Module 8: Whole-Tree Integration & Scaling

From a single stomatal guard cell to the trunk of a giant sequoia spanning 9 orders of magnitude in scale: how do the molecular and cellular processes described in previous modules integrate into whole-tree function? This final module addresses the quantitative laws governing tree architecture, metabolic scaling, carbon balance, and how trees respond to global change. The West-Brown-Enquist (WBE) vascular network theory unifies the pipe model, Kleiber's law, and allometric scaling in a single mechanistic framework.

8.1 Pipe Model Theory: Sapwood & Leaf Area

The pipe model (Shinozaki et al., 1964) states that each unit of leaf area is supported by a corresponding "unit pipe" of xylem (sapwood). This implies a constant proportionality between sapwood cross-sectional area and the leaf area it supplies:

\[ A_{sapwood} = k_p \cdot A_{leaf} \]

where \( k_p \) (the pipe model coefficient) has units of m² sapwood per m² leaf area, typically ranging from 1×10⁻⁴ to 5×10⁻⁴ m² m⁻² in conifers and 0.5–3×10⁻⁴ in hardwoods. The pipe model has been empirically validated across species and can be used to estimate leaf area index (LAI) from increment cores.

The Huber value (HV) is the inverse perspective: sapwood area per leaf area. It is a key hydraulic safety margin — a high HV means each unit of leaf area has a large hydraulic supply, reducing vulnerability to drought-induced hydraulic failure. In xeric species like Pinus halepensis, HV is ~5× higher than in mesic species.

Hydraulic Implication:

The maximum stomatal conductance sustainable without runaway embolism (\( g_{s,max} \)) is directly proportional to HV and the specific hydraulic conductivity \( k_s \)of the sapwood: \( g_{s,max} \propto k_s \cdot HV \cdot |\Delta\Psi| \). Tree species differ by ~10-fold in \( k_s \) and ~5-fold in HV, creating a ~50-fold range in \( g_{s,max} \) across the flora.

8.2 West-Brown-Enquist (WBE) Vascular Scaling Theory

West, Brown & Enquist (1997, 1999) derived metabolic scaling from first principles by modeling the vascular network as a space-filling, area-preserving hierarchical branching system that minimizes the energy cost of fluid distribution. The three key assumptions are:

Space-filling fractal network

The network fills the organism volume: branching is self-similar at all scales from trunk to terminal capillaries (petioles, terminal tracheids)

Area-preserving branching

The sum of daughter-branch cross-sectional areas equals the parent area: r_k² = N · r_{k+1}², ensuring constant fluid velocity and pressure wave speed across branching levels

Invariant service volume

Terminal units (capillaries, terminal branch segments) are approximately constant in size across organisms — the biological clock ticks at the same rate per capillary

Area-Preserving Branching: Mathematical Formulation

At branching level \( k \), a tube of radius \( r_k \) branches into\( N \) daughter tubes of radius \( r_{k+1} \). Area preservation requires:

\[ \pi r_k^2 = N \cdot \pi r_{k+1}^2 \implies r_{k+1} = \frac{r_k}{N^{1/2}} \]

After \( K \) levels of branching, the terminal tube radius is:

\[ r_K = r_0 \cdot N^{-K/2} = r_0 \cdot n_{cap}^{-1/2} \]

where \( n_{cap} = N^K \) is the total number of terminal capillaries (proportional to metabolic rate \( B \)). The total number of capillaries scales with organism mass \( M \). Since capillary density is constant (invariant service volume):\( n_{cap} \propto M \). Therefore:

\[ n_{cap} \propto M \implies B \propto n_{cap} \propto M^1 \quad\text{(naively)} \]

But this ignores the transport efficiency constraint. The Hagen-Poiseuille conductance of level \( k \) is \( K_k \propto r_k^4 / l_k \). With the WBE segment length scaling \( l_{k+1} = l_k \cdot N^{-1/3} \)(from space-filling), and area preservation:

Deriving Kleiber's Law from WBE

The key step is recognizing that the metabolic rate \( B \) is proportional to the total number of capillaries \( N_c \), but the relationship between\( N_c \) and mass \( M \) is not simply 1:1 due to the mass of the transport network itself. Following WBE's exact derivation:

The total volume of the network is:

\[ V_{network} = \sum_{k=0}^{K} N^k \pi r_k^2 l_k \propto r_0^2 l_0 \sum_{k=0}^{K} \left(\frac{N}{N^{2/2} N^{1/3}}\right)^k = r_0^2 l_0 \sum_{k=0}^{K} N^{-k/3} \]

