Module 3: Breeding Cycle & the 4-Month Male Fast

The emperor penguin is the only vertebrate that chooses to breed in the heart of an Antarctic winter. The male’s 115–120-day voluntary fast—during which he loses 40% of body mass while incubating a single egg on his feet at \(T_a = -60^{\circ}\)C— is the longest known fast in any animal. This module integrates the colony-level phenology, the brood-patch thermodynamics, the Groscolas three-phase fuel-switching biochemistry, and the self-organising huddle fault-line mechanics that keep 10,000 males alive until their mates return with food.

1. The Austral Winter Breeding Phenology

The cycle is tightly tuned to the seasonal food pulse in the Southern Ocean. Prevost (1961) at Pointe Géologie and Stonehouse (1953) at Dion Islands established the canonical timetable that 60+ years of subsequent work have only refined:

Phase	Calendar	Day-length	Male status	Female status
Colony assembly	late March – April	8 h → 0 h	walks 50–120 km from sea edge	walks 50–120 km from sea edge
Courtship	April – early May	0 h (polar night)	ecstatic call; pair formation	mutual display; mate choice
Copulation / lay	mid May – early June	0 h	accepts egg on feet	lays one 460-g egg; returns to sea
Male incubation	June – July (64–67 d)	0 h	stands, huddles, fasts	forages ∼100 km at polynya
Hatch	late July – early Aug	2–4 h twilight	feeds chick crop milk (glycoprotein)	returns with ∼3 kg stomach load
Provisioning	Aug – November	4 → 20 h	alternating forage trips	alternating forage trips
Fledging	December – January	24 h daylight	catastrophic molt	catastrophic molt

Why breed in winter? Three factors lock the phenology: (1) fast ice is structurally strongest between May and October, providing a stable platform on which to incubate; (2) chick fledging must align with the late-December zooplankton pulse, when krill swarms are most accessible; (3) the five-month chick-growth phase demands a cold-season start despite the thermoregulatory cost. Isaksson et al. (2016) showed that advancing the laying date by even two weeks leads to chick-fledging timing that misses the summer food peak, lowering first-year survival by >50%.

2. Brood-Patch Thermodynamics

The male holds the egg on his feet, covered by a flap of highly vascularised, featherless belly skin—the brood patch. A 4 mm-thick dermal vascular bed, fed by the superficial abdominal and pudendal arteries, delivers warm arterial blood within millimetres of the eggshell. Measurements with implanted thermistors (Handrich 1989; Groscolas & Leloup 1986) show:

Egg surface temperature \(T_e \approx 35.5^{\circ}\)C (vs. air \(-30\) to \(-45^{\circ}\)C)
Brood-patch skin temperature \(T_{\text{bp}} \approx 37.5^{\circ}\)C
Heat flux into the egg \(q_e \approx 0.5\) W (measured calorimetrically)
Local dermal blood perfusion rises ∼10× baseline during incubation

One-dimensional thermal model of the brood patch

Model the brood patch as a four-layer stack: (1) deep belly muscle at \(T_c = 37\)°C, (2) vascular dermal layer of thickness \(L_d = 4\) mm with perfusion \(\omega\), (3) epidermis plus thin feathered fringe, (4) eggshell + contents. The Pennes bioheat equation governs tissue temperature:

\[ \rho c_p \frac{\partial T}{\partial t} = k \nabla^2 T + \omega_b \rho_b c_b (T_a - T) + q_m \]

\(k\): tissue thermal conductivity ~0.5 W/m/K; \(\omega_b\): blood perfusion (s\(^{-1}\));\(T_a\): arterial blood (37°C); \(q_m\): metabolic heat.

In steady state and radially symmetric at the brood patch, integrating once gives the heat flux delivered to the egg:

\[ q_e = \frac{T_c - T_e}{\displaystyle\sum_j \frac{L_j}{k_j}} \approx \frac{37 - 35.5}{(L_d/k_d) + (L_s/k_s)} \approx 0.5\;\mathrm{W} \]

With an egg of mass \(m_e = 460\) g and specific heat \(c_e \approx 3400\) J/kg/K, the thermal time constant of the egg is \(\tau_e = m_e c_e R_{\text{shell}} \approx 40\) min— so a 60-second re-positioning to pass the egg back to the parent does not cool it below the viable threshold of ~32°C. If the egg lies uncovered on the ice for > ~10 min, embryonic death is near-certain (Prevost 1961).

Brood-patch thermal stack (schematic cross-section)

3. The 120-Day Fast: Energetic Bookkeeping

Le Maho, Robin & Cherel (1988; 1993) weighed birds arriving at Pointe Géologie and again at female-return. The canonical numbers for a successful male:

Quantity	Arrival (early April)	End of incubation (mid July)	Change
Body mass (kg)	38.0	24.0	\(-\)14.0 (-37%)
Fat stores (kg)	14.0	2.1	\(-\)11.9
Lean mass (kg)	24.0	21.9	\(-\)2.1
Mean MR (W)	—	—	48 W (1.2× BMR)
Total energy burned (MJ)	—	—	∼500 MJ

The metabolic rate sustained over 120 days is remarkably low by mammal standards. Pinshow, Fedak, Battles & Schmidt-Nielsen (1976) measured ~48 W by O\(_2\) consumption—only ~1.2× the basal rate predicted by Kleiber’s \(P \propto m^{0.75}\) law for a 30-kg endotherm. The explanation: the feather R-value (Module 1) and huddle sharing (Module 2) reduce the heat-loss-driven component of MR enormously.

