Flow Lab

Syllabus / Module 15: Applications: Biology, Chemistry & Society

Lesson 15.1

Population Models

Exponential, logistic and Gompertz growth; harvesting; Allee effects; fitting models to data.

Written byClaude Sonnet

Prerequisites:3.9 Bifurcations
🌊
Why this matters for simulation: Model fitting to data is the same workflow as calibrating a turbulence model.

By the end of this lesson you can…

  • Explain why exponential growth fails and use the logistic solution, its inflection at $K/2$ and its phase line.
  • Fit $r$, $K$ and $N_0$ by Gauss-Newton on the log residuals and read from the standard errors what the data determine.
  • Solve the Gompertz equation by the substitution $u = \ln(N/K)$, locate its inflection at $K/e$, and compare its fit with the logistic.
  • Compare constant-yield and constant-effort harvesting: the maximum sustainable yield $rK/4$ and the fold at $h = rK/4$.
  • Analyse the strong Allee effect: the threshold $A$, the fold under harvest, and why recovery needs restocking.
  • Connect the logistic map to Euler's method, locate period doubling and chaos numerically, and find the Hopf point $rT = \pi/2$ of Hutchinson's delay equation.
On this page
·
  1. Exponential growth and its failure
  2. Logistic growth, revisited
  3. Fitting a model to data
  4. Gompertz growth
  5. Harvesting and sustainability
  6. The Allee effect
  7. Discrete generations and Euler’s method
  8. Delay equations (a glimpse)
  9. Using the lab
  10. The exercise: Gauss-Newton for the logistic curve
  11. Concept checks
  12. Problems
  13. What comes next
  14. Key Takeaways

14.7 N-Body Simulation closed Module 14, where every model conserved something: energy, momentum, angular momentum. Module 15 turns to systems with nothing to conserve. A population gains by births and loses by deaths, and the books never balance in the mechanical sense. The consequence is the picture of 3.8 Autonomous Equations, Equilibria & Stability: trajectories from different starts converge on an attractor, a state the system forgets its history in. In one dimension an attractor is a fixed point, which is why this lesson leans on the phase line. Not every model in the module settles to a point, and the last two sections show that even one equation can leave “fixed points attract” behind.

