1) 3D Scanning and Virtual Augmentation: The technique requires that a sample of the biogenic habitat-forming organism being approximated is accessible through field or archival sampling. We obtained a staghorn coral skeleton fragment from the Coral Restoration Foundation (CRF) in Key Largo FL, United States. We intentionally selected a fragment where majority of the main structural features were in one plane (Figure 3A); although the 3D file can be manipulated to add/remove features (e.g., branches from the coral), this planar form reduced the need for 3D file manipulation. We scanned the fragment using a 2020i Next Engine Desktop 3D Scanner (NextEngine Inc., Santa Monica, United States) to create a 3D mesh file and manipulated the file using 3D Builder (Microsoft^© Application, 2013) to remove all irregularities, ensure high-resolution quality, and create a “water-tight” mesh (remove any “holes” in the file created during scanning). Users that require large and complex AH to address their research question can modularize the process in the 3D file manipulation stage by breaking the design down into smaller separate objects that are assembled into a final product. The design can also be adapted to change proportions of individual components, adjust size ratio, and/or create attachment features. We formatted the resulting 3D-file to create an object with a thin, flat plane running laterally along the edge of the module, which provided a smooth surface for attachment during the mould making process (see Step 3; Figure 3B).

2) 3D Printing: We printed the resulting 3D-file using two types of extrusion-based printers, the Dremel Digilab 3D45 (Dremel DigiLab, Mt Prospect, United States) and the PRUSA MK 2 and 3 (Prusa Research, Prague, Czech Republic) printers. Both extrude PLA filament (polylactic acid, a common 3D printing filament) by building up consecutive layers of the 3D object, analogous to a hot glue gun extruding liquid plastic that hardens into a firm object. Both of these printers use 1.75 mm PLA filament, with a melting temperature of 175°C, and plate temperature of 60°C. Orienting the object so that the flat plane faced down onto the 3D printer’s build-plate reduced the support material required to hold the structure in place, minimizing print-time (Figure 3C ). Print time, or print speed, ranged between 3-5 h per module, depending on the printer and the number of modules loaded onto the build plate (i.e., total print time of 10–12 h if three modules were printed at once). Print set-up was an iterative process involving continual monitoring and adjustment for the first few print layers to ensure an established print base. Design flaws created during scanning and file manipulation (step 1) may only become apparent during the printing process (See Supplementary Information 1.0).

3) Mould Making: To create moulds from which to cast the artificial habitat modules, we first attached the 3D printed corals to a Plexiglass sheet using modeling clay. After coating the entire surface (plexiglass and coral modules) with spray-on Universal Mold Release (Smooth-On Inc., Macungie, United States), we brushed on the first layer of Dragon Skin ^® 10 Medium Series silicone (Smooth-On Inc., Macungie, United States) onto the 3D printed modules using a 1 cm round-tipped paint brush. This first silicone layer was mixed with a few drops of THI-VEX ^® silicone thickener (Smooth-On Inc., Macungie, United States) to help thicken silicone, ensure adhesion to 3D printed modules, and capture fine details of the organisms’ morphology in the mould (in our case, polyp-level features of the coral; Figure 3D). We poured the next three consecutive layers of silicone onto each mould according to the product’s mixing and pouring directions. The entire mould sat untouched in a cool dry area to set for 24 h. Next, we created a Plaster of Paris (Bondex International, Medina County, United States) casing over top by adding layers of heavy duty paper towel coated in a plaster slurry to give the flexible silicone under-mould structural stability (also called a mother mould/bandage shell mould) The entire mould and plaster casing sat untouched in a cool dry area for an additional 24 h -after which the 3D prints were carefully removed from the silicone layer, resulting in a mould with negative space where the 3D printed corals sat, and a firm support layer.

4) Casting: We used Quikrete ^® countertop mix (The Quikrete Companies, Atlanta, United States) concrete, water, and yellow pigment (Concrete Concrete, Edmonton, Canada) mixed by hand at a ratio of approximately 50:5:1 to approximate the colour of A. cervicornis. After lightly coating the moulds with vegetable oil (to act as a mould release) using a 1 cm round-tip paint brush, we filled and compressed each mould with the concrete mixture (Figure 3E). Since the silicone mold covered almost the entire 3D-printed coral fragment, the concrete filling is not visualized at the bottom of the mold as most of it is in the empty cavity created when the 3D print was removed. After a 24 h setting period, we carefully removed casts from the moulds and left them to cure further in a flat, dry area for 24 hrs minimum. These flexible and strong silicone moulds can be re-used multiple times for casting.

5) Assembly and Deployment: In our example study region, coral restoration projects typically transplant clusters of ‘tripod’ shaped A. cervicornis to reef environments to enhance coral cover because this shape provides multiple points of attachment to the benthos, increases stability, and exposes the coral to adequate water flow (Hollarsmith et al., 2012). We combined the concrete casts into complex 3D structures that mirrored this tripod shape and size (Figure 3F). Assembly involved carefully breaking some of the coral casts into two pieces to create coral “branches”, which were attached to original casts using Apoxie Sculpt Modeling Compound (Aves, Hudson, United States) and left to set for a minimum of 24 hrs (See Supplementary Information 4.1 for a video of this process). Modules were soaked in sea-water for a minimum of 7 days to leach out concrete-associated chemosensory cues before deployment, consistent with practices in other habitat selection studies using artificial habitat structures (Ruhl and Dixson, 2019). Modules were transported in large totes to the field site by boat and transferred underwater to experimental plots in milk crates by scuba divers (See Supplementary Information 4.2 for a video showing deployment process).