For large \( K \), the sum converges and \( V_{network} \propto r_0^2 l_0 \cdot K^0 \)(approximately constant per unit mass). Since organism mass \( M \propto V_{organism} \propto l_0^3 \)and \( r_0 \propto l_0^{1/2} \) (from area preservation):

\[ N_c \propto M \cdot \frac{r_0^2}{r_c^2} \propto M \cdot \frac{r_0^2}{r_c^2} \]

Since capillary radius \( r_c \) is invariant and \( r_0 \propto M^{3/8} \)(from the full WBE geometry):

\[ B \propto N_c \propto M^{3/4} \quad\text{— Kleiber's Law} \]

The 3/4 exponent emerges specifically from the interplay between three-dimensional space filling and the two-dimensional constraint of area preservation. If branching were volume-preserving instead (\( r_k^3 = N r_{k+1}^3 \)), the exponent would be 2/3 (Rubner's surface law). Area preservation — motivated by the requirement for constant fluid velocity (and thus constant shear stress and pressure wave speed) — is the physical origin of the 3/4 law.

8.3 Whole-Tree Carbon Balance: GPP, Respiration & NPP

The carbon economy of a tree is governed by the balance between photosynthetic carbon gain (GPP) and respiratory carbon loss. Net Primary Productivity (NPP) — the carbon available for growth, reproduction, and defense — is the difference:

\[ NPP = GPP - R_{auto} \]

Autotrophic respiration (\( R_{auto} \)) comprises:

Growth respiration (Rg)

~25% of NPP

The ~25% carbon overhead of biosynthesis; fixed by biochemistry regardless of temperature

Maintenance respiration (Rm)

~55–75% of Rauto

Scales with living biomass and temperature (Q10 ≈ 2–2.5); dominated by woody tissue in large trees

Tissue turnover

~10–20% of Rauto

Fine root turnover (50–200% per year), leaf replacement, bark shedding — major hidden cost of sessile life

The carbon use efficiency (CUE = NPP/GPP) averages approximately 0.47 across forest biomes (DeLucia et al., 2007), with a range of 0.3–0.7. Large old trees have lower CUE than young trees because maintenance respiration of their large sapwood volume increases while GPP saturates or declines. This explains the hump-shaped relationship between tree diameter and growth rate, and the empirical finding that net carbon uptake of individual trees peaks at intermediate size (~50 cm DBH for many temperate species).

8.4 Cowan-Farquhar Stomatal Optimization: Lagrangian Derivation

Cowan & Farquhar (1977) proposed that stomata maximize total daily carbon gain (\( A \)) subject to a fixed total daily water loss (\( E \)). This constrained optimization can be solved formally using Lagrange multipliers.

Lagrangian Setup

Integrate over the day: maximize \( \int A \, dt \) subject to\( \int E \, dt = W \) (fixed water budget). The Lagrangian is:

\[ \mathcal{L} = \int \!\left[ A(g_s, \ldots) - \lambda \cdot E(g_s, \ldots) \right] dt \]

where \( g_s \) is stomatal conductance (the control variable),\( \lambda \) is the Lagrange multiplier (marginal water use cost, Pa mol⁻¹ or mol CO₂ mol⁻¹ H₂O), and \( A \), \( E \) are instantaneous assimilation and transpiration. Taking the functional derivative with respect to\( g_s \) and setting to zero:

Cowan-Farquhar Optimality Condition:

\[ \frac{\partial A}{\partial g_s} = \lambda \frac{\partial E}{\partial g_s} \]

This is equivalent to: dA/dE = λ = constant throughout the day. The optimal stomatal strategy maintains a constant marginal water use efficiency. \( \lambda \)has units of mol CO₂ mol⁻¹ H₂O.

Expanding the partial derivatives using Fick's law (\( A = g_s(c_a - c_i) \), \( E = g_s \cdot VPD / P \)) and the Farquhar photosynthesis model (\( A = V_{cmax} c_i / (c_i + K_m) - R_d \)):

\[ \frac{\partial A / \partial c_i}{\partial E / \partial c_i} = \lambda \implies c_i^* = c_a - \sqrt{\frac{\partial E/\partial g_s}{\partial A / \partial g_s} \cdot \lambda} \]

The key prediction is that \( c_i/c_a \) (the ratio of internal to ambient CO₂) should be approximately constant for a given λ and VPD. This is largely confirmed by observations:\( c_i/c_a \approx 0.65–0.75 \) for C3 trees under moderate conditions. Under elevated VPD, stomata partially close to maintain this ratio, trading reduced water loss for reduced carbon gain — the λ = const. condition.