▶ Video Exponential growth and why it fails, the logistic curve, fitting r and K to noisy data, Gompertz growth, harvesting and its tipping point, the Allee threshold, the logistic map and a delay equationcaptions only
Transcript
  1. Exponential growth, N prime equals r N, says a culture of 1000 cells with r equal to 0.7 per hour doubles every 0.990 hours, and after 48 hours the formula gives 3.9 times 10 to the 17 cells. No flask holds that. What fails is the constant per-capita rate: crowding makes it fall as N grows, and the simplest model lets it fall linearly.
  2. Letting the per-capita rate fall linearly gives the logistic equation, N prime equals r N times one minus N over K. With r equal to 0.8, K equal to 100 and N0 equal to 5, the solution is an S-curve whose inflection sits at K over 2, at t equal to 3.68, and it climbs from 10 percent to 90 percent of K in 5.49 time units. On the phase line, 0 is unstable and K is stable, so every positive start ends at K.
  3. Real data are noisy, so here the numbers are synthetic: seed 15, noise 0.08, truth N0 equal to 5, r 0.8, K 100. Fit the log of N by Gauss-Newton and 11 steps give r equal to 0.8291 plus or minus 0.0364 and K equal to 102.14 plus or minus 3.36. The truth lies within 1.5 standard errors. Now keep only the first 6 points, before the curve flattens: K is 106.5 plus or minus 9.9, and with 5 points 100.2 plus or minus 20.1. The steep part pins r; K is known only once the data approach it.
  4. Gompertz growth drops the per-capita rate linearly in the log of N. With u equal to the log of N over K, u prime is minus r u, so it solves in closed form, and its inflection is at K over e, about 37 percent of K against 50 percent for the logistic. On the same 13 points the logistic has rms residual 0.0696 and K equal to 102.1, the Gompertz 0.1049 and K equal to 116.5 plus or minus 8.1. On the first 8 points Gompertz gives K equal to 163 plus or minus 35. When the plateau is unobserved, model choice moves K by a large factor.
  5. Harvest a logistic stock at a constant yield h. The two equilibria are 65.8 and 34.2 at h equal to 18, 61.2 and 38.8 at 19, with recovery time 5.59, and they merge at h equal to 20, the maximum sustainable yield r K over 4, reached at N equal to 50. At h equal to 19, only 5 percent under it, a die-off of 40 percent puts the stock at 36.7, below the threshold of 38.8, and it is extinct at t equal to 11.1. At h equal to 21 the stock reaches zero at 30.2. Constant effort fails only above r.
  6. The strong Allee effect makes the per-capita rate negative below a threshold A, here 20. The system is bistable: extinction and K both attract. Add effort E and at 0.5 the equilibria are 41.3, unstable, and 78.7, stable. They merge at E equal to 0.64, and past it only zero remains. Ramp the effort up and back down: the stock collapses at 0.686 and does not return when E returns to 0. The logistic model under the same ramp collapses at 0.803 and recovers at 0.393. Recovery of an Allee stock means restocking above the threshold.
  7. Many species breed once a year, so the model is a map, x goes to mu x times one minus x, which is also Euler's method on the logistic equation. The fixed point loses stability at mu equal to 3 and the period doubles at 3.449490, 3.544090, 3.564407. The gaps shrink by 4.751, 4.656, 4.668, 4.669, toward the Feigenbaum constant, from memory. The Lyapunov exponent is minus 0.223 at 2.8, plus 0.50 at 3.9, which is chaos, and ln 2 at 4.
  8. A one-dimensional ODE cannot oscillate, but a population that responds to its past density can. In Hutchinson's equation, from memory, the delayed term gives the characteristic equation lambda equals minus r e to the minus lambda T. A pair of imaginary roots is a Hopf bifurcation, at r T equal to pi over 2, with period 4T. With r equal to 1 the real part is minus 0.058 at T equal to 1.4, so the oscillation decays, and plus 0.033 at 1.7: a steady cycle from 0.334 K to 1.943 K, period 6.96. At r T equal to 2.5 the swing runs from 0.0016 K to 4.68 K.
  9. One species is a one-dimensional problem, and even so we met folds, a threshold with hysteresis, and oscillation once generations are discrete or responses delayed. Next, predator and prey couple two species. The Lotka-Volterra equations turn out to have a conserved quantity and closed orbits, a reminder of Module 14 inside a biological model. Holling functional responses then lead to limit cycles and the paradox of enrichment, and competing species to competitive exclusion.

Exponential growth and its failure

If each individual contributes the same births and deaths, N′=rNN' = rN and N=N0ertN = N_0e^{rt}. A culture with N0=1000N_0 = 1000 cells and r=0.7r = 0.7 per hour doubles every ln⁡2/r=0.990\ln 2/r = 0.990 hours, and after 4848 hours the formula says 1000 e33.6=3.9×10171000\,e^{33.6} = 3.9\times10^{17} cells. No flask holds that. What fails is the constant per-capita rate N′/NN'/N: crowding makes it fall as NN grows, and the simplest model lets it fall linearly.

Logistic growth, revisited

Put N′/N=r(1−N/K)N'/N = r(1 - N/K):

N′=rN(1−NK),N(t)=K1+(KN0−1)e−rt.N' = rN\left(1 - \frac{N}{K}\right), \qquad N(t) = \frac{K}{1 + \left(\dfrac{K}{N_0} - 1\right)e^{-rt}}.

3.8 Autonomous Equations, Equilibria & Stability derives the solution by separation. Here rr is the per-capita rate of a nearly empty habitat and KK the level where births balance deaths. The growth rate f(N)=rN(1−N/K)f(N) = rN(1 - N/K) is a parabola, so N′=0N' = 0 at 00 (unstable, f′(0)=rf'(0) = r) and at KK (stable, f′(K)=−rf'(K) = -r), and NN is fastest at K/2K/2, where the curve has its inflection. With r=0.8r = 0.8, K=100K = 100, N0=5N_0 = 5 the inflection is at t=ln⁡19/0.8=3.68t = \ln 19/0.8 = 3.68, and the time to climb from 10%10\% to 90%90\% of KK is 2ln⁡9/r=5.492\ln 9/r = 5.49.

