A Logistic Modeling Problem-Solving Workshop

From a Population Story to a Logistic Model

A population model is useful only when its assumptions fit the situation well enough for the question being asked. In this workshop, the goal is to move from a population scenario to a prediction: identify the model, determine any unknown parameters, calculate the requested population, and interpret what the model says about long-term behavior and growth.

A logistic model is a reasonable choice when a population grows in a setting with limited resources. At low population levels, growth can be approximately proportional to population size. As the population approaches a limiting level, competition and other constraints reduce the per-capita growth rate.

For a population \(P(t)\), measured at time \(t\), the logistic differential equation is

where \(r>0\) is the intrinsic growth-rate parameter and \(K>0\) is the carrying capacity. If time is measured in years, \(r\) has units of inverse years, \(K\) has the same units as population, and \(dP/dt\) has units of population per year.

A Reliable Modeling Workflow

Read the scenario first, then decide what information it supplies. In particular, distinguish a population observation, such as \(P(3)=240\), from a rate observation, such as \(P'(0)=30\). They provide different equations and should not be substituted for one another.

Useful Logistic Relationships

For \(K>0\) and \(r>0\), the equilibrium populations satisfy \(dP/dt=0\), so the equation gives \(P=0\) and \(P=K\). For a positive population below \(K\), the population increases; above \(K\), it decreases. For an initial value \(P(0)=P_0\) with \(P_0>0\), the explicit solution is

The parameter \(r\) can be determined from an initial rate when \(P_0\) and \(K\) are known. Substituting \(t=0\) into the differential equation gives

If \(P_0>0\) and \(P_0\ne K\), this equation can be solved for \(r\). If \(P_0=0\) or \(P_0=K\), the initial rate is zero for every \(r\), so that rate observation does not determine \(r\).

If \(K\), \(P_0\), and a later observation \(P(t_1)=P_1\) are known, the explicit formula gives another useful parameter equation. Provided the observations are consistent with a non-equilibrium logistic trajectory, rearranging yields

This expression requires the logarithm’s argument to be positive. In the usual population setting with \(0<P_0,K\) and \(0<P_1<K\), the observations must also be consistent with movement toward \(K\): an increasing trajectory has \(P_1>P_0\), while a decreasing trajectory above \(K\) must be handled with the corresponding formula and signs checked carefully.

Worked Example: Find the Growth Rate and Predict a Population

Why the Logistic Formula Works

The formula is not merely a convenient curve-fitting expression. It satisfies the differential equation and the initial condition. To check this, write \(A=(K-P_0)/P_0\) and \(P(t)=K/(1+Ae^{-rt})\). Differentiating gives

Meanwhile, using \(P=K/(1+Ae^{-rt})\), we obtain

and therefore

Also, \(P(0)=K/(1+A)=P_0\), because \(1+A=K/P_0\). Thus the formula satisfies both the differential equation and the specified initial value. This check explains why it can be used to make the model’s predictions.

Worked Example: Estimate a Rate from Two Observations

Equilibrium and Fastest Growth

The carrying capacity \(K\) is an equilibrium because the model’s rate is zero at \(P=K\). For a positive initial population below \(K\), the population approaches \(K\) in forward time; this does not mean it reaches \(K\) at a finite time. The model’s instantaneous growth rate as a function of population is

Completing the square shows where this rate is greatest:

Since \(r/K>0\), the squared term is nonnegative and is smallest when \(P=K/2\). Thus the greatest possible growth rate for nonnegative \(P\) is \(rK/4\), attained at \(P=K/2\). Whether a particular trajectory reaches that population is a separate question.

Inflection Depends on the Trajectory

The second derivative helps describe curvature. With \(P'=g(P)\), the chain rule gives \(P''=g'(P)P'\), where

Consequently, \(P''=0\) when \(P=K/2\), for a non-equilibrium trajectory. This identifies a possible inflection population, not a guarantee that the solution changes concavity during the time interval under consideration. A forward-time trajectory with \(0<P_0<K\) reaches \(K/2\) only when \(P_0\le K/2\). If \(K/2<P_0<K\), the population increases toward \(K\) while remaining above \(K/2\), so it does not reach that point in forward time.

Common Mistake: Assuming Every Trajectory Reaches Half the Capacity

Suppose \(K=1200\), \(P_0=1500\), and time moves forward. A mistaken argument might say, “The inflection point is always \(K/2=600\), so this population will eventually reach 600.” That conclusion ignores the direction of motion. Since \(P_0>K\), the rate is negative:

The population decreases toward \(1200\), not toward 600, and stays above \(1200\). Also, \(g'(P)=r(1-2P/1200)<0\) when \(P>1200\), so \(P''=g'(P)P'>0\): the trajectory is concave up rather than changing curvature at \(600\). The correction is to distinguish a population where \(P''=0\) from a population that the given trajectory actually reaches.

A related parameter mistake is dividing the initial-rate equation by \(P_0(1-P_0/K)\) without checking whether it is zero. If \(P_0=0\), then \(P'(0)=r(0)(1-0/K)=0\) for every \(r\); the initial rate cannot identify \(r\). The same problem occurs at \(P_0=K\). Check the conditions before solving for a parameter.