological measurements and found:

Short timescales: the raft is elastic (solid-like). When quickly deformed, it springs back. Storage modulus\(G' \approx 5\)–10 Pa.
Long timescales: the raft flows like a viscous liquid. Ants continuously rearrange, allowing the raft to spread and conform to container shapes. Loss modulus \(G'' \approx 2\)–5 Pa.

This viscoelastic behaviour can be described by a Maxwell model with a characteristic relaxation time \(\tau\):

\[G(t) = G_0 \, e^{-t/\tau}, \qquad \tau = \frac{\eta}{G_0}\]

where \(G_0 \approx 8\) Pa is the instantaneous shear modulus and\(\eta \approx 80\) Pa\(\cdot\)s is the viscosity. The relaxation time is \(\tau \approx 10\) s, meaning the raft behaves as a solid for perturbations faster than ~10 s (waves), but flows for slower perturbations (spreading).

Self-Assembly Mechanism

Raft assembly follows a simple algorithm:

Ants at the colony surface grip their neighbours with tarsal claws and mandibles
Water level rises; connected ants float as a coherent mass
Ants on the bottom (submerged side) actively push outward, spreading the raft
Submerged ants are periodically rotated to the surface (cycling every ~30 minutes) — no individual drowns

The linking force between two ants (via tarsal grip) is approximately\(F_{\text{grip}} \approx 400 \times m_{\text{ant}} \cdot g\), meaning each connection can hold 400 times the ant's body weight. This extraordinary grip strength (relative to body mass) is why the raft holds together even in turbulent water.

Ecological Significance

The rafting ability of S. invicta is a key factor in its success as an invasive species. Rafts can travel for kilometres on flood waters, allowing rapid colonization of new territories. This is believed to be how fire ants spread from South America to the southern United States after arriving at the port of Mobile, Alabama in the 1930s.

4.5 Nest Mound Cross-Section

Cross-section of a wood ant (Formica) nest mound showing the internal chamber structure, temperature gradient from warm interior to cool exterior, and ventilation flow patterns.

4.6 Simulation: Nest Temperature & Fire Ant Raft Buoyancy

This simulation models four aspects of nest physics: (1) daily temperature regulation showing buffered nest temperature vs. fluctuating ambient, (2) seasonal variation with metabolic heating from variable colony size, (3) stack ventilation pressure as a function of temperature differential, and (4) fire ant raft buoyancy as a function of colony size and air content.

Python

script.py146 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

np.random.seed(42)

fig, axes = plt.subplots(2, 2, figsize=(14, 12))
fig.patch.set_facecolor('#0a0a0a')
for ax in axes.flat:
    ax.set_facecolor('#0a0a0a')
    for spine in ax.spines.values():
        spine.set_color('#444')
    ax.tick_params(colors='#aaa', labelsize=9)

# --- Panel 1: Nest Temperature Regulation (daily cycle) ---
ax1 = axes[0, 0]
hours = np.linspace(0, 72, 500)

T_amb_base = 15.0
T_amp_day = 12.0
T_amb = T_amb_base + T_amp_day * np.maximum(0, np.sin(2*np.pi*hours/24 - np.pi/4))

Q_metabolic = 2.5
Q_solar_peak = 5.0
Q_solar = Q_solar_peak * np.maximum(0, np.sin(2*np.pi*hours/24 - np.pi/4))**2
h_coeff = 0.15
C_thermal = 8.0
A_surface = 3.0

T_nest = np.zeros_like(hours)
T_nest[0] = 26.0
dt = hours[1] - hours[0]
for i in range(1, len(hours)):
    dTdt = (Q_metabolic + Q_solar[i] - h_coeff * A_surface * (T_nest[i-1] - T_amb[i])) / C_thermal
    T_nest[i] = T_nest[i-1] + dTdt * dt
    T_nest[i] = np.clip(T_nest[i], 10, 40)

ax1.plot(hours, T_amb, '--', color='#60a5fa', linewidth=1.5, alpha=0.7, label='Ambient T')
ax1.plot(hours, T_nest, '-', color='#f87171', linewidth=2.5, label='Nest interior T')
ax1.axhline(y=25, color='#4ade80', linestyle=':', linewidth=1, alpha=0.5, label='Optimal (25 C)')
ax1.axhline(y=30, color='#4ade80', linestyle=':', linewidth=1, alpha=0.5, label='Optimal (30 C)')
ax1.fill_between(hours, 25, 30, alpha=0.1, color='#4ade80')