The phase line replaces the energy surface of Module 14. Arrows point up on (0,K)(0, K) and down above KK, so every positive start ends at KK. The slope of ff at an equilibrium is the rate of approach, here 1/r1/r as a time scale.

◆ Interactive Logistic model: click the left panel to move the start, change r and K, and watch the phase line and the yield curve

Loading simulation…

Fitting a model to data

A model is useful once its constants are measured. To keep the numbers honest the lab uses synthetic data: numpy default_rng(15), true N0=5N_0 = 5, r=0.8r = 0.8, K=100K = 100, t=0,1,…,12t = 0,1,\dots,12, noise Ni=N(ti)e0.08ziN_i = N(t_i)e^{0.08z_i} with ziz_i standard normal, rounded to 0.10.1 (the values are in the exercise).

Fit ln⁡N\ln N, not NN: the noise is proportional to size, so the residuals ρi=ln⁡m(ti;θ)−ln⁡yi\rho_i = \ln m(t_i;\theta) - \ln y_i have equal spread, whereas residuals in NN would let the big late points dominate. Minimise ∑ρi2\sum\rho_i^2 over θ=(N0,r,K)\theta = (N_0, r, K). Linearise about the current guess, ρ(θ+δ)≈ρ+Jδ\rho(\theta + \delta) \approx \rho + J\delta with Jij=∂ρi/∂θjJ_{ij} = \partial\rho_i/\partial\theta_j, and set the gradient of ∣ρ+Jδ∣2|\rho + J\delta|^2 to zero:

JTJ δ=−JTρ.J^{\mathsf T}J\,\delta = -J^{\mathsf T}\rho.

These normal equations answer the “tall” case of 1.3 Linear Systems, Kernel & Column Space: more data than unknowns, so take the closest solution. Update θ→θ+δ\theta \to \theta + \delta and repeat: Gauss-Newton. With E=e−rtE = e^{-rt}, c=K/N0−1c = K/N_0 - 1 and D=1+cED = 1 + cE, the Jacobian columns are

∂ln⁡m∂N0=KEN02D,∂ln⁡m∂r=ctED,∂ln⁡m∂K=1K−EN0D.\frac{\partial\ln m}{\partial N_0} = \frac{KE}{N_0^2D}, \qquad \frac{\partial\ln m}{\partial r} = \frac{ctE}{D}, \qquad \frac{\partial\ln m}{\partial K} = \frac1K - \frac{E}{N_0D}.

If a step raises the cost, halve it until it does not.

Result. From the start (2,0.3,60)(2, 0.3, 60) the iteration reaches the minimum in 1111 steps (the sim shows each one) at

N0=4.549±0.295,r=0.8291±0.0364,K=102.14±3.36,N_0 = 4.549 \pm 0.295, \qquad r = 0.8291 \pm 0.0364, \qquad K = 102.14 \pm 3.36,

with an rms residual of 0.06960.0696. The ±\pm are standard errors diag⁡ s2(JTJ)−1\sqrt{\operatorname{diag}\,s^2(J^{\mathsf T}J)^{-1}}, s2=SSR/(n−3)s^2 = \text{SSR}/(n-3). A separate scipy least_squares run agrees to five digits, and 20002000 fresh data sets (seed 11) scatter in rr and KK with standard deviations 0.0360.036 and 3.33.3, so the formula is honest. The true values (5,0.8,100)(5, 0.8, 100) lie within about 1.51.5 standard errors.

What the data can and cannot say. Keep only the first 66 points (t≤5t \le 5, before the curve flattens): K=106.5±9.9K = 106.5 \pm 9.9. With 55 points, K=100.2±20.1K = 100.2 \pm 20.1, right by luck and uncertain by a fifth. The 22-standard-deviation band of the fitted curve at t=16t = 16 is ±7%\pm 7\% for all 1313 points and [88.5,128.3][88.5, 128.3] for 66. The steep part pins rr; KK is known only once the data approach it.

