Biological interpretation
Dynamical-systems mathematics tells us how model variables change. Biological interpretation asks what those mathematical results mean for the organisms, populations, infections or physiological processes represented by the model.
Three layers of a model
| Level | Question | Example |
|---|---|---|
| biology | What process is represented? | infection spreads through contact |
| mathematical model | How is it represented? | an infection term such as \(\beta SI/N\) |
| dynamical analysis | What behaviour follows? | infection grows or decays near the disease-free state |
Variables and parameters need meaning
Identify what each variable and parameter represents and its units before interpreting a graph, equilibrium or threshold. The same mathematical value can have different meanings in different models.
Equilibria as biological states
An equilibrium \(\mathbf x^*\) satisfies
\[\frac{d\mathbf x}{dt}=\mathbf0.\]It may represent extinction, disease-free status, endemic infection, persistence, coexistence or a regulated physiological state. Mathematical equilibria must also be checked for biological feasibility.
Stability as response to disturbance
A locally stable equilibrium keeps sufficiently small perturbations close to the equilibrium. A locally asymptotically stable equilibrium additionally attracts nearby trajectories, so sufficiently small perturbations decay toward it.
Biologically, local asymptotic stability can represent local recovery after a small disturbance.
Stable does not mean desirable
Stability is a mathematical property, not a value judgement. An endemic disease state or degraded ecological state can be stable within a model.
Unstable does not mean impossible
An unstable equilibrium is still an equilibrium. A deterministic model beginning exactly there remains there, but arbitrarily small perturbations in an unstable direction can move it away.
Local versus global behaviour
Local asymptotic stability concerns sufficiently small disturbances. A large disturbance may enter another basin of attraction, so local recovery does not imply recovery after every possible disturbance.
Eigenvalues as local response rates
After linearisation, if \(\operatorname{Re}(\lambda)<0\), the corresponding disturbance mode decays; if \(\operatorname{Re}(\lambda)>0\), it grows.
A real part close to zero from below indicates slow local recovery. A more negative real part gives faster decay of that mode.
Complex eigenvalues
For \(\lambda=a\pm ib\), \(a\) controls growth or decay and \(b\) produces oscillation. Thus \(a<0\) gives damped oscillatory return, while \(a>0\) gives growing oscillatory departure.
Nullclines
On an \(x\)-nullcline, \(dx/dt=0\). Only \(x\) is instantaneously unchanged; the whole system need not be stationary. Intersections of all relevant nullclines are equilibria.
Phase planes
A phase plane shows how variables move relative to one another and can reveal attraction, repulsion, thresholds, coexistence and oscillatory motion.
Limit cycles
A stable limit cycle is a periodic orbit that attracts nearby trajectories. The system approaches a repeating rhythm rather than a constant equilibrium.
Repeated waves in data do not alone prove a deterministic limit cycle; forcing and stochasticity can also generate repeated patterns.
Bifurcations
A bifurcation is a qualitative change in dynamics as a parameter changes. Biologically, it can represent invasion, loss of persistence, onset of coexistence, disappearance of a stable state or emergence of sustained oscillation.
Epidemic invasion example
At a disease-free equilibrium, introduce a very small infected population. If that perturbation decays, the disease-free equilibrium is locally asymptotically stable in the infection direction. If it grows, infection can invade under the model assumptions.
In many epidemic models this change is associated with \(R_0=1\), although the exact threshold structure depends on the model.
Population threshold example
In a strong Allee-effect model, an unstable positive equilibrium can separate extinction from persistence. It acts as a critical population threshold within the model.
Predator–prey interpretation
Predator–prey trajectories describe coupled changes in both populations. A lag between peaks can arise because predator growth responds to earlier changes in prey abundance.
Competition interpretation
A locally asymptotically stable coexistence equilibrium means sufficiently small perturbations in species abundances decay. It does not imply coexistence from every initial state.
State disturbances versus parameter changes
A state disturbance changes current variable values. A parameter change changes the rules of the dynamics. Stability mainly concerns state disturbances with parameters fixed; bifurcation analysis concerns parameter changes.
Transient behaviour matters
Models approaching the same equilibrium can have different transient paths, including overshoots, slow recovery or damped oscillations. These differences can matter biologically.
Model assumptions limit conclusions
Mathematical conclusions are conditional on the model. Homogeneous mixing, constant parameters, no spatial structure or simplified interactions can limit biological interpretation.