Module 7: Fasting Biochemistry — 120 Days Without Food

The male emperor penguin fasts for up to 120 days during courtship, incubation, and early chick-rearing — the longest voluntary fast of any vertebrate. He arrives at the colony in March weighing 35–40 kg, transfers the egg to his feet in May, and will not eat again until the female returns around mid-July. René Groscolas and Yvon Le Maho, working at Pointe Géologie since the 1970s, decomposed this fast into three distinct metabolic phases (Groscolas 1986, 1990; Le Maho 1977). Phase III, the protein-catabolism crisis that triggers colony desertion, is arguably the most closely studied starvation syndrome in any vertebrate. This module derives the three-phase fuel-mixing framework, the Spee corticosterone bimodality, and the 22%-body-mass Groscolas abandonment threshold.

1. The Fast in Context

No other vertebrate endures a comparable ordeal. A brown bear hibernates 5 months but conserves water and wakes occasionally; a hummingbird drops into nocturnal torpor but refuels each morning. The male emperor penguin fasts while actively incubating at ambient temperatures of \(-20\) to \(-40\)°C, losing 40% of body mass and producing ~3 kJ/g of metabolic heat over the entire fast. The female sealift-feeds in the open ocean during this time; the male cannot leave the colony without abandoning the egg.

\[ \text{Total fasting days: } \sim 110\;-\;130,\quad \text{mass loss: } 35\%\;-\;45\%,\quad \text{total energy: } \sim 250\;-\;350\;\text{MJ} \]

Reported ranges span years, individual variability, and colony latitude (Groscolas & Robin 2001).

The ability to complete this fast reflects a sophisticated metabolic-endocrine control system: lipid reserves must be mobilised at a controlled rate, glucose supply to the central nervous system must be preserved, and essential protein (heart, brain, diaphragm) must be spared. When these constraints fail, the male abandons the egg; the timing of that decision is the principal focus of the last four decades of physiology research on this species.

2. The Groscolas Three-Phase Model (1986)

Groscolas (1986, Physiol. Zool.) partitioned the emperor male fast into three phases on the basis of plasma metabolites, respiratory quotient (RQ), and body-mass trajectory:

Phase I: Adaptation & lipolysis (days 0–70)

Body mass falls at ~180 g/day (0.5%/day).
Fuel mix: \(\sim 87\%\) of energy from lipid, \(\sim 13\%\) from protein.
RQ plateaus at \(\sim 0.72\) — pure fat oxidation.
Free fatty acids (FFA) elevated 3–5\(\times\) preprandial.
Plasma glucose maintained via glycerol \(\rightarrow\) gluconeogenesis.
Plasma uric acid (index of protein catabolism) is low.

Phase II: Transition (days 70–100)

Lipid reserves approach \(\sim 30\%\) of initial; rate of lipid oxidation declines.
Nonessential protein pool mobilised; ventral abdominal muscles first (Cherel 1988).
RQ drifts up to \(\sim 0.76\) as protein contributes more heavily.
Urea excretion rises; uric acid rises 4–5\(\times\).
Body-mass loss rate steady at \(\sim 170\) g/day.
Behavioural phenotype unchanged: bird continues to incubate, vocalise, thermoregulate.

Phase III: Desperate catabolism (days > 100)

Lipid reserves depleted to < 10% of initial; remaining adipocytes non-accessible or critical-structure.
Essential protein catabolism begins: cardiac, diaphragmatic, and brain proteins at risk.
RQ rises above 0.80.
Uric acid and urea rise sharply; rapid rise in plasma total amino acids.
Corticosterone spike (Spee 2010) from ~15 ng/mL baseline to 50–80 ng/mL.
Body-mass loss rate accelerates to \(\sim 250\) g/day.
Behaviourally: reduced incubation bouts, increased movement, eventual egg abandonment.

Groscolas (2008) formalised the 22% body-mass threshold: when cumulative mass loss reaches 22% of initial mass, the male will abandon the egg even if the female has not returned. This threshold is the evolutionary compromise between parental investment and the male’s own survival.

\[ W(t_{\text{abandon}}) \;\le\; 0.78\,W_0 \]

Groscolas 2008 threshold: at 22% cumulative body-mass loss the male ceases incubation.

3. Schematic: Phase Progression

Simulation 1: Three-Phase Fuel-Mixing ODE with Thermogenic Compensation

Integrate the coupled ODEs for lipid and protein mass over a 130-day fast with a phase-dependent fuel fraction \(f_{\text{lip}}(t, W_{\text{lip}})\). Include a mild thermogenic compensation (5–8% metabolic up-scale) as lipid stores fall. Track body-mass trajectory, RQ, and the 22% abandonment threshold.

Python

script.py209 lines

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

# Three-phase fuel-mixing and body-mass depletion model for the incubating
# male emperor penguin (Groscolas 1986, 1990, 2008).
#
# Notation follows Groscolas & Robin (2001):
#   W(t)    body mass (kg)
#   W_lip   lipid mass (kg)
#   W_prot  protein mass (kg)
#   RQ      respiratory quotient = VCO2 / VO2
#   e_f     specific energy of fat (39.3 kJ/g)
#   e_p     specific energy of protein (17.8 kJ/g)
#
# Phase I (0-70 d):   lipolysis dominant, 87% energy from fat, RQ ~ 0.72
# Phase II (70-100 d): mixed fuels, nonessential protein mobilisation,
#                     RQ drifts toward 0.76
# Phase III (>100 d): desperate protein catabolism, RQ > 0.80, cortisol spike,
#                     behavioural abandonment once W drops to ~0.78 * W0
#
# We integrate a system of ODEs for W_lip(t) and W_prot(t) driven by a
# thermogenic-compensated metabolic rate P_met(t) that rises modestly when
# lipid reserves fall.