A tempting shortcut regresses the per-capita rate (yi+1−yi−1)/2yi(y_{i+1} - y_{i-1})/2y_i on yiy_i (intercept rr, slope −r/K-r/K). Here it gives r=0.829r = 0.829, K=102.7K = 102.7, but over the 20002000 repeats its scatter in rr is 0.0480.048 against 0.0360.036.

◆ Interactive Fitting view: choose the model, use only the first n points, step through the Gauss-Newton iterations and read the standard errors

Loading simulation…

Gompertz growth

Replace the linear drop in per-capita rate with a drop linear in ln⁡N\ln N: N′=rNln⁡(K/N)N' = rN\ln(K/N). Put u=ln⁡(N/K)u = \ln(N/K). Then u′=N′/N=−ruu' = N'/N = -ru, so u=u0e−rtu = u_0e^{-rt} and

N(t)=Kexp⁡ ⁣(ln⁡N0K e−rt).N(t) = K\exp\!\left(\ln\frac{N_0}{K}\,e^{-rt}\right).

Differentiating N′=rNln⁡(K/N)N' = rN\ln(K/N) gives N′′=rN′(ln⁡(K/N)−1)N'' = rN'\left(\ln(K/N) - 1\right), so the inflection is where ln⁡(K/N)=1\ln(K/N) = 1: at N=K/eN = K/e, about 37%37\% of KK against the logistic 50%50\%. The peak growth rate is rK/e=0.368 rKrK/e = 0.368\,rK. The curve is asymmetric where the logistic is symmetric about its inflection. Since (1−xθ)/θ→−ln⁡x(1 - x^\theta)/\theta \to -\ln x as θ→0\theta\to0, the theta-logistic N′=rN (1−(N/K)θ)/θN' = rN\,(1 - (N/K)^\theta)/\theta tends to Gompertz (θ=1\theta = 1 is logistic). The per-capita rate rln⁡(K/N)r\ln(K/N) is unbounded as N→0N\to0.

Against the data. On all thirteen points the logistic has rms residual 0.06960.0696 and K=102.1±3.4K = 102.1\pm3.4, the Gompertz 0.10490.1049 and K=116.5±8.1K = 116.5\pm8.1; the curves differ by at most 9.69.6 (at t=12t = 12). On the first 88 points Gompertz gives K=163±35K = 163\pm35, and on the first 55 it fits fractionally better (0.03280.0328 against 0.03420.0342) with K=737±675K = 737\pm675. Whenever the plateau is unobserved, model choice moves KK by a large factor.

Harvesting and sustainability

Harvest a logistic stock with r=0.8r = 0.8, K=100K = 100. 3.8 Autonomous Equations, Equilibria & Stability analysed the constant yield N′=f(N)−hN' = f(N) - h: the equilibria are K2(1±1−4h/rK)\frac K2\big(1 \pm \sqrt{1 - 4h/rK}\big) and they merge at h=rK/4=20h = rK/4 = 20, the peak of ff, the maximum sustainable yield (MSY), reached at N=K/2=50N = K/2 = 50. The new point is what happens near that edge.

hhstableunstablerecovery time 1/∣f′∣1/\lvert f'\rvert
181865.865.834.234.23.953.95
191961.261.238.838.85.595.59
20205050 (semi-stable)5050no linear rate

At h=19h = 19, only 5%5\% under the MSY, a one-off die-off of 40%40\% from 61.261.2 puts the stock at 36.736.7, below the unstable equilibrium 38.838.8, and it is extinct at t=11.1t = 11.1 (from 3636 it is 9.79.7). At h=21h = 21 the stock starting at KK reaches zero at t=30.2t = 30.2 (from 5050: 15.115.1). This is the fold of 3.9 Bifurcations: slow recovery near the edge is critical slowing down, and above it there is nothing to settle on.

