The experimental results indicated four ways to achieve high survival of Acropora digitifera corals over an eight-month seeding trial: (1) deploy devices with fish-exclusion features; (2) orient corals in the side rather than top positions; (3) use differential survival data to inform appropriate microfragment and spat seeding densities to achieve target numbers; and (4) use ecological variables, like fish abundance, fish diversity and the abundance of nutrient-dense food sources for fish, to help select seeding sites. These four findings can be used to guide coral seeding methods and manage restoration outcomes.
Corals deployed without protection were the most vulnerable to grazing and suffered significantly higher mortality compared to those with protection. Complex device topographies, including crevices, microrefugia and those with high rugosity, have previously been shown to improve coral larval settlement and post-settlement survival11,23,24,30. Thus, it was not unexpected that the engineered seeding device with protective walls was also suitable for coral survival after seeding. Although the caged device still resulted in higher survival of coral microfragments in our study the built-in protection provided by a fish-exclusion device offers a simple and effective solution to grazing that is cheaper and far more scalable than cages. Furthermore, cages are ineffective against fish grazing after removal33. Nevertheless, there is room to improve the design of devices with features to offer more protection and further increase benefits over caged devices.
The orientation of corals in devices had a major influence on post-seeding survival. Coral larvae often preferentially settle on vertical surfaces to reduce the likelihood of disturbance-induced mortality during recruitment63–66. Our findings suggest that corals (especially spat) deployed on top-facing plugs suffer higher rates of grazing and mortality than those in side-facing positions. Many species of parrotfish prefer to feed on horizontal and convex surfaces over vertical and concave surfaces67. Therefore, the vertical orientation of the side-facing plug may offer optimal refuge from some grazers. Our grazing results support the selection of vertical habitats by corals. However, there are also reasons beyond fish predation as to why coral larvae prefer settling in refugia. Coral spat are sensitive to high light, heavy sedimentation and excessive waterflow68, and vertically-orientated structures provide a buffer against such conditions. Fast growing turf, crustose, and macroalgae, that tend to out-compete corals in high-light and sediment-laden environments, can also be less abundant and display different morphologies (e.g., thin- versus thick-crusted CCA) in low-light environments39,69, relieving some competitive pressures. Regardless of the mechanisms, offering vertically oriented refugia for corals was a successful strategy for improving survival.
It was clear that coral life stage impacted survival after seeding, with significantly more A. digitifera microfragments surviving than spat. Spat survival was low (< 50%) on featureless devices past 2 days, and even with device protection. Low survival of spat is common in coral seeding experiments to date. For example, in French Polynesia, < 40% of A. striata spat survived beyond one-week post seeding70 due to grazing-induced mortality, with those in cages performing best. Similarly, in the Philippines, survival of A. tenuis spat reached 25% with partial cages and only 20% with full cages after three months28. On the GBR, 38–65% of A. cytherea spat survived on reefs for one month, with caging increasing survivorship by 22%27. Similarly, partial protection of A. tenuis spat in microrefugia led to 22–39% coral yield (e.g., device-level survival) after one year on the GBR11,12. Taken together, protection from fishes can increase the chance of coral survival during early ontogeny. However, spat survival typically remains well below 50%, and we suspect that: (i) more spat per device need to be deployed to meet required survival outcomes, (ii) other device types should be trialed to optimize the design of protective features, and (iii) further ecological drivers should be studied to gain knowledge that will optimize conditions for spat success after seeding.
While microfragmentation has almost exclusively been performed on corals with plating, massive and sub-massive growth morphologies18,19, our results indicate that the technique is also feasible when applied to branching taxa such as Acropora. Survival of A. digitifera microfragments was exceptionally high, greater than 75% on average and up to 97% in cages. These findings, alongside the result of previous experiments30, confirm the benefits of physical barriers to enhance the survival of microfragmentated corals. In Palau, small fragments of Porites lobata had higher survival when fully protected (> 90%, four-sided crevices on tiles) compared to the result of partial crevices (70%) and exposed controls (28%) after 29 days30. In the same study, Pocillopora damicornis fragments also initially benefited from protection; however, all had died after eight days. These outcomes indicate that tolerance to microfragmentation may be species specific and this will have flow-on effects to survival. Similarly, other factors including fragment health might dictate the likelihood of predation post deployment, with recently dead or dying tissues being more attractive to fishes71. The A. digitifera tested in our study may be more tolerant to microfragmentation and less palatable to fishes, making it a target candidate for seeding over other taxa such as Pocillopora. To ensure positive outcomes after seeding, adequate time should be provided for corals to heal after the microfragmentation process, and the palatability of a range of coral species should be examined.
