The Hidden Math of Chemo, Pesticides, and Deworming: When Repeated Pulses Let Species Survive

Every few weeks, a farmer sprays a field. Every few months, a patient sits in an infusion chair. Every year, a village lines up for a deworming pill. These are not the same act, but they share a hidden structure: biology runs continuously in between interventions, then gets abruptly knocked sideways at a scheduled moment. Chemotherapy shrinks a tumor, the tumor regrows during the drug holiday. Pesticide wipes out a pest outbreak, the pest rebounds between applications. Mass drug administration cuts parasite loads, transmission slowly climbs back before the next round.

Mathematicians model these situations as impulsive differential equations — differential equations that describe continuous growth, interrupted instantaneously by discrete pulses. For decades, most of the work on such systems asked a narrow, practical question: How do we eliminate the bad species? Eradicate the pest, kill the tumor, clear the infection. But elimination is often infeasible. Chemotherapy is capped by patient tolerance. Pesticide overuse breeds resistance. Public-health programs run out of resources. In those settings, a different question becomes urgent: When can we keep multiple populations — even the "bad" ones, in some cases — coexisting stably, instead of tipping into exclusion?

A new paper by Eleanor Courcelle, Jane Shaw MacDonald, and Swati Patel (arXiv, 2026) delivers a general mathematical answer to that question. The researchers developed conditions that guarantee permanence — a formal property meaning that every species stays bounded and never drifts to extinction — for any $n$-species population model subject to periodic multiplicative pulses (Courcelle et al., 2026). The elegance of their result is that it cleanly separates the two forces at work: the continuous ecological dynamics between pulses, and the discrete effect of the pulse itself. And they show their conditions remain robust even when the biology, the pulse strength, or the timing of treatments is slightly perturbed — a crucial feature in the messy, imprecise real world of farming and medicine.

The result is not just theoretical furniture. The authors apply it to two concrete, high-stakes problems: the coexistence of chemotherapy-sensitive and chemotherapy-resistant cancer cells, and the integrated control of an agricultural pest using both pesticides and biological parasitoids. In both cases, the framework reveals something intuitive that was previously hard to prove rigorously: whether populations coexist or one wipes out the other depends on a subtle interplay between how often you intervene, how hard you hit, and how the species interact when left alone.

The Science

The paper operates at the intersection of ecology and dynamical systems theory — the mathematics of how things change over time. The setup is a system of $n$ species whose densities $x_i(t)$ grow continuously according to a per-capita growth rule, but once every $\tau$ time units, each species is multiplied by a factor $1+h_i$. If $h_i$ is negative (between $-1$ and $0$), the pulse shrinks species $i$ — appropriate for a pesticide or a drug. If $h_i$ is positive, the pulse is a birth event. Combined, this yields an impulsive system.

The challenge is that an impulsive system is not a dynamical system in the classical sense. Dynamical systems require a clean semigroup property — that evolving forward in time behaves consistently regardless of how you slice the interval. An impulsive system breaks this: a trajectory that hits a pulse in the middle behaves differently from one that doesn't. The standard workaround is the heavy machinery of skew-product flows. The authors chose a more elegant path.

Their central trick, illustrated in Figure 6, is to transform the pulsed system into a continuous, autonomous one by spreading each instantaneous pulse out over a unit of "extended" time on a circle. The state space gains an extra angular coordinate $\theta$ that rotates on the circle $S^1$. For most of each revolution, the system evolves under the continuous ecological dynamics; for a brief segment, it evolves under a constant growth rate $\ln (1+h_i)$ that reproduces the effect of the pulse.

One revolution takes exactly $\tau +1$ time units, matching the pulse period. Now the dynamics are autonomous — a genuine dynamical system — and a whole toolkit of classical permanence theory becomes available.

The key intellectual ancestors here are the results of Garay and Hofbauer (1998) for continuous systems, and Schreiber's recent "flow-kick" framework for impulsive ones (Schreiber, 2021). The paper's main contribution is to extend the Garay–Hofbauer approach — which ties permanence to weighted growth rates on a Morse decomposition of the extinction set — to the impulsive setting, and to prove robust permanence, something Schreiber had conjectured but not established.

A Morse decomposition is a way of carving an invariant set into a hierarchy of smaller invariant pieces, ordered so that trajectories flow from higher-index pieces down to lower-index ones. On the extinction set — where at least one species has zero density — a Morse decomposition organizes all the ways the system can "get stuck" without all species present. The permanence question then becomes: can the system be coaxed away from every one of those extinction traps?

The researchers prove their main theorem by establishing a precise correspondence between the Morse decompositions of the impulsive discrete map and those of the embedded autonomous system, then applying a permanence result from Patel and Schreiber (2018) for autonomous differential equations with feedbacks. It is a beautiful act of translation: impulsive problems rendered continuous, continuous problems solved by classical theory, permanence transferred back.