Constant effort removes a fraction: N′=rN(1−N/K)−ENN' = rN(1 - N/K) - EN. Now the per-capita rate is r−E−rN/Kr - E - rN/K, a logistic with smaller growth rate, so N∗=K(1−E/r)N^* = K(1 - E/r) and the yield is Y(E)=EN∗=EK(1−E/r)Y(E) = EN^* = EK(1 - E/r), a parabola with the same maximum rK/4rK/4 at E=r/2E = r/2, N∗=K/2N^* = K/2. But the equilibrium moves smoothly with EE and is approached at rate r−Er - E: N∗=50N^* = 50, Y=20Y = 20 at E=0.4E = 0.4 (rate 0.40.4), and N∗=25N^* = 25, Y=15Y = 15 at E=0.6E = 0.6. More effort catches less. Only at E>r=0.8E \gt r = 0.8 does the stock die out, and that is a transcritical, not a fold, crossing. Catch per unit effort is proportional to NN, a warning a fixed quota lacks.

In the sim’s right panel the curve (E,Y)=(p(N),Np(N))(E, Y) = (p(N), Np(N)), pp the per-capita rate, is the equilibrium yield of effort EE, solid where pp decreases (stable). For Gompertz, N∗=Ke−E/rN^* = Ke^{-E/r} never reaches zero, and the MSY is rK/e=29.43rK/e = 29.43 at E=rE = r, N=K/e=36.8N = K/e = 36.8.

The Allee effect

Some species suffer at low density (mates are hard to find). The strong Allee effect makes the per-capita rate negative below a threshold AA:

N′=rN(NA−1)(1−NK).N' = rN\left(\frac NA - 1\right)\left(1 - \frac NK\right).

With r=0.8r = 0.8, K=100K = 100, A=20A = 20 there are three equilibria. f′(0)=−r=−0.8f'(0) = -r = -0.8 (stable: extinction attracts), f′(A)=r(1−A/K)=+0.64f'(A) = r(1 - A/K) = +0.64 (unstable, the threshold) and f′(K)=−r(K/A−1)=−3.2f'(K) = -r(K/A - 1) = -3.2 (stable). The system is bistable, and the threshold AA is a separatrix on the phase line. A start at 1919 is at 0.250.25 by t=10t = 10 and 8.6×10−58.6\times10^{-5} by t=20t = 20; a start at 2121 is at 73.873.8 at t=5t = 5 and at KK to six digits by t=10t = 10. Two units decide the fate. A weak Allee effect, say p=r(1+N/B)(1−N/K)p = r(1 + N/B)(1 - N/K), has no threshold: the per-capita rate is still positive but rises with NN at low density (for B=20B = 20 it peaks at 1.441.44 near N=40N = 40).

Harvesting a strong Allee population. With effort EE, N′=N(p(N)−E)N' = N(p(N) - E) and p(N)=r(N/A−1)(1−N/K)p(N) = r(N/A - 1)(1 - N/K) peaks at N=(A+K)/2=60N = (A + K)/2 = 60 with value pmax⁡=r(K−A)2/(4AK)=0.64p_{\max} = r(K - A)^2/(4AK) = 0.64. For E<0.64E \lt 0.64 there are two positive equilibria: at E=0.5E = 0.5 they are 41.341.3 (unstable) and 78.778.7 (stable). At E=0.64E = 0.64 they merge and for larger EE only 00 remains. The yield on the stable branch peaks at 42.042.0 (at N=70.6N = 70.6, E=0.595E = 0.595), twice the logistic MSY, but that is only 7%7\% of EE below the collapse effort 0.640.64.

Hysteresis. Ramp EE from 00 slowly up to 0.8320.832 and back over 600600 time units. The stock drops below 5%5\% of KK at E=0.686E = 0.686 (just above the fold, a small lag) and does not come back as EE returns to 00, because 00 is stable for every EE. The logistic model under the same ramp collapses at E=0.803E = 0.803 (just past its own r=0.8r = 0.8, where the fold-free transcritical point sits) and recovers above K/2K/2 at E=0.393E = 0.393 on the way back down, a delay but not a permanent loss. Recovery of an Allee stock means restocking above the threshold: above 30.830.8 at E=0.3E = 0.3, and above 2020 with no harvesting. This is the hysteresis of 3.9 Bifurcations: turning the knob back does not undo the jump.