for d in range(3):
    ax1.axvspan(d*24 + 6, d*24 + 18, alpha=0.05, color='#facc15')
    ax1.text(d*24 + 12, 38, f'Day {d+1}', color='#888', fontsize=9, ha='center')

ax1.set_title('Nest Temperature Regulation (3-day cycle)', color='#f87171', fontsize=12, fontweight='bold')
ax1.set_xlabel('Time (hours)', color='#ccc', fontsize=10)
ax1.set_ylabel('Temperature (C)', color='#ccc', fontsize=10)
ax1.legend(fontsize=8, facecolor='#111', edgecolor='#444', labelcolor='#ccc', loc='lower right')
ax1.set_xlim(0, 72)
ax1.grid(True, alpha=0.15)

# --- Panel 2: Seasonal Temperature Variation ---
ax2 = axes[0, 1]
days = np.arange(0, 365)
T_summer = 30 + 5*np.sin(2*np.pi*days/365 - np.pi/2)
T_winter = 5 + 10*np.sin(2*np.pi*days/365 - np.pi/2)
T_amb_seasonal = (T_summer + T_winter) / 2

N_ants_seasonal = 5000 + 3000*np.sin(2*np.pi*days/365 - np.pi/2)
N_ants_seasonal = np.clip(N_ants_seasonal, 1000, 8000)
Q_met_seasonal = N_ants_seasonal * 0.0005

T_nest_seasonal = np.zeros_like(days, dtype=float)
T_nest_seasonal[0] = 20.0
for i in range(1, len(days)):
    dT = (Q_met_seasonal[i] + 2.0*np.maximum(0, np.sin(2*np.pi*days[i]/365 - np.pi/2)) -
          0.1*3.0*(T_nest_seasonal[i-1] - T_amb_seasonal[i])) / C_thermal
    T_nest_seasonal[i] = T_nest_seasonal[i-1] + dT
    T_nest_seasonal[i] = np.clip(T_nest_seasonal[i], 5, 35)

months = ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec']
month_ticks = [0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334]

ax2.plot(days, T_amb_seasonal, '--', color='#60a5fa', linewidth=1.5, alpha=0.7, label='Ambient (mean)')
ax2.plot(days, T_nest_seasonal, '-', color='#f87171', linewidth=2.5, label='Nest interior')
ax2.fill_between(days, 25, 30, alpha=0.1, color='#4ade80', label='Optimal zone')
ax2.set_xticks(month_ticks)
ax2.set_xticklabels(months, fontsize=8, color='#aaa')
ax2.set_title('Seasonal Nest Temperature', color='#f87171', fontsize=12, fontweight='bold')
ax2.set_xlabel('Month', color='#ccc', fontsize=10)
ax2.set_ylabel('Temperature (C)', color='#ccc', fontsize=10)
ax2.legend(fontsize=8, facecolor='#111', edgecolor='#444', labelcolor='#ccc')
ax2.grid(True, alpha=0.15)

# --- Panel 3: Ventilation Stack Effect ---
ax3 = axes[1, 0]
T_ext = np.linspace(10, 40, 100)
T_int = 28.0
h_stack = np.array([0.2, 0.5, 1.0, 2.0])
rho = 1.2
g = 9.81

colors_v = ['#60a5fa', '#4ade80', '#facc15', '#f87171']
for j, h in enumerate(h_stack):
    delta_P = rho * g * h * (T_int - T_ext) / (T_ext + 273.15)
    ax3.plot(T_ext, delta_P, '-', color=colors_v[j], linewidth=2, label=f'h = {h} m')

ax3.axhline(y=0, color='white', linewidth=0.5, alpha=0.3)
ax3.axvline(x=T_int, color='#888', linewidth=1, linestyle='--', alpha=0.5, label=f'T_int = {T_int} C')
ax3.set_title('Stack Ventilation Pressure', color='#f87171', fontsize=12, fontweight='bold')
ax3.set_xlabel('External Temperature (C)', color='#ccc', fontsize=10)
ax3.set_ylabel('Pressure difference (Pa)', color='#ccc', fontsize=10)
ax3.legend(fontsize=8, facecolor='#111', edgecolor='#444', labelcolor='#ccc')
ax3.grid(True, alpha=0.15)

ax3.text(0.02, 0.95, 'Positive: upward flow\nNegative: downward flow',
    transform=ax3.transAxes, color='#888', fontsize=9, va='top',
    bbox=dict(boxstyle='round', facecolor='#1a1a1a', edgecolor='#444'))