# Parameters
W0 = 37.0                 # initial male mass (kg) before fast (pre-incubation)
W_lip0 = 0.35 * W0        # 35% of body mass is lipid (Groscolas 1990)
W_prot0 = 0.17 * W0       # 17% of body mass is protein
W_essential = 0.60 * W0   # structural + organ mass that cannot be catabolised

# Energy densities (kJ/g)
e_f = 39.3
e_p = 17.8

# Basal metabolic rate scaling (mass-specific): 3.6 W/kg
def bmr(W):
    return 3.6 * W  # W (watts)

# Metabolic cost of incubation, including mild cold stress under the huddle
# margin. Thermogenic compensation: as fat stores drop, the bird mobilises
# slight extra heat production (5-8% scale up).
def incub_cost(W, W_lip):
    base = bmr(W)
    frac_lip = W_lip / (W_lip0 + 1e-9)
    thermo_up = 1.0 + 0.08 * np.clip(1.0 - frac_lip, 0.0, 1.0)
    return base * 1.25 * thermo_up

# Fuel mix: phase-dependent fraction of energy from lipid vs protein
def fuel_fraction(t_days, W, W_lip):
    lip_frac_base = W_lip / (W_lip0 + 1e-9)
    if t_days < 70.0 and lip_frac_base > 0.35:
        f_lip = 0.87
    elif t_days < 100.0:
        # Smooth drift as lipid drains (Phase II)
        f_lip = 0.87 - 0.45 * np.clip((t_days - 70.0) / 30.0, 0.0, 1.0)
    else:
        # Phase III: lipid nearly exhausted, protein dominates
        f_lip = max(0.10, lip_frac_base * 0.25)
    return f_lip

# ODE right-hand side
def dYdt(t_days, Y):
    W_lip, W_prot = Y
    W = W_essential + W_lip + W_prot
    P = incub_cost(W, W_lip)  # watts
    # Energy flux in kJ/day
    E_day = P * 86.4 / 1000.0  # W * s/day / 1000 -> kJ/day
    f_lip = fuel_fraction(t_days, W, W_lip)
    # Mass loss rates (g/day)
    dm_lip = - (f_lip * E_day) / e_f
    dm_prot = - ((1.0 - f_lip) * E_day) / e_p
    # Convert g/day -> kg/day
    return np.array([dm_lip * 1e-3, dm_prot * 1e-3])

# RK4 integrator
def rk4(f, t0, t_end, Y0, dt=0.1):
    ts = [t0]
    Ys = [np.array(Y0)]
    t = t0
    Y = np.array(Y0)
    while t < t_end:
        k1 = f(t, Y)
        k2 = f(t + dt / 2, Y + dt / 2 * k1)
        k3 = f(t + dt / 2, Y + dt / 2 * k2)
        k4 = f(t + dt, Y + dt * k3)
        Y = Y + dt / 6 * (k1 + 2 * k2 + 2 * k3 + k4)
        t = t + dt
        ts.append(t)
        Ys.append(Y.copy())
    return np.array(ts), np.array(Ys)

ts, Ys = rk4(dYdt, 0.0, 130.0, [W_lip0, W_prot0], dt=0.1)
Wlip = Ys[:, 0]
Wprot = Ys[:, 1]
W = W_essential + Wlip + Wprot

# Respiratory quotient from instantaneous fuel fraction
f_lips = np.array([fuel_fraction(t, w, wl) for t, w, wl in zip(ts, W, Wlip)])
RQ = f_lips * 0.71 + (1.0 - f_lips) * 0.80
# (fat RQ ~ 0.71; protein RQ ~ 0.80)

# Mass-loss fraction relative to initial mass
loss_frac = (W0 - W) / W0

# Abandonment threshold: Groscolas 2008 -> 22% body-mass loss
abandon_idx = int(np.argmax(loss_frac >= 0.22)) if np.any(loss_frac >= 0.22) else -1
t_abandon = ts[abandon_idx] if abandon_idx >= 0 else ts[-1]

