Module 7: Ornamental Horticulture & Breeding

Flowers are a $40 billion global industry. This module covers the breeding of cultivated ornamentals — from tulip mania in 17th-century Amsterdam to modern transgenic blue roses — and the biochemistry of vase life. We derive the breeder’s equation, the Lockhart law applied to post-harvest turgor, and the ethylene climacteric that ends petal life.

1. Cut-Flower Industry & Tulip Mania

Tulips arrived in the Netherlands in the 1590s via the Ottoman court. The tulip mania of 1636–1637 saw single bulbs trade for the price of Amsterdam canal-houses before the market collapsed in February 1637 — the first recorded asset bubble. The striped “broken” tulips that drove the mania were infected by the aphid-transmitted tulip breaking virus, which silences anthocyanin synthesis in stripes of cells.

Today the industry is industrial: the Netherlands alone auctions over 12 billion stems per year through Royal Flora Holland, Kenya and Ethiopia lead African exports, Colombia and Ecuador supply the North American rose market, and China is the world’s largest producer of cut chrysanthemums.

2. Breeding Techniques

Interspecific hybridization

Modern roses (Rosa × hybrida) derive from ~10 wild species, primarily R. chinensis (recurrent flowering), R. gallica (strong scent), R. gigantea (long stems), and R. moschata (climbing habit). The key 1867 cross that produced ‘La France’ by Guillot marks the beginning of the modern Hybrid Tea class.

Mutation breeding

Gamma-ray irradiation (typically 20–40 Gy to dormant buds) has produced hundreds of chrysanthemum colour mutants — more than 60% of commercial chrysanthemum cultivars in Japan trace to such mutations. Radiation induces point mutations and chromosomal rearrangements that silence or duplicate anthocyanin-pathway loci. The rate of useful variants is low (~1%) but the method is cheap and non-transgenic.

Polyploidy induction with colchicine

Colchicine binds tubulin and blocks spindle formation, producing cells with doubled chromosome number. A 24-hour 0.05% treatment on seedlings of diploid lilies yields 1–5% tetraploid recoveries with larger flowers, thicker leaves, and often increased disease resistance. Hemerocallis (daylily) tetraploid cultivars account for the majority of modern registrations.

Genetic engineering: the blue rose

Roses lack the enzyme flavonoid 3′,5′-hydroxylase (F3′5′H) required to make blue delphinidin pigments. Suntory and Florigene engineered a transgenic rose expressing F3′5′H from pansy plus a rose DFR knockdown, yielding the lavender-blue cultivar Applause released in 2009. Similar pathways produced the Moondust and Moonlite carnations in the 1990s.

Derivation: the breeder’s equation

Suppose a phenotype $P = G + E$ is the sum of an additive genetic component$G$ and an environmental one $E$. The narrow-sense heritability is

$ h^2 = \dfrac{V_A}{V_A + V_E} $

The slope of offspring on mid-parent regression estimates $h^2$.

The expected response to selection in a single generation is

$ R = h^2 \cdot S $

where $S = \bar{P}_{\text{sel}} - \bar{P}$ is the selection differential.

For traits with $h^2 \approx 0.6$ (e.g. flower size in chrysanthemum), selection of the top 10% can advance the population mean by almost one phenotypic standard deviation per generation.

Modern Rose Hybridization Pedigree

3. Vase Life Biochemistry

Cut flowers decline through three interacting processes: ethylene-triggered senescence, water-balance collapse, and carbohydrate depletion. Post-harvest management targets each of these independently.

Ethylene and the climacteric

In ethylene-sensitive species (carnation, Hawaiian orchid, rose, sweet pea) petal senescence begins with a burst of ethylene synthesized by ACC synthase (ACS) and ACC oxidase (ACO):

$ \text{SAM} \xrightarrow{\text{ACS}} \text{ACC} \xrightarrow{\text{ACO}} \text{C}_2\text{H}_4 + \text{HCN} + \text{CO}_2 $

SAM = S-adenosyl methionine, ACC = 1-aminocyclopropane-1-carboxylate.

Ethylene autocatalytically upregulates ACS/ACO transcription, so the process is explosive. STS (silver thiosulphate) and 1-MCP (1-methylcyclopropene) block the ethylene receptor ETR1 and can double vase life.