The optimal stomatal conductance \( g_s^* \) can be written as:

\[ g_s^* = \frac{A}{c_a} \left(1 + \frac{\sqrt{A \cdot VPD}}{\lambda c_a}\right) \]

This forms the basis of the USO (Unified Stomatal Optimization) model and is implemented in most modern land surface models (CLM, ORCHIDEE) replacing the empirical Ball-Berry formulation.

8.5 Functional-Structural Plant Models & Climate Change

Functional-Structural Plant Models (FSPMs)

FSPMs explicitly represent both the 3D geometric structure of a plant (architecture: phytomer topology, branching angles, organ sizes) and the physiological functions running on that structure (photosynthesis, water flow, carbon allocation, senescence). This integration allows emergent phenomena impossible in big-leaf models: light competition within canopies, self-shading, asymmetric growth, effects of tree form on wind drag and stem stress.

L-systems (Lindenmayer)

String rewriting systems for formal description of plant topology. Parametric L-systems encode physiology: each rewriting rule carries biophysical parameters. Implemented in L-Py (Python) and the Vlab environment.

OpenAlea

Python-based modular plant simulation platform. Integrates L-systems, Mockup Turtle (3D rendering), energy budget models, and FSPMs. Used for grapevine, apple, and poplar canopy models.

LIGNUM

Process-based FSPM for coniferous trees. Represents a tree as a collection of tree segments (cylinders), buds, and foliage clusters. Explicitly models pipe model relationships, light interception, and growth allocation per unit.

Climate Change Effects on Trees

CO₂ Fertilization & Water Use Efficiency

Elevated CO₂ (eCO₂) increases photosynthesis through substrate supply (especially for Rubisco-limited photosynthesis in C3 trees) and reduces stomatal conductance by ~20% at doubled CO₂ (via HT1 kinase/guard cell CO₂ signaling). The combination increases intrinsic water use efficiency (iWUE = A/gs) by 20–40%. FACE experiments show a consistent but declining 20–30% GPP enhancement at 2× CO₂, limited by nutrient co-limitation (progressive nitrogen limitation, PNL) as trees grow faster but nitrogen mineralizes at fixed rates.

Hydraulic Failure vs Carbon Starvation Debate

The mechanism of drought-induced tree mortality remains actively debated (McDowell et al., 2008). Hydraulic failure: runaway embolism as xylem water potential falls below P88 → cessation of water flow → desiccation. Evidence: P50 correlates with mortality threshold in Pinus edulis. Carbon starvation: stomatal closure during drought prevents embolism but stops photosynthesis → NSC depletion → inability to maintain membranes, grow fine roots, or fight pathogens. Evidence: declining NSC (non-structural carbohydrates) before death in many species. Current consensus: both mechanisms operate simultaneously, with hydraulic failure dominant in fast-drying events and carbon starvation in prolonged moderate drought.

VPD Effects on Tree Water Balance

Atmospheric vapor pressure deficit (VPD) has increased ~10% globally since 1970 and is projected to increase further at +2°C warming. VPD directly drives transpiration (\( E \propto g_s \cdot VPD \)) and forces stomatal closure via the ABA and hydraulic pathways. Meta-analyses show tree growth correlates negatively with VPD across the boreal and temperate zones, with a threshold around 1.5–2 kPa above which growth is limited even under adequate soil moisture. This "atmospheric drought" effect is distinct from soil drought and is captured by the Cowan-Farquhar optimization: higher VPD raises the carbon cost of water, reducing \( g_s^* \) and photosynthesis.

Metabolic Scaling: Log-Log Plot

Log-log plot of metabolic rate B vs mass M for trees from seedlings (~1 g) to giant sequoia (~500 tonnes). WBE predicts slope = 3/4, distinct from isometric (slope = 1) and Rubner surface scaling (slope = 2/3). The 3/4 slope means large trees are more metabolically efficient per unit mass.