# Thermogenic compensation trajectory
P_metab = np.array([incub_cost(w, wl) for w, wl in zip(W, Wlip)])

# Figure
fig, axes = plt.subplots(2, 3, figsize=(17, 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)

phase_bands = [(0, 70, '#a3e635', 'Phase I'), (70, 100, '#fbbf24', 'Phase II'),
               (100, 130, '#f43f5e', 'Phase III')]

ax = axes[0, 0]
for (ts_start, ts_end, col, lbl) in phase_bands:
    ax.axvspan(ts_start, ts_end, color=col, alpha=0.10)
ax.plot(ts, W, color='#fbbf24', lw=2.4, label='Body mass W(t)')
ax.axhline(W0 * 0.78, color='#f43f5e', ls='--', label='22% loss threshold')
ax.axvline(t_abandon, color='#a3e635', ls=':', label=f'abandon day {t_abandon:.0f}')
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Body mass W (kg)', color='#e2e8f0')
ax.set_title('Body mass trajectory during 120-day fast',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0', fontsize=9)

ax = axes[0, 1]
ax.plot(ts, Wlip, color='#f59e0b', lw=2.4, label='Lipid mass')
ax.plot(ts, Wprot, color='#60a5fa', lw=2.4, label='Protein mass')
for (ts_start, ts_end, col, lbl) in phase_bands:
    ax.axvspan(ts_start, ts_end, color=col, alpha=0.08)
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Component mass (kg)', color='#e2e8f0')
ax.set_title('Lipid and protein depletion',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[0, 2]
ax.plot(ts, f_lips, color='#a3e635', lw=2.4, label='Lipid energy fraction')
ax.plot(ts, 1.0 - f_lips, color='#fb7185', lw=2.4, label='Protein energy fraction')
for (ts_start, ts_end, col, lbl) in phase_bands:
    ax.axvspan(ts_start, ts_end, color=col, alpha=0.08)
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Fuel mix (fraction)', color='#e2e8f0')
ax.set_title('Fuel-mixing across the three phases',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[1, 0]
ax.plot(ts, RQ, color='#fbbf24', lw=2.4)
ax.axhline(0.72, color='#94a3b8', ls=':', label='pure lipid RQ = 0.72')
ax.axhline(0.80, color='#94a3b8', ls=':', label='pure protein RQ = 0.80')
for (ts_start, ts_end, col, lbl) in phase_bands:
    ax.axvspan(ts_start, ts_end, color=col, alpha=0.08)
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Respiratory quotient RQ', color='#e2e8f0')
ax.set_title('RQ trajectory (Le Maho 1977 indicator)',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[1, 1]
ax.plot(ts, P_metab, color='#f43f5e', lw=2.4, label='Incubation metabolic rate')
ax.plot(ts, bmr(W), color='#60a5fa', lw=2.0, ls='--', label='BMR only')
for (ts_start, ts_end, col, lbl) in phase_bands:
    ax.axvspan(ts_start, ts_end, color=col, alpha=0.08)
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Metabolic power (W)', color='#e2e8f0')
ax.set_title('Thermogenic compensation as fuel depletes',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[1, 2]
ax.plot(ts, loss_frac * 100.0, color='#fde68a', lw=2.4)
ax.axhline(22.0, color='#f43f5e', ls='--', label='Groscolas 22% abandon')
for (ts_start, ts_end, col, lbl) in phase_bands:
    ax.axvspan(ts_start, ts_end, color=col, alpha=0.08)
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Body-mass loss (%)', color='#e2e8f0')
ax.set_title('Cumulative mass loss',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

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

# Summary
print('Initial body mass W0 (kg): %.1f' % W0)
print('Initial lipid (kg): %.2f' % W_lip0)
print('Initial protein (kg): %.2f' % W_prot0)
print('Mass at day 70 (kg): %.2f' % W[np.argmin(np.abs(ts - 70))])
print('Mass at day 100 (kg): %.2f' % W[np.argmin(np.abs(ts - 100))])
print('Mass at day 120 (kg): %.2f' % W[np.argmin(np.abs(ts - 120))])
print('Final loss fraction: %.3f' % loss_frac[-1])
print('Day of 22%% abandonment threshold: %.1f' % t_abandon)
print('Integrated energy expended (MJ): %.1f'
      % float(np.trapezoid(P_metab, ts * 86400.0) / 1e6))

Click Run to execute the Python code

Code will be executed with Python 3 on the server

4. Biochemistry of Lipid Mobilisation

Emperor males arrive at the colony with lipid stores concentrated in subcutaneous and visceral white adipose tissue (WAT). Unlike arctic mammals, emperors possess no brown adipose tissue (BAT) (Hohtola 2002): they rely solely on shivering thermogenesis and metabolic heat from substrate oxidation. Lipid breakdown is regulated by the hormone-sensitive lipase (HSL)-perilipin axis in adipocytes, driven by glucagon and catecholamines during fasting.

\[ \text{Triglyceride} \;\xrightarrow{\text{HSL}}\; \text{diglyceride} + \text{FFA} \;\xrightarrow{\text{MGL}}\; \text{monoglyceride} + \text{FFA} \;\xrightarrow{}\; 3\;\text{FFA} + \text{glycerol} \]

Three-step hydrolysis of triacylglycerol. HSL = hormone-sensitive lipase; MGL = monoglyceride lipase. Glycerol enters gluconeogenesis in liver; FFAs enter mitochondrial beta-oxidation.

Free fatty acids (FFA) travel bound to albumin in plasma. They enter hepatocytes and myocytes via CD36 and FATP transporters, are activated to acyl-CoA by long-chain acyl-CoA synthetase, and shuttled into mitochondria by the carnitine palmitoyltransferase (CPT1/CPT2) system.