We applied artificial corals created via 3D-SPMC in a field experiment to evaluate structural and compositional cues driving juvenile reef fish habitat selection in Key Largo, United States from June-July 2019. Living corals were obtained by collaborators at the Coral Restoration Foundation from their Carysfort Reef offshore coral nursery and were cut to the same dimensions as the artificial coral fragments (12 cm³).The experiment involved placing artificial and living coral modules in replicate 1 m² clusters at consistent densities (10 corals/cluster) in three treatments representing different percentages of living coral (but equal structural composition) and in two environmental contexts: high complexity seascape (mean relief of the reef framework ≈ 3 m) and low complexity seascape (mean relief of the reef framework ≈ 0.5 m; N_total = 48 clusters). Divers attached habitat clusters to the bottom with the same epoxy used for module assembly (see step 5) at sites with no previous living staghorn corals present (Figure 4). Artificial habitat modules were equally easy to deploy compared with living coral fragments and required little additional diver training to affix to the benthos. The modules withstood transportation and under-water handling without damage and remained in place for the entire duration of the 2-month study without maintenance or repair.

Following deployment, we conducted SCUBA surveys of the abundance of newly recruited fishes (i.e., < 3 cm total length) at each cluster every 1–4 days over 50 days. Preliminary data visualization from our study suggests when structural features (i.e., habitat size and morphology) are held constant, recruiting fish have differing recruitment patterns over the first 50 days after habitat deployment (Figure 5A) and show a strong associations with habitats comprising 0 and 100% proportions of living coral, compared to disassociation with habitats comprising 50% living coral, with higher association to habitats in low complexity environments regardless of habtiat composition treatments. (i.e., low structural complexity, Figure 5B). These initial results suggest an interaction between habitat composition and environmental context, with implications for restoration site selection: increasing coral cover at low complexity sites may yield benefits that are not realized at high complexity locations in the same reefscape and further research may elucidate patterns of use for different species, trophic groups or trait groups during the first 50 days post-outplanting.

By combining desirable attributes from multiple methods for artificial habitat design and construction into a streamlined workflow, 3D-SPMC is likely to perform on par with or better than eight other individual AH materials and designs used in habitat selection studies (Tables 2, 3). Accessibility: Most of the materials and equipment required in steps 1 and 2 of 3D-SPMC are easily obtained through retail in urban centres; however the 3D printing components require special maintenance (Table 2). While the baseline costs associated with steps 1 and 2 (3D scanning and printing) of 3D-SPMC are high compared with other approaches (Table 3), the cost per AH module decreases substantially with increased production. For example, incorporating printer purchase and printing costs, ten 10 cm³ modules cost $19.37 each, compared to $0.43 each for 500 modules. Once made, 3D files and prints can be repeatedly edited, re-used, and/or shared, facilitating iterative designs and projects with little financial or time investment. A growing number of online tutorials and training centres offer free training in 3D scanning and printing, making the technology more broadly accessible (See Supplementary Information 1.2). While more costly than other moulding materials, Dragon Skin Series silicone material (step 3) is strong and elastic, meaning it can be used multiple times to create hundreds of replicates (Figure 3D).

Scalability: Scaling AH production to the research question and study design at hand requires modules that are durable, easy to deploy, and easy to reproduce (Bortone, 2006). We chose concrete for our casting material as it is durable in aquatic environments, including salt water environments, reducing the risk of collapse and potential for short- and long-term changes to the surrounding environment (Table 2). Other applications of concrete in AH construction result in modules that are large and non-modular, often requiring specialized equipment to deploy (e.g., reef balls; Sherman, 2002), or are created with inflexible moulds that are destroyed during the cast-release process, thus not re-useable (Table 2). Modularizing the AH structure into simple planar components (step 1) that can later be assembled into replicate, complex 3-dimensional configurations (step 5) makes the 3D-SPMC method more scalable -a major benefit compared to 3D printing complex modules (Supplementary Figure S1).

Ecology: While existing designs using concrete only score moderately well in terms of structural realism (Figure 2, Table 2), the application of 3D scanning and printing (steps 1 and 2) within 3D-SPMC facilitates the creation of biogenic habitat replicas with a high degree of morphological detail. Once cast, concrete is chemically neutral, and thus less likely to cause unanticipated chemosensory stimulation of target and non-target organisms (and less environmental harm). While some previous 3D printing methods to create artificial habitat modules have used biodegradable PLA filament (Tarazi et al., 2019; Wolfe and Mumby, 2020), 3D-SPMC avoids this potential source of aquatic plastic pollution, and mitigates against unintentional chemical cues leaching into the environment.