Water relations

Xylem vessels plug with air, mucilage, or bacterial cells in the first hours after cutting. Vase solutions therefore contain acidifiers (to dissolve gas bubbles and inhibit bacteria) and biocides (quaternary ammonium, 8-hydroxyquinoline). The turgor equation for a petal,

$ \Psi_P = \Psi_W - \Psi_\pi $

Turgor pressure = water potential minus osmotic potential; sucrose lowers $\Psi_\pi$ and pulls water in.

Carbohydrate reserve

Petal respiration consumes ~1–2% of petal dry mass per day. A vase solution containing 1–4% sucrose provides external substrate, delaying starvation-induced wilting and improving colour retention.

4. Postharvest Physiology & Logistics

Modern cold chains aim for continuous 2–4 °C storage with >90% relative humidity. Every hour outside the chain halves the shelf life of cut roses. Ethylene contamination from ripening fruit is a common failure mode, so mixed-load shipping of flowers and ethylene-producing fruit (apples, bananas) is avoided.

Pulsing: 20% sucrose + STS for 4 hours before packing.
Hydration: rehydration solutions with citric acid (pH 3.5) and surfactants.
Wet vs dry shipping: roses and lilies ship dry; Gerbera, Anthurium wet.
Modified atmospheres: low-O₂/high-CO₂ bags reduce respiration.
Irradiation & fumigation: phytosanitary treatments at import.

Ethylene Senescence Cascade

Simulation: Polyploid Genome Doubling

Genome size scaling with ploidy, the cubic-root relationship between cell radius and ploidy, the gigas effect on flower diameter, and the sterility cost of odd ploidy.

Python

script.py78 lines

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

# Polyploid genome doubling and its phenotypic consequences.
# Colchicine inhibits tubulin polymerisation -> blocks anaphase ->
# produces tetraploid cells with 4n chromosome content.
# Phenotypic "gigas effect": polyploids typically have larger cells,
# thicker leaves, larger flowers, longer vase life.

np.random.seed(4)

# --- Panel A: genome size multiplication ---
ploidy = np.array([1, 2, 3, 4, 6, 8])
genome_size = 1.5 * ploidy          # pg DNA / nucleus for base genome 1.5 pg
names = ['1x\n(haploid)', '2x\n(diploid)', '3x\n(triploid)',
         '4x\n(tetraploid)', '6x\n(hexaploid)', '8x\n(octoploid)']

# --- Panel B: cell size scales as ploidy^(1/3) (volume scales linearly with genome) ---
cell_volume = ploidy * 100           # arbitrary units
cell_radius = (cell_volume / (4 / 3 * np.pi))**(1 / 3)

# --- Panel C: flower diameter distribution for diploid vs tetraploid ---
diploid = np.random.normal(5.0, 0.6, 3000)
tetraploid = np.random.normal(7.2, 0.9, 3000)
hexaploid = np.random.normal(8.8, 1.1, 3000)

# --- Panel D: fertility vs ploidy (odd ploidies are typically sterile) ---
ploidies = np.arange(1, 9)
fertility = np.where(ploidies % 2 == 0, np.array([0.25, 0.95, 0.55, 0.85, 0.40, 0.70, 0.30, 0.55]), 0.05)

fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(13, 10))
fig.patch.set_facecolor('#0a0a1a')
for ax in (ax1, ax2, ax3, ax4):
    ax.set_facecolor('#0a0a1a')
    ax.tick_params(colors='#94a3b8')
    for sp in ax.spines.values():
        sp.set_color('#334155')
    ax.grid(True, color='#334155', alpha=0.35)

ax1.bar(names, genome_size, color='#f472b6', edgecolor='white', linewidth=0.5)
ax1.set_ylabel('Genome size (pg)', color='#cbd5e1')
ax1.set_title('A. Genome size scales linearly with ploidy',
              color='#f9a8d4', fontweight='bold')

ax2.plot(ploidy, cell_radius, 'o-', color='#fb7185', markersize=9, linewidth=2)
ax2.set_xlabel('Ploidy (x)', color='#cbd5e1')
ax2.set_ylabel('Relative cell radius (r ~ p^{1/3})', color='#cbd5e1')
ax2.set_title('B. Cell radius scales as ploidy^(1/3)',
              color='#f9a8d4', fontweight='bold')