Python: WBE Network, Kleiber Scaling & Carbon Balance

Three quantitative models: (1) WBE branching network simulation — how conduit radius, segment length, and conduit number change across 15 branching levels from trunk to terminal twigs, (2) the metabolic scaling law B ~ M^(3/4) plotted across 9 orders of magnitude of tree mass (seedling to giant sequoia) compared to isometric and Rubner scaling, and (3) the whole-tree carbon balance (GPP, autotrophic respiration, NPP) as a function of tree diameter, showing the hump-shaped NPP curve that peaks at intermediate tree size.

Python

script.py140 lines

import numpy as np
import matplotlib.pyplot as plt
# output saved as output.png

fig, axes = plt.subplots(1, 3, figsize=(17, 6), facecolor='#020f06')
fig.subplots_adjust(wspace=0.45)

# ---- Panel 1: WBE branching network - total conductance from trunk to leaves ----
ax1 = axes[0]
ax1.set_facecolor('#041a0a')

# WBE area-preserving branching: r_k^2 = N * r_{k+1}^2
# So r_{k+1} = r_k / sqrt(N)
# Conductance of a tube: K = pi*r^4/(8*eta*l) [Hagen-Poiseuille]
# l_{k+1} = l_k / n^(1/3) (from WBE)

N = 2        # branching ratio (bifurcation)
eta = 1e-3   # Pa.s  (water viscosity)
r_trunk = 0.10  # m   (10 cm trunk radius)
l_trunk = 20.0  # m   (stem length at level 0)
n_levels = 15   # number of branching levels from trunk to terminal twigs

levels = np.arange(n_levels + 1)
r_vals = r_trunk / (np.sqrt(N) ** levels)       # radii
l_vals = l_trunk / (N ** (1/3)) ** levels        # lengths
n_tubes = N ** levels                            # number of tubes at each level

# Resistance of single tube at level k: R_k = 8*eta*l_k / (pi*r_k^4)
R_single = 8 * eta * l_vals / (np.pi * r_vals**4)

# Total resistance at each level: R_level = R_single / n_tubes (parallel)
R_level = R_single / n_tubes

# Cumulative total resistance from trunk
R_cumulative = np.cumsum(R_level)
K_cumulative = 1.0 / R_cumulative

# Normalize to initial (trunk alone)
K_norm = K_cumulative / K_cumulative[-1] * 100

ax1.plot(levels, r_vals * 100, color='#34d399', lw=2, label='Conduit radius (cm)')
ax1.plot(levels, l_vals, color='#fbbf24', lw=2, label='Segment length (m)')
ax2_twin = ax1.twinx()
ax2_twin.plot(levels, n_tubes, color='#f87171', lw=2, ls='--', label='Number of conduits')
ax2_twin.set_yscale('log')
ax2_twin.tick_params(colors='#f87171', labelsize=7.5)
ax2_twin.set_ylabel('Number of conduits (log scale)', color='#f87171', fontsize=8)
for sp in ax2_twin.spines.values(): sp.set_edgecolor('#34d399')

ax1.set_xlabel('Branching level (0=trunk, N=terminal)', color='#6ee7b7', fontsize=9)
ax1.set_ylabel('Radius (cm) / Length (m)', color='#6ee7b7', fontsize=9)
ax1.set_title('WBE Branching Network\
(area-preserving, N=2)', color='#d9f99d', fontsize=10)
ax1.legend(fontsize=7.5, labelcolor='#d9f99d', facecolor='#041a0a', edgecolor='#34d399', loc='upper right')
ax1.tick_params(colors='#6ee7b7', labelsize=8)
ax1.grid(True, alpha=0.2, color='#34d399')
for sp in ax1.spines.values(): sp.set_edgecolor('#34d399')

# ---- Panel 2: Metabolic scaling law B vs M for trees ----
ax2 = axes[1]
ax2.set_facecolor('#041a0a')

# WBE predicts B = B0 * M^(3/4) - Kleiber's law
# B0 ~ 0.047 W kg^-3/4 for trees (basal metabolic rate, whole tree)
B0 = 0.047   # W kg^(-3/4)  for trees (approximate)
# Mass range: seedling 0.001 kg to giant sequoia ~1.4e6 kg
M_vals = np.logspace(-3, 6, 400)  # kg

B_3_4 = B0 * M_vals ** (3/4)   # WBE prediction
B_1_0 = B0 * M_vals ** 1.0     # isometric scaling (null)
B_2_3 = B0 * M_vals ** (2/3)   # surface area scaling (Rubner)

ax2.loglog(M_vals, B_3_4, color='#34d399', lw=2.5, label='WBE: B ~ M^(3/4)')
ax2.loglog(M_vals, B_1_0, color='#f87171', lw=1.5, ls='--', label='Isometric: B ~ M^1')
ax2.loglog(M_vals, B_2_3, color='#fbbf24', lw=1.5, ls=':', label='Rubner: B ~ M^(2/3)')