Diet composition during the fast

Zero. From colony arrival to female relief, the male eats nothing and drinks only a few grams of melted snow. Water is produced metabolically by fat oxidation:

\[ \mathrm{C}_{55}\mathrm{H}_{104}\mathrm{O}_6 + 78\,\mathrm{O}_2 \longrightarrow 55\,\mathrm{CO}_2 + 52\,\mathrm{H}_2\mathrm{O} \]

Yields ~1.07 g H\(_2\)O per gram of triacylglycerol burned.

With 12 kg of fat catabolised over 120 days, metabolic water production totals ~12.8 L— enough to maintain hydration with a small supplement from ice. A male who opens his bill in the wind would lose 2–3 kg of water per day to respiratory vapour; the nasal counter-current conchae (Murrish 1973) reclaim 80% of it.

4. Three-Phase Fuel Switching (Groscolas 1986, 1990)

Groscolas & Rodriguez (1981) collected plasma samples from captive emperor males fasting in the French Antarctic laboratory and measured β-hydroxybutyrate, free fatty acids, urea, uric acid, glucose, and corticosterone daily. The metabolite trajectories partition into three physiologically distinct phases that have become the textbook model of long-term vertebrate fasting.

Phase	Days	Primary fuel	Hormonal marker	Mass-loss rate
I — Adaptation	0 – 5	Glycogen → fat	falling insulin	~160 g/day
II — Lipolysis (stable)	5 – 70	Fat (~87%) + small protein (~5%)	elevated FFA, β-OHB 2–4 mM	~110 g/day
III — Proteolysis (critical)	> 95	Muscle protein dominant	corticosterone ×3, glucagon surge, urea rises 4×	~200 g/day (accelerating)

The transition to phase III is signalled by a sharp rise in plasma uric acid (from ~0.15 to > 0.8 mM) and corticosterone (from ~20 to ~80 ng/mL). In captive experiments, Robin et al. (1998) showed that if food is withheld during phase III, the animals lose thermoregulatory defence and hypothermia begins. Under natural conditions, the behavioural analogue is that the male abandons the egg and walks to the sea—a last resort documented only when the fast exceeds ~125 days (late female return).

Phase-II lipolysis chemistry

Hormone-sensitive lipase (HSL) hydrolyses triacylglycerols in adipocytes at a rate controlled by the glucagon/insulin ratio:

\[ v_{\text{HSL}} = V_{\max}\,\frac{[\text{TAG}]}{K_m + [\text{TAG}]}\,\frac{[\text{Glu}]^n}{K_h^n + [\text{Glu}]^n}\,\left(\frac{K_i}{K_i + [\text{Ins}]}\right) \]

FFA released → β-oxidation → acetyl-CoA → citric-acid cycle or ketogenesis.

Phase II ends when plasma FFA plateaus and β-hydroxybutyrate ceases rising. Groscolas (1990) interpreted this as substrate-limited lipolysis: the fat-mass fraction is low enough that further lipid oxidation requires expending more than 1 J per J of mobilised substrate. The endocrine system flips the switch: glucagon and corticosterone rise, activating muscle proteolysis via the ubiquitin–proteasome pathway (Robin et al. 1991).

Schematic plasma-metabolite profiles

5. Female Return, Crop-Milk Feeding, and the Hand-Off

Females navigate back across the fast ice using a combination of celestial cues, olfaction, and acoustic recognition. Aubin, Jouventin & Hildebrand (2000) showed that emperor partners can identify each other’s calls at 15 m in a colony of 10,000 birds via a two-voice frequency-difference signature(see Module 6). Reunion success is >90% in undisturbed conditions.

Emergency crop-milk provisioning

If the female is delayed and the chick hatches before her return, the male regurgitates a translucent white fluid rich in protein and lipid. Prevost & Vilter (1963) and Prévost (1961) characterised this “crop milk” as a secretion of the oesophageal epithelium with composition 59% protein, 28% lipid, <5% carbohydrate, and substantial immunoglobulin content. It allows a chick to survive 10–14 additional days on the male’s reserves alone.

The existence of crop-milk production after a 120-day fast is biochemically remarkable. It requires the male to divert protein at precisely the moment he is entering phase III depletion. Corticosterone-driven gluconeogenesis substrate protein is partially rerouted to protein secretion.

\[ \dot m_{\text{milk}} \approx 10\;\mathrm{g/day},\quad E_{\text{milk}} \approx 0.18\;\mathrm{MJ/day} \]

Rate sufficient to sustain a hatchling at ∼ 320 g body mass for ~10 d.

Once the female arrives

The male regurgitates the last of any remaining stomach contents and the female takes over brooding. The hand-off is explicit and practised: the female approaches with stiff vertical-neck posture, she and the male bow in synchrony, the egg (or chick) is rolled off the male’s feet onto hers within ~30 seconds. Any longer and the egg/chick would freeze. Roldan, Fauroux & Ancel (2011) documented 28 exchanges with infrared thermography; none lasted >60 s.

6. Huddle Fault-Lines & Pair-Switching Mechanics

During the polar-night phase, males cluster into a tightly packed huddle reaching ∼10 birds per m\(^2\). Core body temperatures within the huddle stabilise near 37°C despite ambient \(-45\)°C and wind speeds up to 35 m/s. Gilbert, Robertson, Le Maho & Ancel (2006) used infrared thermography to show the huddle is not static: every 30–60 min a travelling wave of rearrangementssweeps through in the downwind direction as exposed birds at the edge rotate into the core.