What They Found

The paper's central result, Theorem 3.1, states a sufficient condition for permanence. The statement involves a weighted sum of growth rates, but its meaning is intuitive. For each extinction piece $M_k$, the researchers require the existence of positive weights $p_{k1},\dots ,p_{kn}$ — one per species — such that, averaged over some number of pulse periods, the weighted total growth rate is positive.

Crucially, that growth rate splits into two explicit contributions:

$$\sum _{i=1}^n{p}_{ki}\sum _{j=0}^{{\ell }_x}{\int }_0^{\tau }\left[f_i(\Phi (\pi (x,j),s))+\frac{\ln (1+h_i)}{\tau }\right]ds>0$$

The first term inside the integral, $f_i$, is the continuous per-capita growth rate of species $i$ between pulses — the ecological competition, predation, or cooperation at work. The second term, $\frac{\ln (1+h_i)}{\tau }$, is the per-period logarithmic effect of the pulse on species $i$. Because the natural log of a number less than one is negative, a negative $h_i$ (a killing pulse) contributes a negative term. But the pulse also has an indirect effect: by changing population densities, it alters the continuous trajectories that feed into the first term.

This decomposition is the paper's most practically valuable feature. It lets a manager or a clinician see exactly how the intervention and the underlying ecology trade off. A strong pesticide pulse on the pest contributes a large negative direct term; but if pest numbers crash, the pest's natural enemies — the parasitoids — may also suffer or the pest's own density-dependent competition may relax, shifting the indirect term.

The authors then apply the framework to three cancer scenarios and one pest-control scenario.

The first cancer case

is a system that is permanent in the absence of treatment — sensitive and resistant cells coexist naturally. The question is whether pulsed chemo breaks that coexistence. The researchers find a bifurcation structure: for a given pulse period, there is a threshold of treatment intensity above which permanence is lost.

The second case

is more subtle and more interesting. In the untreated system, the sensitive cells actually have the competitive advantage — resistant cells would lose in a fair fight. Yet pulsed treatment, by selectively sparing the resistant type, can create a priority effect: depending on initial conditions, either the sensitive or the resistant population gets driven to extinction. The outcome depends not just on the parameters, but on where you start — a sobering reminder that the same treatment protocol can lead to opposite outcomes in different patients depending on their initial tumor composition.

The third cancer case

is the grim one: resistant cells have the fitness advantage even untreated, and would exclude the sensitive type on their own. Curiously, the researchers find that concave pulse effects — where the resistant type is hit proportionally less hard than the linear case — can help the sensitive type hang on, because the sensitive cells experience a survival benefit from the treatment pressure that keeps the resistant population in check.

Why This Changes Things

The two

Pulsed chemo can shift coexistence regimes

Label	Value
Untreated coexistence (permanent)	1
Untreated: sensitive excludes resistant	2
Untreated: resistant excludes sensitive	3

Synthesized from the three cancer scenarios (Figs 2-4). Untreated coexistence can persist or be broken by pulses; untreated sensitive-advantage can develop a priority effect; untreated resistant-advantage can be tempered by concave pulses (Courcelle et al., 2026).

and

Intervention frequency gates permanence outcomes

Label	Value
Low frequency	1
Moderate frequency	2
High frequency	3

Interpretive summary of the bifurcation plots (Figs 2-5). Higher intervention frequency generally strengthens the pulse's selective and indirect effects, which can either preserve or destroy coexistence depending on the ecological setting (Courcelle et al., 2026).

below summarize the qualitative regimes the framework identifies across these applications.

Intervention frequency gates permanence outcomes

Label	Value
Low frequency	1
Moderate frequency	2
High frequency	3

Interpretive summary of the bifurcation plots (Figs 2-5). Higher intervention frequency generally strengthens the pulse's selective and indirect effects, which can either preserve or destroy coexistence depending on the ecological setting (Courcelle et al., 2026).

Pulsed chemo can shift coexistence regimes

Label	Value
Untreated coexistence (permanent)	1
Untreated: sensitive excludes resistant	2
Untreated: resistant excludes sensitive	3

Synthesized from the three cancer scenarios (Figs 2-4). Untreated coexistence can persist or be broken by pulses; untreated sensitive-advantage can develop a priority effect; untreated resistant-advantage can be tempered by concave pulses (Courcelle et al., 2026).

The most consequential implication concerns the long-standing puzzle of why chemo-resistant cancer cells coexist with their sensitive counterparts — and what that means for treatment design. Textbook intuition suggests that treatment should push hard toward eliminating the sensitive cells and hope the resistant ones stay rare. But this framework shows the mathematics is more interesting: pulsed treatment can actually stabilize coexistence of two cell types in regimes where the untreated system would collapse to one. That is not necessarily bad news; a stable mixture means the resistant population stays in check, the tumor stays manageable, and the evolutionary pressure toward full resistance remains blunted.