ax3.hist(diploid, bins=50, color='#fb7185', alpha=0.55, label='diploid',
         edgecolor='white', linewidth=0.3)
ax3.hist(tetraploid, bins=50, color='#f472b6', alpha=0.55, label='tetraploid',
         edgecolor='white', linewidth=0.3)
ax3.hist(hexaploid, bins=50, color='#fde68a', alpha=0.55, label='hexaploid',
         edgecolor='white', linewidth=0.3)
ax3.set_xlabel('Flower diameter (cm)', color='#cbd5e1')
ax3.set_ylabel('Count', color='#cbd5e1')
ax3.set_title('C. Gigas effect on flower size',
              color='#f9a8d4', fontweight='bold')
ax3.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

ax4.bar(ploidies, fertility, color=np.where(ploidies % 2 == 0,
         '#f472b6', '#fb7185'), edgecolor='white', linewidth=0.5)
ax4.set_xlabel('Ploidy level (x)', color='#cbd5e1')
ax4.set_ylabel('Seed fertility (fraction)', color='#cbd5e1')
ax4.set_title('D. Odd ploidies are sterile (triploid rose, banana...)',
              color='#f9a8d4', fontweight='bold')

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#0a0a1a')
print('Genome doubling produces larger cells, flowers, and often thicker leaves.')
print(f'Colchicine converts 2x -> 4x at ~1% efficiency (24h 0.05% treatment).')
print(f'Famous polyploid ornamentals: strawberry (8x), chrysanthemum (6x),')
print(f'sugarcane (12x), Phalaenopsis orchid (2x, 3x, 4x cultivars).')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Simulation: Vase Life Model

Coupled ethylene climacteric, hydraulic conductance decay, and petal carbohydrate dynamics, with STS, 1-MCP, and sucrose interventions.

Python

script.py104 lines

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

# Vase life model: ethylene, water, and carbohydrate dynamics.
# Petal senescence = climacteric burst of ethylene triggered by ACC synthase
# and ACC oxidase. 1-MCP or STS block ethylene perception and extend life.
# Water relations: stem hydraulic conductance declines as xylem blocks.

np.random.seed(8)

t = np.linspace(0, 14, 300)  # days

# 1. Untreated flower: exponential decline in petal turgor + ethylene burst
def model(t, sts=False, mcp=False, water_change=True, sugar=2.0):
    # Ethylene production
    E = np.zeros_like(t)
    for i, ti in enumerate(t):
        ACC = 0.2 * (1 + 0.3 * np.sin(ti))
        ACC_ox_activity = 1.0 if not mcp else 0.15
        burst = 0.3 * np.exp((ti - 6) / 1.5) if ti > 4 else 0.02
        if sts:
            burst *= 0.2
        E[i] = ACC * ACC_ox_activity + burst
    # Water (turgor) decline
    hydraulic_k = 1.0 - 0.12 * t
    if water_change:
        hydraulic_k = np.maximum(hydraulic_k + 0.05 * np.round(t / 2),
                                 1.0 - 0.03 * t)
    # Sugar reserve
    S = np.maximum(sugar - 0.2 * t, 0)
    # Petal quality index (0-1)
    Q = np.clip(hydraulic_k * np.exp(-0.4 * E) * (0.5 + 0.3 * S / 2), 0, 1)
    return E, hydraulic_k, S, Q

E_c, k_c, S_c, Q_c = model(t)
E_s, k_s, S_s, Q_s = model(t, sts=True)
E_m, k_m, S_m, Q_m = model(t, mcp=True)
E_f, k_f, S_f, Q_f = model(t, sts=True, sugar=4.0)

# Vase life = time until Q < 0.3
def vlife(t, Q):
    idx = np.where(Q < 0.3)[0]
    return t[idx[0]] if len(idx) > 0 else t[-1]

lives = {
    'control': vlife(t, Q_c),
    'STS': vlife(t, Q_s),
    '1-MCP': vlife(t, Q_m),
    'STS + sucrose': vlife(t, Q_f),
}

ax1.plot(t, E_c, color='#fb7185', linewidth=2, label='control')
ax1.plot(t, E_s, color='#f472b6', linewidth=2, label='+STS')
ax1.plot(t, E_m, color='#fde68a', linewidth=2, label='+1-MCP')
ax1.set_xlabel('Days in vase', color='#cbd5e1')
ax1.set_ylabel('Ethylene production (relative)', color='#cbd5e1')
ax1.set_title('A. Ethylene climacteric',
              color='#f9a8d4', fontweight='bold')
ax1.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