◆ Interactive Strong Allee effect with constant-effort harvest: press the sweep button and watch the yield fall to zero and stay there

Loading simulation…

Discrete generations and Euler’s method

Many species breed once a year, so the natural model is a map, xn+1=μxn(1−xn)x_{n+1} = \mu x_n(1 - x_n) with xx the population as a fraction of its maximum. The same map appeared in 3.8 Autonomous Equations, Equilibria & Stability: Euler’s method on the logistic ODE with step hh is exactly this map with μ=1+hr\mu = 1 + hr. For an ODE the oscillation is an artifact of the method; for a population with discrete generations it is real.

The fixed point x∗=1−1/μx^* = 1 - 1/\mu has multiplier 2−μ2 - \mu, so it is stable for 1<μ<31 \lt \mu \lt 3. A 2-cycle takes over at μ=3\mu = 3 and loses stability at 1+6=3.44951 + \sqrt6 = 3.4495. Solving numerically for the parameter where each 2k2^k-cycle has multiplier −1-1 (Newton on (x,μ)(x, \mu), and independently with scipy fsolve) gives the period-doubling points

μ=3,  3.449490,  3.544090,  3.564407,  3.568759,  3.569692,\mu = 3,\; 3.449490,\; 3.544090,\; 3.564407,\; 3.568759,\; 3.569692,

with ratios of successive gaps 4.751, 4.656, 4.668, 4.6694.751,\ 4.656,\ 4.668,\ 4.669, approaching the Feigenbaum constant 4.6692…4.6692\ldots (the digits from memory). The doublings pile up near μ∞≈3.56995\mu_\infty \approx 3.56995 (extrapolating with δ≈4.669\delta \approx 4.669; from memory 3.56994563.5699456). Beyond it the Lyapunov exponent λ\lambda, the mean of ln⁡∣μ(1−2x)∣\ln|\mu(1-2x)| along an orbit, can be positive, which means chaos.

μ\muattractorλ\lambda
2.82.8fixed point−0.223-0.223
3.53.54-cycle−0.87-0.87
3.573.57chaotic+0.01+0.01
3.833.833-cycle (window opens at 1+8=3.82841 + \sqrt8 = 3.8284, checked)−0.37-0.37
3.93.9chaotic+0.50+0.50
44chaoticln⁡2=0.693\ln2 = 0.693

For Euler on the logistic ODE chaos starts at hr≈2.57hr \approx 2.57, i.e. h≈3.2h \approx 3.2 for r=0.8r = 0.8. (See 4.5 Stability Regions & Stiffness for stepping limits.)

◆ Interactive The logistic map: cobweb, bifurcation diagram and Lyapunov exponent; drag mu through the doublings, into chaos and the period-3 window

Loading simulation…

Delay equations (a glimpse)

A one-dimensional ODE cannot oscillate, but real populations respond to past density: maturation takes time. Hutchinson’s equation (from memory, 1948) is

N′(t)=rN(t)(1−N(t−T)K).N'(t) = rN(t)\left(1 - \frac{N(t - T)}{K}\right).

The state is now the whole history on [t−T,t][t - T, t]. KK is still an equilibrium; with N=K+uN = K + u, u′≈−r u(t−T)u' \approx -r\,u(t - T). Try u=eλtu = e^{\lambda t}:

λ=−r e−λT.\lambda = -r\,e^{-\lambda T}.

A pair of imaginary roots λ=±iω\lambda = \pm i\omega marks a Hopf bifurcation. The real part of iω=−r(cos⁡ωT−isin⁡ωT)i\omega = -r(\cos\omega T - i\sin\omega T) gives cos⁡ωT=0\cos\omega T = 0, so ωT=π/2\omega T = \pi/2; the imaginary part gives ω=rsin⁡ωT=r\omega = r\sin\omega T = r. The stability limit is

rT=π2,rT = \frac{\pi}{2},

with period 2π/ω=4T2\pi/\omega = 4T at onset. For rT<1/e=0.368rT \lt 1/e = 0.368 the roots are real (they merge where the line is tangent to the exponential) and the return to KK is monotone; between 1/e1/e and π/2\pi/2 it is a damped oscillation. With r=1r = 1, the real part of the leading root is −0.058-0.058 at T=1.4T = 1.4 and +0.033+0.033 at 1.71.7 (roots from the Lambert W function, scipy).

