Scleractinian (hard) corals often dominate shallow coral reef communities. Sponges can also form an important component of the coral reef ecosystems. Accordingly, we selected the corals Acropora millepora and Ctenactis crassa, and the sponges Carteriospongia foliascens, Ircinia sp., and Luffariella variabilis. A. millepora and L. variabilis were important inclusions because both species have previously shown strong preferences for settlement to chemical based cues. A. millepora larvae settle to CCA cues [24] and L. variabilis to conspecific cues [17].

A. millepora and C. crassa are locally abundant corals found throughout the shallow waters of the Great Barrier Reef. A. millepora is hermaphroditic and C. crassa is gonochoric. Both species are broadcast-spawning corals with non-feeding larvae that lack zooxanthellae when released.

Larvae were cultured from three colonies of A. millepora and one female and four male specimens of C. crassa. All corals were collected from the fringing reefs of Orpheus Island, central Great Barrier Reef, two days prior to the full moon in November 2012, transported to Orpheus Island Research Station (OIRS), and maintained in flow through aquaria until spawning occurred in December 2012. Gametes from all A. millepora colonies were collected, combined, and left for 2 h to fertilize. Eggs of the female C. crassa polyp were collected on the 4^th of December 2012, combined with sperm from 3 male polyps, and left for fertilization for 1.5 hours. After fertilization, developing embryos were transferred to 500-L aquaria containing 1 μm filtered seawater (FSW). The aquaria were continuously supplied with flow through FSW, at a flow rate of approximately 2 L min^-1. The aquaria were maintained for eight days in a temperature controlled room at 27C with a 12:12 h light: dark cycle. The majority of larvae showed settlement behaviour (cork-screw like swimming towards bottom of aquaria) at six days after spawning and were then used in settlement assays.

Luffariella variabilis, Carteriospongia foliascens and Ircinia sp. are brooding dictyoceratid sponges that are locally abundant within the Great Barrier Reef [49,50]. These three species dribble spawn parenchymella larvae daily for several weeks during the Austral summer and are competent to settle within three days of release [17,51,52] (Whalan unpublished data). Up to 10 individual sponges from each species were collected from the fringing reefs of Orpheus Island in December 2012 and transported to OIRS where they were maintained in flow through aquaria. Larvae were collected by placing mesh traps over sponges following Whalan et al [33], and immediately used in larval experiments.

Larval size is important in interpreting the effect that specific sizes of microtopography have on larval settlement choices. Therefore, 50 larvae from each of the five species were collected at spawning (sponges), or just prior to use in settlement assays (corals), and fixed in 2.5% glutaraldehyde to enable measurements of larval width and length.

Artificial surfaces engineered with a range of different sized surface microtopography were used to test the effect of surface microtopography on larval settlement. Surfaces comprised 50 mm × 50 mm × 5 mm polycarbonate tiles (Makrolon UV, Bayer Material Science). Surface treatments were prepared by drilling 200 μm deep holes at 400 μm, 700 μm or 1000 μm diameters using a Hartford VMC 1020 numerically controlled mill with Heidenhain controller at a drill speed of 8000 rpm. Surfaces were designed with a 15 × 15 grid of holes (see Fig. 1a for an example of 400 μm treatments). The distances between each hole therefore provided an inter-hole, flat surface. Control surfaces included flat tile surfaces without microtopography, although it is noted that the crevice between the tile and vial also presents a form of microtopography, both within controls and all other treatments.

Larval settlement assays for each species were undertaken by pipetting 20 larvae into 20 mL glass scintillation vials holding 15 mL of 10 μm filtered sea water (FSW), placing the artificial settlement tile over the opening of the scintillation vial, and carefully inverting the vial so that larvae were in direct contact with the test surface. Vials were then secured to surfaces with rubber bands and maintained in temperature controlled rooms with 12:12 photo period cycles. A total of 10 replicate tiles for each of the surface sizes were used for each species. Larval metamorphosis was scored every six hours using an inverted microscope. Larval metamorphosis was scored in both holes and on the flat surface between holes for all surface treatments, and the flat surface of controls. While we acknowledge there are differences between the terms of “settlement” and “metamorphosis” [21] for clarity we use the term “settlement” as a descriptor of the process of larval settlement through to larval metamorphosis from this point on.

The effect of surface microtopography on settlement was established with several different analyses. Firstly, to explore the effect of larval size and settlement to sizes of microtopography (among all five species) data was analysed using Principal Coordinate Analysis (Primer E). Secondly, we analysed the effect of surfaces on settlement, treating surface microtopography as a fixed factor, using one way Analysis of Variance (ANOVA). Repeated measures ANOVA was used to analyse the effect of surface microtopography on settlement over time, with time as the within factor variable and surface treatment as the (fixed) between factor variable. Finally, to establish if settlement was occurring randomly to holes or non-holes, for each surface microtopography treatment, we tested the data against a goodness of fit binomial probability model (settlement to a hole versus non hole) based on the respective surface areas of holes and no holes within each tile. Because each tile treatment comprised holes at different diameters the ratio of surface area between “holes” and “no holes” differed between surface treatments. This difference was accounted for in the analysis by including the ratio of the surface area taken up by holes to no holes in the probability model as follows: 400 μm—0.25:0.75; 700 μm—0.2:0.8; 1000 μm—0.55:0.45; for holes to no holes respectively. Unless otherwise stated all data are reported as means ± one standard error.