Ketogenesis

Under sustained lipolysis, hepatic acetyl-CoA accumulates and is channelled into ketogenesis. Robin, Frain, Sardet, Groscolas & Le Maho (1988) measured plasma beta-hydroxybutyrate (BHB) in fasting emperor males, finding a rise from ~0.1 mM pre-fast to 2–3 mM at mid-fast (Phase I–II transition):

\[ 2\;\text{acetyl-CoA} \;\rightarrow\; \text{acetoacetate} \;\rightarrow\; \beta\text{-hydroxybutyrate} + \text{acetone} \]

Ketogenesis in liver: HMG-CoA synthase, HMG-CoA lyase, and beta-hydroxybutyrate dehydrogenase. BHB is the dominant circulating ketone body.

Ketone bodies (primarily BHB) cross the blood-brain barrier via monocarboxylate transporters (MCT1) and serve as the principal brain fuel during prolonged fasts. This spares glucose — and therefore protein carbon for gluconeogenesis. Le Maho’s group measured that by day 30 of fasting, ketones provide \(\sim 60\%\) of brain energy.

5. Gluconeogenesis and Glucose Sparing

Plasma glucose must be maintained above \(\sim 4\) mM to prevent central nervous system hypoglycaemia. In the fasting emperor, glucose arises from three sources, in phase-dependent proportions:

Substrate	Phase I (%)	Phase II (%)	Phase III (%)
Glycerol (from lipolysis)	60	35	10
Alanine (muscle)	25	45	60
Lactate (Cori cycle)	15	20	30

In Phase I, glycerol is the dominant glucogenic substrate, because lipolysis is high and one triacylglycerol yields three FFAs and one glycerol. As lipid stores deplete, the alanine-glucose cycle (muscle-liver) takes over, releasing NH3 that is excreted as uric acid (birds’ nitrogen-waste pathway).

\[ \text{Alanine} \;\xrightarrow{\text{ALT}}\; \text{pyruvate} + \text{glutamate},\quad \text{pyruvate} \;\xrightarrow{\text{PC, PEPCK}}\; \text{glucose} \]

Alanine transamination in liver, then classical gluconeogenic pathway. ALT = alanine aminotransferase; PC = pyruvate carboxylase; PEPCK = phosphoenolpyruvate carboxykinase.

6. Protein Catabolism: Nonessential First, Essential Last

The order of protein mobilisation is not random. Cherel, Robin & Le Maho (1988) used sequential biopsies to map which tissues lose mass first under fasting:

Phase II mobilisation: ventral abdominal muscles (M. obliquus abdominis externus), then superficial pectoralis, leg muscles (gastrocnemius). These tissues support incubation posture but not flight or key organ function.
Phase III mobilisation: deep pectoralis (flight-swim muscles), cardiac muscle, liver, diaphragm. Losing these begins to impair thermogenesis (shivering depends on intact muscle) and cardiac output.
Spared until death: brain protein, cardiac interventricular septum, intestinal mucosa.

The regulatory mechanism is the ubiquitin-proteasome pathway (UPP), modulated by glucocorticoids and insulin. Fasting-induced corticosterone rise upregulates MuRF1 and MAFbx E3 ubiquitin ligases, tagging proteins for degradation. Low insulin signalling (via AKT/FoxO) releases FoxO transcription factors that drive atrogin-1 and related muscle-atrophy genes.

\[ \text{Fasting}\;\rightarrow\;\uparrow\text{CORT} \;\rightarrow\; \text{FoxO nuclear translocation} \;\rightarrow\; \uparrow\text{MuRF1, atrogin-1} \;\rightarrow\; \text{muscle proteolysis} \]

Glucocorticoid-induced muscle atrophy pathway. Bodine & Baehr (2014) review.

The “essential-vs-nonessential” ordering is thus not morphological pre-programming but a consequence of relative expression of the proteolysis machinery across tissues. Heart and brain are protected because their cells maintain high autophagy/proteasome homeostasis and low FoxO susceptibility.

7. The Hormonal Axis: T3, Glucagon, Insulin, Corticosterone

The emperor endocrine response follows the canonical fasted-vertebrate pattern but with two unusual features: a very gradual T3 decline and a remarkable bimodal corticosterone profile. Cherel, Robin, Walch, Karmann, Netchitailo & Le Maho (1988) reported longitudinal hormone measurements across the 120-day fast:

Hormone	Day 0	Day 60	Day 100	Day 120 (abandon)
T3 (ng/mL)	2.8	2.1	1.7	1.4
T4 (ng/mL)	16	14	11	9
Glucagon (pg/mL)	120	210	260	280
Insulin (muU/mL)	11	7	6	5
Corticosterone (ng/mL)	10	14	22	65

Thyroid axis suppression

T3 and T4 fall throughout the fast. The emperor’s T3 decline of ~50% is less severe than that reported in other fasting birds (penguin chicks drop ~70%). The gradual decline is adaptive: reducing basal metabolic rate conserves energy while maintaining the thermogenic capacity required for incubation in subzero temperatures.

Glucagon-insulin axis

Glucagon rises 2–3\(\times\), driving lipolysis and hepatic glucose output. Insulin drops to basal, enabling adipose-tissue lipolysis and muscle proteolysis. The ratio I/G falls from ~0.1 pre-fast to 0.02 at abandonment.

