Larval seedboxes: A modular and effective tool for scaling coral reef restoration

Abstract

Natural recovery of degraded coral reefs is constrained by low larval recruitment, limiting restoration at ecologically meaningful scales. While propagule-based approaches have proven effective in plant-dominated systems, scaling larval restoration for sessile invertebrates like corals remains challenging. Traditional coral larval methods rely on net enclosures, restricting impact to small areas (<75 m²). We developed and tested a modular, passive larval delivery system-the larval seedbox-to overcome these spatial constraints. Each unit (600 × 500 × 300 mm; 11 kg) enables delayed release of competent larvae near the benthos, enhancing substrate encounter rates over broader areas. At Lizard Island (Great Barrier Reef), five seedboxes delivered ~14 million larvae across ~2 ha of degraded reef. Larval release coincided with slack currents to facilitate local retention and subsequent dispersal. Settlement was assessed on 234 tiles placed in concentric arrays around each seedbox. After 48 h, 85% of tiles had settlers (up to 1041 per tile), with mean densities 24 times greater than background levels. Enhanced settlement was directly quantified across ~470 m², with spatial modeling estimating >3000 m² via tidally driven dispersal. The larval seedbox enables unrestrained, scalable coral larval seeding and represents a practical advance toward broad-scale reef restoration.

Keywords: conservation; dispersal; propagule; recruitment; resilience; settlement.

FIGURE 1

(a) Top and sides of a larval seedbox showing placement of eight, 100 mm holes and a central 125‐mm hole on top for inspection ports. (b) Bottom side showing the drill pattern for the 546 holes, each tapering from 10.5 to 2.5 mm to encourage larval escape, and (c) the hole cover guides on the base and a 550 × 500 mm cut section of 6‐mm polycarbonate sheet that slides over and covers the holes in the base. (d) Larval seedbox (left) prior to larval deployment showing it raised 50 mm from the substrate on aluminum legs and (right) showing the properties of the rhodamine dye coming out from the bottom and dispersing laterally and vertically. Image credits: George Roff (d, left) and Geoffrey Carlin (d, right).

FIGURE 2

(a) Lizard Island, northern Great Barrier Reef, (b) study location at Watsons Bay and control location (marked with red star), (c) orthomosaic of study location showing the five seedboxes with georeferenced tiles (white circles), (d) tile array surrounding each seedbox, and (e) inset example of georeferenced tiles from orthomosaic. Images credit: George Roff.

FIGURE 3

(a) Spatial map of tiles within each seedbox array colored by settler density (where “x” indicates zero settlement); (b) maximum settlement density within each seedbox for all tiles (including tiles immediately adjacent to the seedbox) and for outer tiles (>50 cm from each seedbox); (c) mean settlement density per tile for the reference site and five seedboxes (horizontal line = mean settlement across all seedboxes, 3.65 larvae tile⁻¹); and (d) Bayesian posterior median estimates (with 95% credible intervals) from the zero‐inflated Poisson model of settlement within seedbox tiles relative to the baseline “natural settlement” at the reference site (with exp(β) or x‐fold differences above).

FIGURE 4

(a) Spatial map of seedbox site including orthomosaic, tile arrays, and interpolated thin‐plate spline map of settlement density; (b) rhodamine cloud 60 min after release time; (c) rhodamine cloud 120 min after release time showing NNW dispersal from the seedbox site across adjacent reef habitat, with inset flow data supporting a NNW dispersal of larvae for the 12‐h period following release. Images credits: George Roff.