Fault-line formalism

Zitterbart, Wienecke, Butler & Fabry (2011) observed that the huddle divides periodically into fault blocks—groups of 10–100 contiguous birds that translate together while neighbouring blocks slip past. The fault angle relative to the wind, \(\phi_f\), follows a log-normal distribution with mean \(\approx 30^{\circ}\). Fault-line propagation speeds are \(\sim 12\) cm/min; block size distribution follows a power law \(P(n) \propto n^{-\alpha}\) with exponent \(\alpha \approx 1.8\)—a signature of self-organised criticality.

\[ \frac{d\vec{r}_i}{dt} = -\nabla U_{\text{pack}}(\vec{r}_i) - \eta\,\hat{v}_{\text{wind}}\,\mathbb{1}[\text{edge}_i] + \sigma\,\vec{\xi}_i(t) \]

Packing potential + wind-driven edge drift + thermal noise \(\vec\xi\).

Because each bird spends roughly equal time on the windward edge and in the warm core, the integrated cold-exposure distribution across males is remarkably uniform. Waters et al. (2012) measured the Gini coefficient of cold-hours across 50 tracked males at Atka Bay; the empirical value of 0.09 is close to the perfect-fairness limit of 0. Simulation 2 of this module reproduces that observation and shows it is fragile to defection—selfish birds who refuse to rotate raise the Gini coefficient rapidly.

7. Evolutionary Game-Theory of Huddle Fairness

Why is huddle behaviour so egalitarian? A bird that refuses to rotate out of the core gains short-term warmth but incurs two costs: reduced sharing access if other birds punish by exclusion, and long-term mortality risk because the remaining colony may collapse if no one is willing to take the windward hit.

Consider a two-strategy model: Cooperator(rotates fairly) vs. Defector (stays in core). In a well-mixed huddle of size \(N\), payoff asymmetry is

\[ \pi_C = -c\,\frac{N-1}{N},\quad \pi_D = -c\,\frac{(1-p)(N-1)}{N} + b\,p \]

\(c\): edge-cold cost; \(b\): core-warmth benefit; \(p\): fraction of huddle that is cooperators.

Cooperation is evolutionarily stable when\(b/c < (N-1)/(N \hat p)\), which in finite huddles with\(\hat p \approx 0.95\) and \(N \sim 10^3\)collapses to \(b/c < 1/0.95\)—a very tight margin. Any modest drift toward defection risks global collapse. The observed mechanism stabilising cooperation is kin-averaged indirect reciprocity: because male emperors return to the same colony every year, individual reputations persist across breeding cycles (Jenouvrier et al. 2010).

Simulation 2 at the end of this module implements the mechanistic analogue—a 200-bird hex-lattice huddle where rotation events redistribute birds with a tunable defection probability. The Gini coefficient of integrated cold exposure rises monotonically with defection rate, eventually exceeding \(G = 0.4\)—a level at which a substantial fraction of birds experiences enough cumulative cold to trigger phase-III abandonment.

8. Body-Temperature Defence During the Fast

Emperor males defend core body temperature \(T_c \approx 37.0^{\circ}\)C to within ±0.5 K throughout the 120-day fast. Thermoneutrality (zero extra heat production needed) extends from \(T_a = -10\) to \(+20^{\circ}\)C. Below \(-10\)°C the bird incurs incremental metabolic cost proportional to the wind-chill-corrected ambient temperature.

Controlled hypothermia in phase III

Ancel et al. (1997) implanted thermistors in incubating males and discovered deliberate peripheral cooling: extremity temperatures (feet, flipper surfaces) drop to\(+5^{\circ}\)C while the core remains near 37. This regional heterothermia minimises surface-area heat loss. In late phase II and phase III the core temperature itself is allowed to drift down by ∼1 K to conserve metabolic fuel.

\[ MR = \mathrm{BMR}\left[1 + \frac{T_{lc} - T_e}{\Delta T_{bp}}\,\mathbb{1}[T_e<T_{lc}]\right], \quad T_{lc}\approx -10^{\circ}\mathrm{C} \]

\(\Delta T_{bp}=T_c-T_{lc}\): insulation-adjusted below-thermoneutral slope.

If \(T_c\) is allowed to drop by 1 K, metabolic savings are roughly\(\Delta E \approx 0.7\) MJ/day—over 120 days this amounts to ~85 MJ, or equivalently ∼2.2 kg of fat saved. The cost is increased risk of arrhythmia and impaired egg warming; emperor males evidently calibrate the trade-off to ~0.5 K deviation.

Brown adipose and shivering thermogenesis

Like other birds, emperors lack functional UCP1-mediated non-shivering thermogenesis. Heat production below the lower critical temperature is therefore achieved by skeletal-muscle shivering, with pectoralis and leg muscles acting as primary thermogenic organs. Because the male fasts, his shivering capacity erodes with muscle protein loss; in phase III shivering is functionally impaired, forcing abandonment or death.

Simulation 1: 120-Day Fast ODE with Temperature Coupling

Couple fat, protein, and glycogen compartments via Groscolas three-phase fuel switching, with thermogenic heat demand set by the seasonal ambient-temperature profile at Pointe Géologie. Track body mass, respiratory quotient, and the abandonment criterion (\(m<23\) kg in phase III).