The sim integrates the equation by RK4 with a cubic-Hermite history. A separate method-of-steps solution in scipy (solve_ivp, tolerance 10−1110^{-11}) agrees to better than 10−710^{-7} in the amplitudes below and 10−410^{-4} in the periods. At rT=1.4rT = 1.4 the solution stays within 4×10−84\times10^{-8} of KK from t=280t = 280 to 400400. At rT=1.7rT = 1.7 it settles into a steady cycle between 0.334K0.334K and 1.943K1.943K, period 6.96=4.10 T6.96 = 4.10\,T. At 1.81.8: 0.197K0.197K to 2.299K2.299K, period 7.53=4.18 T7.53 = 4.18\,T. At 2.52.5 the cycle swings from 0.0016K0.0016K to 4.68K4.68K with period 13.4=5.36 T13.4 = 5.36\,T, so the model population falls to 0.2%0.2\% of KK every cycle.

◆ Interactive Hutchinson's delay equation: move r T across pi/2 and watch the decaying oscillation become a limit cycle; the right panel is the linear growth rate

Loading simulation…

Using the lab

  1. Logistic: the dotted line is at K/2K/2; select Gompertz and it moves to K/eK/e.
  2. Constant yield: slide hh from 1818 to 2121 and watch the equilibria merge.
  3. Strong Allee: start at 1919 and 2121, then add effort and press the sweep button.
  4. Fit: use 55 points, then 1313, then Gompertz. Map: jump to 3.833.83, then 3.93.9. Delay: cross π/2\pi/2.

The exercise: Gauss-Newton for the logistic curve

The playground runs the harvested stock with RK4 and the Hutchinson equation with Euler on a grid that holds t−Tt - T exactly. In the exercise you write the model, its Jacobian and the damped Gauss-Newton loop; the tests check the numbers above.

🐍 PythonConstant yield near the edge and a delay equation
⌘/Ctrl + Enter
🧪 Coding exerciseExercise: fit the logistic curve by Gauss-Newton
⌘/Ctrl + Enter

Concept checks

✅ Concept check

A logistic population has $r = 0.5$ per year. How many years does it take to climb from $10\%$ to $90\%$ of $K$?

✅ Concept check

A fishery holds a logistic stock at a constant yield $h$ just above $rK/4$. What happens?

✅ Concept check

In the strong Allee model with $A = 20$, $K = 100$, what happens to a population started at $N_0 = 19$?

✅ Concept checkselect all that apply

Select all true statements about the Gompertz equation $N' = rN\ln(K/N)$.

✅ Concept check

For $N' = rN(1 - N(t - T)/K)$ with $r = 0.5$, at which delay $T$ does $N = K$ lose stability?

Problems

✏️ Pen & paperGompertz growth rate●○○

For N′=rNln⁡(K/N)N' = rN\ln(K/N) with r=0.8r = 0.8 and K=100K = 100, find the largest growth rate and where it occurs. Compare it with the logistic with the same rr and KK.

Show worked solution

f(N)=rNln⁡(K/N)f(N) = rN\ln(K/N) has f′=rln⁡(K/N)−r=0f' = r\ln(K/N) - r = 0 at N=K/e=36.79N = K/e = 36.79, where f=rK/e=80/e=29.43f = rK/e = 80/e = 29.43. The logistic peaks at rK/4=20rK/4 = 20 at N=50N = 50. At the peak the per-capita rate is rln⁡e=0.8r\ln e = 0.8 for Gompertz and r/2=0.4r/2 = 0.4 for the logistic, so the Gompertz peak is higher (36.79×0.836.79\times0.8 against 50×0.450\times0.4), though the two rr‘s do not mean the same thing.