Corticosterone bimodality (Spee 2010)

Spee et al. (2010, 2011) discovered that fasting emperor CORT is bimodal: a low baseline (\(\sim 10\)–\(20\) ng/mL) throughout Phase I–II, then an acute spike to 50–80 ng/mL at Phase III onset that coincides with the behavioural abandonment decision. The spike is not a gradual rise; it occurs over 1–3 days. This provides the physiological correlate of the Groscolas 22% mass threshold.

\[ \text{CORT}(t) \;=\; C_{\text{base}}(t) \;+\; C_{\text{spike}}\,\Theta(t - t_{\text{crisis}}) \]

Bimodal model: slow baseline rise + sigmoid spike near Phase III onset. The spike upregulates gluconeogenic enzymes and muscle-atrophy genes, and also induces foraging behaviour (Kitaysky 2001).

Simulation 2: Hormonal Axis Dynamics and Abandonment Trigger

Integrate the coupled T3, glucagon, and insulin axes with a first-order tracking model; add a bimodal corticosterone trajectory with baseline drift and a sigmoidal Phase III spike. Apply a dual-criterion abandonment rule (\(\text{CORT} > 40\) ng/mL AND\(d\text{CORT}/dt > 0.5\) ng/mL/d) and report the predicted abandonment day.

Python

script.py197 lines

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

# Hormonal axis dynamics in the fasting emperor penguin.
#
# Based on Cherel, Robin, Walch, Karmann, Netchitailo & Le Maho (1988), Robin,
# Frain, Sardet, Groscolas & Le Maho (1988), Groscolas & Robin (2001), and
# the corticosterone measurements of Spee, Beaulieu, Dervaux, Chastel, Le Maho,
# Raclot (2010). Four coupled hormones drive fuel mobilisation and the
# behavioural abandonment decision:
#
#   T3 (triiodothyronine): thyroid hormone; sets resting metabolic rate;
#                          drops ~40% from Phase I to Phase III.
#   Glucagon:             pancreatic alpha-cell; promotes lipolysis + glucose
#                          production; rises 2x early, stays elevated.
#   Insulin:              pancreatic beta-cell; suppressed to basal levels.
#   Corticosterone (CORT): adrenal; stress steroid; bimodal pattern -> baseline
#                          modestly low in Phase I, then large spike in late
#                          Phase III that correlates with colony desertion.
#
# We model these with coupled ODEs:
#   T3' = -k_T3 * (T3 - T3_set(t))
#   Glucagon' = k_glu * (Glu_set(lipid, t) - Glu)
#   CORT' = baseline + spike(t) + noise
# and use the CORT time series to define an abandonment decision rule.

from scipy.integrate import odeint

t_days = np.linspace(0.0, 135.0, 27000)

# Lipid fraction trajectory (from fuel-mixing model, simplified)
def lipid_frac(t):
    # approximate monotone decay: Phase I slow, Phase II faster, Phase III near zero
    if t < 70.0:
        return 1.0 - 0.4 * (t / 70.0)
    elif t < 100.0:
        return 0.6 - 0.4 * ((t - 70.0) / 30.0)
    else:
        return max(0.02, 0.2 - 0.15 * ((t - 100.0) / 35.0))

L = np.array([lipid_frac(t) for t in t_days])

# T3 dynamics: set point drops as lipid depletes (Cherel 1988 measurements)
T3_full = 2.8      # ng/mL at start
T3_low = 1.4       # ng/mL at abandonment
def T3_setpoint(t):
    f = lipid_frac(t)
    return T3_low + (T3_full - T3_low) * f
k_T3 = 0.03  # day^-1

# Glucagon: set point rises when lipid drops
G_base = 120.0   # pg/mL baseline
G_max = 280.0    # pg/mL crisis
def Glucagon_setpoint(t):
    f = lipid_frac(t)
    return G_base + (G_max - G_base) * (1.0 - f) ** 1.6
k_G = 0.05

# Insulin: low plateau with occasional small pulses
# basal 6 muU/mL, saturating at around 10 muU/mL
def Insulin(t):
    # fasted: suppressed; no postprandial peaks
    return 6.0 + 1.5 * np.sin(2.0 * np.pi * t / 24.0) * 0.02 + 0.4 * np.random.randn()

# Corticosterone: bimodal pattern
# - Phase I: baseline 12 ng/mL (low for incubating bird)
# - Phase II: drift up to 18 ng/mL
# - Phase III: spike to 70 ng/mL coincident with behavioural abandonment
def CORT_signal(t, t_abandon=110.0):
    base = 12.0 + 6.0 * np.clip((t - 70.0) / 30.0, 0.0, 1.0)
    spike = 55.0 / (1.0 + np.exp(-(t - t_abandon) / 1.5))
    return base + spike * (t > (t_abandon - 5.0)).astype(float)

def axis_rhs(Y, t):
    T3, G = Y
    dT3 = -k_T3 * (T3 - T3_setpoint(t))
    dG = k_G * (Glucagon_setpoint(t) - G)
    return [dT3, dG]

Y0 = [T3_full, G_base]
sol = odeint(axis_rhs, Y0, t_days, atol=1e-8, rtol=1e-6)
T3 = sol[:, 0]
G = sol[:, 1]
I = np.array([Insulin(t) for t in t_days])

np.random.seed(17)
CORT = CORT_signal(t_days, t_abandon=110.0) + 1.2 * np.random.randn(len(t_days))