We found that coral spat were also targeted for grazing more often than microfragments. Parrotfish, mainly Scarus globiceps, were identified as the most common and destructive grazers of seeded corals, especially spat. However, most scraping parrotfish are nominally herbivorous and therefore we hypothesize that the grazing-induced mortality of coral spat in our study is likely accidental and indirect. The morphological description of the jaw of a scraping parrotfish like S. globiceps, confirms their ability to graze early successional bacterial-algal biofilms growing on reef substrata67,72. The artificial plugs used to grow corals in our study were conditioned with similar biofilms (e.g., thin-crusted CCA, microalgae and mixed bacteria) and these are known to be palatable to many herbivorous parrotfishes73,74. Thus, it is likely parrotfish were targeting the benthic community growing on the plug rather than the coral itself. Furthermore, spat plugs were held in aquaria longer than microfragment plugs prior to deployment; this likely contributed to significant differences in grazing among the coral life stage treatment. The use of less palatable benthic organisms for plug conditioning and the incorporation of an antifoulant (i.e., chemical deterrent) on plugs or devices are potential direct and indirect solutions to deter parrotfish grazing. However, we suggest these methods must be balanced against the requirement for benthic communities that: (i) are inductive for the larvae during settlement, (ii) reduce competition against spat and microfragments during recruitment, and (iii) promote coral survival to maturity.
Lastly, the influence of grazing on survival is strongly site-specific, and we may be able use this information to guide the selection of sites for successful seeding. Foremost, our results support growing evidence that herbivorous parrotfish may preferentially target nutrient-rich food sources (e.g., algal mats or turfs dominated by cyanobacteria), when available75,76. However, on reefs with high coral cover and low abundance of cyanobacteria, opportunistic grazing on less preferred food sources such as deployed corals may occur. For example, Site 2 exhibited the least grazing on coral plugs but the highest number of parrotfish bites in experimental plots, and concurrently, a notably higher (12 x) abundance of mat-forming cyanobacteria. Parrotfish consumption of nutrient-dense foods, such as endolithic algae and cyanobacteria, has been documented in the Caribbean75 and the northern GBR76. Field observations performed by Cissell et al (2019) found that greater than 15% of the diet of parrotfishes S. iseri and S. coeruleus consisted of mat-forming cyanobacteria75. Similarly, stable isotope analysis of 22 species of GBR-dwelling parrotfish provided strong evidence to support the consumption and preference of microscopic cyanobacteria over other algal substrates during feeding76. From this, we propose a new hypothesis that nutritious foods like cyanobacteria can offer natural forms of refuge to corals from grazing after seeding.
En masse, deploying to sites with high-nutrient food sources for fish may limit both direct and indirect grazing pressure on corals, increasing the likelihood of successful restoration. However, the composition of modern reefs is expected to shift as the frequency of mild to severe disturbances increases over time. For example, sudden spikes in the abundance of mat-forming cyanobacteria are common characteristics of recently disturbed reefs77,78. We suspect slight increases in cyanobacteria (< 20% in 0.25 m2) can aid coral survival after seeding; however, too much cyanobacteria can negatively impact coral dominance on disturbed reefs. For A. digitifera spat, we found that a minimum cover of closely related Scleractinian corals (10%) and bare rock (20%) may reduce the impact of grazing by 10 to 50%, respectively. Indeed, other benthic constituents, like sponges and epilithic algal communities, have been correlated with high survival of seeded corals on the GBR12. Therefore, a diverse group of grazing fishes at a minimum biomass will be required to mediate competition between corals and fast-growing benthic organisms after seeding. A relatively low biomass of herbivores (177 kg ha [17.7 g m2]) sustains community equilibrium and reduces the risk of regime shifts occurring in the Indian Ocean79. Defining a biomass threshold for parrotfish (e.g., where the net benefit towards coral competition outweighs the negative effects of accidental grazing), represents one tool that can be used to guide or tailor deployments of corals to specific sites. The direct manipulation of herbivore abundance could also be considered at sites with significant regime shifts. To inform the selection of candidate sites for reef restoration a more holistic approach to coral seeding should be explored.