Shifts from cooperative to individual-based predation defense determine microbial predator-prey dynamics

The study investigates how alternative prey defenses shape long-term predator-prey dynamics in a microbial system. Classical predator-prey theory predicts equilibria or oscillations, but inducible defenses and evolutionary change can add complexity and alter trajectories. Bacteria exhibit diverse defense strategies differing in mode (attack vs. elusion), scope (individual vs. cooperative), and reversibility (inducible vs. permanent). Cooperative defenses (e.g., toxin production) can be highly effective but are vulnerable to cheating; permanent defenses can arise via de novo mutations or selection on standing variation. Prior work often examined single defense mechanisms over short timescales, leaving the ecological and evolutionary interplay of multiple defenses over many generations unresolved. The research question is whether and how prey populations shift between cooperative, inducible defenses and individual, genetically fixed defenses under sustained predation, and what ecological and evolutionary mechanisms drive such successions. The purpose is to track in situ dynamics over weeks, identify defense mechanisms, test for predator adaptation, and develop a mechanistic model to explain observed patterns. The importance lies in revealing how individual-level selection can undermine population-level cooperative defenses, reshaping community dynamics.

Background literature documents numerous bacterial grazing defenses, including chemical inhibition, morphological changes (e.g., filamentation), and aggregation/biofilms. Defenses vary by inducibility and social scope; community defenses (public goods like extracellular toxins or siderophores) are effective at high densities but susceptible to cheaters. Inducible phenotypes can be switched off when costs outweigh benefits, whereas permanent genetic defenses may entail fitness costs but confer stable protection. Studies have shown rapid evolution can alter predator-prey dynamics and that siderophore-mediated interactions influence cooperation and virulence. However, most experiments focus on single defenses over short periods, providing limited insight into how multiple defenses trade off and succeed one another under continuous predation and dilution, and how individual selection can erode cooperative defenses.

Organisms and medium: Predator-prey system comprised Pseudomonas putida KT2440 (prey) and the bacterivorous flagellate Poteriospumella lacustris JBM 10 (predator). Cultures were maintained in wheat grass medium prepared from a 25 g L−1 wheat grass infusion, filtered (0.2 µm), supplemented with balanced salts and phosphate, and autoclaved.

Experimental design: Semi-continuous planktonic co-cultures were run for five to six weeks at 19 °C, shaking at 120 rpm, with daily 50% dilution into fresh medium. At least 10 mL were transferred at each dilution. Sampling occurred every 24 h to quantify prey and predator abundances microscopically. Replicated experiments (n = 16 total across time points) were initiated under different predation pressures: high (1×10^5 flagellates mL−1; 1×10^5 bacteria mL−1), low (1×10^2 flagellates mL−1; 1×10^4 bacteria mL−1), and predator-free controls (bacteria 1×10^4 mL−1). Bacterial inocula were derived from a single KT2440 colony; flagellates were washed twice in wheat grass medium prior to inoculation.

Cell quantification and morphology: Bacteria were stained with SYBR Green I (1×, 15 min, dark), filtered onto 0.2 µm black polycarbonate filters, and counted by epifluorescence microscopy (≥250 cells/sample). Filament lengths (n=100 per sample/day) were measured in selected replicates. Flagellates were counted with Neubauer improved chambers (technical duplicates, 10 µL load). Filament biovolume and size distributions were recorded.

Genomics of filamentous isolates: Filamentous KT2440 isolates from independent co-cultures and a single-celled control were whole-genome sequenced (Illumina NovaSeq 6000 PE; ~5 million paired reads; ≥100× coverage). DNA extraction used Qiagen PowerSoil. Reads were mapped to the KT2440 reference. CDS were annotated with Prokka; variants were classified into synonymous and non-synonymous. Mutations were examined for involvement in septation genes (e.g., ftsQ, ftsA, minC). Sequences were deposited (accession NP_1380374; note: SRA accession also provided elsewhere as PRNJK03374).

Filtrate exposure assays: Sterile filtrates harvested from different phases of co-cultures or bacteria-only controls were used to assess effects on flagellate growth rates via exposure assays; statistics included one-way ANOVA, Tukey HSD, and Wilcoxon rank-sum tests, depending on comparison.

Mathematical modeling: An ODE-based, semi-mechanistic model (implemented in R/RStudio) simulated seven state variables: resource R; undefined single-celled bacteria Bo; metabolite-producing single-celled bacteria Bx; filamentous bacteria Bff; single-celled offspring of filamentous genotype Bfs; flagellates F; metabolites X. Nine processes captured growth, grazing, metabolite regulation (up/down), excretion, filamentation, and asymmetric division. Daily dilution/replenishment was explicitly applied. Process rates used Monod-type functions, step functions for induction thresholds (predator density and quorum), and inhibition functions for predator suppression by metabolites. Key parameters included bacterial and flagellate growth/ingestion constants, metabolite thresholds for upregulation and toxicity, mutation-to-filament rate (conditioned on high X), fitness costs for Bx (≈11%) and Bff (≈5% growth disadvantage), probability of asymmetric division (p = 0.6), and assumption that filaments do not produce metabolites at inhibitory levels. Model outputs were compared qualitatively and quantitatively to observed dynamics; code and data were made available via GitHub.