# Ketone bodies (beta-hydroxybutyrate): rise with fat oxidation
# Robin et al. 1988: BHB rises from 0.1 mM pre-fast to ~2.5 mM mid-fast, falls
# to <0.5 mM in Phase III as lipid depletes.
BHB = 0.1 + 2.4 * np.exp(-((t_days - 60.0) / 30.0) ** 2)

# Glycemia: maintained via gluconeogenesis until Phase III
glucose = 11.0 - 3.0 * np.clip((t_days - 100.0) / 20.0, 0.0, 1.0) + 0.2 * np.random.randn(len(t_days))

# Behavioural abandonment decision
# Rule: abandon when dCORT/dt > threshold AND CORT > 40 ng/mL
dCORT_dt = np.gradient(CORT, t_days)
abandon_mask = (CORT > 40.0) & (dCORT_dt > 0.5)
abandon_idx = np.argmax(abandon_mask) if np.any(abandon_mask) else -1
t_abandon = t_days[abandon_idx] if abandon_idx >= 0 else np.nan

phase_bands = [(0, 70, '#a3e635', 'Phase I'), (70, 100, '#fbbf24', 'Phase II'),
               (100, 135, '#f43f5e', 'Phase III')]

ax = axes[0, 0]
for (a, b, c, _) in phase_bands:
    ax.axvspan(a, b, color=c, alpha=0.08)
ax.plot(t_days, T3, color='#fbbf24', lw=2.2, label='T3')
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('T3 (ng/mL)', color='#e2e8f0')
ax.set_title('Thyroid hormone depression (Cherel 1988)',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[0, 1]
for (a, b, c, _) in phase_bands:
    ax.axvspan(a, b, color=c, alpha=0.08)
ax.plot(t_days, G, color='#a3e635', lw=2.2, label='Glucagon')
ax.plot(t_days, I, color='#60a5fa', lw=1.4, alpha=0.7, label='Insulin (low plateau)')
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Hormone (pg/mL or muU/mL)', color='#e2e8f0')
ax.set_title('Pancreatic axis: glucagon rises, insulin suppressed',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[0, 2]
for (a, b, c, _) in phase_bands:
    ax.axvspan(a, b, color=c, alpha=0.08)
ax.plot(t_days, CORT, color='#fb7185', lw=1.4, label='Corticosterone')
ax.axhline(40.0, color='#f43f5e', ls='--', alpha=0.7,
           label='abandonment criterion')
if abandon_idx >= 0:
    ax.axvline(t_abandon, color='#fbbf24', ls=':',
               label=f'abandon at t = {t_abandon:.1f} d')
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('CORT (ng/mL)', color='#e2e8f0')
ax.set_title('Corticosterone bimodal profile + abandonment spike',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0', fontsize=9)

ax = axes[1, 0]
for (a, b, c, _) in phase_bands:
    ax.axvspan(a, b, color=c, alpha=0.08)
ax.plot(t_days, BHB, color='#fde68a', lw=2.2)
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('beta-hydroxybutyrate (mM)', color='#e2e8f0')
ax.set_title('Ketone bodies: brain fuel in prolonged fast',
             color='#fde68a', fontsize=12, fontweight='bold')

ax = axes[1, 1]
for (a, b, c, _) in phase_bands:
    ax.axvspan(a, b, color=c, alpha=0.08)
ax.plot(t_days, glucose, color='#60a5fa', lw=1.8)
ax.axhline(4.0, color='#f43f5e', ls='--', alpha=0.5,
           label='hypoglycaemia threshold')
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Plasma glucose (mM)', color='#e2e8f0')
ax.set_title('Glycaemia maintained by gluconeogenesis',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

ax = axes[1, 2]
for (a, b, c, _) in phase_bands:
    ax.axvspan(a, b, color=c, alpha=0.08)
ax.plot(t_days, L, color='#fbbf24', lw=2.2, label='Lipid fraction')
ax.set_xlabel('Time (days)', color='#e2e8f0')
ax.set_ylabel('Lipid reserves (norm.)', color='#e2e8f0')
ax.set_title('Lipid depletion driving the hormonal axis',
             color='#fde68a', fontsize=12, fontweight='bold')
ax.legend(facecolor='#1e293b', edgecolor='#475569', labelcolor='#e2e8f0')

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

print('T3 at day 0:   %.2f ng/mL' % T3[0])
print('T3 at day 120: %.2f ng/mL' % T3[np.argmin(np.abs(t_days - 120))])
print('Glucagon peak: %.1f pg/mL' % G.max())
print('CORT peak:     %.1f ng/mL' % CORT.max())
print('BHB peak day:  %.1f'       % t_days[np.argmax(BHB)])
print('Predicted abandonment day: %.1f' % t_abandon)
print('Integrated corticosterone AUC day 0-130 (ng*d/mL): %.1f'
      % float(np.trapezoid(CORT, t_days)))

Click Run to execute the Python code

Code will be executed with Python 3 on the server

8. Thermogenesis Without Brown Fat

Most endotherms in cold environments possess brown adipose tissue (BAT), enriched in uncoupling protein 1 (UCP1) and capable of non-shivering thermogenesis. Emperor penguins (like adult birds generally) lack BAT(Hohtola 2002). Thermogenic heat comes from two sources:

Shivering thermogenesis: asynchronous contraction of pectoralis and leg muscles. Accounts for 60–80% of thermogenic heat in cold stress.
Substrate oxidation obligate heat: ATP yield from FFA oxidation is ~40% efficient; the remaining 60% of chemical energy appears as heat. Accounts for 20–40% of total heat, proportional to metabolic rate.