Python

script.py175 lines

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

# Fat-protein depletion ODE for a fasting male emperor penguin
# Coupled to thermogenic demand over 120 days of courtship + incubation + early chick rearing
#
# Compartments:
#   F(t) [kg]   = lipid stores (triacylglycerol)
#   P(t) [kg]   = protein stores (skeletal muscle, viscera dry)
#   G(t) [kg]   = glycogen (small, transient)
#   T_b(t) [C]  = deep-body temperature (defended at 37 C by thermogenesis)
#
# Metabolic references:
#   Groscolas (1986, 1990) J. Comp. Physiol. B: plasma metabolite phases
#   Cherel et al. (1988)   Am. J. Physiol.   : phase I/II/III fuel switching
#   Le Maho et al. (1993)  American Zoologist : 120-d fast bioenergetics
#   Robin et al. (1998)    Am. J. Physiol.   : corticosterone / glucagon surge
#
# Phase I  : 0-70 d   lipid dominant, 87% fat oxidation, protein spared
# Phase II : 70-105 d protein catabolism accelerates, RQ rises 0.72->0.77
# Phase III: 105 d+   critical lean mass, corticosterone surge, abandon nest

# Energy densities (MJ per kg substrate)
e_F = 39.3      # fat  (9.4 kcal/g)
e_P = 17.8      # protein (4.3 kcal/g, accounting for water-bound loss)
e_G = 16.7      # glycogen

# Thermogenic demand (W) at ambient T_a with plumage R = 0.18 m^2 K / W
def thermogenic_W(Tb, Ta, m):
    A = 0.1 * m ** 0.67
    R = 0.18
    Q = A * (Tb - Ta) / R
    return max(5.0, Q)

# Parameters
t_max = 120.0
dt = 0.05
N = int(t_max / dt)
t = np.linspace(0.0, t_max, N + 1)

# Initial state (robust 38-kg courtship male)
F0 = 14.0
P0 = 7.0
G0 = 0.08
m0 = 38.0
Tb0 = 37.0

def T_ambient(tt):
    return -20.0 - 30.0 * np.exp(-((tt - 65.0) / 25.0) ** 2)

def phi_protein(F_frac):
    return 1.0 / (1.0 + np.exp(12.0 * (F_frac - 0.13)))

F = np.zeros_like(t); F[0] = F0
P = np.zeros_like(t); P[0] = P0
G = np.zeros_like(t); G[0] = G0
Tb = np.zeros_like(t); Tb[0] = Tb0
m = np.zeros_like(t); m[0] = m0
Ta_trace = np.zeros_like(t)
Q_trace = np.zeros_like(t)
phase = np.zeros_like(t, dtype=int)
RQ = np.zeros_like(t)
abandon_day = None

for i in range(N):
    Ta = T_ambient(t[i])
    Ta_trace[i] = Ta
    Q = thermogenic_W(Tb[i], Ta, m[i])
    Q_trace[i] = Q
    E_day = Q * 86400.0 * 1e-6
    f_frac = F[i] / F0
    p_share = phi_protein(f_frac)
    dG_day = -0.004 * (1.0 - p_share) - 0.001 * p_share
    if G[i] <= 0:
        dG_day = 0.0
    E_from_G = -dG_day * e_G
    E_remainder = max(0.0, E_day - E_from_G)
    E_from_F = (1.0 - p_share) * E_remainder
    E_from_P = p_share * E_remainder
    dF_day = -E_from_F / e_F
    dP_day = -E_from_P / e_P
    if p_share < 0.1:
        phase[i] = 1
    elif p_share < 0.6:
        phase[i] = 2
    else:
        phase[i] = 3
    RQ[i] = 0.71 * (1.0 - p_share) + 0.80 * p_share
    F[i + 1] = max(0.0, F[i] + dF_day * dt)
    P[i + 1] = max(0.0, P[i] + dP_day * dt)
    G[i + 1] = max(0.0, G[i] + dG_day * dt)
    dTb_day = -0.004 * p_share
    Tb[i + 1] = max(35.5, Tb[i] + dTb_day * dt)
    m[i + 1] = m[i] + dt * (dF_day + dP_day + dG_day)
    if abandon_day is None and m[i + 1] < 23.0 and p_share > 0.6:
        abandon_day = t[i + 1]

Ta_trace[-1] = T_ambient(t[-1])
Q_trace[-1] = thermogenic_W(Tb[-1], Ta_trace[-1], m[-1])
RQ[-1] = RQ[-2]
phase[-1] = phase[-2]

fig, axes = plt.subplots(2, 2, figsize=(15, 10))
fig.patch.set_facecolor('#0a0a1a')
for ax in axes.flat:
    ax.set_facecolor('#0a0a1a')
    ax.tick_params(colors='#e2e8f0')
    for spine in ax.spines.values():
        spine.set_color('#475569')
    ax.grid(True, color='#334155', alpha=0.35)

ax = axes[0, 0]
ax.plot(t, F, color='#fbbf24', lw=2.4, label='Fat F(t)')
ax.plot(t, P, color='#f43f5e', lw=2.4, label='Protein P(t)')
ax.plot(t, G * 50, color='#60a5fa', lw=1.6, ls='--', label='Glycogen x50')
ax.axvline(70, color='#94a3b8', ls=':', alpha=0.7)
ax.axvline(105, color='#94a3b8', ls=':', alpha=0.7)
ax.text(35, 13.5, 'Phase I', color='#fde68a', fontsize=11, ha='center')
ax.text(87, 13.5, 'Phase II', color='#fde68a', fontsize=11, ha='center')
ax.text(112, 13.5, 'III', color='#fecaca', fontsize=11, ha='center')
ax.set_xlabel('Day of fast', color='#e2e8f0')
ax.set_ylabel('Store (kg)', color='#e2e8f0')
ax.set_title('Groscolas 3-phase fuel depletion',
             color='#fde68a', fontsize=13, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[0, 1]
ax.plot(t, m, color='#fcd34d', lw=2.6)
ax.axhline(23.0, color='#f43f5e', ls='--', alpha=0.6, label='Abandonment threshold 23 kg')
ax.fill_between(t, m, m0, color='#f59e0b', alpha=0.08)
ax.set_xlabel('Day of fast', color='#e2e8f0')
ax.set_ylabel('Body mass (kg)', color='#e2e8f0')
ax.set_title('Total body mass: 38 -> 24 kg typical 120-day trajectory',
             color='#fde68a', fontsize=13, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[1, 0]
ax.plot(t, Ta_trace, color='#60a5fa', lw=2.0, label='Ambient T_a')
ax2 = ax.twinx()
ax2.plot(t, Q_trace, color='#fbbf24', lw=2.0, label='Thermogenic load Q (W)')
ax2.tick_params(colors='#fbbf24')
ax.set_xlabel('Day of fast', color='#e2e8f0')
ax.set_ylabel('T_a (C)', color='#60a5fa')
ax2.set_ylabel('Q (W)', color='#fbbf24')
ax.set_title('Ambient temperature and metabolic heat demand',
             color='#fde68a', fontsize=13, fontweight='bold')