• Consistent defense succession in all 16 replicate co-cultures over ~5 weeks (~35 predator generations): initial cooperative chemical defense (toxic metabolite production) that nearly collapsed predator populations, followed by replacement with a genetically fixed individual defense (filamentation).
• Initial community defense: Upregulated toxin production at high predator and bacterial densities (consistent with joint predator recognition and quorum sensing). Filtrate exposure experiments showed significant inhibition of flagellate growth during the predator-collapse phase; inhibition mirrored the semi-continuous cultures. Bacterial abundances during this phase exceeded predator-free carrying capacity, implying a dual function of metabolites (growth enhancement and predator suppression), consistent with siderophore/pyoverdine-like activity.
• Predator adaptation not detected: Flagellates sampled at the onset of the filamentous phase remained strongly inhibited by filtrates harvested from the earlier collapse phase, indicating no detectable evolution of toxin resistance by predators.
• Emergence and dominance of filamentation: Around day ~20, filamentous bacteria appeared; by day ~25 their biovolume matched single-celled bacteria (~1.5×10^6 µm³ each), then became predominant. Filaments reached up to ~200 µm; >97% exceeded ~5 µm, rendering them effectively grazing resistant. Filamentation was non-reversible in predator-free culture, consistent with a genetic basis.
• Genetic basis of filamentation: Whole-genome sequencing of 9 filamentous isolates revealed non-synonymous substitutions; in 4/9 isolates mutations mapped to septation-related genes (ftsQ, ftsA, minC), consistent with a mechanistic link to filamentous morphology.
• Fitness trade-offs: Filamentous isolates exhibited ~5% lower growth rate than undefended single-celled bacteria (Wilcoxon p < 0.05). Metabolite production incurred an estimated ~11% fitness cost (model fit; consistent with literature on pyoverdine costs).
• Modeling results: The ODE model reproduced the observed shift from high-amplitude cycles under toxin defense to a steady state dominated by filaments with sustained predator presence. It attributed stable predator persistence to continuous release of single-celled offspring via asymmetric filament division. The model indicated the filamentous genotype arises de novo during the experiment (not from standing variation), and would not predominate early unless seeded. Even when filamentation had higher costs than toxin production, selection at the individual level favored its eventual dominance, explaining the replacement of a superior population-level defense (cooperative toxin) by an inferior one (filaments) from the population perspective.
• Mechanistic insights: Mutation rates leading to filamentation were best explained as conditional on stress (high metabolite concentrations), aligning the timing of emergence across replicates. Filaments were assumed and empirically supported to not produce inhibitory metabolites at toxic levels.

The findings show that in microbial predator-prey systems, cooperative, inducible defenses that maximize population-level benefits can be superseded by individual, genetically fixed defenses that maximize individual fitness despite offering less community-level protection. Initially, P. putida upregulated production of secondary metabolites (likely siderophore/pyoverdine-associated), simultaneously suppressing flagellate growth and boosting bacterial resource exploitation at high densities. However, this cooperative defense carries production costs and is vulnerable to non-producers and to environmental fluctuations that turn it off when predator density drops. Rapidly evolving filamentous mutants, though incurring a modest growth penalty and providing no benefit to conspecifics, secure robust, heritable protection and persist under dilution. The ODE model clarifies that sustained predator numbers in the filament-dominated steady state result from asymmetric division that continually supplies grazeable single cells. The absence of detectable predator adaptation to toxins suggests the defense transition was driven by prey evolution rather than an arms race. Joint control of toxin expression by predator recognition and quorum sensing explains the strong initial inhibition and subsequent downregulation that created ecological space for filamentous mutants. Overall, selection acting at the individual level can dismantle successful social defenses, reshaping long-term dynamics and leading to stable coexistence at altered community states.

This work demonstrates a reproducible succession from a cooperative, inducible chemical defense to an individual, genetically fixed morphological defense in a microbial predator-prey system. Experimentation and mechanistic modeling together reveal that individual-level selection can override population-level optimization, leading to the breakdown of social defenses and establishment of a filament-dominated steady state that sustains predators via continual release of single cells. Contributions include: (i) documenting defense succession across many generations and replicates; (ii) providing evidence for a dual-function metabolite-based defense and its regulation; (iii) establishing the genetic basis and costs of filamentation; and (iv) presenting an ODE framework that recapitulates observed dynamics and explains persistence mechanisms. Future research should identify and quantify the specific metabolites/toxins, map the full genetic architecture of filamentation and its regulatory pathways, assess how spatial structure and community complexity influence cooperation breakdown, explore generality across prey and predator taxa, and test how environmental regimes (e.g., resource pulses, dilution rates) modulate defense transitions.

• Metabolite identity not directly confirmed; inference of siderophore/pyoverdine involvement is based on genomic potential and phenomenology, not direct chemical characterization.
• Genetic causality for filamentation was pinpointed in only a subset of isolates; broader mutational landscapes and potential parallel pathways remain to be resolved.
• Semi-continuous, well-mixed laboratory conditions lack spatial structure, potentially affecting the stability of cooperation and generalizability to natural ecosystems.
• Model includes assumptions (e.g., conditional mutation under high metabolite levels, no metabolite production by filaments) that, while consistent with data, require further experimental validation.
• Parameter estimates (e.g., fitness costs) are context-dependent (medium, temperature) and may vary across environments or strains.
• Predator adaptation was assessed via filtrate exposure; other adaptive responses (e.g., behavioral or ingestion threshold changes) might have been undetected at the resolution used.