The absence of BAT has two consequences for the fasting male. First, the muscle mass that drives shivering cannot be catabolised as readily as it could in a BAT-equipped animal — protecting essential shivering-capable muscle delays protein catabolism onset. Second, there is no fast-acting UCP1-based heat source for extreme cold pulses; the emperor instead relies on the behavioural strategy of huddling (Module 2) to minimise individual heat loss rather than dialling up thermogenesis.

\[ P_{\text{therm}} \;=\; P_{\text{shiver}} + (1 - \eta_{\text{ox}}) \cdot P_{\text{metab}} \]

Total heat dissipation: shivering contribution + obligate heat from substrate oxidation (oxidative efficiency \(\eta_{\text{ox}} \approx 0.4\)).

9. Comparative Metabolic Challenges

Species	Duration	Mass loss (%)	T_body	Key mechanism
Emperor male	120 d	35–45	Euthermic 38°C	3-phase lipolysis
Brown bear	150 d	20–30	32°C (mild torpor)	Torpor + urea recycling
Ground squirrel	180 d	30–40	0–5°C	Deep hibernation
Hummingbird	< 12 h	10–20	10–20°C	Nocturnal torpor
Human (starvation)	60–90 d	30–40	Euthermic	Ketogenesis + protein sparing
Elephant seal	70 d	25–35	Euthermic	Molt + lactation

The emperor’s combination of long duration, full euthermy, cold exposure, and simultaneous incubation sets a vertebrate benchmark. Bears achieve similar duration via metabolic suppression (reducing BMR by 75% during torpor); ground squirrels via deep hibernation (reducing body temperature by 35°C). The emperor maintains normal body temperature and incubation behaviour through the entire fast, powered only by the lipid depot he accumulated during the late-summer pelagic foraging period.

10. Water Balance: Fasting Without Drinking

The emperor also fasts from water: liquid water is unavailable in the winter colony (ambient \(\sim -30\)°C). Water comes from three sources:

Metabolic water: oxidation of 1 g fat produces ~1.07 g water. Oxidation of 1 g protein produces ~0.41 g water. Over 120 days of fasting, total metabolic water production is ~10–12 L.
Snow ingestion: emperors occasionally consume fresh snow blown onto their feet, but the thermoregulatory cost (melting 1 g of ice requires 334 J) limits this.
Recycling from breath and urine: the nasal salt-gland/kidney system concentrates waste, reducing water loss.

Water loss is dominated by respiratory evaporation. Emperors minimise it via the nasal counter-current heat exchanger (Murrish 1973), which recovers up to 80% of exhaled water vapour by cooling the nasal turbinates below core temperature. Respiratory water loss is an unavoidable cost of incubation in cold, dry air.

\[ \dot{m}_{w,\text{resp}} \;=\; \dot{V}_E \,(\chi_{\text{exhale}} - \chi_{\text{inhale}}) \]

Respiratory water loss rate: product of ventilation rate and the difference in humidity ratio \(\chi\) (kg H2O/kg dry air) between exhaled and inhaled air. Nasal counter-current exchange reduces the humidity difference.

11. Refeeding: Rapid Lipid Anabolism

When the female returns and takes over incubation, the male heads to the ice edge, traverses 50–150 km to the ocean, and begins feeding. Robin, Boucontet, Chillet & Groscolas (1998) measured the refeeding kinetics: a 40%-depleted emperor regains 200–300 g/day during initial refeeding, reaching pre-fast mass within 30–50 days. Most of the regained mass is lipid (~60%), proportional to the triacylglycerol fraction of the consumed fish.

The rapid anabolic response requires reactivation of the insulin-AKT-mTOR axis, induction of fatty-acid synthase (FAS) and acetyl-CoA carboxylase (ACC), and restoration of glycogen stores. Plasma insulin rises 4–5\(\times\) within 24 h of first meal; corticosterone falls to pre-fast baseline within a week.

12. Synthesis: The Biochemical Arithmetic of a 120-Day Fast

An emperor male arrives at the colony with 12–14 kg of lipid and 6–7 kg of protein (lean-tissue pool, exclusive of essential structure). His average metabolic power over the fast is \(\sim 110\) W (130 W in Phase I cold pulses, 100 W during thermal-neutral huddles). Over 120 days that is\(\sim 110 \times 120 \times 86400 \approx 1.14\times 10^9\) J = 1140 MJ, but the conversion to energy stores implies only ~290 MJ of substrate consumed because mass loss (~13 kg fat + ~3 kg protein) * energy density (~40 kJ/g fat, ~18 kJ/g protein) = ~570 MJ of substrate. The gap reflects the fact that the huddle strategy and thermogenic efficiency keep the effective average power closer to ~60–70 W rather than the nominal 110 W.