✏️ Pen & paperConstant effort●●○

A logistic stock is harvested at constant effort E<rE \lt r. Find the equilibrium, the yield Y(E)Y(E), the effort that maximises it, and the rate at which the stock returns after a small disturbance. Evaluate for r=0.8r = 0.8, K=100K = 100, E=0.6E = 0.6.

Show worked solution

N′=N(r−E−rN/K)N' = N(r - E - rN/K) has the stable equilibrium N∗=K(1−E/r)N^* = K(1 - E/r). The yield is Y=EN∗=EK(1−E/r)Y = EN^* = EK(1 - E/r), with Y′=K(1−2E/r)=0Y' = K(1 - 2E/r) = 0 at E=r/2E = r/2, so Ymax⁡=rK/4Y_{\max} = rK/4 at N∗=K/2N^* = K/2. The slope of the right-hand side at N∗N^* is −(r−E)-(r - E), so a disturbance decays like e−(r−E)te^{-(r-E)t}. For E=0.6E = 0.6: N∗=25N^* = 25, Y=15Y = 15, rate 0.20.2 (time constant 55).

✏️ Pen & paperThe delay at onset●●●

For Hutchinson’s equation with r=0.5r = 0.5 find the critical delay TcT_c and the period of the oscillation at onset, and check the characteristic equation at λ=iω\lambda = i\omega.

Show worked solution

The characteristic equation is λ=−re−λT\lambda = -r e^{-\lambda T}. At onset ω=r=0.5\omega = r = 0.5 and ωTc=π/2\omega T_c = \pi/2, so Tc=πT_c = \pi. The check: −re−iπ/2=−0.5(cos⁡(π/2)−isin⁡(π/2))=0.5i=iω-r e^{-i\pi/2} = -0.5(\cos(\pi/2) - i\sin(\pi/2)) = 0.5i = i\omega. The period is 2π/ω=4π=12.57=4Tc2\pi/\omega = 4\pi = 12.57 = 4T_c.

What comes next

One species is a one-dimensional problem, and even so we met folds, a threshold with hysteresis, and oscillation once generations are discrete or responses delayed. 15.2 Predator–Prey & Competing Species couples two species. The Lotka-Volterra equations turn out to have a conserved quantity and closed orbits, a reminder of Module 14 inside a biological model; Holling functional responses then lead to the limit cycles and the paradox of enrichment of the Rosenzweig-MacArthur model, and competing species to competitive exclusion.

Key Takeaways

  • Exponential growth fails because the per-capita rate falls with crowding; logistic growth has its inflection at K/2K/2 and climbs from 10%10\% to 90%90\% of KK in 2ln⁡9/r2\ln9/r.
  • Fit ln⁡N\ln N by Gauss-Newton (JTJ δ=−JTρJ^{\mathsf T}J\,\delta = -J^{\mathsf T}\rho, step halving): r=0.829±0.036r = 0.829\pm0.036, K=102.1±3.4K = 102.1\pm3.4 from 1313 points, K=106.5±9.9K = 106.5\pm9.9 from 66.
  • Gompertz solves by u=ln⁡(N/K)u = \ln(N/K) and has its inflection at K/eK/e, peak rate rK/erK/e.
  • Constant yield and constant effort share the maximum rK/4rK/4, but yield has a fold at h=rK/4h = rK/4 (finite-time extinction above it) while effort fails only at E>rE \gt r.
  • A strong Allee effect gives bistability with threshold AA; harvesting folds at E=r(K−A)2/(4AK)E = r(K-A)^2/(4AK), and collapse does not reverse when effort returns.
  • Euler on the logistic ODE is the map μx(1−x)\mu x(1-x), μ=1+hr\mu = 1 + hr: doublings at 3,3.449,3.544,…3, 3.449, 3.544, \dots (ratio →4.669\to 4.669), chaos beyond 3.573.57.
  • A delay turns the logistic into an oscillator: the Hopf point of N′=rN(1−N(t−T)/K)N' = rN(1 - N(t-T)/K) is rT=π/2rT = \pi/2, with onset period 4T4T.
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.