# Mark specific tree sizes
trees = [
    ("Seedling\n1g", 0.001),
    ("Sapling\n10kg", 10),
    ("Oak\n5t", 5000),
    ("Redwood\n500t", 5e5),
]
for name, M in trees:
    B = B0 * M ** (3/4)
    ax2.scatter([M], [B], s=60, color='#d9f99d', zorder=5)
    ax2.annotate(name, (M, B), textcoords="offset points", xytext=(5, 5), fontsize=6.5, color='#d9f99d')

ax2.set_xlabel('Tree mass M (kg)', color='#6ee7b7', fontsize=9)
ax2.set_ylabel('Basal metabolic rate B (W)', color='#6ee7b7', fontsize=9)
ax2.set_title('Metabolic Scaling: Kleiber Law\
(B = B0 * M^(3/4), WBE prediction)', color='#d9f99d', fontsize=10)
ax2.legend(fontsize=7.5, labelcolor='#d9f99d', facecolor='#041a0a', edgecolor='#34d399')
ax2.tick_params(colors='#6ee7b7', labelsize=8)
ax2.grid(True, alpha=0.2, color='#34d399', which='both')
for sp in ax2.spines.values(): sp.set_edgecolor('#34d399')

# ---- Panel 3: GPP, Rauto, NPP partitioning across tree size classes ----
ax3 = axes[2]
ax3.set_facecolor('#041a0a')

# Autotrophic respiration fraction ~ 0.47 of GPP (average for forests)
# NPP/GPP ratio varies with tree size and temperature
# Model: NPP = GPP * (1 - CUE_auto) where CUE_auto increases with size
# GPP scales as: GPP = Gmax * (D/Dmax)^alpha * exp(-(D-Dopt)^2/sigma^2)
D_vals = np.linspace(0, 120, 300)  # DBH in cm
Dmax = 120; Dopt = 50; sigma = 40; Gmax = 45.0; alpha = 0.5

GPP = Gmax * (D_vals / Dmax) ** alpha * np.exp(-((D_vals - Dopt)**2) / (2 * sigma**2))
# Autotrophic respiration increases as fraction of GPP with size (maintenance cost)
R_frac = 0.35 + 0.15 * (D_vals / Dmax)   # fraction, 35-50%
R_auto = GPP * R_frac
NPP    = GPP - R_auto

ax3.fill_between(D_vals, GPP, color='#34d399', alpha=0.25, label='GPP')
ax3.fill_between(D_vals, NPP, color='#4ade80', alpha=0.5, label='NPP')
ax3.fill_between(D_vals, R_auto, color='#f87171', alpha=0.4, label='R_auto')
ax3.plot(D_vals, GPP,   color='#34d399', lw=2)
ax3.plot(D_vals, NPP,   color='#86efac', lw=2, ls='--')
ax3.plot(D_vals, R_auto, color='#f87171', lw=1.5, ls=':')
ax3.axvline(Dopt, color='#fbbf24', lw=1, ls=':', alpha=0.7)
ax3.text(Dopt+2, 30, 'GPP peak\
~50 cm DBH', color='#fbbf24', fontsize=7.5)

ax3.set_xlabel('Stem diameter DBH (cm)', color='#6ee7b7', fontsize=9)
ax3.set_ylabel('Carbon flux (Mg C ha-1 yr-1)', color='#6ee7b7', fontsize=9)
ax3.set_title('Carbon Balance vs Tree Size\
(GPP, R_auto, NPP)', color='#d9f99d', fontsize=10)
ax3.legend(fontsize=8, labelcolor='#d9f99d', facecolor='#041a0a', edgecolor='#34d399')
ax3.tick_params(colors='#6ee7b7', labelsize=8)
ax3.grid(True, alpha=0.2, color='#34d399')
for sp in ax3.spines.values(): sp.set_edgecolor('#34d399')

fig.suptitle('Whole-Tree Scaling: WBE Network, Metabolic Scaling Law & Carbon Balance', color='#d9f99d', fontsize=11, fontweight='bold')

plt.savefig('output.png', bbox_inches='tight', facecolor=fig.get_facecolor(), dpi=120)
plt.close()
print('Simulation complete.')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

←Stress Biophysics Course Overview→