Sea Star Chemotaxis on a Reef Patch in 3D: Scent Plume, Arm-Tip Receptors and Energy Budget

Starfish smell mussel beds with five arm tips, climb the scent gradient across a Blender-built reef and spend an energy budget doing it.

2D top view

Presets

Playback

Parameters

Layers

Camera

Quality

Model checks

The science

What you see. Mussel beds leak a scent that the bottom current carries and mixing spreads out. Each starfish has no head, but five arms with chemical receptors at the tips. It compares what the five tips smell, turns toward the strongest side, crawls in the direction of the nearest arm, and eats when it reaches a bed. The reserve falls with every minute of living and every metre of crawling, and rises while feeding.

Scent field. Depth-averaged concentration C(x, y, t) in arbitrary units (au), on a 48 x 48 grid over a 3 x 3 m reef patch: ∂C/∂t + u·∇C = D ∇2C − k C + Σ Qj δ(x − xj). u = bottom current (0-3 cm/s), D = turbulent mixing (5-80 cm2/s, an effective eddy diffusivity), k = 1/lifetime of the scent, Qj = emission of bed j = Q0 Bj/(Bj + Bh), so a bed that is eaten down smells weaker. The scheme is a conservative finite-volume method: upwind advection, central diffusion, explicit time step limited by dt ≤ 0.8 / (4D/Δx2 + (|ux| + |uy|)/Δx + k). Clean water enters at the upstream walls and scent leaves at the downstream walls. In still water the steady plume is C(r) = Q/(2πD) K0(r/λ) with λ = √(D/k), which the page uses as a check.

Chemotaxis of one starfish. Every 10 s the five arm tips (radius 8 cm, at body angles φ + 72° i) sample C, each with relative noise σ. The gradient estimate is g = (2/(5R)) Σi (ci − mean c) ei (exact for a linear field). If the mean exceeds the detection threshold and the animal is hungry, the heading θ relaxes toward the direction of g: dθ = ω η wrap(arg g − θ) dt + √(2 Dθ dt) N(0,1), with ω = 0.02 s-1 and η falling from 1 to 0 as the reserve rises from 88 % to 98 % of full (a satiated star stops following the scent). Dθ = 1/(persistence time) gives the random wandering. The body turns slowly so that the nearest arm points along the heading (the “lead arm”). Speed is v = v0 / (1 + (s/0.8)2) with s = local slope of the rocks (the domes are h (1 − (d/r)2)2, the same function the Blender rocks are modelled from), so steep rock faces slow it down. Within 10 cm of a bed with prey left the star stops and feeds.

Energy budget (illustrative numbers). dE/dt = ηa I(B) − Pbasal − cmove (1 + s2) v with I = Imax B/(B + Bh), full reserve 200 J, assimilation ηa = 0.5, Pbasal = 5 mW, cmove = 2 J/m, Imax = 30 mW, Bh = 40 J. These are round numbers chosen so that behaviour unfolds within a few simulated hours, not measurements. At 0 J the animal stops (“starved”). The reserve closes exactly: start + assimilated − basal − walking = present value.

Speeds and ranges. Reported crawling speeds of common sea stars are of the order of a few centimetres per minute, with large differences between species, temperature and food (some big species are much faster). The slider spans 1-15 cm/min as a typical order of magnitude, not a fit to a species. The turbulent diffusivity range, the scent lifetime, the detection threshold and the receptor noise are plausible illustrative values; real odour plumes in the sea are patchy filaments, not smooth fields, and the stars here respond to the mean field plus noise.

Simplifications. Two-dimensional, depth-averaged; the current is uniform and constant (no tide, waves, boundary layer or obstacle wakes); rocks change speed and cost only, not the water flow or the scent; no light, no predator, no competition for a prey item beyond sharing the bed; arm motion in the 3D view is a decorative rigid-part animation (tube-foot waves are not simulated). Coral and sand are scenery, and the starfish are drawn 1.3 times larger than life so that they stay visible.

Checks on this page. Seven independent checks run when the page loads: Gaussian diffusion against the exact solution, scent mass closure with current, decay and open walls, exact advection of the plume centre, the steady plume against the Bessel K0 solution, the mean squared displacement of 4000 random walkers against the persistent-random-walk formula ⟨r2⟩ = 2 v2 τ [t − τ(1 − e−t/τ)], exact energy and prey bookkeeping, and noise-free gradient climbing against the straight-line travel time.

Sources and what to verify. Chemoreception and the terminal-tentacle sensory organ of sea star arms; the keystone role of a starfish predator on mussel beds (Paine, 1966); eddy-diffusion models of odour plumes (Murlis, Elkinton and Cardé, 1992; Weissburg, 2000); Abramowitz and Stegun for K0. Citations are written from memory: please verify before quoting. Model, textures and sprites were made for this page; the Blender models are scenery based on common reef shapes, not a particular species.

Loading model...
Loading Blender model...
Energy reserve of each starfish against time
Scent at the arm tips of starfish 1 against the detection threshold
Distance of each starfish to the nearest mussel bed
Made with 🟠 Blender · Three.js