ES
Eco-Spine Ecosystem LabAGENT-BASED CONNECTIVITY SIMULATION · REV. 1.0
Research visualisation · mechanistic model

Does an EcoFrame actually change the ecological outcome?

A custom agent-based simulation of bees, butterflies and generalist pollinators moving through a fragmented urban landscape. Instead of treating an EcoFrame as an instant energy “refill”, the model links movement, resource depletion, flowering, nesting, weather, disturbance, microclimate, mortality and reproduction.

Interpretation rule: This is a research-informed exploratory model, not a validated prediction of real Singapore pollinator populations. Distances, trait values and response functions are explicitly parameterised so your team can replace assumptions with field measurements as the project develops.
01

Interactive experiment

READY · DAY 0 · POLLINATOR
Habitat EcoFrame Bee Butterfly Generalist line = most likely movement route
02

Population & connectivity trajectories

Population over time

Cumulative successful crossings

Resource availability at EcoFrames

Ecological trap risk

03

What the model is actually doing

1 · Agents

Each pollinator has energy, age, species-specific resource preferences, movement cost, thermal tolerance, disturbance sensitivity and a probability of reproduction.

2 · Nodes

Habitats and EcoFrames carry dynamic nectar, host-plant, nesting, shelter and water resources. Resources replenish and are depleted through use.

3 · Decisions

Pollinators choose the next node based on resource quality, distance, disturbance, microclimate and remaining energy instead of simply teleporting between hubs.

4 · Feedback

Flowering, rainfall, maintenance, heat and disturbance change resources; resource availability changes survival, reproduction and movement; population pressure feeds back into depletion.

Fig. 06 — Parameter interrogation

Which design variables matter most?

A simple local sensitivity sweep perturbs each parameter while holding the others constant. This is not a causal statistical model; it is a way to expose which assumptions are doing most of the work in the simulation.

READING

How to use this

High sensitivity

Prioritise field measurement or expert validation. If a small change strongly shifts the result, you should not treat that parameter as a casual assumption.

Low sensitivity

The result is comparatively robust to that assumption within the tested range. It may be lower priority for early calibration.

Scenario comparison

Compare Core, Pollinator and Smart configurations under identical landscape conditions to show how added ecological functions change outcomes.

Field calibration

Replace assumed response functions with measured visitation, plant survival, water use, microclimate and maintenance data as the physical prototype matures.

Fig. 07 — Method & evidence discipline

Research-informed, explicitly provisional

This page explains how the simulation connects to your project evidence without pretending that a classroom model is a validated ecological forecast.

MODEL LAYERS

Mechanisms represented

LayerWhat is representedHow to calibrate later
LandscapeGap distance, background green cover and disturbanceGIS, site survey, traffic/noise measurements and existing greenery mapping
ResourcesNectar/pollen, butterfly host plants, nesting, shelter and waterPlant inventory, flowering phenology, habitat occupancy and irrigation logs
MovementDistance cost + resource attraction + disturbance + energy budgetObserved visitation and movement between tagged/observed nodes
MicroclimateTemperature, shade/humidity and weather scenario effectsSensor data from prototype vs control pillars
PopulationSurvival and simplified reproductionObserved abundance, occupancy, life-stage observations and repeated counts
OperationsMaintenance quality, water demand and module burdenActual maintenance hours, water use and replacement records
ECOLOGY

Why nectar + host plants are separate

Butterflies are not supported by flowers alone. NParks explains that nectar plants provide food for adult butterflies, while host plants provide food and habitat for caterpillars. The simulation therefore gives butterflies two separate resource channels. This means a module can look “flower-rich” but still have low reproductive support if host-plant provision is poor.

Bees are represented as a broad guild, not a species. Singapore has documented more than 130 bee species, with diverse urban habitat use. The model therefore avoids claiming one universal bee flight range and instead uses a configurable energy/travel budget that can be calibrated later for particular target species.

SINGAPORE CONTEXT

Evidence anchors

NParks — Nature Corridors & Nature Ways

Official overview of how Nature Ways use multi-tier planting to facilitate wildlife movement between parks and nature reserves.

Open source ↗
NParks — DIY Butterfly Garden

Provides Singapore-relevant examples of butterfly host plants and reinforces the distinction between adult nectar resources and larval host plants.

Open source ↗
AVS — Bees and Wasps in Singapore

Records more than 130 bee species in Singapore and describes their use of urban parks and gardens.

Open source ↗
2025 Singapore road-verge butterfly study

Reports positive associations between butterfly diversity and nectar-floral diversity / vegetation structural complexity, and a positive landscape-scale association with greenness at ≥500 m; traffic density was negatively associated with butterfly diversity.

Open source ↗
LIMITATIONS

What this simulation cannot prove

  • It cannot prove that real bees or butterflies will use EcoFrames at the predicted rates.
  • It does not currently use species-level movement data for a particular Singapore bee or butterfly.
  • It does not calculate real aerodynamic or structural effects of the MRT viaduct.
  • It does not model every predator, parasitoid, pathogen or plant–insect interaction.
  • Its population dynamics are deliberately simplified and should be calibrated against field observations before quantitative claims are made.