The Island With No Predators

Natural selection when predators are absent

What happens to speed when nothing hunts you? In this evolution simulation, faster grazers can escape more predators, but maintaining a high top speed costs energy. On the island without predators, that escape benefit disappears while the upkeep cost remains. Slower variants can then leave more offspring.

This is a model of natural selection: inherited differences affect survival and reproduction, changing the population over time. An individual grazer does not decide to evolve or become slower because it feels safe. The two populations change as grazers with different inherited speeds leave different numbers of descendants.

Created and published by EvoEco Studio · Explanation and results checked September 25, 2026.

What is the setup?

Each run starts with two copies of the same island, grass and 90 grazers. The paired populations have identical starting positions, ages, energy and speed values. Their founders have small differences in speed around a nominal reference of 1.0; they are not all equally fast. The initial speed variation uses a normal distribution with a standard deviation of 0.05 before bounds are applied.

• Only grazer speed evolves. Body size and sensing range stay fixed. A grazer reproduces alone after reaching the minimum age and energy threshold. Its offspring inherits its speed with a random change, normally distributed with standard deviation 0.05, bounded between 0.2 and 3.0.
• One island starts with four predators. Their traits do not evolve. Predator speeds are assigned across a configured 0.9–1.7 range; they eat, reproduce and die. Their population has a cap of 10, with arrivals every 600 ticks when fewer than two remain. It is not a constant group of four.
• The other island stays predator-free. Grass regrows under the same rules on both islands, and grazers can die from predation, starvation or old age. There is no later invasion in this playable.

Each seed chooses a reproducible random history under these rules. Sharing a seed gives the islands identical founders; it does not keep their later random events synchronized once predation changes their histories.

Why can slower grazers succeed?

The model charges every grazer for maintaining its top speed on every tick, including while resting. With body size fixed at 1, that upkeep is 0.0025 × top-speed value squared energy units per tick. A speed value of 1.2 therefore costs 0.0036 units in upkeep; a value of 0.8 costs 0.0016. Moving also costs energy, separately from this upkeep.

Grazers normally forage at a shared walking pace, capped by their own top speed. Extra sprint capacity helps them flee; it does not automatically make ordinary foraging faster. Without predators, reduced upkeep can leave more energy for offspring. Under predation, the escape advantage can outweigh that cost. These numerical rules are design choices for this model, not measured metabolic equations for a real species.

How to explore the simulation

• Show me around / Guided tour starts a guided run using seed 42. The captions explain the creatures and read the changing results from that run.
• Pause / Play stops or resumes the simulation. Normal, Fast and Faster change how quickly ticks are displayed, not the ecological rules.
• New run restarts with a new random seed. It can follow a different history from the guided run and the examples below.
• Hover or tap a creature to inspect it. On a phone, use Chart to show the mean-speed graph. The colour key runs from blue (slower) through yellow (near the nominal starting reference) to red (faster).

The public controls let you observe, compare and restart. They do not edit predator counts, mutation rates or energy costs. The generation display follows the average lineage generation of living grazers on the hunted island; generations overlap and do not advance on a fixed clock.

What happened in eight repeated runs?

We reran the base E01 experiment with eight preselected seeds for 48,000 simulation ticks per island. Every paired run ended with a higher mean top speed on the hunted island. Mean speed across the eight runs was 1.24 with predators and 0.74 without predators.

How to read the numbers: 1.0 is the model's nominal speed reference. These values average the top-speed trait of living grazers; they are not their current movement speed. Because founders vary, 1.0 is not exactly the starting population mean: the eight actual founder means ranged from 0.9930 to 1.0163. The downloads include each seed's actual founder mean and the endpoint-to-founder ratio separately.

These results use EvoEco simulation engine 0.1.0 (source fingerprint 263ea5c334b2) and one fixed configuration. The data summary records the engine source fingerprint, complete parameters, run command, runtime and metric definitions. We checked that each pair started with identical founders and grass, and sampled each endpoint at exactly tick 48,000. Download the eight-run results (CSV) or the results and reproduction details (JSON).

What does this model leave out?

• One inherited trait and asexual reproduction. There is no sexual reproduction, recombination, learning or evolution of body size and senses in E01.
• Predator traits and environmental rules are fixed. Predator numbers change, but predators do not evolve in response. There are no seasons, disease, other prey species or migration of grazers.
• The energy tradeoff is built into the rules. Different costs, food supplies, movement rules or mutation rates could change the outcome. Removing predators alone does not prove that speed must decline in every biological system.
• Eight seeds describe this small set of runs. They do not establish how often the result occurs over all seeds or parameter choices, and there is no conversion from simulation ticks to years.

The experiment illustrates a possible evolutionary mechanism. It is not a reconstruction of the dodo's history, a forecast for a particular island, or evidence that all predator-free animals become slower.

Scientific context and sources

When a source of selection weakens or disappears, biologists call this relaxed selection. Whether a trait then shrinks, persists or changes in another way depends on its remaining costs, benefits and genetic context. The papers below support that biological context; they do not validate this simulation's chosen parameters or numerical results.

1. Lahti, D. C. et al. (2009), Relaxed selection in the wild. Trends in Ecology & Evolution, 24, 487–496. A review of how traits respond when a source of selection is weakened or removed. DOI: 10.1016/j.tree.2009.03.010.
2. Reznick, D. A., Bryga, H. & Endler, J. A. (1990), Experimentally induced life-history evolution in a natural population. Nature, 346, 357–359. A field experiment in guppies found evolutionary changes in life-history traits after changing which age classes faced predation. It studied maturation and reproduction, not sprint-speed loss on predator-free islands. DOI: 10.1038/346357a0.