# --- Panel 4: Fire Ant Raft Buoyancy ---
ax4 = axes[1, 1]
N_ants_raft = np.linspace(100, 50000, 200)
m_ant = 1.5e-3
rho_water = 1000.0
rho_raft_vals = [400, 600, 800]
colors_r = ['#4ade80', '#facc15', '#f87171']

for j, rho_r in enumerate(rho_raft_vals):
    total_mass = N_ants_raft * m_ant
    V_total = total_mass / rho_r
    V_submerged = total_mass / rho_water
    fraction_submerged = V_submerged / V_total
    fraction_submerged = np.clip(fraction_submerged, 0, 1)
    ax4.plot(N_ants_raft/1000, fraction_submerged*100, '-', color=colors_r[j],
        linewidth=2, label=f'rho_raft = {rho_r} kg/m3')

ax4.axhline(y=100, color='white', linewidth=1, linestyle='--', alpha=0.3, label='Sinking threshold')
ax4.set_title('Fire Ant Raft: Fraction Submerged', color='#f87171', fontsize=12, fontweight='bold')
ax4.set_xlabel('Number of ants (thousands)', color='#ccc', fontsize=10)
ax4.set_ylabel('Fraction submerged (%)', color='#ccc', fontsize=10)
ax4.legend(fontsize=8, facecolor='#111', edgecolor='#444', labelcolor='#ccc')
ax4.grid(True, alpha=0.15)

ax4.text(0.5, 0.55, 'Air layer trapped between\nhydrophobic ant bodies\nreduces effective density',
    transform=ax4.transAxes, color='#888', fontsize=9, ha='center',
    bbox=dict(boxstyle='round', facecolor='#1a1a1a', edgecolor='#444'))

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#0a0a0a')
plt.close()

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Key Observations

Daily regulation: the nest interior (red) remains within the 25–30\(^\circ\)C optimal zone (green band) while ambient temperature swings from 15\(^\circ\)C to 27\(^\circ\)C.
Seasonal variation: metabolic heat from colony size changes helps buffer winter temperatures, but regulation is less effective in the coldest months.
Stack ventilation: pressure difference reverses sign when external temperature exceeds internal — explaining why nest ventilation patterns change on hot summer days.
Fire ant raft: the fraction submerged is independent of colony size (a dimensional analysis result) and depends only on the effective raft density. Air trapping reduces this below 100%, ensuring flotation.

References

Tschinkel, W. R. (2004). The nest architecture of the Florida harvester ant, Pogonomyrmex badius. Journal of Insect Science, 4(1), 21.
Mlot, N. J., Tovey, C. A., & Hu, D. L. (2011). Fire ants self-assemble into waterproof rafts to survive floods. Proceedings of the National Academy of Sciences, 108(19), 7669–7673.
Hu, D. L., Phonekeo, S., Alvarado, E., & Goldberg, D. (2016). Rheology of fire ant aggregations. Proceedings of the National Academy of Sciences, 113, 2017.
Rosengren, R., Fortelius, W., Lindström, K., & Luther, A. (1987). Phenology and causation of nest heating and thermoregulation in red wood ants of the Formica rufa group studied in coniferous forest habitats in southern Finland. Annales Zoologici Fennici, 24, 147–155.
Kleineidam, C., Ernst, R., & Roces, F. (2001). Wind-induced ventilation of the giant nests of the leaf-cutting ant Atta vollenweideri. Naturwissenschaften, 88(7), 301–305.
Bollazzi, M., & Roces, F. (2007). To build or not to build: circulating dry air organizes collective building for climate control in the leaf-cutting ant Acromyrmex ambiguus. Animal Behaviour, 74(5), 1349–1355.
King, H., Ocko, S., & Mahadevan, L. (2015). Termite mounds harness diurnal temperature oscillations for ventilation. Proceedings of the National Academy of Sciences, 112(37), 11589–11593.
Mueller, U. G., Gerardo, N. M., Aanen, D. K., Six, D. L., & Schultz, T. R. (2005). The evolution of agriculture in insects. Annual Review of Ecology, Evolution, and Systematics, 36, 563–595.

←M3: Navigation & Orientation M5: Colony Organization→