Every phase of the fast is orchestrated: the slow T3 decline protects thermogenesis; the glycerol-glucose pathway spares protein; the corticosterone bimodal profile times the abandonment decision to preserve the male’s own survival. The 22% mass threshold is not an arbitrary physiological failure point but an evolved behavioural criterion selected over millennia to balance reproductive success against survival risk.

When Module 5 asked how a penguin swims at 30 km/h and Module 6 asked how he recognises his chick in a 10 000-bird colony, Module 7 asks how he stands still on a sea-ice pack at\(-30\)°C for four months without eating. The emperor is the most biochemically-committed parent among vertebrates.

Discussion & Graduate Exercises

Derive the total glucose production required from gluconeogenesis in Phase I if the brain consumes 12 g glucose/day. Assuming 60% of it comes from glycerol, compute the lipolysis rate (g TAG/day) required.
Modify Simulation 1 to include a stochastic cold-stress perturbation (a 5% metabolic spike at random days). How much earlier is the 22% mass threshold reached?
Fit the Spee (2010) bimodal corticosterone dataset with a two-component sigmoid+baseline model. Compare predicted abandonment to observed behaviour in individual tracked birds.
Compare the emperor’s lipid-to-protein loss ratio (3:1) with that of a fasting human (10:1) and a hibernating ground squirrel (30:1). What physiological factors drive the differences?
Compute the fraction of total fasting metabolic rate accounted for by obligate oxidative heat vs. active shivering. Use fat and protein oxidation ratios and a 40% ATP-yield efficiency.
The Groscolas 22% threshold assumes a fixed critical mass. Propose an experimental test that would differentiate (a) mass-threshold vs. (b) metabolite-threshold (plasma amino acid spike) vs. (c) CORT-threshold as the primary abandonment trigger.

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., Robin, J.-P. (2001). “Long-term fasting and re-feeding in penguins.” Comp. Biochem. Physiol. A 128, 645–655.

• Groscolas, R., Lacroix, A., Robin, J.-P. (2008). “Spontaneous egg or chick abandonment in energy-depleted emperor penguins: a role for corticosterone and prolactin?” Horm. Behav. 53, 51–60.

• Le Maho, Y. (1977). “The emperor penguin: a strategy to live and breed in the cold.” American Scientist 65, 680–693.

• Cherel, Y., Robin, J.-P., Walch, O., Karmann, H., Netchitailo, P., Le Maho, Y. (1988). “Fasting in king penguin. I. Hormonal and metabolic changes during breeding.” Am. J. Physiol. 254, R170–R177.

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

• Robin, J.-P., Frain, M., Sardet, C., Groscolas, R., Le Maho, Y. (1988). “Protein and lipid utilization during long-term fasting in emperor penguins.” Am. J. Physiol. 254, R61–R68.

• Robin, J.-P., Boucontet, L., Chillet, P., Groscolas, R. (1998). “Behavioral changes in fasting emperor penguins: evidence for a ‘refeeding signal’.” Am. J. Physiol. 274, R746–R753.

• Spee, M., Beaulieu, M., Dervaux, A., Chastel, O., Le Maho, Y., Raclot, T. (2010). “Should I stay or should I go? Hormonal control of nest abandonment in a long-lived bird, the emperor penguin.” Horm. Behav. 58, 762–768.

• Spee, M., Marchal, L., Thierry, A.-M., Chastel, O., Enstipp, M., Le Maho, Y., Beaulieu, M., Raclot, T. (2011). “Exogenous corticosterone mimics a late fasting stage in captive emperor penguins.” Am. J. Physiol. 300, R1241–R1249.

• Hohtola, E. (2002). “Facultative and obligatory thermogenesis in young birds.” Comp. Biochem. Physiol. A 131, 733–739.

• Bodine, S.C., Baehr, L.M. (2014). “Skeletal muscle atrophy and the E3 ubiquitin ligases MuRF1 and MAFbx/atrogin-1.” Am. J. Physiol. Endocrinol. Metab. 307, E469–E484.

• Kitaysky, A.S., Wingfield, J.C., Piatt, J.F. (2001). “Corticosterone facilitates begging and affects resource allocation in the black-legged kittiwake.” Behav. Ecol. 12, 619–625.

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

• Cahill, G.F. Jr. (2006). “Fuel metabolism in starvation.” Annu. Rev. Nutr. 26, 1–22.

• Castellini, M.A., Rea, L.D. (1992). “The biochemistry of natural fasting at its limits.” Experientia 48, 575–582.

Synthesis & Bridge to Module 8

The 120-day fast stands as the single most extreme metabolic feat among vertebrates: 40% body-mass loss while maintaining euthermia, incubating an egg, and producing individually-coded calls. The three-phase Groscolas model, the Spee corticosterone bimodality, and the 22% abandonment threshold together describe a finely-tuned physiological system pushed to the edge of its operating range every winter.

Module 8 confronts the sobering climatic reality: a changing Southern Ocean, retreating fast-ice, H5N1 outbreaks, and the demographic projections of Jenouvrier (2014, 2021) for a species whose life history is so intimately tied to the timing and extent of Antarctic sea ice. We close the course with the conservation biology that emerges from all the physiology we have studied — and the question of what, if anything, can be done.

← Module 6: Two-Voice Syrinx Module 8: Climate & Conservation →