ax2.plot(t, k_c, color='#fb7185', linewidth=2, label='control')
ax2.plot(t, k_s, color='#f472b6', linewidth=2, label='+STS')
ax2.set_xlabel('Days in vase', color='#cbd5e1')
ax2.set_ylabel('Xylem hydraulic conductance (rel.)', color='#cbd5e1')
ax2.set_title('B. Water relations', color='#f9a8d4', fontweight='bold')
ax2.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

ax3.plot(t, S_c, color='#fb7185', linewidth=2, label='no sucrose')
ax3.plot(t, S_f, color='#f472b6', linewidth=2, label='+2% sucrose')
ax3.set_xlabel('Days in vase', color='#cbd5e1')
ax3.set_ylabel('Petal sugar reserve (rel.)', color='#cbd5e1')
ax3.set_title('C. Carbohydrate dynamics',
              color='#f9a8d4', fontweight='bold')
ax3.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

ax4.plot(t, Q_c, color='#fb7185', linewidth=2, label='control')
ax4.plot(t, Q_s, color='#f472b6', linewidth=2, label='STS')
ax4.plot(t, Q_m, color='#fde68a', linewidth=2, label='1-MCP')
ax4.plot(t, Q_f, color='#a7f3d0', linewidth=2, label='STS + sucrose')
ax4.axhline(0.3, color='#fde68a', linestyle='--', alpha=0.6, label='saleable threshold')
ax4.set_xlabel('Days in vase', color='#cbd5e1')
ax4.set_ylabel('Petal quality Q', color='#cbd5e1')
ax4.set_title('D. Vase life prediction',
              color='#f9a8d4', fontweight='bold')
ax4.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#0a0a1a')
for name, l in lives.items():
    print(f'  vase life with {name}: {l:.1f} days')
print('STS (silver thiosulphate) blocks ethylene receptors; 1-MCP is a safer')
print('reversible analogue. Carnation vase life can be doubled (6 -> 12 days).')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Simulation: Heritability of Flower Colour

Additive genetic model across 8 anthocyanin loci, parent-offspring regression estimate of $h^2$, breeder’s-equation response to selection across 20 generations, and sensitivity of $h^2$ to environmental variance.

Python

script.py101 lines

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

# Heritability of flower colour and response to selection.
# Colour score = sum of minor allele effects at multiple anthocyanin loci
# (DFR, ANS, F3'H, F3'5'H, UFGT) + environmental noise.
# Breeder's equation:  R = h^2 * S

np.random.seed(2)

n_loci = 8
n_indiv = 2000
effects = np.random.normal(0, 1, n_loci)
genotypes = np.random.binomial(2, 0.3, (n_indiv, n_loci))
G = genotypes @ effects                # additive genetic value
# Environmental noise
sigma_E = np.std(G) * 0.6
E = np.random.normal(0, sigma_E, n_indiv)
P = G + E
h2 = np.var(G) / np.var(P)

# Selection: top 10% selected
threshold = np.quantile(P, 0.9)
selected = P >= threshold
S = np.mean(P[selected]) - np.mean(P)
R_pred = h2 * S
# Simulate next generation mean
offspring_G = np.random.choice(G[selected], n_indiv) + np.random.normal(0, np.std(G) / np.sqrt(2), n_indiv)
offspring_P = offspring_G + np.random.normal(0, sigma_E, n_indiv)
R_obs = np.mean(offspring_P) - np.mean(P)

# Response across generations (breeder's equation iteration)
n_gens = 20
means = [np.mean(P)]
current = P.copy()
for _ in range(n_gens):
    sel = current >= np.quantile(current, 0.9)
    g_sel = current[sel]
    # h^2 decays slowly as variance shrinks
    new = np.random.choice(g_sel, n_indiv) + np.random.normal(0, np.std(g_sel), n_indiv)
    current = new
    means.append(np.mean(current))