ax = axes[1, 1]
ax.plot(t, RQ, color='#a3e635', lw=2.2)
ax.set_ylim(0.68, 0.82)
ax.axhline(0.71, color='#94a3b8', ls=':', alpha=0.6)
ax.axhline(0.80, color='#94a3b8', ls=':', alpha=0.6)
ax.set_xlabel('Day of fast', color='#e2e8f0')
ax.set_ylabel('Respiratory quotient RQ', color='#e2e8f0')
ax.set_title('RQ: 0.71 (fat) -> 0.80 (mixed) signals phase transition',
             color='#fde68a', fontsize=13, fontweight='bold')

plt.tight_layout()
plt.savefig('output.png', dpi=120, bbox_inches='tight', facecolor='#0a0a1a')

print('Final body mass at day 120 (kg):', round(float(m[-1]), 2))
print('Mass loss fraction:', round(float(1 - m[-1] / m0), 3))
print('Fat depletion (kg):', round(float(F0 - F[-1]), 2))
print('Protein catabolised (kg):', round(float(P0 - P[-1]), 2))
print('Phase II onset near day:', int(t[np.argmax(phase >= 2)]))
print('Phase III onset near day:', int(t[np.argmax(phase >= 3)]) if np.any(phase >= 3) else 'not reached')
if abandon_day is not None:
    print('Nest abandonment at day ~', round(float(abandon_day), 1))
else:
    print('Fast completed without abandonment (typical successful male)')
print('Total thermogenic energy expended (MJ):',
      round(float(np.trapezoid(Q_trace, t) * 86400.0 * 1e-6), 1))

Click Run to execute the Python code

Code will be executed with Python 3 on the server

9. Hatch-Timing Optimisation

The hatch date is a control variable with a quadratic cost function. Too early and chicks experience lethal cold; too late and they miss the summer krill peak. Define a cost

\[ C(t_h) = \alpha_1 (t_h - t_{T\min})^2 + \alpha_2 (t_h - t_{\text{food}})^2 \]

Sum of thermal and nutritional mistiming costs; optimum balances\(t^{*}_h = \tfrac{\alpha_1 t_{T\min} + \alpha_2 t_{\text{food}}}{\alpha_1+\alpha_2}\).

Because thermal cost \(\alpha_1\) exceeds nutritional cost \(\alpha_2\)at the Pointe Géologie latitude, optimum hatch is mid-July—~6 weeks before the austral-summer food peak. The chick then benefits from parental provisioning across a 6-week thermoneutrality buffer. In warming climates \(\alpha_1\) decreases; the optimum shifts later. Le Bohec et al. (2008) reported that A. patagonicus has shifted hatch dates by ~14 days later per decade—an early demonstration of this mechanism in a penguin.

Why only a single egg?

Allometric consideration: a two-egg clutch would require the male to hold two eggs on his feet—geometrically infeasible without dropping one. The single-egg strategy is locked in by the foot-brooding morphology. Evolutionarily, the lineage trades clutch size for per-egg survival: a 460-g emperor egg represents 1% of female body mass, vs. 3–5% for similar-sized seabirds producing two-egg clutches.

10. Comparative Context: Emperor vs. Other Penguins

Species	Clutch	Incubating parent	Incubation days	Max male fast (d)
A. forsteri (emperor)	1	Male only	64–67	115–120
A. patagonicus (king)	1	Both, alternating	53–55	~30
Pygoscelis adeliae (Adélie)	2	Both, alternating	35	~10
Spheniscus demersus (African)	2	Both, alternating	38–41	~3
Eudyptula minor (little)	2	Both, alternating	33–38	<1

The emperor is the unique case of uni-parental male incubation with an associated 100+ day fast. The pattern is feasible only because of (i) the huge lipid reserve built during the pre-breeding foraging bout (Nov–Feb), (ii) huddle sharing of thermal cost, (iii) insulation of the feather layer with vapour-trap value rivalling any mammal pelt, and (iv) the ability to mobilise crop milk if the female is delayed.

Simulation 2: Stochastic Huddle Pair-Switching & Fairness

Simulate a 200-bird hex-lattice huddle under windward cooling. Periodic rotation events move birds with highest cumulative cold into the core; a tunable defection rate lets selfish birds refuse rotation. Track the Gini coefficient of cold exposure and show that fairness collapses as defection grows.