The pest-control example

Integrated control can preserve pest-parasitoid coexistence

Label	Value
Pesticide only	1
Parasitoid alone	2
Integrated (pesticide + parasitoid)	3

Interpretive summary of the pest-parasitoid bifurcation (Fig 5, equation 22). Tuning pulse period and treatment specificity lets the parasitoid rebound between sprays, sustaining coexistence (Courcelle et al., 2026).

carries the same lesson for agriculture. Integrated pest management — combining pesticides with natural enemies like parasitoid wasps — is often recommended as a resistance-management strategy. The framework formalizes when that combination works: when the pulse period and pesticide specificity are tuned so that the parasitoid's capacity to rebound between applications offsets the direct kill.

Spray too often or too unselectively, and the parasitoids themselves crash, leaving the pest to rebound unchecked — the worst of both worlds.

There is also a fundamental conceptual payoff. The paper answers, in the affirmative, an open conjecture from Schreiber (2021) about robust permanence of impulsive systems. "Robust" here is not a rhetorical flourish. It means the permanence property survives not only the exact model, but any sufficiently small perturbation of the continuous dynamics, the pulse period, and the pulse effects. In an applied context this is indispensable: no real pesticide dose is exactly what you prescribed, no chemotherapy schedule is perfectly regular, no biological interaction is precisely what the equations assume. A permanence result that evaporates under the smallest perturbation is a mathematical artifact, not a tool. By establishing robustness, the authors turn a theoretical condition into something a practitioner can actually rely on.

The proof strategy itself — embedding an impulsive system into an autonomous one with a rotating angular coordinate — is likely to be reused. Impulsive systems are notoriously awkward because they sit awkwardly between discrete and continuous mathematics. This embedding dissolves that awkwardness, opening the door to importing a century of autonomous-system theory into the impulsive world.

What's Next

The paper is a framework, not a finished recipe, and the authors are candid about its limits. The permanence condition is sufficient but not necessary — a system can be permanent even when the theorem's condition fails, so the framework gives you one lens, not the whole picture. The results also assume pulses are perfectly periodic and perfectly synchronized across species; real interventions drift, and real treatments often affect different cell types or pest stages with different timing lags.

The cancer application also carries a caveat that the authors do not hide. Permanence of sensitive and resistant cell types means the tumor persists — bounded, stable, but present. That is a management-acceptance framework, not a cure. Whether that's the right clinical objective is a values question as much as a mathematical one. What the math offers is the intellectual honesty to describe what coexistence regimes actually look like before anyone decides what to do about them.

Several directions now look ripe. The pulse effects in this paper are multiplicative scalings; many real interventions are better modeled as saturating or concentration-dependent kills, and extending the framework there would broaden its reach. The proof machinery could also be adapted to stochastic pulses — treatments whose timing or intensity genuinely varies — though the clean autonomous embedding would need modification. And the two-species applications here are deliberately minimal; the real payoff will come in exploring higher-dimensional communities where the interplay of several pulses and several interactions becomes combinatorially rich.

There is something quietly hopeful in this work. So much of the language around control — eradicate, eliminate, wipe out — imagines a world where we can simply remove the unwanted. This paper is an argument for a humbler, but more realistic, ambition: stable coexistence with the things we fight, managed within bounds we can live with. In cancer, in agriculture, in disease control, that may be the goal that is actually achievable — and the mathematics now has a language for it.

Pulsed chemo can shift coexistence regimes

Label	Value
Untreated coexistence (permanent)	1
Untreated: sensitive excludes resistant	2
Untreated: resistant excludes sensitive	3

Synthesized from the three cancer scenarios (Figs 2-4). Untreated coexistence can persist or be broken by pulses; untreated sensitive-advantage can develop a priority effect; untreated resistant-advantage can be tempered by concave pulses (Courcelle et al., 2026).

Intervention frequency gates permanence outcomes

Label	Value
Low frequency	1
Moderate frequency	2
High frequency	3

Interpretive summary of the bifurcation plots (Figs 2-5). Higher intervention frequency generally strengthens the pulse's selective and indirect effects, which can either preserve or destroy coexistence depending on the ecological setting (Courcelle et al., 2026).

Integrated control can preserve pest-parasitoid coexistence

Label	Value
Pesticide only	1
Parasitoid alone	2
Integrated (pesticide + parasitoid)	3

Interpretive summary of the pest-parasitoid bifurcation (Fig 5, equation 22). Tuning pulse period and treatment specificity lets the parasitoid rebound between sprays, sustaining coexistence (Courcelle et al., 2026).