# h^2 vs environmental noise
noise_levels = np.linspace(0.2, 2.0, 50) * np.std(G)
h2_curve = np.var(G) / (np.var(G) + noise_levels**2)

ax1.hist(P, bins=50, color='#f472b6', alpha=0.7, edgecolor='white', linewidth=0.3)
ax1.axvline(threshold, color='#fde68a', linestyle='--', linewidth=2,
            label=f'selection threshold (top 10%)')
ax1.set_xlabel('Phenotype (colour intensity)', color='#cbd5e1')
ax1.set_ylabel('Count', color='#cbd5e1')
ax1.set_title('A. Colour phenotype distribution',
              color='#f9a8d4', fontweight='bold')
ax1.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

# Parent vs offspring regression (slope = h^2)
parents_mean = (G + E) / 2
offspring = G / 2 + np.random.normal(0, sigma_E / np.sqrt(2), n_indiv)
ax2.scatter(parents_mean[:500], offspring[:500], s=10, alpha=0.6, color='#fb7185')
coeffs = np.polyfit(parents_mean, offspring, 1)
xs = np.linspace(parents_mean.min(), parents_mean.max(), 50)
ax2.plot(xs, coeffs[0] * xs + coeffs[1], color='#fde68a', linewidth=2,
         label=f'slope = {coeffs[0]:.2f} (~h^2)')
ax2.set_xlabel('Mid-parent phenotype', color='#cbd5e1')
ax2.set_ylabel('Offspring phenotype', color='#cbd5e1')
ax2.set_title('B. Parent-offspring regression',
              color='#f9a8d4', fontweight='bold')
ax2.legend(facecolor='#1e293b', edgecolor='#334155', labelcolor='#e2e8f0')

ax3.plot(range(len(means)), means, 'o-', color='#f472b6', linewidth=2, markersize=8)
ax3.set_xlabel('Generation', color='#cbd5e1')
ax3.set_ylabel('Mean colour phenotype', color='#cbd5e1')
ax3.set_title('C. Selection response over 20 generations',
              color='#f9a8d4', fontweight='bold')

ax4.plot(noise_levels / np.std(G), h2_curve, color='#fb7185', linewidth=2.5)
ax4.fill_between(noise_levels / np.std(G), h2_curve, alpha=0.2, color='#fb7185')
ax4.set_xlabel('Environmental noise (sigma_E / sigma_G)', color='#cbd5e1')
ax4.set_ylabel('Heritability h^2', color='#cbd5e1')
ax4.set_title('D. h^2 vs environmental variance',
              color='#f9a8d4', fontweight='bold')

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#0a0a1a')
print(f'Heritability h^2 = {h2:.3f}')
print(f'Selection differential S = {S:.3f}')
print(f'Predicted response R = h^2 * S = {R_pred:.3f}')
print(f'Observed response after one generation = {R_obs:.3f}')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Key References

• Goldgar, A. (2007). Tulipmania: Money, Honor, and Knowledge in the Dutch Golden Age. Univ. Chicago Press.

• Dekker, R. J. et al. (2001). “The tulip-breaking virus.” Plant Pathology, 50, 540–549.

• Woltering, E. J. & van Doorn, W. G. (1988). “Role of ethylene in senescence of petals.” J. Exp. Bot., 39, 1605–1616.

• Serek, M., Sisler, E. C. & Reid, M. S. (1995). “1-Methylcyclopropene: novel gaseous inhibitor of ethylene action.” HortScience, 30, 1310–1314.

• Katsumoto, Y. et al. (2007). “Engineering of the rose flavonoid biosynthetic pathway successfully generated blue-hued flowers accumulating delphinidin.” Plant Cell Physiol., 48, 1589–1600.

• van Doorn, W. G. (2012). “Water relations of cut flowers: an update.” Horticultural Reviews, 40, 55–106.

• Falconer, D. S. & Mackay, T. F. C. (1996). Introduction to Quantitative Genetics (4th ed.). Longman.

• Eeckhaut, T. et al. (2013). “Progress in plant protoplast research.” Planta, 238, 991–1003.

• Debener, T. & Linde, M. (2009). “Exploring complex ornamental genomes: the rose as a model plant.” Critical Reviews in Plant Sciences, 28, 267–280.

• Reid, M. S. & Jiang, C.-Z. (2012). “Postharvest biology and technology of cut flowers and potted plants.” Horticultural Reviews, 40, 1–54.

• Rout, G. R. & Mohapatra, A. (2006). “Use of molecular markers in ornamental plants: a critical reappraisal.” European Journal of Horticultural Science, 71, 53–68.

← Module 6: Phylogeny & Evolution Module 8: Flowers & Climate Change →