Python

script.py174 lines

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

# Stochastic pair-switching huddle simulation
# Based on Gilbert et al. (2006, 2008) Biol. Lett. 'Fault-lines in huddles'
# and Zitterbart et al. (2011) PLOS ONE 'Coordinated movements'

np.random.seed(11)

N_target = 200
T_env = -45.0
T_core_target = 36.0
T_skin_huddle = 22.0
dt = 0.1
T_sim = 48.0
steps = int(T_sim / dt)

R = 4.5
positions = []
for i in range(-8, 9):
    for j in range(-8, 9):
        x = i + 0.5 * j
        y = j * (np.sqrt(3) / 2.0)
        if x * x + y * y <= R * R:
            positions.append([x, y])
positions = np.array(positions)
if len(positions) >= N_target:
    idx = np.random.choice(len(positions), N_target, replace=False)
    positions = positions[idx]
N = len(positions)
print('Birds simulated:', N)

defection_rate = 0.12
cumulative_cold = np.zeros(N)
wind = np.array([-1.0, 0.0])

def local_temperature(positions, i, T_env):
    r = np.linalg.norm(positions[i])
    neighbors = 0
    for j in range(len(positions)):
        if i == j:
            continue
        d = np.linalg.norm(positions[i] - positions[j])
        if d < 0.75:
            neighbors += 1
    frac_insulation = min(1.0, neighbors / 6.0)
    T_local = T_env + (T_skin_huddle - T_env) * frac_insulation
    dot = np.dot(positions[i] / (r + 1e-6), wind)
    if dot > 0.3 and r > 2.0:
        T_local -= 6.0
    return T_local

T_hist = []
positions_hist = [positions.copy()]
gini_hist = []

for step in range(steps):
    T_local = np.array([local_temperature(positions, i, T_env) for i in range(N)])
    cumulative_cold += np.maximum(0.0, T_core_target - T_local) * dt

if step > 0 and step % int(2.0 / dt) == 0:
        ranks = np.argsort(-cumulative_cold)
        n_rotate = int(0.10 * N)
        for k in range(n_rotate):
            bird = ranks[k]
            warm_bird = ranks[-1 - (k % 10)]
            if np.random.rand() < defection_rate:
                continue
            tmp = positions[bird].copy()
            positions[bird] = positions[warm_bird]
            positions[warm_bird] = tmp
        positions += 0.03 * np.random.randn(*positions.shape)

if step % 20 == 0:
        positions_hist.append(positions.copy())
        T_hist.append(T_local.copy())
        sorted_c = np.sort(cumulative_cold)
        cum = np.cumsum(sorted_c)
        if cum[-1] > 0:
            gini = (2.0 * np.sum((np.arange(1, N + 1)) * sorted_c) / (N * cum[-1])) - (N + 1) / N
        else:
            gini = 0.0
        gini_hist.append(gini)

fig = plt.figure(figsize=(16, 10))
fig.patch.set_facecolor('#0a0a1a')
gs = fig.add_gridspec(2, 3, hspace=0.35, wspace=0.35)
ax_a = fig.add_subplot(gs[0, 0])
ax_b = fig.add_subplot(gs[0, 1])
ax_c = fig.add_subplot(gs[0, 2])
ax_d = fig.add_subplot(gs[1, 0])
ax_e = fig.add_subplot(gs[1, 1])
ax_f = fig.add_subplot(gs[1, 2])

for ax in [ax_a, ax_b, ax_c, ax_d, ax_e, ax_f]:
    ax.set_facecolor('#0a0a1a')
    ax.tick_params(colors='#e2e8f0')
    for spine in ax.spines.values():
        spine.set_color('#475569')

for axs, snap_idx, title in [
        (ax_a, 0, 't = 0 h'),
        (ax_b, len(positions_hist) // 2, 't = ~24 h'),
        (ax_c, len(positions_hist) - 1, 't = 48 h'),
    ]:
    p = positions_hist[snap_idx]
    if snap_idx == len(positions_hist) - 1:
        color_vals = cumulative_cold
    else:
        color_vals = np.linalg.norm(p, axis=1)
    axs.scatter(p[:, 0], p[:, 1], c=color_vals, cmap='inferno', s=45, edgecolor='white', lw=0.3)
    axs.arrow(-5.5, 4.2, 1.3, 0.0, head_width=0.25, color='#60a5fa')
    axs.text(-5.5, 4.7, 'wind', color='#bfdbfe', fontsize=9)
    axs.set_xlim(-6, 6); axs.set_ylim(-6, 6)
    axs.set_aspect('equal')
    axs.set_title(title, color='#fde68a', fontsize=12, fontweight='bold')

ax_d.plot(np.linspace(0, T_sim, len(gini_hist)), gini_hist, color='#fbbf24', lw=2.2)
ax_d.set_xlabel('Time (hr)', color='#e2e8f0')
ax_d.set_ylabel('Gini coefficient of cold burden', color='#e2e8f0')
ax_d.set_title('Cold-burden inequality (0 = perfect fairness)',
               color='#fde68a', fontsize=12, fontweight='bold')
ax_d.grid(True, color='#334155', alpha=0.35)

ax_e.hist(cumulative_cold, bins=24, color='#f59e0b', edgecolor='#fde68a')
ax_e.set_xlabel('Cumulative cold burden (C-hr)', color='#e2e8f0')
ax_e.set_ylabel('Number of birds', color='#e2e8f0')
ax_e.set_title('Distribution of cold exposure at 48 h',
               color='#fde68a', fontsize=12, fontweight='bold')
ax_e.grid(True, color='#334155', alpha=0.35)

