Deriving Exponential Cell Growth

From a Per-Capita Rate to a Population Trajectory

A per-capita rate describes change relative to the current population. If each cell contributes the same constant net rate of change, the number of cells added per unit time increases as the population grows. This proportionality is the defining feature of exponential growth: a population that is larger changes faster, even though the rate per cell stays fixed.

Let \(N(t)\) be the number of cells at time \(t\), and let \(r\) be the constant net per-capita growth rate, measured in inverse time. When division and loss each occur at constant per-cell rates, their contributions can be combined into \(r\). For example, if division adds one net cell per division event and loss removes one cell per loss event, then \(r=\lambda-\delta\), where \(\lambda\) and \(\delta\) are the respective per-cell rates. The growth equation concerns net cell-count change; it does not imply that division and loss occur separately at rate \(r\).

This model assumes a closed population over the time interval being considered, a constant net rate per cell, and no dependence of that rate on the population size or elapsed time. It treats \(N(t)\) as a continuous quantity, so its predictions are modelled cell counts and need not be integers. It is most useful when these assumptions are reasonable over the time window of interest, not as a claim that populations grow exponentially indefinitely.

Deriving the Exponential Solution

For \(N(t)>0\), divide the growth equation by \(N(t)\) and separate the time and population terms. Integrating from time zero to time \(t\) gives the solution:

The initial condition is built into the solution: at \(t=0\), \(e^{r\cdot 0}=1\), so \(N(0)=N_0\). A direct check also confirms the result. Differentiating \(N_0e^{rt}\) gives \(rN_0e^{rt}=rN(t)\), as required by the model. The exponent \(rt\) must be dimensionless: if \(r\) is in days\(^{-1}\), then \(t\) must be in days.

The same derivation shows why a plot of \(\ln N(t)\) against time is a useful diagnostic under this model. Taking logarithms gives \(\ln N(t)=\ln N_0+rt\), a straight line with intercept \(\ln N_0\) and slope \(r\). A visibly curved pattern on that scale suggests that a constant net per-capita rate may not describe the observations over the interval.

Calculating Population Size and Doubling Time

For \(r>0\), doubling time is the time required for the population to reach twice its current size. Under constant \(r\), the time to double is the same no matter when the doubling starts. Set \(N(t+T_d)=2N(t)\) in the exponential solution; the initial population cancels, leaving \(e^{rT_d}=2\). Thus:

The doubling time has units of time. It is not defined as a positive growth doubling time when \(r=0\) or \(r<0\): in those cases the population does not grow to twice its size under this model. When \(r<0\), the time to halve the population is \(\ln 2/|r|\).

Checking the Model and Choosing an Approach

A calculated trajectory is only as credible as its assumptions. Useful checks include plotting cell counts over time and, where counts are positive, plotting their logarithms against time. Approximately linear log-counts support a constant-rate description over the measured interval; systematic curvature indicates that the net rate may be changing. Measurement error, sparse sampling, or variability between replicate cultures can also make a true pattern hard to assess, so a straight-looking plot is not proof that the assumptions hold.

The exponential equation is the continuous-time counterpart of repeated proportional multiplication. A discrete approximation that doubles the population every \(T_d\) time units uses \(N(t)=N_0\,2^{t/T_d}\). When its doubling time represents the same constant growth rate, this expression is algebraically equivalent to \(N_0e^{rt}\), not a different growth mechanism. A simple step-by-step doubling rule is often convenient for mental estimates, but the exponential form directly handles any elapsed time.

If measurements suggest that the rate varies with time, one constant \(r\) may not summarize the whole interval. If the rate depends on population size, a density-dependent model may be more appropriate; such a model requires a different equation and its own assumptions. For complex rates or model structures without a simple closed-form solution, numerical integration can approximate the trajectory, but it does not fix an inappropriate model.

A second limitation is extrapolation. The equation predicts continued exponential growth for every future time if \(r>0\); a real cell population may not maintain that rate indefinitely. Predictions far beyond the observations should therefore be labelled as model-based projections, not as guaranteed outcomes. In addition, an estimated rate from two measurements is sensitive to counting errors, while additional time points can reveal whether a constant-rate fit is plausible.