Restoration planning has only recently started to include metrics of ecosystem function (Suding and Hobbs, 2009), and/or focus on key species that may accomplish broader co-beneficial goals of restoration (Peterson et al., 2003; Jones and Davidson, 2016; Ladd et al., 2018). Information on key structural and compositional attributes of biogenic habitats that promote habitat selection and use by resident species is essential for conservation and restoration planning; manipulative field and laboratory experiments using this method to create artificial habitats (such as in the experiment described above) can provide insights into the composition and optimal placement of biogenic habitats to bolster ecosystem function and sustainability (Ferrario et al., 2016). As seen in our case study, habitat composition affects fish recruitment, but environmental context also mediates the direction and magnitude of ecological response. Crucially, we only detected this effect by using 3D-SPMC to create modules for an experimental design in which we replicated a high degree of morphological realism across habitat modules varying in material composition (live coral vs concrete). In some aquatic ecosystems, employing artificial structures in restoration itself has resulted in increased fish recruitment (Harding and Mann, 2001; Green et al., 2015) however they are no substitute for the suite of ecosystem services created by living biogenic habitats (e.g., Côté and Darling, 2010; Bruno et al., 2019; Supplementary Information 3.0).

A persistent challenge for testing habitat selection cues has been designing habitat modules that enable researchers to isolate and manipulate structural and compositional attributes, thus disentangling their relative influence on the habitat selection process (Harborne et al., 2011; Coker et al., 2012). 3D-SPMC offers a flexible means to design AH modules that manipulate structure and composition of focal biogenic habitat-forming organisms, and could be employed to provide insights into factors affecting habitat selection by resident biota in environments ranging from coastal oyster reefs and mangroves (Ellis and Bell, 2004; Beck et al., 2011), to woody vegetation in freshwater bodies, to mesophotic glass sponge reefs (Dunham et al., 2018).

Scale is an important consideration when designing AHs for selection cue studies; the spatial and temporal scale of the ecological process being examined and the type(s) of cues to be manipulated influences decisions around study type (i.e., in situ vs. ex situ) and duration, focal habitat size and configuration, ecosystem connectivity, placement and response variable selection (see Supplementary Information 5.0 for a list of key research design questions the researcher should address in selecting appropriate AH configuration). We used coral fragment-sized modules in dense clusters at a spatial scale which fish recruitment is likely to vary, that mimics the design of reef restoration projects, and because recent evidence suggests fine-scale morphology affects larger-scale ecological processes (Urbina-Barreto et al., 2020). We also chose to measure colonization processes (recruitment and retention) that occur at the temporal scale most artificial habitat selection studies typically investigate, however metrics could be expanded to include longer temporal-scale processes like reproduction which could affect the overall abundance of organisms available to be recruited and retained to habitats over longer periods of time. However, 3D-SPMC could be used to create larger and/or more complex habitat patches by generating larger 3D prints (and mould/casts), increasing habitat complexity in the assembly phase by combining multiple casts into a single module, and/or combining multiple habitat modules into a final structure. While the effect of patch size on species colonization and habitat use has been relatively well studied compared to habitat composition (Bohnsack et al., 1994) our method allows researchers to incorporate both aspects into their study design to evaluate habitat composition characteristics.

By manipulating the composition of casting material (step 4), researchers can use 3D-SPMC to study compositional features hypothesized to affect chemosensory stimulation. For example, studies aiming to study predator, prey, and competitor detection by focal organisms could directly incorporate homogenized tissue, body fluids, or key chemical components (e.g., pheromones, hormones) of con- and hetero-specifics into the casting material. One could also test multiple recruitment cue responses (Huijbers et al., 2012) by deploying AHs in combination with other cues (i.e., acoustic cues).

3D-SPMC could also be used to study epi-biotic habitat colonization by invertebrates like corals, sponge, or oyster spats by altering the configuration and/or casting composition of the AH modules. For example, oyster shells incorporated into structure provided attractant cues to larval oyster and saw higher spat recruitment (Ortego, 2006). Ceramic modules with tighter surface-pore densities may reduce biofouling and/or enhance targeted species-specific settlement (Johari et al., 2010). Companies are already creating “ecologically active” concrete materials that modify composition and surface texture to support specific marine fauna and flora (Perkol-Finkel and Sella, 2014), lowers the carbon footprint of artificial habitat construction (Dennis et al., 2018), and addresses the concern of concrete waste in aquatic ecosytems (Cooke et al., 2020). One could even consider expanding and adapting this method to test biofilm or anti-biofouling coatings that reduce or promote targeted biotic build-up (Tamburri et al., 2008).

This method can also be adapted to evaluate the structural characteristics of soft-bodied biogenic habitat-forming organisms such as sea-fans and seagrasses by using flexible casting material or 3D printing using flexible or biogenic printing material (Yirmibesoglu et al., 2018; Wangpraseurt et al., 2020), with implications to bio-mechanic studies and the contribution of biogenic organisms to shoreline protection (Christianen et al., 2013). Note that specialized printers may be a more expensive option that may limit affordability and accessibility.