Larvae from species of both corals and sponges exhibited an ovoid shape at release and during their planktonic phase. Larval behaviours were similar amongst species. During the settlement phase larvae actively explored surfaces and periodically made contact with a surface for several seconds, often spending extended durations in holes prior to settlement. In general larvae exhibited a behaviour whereby they rotated on their posterior-anterior pole during settlement and prior to metamorphosis. Metamorphosis included a transition from a typical larval form to a flattened disk-like morphology (sponges) or distinctive polyp form (corals) (Fig. 1 a-c).

Given the preference for larvae to commence settlement by rotating on their aboral-oral axis the relevant metric of larval size in relation to the holes defining the surface microtopography is larval width. Mean widths of larvae ranged from 273 μm for the smallest species (C. crassa) to 504 μm for C. foliascens (Table 1). The range of surface microtopography constituted surfaces with holes having diameters of 400 μm, 700 μm and 1000 μm. The 400 μm treatment closely resembled larval width for A. millepora, L. variabilis and Ircinia sp., with larger sized holes (700 μm and 1000 μm) being 200–400 μm larger than the larval widths of all five species.

A. millepora, C. crassa and L. variabilis larvae showed clear responses of selective settlement to 400 μm holes, whilst the other remaining sponge species, Ircinia sp., and C. foliascens, showed less specific responses to surface microtopography. This broad trend is clear within the PCA ordination with C. crassa, L. variabilis, and A. millepora grouping near 400 μm holes (Fig. 2). In contrast, C. foliascens and Ircinia sp., settlement occurred across the spectrum of surface microtopographies.

There were significant effects of surface microtopography on larval settlement for the two coral species, A. millepora and C. crassa, and the sponge L. variabilis at the completion point of each experiment (Fig. 3 a-b; Fig. 4a, Table 2). For A. millepora, settlement in 400 μm treatments reached a mean of 20 ± 4.49%, and this sized treatment also corresponded with the mean larval width of this species (Table 1). There was no settlement of A. millepora in controls. A. millepora settlement in the 400 μm treatments was significantly higher than larval settlement in 700 μm (5 ± 1.63%) and 1000 μm (15 ± 4.08%) treatments, both of which showed consistent levels of settlement (Fig. 3a). There was no settlement of C. crassa to 1000 μm treatments and larval metamorphosis to both 400 μm (19.44 ± 4.4%) and 700 μm (6.52 ± 1.1%) treatments was significantly higher compared to control surfaces (2.22 ± 1.22%); noting that the mean larval width of C. crassa, at 272 μm, was smaller than all hole treatments tested (Fig. 3b; Table 1). L. variabilis had increased settlement in 400 μm treatments and this size topography closely matched mean larval width (Table 1). L. variabilis settlement reached 65 ± 5.82% in 400 μm compared to 35 ± 4.53%, 24.44 ± 3.76 and 32 ± 5.12% in 700 μm, 1000 μm and controls respectively (Fig. 4a).

For the sponges Ircinia sp. and C. foliascens there was no effect of surface treatment on larval settlement. Both these species showed consistent levels of settlement between holes and no-holes (Table 2; Fig. 4 b-c) irrespective of surface treatments.

Settlement also occurred on the walls of vials, or at the vial-tile interface for all three sponge species. Settlement to vial walls was highly inconsistent among species, and surface treatments, and in many instances occurred in a single replicate. These results preclude reliable statistical analysis beyond reporting a summary of cases where settlement occurred (Table 3). No larval settlement was recorded on the walls of the vials for the two coral species.

A. millepora, L. variabilis, C. foliascens and Ircinia sp. larvae completed settlement within three days, and C. crassa settlement occurred over an 11 day period (Fig. 3 a-b, Fig. 4 b-c). A. millepora, C. crassa and L. variabilis larvae all showed significant interactions between time and treatment effects (A. millepora—Pseudo F _{12, 177} = 2.74, p < 0.01; C. crassa—Pseudo F _39,495 = 10.06, p < 0.01; L. variabilis—Pseudo F _{5, 210} = 3.73, p < 0.01). Settlement for the remaining two sponges, C. foliascens and Ircinia sp., was affected by time only (C. foliascens—Pseudo F _{4, 186} = 14.46, p < 0.01; Ircinia sp., Pseudo F _{3, 144} = 73.94, p < 0.01), confirming, in part, the results at the completion point of settlement assays (Fig. 4 b-c).

The outcome of whether settlement was a reflection of randomly selecting a hole or the flat surfaces in between holes varied among species (Fig. 5 a-b, Fig. 6 a-c). A. millepora showed clear patterns of selectivity for treatments of 400 μm and 700 μm holes with 100% of larval settlement occurring in holes for both these sized treatments (Fig. 5a—no inferential statistics computed). However, for 1000 μm treatments there was no significant difference between A. millepora settlement to holes or to flat surfaces (Fig. 5a, Binomial test p = 0.09). C. crassa larvae selectively settled in 400 μm holes (Fig. 5b, Binomial test 400; p < 0.01), but showed no pattern of selective settlement in treatments containing larger holes (700 μm and 1000 μm).

L. variabilis also engaged holes to undergo settlement irrespective of surface (size) treatment (Fig. 6a, Binomial Test 400 μm p < 0.01, 700 μm p < 0.01, 1000 p < 0.01). C. foliascens larvae showed a significant preference to settle on the flat surfaces of 400 and 700 μm treatments but no distinction was observed in 1000 μm treatments (Fig. 6b, Binomial Test: 400 & 700 μm, p < 0.01; 1000 p = 0.09). Ircinia sp. also showed significant biases to settle on flat (inter hole) surfaces in 400 μm and 700 μm treatments but not in 1000 μm treatments (Fig. 6c, Binomial Test: 400 p < 0.01; 700 μm p < 0.03; and 1000 μm p = 0.26).