def_rates = np.linspace(0.0, 0.5, 8)
ginis_scan = []
for dr in def_rates:
    positions2 = positions_hist[0].copy()
    cc = np.zeros(N)
    for s in range(int(T_sim / 0.2)):
        T_local = np.array([local_temperature(positions2, i, T_env) for i in range(N)])
        cc += np.maximum(0.0, T_core_target - T_local) * 0.2
        if s % 10 == 0 and s > 0:
            ranks = np.argsort(-cc)
            for k in range(int(0.1 * N)):
                if np.random.rand() < dr:
                    continue
                warm_b = ranks[-1 - (k % 10)]
                tmp = positions2[ranks[k]].copy()
                positions2[ranks[k]] = positions2[warm_b]
                positions2[warm_b] = tmp
            positions2 += 0.03 * np.random.randn(*positions2.shape)
    sorted_c = np.sort(cc)
    cumv = np.cumsum(sorted_c)
    if cumv[-1] > 0:
        g = (2.0 * np.sum((np.arange(1, N + 1)) * sorted_c) / (N * cumv[-1])) - (N + 1) / N
    else:
        g = 0.0
    ginis_scan.append(g)

ax_f.plot(def_rates, ginis_scan, color='#f43f5e', lw=2.4, marker='o')
ax_f.set_xlabel('Defection rate', color='#e2e8f0')
ax_f.set_ylabel('Final Gini', color='#e2e8f0')
ax_f.set_title('Inequality grows with defection',
               color='#fde68a', fontsize=12, fontweight='bold')
ax_f.grid(True, color='#334155', alpha=0.35)

plt.savefig('output.png', dpi=120, bbox_inches='tight', facecolor='#0a0a1a')

print('Mean cumulative cold burden (C-hr):', round(float(cumulative_cold.mean()), 2))
print('Max cumulative cold burden (C-hr):', round(float(cumulative_cold.max()), 2))
print('Min cumulative cold burden (C-hr):', round(float(cumulative_cold.min()), 2))
print('Final Gini coefficient:', round(float(gini_hist[-1]), 3))
print('Gini at defection 0.0 vs 0.5:', round(ginis_scan[0], 3), 'vs', round(ginis_scan[-1], 3))

Click Run to execute the Python code

Code will be executed with Python 3 on the server

11. The Hormonal Architecture of a Long Fast

Groscolas, Schreiber & Morin (1986) and later Robin, Cherel, Girard & Le Maho (1998) established the hormone time-course that orchestrates the three phases. Key players:

Insulin: falls from 20 μU/mL arrival to <3 μU/mL by day 10, permitting unbridled lipolysis.
Glucagon: rises gradually from 40 to 120 pg/mL across phase II, spikes to 250 pg/mL at phase III onset.
Corticosterone: baseline 20 ng/mL; elevated 3× at phase III onset; activates ubiquitin-proteasome muscle proteolysis.
T3/T4 (thyroid): fall by 40% in phase II, remarkable for a cold-exposed endotherm—a “euthyroid sick”-like adaptive down-regulation of non-essential metabolism (Cherel & Groscolas 1999).
Leptin: unusually absent in birds; the equivalent satiety signal is likely ghrelin-driven.

Phase III is metabolically distinct, not an extension of II

The transition is signalled molecularly by a ~10× increase in hepatic PEPCK expression, a 4× rise in plasma urea, and a shift in RQ from 0.72 (pure fat) to 0.77 (mixed fuel). Neural control involves an AgRP-like hypothalamic circuit that senses FFA and β-OHB; when ketone production falls, the circuit triggers the corticosterone cascade and behavioural drive to break the fast. In captive birds the drive is physiologically irrepressible: even if food is denied, corticosterone continues to rise and wing-flapping escape behaviour intensifies.

12. Mortality, Failed Clutches, and Colony-Level Consequences

In normal years at Pointe Géologie, ~5% of males die during the fast, mostly in phase III. Lethal events cluster during episodes of prolonged severe cold (<\(-55\)°C) coinciding with high wind speeds that overwhelm huddle insulation. An additional ~20% of clutches fail due to egg loss during the hand-off, embryonic death, or post-hatch predation/exposure—making single-season breeding success on average 60–70%.

Long-term breeding success (probability of a chick fledging given that a pair formed) is 40–50% under stable sea-ice conditions, dropping to near zero at sites that lose fast ice before November. Halley Bay, the world’s former second-largest colony, suffered near-total breeding failure from 2016–2018 due to loss of fast ice during the chick-rearing phase (Fretwell & Trathan, 2019).

\[ \text{Chick survival} = \mathrm{Pr}(\text{fast ice until day 260}) \times s_{\text{abiotic}} \times s_{\text{provisioning}} \]

Each factor drops with warming: Jenouvrier et al. (2021) estimate 0.42 current → 0.08 by 2100 under SSP5-8.5.

Discussion & Graduate Exercises

Derive analytically the time to phase III for a male of initial fat mass\(F_0\) maintaining metabolic rate \(\dot E_0\) at thermoneutrality. Show that \(t_{II\to III} \approx F_0\,e_F / \dot E_0\)under the pure-lipolysis approximation, and compute its value for\(F_0=14\) kg, \(\dot E_0=50\) W.
From the bioheat equation, compute the brood-patch heat flux \(q_e\) if the dermal perfusion rate drops by 50% (as may occur in phase III). Show that the egg temperature can no longer be maintained at 35°C when ambient is below \(-30\)°C.
Show that the critical b/c ratio for stable cooperation in a huddle of\(N=10^3\) birds with \(\hat p=0.95\) is\(b/c < 1.053\). Discuss why emperor huddles persist despite this fragile margin (hint: iterated game + kin selection).
Using the RQ output of Simulation 1, identify the day at which RQ crosses 0.75. Compare to Groscolas’ empirical observation of day ~70. What adjustment to the glucagon response parameter \(\phi\) would bring the simulated day into better agreement?
Modify Simulation 2 to include a spatially graded wind field (linear in y). Does the resulting fault-line angle distribution still peak at 30°? What happens to the Gini coefficient if a small fraction of birds (<5%) are chronic defectors?
Compute the metabolic water produced from 12 kg of fat using the stoichiometry of the combustion of tripalmitin. Confirm ~12.8 L. If the male loses an additional 2.8 L to respiratory vapour, what minimum ambient humidity is required to avoid dehydration?

Key References

• Groscolas, R. (1986). “Changes in body mass, body temperature and plasma fuel levels during the natural breeding fast in male and female emperor penguins.” J. Comp. Physiol. B 156, 521–527.

• Groscolas, R. (1990). “Metabolic adaptations to fasting in emperor and king penguins.” Penguin Biology, 269–296. Academic Press.

• Groscolas, R., Leloup, J. (1986). “The endocrine control of reproduction and moult in male and female emperor penguins.” Gen. Comp. Endocrinol. 62, 43–53.

• Robin, J.P., Cherel, Y., Girard, H., et al. (1998). “Plasma metabolites, glucagon, and insulin levels during the fast of emperor penguins.” Am. J. Physiol. 274, R1101–R1108.

• Le Maho, Y., Delclitte, P., Chatonnet, J. (1976). “Thermoregulation in fasting emperor penguins under natural conditions.” Am. J. Physiol. 231, 913–922.

• Le Maho, Y., Robin, J.P., Cherel, Y. (1993). “Body fuel metabolism during long-term fasting in birds.” American Zoologist 33, 128–139.

• Cherel, Y., Groscolas, R. (1999). “Relationships between nutrient storage and nutrient utilisation in long-term fasting birds and mammals.” Proc. 22nd Int. Ornithol. Congress 17–34.

• Cherel, Y., Robin, J.P., Le Maho, Y. (1988). “Physiology and biochemistry of long-term fasting in birds.” Can. J. Zool. 66, 159–166.

• Gilbert, C., Robertson, G., Le Maho, Y., Naito, Y., Ancel, A. (2006). “Huddling behavior in emperor penguins: dynamics of huddling.” Physiology & Behavior 88, 479–488.

• Gilbert, C., Le Maho, Y., Perret, M., Ancel, A. (2007). “Body temperature changes induced by huddling in breeding male emperor penguins.” Am. J. Physiol. 292, R176–R185.

• Zitterbart, D.P., Wienecke, B., Butler, J.P., Fabry, B. (2011). “Coordinated movements prevent jamming in an emperor penguin huddle.” PLOS ONE 6, e20260.

• Waters, A., Blanchette, F., Kim, A.D. (2012). “Modeling huddling penguins.” PLOS ONE 7, e50277.

• Ancel, A., Visser, H., Handrich, Y., Masman, D., Le Maho, Y. (1997). “Energy saving in huddling penguins.” Nature 385, 304–305.

• Pinshow, B., Fedak, M.A., Battles, D.R., Schmidt-Nielsen, K. (1976). “Energy expenditure for thermoregulation and locomotion in emperor penguins.” Am. J. Physiol. 231, 903–912.

• Prevost, J. (1961). Ecologie du Manchot Empereur. Paris: Hermann.

• Prevost, J., Vilter, V. (1963). “Histologie de la sécrétion oesophagienne du manchot empereur.” Proc. XIII Int. Ornith. Congr. 2, 1085–1094.

• Handrich, Y. (1989). “Incubation water loss in king penguin egg: causes and consequences.” J. Comp. Physiol. B 159, 315–321.

• Stonehouse, B. (1953). “The emperor penguin Aptenodytes forsteri Gray. I. Breeding behaviour and development.” FIDS Sci. Rep. 6, 1–33.

• Aubin, T., Jouventin, P., Hildebrand, C. (2000). “Penguins use the two-voice system to recognize each other.” Proc. R. Soc. B 267, 1081–1087.

• Isaksson, N., Evans, T.J., Shamoun-Baranes, J., Åkesson, S. (2016). “Land or sea? Foraging area choice during breeding by an omnivorous gull.” Mov. Ecol. 4, 11.

• Fretwell, P.T., Trathan, P.N. (2019). “Emperors on thin ice: three years of breeding failure at Halley Bay.” Antarctic Science 31, 133–138.

• Jenouvrier, S., Caswell, H., Barbraud, C., et al. (2010). “Demographic models and IPCC climate projections predict the decline of an emperor penguin population.” PNAS 106, 1844–1847.

• Murrish, D.E. (1973). “Respiratory heat and water exchange in penguins.” Respir. Physiol. 19, 262–270.

• Le Bohec, C., Durant, J.M., Gauthier-Clerc, M., et al. (2008). “King penguin population threatened by Southern Ocean warming.” PNAS 105, 2493–2497.

Synthesis & Bridge to Module 4

The breeding cycle weaves together every physical and biological scale we have touched. Feather insulation (M1), huddle thermodynamics (M2), and the 120-day fast together lower the male’s metabolic rate to ~1.2× basal—enabling the last of a vertebrate line to breed through polar winter. The three-phase biochemistry couples plasma hormones to behaviour: once phase III fires, abandonment follows within days unless the female returns.

Module 4 turns from the land cycle to the ocean half of the life. After his fast, the recovering male walks back to the sea edge and begins the diving bouts that will refill his lipid stores for next year’s breeding. The physiology of emperor diving is as extreme as his fasting: 565 m depth, 22 minute breath-holds, bradycardia to 6 bpm, and blood-oxygen carrying capacity that exceeds any other bird. We derive the aerobic dive limit, model compartmental oxygen depletion during a 500 m dive, and characterise the neural reflex that triggers the bradycardic response.

← Module 2: Huddle Dynamics Module 4: Diving Biophysics (565 m) →